United States
Department of
Agriculture
Forest Service
July 1994
Miscellaneous
Publication 1520
EPA - 8?1 -0-9V- -00FB
Evaluating the Effectiveness
of Forestry Best Management
Practices in Meeting Water Quality
Goals or Standards
-------
This report has been reviewed by the Environmental Protection Agency as a technical reference document. Review and
publication do not signify that the contents necessarily represent or reflect the policies and program guidance of the
Environmental Protection Agency. Any trade names or product names mentioned in this publication do not imply
endorsement by authors or the sponsoring institutions.
The United States Department of Agriculture (USDA) prohibits discrimination in its programs on the basis of race, color,
national origin, sex, religion, age, disability, political beliefs, and marital or familial status. (Not all prohibited bases apply
to aN programs.) Persons with disabilities who require alternative means for communication of program information
(Braille, large print, audiotape, etc.) should contact the USDA Office of Communications at (202) 720-5881 (voice) or (202)
720-7808 (TDD).
To file a complaint, write the Secretary of Agriculture, U.S. Department of Agriculture, Washington, D.C. 20250, or call
(202) 720-7327 (votes) or (202) 720-1127 (TDD). USDA is an equal employment opportunity employer.
July 1994
-------
Evaluating the Effectiveness
of Forestry Best Management
Practices in Meeting Water Quality
Goals or Standards
George E. Dissmeyer
USDA Forest Service
Southern Region
1720 Peachtree Road NW
Atlanta, GA 30367
-------
-------
Contents
iv List of Figures
v List of Tables
vii Acknowledgements
1 Introduction
3 Background
5 Planning Monitoring Projects
15 Quality Assurance/Quality Control
19 Statistical Considerations
Statistical Design in Water Quality Monitoring
General Design and Replication
Benefits of Proper Statistical Design
Design Problems and Constraints
Principles of Sampling
Principles of Statistical Testing
Assumptions and Distributions
Statistical Compromises
Sample Size
Variability
Level of Significance
Power
Minimum Detectable Effect
41 Selecting the Appropriate BMP Effectiveness Monitoring Level
Monitoring Levels
Monitoring Objectives
Skills Needed
Reference Condition
Amount of Data Collected
Number of Streams Evaluated
Length of Study
Quality of Data for Decisions
Quality of Decisions and Risk
57 Monitoring Methods
On-Slope Monitoring Methods
On-Slope Methods by Monitoring Level
BMP Implementation I
Florida's BMP Implementation Monitoring
BMP Implementation II
BMP Implementation III
Montana's Forestry BMP Implementation Monitoring
Best Management Practices Evaluation Program
Erosion Indicators
Universal Soil Loss Equation
Fabric Dams
-------
Flumes and Samplers
Rainfall Simulators
Herbicide Cards
Herbicide Indicators
Chemical and Physical Monitoring Methods
Chemical and Physical Methods by Monitoring Level
Channel Geomorphology Monitoring Methods
Channel Geomorphology Methods by Monitoring Level
Empirical/Perceptual
Pfankuch Channel Stability
Rosgen's Stream Channel Classification
Alaska Channel Classification
V* Pool Sediment
Rosgen Channel Monitoring
Pebble Counts
Habitat Monitoring Methods
Habitat Monitoring Methods by Monitoring Level
Empirical/Perceptual
Rapid Bioassessment Protocol—Habitat
Environmental Monitoring and Assessment Program
Streamwalk
T-Walk
Bisson Habitat
Hankin-Reeves Habitat
Basinwide Estimation of Habitat
USDA Forest Service R-5, R-1, R-4 FHR Stream Habitat Classification
and Inventory Procedures
Ohio-QHEl
Platts Riparian Habitat
Biomonitoring Methods
Biological Monitoring Methods by Monitoring Level
Empirical/Perceptual
Rapid Bioassessment Protocol I
Rapid Bioassessment Protocol II
Rapid Bioassessment Protocol III
Index of Biotic Integrity—Fish (RBP V)
Hankin-Reeves
Izaak Walton League
Vermont Guide to Macroinvertebrate Sampling
Basin Area Stream Survey
Ohio EPA Macroinvertebrate Sampling
Ohio EPA Fish Sampling
Snorkel Methods for Fish
Streamwalk
Streamwalk II
Hilsenhoff Biotic Index
North Carolina Standardized Ten Sample Method
North Carolina Rapid Bioassessment Method (Four Sample)
Aquatic Vegetation and Zooplankton Monitoring Methods
Periphyton and Macrophyte Monitoring Methods by Monitoring Level
Protocol I - Periphyton
Protocol II - Periphyton
-------
Protocol III - Periphyton
Protocol I - Aquatic Macrophytes
Protocol II - Aquatic Macrophytes
Protocol III - Aquatic Macrophytes
Montana Periphyton Protocol I
Montana Periphyton Protocol II
153 Deciding on BMP Effectiveness: Some Case Histories
Implementation and Effectiveness Monitoring in South Carolina
A Water Quality Criteria Approach
Multi-Factor Approach/Stream Continuum
BMP Effectiveness by Stream/Ecosystem Class
An Example of BMP Effectiveness I Monitoring
Determining Effective Life of Best Management Practices
Major Storms and BMP Effectiveness
174 References
179 Appendix
-------
List of Figures
Figure 1. Development of a monitoring project
Figure 2. Schematic representation of the trade-offs among levels of signifi-
cance, power, and variability for two normally distributed populations
Figure 3. Maximum allowable coefficient of variation to detect changes
Figure 4. Ecological integrity
Figure 5. Reference reach channel classification similar to study reach
Figure 6. Above and below monitoring using reaches with the same stream
channel classification
Figure 7. Before and after comparison on the same stream reach
Figure 8. The area of discernment of water quality condition by BMP effective-
ness monitoring level
Figure 9. Example form developed for the BMPEP monitoring program
Figure 10. Fabric dam construction
Figure 11. Rosgen Class A Streams
Figure 12. Rosgen Class G Streams
Figure 13. Rosgen Class F Streams
Figure 14. Rosgen Class B Streams
Figure 15. Rosgen Class E Streams
Figure 16. Rosgen Class C Streams
Figure 17. Rosgen Class D Streams
Figure 18. Example of a permanent cross-section with bench mark locations
and points of measurement
Figure 19. Measuring bank erosion with bank pins
Figure 20. Bank profile procedure using toe pin and survey rod
Figure 21. Comparison of relative elevations for cross-sections for a 2-year
period
Figure 22. Combined cross-section and bank erosion study for a 2-year period
Figure 23. Comparison of two hypothetical cumulative sediment size curves
between a reference and study stream
Figure 24. Effectiveness of logging practices in protecting benthic
macroinvertebrates and habitat
Figure 25. Benthic macroinvertebrate response to logging with and without
BMP's, Piedmont, S. Carolina
Figure 26. Monitoring effectiveness along a stream continuum
Figure 27. Map of Silver Creek watersheds
Figure 28. Relationship between percent substrate fines and BMP effective-
ness rating for sediment control, Silver Creek watersheds
Figure 29. Relationship of percent EPT to percent fine sediment in substrate,
Silver Creek watersheds
Figure 30. Relationship of percent Peltoperlidae to percent fine sediment in
substrate, Silver Creek watersheds
Figure 31. Relationship of percent Chironomidae (Midges) to percent fine
sediment in substrate, Silver Creek watersheds
IV
-------
List of Tables
i
Table 1. Qualitative assessment of the effects of water quality parameters on
the major designated uses of water from forested watersheds
Table 2. Sensitivity of water quality parameters to management activities
Table 3. Effectiveness monitoring process matrix
Table 4. Factors that can influence the physical, chemical, and biological
integrity of water
Table 5. Monitoring methods by effectiveness monitoring level
Table 6. Potential subfactors by disturbance category
Table 7. The 14 forest resource management water-quality-related para-
meters covered in this discussion
Tables. RBP Habitat Rating
Table 9. Comparability Assessment
Table 10. Metric Scores (IBI)
Table 11. Index Score Interpretation
Table 12. Invertebrate Community Index (ICI) metrics and scoring criteria
based on macroinvertebrate community data from 247 reference
sites throughout Ohio
Table 13. Tolerance to pollution—Common aquatic insects for Idaho (toler-
ance is based on the Hilsenhoff Biotic Index)
Table 14. Methods used for the standardized qualitative collection techniques
Table 15. Spearman Rank Correlations—BMP effectiveness in the Silver
Creek watersheds
Table 16. Spearman Rank Correlations—Percent fine sediment in the Silver
Creek watersheds
Table 17. Management and BMP effectiveness evaluation of the 6 developed
and the control Silver Creek watersheds. Bold values indicate
departure from reference conditions
v
-------
-------
Acknowledgements
The National Association of State Foresters, the Washington Office of the USDA
Forest Service, and the Headquarters of the U.S. Environmental Protection Agency
all played vital roles in the development of this methodology. The Water Re-
sources Committee of the National Association of State Foresters, under the
leadership of Gary Brown, Retired State Forester in Montana, identified the need for
this methodology to help State Foresters answer questions concerning the effec-
tiveness of best management practices in meeting water quality goals. The Forest
Service and the Environmental Protection Agency also recognized the need to
develop this methodology. Therefore, the joint sponsorship of these three organiza-
tions led to the initiation of this project. The project was funded by the Washington
Offices of the Forest Service and the Environmental Protection Agency, which was
greatly appreciated and without which the project could not have been'done.
There was a wide base of support for the project as reflected in the composition of
the Steering Committee and the Task Force. The project has required substantial
commitment of time and effort from representatives for forest industry, state water
quality agencies, State foresters, national forests, Forest Service Research, State
fisheries agencies, the EPA, State and national forestry associations, and universi-
ties. The interest and support from these people made this project possible and
was greatly appreciated.
The Steering Committee provided guidance to the Task Force by outlining their
expectations for the project and guidance to avoid conflicts with existing policy and
direction concerning water quality. The Steering Committee members included
Gary Brown, State Forester, Missoula, Montana; Stan Adams, State Forester,
Raleigh, North Carolina; Jeff Vowell, Hydrologist, Florida Division of Forestry,
Tallahassee, Florida; George Gibson, U. S. Environmental Protection Agency,
Washington, D.C.; John Cannell, U. S. Environmental Protection Agency, Washing-
ton, D.C.; George Ice, National Council of the Paper Industry for Air and Stream
Improvement, Corvallis, Oregon; Warren Harper, Watershed and Air Management
Staff, Forest Service, Washington, D.C.; Gordon Stuart, Cooperative Forestry Staff,
Forest Service, Washington, D.C.; Robert Zimmerman, Delaware Surface Water
Management, Dover, DE; Robert Olszewski, Georgia Pacific, Atlanta, GA; and
Mitchell Dubensky, API/NFPA, Washington, D.C.
The Task Force was composed of representatives from various regions of the
country and from various disciplines. The Task Force consisted of hydrology,
fisheries, benthic macroinvertebrate, channel geomorphology, aquatic vegetation,
and forestry professionals and researchers. The Task Force worked together to
reach a consensus on the approach, supplied literature and methods, and reviewed
and commented on the methodology to this author. The Task Force support and
input was greatly appreciated and without it, this product would not have been
possible. The Task Force members were Jeff Vowell, Hydrologist, Florida Division
of Forestry, Tallahassee, Florida; Dr. George Ice, National Council of the Paper
Industry for Air and Stream Improvement, Corvallis, Oregon; Alan Clingenpeel,
Hydrologist, Ouachita National Forest, Forest Service, Hot Springs, Arkansas; Glen
Chen, Aquatic Monitoring Center, Forest Service, Logan, Utah; Bill Schultz, Hy-
drologist, Montana Forestry Division, Missoula, Montana; Steve Filipek, Research
Fisheries Biologist, Arkansas Game and Fish and representing the American
vii
-------
Fisheries Society, Little Rock, Arkansas; Dr. James Hornbeck, Hydrologist, Forest
Service, Northeastern Forest Experiment Station, Durham, New Hampshire; Jim
Lazorchak, Chief Aquatic Biology Branch, U.S. Environmental Protection Agency,
Cincinnati, Ohio; Dr. Kate Sullivan, Geomorphologist, Weyerhaeuser Company,
Tacoma, Washington; Jim Maxwell, Hydrologist, Forest Service, Denver, Colorado;
Danny Ebert, Fisheries Biologist, Staff Officer, Boise National Forest, Boise, Idaho;
Tim Burton, Fisheries Biologist, Boise National Forest, Boise, Idaho; Robert
Olszewski, Georgia Pacific Corp, Atlanta, Georgia; Dr. Wade Nutter, Hydrologist,
University of Georgia, Athens, Georgia; John Cannell, U.S. Environmental Protec-
tion Agency, Washington, D.C.; Chris Yoder, Division of Water Quality Planning
and Assessment, Ohio EPA, Columbus, Ohio; Dr. Lee MacDonald, Hydrologist,
Colorado State University, Fort Collins, Colorado; Mike Kuehn, Hydrologist, Forest
Service, Juneau, Alaska; Steve Bauer, Consultant, Boise, Idaho; and Steve Foster,
Alabama Department of Environmental Management, Montgomery, Alabama.
In addition to the Task Force, appreciation is extended to Dave Rosgen, Hydrolo-
gist, Wildland Hydrology, Pagosa Springs, Colorado, for supplying a requested
chanhel erosion methodology and latest information of channel classification.
viii
-------
Introduction
Purpose
The purpose of this document is to help forest managers and their staffs develop
water-quality monitoring plans to evaluate the effectiveness of forestry best
management practices (BMP's) in meeting water-quality goals or standards in
the United States. The term "water quality" refers to the chemical, physical,
biological, and habitat condition of streams. Approaches and methods described
here can also be used to evaluate the effectiveness of grazing, mining, and other
nonpoint source (NPS) best management practices.
The document is a reference, not a national protocol. It deals with the design of
monitoring projects and the selection of variables and methods for monitoring
them given (1) the designated beneficial use, (2) the type and intensity of man-
agement, (3) the stream/ecosystem setting, and (4) monitoring objectives or
questions related to NPS pollution. The methods apply only to streams of the 4th
order and smaller that can be waded during low-flow periods.
With a few exceptions, forest management affects naturally occurring water-
quality variables rather than in the introducing unnatural pollutants. This docu-
ment describes monitoring projects designed to isolate the effect of forest man-
agement from other nonpoint sources of pollution and determine whether BMP's
or management activities meet NPS goals or standards. BMP's are practices
designed to meet water quality goals or standards. Some forest treatments have
not been identified as BMP's and are called management activities in this report.
The specific objectives are to: (1) ensure that monitoring projects consider all
factors influencing NPS pollution, (2) design monitoring systems that isolate the
impacts of BMP's or management activities, (3) review variables to consider
when monitoring BMP effectiveness, (4) summarize representative methods
appropriate for monitoring NPS variables, and (5) provide some examples of
BMP effectiveness monitoring and decisions.
The document is intended for State foresters, managers of national forests, forest
industry, State water quality agencies, State conservation agencies, universities,
consultants, and any other organizations or individuals involved in evaluating the
effectiveness of forest management in meeting water-quality goals or standards.
-------
-------
Background
In 1972 Congress passed the Federal Water Pollution Control Act (P.L. 92-500),
commonly called the Clean Water Act. The Clean Water Act's goal is to restore
and maintain the chemical, physical, and biological integrity of the Nation's
waters. An interim goal of the Act is the protection and propagation of fish,
shellfish, and wildlife. Propagation includes the full range of biological conditions
necessary to support reproducing populations of all forms of aquatic life and
other life that depend on aquatic systems (EPA 1990). EPA's response to this
Act is an ecological approach to water-quality monitoring, including monitoring
habitat condition.
Under this Act, States have established water-quality standards for streams and
other water bodies. Standards vary among States, but consist of a designated
beneficial use, criteria to protect the beneficial use, and an antidegradation policy
statement.
The States were to adopt narrative biological criteria into their water-quality
standards during the period 1991-93 (EPA 1990). EPA expects numeric biologi-
cal criteria to be adopted in the next few years and is in the process of refining its
guidance for biological criteria for streams (EPA 1991). EPA has issued guid-
ance for monitpring the physical, chemical, and biological integrity of streams
(MacDonald et'al. 1991, Plafkin et al. 1989, and EPA 1990). The EPA guidance
clearly demonstrates an aquatic ecosystem approach to standards and monitor-
ing, including the use of integrated physical, chemical, and biological assessment
techniques.
s
BMP's effectiveness monitoring will determine if BMP's protect water quality, that
is, yield water that does not contribute to degradation of water quality in the
receiving stream. BMP's will be designed, implemented, monitored, and revised
(if needed) to assure protection of water quality. In order to evaluate the effec-
tiveness of forestry BMP's, many factors need to be considered.
Recent EPA guidance for monitoring the physical, chemical, and biological
integrity of water will directly influence monitoring in forest streams, including the
selection of reference sites (streams), which are assessed to develop a reference
condition that will provide an attainable measure of ecological health. The
reference condition defines the range and variability of physical, chemical, bio-
logical, and habitat conditions. It is against the reference condition that the
health of other streams is compared.
A stream/ecosystem classification system is needed to ensure proper compari-
son of study streams with reference streams. Many factors must be evaluated to
determine if forest management is impacting water quality and if water quality
goals or standards have been met.
Most NPS responses to forestry practices occur during high rainfall or snowmelt
runoff events, yet impacts may occur during low-flow periods. Suspended sedi-
ment, pesticides, and other pollutants are discharged from the stream system
during high streamflow. However, some pollutants settle out during waning
stages of runoff and accumulate in the streambed—for example, fine sediments,
soil aggregates, and attached pollutants.
-------
Forest management should protect the designated beneficial use: namely cold-
or warm-water fisheries; recreation; and municipal water supply; or potentially
endangered, threatened, or sensitive aquatic species. Usually, cold- and warm-
water fisheries will be the most limiting beneficial use. States are developing
biological criteria, including habitat, for protecting this beneficial use. To ensure
comparability of present and future monitoring data with standards, forestry
should use the same monitoring and analysis methods as the State.
By definition, NPS's are diffuse, coming from the landscape in a dispersed
fashion over the surface or through subsurface flow. Therefore, for forest NFS,
monitoring methods must determine if what is measured or observed in the
stream is influenced by BMP's or management activities on the land. This will
require on-land monitoring to develop the linkage. Part of this analysis should
explain why in-stream conditions change or remain unchanged.
On-land monitoring includes: (1) evaluating the implementation of BMP's,
(2) evaluating the effectiveness of BMP's in reducing erosion and sediment
leaving the slopes, (3) measuring the amount of solar radiation through the
riparian canopy, (4) determining if pesticide drift has reached the stream, and
(5) determining if logging slash has entered the stream. Models, such as the
Universal Soil Loss Equation, can be used to compare on-slope erosion between
treatments or BMP's and help explain in-stream responses.
Monitoring in-stream conditions above and below a management area might
reveal a change in aquatic life and habitat. Does this change fall within the range
defined by the reference condition? Is this change really linked to management
on the slope or is it natural variation in abundance of aquatic life and habitat
quality along the stream? Is it a result of stream-channel morphology changes,
effects of prior land use, etc.? Some of these questions can be answered by
proper selection and classification of reference streams within ecosystems. The
rest must be addressed in a monitoring project designed to link on-slope activi-
ties with in-stream changes in physical, chemical, and/or biological variables.
To sum up, effectiveness monitoring answers the questions: Did we get the
desired result? Did we protect water quality? Did BMP's meet specific water-
quality criteria? Have BMP's been selected, implemented, and maintained
properly?
Finally, the accepted practice in water-quality assessment is to use the term
"parameter" to refer to chemical and physical constituents of water. In this book,
the term 'Variable" is used in place of "parameter" because many channel mor-
phology, habitat, etc., variables are added to the list of potential measurements
to determine the physical, chemical, biological, and habitat conditions. It is best
to call all these measurements 'Variables." However, some excerpts from meth-
ods will use the term "parameter."
-------
Planning Monitoring Projects
Effectiveness monitoring in this document seeks to answer five questions:
1. Do properly selected, implemented, and maintained BMP's meet
water-quality goals or standards? (Various levels of BMP can be installed
and monitored to assess in-stream physical, chemical, and biological
responses.)
2. Does a management activity jeopardize water quality?
3. Which management system best protects water quality?
4. How long do BMP's remain effective?
5. How effective are BMP's under major climatic events?
The acceptance or rejection of the null hypothesis is basic to all five questions.
Some examples of null hypotheses include:
- BMP's produce water meeting temperature criteria.
- Benthic invertebrates populations below logged areas with BMP's are the
same as benthic invertebrate populations in reference streams.
- The,re is no difference in aquatic habitat between the study stream seg-
ment and reference stream.
- All management activities with BMP's protect water quality equally.
The basics for monitoring project planning will be discussed first, followed by
some comments appropriate to each of the five questions.
MacDonald et al. (1991) provides a good summary of key steps for designing
and implementing water-quality monitoring projects:
Although the definition of specific steps in developing a monitoring projects
tends to vary according to the author of the guidelines and the particular
monitoring situation, the key steps are as follows:'
- propose—together with the managers—the general objectives;
- define the approximate budget and personnel constraints;
- review existing data;
- determine monitoring parameters, sampling locations, sampling proce-
dures, and analytic techniques;
- evaluate hypothetical or real data;
- reassess monitoring objectives and compatibility with existing re-
sources;
- initiate monitoring activities on a pilot basis;
- analyze and evaluate data;
- reassess monitoring objectives and compatibility with existing re-
sources;
- modify monitoring project as necessary;
- continue monitoring;
- prepare regular reports and recommendations.
Figure 1 is a schematic representation of these key steps, and it also indi-
cates some of the critical feedback loops in. developing and implementing a
water quality monitoring project. In most cases, however, the key steps are
not nearly as distinct and sequential as indicated in Figure 1. Decisions
-------
made at each step often have repercussions for the entire monitoring project,
and sometimes this may force a reassessment of previous steps. For ex-
ample, preliminary identification of the possible sampling locations may
necessitate a review of the budget constraints or monitoring objectives.
Hence the feedback loops shown represent only the most critical pathways,
and each step may not always be completed in the order indicated. What is
essential is that each key step be explicitly addressed, and the sequence
indicated in Figure 1 is one approach to optimize the process of developing a
monitoring project.
The first step is to identify the general objectives, and this best done by the
managers in consultation with the technical staff. Once the general objec-
tives have been determined, the approximate personnel and budgetary
constraints must be specified in order to ensure that the subsequent monitor-
ing plan is realistic. The availability of past data also must be assessed. If
past data are available, it may be possible to evaluate changes over time
provided the same measurement techniques and sampling locations are
employed. If past data are unavailable, changes probably will have to be
assessed by site comparisons, and this often leads to greater flexibility in the
selection of both the monitoring parameters and the sampling locations.
The next step is to formulate the specific objectives. This requires the partici-
pation of both the managers and the technical staff in order to ensure that
the specific objectives are technically and financially feasible. The impor-
tance of this interaction is often overlooked, and a failure in communication
can lead to a variety of problems. For example, if the manager is unaware of
the potential benefits of the monitoring project, obtaining the necessary
resources to carry out the project may be difficult. Alternatively, if the techni-
cal specialist does not listen to the manager, the specialist may design a ,
monitoring project that will not provide the necessary guidance for manage-
ment decisions. Input from both the managers and the specialists is needed
to balance the need for more data and the cost of acquiring the data. Both
sides also must be explicitly aware of the risks and uncertainties associated
with monitoring in a highly variable environment.
Often the technical specialist will need to take the lead role in formulating the
specific objectives because the specialist will be more familiar with previous
monitoring efforts and the likely impacts of management activities on water
quality and aquatic resources. Formulation of specific objectives also re-
quires some knowledge of the fluvial system to be monitored and the likely
impact on management activities.
Careful identification of the specific objectives probably is the most crucial
step in the entire process, as a set of precise.objectives will largely define the
remainder of the monitoring project, including the approximate cost, monitor-
ing parameters, sampling locations, sampling frequency, and data analysis
techniques.
-------
Define personnel and budgetary constraints
Define monitoring parameters,
sampling frequency, sampling
location, and analytic procedures
Evaluate hypothetical
or, if available, real data
Will the data meet the
proposed monitoring objectives?
Yes
No
Is the proposed monitoring
program compatible with
available resources?
Yes
No
Initiate monitoring activities on a pilot basis
Analyze and evaluate data
Does the pilot project meet
the monitoring objectives?
Yes
No
Continue monitoring and data analysis
Reports and recommendations
Revise the
objectives
or the
monitoring
procedures
Revise
monitoring
plan as
needed
Figure 1. Development of a monitoring project.
-------
Once the specific objectives have been formulated, the next step is to select
the parameters to be measured and set out the protocols for collecting data
and analyzing field samples. Provision for outside analyses and repetitive
samples is needed for quality assurance and quality control. The frequency,
duration, and location of measurements will be determined by the objectives
and the decisions with regard to the trade-offs among sample size, variability,
risk, and uncertainty.
Probably the best means to evaluate the feasibility of the objectives is to
develop and test a set of hypothetical or—if available—real data. This is
rarely done, but it can be extremely helpful in terms of crystallizing the proce-
dures and attainable objectives. Problems at this stage may necessitate a
rethinking of the objectives, a change in the parameters to be monitored, or
alterations in the sampling design. If the data are consistent with the other
components of the monitoring plan, a final check should be made to ensure
that the resources are available to carry out the work, and that responsibilities
for each aspect of the monitoring plan are clearly defined.
If the specific objectives are determined to be feasible, the next step is to
obtain a final cost estimate in terms of staff time, equipment, and outside
expenditures, .pelaying the final cost estimates until this step is unusual, but
the advantages are (1) the managers already have bought into the monitoring
project by helping to define it, and this makes it easier to obtain the necessary
support; and (2) the monitoring objectives play a more prominent role in
designing the monitoring plan, rather than the monitoring plan being primarily
a function of the available staff and expertise. Thus, as indicated in Figure 1,
the balancing of monitoring needs and budgetary constraints should be a two-
step, iterative process. The first step is simply to ensure that the objectives
and scope of the monitoring plan are generally realistic with regard to the
available personnel and budget. From that point, however, the planning
process should emphasize the optimal achievement of the monitoring objec-
tives. A final synthesis occurs when the monitoring plan has been fully
conceptualized.
If at this stage the proposed monitoring plan substantially exceeds the avail-
able resources, it may be necessary to revise the monitoring objectives.
Alternatively, a smaller reduction in costs might be achieved by reducing the
number of sampling sites, reducing the number of parameters to be moni-
tored, or reducing the frequency of sampling. The danger of adjusting sam-
pling intensity rather than the objectives is that the expectations may remain
unchanged while the capability or sensitivity of the monitoring project is
reduced. By having managers participate in the planning process, they will
be much more aware of how additional personnel and budget constraints will
alter the anticipated results of the water quality monitoring project.
At this point the proposed monitoring project is ready for data collection to
begin. Generally it is best to consider the first field season or set of data
collection activities as a pilot project. This allows more flexibility to adapt the
methodology to the conditions and variability found in the field. It also pro-
vides more impetus to the rapid analysis of field data and subsequent
-------
modifications of the monitoring plan. All too often the monitoring plan is
considered as a final, fixed document, and then there is not as much incen-
tive to analyze the data as they are collected. In such cases the data tend to
simply accumulate, and it is not until the end of the project that somebody
recognizes that the efficiency and quality of data collection could be im-
proved, or that the original monitoring objectives cannot be fully achieved.
Designation of the first phase of data collection as a pilot project greatly
enhances the potential for communication among all those involved in the
monitoring project—technicians, statisticians, managers, and technical
specialists.
As shown in Figure 1, the results of the pilot project can lead either to a
revision of the monitoring project or to continued monitoring. In most cases a
pilot project, if properly formulated, will result in some modifications in the
monitoring procedures, but will not alter the basic structure or objectives of
the overall monitoring project. Continued monitoring will then lead to the
accumulation of data that must be checked, stored, and analyzed. A descrip-
tion of these steps is beyond the scope oflhis document, but data storage
and retrieval is another key aspect of monitoring that is often neglected in the
planning phase.
The final step in Figure 1 is the preparation of reports and recommendations.
For a variety of reasons many monitoring projects do not follow through to
this step, and in such cases the worth of conducting the project must be
questioned. In general, the multiple demands on staff time mean that the
monitoring data will be used only if they are summarized and interpreted. If
the results are clearly presented, the information will be much more widely
disseminated, and this will reflect favorably on those responsible for the
monitoring project. More importantly, the data are more likely to be evaluated
by managers and used for the original purpose, namely the guidance of
management decisions. Failure to follow through to this final step implies a
basic failure in achieving the objectives of a monitoring project.
Monitoring a watershed is analogous to a person getting a physical exam. Dur-
ing the physical, the doctor takes a blood sample, which is analyzed for sugar,
hormones, cell counts, cholesterol, etc. Like the circulatory system is to a hu-
man, the stream is to a watershed. Based upon the blood sample, the doctor
can judge whether the person is healthy or not. Similarly, a water-quality sample
from a stream can indicate the health of a watershed.
If the doctor finds the cell counts out of balance, she runs other tests in an at-
tempt to pinpoint the source of the problem. The doctor uses multiple tests to
make a diagnosis before starting treatment.
The same kind of careful efforts are needed to diagnose whether BMP's are
effective. The stream sample may reveal adverse impacts on fish, but further
investigation needs to pinpoint the source of the problem. All pertinent factors
associated with the stream and management practices need to be examined.
Table 5 (page 58) identifies various methods to monitor reference and study
streams: biology, habitat, channel geomorphology, aquatic vegetation, chemical
and physical attributes, and on-slope conditions. All factors need not be
-------
evaluated in every situation; however, all factors should be considered for evalu-
ation. Enough of them must be evaluated to reach a good diagnosis.
The scheduling of monitoring depends on when information on BMP effective-
ness can be most effectively and precisely determined. Monitoring can be
scheduled within a season, within the period of a management activity, or over a
longer period. Often, it is most appropriate to collect water-quality samples
before, during, or immediately after a management activity. For example, sus-
pended sediment from constructing a road stream-crossing should be monitored
during construction. Herbicide monitoring should be done during the application
and/or through the next series of major storms. Seasonal differences can also
influence the scheduling of monitoring; for example, monitoring during low-flow
periods is best done when dissolved oxygen is being assessed for BMP effec-
tiveness. Benthic macroinvertebrate diversity and abundance are seasonally .
sensitive, and the best season to compare biological responses will vary by
region.
The timing of on-Iand, channel, or biological assessments can vary with the
region and hydrologic/geomorphic processes being evaluated. Monitoring of on-
Iand BMP effectiveness for surface erosion in the South must be conducted soon
after treatment because of rapid recovery of sites, usually 2 to 4 years. On the
other hand, monitoring the effect of large wood in road fills on landslides would
not be appropriate until the woody material has decayed and the road has been
exposed to a "testing" storm.
Following are comments on planning monitoring projects for each of the five
effectiveness monitoring questions previously stated. Again, it is critical to
monitor factors that can link in-stream observations with management activities
for all five effectiveness questions.
QUESTION 1. Do properly selected, implemented, and maintained BMP's
meet water quality goals or standards?
This question can be approached in two ways. First, specific BMP's can be
installed adjacent to study streams. This allows design criteria, such as varying
streamside management zone widths or varying percent of canopy retained, to
be tested. A BMP design and implementation phase would be required. Similar
reference and study streams would have to be preselected.
The second approach locates management areas where BMP's have been or
are being installed and maintained properly. These tracts can be located through
implementation monitoring. Potential areas to be monitored must be screened to
ensure that upstream natural and/or human influences will not mask potential
water-quality responses to the activities in question. Upstream conditions should
approximate reference conditions. Such screening requires some time, but
should be faster than planning and implementing management activities. How-
ever, under this alternative, it would be difficult to evaluate BMP design criteria.
QUESTION 2. Does a management activity jeopardize water quality?
When concern about a specific management activity is expressed, a quick
monitoring effort is required to determine if water quality is being impaired and, if
10
-------
needed, assurance that remedial action is prompt. Above and below monitoring
sites are commonly used. The stream segment above the management area
must be assessed to determine if it approximates reference conditions or if it is
heavily impacted by human activities that would mask any impact from the
management area in question. Several stream sites above are assessed to
determine the range and variability of biological and habitat conditions, which are
then compared to the reference conditions, if available. If reference condition
data are not available, the least disturbed reference sites are located, assessed,
and analyzed to define a reference condition.
Several sites below the management area are monitored to determine if there is
impairment compared to the upstream (reference) conditions. The number of
sites above and below is determined by statistical needs to make an inference as
to whether there are significant differences.
If the State has defined reference conditions and associated water-quality crite-
ria, monitoring stream segments adjacent to and below the management area will
provide data for comparison with reference condition and established criteria.
QUESTION 3. Which management system best protects water quality?
In many States, BMP's do not prescribe such management activities as the type
of site preparation to use for specific watershed conditions or the type of logging
system to use. When these management activities are Implemented, some
BMP's may be installed—for example, streamside management zones may be
left between a prepared site and the stream. The question becomes, which
management treatment with associated BMP's best protects water quality?
Planning monitoring projects to answer this question can take the same two
approaches as for question 1. First, several management treatments with asso-
ciated BMP's are planned, implemented, and monitored near streams. These
treatments are replicated to define range and variability of responses. The '
second approach is to locate a range of existing treatments with proper imple-
mentation of associated BMP's, find replicates, and monitor to define range and
variability of responses. In both approaches, comparisons are made with refer-
ence conditions.
QUESTION 4. How long do BMP's remain effective?
Question 4 poses difference challenges. Once installed, do BMP's remain
effective.through the recovery period of the treatment area? If not, how long do
they last and under what climatic conditions? The answers to these questions
would help define BMP maintenance programs.
Again, there are two approaches to this question. One approach is to implement
a treatment, install associated BMP's, and monitor in-stream and on-slope vari-
ables until the treatment area stabilizes. In some regions, this may take just a
few years, while other regions may recover very slowly, requiring more than 10
years. The latter requires an extended commitment of time and resources and
the answer to questions would be slow in coming.
11
-------
The second approach is to locate management activities with properly installed
BMP's of various ages near streams. Treatment areas should range from new to
the age when the treatment area would be stabilized. In-stream and on-slope
monitoring are planned to determine trends in recovery and in the effectiveness
of BMP's over time. This approach yields answers much sooner than the first
approach. The difficulty is determining if BMP's were properly installed on the
old treatments. This problem can be partially overcome by having access to
good implementation monitoring records or by inspecting treatment areas.
QUESTION 5. How do major climatic events affect BMP effectiveness?
Answering this requires a flexible, opportunistic approach to monitoring. It is not
practical to install BMP's and hope that a major rainfall or snowmelt event will
occur during a short recovery period. Therefore, the monitoring project design
must take advantage of major climatic events that occur within reasonable travel
distance. The plan would be to go to the affected area, find treatment areas with
properly installed BMP's, and monitor associated streams and reference
streams. This type of monitoring determines if in-stream responses are associ-
ated with management and associated BMP's, with the climatic event alone, or a
combination of the two. On-slope monitoring will determine if BMP's survived the
storm, or if they were effective in minimizing impacts on streams.
When one plans to monitor the effectiveness of BMP's in meeting water-quality
goals or standards, the components monitored must be related to protecting the
designated beneficial use or those set as water-quality criteria by the State to
protect the beneficial use and can be affected by forest management. Poorly
planned monitoring will sample variables that are not related to beneficial uses or
water-quality criteria or are not affected by forestry. There is an ongoing effort to
develop biological and habitat metrics to use in monitoring water quality. Such
metrics will vary by region and the aquatic community, so local knowledge and
research need to be used in designing monitoring projects. The State water-
quality agency is a good source of this information.
MacDonald et al. (1991) discussed variables most commonly related to a major
designated beneficial use (Table 1) and those commonly affected by forest
management activities (Table 2). These variables were first defined for the
Pacific Northwest and Alaska but are generally applicable to the rest of the
country. The sensitivity ratings have been revised to reflect national conditions.
These sensitivity ratings are very general and local conditions may cause widely
different sensitivities. The parameters and associated sensitivity identified in
Tables 1 and 2 are for general guidance and will vary greatly among ecoregions.
For example, the sensitivity of pH to roads is rated as "indirectly affected and not
very sensitive" in Table 2. However, if roads are cut through rock formations
containing iron pyrite, exposing the pyrite to rain will result in formation of acid
runoff, which can significantly lower stream pH.
MacDonald et al. (1991) did not address site preparation for tree planting, so site-
preparation columns are added to Table 2. These tables can help in selecting
variables to monitor, but selection must always be tempered by knowledge of
local conditions and management activities.
12
-------
Table 1. Qualitative assessment of the effects of water quality parameters on the major designated uses of water from forested
watersheds in the Pacific Northwest and Alaska. 1 = designated use is directly related and highly sensitive to the parameter in almost all
cases; 2 = designated use is closely related and somewhat sensitive to the parameter in most cases; 3 = designated use is indirectly
related and not very sensitive to the parameter in most cases;- 4 = designated use is largely unrelated to the parameter; V = relationship
between the parameter and the designated use is highly variable
Designated uses affected by water quality parameter
Water quality
parameters
Water column
Temperature
pH
Conductivity
Dissolved oxygen
Intergravel DO
Nitrogen
Phosphorus
Herbicides and
pesticides
Flow
Peak flows
Low flows
Water yield
Sediment
Suspended
Turbidity
Bedload
Channel
Characteristics
Cross-sections
Width/width-
depth ratio
Pool parameters
Thalweg profile
Habitat units
Bed material
Size
Embeddedness
Surface vs.
subsurface
Large woody
debris
Bank stability
Riparian
Riparian canopy
opening
Riparian vegetation
Aquatic organisms
Bacteria
Algae
Invertebrates
Fish
Domestic
water
supply
3
1
1
2
4
2
2
1
4
2
2
1
1
3
4
4
4
4
4
3
4
4
4
3
4
4
1
2
4
4
Agricultural
water supply
4
1
1
3
4
2
2
1
4
1
1
1
2
3
4
4
4
4
4
4
4
•
4
4
3
4
4
3
3
4
4
Hydroelectric
generation
4
4
4
4
4
4
4
4
1
1
1
1
1
2
4
4
4
4
4
4
4
4
4
3
4
4
4
4
4
V
Recreation
2
3
4
2
3
2
2
2
3
• 2
3
2
1
3
3
2
2
3
3
3
3
4
2
2
2
2
1
1
3
1
Warm-
water
fishes
2
3
4
1
2
3
3
3
3
2
4
2
1
2
3
2
1
2
1
1
2
2
1
2
2
2
4
2
1
1
Cold-
water
fishes
1
3
4
1
1
3
3
3
2
2
4
2
1
2
3
2
1
2
1
1
1
2
1
2
2
2
4
2
1
1
Biological
integrity
1
3
4
1
1
2
2
1
2
3
4
1
1
2
3
2
2
3
2
1
1
2
2
2
1
1
4
1
1
1
13
-------
Table 2, Sensitivity of water quality parameters to management activities, assuming average management activities: 1 = directly affected
and highly sensitive; 2 = moderately affected and somewhat sensitive; 3 = Indirectly affected and not very sensitive; 4 = largely
unaffected
Forest management
Parameter Harvest
Water column
Temperature
pH
Conductivity
Dissolved oxygen (DO)
Intergravel DO
Nitrogen
Phosphorus
Herbicides and pesticides
Flow
Peak flows
Low flows
Water yield
Sediment
Suspended
Turbidity
Bodload
Channel characteristics
Cross-sections
Width/width-depth ratio
Pool parameters
Thalweg profile
Habitat units
Bed material size
Embeddedness
Surface vs. subsurface
Large woody debris
Bank stability
Riparian
Canopy opening
Vegetation
Aquatic organisms
Bacteria
Algae
Invertebrates
Fish
1-2
3
3
2-3
2
2
2
4
1-2
1
1
1-3
1-3
1-3
1-3
1-3
1-3
1-3
2
2
1-3
2
1
1-3
1-3
1-3
4
1
1
2
Road building
activities
Applications
& maintenance Fertilizer Herbicides Pesticides
3
3
3
3
2
3
3
3-4
1
3
- 3
1
1
1
1
1
1
1
1
1
1
1
4
1
1-2
3
4
2-3
1
1
4
3
3
2
3
1
1-3
4
4
3
3
3
3
3
4
4
4
4
4
4
4
4
4
3
3
3
4
2
3
3
3
3
3
3
3
3
3
1
3
3
3
3
3
3
3
3
3
3
4
3
3
3
3
2
1
1
4
2
3
3
4
4
4
4
4
4
4
1
4
4
4
4
4
4
4
4
4
4
4
4
4
4
3
4
4
4
4
3-4
2
2-3
Intensive
site
prep."
1-2
2-3
3
3
1-2
2
2
3
-2
-2
-2
2
2
1-2
1-2
1-2
1-2
1-2
1-2
1
1-2
1-3
1-3
4
1
1
2
Low Int.
site
prep."
1-2
2-3
3
3
2-3
2
2
3-4
1-2
1
1
1-3
1-3
2-3
2
2
2
2
2
2
2
2
1
2
1-3
1-3
4
1
1
1
Grazing Recreation
2
3
3
1
2
1
1
4
3
2
3
2
2
2
1
1
2
2
2
2
2
2
4
1
2
1
1
1
1
1-2
4
4
4
4
3
. 3
3
3
3
4
4
3
3
4
4
4
4
4
4
4
4
4
4
1-3
3
3
1
3
3
1-3
" Mechanical and other methods that expose more than 30 percent of the soil.
* Mechanical and other methods that expose less than 30 percent of the soil.
14
-------
Quality Assurance/Quality Control
Quality assurance/quality control (QA/QC) planning and implementation are vital
parts of any monitoring project. The QA/QC program will vary by level of effec-
tiveness monitoring and the variable(s) being monitored. The following are some
basic ingredients of a QA/QC program (EPA 1992).
Project Description: The specific monitoring goals and objectives should be
explicitly described in the project description. The project description should
describe how the data will be collected to meet the goals and objectives.
Project Organization and Responsibility: The organization of a monitoring
project defines how the project will be conducted. A description of organizations
and their functions should strengthen monitoring performance and compliance
with QA/QC procedures. The key individuals responsible for collecting data and
for ensuring precision and accuracy of data analyses should be identified by title,
level of expertise, and a brief description of their responsibilities. An individual's
level of expertise is especially important when an activity relies on professional
judgment.
Quality Assurance Objectives for Measurement Data: For each variable
measurement or value, the following QA objectives should be presented.
a. Precision: Precision is a measure of mutual agreement among individual
measurements or values of a variable taken under similar conditions.
b. Accuracy. Accuracy is the degree of agreement between a measured
value and the true or expected value for the variable.
c. Completeness: Completeness is the percentage of measurements made
that are judged to be valid.
d. Representativeness: Representativeness is the degree to which data
accurately and precisely represent a characteristic of a population, varia-
tions at a sampling point, or an environmental condition. The program
should be designed so that the samples collected are as representative as
possible of the habitat or population and a sufficient number of samples is
collected.
e. Comparability. Comparability is a measure of the confidence with which
one data set can be compared to another. Comparability is not quantifi-
able. However, it must be considered when designing sampling plans,
analysis procedures, quality control, and data reporting. Employing
consistent data forms and survey protocols will maximize comparability.
Laboratory Analytical Procedures: Methods of sample and data analysis
should be specified. For EPA-approved or standard methods, pertinent literature
should be referenced. For nonstandard or modified methods, detailed operating
procedures should be provided, including sample preparation and analytical
procedures.
15
-------
Field Sampling Procedures: Field sampling procedures should include steps to
be taken to ensure the quality of samples and sample data. The sampling
procedures will vary by variable. The following are examples of procedures to
documented:
a. Specific physical, chemical, biological, and habitat variables to sample
b. For biological variables, identify target assemblages
c. Sampling methodology
d. Habitat-assessment methodology
e. Details of sample preservation
f. Use and calibration of instruments
g. Replication and QC requirements
h. Sampling site selection
Sample Custody: A chain-of-custody procedure should be developed to ensure
a written record traces the possession of the sample from the time of collection
through data analysis.
Calibration Procedures and Frequency: A program of calibration procedures
should be written to ensure that field and laboratory equipment is functioning at
an optimal level.
Preventive Maintenance: A plan for routine inspection and preventive mainte-
nance should be developed for all field and laboratory equipment and facilities to
ensure data of consistently high quality.
Data Reduction, Validation, and Reporting: This part of the QA plan is de-
signed to ensure good data by maintaining data quality throughout data reduc-
tion, transfer, storage, retrieval, and reporting. For example, biotic samples
should be checked for proper taxonomic identification and forms checked for
completeness, recording errors, plausibility, and consistency.
Internal Quality Control Checks: All personnel participating in monitoring
activities must be trained in the proper use and maintenance of sampling equip-
ment. They must be aware of the limitations of each piece of equipment. Inter-
nal quality-control checks can include replicate samples at stations to check
consistency of collection, repeat field collections by separate crews, etc.
Data Precision and Accuracy Procedures: The best approach to ensure
precision and accuracy is ensure that project personnel have received adequate
training and gained the necessary experience to conduct the project. Adequate
training of personnel in methods application is best way to ensure consistency,
repeatability, and precision.
Performance Audits, Systems Audits: Quality control checks on the proce-
dures used by field personnel should be done periodically during the field sea-
son. If any problems are found, corrective action must be taken immediately.
16
-------
Corrective Action: A corrective-action program capable of detecting errors at
any point in the project implementation process should be developed. The pro-
gram must be able to identify problems and their source, implement action to
correct them, document results of corrective action, and continue the process
until each problem is eliminated.
Quality Assurance Reports: A formal report should be written to inform appro-
priate managers about the performance and progress of the monitoring plan.
17
-------
-------
Statistical Considerations
In designing a water-quality monitoring effort, statistical design is vitally important
in drawing sound conclusions from the data. Statistical design will guide the
sampling, the number of sites to sample, the statistical analyses to use, and the
confidence level of the results. The following statistical discussion, excerpted
from MacDonald et al. (1991), is lengthy but is necessary to demonstrate the
relevance of statistics to water-quality monitoring.
Statistics are an inherent component of nearly all water quality monitoring
programs. In most case a precise formulation of the monitoring objectives
results in a question that is best answered on a statistical basis. For ex-
ample, a common objective of water quality monitoring plans is to determine
if a particular management activity is causing an adverse change in water
quality. To answer this question in a quantitative manner, it is necessary to
acquire data and make a comparison to other site(s), or to data from the
same site prior to the management activity. If the monitoring plan is properly
designed and replicated, data analysis will yield specific conclusions with an
identified level of risk.
Statistics provides the scientific basis and procedures for studying numerical
data and making inferences about a population based on a sample of the
population (Mendenhall, 1971; Sokal and Rohlf, 1981).
By its very nature, water quality monitoring is a sampling procedure. It simply
is not possible to make continuous measurements of all parameters at all
locations. This means that before any data are collected one must address
questions such as:
- How many samples are likely to be needed to characterize a param-
eter with a specified degree of uncertainty?
- How many samples are likely to be needed to determine if there is a
difference between locations, or a change over time?
Where and when should samples be taken?
Which parameters should be measured?
- How will the precision and accuracy of the data be assured?
As the monitoring plan develops and data are collected, there is a continuing
need to analyze the data, evaluate whether the data are meeting the objec-
tives, and determine whether the timing and location of sampling is optimal.
All these aspects of a monitoring program either require or involve statistics.
Many people react negatively to the use of statistics. Typically this is due to a
lack of understanding about the role of statistics in water quality monitoring,
or past experiences in which the application of statistics led to unexpected
conflict or uncertainty. Statistics can make a strong positive contribution to
water quality monitoring by:
- providing an overall design for collecting and analyzing data;
- facilitating the precise specification of objectives, including an explicit
recognition of the uncertainty and potential errors;
- providing a quantitative means to optimize the location and times of
sampling, and thereby reduce costs;
19
-------
- providing a rigorous set of procedures for analyzing the data collected
in a water quality monitoring program; and
- providing a quantitative basis for making inferences about the charac-
teristics and response of the population being sampled.
To take full advantage of these potential benefits, those responsible for
preparing monitoring plans should consult with a statistician early and often.
Too often a statistician is consulted after the data have been collected, and
the statistician's tools are unable to salvage inconsistent or unreplicated data.
These five contributions imply that statistical procedures are critical tools in
water quality monitoring, but they are not a substitute for decision-making.
Averett (1979) states "data interpretation is an intellectual activity; statistical
applications is a mechanical activity." Those responsible for a water quality
monitoring program still must decide how much uncertainty can be tolerated
and balanced the relative risks and costs associated with different types of
errors. The manager and technical staff must also determine what type of
monitoring design is most appropriate, which parameters to measure, and the
initial times and locations for sampling.
This chapter presents some of the key statistical principles which must be
considered in developing a water quality monitoring program. The overall
goal is to demystify the role of statistics and statisticians in water quality
monitoring programs. The specific objectives are to: (1) explain how statisti-
cal considerations should be taken into account in designing and implement-
ing water quality monitoring programs; and (2) explicitly discuss the trade-offs
between sample size, inherent variability, level of significance, statistical
power, and the minimum detectable effect.
Specific guidance on the selection and use of statistical test is not addressed
in this document.
Statistical Design in Water Quality Monitoring
General Design and Replication
The overall design of a monitoring project is largely determined by the moni-
toring objectives and closely tied to the type of monitoring. In many cases the
design of the monitoring plan will determine the statistical procedures used to
analyze data.
Standard statistical designs are based upon a series of experimental units.
Experimental units are defined as the objects upon which measurements are
made (Mendenhall, 1971), and in water quality monitoring these are usually
sample sites. In an idealized, simple experiment, the experimental units
would be randomly selected and half assigned to some treatment, while the
other half would be left as untreated controls. Both the treated and the
untreated experimental units usually are considered to be representative
samples of much larger populations. Repeated measurements on the experi-
mental units generate the data used to describe the sampled populations,
20
-------
and to draw inferences about the larger population from which the experimen-
tal units were drawn.
Multi-factor designs allow this simple experimental design to be expanded to
multiple levels of treatments and their interactions. Multivariate statistics are
used to analyze data and test hypotheses when several independent (causal)
or dependent (response) factors exist. Multi-factor designs and multivariate
statistics will not be discussed here as they are based on the same principles
as the simpler, univariate methods.
Nonparametric statistics originally were developed to analyze qualitative or
ranked data, and they also can be used when the underlying distribution of
the data is not normal. Hence they are more broadly applicable and more
robust in terms of requiring fewer assumptions, but they generally are less
sensitive. Recent advances are likely to greatly increase the use of nonpara-
metric procedures. In this chapter parametric statistics are emphasized
because most water quality data are numerical and can be transformed into
an approximately normal distribution.
The idealized simple experiment outlined above illustrates several key ele-
ments common to all statistical designs. First, a population is defined, and
samples are drawn from that population. The population might be defined as
a particular fish species in a stream reach, in pools of a certain size, or in a
certain type of stream. Second, some treatment is applied to the designate
experimental units, and this might be timber harvest, forest fertilization, or
gravel extraction. Third, this treatment is applied to two or more experimental
units, and two or more experimental units are left as controls. Fourth, a
series of measurements are made, and these provide the raw data for the
statistical analysis.
Unfortunately most water quality monitoring plans do not fit this idealized
design. A typical objective of water quality monitoring plans is to determine
whether the value of a parameter has changed over time at a particular site.
The two most common approaches used to address this question are (1) to
measure the selected parameter over time at the site of interest, as in trend
monitoring; and (2) to compare data from a treated site with an untreated site,
as in project monitoring.
With regard to the first case, there is only one experimental unit, and the data
collected are a sample of all possible measurements in time. In some cases
the onset of a management activity can be used to separate the data into two
groups (i.e., before and after), and one can test for significant change over
time by comparing the means and variances over the initial period (baseline
data) to the means and variances following the onset of the management
activity. Often, however, this straightforward approach is not valid because
the data are serially correlated (i.e., the value of any given data point is
related to the previous value), or the data vary according to season, dis-
charge, or other variables.
21
-------
The approach to detecting trends will depend on the number of data points
available and the type of trends or correlations present in the data. Graphing
the data is the first and probably most important step in identifying the compli-
cating factors and determining the appropriate statistical approach (Gilbert,
1987). A basic choice is either to attempt to remove the trend or correlation
and then use parametric statistics, or use nonparametric statistics on the
original data. Gilbert (1987) provides a useful guide to trend analysis tech-
niques, and he references Harnes etal. (1981) for analyzing discharge-
related parameters and Montgomery and Reckhow (1984) for analyzing
serially correlated data....
The first and most common design to evaluate changes over time is to moni-
tor a single site. This approach is useful to detect seasonal or other trends,
but a basic problem is that statistical inferences cannot be made either about
the cause of and observed change at the monitoring site or about the cause
of similar changes observed at other sites. Data from other sites are neces-
sary for making inferences about other locations (Hurlbert, 1984).
The paired-site approach is the second design which often is used to evalu-
ate change. In this design two sites are monitored, and a statistical relation-
ship between the sites is established for the parameter(s) on interest. After
this initial calibration period, one site is subjected to a treatment (e.g., timber
harvest), and the other is left as a control. A significant change in the statisti-
cal relationship between sites is used to indicate a treatment effect. This is
the basic concept behind paired-watershed experiments (e.g., Bosch and
Hewlett, 1982).
The advantage of the paired-site approach is that the untreated or control site
provides a basis for separating the treatment effect from other extraneous
factors (e.g., climatic events). Nevertheless, this design still shares the same
major flaw as the single-site approach, namely the lack of replication. In the
absence of replicated treated and control sites, there is no information on the
spatial variability of the parameters being measured. An estimate of the
variability is necessary to make any statistically based inference about the
cause of observed difference between the treated and control sites. Since in
most case sites are not replicate, claims of cause and effect must be based
on other information and not statistical testing (Hurlbert, 1984).
Multiple pairs of treated and control sites, although costly, usually result in a
high sensitivity to change. Both the control and treated sites are subjected to
the same extraneous factors, so the exclusion of these factors greatly in-
creases the likelihood of detecting a treatment effect.
The paired-site approach is commonly used in project monitoring. Typically
water quality is measured upstream and downstream of a particular activity,
and the observed differences between sites are presumed to be due to the
particular project or activity. However, the known differences in water quality
and stream characteristics between upstream and downstream locations
(e.g., Hainman et a/., 1991) means that a pre-project calibration period is
essential for unreplicated sites. As in the paired-site approach, the absence
22
-------
of multiple treated (downstream) and control (upstream) sites means that the
inference of cause-and-effect must be based on qualitative evaluation rather
than statistical testing.
As suggested above, neither the single-site nor the paired-site approach fit
into the traditional randomized designs described in statistical texts. In most
water quality monitoring plans, the experimental units are streams, lakes, or
sampling sites, and these cannot be randomly allocated among treatments
such as clearcuttingor road building. Typically the experimental units and
treatment(s) are already specified, and the objective of the monitoring pro-
gram is to determine if change has occurred. Sampling sites are often fixed
by the presence of a bridge or other structure from which samples can be
safely taken at high flows, or by assess to the drainage network.
The randomized block design may be the most relevant to water quality
monitoring. Each block includes all of the treatments as well as a control.
Treatments are randomly assigned to the experimental units within a block.
Analysis of variance procedures are used to evaluate the differences be-
tween treatments in one or more blocks, regardless of variation, among
blocks, Th.us the primary advantage of this design is to exclude extraneous
factors (such as site differences, which occur between blocks) and focus on
the differences between treatments within blocks. This makes the design
statistically more robust (i.e., the results are reliable over a wider range of
conditions).
Paired watersheds and upstream-downstream comparisons represent two of
the simplest forms of a block design. The combination of a treated watershed
and a control watershed form one, unreplicated block. Additional paired
watersheds undergoing identical treatments result in additional blocks. To
the extent that treatments are randomly assigned to each experimental
watershed within a block, this yields a randomized block design, and statisti-
cal inference can be made regarding (1) the cause of any observed differ-
ences, and (2) the likely result of a similar treatment on other unmonitored
sites that are part of the same population.
In the case of upstream-downstream comparisons, the upstream site usually
acts as the control, and the downstream site serves as the treated site.
Again the addition of paired upstream-downstream sites generates additional
blocks. The problem with this design is that the treatments are not randomly
assigned within each block, but are set according to a pre-determined and
recognized site difference. The statistical design to resolve this problem is to
replicate pairs without any treatment or project, and compare the upstream-
downstream differences between these untreated pairs to the differences for
the pairs where there actually is management activity. Alternatively, a rela-
tionship between each upstream and downstream location could be estab-
lished during a calibration period, and a change in this relationship following
management activities would indicate a treatment effect rather than a site
difference.
23
-------
In practice it is often assumed that natural variability overwhelms a consistent
site effect between the upstream and downstream locations. In this case, a
calibration period might not be necessary, and any difference between the
sites should be due solely to the treatment being studied. Such an approach
is inconsistent with basic statistical principles as indicated above.
The problem of separating site differences from treatment differences is
particularly acute for many of the channel parameters... Channel cross-
sections, pool parameters, and bed material particle size are all sensitive to
environmental factors such as local geology and landforms, and they may
exhibit considerable variation over relatively short distances. ...the problem
of spatial variation can at least be alleviated by carefully identifying the sites
to be monitored, and.monitoring prior to initiating management activities.
This available of pre-project data is critical to inferring management effects,
and also influences the choice of statistical test(s).
As indicated earlier, more complicated designs have been developed to
analyze multiple factors and the interactions between them. The primary
problem associated with these designs in that the number of blocks usually is
. a product of the number of different levels for each factor, and the desire to
examine several factors at once rapidly leads to a large number of experi-
mental units. For example, an evaluation of the effects of two levels of
nitrogen and two levels of phosphorous in streams with three repetitions of
each combination and a set of controls requires a total of 15 experimental
units. Adding in a three-level, qualitative factor such as the type of riparian
vegetation (e.g., coniferous, deciduous, or no tree cover) increases the
number of experimental units to 45. The use of these more complex designs
will depend on factors such as the availability of experimental units, the
estimated importance of the interactions, the monitoring budget, and the
statistical trade-offs.
Benefits of Proper Statistical Design
The importance of the statistical design can be illustrated by an example.
Suppose a land manager needed to determine if clearcutting increases the
number of landslides. The staff reviewed a recent set of aerial photos and
determined that there were 25 landslides on 5,000 acres of clearcuts, and 50
landslides on 20,000 acres of mature forest. Although the number of land-
slides per unit area proportionally was twice as high on clearcuts as mature
forest, it is not possible to make any generalizations on statistical conclusions
because data on the variability of the number of landslides on clearcut and
forested areas was unavailable. Statistical analyses require multiple mea-
surements in time or space, but the results cited above are from a single
survey of one control and one treated experimental unit.
One might argue that statistical rigor may not be important in cases where the
difference is relatively large, but what is the certainty associated with the
statement that clearcutting increases landslides if there were 70, 80, or 90
landslides in the uncut forest? The overall trend in land management is
towards increasing regulation to eliminate the obvious adverse impacts.
24
-------
Hence increasingly sensitive techniques will be required to evaluate future
management effects, and this will require proper statistical designs.
Collection of essentially the same data in the context of an overall statistical
design can greatly increase the value of the data. In the above example,
instead of treating the entire area as one experimental unit with two treat-
ments (clearcut and forested), the staff should have identified the population
of clearcut units and potentially harvestable forested units. Then data on the
area and number of landslides in each clearcut and each forested unit should
have been collected. This procedure would yield a data set consisting of the
number of landslides per unit area of each potentially harvestable forested
unit and each clearcut unit. Since these data represent a sample of a much
larger population of clearcut and potentially harvestable forested areas,
statistical techniques can be used to determine the following:
- the mean and variance of the number of landslides in each land use
type;
- the approximate shape of the underlying distribution of the data (e.g.,
normal, binomial, Poisson, lognormal);
- the significance probability associated with the observed difference in
the number of landslides between the two land use types; and
- the likelihood of obtaining a false conclusion.
A proper statistical design can greatly improve the efficiency of data collec-
tion. For example, if the staff thought that the difference in the number of
landslides in the clearcut and forested units was relatively large, measure-
ment might only be made on a random sample of 10% of the units. If this
sample indicated a statistical significant difference, measuring the remaining
units might be unnecessary, and a substantial saving in the cost of the study
would be realized.
The potential benefits of an adequate statistical design are even more appar-
ent if there are several sources of variation. In the above example, the
frequency of landslides might be strongly influenced by slope steepness or
the type of bedrock. If the sample size is sufficiently large, statistical proce-
dures can be used to separate these factors. Such information is extremely
useful for developing practical management procedures, such as identifying
high-risk areas or predicting sediment to streams from landslides.
In many cases the critical factors are known prior to initiating the monitoring
project. This a priori information can be used to construct strata to improve
the sampling efficiency and the sensitivity of the statistical tests. The basic
principle is that the strata should remove as much of the within-stratum
variability as possible, thereby allowing the treatment effect to stand out from
the "noise" of the data. Continuing with the previous example, the forested
and clearcut units might be stratified by geologic type. A random sample of
the forested and the clearcut units would be taken from each geologic type
(stratum), and an analysis of variance procedure, would be used to detect
differences in landslide frequency among strata (e.g., sandstone or shale)
and treatment (e.g., clearcut or forested). If prior information is lacking, a
pilot study can be extremely helpful in determining an appropriate statistical
design, identifying strata, and estimating sample size.
25
-------
Design Problems and Constraints
Some of the major problems and constraints associated with developing
water quality monitoring plans include:
- lack of adequate information prior to initiating a monitoring project,
- difficulty in distinguishing between the effects of management activities
and natural events,
- difficulty in distinguishing among the relative effects of multiple man-
agement activities,
- the possible time lag between an action and its effect on water quality,
and
the random nature of climatic events.
The lack of adequate information about the parameters to be monitored is an
important limitation to the development of a monitoring plan.' In many cases
key sampling decisions must be made with little or no data on the diurnal and
seasonal fluctuations of different parameters, the dependence of the param-
eters on flow and site conditions, and the spatial variability. This is why it is
so important to consider monitoring as an iterative process. As monitoring
proceeds more of the temporal and spatial variability is captured in the data
set, and the quality of the information increases. This may make the initial
design inappropriate, and the monitoring plan should be revised. Unfortu-
nately any change in monitoring sites or methodology may preclude compari-
sons with the earlier data, thus making it important to conduct a pilot monitor-
ing project before initiating large-scale or long-term monitoring activities.
Even if the overall design is known to be satisfactory, a regular review of the
monitoring objectives and data is needed to maximize the efficiency of data
collection and analysis.
One example of how inadequate information might seriously hinder monitor-
ing efforts may be found in the use of habitat types. At present almost no
information is available on changes in habitat units over time, particularly in
response to management activities. Similarly, few data have been published
on the accuracy of habitat unit surveys under different flow conditions or by
different survey teams. Nevertheless, extensive efforts are underway to
characterize and monitor habitat types in streams, and arbitrary limits on the
amount of allowable change are being established. A few years' experience
with a series of pilot projects might be preferable to determine the spatial and
temporal variability, help identify allowable limits for change, and determine
whether a multi-stage sampling scheme could reduce survey costs.
A second constraint on the use of standard statistical procedures in water
quality monitoring is the need to separate the effects of management activi-
ties from natural events. This is best done by comparing changes in water
quality at unmanaged controls sites to changes at sites with management
activities. The unmanaged control sites provide an index of change due to
some key factors, such as climate, and the removal of these factors increase
the sensitivity of the monitoring. Problems with this approach include the
additional cost of monitoring the control sites, and the fact that in many areas
it is becoming increasingly difficult to find sites that have not been subjected
26
-------
to extensive management activities. Valid control sites also must be left
undisturbed for the duration of the monitoring program. Although it may be
possible to find adequate control sites for small headwater streams in some
areas, most of the control sites for larger streams will be found only in parks
or wilderness areas. This suggests an increasing distance between the
treated and the control sites, and an increasingly tenuous assumption of
comparability. With an weaker statistical relationship between sites, there is
a declining ability to detect significant changes due to management activities.
To a certain extent additional observations can substitute for a rigid statistical
design. Qualitative or quantitative data that document the processes linking
management activities to water quality can greatly enhance the validity of any
observed trends in water quality. For example, direct observations during
storm events can show how a particular road or clearcut is affecting water
quality, and this can help compensate for the absence of replicated control
sites.
A third basic limitation to using standard statistical procedures is the problem
of overlapping activities. Relatively few watersheds are subject to just one
management activity. Often one type of management activity, such as forest
harvest, incurs other activities, such as increased road traffic, additional road
maintenance, and road construction. Larger watersheds, which are the focus
of most trend monitoring programs, tend to have more numerous and diverse
land uses. Since in-channel water quality measurements integrate the effects
of all the upstream management activities and a myriad of natural processes,
in many cases a change in water quality cannot be directly related to a
specific management activity or a specific set of BMP's. This may not be
important if the monitoring objective is simply to determine the overall condi-
tion and trend in Water quality. However, the common objectives of most
water quality monitoring programs are to identify problems, minimize adverse
impacts, and guide future management. The limited capacity of inchannel
measurements to distinguish among overlapping management activities
means that these objectives may not be met through inchannel measure-
ments. Multivariate techniques offer the potential to resolve the effects of
several independent (causal) variables, but their use in water quality monitor-
ing is hampered by the large number of •• enables needing consideration, and
the limited number of monitoring sites that can provide the necessary depen-
dent (effect) data.
The fourth limitation to using standard statistical procedures is the potential
lag time between a management activity and its effect on water quality. For
example, several studies have shown that the hazard of landslides in clearcut
areas is maximized 4-15 years after the harvest is completed (e.g., Gray and
Megahan, 1981; Swanston, 1969). Similarly, the time lag between the de-
tachment of a soil particle and its delivery to the stream channel is often
substantial (Swanston et at., 1982). This suggests that the typical project
monitoring period of up to 3 years (Ponce, 1980a) may not be adequate to
evaluate the long-term or delayed effects of certain management activities on
water quality.
27
-------
Climatic variability poses a similar problem with regard to the design of water
quality monitoring projects. The relative impact of a particular management
activity may vary according to the severity of the climatic conditions during
the period of maximum management impact. This suggests that a longer
monitoring period may be necessary if one wishes to evaluate or contrast
management impacts with the more sporadic or extreme natural events. In
some cases extreme events, such as floods or debris flows, can completely
overwhelm the changes in water quality due to management activities (e.g.,
Lisle, 1982; Griggs, 1988). Statistics can be used to evaluate the likelihood
of experiencing a certain event within a specified time period, and this infor-
mation can be helpful in the initial formulation of a monitoring plan. Similarly,
the results of a water quality monitoring project must be evaluated in the
context of the climatic events experienced during the monitoring period.
For some water quality monitoring objectives, a statistical design may not be
necessary. A source-search methodology often can be used to qualitatively
identify the cause of a water quality problem, or the most likely locations for
sampling. The basic procedure is to make systematic observations, usually
in the upstream direction, in order to identify the source of potential or exist-
ing water quality problems. Often observations or water samples are taken at
each major tributary. This procedure is most effective if there is a localized
pollution source that is having a substantial impact on water quality, and
when measurements can be made in the field.
'Similar procedures.can be followed to rapidly and qualitatively evaluate
management practices and impacts. Walking or driving a road network
during runoff events, for example, can provide a useful, qualitative review of
BMP's, and indicate where road-related water quality problems are develop-
ing. This type of reconnaissance can be extremely cost-effective as it facili-
tates the early identification of problems without embarking on a costly moni-
toring scheme. Such activities also offer the potential to resolve adverse
impacts at an early stage, and thereby reduce the costs of repairs or future
mitigation measure.
In short, there is a need to complement any instream monitoring program with
additional observations or measurements. These should aim at (1) providing
a direct link between upslope or riparian management activities and instream
water quality; and (2) enhancing one's understanding of watershed pro-
cesses. Such information is not only helpful in alleviating any problems with
the statistical design, but also is essential for helping to guide future research
and management.
Principles of Sampling
Many of the most important principles of sampling are similar to the principles
of statistical design, and these are discussed in most statistical texts. The
three basic types of sampling are random, systematic, and stratified, and
each can be applied in space or over time.
The procedure for simple random sampling is to clearly identify the universe
of potential sampling times or locations, and then select individual times or
28
-------
locations for sampling according to a random number table or any random
procedure. If information on the variability of the parameter is known, then
the number of samples needed to achieve a certain confidence interval can
be calculated. For example, simple random sampling might be used to select
the particular day for measuring pH in a large river.
Using simple random sampling to select monitoring sites may prove difficult in
practice because it requires identifying all possible sampling sites (i.e., the
sampling frame). This may not be a problem if the precise location of the
sample is not important. The sites for monitoring many water column param-
eters, for example, could be randomly selected from the population of river
miles. Simple random sampling could be very time-consuming if one wishes
to sample only certain habitat types (Hankin and Reeves, 1988).
Systematic sampling consists of randomly selecting the first sample, and then
selecting all subsequent samples by applying a constant interval. Systematic
sampling can result in a biased sample if there is a systematic variation in the
population being measured. For example, if the timing of the first sample in a
given year was determined randomly, and subsequent samples were taken at
exactly 6-month intervals, this might not represent a true long-term average
because all the samples would be taken in two different seasons. Hankin
and Reeves (1988) discuss the merits of different sampling schemes to
estimate fish abundance and habitat areas in small streams. They advocate
systematic sampling of individual habitat units (e.g., measuring the area of
every tenth glide, or counting the fish in every fifth plunge pool) because it is
the most practical and is unlikely to significantly bias the results. Systematic
sampling along a river or stream can be an efficient means to detect distinct
but unknown sources of pollution (Gilbert, 1987).
Stratified random sampling involves some grouping of the population of
interest, and then randomly sampling each group or stratum (Ponce, 1980a).
This procedure is often used in water quality sampling because certain
parameters are known to vary by the time of day, season, discharge, or some
other factor. The different strata can be sampled at different frequencies
according to the estimated size of the population (proportional sampling) or
the variability within the different strata (optimal sampling). Optimal sampling
generally is preferable for flow-dependent parameters, whereas proportional
sampling maybe equally efficient for time-dependent (e.g., seasonally vary-
ing) parameters.
The advantages of stratified random sampling are similar to the advantages
of a randomized block design, in that it can (1) improve the efficiency of
sampling, (2) provide separate data on each stratum, and (3) enhance the
sensitivity of statistical tests by separating the variability among strata from
the variation within strata. The information needed to construct the strata and
estimate the sampling frequency must either be known prior to sampling or
obtained through a pilot study.
Seasonal strata are often used for sampling invertebrates and fish, while
discharge is often used to establish strata for sampling sediment and the
other physical and chemical constituents of water. Stratification by discharge
also helps ensure that high flows are sufficiently sampled.
-------
Sub-strata can also be defined. If discharge is used to define the primary
strata, for example, the high flow stratum might then be sub-stratified accord-
ing to the position of the hydrograph (e.g., rising limb or falling limb) or by the
cause of the high discharge (e.g., snowmelt or rainstorm). The former might
be appropriate in the case of turbidity or suspended sediment, while the latter
might be more applicable for specific conductance, pH, or nutrient concentra-
tions. Although the value of stratification can be statistically tested, a plot of
the data provides a quick and qualitative indication of the value of stratifying.
The statistical benefits of a reduction in unexplained variability must then be
weighed against the cost of the sampling scheme needed to adequately
characterize each stratum.
For water quality programs the decision regarding where to sample is largely
determined by the objectives of the monitoring program, location of the
management activities, layout of the catchment(s) to be monitored, access to
the monitoring sites, and the design of the monitoring program. These issues
are discussed in greater detail in Gilbert (1987), Kunkle etal. (1987), Ponce
(1980a,b) and Sanders et al. (1987). For project monitoring the general
principle is to locate the sampling sites as close to the actual project as
practicable, as the largest water quality impact will be immediately down-
stream of the activity. Minimizing the distance between the upstream (con-
trol) and downstream (treated) sites will help minimize confounding site
differences....
Empirical knowledge of the basin to be monitored is extremely helpful in
developing a monitoring program (Ponce, 1980a). Even a cursory inspection
can indicate the types of adverse change that are likely to be encountered
and the spatial distribution of management activities. This type of spatial data
provides much of the guidance needed to establish sites and direct the
monitoring activities towards the problem areas.
Monitoring of channel characteristics or water quality within a particular basin
or region may best be achieved with a spatial stratified sampling scheme. In
keeping with the principles of stratified random sampling, the best approach
is to classify stream segments and subsample these. Rosgen (1985) and
Cupp (1989) are probably the two most widely used stream classifications
systems at this time. Once the stream segments have been identified, an
additional stratification into habitat units may be desirable (Hankin and
Reeves, 1988). Specific details for laying out such nested sampling schemes
are beyond the scope of this document, but the principles and references
cited in this section can provide the necessary guidance.
In turbulent streams many of the physical and chemical constituents of water
are relatively insensitive to the precise monitoring location. For these param-
eters a general site description, such as just upstream of a particular tribu-
tary, usually is sufficient. In less turbulent reaches, some parameters, such
as suspended sediment, can vary considerably with depth, and the reader
should refer to the appropriate U.S. Geological Survey publication for detailed
sampling guidelines.
30
-------
Other parameters, particularly those that pertain to channel geomorphology,
may exhibit a great deal of variation over a few meters. For these param-
eters a more precise site description is needed, and this usually is based on
prior knowledge. Bed material particle size, for example, might be evaluated
in certain, geomorphologically determined locations like the downstream
edge of a point bar. Embeddedness often is measured in riffles with certain
characteristics, although the precise location of each hoop sample is random.
Even after determining the general time and location of sampling, another
series of sampling questions must be addressed. Is a grab sample close to
the bank adequate, or should a series of depth-integrated samples be taken
across a particular cross-section? Is one sample in time adequate, or should
several samples be taken? If several samples are taken across a channel or
over a relatively short time, should these samples be kept separate or com-
bined?
The answer to these questions largely depends on the objectives of the
study, the parameter being measured, and the site characteristics. Certainly
the same statistical principles apply to these fine-scale questions of sampling
' as they do to the larger-scale questions of location and timing discussed
above. Composite samples over time or space can represent a substantial
saving in analytic costs, but this reduces the resolution of the data. Samples
for analyzing bacterial contamination should never be composited. If only a .
single grab sample can be taken, this should be taken in the middle of the
stream at the 0.6 depth (Ponce, 1980a). Specific recommendations for
sampling various parameters can be found in APHA (1989), Greeson et al.
(1977), Guy (1970) and Guy and Norman (1970). Once a monitoring project
has been initiated, any change in sampling procedure should be undertaken
very cautiously, as this may preclude any comparisons to data collected
using any other procedure.
Principles of Statistical Testing
Assumptions and Distributions
Most of the better-known statistical tests have been developed for sealer data
that follow a bell-shaped or "normal" distribution. Data with these attributes
normally are analyzed with parametric statistics. Nonparametric statistics are
applied to data which have an unknown population distribution, or that are
ranked or categorized rather than measured. For normally distributed data,
nonparametric statistics are less efficient than parametric statistics (i.e., they
are more likely to yield false conclusions) (Mendenhall, 1971). The lower
efficiency of nonparametric statistics is a primary reason why transformations
and other methods are used to obtain data that approximate a normal distri-
bution. The advantage of nonparametric statistics for water quality monitor-
ing is that they require fewer assumptions about the underlying distribution of
data. The focus of the following section is on parametric statistics, but most
of the principles also apply to nonparametric statistics.
31
-------
The key assumptions for the use of parametric statistics are as follows:
1. the data are normally distributed,
2. the data are a random sample of the population,
3. the observations are spatially and temporally independent, and
4. the errors in the data are randomly distributed.
Each of these assumptions can be tested, but only major violations can be
identified. Although these assumptions may not be strictly true in many
cases, the relevant question is whether a violation of these assumptions
substantially affects the probability statements being drawn from the data
(Ponce, 1980b). Different statistical tests vary in their sensitivity to each of
these assumptions. In uncertain situations nonparametric statistics can be
used to bolster or supplement any conclusions developed from the use of
parametric statistics. The following paragraphs briefly discuss these four
assumptions for parametric statistics in the context of water quality
monitoring.
1. The first step in determining if the distribution of a data set is normal is
to plot it. Data can be plotted over time, against a controlling variable
such as discharge, or as frequency distribution. A plot of the raw data
in important to visualize the distribution, as this allows a quick and
qualitative check for patterns, extreme values, and obvious errors.
Often a frequency distribution of water quality data shows a distinct
clumping to the left with a long tail of extreme values to the right. This
type of distribution is known as a lognormal distribution, and it usually
can be converted to the normal, bell-shaped distribution by converting
the data to base 10 or natural logarithms. If zero values are present,
the data are transformed by adding 1 to each value and then taking the
logarithm (Ponce, 1980b). After transformation the data should always
be plotted again, as this provides a familiarity with the data and an
intuitive check on the transformations and any subsequent calculations.
Numerous other transformations can be used, and two of the most
common are the square root and cube root, respectively. Both transfor-
mations and normalization procedures are discussed in most statistics
texts along with the tests necessary to check on the normality of the
data. One "quick and dirty" test for the normality of a frequency distri-
bution is to determine whether 2/3 of, the data falls within on standard
deviation of the mean, and 99.9% of the data is within three standard
deviations of the mean. If this is the case, the data are likely to be
considered normal for statistical testing purposes.
2. The second key assumption for parametric statistics is that the data are
a random sample of the population of interest...
3. The third assumption of spatial and temporal independence is best met
by establishing a proper design for collecting data. Daily streamflow
data, for example, are not independent in time, as the streamflow for
any given day is partly dependent on the amount of flow in the previous
32
-------
day (i.e., they are serially correlated). One could, however, randomly
sample from a population consisting of all the daily streamflow values.
Similarly, there is often a strong correlation between data collected in
adjacent basins, and this relationship forms the theoretical basis for the
paired-watershed approach. Usually this problem is best addressed by
properly defining the experimental units and the population to be
sampled. For example, the rainfall and streamflow data for adjacent
small basins will be highly correlated, but the rainfall for a particular
day from a series of gages in comparable locations should be normally
distributed. Thus an acceptable population for sampling might be the
precipitation data for a particular day from a series of gages. However,
statistical analysis of daily precipitation at a site would have to account
for the autocorrelation between daily values, and this can be done
through time series analysis (e.g., Box and Jenkins, 1976).
4. The final assumption—the random distribution of errors in the data—
also can be satisfied by ensuring that the design and sampling proce-
dures do not contain systematic errors. This is done by randomly
assigning treatments to the experimental units, by random sampling,
and by careful attention to measurement procedures (Sokal and Rohlf,
1981). Systematic errors can be removed if the cause can be identi-
fied, and the removal of systematic errors is a key procedure in Hankin
and Reeves' (1988) methodology for measuring habitat types. The
danger is that systematic errors are not recognized, and these could
easily result from a change in personnel, equipment, or measuring
techniques. The possibility of systematic errors is a particularly impor-
tant consideration when analyzing trend data from a single site. Quality
assurance and quality assurance techniques (e.g., EPA, 1983) are an
important means for reducing the possibility on non-random errors.
Plots of the data and residuals can help identify unusual trends, and
nonparametric tests can be used to assess whether the errors are
significantly non-random.
Statistical Compromises
Probably the most commonly asked question in water quality monitoring is
whether significant change has occurred. The use of inferential statistics
requires that this question be more properly phrased as "Can we conclude
that there is a difference between population A and population B based on
samples drawn from those two populations?" If the data are collected in the
context of a proper statistical design and meet the criteria discussed in the
previous section, then our ability to answer this question depends on five
interacting factors:
1. sample size,
2. variability,
3. level of significance,
4. power (i.e., the probability of detecting a difference when on exists),
and
5. minimum detectable effect.
33
-------
The important aspect of these five factors is that a change in one factor will
affect some or all of the other factors. In general, any improvement in one
factor will come at some cost. This cost could be with regard to one or more
of the other factors, increased sampling costs due to an increase in the
sample size, or a change in the statistical design. The technical specialist
and the manager must realize that there is no perfect solution in statistics,
and an explicit recognition of these trade-offs is a necessary stage in design-
ing a statistically-sound monitoring plan. The following paragraphs discuss
each of these five factors and their relative effects on monitoring costs,
uncertainty, and risk. Note that these trade-offs are discussed with regard to
a comparison of the means from two populations using parametric statistics,
but the same principles apply to all statistical tests using both parametric and
nonparametric statistics.
Sample size. The relationship between sample size, accuracy, and uncer-
tainty is generally understood. For example, a larger sample size will reduce
the difference between the sample mean and the true population mean. A
larger sample size also will reduce the standard error of an estimated param-
eter. (The standard error is the standard deviation of a particular descriptive
statistic, such as the sample mean. Usually it is estimated for a population
from a single sample.) It follows that a larger sample size increases the
ability to detect a difference between two populations because less uncer-
tainty is associated with the estimated population means. Unfortunately this
increased ability to detect a difference between two populations is a logarith-
mic function of the sample size rather than a linear function. This means that
increasing the sample size may make a substantial difference if there are
very few samples (e.g., less than five or ten), but the benefits of increasing
the sample size beyond about thirty or forty generally are very small unless
the parameter is highly variable.
The statistical trade-off associated with a larger sample size is that it increase
the likelihood of concluding there is a statistical difference when in fact there
is no difference (Type I error; see discussion on power below). For most
water quality monitoring programs Type I error is not a major problem. Typi-
cally the costs of sampling and the inherent variability mean that one is more
likely not to detect a difference when in fact there is a significant difference
(Type II error; see discussion on level of significance below). The other
problem associated with increasing the sample size is that each additional
sample has certain cost, and one must evaluate the marginal benefits of an
additional sample as compared to its cost in terms of drawing resources away
from other activities. If this marginal cost is acceptable, it usually is statisti-
cally advantageous to increase the sample size.
A variety of statistical techniques can be used to estimate the appropriate
sample size give the statistical objectives and the known variability of the
parameter. Often information on the variability of parameter can be obtained
from previous studies. If no information is available, usually it is best to
conduct a small pilot study. This will not only provide some basic information
on the parameter(s) of interest, but also provide an opportunity to refine the
34
-------
objectives and techniques. If no information is available and a pilot study is
not possible, a sample size of "about 10" often represents a reasonable
compromise between the cost of sampling and the need to reduce the uncer-
tainty of the population estimates. This estimate is based on the minimum
detectable effect for the t-test (Figure 3).
Variability. The most common statistic used to describe the variability of a
sample is the.standard deviation. The square of the standard deviation is the
variance. Both the standard deviation and the variance are expressed in the
same units as the mean. Dividing the standard deviation by the mean yields
the coefficient of variation, and multiplying this by 100% provides a standard-
ized measure of the relative variability in percent of the mean.
As suggested above, the variability of a parameter in time and space is
inversely related to the ability to detect significant change. Increasing the
sample size can help compensate for a high degree of variability, but since
the standard deviation is a square root function this is subject to diminishing
returns. The other, more difficult means to decrease the variability is by
improving measurement techniques and sampling methodology. Errors in
measurement can spring from a wide variety of sources, and these include
equipment problems, actual measurement problems, transcription problems,
and inaccurate data entry. Reducing these errors can be very time-consum-
ing, but it is a necessary part of the quality assurance and quality control
aspects of any monitoring project.
The other effective means to reduce the variability of the data is to modify the
sampling scheme. If some of the factors causing the variability can be
identified, then the data can be more efficiently stratified. A series of statisti-
cal techniques can be applied to evaluate the efficiency of existing strata, and
to optimize sampling among the different strata.
In some cases a reduction in variability can be gained only by narrowing the
scope of the investigation or objectives. Often the natural variability in the
streams being monitored makes it difficult to answer broad management
questions. Narrowing the question allows a more focused investigation, a
reduction in unaccounted variability, and improved statistical resolution. For
example, bed material particle size might be evaluated only in a very specific
location, such as the deepest portion of certain types of pools, or the down-
stream edge of point bars. A change in fish populations might be more
narrowly defined as a change in the number of 1 + steelhead. Basins subject
to landslides or debris flows might be considered separately. One also might
reevaluate the parameters being measured. In contrast to simply increasing
the sample size, all these approaches require more information about the
streams and the parameters being monitored, and hence more involvement
of the technical staff. This has advantages for interpreting the monitoring
results, and for designing future monitoring plans.
Level of significance. The level of significance refers to the probability that
an apparently significant difference is not real but simply due to chance. This
35
-------
/s the a value listed in statistical tables and shown graphically in Figure 2.
Convention has the a value set at 0.05 for most statistical tests, and this
means that there is only a 1 in 20 chance that an observed difference is due
to chance (Type I error). A stronger level of significance (i.e., lower a) indi-
cates a higher lever of confidence that the difference is real, and a lower
probability that it was due to chance. The most common ways of obtaining as
stronger level of significance are to: (1) increase the sample size, and
(2) reduce the variability by altering the measurement techniques or sampling
design.
Figure 2A graphically shows that a comparison of samples from two normally
distributed populations with quite different means has both a strong level of
significance as indicated by the shaded area in Population A where a = 0.05,
and a high power (i.e., a high probability of detecting a difference when in fact
there is a difference) as indicated by the large unshaded area in Population
B. Figure 2B indicates that as the two populations become more similar,
there is a decreasing ability (i.e., declining power) to detect a significant
difference when the level of significance is kept at 0.05. Figure 2C is similar
to Figure 2B, except that the variability of population A and B has been
reduced (i.e., the estimated means of the two populations have less uncer-
tainty). This increases the power for the same level of significance.
The selection of an a level is purely arbitrary and should reflect the values
and risks associated with each of the other four factors discussed here. In
most cases a is set at 0.05, but for many water quality applications a higher a
level (or weaker level of significance) may be more appropriate. One justifi-
cation for a higher a is that an objective of most monitoring programs is to
identify changes in water quality due to management. However, by the time
an adverse change has been detected, adverse effects on the designated
uses may already occurred, and restoration or recovery may be a long or
costly process. Hence it may be preferable to try and identify change earlier
by decreasing the level of significance, even though this will simultaneously
increase the likelihood of identifying change when it has not actually oc-
curred. In other words, the cost of not identifying adverse changes is greater
than the cost of erroneously detecting change, and this is the basis for the
trade-off between the level of significance (a) and power (1 - (3).
Figure 2D illustrates that a weaker level of significance will increase power. If
there is considerable overlap between the two populations, a substantial
increase in power can result from a relatively small increase (weakening) in
the level of significance.
An example of a situation where a weaker level of significance might be
appropriate in monitoring for bacterial contamination is a stream used for
domestic water supplies (Ponce, 1980b). In this case the cost of failing to
detect contamination is quite high, and it usually is better to have more false
warnings that to miss a contamination episode. Other reasons for adopting a
weaker level of significance include the high variability of water quality param-
eters, and the costs associated with increasing the sample size in order to
achieve a stronger lever of significance.
36
-------
Population A
a = 0.05
A large difference in the estimated mean values
of a parameter for population A and population
B results in both high power (1-P) and a strong
level of significance. This is the ideal situation.
Population B
Population A
B.
a = 0.05
A smaller difference between the estimated
mean values for populations A and B results
in lower power at the same level of
significance.
Population B
Population A
a = 0.05
C.
Reduction in the variability of populations
A and B results in more power at the
same level of significance, even though
the population means are unchanged
from Figure 5B.
Population B
Population A
= 0.10
A slightly weaker level of significance
substantially increases the power for
the same population means and
variances as in Figure 5B.
Population B
Figure 2. Schematic representation of the trade-offs among level of significance, power, and variability for two
normally distributed populations. The figures assume a one-tailed t-test is being used to determine if a
significant difference exists between the two populations.
37
-------
One should also keep in mind the distinction between the statistical level of
significance and the level of significance relevant to the designated uses of
water. A 25% decline in outmigrating salmonids may not be statistically
significant because of the large interannual variability, but a loss of 25% of
the outmigrating salmonids due to poor water quality or habitat deterioration
is a serious impairment of the designated use for fisheries. Conversely, a
level of significance of 0.05 for a given water quality parameter does not
necessarily mean that a designated use is impaired. The point is that the
statistical results must be interpreted by the specialist and the manager, and
this requires an understanding of the physical and biological functioning of
the water body being monitored.
Power. The power of a statistical test is the probability of detecting a differ-
ence when in fact there is a difference (Mendenhall, 1971). This probability is
usually designated as 1 - p. When comparing two sample means, the quan-
tity of 13 is commonly known as Type II error. Type II error can be described
as the probability of incorrectly concluding that two populations are the same
when in fact they are different. Since power is closely related to the level of
significance, it exhibits a similar response. An increase in sample size usu-
ally will increase the power of a test by reducing the uncertainty around the
mean. The greatest increase in power occurs at small sample sizes. More
variability increases the overlap between two populations, thereby decreasing
the power of a test. Decreasing the minimum detection limit will increase the
allowable overlap between two populations. This enhances the possibility of
Type II error and correspondingly reduces the power.
Minimum detectable effect (MDE). A key factor that should be considered
in the design phase of most monitoring programs is the minimum change one
wishes to detect. Usually this question is not explicitly considered, even
though it is directly related to sample size, parameter variability, level of
significance, and statistical power (Green, 1989). Any decision regarding the
desired minimum detectable effect also must consider the sensitivity of the
different designated uses in the streams being monitored. In other words,
how much change is acceptable in the parameter being monitored before a
designated use is impaired? Although in some cases the answer might be
that no change is acceptable, this not a statistically acceptable answer be-
cause no monitoring program can detect an infinitesimal change. An explicit
discussion of the MDE is helpful in forcing the manager and the technical
staff to agree on specific, quantitative objectives for the monitoring plan and,
by implication, for management impacts.
For some parameters, such as pH, a change in either direction is significant,
and this expands the allowable zone of change if the level of significance is
kept constant (i.e., a two-tailed test instead of a one-tailed test). If the con-
cern over possible change is only in one direction, such as an increase in
suspended sediment concentration, then the limit of allowable change will be
slightly less (i.e., a one-tailed test).
38
-------
I
(0
o
o
For most trend, project, and effectiveness monitoring, an explicit MDE target
should be established. Setting an MDE in validation monitoring may be
helpful in defining the uncertainty in the model being validated. The MDE
may be an important issue in compliance monitoring it the numerical standard
is expressed in terms of percent change above background. Monitoring for
such a standard is often difficult because of the need to first determine back-
ground, and then to evaluate the numerical change in a sometimes highly
variable parameter. For example, some States permit forest harvest activities
to increase turbidity by only 10 or 20% above background. In view of the
temporal variability associated with turbidity, this standard becomes very
difficult to enforce.
Figure 3 graphically shows the percent change detectable according to
sample size and the coefficient of variation. This assumes a constant signifi-
cance level of 0.05 and a power of 80%. Changes in the significance level
(e.g., to 0.10) or power (e.g., to 90%) would not substantially alter the form or
values in Figure 3. In very rough terms, Figure 3 indicates that the treatment
effect, expressed as a percent of the mean, will have to be somewhat larger
than the coefficient of variation for detection when a = 0.05. Increasing the
sample.size is an increasingly effective tactic to decrease the MDE as the
coefficient of variation increases. If a relatively small MDE is desired, one
must have a correspondingly low coefficient of variation, and this has impor-
tant implications for selecting parameters.
300 -
250
200
c
0)
'o
150
100
CO
I
1
J 50
10
15
20
150%
125%
100%
75%
50%
25%
15%
25
30
Sample size (per group)
Figure 3. Maximum allowable coefficient of variation to detect changes ranging from 15 to 150%. Figure
assumes a two-sample t-test is being used to detect change at a 5% level of significance and a power of 80%.
The labeled curves show the minimum percent change that can be detected given a particular coefficient of
variation for the parameter being measured and population sample size (figure courtesy ofL. Conquest, Center
of Quantitative Studies, University of Washington).
39
-------
-------
Selecting the Appropriate BMP Effectiveness Monitoring
Level
Monitoring Levels
The monitoring level selected must be appropriate to the objective of the monitor-
ing program. Four levels of monitoring are used for evaluating the effectiveness
of forestry BMP's in meeting water-quality goals or standards: BMP I, II, III and IV
(Table 3). They are defined as follows:
BMP Effectiveness I uses empirical observations and a limited amount of
qualitative data to evaluate effectiveness. A heavy reliance on experience is
used to judge whether land management or BMP's are effective in protecting
water quality. It is a professional judgment as to whether there is an obvious
problem or not.
.BMP Effectiveness II depends mainly on qualitative data and limited quanti-
tative data for analysis leading to the evaluation of effectiveness.
BMP Effectiveness III relies predominantly on quantitative data, with some
qualitative data for detailed analysis to evaluate effectiveness.
BMP Effectiveness IV relies predominantly on quantitative data, with some
qualitative data for detailed analysis to evaluate effectiveness, to establish
water-quality criteria, goals, or cause and effect. This level of monitoring is
usually associated with research.
Low-level monitoring (BMP Effectiveness I and II) identifies high-level needs and
also extends the information developed by high-level monitoring. High-level
monitoring (BMP Effectiveness III and IV) develops information that supports or
validates decisions made in low-level monitoring.
The level selected is determined by:
1. The type of question or issue to be addressed.
2. The type, number, and skill of people required.
3. The number and stratification of reference and study streams.
4. Representative monitoring methods available for evaluating: (a) on-slope
conditions, (b) chemical and physical components, (c) channel geomor-
phology and stability, (d) biology, (e) habitat, and (f) aquatic vegetation
(TableS).
5. Amount of data to be collected, time required, and number of streams that
can be monitored,
6. The quality of data needed for making decisions.
7. The quality of decisions needed and the potential risk in making the wrong
judgments.
Tables 3 and 5 should be viewed as guides only, not to be rigidly followed in
selecting monitoring methods to use. For example, a user can use a level 3
method at level 2 to answer a specific question.
41
-------
are
, with high
g
ion
Decisions
subjective,
risk of mak
wrong decisi
for grey area
situations.
(0 - C CD
c 5 ^5
-
5 g ^5 c -o
.2 » E •- o w CD .
•e s s>.°i « s s
lilill
Sf
ll
uality
Data For
Decision
•s-gsE-ga
Ł CO
a .Wg 2 =g "g -5
co
co -o 5.-S .9 5 0 I
s -5 &
Ti to
Decision
A few hours to
or 2 days to a
week.
co »
I
I
5*1
Is!
< °
g E
E co
0 g>
CD —
§
u w
CO (Q
853
ceo 85
Ł >,=
O 73
"S3 3
ecw
o
n CD ,
CO To 0 2 8 To
CJ IN "3. to
«
si
(/}
0)
Q..E
^12 -5
Ł|rfS||
•n — .g "= CD =
CM
..E -a J
2 Ł
S
Qution
or
Issues
I
11
5 08 ^ -o .2 .S o-
g-S Ł Ł=5 g-s
i §
O>
Ii
42
-------
There are two ways to interpret Table 3—horizontally or vertically. The horizontal
presentation shows the various factors involved in the monitoring process. The
vertical approach documents the differences among levels of monitoring and
shows the progression from low to higher levels of monitoring, allowing for
rational discussion of monitoring subjects. The vertical approach is also used for
discussions of representative monitoring methods (Chapter 6).
Monitoring Objectives
The type of question or issue raised will define monitoring objectives. These
objectives, and the associated significance to society and to the manager, will
determine the monitoring level to use. Certain kinds of questions or issues call
for specific monitoring levels. The following list of questions or issues is not
exhaustive, but suggests the level or nature of the objective(s) suited to a specific
level of monitoring.
BMP Effectiveness I
A. Issue: A complaint about a management practice's impact on a stream.
This monitoring response would quickly determine if there is an obvious
impact. .If effectiveness is not empirically obvious, a higher level of moni-
toring might be needed to determine whether management is protecting
the physical, chemical, and biological integrity of streams.
B. Issue: Need a quick and inexpensive way to determine which BMP's are
obviously meeting water-quality goals.
C. Issue: Need early warning as to which BMP's are effective.
Caution: Low levels of monitoring will detect which management activities,
with proper implementation of BMP's, obviously have or have not protected
water quality, but uncertainty will prevail in many cases. These will require
more detailed data collection and analysis to determine effectiveness and
could call for a higher level of monitoring. Careless or poorly trained observ-
ers will tend to classify BMP's as effective when in fact they may not be.
The key to accurate professional judgment is experience and knowledge
gained from monitoring streams in the ecoregion at higher levels and apply-
ing that experience and knowledge at this level in the same ecoregion. An
alternative is sound training by individuals experienced in the ecoregion.
BMP Effectiveness II
A. Issue: Need to respond to a complaint concerning a high-value stream.
B. Issue: Need to respond to a concern by a water-quality agency.
C. Issue: Trends of water quality.'
43
-------
BMP Effectiveness III
A. Issue: Need to respond to a complaint about management impacts on a
high-value stream with high public interest, such as outstanding national
resource waters or water-quality-limited streams.
B. Issue: Need to define biological criteria for water-quality standards.
C. Issue: Need to define desired future conditions under Forest Land Man-
agement Plans.
D. Issue: Forestry compliance with State water-quality standards.
E. Issue: Determining the need to modify BMP's.
BMP Effectiveness IV
A. Issue: Need to determine the cause and effect of management on the
physical, chemical, and biological integrity of streams.
B. Issue: Determining the need to modify BMP's.
C. Issue: Need to develop data and information to respond to a law suit.
Skills Needed
The professional skills needed to monitor effectiveness increase with level.
Likewise, the number of professionals and technicians needed increases as well
as the number and types of expertise. An agency may have the skills needed at
low levels, but may not have them at higher levels. Interdisciplinary monitoring is
usually required and may require interagency cooperation to do the monitoring.
Only professional, qualified, trained, or supervised people should be involved in
effectiveness monitoring.
, BMP Effectiveness I
Needed: One professional, one professional and a technician, or two profes-
sionals. They must be trained in the empirical/qualitative methods used,
have the experience and knowledge to judge whether BMP's are clearly
effective or not, and be able to discern when uncertainty occurs (called a grey
area) and higher level methods and analyses are needed. Two people are
often-needed to take some measurements and record.
BMP Effectiveness II
Needed: Two professionals, trained in hydrology, fisheries, benthic inverte-
brates, and habitat, plus technicians. They must be trained and experienced
in the qualitative methods used and the analysis procedures, have the experi-
ence to judge from the qualitative data whether BMP's are effective or not,
and be able to discern grey-area situations. Interdisciplinary and interagency
cooperation may be needed to bring together the necessary skills.
44
-------
BMP Effectiveness III
Needed: Professionals in statistics, hydrology, fisheries, benthic inverte-
brates, aquatic habitat, channel geomorphology, and other professions.
Which of these will be needed depends upon the situation and the stream/
ecosystem/upland elements being monitored. Quantitative assessment
clearly requires interdisciplinary effort and may require interagency co-opera-
tion to bring together the professionals needed.
BMP Effectiveness IV
This level of monitoring requires the same professionals as BMP Effective-
ness III, but these people will likely be researchers from various agencies,
universities, and other organizations.
Reference Condition
Reference streams or stream segments serve as controls and are essential to
determining the effectiveness of management or BMP's in meeting the physical,
chemical, and biological integrity of water. Reference streams are being used by
many State water-quality agencies to define the reference condition for biological
criteria for water-quality standards. Reference streams are necessary whether or
not biological criteria have been established by the State. If the State has not
established reference streams for the ecoregion in question, then the monitoring
project plan and design must include the selection and establishment of one or
more reference streams and a similar number of study streams. For the latter
situation, the classification and selection of reference and study streams ought to
be done with the full cooperation and support of the State water-quality agency.
The reference condition defines the range of biotic, habitat, chemical, and physi-
cal variables in reference streams. Biotic, habitat, and other water-quality vari-
ables vary temporally and spatially in reference streams. Several reference
segments (10+) along a reference stream or in several separate reference
streams are assessed for their physical, chemical, biological, and habitat condi-
tion. The data are analyzed to determine the variability of water-quality variables,
thus the reference condition.
Reference condition should be determined by a professional biologist in coopera-
tion with other specialists. It is not a simple task. In defining reference condi-
tions, experience has found they are not just a product of abiotic conditions. For
example, even with the best physical habitat, the site must be assessed and the
biotic community data analyzed to determine if the site contributes to defining the
reference condition.
The reference condition should be established using the highest monitoring level
possible. The ability to define classes of water quality and ranges in parameters
and to set water-quality criteria depends on methods used, number of samples,
sample locations, accuracy of measurements, metrics, and analysis procedures.
Using a high-level monitoring may yield more classes of water quality, a narrower
range of reference conditions, and better water-quality criteria than using a lower
level with simpler methods.
45
-------
It is important to consider reference streams and their role in water-quality man-
agement under the Clean Water Act, EPA guidance to States, and in developing
biological criteria. The expression "biological integrity" is used in the Clean
Water Act to define the Nation's objective for water quality (Karr and Dudley,
1981) (EPA 1990). EPA (1990) defines biological integrity as follows: "According
to Webster's New World Dictionary (1966), integrity is, 'the quality or state of
being complete; unimpaired.' Biological integrity has been defined as 'the ability
of an aquatic ecosystem to support and maintain a balanced, integrated, adap-
tive community of organisms having a species composition, diversity, and func-
tional organization comparable to that of the natural habitats within a region'
(Karr and Dudley 1981). For the purposes of biological criteria, these concepts
are combined to develop a functional definition for evaluating biological integrity
in water-quality programs. Thus, biological integrity is functionally defined as
"the condition of the aquatic community inhabiting the unimpaired water bodies of
specific habitat as measured by community structure and function." Ecological
integrity is attainable when physical, chemical, and biological integrity occur
simultaneously (Figure 4) (EPA 1990). Biological integrity includes habitat.
EPA (1990) recognizes the difficulty in finding unimpaired water to define biologi-
cal integrity and to establish reference conditions. However, the structure and
function of aquatic communities in high-quality streams can be approximated in
several ways. First, characterize aquatic communities in the most protected
waters in the ecoregion. In regions where few or no unimpaired sites are avail-
able, characterize the least impaired systems. In greatly disturbed regions, least
impaired streams are those with the least-impact from human activities. Because
least-impaired systems will be used as references, the limitations on biological,
chemical, and physical integrity can be considered and incorporated into goals
for those waters and the program to improve water quality.
EPA is using Omernik's (1987) ecoregion procedure for grouping streams.
Within these ecoregions, reference streams are being selected and assessed to
characterize the biological, physical, and chemical integrity for water quality
criteria. Omernik's system uses an ecoregion framework for interpreting spatial
patterns in State and national water quality data. His ecosystem classification is
based on regional patterns in land-surface form, soils, potential vegetation, and
land use. He recognizes the need to refine the classification within the broad
ecoregions by considering stream size, hydrologic regime, and riparian vegeta-
tion. In addition to Omernik's (1987) factors for grouping streams in ecoregions,
there are several more factors that can influence the physical, chemical, and
biological integrity of water (see Table 4).
46
-------
Figure 4. Ecological integrity is attainable when chemical, physical, and biological integrity occur
simultaneously (EPA 1990).
Table 4. Factors that can influence, the physical, chemical, and biological integrity of water
1. Prior land use
2. Stream order or size
3. Drainage area
4. Geology, soils
5. Stream gradient
6. Sinuosity
7. Aquatic vegetation and cover
8. Vegetation on slopes
9. Riparian vegetation
10. Amount of canopy over or shading stream
11. Sensitivity of watershed to change
12. Aspect ,
13. Fishing pressure
14. Hyporheic zones
15. Wetlands adjacent to streams
16. Elevation ;
17. Beaver dams and ponds
18. Water withdrawal or augmentation
19. Presence or absence of large woody debris
20. Volume and quality of ground water
21. Substrate type and heterogeneity
22. Substrate stability
23. Acid rain
24. Introduced aquatic species
47
-------
Understanding how these factors influence the physical, chemical, and biological
integrity of streams can help to fine tune classification of reference and study
streams. For example, sediment from past abusive land use has been stored in
stream channels and flood plains, but is now being removed from storage.
Attached to these sediments are pollutants from past discharges. Also, prior land
use affects the quality of ground water being discharged to streams today.
Fish and benthic invertebrate communities are influenced by stream size, stream
order, watershed area, elevation, stream gradient, and substrate type, stability,
and heterogeneity. Organ energy and functional feeding-group communities are
influenced by the type of detritus from riparian vegetation, the amount of the
stream's solar exposure, the presence of large woody debris, the type of detritus
inputs from ephemeral streams from vegetated slopes, and the type and hetero-
geneity of substrate. The physical and chemical character of water is affected by
geology, soils, riparian vegetation, slope vegetation, presence of wetlands, and
water addition and withdrawal. Geology and soils influence the sensitivity of
watersheds to change and affect landslide risk. Beaver dams and ponds change
the hydrology and water quality characteristics of streams.
National and State efforts to classify streams by ecoregions need to be refined to
focus on the influences affecting the physical, chemical, and biological integrity in
the forest environment. A refined stream/ecosystem classification will eliminate a
lot of the "noise" in the system and minimize the variation in conditions being
compared between the reference and study streams. A good understanding of
the forest stream/ecosystem will lead to better monitoring design, better data
collection, better interpretation of data, and better water-quality management.
Barbour, M. T. et al. (1991) discuss the role of stream classification in bioassess-
ment. "One function of classification is to increase the resolving power or sensi-
tivity to biological surveys to detect impairment by partitioning variation within
selected environmental parameters or among sites. The importance of minimiz-
ing variation...Clearly it is easier to distinguish impairment if the parameters have
low variability. Formal statistical tests (parametric and nonparametric) indicate
greater resolution and power exist if there is low variance within elements being
compared. Effective classification leads to improving resolving power by parti-
tioning or accounting for variability. A coarse classification yields higher variance
and therefore lower resolving power; vice versa for finer classifications."
The USDA Forest Service is refining stream/ecosystem classifications for
streams and watersheds on national forests. This effort will lead to finer stream
classifications than Omernik's for ecosystem and water resource management
on national forests. This refined classification may also be useful to State water-
quality agencies in classifying some of their streams and ecoregions.
Stream/ecosystem classifications need to reviewed and refined after streams and
watersheds have been subjected to major streamf low events that alter stream
channels. To re-establish reference conditions after major storms, reference
and study stream segments may need to be reclassified to ensure similarity and
detect different responses between reference and study streams subjected to the
same storm.
48
-------
There are three ways that reference streams can be used in monitoring:
1. Reference Reach—A stream reach that has the same stream type and is
in similar condition as the study reach being monitored for management
impacts is called the "reference reach" and serves as the control. Refer-
ence and study reaches need to be in neighboring watersheds. It is
assumed that the same storms, snowmelt rate, etc. and resulting
streamflow will prevail on both the reference watershed and reach and the
study watershed and reach (Figure 5).
2. Above vs. Below—Reference reaches on the same stream above a
management activity are compared to study reaches below (Figure 6).
Reference and study reaches need to be in near each other. Measure-
ments are taken concurrently on both reaches.
3. Before vs. After—Calibrating of individual reach conditions before imple-
menting a potentially impacting activity establishes a "baseline" against
which the effect of the activity can be measured (Figure 7). Ideally, the
period of evaluation will span similar weather-patterns (dry vs. wet years,
unique storms, etc.)
Study
Reach
Reference
Reach
Figure S. Reference reach channel classification similar to study reach.
49
-------
Management Area
Reference
Reach
above
Study
Reach
below
Figure 6. Above and below monitoring using reaches with the same stream channel classification.
Before vs. after
comparison on
same stream reach
Figure 7. Before and after comparison on the same stream reach.
50
-------
Reference streams are usually in separate watersheds from the study streams,
especially for establishing biological criteria. However, reference stream seg-
ments above a management activity can be used to evaluate study segments
adjacent to and below the activity. Again, the reference and study segments
should be as similar as possible. Several reference segments representing least-
disturbed conditions should be assessed to define a reference condition.
Guidelines for selecting reference and study streams vary by BMP effectiveness
level:
BMP Effectiveness I
At this level of monitoring, there are several alternatives for selecting refer-
ence and study streams or segments: .
1. If State reference streams for the ecoregion are similar to the proposed
study streams, the empirical/qualitative monitoring can proceed directly to
comparing the two.
2. If a State reference stream does not exist for the ecoregion, reference
streams or segments similar to the study stream must be selected. A
nearby stream in a least disturbed condition could be selected, assessed,
and viewed to develop a reference condition for judging effectiveness. A
mental picture of the habitat and biological health, and/or empirical evi-
dence and qualitative data indicating the physical and chemical integrity of
the reference stream would serve as the basis for comparison to the study
stream.
3. A series of photographs of reference conditions can be compiled in a
notebook and used for comparing study stream conditions: for example,
pictures of stream types, riparian vegetation, stream bank stability, habitat
types, and life forms associated with reference conditions. This tool would
be useful in judging the effectiveness of BMP's and management prac-
tices for obvious yes-or-no in meeting water-quality integrity.
Reference segments selected above the managed area must represent least-
disturbed conditions, not impaired waters. Making such a selection requires
knowledge of watershed and stream conditions above the activity. To get this
knowledge, the team would have to walk out the stream or use aerial photo-
graphs to judge whether the stream segment above is least impaired. Im-
paired segments must not be used as references, because (1) the impacts
from above may obscure the impacts associated with the activity under study,
(2) such references would define a low level of physical, chemical, and
biological integrity as acceptable, which could lead to a management activity
or BMP being judged as effective in meeting water-quality goals or standards
when in fact it impairs the integrity of water.
The criteria to use in selecting references will depend upon the ecoregion and
the key habitat needs of the aquatic species. Some general factors to con-
sider include stream order, channel classification (for example, Rosgen's or
Alaska channel classification, discussed under Channel Geomorphology
, 51
-------
Monitoring Methods), type of riparian vegetation, etc. Using a few additional
factors will lead to more similar streams being compared.
BMP Effectiveness II
The issue, monitoring design, and statistical requirements determine the
number of reference and study streams, if the State has not defined refer-
ence conditions for the ecoregion. Qualitative and some quantitative data are
collected for analysis of effectiveness. BMP Effectiveness II issues require
detecting significant changes in the physical, chemical, and biological integ-
rity of water. Therefore, the reference and study streams need to be similar
to minimize variation and to increase ability to detect impacts of manage-
ment Reference and study streams must be in the same ecoregion. Addi-
tional classification factors to be considered include: stream size, channel
classification (for example, Rosgen's or Alaska channel classification, dis-
cussed under Channel Geomorphology Monitoring Methods), type of riparian
vegetation, type of vegetation of watershed slopes, type of substrate, and
possibly aquatic habitat units.
BMP Effectiveness III
Again, the issue, monitoring design, and statistical requirements will deter-
mine the number of reference and study streams, which need to be as similar
as possible. Stream similarity will be achieved using a detailed stream/
ecosystem stratification system. If key habitat elements are known for intoler-
ant aquatic life forms, these can be used in stratifying streams. If the key
habitat elements need to be determined, the stratification will need to be
detailed enough to minimize variation and allow identifying the key habitat
elements from the data. Experts in the various aquatic life forms can help in
selecting key factors to use in the stratification process.
A statistician needs to be involved in monitoring and can assist in determining
the number of reference and study streams needed, which will depend on
variability, level of change needed to be detected, and level of significance.
BMP Effectiveness IV
The stream/ecosystem stratification detail for this level is the same or slightly
more detailed than for BMP Effectiveness III.
Amount of Data Collected
The amount and type of data needed vary significantly among monitoring levels,
reflecting the number and complexity of methods used. It is important at all
levels of monitoring that data be collected systematically according to a plan that
specifies parameters. The observers must be properly trained to ensure consis-
tent interpretation of conditions, measurements, and sample collection. Field,
laboratory, and analysis forms need to be developed to ensure nothing is over-
looked. Simple forms will suffice for low levels of monitoring, but will increase in
complexity and detail with higher levels. Recording data on forms facilitates
entry into a database. Data have lasting value only if stored in a database and
can be readily accessed and analyzed. Analysis of a monitoring database can
identify BMP's that are consistently effective, those that are only effective in
52
-------
specific situations, and those that need modification and can reveal in-stream
responses to BMP's implemented in various ecoregions. ••'•
Air quality assurance/quality control (QA/QC) program is needed at all levels of
monitoring to ensure consistency of interpretation and accuracy of data. This
includes checking performance of observers. It will vary by monitoring level and
the variable being assessed. See Chapter 3 for more details.
Following is a general discussion of type and amount of data to be collected by
monitoring level:
BMP Effectiveness I
Qualitative data will dominate, with some empirical observations. A small to
moderate amount of data is recorded on a field form, which can be completed
quickly. Recorded observations on reference and study streams will permit
judging whether the BMP's are obviously effective or not in meeting water-
quality goals or standards.
BMP Effectiveness II
Qualitative data will dominate, with a few measured observations. A moder-
ate amount of data is collected.
BMP Effectiveness III
Quantitative data will dominate, with limited qualitative data. The amount of
data collected can be large and the forms more complex.
BMP Effectiveness IV
Quantitative data will dominate, with limited qualitative data. The volume of
data will be very large, reflecting the number and detail of monitoring meth-
ods used. The forms will be lengthy and complex.
Number of Streams Evaluated
The number of streams that can be evaluated will depend upon the time and
resources available for monitoring. Given limited resources, as monitoring
intensity increases from BMP Effectiveness I to IV, the number of streams and
management activities that can be evaluated decreases significantly.
Length of Study
The study begins with planning monitoring and ends when a decision is made
concerning BMP effectiveness in meeting water-quality goals or standards.
Exact time frames cannot be predicted for each monitoring level, but general
estimates can be made. The complete job includes defining the issues, selecting
reference and study streams, designing and implementing the monitoring pro-
gram, collecting and analyzing the data, entering data into a database, reporting
results, and making the decision and recommendations.
53
-------
54
BMP Effectiveness I
Usually a single site is visited to assess BMP effectiveness. Depending on
the stream ecosystem classification used, the number of in-stream and on-
slope observations made, and data and information analyzed, the time
required may be a few hours to a week per site.
BMP Effectiveness II
Assuming more than one reference stream and study stream are being
monitored, and depending on the amount of data collected per site, it may
take 2 weeks to a month to reach a decision. In some cases, repeated visits
to the sites may be needed to assess trends in recovery or to sample popula-
tions when they are at their peak. In such cases, the decision may be de-
layed.
BMP Effectiveness III
One-time visits may be made to sites on study streams for comparison with
reference conditions. Even so, the amount of time to decision can be several
weeks because of the number of streams, the detailed classification, the
amount of data collected and analyzed, and detailed report preparation. Two
to three months may be needed to do the complete job.
BMP Effectiveness III can involve monitoring management practices in
several study watersheds and comparing them with several reference
streams. One to three years may be required before decisions are made with
this type of monitoring design.
A stream may be evaluated by segments or its whole length. The intensity of
monitoring and length of streams will determine the time to decision.
When the stakes are high, decisions about specific management practices
often need to be made quickly at this level. Detailed BMP Effectiveness III
monitoring can be done by concentrating manpower and resources in a short
period of time, thereby decisions can be reached in a few days or weeks.
BMP Effectiveness IV
Because this level usually involves research, the length of study will likely be
at least 2 years.
Quality of Data for Decisions
The type and quality of data vary by monitoring level, methods used, and the skill
of the people involved. Three types of data are taken: (1) empirical, (2) qualita-
tive, and (3) quantitative. Quality is determined by the precision of measure-
ments or observations.
The methods used at each level do not always clearly reveal whether goals or
standards have been met. This uncertainty is termed the "grey area." As the
monitoring level increases, the grey area decreases (Figure 8). Within these
grey areas, the tendency is to assume the goals or standards have been met,
even though higher levels of monitoring might reveal impairment. Uncertainty
occurring with BMP Effectiveness I and II monitoring could trigger the use of
higher levels of monitoring to resolve grey-area issues.
-------
Monitoring
Level
IV
Area of
Discernment
Area of
Uncertainty
(Grey Area)
Area of
Discernment
.' ' | Unimpaired | ? | Impaired |
Water Quality Condition
Figure 8. The area of discernment of water quality condition by BMP effectiveness monitoring level.
BMP Effectiveness I
Qualitative data dominates, with some empirical observations and quanti-
tative data. Because BMP Effectiveness I relies on knowledge and experi-
ence gained from higher levels of monitoring, the observations and data
should permit discerning when BMP's have clearly met water-quality goals or
standards and clearly when they have not. However, the grey area may be
large and should be recognized.
BMP Effectiveness II
Most of the data are qualitative, with moderate amounts of quantitative data.
Level II monitoring should also benefit from knowledge and experience
gained from higher levels of monitoring so the observations and data should
be good for discerning when BMP's have clearly met water-quality goals or
standards. The amount of data collected at this level is significantly more
than for level 1. The precision of data also increases. The grey area is
moderately sized. A wider range of effectiveness can be judged.
BMP Effectiveness III
Quantitative data with a high level of precision will dominate. Qualitative data
are limited, but sound. The grey area is small, with most significant impacts
detected.
BMP Effectiveness IV
Carefully collected quantitative data are coupled with a small amount of
qualitative data. The precision is excellent. Methods used can detect major
impacts and significant subtle changes in the physical, chemical, and biologi-
cal integrity of water.
55
-------
Quality of Decisions and Risk
There is risk in making the wrong decision about the effectiveness of BMP's.
When the stakes are high, the risk needs to be minimized. Risk can be mini-
mized by using a higher level of monitoring or recognizing the limits of the meth-
ods used at each level of monitoring. Usually the obviously effective and ineffec-
tive BMP's and management practices are easily recognized. The zone in
between is the grey area and too often these situations are judged as effective,
when they are not.
Decisions are subjective or objective. Subjective decisions, primarily based upon
perception, judgment, and experience, should be backed by data and information
derived from higher levels of monitoring. Objective decisions are based upon
measurements, data, and analyses, although some judgment is involved.
BMP Effectiveness I
Decisions are subjective, with high risk in grey areas. The quality of informa-
tion depends on the quality of training and skill of the observers. BMP Effec-
tiveness I is most useful when BMP Effectiveness III and IV assessments are
available to guide BMP Effectiveness I observations.
BMP Effectiveness II
Enough data are available to make objective decisions, yet judgment plays a
major role in the decision. The grey area is reduced, thus reducing risk.
BMP Effectiveness III
Objective and informed decisions are based upon data. The risk is low and
the decisions defensible.
BMP Effectiveness IV
Objective, data-based decisions are very defensible. The risk is low.
56
-------
Monitoring Methods
Monitoring methods are organized into six categories: (1) on-slope, (2) chemical
and physical, (3) channel geomorphology and stability, (4) biological, (5) habitat,
and (6) aquatic vegetation (Table 5). The methods listed under each heading in
Table 5 merely represent the techniques available. It is not the purpose here to
present an exhaustive summary of methods. However, the list illustrates the,
types of methods needed to determine the effectiveness of BMP's in meeting
water-quality goals or standards atvarious levels of effectiveness monitoring.
The categories of monitoring methods in the table illustrate the types of questions
that must be addressed to sort out water-quality responses to BMP's from other
influences on water quality. Several items under each category may be needed
to evaluate effectiveness.
Within each category, appropriate monitoring methods are identified for each
monitoring level. The organization of methods by monitoring level is not rigid.
For example, a monitoring plan may be oriented to a Level III assessment, but
have one question addressed by a Level II method.
On-Slope Monitoring Methods
In-stream monitoring must be coupled with on-land monitoring to develop the
linkage between what is observed in the stream and the management activity on
the land. To evaluate NFS pollution from forest management requires evidence
linking the activity to the stream impacts. So, on-land management activities
need to be evaluated for their effectiveness in (1) preventing sediment, nutrients,
pesticides, and organic debris from reaching the stream; (2) protecting stream
banks; (3) shading the stream; (4) controlling surface runoff; and (5) minimizing
soil exposure and compaction. In addition, the Ideation of practices, the quality of
installation, the timing of management activities, the appropriateness of BMP's
selected for the site, the completeness of BMP's implementation on the site, and
the timing of BMP implementation need to be considered.
The effectiveness of one set of management practices versus another can be
estimated by using available models, installing plots and instrumented flumes on
the slope, establishing sediment traps, and other methods. Such on-land com-
parisons reveal the relative movement of potential pollutants on the slope toward
the stream. This information can identify management practices that will mini-
mize the amount of sediment, nutrients, pesticides, and surface runoff from
entering the stream.
On-Slope Methods by Monitoring Level
Comparing reference and study on-slope conditions is needed to link on-slope
conditions with observations made in streams. Also, on-slope comparison
between BMP's or management practices provides information on relative effec-
tiveness of treatments.
For each effectiveness level, appropriate on-land monitoring methods are sug-
gested. The page number beside each method identifies the location of the
summary in the text.
57
-------
.2 •§
a 01 o
Mi
«.
nt
OC DC
'
i— _* J3 »—
ŁŁfŁ =
ills I
~ = O CO •*
-. _ m *^ O
• J .si
:ft
en m •
DC DC
If*
'So o
-
LU .N I
il 1 1
ffl _
cccc
T
^^"
Q.Q..2U-
OCCC O
CW
w
m
o>.
.2 S s
"
S
1C
1
m
.CC.fr
U/ .ii
arm
.S
C - O a.a:
DC
8-
Fia
O CO .E
*^ 5« *•"
^a-^f
ir
2
» SS
•T? r=r *Ł
® 2
3-I^^II
c^ -s t- en ?9 t±
I"0
•s-5 = o>Ł
'8 8
•=-3 S •
COO
tl
-
.2 2 3
U-C3O
•o cm
m
m
m
m
58
-------
BMP Effectiveness I
BMP Implementation I
Erosion Indicators
Universal Soil Loss Equation
Herbicide Indicators
Herbicide Cards
Florida's BMP Implementation Monitoring
BMP Effectiveness fl
BMP Implementation II
Best Management Practices Evaluation
Program
Erosion Indicators
Universal Soil Loss Equation
Fabric Dams
Herbicide Indicators
Herbicide Cards
BMP Effectiveness III
BMP Implementation III
Best Management Practices Evaluation
Program
Fabric Dams
Flumes and Samplers
Herbicide Cards
Montana BMP Implementation Monitoring
BMP Effectiveness IV
BMP Implementation III
Fabric Dams
Flumes and Samplers
Rainfall Simulators
Herbicide Cards
Page
59
70
72
79
78
60
65
68
70
72
76
79
78
65
68
76
78
79
65
65
76
78
78
78
BMP Implementation I
At the BMP Effectiveness I level, one individual can inspect the management
area to determine if the BMP's were properly selected and implemented. A short
inspection form can be constructed reflecting the State BMP guidance or require-
ments. Using such a form will ensure that the inspection considers all appropri-
ate BMP's for the management situation. Inspection observations are recorded
on the form for later reference and for entering into a database. The database
can be analyzed to determine trends in BMP compliance and identify those
BMP's that have poor compliance.
The observer should note any specific BMP's that were not properly implemented
and those that appear to have failed. If fresh sediment is found in the stream, the
observer should attempt to trace it to its source to determine if it came from the
management area.
59
-------
Florida's BMP Implementation Monitoring
A good example of BMP Implementation I monitoring is Vowell (1990). He
reports that in 1981 the Florida Division of Forestry (DOF) developed the Best
Management Practices compliance survey to help direct BMP implementation in
the State. The survey, which is conducted every 2 years, was designed to
(1) measure the compliance of forestry operations with State-approved BMP's,
and (2) identify specific implementation and practice deficiencies. Vowell de-
scribes the Florida BMP Compliance Survey, including (1) the method of site
selection, (2) the field procedures, and (3) follow-up activities:
Site Selection
During a survey year approximately 150 individual forestry operations (survey
sites) are selected by county for compliance monitoring. The number of sites
per county is based on the USDA Forest Service Timber Survey estimated of
harvest levels. Survey sites are identified from aerial observations, using
fixed wing aircraft. Aerial site identification helps maintain a more representa-
tive cross section of ownerships, forest types and physiographic regions by
minimizing bias due to ground accessibility, and by providing a more random
site distribution within each county.
Survey sites are selected on the basis of two criteria: (1) length of time since
the forestry operation took place, and (2) proximity of the operation to a body
of water. In order for a site to qualify for the survey, the forestry operation
must have occurred within 2 years and be located within 300 feet of an
intermittent or perennial stream or a lake 10 acres in size or larger. The
purpose of these criteria is to limit the number of sites to those in which the
greatest potential for impacts exists, and to increase the probability that any
impacts resulting from forestry activities are still visible.
Data Collection
Once a survey site has been selected, a trained professional forester visits
the site and evaluates the operation for compliance with BMP's. The evalua-
tion involves a detailed field inspection and completion of an 85 point ques-
tionnaire. The questionnaire, which focuses on roads, streamside manage-
ment zones and site preparation, was designed with several objectives in
mind. First, to encourage the observer in the field to "get out and look" rather
than conduct a "windshield survey." In responding to the questionnaire, the
observer will consider virtually every aspect of the operation that took place
as well as the local topography, soils, timing, ownership and whether or not
there was technical assistance involved. However, as with any exercise of
this nature there are always some "gray" areas which can result in subjective
judgment. Hence, the questionnaire consists primarily of yes/no and multiple
choice type questions, which minimize the potential for biased re-sponses
and also makes the data more compatible for computer analysis. The final
entry in the questionnaire asks for an overall "pass/fail" rating of the site. This
rating is based solely on the information from the questionnaire and takes into
consideration all aspects of the operation rather than any single and/or
perhaps more visible practice or problem.
60
-------
Division of Forestry
Department of Agriculture & Consumer Services
BMP Compliance Check
Survey Number.
I. General Information
1. County,
Range_
Section
. Township,
2. Approx. No. of acres treated (logged, planted and or site
prepared)
3. Ownership (check one) Government, Industry,
non-ind. ,
Private
4.
5.
If private non-industrial, was technical (professional) forestry assistance
provided by (check one): _DOF Forester, .Consultant,
Industry Forester, ..None at all.
Dominant site type (before treatment): Natural pine, Pine
Plantation, Mixed Pine/Hardwood, Bottomland Hardwood,
.Upland Hardwood,
.Field or Pasture.
6. Type of Treatment (check one or more): Clearcut, _
or shelterwood, Seed tree cut, Site preparation for natural
Selective cut
regeneration. Site preparation for hand planting, Site
preparation for machine planting, _Chemical site preparation..
II. Various Site Characteristics
1. Terrain: Bottomland, Flatwoods, Upland Clay Hills,
Sandhill.
2. Principle soil texture:, Clay, Loam, _
3. Highest slopes approaching stream bottoms:.
4. Soil type (from soil survey, if available)
5. Soil credibility (K factor): High, .Medium,
III. Water Conditions in Area
Sand.
%.
Low.
1. What type of water body (bodies) is on or border the site:
Perennial stream, Intermittent stream, Lake/pond (10
acres or greater in size), Canal, None of the above. Explain:
(Name of water body if known)
2. How close did the following activities get to the water's edge, if
applicable? Site preparation feet, Clearcut feet.
Mechanical tree planting feet.
3. Evidence of damage to stream channel: (may check more than one)
Debris left in stream channel from temporary crossing.
61
-------
Slash, other debris randomly left in stream.
. Erosion and/or failure of stream channel banks.
. Sediment is being deposited in stream.
. None
. Other (specify)
YES
NO
IV. Roads/Skid Trails
1. Roads systems on site. If no, then go to #77 _
2. Does any newly constructed road system (as a result
of the operation) occur within 300 ft. of a watercourse? _
3. Is there sediment from any road system being
deposited into the watercourse? _
4. Are road systems generally located to avoid steep
slopes and gullies? _
5. Are roads locate to avoid streams, and
depressional areas as much as possible? _
6. Are unneeded access roads closed to vehicular use? _
7. Are any stream crossings on site? If no proceed to #12._
8. Are stream crossings adequately stabilized? _
9. Is harvesting equipment (skidders) randomly
crossing streams? _
10. Is there sediment being deposited
into stream from skid trails? _
11 Are planned stream crossing made at right
angles to the watercourse? _
12. Any culvert or bridge washouts. _
13. Are any roadside ditches pulled (connected)
directly into watercourse? _
14. Were oil and trash properly disposed of? _
15. Evidence of BMP's applied to roads (check one or more)
Waterbars installed
Broad-based or rolling dips installed
Mulching, seeding, and/or fertilizing
Water turnouts or wing ditches
Effort made to minimize slopes on road
Roads exist, but no BMP's were applied to them
No roads on site
Other (specify)
16. Evidence of BMP's applied to skid trails:
Seeded
62
-------
.Waterbars
. Slopes minimized
Skid trails exist, but no BMP's were applied
. No skid trails apparent on site
. Other (specify) : '.
17. If new roads were constructed, who was responsible for layout and
construction? (maybe more than one)
Government crew
Landowner
Forest industry crew
Private vendor
Information not available
Not applicable
Other (specify)
18. Were erosion control efforts on road systems generally effective and
adequate? Yes No
Evidence is: (check one or more)
^ Water is diverted off roads at intervals which are successfully
controlling surface erosion.
Water is diverted away from road prior to reaching stream channels.
None installed, erosion occurring
None installed, no evidence of erosion
19. Were landings and decks properly located, adequately cleaned up, and
stabilized? Yes No
V. Site Preparation and Regeneration
1. Method of site preparation (check one or more): Bulldozed,
Disk/harrow, Shear, Chop, Burn, Bed,
Rake, Herbicide, Not applicable, Other (specify)
Root
Yes No N/A
2. Was site planted or to be planted by hand?
3. Was site planted or to be planted by machine?
4. Planting follows natural land contours? If not, go to #8
5. Was slash and debris windrowed/piled?
6. Is there excessive topsoil pushed into windrows/piles?
7. Do windrows follow natural land contours?
8. Were any new lateral drainage (field)
ditches constructed?
63
-------
9. Do established lateral field ditches occur on site?
10. Do firelines exist on the site? If no, go to #14.
11. Does erosion appear to be associated with firelines?
12. Any waterbars, other water diversions
used on firelines?
13. Do firelines avoid streamside management
zones (SMZ)?
14. If a perennial stream or lake (10+ acres) is adjacent
was a selective cut 35' wide Primary SMZ maintained
throughout? (Indicate N/A if no perennial
water is present)
15. If necessary (according to guidelines for perennial
streams or lakes 10+ acres), does a secondary SMZ
appear and does it meet guidelines? (Indicate N/A if
no perennial water is present)
16. If an intermittent stream, was mechanical site
preparation avoided within the required SMZ?
(Indicate N/A is no intermittent water is present or if
site had not site prepared).
17. Type of operator responsible for logging/site preparation work on this site:
Private landowner, using own equipment, Government crew,
Vendor contracted by forest industry, Forest industry crew,
Vendor contracted by private landowner, Other (specify)
18. Additional comments/summary concerning this
site:
19. General grade - Do you feel there was generally good compliance with
208 guidelines on this site? Yes, No. If not, briefly state
primary reason for site failure..
Forester
Landowner
Telephone
Address
Date
City, State, ZIP Code
64
-------
BMP Implementation II
At the BMP Effectiveness II level, one observer visits the management area and
evaluates in detail the BMP implementation and effectiveness. Roads and skid
trails are carefully examined for waterbars, proper drainage, proper bridge and
culvert installation, cut-and-fill slope revegetation success, proper location, etc.
For example, the effectiveness of waterbars on a skid trail would be evaluated by
tracing the path of runoff below the waterbar to determine if the runoff reached
the stream and discharged sediment into it. As another example; runoff and
eroded soil from a site-prepared area could be traced through the filter strip to
determine if runoff and sediment reached the stream. If fresh sediment is found
in the stream, the observer would attempt to trace it to its source and determine if
it was caused by a specific practice or disturbance in the management area.
BMP Implementation III
This method is applicable to both BMP Effectiveness 111 and IV assessments. An
interdisciplinary team conducts a detailed inspection of BMP implementation and
effectiveness as outlined in BMP Implementation II. An interdisciplinary team
may have a forester, a road engineer, a logging engineer, a State water-quality
representative, a soil scientist, a hydrologist, and/or a fisheries biologist. The
team can be interagency, too. Each member evaluates specific BMP's in terms
of his or her expertise, and records his/her compliance and effectiveness obser-
vations on a field form. Team members assemble at the end of the inspection,
discuss their observations, and come to a consensus as to the level of compli-
ance, the quality of implementation, and the effectiveness of BMP's. The con-
sensus results are recorded on a form for entering into a database. The results
of the team inspection should be discussed with the landowner and operators.
The landowner and operator can provide information as to conditions at the time
of the management activity, discuss the management plan and the problems they
encountered, and explain^why they chose to deviate from the plan. These
discussions will help the team members interpret their observations and rate
compliance and effectiveness;
Montana's Forestry BMP Implementation Monitoring
A good example of Level III implementation monitoring is Montana's Forestry
Best Management Practices Implementation Monitoring (Schultz 1992). This
process is used to:
1. determine if BMP's are being applied on timber harvest operations.
2. evaluate the general effectiveness of BMP's in protecting soil and water
resources.
3. provide information on the need to revise, clarify, or strengthen BMP's.
4. provide information to-focus future study efforts by identifying subjects and
geographic areas in need of further investigation.
For the 1992 field audit, Montana used three audit teams, each assigned a
region of the State (northwest, west central, and southwest/east regions of State)
(Schultz 1992). Each team was composed of six members: a fisheries biologist,
a forester, a hydrologist, a representative of a conservation group, a road engi-
neer, and a soil scientist. One member of each audit team was assigned to be
leader and was responsible for filling out the consensus rating form and for
logistics. Team members were Federal, State, and industry employees, and
consultants.
65
-------
Implementation monitoring was limited to those timber harvests that were most
likely to impact water quality. The initial criteria for site selection were:
1. Sites harvested in 1990,1991, or 1992 (last 3 years). Priority was given
to the most recently harvested sites.
2. Part of the harvest area needed to be within 200 feet of a stream.
3. Minimum size of harvest unit to audit was 5 acres. Minimum timber
removal was 7,000 board feet per acre. Some flexibility was allowed if
smaller areas or low-volume harvests met other criteria.
4. In aggregate, sites needed to reflect the distribution of timber harvest in
Montana, geographically and by ownership. Sites were classified into four
land-ownership categories: nonindustrial private, industrial, Federal, and
State. The number of monitoring sites per type of ownership was propor-
tional to the volume of timber harvested by each class in 1990, the latest
complete recording period. At least five sites per ownership class were
audited.
Sites that met the initial criteria were further stratified by the Montana Department
of State Lands according to the following priorities (Schultz 1992):
1. Sites with road construction or reconstruction and slash disposal com-
pleted.
2. Sites with road construction or reconstruction but slash disposal not
completed.
3. Sites with slash disposal completed but no road construction.
Sites were further stratified by erosion-hazard. Two-thirds of the audits were
located on high-erosion-hazard sites; one-third were located on low-to-medium-
erosion-hazard sites. This stratification was subject to availability of sites in the
various erosion-hazard classes. Erosion hazard is a function of slope steepness
and erosivity, based on geologic parent material and soil texture. Care was
taken that a disproportionate number of high-hazard sites did not fall in any one
ownership class.
The audit team's rating form covered 58 potential practices at each site, rating
application and effectiveness of each practice on a 5-point scale (Schultz 1992).
The team rated a BMP by noting first if it was applicable to the site, and if so,
whether it was applied correctly and in the proper locations. Lack of adequate
application or misapplication were regarded as departures from BMP's. The
rating guide for application of BMP's was:
5—Operation exceeds BMP requirements.
4—Operation meets BMP requirements.
3—Minor departure from intent of BMP.
2—Major departure from intent of BMP.
1—Gross neglect of BMP.
Ratings of 5 and 4 are self-explanatory. The 3 rating applies to small departures
distributed over a localized area or over a large area where potential for impact is
low. The 2 rating applies to large departures or BMP's repeatedly neglected.
The 1 rating applies where risks to soil and water resources were obvious and
yet there was no evidence indicating that operators had applied BMP's to protect
these resources (Ehinger and Potts 1990).
66
-------
The effectiveness rating answered questions concerning impacts (Schultz 1992).
For example, "Has the application or misapplication of a particular forest practice
increased the likelihood or actual occurrence of surface sediment in the stream
channel?" The rating guide for effectiveness was:
5—-Improved protection of soil and water resources over pre-project condi-
tions.
4—Adequate protection of soil and water resources.
3—Minor and temporary impacts on soil and water resources.
2—Major and temporary, or minor prolonged, impacts on soil and water
resources.
1—Major and prolonged impacts on soil and water resources.
Before the audit, the work group defined the above terms prior to facilitate con-
sistency:
Adequate—Small amount of material eroded; material does not reach draws,
channels, or flood plain.
Minor—Some material erodes and is delivered to draws but not to stream.
Major—Material erodes and is delivered to stream or annual floodplain.
Temporary—Impacts lasting 1 year or less, no more that one runoff season.
Prolonged—Impacts lasting more than 1 year.
The teams conducted the audits July through September of 1992. Usually a
representative of the landowner briefed the team by giving background informa-
tion on the silvicultural prescription, time of operation, and associated practices.
Team members walked the site, reviewing potential impact areas such as roads,
streams, streamside management zones, skid trails, and firelines. Teams typi-
cally spent about two hours inspecting each site.
After finishing the inspection, the team gathered to discuss the rating while still
on site. Each team completed a single rating form for each site. If the team
could not reach consensus, the rating was determined by vote. The team leader
noted dissenting opinions in the comments section. The results were discussed
with the landowner representative, if available.
The results of the 1992 audits are summarized below (Schultz 1992).
Application of BMP's (All Ownerships Combined)
Meets or exceeds
Minor departures
Major departures
Gross neglect
87%
8%
3%
2%
Effectiveness of BMP (All Ownerships Combined)
Adequate protection
Minor/temporary impacts
Major/temp, Minor/prolong
Major/Prolong
90%
6%
3%
1%
67
-------
Best Management Practices Evaluation Program—BMPEP
Roby, Rector, and Furniss (1992) have developed an interdisciplinary, detailed
BMP implementation/effectiveness monitoring methodology for the California
Region, USDA Forest Service that is applicable to both BMP Effectiveness II and
III levels. Through a iterative process of BMP implementation, monitoring BMP
compliance and effectiveness, and BMP development/modification, BMPEP
strives to Improve BMP's so that water-quality goals will be met.
The objectives of the BMPEP are to:
1. Assess the degree of implementation of BMP's.
2. Determine which BMP's are effective.
3. Determine which BMP's need improvement or development.
4. Fulfill Forest Land and Resource Management Plan BMP monitoring
commitments.
5. Record performance of NPS pollution management in the California
Region.
The BMPEP has three primary components:
1. Administrative Evaluations
2. On-Site Evaluations
3. In-Channel Evaluations
Administrative Evaluations are broad-scale, subjective assessments of multiple
BMP's at the project level. These assessments look at the total project, from the
planning and design phases to implementation, and subjectively evaluate effec-
tiveness of the resulting project in attaining desired water-quality protection. The
assessment evaluates factors not observable or measurable at a project site,
such as the Interdisciplinary Team (ID) process and communication during
project development and execution.
The Interdisciplinary Team process is normally evaluated by members of review
teams during project and activity program reviews. The review assesses the
process used on the Forest to identify and schedule harvest activities, evaluates
the quality of ID teamwork and NEPA documents, and rates how well the process
is being used and how well it is working.
On-site Evaluations provide a means of gathering objective data at the site of
BMP implementation for specific practices. The evaluations are based on actual
measurements of key criteria (ground cover, canopy closure, etc.) and ocular
estimates (presence or absence of rills, debris in culvert inlets, etc.)
There are 28 different On-Site Evaluation Procedures, each dealing with one of
the following:
1. Streamside-management zones (SMZ)
2. Skid trails
3. Suspended yarding
4. Landings
5. Timber administration
6. Special erosion control and revegetation
68
-------
7. Meadow protection
8. Road surface, drainage, and slope protection
9. Stream crossings
10. Control of sidecast material ,
11. Servicing and refueling '
12. In-channel construction practices
13. Temporary roads
14. Rip rap composition
15. Water-source development
16. Snow removal
17. Pioneer road construction
18. Restoration of borrow pits and quarries
19. Protection of roads during wet periods
20. Designated swimming areas
21. Developed recreation sites
22. Location of stock facilities in wilderness
23. Range management
24. Prescribed fire
25. Mining operations
26. Common variety minerals
27. Vegetation manipulation
28. Revegetation of surface-disturbed areas
A detailed assessment of BMP implementation is conducted. Rating involves
reviewing project plans, environmental assessments, and the actual practices on-
the-ground to gauge how well the implemented practices match what was
planned.
The extent of BMP effectiveness is evaluated based upon observation, on the
slope. Based on experience and judgment, the effectiveness is rated subjec-
tively by such comments as:
- Effects are to a perennial stream and are major and prolonged ("major"
meaning that beneficial uses of water could be affected; and "prolonged"
meaning more than a year).
- Effects are to an intermittent or ephemeral stream and are major and pro-
longed.
- Effects are to a perennial stream and are either minor and prolonged or
major and temporary.
- Effects are to an intermittent or ephemeral stream and are either minor
and prolonged or major and temporary.
- Effects are to perennial stream and are minor and temporary.
- Effects are to an intermittent or ephemeral stream and are minor and
temporary.
- Effects extend to the SMZ, but not to the channel of a perennial stream.
- Effects extend to the SMZ, but not to the channel of an intermittent or
ephemeral stream7.
69
-------
70
The data are recorded on a field form and entered into a database for subse-
quent analysis and report writing. The skid-trail form is an example of the forms
developed for this method (Figure 9).
Erosion Indicators
Evidence of sheet erosion, the amount of bare soil, the presence of rills, gullies,
and slumps, rutting, and compaction are on-site indicators of management and
BMP effectiveness. Several factors affect surface-erosion rates in addition to
amount of bare soil: slope, slope length, rainfall energy, and the inherent erodibil-
ity of the soil. When comparing the effect of two different practices on erosion,
the two areas should be similar in slope, soils, rainfall, and slope length. The
amount of bare soil can be significantly different. However, a large amount of
bare soil on a gentle slope may yield as much erosion as a small amount on a
steep slope.
On-site erosion, as evidenced by erosion pedestals, is not usually visible until
erosion rate exceeds 10 tons per acre per year or approximately 1/16th of an
inch of soil. Erosion pedestals capped with small stones, leaves or twigs mea-
suring a 1/16th inch and shorter are difficult to detect. This explains why many
observers see no evidence of erosion when in fact significant sheet erosion is
occurring.
On-site detachment, as evidenced by erosion pedestals, does not automatically
mean that all detached soil leaves the slope; most detached soil is trapped/
stored on the slope before it reaches the toe of the slope.
However, if erosion pedestals are clearly evident, the effect of two treatments on
erosion can be compared, using the rule-of-thumb that an area inch equals
approximately 160 tons per acre. If one management area has 50 percent bare
soil and the average pedestal height is 1/4th inch in the bare areas, on-site soil
detachment is estimated at 20 tons per acre (160 tons/acre inch x .25 inch x 0.5).
If a second area of the same age received about the same rainfall, has 15 per-
cent bare soil, and pedestal height averages 0.2 inches, the on-site detachment
for the second area is approximately 5 tons per acre.
The presence of rills or gullies suggests significant surface runoff and movement
of eroded soil down the slope. Rills are rivulets on the slope generally less than
one square foot in cross section; rill paths generally are parallel. When rills
converge or the cross sectional area exceeds one square foot, they become
gullies. The frequency and size of rills and gullies on the slope can be used to
compare erosion between management areas.
The amount of rill and gully erosion area can be approximated by estimating the
percent of the area occupied by rills and gullies and their average cross sectional
area. For example, if rills occupy 5 percent of the area and have an average
cross sectional area of 16 square inches, the rill erosion rate is estimated at
about 10 tons per acre: ((.05 X 43560 ft2/acre X 85 Ibs/ft3 of soil X 16 in2/144 in2)/
(2,000 Ibs/ton) = 10.3 tons/acre). The same method can be used for estimating
small landslides and slumps. The total estimated erosion on a site would be the
sum of sheet and rill erosion, plus measured gully erosion, plus measured land-
slide erosion. If the time since management began is known, rill and gully ero-
sion rates per year can be estimated.
-------
[MCooxiliuM Za» _
Best Management Practices Evaluation
Form TM: Skid Trail*
(BMP 1.10 41.17)
_v«ttU>«*SOc<
IMFUMINTATION
SMTrilWcompiywltliPSH«nd»r AM no «««*»
QSUdtnltdMuibl^Ui«)10%ofw« DSUdtnflidWiat»10%bal<13%ofnnlt nsUdtnOldlMuib>15%ofuaft
nuttW(»m«vUmotoM.buooii>oii<20%ofiUi
Q Uttk or no cvfckact o( rmttig
Q < 10% of wttetan (cfl to divot flow
ortofiUdlid
dSedlniHdqnJtloaitiMordoiin*
MendtorondouMtonifOl
t)»tin>rilit|>od>lirlon(
a>1«bU<2l%20»o(wt«»n MB divot flow
rknofflUJMll (raniildmU
ONonUXKlofnnioontoSMZ
nriuiriniii) ili|i mlli'l lnllTlilil nnl In ftiinnil 0 Evidence of Mdtewnttnraipart to or
FormfTQZ Venkxi4/92.
9TCCTIWNW \CCttnJtUtd)
Form TOb Skid Tralto
-------
Universal Soil Loss Equation
For sheet and rill erosion, the Universal Soil Loss Equation (USLE) has been
adapted to forest situations in Eastern and Western United States, and is appli-
cable for BMP Effectiveness I and II levels. The USLE estimates soil erosion
from rainfall but not snow melt.
Despite being called "Universal," the equation cannot be applied to new areas
without adaptation. Vegetation characteristics, soils, rainfall, and other condi-
tions vary among regions of the country. The factors and subfactors in the USLE
can be modified to increase the accuracy of erosion estimates for local conditions
(Dissmeyer 1982a and 1982c, and Dissmeyer and Foster 1981). This has been
done for the Eastern United States.
The USLE predicts erosion on watershed slopes from the top of a slope to level
ground where deposition occurs, or where the surface runoff flows into a depres-
sion, road ditch, skid trail, gully, or stream. In the forest environment, deposition
often occurs at the junction of a management area and the filter strip, which
defines the end of the slope for the USLE and where erosion stops.
Erosion estimates made with the USLE do not be reflect the amount of
sediment reaching a stream or the sediment yield from a watershed. In
most forest situations only a small portion of the sheet and rill erosion reaches a
stream. Much of the eroded soil is trapped behind debris, in litter on the slope, or
in intervening vegetation. It is important to trace soil movement down slope to
determine it runoff has discharged sediment into a stream. Only a portion of the
management area will yield sediment. Using the area contributing sediment and
observations on the amount of deposition on the slope, skilled observers can
estimate the percent of erosion reaching a stream. This is very subjective, how-
ever, and requires a great deal of experience and testing estimates against
measured sediment yields.
The Wischmeier and Smith (1978) USLE procedure significantly overestimated
erosion for undisturbed and managed forest situations in the Eastern United
States. This led to a modification by Dissmeyer and Foster (1984). Their meth-
odology was tested on 39 research watersheds treated with a wide range of
common forestry practices, from undisturbed to logging to the most intensive
mechanical site preparation for tree planting. The results confirm that their
procedure gives reasonable estimates of erosion on managed forests. The
Dissmeyer and Foster (1984) modification is as follows:
The Universal Soil Loss Equation is: A = RKLSCP
Where:
A is the computed soil loss per unit area, expressed in units selected
for K and for the period selected for R. In practice, these are usually
so selected that they compute A in tons per acre per year, but other
units can be selected.
R, the rainfall and runoff factor, is the number of rainfall erosion index
units. The R factor is usually read from an average annual rainfall
index map. For specific years or locations, the R factor can be com-
puted from recording rain gage records.
72
-------
K, the soil erodibility factor, is the soil loss per unit of R for a specific
soil as measured on a unit plot, which is defined as a 72.6 foot length
of uniform 9 percent slope continuously in a clean-tilled fallow. Clean-
tilled fallow is an agricultural condition where the soil is kept bare and
disked up and down slope—the most erosive condition for agricultural
land. The C cover management factor has subfactors that adjust K to
reflect unfilled forest management conditions.
L, the slope-length factor, is the ratio of soil loss from the field slope
length to that from a 72.6 foot length under identical conditions.
S, the slope-steepness factor, it the ratio of soil from the field slope
gradient to that from a 9 percent slope under otherwise identical
conditions.
C, the cover-management factor, is the ratio of soil loss from an area
with specific cover and management to that from an identical area in
tilled, continuous fallow.
P, the support practice factor, is the ratio of soil loss with a support
practice like contour disking to that with straight-row farming up and
down slope.
It should be noted that the USLE continues to be revised and improved, recently
with improvements in the R, K, and LS factors (Reinard et al., 1991). A revised
R-factor map better reflects both western and eastern conditions. The K factor
now reflects seasonal differences in erodibility. The LS guidance has been
revised to help the user better select slope lengths appropriate to field conditions.
These revisions have refined erosion estimates with the Dissmeyer and Foster
(1984) procedure.
The C and P are combined for forestry application and called the cover-
management factor (CP) (Dissmeyer and Foster 1984). The forestry modifi-
cations to the USLE were modifications in the subfactors of CP and the
addition of a subfactor. The major subfactors operating in the forest environ-
ment are (1) amount of bare soil, or conversely, ground cover, (2) canopy,
(3) soil reconsolidation, (4) soil organic content, (5) fine roots, (6) residual
binding effect, (7) on-site storage, (8) steps, and (9) contour tillage.
Subfactors 1,2, 3, 5, 6 and 7 have direct counterparts in agricultural prac-
tices. Steps do not occur in most agricultural settings. Contour tillage is the
supporting practices factor P.
Bare Soil Subfactor: Erosion is a function of the amount of exposed
soil. Cover such as litter, slash, logs, and surface rock protects soil
- from the erosive forces of raindrop impact and runoff. Exposed forest
soils are subject to soil detachment by raindrop impact. Also, they
yield surface runoff, which potentially erodes soil and transports de-
tached soil from the slope. The observer estimates the amount of bare
soil and enters that percentage into a relationship to get a subfactor
value for bare soil.
73
-------
If the forest floor is completely covered with litter, entering zero percent
bare soil into relationship yields a zero for the subfactor value. Pro-
tected and undisturbed forest soils have infiltration rates that usually
exceed rainfall intensity. Entering a zero into the equation and multi-
plying it against the other factor and subfactor values will result in a
zero erosion estimate. This is consistent with research results and
observations made on undisturbed forest. Undisturbed and protected
forest soils will infiltrate all but the rarest rainfall intensities, yield no
surface runoff, and experience no soil detachment or transport over the
soil surface.
Canopy Subfactor: Vegetal canopy intercepts rainfall and collects
water on its foliage. Water drops form and fall.to the ground. Drops
falling from the canopy may be larger than the original raindrops, but
they fall from a low canopy; the energy of the drops reaching the soil
surface is less than that of rainfall in open areas. Some of the inter-
cepted rainfall never reaches the ground, but is evaporated during and
after the storm. Some of the intercepted rainfall reaches the ground as
stemflow and may contribute to runoff. Wischmeier (1975) developed
values for the canopy subfactor that depend of foliage density and
average drop height. The average drop height is approximately the
midpoint in many types of canopy.
Soil Reconsolidation Subfactor: Soil reconsolidates and becomes
less erodible over time after land is retired from tillage. At Zanesville,
Ohio, after 7 years erosion on plots retired from tillage and kept bare
reduced to 45 percent of the erosion when they were maintained in
tilled, continuous fallow (Wischmeier 1973). The 0.45 value corre-
sponds to the C factor for undisturbed land with no cover (Wischmeier
and Smith 1978). This soil-type subfactor is necessary because the
soil erodibility factor K is derived from tilled soils in continuous fallow;
that is, continuously void of vegetative cover. A relationship for de-
creasing erodibility over time as soil reconsolidates is presented.
For unfilled forest soils, the soil reconsolidation subfactor is 0.45.
However, if the soil is tilled by disking, or rootraking 2 inches or more
deep, this subfactor begins at 1.0 and decreases with time after tillage.
To evaluate soil reconsolidation, the observer determines whether the
soil has been tilled or not, and if tilled, how long ago.
High Organic Content Subfactor: Under permanent forest, topsoil
accumulates a high organic matter content and in some cases it is
higher than that used in the USLE soil erodibility nomograph, which
only goes as high as 4 percent organic matter. For those forest soils
that have more than 4 percent, an adjustment factor of 0.7 is assigned
this subfactor.
Fine Root Subfactor: A dense mat of fine roots is usually present in
the top 2 inches of forest soils. Even after the trees are removed, the
residual root mat will partially protect soil from erosive forces of rainfall
74
-------
and runoff by holding the soil in place. Sometimes the site is exposed
by removal of the surface organic material, while the topsoil with its
fine root mat is left in place. Where equipment has removed the
topsoil, the fine root mat is usually eliminated. Careful examination is
often required to see fine roots. The subfactor is evaluated by estimat-
ing the percent of the bare soil with fine roots present and entering into
a relationship to get the subfactor value.
Residual Binding Effect Subfactor: The erosion response of a soil
depends on the soil's recent history. That is, there is a residual or
carryover effect when the land use or condition changes. When a soil
that has not been tilled for some time is cultivated, erosion immediately
after it is first tilled may be much less than it will be 2 or 3 years later.
At first the soil has a fairly good structure and fine roots and organic
matter bind soil into more stable aggregates. With time, this effect
decays and the soil becomes more erodible.
The magnitude of the effect, and its duration, is a function of the
amount of roots and organic matter in the soil at the time of tillage, plus
structure and permeability of the subsoil.
Onsite Storage Subfactor: Not all detached soil may be delivered to
the toe of the slope; a portion may be stored locally in depressions,
behind slash and logs, and in litter on the slope. Values range from 0
to 1.0 for forest conditions. A "0" means all detached soil is stored
locally on the slope; none reaching the toe of the slope. A "1.0" means
there is no storage on the slope. The observer evaluates onsite
storage by estimating the proportion of the existing onsite erosion that
will be trapped locally.
Step Subfactor: Surface runoff often washes debris down slope until
it lodges. This debris accumulates and forms miniature dams which
pond water and collect sediment. When these miniature dams are
filled with sediment, they form steps. Steps also form behind roots,
clumps of vegetation, and other obstacles, and when depressions fill
with sediment. On steep slopes, cattle trails on the contour can form
steps.
Steps reduce slope steepness on the area occupied by steps. The
step subfactor is evaluated by estimating the percentage of the slope
occupied by steps, measuring the slope gradient, and entering these
observations into a relationship.
Contour Tillage Subfactor: Disking on the contour generally reduces
sheet and rill erosion by reducing runoff amount and velocity in com-
parison to tillage up and down slope. Disking equipment should be
operated on the contour, but this in not always practical, resulting in
ridges and furrows being oriented at an angle to the contour. As
furrows and ridges increasingly deviate from the contour, their effec-
tiveness decreases. As the grade along the furrows increases, the
amount of material deposited in furrows quickly decreases. The value
75
-------
for this subfactor is a function of degrees off contour by the furrows
and land slope.
The potential subfactors by disturbance category are presented in Table 6.
Not all subfactors will be operating on a site. Therefore, the observer must
determine which subfactors are operating and make the necessary observa-
tions to develop subfactor values. These subfactor values are multiplied
together to product the CP factor for the equation. The product of factor
values for R, K, LS and CP equals the estimated erosion rate.
The use of the USLE should be limited to comparing sheet and rill erosion
between management practices and the effectiveness ofBMP's in minimizing
sheet and rill erosion. The best use of the USLE is to estimate long-term
average erosion rates for the management condition and use it as a planning
tool. The reason for the long term rate is that the R (rainfall) factor is the
average rainfall energy computed for precipitation records exceeding 20
years in length.
Rainfall energy can be computed from recording rain gage records for a
individual year or a season. Based upon the compute R, erosion can be
estimated for a year or season.
Table 6. Potential subfactors by disturbance category
Subfactor
Disturbance Category
Tilled Unfilled
Bare soil
Canopy
Soil reconsolidation
Soil organic content
Fine roots
Residual binding
Onsite storage
Steps
Contour tillage
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Fabric Dams
Dissmeyer (1982b) developed the use of fabric dams to measure erosion on
slopes and demonstrate the effectiveness of BMP's in reducing slope erosion.
These have been used in erosion research. Fabric dams can be used in BMP
Effectiveness II, III and IV monitoring.
Fabric dams are relatively easy and inexpensive to construct (Figure 10). The
materials include a rot-resistant and water-permeable fabric, treated posts,
lumber, hog wire, staples, and nails. Posts are set 4 to 8 feet apart 18 to 24
inches into the ground. The posts are braced to prevent the weight of water and
sediment from collapsing the dam. Posts should be 3 feet above the ground. A
2 x 4 is nailed across the tops of the posts to form a top rail. The dam should be
about 30 feet wide on the contour and have wings at the ends extending straight
up-slope to keep runoff and sediment from flowing around the end of the dam.
On steep slopes the wings extend up slope to a point equal in elevation to the top
rail on the dam. On gentle slopes the wings should extend 10 feet up slope.
76
-------
Benchmark
•v
^ f
Survey Pins ~~ ^ j- r < •<
/- r r fJf
Filter Fabric —,
Hogwire Fence —
Staple
Top Rail
Brace
Figure 10. Fabric dam construction.
A 1- by 1-foot grid of flagged pins is installed immediately up-slope of the dam to
facilitate measuring the volume of trapped sediment. A bench mark is estab-
lished near the dam and the ground elevation at each pin in the grid is measured
with a rod and level. As sediment accumulates, its elevation at each pin in-
creases. The volume of sediment trapped is determined by measuring the eleva-
tion of sediment at each pin, computing the elevation change at each pin, and
multiplying the average depth of deposit by the area of sediment deposition on
the grid. To translate volume of sediment into weight, the bulk density of the
trapped sediment is determined and multiplied by the volume. Sediment can be
measured annually, seasonally, or after a major storm.
The plot boundary above the dam needs to be established by building a berm and
covering it with litter or by installing plastic garden edging. The recommended
size of plot is 70 to 100 feet long by 25 to 35 feet wide on slopes. Smaller dams
can be used to measure erosion from a road dip or a waterbar on a skid trail.
Fabric dams are not recommended for streams because the dams can be over-
topped or stream flow can be diverted around the end, causing stream-bank
erosion.
77
-------
Flumes and Samplers
Flumes and samplers are appropriate for the BMP Effectiveness IV and occasion-
ally at level III. Researchers frequently install flumes and samplers to measure
erosion, runoff, and water quality from watershed slopes, small drainages, or
specific management areas, such as road drainages, road fills, road banks, etc.
For these purposes, the flumes and samplers need to be located on the slope
with defined plots or in ephemeral drains. Watershed or plot boundaries are
mapped or established to define the area above the flume.
The volume of runoff is measured by stage gages in the flumes or by proportional
samplers. The proportional samplers collect a prescribed percent of the water
and sediment. The sampled water can be analyzed for nutrients and other con-
stituents.
Pump samplers are often used to collect bottles of water at fixed time intervals
during runoff events. Pump samplers can be turned on by a rise of water in the
flume so that sampling intensity is increased during rainfall/runoff events. Each
water sample collected is stored in a separate bottle, allowing sediment concen-
trations to be analyzed at various times during the study.
These installations require periodic maintenance and collection of samples for
analysis. They are usually accompanied by recording raingages and weather
stations to facilitate analysis of runoff/precipitation relationships. Servicing all the
equipment is labor intensive and expensive.
Rainfall Simulators
Rainfall simulators have been used for evaluating erosion and refining erosion
prediction methods (e.g., USLE). They are research tools to evaluate different
management practices and on-ground conditions, thus appropriate for BMP
Effectiveness IV assessment. Simulators are quick way to test the effectiveness
of BMP's in reducing erosion and surface runoff.
Rainfall simulators take various forms, but have common features. A rainfall
simulator applies a known amount of "rain" to a plot of fixed size. The apparatus
includes sprinklers at a fixed elevation over the plot, a water source or tank trucks
of water, and pumps to transport water from the source to the sprinklers. The
sprinkler nozzles can produce droplets of desired sizes to simulate raindrops.
The plots will vary in width and length depending on the apparatus used and the
conditions being evaluated. The plot boundary is defined by metal or plastic
edging. At the base of the plot is a small flume and samplers to measure runoff
and sediment.
Herbicide Cards
Cards with surfaces that react to herbicides and are used to monitor drift from
aerial applications. Herbicide cards placed at various distances from a stream in
buffer strips before spraying reveal if and how far into the buffer drift occurred and
if drift deposited herbicides directly into the stream. Herbicide cards can be used
at all levels of monitoring. They are used to refine BMP's and to monitor their
effectiveness in preventing herbicide drift from reaching the stream.
78
-------
Herbicide Indicators
For BMP Effectiveness I, empirical observations can determine if the herbicide
drift reached a stream. Dead vegetation in the buffer strip reveals how far lethal
concentrations drifted. Drift of sublethal concentrations are revealed by close
inspection of green leaves, looking for spots and other symptoms of herbicide
damage. These observations are made after the herbicide has had time to
damage or kill the vegetation.
f
Chemical and Physical Monitoring Methods
Decades of water-quality research on the impacts of forest management prac-
tices have revealed that there are a limited number of variables affected by forest
management. MacDonald et al. (1991) present an excellent summary of forest
management interactions with 29 of the most commonly affected water-quality
variables (see Appendix). The discussions in this section will be limited to 14
chemical and physical variables; most of the remaining 15 are discussed in the
biological, habitat, channel geomorphology, and aquatic vegetation sections.
Generally, only a few chemical and physical variables need to be considered in
evaluating the effectiveness of forest management practices in meeting water-
quality goals or standards (Table 7). Too often, forest monitoring efforts have
analyzed water samples for metals and other nonforest-management-related
variables. Monitoring such variables should be restricted to other land uses, for
example, oil wells, mines, hazardous waste sites, and road construction exposing
geologic materials that contribute to pollution or atmospheric deposition. If
woodland grazing or developed recreation area occur in the study watershed,
fecal coliform would be added to the list of variables measured. A detailed
discussion of the significance, measurement, and analytical methods for each of
these parameters is presented in MacDonald et al. (1991) in the Appendix.
Table 7. The 14 forest resource management water-quality-related parameters covered in this discussion
Water column
Flow
Sediment
Temperature
PH
Conductivity
Dissolved oxygen
Intergravel DO
Nitrogen
Phosphorus
Herbicides and pesticides
Peak flows
Low flows
Water yield
Suspended
Turbidity
Bedload
79
-------
Chemical and Physical Methods by Monitoring Level
The monitoring approach for each level is discussed in general terms. Not all
variables need to be monitored for each level of monitoring. The variables
monitored depend on the situation and the issues being addressed.
BMP Effectiveness I
Some empirical evidence can be used to compare the reference stream with the
study stream, including: (1) water clarity, color, and odor, (2) presence of scum,
(3) turbidity, and (4) brightness of substrate material. The reference and study
stream must be observed within a short period of time to make the comparisons
valid. The reference and study stream need to be experiencing the same
weather at the time of comparison.
Field kits and meters can be used at this level of monitoring. Temperature can be
measured by a thermometer. Conductivity, dissolved oxygen, turbidity, and pH
are measured by meters in situ. Nitrogen can be analyzed using a field kit. A
grab sample can be sent to a laboratory for analysis for suspended sediment and
fecal coliform. Given the limited precision of the field kits, this level of monitoring
will detect only major differences between reference and study streams.
At this level of monitoring, stream flow comparisons between reference and study
streams will be very difficult and unreliable. However, at points with equal water-
shed area above, high-water marks could be compared. If the reference stream
stayed within its banks and the study stream did not, the cumulative effect of
activity in the study watershed may have increased peak flows. Judgment on
differences in flow can be supported by differences in channel morphology and/or
stability.
If management in the study watershed is suspected to have altered low flow, the
following procedure can be used. First, locate points on the reference and study
streams that are similar in classification and watershed area. Using a current
meter in a measured stream profile, measure stream flow during the low-flow
period. The flow measurements should be made same day, one immediately
after the other. If the flow cannot be measured at points with essentially the
same watershed area, converting flow measures to cubic feet per second per
square mile will allow comparison.
BMP Effectiveness II
The same basic procedures in Level 1 are applied in BMP Effectiveness II.
However, the number of samples and sample points would increase. Samples
are collected several times over a period of a few weeks to a couple of months,
depending upon the issue and management activity being monitored.
BMP Effectiveness III
Water quality is sampled spatially and temporally in reference and study streams,
following a detailed, statistically designed monitoring plan. Depth-integrated
sediment sampling, automated pump samplers, recording temperature gages,
and similar equipment may be used at this level. Samples are preserved and
sent to an analytical lab using USEPA quality assurance/quality control protocols
80
-------
for analysis. The resulting data are entered and stored in a computer database
for management and analysis. Established statistical methods for water-quality
and stream-flow analysis are used; This is an appropriate level to monitor herbi-
cides and pesticides because of the importance of timing of sampling with rain-
storms and the need for analytical labs to process the samples.
Monitoring ground water quality helps to answer some questions and is appropri-
ate at this level. A couple of examples where ground-water monitoring may be
needed are: (1) tracing herbicide movement through the soil to ground water or
to a stream, and (2) evaluating the quality of ground water under forested land
that was formerly in agriculture to determine if agricultural chemicals are still
affecting stream quality.
Raingages need to be located in each of the reference and study watersheds so
that precipitation can be measured. Specialized equipment can be installed
where atmospheric deposition is an issue. These can be storage gages and/or
recording gages, depending upon the issues addressed.
Recording stream gaging stations are often needed to translate concentration
data into total yields for a storm, a season, or a year. Stream gages measure
stage heights, which can be translated into discharge using rating curves. Long-
term flow records can be used to compare peak flows, low flows, and annual
water yield between the reference and study streams. It may be necessary to
establish recording stream-gaging stations if USGS gaging stations are not
available on the reference and study streams.
BMP Effectiveness IV
In addition to the procedures followed in Effectiveness III, five others may be
needed:
(1) Automated sampling during the rising and falling stages of runoff for some
variables, for example, suspended sediment. Flumes with proportional
samplers might be used to sample sediment yields.
(2) Monitoring herbicides and pesticides intensively for an extended period
after application to determine the peak concentration, when it occurred,
and under what climatic conditions.
(3) Monitoring ground-water quality to trace chemical/nutrient movement
through the soil to ground water or to a stream, to measure the peak
concentration and determine when it occurred after application, and to
determine how long the chemical/nutrient content in soil moisture is
elevated.
(4) Rain gage networks, coupled with complete weather stations, should be
located in each of the reference and study watersheds. If atmospheric
deposition is an issue, it should be sampled and assessed.
81
-------
(5) Weirs may need to be constructed across streams to increase the accu-
racy of flow, sediment, nutrient, etc., measurements so watershed bud-
gets can be constructed for each constituent. Long-term, accurate flow
records can be used to detect differences in peak flows, low flows, and
annual water yield between the reference and study streams.
Channel Geomorphology Monitoring Methods
The geomorphology of stream channels responds to influences of management.
Some of these influences are addressed in the discussion of habitat. Here the
focus is on channel geomorphologic characteristics: channel cross-sections,
channel width/depth ratios, thalweg profile (longitudinal profile of the stream bed),
bed material, sinuosity, and bank stability. Another issue addressed is classify-
ing reference and study streams to ensure they are as similar as possible.
Channel Geomorphology Methods by Monitoring Level
The channel geomorphology monitoring methods abstracted below apply to all
levels of monitoring. The differences among monitoring levels are merely the
degree of implementation. At the BMP Effectiveness I level, empirical compari-
sons of stability and channel classification are made using knowledge of refer-
ence conditions. At the BMP Effectiveness II level, the Pfankuch stability rating
is used, along with a detailed Rosgen's channel classification of reference and
study streams. At the BMP Effectiveness III and IV levels, the Pfankuch method
is appropriate for stability rating analysis between similar reference and study
streams. A detailed Rosgen channel classification and mapping, the develop-
ment and use of bankfull width, average bankfull depth and cross-sectional area
to watershed area relationships, and particle size distributions for reference
stream comparisons to study streams are appropriate. The V* Pool Sediment or
Rosgen Channel Monitoring are appropriate at levels III and IV.
The various methods are listed below and discussed in detail as indicated.
Page
Empirical/Perceptual 82
Pfankuch Channel Stability 83
Rosgen's Stream Channel Classification, plus
channel geomorphologic relationships 95
Pebble Counts 112
V* Pool Sediment 107
Rosgen Channel Monitoring 108
Alaska Channel Classification 105
Empirical/Perceptual
This method is used only for BMP Effectiveness I. The trained observer must be
knowledgeable in stream-channel classification and in characteristics of stable
and unstable channels. The observer must be also trained in Rosgen's (1985)
stream-channel classification system and/or Pfankuch (1978) channel-stability-
evaluation methods. He/she should have a mental picture of the channel classes
and stability of the reference streams for comparison with the study streams.
82
-------
Reference and study streams should be as similar as possible, including channel
classification and stability. As a study stream passes a management area, there
may be a significant change in channel class or stability that is clearly inconsis-
tent with the reference stream. For example, the upstream reach of the study
stream may have a stable, single channel. As the stream passes the manage-
ment area, it may become a multi-channel, braided stream. The observer notes
that the bank stabilizing vegetation has been removed by management, resulting
in an unstable, braided channel that is atypical of the reference stream.
Streambed material is used in classifying stream channels. The observer may
notice a major shift in streambed material (for example, cobble to sand) as the
stream passes a road or a management area. If the change is inconsistent with
the reference stream conditions and sediment loading can be traced to the
management condition, then channel geomorphological changes can be associ-
ated with management.
The observer visually compares bank stability, width/depth ratios, streambed
stability, aggradation, and degradation between reference and study streams.
Pfankuch Channel Stability
Pfankuch (1978) developed a qualitative method for evaluating the stability of
stream channels that can be used at all levels of monitoring. The method evalu-
ates 15 factors and rates their quality based upon narrative descriptions. De-
pending upon which description best applies to the condition found, a score for
that factor is given. The scores for all factors are totalled and the total score is
used to rate stability as excellent, good, fair, or poor.
The Pfankuch (1978) method uses a field form containing the following factors,
narrative descriptions, and rating scores:
/. Upper Channel Banks
The land area immediately adjacent to the stream channel is normally and
typically a terrestrial environment. Landforms vary from wide, flat, alluvial
flood plains to the narrow, steep termini of mountain slopes. Intermittently
this dry land flood plain becomes a part of the water course. Forces of
velocity and turbulence tear at the vegetation and land. These hydraulic
forces, while relatively short lived, have great potential for producing on-site
enlargements of the stream channel and downstream sedimentation damage.
Resistance of the component elements on and in the bank are highly vari- •
able. This section is designed to aid in rating this relative resistance to
detachment and transport by floods.
1. Land form slope: The steepness of the land adjacent to the stream
channel determines the lateral extent and ease to which banks can be
eroded and the potential volume of slough which can enter the water.
All other factors being equal, the steeper the land adjacent to the
stream, the greater the potential volume of slough materials.
A. Excellent: Side slopes to the channel are generally less than 30
percent on both banks. (2 points)
83
-------
B. Good: Side slopes up to 40% on one or occasionally both banks.
(4 points)
C. Fair: Side slopes 40 to 60% common on one or both banks.
(6 points)
D. Poor: Steep slopes over 60%. (8 points)
2. Mass Wasting Hazard: This rating involves existing or potential
detachment from the soil mantle and downslope movement into water-
ways of relatively large masses of soil mantle. Mass movement of
banks by slumping or sliding introduces large volumes of soil and
debris into the channel suddenly, causing constrictions or complete
damming followed by increased stream flow velocities, cutting power
and sedimentation rates. Conditions deteriorate in this factor with
proximity, frequency and size of the mass wasting areas and with
progressively poorer internal drainage and steeper terrain.
A. Excellent: There is no evidence of mass wasting that has or could
reach the stream channel. (3 points)
B. Good:, There is evidence of infrequent and/or very small slumps.
Those that exist may occasionally be "raw," but predominately the
areas are revegetated and relatively stable. (6 points)
C. Fair: Frequency and/or magnitude of the mass wasting situation
increases to the point where normal high water aggravates the
problem of channel changes and subsequent undercutting of
unstable areas with increased sedimentation. (9 points)
D. Poor: Mass wasting is not difficult to detect because of the fre-
quency and/or size of existing problem areas or the proximity of
banks are so close to potential slides that any increases in the flow
would cut the toe and may trigger slides of significant size to cause
downstream water quality problems for many years. (12 points)
3. Debris Jam Potential: Floatable objects are deposited on stream
banks by man and as a natural process. By far, the bulk of this debris
in natural in origin. Tree trunks, limbs, twigs, and leaves reaching the
bulk of the obstructions, flow deflectors, and sediment traps are rated.
This factor assesses the potential for increasing these impediments to
the natural direction and force of flow where they now lay. Also, it
includes the possibility of creating new debris jams under certain flow
conditions.
Caution: Woody debris provides vital habitat for fish and benthic
macroinvertebrates and is part of productive natural streams.
Because woody debris is associated with some stability problems
does not mean it has to be removed.
84
-------
A. Excellent: Debris may be present on the banks, but is so situated
or is.pfsuch size, that the stream is not able to push or float it into
the channel and, therefore, for all intents and purposes, it is absent.
In truth, there may be none physically present. Both situations are
rated the same. (2 points)
B. Good: The debris present offers some bank protection fora while
but is small enough to be floated away in time. Only small jams
could be formed with this material alone. (4 points)
C. Fair. There is a noticeable accumulation of all sizes and the
stream is large enough to float it away, at certain times, thus
decreasing the bank protection and adding to the debris jam poten-
tial downstream. (6 points)
D. Poor: Moderate to heavy accumulations are present due to fires,
insect attack, disease mortality, windthrow, or logging slash. High
flows will float some debris away and the remainder will cause
channel changes. (8 points)
4. Vegetative Bank Protection: The soil in banks is held in place largely
by plant roots. Riparian plants have almost unlimited water for both
crown and root development. Their root mats generally increase in
density with proximity to the open channel. Trees and shrubs gener-
ally have deeper root systems than grasses and forbs. Roots seldom
extend far into the water table, however, and near the shore of lakes
and streams they may be compartively shallow rooted. Some species
are, therefore, subject to windthrow.
In addition to the benefits of the root mat in stabilizing the banks, the
stems help to reduce the velocity of flood flows. Turbulence is gener-
ated by stems in what may have been laminar flow. The seriousness
of this energy release depends on the density of both overstory and
understory vegetation. The greater the density of both, the more
resistance displayed. Damage from turbulence is greatest at the bank
edge and diminishes with distance from the normal channel. Other
factors to consider, in addition to the density of stems, are the varieties
of vegetation, the vigor of growth, and the reproduction processes.
Vegetal variety is more desirable than a monotypic plant community.
Young plants, growing and reproducing vigorously, are better than old,
decadent stands.
A. Excellent: Trees, shrubs, grass and forbs combined cover more
than 90 percent of the ground. Openings in this nearly complete
cover are small and evenly dispersed. A variety of species and
age classes are represented. Growth is vigorous and reproduction
of species in both the under- and over-story is proceeding at a rate
to insure continued ground cover conditions. A deep, dense root
mat is inferred. (3 points)
85
-------
B. Good: Plants cover 70 to 90 percent of the ground. Shrub species
are more prevalent than trees. Openings in the tree canopy are
larger than the space resulting from the loss of a single mature
individual. While the growth vigor is generally good for all species,
advanced reproduction may be sparse or lacking entirely. A deep
root mat is not continuous, and more serious erosive incursions are
possible in the openings. (6 points)
C. Fair: Plant cover ranges from 50 to 70 percent. Lack of vigor is
evident in some individuals and/or species. Seedling reproduction
is nil. This condition is ranked fair, based primarily on the percent
of the area not covered by vegetation with a deep root mat poten-
tial and less on the kind of plants that make up the overstory.
(9 points)
D. Poor: Less than 50 percent of the ground is covered. Trees are
essentially absent. Shrubs largely exist in scattered clumps.
Growth and reproduction vigor is generally poor. Root mats dis-
continuous and shallow. (12 points)
II. Lower Channel Banks
The channel zone is located between the normal high water and low water
lines. Both aquatic and terrestrial plants may grow here but normally their
density is sparse.
The lower channel banks define the present stream width. Stability of these
channel banks is indicated under a given flow regimen by minor and almost
imperceptible changes in channel width from year to year. In other words,
encroachment of the water environment into the land environment is nil.
Under conditions of increasing channel flow, the banks may weaken and both
cutting (bank encroachment) and deposition (bank extension) begin, usually
at bends and points of constriction. Cutting is evidenced by steepening of the
lower banks. Eventually the banks are undercut, followed by cracking and
slumping. Deposition behind rocks or bank protrusions increase in length
and depth.
As the channel is widened, it may also be deepened to accommodate the
increased volume of flow. For convenience only, changes of channel bot-
toms are observed separately and last in this evaluation scheme.
5. Channel Capacity: Channel width, depth, gradient, and roughness
determine the volume of water which can be transmitted. Over time,
channel capacity has adjusted to the size of watershed above the
reach rated, to climate, and to changes of vegetation. Some indicators
of change are widening and/or deepening of the channel, which affects
the ratio of width to depth. When the capacity is exceeded, deposits of
soil are found on the backs and organic debris may be found hung up
in the bank vegetation. These are expressions of the most recent flood
event. Indicators of conditions as recent as a year or two ago may be
difficult or impossible to find, but do your best to estimate what normal
86
-------
peak flows are and whether the present cross section is adequate to
handle the flow without bank deterioration.
A. Excellent: Cross sectional area is ample for present peak volumes
plus some additional, if needed. Over-bank floods are rare. Width
to depth ratio less than 7. (1 point)
B. Good: Adequate cross sectional arcds contain most peak flows.
Width to depth ratio 8 to 15. (2 points)
C. Fair: Channel barely contains the peak runoff in average years or
less. Width to depth ratios range from 15 to 25. (3 points)
D. Poor: Channel capacity generally inadequate. Overbank floods
quite common as indicated by kind and conditions of bank plants
and the position and accumulation of debris. Width to depth ratio
25 or more. (4 points)
6. Bank Rock Content: Examination of the materials that make up the
channel bank will reveal the relative resistance of this component to
detachment by flow forces. Since the banks are perennially and
intermittently both aquatic and terrestrial environments, these sites are
harsh for most plants that make up both types. Vegetation is therefore
generally lacking, and it is the volume, size, and shape of the rock
component that primarily determine the resistance to flow forces.
A soil pit need not be dug. Surface rock and exposed cut banks will
enable categorization of this item as percentage ranges.
A. Excellent: Rock makes up 65% or more of the volume of the
banks. Within this rock matrix large angular boulders 12+ inches
(on their largest axis) are numerous. (2 points)
B. Good: Banks 40-65% rock that are mostly small boulders and
cobble ranging in size from 6-12" mean diameter. Some may be
rounded while others are angular. (4 points)
C. Fair: 20-40 of bank volume rock. While some big rock may be
present, most fall into the 3-6" diameter class. (6 points)
D. Poor: Less than 20% rock fragments, mostly of gravel sizes 1-3" in
diameter. (8 points)
7. Obstructions and Flow Deflectors: Objects within the stream chan-
nel, like large rocks, embedded logs, bridge pilings, etc., change the
direction of flow and sometimes the velocity as well. Obstructions may
produce adverse stability effects when they increase the velocity and
deflect the flow into unstable and unprotected banks and across
unstable bottom materials. They also may produce favorable impacts
when velocity is decreased by turbulence and pools are formed.
87
-------
Sediment Traps: Channel obstructions that dam the flow partly or
wholly form pools or slack water areas. The pools lower the channel
gradient. With this loss of energy, the sediment transport power is
greatly reduced. Coarse particles drop out first at the head of the pool.
Some or all the fine suspended particles may carry on through.
Embedded logs and large boulders can produce very stable natural
dams that do not add to channel instability. Some debris dams and
beaver dams, however, are quite unstable and only serve to increase
the severity of channel damage when they break up.
The effectiveness of these sediment traps depends on pool length
relative to entrance velocity. The swifter the current, the longer the
pool needed to reach zero velocity. Turbulence caused by falls at the
head of the pool shortens the length required to reach zero velocity.
How long these traps are effective depends on depth and width as well
as pool length and, of course, the rate of sediment accretion.
Types of vegetation growing in the water, like alders, willows, cattails,
reeds, and sedges, are also effective traps in some locations and
reduce flow velocity and sediment carrying power.
A. Excellent: Logs, rocks, and other obstructions to flow are firmly
embedded and produce a pattern of flow that does not erode the
banks and bottom or cause sediment buildups. Pool riffle relation-
ship stable. (2 points)
B. Good: Obstructions to flow and sediment traps are present, caus-
ing cross currents that create some minor bank and bottom ero-
sion. Some of the obstructions are newer, are not firmly embed-
ded, and move to new locations during high flows. Some sediment
is trapped in pools, decreasing their capacity. (4 points)
C. Fair: Moderately frequent and quite often unstable obstructions
cause noticeable seasonal erosion of the channel. Considerable
sediment accumulates behind obstructions. (6 points)
D. Poor: Obstructions and traps are so frequent they overlap, are
often unstable to movement, and cause a continual shift of sedi-
ments at all seasons. Since traps are filled as soon as formed, the
channel migrates and widens. (8 points)
Cutting and Deposition: are processes that occur at the same time, but at
different locations along a stream. You can not have one without the other.
However, it is possible for each to be taking place in different reaches of the
same stream at the same time, and hence the separation for classification
purposes.
88
-------
8. Cutting: One of the first signs of channel degradation is a loss of
aquatic vegetation by scouring or uprootipg. Some channels are
naturally devoid of aquatic plants, and here the first stages would be
an increase in the steepness of the channel banks. Beginning near the
top, and later extending in serious cases to the total depth, the lower
channel bank becomes a near vertical wall.
If plant roots bind the surface horizon of the adjacent upper bank into a
cohesive mass, undercutting will follow. This process continues until
the weight of overhang causes the sod to crack and subsequently
slump into the channel. Differential horizontal compaction and texture
could also result in undercut banks even with an absence of vegetative
cover. There are some loosely consolidated banks that with or without
vegetation are literally nibbled away, never developing much, if any,
overhang.
A. Excellent: Very little or no cutting is evident. Raw, eroding banks
are infrequent, short, and predominately less than 6" high.
(4 points)
B. Good: Some intermittent cutting along channel outcurves and at
prominent constrictions. Eroded areas are equivalent in length to
one channel width or less, and the vertical cuts are predominately
less than 12". (8 points)
C. Fair: Significant bank cutting occurs frequently in the reach. Raw
vertical banks 12" to 24" high are prevalent as are root mat over-
hangs and sloughing. (12 points)
D. Poor: Nearly continuous bank cutting. Some reaches have vertical
cut faces over 2 feet high. Undercutting, sod-root overhangs, and
vertical side failures may also be frequent in the rated reach .
(16 points)
9. Deposition: Lower bank channel areas are generally the steeper
portions of the wetted perimeter and may be rather narrow strips of
land that offer slight opportunity for deposition. Exceptions to this
abound, because deposition is often noted on the lee side of large
rocks and log deflectors that form natural jetties. However, these
deposits tend to be short and narrow. On the less steep, lower banks,
deposition during recession from peak flows can be quite large. The
appearance of sand and gravel bars where they did not previously
exist may be one of the first signs of upstream erosion. These bars
tend to grow, primarily in depth and length, with continued watershed
disturbancefs). Width changes are in a shoreward direction as over-
flow deposition takes place on the upper banks. Dimensional deposi-
tion "growth" is limited by the size and orientation of the obstructions to
flow along the channel banks, flow velocity, and a continuing upstream
sediment supply.
89
-------
Deposition may also occur on the inside radii of bends, particularly if
active cutting is taking place on the opposite shore. Also, deposits are
found below constrictions or where there is a sudden flattening of
stream gradient as occurs upstream of geologic nick points.
A. Excellent: Very little or no deposition of fresh silt, sand, or gravel in
channel bars in straight reaches or point bars on the inside banks
of curved reaches. (4 points)
B. Good: Some fresh deposits on bars and behind obstructions.
Sizes tend to be predominately from the larger size classes—
coarse gravels. (8 points)
C. Fair: Deposits of fresh, coarse sands and gravels observed with
moderate frequency. Bars are enlarging and pools are filling so
riffle areas predominate. (12 points)
D. Poor: Extensive deposits of predominately fresh; fine sands, some
silts, and small gravels. Accelerated bar development common.
Storage areas are now full and sediments are moving even during
low flow periods. (16 points)
III. Channel Bottom
Water flows over the channel bottom nearly all of the time in perennial
streams. It is, therefore, almost totally an aquatic environment, composed of
inorganic rock constituents found in an infinite variety of kinds, shapes, and
sizes. It is also a complex biological community of plant and animal life.
Both components, by their appearance alone and in combination, offer clues
to the stability of the stream bottom. They are arbitrarily separated and
individually rated for convenience and emphasis during the evaluation pro-
cess. Because of the high reliance on the visual sense, inventory work is
best accomplished during the low flow season and when the water is free of
suspended or dissolved substances.
10. Rock Angularity: Rocks from stratified, metamorphic formations
break out and work their way into channels as angular fragments that
resist tumbling. Their sharp corners and edges wear and are rounded
in time, but they resist the tumbling motion. These angular rocks pack
together well and may orient themselves like shingles (imbricated). In
this configuration they are resistant to detachment.
In contrast, igneous rocks often produce fragments that round up
quickly, pack poorly, and are easily detached and moved downstream.
Excellent to poor ratings relate to the amount of rounding exhibited
and, secondarily, the smoothness or polish the surfaces have
achieved. Some rocks never do smooth up in the natural environment,
90
-------
but most round up in time. Both conditions, of course, are relative
within the inherent capability of the respective rock types.
A. Excellent: Sharp edges and corners, plane surfaces roughened.
(1 point)
B. Good: Rounded corners and edges, surfaces smooth and flat.
(2 points)
C. Fair: Corners and edges well rounded in two dimensions.
(3 points)
D. Poor: Well rounded in all dimensions, surfaces smooth. (4 points)
11. Brightness: Rocks in motion "gather no moss", algae, or stain.
They become polished by frequent tumbling and, as a general rule,
appear brighter in their chroma values than similar rocks that have
remained stationary. The degree of staining and vegetative growths
relate also to water temperature, seasons, nutrient levels, etc. In some
areas a "bright" rock will be "dulled" in a matter of weeks or months. In
another it may take years to achieve the same results. Nevertheless,
even slight changes during the spring runoff should be detectable
during the next summer's survey. Look for changes in the sands and
gravels.
A. Excellent: Less than 5% of the total bottom should be bright, newly
polished and exposed surfaces. Most will be covered by growths
or a film of organic stain. Stains may also be from minerals dis-
solved in the water. (1 point)
B. Good: 5 to 35% of the bottom appears brighter, some of which
may be on the larger rock sizes. (2 points)
C. Fair: About a 50-50 mixture of bright and dull with a 15% leeway in
either direction (i.e., a range of 35 to 65% bright materials).
(3 points)
D. Poor: Bright, freshly exposed rock surfaces predominate with more
than 65% bright materials. (4 points)
12. Consolidation (Particle Packing): Under stable conditions, the array
of rock and soil particle sizes pack together. Voids are filled. Larger
components tend to overlap like shingles (imbricate). So arranged, the
bottom is quite resistant to even exceptional flow forces. Some rock
types (granitics) are less amenable to this packing process and never
reach the stable state of other like Belt Series rocks.
91
-------
A. Excellent: An array of sizes are tightly packed and wedged with
much overlapping which makes it difficult to dislodge by kicking.
(2 points)
B. Good: Moderately tight packing of particles with fast water parts of
the cross section protected by overlapping rocks. These might be
dislodged by higher than average flow conditions. (4 points)
C. Fair: Moderately loose without any pattern of overlapping. Most
elements might be moved by average high flow conditions.
(6 points)
D. Poor: Rocks in loose array, moved easily by less than high flow
conditions and move underfoot while walking across the bottom.
The shape of these rocks tends to be predominantly round and
sorted so that most are of similar size. (8 points)
13. Bottom Size Distribution and Percent Stable Materials: Rocks
remaining on a stream's bottom reflect the geologic sources within the
basin and the flow forces of the past. Normally, there is an array of
sizes expected to be seen in a given local. After a little experience,
observer will be sensitive to abnormal situations. For example, in the
mature topography typical of the mountainous West, the flow in small,
steep streams is sufficient to wash fine sediments and gravels away.
What remains is a gravelly, cobble stream bottom. In the lower
reaches where the gradient is less and flow is often slower, deposition
of the "fines" eroded above begin to drop out. The separates of sand,
silt, and some clay begin to cover the coarser elements. Except where
trapped in still water areas, these fines tend to be in constant motion to
ever lower elevations.
Two elements of bottom stability are rated in this item: (1) Changes or
shifts from the natural variation of component size classes and (2) the
percentage of all components that are judged to be stable materials.
Bedrock, large boulders, and cobblestones ranging in size from one to
three feet or more in diameter are considered "stable" elements in the
average situation. Obviously, smaller rocks in smaller channels might
also be classed as stable. The sizes are given only to guide thought.
Bedrock as a major component of bottom and banks, no matter what
size the channel or how the other elements rate, always results in an
excellent classification of that reach.
A. Excellent: There is no noticeable change in size distribution. The
rock mixture appears to be normal for the kind of geologic sources
in the basin and the flow forces of streams of this size and location
in the watershed.
92
-------
If a shift or change has taken place so there are greater percent-
ages of large rock in the small streams and smaller sizes in large
streams, the condition class most appropriate should be checked.
It is a matter of degree as follows:
(Stable materials 80-100%) (4 points)
B. Good: Slight shift in either direction. Stable materials 5-80%.
(8 points) .
C. Fair: Moderate shift in size classes. Stable materials 20-50%.
(12 points)
D. Poor: Marked, a pronounced shift. Stable materials less than
20%. (16 points)
14. Scouring and/or Deposition: Items of size, angularity, and bright-
ness rated above leads to some conclusions as to the amount of
scouring and/or deposition that is taking place along the channel
bottom.
A. Excellent: Neither scouring nor deposition is much in evidence.
Up to 5% of either or a combination of both may be present along
the length of the reach. (6 points)
B. Good: Affected length ranges from 5 to 30%. Cuts are found
mostly at channel constrictions or where the gradient steepens.
Deposition is in pools and backwater areas. Sediment in pools
tends to move on through so pools change only slightly in depth but
greatly in composition of their size classes. (12 points)
C. Fair: Moderate changes are occurring. 30 to 50% of the bottom is
in a state of flux. Cutting is taking place below obstructions, at
constrictions and on steep grades. Deposits in pools now tend to
fill the pool and decrease their size. (18 points)
D. Poor: Both cutting and deposition are common; 50% plus of the
bottom is moving not only during high flow periods but at most
seasons of the year. (24 points)
15. Aquatic Vegetation: When some measure of stabilization of the soil-
rock components is achieved, the channel bottom becomes fit habitat
for plant and animal life. This process begins in the slack water areas
and eventually may include the swift water portions of the stream cross
section. With a change in volume of flow and/or sedimentation rates,
there may also be a temporary loss of the living elements in the
aquatic environment. This last item attempts to assess the one macro-
aquatic biomass indicator found to best express a change in channel
stability.
93
-------
Clinging Moss and Algae: These lower plant forms do not have roots
but cling to the substrate. They are low growing and may first appear
as a green to yellow-green slick spot on the bottom rocks. Moss plants
continue with slight variation in color but no great change in mass from
season to season. Algae by contrast have a peak of growth activity
and then die off in great numbers. The slippery conditions they pro-
duce persists after death, however.
Both algae and moss inhabit the swift water areas as well as the quiet
pools and backwater portions of the stream bottom.
A. Excellent: Clinging plants are abundant throughout the reach from
bank to bank. A continuous mat of vegetation is not required but
moss and/or algae are readily seen in all directions across the
stream. (1 point)
B. Good: Plants are quite common in the slower portions of the reach
but thin out or are absent in the swift flowing portions of the stream.
(2 points)
C. Fair: Plants are found but their occurrence is spotty. They are
almost totally absent from rocks in swifter portions of the reach and
may also be absent in some of the slow and still water areas.
(3 points)
D. Poor: Clinging plants are rarely found anywhere in the reach. (This
is an unusual situation, but it could happen under a combination of
adverse environmental conditions). (4 points)
The rating scores for all elements are added to get a total, stability rating
score. Based upon the total score, the stability of the reach is rated:
<38 = excellent, 39-76 = good, 77-114 = fair, and 115+ = poor.
Pfankuch's method can be used to compare study and reference stream stability.
Again, it is cr/Y/ca/that the reference and study streams have the same stream
channel classification for valid comparisons of channel stability. Rosgen's (1993)
channel classification system stratifies streams by geomorphologic characteris-
tics, which translates into an evaluation of relative stability. The degree of chan-
nel stability varies by channel type, i.e., the stage of geomorphologic develop-
ment of the channel and land forms. The best channel stability rating for an
unstable stream class may be 85; for a stable stream class it may be 50. Com-
paring different channel classes leads to erroneous judgments of management
impacts on channel stability.
If there is a significant difference in the total score between study and reference
streams, comparing individual element scores should reveal what elements are
different. The next step is to determine if an impaired element in the study
94
-------
stream can be linked to a management activity or some other cause in the
watershed.
Rosgen's Stream Channel Classification
Rosgen (1985) developed a stream classification system, which is primarily used
for classifying reference and study streams. However, it can be used to monitor
changes in channel geomorphoiogy. The narrative below comes from Rosgen's
1985 paper; the description of the stream classes reflect 1993 revisions (per-
sonal communication). This method can be used at all levels of monitoring, but
the level of detail will vary.
It has long been a goal for individuals working with rivers to define and under-
stand the processes which influence the pattern and character of river sys-
tems. Their differences as well as their similarities under diverse setting pose
a real challenge for study. One consistent axiom associated with rivers is
that what initially appears complex is even more so under further investiga-
tion.
Obviously, over-simplification of such complex systems may appear pre-
sumptuous. However, the need to categorize stream and river systems by
channel morphology is apparent for the following reasons: (1) the need to
predict a river's behavior from its appearance; (2) the need to extrapolate
specific data collected on a given river reach to another of similar character
and; (3) the need to provide a consistent and reproducible frame of reference
for those working with river systems.
Stream Classification Criteria
The purpose of this classification scheme is to categorize natural stream
channels on the basis of measurable morphological features. Thus, consis-
tent and reproducible descriptions and interpretations can be readily obtained
over a wide range of hydrophysiographic regimes.
There are many observable stream channel features governed by the laws of
physics which operate to form the morphology of the present day channel.
Stream morphology and related channel patters are directly influenced by
eight major variables including width, depth, velocity, discharge, slope, rough-
ness of channel materials, sediment load and sediment size (Leopold et al.,
1964). A change in any one of these variables sets up a series of concurrent
changes in the others, resulting in altered channel patterns. Since stream
morphology is a result of an integrative process of mutually adjusting vari-
ables, those most directly measurable have been incorporated into the
delineative criteria for stream types. Selection of the delineative criteria for
stream classification was developed from detailed analysis of hundreds of
streams over many hydrophysiographic regions and from portions of existing
classification schemes.
The stream type classification is summarized below, displayed in figures, and
includes the following criteria:
95
-------
1.
2.
3.
4.
5.
6.
Channel gradient (measured as energy slope of the water surface),
Sinuosity (ratio of channel length to valley length),
Width/depth ratio (width at bankful stage divided by bankful depth),
Dominant particle size of bed and bank materials,
Entrenchment of channel and confinement of channel in valley (ratio of
flood prone width divided by bankful width),
Landform features, soil credibility and stability.
Stream Classification Key
Single Thread Channels
- Well Entrenched (Ratio <1.4)
- Low Width/Depth Ratio (<12)
- Low sinuosity (<1.2)
- Slope Range >. 10 (Figure 11)
Channel Materials
- Bedrock
- Boulders
- Cobble
- Gravel
-Sand
- Silt/Clay
- Slope Range
Channel Materials
-Bedrock
- Boulders
- Cobble
- Gravel
-Sand
- Silt/Clay
- Slope Range .02-.039
Channel Materials
- Bedrock
-Boulders
- Cobble
- Gravel
- Sand
- Silt/Clay
Moderate Sinuosity (>1.2)
- Slope Range .02-.039 (Figure 12)
Channel Materials
- Bedrock
- Boulders
- Cobble
- Gravel
-Sand
- Silt/Clay
Channel Class
A1a+
A2a+
A3a+
A4a+
A5a+
A6a+
.04-.099
A1
A2
A3
A4
A5
A6
A1b
A2b
A3b
A4b
A5b
A6b
G1
G2
G3
G4
G5
G6
96
-------
- Slope Range <.02
Channel Materials
- Bedrock
-Boulders
- Cobble
- Gravel
- Sand
- Silt/Clay
• Moderate-High Width/Depth Ratio (>12)
- High Sinuosity (>1.4)
- Slope Range .02-.039 (Figure 13)
Channel Materials
- Bedrock
- Boulders
- Cobble
- Gravel
-Sand
- Silt/Clay
- Slope Range <.02
Channel Material
- Bedrock
- Boulders
- Cobble
- Gravel
-Sand
- Silt/Clay
G1c
G2c
G3c
G4c
G5c
G6c
F1b
F2b
F3b
F4b
F5b
F6b
F1
F2
F3
F4
F5
F6
97
-------
Stream Type
Floodprone Width
Bankfull Width
Figure 11. Rosgen Class A Streams.
Stream Type
G
Figure 12. Rosgen Class G Streams.
Stream Type
F
Figure 13. Rosgen Class F Streams.
98
-------
Moderately Entrenched (Ratio, 1.4-2.2)
- Moderate Width/Depth Ratio (>12)
- Moderate Sinuosity (>1.2)
- Slope Range (.04-.099) (Figure 14)
Channel Material
- Bedrock
- Boulders
- Cobble
- Gravel
- Sand
- Silt/Clay
- Slope Range (.02-.039)
Channel Material
- Bedrock
- Boulders
- Cobble
- Gravel
-Sand
- Silt/Clay
- Slope Range (<.02)
Channel Material
- Bedrock
- Boulders
- Cobble
- Gravel
- Sand
-Silt/Clay
- Not Entrenched (Ratio >2.2)
- Very Low Width/Depth Ratio (<12)
- Very High Sinuosity (>1.5),
- Slope Range (.02-.039) (Figure 15)
Channel Material
- Cobble
- Gravel
- Sand
- Silt/Clay
- Slope Range (<.02)
Channel Material
- Cobble
- Gravel
- Sand
-Silt/Clay
B1a
B2a
B3a
B4a
B5a
B6a
B1
B2
B3
B4
B5
B6
B1c
B2c
B3c
B4c
B5c
B6c
E3b
E4b
E5b
E6b
E3
E4
E5
E6
99
-------
Stream Type
Figure 14. Rosgen Class B Streams.
Stream Type
E
Figure 15. Rosgen Class E Streams.
100
-------
Moderate-High Width/Depth Ratio (>12)
- High Sinuosity (>1.4)
- Slope Range (.02 - .039) (Figure 16)
Channel Material
- Bedrock
- Boulders
- Cobble
- Gravel
- Sand
-Silt/Clay
- Slope Range (<.02)
Channel Material
- Bedrock
- Boulders
- Cobble
- Gravel
- Sand
- Silt/Clay
C1b
C2b
C3b
C4b
C5b
C6b
C1
C2
C3
C4
C5
C6
Stream Type
C
Figure 16. Rosgen Class C Streams.
101
-------
Multiple Channels
- Very High Width/Depth Ratio (>50)
- Unstable
- Slope Range (. 02-. 039) (Figure 17)
Channel Material
- Cobble
- Gravel
-Sand
- Silt/Clay
- Slope Range (<.02)
Channel Material
- Cobble
- Gravel
-Sand
- Silt/Clay
' -Stable
- Slope Range (<.005)
Channel Material
- Gravel
-Sand
-Silt/Clay
D3b
D4b
D5b
D6b
D3
D4
D5
D6
DA4
DAS
DAG
Stream Type
D
Figure 17. Rosgen Class D Streams.
102
-------
Bankfull channel width, average channel depth, and channel cross-sectional area
can be compared for reference and study streams with the same channel classifi-
cation. The following discussion is based upon material presented by D. Rosgen
and L B. Leopold in their Short Course on Rivers and Operational Hydrology
(1991):
The bankfull discharge is one of the most important characteristics of a channel
location. It is not only the discharge that carries over a long time the maximum
amount of sediment, but is it the channel forming (channel maintaining) dis-
charge. It is a quantity that integrates all the parameters of the basis and is in a
way the fingerprint of the basin.
Therefore the level or stage of the bankfull condition is crucial to an understand-
ing of the river. Keep in mind that the bankfull stage is a geomorphic not a
hydraulic feature.
The movement or migration of the channel across the valley floor results in the
formation of a flood plain. The flood plain is defined as the flat area near the
channel constructed by the river in the present climate and inundated at times of
high flow.
This construction is carried out primarily by the growth or extension of point bars
and so the top of the point bar is the elevation of the flood plain and is the eleva-
tion of the bankfull stage. Thus the elevation of the point bar is on of the best
indicators of bankfull condition.
The next most important indicator of change from point bar to flood plain is a
change in vegetation. A change in slope is an indicator, for example where a
bank sloping upward becomes a vertical bank, or where a vertical bank grades
upward to a sloping surface. Another indicator is a change in size or size-distri-
bution of bank material, for example, a change from cobble to mixture of sand,
cobbles and fine gravel.
In most regions of the country, there is an abandoned flood plain above the flood
plain which is called a terrace.
Of all the parameters, bankfull width is the most easily adjusted and is sensitive
to the input of water and sediment. The measurement of bankfull width should
be done on a straight stream reach. At the same location, the average bankfull
depth needs to be measured.
In several regions of the country, studies have document relationships between
channel parameters and watershed area. Bankfull width, bankfull average depth
and cross sectional area have been correlated to watershed area.
Relationships of bankfull width, average bankfull depth and cross sectional area
to watershed area can be developed for reference streams. At 10 to 20 straight
reaches located from the headwaters to larger order streams lower in the water-
shed, bankfull widths, average bankfull depths and cross sectional area are
103
-------
measured in the field. The points of measurement are carefully located on .
topographic maps. The watershed boundaries above measure points can be
drawn on the topographic maps and the watershed area estimated from the
maps. Bankfull widths, average bankfull depths and cross sectional areas are
plotted against watershed areas to develop the relationships.
These relationships can be used to evaluate the impacts of management on
channel characteristics in study streams. A couple hypothetical examples will
demonstrate the use of these relationships. First, a study reach extends from
above to below a management activity that has removed the riparian vegetation
and the stream banks are unstable. Bankfull widths, average bankfull depth and
cross sectional are measured at straight reaches above, adjacent and below the
management activity. These points are located on a topographic map and the
watershed area above each study point is determined.
The watershed area for the point above the management area is entered into the
reference relationships to get expected bankfull width, average bankfull depth
and cross sectional area. The study stream width, depth and cross sectional
area are compared with reference values. In this example, we will assume they
are essentially the same.
However, the values for the adjacent and below reaches are significantly different
from what would be expected for their respective watershed areas. The stream
is wider and shallower than expected. This would suggest channel aggradation
is occurring, heavy sediment deposition from the management activity or channel
instability from the removal of riparian vegetation. Further investigation will be
needed to determine the causefs). The upstream reach was as expected sug-
gesting that the impact is a reflection of the management activity.
Further investigation finds a low water bridge with a small culvert below the
management area that is functioning as a sediment trap, the head of the sedi-
ment deposit is located near the top of the reach adjacent to the management
area, and the removal of the riparian vegetation is facilitating the widening and
shallowing of the stream. So, in this example, the problem rests with a poorly
designed stream crossing and with the removal of riparian vegetation.
The second example addresses the concern about cumulative impacts of several
management areas in a watershed. In this case, the relationships of bankfull
width, average bankfull depth and cross sectional area are developed for the
study stream. The relationships between the reference and study streams are
compared to determine whether cumulative impacts have occurred or not.
In many streams, the comparison of particle size distributions within stream
habitat units can reveal impacts. To do this requires the study reach have the
same stream class as the reference reach. The particle size distribution for the
reference stream is developed by taking a 100 count sample ofstreambed
materials in a cross section of the stream and tallying them by particle size. The
percent composition of each particle type (bedrock, boulders, cobble, gravel,
104
-------
sand and silt/clay) is computed from the tally. The same process is used in the
study reach (the study and reference reach need to be as similar as possible).
Shifts in percentages in particle sizes can reveal impacts, for example a low
percentage of sand in the reference site versus a high percentage of sand in the
study site. (End of reference to Rosgen and Leopold 1991).
The detailed evaluation and comparison of bankfull widths, average bankfull
depth, cross-sectional area, and particle size distributions would be used at the
BMP Effectiveness III and IV levels of monitoring.
Also, BMP Effectiveness III and IV assessments could use detailed classification
and mapping of channel types from headwaters to the mouth of the watershed.
The channel type will vary with location on the stream. The reference stream will
set the pattern of expected channel types as the stream descends to the mouth.
If at a similar location along the study stream an unexpected channel class
occurs in association with management, an impact of management might be
indicated. For example, a D4 (braided) stream could be found where a C4
(single channel) stream would be expected. Or C5 (sand bottom) reach could be
found where a C3 (cobble bottom) is expected.
At the BMP Effectiveness I and II levels of monitoring, comparing channels
above, adjacent to, and below a management area may reveal changes in
classes, which, through additional observations, can be linked to an activity. An
observer knowledgeable about reference stream conditions may know what
channel types should be present given the location in the study watershed.
Alaska Channel Classification
Channel types are used to distinguish the various parts of a stream system.
They allow the user to define the characteristics of the channel and to accurately
predict probable responses to natural and human influences, and thus are tools
intended to complement a holistic approach to watershed management. How-
ever, channel types cannot be managed as isolated segments. A stream reach
in one part of a watershed can be affected by activities taking place in a different
part of the watershed, either upstream, downstream, or on adjacent land areas.
The Alaska Region Channel Type Classification is defined according to nine
basic fluvial process groups (Paustian et al. 1992). These process groups
describe the interrelationship among watershed runoff, landform, geology, and
glacial or tidal influences on fluvial erosion and deposition processes. Individual
channel type classification units within each process group are defined by physi-
cal attributes, such as channel gradient, channel pattern, stream bank incision
and containment, and riparian plant community composition.
The Channel Type Classification System was developed with water-resource
management needs in mind. Propagation of anadromous fisheries is the major
beneficial use of water resources in Southeast Alaska (Paustian et al. 1992).
Channel-type inventories provide key information on fish habitat utilization and
capability and fisheries enhancement options in survey area watersheds.
105
-------
Channel types also provide information on suitable stream-crossing locations and
design criteria for road drainage structure. Channel types are used to evaluate
potential sediment delivery and retention for cumulative watershed-effects analy-
sis.
Classes are designated by a channel type name and the process group name.
Process Group Name. Channels bear the signatures of the processes that
formed them. Channels formed and maintained by the same or similar fluvial
processes are grouped for taxonomic purposes. Process groups reflect the long-
term interaction of geology, landform, climate, and riparian vegetation. They
characterize the basic interrelationships among the runoff, sediment transport,
and vegetation patterns along stream banks.
Channel Type Name. Within each process group are a number of channel
types that further define differences and describe individual channels. Channel
types have fewer variables than the process group. This allows for more site-
specific analysis and prescription for project plans. Some channel types may
have one or more phases, or common variants, which influence management
interpretations.
The following are the Process Group Names and associated Channel Types:
Estuarine Process Group
- Silt Substrate Estuarine Channel or Slough
- Narrow Small Substrate Estuarine Channel
- Narrow Large Substrate Estuarine Channel
Large Estuarine Channel
- Broad Braided Glacial Outwash Estuarine Channel
Palustrine Process Group
- Narrow Placid Flow Channel
- Moderate Width Placid Flow Channel
- Shallow Ground Water Fed Slough
- Flood Plain Backwater Slough
- Beaver Dam/Pond Channel
Flood Plain Process Group
Uplifted Beach Channel
Foreland Uplift Estuarine Channel
Narrow Low Gradient Flood Plain Channel
Low Gradient Flood Plain Channel
Wide Low Gradient Flood Plain Channel
Glacial Outwash Process Group
- Glacial Outwash Flood Plain Side Channel
- Large Meandering Glacial Outwash Channel
- Large Braided Glacial Outwash Channel
- Moderate Width Glacial Channel
- Cirque Channel
106
-------
Alluvial Fan Process Group
Moderate Gradient Alluvial Fan Channel
High Gradient Alluvial Cone Channel
Glacial Alluvial Cone Channel
Large Contained Process Group
- Low Gradient Contained Channel
Moderate Gradient Contained Narrow Valley Channel
Moderate Gradient Mixed Control Process Group
Narrow Mixed Control Channel
Moderate Width Mixed Control Channel .
Moderate Gradient Contained Process Group
Narrow Shallow Contained Channel
- Moderate Width and Incision, Contained Channel
- Deeply Incised Contained Channel
High Gradient Contained Process Group
- Shallow Incised Muskeg Channel
Shallow to Moderately Incised Footslope Channel
- Deeply Incised Upper Valley Channel
- Deeply Incised Muskeg Channel
Shallow Incised Very High Gradient Channel
Deeply Incised Mountainslope Channel
Moderate/High Gradient Glacial Cascade Channel
High Gradient Incised Glacial Torrent Channel
V* Pool Sediment
Lisle and Hilton (1992) developed a method to evaluate and monitor the supply
of mobile sediment in gravel-bed streams and to detect and evaluate sediment
inputs along a channel network by measuring fine sediment in pools. This
method applies primarily to the BMP Effectiveness HI and IV levels, and occa-
sionally to Level II when coupled with other qualitative methods.
Mobile sediment tends to be concentrated in pools; thus the fraction of the pool
filled indicates the supply of mobile sediment in natural, gravel-bed channels.
During high flow, fine sediments in scour pools are remobilized. During waning
flows, fine sediment is selectively transported from zones of relatively fast-
moving water, such as riffles, and deposited in zones of slow-moving water, such
as pools, where they cover a coarser substrate.
Fine sediments tend to move fast and can be flushed rapidly from streams.
Therefore, their high concentration on a streambed can indicate widespread,
chronic supplies or recent, local inputs (Platts and Megahan 1975). Abundant
fines on the bed surface may indicate an increase in sediment supply (Dietrich et
al. 1989). Finally, fine-sediment abundance can indicate a reduction in transport
capacity without a compensating decrease in sediment supply.
107
-------
Study reaches should have gentler channel gradients than adjacent reaches
because gentle reaches tend to store more sediment and are sensitive to
changes in sediment inputs. All study reaches should have well-developed riffle-
pool sequences. The streambed surfaces should be covered by a layer of
gravel, cobble and/or boulders. The procedure will hot work well if the stream
beds were originally sand, silt, or clay.
Fine sediment deposits and residual pool volumes are measured during low-flow
periods. All depth measurements are made with a graduated 0.5-inch rod. The
datum is the riffle crest below the pool. The water depth over the riffle crest is
measured to define the datum. The residual pool is that pool volume remaining
after a portion of the scour pool has been filled with fine sediment below the
datum or riffle crest.
A series of transects is established across the pool and the depth to the stream-
bed is measured at points along each transect. The residual pool depth is
measured at the intercept of the rod with the water surface, minus the average
water depth over the riffle. For example, if water depth is 2.5 feet and riffle water
depth is 0.5 feet, the residual water depth is 2.0 feet.
At the same points, the rod is driven carefully into the fine-sediment deposit until
it hits the buried substrate. The rod will stop or the observer will feel a vibration
or resistance in the rod when it hits larger particles. The depth of fine sediment is
the difference between depth to streambed and the depth to substrate. With a
series of transects, the profile of the residual pool and fine sediment deposits can
be plotted and the volume of each computed.
V* equals the volume of fine sediment divided by the sum of the residual pool
volume and fine sediment volume, i.e., the fraction of the scour pool (the sum of
the residual pool volume and fine sediment volume) filled with fine sediment. V*
can be averaged for a series of pools in a reach below a management area and
compared with the average V* for the reference reach above or in a nearby
watershed. For example a V* for the reference reach of 0.1 and V* for the study
reach of 0.3 would suggest the management activity supplied sediment to the
stream.
Rosgen Channel Monitoring
Rosgen (personal communication, 1993) uses the following method to monitor
channel and stream-bank erosion, which is appropriate for BMP Effectiveness III
and IV monitoring. To isolate changes in channel stability and sediment supply
caused by climate, geology, and morphology from those caused by manage-
ment, monitoring should use one of the following three approaches:
1. Compare study reach with a reference reach (control).
2. Compare above versus below impact area for same stream type.
3. Compare before versus after management activity.
108
-------
The field methods for evaluating channel stability and sediment supply are:
Channel Stability—Stability is assessed by measuring vertical and lateral
changes in the stream channel.
Vertical Stability (Aggradation/degradation)
To determine if the stream is either down cutting (degrading), filling (aggrad-
ing), or is stable, permanently monumented cross-sections are used on at
least one riffle and pool segment of a reach (or step vs pool segment if a
step/pool stream type).
The bench mark for the cross-section should be located on a stable site above
the active channel (Figure 18). An elevation cross-reference is often needed if
one side of the cross-section is located on an unstable slope. An elevation
bench mark is set, not necessarily tied into an absolute elevation, but rather a
relative elevation set at 100 feet. To mark the ends of the transect, two holes are
dug and filled with half a bag of premixed concrete. A 10-inch stove bolt is
placed flush in the concrete before it sets. The transect markers need to be at
the same elevation.
The profile of the stream cross-section is measured with a rod from a "leveled"
tape line (Grant et al. 1992). The cross-section should be re-surveyed annually
and/or following stormflow/snowmelt runoff events. Any "high water" marks need
to be indicated on the cross-section. The cross-section profile is plotted for each
measurement and compared to previous profiles.
Use Sag-tape Method as an alternative to using level/tripod survey instruments
(Grant et al. 1993) as follows:
1. Locate permanent bench mark on both sides of stream.
2. Stretch tape tightly with spring clamp and level tape, i.e. tape at same
elevation as reference bolt on bench mark.
3. Read distance and elevation reading of rod intercept with tape.
4. Measure major features such as:
a. bench marks
b. terraces/floodplains
c. bankfull marks
d. banks
e. edges of water
f. differences in bed configurations across bed
g. thalweg
h. inner-berm features
109
-------
Stretched Tape
Rod intercepts with streched tape
to measure profile of stream cross-
section.
RBM
RBF
Left vs. right is determined.
by looking downstream.
Figure 18. Example of a permanent crbss-section with bench mark locations and points of measurement.
Lateral Stability
To determine the rate and magnitude of bank erosion, bank "pins" and/or
bank profiles are installed on sites representative of the stream banks on the
river (Figure 19). These should be stratified and installed on outside bends
and straight reaches with different erodibility along stream segments of the
same stream type. Erosion rates can be expressed in feet/year, cubic yards/
year, or total tons/stream reach.
The erosion-pin method involves inserting two or three smooth rods 4 to 5
feet long and 0.3 to 0.5 inch in diameter, horizontally into the banks at a
permanent cross-section (Figure 19). Periodically, or at least before and after
storms, the distances between the end of the rods and the bank are mea-
sured. This not only reveals the rate of lateral migration, but also indicates if
dimensions (such as bankfull width) and elevations are changing.
The bank prof He procedure involves (Figure 20):
(1) Installing a permanent cross-section over the bank profile site.
(2) Installing a permanent toe pin (rod) offset and directly adjacent to the
study bank.
(3) Placing rod level on survey rod that is set on toe pin.
(4) Stabilizing rod with either a tripod or a frame to attach to bank to hold
rod "plumb."
(5) Measuring horizontally with tape rule from vertical rod to bank. (Mea-
sure at frequent intervals to describe bank dimensions and features.)
(6) Plotting data to display profile for each survey.
(7) Comparing with previous surveys annually or following storms (Figures
21 and 22).
(8) Computing mean erosion rate.
110
-------
IBM
Figure 19. Measuring bank erosion with bank pins.
Rod-
Toe Pin -
Figure 20. Bank profile procedure using toe pin and survey rod.
111
-------
Bed Material Size Distribution
To observe a shift in bed-material size distribution use one of the following:
(1) Transect to obtain existing and departures from particle size distributions
by measuring 100 particles from bed material at riffles, pools, and spe-
cial features of interests (spawning reds, fish-habitat structures, etc.).
(2) Riffle Armor Index (Kappesser 1993) where fine material deposition
changes can be quantified.
(3) Freeze core sampling and other measures documented in MacDonald
etal. (1991).
If a pebble-count method is used without a permanently established transect for
resurvey, then samples should be taken in proportion to the frequency of riffles
and pools. For example, if 80 percent of the channel is composed of riffles, then
80 percent of the cross-sections sampled should be in riffles. Permanent
transects are preferred for a higher level of detail, to make replication easier, and
to reduce sample error.
Pebble Counts
Pebble counts can be used at all levels of monitoring to determine if the composi-
tion of stream-bed surfaces differs between reference and study streams, which
can reflect differences in sediment supply. The pebble count determines the
proportion of bed material less than a given size. It can be done in wadable
streams with beds of coa^ material. A pebble count consists of selecting at
least 100 particles located by the toe of the observer's boot as he/she walks
along a transect from bankfull to bankfull stage (Leopold 1970). A particle has
three axes that are mutually perpendicular: long, short, and intermediate in
length. The intermediate axis of each particle is measured. Pebbles down to
2 mm can be easily measured, but smaller sizes are judged by feel, using soil-
classification methods.
Particle sizes are tallied by the following classes: <:2 mm, 2-4 mm, 4-8 mm,
8-16 mm, 16-32 mm, 32-64 mm, 64-128mm, 128-256 mm, 256-512 mm,
512-1024 mm, 1024-2048 mm and > 2048 mm. A cumulative size distribution
curve is created by plotting the cumulative percent fines over particle size class.
Reference and study reaches should be compared where reaches are of the
same stream size and expected geomorphic classification. Two hypothetical
cumulative size-distribution curves are presented in Figure 23. The study stream
contains many more fine particles than the reference stream. If on-slope moni-
toring of the management activity traces sediment to the stream and no other
source can be identified, then the differences can be attributed to the manage-
ment activity.
112
-------
Figure 21. Comparison of relative elevations for cross-sections for a 2-year period.
Toe Pin
Figure 22. Combined cross-section and bank erosion study fora 2-year period.
113
-------
Percent
100
10
Particle Size - mm
100
1,000
Figure 23. Comparison of two hypothetical cumulative sediment size curves between a reference and study
stream.
114
-------
Habitat Monitoring Methods
This section summarizes various habitat monitoring and classification methods to
assist the readers in selecting the method most appropriate to their needs. In
addition to delineating habitat types for comparison in habitat monitoring, habitat
classification defines habitat niches to sample for benthic macroinvertebrates,
fish, and aquatic vegetation. Efforts to evaluate habitat should describe current
habitat and the extent to which it resulted from human activity as opposed to
natural processes at the site.
For more detailed information, the reader is referred to the original literature. As
with the biological monitoring methods, some of the discussions are excerpts
from the source papers and others are summary statements. No distinction is
made, but all are clearly documented.
Habitat Monitoring Methods by Monitoring Level
The habitat monitoring methods abstracted here evaluate aquatic and terrestrial
habitat conditions associated with benthic macroinvertebrates and fish. Specific
monitoring methods are recommended for each level of monitoring:
BMP Effectiveness I
Empirical/perceptual
Rapid Bioassessment Protocol
Streamwalk
T-Walk
BMP Effectiveness II
Rapid Bioassessment Protocol
Basinwide Estimation of Habitat
Hankin-Reeves
BMP Effectiveness III
Rapid Bioassessment Protocol
Bjsson
Hankin-Reeves
Forest Service R-5, R-1, R-4 FHR
Platts Riparian Habitat
Ohio-QHEl
EMAP
BMP Effectiveness IV
Rapid Bioassessment Protocol
Hankin-Reeves
Bisson
Forest Service R-5, R-1, R-4 FHR
Platts Riparian Habitat
Ohio-QHEl
EMAP
Page
116
116
118
119
116
123
123
116
120
123
124
126
126
118
116
123
120
124
126
126
118
115
-------
Empirical/Perceptual
At the BMP Effectiveness I level, a trained professional needs to know the key
habitat needs of invertebrates and fish, the habitat types, the prevalence of each
habitat type, and the quality of the reference streams. The observer also needs
to know the kinds and distribution of habitat types; the amount, type, and over-
hang of stream bank vegetation; bank angle and stability; pool depths; amount
and distribution of large woody debris; the amount of stream surface shading;
type and density of riparian vegetation; substrate type; percent embeddedness or
percent fines in the substrate; etc. To supplement the visual picture, the ob-
server may take notes or complete a form on reference habitat conditions and
may even take photographs of the reference habitat conditions to assist in the
comparing the study stream with the reference stream.
The key habitat elements will vary among ecoregions and the types of fish and
benthic macroinvertebrate communities. In some ecoregions and for some
fisheries, key habitat requirements have been developed that can be used in
assessing the habitat quality of reference and study streams. As State water-
quality agencies develop narrative and numeric criteria, key habitat criteria will be
identified and should be used.
The observer mentally compares the study stream habitat with reference stream
habitat, judging whether they are similar or whether the study stream habitat is
impaired. Then the observer tries to determine whether the impairment was
caused by the management situation under evaluation.
Rapid Bioassessment Protocol—Habitat
The Rapid Bioassessment Protocol—Habitat (RBP) is being adapted and
adopted by State water-quality agencies. RBP Habitat applies to each RBP
biological monitoring level because it is flexible. The RBP Habitat Assessment
will be adapted to the ecoregions, invertebrates and fisheries of each State. If
the State has developed RBP habitat elements, forest water-quality studies
should use them in evaluating habitat responses to management.
According to Plafkin et al. (1989), the habitat assessment matrix (Table 8) is
based on the "Stream Classification Guidelines for Wisconsin" developed by Ball
(1982) and "Methods of Evaluating Stream, Riparian, and Biotic Conditions"
developed by Platts et al. (1983). Because this habitat assessment is intended
to support biosurvey analysis, the various habitat variables are weighted to
emphasize those most biologically significant. All variables are evaluated for
each station studied and rated according to a numerical system. The ratings are
totalled and compared to a reference for final habitat ranking. Scores increase
as habitat quality increases. To ensure consistency in the evaluation procedure,
descriptions of the physical variables and related criteria are included in the
rating form.
Reference conditions are used to normalize the assessment to the "best attain-
able" situation. This is critical because stream characteristics vary dramatically
among ecoregions. Other habitat assessment systems may be used, or a more
precise quantitative approach may be taken. However, the importance of a
holistic habitat assessment to enhance the interpretation of biological data
cannot be overemphasized.
116
-------
Habitat variables pertinent to the assessment of habitat quality are separated into
three principal categories: primary, secondary, and tertiary. Primary variables
are those that characterize the stream "microscale" habitat and exert the greatest
direct influence on the structure of the indigenous communities. These variables,
which comprise characterization of the bottom substrate and available cover,
and estimates of embeddedness, and flow or velocity and depth regime, have the
widest score range (0-20) reflecting their contribution to habitat quality. The
secondary variables define the "macroscale" habitat, i.e., channel morphology
characteristics such as channel alteration, bottom scouring and deposition, and
stream sinuosity. These variables have a score range of 0-15. Tertiary variables
define riparian and bank structure and include: bank stability, bank vegetation,
and streamside cover, variables most often ignored in biosurveys. These vari-
ables have a score range of 1-10.
Instream habitat is evaluated first, followed by channel morphology, and finally
structural features of the bank and riparian vegetation. Stream segment length
or area assessed will vary with each site. Generally, primary variables are
evaluated within the first riffle/pool sequence or the immediate sampling area,
such as in fish sampling.
Secondary and tertiary variables are evaluated over a larger stream area, prima-
rily upstream, where conditions will have the greatest impact on the community
being studied. The actual habitat assessment process involves rating the nine
variables as excellent, good, fair or poor, based upon criteria included on the
Habitat Assessment Field Data Sheet.
Table 8. RBP Habitat Rating
Condition
Habitat Variables
Excellent Good
Fair Poor
Primary-Substrate & Instream Cover
1. Bottom substrate and available cover
2. Embeddedness
3. Flow/velocity
Secondary-Channel Morphology
4. Channel alteration
5. Bottom scouring and deposition
6. Pool/riffle, run/bend ratio
Tertiary-Riparian and Band Structure
7. Bank stability
8. Bank vegetation
9. Streamside cover
16-20
16-20
16-20
12-15
12-15
12-15
9-10
9-10
9-10
11-15
11-15
11-15
8-11
8-11
8-11
6-8
6-8
6-8
6-10
6-10
6-10
4-7
4-7
4-7
3-5
3-5
3-5
0-5
0-5
0-5
0-3
0-3
0-3
0-2
0-2
0-2
117
-------
A total score is obtained for each study stream and compared to a reference
stream. The score for the study stream is expressed as a percent of that for the
reference stream and the study stream is then classified on the basis of its
apparent potential to support an acceptable level of biological conditions
(Table 9).
Table 9. Comparability Assessment
Assessment Category Percent of Reference
Comparable to Reference
Supporting
Partially Supporting
Non-Supporting
> 90%
75-88%
60-73%
< 59%
Environmental Monitoring and Assessment Program (EMAP)
The USEPA (1993) is in the process of developing and testing a more quantita-
tive system for evaluating habitat and the chemical, physical, and biological
condition of streams. Occasionally, this system may be used in BMP Effective-
ness III and IV levels. EMAP is mentioned to make the reader aware that a more
detailed habitat assessment system is available than RBP.
Streamwalk
Streamwalk is a BMP Effectiveness I method that provides a qualitative assess-
ment of habitat conditions (Idaho WRR11992). The variables may vary, but in
Idaho the following are evaluated:
STREAM CHARACTERISTICS (Measured or estimated)
- Depth (feet)
- Width (feet)
- Gradient from map (Feet/100 feet)
- Pools and riffles (percent of stream)
STREAM BANK (Estimated percent of all banks in each category)
- % of banks mostly covered with vegetation
- % of banks mostly not covered with vegetation
- % of banks stable (not eroding, slumping, or sloughing)
- % of banks unstable (eroding, slumping or sloughing)
STREAM BOTTOM (Estimate percent in each category)
- % Silt/Clay/Mud
- % Sand (up to .1")
- % Gravel (.1 to 2")
-% Cobble (2 to 10")
-% Boulders (over 10")
- % Bedrock (solid)
118
-------
STREAMSIDE RIPARIAN VEGETATION WIDTH (Feet)
(Estimate to nearest 10 feet)
Left (looking downstream) Right
STREAMSIDE VEGETATION TYPE (check appropriate boxes)
None/Sparse Occasional Common
Conifers [] [] []
Deciduous [] [] []
Shrubs [] [] []
Grasses [] [] []
Vegetation Appears: Natural [ ] or Cultivated [ ]
Extent of vegetative canopy shading the water surface
[]0-25% [] 25-50% [] 50-75% [] 75-100%
Extent of artificial bank protection
[]0-25% [] 25-50% [.] 50-75% [] 75-100%
Presence of logs or large woody debris in stream
[ ] None [ ] Occasional [ ] Common
Presence of other organic debris (small limbs, etc.) in stream
[ ] None [ ] Occasional [ ] Common
By applying the procedure to both study and reference streams, quality of habitat
can be compared. Based upon experience, the observer can judge whether the
study and reference streams are similar or whether habitat is impaired in the
study stream compared to the reference stream.
Ocular estimation of some variables may be difficult and results have
varied among observers and sampling periods. This can influence reliabil-
ity of the data and repeatability of the methods.
T-Walk
Thalweg Walk (T-Walk) uses qualitative data to evaluate the effectiveness of
management in protecting habitat, channel stability, and macroinvertebrates
(Ohlander 1991). The method deals chiefly with stream habitat condition and
channel stability, coupled with qualitative sampling of invertebrates and an
estimate of invertebrates as fish food, based upon substrate size distribution.
The method primarily involves monitoring above and below the management site
and occasionally includes reference streams. The recommended length of
stream segments to evaluate is 600 feet. Substrate size, sand infill, vegetation,
maximum and minimum water depth, and flow velocity are measure^ at 30 to
100 points along the segment.
119
-------
Based upon the substrate composition, a Tarzwell substrate ratio is determined.
The Tarzwell substrate ratio is a dimensionless measure of fish-food organisms
produced on different substrates, indexed to sand as the least productive.
Tarzwell (1937) quantified fish food production for 25 different substrate condi-.
tions.
Channel morphology is characterized by evaluating stream slope, sinuosity,
bankfull width and depth, channel depositional features (types of bars), bank
conditions, and stability.
Riffle insects are evaluated by sampling 10 cobbles, and identifying and enumer-
ating the various invertebrates.
Adjacent land uses and conditions are assessed.
Compared with reference conditions, the study stream characteristics are rated
as: robust, adequate, diminished, impaired, precarious, or catastrophic. These
qualitative ratings are used to judge whether management has been effective.
The system is designed to identify early those management activities that are
clearly not effective.
Bisson Habitat
Bisson et al. (1981) presented a system for classifying habitat in small streams
that can be used in BMP Effectiveness III and IV assessments. Fish habitat is
classified according to location within the channel, pattern of water flow, and
nature of flow controlling structures. Riffles are divided into three habitat types:
low gradient riffles, rapids, and cascades. Pools are divided into six types:
secondary channel pools, backwater pools, trench pools, plunge pools, lateral
scour pools, and dammed pools. Glides, the last habitat type, are intermediate in
many characteristics between riffles and pools.
Bisson etal. (1981) habitat classification system is as follows:
Low gradient riffles are shallow (< 20 cm deep) stream reaches with moder-
ate velocity (20-50 cm/sec) and moderate turbulence. Substrate is usually
composed of gravel, pebble, and cobble-sized particles (2-256 mm). An
upper gradient limit for this habitat type was arbitrarily set at 4 percent.
Rapids possess a gradient greater than 4 percent with swiftly flowing water
(> 50 cm/sec) having considerable turbulence. The substrate of rapids is
generally coarser than the substrate for low gradient riffles, and during low
streamflow conditions large boulders typically protrude through the surface.
Cascades are the steepest. Unlike rapids, which have an even gradient,
cascades consist of a series of small steps of alternating small waterfalls and
shallow pools. The usual substrate is bedrock or an accumulation of large
boulders; however, this habitat type is occasionally found on the downstream
side of woody debris.
120
-------
During low streamflow conditions there are six pool types, which are associ-
ated with the presence of bedrock outcroppings, large rocks, or large tree
stems and rootwads in the channel.
Secondary channel pools are those that remained within the baneful mar-
gins of the stream after freshets. During the summer, most of these pools will
usually disappear, and those remaining have little flow through them. Sec-
ondary pools are usually associated with gravel bars, but many contain sand
and silt substrates.
Backwater pools are found along channel margins and are caused by
eddies behind large obstructions such as rootwads and boulders. This pool
type is often quite shallow (< 30 cm) and tend to be dominated by fine
grained substrates. Like secondary channel pools, backwater pools possess
very low current velocities.
Trench pools are long, generally deep slots in a stable substrate. Channel
cross sections are typically U-shaped with coarse-grained bottom flanked by
bedrock walls. Current velocities in trench pools are the swiftest of any pool
type and the direction of flow is the most uniform.
Plunge pools occur where the stream passes over a complete or nearly
complete channel obstruction and drops vertically into the streambed below,
scouring out a depression. This type of pool is often large, quite deep
(> 1 m), and possesses a complex flow pattern radiating from the point of
water entry.
Lateral scour pools differ from plunge pools in that the flow is directed to
one side of the stream by a partial channel obstruction. Often an undercut
bank is associated with this type of pool.
Dammed pools consist of water impounded upstream from a complete or
nearly complete channel blockage. Typical causes of dammed pools are
debris jams, rock landslides, or beaver dams. Depending on the size of the
blockage, dammed pools could be very large. Water velocity in dammed
pools is characteristically slow and substrates tend toward smaller gravels
and sand.
Glides are a third general habitat category which possesses attributes of
both riffles and pools. Glides are characterized by moderately shallow water
(10-30 cm) with an even flow that lacks pronounced turbulence. They are
usually located at the transition between a pool and the head of a riffle.
Glides are occasionally found in long, low gradient stream reaches with
stable banks and no major flow obstructions. The typical substrates are
gravel and cobbles.
Cover includes eight distinct kinds of cover for fish. These include three
kinds of wood debris—rootwads, large debris (tree stems), and small debris
(branches, twigs,etc.)—that differ in the amount of overhead cover and flow
121
-------
modification they provide within the channel. Overhanging terrestrial vegeta-
tion and undercut banks are two kinds of cover that are governed by the
condition of the riparian zone. Water turbulence acts as cover when the
presence of bubbles prevent a clear view of water beneath. Rocks function
as cover in two ways; (1) by providing overhanging ledges and (2) by provid-
ing crevices for hiding. Finally, deep water is itself a form of cover from non-
diving terrestrial predators. The primary function of cover during the summer
is assumed to be protection from predation.
Each stream reach is surveyed on foot and the location of different habitat
types, as well as significant flow controlling structures, is drawn to scale on a
map. Contour lines based on depth measurements are drawn within pools to
enable volume estimation. Wetted surface areas are determined by counting
squares on gridded paper that is superimposed on the maps. Axial length is
figured as the distance along the thalweg or greatest linear dimension of a
habitat unit parallel to the direction of flow. Reach summaries are con-
structed by summing the lengths, areas, and volumes of each habitat type
and expressing each group as a percentage of the total.
The amount of cover in each habitat is rated on a relative abundance scale of
0-3, where a score of zero indicates that the particular kind of cover is essen-
tially absent and a score of three indicates a very abundant condition. Sub-
strate is noted as according to the predominant type.
Fish populations are sampled by isolating individual habitat types with block ,
nets and electrofishing the habitat three times. The fish population is charac-
terized as to species, biomass, and age.
The utilization of habitat types by species and age classes is evaluated. In
order to quantify habitat utilization by species and individual age classes, it is
, necessary to relate the fraction of the pollution found within a particular
habitat type to the relative abundance of that habitat type in the stream. The
formula used is based upon the electivity index of Ivlev (1961):
Utilization = habitat specific density - average total density
average total density
Where:
habitat specific density = average density in the habitat type in question
average total density = average density over the entire stream reach, all
habitats combined
Values for habitat utilization coefficients can theoretically range from minus
one, indicating total non-use of a habitat type, to positive infinity as greater
proportion of the population resides in the habitat type on interest. A value of
zero indicates that the population occurs in the habitat type in proportion to
that type's abundance in the stream.
122
-------
Hankin-Reeves Habitat
Hankin-Reeves visual estimation techniques allow surveyors to inventory a
preselected set of habitat characteristics in an entire watershed. Habitat sam-
pling is done two steps (Hankin-Reeves 1988): (1) individual habitat units are
classified and visually estimated by type and habitat characteristics (e.g., surface
area, substrate composition), (2) visual estimates are paired with precise mea-
surements taken at a predetermined number of units (at least 10 for each habitat
type) to develop calibration ratios. Sampling teams typically consist of two
people: an estimator and a recorder.
Before heading for the field, sampling teams first: (1) select the classification
system they will use to identify habitat types (e.g., Bisson et al. 1981); (2) deter-
mine the habitat characteristics they will survey; and (3) stratify the study area
into survey units (reaches) based on gradient, confluence of same-order chan-
nels, or other distinctive features. The team then decides how many of the
habitat units in each habitat type will be measured. The number of units to be
measured is based on the consistency between estimates and measurements of
habitat areas and on the expected number of units of each habitat.type. For
example, a smaller number of pools would need to be measured if the team
always overestimated pool areas than if it sometimes overestimated and other
times underestimated. And if the habitat type is rare, the team may need to
measure most or even all the units of that particular type to ensure that the
number of paired measurements and estimates is at least the recommended
minimum of 10.
The area of measured habitat units is calculated as the product of mean width
and length. Mean width is derived from measurements taken at three or more
locations parallel to the thalweg along the length of the unit. The interval
between width measurements will depend on the complexity of the unit. A
5-meter interval may be appropriate for a 15-meter stretch of straight riffle.
However, that same length of stream in an irregularly shaped pool might require
a 2-meter measurement interval. Estimates and variances of total habitat area for
each habitat type within a stream reach are calculated using equations devel-
oped by Hankin and Reeves (1988).
The habitat distribution can be compared between study and reference streams.
The Hankin-Reeves method would be used to determine the habitat types and
areas, the relative mix, and habitat distribution along the stream. If key habitat
types are lower in quantity and quality in the study stream than in the reference
stream, habitat may have been impacted by management in the study water-
shed. Indications of habitat changes might include differences in the riffle/pool
ratio, number and size of pools, etc. Evaluating such shifts in habitat can reveal
the types of management impacts.
Basinwide Estimation of Habitat
Dolloff et al. (1993) modified the Hankin-Reeves method to achieve more consis-
tent interpretations of habitat types. This method is best suited for BMP Effec-
tiveness II. They limited habitat types to four: pool, riffle, cascade, and complex.
Using the Hankin and Reeves (1988) more detailed classifications leaves it to
123
-------
personal judgment to determine where habitat types change. In this method, if a
pool and riffle exist side by side along the stream, the dominant habitat type is
assigned to the stream segment.
The survey reaches are defined by gradient, confluence of same-order channels,
or other distinctive features. The habitat units are estimated for area and a
subset of habitat units is measured to compute area.
Additional habitat information is collected, including average and maximum
depth, dominant substrate, degree of embeddedness, number of pieces of large
woody debris by size classes, and features that are likely to influence fish popu-
lations, such as landslides, tributary junctions, bridges, trail crossings, and
significant changes in riparian vegetation. The total area of each habitat type is
estimated using equations from Hankin and Reeves (1988) and Dolloff et al.
(1993).
USDA Forest Service Region 5/Region 1/Region 4 FHR Stream Habitat
Classification and Inventory Procedures
This method is designed to survey both fish habitat and fish populations (Overton
et al. 1990). It is applicable to BMP Effectiveness III and IV assessments. Popu-
lation is estimated as described in Hankin and Reeves (see page 136). The
method presents a classification system that has 22 habitat types. Many are the
same as the Bisson et al. (1981) habitat types, including:
1. Low gradient riffles
2. High gradient riffles (rapids)
3. Cascade
4. Secondary channel pool
5. Trench/chute
6. Plunge pool
7. Dammed pool
8. Glide
Several of the Overton et al. (1990) habitat types are refinements of the Bisson
types, reflecting the channel feature associated with the type.
9. Backwater pool—boulder formed
10. Backwater pool—root wad formed
11. Backwater pool—log formed
12. Lateral scour pool—log formed
13. Lateral scour pool—root wad formed
14. Lateral scour pool—bedrock formed
15. Lateral scour pool—boulder formed
Overton et.al. (1990) identified these additional habitat types:
16. Run—Swiftly flowing reaches with little surface agitation and no major
flow obstructions. Often appears as flooded riffles. Typical substrates
are gravel, cobble, and boulders.
17. Step run—A sequence of runs separated by short riffle steps. Sub-
strates are usually cobble and boulders.
124
-------
18. Mid-Channel poof—Large pools formed by mid-channel scour. The
scour hole encompasses more than 60% of the wetted channel. Water
velocity is slow and the substrate is highly variable.
19. Edgewater— Quiet, shallow area found along the margins of the stream,
typically associated with riffles. Water velocity is low and sometimes
lacking. Substrate varies from cobbles to boulders,
20. Channel confluence pool—Large pools formed at the confluence of two
or more channels. Scour can be due to plunges, lateral obstructions or
downscour at the channel intersections. Velocity and turbulence are
usually greater than those in other pool types.
21. Pocket water—A: section of swift stream containing numerous boulders
and other large obstructions that create eddies or scour holes (pockets)
behind obstructions.
22. Corner pool—Lateral scour pools formed at a bend in the channel.
These pools are common in lowland valley bottoms where stream banks
consist of alluvium and lack hard obstructions.
This method uses both Bisson habitat typing and Hankin and Reeves estimation/
measurement correction factors. The primary difference between this method
and Hankin and Reeves is the spatial scale of interest. USFS R5/R1/R4 FHR
was designed primarily to distinguish microhabitat features and associated fish
usage. For instance, Overton found that salmonid fish species distribution was
distinctly different between the various pool types surveyed. He developed the
method originally for anodromous salmonid streams in northern California coastal
watersheds but has expanded its usage to high Sierra Nevada, Rocky Mountain,
and intermountain western watersheds containing both resident and anadromous
salmonids.
In assessing habitat for a stream reach or an entire basin, the intent is to gather
information that will adequately describe the area of interest. Conducting a
habitat inventory can be time-consuming, so work must be carried out quickly
and efficiently. The level or scale of inventory to be used depends on the project
objectives. The system is used at two scales: streams in a watershed or a
stream segment associated with a project. Basin^level habitat classification is on
the scale of a stream's naturally occurring pool-riffle-run units, where habitat unit
size depends on stream size and order. Generally in a basin-level inventory,
homogeneous areas of habitat that are as long as or longer than one channel
width are recognized as distinct habitat units. On the other hand, project-level
habitat assessment uses a scale of less than one channel width for use on
reaches of intense management or study. Project-level habitat typing is used to
quantify changes in habitat as the result of fish-habitat-enhancement projects or
to evaluate management impacts on habitat.
125
-------
Habitat typing can be done efficiently by two or three field people. However,
describing and measuring all 22 habitat types is labor intensive; an average of
one mile per day can be completed by trained surveyors. Decisions are best
reached by a consensus among the team members after a discussion of the
facts, thereby balancing out the biases inherent in each observer and ensuring
quality data.
The basic method of habitat typing is relatively simple. Start at the mouth of a
stream and work upstream to establish a known starting point. Measure mean
length and width of each unit using a tape, rod, optical rangefinder, or hip chain.
Three to five width measurements are sufficient. Along each width measurement
transect, use a graduated leveling rod (or similar device) to take several depth
measurements from bank to bank and estimate mean depth. If a significant
portion (>10%) of the measured habitat includes exposed boulders and/or is-
lands, that portion should be estimated and subtracted from calculations of area
(total area - exposed area = wetted area). Data on other variables, such as
stream substrate, in-stream cover elements and abundance, canopy cover,
riparian quality, etc., can be collected along with the habitat type data.
The habitat data can be used to compare reference and study streams as previ-
ously described in the Hankin/Reeves and Dolloff/Hankin/Reeves discussions.
Again, for this to be used as a monitoring tool, it is important to link habitat
changes with the specific effects of particular management activities.
Ohio-QHEl
The Ohio Environmental Protection Agency has developed a qualitative habitat
evaluation index (QHEI) to evaluate habitat quality for fish and macro-
invertebrates (Rankin 1989). Ohio-QHEl is an example of habitat monitoring
methods being developed by State water-quality agencies. It is being used at the
BMP Effectiveness III level but is also applicable to level IV.
Habitat measurements taken include: substrate type and quality, instream cover
type and amount, channel quality (sinuosity, development, channelization, and
stability), riparian/erosion (width, floodplain quality, and bank erosion), pool/riffle
(maximum depth, current available, pool morphology, riffle/run depth, riffle sub-
strate stability, and riffle embeddedness), and gradient. Each measurement is
scored and the sum of the scores is the habitat quality index. The QHEI has
been correlated with biological indexes.
Platts Riparian Habitat
Riparian habitat is closely linked to stream-channel habitat and directly influences
the biotic health of streams. Platts et al. (1987) developed a monitoring method-
ology for evaluating riparian habitat that is applicable to BMP Effectiveness III
and IV assessments. Vegetation in the riparian ecosystem, including vegetation
on the stream bank and flood plain, is evaluated because it helps stabilize stream
banks, control nutrient cycling, reduce water velocity, provide cover and food for
aquatic biota, and shade the stream.
126
-------
Riparian vegetation is characterized by type, density, use by animals, overhang-
ing vegetation, streamside stability, and cover. Riparian soils are classified
according to flood plain geomorphology, valley-bottom conformations, soil mor-
phology, stream bars, channel levees, adjacent wetlands, and riparian meadows.
Vegetative canopy closure and density are evaluated using a concave spherical
densiometer. Light intensity and stream-surface shading are estimated to gauge
the inputs of solar energy inputs into the stream.
Stream bank and channel aggradation, degradation, and morphology are mea-
sured using channel cross-sections. Banks, active channel, bank undercuts,
stream bottom, and bank heights are some of the features evaluated.
Organic debris is mapped and measured. Large woody debris is classified by
size:
•j pines—needles, twigs, and pieces < 5 cm average diameter
2. Coarse—branches, limbs and pieces 5-20 cm in diameter and up to 2.5 m
in length
3. Heavy—logs, trees, branches, stumps and pieces > 20 cm in diameter
4. Debris jams
Large woody debris (heavy) is further classified as complete bridge, collapsed
bridge, ramp, and drift.
Platts et al. (1987) discuss in detail monitoring and evaluating methods, statistical
design, and analysis of data.
Biomonitoring Methods
Biological monitoring methods are concerned with fish and benthic
macroinvertebrates. Aquatic life is continually exposed to influences on streams
and its abundance and diversity reflect these impacts.
This section acquaints the reader with several of the biomonitoring methods
available as an aid in selecting the method appropriate to his or her monitoring
effort. For more detailed information, the reader is referred to the original litera-
ture.
If the State agency has established biological monitoring methods, these should
be used to ensure comparable data. Many State water-quality agencies use or
modify Rapid Bioassessment Protocols (RBP) (Plafkin et al. 1989), while others
develop their own methods.
Many fish and invertebrate metrics are being used to evaluate water-quality
impacts (EPA 1993). As a help in preparing this document, two workshops were
held at which experts in fish and invertebrate metrics offered their best profes-
sional judgment of metrics to consider for small forest streams. The following
metrics for invertebrates were identified by both EPA (1993) and the workshops:
127
-------
EPT/Chironomid + EPT
EPT Index (number of EPT taxa)
Percent EPT taxa
Total Taxa Richness
Chironomid Richness
Hilsenhoff Biotic Index
Percent dominant taxa
Index of Community Integrity
Community Loss Index
Jaccard Coefficient of Community
Percent Shredders
Percent Filters
Percent Scrapers
Ratio Scrapers/Filter Collectors
Total Abundance
The reader is referred to EPA (1993) for information on these metrics. The
workshops also identified additional metrics judged as having potential for forest
streams:
Biotic Condition Index (Winget and Mangum 1979)
Biomass
Voltinism (Number of generations annually for a species)
Percent Tolerant taxa
Percent Intolerant taxa
For invertebrate monitoring, biological expertise should be consulted to identify
metrics to consider for a monitoring project. The key to invertebrate monitoring
and analyses is sampling procedures and identification of taxa. The metrics are
results of data analysis. The response of metrics to forestry activities should be
compared to reference conditions to evaluate effectiveness.
EPA (1993) presents a long list of fish metrics for large western rivers,
depauperated fish assemblages, and others to consider. Many of these fish and
invertebrate metrics are different than those used in the Rapid Bioassessment
Protocol discussed in this chapter. These metrics demonstrate the need to adapt
the Rapid Bioassessment Protocol or other biomonitoring methods to ecoregions.
Biological Biomonitoring Methods by Monitoring Level
Monitoring methods are listed below, followed by abstracts of the various meth-
ods recommended for each monitoring level. These methods represent a cross-
section of biological monitoring techniques being used.
128
-------
BMP Effectiveness I
Empirical/perceptual
Rapid Bioassessment Protocol I
(Macroinvertebrates)
Streamwalk (Macroinvertebrates)
Izaak Walton (Macroinvertebrates)
Hilsenhoff
Vermont Guide (Macroinvertebrates)
BMP Effectiveness II
Rapid Bioassessment Protocol II
(Macroinvertebrates)
Index of Biological Integrity (RBP V)(Fish)
Streamwalk II (Macroinvertebrates)
Snorkel Methods (Fish)
North Carolina Rapid Bioassessment
Method (Four Sample)
BMP Effectiveness III
Rapid Bioassessment Protocol III
(Macroinvertebrates)
Index of Biological Integrity (RBP V) (Fish)
Hankin-Reeves (Fish)
Basin Area Stream Survey
(Macroinvertebrates and Fish)
Ohio EPA (Macroinvertebrates and Fish)
Snorkel Methods (Fish)
NC Standardized Ten Sample Method
NC Rapid Bioassessment Method
(Four Sample)
Vermont Guide (Macroinvertebrates)
BMP Effectiveness IV
Rapid Bioassessment Protocol III
(Macroinvertebrates)
Index of Biological Integrity (RBP V) (Fish)
Ohio EPA (Macroinvertebrates and Fish)
Hankin-Reeves (Fish)
North Carolina Standardized
Ten Sample Method
Page
129
130
142
136
143
138
130
133
142
142
146
132
133
136
138
139/141
142
145
146
138
132
133
139/141
136
145
Empirical/Perceptual
At the BMP Effectiveness I level, a trained observer needs to know the benthic
invertebrate communities found in the reference streams. The observer also
needs to know which families of invertebrates are intolerant or tolerant to NPS
impacts from forestry as well as the functional feeding groups (shredders, scrap-
ers, etc.) and their habitat niches.
129
-------
Empirical/perceptual monitoring of invertebrates relies on visual observations.
The observer picks up rocks, leaves, and woody debris from key habitat in the
stream bed and determines which families of invertebrates are present. Compar-
ing these observations with knowledge of life forms in the reference streams, the
observer judges whether the study stream has been obviously impaired or not.
Indicators of impairment will be the absence of intolerant invertebrates, the
dominance of tolerant species, the relative abundance of invertebrates, and the
presence or absence of functional feeding groups.
Rapid Bioassessment Protocol I—Benthic
Macroinvertebrates (RBP I)
RBP I is a screening or reconnaissance assessment of benthic macroinverte-
brates that involves systematically documenting specific visual observations
made in the field by trained professionals (Plafkin et al. 1989). This system is
appropriate for the BMP Effectiveness I level. RBP I is used to identify obviously
impacted and non-impacted streams and those requiring further investigation—in
other words, the grey area between the obviously impaired and the obviously
non-impaired conditions. To assess whether impairment exists once a grey area
has been identified, RBP II or RBP III (discussed below) can be used to assess
the integrity of the benthic invertebrate community.
The biosurvey component of RBP I includes qualitative sampling of benthic
macroinvertebrates, supplemented by a preliminary field examination of other
biota (periphyton, macrophytes, slimes, and fish) (Plafkin et al. 1989). Samples
are collected from all available habitats using a dip net or kick net, or by hand.
To use this method, persons must be able to identify aquatic macroinvertebrate
taxa in the field. It can be difficult to identify organisms to the family level without
keys or the aid of a laboratory microscope. Benthic macroinvertebrate orders/
families collected are listed on a Biosurvey Field Date Sheet, with an estimate of
their relative abundance. Relative abundance of other aquatic biota is also
recorded, providing additional clues as to impairment.
Impairment may be indicated by the absence of generally pollution-sensitive
taxa, dominance of pollution-tolerant taxa, or low abundance and taxa richness.
Impairment may also be indicated by an overabundance of slimes or algae and
the absence of expected fish species. On the basis of this qualitative biosurvey
and observations made on habitat, water quality, and physical characteristics, the
investigator determines whether the stream is impaired (Plafkin et al. 1989).
Rapid Bioassessment Protocol II—Benthic
Macroinvertebrates (RBP II)
RBP II is a BMP Effectiveness II method that systematically collects and ana-
lyzes of major benthic taxa in the field, providing a more intensive assessment
than RBP I with relatively little additional time and effort (Plafkin et al. 1989).
RBP II can be used to select sites for more intensive evaluation (i.e., RBP III,
replicate sampling, etc.) or it can be used in lieu of RBP I as a screening tech-
nique. RBP II involves benthic analysis at the family level and uses field sorting
and identification.
130
-------
In the field, the RBP II biosurvey focuses on standardized sampling of benthic
macroinvertebrates, supplemented by a cursory field observation of other aquatic
biota (periphyton, macrophytes, slimes, and fish). Samples of the
macroinvertebrates from comparable habitat at all stations along the stream are
collected, plus separate Coarse Particulate Organic Matter (CPOM) samples.
RBP II concentrates on riffle/run habitat because it is the most productive habitat
available in streams and includes many pollution-sensitive taxa of the scraper
and filtering collector functional feeding groups. The CPOM sample provides a
measure of the effects of pollution on shredders.
Riffle areas with moderately fast currents and cobble and gravel substrates
generally have the most diverse community of benthic macroinvertebrates. The
recommended sampling method is to use a kick net to collect from an approxi-
mate 1 m2 area of riffle. Two 1 -m2 riffle samples are collected at each station:
one from an area of fast current and one from an area with slower current. The
two samples are combined for processing. In streams without riffles, run areas
with cobble and gravel substrate can be kick-net sampled. But where riffle/run
cobble and gravel habitat does not exist, other types of habitat (submerged
boulders, logs, bridge abutments) should be sampled for biota by hand picking.
Such habitats are used by scrapers and filtering collectors (Plafkin et at. 1989).
Submerged logs and root wads are important habitat in sand-bottomed Coastal
Plain streams.
The CPOM is habitat for shredders and is sampled to reveal the abundance of
shredders, important in forested stream ecosystems ranging from 1st- to 4th-
order streams (Minshall et al. 1985). The absence of shredders is characteristic
. of unstable headwater streams in disturbed watersheds, which to not retain much
CPOM.
The CPOM sample is processed separately from the riffle/run sample and the
functional feeding groups are identified (Plafkin et al, 1989). Potential sample
sources include leaf packs, shorelines, and other depositional areas where
CPOM accumulates. A variety of CPOM types should be collected; samples may
be washed in a dip net or a sieve bucket.
Shredders are most abundant when the CPOM is about 50 percent decomposed
(Cummins et al. 1989). Avoid collecting fresh or fully decomposed leaf litter to
optimize collection of the shredder community. The type of leaf litter will affect
the seasonality of shredder abundance.
It is recommended that the riffle/run sample be sorted and enumerated in the
field to obtain a 100-count benthic subsample. All invertebrates in the subsample
are classified according to functional feeding group. The scraper and filter-
collector functional groups are the key indicators in the riffle/run community.
Numbers of individuals in these two groups are recorded. All organisms in the
subsample are identified as to family or order, enumerated," and recorded along
with any observations on abundance of other aquatic biota.
131
-------
Organisms collected in the CPOM sample are classified as shredders or
nonshredders. As with the RBP I method, ability to identify aquatic
macroinyertebrates to order and/or family is needed to classify organisms into
functional feeding groups in the field. Numbers of individuals representing the
Shredder Functional Group, as well as the total number of macroinvertebrates
collected in the sample, are recorded.
Eight metrics are used to characterize the biological condition of the study stream
versus the reference stream:
1. Taxa Richness
2. Family Biotic Index (pollution tolerance index for families)
3. Ratio of Scrapers/Filter Collectors
4. Ratio of EPT and Chironomid Abundance
5. Percent Contribution of Dominant Family
6. EPT Index
7. Community Loss Index
8. Ratio of Shredders/Total
Each metric is scored based on its percent comparability to a reference stream.
Scores are totaled and compared to the total score for the reference stream,
providing an evaluation of the biological condition of the stuay stream.
Caution: Some metrics may be applicable to specific ecoregions and others
may not. Metrics 3 and 4 are currently questionable (Karr, personal com-
munication).
Rapid Bioassessment Protocol III—Benthic
Macroinvertebrates (RBP III)
RBP III is a more rigorous bioassessment technique than RBP II, involving
systematic field collection and subsequent lab analysis, allowing detection of
more subtle degrees of impairment (Plafkin et al. 1989). RBP III can be used in
BMP Effectiveness III and IV assessments. Invertebrate samples are analyzed
in more detail for RBP III, which requires taxonomic identification to genera.
Samples must be processed at laboratories specifically designed for such pur-
poses. Use of these lab services is cost-efficient and also ensures adequate
quality control. In addition, this protocol provides a basis for monitoring trends or
cumulative impacts over time.
Sample collection for RBP III is essentially the same as for RBP II. The riffle/run
habitat is the site for sampling scrapers and filtering collectors, supplemented by
CPOM samples to collect shredders. Again, submerged boulders, logs, bridge
abutments, and pilings are hand picked if riffle/run habitat is absent.
Subsamples of 100 are usually sufficient, but subsarnple sizes can be 200-300
specimens. All benthic macroinvertebrates in the subsarnple should be identified
to genus or species, if possible, and then classified by functional feeding groups.
132
-------
The data are analyzed is performed as follows. Using the raw benthic data, a
numerical value is calculated for each metric. Calculated values are then com-
pared to values from reference streams or segments on the same stream. Each
metric is then assigned a score according to the comparability (percent similarity)
of calculated and reference values. Scores for the eight metrics are totaled and
compared to the total score for the reference stream. The percent comparison
between the total scores provides a final evaluation of biological condition.
The eight metrics are the same as in RBP II.
The study stream can now be classified into one of four biological conditions:
1. Non-Impaired—Comparable to the best situation to be expected within an
ecoregion. Balanced trophic structure. Optimum community structure
(composition and dominance) for stream size and habitat quality.
2. Slightly Impaired—Community structure less than expected. Composition
(species richness) lower than expected due to loss of some intolerant
forms. Percent contribution of tolerant forms increases.
3. Moderately Impaired—Fewer species due to loss of most intolerant forms.
Reduction in EPT index.
4. Severely Impaired—Few species present. Organisms present are domi-
nated by one or two taxa. Only tolerant organisms present.
Index of Biotic Integrity—Fish (Rapid Bioassessment Protocol V)
The Index of Biotic Integrity (IBI), a fish-community-assessment approach, was
developed by Karr (1981). Substantial additional developmental work and
regional testing is behind IBI (Fausch et al. 1990, Karr 1991 and 1993, Karr et al.
1986, Lyons 1992, and Miller et al. 1988). It is a rigorous system that should be
used primarily at the BMP Effectiveness III and IV levels, but can be adapted to
BMP Effectiveness II. .
IBI involves careful, standardized field collection, species identification and
enumeration, and community analyses using biological indices or quantification
of the biomass, and numbers of key species. An IBI survey yields an objective,
discrete measure of the health of the fish community that usually can be com-
pleted on site by qualified fish biologists. Data collected can serve to assess
whether the beneficial use for the stream is attained, develop biological criteria,
prioritize sites for further evaluation, provide a reproducible impact assessment,
and assess fish community status and trends. IBI is the system used in Rapid
Bioassessment Protocol V (Plafkin et al. 1989).
RBP V involves field evaluation of the same physical, chemical, and habitat
characteristics as RBP's I, II, and III, a similar impairment assessment, and a fish
community biosurvey. Because they provide critical information for evaluating
the cause and source of impairment, these data are essential to RBP V.
133
-------
Electrofishing, the most widely applicable system for sampling stream habitats, is
recommended for use with IBI. The biosurvey data are intended to be represen-
tative of the fish community at all station habitats. Each sampling station should
be representative of the reach, incorporating at least one (preferably two) riffle(s),
run(s), and pool(s) if these habitats are typical of the stream in question. Sam-
pling of most species is best done near shore and cover (macrophytes, boulders,
snags, brush). The biosurvey is not an exhaustive inventory, but provides a
realistic sample of fishes likely to be encountered in the stream.
Typical sampling stations are 100-200 meters long for small streams and 500-
1,000 meters for rivers, but lengths are best determined by pilot studies. The
reference station should be long enough to produce 100-1,000 individuals and
80-90 percent of the species expected if the station length is increased by 50
percent. Block nets set at both ends of the reach increase sampling efficiency for
large, mobile species sampled in small streams. The RBP V fish community
assessment requires that all fish species (not just gamefish) be collected.
A field collection data sheet is completed for each sample. Sample duration and
area are recorded in order to determine level of effort. Species may be sepa-
rated into adults and juveniles by size and color; then total numbers and weights
and the incidence of external anomalies, such as tumors, deformities, and dam-
aged fins, are recorded for each group.
The steps in determining the IBI are:
1. Select a site.
2. Identify regional fish fauna.
3. Assign species to trophic, tolerance, and origin guilds.
4. Assess available data for metric suitability and stream-size patterns.
5. Develop scoring criteria from reference streams.
6. Quantitatively sample fish.
7. Record abundance of species, hybrids, and anomalies.
8. Calculate and score metric values (Table 10).
9. Add scores to get IBI score.
10. Make recommendations (Table 11).
134
-------
Table 10. Metric Scores (IBI) (Karr 1981)
Scoring Criteria
Metric
1 . Number of native fish species
2. Number of darter or benthic species
3. Number of sunfish or pool species
4. Number of sucker or long-lived species
5. Number of intolerant species
6. Proportion of green sunfish or tolerant
individuals
7. Proportion omnivorous individuals
8. Proportion insectivorous
9. Proportion top carnivores
10. Total number of individuals
1 1 . Proportion hybrids or exotics
12. Proportion with disease/anomalies
>67%
>67%
>67%
>67%
>67%
<10%
<20%
>45%
>5%
>67%
0%
<1%
33-67%
33-67%
33-67%
33-67%
33-67%
10-25%
20-45%
20-45%
1-5%
33-67%
0-1%
1-5%
<33%
<33%
<33%
<33%
<33%
>25%
>45%
<20%
<1%
<33%
>1%
>5%
Note: Metrics 1 -5 are scored relative to the maximum species richness line.
Metric 10 is drawn from reference site data.
Table 11. Index Score Interpretation (Karr 1981)
IBI
Integrity Class Characteristics
58-60
48-52
40-44
28-34
12-22
Excellent
Good
Fair
Poor
Very Poor
Comparable to pristine conditions,
exceptional assemblage of species.
Decreased species richness, intolerant
species in particular decreased; sensitive
species present.
Intolerant and sensitive species
absent, skewed trophic structure.
Top carnivores and many expected
species absent or rare; omnivores
and tolerant species dominant.
Few species and individuals present;
tolerant species dominant;
diseased fish frequent.
Metrics will vary by ecoregion and fisheries. The metrics and key habitat
elements need to be verified or validated by ecoregion. The above metrics
were developed for midwestern streams and may not be applicable to other
ecoregions.
135
-------
Hankin-Reeves—Fish Population Estimation and Community
Composition
The Hankin and Reeves (1988) method can be used at'the BMP III and IV levels.
Visual fish surveys rely on the same premise as visual habitat surveys: if divers
count a consistent fraction of the fish present, then there will be a strong correla-
tion between visual estimates and the "true" number of fish. Divers observe the
species, number, and usually the size classes of fish in sample habitat units from
an entire reach. Like the habitat survey, wherein visual estimates of surface area
of habitat units are calibrated by periodic measurements, visual estimates of fish
populations are calibrated, but by a more accurate method (such as multi-pass
depletion electrofishing) in a fraction of the sampled habitat units.
If a Hankin and Reeves habitat survey was done earlier, that information should
be used. If not, the team selects a classification system for identifying habitat
types (e.g., Bisson et al. 1981) and stratifies the study area into reaches based
on gradient, confluence of same-order channels, or other distinctive feature. The
team then decides how many units of etch habitat type that will be sampled.
The fraction of units sampled need not be the same for every habitat type and
should depend on the objectives of the survey. As a general rule, a higher
percentage of units should be sampled in habitat types that are preferred by the
species of interest.' For example, if the species of interest prefers pools, the
surveyors might sample 25 percent of the pools and only 10 percent of the riffles
and cascades in a study area. However, no habitat type should be eliminated
from sampling just because surveyors do not expect fish to be present. There is
only one way to confirm the presence or absence of fish in a habitat type:
sample it
The next step is to select the percentage of units to be sampled for calibration of
diver counts. As a "rule of thumb," at least 10 percent of the units sampled by
divers should be sampled by a more accurate method of counting fish (typically
multiple-pass depletion electrofishing). The surveyors randomly select the
starting unit and interval between units designated for diver calibration.
Population estimates and confidence intervals for species and age-classes in a
habitat type are calculated according to Hankin and Reeves (1988) (2 divers
count fish in each habitat unit) or Dolloff et al. (1993) (single diver counts fish in
each habitat unit).
Using the Hankin and Reeves method on both reference and study streams will
allow comparing fish abundance and functional feeding groups between streams.
This will reveal major differences in total fish abundance and community compo-
sition.
Izaak Walton League—Benthic Macroinvertebrates
The Izaak Walton League (1990) method is a qualitative assessment of benthic
invertebrates, coupled with empirical observations of water quality, habitat, fish
and land use. This methodology is best used for evaluating invertebrates at the
BMP Effectiveness I level. Invertebrates are sampled in three riffles at depths of
3-6 inches and in a 3-foot-square area with a kick net. The invertebrates are
136
-------
washed into a white pan for identification to order and suborder. Then they are
classified as pollution sensitive, somewhat tolerant, or tolerant, using a simple
chart with taxonomic drawings of invertebrates that are keyed to pollution
tolerance.
The numbers of individuals in each order found in the 3-foot square are reduced
to a code: A = 1-9, B = 10-99, and C = 100+ . The appropriate letter is assigned
to each order and suborder, and recorded on the form below.
MACROINVERTEBRATE COUNT
Water Quality Indicator
Good
Fair
Poor
Caddisfly
Dobsonfly
Mayfly
Stonefly
Other snails
Riffle beetles
Water penny
# of letters
times 3 =
Index value+
Beetle larvae
Clams
Crane fly
Crayfish
Damselfly
Dragonfly
Scuds
Sowbugs
Antherix
# of letters
times 2 =
Index value +
Aquatic worms
Blackflv
Leeches
Mido.es
Pouch snails
# of letters
times 1 =
Index value
Then the number of letters in each column is multiplied by the indicated index.
The three index values are added to get a total index value, which is then classi-
fied as: Excellent (> 22), Good (17-22), Fair (11 -16), or Poor (< 11). This rating
of excellent, good, fair, or poor is a qualitative assessment of water quality as
reflected in the benthic community. This method is usually used to monitor the
condition or trends in one stream, or above and below a management area.
To evaluate the effectiveness of BMP's, study streams must be compared to
reference streams. If the ecoregion has been impacted by prior land use, the
reference stream may be assessed as only fair. Without the reference assess-
ment, BMP's near a study stream rated as fair might be erroneously judged as
ineffective when indeed they could be effective.
Because invertebrate numbers are lumped into only three classes, study and
reference streams may receive the same qualitative rating, when in fact there
may be significant differences in the number and taxa of invertebrates. So the
simple rating of excellent, good, fair, or poor can be misleading as can comparing
total index values between study and reference streams. However, by compar-
ing the occurrence of orders of invertebrates and of numbers of individuals
between study and reference stream, qualitative assessments of BMP effective-
ness can be made with the Izaak Walton League method.
137
-------
Vermont Guide to Macroinvertebrate Sampling
The River Watch Network in Vermont (1992) has developed a guide for
macroinvertebrate sampling which was adapted from Rapid Bioassessment
Protocols (Plafkin 1989) for local conditions. It uses two levels of data analysis.
Level 1 classifies invertebrates into major groups (mostly orders) to reveal the
basic composition of the macroinvertebrate community as well as serious
changes resulting from changes in the stream ecosystem. The metrics used in
this analysis are taxa richness, organism density per sample, EPT richness
(orders), and EPT-to-Chironomidae ratio. This level of analysis is appropriate for
BMP Effectiveness I and II.
Level 2 identifies macroinvertebrates to the family level, refines one data analysis
procedure in Level 1, and adds two others, all of which reveal more subtle differ-
ences among sites. The refined analysis is EPT richness at the family level, and
the added analyses are the modified family biotic index (based upon Hilsenhoff
tolerance indices to pollution) and Pinkham and Pearson community similarity
index. This is a BMP Effectiveness III method.
The stream conditions measured or estimated are: water temperature, air tem-
perature, average width, average depth, relative flow compared to average
annual flow, sediment deposits, smell of water, water color, algal growth,
channelization, upstream dam, point discharges, stream bank characteristics,
substrate composition, embeddedness, and overhead canopy.
Basin Area Stream Survey (BASS)
Clingenpeel and Cochran (1992) describe a process for evaluating the physical,
chemical, and biological indicators of water quality on the Ouachita National
Forest in Arkansas called the BASS methods. BASS elaborates on methods
developed by Pfankuch (1975), Hankin (1984), Bisson et al. (1982), Ebert et al.
(1987 and 1991), and Filipek et al. (forthcoming). It includes habitat typing and
geographical-information-system mapping, evaluating habitat and water quality,
and sampling fish and benthic invertebrates. BASS uses reference and study
streams within ecoregions. It is a BMP Effectiveness III method.
The biological inventory is based upon a 10-percent sample of habitat types
along study and reference streams. For example, if 27 main channel pools are
identified within a stream, three of these pools are sampled.
Fish populations in each reach are isolated with block nets before sampling by
electrofishing equipment. Fish in the sample area are collected using at least
two but preferably three or more passes for depletion sampling. With each pass,
shocked fish are collected, reducing the number of fish remaining in the stream.
Depletion sampling will collect most of the fish in the stream segment.
Electrofishing passes cover the entire reach from the downstream net up, with
equal effort applied on each pass. The downstream net is checked for fish after
each pass and captured fish are included with the sample for that pass. Each
pass comprises a sample and the fish are placed in separate containers. Game,
endangered, threatened, and sensitive species are identified, measured, and
138
-------
weighed in the field and returned to the stream. The remaining fish are pre-
served in 10-percent formalin and labeled to allow follow-up analysis, if needed.
The preserved fish are identified, sorted, and measured in a laboratory. Popula-
tion of each species is estimated using the maximum-likelihood method of
VanDeventer and Platts (1990).
Macroinvertebrate samples are collected in the same reaches as sampled for
fish, using a 5-minute kick-net sample. Collecting is done by shuffling or kicking
the substrate directly above the kick net. All microhabitats (woody debris, leaf
packs, etc.) within the reach are sampled. In addition, a 5-minute wash sample
of the substrate is taken. A dip net is placed in the stream and individual cobbles
above it are scrubbed with a soft-bristle brush into the net. The wash sample is
combined with the kick-net sample. Large organic matter and leaves are washed
and removed from the sample. The invertebrate samples are preserved in
70-percent ethanol and labeled. Identification and sorting are done in a labora-
tory.
Ohio EPA Macroinvertebrate Sampling
As an example of BMP Effectiveness III and IV methods being developed by
State water-quality agencies, Ohio EPA (1987 and 1989) developed a method for
assessing benthic macroinvertebrates using artificial substrates supplemented
with a qualitative sample of the natural substrate. The primary sampling equip-
ment used for collecting benthic macroinvertebrates is the modified Hester-
Dendy, multiple-plate, artificial-substrate sampler, which is placed in the stream
for a period of time to allow invertebrates to colonize it. The sampler is con-
structed of 1/8-inch tempered hardboard cut into 3-inch-square plates and 1-inch
square spacers, which are held together at the center by a 1/4-inch eye-bolt.
Eight plates and twelve spacers are used for each sampler. The plates and
spacers are arranged on the eyebolt so that there are three single spaces, three
double spaces and one triple space between plates.
Before the samplers are placed in the stream, they are tied to a concrete block,
which anchors them in place and prevents the plates from touching the natural
substrate. In water deeper than 4 feet, a float is attached to the samplers to keep
them within 4 feet of the surface. Whenever possible, the samplers are placed in
runs rather than pools or riffles and stations are established in as similar an
ecological situation as possible. The samplers are exposed for 6 weeks.
The samplers are retrieved by cutting them from the block and placing them in
1-quart plastic containers while still submerged. Care is taken to avoid disturbing
the samplers and thereby dislodging any organisms. Enough formalin is added
to each container to approximate a 10-percent solution.
Qualitative samples of macroinvertebrates inhabiting natural substrates are
collected at the time of sampler retrieval. In shallow water, samples are taken in
a stream segment covering all available habitats in the vicinity of the samplers.
Samples are collected using triangular-ring-frame, 30-mesh dip nets and hand
picking with forceps. The sampling continues until, by visual examination, no
new taxa are being taken.
139
-------
Where quantitative biological samples are collected from the natural substrate,
the collector stands on the downstream side of a Surber square-foot sampler
(30-mesh netting) sampler and works the substrate using a 2-tined hand cultiva-
tor. Large rocks are gently scrubbed with a brush. The material collected is
placed in sealed containers, preserved in 10-percent formalin, and transported to
the laboratory. Three to five Surber samples are taken at each site.
Laboratory methods for quantitative sampling include taking composite samples
with five multi-plate samplers to be used in station evaluations for routine moni-
toring. For litigation and other needs, replicate sampling and larger composite
samples can be used. The Ohio EPA has established a laboratory protocol for
identifying and enumerating invertebrates, generally to the genus, or species
level. For the qualitative sample, Ephemeroptera, Plecoptera, and Trichoptera
are enumerated.
The principal measure of overall macroinvertebrate community condition is the
Invertebrate Community Index (ICI). The ICI is a modification of the Index of
Biotic Integrity (IBI) for fish developed by Karr (1981). It consists of ten structural
community metrics, each with four scoring categories of 6, 4, 2, and 0 points
(Table 12).
The point system evaluates a sample against a database of 247 relatively undis-
turbed reference sites throughout Ohio. Six points are given if a metric has a
value comparable to exceptional stream communities, 4 points for those charac-
teristic of more typical good communities, 2 points for metric values slightly
deviating from the expected range of good values, and 0 points for metric values
strongly deviating from the expected range of good values. A community-similar-
ity index and rank correlation are used to test similarity between study streams
and reference streams.
Table 12. Invertebrate Community Index (ICI) metrics and scoring criteria based on macroinvertebrate
community data from 247 reference sites throughout Ohio
Metric
Scoring Criteria
0246
1. Number of Taxa
2. Total Number of Mayfly Taxa
3. Total Number of Caddisfly Taxa
4. Total Number of Diptera Taxa
5. Percent Mayflies
6. Percent Caddisflies
7. Percent Tribe Tanytarsini Midges
8. Percent other Dipterans and
Noninsects
9. Percent Tolerant Organisms
10. Total Number of Qualitative
Ephemeroptera, Plecoptera, and
Trichoptera (EPT) Taxa
Scoring of each metric
varies with drainage
area (Ohio 1987).
140
-------
Ohio EPA Fish Sampling
The Ohio EPA samples fish using several methods to calculate the Index of
Biotic Integrity (IBI) and the Modified Index of Wellness (IBW) from which aquatic
life use attainment is partially judged. This is an example of a BMP Effectiveness
III method developed by State water-quality agencies for fish.
Fish sampling sites are selected according to several factors including, but not
limited to, the following:
1. location of point source dischargers
2. stream use designation evaluation
3. location of physical habitat features (e.g., dams, changes in geology,
changes in stream order, presence of a stream confluence, etc.)
4. location of nonpoint sources of pollution
5. variations in macrohabitat
6. proximity to ecoregion boundaries
Each study area is allocated a number of biological sampling sites allocated
based on the number and complexity of the priority issues requiring field evalua-
tion. The principle objectives of each survey determine where sampling sites will
be located. Placement of sampling sites is influenced by practical access and
resource constraints. Generally, sites are located upstream from all pollution
sources to determine the background condition for the study area. Should the
upstream portion of the stream be impacted, an alternate site may be chosen on
an adjacent stream with similar watershed characteristics. Reference sites within
the same ecoregion may also be used. The role of upstream sites is not neces-
sarily to provide a biological performance level against which downstream sites
are compared since the ecoregion biocriteria serve this purpose for aquatic life
use designations. Upstream sites are, however, important in defining any site or
watershed specific background conditions that might temporarily or permanently
influence eventual aquatic life use attainment in the downstream reaches. The
most typical habitat available is selected for sampling within a segment in an
effort to represent the current potential of that segment. An attempt should be
made to sample similar macrohabitats at all sampling sites within the study area.
The Ohio EPA uses mainly pulsed, direct-current electrofishing to measure fish
abundance and distribution. Boat mounted electrofishers are used in large
streams and rivers; backpack electrofishers are used in wadable streams.
Fish are best sampled between mid-June and early October, when stream and
river flows are generally low, pollution stresses are potentially the greatest, and
the fish community is most stable and sedentary. Since the ability of the netter to
see stunned fish is critical, sampling should take place only during periods of
"normal" water clarity and flow. "Normal" water conditions in Ohio occur during
below annual average river discharge levels when the surface of the water
generally has a "placid" appearance. Abnormal turbidity is to be avoided as are
high flow and rapid current. All these reduce sampling efficiency and may dis-
qualify data for calculating the modified IWB and IBI.
141
-------
Captured fish are immediately placed in an on-board livewell for later processing
when sampling each site is complete or when the livewell is full. Water in the
livewell should be changed regularly to minimize mortality of captured fish. Fish
are released immediately after they are identified as to species, examined for
external anomalies, and, if necessary, weighed. Most captured fish can be
identified to species in the field; however, if there is any uncertainty, the fish must
be preserved for later laboratory identification. The data are used to develop IBI
and IWB scores for comparing study streams with reference conditions.
Snorkel Methods for Fish
Thurow (1991) describes a snorkeling method, suited for BMP Effectiveness II
and III assessments, to visually determine fish species, age, and size. The
method is described for evaluating salmonids. Observations should be made
after the stream temperature exceeds 9 °C. Divers must be trained to judge
species, age, and size of fish. Snorkelers enter the water downstream of the
habitat to be surveyed and proceed slowly upstream. Data are recorded on a
PVC sleeve or slate with a lead pencil.
Streamwalk
The USEPA Region 10 has begun a citizens monitoring program called
Streamwalk. Used by a professional, it can yield meaningful results at the BMP
Effectiveness I level. The empirical observations include stream characteristics,
stream-bottom type, width of streamside corridor, streamside vegetation, canopy
cover, presence of large woody debris, adjacent land use, stream-bank condi-
tions, and others. Streamwalk is used to identify potential problem areas and
monitor trends.
Streamwalk II
In Idaho, Streamwalk has been expanded to be more rigorous and more de-
manding of participants, and thus has become Streamwalk II (Rabe 1992).
Streamwalk II assesses habitat, chemical and physical conditions, and biology.
Conditions in study streams are compared to reference-stream conditions. This
method is applicable to the BMP Effectiveness II level of monitoring.
The habitat assessment procedure was adapted from the Rapid Bioassessment
Protocols (Plafkin 1989) and information provided by the State DEQ and univer-
sity personnel from Idaho. Variables are weighted to emphasize biological
criteria. After the study reach has been surveyed, ratings are totalled and com-
pared to a reference. The study reach score is divided by the reference score to
calculate a percent comparability score for each variable and total score. Refer-
ence streams represent the best habitat conditions attainable for a region.
The primary habitat variables (bottom substrate, embeddedness, and velocity/
depth) are evaluated in a riffle-pool sequence and scored from 0-20 points.
Secondary variables consist of wetted channel shape, pool/riffle ratio, and chan-
nel alteration (0-15 points). Tertiary variables include lower-bank stability, bank
vegetation protection, canopy cover, and width of riparian zone (0-10 points).
142
-------
The chemical and physical quality of water is analyzed and a Water Quality Index
determined. Ten variables are assessed: dissolved oxygen, pH, alkalinity, and
conductivity are analyzed in the field; nitrates, phosphates, total solids, turbidity,
fecal coliforms, biochemical oxygen demand, and pH are analyzed in a lab.
Biomonitoring includes systematic sampling of,habitat types with a variety of
sampling gear. Comparisons are made within and between study and reference
streams to detect the presence of macroinvertebrates that are sensitive or toler-
ant to water-quality impacts and to determine taxa richness. As many taxa as
possible are collected in a reasonable period of time. The habitat types sampled
and the associated organisms are recorded. Habitat types include riffles, pools,
aquatic plants, coarse particulate material, and soft sediments.
The invertebrate samples are processed and the following biological metrics are
developed:
1. Species richness—the total number of different taxa present in the
sample.
2. EPT index—the total number of Ephemeroptera (mayflies), Plecoptera
(stoneflies), and Trichoptera (caddisworms) found in the sample. Insects
comprising these three orders are generally sensitive to pollution. A high
EPT index is associated with clean water and unimpacted habitat.
3. Biotic index—a value given to an organism depending upon its tolerance
to pollution. Low values indicate organisms sensitive to pollution. High
values indicate organisms more tolerant and less sensitive to stress. The
Hilsenoff index is used.
4. Dominance—the relative number of individuals belonging to one or two
species in the community. The dominance of one or two species usually
relates to an impacted habitat or poor water quality.
5. Functional feeding groups—functions or activities are associated with
different macroinvertebrates in a stream. If several different types are
present, then this condition reflects a healthy ecosystem as compared to
a stream with only a few functional types.
6. Index of similarity—the similarity of taxa between study and reference
streams.
Hilsenhoff Biotic Index
Water-quality condition can be determined simply using benthic macroinverte-
brate sampling and the Hilsenhoff Biotic Index (1987). The following was devel-
oped in Idaho and represents a BMP Effectiveness I technique.
The relative health of a stream can be estimated by collecting a
macroinvertebrate sample from a stream riffle section, identifying the basic
composition of invertebrate taxa in the sample, and averaging Hilsenhoff
tolerance values over all specimens examined. The procedure is as follows:
1. Obtain a 3' x 3' Section of fine mesh screen material and attach two
2.5' lengths of 1" diameter wood dowels to opposite edges of the
screen to make the kick net.
143
-------
Table 14. Methods used.for the standardized qualitative collection techniques (Lenat 1988)
Sampler
Coarse-mesh"
Kick net (high current
with structure13)
Dip net (low current
with structure)
Wash bucket (leaves)
Fine-meshc
Plastic basin + PCV
cylinder (aufwuchs)
Nitex bag (sand)
Visual collections
Habitat
Usually riffles, also
snags in sandy streams
Usually banks
Leaf packs
Rocks, logs
Sand
Large rocks & logs
Number of
Samples
2
3
1
2
1
1
Type
Single,
disturbance
Composite,
disturbance
Composite,
wash
Composite,
wash
Composite (3),
disturbance
Composite
•0.5-1.0 mm mesh.
b Substrate (rocks, logs, roots) that offers protection from scour or place of attachment.
c 0.3 mm Nitex mesh.
North Carolina Rapid Bioassessment Method (Four Sample)
North Carolina developed a four-sample technique as a rapid bioassessment
method (Lenat 1988) to satisfy a need for a more rapid survey technique than
their standard qualitative collection method (10-sample). This techniques is
applicable to BMP Effectiveness II and III levels. A modification of their 10-
sample technique, the four samples consist of: a kick-net collection, a sweep-net
collection (bank area), a leaf pack, and picking insects from large rocks and logs.
This method focuses on EPT taxa richness. Eaton and Lenat (1991) report a
strong correlation between EPT taxa per the rapid method and the standard
method, and that the rapid EPT method does nearly as good a job of predicting
water quality as standard qualitative methods. Sampling and processing take
half as long as standard qualitative methods.
They caution using the rapid method if EPT taxa richness is low, such as in very
small streams, some swampy coastal areas, and areas where the "control" site is
already severely polluted (Eaton and Lenat 1991). In these situations, the
10-sample method should be used. They suggest that EPT collections be
supplemented by notes on the abundance of obvious pollution-indicator groups.
146
-------
Aquatic Vegetation and Zooplankton Monitoring Methods
Aquatic vegetation and zooplankton are important elements of stream ecosys-
tems and can be used as indicators of stream health, EPA's draft guidance
"Biological Criteria: Technical Guidance for Streams" (Barbour et al. 1991)
discusses periphyton, macrophytes, and zooplankton as possible biological
criteria for inclusion in State water-quality standards. The following is quoted
from their draft guidance (for references cited in this excerpt, see Barbour et al.
1991).
Periphyton: The broad organismal assemblage known as periphyton is
composed ofbenthic algae, bacteria, their secretions, associated detritus and
various species of microinvertebrates (Laberti and Moore 1984). Periphyton
is an important energy base for many lotic situations (Steinman and Parker
1990; Minshall 1978; Dudley et al. 1986) serving as the primary nutrient
source for many stream organisms (Lambert! and Moore 1984). The capacity
of these benthic assemblages to colonize and increase in biomass is influ-
enced by variability in stream channel geomorphology, flow rates, herbivore
grazing pressure, light intensity, seasonality, and random processes
(Stevenson 1990; Grimm and Fisher 1989; Coleman andDahm 1990; Poffet
al. 1990; Korte and Blinn 1983; Hamilton and Duthie 1984; Lambent et al.
1987; Steinman and Mclntire 1986, 1987; Steinman etal. 1987). The dy-
namic position of the periphyton assemblage within most stream ecosystems
makes it a prime for consideration as a bioassessment/biosurvey target.
More specific advantages given by Plafkin etal. (1989) are that:
• rapid algal reproduction rates and short life cycles make them valuable
as indicators of short-term impacts;
• as the periphyton assemblage is a function of primary production,
physical and chemical factors have direct effect;
• sampling methods are straightforward and sampling effort is easily
quantified and standardized;
• methods have been relatively standardized for functional and
nontaxonomic characteristics of periphyton communities such as biom-
ass and chlorophyll measurements; and
• algal components of periphyton are sensitive to some pollutants to
which other organisms may be relatively tolerant.
Macrophytes: Both emergent and submergent vegetation provide numerous
features to lotic situations which enable them to support healthy, dynamic,
biological communities ( Miller et al. 1989; Campbell and Clark 1983; Hurley
1990). Spatio-temporal distributional characteristics and some understanding
of the effecting environmental conditions of macrophytes (Hynes 1970) allow
their use in bioassessment strategies. Hynes (1970) and Westlake (1975)
discuss differences in lotic macrophyte assemblages based on habitat factors
such as water hardness, pH, gradient, and propensity for siltation. Other
workers have emphasized the capacity of macrophytes to influence habitat
structure (Miller et al. 1989; Gregg and Rose 1982; McDermid and Naiman
1983; Pandit 1984; Carpenter and Lodge 1986), water chemistry, nutrient
147
-------
cycling, and macroinvertebrate colonization (Miller et al. 1989; McDermidand
Naiman 1983). Seddon (1972), Westlake (1975) and Pandit (1984) point out
the utility of macrophytes as an indicator assemblage in lotic situations.
Aquatic macrophytes are important as a food source for birds and mammals.
Fassett (1957) lists 36 species of waterfowl, nine marshbirds, four shorebirds
and nine upland game birds feeding on these plants. He also lists beaver,
deer, moose, muskrat, and porcupines as aquatic macrophyte herbivores.
From these and other considerations, a number of advantages are apparent
for use of macrophytes in bioassessment programs:
• taxonomy to the generic level is relatively straightforward;
• establishment of populations in a specific habitat may be somewhat
dependent on local environmental conditions; thus, there is potential as
site-specific indicators;
• specific microhabitat structure may not limit germination; thus, there is
potential for high population density;
• growth patterns of individual plants are directly influenced by herbivore
activity;
'• longevity, distribution, and rate of population growth may directly reflect
prevailing conditions.
Zooplankton: These organisms are generally considered the second link
(primary consumers) in aquatic food chains, although certain common lim-
netic taxa such as Cyclops species prey upon zooplankton and are regarded
as secondary consumers. Zooplankton that are primary consumers feed on
phytoplankton, bacteria, protozoa, and organic detritus and are in turn utilized
by aquatic macroinvertebrates and fish (secondary and tertiary consumers).
Many fish species at some stage of development use zooplankton as a food
source. Availability of sufficient quantities of zooplankton at critical periods
may affect survival of fry and fingerling stages of fish species.
Alterations of the environment, such as increases in water temperature,
turbidity, or changes in the water circulation pattern could influence densities
and species composition of indigenous zooplankton populations, thereby
affecting zooplankton-dependent organisms at higher trophic levels.
Advantages in use of zooplankton in biocriteria are the following:
• high diversity,
• can attain high abundance levels in low gradient river systems,
• likely to encompass broad range of sensitivities,
• brief life cycles compared to higher trophic levels, and thus are respon-
sive to short-term perturbations,
• field sampling methods are cost-effective and straight forward.
Periphyton are algae, fungi, bacteria, protozoa, and associated organic
matter attached to channel substrates (Klemm and Lazorchak, 1993). Per-
iphyton are useful indicators of environmental conditions because they
respond rapidly, are non-mobile, and are sensitive to a number ofanthropor
genie disturbances, including contamination by nutrients, metals, herbicides,
148
-------
hydrocarbons, and acids. Periphyton represent another trophic level, exhibit
a different range of sensitivities and will often indicate effects only indirectly
observed in the benthic macroinvertebrates and fish communities (Plafkin
etal. 1989, Hayslip 1993). Hayslip (1993) recommends considering the
following periphyton analyses:
1. Chlorophyll a
2. Ash free dry mass (AFDM)
3. Diversity indices
4. Taxa richness
5. Indicator species
Klemm and Lazorchak (1993) describe periphyton sampling methods, labora-
tory and quality assurance/quality control procedures.
Periphyton and Macrophyte Monitoring Methods by Monitoring Level
As with all elements being monitored, comparisons are made between reference
and study stream periphyton and macrophyte communities. Monitoring methods
are listed below, followed by abstracts of the various recommended methods for
each monitoring level.
BMP Effectiveness I
Protocol I—Periphyton
Protocol I—Aquatic Macrbphytes
Montana Periphyton Protocol I
BMP Effectiveness II
Protocol II—Periphyton
Protocol II—Aquatic Macrophytes
Montana Periphyton Protocol I or II
BMP Effectiveness III
Protocol II or III—Periphyton
Protocol III—Aquatic Macrophytes
Montana Periphyton Protocol II
BMP Effectiveness IV
Protocol III—Periphyton
Protocol III—Aquatic Macrophytes
Page
150
150
151
150
150
151/152
150
150
152
150
151
The methods for evaluating periphyton and macrophytes are in the early stages
of development for inclusion in the Rapid Bioassessment Protocols (personal
communication from Brian H. Hill and James M. Lazorchak, USEPA Environmen-
tal Monitoring Systems Lab, Cincinnati). Their suggested RBP approach for
aquatic plants is as follows:
149
-------
Protocol I—Periphyton
Protocol I is a cursory survey for the presence and relative abundance of per-
iphyton, including attached filamentous algae and blue-green algae. Presence of
a periphyton community is determined by visual observation (growth on sub-
strate, texture, color) and feeling the substrate for "sliminess". Filamentous algae
are visible "streamers" attached to substrates. Blue-green algae are notable only
when they are at nuisance levels and are detectable by color and odor. Periphy-
ton, exclusive of filamentous algae and blue-green algae, are scored as being
absent or not apparent, rare, common, abundant, or dominant. This method can
be used at the BMP Effectiveness I level.
Protocol II—Periphyton
Protocol II requires qualitative sampling (e.g., scraping from rocks and collecting
of fine sediments in depositional areas) of representative substrates for analysis
in the laboratory. Laboratory analysis involves viewing an aliquot of suspended
periphyton under high magnification (400-1 OOOx) to determine the proportions of
diatoms, blue-green algae, and other algae. Only gross morphology is used to
separate diatoms and blue-green algae from the other algal taxon. This is a
BMP Effectiveness II method.
Protocol III—Periphyton
At a BMP Effectiveness III or IV level, Protocol III requires a minimum of three
quantitative samples (e.g., scraping a given area) of the periphyton community..
This number needs to multiplied by the number of replicate samples that will be
taken at each site. Samples are analyzed for species composition density,
chlorophyll concentration, and biomass (ash free dry mass, AFDM). One of the
three scrapings is sub-sampled and processed according to standard procedures
to allow for species determination and counts of organisms present. The second
scraping is filtered and analyzed for chlorophyll concentration. The third is
filtered, dried, weighed, and ashed to determine AFDM.
Protocol I—Aquatic Macrophytes
Protocol I includes qualitative estimates of aquatic macrophytes and the propor-
tion of the community composed of submerged, floating-leaved, and emergent
life forms. In addition, the percent of the stream occupied by these macrophytes
is estimated. BMP Effectiveness I can use this method. A more advanced
Protocol I involves collecting and field-identifying (to the lowest possible taxon)
aquatic macrophytes present.
Protocol II—Aquatic Macrophytes
Protocol II involves qualitative (line or belt transects) to estimate the relative
abundance of species present. Samples may be sorted, identified, and counted
in the field or returned to the lab for processing. BMP Effectiveness II can use
this method.
150
-------
Protocol III—Aquatic Macrophytes
Protocol III is based on quantitative sampling. Samples are collected to deter-
mine relative abundance and biomass by species. Samples may be sorted,
identified, counted, and weighed in the field or returned to the laboratory for
processing. Level III sampling on successive dates (e.g., monthly during the
growing season) reveals aquatic macrophyte productivity. RBP III is applicable
to BMP Effectiveness III and IV levels.
\
Montana Periphyton Protocol I
The Montana Water Quality Bureau has developed a manual that presents
guidelines for using the composition and structure of periphyton communities to
assess biological integrity and impairment of aquatic life in Montana streams
(Bahls 1993). The manual recommends: (1) methods for collecting, processing,
and analyzing periphyton samples; (2) measurements and metrics for evaluating
periphyton communities; and (3) biocriteria and protocols for assessing biological
integrity and aquatic-life impairment. The document also discusses variability in
metrics.
The manual describes two protocols: Protocol I compares study-stream impair-
ment compared to ecoregion reference conditions; Protocol II assesses impair-
ment based upon comparisons with upstream or side stream conditions. Both
protocols distinguish among four levels of biological integrity and impairment of
aquatic life.
Summer (June 21 to September 21) is the preferred time for collecting periphyton
from Montana streams and is the index period for establishing ecoregion refer-
ence conditions. In Montana, periphyton diversity peaks in summer and early
fall. If sampling must be done outside this period, Protocol II should be used.
Periphyton are collected from natural substrates (rocks, logs, moss, mud) in
proportion to the abundance of these substrates at a given site and then pooled
in a common container. Microalgae are collected by scraping the entire surface
of several rocks of different sizes selected at random. Portions of macroalgae
(algae growing in large filaments or colonies visible to the naked eye) are col-
lected in proportion to their abundance at the site and added to the common
container. Sampling is usually concentrated in riffles, but other macrohabitats
(pools and runs) are also sampled if they support algal growth.
A two step analysis is used: (1) a ranking of non-diatom (soft-bodied) genera by
relative volume (biomass), and (2) relative abundance of diatom species. Soft-
bodied algae are classified by (1) dominant phylum, (2) indicator taxa, and
(3) number of genera. Dominance of blue-green algae may reflect a low level of
inorganic nitrogen in mountain streams, while green algae may indicate nitrogen
enrichment. Certain genera of non-diatom algae can be used as indicators of
different levels and causes of pollution. The number of non-diatom genera is
inversely proportional to the degree of pollution. However, in infertile mountain
streams a small increase in nutrients can increase the number of genera.
151
-------
For diatom species, four metrics are used: (1) diversity index, (2) pollution index,
(3) siltation index, and (4) similarity index. The Shannon diversity index incorpo-
rates dominance and species richness and is sensitive to water-quality impacts.
Diatoms are rated as tolerant, less tolerant, and sensitive to pollution. The
siltation index is based on the abundance of species that are sensitive to sedi-
mentation.
Protocol I is used to compare study streams with ecoregion reference conditions
and can only be used during the summer. This protocol would be appropriate for
BMP Effectiveness I and II.
Montana Periphyton Protocol II
This protocol compares metric values for a study site to those for upstream
reaches or side-stream reference sites (Bahls 1993). The reference site must
have the same stream order as the study reach. This protocol uses the same
three diatom indexes as Protocol I plus the percent-similarity index. Protocol II is
more sensitive than Protocol I because it compares study streams with local
reference conditions. And, unlike Protocol I, it can be applied year around. The
siltation rating method used in Protocol II puts a greater penalty on sediment
increases at the lower end of the siltation index. This method applies to BMP
Effectiveness II and III.
152
-------
Deciding on BMP Effectiveness:
Some Case Histories
The monitoring methods just described can be applied in various ways. The
examples reviewed below, real and hypothetical, illustrate how some of these
methods can be used. These examples demonstrate multi-factor monitoring to
collect, analyze, and present data and information for decisions on whether
BMP's are effective in protecting the physical, chemical, and biological integrity of
streams.
Implementation and Effectiveness Monitoring in South Carolina
T. Adams (1992) monitored the implementation and effectiveness of harvesting
BMP's in South Carolina using RBP III at the BMP Effectiveness III level. The
study involved 27 harvest areas associated with perennial streams. He rated
BMP implementation as adequate or inadequate, traced soil movement from its
source to the stream, and identified sources of sediment found in the stream.
Reference stream segments above the harvest area or in nearby watersheds
were carefully selected to represent least-impaired streams. Similar stream
segments below the harvest area were located for sampling. He used Rapid
Bioassessment Protocol III (RBP III) to sample and identify benthic invertebrates
to the species level and to assess the biological integrity of the reference and
study streams. The RBP habitat evaluation was used to compare habitat quality
with the reference segments. The biological and habitat metrics were analyzed
to determine which ones were most important for the stream/ecosystems being
monitored.
The habitat and biological scores for stream segments below logged areas were
divided by reference scores to obtain the percent of reference and plotted (Figure
24). Based upon analysis of data, biological condition in excess of 75% of the
reference condition was judged as non-impacted. Habitat was judged as non-
impaired if it exceeded 77% of the reference. These thresholds closely corre-
spond to RBP guidance (EPA 1989).
Of the 16 logged areas with adequate BMP implementation, 13 stream segments
below have unimpacted habitat and biological conditions, 2 have unimpacted
habitat but impacted biological condition, and 1 has impacted habitat and
unimpacted biological condition.
Stream segments below 10 of 11 logged areas without adequate BMP implemen-
tation generally had impacted habitat and/or biological conditions (Figure 24).
However, one logged area that had a road crossing a stream with a very small
culvert appeared unimpacted because a high-flow event washed out the road fill
and apparently carried the sediment far below the study reach. ,
Harvest areas rated as having inadequate BMP compliance had been logged
under wet conditions and had inadequate streamside management zones, poorly
designed skid trail stream crossings, poorly located skid trails, inadequate sedi-
ment control from roads, poor road crossings of streams, excessive rutting, etc.
Stream habitat was impaired primarily by excessive amounts of large woody
debris and sediment. The increase in cobble embeddedness, the loss of avail-
able macroinvertebrate habitat, and the loss of riparian canopy all adversely
153
-------
impacted the streams. The health of the invertebrate community increased with
the level of BMP compliance. Multi-factor monitoring linked poor logging prac-
tices with impaired benthic invertebrate condition and aquatic habitat. In short,
the diagnosis is: poor logging impairs water quality. Generally, water quality is
adequately protected where the logger did a good to excellent job of implement-
ing BMP's.
Biological - % of Reference
120
100
80
60
40
20
Non Impacted
Impacted
+ +
-f +
Impacted
77%
Non Impacted
0 10 20 30 40 50 60 70 80 90 100110120130
Figure 24. Effectiveness of logging practices in protecting benthic macroinvertebrates and habitat.
154
-------
A Water-Quality Criteria Approach
Using Adams' (1992) data for the Piedmont region of South Carolina, this ex-
ample demonstrates how water-quality criteria can be used to evaluate streams
at the BMP Effectiveness 111 level of monitoring. For this example, only biological
conditions will be discussed. The Rapid Bioassessment Protocol III for benthic
macroinvertebrates was used to determine the biological condition for each
reference and study stream. Sixteen reference streams were assessed and the
biological-condition scores were plotted (Figure 25). The reference scores for
the Piedmont ecoregion ranged from 36 to 42, with an average of 40, which
defines a reference condition. About 75 percent of the scores were above 39. A
State might therefore set the criterion for biological condition at 39 but not take
enforcement action if the stream under investigation fell within the reference
condition range.
Stream segments below 11 logging areas with BMP's had biological conditions
ranging from 30 to 42 and averaging 37. Nine of these segments had scores
within the range of the reference condition. Two had scores of 32 and 30, well
below the reference condition. Follow-up field investigation of these two logging
areas might find them in general compliance with BMP's, although one or two
BMP's might be only partially implemented or lacking, or the BMP's might not be
totally effective in those situations.
Four logged areas with poor BMP implementation had scores ranging from 24 to
36, with an average of 32 (Figure 25). The BMP compliance evaluation found
several poor logging practices that resulted in organic debris and sediment
loading of streams and reduced stream shade. The 36 score falls at the bottom
of the reference range, but it is not unexpected for poor logging to meet refer-
ence conditions in specific situations. However, three study segments fell well
below the reference range, indicating these violated criteria for biological condi-
tion.
Multi-Factor Approach/Stream Continuum
For a hypothetical situation at the BMP Effectiveness III level of monitoring, a
continuum might be used to evaluate the effectiveness of BMP's. In a reference
stream, the index of biological integrity for fish (IBI), embeddedness, and stream
temperature are evaluated along the continuum from the headwaters down-
stream through a fourth order segment (Figure 26). In the figure, the stream
flows from left to right. Stream order is on the horizonal axis. Variations in the
index of biological integrity, embeddedness, and stream temperature are plotted,
trends established, and data analyzed.
155
-------
50
Biological Condition
40 -
30 -
t
Reference
w/BMP's
w/o BMP's
Figure 25. Benthlc macroinvertebrate response to logging with and without BMP's, Piedmont, S.
Carolina. (Vertical line depicts range in values. Horizonal line represents the mean value.)
The same variables are evaluated for the study stream (Figure 26). The study
watershed features logging with and without BMP's. The trends and variation in
IBI, embeddedness, and temperature can be compared between study and
reference streams and along streams.
Above the logging with BMP's, IBI, embeddedness, and temperature variation
and trends in the study stream appear to be very similar to the reference condi-
tions. As the study stream proceeds past the logging with BMP's, IBI decreases
slightly, embeddedness is unchanged, and temperature increases slightly—
variations and trends similar to those for a comparable segment of the reference
stream. On-slope monitoring of the logging found that BMP's had been installed
correctly, no sediment reached the stream, and the riparian canopy density was
maintained at prescribed levels. The evidence strongly suggests that the BMP's
were effective in protecting water quality to this point along the stream.
As the stream proceeds past the logging without BMP's, IBI decreases signifi-
cantly (Figure 26), stream temperature increases to a near critical level, and
embeddedness in the stream increases significantly. Moreover, 75 percent of
the riparian canopy was removed by the logging. Soil erosion from skid trails,
logging decks, and roads is traced through the riparian area to the stream. All
three in-stream responses can be linked to the poor logging and are clear depar-
tures from reference conditions and natural variation, leading to a diagnosis that
the poor logging has impaired water quality. The next step may be comparing
the impacts with State water-quality criteria (if available) to determine if violations
have occurred.
156
-------
Logging
w/BMP's
Logging
w/o BMP's
index of Biological Integrity
Embeddedness
Temperature
2 3
Stream Order
Reference
Study Stream — — — —
Figure 26. Monitoring effectiveness along a stream continuum.
157
-------
BMP Effectiveness by Stream/Ecosystem Class
At the BMP Effectiveness III and IV levels, another approach to evaluating BMP
effectiveness might be by stream ecosystem class. Some of the methods sug-
gested in this report were tested in a forest setting on streams with the same
stream/ecosystem classification. The test was conducted on the Silver Creek
Research watershed in the Boise National Forest. Tim Burton, Boise National
Forest in Idaho, agreed to test the method and prepare a report (Burton 1993),
which is excerpted below (his citations are not included in References):
The Silver Creek study was established in the early 1960's to obtain informa-
tion on the environmental impacts of logging and associated road construc-
tion. The study encompasses 7 small watersheds located in the headwaters
of Silver Creek, a tributary of the Middle Fork Payette River, on the Boise
National Forest (Figure 27). These third order watersheds are all similar in
size, aspect, and elevation, and ideal for evaluating the effectiveness of
various silvicultural management practices to protect water quality and sup-
porting beneficial uses of water.
To date, most research in Silver Creek has focused on measures of erosion,
sedimentation, and water yield and their responses to silvicultural treatments
and road building. Sediment collection dams, stream gages, and precipita-
tion gages were established in the mid 1960's. Measurements of sediment
and water yield have been continuous since that time. The watersheds were
calibrated for a period of 10 years prior to logging treatments, which began in
1976. Road construction was initiated in 1980 on three of the watersheds,
the others remained unroaded. Three unroaded watersheds were logged by
helicopter. Logging continued on various watersheds until 1985.
A wide variety of treatments, including ground and helicopter yarding, various
stand opening sizes, different slash disposal methods, and ranges of road
Standards were applied to the various watersheds. One watershed, Eggers
Creek was never entered and was retained as a control for comparison
purposes.
Benthic macroinvertebrates and aquatic habitat characteristics were moni-
tored prior to logging in 1973 and 1974. These baseline data provide a
unique opportunity to assess the relative logging impacts on biological integ-
rity. Most of the same stations visited in the 1970's were re-visited in 1992 to
evaluate the aquatic impacts of the various silvicultural treatments. Benthic
macroinvertebrate indicators reflect changes in the aquatic ecosystem and
integrate environmental impacts that have occurred over the period of logging
and application of best management practices.
158
-------
Monitoring Strategy
A total of 17 monitoring stations were sampled in the fall of 1992 at approxi-
mately the same locations as those of Hart and Brusen (1976). Within the 7
study watersheds, stations were located near the mouth, in the reach up-
stream of each watershed's sediment collection dam. In two watersheds,
stations were also located upstream of the locations of road construction, at
sites not previously sampled by Hart and Brusven (1976). Also, several sites
on the mainstem of Silver Creek were sampled to evaluate habitat changes
downstream. Locations of all stations are displayed on the watershed map
(Figure 27).
SILVER CREEK WATESHEDS
Boise National Forest
Eggers
Control Creek
N
Streams
- - — • Roads
17 Monitoring
station
KILOMETERS
0 0.5 1.0
I I i
Figure 27. Map of Silver Creek watersheds.
159
-------
Samples collected in the fall of 1992 on treatment watersheds are compare to
the fall samples before treatment (1974), as well as those in the control
watershed after treatment to assess relative effects on the biological health of
the streams. Thus, three cause-and-effect strategies are used: (1) Before-
after; (2) Treatment-reference; and (3) Upstream-downstream.
A variety of silvicultural BMP's have been applied to the research watersheds
during the past 12 years. This project evaluates the relative effectiveness of
these treatments in protecting aquatic ecosystem health. Details of the
BMP's and research observations on erosion and sediment production
through the early 1980's are contained in cited literature. Recent sediment
production estimates are not available for the Silver Creek watersheds, but
will be reported in 1994.
Description of Treatments and Their Effect on Sediment Yield
The following is a watershed-by-watershed summary of logging and reading
treatments applied over the ten year period (1976-1985) and associated
sediment yields.
1.
2.
"C" Creek Watershed. This basin was logged in 1982. The harvest-
ing method consisted of logging very small patches and yarding by
helicopter. No roads were constructed. Slash was hand piled and not
burned. Sediment yields were not significantly increased as a result of
these treatments.
"D" Creek Watershed. This watershed was logged by clearcutting 33
percent of the area in 2 to 25 acre patches. Buffer strips of about 75
feet were left adjacent to perennial streams in the watershed. These
buffer strips were left undisturbed, except for the removal of any trees
expected to die before the next scheduled harvest. Any logging slash
that fell in streams was removed. Yarding was accomplished by
helicopter, slash was lopped and scattered and not burned, and no
roads or skid trails were constructed in the basin.
Although the overall sediment yield in the basin was not significantly
increased by these activities, a sapping (piping) failure was detected
below one of the clearcut patches in this watershed (Megahan and
Bohn, 1989). This type of mass failure is caused by increased soil
water seepage into headwater, zero-order channels. The reduction in
evapotranspiration within the clearcut increases ground water move-
ment downslope, causing soil in adjacent swails to become saturated.
Mass wasting results when largevolumes of material become satu-
rated and begin to flow downslope in liquid or slurry form below the
sapping zone. A new channel is created by headward migration of the
failure scrap orheadcut
The failure in this watershed commenced in 1984 and produced 38
cubic meters of sediment in 1988. In 1986, this particular sapping
failure accounted for 36 percent of the total annual sediment yield from
160
-------
the watershed. Channel sediment may have increased slightly down-
stream after logging as a result of the mass failure.
3. Eggers Creek Watershed. This watershed has not been entered for
logging or road construction. The watershed is used as a control for
comparison purposes.
4. Ditch Creek Watershed. Ditch Creek has a long history of develop-
ment. In 1933 a low standard road of about 9 miles was constructed
entirely within the 250 acre watershed. Megahan et al. (1983) esti-
mated erosion rates on this road of 225 tons/hectare/year (the 45 year
average). This is more than 1000 times greater than natural slope
erosion on undisturbed forest slopes in the Silver Creek area. During
the calibration period, prior to logging and roading treatments in the
study, sediment yields were consistently higher here than other water-
sheds in study (averaging. 1 cubic meters/hectare/year as compared
with .01 to .07 on undisturbed watersheds).
A .6 mile connector road was constructed in Ditch Creek in 1980 using
routine erosion control measures for that time. Routine erosion con-
trols included: no road surfacing, uncompacted fills, no downspouts,
no treatment of cutslopes, and dry seeding of fill slopes. The road was
about 25 wide, had an average cutslope length of 40 feet and fill
slopes of about 35 feet, and average grade of 8 percent (Megahan,
et al., 1986). During construction, sediment yields were more than 5
times the average for pre-logging years. Most of the eroded sediment
(77%) was stored on the slopes below the road. About 10 percent of
the road sediment became channel sediment which has continued to
be reflected in increased sediment yield and affect aquatic habitat
since road construction.
Small clearcut logging (1-5 acres) occurred in Ditch Creek in 1987.
About 26 percent of'the area was logged. Ground skidding using
caterpillar tractor and rubber tire skidder was used on gentle slopes,
and helicopter logging was used elsewhere. The slash was machine
piled and burned.
Because of the development history in this watershed, it has produced
significant increases in sediment yield at the watershed outlet.
5. No Name Creek Watershed. About one mile each of connector and
local roads were constructed in the watershed in 1980. Routine ero-
sion control practices, as described for Ditch Creek Watershed were
used on roads in this watershed. Sediment yield increases were
equally significantly after logging: about 5 times the pre-logging sedi-
ment yields.
No Name Watershed was selection logged in 1985. Skidding was by a
combination of ground and helicopter depending on proximity to roads.
Routine slash disposal practices were used.
161
-------
Significant increases in sediment were produced mostly as a result of
road system development.
6. Cabin Creek Watershed. This drainage was logged and roaded using
maximum erosion control efforts. About .5 miles of connector and 1.3
miles of local roads were constructed in the basin in 1980. Road
erosion control included: asphalt surfacing on the connector road, rock
surfacing on the local roads, mulching and revegetation of the fill
slopes, energy dissipators on culverts, road location following contours
(minimizing cuts and fills), rolling grade (drainage controls), and few
culverts (thus little concentrated drainage from the road). Sediment
production at the drainage outlet in Cabin Creek, however was signifi-
cantly greater than the pre-development average, but only by one
order of magnitude. This compares to 5 orders of magnitude in Ditch
and No Name watersheds.
Cabin Creek Watershed was logged in 1987. Both ground and heli-
copter skidding were employed in the watershed. About 28 percent of ,
the area was harvested in small clearcuts (1-4 acres). Slash was piled
but not burned.
This watershed has not experienced the severe sediment increases
observed in several of the other developed watersheds. Maximum
erosion control efforts minimized the adverse affects, even though
development has been at a comparable level.
7. Control Creek Watershed. Control Creek was clearcut in 1976.
Three clearcuts totaling 100 acres of old growth ponderosa pine and
Douglas-fir were helicopter yarded. This represented 23 percent of the
watershed area, and all logging was on south-facing slopes, which in
thebatholith are sensitive to erosion and sedimentation.
Logging in Control Creek was not associated with any road building or
construction of landings. Trees were removed from areas located at
least 100 feet from perennial streams. Buffers adjacent to streams
were left undisturbed, except for removal of some dying trees. Slash
was lopped, scattered, and then broadcast burned in the fall of 1976
and winter of 1977. The fire severity was not great and only 50 per-
cent of the clearcuts were actually burned.
Significant increases in sediment production occurred on this water-
shed after logging. Megahan (1987) indicated that sediment yields
showed a statistically significant increase of around 100% over previ-
ous levels resulting from timber harvest. Accelerated sediment pro-
duction persisted at least 10 years after logging, and showed no signs
of abating throughout the decade. Research showed that 6 percent of
the sediment production resulted from mass erosion associated with
the clearcuts, and that 94 percent of the increased sediment yield was
from surface erosion caused by long-term exposure of bare soil in
logged and burned clearcuts. The harsh site conditions on these south
162
-------
slopes restricted vegetation regrowth allowing accelerated erosion
conditions over long periods of time.
Prescribed burning on these sites negated the beneficial effects of
helicopter logging. In such situations, no burning is advisable. Erosion
on these fragile landscapes increased about 80 times over pre-logging
erosion rates causing serious increases in sediment yield.
Summary: Only 2 of the 7 watersheds showed no significant increases in
sediment production after 1976, the control watershed Eggers Creek and "C"
Creek. The latter watershed was logged, but no roads were constructed,
trees were removed by selection logging, and slash was not burned. The "2"
and Cabin Creek watersheds experienced relatively small increases in sedi-
ment yield. On these, intensive application ofBMP's seemed to prevent
severe sedimentation increases. Trees were yarded using helicopter meth-
ods mostly. Reading in Cabin Creek was associated with maximum erosion
controls. The three other watersheds experienced significant sediment yield
increases resulting from clearcut logging, burning of slash, tractor yarding,
. and normal road construction practices.
Based on the research results, Burton rated the effectiveness of BMP appli-
cation within the various watersheds as follows:
Undeveloped:
Highly effective:
Moderately effective:
Ineffective:
METHODS
Eggers Creek
"C" Creek
"D" Creek
Cabin Creek
Ditch Creek
No Name Creek
Control Creek
Rating = 3
Rating = 2
Rating = 1
Rating = 0
At each of the stations shown on the map (Figure 27), biological monitoring
was conducted to evaluate BMP effectiveness. Monitoring was conducted at
the BMP Effectiveness III level. This project employed the basic protocols for
rapid bioassessment (Plafkin et al. 1989) and uses local adaptations for
metrics assessment (Clark 1992).
At each station, three replicate samples were collected using a standard
Hess sampler at a series of randomly selected riffles. One hundred organ-
isms were composited from the samples for rapid bioassessment. Identifica-
tions were made on several samples in the field and all samples were identi-
fied in greater detail at the Idaho Department of Health and Welfare labora-
tory in Boise. Field identification was conducted to compare metric results
from lab identification and determine relative cost effectiveness of the two
methods. At several sites, two composite samples were collected and trans-
ported to the lab to evaluate spatial variability and sample size effectiveness.
163
-------
Biotic indices were computed for the macroinvertebrate samples at each
station. These included: Total number (abundance), total taxa, total EPT
taxa, % EPT, % chironomids, % sediment indicators, % clean water indica-
tors, HSI tolerance, BCI tolerance, % scrappers, % filterers, % shredders,
ratio of scrappers/filterers, and ratio of % EPT/chironomids.
Habitat evaluations were performed at each site using the standard RBP
habitat assessment approach as modified by Clark (1993). Such habitat
evaluations are based on qualitative ratings of cover, substrate
embeddedness, canopy, channel condition, pool-riffle ratio, bank stability,
and riparian vegetation condition. Because sediment is so important in the
present evaluation, a simple quantitative technique was applied to estimating
surface fine sediments. On each plot sampled for macroinvertebrates, a
gridded plexiglass plate was placed on the water surface and each grid
intersection over fine sediment was tallied to determine percent surface fines.
Plots were averaged to determine site conditions.
RESULTS
Spearman ranked correlations were used to evaluate possible relationships
between BMP effectiveness and habitat and biological parameters. Results
are presented in Table 15. Based on these data, the effectiveness ofsilvicul-
tural management practices can best be evaluated by substrate fine sedi-
ment, habitat condition (score), percent of EPT insects (stoneflies,
caddisflies, and mayflies), percent shredders, percent filterers, and percent
scrapers.
Among the habitat factors evaluated, percent fine sediment had the strongest
correlation to BMP effectiveness. The relationship is displayed in Figure 28.
Ineffective BMP's (ratings less than 2) were associated with higher sedimen-
tation, and substrate fines greater than 25 percent.
Table 15. Spearman Rank Correlations—BMP effectiveness in the Silver Creek watersheds
PARAMETER 1
PARAMETER 2
rho
BMP effectiveness
BMP effectiveness
BMP effectiveness
BMP effectiveness
BMP effectiveness
BMP effectiveness
BMP effectiveness
BMP effectiveness
BMP effectiveness
BMP effectiveness
% Fine Sediment -.82
Habitat score .69
Pool/Riffle ratio .77
Bank Stability .69
Channel alteration rating . 77
Substrate Stability .77
% EPT insects .64
% Filterers -.66
% Shredders .76
Ratio: Scraps/Fills .66
164
-------
The relationship between EPT insects and BMP effectiveness reflects the
impact of sedimentation on this taxonomic group. As sediment production
increases, the numbers of these insects decline while the numbers ofDiptera
(especially the midges) increase. As shown in Figure 28, as substrate fines
increase above 30 percent, the numbers of EPT decline to less than 55
percent of the sample. The effect on EPT relates to a shift in composition as
shown in Figures 29 and 30. Figure 29 indicates that as fine sediments
increase, the abundance of the stonefly, Peltoperlidae decreases. Figure 30
shows an opposite relationship to Chironomidae.
The stonefly Peltoperlidae has an affinity to forested streams with abundant
shade and canopy cover as well as course substrates. It was suggested as
an excellent indicator of the effect of logging practices in earlier Silver Creek
studies by Brusven (1978). This insect, along with the Chironomidae
(midges) are easily recognized in the field. Because of their relative abun-
dance, they constitute major factors in computation of the EPT index.
Burton evaluated the relationship between biological factors and substrate
fine sediments using the Spearman rank correlation. Results are presented
in Table 16. Percent fines have a negative correlation with pool/riffle ration,
bank stability, channel alteration, and the number of EPT taxa. These rela-
tions suggest that as sedimentation increases, adverse channel changes are
concurrent with declines in EPT taxa.
Table 16. Spearman Rank Correlations—Percent fine sediment in the Silver Creek watersheds
% Fine Sediment Pool/Riffle ratio
-.79
% Fine Sediment Bank Stability -.80
% Fine Sediment Channel alteration -.88
% Fine Sediment % EPT, Taxa
-.60
Percent EPT also indicated a high correlation to habitat score, bottom cover,
pool/riffle ration, channel alteration, and substrate stability. No other biotic
indices correlated significantly with the habitat parameters. Because EPT
insects can be estimated in the field, this index has tremendous potential for
evaluating BMP's in a cost-effective monitoring program.
165
-------
1992 data: + Pre:-logging data
8 16 28 38 48 58 68 78 88
Substrate surface fines (percent)
Figure 28. Relationship between percent substrate fines and BMP effectiveness rating for sediment control
Silver Creek watersheds.
' t
LU
• 1992 data: +: Pre-loading data
0 10 20 30 40 50 60 70 80 90 100
% Fines < 6.4 MM
Figure 29. Relationship of percent EPT to percent fine sediment in substrate, Silver Creek watersheds.
166
-------
o>
Q.
3
I
50
45
40
35
30
25
20
15
10
5
10 20 30 40 50 60
% Fines < 6.4 MM
70
80
90 100
Figure 30. Relationship of percent Peltopertidae to percent fine sediment in substrate, Silver Creek watersheds.
50
45
40
35
I 30
| 25
6 20
<Ł
15
10
5
10 20 30 40 50 60
% Fines < 6.4 MM
70
80 90
100
Figure 31. Relationship of percent Chironomidae (Midges) to percent fine sediment in substrate, Silver Creek
watersheds.
167
-------
Reference conditions for percent fines, percent EPT, percent Peltoperlidae, and
percent Chironomidae can be gleaned from Burton's data in Figures 28 to 31.
The prelogging data plus data from Eggers Creek (control) could represent
reference site values for the four variables and the range of these scores can be
used to define reference conditions. The reference condition for fine sediments
is 12-25%, for percent EPT it is 50-77%, for percent Peltoperlidae it is 22-48%,
and for percent Chironomidae it is 1-10%. If the study stream score for percent
fine sediment and percent Chironomidae exceed the corresponding reference
condition, then management activities in the watersheds impaired these vari-
ables. If the study stream scores for percent EPT and for percent Peltoperlidae
fall below their reference conditions, impairment has occurred.
The effectiveness of management activities in the 6 developed watersheds can
be compared to these reference conditions (Table 17). The small-patch
clearcuts yarded by helicopter without reading in "C" Creek resulted in the 4
variables falling within reference conditions, suggesting that this form of logging
and associated BMP's protected water quality.
Cabin Creek had well-designed and well-located access roads that incorporated
sound road surfacing and stabilization BMP's. Twenty-eight percent of the
watershed was clearcut and slash was not burned. Sediment yield was in-
creased some, but the 4 water-quality scores fell within the reference conditions.
Diagnosis: water quality was protected by this type of management and quality of
BMP's.
Table 17. Management and BMP effectiveness evaluation of the 6 developed and the control Silver Creek
watersheds (bold values indicate departure from reference conditions)
Sediment Percent Percent Percent
Watershed Yield Increase Fines EPT Peltoperlidae
"C" Creek None 14 51 25
"D" Creek Some 39 55 19
Egger Creek
(Control) None 19 77 48
Percent
Chironomidae
5
7
1
Ditch Creek
No Name
Creek
Cabin
Creek
Control
Creek
5 Times
5 Times
Doubled
80 Times
75
83
25
47
42
55
49
27
24
35
19
38
12
11
168
-------
"D" Creek was logged by helicopter and was unroaded. A buffer strip was left
between the logged area and the stream. The land form and reduced evapo-
transpiration resulted in a small landslide after logging. An increase in sediment
yield from the landslide resulted in the percent fine sediment and the percent
Petloperlidae scores departing from reference conditions (see bold values in
Table 17). These responses would suggest impairment. "C" Creek was man-
aged essentially the same way, with good results. "D" Creek results suggest that
some refinement in BMP's to minimize the potential of landslides may be in
order.
In Ditch and No Name Creeks, the lack of erosion control on roads clearly in-
creased sediment yields and resulted in increased percent fines and reduced
EPT, both of which are outside reference conditions. The combination of man-
agement activities and lack of implementation of road BMP's appears to have
impaired water quality significantly.
Logging a south aspect and broadcast burning in Control Creek also resulted in
water-quality impairment, based upon water-quality variables.
The Silver Creek project clearly demonstrates multi-factor monitoring and how
various management activities can be evaluated for effectiveness within an
ecoregion and for a stream/ecosystem class.
An Example of BMP Effectiveness I Monitoring
Walsh (1992) evaluated the change in biological diversity from above to below
harvest areas using the Izaak Walton League (1990) biological diversity index.
Above the harvest site, the biological diversity index was 26, but below it was 5.
Both samples were taken in similar riffles. Temperature, pH, conductivity, total
dissolved solids, turbidity, nitrate, and ortho-phosphate concentrations were
similar at both stream locations. Walch states: "The only observable impact on
stream quality was a large haul road that began far up the slope and emptied
directly into the small stream. No water bars, seeding, pipe culvert or means for
stream crossing protection were present and heavy sediment flow had occurred
in the past, evidenced by the road condition. Possibly the erosion from this road
had contributed significantly to stream degradation."
Determining Effective Life of Best Management Practices
Questions often asked include: Are forestry best management practices effective
for as long as it takes for a management area to recover and stabilize? How long
are BMP's effective? How can we determine how long they are effective?
Several factors determine how long BMP's are effective, including: (1) quality of
BMP implementation, (2) climate, (3) rate of vegetative recovery, (4) type of soils,
and (5) the occurrence of large storms.
169
-------
The quality of BMP implementation refers to meeting design specifications.
Waterbar installation is a good example. Are they the right height? Right angle?
Do they discharge onto undisturbed soil? To accurately evaluate the design life
of BMP's, they must be installed according to design specifications. If not, and
they fail, it is not the fault of the design.
Climate greatly influences the effective life of BMP's. Snow-melt hydrology is
different from rainfall hydrology. Rainfall amounts and erosive energy vary
widely from region to region. Mediterranean climates have wet winters and
droughty summers, whereas in other regions precipitation occurs relatively
uniformly throughout the year. Some regions have intensive thunderstorms that
can produce several inches of rain in a short period. The form of precipitation,
associated erosion, and surface runoff influence the life of BMP's.
Closely associated with climate is the rate of vegetative recovery. In the arid
West, vegetation may take decades to invade and stabilize disturbed areas,
while in the humid South it may take only a matter of months to a few years,
depending upon the disturbance.
For example, the filtering capability of streamside management zones (SMZ) is
related to the rate of vegetative recovery above them. SMZ's are designed to
absorb surface runoff and trap eroded soil before it reaches a stream. SMZ's
can only trap so much sediment before they become ineffective. The period of
vegetative recovery determines how long SMZ's are subjected to accelerated
sedimentation. Establishing SMZ's to specific design specifications may be
effective in the South where vegetative recovery is fast, but these specifications
may not be effective in the arid West where recovery is slow.
The type of soil also helps determine the effective life of BMP's. Soil stability,
credibility, compactibility, moisture-holding capacity, nutrient content, permeabil-
ity, and hydrophobic conditions all influence BMP's effective life. Soil stability
refers to risk of mass movement. Soil credibility influences the rate of soil de-
tachment and amount of sediment from a site. Soil-moisture holding capacity
and nutrient content influence the rate of vegetative recovery. The susceptibility
of a soil to compaction influences surface runoff and the rate of vegetative recov-
ery. Some soils have high infiltration rates, others low, and some are hydropho-
bic (water repellent).
Major storms can shorten the life of BMP's. A major storm can overload filter
strips with sediment, wash out drainage structures and waterbars, cause land-
slides, etc. With expected precipitation and storms, effective life is usually a
matter of a few years. However, if a rare storm occurs during the recovery
period, the BMP's may fail.
Determining Effective Life Under Normal Weather Patterns: A BMP is con-
sidered effective if, throughout its life, it meets the physical, chemical, and bio-
logical goals or standards for water. Some BMP's may be effective for the first
year or so but gradually fail in subsequent years. The question is, how long is a
BMP effective? The following monitoring procedure will answer this question for
170
-------
normal weather patterns. Evaluating BMP effectiveness when rare, major storms
occur is discussed later.
When and how long to monitor depends on response lag, that is, how long after
the management activity the expected response occurs and on the expected
duration of the management impact and the longevity of the BMP's. For ex-
ample, monitoring sediment from surface erosion could begin with the initiation of
the activity and continue for a few years, while landslides may not develop until 5
years after logging so monitoring for landslide sediment could be delayed.
The methods for determining effective life and BMP effectiveness are the same.
The choice of monitoring level again depends on the issues or need; effective life
can be determined at all four levels of monitoring. The procedure is as follows:
1. Use reference and study streams, and classify the stream/ecosystem to
ensure streams or stream segments are as similar as possible.
2. Select precipitation gages near reference and study streams. Analyze the
precipitation record to determine if the rainfall and snowstorms are repre-
sentative of those expected during a normal recovery period for the
management activity. If a rare storm has occurred, the study may need to
be moved to area of "typical" precipitation for the normal vegetative recov-
ery period.
3. Locate management areas of various ages adjacent to streams.
Logged areas in the South require 3 to 4 years to revegetate and stabilize,
so areas 1, 2, 3, and possibly 4 years old should be located for study. In
the West, it may take 10 years to recover, so logged areas 1-10 years old
could be selected for study, or areas 1, 3, 5, 7 and 9 years old could be
evaluated. The management areas need to be similar in geology, slope,
soils, aspect, and vegetation.
The alternative is to implement a management activity and BMP's and
monitor conditions on-slope and in-stream until the area is completely
healed.
4. Using on-slope monitoring methods, evaluate the quality of BMP imple-
mentation, trace sediment to point of deposition or to the stream, and
estimate or measure soil erosion, etc. Evaluating on-slope data over time
reveals trends in the amount of bare soil, litter decomposition, revegeta-
tion, rill and gully formation, and sediment in filter strips, etc.
5. Monitor the physical, chemical, and biological integrity of streams to
document the health of the streams associated with each management
age. Compare study streams with reference streams to evaluate the
amount of impact on the physical, chemical, and biological (PCB) integrity.
The effective life of BMP's is time between implementation to when BMP's
break down and result in unacceptable impacts to streams.
171
-------
In-stream impacts may not occur until a disturbance is a few years old.
Coupled with on-slope monitoring data, this information may document
when BMP's break down and begin to cause unacceptable impacts.
6. If the apparent effective life is not at least as long as the recovery period,
review and revise the BMP implementation standards. The on-slope
information should show what, where, and how BMP's fail and suggest
how to modify them to extend their effective life.
Why is extending effective life important? Water-quality agencies usually expect
water-quality goals or standards to be met at nearly all times, which means they
will accept only very short-term departures, meaning days or weeks, not years.
Major Storms and BMP Effectiveness
Foresters often wonder how large a storm BMP's will'withstand. Some BMP's
have storm size considered; for example, culverts are typically designed to
handle a storm with a 25- to 100-year-return period. But what about other ,
BMP's? How do we determine how large a storm waterbars, filter strips, road
drainage, etc. can withstand? Will filter strips handle runoff and sediment from a
storm expected to occur annually, every 2 years, every 5 years, every 10 years,
every 25 years, every 50 years, or every 100 years? Management cannot wait
10 to 100 years to find out.
The solution to this problem is to keep track of weather events in the region or
State. Usually within the region or State, a large storm will occur every few
years. Using data on the precipitation amount and duration, the investigator can
determine the return period for the storm from weather maps, by calculating
storm-frequency distributions, or by referring to State climatical publications or
Weather Bureau publication TP-40.
The area affected by the storm can be mapped. Within the affected area, forest
management areas with BMP's can be located by contacting local foresters. The
same monitoring logic should be used as for evaluating the effective life of
BMP's: age of management areas and BMP's needs to be considered; reference
and study streams need to be assessed for physical, chemical, and biological
integrity. The on-slope evaluation of BMP's needs to be fairly detailed; tracing
soil movement and runoff is vital because runoff, erosion, and sediment are the
primary variables related to with storm size and BMP effectiveness.
On-slope monitoring will determine if waterbars survived and are functional, if
runoff and sediment overwhelmed filter strips, if revegetation treatments held, if
landslides occurred and in association with what practices or land features, etc.
In-stream concerns may be channel and stream-bank stability, transport of large
woody debris, substrate stability, habitat changes or damage, flushing out of
benthic invertebrates and fish, etc.
172
-------
Invertebrate populations can be dramatically reduced with substrate movement
and habitat loss from high-flow events. If a study stream has low benthic inverte-
brate and fish populations, reduced habitat, and impaired channel stability, is it
due to BMP failure or natural processes associated with major a major storm?
Assessing the reference stream may reveal similar conditions, suggesting the
storm as the culprit.
Investigating areas subjected to various storm sizes and building a database of
on-slope and in-stream information can help determine the size of storms indi-
vidual BMP's can withstand in a region. Some BMP's may work in storms with
5- or 10-year returns, while others fail. A 25-year storm may wash out or breach
waterbars, while rolling dips work; filter strips may not keep runoff and sediment
from entering the stream. With a 100-year storm, all BMP's may fail.
But does the apparent on-slope failure of BMP's really matter? Reference and
study streams may have essentially the same biota, habitat, and channel condi-
tions. In such a case, management impacts are insignificant and are masked by
the huge impacts from the storm.
Monitoring areas exposed to 1- to 10-year storms may reveal that proper imple-
mentation of BMP's does protect the water quality during such storms. Maybe
10- to 25-years storms cause some BMP's to fail, resulting in stream impairment.
For storms of intervals greater than 25 years, all BMP's may fail, but their impacts
may or may not be obscured by the natural processes.
173
-------
References
APHA. 1989. Standard methods for the examination of water and wastewater. 17th ed. Prepared and published
jointly by American Public Health Association, American Water Works Association, Water Pollution
Control Federation. Joint Editorial Board, Clesceri, Lenore S.; Greenburg, Arnold E.; Trussel, R. \
Rhodes; Franson, Mary Ann H., managing editor. Washington: American Public Health Association.
1460 p.
Adams, Timothy.0.1992. Implementation and effectiveness monitoring of silvicultural best management
practices on harvested sites in South Carolina. Clemson, SC:.CIemson Univ., Forest Resources. 196 p.
Dissertation
Averett, R.C. 1979. The use of select parametric statistical methods for the analysis of water quality data.
Presented at USGS-BLM Conference on Water Quality in Energy Areas. January 10-11,1979, Denver,
CO. 16 p.
Bahls, Loren U, 1993. Periphyton bioassessment methods for Montana streams. Helena, MT: Montana Dept.
of Health and Environmental Sciences, Water Quality Bureau. 60 p.
Bali, J. 1982. Stream classification guidelines for Wisconsin. Wisconsin Dept. of Natural Resources Technical
Bulletin. Madison, Wl: Wisconsin Department of Natural Resources. (Draft) 12 p.
Barbour, Michael T.; Stribling James B.; and Karr, James R. 1991. Biological criteria: technical guidance for
streams - 4th draft. EPA Contract No. 68-CO-0093. Washington: U.S. Environmental Protection Agency,
Office of Science and Technology. 192 p.
Bisson, Peter A.; Nielsen, Jennifer L; Palmason, Ray A.; Grove, Larry E. 1981. A system of naming habitat
types in small stream, with examples of habitat utilization by salmonids during low streamflows. In:
Armanfront, Neil B. ed. Acquisition and utilization of aquatic habitat inventory information, western
division. Am. Fish. Soc.: Proceedings of a symposium; 1981 October 28-30; Portland, OR: 62-73.
Bosch, J.M.; Hewlett, J.D. 1982. Review of catchment experiments to determine the effect of vegetation
changes on water yield and evapotranspiration. J. Hydrol. 55:3-23.
Box, G.E.P.; Jenkins, G.M. 1976. Time series analysis: forecasting and control, Holden-Day series in time
series analysis and digital processing. San Francisco: Holden-Day. 575 p.
Burton, Timothy. 1993. Draft report. Evaluating the effectiveness of forestry best management practices using
rapid bioassessment procedures: Silver Creek, Boise National Forest. Boise, ID: U.S. Department of
Agriculture, Forest Service. 14 p.
Clingenpeel, J. Alan; Cochran, B. G. 1992. Using physical, chemical, and biological indicators to assess water
quality on the Ouachita National Forest utilizing Basin Area Stream Survey (BASS) methods. Hot
Springs, AR: U.S. Department of Agriculture, Forest Service, Ouachita National Forest. 6 p.
Cummins, Kenneth W.; Wilzbach, M.A.; Gates, D.M.; Perry, J.B.; Taliaferro, W.B. 1989. Shredders and riparian
vegetation. Bioscience. 39(1):24-30.
Cupp, C.E. 1989. Stream corridor classification for forested lands of Washington. Bellevue, WA: Hosey and
Assoc. 46 p.
Dissmeyer, George E. 1982a. Developing a USLE cover-management (C) factor procedure for forest condi-
tions. In Proceedings of the workshop on estimating erosion and sediment yields on rangelands; 1981
March 7-9; Tucson, AZ. ARM-W-26. U.S. Department of Agriculture, Agricultural Research Service:
166-186.
Dissmeyer, G.E. 1982b. How to use fabric dams to compare erosion from forestry practices. Forestry Report
SA-FR 13. Atlanta: U.S. Department of Agriculture, Forest Service, Southern Region. 9 p.
Dissmeyer, George E. 1982c. Report on application of the USLE to desert brush and rangeland. In Proceedings
of the workshop on estimating erosion and sediment yields on rangelands. Tucson, AZ. 1981 March 7-9;
Tucson, AZ. ARM-W-26. U.S. Department of Agriculture, Agricultural Research Service: 214-225.
Dissmeyer, George E.; Foster, George R. 1981. Estimating the cover-management factor (C) in the universal
soil loss equation for forest conditions. J. Soil and Water Conservation. 36(4):235-240.
174
-------
Dissmeyer, George E.; Foster, George R. 1984. A guide for predicting sheet and rill erosion on forest land.
Tech. Pub. R8-TP 6, Atlanta: U.S. Department of Agriculture, Forest Service, Southern Region. 40 p.
Dolloff, C. Andrew; Hankins, David G.; Reeves, Gordon H'. 1993. Basinwide estimation of habitat and fish
populations in streams. Gen. Tech. Rep. SE-83. Asheville, NC: U.S. Department of Agriculture, Forest
Service, Southeastern Forest Experiment Station. 25 p.
Eaton, Lawrence E.; Lenat, David R. 1991. Comparison of a rapid bioassessment with method North Carolina's
qualitative macroinvertebrate collection method. Journal North American Benthological Society
10(3):335-338.
Ebert, Danny J.; Filipek, Stephen P.; Patton, Richard D. 1987. A basic habitat analysis guide for Arkansas
streams. Russellville, AR: U.S. Department of Agiculture, Forest Service. Unpublished Report. 46 p.
Ehinger, W.; D. Potts, D. 1990. On-site assessment of Best Management Practices as an indicator of cumula-
tive watershed effects in the Flathead Basin. Flathead Basin Water Quality and Fisheries Cooperative.
Missoula, MT: University of Montana, School of Forestry. 137 p.
. EPA, 1983. Guidelines and specifications for preparing Quality Assurance Program plans. Washington: U.S:
Environmental Protection Agency, Off. Res. Develop., EPA-600/8-83-024. 22 p.
EPA. 1990. Biological criteria: National program guidance for surface waters. Washington: U.S. Environmental
Protection Agency, Office of Water Regulations and Standards (WH-585), EPA-440/5-90-004.57 p.
EPA. 1993. Environmental monitoring and assessment program, surface water and Region 3 regional environ-
mental monitoring and assessment program: 1993 Pilot field operations and methods manual-—streams.
Cincinnati, OH: U.S. Environmental Protection Agency, Environmental Monitoring Systems Laboratory.
233 p.
Fausch, K.D.; Lyons, J.; Karr, J.R.; Angermeier, P.L. 1990. Fish communities as indicators of environmental
degradation. In: Adams, S.M. ed. Biological indicators of stress in fish. Bethesda, MD: American
Fisheries Society. American Fisheries Society Symposium 8:123-144.
U.S. Department of Agriculture, Forest Service, 1992. Investigating water quality in the Pacific Southwest
Region: BMPEP—Best management practices evaluation program: A user's guide. San Francisco:
USDA Forest Service, Region 5.242 p.
Gilbert, R.0.1987. Statistical methods for environmental pollution monitoring. New York: Van Nostrand
Reinhold. 320 p.
Grant, Gordon E.; Duval, Joseph E.; Koerper, Greg J.; Fogg, James L. XSPRO: A channel cross-section
analyzer. Tech. Note 387. Denver: Bureau of Land Management. 53 p.
Gray, Donald H.; Megahan, Walter F, 1981. Forest vegetation removal and slope stability in the Idaho batholith.
Res. Pap. INT-271. Ogden, UT: U.S. Department of Agriculture, Forest Service, Intermount. Forest and
Range Exper. Sta. 23 p.
Green, R.H. 1989. Power analysis and practical strategies for environmental monitoring. Environ. Res. 50:195-
205.
Griggs, G. 1988. Impact of the January 1982 flood in Santa Cruz County. In: Landslides, floods, and marine
effects of the storm of Jan. 3-5,1983, in the San Francisco Bay Region, California. USGS Prof. Pap.
1434. Washington: U.S. Geological Survey:205-228.
Guy, H.P. 1970. Fluvial sediment concepts. Chapter C1 In Techniques of water-resource investigations of the
United States Geological Survey. Applications of hydraulics, Book 3. Washington: U.S. Geological
Survey. 55 p.
Guy, H.P.; Norman, V.W. 1970. Field methods for measurement of fluvial sediment. Chapter C2 in Techniques
of water-resources Investigations of the United States Geological Survey. Applications of hydraulics,
Book 3. Washington: U.S. Geological Survey. 59 p.
Hankin, David G. 1984. Multistage sampling design in fisheries research: applications in small streams. Can. J.
Fisheries and Aquatic Sci. 41:1575-1591.
175
-------
Hankin, David G.; Reeves, Gordon H. 1988. Estimating total fish abundance and total habitat area in small
streams based on visual estimation methods. Can. J. Fish. Aquat. Sci. 45(5):834-844.
Hayslip, Gretchen A. 1993. Final Draft—EPA Region 10 in-stream biological monitoring handbook for wadable
streams in the Pacific Northwest. EPA 910/9-92-013. Seattle: EPA Region 10. 75 p.
Hilsenhoff, W.L. 1987. An improved biotic index of organic stream pollution. Great Lakes Entomologist 20(1 ):31-
39.
Hurlbert, S.H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs
54(2):187-211.
Idaho WRRI, 1992. Idaho streamwalk: a stream monitoring tool for citizens. Moscow, ID: Idaho Water Re-
sources Research Institute. University of Idaho. 12 p.
Izaak Walton League. 1990. Save our streams: stream quality survey. Izaak Walton League, 1401 Wilson
Boulevard, Level B, Arlington, VA 22209.
Kappesser, Gary B. 1993. Riffle stability index. Coeur d'Alene, ID: U.S. Department of Agriculture, Forest
Service, Idaho Panhandle National Forests. 10 p.
Karr, James R. 1981. Assessment of biotic integrity using fish communities. Fisheries. 6(6):21-27.
Karr, James R. 1991. Biological integrity: a long neglected aspect of water resource management. Ecol. Applic.
1:66-84.
Karr, James R. 1993. Defining and assessing ecological integrity: beyond water quality. Environmental
Toxicology and Chemistry 12(9): 1521-1531.
Karr, J.R.; Fausch, K.D.; Angermeier, P.L.; P.R. Yant, P.R.; Schlosser, I.J. 1986. Assessing biological integrity
in running waters: a method and its rationale. Special Publication No. 5. Campaign, IL: Illinois Natural
History Survey. 28 p.
Klemm, Donald J.; Lazorchak, James M. 1993. Environmental monitoring and assessment program, surface
waters and Region 3 regional environmental monitoring and assessment program, 1993 pilot field
operations and methods manual, streams. Cincinnati, OH: EPA Environmental Monitoring Systems
Laboratory. 221 p.
Kunkle, Samuel; Johnson, W.S.; Rora, M. 1987. Monitoring stream water for land-use impacts. Fort Collins,
CO: Nat. Park Serv., Water Resources Div.
Lenat, David, R. 1988. Water quality assessment of streams using a qualitative collection method for benthic
macroinvertebrates. J. N. Am. Benthological Soc. 7(3):222-233.
Leopold, Luna B. 1970. An improved method for size distribution of stream-bed gravel. Water Resources.
Research 6(5): 1357-1366.
Lisle, Thomas E/1982. Effects of aggradation and degradation on riffle-pool morphology in natural gravel
channels, northwestern California. Water Resour. Res. 18(6):1643-1651.
Lisle, Tomas E.; Hilton, Susan J. 1992. The volume of fine sediment in pools: an index of sediment supply in
gravel-bed streams. Water Res. Bui. 28(2):371 -383.
Lyons, John. 1992. Using the index of biotic integrity (IBI) to measure environmental quality in warmwater
streams in Wisconsin. Technical Report NC-149. St. Paul, MN: U.S. Department of Agriculture, Forest
Service, North Central Forest Experiment Station. 51 p.
MacDonald, Lee H.: Smart, Alan W.; Wissmar, Robert C.1991. Monitoring guidelines to evaluate effects of
forestry activities on streams in the Pacific Northwest and Alaska. EPA/910/9-91-001. Seattle, WA: U.S.
Environmental Protection Agency. 166 p.
Mendenhall, W., 1971. Introduction to probability and statistics. 3rd. ed. Belmont, CA: Duxbury Press. 466 p.
176
-------
Miller, D.L.; Leonard, R.M; Hughes, R.M.; et al. 1988. Regional applications of an index of biotic integrity for use
in water resource management. Fisheries. 13(5):12-20.
Minshall, G.Wayne; Cummins, Kenneth W.; Petersen, R.C.; etal. 1985. Developments in stream ecosystem
theory. Can. J. Fish. Aquat. Sci. 42(5):1045-1055.
Montgomery, Robert H.; Reckhow, Kenneth H. 1984. Techniques for detecting trends in lake water quality. Wat.
Resour. Bull. 20(1 ):43-52.
Ohio EPA. 1987. Biological criteria for protection of aquatic life. Vol. 2. Users manual for biological field
assessment of Ohio surface waters. Columbus OH: Ohio Environmental Protection Agency, Division of
Water Quality Planning and Assessment. 281 p.
Ohio EPA. 1989. Biological criteria for the protection of aquatic life. Vol. 3. Standardized biological field
sampling and laboratory methods for assessing fish and macroinvertebrate communities. Columbus,
OH: Ohio Environmental Protection Agency, Ecological Assessment Section, Division of Water Quality
Planning and Assessment. 58 p.
Ohlander, Coryell A. 1991. Water resources analyses: T-Walk: Water quality monitoring field manual and
tables. Lakewood, CO: U.S. Department of Agriculture, Forest Service, Region 2. 23 p.
Omernik, James M. 1987. Map supplement: ecoregions of the conterminous United States. Annals Association
of American Geographers. 77(1 ):118-125.
Overton, Kerry; McCain, Mike; Fuller, Dave; Decker, Lynn. 1990. R-5 fish habitat relationships. FHR Technical
Bulletin No. 1. Eureka, CA: U.S. Department of Agriculture, Forest Service, Six Rivers National Forest.
15 p.
Paustian, Steve J.; Anderson, Kimberiy; Blanchet, Dave; et al. 1992. A channel type users guide for the
Tongass National Forest, Southeast Alaska. R10 Tech. Pap. 26. Juneau, AK: U.S. Department of
Agriculture, Forest Service, Alaska Region. 187 p.
Pfankuch, Dale J., 1978. Stream reach inventory and channel stability evaluation: a watershed management
procedure. Missoula, MT: U.S. Department of Agriculture, Forest Service, Lolo National Forest, 26 p.
Plafkin, James L.; Barbour, Michael T.; Porter, Kimberiy D. et al. 1989. Rapid bioassessment protocols for use
in streams and rivers: benthic macroinvertebrates and fish. EPA/444/4-89-001. Washington: Environ-
mental Protection Agency, Office of Water (WH-553). 174 p.
Platts, W.S.; Armour, C.; Booth, G. D.; et al. 1987. Methods for evaluating riparian habitats with applications to
management. GTR INT-221. Ogden, UT: U.S. Department of Agriculture, Forest Service, Intel-mountain
Forest and Range Experiment Station. 177 p.
Platts, William S.; Megahan, Walter F.; Minshall, G. Wayne. 1983. Methods for evaluating stream, riparian, and
biotic conditions. Gen. Tech. Report INT-138. Ogden, UT: U.S. Department of Agriculture, Forest
Service, Intermountain Forest and Range Experiment Station. 70 p.
Ponce, Stanley L., 1980b. Statistical methods commonly used in water quality data analysis. Tech. Pap.
WSDG-TP-00001. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain
Forest and Range Experiment Station. 136 p.
Ponce, Stanley L. 1980a. Water quality monitoring programs. Tech. Pap. WSDG-TP-00002. Fort Collins, CO:
U.S. Department of Agriculture, Forest Service, Rocky Mountain Forest and Range Experiment Station.
66 p.
Rabe, Fred W. 1992. Streamwalk II: Learning how to monitor our streams. Moscow, ID: Idaho Water Resources
Research Institute, University of Idaho. 59 p.
Rankin, Edward T. 1989. The qualitative habitat evaluation index (QHEI): rationale, methods, and application.
Columbus, OH: Ohio Environ. Protection Agency, Div. of Water Quality Planning and Assessment. 72 p.
Renard, Kenneth G.; Foster, George R.; Weesie, G.A.; et al. 1991. RUSLE: Revised universal soil loss
equation. J. Soil and Water Conservation. 46(1 ):30-33.
177
-------
River Watch Network. 1992. Guide to macroinvertebrate sampling. Montpelier, VT: River Watch Network. 98 p.
Rosgen, David L., 1985. A stream classification system. In: Johnson, R.R.; Zeibell, C.D.; Patton, D.R.; et al eds.
Riparian ecosystems and their management: reconciling conflicting uses. Gen. Tech. Rep. RM-120. Fort
Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Forest and Range
Experiment Station: 91-95.
Sanders, T.G.; Ward, R.C.; Loftis, J.C.; et al. 1987. Design of networks for monitoring water quality, 2nd ed.
Littleton, CO: Water Resource Publ. 328 p.
Schultz, Bill. 1992. Montana forestry best management practices irnplementatiqnmonitoring, the 1992 forestry
BMP audits, final report. Missoula, MT: Montana Department of State Lands, Forestry Division. 52 p.
Siegel, S. 1956. Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill. 312 p.
Sokal, R.R.; Rohlf, F.J. 1981. Biometry. San Francisco: W.H. Freeman and Co. 859 p.
Swanston, F.J.; Janda, R.J.; Dunne, R.J.; Swanston, D.N. 1982. Sediment budgets and routing in forested
drainage basins. Gen. Tech. Rep. PNW-141. Portland, OR: U.S. Department of Agriculture, Forest
Service, Pacific Northwest Forest and Range Experiment Station. 23 p.
Tarzwell, C.M. 1937. Experimental evidence on the values of trout stream improvement in Michigan. Trans. Am.
Fish. Soc. 66:177-187.
Thurow, Russ. 1991. Protocols for underwater visual estimates of fish abundancein the Salmon River Basin.
Boise, ID: U.S. Department of Agriculture, Forest Service, Intermountain Forest and Range Experiment
Station. 6 p.
VanDeventer, J.S.; Plaits, W.S. 1990. Microcomputer software system for generating population statistics from
electrofishing data - users guide for Microfish 3.0. Gen. Tech. Rep. INT-254. Ogden, UT: U.S. Depart-
ment of Agriculture, Forest Service, Intermountain Forest and Range Experiment Station.
Vowell, Jeff. 1990. Monitoring best management practices in Florida. Workshop on implementation monitoring
of forestry best management practices. Atlanta, GA: U.S. Department of Agriculture, Forest Service,
Southern Region: 25-37.
Walsh, Jennifer. 1992. Effects of timber harvesting on benthic macroinvertebrate populations in southwestern
Virginia. Charlottesviile, VA: Virginia Department of Forestry. 53 p.
Winget, Robert N.; Mangum, Fred A. 1979. Biotic condition index: integrated biological, physical, and chemical
stream parameters for management. Ogden, UT: U.S. Department of Agriculture, Forest Service,
Intermountain Region. 51 p.
Wischmeier, Walter H.; Smith, Dwight D. 1978. Predicting rainfall erosion losses -a guide to conservation
planning. Agric. Handb. No. 537. Washington: U.S. Department of Agriculture. 58 p.
178
-------
Appendix
The following material was taken directly from MacDonald et al. 1991. This
material reviews the 29 parameters affected by forest management activities and
was developed for the Pacific Northwest. It is presented without modification.
The material has been reviewed by the Task Force and the general principles
apply to most situations in the rest of the country. However, the discussion of the
relationship of variables to beneficial uses may vary in other regions because of
differences in warmwater fisheries, wetlands, wildlife, and other uses.
The discussion of water-quality standards is general, and the user of methods in
this document should use State water-quality standards to assess BMP effective-
ness. Also, the States have established methods for sampling various param-
eters and these methods should be used whenever possible in evaluating BMP
effectiveness. This will result in data comparable with State water-quality criteria.
179
-------
-------
EPA 910/9-91-001
MONITORING GUIDELINES TO
EVALUATE EFFECTS OF FORESTRY
ACTIVITIES ON STREAMS IN THE
PACIFIC NORTHWEST AND ALASKA
LEE H. MACDONALD
WITH
ALAN W. SMART AND ROBERT C. WISSMAR
These Guidelines were developed for Region 10, U.S. Environmental Protection Agency,
Seattle, Washington, under EPA Assistance No. CX-816031-01-0
with the
Center for Streamside Studies in Forestry, Fisheries & Wildlife
College of Forest Resources/College of Ocean and Fishery Sciences
University of Washington
Seattle, Washington
1991
CSS/EPA
-------
This report has been reviewed by the Nonpoint Source Section, Water Division, Region 10, EPA, Seattle, WA and
approved for copying and dissemination. Approval does not signify that the contents necessarily reflect the views
and policies of the Environmental Protection Agency. Any trade names or product names mentioned in this
publication do not imply endorsement by the authors or the sponsoring institutions.
Library of Congress Catalog Card Number 91-73312
Monitoring guidelines.
Additional copies of this publication may be obtained from the U.S. Environmental Protection Agency, Region 10,
NFS Section, WD-139,1200 Sixth Ave., Seattle, WA 98101. Copies of the expert system may be obtained by
sending a diskette formatted in MS-DOS to the same address.
-------
PART II
REVIEW OF MONITORING PARAMETERS
-------
-------
1- INTRODUCTION
1.1 PURPOSE AND USE OF PART II
Part I provided guidance on the design of monitoring
projects and the selection of monitoring parameters. The
selection of monitoring parameters was presented as a
function of the designated uses, management activities, and
monitoring costs. The importance of other factors, such as
geology, soils, and climate, was acknowledged, but these
site-specific factors must be considered on a project-by-
project basis and could not be incorporated into the tables
developed in Chapter 4. An interactive, PC-based expert
system, based on essentially the same selection process in
Tables 2-4 in Part I (pages 39,41 and 43, respectively), has
been developed and is available from EPA's regional office
in Seattle (Part I, Section 5.2). '
Both the tables in Chapter 4 and the expert system result
in a set of recommended parameters (Section 5.1 in Part I).
The recommended parameters can be characterized as those
parameters that are (1) relatively sensitive to the various
management activities listed in Table 5 (pages 50-51), (2)
closely related to the most common designated uses of water
in the Pacific Northwest and Alaska, and (3) cost-efficient
The expert system is both more flexible and more specific
than Table 5,in that the user can simultaneously selectmulu'ple
management activities and designated uses, and specify the
frequency of monitoring, access during high flows, and
allowable costs of data collection and analysis. This addi-
tional information results in a list of recommended param-
eters that is more directly applicable to particular situations.
At the end of the initial selection process, the recom-
mended monitoring parameters—either from Table 5 or the
expert system—must then be evaluated individually and
collectively to determine which parameters should be in-
corporated into the monitoring project This "final" selec-
tion of monitoring parameters must draw upon professional
judgment and consider the availability of existing data. (As
discussed in Chapter 2 of Part I, the development of an
effective monitoring project is an iterative process, and the
initial selection may need to be modified as data and
experience are accumulated.)
Often it will not be desirable or cost-effective to monitor
all the parameters suggested by the tables or the expert
system. Many of the 30 monitoring parameters are closely
related, and in such cases an explicit choice should be made.
Some of the more common pairs or groups of parameters
that may overlap are suspended sediment and turbidity;
nitrogen and phosphorus; channel characteristics (e.g., pool
parameters and thalweg profile); and bank stability and
riparian canopy opening. The potential overlap between
parameters was a major rationale for preparing Tables 6a
and 6b (pages 62-65). These tables qualitatively evaluate
how changes in each parameter affect all the other param-
eters. Hence Tables 6a and 6b provide one means to both
help determine the extent of the interactions between two
parameters, and obtain a preliminary indication of the
possible redundancy between any two parameters.
One can argue that no two parameters are completely
redundant, and the inherent problems and uncertainties
associated with collecting, analyzing, and interpreting field
data mean that all the parameters recommended in Table 5
or by the expert system should be used. While this argument
has some validity, it does not recognize the very real con-
straints of time and money. Clearly more data from more
parameters will facilitate understanding and a more precise
evaluation of changes in the stream channel and in water
quality, but no monitoring project is free of cost constraints.
Hence there is a need to balance idealized data needs with
real-world constraints of personnel time and external costs.
The review of individual monitoring parameters that
comprises Part II is a second and more comprehensive
means to facilitate the selection of the most appropriate
monitoring parameters. Although reading the section on
each recommended parameter requires more effort on the
part of the user, the additional information should lead to a
-------
Part II
more informed and better decision.
In addition to furthering theparameter selection process,
ascGondpurposeofPartllistosummarizecurrentknowledge
on each parameter. This should help the reader understand
the rationale, possibilities, and constraints for monitoring
each of the 30 parameters reviewed. In order to facilitate the
use of the Guidelines as a quick reference document, the
review of each parameter is divided into seven subsections:
(1) definition, (2) effects on designated uses, (3) response to
management activities, (4) measurement concepts, (5)
standards, (6) current uses, and (7) assessment. The last
subsection for each parameter—Assessment—is a rela-
tively brief, qualitative evaluation of the potential role of the
parameter in monitoring. Hence the Assessment section can
be read on its own as a summary of each parameter, and the
reader then can refer back to the other subsections as needed
for more information. Extensive references are provided in
each of the first six subsections in order to direct the reader
to key studies and more in-depth sources on any particular
topic.
The parameter reviews comprising Part II do not detail
field techniques and analytic procedures, as inclusion of this
material was beyond the scope of the project and would
greatly increase the size of the Guidelines. Instead, the
Measurement Concepts subsection outlines someof the key
considerations associated with measuring particular pa-
rameters, such as spatial and temporal variability, and the
types of measurements that might be made within the more
broadly definedparameters such as fish orriparian vegetation.
Again the references cited will direct the reader to more
detailed sources of information.
1.2 SELECTION AND ORGANIZATION OF THE
PARAMETERS IN PART II
The 30 parameters are grouped into six categories
(chapters):
1. physical and chemical constituents,
2. flow,
3. sediment,
4. channel characteristics,
5. riparian, and
6. aquatic organisms.
Each chapter includes reviews of 2 to 10 parameters that
may vary widely in scope. Fish, for example, are considered
within one section (i.e., as one parameter), even though there
are many possible measurements which could be used in
monitoring projects (e.g., species diversity, productivity,
density, etc.). On the other hand, the 10 different parameters
within the chapter on channel characteristics are much more
narrowly defined.
There are two main reasons why parameters are in-
cluded and grouped in what may appear to be an arbitrary or
uneven manner. First, the Guidelines emphasize those
monitoring parameters which are less known. It did not
seem productive to duplicate the extensive literature on the
more common and obvious water quality monitoring pa-
rameters, such as the chemical and physical characteristics
of water. Second, the Guidelines focus on those parameters
thatappeartohaveconsiderablepotentialformonitoringthe
effects of forestry activities on streams, but which are not yet
widely utilized. There is a strong and natural tendency to
monitor those parameters with which one is familiar, and
part of the rationale for these Guidelines is to take a fresh
look at the entire range of monitoring parameters.
For many of the less well-known parameters, the poten-
tial for monitoring still needs to be rigorously evaluated.
Often there is strong theoretical and practical justification,
but relatively little experience or data to validate the use of
a particular parameter for monitoring. This is the case for
many of the channel characteristics, and the development of
biological criteria is only now being addressed by the states.
To a certain extent the differing emphasis on the various
parameters reviewed in Part II reflects our attempt to an-
ticipate future trends in water quality monitoring in forested
areas. As more data are collected, modifications to the
ranking and evaluation of different parameters will be
necessary. Embeddedness (Section 5.6.2.) is a good example
of a parameter that has undergone a rapid evolution over the
past 5 years, and which is beginning to be more widely
applied even though its usefulness and measurement tech-
niques are still being debated.
Given the 1-year time frame for preparing these
Guidelines, often it was not possible to review all the studies
pertaining to a particular parameter. There also was a need
to keep the individual review sections brief enough to be
easily accessible but comprehensive enough to present the
key elements. Inevitably there will be some dissent over the
coverage or evaluation of a particular parameter, but it is our
hope that such feelings will be channeled into a critical
review of one's own experience and values, and that a
perusal of Part II will lead to an improved understanding and
formulation of water quality monitoring efforts in forested
areas.
-------
2. PHYSICAL AND CHEMICAL CONSTITUENTS
INTRODUCTION
The physical properties and chemical constituents of
water traditionally have served as the primary means for
monitoring and evaluating water quality. Parameters such
as pH, dissolved oxygen, conductivity, alkalinity, nitrite-
nitrogen, and biochemical oxygen demand are most com-
monly measured, and this is due both to their sensitivity to
municipal and industrial pollution, and their importance in
aquatic ecosystems. However, these same parameters may
not be as useful in forested areas because of differences in
metypeofpollution,theratesofchemicalandphysicalprocesses
within the stream, and the designated uses of the water body.
The water column parameters included in these
Guidelines were selected because (1) they are sensitive to
forest management activities and can be related to the des-
ignated uses, or (2) they are commonly monitored in forest
streams. A number of other physical properties and chemi-
cal constituents could help characterize water quality, and
thereby facilitate an understanding of the aquatic system,
but the focus of the Guidelines is on parameters useful for
monitoring the effects of forestry activities on streams.
Temperature is akey parameter that can be significantly
altered as aresult of timber harvest immediately adjacent to
the stream channel. Increases in peak summer water tem-
peratures can directly affect coldwater fishes. Nitrogen and
phosphorus are often limiting in aquatic ecosystems, and
there are several means by which forest management activi-
ties—including forest fertilization—can increase nitrogen
or phosphorus concentrations. Dissolved oxygen is another
parameter that.is critical to the health of aquatic ecosys-
tems, but for a variety of reasons intergravel dissolved
oxygen is more likely to serve as a useful parameter for
monitoring the effects of forestry activities. Herbicide and
pesticide concentrations generally need to be monitored
when these chemicals are applied because of their potential
effects on non-target organisms.
For each of these parameters, one or more surrogates
can be monitored. The width of the riparian canopy opening
(Section 6.1) is an important control on the amount of
incoming solar radiation, and incoming solar radiation can
be used to predict stream temperatures. Algal production
(Section 7.2) may be related to nitrogen or phosphorus
concentrations, while intergravel dissolved oxygen may be
reducedby high levels of suspended sedimentor fine bedload
(Chapter 4). The point is that instead of directly measuring
the parameter of interest, one can choose to monitor either
those parameters which act as controlling factors, or those
parameters which are sensitive to changes in the parameter
of interest.
Conductivity and pH are included primarily because
they are sooften included in water qualitymonitoringprojects.
Both parameters are important indicators of the chemical
and physical status of water, but they generally are much
less sensitive to forest management activities than the other
parameters mentioned above. They also are rarely limiting
to the primary designated uses. Direct monitoring of pH and
conductivity is important when other issues, such as acid
precipitation, or other management activities, such as hard-
rock mining, are of concern.
2.1 TEMPERATURE
Definition
Water temperature is an easily measured parameter that
has considerable chemical and biological significance. It is
measured on a linear scale in either degrees Fahrenheit (°F)
or degrees Celsius (°C). Celsius is increasingly preferred
and can be obtained easily from °F by the equation:
°C = 5/9 (°F - 32)
-------
Part II
Stream temperatures are the net result of a variety of
energy transfer processes, including radiation inputs,
evaporation, convection, conduction, and advection (Brown,
1983). Stream temperaturesreflectboth the seasonal change
in net radiation and the daily changes in air temperature.
These patterns of energy inputs and outputs are modified by
stream characteristics such as the flow velocity, flow depth,
and groundwater inflow. Typically peakdaily temperatures
occur in the late afternoon, and daily minima occur just
before dawn. The seasonal pattern of stream temperatures
generally is similar to the pattern of incoming solar radia-
tion, but with a lag of 1 to 2 months (Beschta et al., 1987).
Relation to Designated Uses
Increased water temperatures are known to increase
biological activity. A rough rule of thumb is that a 10°C
increase in temperature will double the metabolic rate of
cold-blooded organisms (Keeton, 1967). Salmonideggand
alevin development, and subsequent timing of emergence
from gravel, have been shown to be closely associated with
stream temperatures (Alderdice and Velsen, 1978). A rise
in summertime water temperature resulting from forest
harvest may increase the growth rate and productivity of
many aquatic organisms (Beschta et al., 1987).
The optimal temperature range for most salmonid spe-
cies is approximately 12-14°C. Lethal levels for adult
salmonids will vary according to factors such as the accli-
mation temperature and the duration of the temperature
increase, but they generally are in the range of 20-25°C.
Salmonid eggs and juveniles are much more sensitive to
high temperatures. Combs (1965) found the lethal limit of
sockeye salmon eggs to be 13.5°C. Spawning coho and
steelhead may be intolerant of temperatures above 10°C
(Beschta etal., 1987).
Acute effects of high temperatures on fish have been
well documented in laboratory studies, but little informa-
tion is available on the long-term exposure of salmonids to
sub-lethal temperatures. Similarly, the sub-lethal effects of
altered thermal regimes due to forest harvest have seldom
been documented for salmonid species. Recent studies by
Holtby (1988) andBerman andQuinn (1990) are beginning
to address these sub-lethal effects.
Stream temperature also can affect the behavior of
aquatic organisms, but these behavioral effects generally
are poorly understood or have been documented for only a
few species. For example, attemperatures below about 59C,
juvenile salmonids tend to move into the gravel or other
protected areas. This behavioral thermoregulation allows
salmon and other fish to minimize body temperature fluc-
tuations despite wide variations in stream temperatures
(Coutant, 1969).
Temperature controls the rate of many chemical reac-
tions. A general rule of thumb is that the rate of a chemical
reaction proceeding at room temperature will double with a
10°Cincreaseintemperature(Eastman, 1970). Theequilib-
rium between ammonium and unionized ammonia, for ex-
ample, is highly dependent upon temperature and can have
a series of repercussions with regard to nitrogen cycling and
water quality (Section 2.5.1). In contrast, the equilibrium
concentration of dissolved carbon dioxide and oxygen in
water is inversely proportional to water temperature (Sec-
tions 2.2 and 2.4, respectively).
Response to Management Activities
In many areas of the Pacific Northwest and Alaska, the
forest cover provides substantial shade to streams and other
water bodies. A reduction in the forest cover along streams
can increase the incident solar radiation and hence peak
summer stream temperatures. Complete removal of the forest
canopy in the Pacific Northwest has been shown to increase
the highest daily stream temperatures in the summer by 3-
8°C, although daily summer minima are increased by only
1-2°C (Beschta et al., 1987).
These temperature increases are due almost entirely to
the additional input of incoming shortwave radiation. Hence
elevated stream temperatures may not return to pre-logging
levels until the stream banks become revegetated and the
input of shortwave radiation has been reduced to pre-
logging levels (Moring 1975; Holtby, 1988). The thermal
energy in streams is not easily lost through reradiation,
convection, advection, and conduction. This means that
increases in stream temperature generally are additive, and
an alternation of shaded and unshaded reaches is not an
effective strategy to minimize increased summer tempera-
tures due to forest harvest (Beschta et al., 1987).
Removal of the forest canopy may decrease the mini-
mum nighttime temperature in winter by allowing more
radiation heat loss. In coastal areas this possible effect is
likely to be minimal, but in colder locations clearing the
riparian zone may cause increased incidence of anchor ice
or freeze-up (Beschta et al., 1987). The largest changes in
winter minima will occur in small, shallow, slow-flowing
streams that do not have significant groundwater inflow.
Although the greatest effect of forest harvest is on sum-
mer maxima, smaller temperature changes in other seasons
can have greater biological significance. On Carnation
Creek in coastal British Columbia, for example, coho smolt
numbers, size, and migration were affected more by small
changes in late winter and spring temperatures than by the
larger changes in summer temperatures (Hartman et al.,
1987). Both this research and recent models indicate that
alterations in stream temperatures can have a series of
complex, interacting effects that we are only beginning to
unravel for single-species systems. Holtby (1988) and
Holtby et al. (1989) reported that habitat changes, like
temperature elevation, can affect more than one life history
stage and persist throughout the life cycle.
-------
CHAPTER 2. PHYSICAL AND CHEMICAL CONSTITUENTS
Measurement Concepts
Temperature can be measured by either a thermometer
or an electronic sensor. Thermometers are relatively inex-
pensive but should be calibrated if accurate measurements
(e.g., within 1°C) are required. Inexpensive thermometers
may have measurement errors as large as 3°C (APHA,
1976). Electronic sensors have the advantage of allowing
continuous monitoring.
To obtain average stream temperatures, measurements
should be made in more turbulent reaches. Water tempera-
tures near the bottom of pools can be 5- 10°C cooler than the
surface water (e.g., Bilby, 1984a). Usually thermal varia-
tions within a stream result from inflows of cool water
sources, such as groundwater or intergravel water, into
slow-moving reaches, pools, or backwater areas. In such
cases a single surface temperature can be misleading. The
daily fluctuations in stream water temperature also must be
considered if instantaneous rather than continuous tempera-
ture measurements are being made.
Standards
EPA has established a general national criteria for
coldwater fisheries. This states that the weekly average
warm season temperature should (1) meet site-specific
requirements for successful migration, spawning, egg incu-
bation, fry rearing, and other reproductive functions of
important species; (2) preserve normal species diversity or
prevent appearance of nuisance organisms; and (3) not
exceed a value more than one-third of the difference be-
tween the optimum and the lethal temperature for sensitive
species (EPA, 1986b). Specific temperature standards to
satisfy these criteria are left to the individual states.
Many aquatic organisms respond more to the magnitude
of temperature variations and amount of time spent at a
particular temperature than to an average value. For this
reason temperature criteria should not only specify the
maximum allowable increase in the weekly average, but
also the maximum increase for shorter periods of time.
Current Uses
The dependence of stream temperatures on energy
transfer processes suggests that changes in water tempera-
tures due to forest harvest can be modeled and predicted.
For reaches <1000 m in length, the change in maximum
daily temperature can be predicted from the change in
incoming direct solar radiation. The change in shading can
be determined by evaluating the change in angular canopy
density, and the procedure for doing this is discussed in
detail by Brown (1983). This methodology also provides a
basis for determining the width of a buffer strip needed to
minimize changes in peak summer temperatures. Predic-
tion of the change in stream temperatures due to partial
removal of the streamside canopy is considerably more
difficult .
The predictability of temperature increases due to forest
harvest has recently led to the development of a model
intended to be used for forest management purposes in
Washington. The close relationship between mean stream
and air temperatures is used as the core of a heat transfer
model. Other factors, such as the riparian canopy, stream
depth, and groundwater inflow, are incorporated as factors
that affect heat inputs and outputs. The balance of these
factors determines stream temperature and can permit the
prediction of temperature patterns at the basin scale (Adams
and Sullivan, 1988). Physical models that project changes
in stream temperature due to management activities can
then be used to evaluate the potential effects on fish, other
aquatic organisms, and designated uses such as recreation.
At present, however, the width and canopy cover of
buffer strips usually is fixed. Actual measurements rather
than a model are used to determine whether a change in
temperature has occurred. Temperature is often included in
monitoring projects because it is relatively easy and inex-
pensive to monitor, and there is a widespread awareness of
the lethal effects of high temperatures on coldwater fisher-
ies. The scanty but increasing evidence for sublethal effects
suggests that temperature monitoring should not be limited
to those situations where forest harvest and other manage-
ment activities are likely to result in near-lethal temperature
increases.
Assessment
Measurement of summerand winter water temperatures
is a useful approach to assessing the thermal suitability of a
stream for fish. In contrast to most of the Other parameters
discussed in these Guidelines, temperature monitoring is
relatively straightforward and inexpensive. In turbulent
forest streams that are well shaded by riparian vegetation,
relatively few measurements may be required because of the
limited spatial and temporal variability. In pools and
backwater areas, however, additional measurements may
be necessary to determine whether these areas experience
thermal stratification or are subject to cooUwater inputs.
Similarly, the timing and frequency of temperature mea-
surements should be determined only after data have been
collected on the diurnal fluctuations in temperature and the
sensitivity of daily peak stream temperatures to short-term
fluctuations in air temperature. Use of continuous recording
devices eliminates the sampling problems caused by tempo-
ral variability.
Beschta et al. (1987) concluded that logging-related
temperature increases generally have not resulted in signifi-
cant mortality of resident salmonids. However, research has
suggested that a variety of sub-lethal adverse effects may
occur as a result of forest harvest, and this suggests that
continued efforts to monitor stream temperature changes
-------
Part II
may be desirable. The difficulty is not in monitoring these
changes, but in predicting the biological effects in complex
ecosystems.
Similarly, the postulated decline in nighttime winter
stream temperatures due to forest harvest have not been
verified. Although the magnitude of the change is likely to
be relatively small in most cases, it may have important
implications for stream icing in colder locations.
The additive nature of temperature increases and the
likely importance of sub-lethal effects suggest that monitor-
ing is needed when (1) the potential exists for large changes
in water temperatures due to management activities, (2)
water temperatures already are in the upper range of the
acceptable temperatures, and (3) there is a potential for
significant temperature increases due to the additive effects
of numerous smaller increases. Care also needs to be taken
in distinguishing temperature effects on aquatic organisms
from other changes due to opening up the forest canopy such
as increased light, increased nutrients, greater primary pro-
ductivity, and alterations in the amount of large woody
debris.
2.2 pH
Definition
pH is defined as the concentration of hydrogen ions in
water in moles per liter (moles L-1). Because the range of
hydrogen ion concentrations in water can range over 14
orders of magnitude, pH is defined on a logarithmic scale
as:
pH=logl/[H+] = -log[H+]
where [H+] refers to the concentration of hydrogen ions in
moles L-1.
For practical purposes the parameter of interest is not
the absolute concentration of hydrogen ions, but the chemi-
cal activity of those ions. In very dilute solutions the activity
and concentration of hydrogen ions may be nearly equal, but
this is less true as other ions are introduced into the sample.
The common measurement techniques for pH are based on
hydrogen ion activity, and do not directly measure hydrogen
ion concentration.
Hydrogen ion activity varies with temperature, but it is
not a simple linear relationship. At 24°C pure water has a
hydrogen ion activity of 1 x 10"7 moles L-1, so its pH is 7.0.
Decreasing the temperature to 0°C decreases the hydrogen
ion activity and increases the pH to 7.5. Increasing the
temperature of pure water to 60°C increases the hydrogen
ion activity and decreases the pH to 6.5 (APHA, 1980).
Solutions with a higher ion hydrogen activity than pure
water at 24°C have a pH <7.0 and are termed acidic.
Solutions with less hydrogen ion activity have a higher
pH and are called alkaline or basic.
It is important to understand that alkalinity and acidic
factors refer not to the pH, but rather to the ability of a
solution to neutralize acids and bases, respectively (Stumm
and Morgan, 1981). In many cases both alkalinity and pH
must be measured to properly evaluate changes in water
chemistry due to natural events (e.g., erosion, variations in
discharge) and human activities. In natural waters alkalinity
is produced by anions or weak acids that are fully dissoci-
ated above a pH of 4.5. Methods for measuring alkalinity
can be found in standard reference texts such as APHA
(1989).
The most important buffering system in natural waters
involves the dissolution of carbon dioxide (CC>2). Com-
pared to most other atmospheric gases, carbon dioxide is
relatively soluble, and in solution it combines with water to
form carbonic acid (HaCOa). The equilibrium between car-
bonic acid and its component ions (H+ and HCOs") depends
on the water temperature as well as the type and concentra-
tion of other ions. This carbonate system play s a critical role
in water chemistry, and it is the presence of the dissociated
carbonic acid that causes the equilibrium pH of pure water
in contact with the atmosphere to be mildly acidic (approxi-
mately 5.7 pH units) (Hem, 1970).
pH generally shows a weak inverse relationship to
discharge. At higher discharges rainfall or snowmelt is
rapidly converted into runoff, and this reduces the concen-
tration of base minerals. At low flows the incorporation of
more dissolved materials tends to increase pH (e.g., Aumen
etal., 1989).
Relation to Designated Uses
pH can have direct and indirect effects on stream water
chemistry and the biota of aquatic ecosystems. A pH range
from 5 to 9 is not directly toxic to fish, but a decline in pH
from 6.5 to 5.0 resulted in a progressive reduction in
salmonid egg production and hatching success (EPA, 1986b).
The emergence of certain aquatic insects also declines
below a pH of 6.5. From this and other data, EPA has
concluded that pH should range between 6.5 and 9.0 in order
to protect aquatic life (EPA, 1986b).
Indirect effects of pH on stream chemistry result from
the hydrogen ion activity and the interactions between pH
and a variety of other chemical equilibria. For example, at
5°C the equilibrium concentration of un-ionized ammonia
can increase tenfold with a change in pH from 6.5 to 7.5
(Section 2.5.1). Similarly, the solubility of many metal
compounds changes greatly with pH, and this is of critical
importance in areas with high levels of heavy metals in
bottom sediments. Carbonic acid in cool, CO2-saturated
streams can stimulate a wide range of weathering reactions,
and this will affect the aqueous concentration of a number
of dissolved ions (Reynolds and Johnson, 1972).
-------
CHAPTER 2. PHYSICAL AND CHEMICAL CONSTITUENTS
Response to Management Activities
Rigorous studies assessing the effects of forest manage-
ment activities on pH are surprisingly scarce. The available
data indicate that pH is not sensitive to most forest manage-
ment activities. In two small watersheds in northwestern
Oregon, for example, anion and cation concentrations ex-
hibited virtually no change as a result of partial clearcutting
and broadcast burning. Two-thirds of the total anionic load
was due to the dissolution of carbon dioxide from the
atmosphere (Harr andFredriksen, 1988). In many cases the
buffering capacity of the soil ensures that activities such as
forest harvest, forest fertilization, and road building do not
affect stream pH (e.g., Stottlemeyer, 1987).
Forest management activities can indirectly af fectpH in
several different ways. The introduction of large amounts of
bark and other organic debris, for example, can influence pH
by increasing the concentration of organic acids, increasing
oxygen demand, and increasing CC>2 inputs due to respira-
tion (Peters et al., 1976). The stimulation of primary pro-
duction by increased light or nutrient loading can increase
the diurnal variation in pH. Changes in the timing and
volume of runoff (Section 3) can have a minor effect on pH.
Erosion increases the concentration of dissolved solids and
may alter pH, but conductivity (Section 2.3) and alkalinity
are much more sensitive measures of this effect than pH.
Hard rock mining is the management activity most
likely to substantially alter the pH of streams and lakes
(Kunkleetal., 1987). Thevariationinminingandextraction
methods makes generalization difficult, but highly acidic
water is most likely to emanate from mine tailings and
settling ponds. A reduction in pH exacerbates the problems
associated with heavy metals by increasing their solubility
and hence their mobility and rate of biologic uptake. Other
types of mining, such as quarries, may alter pH if they
increase the exposure of certain rock types, such as lime-
stone, to weathering (Kunkle et al., 1987).
Measurement Concepts
pHcan be measured either colorimetrically or electroni-
cally. Since colorimetric methods are subject to interfer-
ence from turbidity, color, colloidal matter, oxidants, and
reductants, they are suitableonly for rough estimates (APHA,
1976). Usually pH is measured electronically with a pah- of
electrodes. One electrode is a constant-potential electrode
(e.g., calomel or silver-silver chloride), and the indicating
electrode usually is glass because it is relatively free from
interference (APHA, 1989).
The variation of pH with temperature and carbon diox-
ide concentrations means that measurements should be
made in the field immediately after taking the water
sample. For accurate readings .the pH meter must be
temperature-compensated, and the sample should be thor-
oughly mixed between readings. The temperature of the
samplealso needs toberecorded, as the equilibrium concen-
trations of the different ions are temperature-dependent, and
the pH meter cannot compensate for temperature-related
shifts in the chemistry of the sample. The pH meter and
electrodes must be regularly calibrated using solutions of
known pH. In general, readings generally should be consid-
ered accurate only to the nearest 0.1 pHunit(APHA, 1989),
and often should be assumed no more accurate than 0.5 pH
units.
Accurate measurement of pH is particularly difficult in
many forested areas because of the very low concentrations
of dissolved solids. To obtain reliable data, the following
points must be considered. First, the electrodes must be
designed to function in waters with a low specific conduc-
tance. Second, pH electrodes tend to react more slowly in
very dilute solutions, so a longer period of time is needed to
obtain a stable reading. Third, waters with a very low
concentration of dissolved solids (e.g., <50 fiS) should not
be stirred while readings are being taken because the stirring
creates a streaming potential. Finally, the pH meter should
be calibrated in standard solutions with a low concentration
of dissolved solids. Calibration of the electrodes in buffer
solutions of high ionic strength can lead to false readings.
The known equilibrium of carbon dioxide in water means
that distilled water saturated with air can be used to check or
calibrate pH measurements in water with a very low ionic
strength (S. McKenzie, U.S. Geological'Survey, pers.
comm.).
Water bodies with high algal growth can exhibit consid-
erable variation in pH over a 24-hr period. Maximum pH
values usually occur in the afternoon when photosynthetic
activity consumes CO2 and dissolved oxygen concentra-
tions are at a maximum. Minimum pH values are observed
at night when carbon dioxide is being released by algal
respiration. In some cases it may be possible to use this
diurnal variation in pH to estimate primary production.
Standards
EPA has set apHrange of 5.0-9.0 as the national criteria
for domestic water supplies. A pH range of 6.5 to 9.0 has
been established as the criterianecessary to protectfreshwa-
ter aquatic life (EPA, 1986b).
Current Use
pH is included in many water quality monitoring pro-
grams because it is well recognized and perceived as easy to
measure. Often, however, less data is available forpH than
for other water quality constituents because it must be
measured in the field immediately after taking the water
sample.
Probably the most intensive program to monitor pH is
the ongoing effort to assess the prevalence and effects of wet
(e.g., rain, snow, and fog) and dry acid deposition. High-
-------
Part II
elevation areas are of particular concern because they typi-
cally have thinner soils and less buffering capacity.
Regular monitoring of pH also is being conducted in
conjunction with mining operations. This includes not only
current operations, but also old tailings where an increase in
acidity could adversely affect drinking water quality, fish-
cries, and the use of water for irrigation.
Assessment
The hydrogen ion activity or pH of a stream is an
important water quality parameter. pH affects a wide variety
of chemical reactions. pH levels above 9.0 and below 6.5
have an adverse effect on some life cycle stages of certain
salmonids and aquatic macroinvertebrates. pHis of particu-
lar concern in areas contaminated with heavy metals, as a
decline in pH can greatly increase their mobility.
pH generally is not sensitive to forest management
activities. Hard rock mining is the management activity
which is most likely to affect pH in aquatic systems. Aban-
doned mine tailings and certain other types of mining also
can affect pH, and the intensity of monitoring will depend
upon factors such as the type of rock being mined or
disturbed, the designated water uses, and the amount of
drainage from the mine or spoils.
Monitoring pH in forested areas may necessitate special
procedures and equipment because most surface waters
have very low concentrations of dissolved solids. Failure to
acknowledge these special considerations can easily lead to
unreliable data. Synoptic measurements can indicate the
spatial variability of pH. Some of the differences between
streams can be related back to physical factors such as
climate and geology.
Temporal variation can occur on different scales. Diur-
nal variation is often due to primary production, while
monthly and seasonal variation results from factors such as
fractionation during snowmelt, changes in runoff processes,
and changes in atmospheric deposition. The potential
linkage between pH and discharge means that simultaneous
flow measurements are needed for thorough data analysis.
2.3 CONDUCTIVITY
Definition
Conductivity (or specific conductance) refers to the
ability of a substance to conduct an electric current The
conductivity of a water sample is a function of the water
temperature and the concentration of dissolved ions. Con-
ductivity may not be directly proportional to the concentra-
tion of dissolved ions, as ion mobility, ionic charge, and
ionic concentrations may affect conductivity in anon-linear
manner (APHA, 1976). The relationship between conduc-
tivity and temperature also is slightly non-linear, as the
dissociation constants of different ions vary with tempera-
ture. For dilute solutions, a 1°C increase in temperature
increases conductivity by approximately 2% (Hem, 1970).
Conductance is the inverse of resistance and is mea-
sured in the reciprocal of ohms, or mhos. Conductivity is
measured in terms of conductance per unit length, or mhos/
cm. These units are too large for most natural waters, so the
usual unitis|imhos/cm, where ^(imhosisequal to 1 mhos.
Conductivity may also be reported in millisiemens/meter,
with 1 millisiemen/m equal to 0.1 nmhos/cm (APHA, 1976).
Pure water not in contact with the atmosphere has a
conductivity of approximately 0.05 ^mhos/cm. Normal
distilled or deionized water has a conductivity of at least 1.0
(imho/cm, and this is largely due to the dissolution of carbon
dioxide in water (Section 2.2). Melted snow in the western
UnitedStateshasaconductivityof2to42|jmhos/cm(Hem,
1970). The range for potable water in the U.S. is 30 to 1500
jimhos/cm. The conductivity of streams emanating from
forested areas in the Pacific Northwest almost always falls
at the low end of that range (e.g., Aumen et al., 1989).
Relation to Designated Uses
Conductivity is an indication of the number of dissolved
ions in the water. This makes it very useful for quickly
assessing the quality of water for irrigation or water supply
purposes, and for monitoring the total concentration of dis-
solved ions in wastewaters. Often a linear relationship can be
established between conductivity and the major ionic species.
Using conductivity as a surrogate for other ions can reduce
the amount of laboratory work needed to characterize a
sample and facilitate continuous monitoring (Hem, 1970).
Conductivity is atleastas useful as total dissolved solids
(TDS) for assessing the effect of diverse ions on chemical
equilibria, corrosion rates, etc. In most cases TDS in milli-
grams per liter can be estimated by multiplying conductivity
by an empirical factor. For natural waters this conversion
factor ranges from 0.54 to 0.96, with most values falling
between 0.55 and 0.75 (Hem, 1970).
For natural waters in the Pacific Northwest and Alaska,
conductivity has no apparent effect on the designated uses
of water. Conductivity is most likely to pose problems for
irrigation and water supply purposes in downstream reaches
subject to withdrawals of high quality water and inputs of
poor quality return flows from agriculture and industry. The
relative insensitivity of aquatic biota to conductivity is
illustrated by the absence of an EPA-recommended criteria
(EPA, 1986b).
Response to Management Activities
In the Carnation Creek study in southwestern British
Columbia, conductivities increased in the sub-catchment,
which was intensively logged and burned, and in the main
-------
CHAPTER 2. PHYSICAL AND CHEMICAL CONSTITUENTS
stem by a maximum of 90% and 50%, respectively. These
increases were restricted to higher flows in the first 2 years
after logging (Scrivener, 1988). In absolute terms, the
increase in conductivity for high flow events from 20 to 40
Hmhos/cm was well below the range of 50-120 nmhos
observed during moderate and low flows. Other studies
have found that forest harvest caused little or no change in
the concentration of some of the major ions which contrib-
ute to conductivity, but they did not report on changes in
conductivity per se (Brown etal., 1973; HarrandFredriksen,
1988).
The management activity in forested areas which is
most likely to affect conductivity is hard rock mining.
Kunkle et al. (1987) recommend using conductivity as one
of the key indicators of water quality. A substantial change
in conductivity, or unusually high values, should spur more
detailed analyses.
Measurement Concepts
Conductivity often is measured with temperature-com-
pensated electrodes mounted to maintain a fixed distance
between them. As with any electrodes, proper maintenance
and calibration are essential for accurate measurements.
APHA (1980) reports that measurements made by a trained
operator should be within 1 % of the true value, but tests of
unknown samples resulted in a relative standard deviation
of nearly 10%. Because conductivity is very sensitive to
water temperature, the sample temperature should be re-
corded along with the conductance, or the conductance
should be corrected to reflect a standard temperature such as
25°C.
Usually there is an inverse relationship between con-
ductivity and discharge (e.g., Keller et al., 1986; Aumen et
al., 1989). Water that is slowly transmitted to the stream
(baseflow) has more opportunities to pick up dissolved ions
through weathering and other chemical reactions. Water
that is quickly transformed from precipitation to runoff
(quickflow) tends to have fewer dissolved ions, thus causing
a corresponding decline in conductivity at high discharges.
This relationship between conductivity and discharge means
that simultaneous discharge measurements are needed to
properly interpret conductivity data.
Standards
No standards for conductivity have been established or
proposed.
Current Uses
Conductivity is often included in water quality monitor-
ing projects, but its use in forested areas needs to be further
evaluated. In areas where conductivity between surface and
groundwater differs significantly, a change in conductivity
can be a sensitive indicator of groundwater seepage into
stream channels. Conductivity is an excellent indicator of
the total concentration of dissolved ions and thus can be a
very useful indicator of mining impacts or agricultural water
quality (Kunkle et al., 1987).
In forested areas the ions of primary concern, such as
nitrates and dissolved plant-available phosphorus, gener-
ally are present in such low concentrations that they do not
make a substantial contribution to the electrical conductivity.
This makes it difficult to use conductivity as a surrogate for
the more costly analyses of nitrogen and phosphorus.
Conductivity data can help characterize overall stream
chemistry. Such data are particularly useful for interpreting
pH measurements when both pH and conductivity are
controlled by dissolved inorganic ions (e.g., in the bicarbon-
ate-type waters that dominate in the Pacific Northwest).
The relationship between pH and conductivity may be quite
different in waters with high concentrations of dissolved
organic matter and low concentrations of major ions
(Wissmar etal., 1990).
Assessment
At the levels commonly found in forested areas, electri-
cal conductivity alone has little or no direct effects on
aquatic life. Conductivity is essentially a sum of the con-
ductances of all-the individual ionic species, so the signifi-
cance of a change in conductivity depends on which ions
were responsible for that change.
Forest activities can affect certain nutrients, such as
nitrogen and phosphorus (Section 2.5), but these are rela-
tively minor components of the total conductivity and
generally should be measured separately. Forest activities
also can affect specific, conductance by altering rates of
erosion and mineralization, and this proportionally increases
the concentration of some of the major cations and anions
(Wetzel 1975). Conductivity also can be increased by the
extensive use of deicing salts or dust-reduction compounds.
Although the effects of these ions on the aquatic system are
believed to be negligible at the concentrations usually
observed, this isone situation whereconductivity monitoring
may be appropriate.
The primary value of systematic conductivity measure-
ments in forested areas is to help classify streams within a
particular region or to compare streams from different
regions. Such data collection efforts fall into the category of
baseline monitoring and are quite distinct from monitoring
to assess the effects of forest management.
Conductivity can be a useful parameter for monitoring
mining impacts. It is easily measured and can serve as a
surrogate for total dissolved solids or some of the major
ions. If a change in conductivity is detected, more specific
measurements of individual ions will be needed to deter-
mine the specific cause and predict the potential effect.
-------
Part II
2.4 DISSOLVED OXYGEN
Definition
Dissolved oxygen concentration refers to the amount of
oxygen dissolved in water. Oxygen is a sparingly soluble
gas and its concentration in water is usually measured in
ppm or mg Is1. The capacity of water to hold oxygen in
solution (dissolved oxygen saturation) is inversely propor-
tional to the water temperature. Increased water tempera-
ture lowers the concentration of dissolved oxygen at satura-
tion (i.e., equilibrium with the atmosphere).1
The actual concentration of dissolved oxygen (DO) in
water depends not only on the saturation concentration but
also on oxygen sinks and sources. The primary oxygen
sinks are respiration and the biochemical oxygen demand
(BOD) of substances in the water. Major oxygen sources
include photosynthesis and the dissolution of atmospheric
oxygen in water as oxygen levels are depleted (reaeration).
Higher water temperatures not only depress the concentra-
tion of dissolved oxygen in water at saturation, but also
increase the rate of BOD. In general, most forest streams
have cool temperatures, rapid reaeration rates, and rela-
tively low oxygen demand; thus stream water normally is
close to or at saturation. Situations in which stream water
may notbe near saturation include: very slow, low-gradient
streams where the rate of reaeration is low; sites where fresh
organic debris causes a large BOD; warm eutrophic streams
where high levels of photosynthesis and respiration cause
diurnal fluctuations in dissolved oxygen; and ponded sites
such as those formed by beavers.
DO concentrations also can vary between the surface
stream water and the water flowing through alluvial mate-
rials in the stream bed. DO within these alluvial materials
is termed intergravel dissolved oxygen or intergravel DO.
Oxygen replenishment to these intergravel waters comes
primarily from the exchange of well-aerated surface waters
with oxygen-impoverished intergravel waters. The impor-
tance of this oxygen exchange between surface and inter-
gravel waters is a primary reason why the clogging of
gravels with fines is of such concern.
Intergravel DO is controlled by the same factors as
surface water, but there is no photosynthesis or reaeration.
Oxygen demand comes from the fine organic debris en-
trained in the gravels and from the respiration of organisms
living within the alluvial interstices. In spawning streams
the tens or hundreds of thousands offish eggs also can exert
a measurable oxygen demand. Groundwater usually has a
low concentration of DO, and areas with substantial ground-
water seepage are likely to have lower concentrations of
intergravel DO. For these reasons the DO concentration
typically is lower within the streambed than in the adjacent
stream water.
Relation to Designated Uses
DO is critical to the biological community in the stream
and to the breakdown of organic material. Table 7 summa-
rizes the biological effects of different DO concentrations in
salmonid and non-salmonid waters. In salmonid streams,
intergravel DO should be near saturation, or at least above
minimum concentrations, to ensure normal growth and
survival of eggs and alevin (Chapman and McLeod, 1987).
As indicated in Table 7, high DO levels in streams and
intergravel areas also are needed to sustain the more sensitive
macroinvertebrates (EPA, 1986a).
Intergravel DO has been used as a surrogate for the
amount of interstitial fines and as an indication of the
suitability of streambed gravels for fish spawning. Note,
however, that fish can greatly modify spawning site condi-
tions, particularly the amount of interstitial fines, through
the redd building process. Monitoring sites must be care-
fully selected to represent the actual DO concentrations
that the fish eggs will experience (Chapman and McLeod,
1987).
Response to Management Activities
The Alsea watershed study in coastal Oregon indicated
that heavy inputs of fine, fresh organic material, when
combined with sedimentation, reduced reaeration, and in-
creased water temperature, could severely deplete DO in
small forest streams (Hall and Lantz, 1969; Wringler and
Hall 1975). Subsequent research has shown that the charac-
teristically high turbulence of forest streams rapidly replen-
ishes DO (Ice, 1978). Current forestmanagement techniques
in the Pacific Northwest normally do not introduce large
amounts of fine organic material into streams (Skaugset and
Ice, 1989).
Low DO in streams is most commonly associated with
major point sources such as pulp mills or municipal waste
treatment facilities. Only a few examples of depressed DO
related to forest management are available. In one Canadian
study, for example, a stream with a slope gradient of <1%
was loaded with logging debris of sufficient size and quan-
tity to impound the stream. The fresh slash and low reaera-
tion rate for this stream caused the DO concentration to
drop to zero (Plamondon et al., 1982). Present logging prac-
tices and the increased protection for the major stream chan-
1Thc amount of oxygen that can dissolve in water increases with increasing atmospheric pressure. Dissolved oxygen saturation values (C) for
different water temperatures are reported for 1 atm barometric pressure (760 mm Hg). A close approximation of actual saturation value at any
pressure can be made using the equation: Cp=C x (p/760) where Cp is the saturation concentration at atmospheric pressure p (mm Hg). Generality
this pressure correction can be ignored, but it may be important for some high elevation sites. The relationship between gas solubility and pressure
also may be important when effects of dam spill-ways on supersaturation disease in fish are considered.
-------
CHAPTER 2. PHYSICAL AND CHEMICAL CONSTITUENTS
Table 7. Biologic effects of decreasing dissolved oxygen (DO)
levels on salmonids, non-salmonid fish, and aquatic invertebrates.
The instream values for embryo and larval stages of salmonids
were obtained by assuming that a difference of 3 mg L-1 between
intergravel and instream DO would adequately maintain DO levels
within the gravel (EPA, 1986a).
~~~ Dissolved
oxygen (mg L-1)
Inter-
Instream gravel
8
6
5
4
3
I. Salmonid waters
A. Embryo and larval stages
No production impairment 11
Slight production impairment 9
Moderate production impairment 8
Severe production impairment 7
Limit to avoid acute mortality 6
B. Other life stages
No production impairment 8
Slight production impairment 6
Moderate production impairment 5
Severe production impairment 4
Limit to avoid acute mortality 3
II. Non-salmonid waters
A. Early life stages
No production impairment 6.5
Slight production impairment 5.5
Moderate production impairment 5
Severe production impairment 4.5
Limit to avoid acute mortality 4
B. Other life stages
No production impairment 6
Slight production impairment 5
Moderate production impairment 4
Severe production impairment 3.5
Limit to avoid acute mortality 3
III. Invertebrates
No production impairment 8
Some production impairment 5
Limit to avoid acute mortality 4
nels suggest that management-induced depletion of DO in
stream water will occur only under unusual circumstances.
Forest management activities are more likely to affect
intergravel DO through the increase in fine sediment Everest
et al. (1987) recently reviewed the linkage between fine
sediment, management activities, and aquatic organisms,
but provided little data on DO. An extensive review of the
oxygen requirements of aquatic organisms is found in and
Chapman and McLeod (1987) and EPA (1986b),T>ut these
do not relate changes in intergravel DO to management
activities. Hence thecause-and-effectchain of management
activities increasing fine sediment, which then decreases
gravel permeability and decreases DO, must be largely
inferred. Nevertheless, the linkage is sufficiently strong that
Idaho has proposed intergravel DO as a sediment criteria
(Harvey, 1989), and EPA has incorporated intergravel DO
values into their criteria for DO (EPA, 1986a).
Measurement Concepts
Either chemical or potentiometric methods can be used
to measure DO (APHA, 1989). The standard chemical
method, known as the Winkler method, is based upon the
oxidation of manganese, the liberation of iodine in propor-
tion to the DO present in the sample, and then the titration
of the iodine with thiosulfate. The Winkler method is very
accurate provided there is no interference from suspended
solids, other oxidizing agents, or certain organic com-
pounds. Modified methods exist to reduce or eliminate each
of these potential problems (APHA, 1989). The standard
deviation of measurements using a standard or modified
Winkler method is between 0.02 to 0.1 mg L-1. Since the
titration can not be performed in situ, it is important that the
sample be collected in a manner that minimizes disturbance
and gas exchange. Designs for DO sample collection
devices are available (APHA, 1976).
Electrical (potentiometric) methods are based on the
rate of diffusion of dissolved (molecular) oxygen across a
membrane, and the resulting generation of an electrical
signal. The measurement of DO by membrane electrodes is
affected by both temperature and salinity, but nearly all
commercially available electrodes have built-in thermistors
for temperature compensation. Salinity generally is not a
problem in forested areas but may need to be considered in
estuaries or in streams where return-flows from irrigation
result in a high concentration of dissolved solids. Provided
the electrodes are properly maintained and calibrated, the
potentiometric method is sufficiently accurate for nearly all
field monitoring projects (accuracy of approximately ±0.1,
mg L-1; precision of ±0.05 mg L-1) (APHA, 1989). These
considerations, together with the fact that measurements
can be made in situ, make potentiometric methods the pre-
ferred field technique. ,
The timing of the measurement can be important. Dur-
ing the day, warming of the stream water can depress the
saturation concentration for DO and accelerate the rate of
oxygen uptake. For slow-moving streams and rivers with
high primary productivity, large diurnal fluctuations in DO
concentration can result from algal photosynthesis and
respiration. During the day photosynthesis in excess of
respiration is a source of oxygen. At night photosynthesis
ceases and respiration becomes an oxygen sink. The rela-
tive importance of the various,oxygen sources and sinks
must be evaluated when designing a monitoring project.
-------
Part II
The same techniques are used to measure intergravel
DO, but the collection of a representative water sample
presents sample collection problems. Typically there is a
great deal of spatial variability, and there is always a
question as to whether one should sample from within a
redd, which is mostdirectly applicable to salmonid survival,
or from a "representative" riffle, run, or pool. Moring
(1975) found intergravel DO concentrations to vary from 4
to over 9 mg L-1 on the same day at different locations in a
small, undisturbed coastal stream.
Usually intergravel water samples are obtained by plac-
ing a standpipe into the gravel some weeks or months prior
to the sampling (Hoffman, 1986; Moring, 1975). Once the
standpipe has been installed, a siphon can be used to remove
water samples, or measurements can be made in situ using
potentiometric methods. Skaugset (1980) used a syringe
technique to rapidly extract water samples with minimal
disturbance to the streambed. Two of the key principles
associated with the collection of intergravel water samples
are (1) minimize disturbance and gas exchange for the
sample being collected, and (2) avoid disturbance to the
streambed, which causes increased or decreased mixing of
intergravel waters with surface waters.
Standards
Standards can either be absolute (mg L-1) or expressed
as a percent of saturation. Recent EPA reports discuss both
the biological effects of reduced DO (EPA, 1986a) and
summarize the existing state and national criteria (EPA,
1988a). The more stringent criteria are applied to those
waters containing a salmonid fish population, and these
state that the 1-day minimum and a 7-day mean DO con-
centration should beS.O and 9.5 mgL-1, respectively. These
criteria are based on the assumption that intergravel DO.is
about 3 mg L-1 less than the DO concentration in surface
water, and this makes the 1-day minimum and 7-day mean
intcrgravel DO concentration 5.0 and 6.5 mg L-1, respec-
tively. Less stringent criteria apply if only adult salmonids
are present (EPA, 1986a).
State standards for DO concentrations in surface water
have been established in Alaska, Idaho, Oregon, and Wash-
ington, and these vary according to the location and desig-
nated use. For basins in western Oregon with salmonids
"fresh waters shall not be less than 90 percent of saturation
at the seasonal low, or less than 95 percent of saturation in
spawning areas during spawning, incubation, hatching, and
fry stages of salmonid fish." For western Oregon basins
with non-salmonids "dissolved oxygen concentration shall
not be less than 6 mg L-1." Some states, particularly Idaho,
are considering the promulgation of a water quality criteria
for intergravel DO.
Current Uses
Concern about DO is justified primarily in situations
where (1) water flow is low and temperature is high; (2) the
rate of energy dissipation, which accelerates reaeration, is
low; and (3) oxygen sinks are high. Some examples of
where this can occur are (1) slow-moving, warm streams
and rivers; (2) off-channel habitat (where there is a low
exchange rate for water); (3) ponded sites where water flow
is slow; (4) large lakes where there is extensive log transport
or storage; and (5) spawning areas and alluvial channels
where the gravels are subject to high rates of sedimentation.
The conditions that cause streams to be sensitive to
management impacts also make streams sensitive to natural
inputs of organic material. For example, autumn leaf fall
from red alder has caused a 6 mg L-1 oxygen deficit in a
small stream in coastal Oregon (Ice, 1991).
Reductions in intergravel DO are directly related to the
rate of intergravel respiration and the permeability of the
gravel. Gravel permeability is sensitive to the amount of
fine sediment, and this linkage has led the state of Idaho to
propose intergravel DO in artificial redds as a sediment
criteria in Idaho's water quality standards (Harvey, 1988).
The implementation of this criteria may prove difficult
because of the problems in defining key factors such as the
location and size of the gravel to be used in the artificial
redd. Research has shown that small changes in bed
topography and gravel permeability can greatly alter the
susceptibili ty of aredd to siltation (Cooper, 1965; Chapman
and McLeod, 1987).
Assessment
Dissolved oxygen (DO) is another parameter that is
easily measured and often included in monitoring efforts.
While it is critical for sustaining fish and invertebrates, DO
concentrations in streams are rarely a limiting factor. Forest
management and harvesting activities that avoid the intro-
duction of fresh slash into streams generally do not generate
a sufficient instream oxygen demand to deplete stream
oxygen concentrations. Similarly, forest practices that
minimize temperature increases will help maximize abso-
lute concentrations of DO. Conditions that contribute to a
reduced concentration of DO include low flows, warm
temperatures, shallow stream gradients, fresh organic mat-
ter inputs, and high respiration rates. The presence of one or
more of these factors should signify a possible need to
monitor DO concentrations within the water column.
Intergravel DO is more sensitive to management activi-
ties and hence potentially more useful as a monitoring
technique. At least in gravel-bedded streams, any increase
in the amount of fine particles is likely to adversely affect the
subsurface permeability of the streambed. This reduces the
rate of exchange between the intergravel and surface waters.
In the absence of any change in oxygen demand in the
-------
CHAPTER 2. PHYSICAL AND CHEMICAL CONSTITUENTS
intergravel layer, the reduced exchange could result in a
decline in intergravel DO concentrations. Numerous stud-
ies have shown that even a small decline in DO concentra-
tions in the intergravel layer can adversely affect the repro-
ductive success of salmonid species and the viability of
other aquatic organisms. Although intergravel DO criteria
have been established, very few projects have utilized
intergravel DO as a monitoring parameter. This reluctance
to use intergravel DO stems partly from the uncertainty
regarding sample locations, and partly from the difficulty in
developing acceptable sampling techniques. Nevertheless,
the importance of intergravel dissolved oxygen to stream
health suggests that further testing and development of this
parameter is warranted.
2.5 NUTRIENTS
2.5.1 NITROGEN
Definition
Nitrogen in aquatic ecosystems can be partitioned into
dissolved and paniculate nitrogen. Most water quality
monitoring programs focus on dissolved nitrogen, as this is
much more readily available for both biological uptake and
chemical transformations. Both dissolved and paniculate
nitrogen can be separated into inorganic and organic com-
ponents. The primary inorganic forms of nitrogen are
ammonium (NH4+), nitrate (NO3-), and nitrite (NO2-).
Under certain conditions un-ionized ammonia (NH3) also
can be present.
In terrestrial ecosystems most of the soil nitrogen is
associated with organic matter and is relatively immobile.
Mineralization of the organic nitrogen usually converts it to
ammonia (NH3+). Ammonium (NH4+) is the soluble form,
and it can be taken up by plants, lost through leaching,
oxidized, or fixed by exchange reactions. Normally nitro-
gen does not persist in the soil in the ammonia form, as it is
oxidizedby microbes (nitrification) firstto nitrite (NO2-) and
then to nitrate (NO3-)- Although both nitrite and nitrate are
soluble and thus subject to leaching and biological uptake,
the nitrite form is relatively transient. In the undisturbed
forest ecosystems of the Pacific Northwest, nearly all of the
nitrate is converted into organic nitrogen by microorgan-
isms or plants, and this completes the basic terrestrial
nitrogen cycle. Losses of nitrate can occur by leaching, or
by microbial reduction (denitrification) to gaseous N2 if urea
is present or conditions are anoxic (Brady, 1974; Doelle,
1975). Most of the nitrogen losses from forests to streams
is in the form of nitrate (Vitousek et al., 1979), but these
losses are relatively small for most undisturbed forest eco-
systems (Cole, 1979; Triska et al., 1984).
In aquatic systems the inorganic forms of nitrogen
(NH4+, NO2-, and NO3-) are subject to many of the'same
transformations and processes as in terrestrial ecosystems
(Triska et al., 1984; Wissmar et al., 1987; Meyer et al.,
1988). Nitrate is the predominant form in unpolluted
waters. Un-ionized ammonia (NHs) also may be present as
an intermediate breakdown product of organic nitrogen,
fertilizers, and animal wastes. Predictions of un-ionized
ammonia concentrations are difficult because it is an inter-
mediate breakdown product, and it is in a non-linear tem-
perature- and pH-dependent equilibrium with ammonium'
(NH4+). Both ammonium and nitrate are readily taken up by
aquatic biota, so an increase in nitrate concentrations tends
to diminish rapidly in the downstream direction.
The riparian zone plays a critical role in nitrogen trans-
formations as both aerobic and anaerobic conditions are
present Recent research indicates that riparian zones are
important sites for denitrification (Green and Kauffman,
1989). Certain riparian plants, such as alder, can add
nitrogen to the system by fixing atmospheric nitrogen, and
this further complicates the interactions between the terres-
trial and aquatic nitrogen cycles.
Relation to Designated Uses
Certain nitrogen compounds have toxic effects at rela-
tively low aqueous concentrations. Nitrate has been linked
to methemoglobinemia (blue-baby) syndrome in human
infants at concentrations of 10 mg L-1 of nitrate-nitrogen
(EPA, 1986b). Nitrite also will react with hemoglobin, and
this can be hazardous for infants. Trout and salmon species
are not as sensitive to nitrates as human infants, but nitrite-
nitrogen concentrations as low as 0.5 mg L-1 havebeen shown
to be toxic to rainbow trout (EPA, 1986b).
Ammonia (NH3-) is toxic to some aquatic invertebrates
and fish at concentrations as low as 0.08 mg L-1, while
chronic effects occur at concentrations of only 0.002 mg
L-1 (EPA, 1986b). The toxicity of ammonia is affected by
other factors such as the concentration of dissolved oxygen,
temperature, pH, salinity, and the carbon dioxide-carbonic
acid equilibrium. The same factors affect aqueous NHs
concentrations by influencing the chemical equilibrium
between NH3 and NH4+.
Nitrogen is one of the most important nutrients in aquatic
systems. Most of the non-toxic effects of nitrogen result
from the fact that increased inorganic nitrogen stimulates
primary production (e.g., bacteria and algae) and possibly
secondary production (e.g., macroinvertebrates and fish).
However, few studies have documented an increase in
primary production due to the effects of forest management
on the aquatic nitrogen cycle. Studies that have attempted
to analyze these more subtle effects suggest that an increase
in plant-available nitrogen will increase primary productiv-
ity only if the algae are not limited by light or other nutrients
such as phosphorus (Bisson, 1982). Both Lyford and
-------
Part II
Gregory (1975) and Busch (1978) have found that nitrogen
enrichment in heavily-shaded streams in the western Cas-
cades did not enhance primary productivity. In contrast,
nitrogen can be limiting in large unshaded systems like the
McKenzie River in Oregon (Bothwell and Stockner, 1980).
The desirability of increased biotic production depends
on the local and downstream designated uses. For many
forest streams a small or moderate increase in primary pro-
duction might be considered beneficial as it is likely to
increase fish production. However, if plant respiration
Txjgins to deplete dissolved oxygen or results in unsightly
growth of aquatic plants, this probably would be considered
an adverse effect.
Increased nitrogen loading in lakes is potentially much
more serious than an increase in stream nitrogen because of
the potential accumulation of nutrients (Schindler et al.,
1976). Over time the accumulation of relatively small
nitrogen inputs may stimulate algal growth to the point
where eutrophication begins and the beneficial uses such as
recreation and fishing are impaired (Brown, 1988). This
scenario has been documented for lakes that had a variety of
nutrient inputs, but apparently not for lakes that have been
subjected only to forest management activities. Studies at
Lake Tahoe on the California-Nevada border, for example,
indicate that logging had relatively little impact as com-
pared to changes in land use (Goldman and Byron, 1986).
Response to Management Activities
Forestmanagementactivities can alter many parts of the
nitrogen cycle, and this makes it difficult to generalize about
the effects of logging, fire, erosion, and forest fertilization.
Logging affects stream nitrogen by introducing organic
material and sediment, and may also increase the inputs of
inorganic nitrogen. In coastal British Columbia, for ex-
ample, logging increased the concentration of nitrate in
Carnation Creek by a factor of 2, to a maximum of 0.15 mg
L-i (Scrivener, 1988). Clearcutting and burning the Needle
Branch catchment in coastal Oregon resulted in a fivefold
increase in nitrate concentrations. However, no increase in
nitrate was observed following patch cutting in the adjacent
Deer Creek catchment. Maximum values in Needle Branch,
Deer Creek, and the adjacent control stream were all about
3 mg L-1 of nitrate-nitrogen (Brown et al., 1983).
In the Bull Run watershed in Oregon, partial clearcut-
ting caused a fourfold increase in nitrate-nitrogen when the
slash was broadcast burned and a sixfold increase when the
slash was allowed to decompose naturally. Maximum values
followed the same pattern, with a high of 0.08 mg L-1 when
the slash was broadcast burned and 0.27 mg L-1 when the
slash was leftto decompose(HarrandFredriksen, 1988). In
the Carnation Creek, Needle Branch, and Bull Run studies,
nitrate-nitrogen concentrations returned to pre-logging lev-
els after approximately 5 years. A more recent study has
documented that, despite the relatively large increases in
nitrate-nitrogen following timber harvests, the total loss of
nitrogen is less than the annual input of nitrogen through
precipitation (Martin and Harr, 1989).
Relatively little data is available on the indirect lossesof
nitrogen associated with logging. Fredriksen (1971) found
that the amount of nitrogen lost in association with inor-
ganic sediment (i.e., erosion) was larger than the amount of
nitrogen lost in solution. Paniculate nitrogen and dissolved
organic nitrogen accounted for the majority of nitrogen lost
from the experimental Fox Creek watershed following
logging (Harr and Frederiksen, 1988). Generally the nitro-
gen associated with sediment (i.e., paniculate inorganic
nitrogen) is not readily available to the stream biota.
Fire also has a series of direct and indirect effects on the
terrestrial nitrogen cycle (Brown et al., 1973). In general,
the amount leached into the aquatic system following major
fires appears to be roughly comparable to the increases due
to logging (Wright, 1981). In north central Washington
only traces of nitrogen were lost through leaching after a
severe wildfire (Grier, 1975). A greater increase in aquatic
nitrogen concentrations might be expected if substantial
amounts of burned material enter the stream channel. In
other cases the largest source of nitrogen following a fire
could be due to increased erosion.
Plant-available nitrogen has been demonstrated to be
the limiting nutrient for forest productivity in the Pacific
Northwest (Gessel et al., 1979), and this has led to a number
of forest fertilization programs. Mostoftheseareon private
timberland, and virtually all programs apply pelletized urea
from the air at concentrations of around 200 kg/ha (Moore
and Norris, 1974, cited in Norris et al., 1983). The use of
pellets minimizes drift, so the delivery of fertilizer to the
aquatic ecosystem occurs either from direct application, or
transport by surface and subsurface runoff (Cline, 1973).
Organic nitrogen in the form of urea is subject to the various
nitrogen transformations, but it can also be lost as gaseous
N2 through denitrification and volatilization.
Fredriksen et al. (1975) summarized the results of
several studies that monitored water quality following the
application of urea. Ammonia, urea, and nitrate concentra-
tions each peaked within a couple of days after application,
although the nitrate showed a slight time lag as compared to
ammonia and urea.' These peaks result from the direct
application of the urea pellets into the stream channels, and
the relatively rapid transformation from urea to ammonia,
nitrite, and nitrate. With the advent of the rainy season, a
second nitrate peak was observed. The total amount of
nitrogen lost to theaquatic ecosystem was estimatedatO.5%
of the total applied (Fredriksen et al., 1975). More recent
studies have shown that a wet season application of urea can
result in a large, short-term increase in the concentration of
ammonia and total nitrogen, and smaller increases in nitrate
concentrations (Bisson, 1988). In more frequently fertilized
watersheds, the total losses ranged up to nearly 10% of the
amount applied (Bisson, 1982).
-------
CHAPTER 2. PHYSICAL AND CHEMICAL CONSTITUENTS
Some of the other ways in which humans increase
nitrogen concentrations in streams and lakes include inad-
equate human waste disposal, livestock, and atmospheric
fallout. Inadequate human waste disposal can result from
dispersed recreation, septic tanks, and municipal wastewa-
ter treatment plants. Both dispersed recreation and septic
tanks are considered nonpoint sources and require relatively
intensive wet season monitoring to determine their effect on
water quality. Municipal wastewaterplants are pointsources
and therefore easier to monitor. The problem is that small
rural wastewater treatment plants often cannot afford the
additional treatment necessary to remove most of the nutri-
ents, and must rely on dilution to minimize adverse effects.
These considerations are taken in account when point dis-
charge (NPDES) permits are written.
Measurement Concepts
Methods for measuring the concentration of the differ-
ent nitrogen compounds in water are well known and
detailed elsewhere (APHA, 1989; Stednick, 1991). An
important step is to determine which nitrogen species are of
most interest and to identify the measurement technique
most appropriate to those species. Kjeldahl nitrogen com-
bines both organic nitrogen and total ammonia. Total
ammonia includes both ionized (NH4+)andun-ionizedforms
(NHs). Dissolved nitrite and nitrate are often combined, as
the concentration of nitrite in forested streams generally is
very small. Dissolved organic nitrogen, can be obtained
from the difference between Kjeldahl nitrogen and total
ammonia. Adding Kjeldahl nitrogen to dissolved nitrate
and nitrite yields total dissolved nitrogen.
, Attention also must be given to the method of express-
ing concentrations of the various nitrogen species. For
example, a concentration of 10 mgL-1 of nitrate includes the
weight of both the nitrogen and the oxygen atoms in the
nitrate molecule, while a concentration of 10 mg L-1 of ni-
trate-nitrogen refers only to the mass of elemental nitrogen
present as nitrate. The difference in the molecular weight of
nitrate and nitrogen means that 10.00 mg L-1 of nitrate is
only 2.26 mg L'1 of nitrate-nitrogen.
A distinction should be made between monitoring for
water quality standards and monitoring to estimate total
load. Monitoring for water quality standards is primarily a
matter of taking samples at the times and locations where
peak concentrations are expected to occur. Simultaneous
discharge data is necessary for proper interpretation of the
data, but not for determining whether standards are being
met Monitoring for total load requires monitoring of total
dissolved and paniculate nitrogen and continuous discharge
measurements. As with any total load calculation, it is
critical to adequately sample the high flows when the bulk
of the nitrogen is being transported past the monitoring
station.
Standards
The national drinking water standard for nitrate-nitro-
gen is 10 mg L-1 (EPA, 1987). A standard for nitrite-ni-
trogen has not been established because nitrite is such a
transientform. Water bodies with highnitrite concentrations
are likely to be highly polluted and not meet existing
standards for other constituents such as bacterial contami-
nation and dissolved oxygen (EPA, 1986b).
For ammonia, national criteria have been established to
prevent "unacceptable" effects on freshwater organisms
and their uses (EPA, 1986b). The dynamic equilibrium of
ammonia with other chemical species is calculated for 1-hr
and 4-day mean concentrations using formulas based on pH,
temperature, and the presence or absence of salmonid spe-
cies. These formulas are applicable for a temperature range
of 0-30°C and a pH range of 6.5-9.0 (EPA, 1986b). Table
8 lists total ammonia concentrations that correspond to an
un-ionized ammonia concentration of 0.020 mg L-1 for a
range of common temperature and pH values (from Bisson,
1982, adapted from Thurston et al., 1974). Thew data
indicate that the proportion of un-ionized ammonia is ex-
tremely sensitive to pH and less sensitive to temperature.
Although no national standards have been established,
Cline (1973) indicated that a nitrate concentration of <0.3
mg L-1 would probably prevent eutrophication. In basins
that have been designated as impaired, strict limitations on
the total nitrogen load may be imposed (Parti, Section 1.4).
Current Uses
Many water quality monitoring programs regularly
measure concentrations of one or more species of nitrogen.'
In undisturbed basins these data provide a baseline for com-
parison and an indication of long-term trends. In actively
managed forested basins, forest harvest disrupts the terres-
trial nitrogen cycle by increasing the amount of decompos-
ing organic material, reducing root uptake, and changing
the soil moisture regime. This can greatly increase the
concentration of dissolved inorganic nitrogen—primarily
nitrates and ammonium—in the stream. In many cases,
however, increased leaching of nitrogen to the stream will
be attenuated or completely obscured as a result of increased
Table 8. Equilibrium concentration of un-ionized ammonia in
mg L-1 as a function of temperature and pH.
Temperature (°C)
pH
6.5
7.0
7.5
5
10
15
20
26
51
34
23
16
11
16
11
7.3
5.1
3.6
5.1
3.4
2.3
1.6
1.1
cation exchange reactions (Brady, 1974). Hence the princi-
pal means by which humans can increase phosphate levels
in aquatic systems is by altering rates of erosion and organic
matter inputs.
(Brown et al., 1973; Harr and Fredriksen, 1988). Water
quality monitoring in the Carnation Creek study in south-
western British Columbia demonstrated that phosphate
concentrations were unrelated to streamflow, season, or
logging. Higher phosphate ion concentrations were ob-
served after fires, but maximum concentrations of
phosphate-phosphorus werestill
-------
Part II
uptake by aquatic plants. The complexity and interactions
of the terrestrial and aquatic nitrogen cycles must be consid-
ered when attempting to relate a change in stream chemistry
to a particular management activity (Meyer et al., 1988).
Monitoring of nitrogen species can be useful when
undertaking a forest fertilization program. Past studies sug-
validity of the water quality model, and changes due to
management actions.
Assessment
Almost any forest management activity will affect some
Part II
On the other hand, aerial application of herbicides and
pesticides can have serious implications for non-target
organisms. Herbicides, for example, are usually directed at
broad-leaved species, and many riparian species are suscep-
tible to, if not the target of, the commonly used herbicides.
Killing the riparian vegetation can have a wide range of
secondary effects, including destabilizing the stream chan-
nel, reducing the input of both fine organic material and
large woody debris, and increasing stream temperatures.
Similarly, the aerial spraying of pesticides can adversely
affect the riparian fauna, and this can reduce the availability
of terrestrial insects for fish populations.
Response to Management Activities
Herbicides, pesticides, and their intermediate break-
down products are only present as a result of human at-
tempts to control unwanted vegetation or animal pests.
Ground-based programs greatly reduce the likelihood of
direct contamination of streams arid lakes provided proper
care is taken in the transport, mixing, application, and
disposal of the herbicides and pesticides.
For aerial applications the use of spray buffer, strips
along the stream channels greatly reduces the exposure of
the riparian zones and the stream channels. For spray buffer
strips to be effective, careful consideration must be given to
factors such as the droplet size, height of application, wind
speed, and flight path. The requirement of spray buffer
strips along fish-bearing streams usually minimizes overt
damage to the riparian vegetation. In tributary channels or
along unprotected streams, however, herbicide use may kill
off the riparian vegetation and initiate a series of adverse
effects on aquatic organisms and the stream channel.
Measurement Concepts
The critical aspect in monitoring herbicide and pesticide
applications is the selection of monitoring locations and the
timing of the water samples. State forest practice regula-
tions generally haveestablishedprotocols for sampling, and
these represent a compromise between the need for compre-
hensive sampling and the costs of collection and analyses.
The usual procedure is to take one sample immediately
prior to application and a series of samples at various times
after application. Often an attempt is made to sample peak
concentrations by estimating the average velocity in the
stream and then using this to estimate when the peak con-
centration would occur at a sampling point 200-500 ft
downstream of the spray boundary. Some states specify that
samples are to be taken at specified times (e.g., 0.5,1.0 and
2.0 hr after the cessation of spraying). To minimize the costs
of analysis, some states allow a portion of each short-term
sampletobecombinedintoacompositesample.Thecomposite
sample is then analyzed, and the remaining portions of the
individual samples are tested only if the composite sample
exceeds some threshold.
Many states also have a procedure to qualitatively
evaluate the relative risk of the application to impair water
quality. The intensity of sampling is then adjusted to reflect
the estimated risk. The factors used to assess this risk
include the type of chemical (toxicity, persistence, and
mobility), the potential drift (this is a function of the slope,
slope length, irregularity of the landscape, the stream length
exposed to the application, the riparian cover, droplet size,
and the weather conditions), and the beneficial uses of the
stream (Oregon State Department of Forestry, 1979, in
Appendix A, NCASI, 1984a).
One approach to the problem of sample timing is to take
a 24-hr composite sample. The 24-hr composite sample is
effective as long as the subsampling interval is short enough
to adequately capture higher concentrations and the analytic
technique is sufficiently sensitive. Combining concentra-
tion data with discharge data allows the 24-hr mean concen-
. tration to be calculated. The 24-hr mean concentration is the
basis for many of the state standards for pesticide and
herbicide concentrations (EPA, 1977).
Another approach is to continuously sample the stream
using a trace enrichment cartridge (NCASI, 1984b). In this
method a constant flow of water is run through a cartridge
designed to capture the chemical of interest. Analysis of the
cartridge at the end of the sampling period, when combined
with discharge data, provides a mean concentration over the
entire sampling period. The technology is still being tested,
but concern exists that some of the subject chemical may be
slowly lost from the cartridge. Preliminary data indicate
that losses from trace enrichment cartridges are a function of
stream pH, the flow rate through the cartridge, and the
relative concentrations of the chemical in the stream and the
cartridge (G. Ice, Nat. Council for Air and Stream Improve-
ment, Corvallis, pers. comm.).
Another approach to assess overspray and drift is through
the use of spray cards or tracers. Spray cards are simply flat
cards which are set out prior to aerial spraying, and visual
inspection provides a qualitative indication of the amount of
chemicals that reached the surface of that particular site.
The short lag between application and observation means
that this method can be used as a near real-time monitoring
technique. Disadvantages include an increased exposure to
monitoring personnel and an indication of the total input,
rather than the maximum concentration, of chemicals into
the stream system.
Another approach is to mix a fluorescent dye with the
pesticide or herbicide and monitor dye concentrations in the
stream of interest. This allows direct, real-time monitoring
provided that a definable relationship exists between the
chemical of interest and the dye (NCASI, 1984a). Smart and
Laidlaw (1977) reviewed the use and measurement of fluo-
rescent dyes as hydrologic tracers, and they noted that a
variety of different factors can influence the measurements.
-------
CHAPTER 2. PHYSICAL AND CHEMICAL CONSTITUENTS
One of the few field studies to measure the aqueous
concentration of both a herbicide (2,4-D) and a fluorescent
dye (rhodamine WT) found that the dye peak was much
sharper and higher than the herbicide peak. This-differeiice
was attributed to greater sorption of the herbicide by the
organic material in the stream (NCASI, 1984a). Concerns
over the mutagenic activity of rhodamine WT may limit its
use as a marker for herbicides and pesticides (NCASI,
1984a).
Standards
EPA has recommended maximum allowable mean con-
centrations over a 24-hr period for silvicultural chemicals.
These concentrations vary according to the size of the
stream and the designated uses of the water body (EPA,
1977). The maximum allowable mean concentrations are
based on a combination of the acute toxicity as defined by
the LC-50 and a safety factor. (LC-50 refers to the concen-
tration at which 50% of the target organisms perish within
the testingperiod.) Recommended maxima range from one-
fifth to one one-hundred thousandth of the LC-50. Most
states have adopted standards based on the EPA recommen-
dations.
Current Uses
Considerable variation exists in the intensity and type of
water quality monitoring associated with the application of
herbicides and pesticides. Public agencies tend to test more
regularly, but they are more constrained with regard to the
aerial application of forest chemicals. Testing by private
industry depends upon state regulations and the perceived
risk.
In most cases the procedure is to take samples at a
location assumed to be well mixed and therefore represen-
tative of the entire stream cross-section. These data are
useful for (1) documenting the level of chemicals in the
stream system, and thereby limit future liability; and (2)
evaluating the effectiveness of the application techniques in
minimizingtheamountof chemicals releasedintotheaquatic
system.
The relative absence of articles documenting adverse
water quality effects suggests that current application pro-
cedures are effective in minimizing adverse impacts. Many
of the more toxic or persistent chemicals are no longer used,
and this significantly reduces the possible level of exposure.
Continuing attention must be paid to minimizing drift,
which can be achieved by using buffer strips and spray
delivery systems that generate appropriate droplet sizes, as
well as by spraying under low wind conditions (EPA, 1977).
The use of buffer strips along streams and lakes is the single
most effective means for minimizing both the direct and
indirect adverse effects of herbicides and pesticides on
water quality.
Assessment
Contamination of streams and lakes by herbicides and
pesticides is unlikely except in the case of accidental spills
or aerial application. Monitoring of aerially applied chemi- '
cals is sporadic, although some states have established
procedures to determine if water quality monitoring is
necessary. It may be questioned, however, whether the
typical monitoring procedures will achieve the overt moni-
toring objectives.
The first objective of monitoring—to document the
amount of unwanted chemicals entering the aquatic sys-
tem—is probably rarely achieved because of the temporal
variation in pesticide and herbicide concentrations. A few
grab or pump samples may or may not capture the peak
concentration. In the absence of information on the shape.of
the concentration curve over time, the reliability of the grab
samples is very low, and minor changes in the sampling
location or time could dramatically affect the observed
concentration.
The second use of monitoring data is to evaluate the
effectiveness of the application procedures and Best Man-
agement Practices in minimizing chemical inputs to the
aquatic system. This can be done only if (1) the sampling
was sufficient to determine that the concentrations did not
exceed some designated level, and (2) data are available to
document the conditions and methods of application. In
other words, downstream data only indicate whether there
was a problem. Identifying the cause of the contamination
will require data on all the factors that would have affected
the delivery of the herbicide or pesticide into the aquatic
system. Both types of data are needed to iteratively improve
the application procedures, but mostagency reportingforms
do not request sufficient information to carry out this kind
of evaluation.
Currently available information suggests that the use of
herbicides and pesticides in forested areas generally does
not adversely affect the designated uses of water. The
relative absence of adverse effects is at least partially due to
the infrequent use of these chemicals in forested areas as
compared to croplands. Most silvicultural prescriptions call
for no more than one or two applications of herbicides over
the entire rotation period of 60-120 years. Pesticides tend to
be applied only as need requires. Although aerial applica-
tions can adversely affect aquatic and riparian ecosystems,
the careful application of Best Management Practices and
buffer strips should minimize the impact on most water
bodies.
-------
3. CHANGES IN FLOW
INTRODUCTION
Changes in the size of peak flows, the discharge at low
flows, or annual water yield usually are not considered as
water quality parameters. Nevertheless, forestharvest, road
building, and other management activities can result in
substantial changes in the volume and timing of runoff, and
this has longbeen a source of public concern. Changes in the
size of peak flows can have important implications for the
stability of the stream channel, size and quantity of the bed
material, and sediment transport rates. An increase in low
flows generally will reduce peak summer temperatures and
increase the available fish habitat. Changes in water yield
typically are too small to be measured, but in high elevation
basins with extensive hydropower development the theo-
retical increase in water yield can have substantial economic
value. In some areas the evaluation of cumulative effects is
based largely on the estimated capability of the stream
channel to accommodate an increase in discharge.
Flowparameters were included in theGM/rfe/j/zey because
of theirpotential sensitivity to forest management activities,
their relationship to designated uses, and general public
concern. Even if a flow parameter is not explicitly included
in a monitoring project, discharge measurements are needed
to interpret other data, such as turbidity and conductivity,
and to calculate the total flux of nutrients, sediment, and
other materials being transported by streams.
In summary, the patterns and values of discharge are
important characteristics of forest streams, and they inte-
grate all the different effects of specific management
activities on the hydrologic cycle. Maintaining favorable
conditions of flow was an important justification for estab-
lishing the National Forest system, and this concern per-
sists to the present day. Forest management activities can
affect discharge through a variety of individual processes,
and this chapter reviews the three parameters of greatest
concern.
3.1 INCREASES IN THE SIZE
OF PEAK FLOWS
Definition
Peak flows refer to the instantaneous maximum dis-
charge associated with individual storm or snowmelt events.
The diversity of climates in EPA's Region 10 means that
peak flows can result from several different types of cli-
matic events. In the low-lying, coastal basins in the Pacific
Northwest, for example, winter rainfall is the primary cause
of peak flows. In many of the higher-elevation and interior
areas, peak flows are generated by spring snowmelt. Other
possible causes of peak flow events are summer thunder-
showers and rain-on-snow events. Both of these latter causes
may be less common and less predictable, but in certain
basins they may be responsible for the largest runoff events.
Many basins may be exposed to more than one cause of
peak flows. For example, spring snowmelt may generate
the peak discharge in most years for a given basin, but less
common rain-on-snow events may be responsible for the
largest discharge events. Prediction of the effects of forest
management on the size of peak flows is complicated by the
fact that forest management will have quite different effects
on the size of peak flows depending upon whether the peak
flows are caused by spring snowmelt, high-intensity rain
storms, or rain-on-snow events. The effect of forest harvest
and other management activities also will vary according to
factors such as the type of yarding (tractor or cable), the
density of skid trails and landings, soil type, and soil
moisture content. Prediction of the effect of management on
the size of peak flows therefore requires (1) knowledge of
the climatological events that cause the peak flows in the
basin of interest, (2) specification of the peak flows of
concern (e.g., the mean annual flood or more extreme events
such as the 50-year flood), and (3) specific knowledge on
-------
CHAPTER 3. CHANGES IN FLOW
how the management activities are likely to affect each of
the major components of the hydrologic cycle (interception,
infiltration, evapotranspiration, and snowmelt).
Relation to Designated Uses
Peak flows have important effects on stream channel
morphology and bed material particle size (Chapter 5).
Specifically, since higher flows move larger particles, peak
flows determine the stable particle size in the bed material
(Grant, 1987). Large, stable particles provide important
habitat niches for invertebrates and small fish. A highly
unstable bed will reduce periphyton and invertebrate pro-
duction (Hynes, 1970). The size of peak flows also is im-
portant in determining the stability of large woody debris
and the rate of bank erosion. Increased bank erosion and
channel migration will affect the riparian vegetation and
alter the amount of active sediment in the stream channel.
Periods of high flow also are periods of bank building and
deposition on active floodplains, especially in areas with
dense riparian vegetation.
The vast majority of the sediment transport occurs during
peak flows, as sediment transport capacity increases loga-
rithmically with discharge (Ritter, 1978; Garde and Ranga
Raju, 1985). The ability of the stream to transport the
incoming sediment will help determine whether there is
deposition or erosion within the active stream channel. The
relationship between sediment load and sediment transport
capacity will affect thedistribution of habitat types, channel
morphology, and bed material particle size (Chapter 5).
Increased size of peak flows due to urbanization have been
shown to cause rapid channel incision and severe decline in
fish habitat quality (Booth, 1990).
A change in the size of peak flows can have important
consequences for human life and property. Structures such
as bridges, dams, and levees are designed according to a
presumed distribution of peak flows. If the size of peak
flows is increased, this could reduce the factor of safety and
lead to more frequent and severe damage.
Response to Management Activities
Forest management activities can increase the size of
peak flows by a variety of mechanisms, and these include
the following:
1. road-building (due to both the impervious surface and
the interruption of subsurface lateral flow);
2. reduction of infiltration rates and soil moisture storage
capacity by compaction;
3. reduced rain and snow interception due to removal of
the forest canopy;
4. higher soil moisture levels due to the reduction of
evapotranspiration;
5. increased rate of snowmelt; and
6. any change in the timing of flows that results in a
synchronization of previously unsynchronized flows.
By the same logic indicated in item 6 above, forest harvest
may reduce the size of peak flows by desynchronizing
runoff peaks (Harr, 1989). Under certain conditions forest
harvest also can reduce the size of the smaller peak flows by
reducing fog drip, thereby reducing the amount of soil
moisture storage prior to some storm events.
Each of these mechanisms will have different effects in
different seasons and in storms of different magnitudes.
Sufficient care in the layout and execution of roads and
timber harvest will minimize the changes in the size of peak
flows from the first four runoff processes identified above.
Thus in the absence of rain-on-snow events, the most
dramatic changes in the size of peak flows are observed in
the smaller storms in autumn or early winter, when less
precipitation is needed to recharge soil moisture (e.g., Harr
et al., 1975; Ziemer, 1981). Forest management activities
can have a relatively negligible effect on the peak flows
associated with major floods if very little of the catchment
has been subjected to compaction or converted to an imper-
vious surface.
The effects of forest management on peak flow size are
quite different when the largest floods are caused by rain-
on-snow events. In these areas, forest management—by
increasing snowpack accumulations in openings and in-
creasing the rate of snowmelt in clearcuts and young plan-
tations (Berris and Harr, 1987)—can increase the size of
peak flows in major flood events.
The effects of forest management activities on the size
of peak flows have been studied in a number of paired
watershed experiments in the Pacific Northwest and else-
where (e.g., Harr, 1983; Bosch and Hewlett, 1982). In most
cases forest harvest has been found to increase the magni-
tude of peak flows, and this has been attributed to soil
disturbance reducing infiltration and subsurface stormflow
(Cheng etal., 1975), changes in short-term snow accumula-
tion and melt (Harr and McCorison, 1979), and soil com-
paction (Harr etal., 1979).
A few studies have shown no significant changes in the
frequency or magnitude of peak flows (Harr, 1980; Harr et
al., 1982; Wright etal., 1990). In one case the absence of an
increase in the size of peak flows was due at least in part to
a reduction in fog drip; one must also assume there was
minimal soil compaction and soil disturbance. The lesson
from these studies is that forest management can have a
variety of interacting hydrologic effects, and the sum of
these effects will determine whether an increase in the size
of peak flows is likely (Harr et al., 1982).
Measurement Concepts
Peak flows can be identified either by continuous mea-
surement of stage (water surface elevation) or by the use of
crest stage recorders. Usually stage is converted to dis-
-------
Part II
charge by periodically surveying the stream cross-section
and measuring stream velocity at various water surface
elevations. The calculated discharge is then plotted against
stage to obtain a rating curve (Buchanan and Somers, 1969).
The conversion of stage to discharge is needed in order
to establish a quantitative relationship between peak flows
in two or more basins. Changes in the size of peak flows can
then be detected by a change in this relationship. Direct
comparisons of stage heights between basins is not appro-
priate because the relationship between stage and discharge
is unique for each location and may change over time as the
channel erodes, aggrades, or shifts laterally.
The comparison of discharge from similar, adjacent
catchments is the most sensitive means to detect changes in
the size of peak flows. Usually at least 3 years of calibration
data are needed to establish a relationship capable of pre-
dicting about 70-85% of the variance in discharge. A pro-
portionally longer calibration period will be needed to es-
tablish a valid statistical relationship for peak flows with
longer recurrence intervals. The pre-disturbance discharge
relationship is then used to determine if there is a statisti-
cally significant change in discharge due to management
activities in one of the catchments.
An alternative to the paired-catchment approach is to
relate the stage or discharge at one location to precipitation,
and then assess how this rainfall-runoff relationship changes
with management. The difficulty with this technique is that
rainfall-runoff models are relatively crude, and the uncer-
tainty associated with rainfall-runoff model predictions
generally increases with increasing discharge. This uncer-
tainty then makes it very difficult to identify a change in the
size of peak flows due to management activities.
Direct measurement of peak flows can be obtained by
continuous measurements of water level or by crest-stage
recorders. Continuous measurement of discharge usually
requires constructing a stilling well and establishing a stage-
discharge relationship. This is relatively expensive and
requires a continuing inputof staff time to check on the stage
recorder, establish a stage-discharge relationship, and trans-
form the stage data to discharge.
Crest-stage recorders are much simpler, as they only
record the maximum water level. In the absence of a stage-
discharge relationship, the values may be difficult to inter-
pret, as changes in channel morphology can alter the observed
crest from events with identical peak discharges. Typical
crest stage recorders consist of vertical tubes containing
powdered cork. Small holes in the tube allow water to enter
and leave the crest gages, and a ring of powdered cork is left
at the highest water level occurring between observations.
A major problem in monitoring changes in the size of
peak flows is the infrequent nature of high flow events.
Hence sample sizes are small, and the capability to detect a
statistically significant change is low. For this reason most
research addressing changes in peak flows have focused on
runoff events that occur several times each year. Monitor-
ing changes in the size of peak flows associated with storms
with longer recurrence intervals is much more difficult. A
5-year storm, for example, only has a 20% chance of occur-
ring in a given year, and only a 67% chance of occurring
within a specified 5-year period. Hence a very long calibra-
tion period is needed for these rarer events, and the post-
harvest monitoring period is limited by the hydrologic
recovery of the site to pre-harvest conditions. For this
reason changes in the size of the larger peak flows generally
cannot be measured directly.
Monitoring changes in the size of peak flows is also
limited by the cost of establishing and maintaining stations
to measure peak discharges. Continuously recording gag-
ing stations are relatively costly. Discharge measurements
during high flow events require some access to the site and
a structure from which one can safely measure velocity.
Crest-stage recorders are relatively simple and inexpensive,
but they have a much lower sensitivity.
Standards
No standards for changes in the size of peak flows have
been established or proposed.
Current Uses
The difficulties in determining a change in the size of
peak flows means that this parameter is rarely included in
most monitoring projects. Nevertheless, potential changes
in the size of peak flows can be an important constraint to
forest management (Grant, 1987), particularly in areas sub-
ject to rain-on-snow events. Hence most environmental
assessments and other planning documents evaluate pro-
jected changes in the size of peak flows by extrapolating
from the limited number of paired-catchment experiments
that have examined the issue.
It is important to note that any change in the size of peak
flows is most likely to decline in magnitude as one moves
downstream. This is due to both a dispersion of the flood
wave in time and the lack of change in other tributaries (i.e.,
a dilution effect) (Linsley et al., 1982). Proportionally
larger increases in the size of peak flows will occur down-
stream only if the timing of peak runoff in the managed basin
is altered in such a way that it becomes synchronized with
peak runoff in other tributaries (Harr, 1989).
Assessment
Forest management activities can increase the size of
peak flows by transforming subsurface flow to surface
flow, reducing infiltration rates and soil moisture storage
capacity, reducing interception losses, increasing soil mois-
ture, and altering rates of snowmelt. The relative effects of
these changes will vary by season, site, and storm size.
Careful management and post-harvest rehabilitation mea-
-------
CHAPTER 3. CHANGES IN FLOW
sures can largely alleviate changes in the size of peak flows
due to compaction, disruption of subsurface flow paths, and
reduced infiltration rates. This means that in areas not
subject to rain-on-snow events, the largest change in the size
of peak flows can be limited to the firstfew storms following
the growing season, when thehighersoil moisture carryover
results in a greater proportion of runoff . Major floods, such
as those with a return interval of 50 years or more, should not
be as greatly affected by forest management activities, as the
total rainfall is normally sufficient to make up any initial
differences in soil moisture content. However, if forest
harvest and other management activities substantially in-
crease the amount of compacted or impervious areas (e.g.,
roads, landings, and skid trails), the size of peak flows from
all storms is likely to increase (Harr et al., 1979). ,
Forest harvest can increase the size of the largest peak
flows in areas where the largest floods are caused by rain-
on-snow events. This increase in the size of peak flows is
due to the combination of increased snowpack (caused by a
reduction in interception losses) and an increase in snowmelt
due to increased turbulent heat transfer. Recent research in
the Washington Cascades has indicated that harvested plots
can yield up to 95% more runoff than unharvested areas, and
runoff from 18- to 20-year-old plantations is around "'40%
higher (R.D. Harr, U.S.F.S. Pac. Northw. Res. Sta., Seattle,
pers. comm.).
In summary, the effects of forest harvest on the size of
peak flows is difficult to predict and measure. Providing
that soil disturbance and compaction are kept to a minimum,
concern over increases in the size of peak flows is appropri-
ate primarily in areas where rain-on-snow events generate
the largest flood peaks. Careful monitoring of changes in
the size of peak flows could help provide some insight into
the hydrologic behavior of a basin, but there are more direct
and efficient ways to monitor most of the physical effects
that lead to a change in peak flows.
Monitoring of changes in the size of peak flows is
difficult because it requires a long-term commitment and
the matching of the basin of interest to one with no land use
changes or management activities. Data from past studies
on small catchments indicate that monitoring the size of
peak flows provides little understanding unless it is accom-
panied by studies documenting the probable cause(s) of any
observed change. Hence, monitoring the size of peak flows
is more appropriate as part of an applied research project
than as a standard monitoring practice.
3.2 CHANGES IN Low FLOWS
Definition
In most of the western U.S., the minimum streamflow is
observed during the late summer and early autumn. This
decline in discharge is due to a combination of low precipi-
tation, reduced drainage from the soil and bedrock, and
sustained high evapotranspiration. Removal of the forest or
other vegetative cover usually results in an increase in low
flows by reducing evapotranspiration (e.g., Harr et al.,
1979) and secondarily, interception.
Relation to Designated Uses
Summer low flows are important primarily for main-
taining aquatic habitat. An increase in low flows will
increase the wetted perimeter and flow depth, and thereby
provide more habitat Increased flows will also reduce the
magnitude of any temperature increase due to forest harvest,
as temperature increases are highly dependent on the in-
crease in incoming net radiation relative to total discharge
(Section 2.1).
Response to Management Activities
In mostsmall catchment studies in thePacificNorthwest
forest harvest has been shown to increase summer low flows
by up to 300% (Anderson, 1963; Rothacher, 1970). Although
this is a large relative increase, the absolute volume of the
increase is small relative to the total annual water yield (Harr
et al., 1982). However, in areas where fog drip is a major
hydrologic input, forest harvest can cause a decline in
summer low flows (e.g., Harr, 1980). Studies in the drier,
snowmelt-dominated areas of the Rocky Mountains have
shown low flow increases of only 0-12% following forest
harvest (Bates and Henry, 1928; Troendle, 1983; Van
Haveren, 1988). The presence of a low flow increase in
these more arid environments may depend on whether
summer precipitation is sufficient to generate a response in
streamflow.
As forest regrowth occurs the increase in low flows is
diminished, and the rate at which low flows return to pre-
harvest conditions can be highly variable. In coastal Oregon
the harvest of a mature coniferous forest was followed by
the rapid establishment of phreatophytic vegetation (red
alder, cottonwoods, and willows) in and adjacent to the
stream channel. Within 10 years the measured summer low
flows showed no increase relative to pre-harvest conditions,
and in subsequent years the summer low flows were less
than predicted by the pre-harvest calibration equation. This
reduction in low flows can be expected to continue until the
phreatophytic vegetation is overtopped by the less water-
consumptive coniferous species (Harr, 1983). Hydrologic
recovery from thinning, understory removal, or burning of
brush also is likely to require less than a decade.
Measurement Concepts
As was the case for peak flows, the most sensitive means
for detecting a change in low flows is to establish a statistical
-------
Part II
relationship between the discharge of adjacent catchments.
A change in the relationship between the two catchments is
used to demonstrate a change in low flows. The need to
accurately measure relatively small discharges means the
gaging stations must be carefully placed to minimize seep-
age, and the width-depth ratio should be as low as possible.
In small streams some type of weir or flume structure is
likely to be needed to obtain the necessary accuracy.
Changes in low flows generally will be more difficult to
detect in larger catchments because a smaller proportion of
the catchment will be harvested over a relatively short time
period. Hence any increase in low flows will be subject to
a dilution effect from other sub-catchments which do not
have a hydrologically altered vegetation canopy.
Standards
No standards for changes in low flows have been
established or proposed.
Current Uses
Monitoring stream discharge is an important compo-
ncntof most water quality monitoring programs. However,
low flows are relatively unimportant in terms of their
contribution to constituent load, sediment load, and water
yield. Paired-catchment experiments have shown that 20-
30% of a catchment must be cleared to obtain a measurable
increase in water yield (Bosch and Hewlett, 1982). Since
most long-term gaging stations are on larger catchments that
do not experiencesuch heavy harvestlevelsoverarelatively
short time period, changes in low flows are unlikely to be
observed at existing gaging stations.
Little attention has been paid to monitoring changes in
low flows because there is very little scope for management.
Removal of the riparian vegetation usually is not a viable
option because of concerns over wildlife and fisheries
habitats, sediment and nutrient inputs, bank erosion, and
stream temperatures (Section 6.2). Forest harvest is known
to decrease evapotranspiration, and some of this water will
be expressed as an increase in streamflow, but we have very
limited control over the amount and timing of this increase.
Although this increase in low flows may be significant in
terms of increased habitat area—particularly in small
streams—on larger streams the increase generally is too
small to be measured. For these reasons most monitoring
projects do not explicitly attempt to document any change
in low flows.
Assessment
Forest harvest can cause a substantial increase in sum-
mer low flows, and this will provide additional habitat for
stream biota. Increased low flows also may reduce the
susceptibility of the stream to adverse temperature changes
resulting from removal of the riparian canopy. Thus changes
in low flows may be beneficial and of interest to managers,
but low flows generally cannot be used as an indicator of
water quality. To date, water rights courts have not ad-
dressed the allocation of any increase in water yield due to
forest harvest. The absence of any institutional mechanism
to capture the economic benefits of increased low flows, and
the difficulty of measuring small increases on large basins,
indicates that low flow monitoring is rarely appropriate.
3.3 WATER YIELD
Definition
A change in water yield represents the sum of all the
individual changesin runoff over a water year. Most paired-
watershed experiments have focused on changes in the total
annual water yield, so there is much more data on changes
in water yield than on changes in low flows or the size of
peak flows.
Relation to Designated Uses
The importance of an increase in water yield depends on
the timing of the increase, the uses of the water, and the
extent to which the increase can be captured by storage
facilities. In rain-dominated or warm snow environments,
the largest relative increases in water yield usually occur
during the summer and first autumn storms (Harr, 1983).
The largest absolute increases occur during the fall-winter
rainy season (Harr et al., 1982).
In colder, snow-dominated environments most of the
increase in water yield will occur early in the spring snowmelt
period because less snowmelt is needed to recharge soil
moisture (e.g., Troendle and King, 1985). If there is suffi-
cient precipitation during the summer and fall to generate
substantial amounts of streamflow and maintain high levels
of soil moisture, water yield increases also may be detected
in these periods (e.g., Swanson and Hillman, 1977).
The significance of an increase in low flows was dis-
cussed in Section 3.2; the likelihood and significance of
increasing peak flows was discussed in Section 3.1. Other
than the possible increase in the size of the larger peak flows
due to rain-on-snow events, the increase in fall and winter
discharge from forest activities is likely to have little bio-
logical or physical significance. However, any increase in
flow may be beneficial if it can be captured in a downstream
reservoir and used for generating electricity, irrigation, or
water supply purposes.
-------
CHAPTER 3. CHANGES IN FLOW
Response to Management Activities
Bosch and Hewlett (1982) summarized the results of 94
paired watershed experiments worldwide and found that (1)
in areas with over 450 mm of annual precipitation, clearing
at least 20% of the forest cover resulted in a water yield
increase; (2) the increase in water yield was proportional to
the average annual precipitation; and (3) the increase in
water yield was quite variable but was larger in wet years,
particularly in dry areas. The magnitude of the observed
increases ranged from 0-660 mm per year.
In general the increase in water yield due to forest
management activities will be too small to be measured.
U.S. Geological Survey gaging records are regarded as
excellent if they are accurate to within 5%, and most local
discharge measurements will be less accurate. The impre-
cise nature of discharge measurements, particularly at high
flows, and the fact that measurable increases occur only
when at least 20% of the forest cover has been removed
(Bosch and Hewlett, 1982), suggest that increases in water
yield can be reliably detected only when a large proportion
of the forest cover has been harvested over a relatively short
time period. As one moves downstream these individual
increases will be smoothed out over time and increasingly.
diluted (MacDonald, 1989). This is why the sustainable
average increase in annual water yield in western Washing-
ton and Oregon has been estimated at <3-6% of the
unaugmented flows, while the maximum annual increase in
water yield from small clearcut catchments has ranged up to
600 mm/yr per unit area (Harr, 1983).
Measurement Concepts
By definition, water yield increases must be determined
by continuous stream gaging and conversion of the ob-
served stage to discharge. A paired-watershed approach is
essential to remove the effects of climatic variability and
obtain the necessary sensitivity.
Since water yield increases will tend to be lost in the
downstream direction, stream gaging should be conducted
as high in the watershed and as close to the management
activity as possible. Accurate measurements are essential,
and measurements must be made during high flow periods
when the bulk of the runoff is occurring. The logistical
difficulties of accurately measuring streamflow during high
runoff periods in remote sites cannot be overstated.
Standards
No standards for changes in water yield have been
established or proposed.
Current Uses
A change in water yield integrates all the changes that
have occurred over the designated time period (usually one
water year). As such, it provides little information on the
physical processes causing the observed change in water
yield and hence little information useful to land managers.
Changes in water yield may be important for water supply
purposes, but the absolute amount in larger streams is very
small due to the dispersed nature of forest management.
Assessment
In general, changes in water yield are detectable only in
the immediate proximity of the harvested units. Measure-
ment errors, the lack of aperfect relationship between paired
basins, downstream dilution, and the small change in total
volume all preclude the detection of a change in water yield
in moderate-to-large streams (e.g., larger than second or
third order). This, plus the extensive information already
available, suggests that monitoring of water yield is not
necessary under most circumstances.
On the other hand, continuous discharge measurements
may be needed to calculate the total load of critical nutrients,
or as part of a project to monitor turbidity or suspended
sediment. If these data are being compared with an adjacent
unmanaged basin, it then may be possible to utilize these
discharge data to estimate the change in water yield. How-
ever, discharge and constituent data usually are collected at
only a few sites in order to estimate the total load from
different sub-basins, or they are being collected upstream
and downstream of a particular project. Rarely are com-
parable data available from an undisturbed watershed. These
limitations in the statistical design of most monitoring
projects, together with the absence of an unmanaged control
and the difficulties in accurately measuring discharge,
preclude a rigorous estimation of the change in water yield
due to forest management activities.
-------
4. SEDIMENT
INTRODUCTION
An increased sediment load is often the most important
• adverse effect of forest management activities on streams.
Large increases in the amount of sediment delivered to the
stream channel can greatly impair, or even eliminate, fish
and aquatic invertebrate habitat, and alter the structure and
width of the streambanks and adjacent riparian zone.
The physical effects of increased sediment load can be
equally far-reaching. Fine sediment can impair the use of
water formunicipal or agricultural purposes. The amountof
sediment can affect channel shape, sinuosity, and the rela-
tive balance between pools and riffles. Changes in the
sediment load also will affect the bed material size, and this
in turn can alter both the quantity and the quality of the
habitat for fish and benthic invertebrates.
Many nutrients and other chemical constituents are
sorbed onto fine particles, so sediment loads are often
directly related to the load of these constituents. Indirect
effects of increased sediment loads may include increased
stream temperatures and decreased intergravel dissolved
oxygen (DO).
These wide-ranging effects suggest that there are an
equally broad range of techniques that can be used to assess
the quantity and impact of the sediment load in a particular
stream. Direct measurements include suspended sediment
concentration, turbidity, and bedload. Indirect methods
include measurements of channel characteristics such as the
width-depth ratio, residual pool depth, bed material particle
size, or the width of the riparian canopy opening (Sections,
5.2,5.3,5.6, and 6.1, respectively). This chapter discusses
only the parameters of suspended sediment, turbidity, and
bedload.
4.1 SUSPENDED SEDIMENT
Definition
Suspended sediment refers to that portion of the sedi-
ment load suspended in the water column. This, at least
conceptually, is distinct from bedload, which is defined as
material rolling along the bed. The relative size of particles
transported as bedload and suspended sediment will vary
with the flow characteristics (e.g., velocity, bed forms,
turbulence, gradient) and the characteristics of the material
being transported (e.g., density, shape). For the Pacific
Northwest and Alaska, particles <0.1 mm in diameter (clays,
silts, and very fine sands) are typically transported as sus-
pended sediment, while particles > 1 mm in diameter (coarse
sand and larger) typically are transported as bedload (Everest
et al., 1987). Particles between 0.1 and 1 mm are usually
transported as bedload, but can be transported as suspended
load during turbulent, high flow events (Sullivan et al.,
1987). The process of saltation, in which particles bounce
from the bed up into the water column, blurs the distinction
between these two terms. Local hydraulic conditions also
can cause shifts in the relative proportion and size classes of
bedload and suspended sediment.
Suspended sediment also should be distinguished from
wash load. The latter term refers to particles that are washed
into the stream during runoff events, and that are finer than
the particles found in the stream bed (Ritter, 1978). By
definition the wash load is finer than the bed material load,
and the wash load is considered to remain suspended the
length of the fluvial system (Linsleyetal., 1982). Normally
the wash load is defined as particles smaller than 0.062 mm
(silts and clays). The concept of wash load is rarely used by
fluvial geomorphologists or fish biologists, and it is difficult
to apply in the type of monitoring studies addressed in these
Guidelines.
-------
CHAPTER 4. SEDIMENT
Relation to Designated Uses
Numerous laboratory studies have documented the ad-
verse impacts of fine sediment on benthic invertebrates as
well as salmonid reproduction and growth (Chapman and
McLeod, 1987). Hynes (1970) characterizes streams with
sandy beds as having the lowest species diversity and
aquatic productivity. As noted in Section 2.4, fine sediments
tend to fill the interstices between coarser particles, and this
reduces the habitat space for smali fish, invertebrates, and
other organisms. An intrusion of fine particles into the bed
material also reduces the permeability of the bed material,
and this often results in a decline in the concentration of
intergravel DO (Section 2.4). Certain invertebrate species
are very sensitive to even small declines in DO, and the EPA
standards for DO within the water column are set in part
because of the sensitivity of invertebrates and salmonid
reproduction to the concentration of intergravel DO (EPA,
1986b).
Reduced gravel permeability can inhibit salmonid re-
production by reducing the concentration of DO and by
entrapping alevins or fry. In a laboratory study a substrate
containing 20% fines was found to reduce emergence suc-
cess by 30-40% (Phillips et al., 1975). Although other field
observations support the basic link between fine sediment
and a decline in salmonid reproduction, direct extrapolation
of laboratory studies to the field is difficult because (1)
changes in suspended sediment typically are accompanied
by changes in other environmental factors; (2) different
species have varying sensitivity to sediment at different life
stages and under different environmental conditions; and
(3) changes in behavior may help alleviate the adverse
effects of increased sediment (Everest et al., 1987). These
same constraints apply to studies relating the concentration
of fine sediment to the growth and survival of salmonid
juveniles and adults.
An excess of fine sediment can adversely affect habitat
availability. The case study of the South Fork of the Salmon
River (Box 3, page 17) provides one example, and similar
observations have been made on other streams (e.g., Grant,
1986; Cederholm and Reid, 1987; Sullivan et al., 1987).
Often, however, pool infilling is due to sand-sized particles
which are considered fines by fisheries biologists, but may
not be transported as suspended sediment. Thus an increase
in the concentration of suspended sediment may not nec-
essarily be correlated with a decreasing bed material particle
size.
Direct effects of suspended sediment on salmonids
occur only at relatively high concentrations. For example,
Noggle (1978) found that the ability of coho salmon fin-
gerlings to capture prey was reduced at suspended sediment
concentrations of 300-400 mg IA Mortality of salmonids
occurs only at concentrations greater than 20,000 mg L~l
(Everest et al., 1987).
An increase in suspended sediment concentration will
reduce the penetration of light, and a sustained high concen-
tration of suspended sediment could reduce primary pro-
duction if otherfactors are notlimiting(Gregoryetal., 1987;
Section 7.2). The effect of suspended sediment on water
temperature has not been well documented. EPA's Quality
Criteria for Water notes that suspended materials will in-
crease heat absorption, particularly in the surface layer, and
inhibit mixing between the warmer surface layer and the
cooler underlying waters (EPA, 1986b). Others believe that
the additional heating due to suspended sediment is negli-
gible because turbid waters have a higher reflectance. The
reduced penetration of solar energy caused by an increase in
suspended sediment concentration could reduce the solar
heating of the bed material, but the attenuation of light
energy in water is so rapid that any difference in heating
would occur only in areas where the water is less than about
10 cm deep. The practical implications of an increased
suspended sediment load on stream temperatures and mixing
are limited by the fact that (1) most forest streams are very
well mixed, and (2) suspended sediment concentrations
typically are very low in summer, which is when high water
temperatures are of most concern.
The concentration of suspended sediment also can af-
fect the morphology of alluvial channels. Schumm (1972)
classified alluvial streams by the proportion of bedload to
suspended load. Streams with 97% or more of the total
sediment load as suspended sediment had width-depth ra-
tios < 10, and sinuosities >2. In such channels an increase in
the suspended load would tend, at least initially, to narrow
the channel as the fine sediment is deposited along the
banks. Flumestudieshaveshown that an increase in suspended
sediment concentrations causes an increase in velocity and
a steeper channel gradient (Chang, 1988). An increase in
fine sediment may also delay the initiation of bedload
transport(Beschtaand Jackson, 1979). In general, however,
the concentration of suspended sediment has little influence
in shaping stream channels (Everest et al., 1987).
Suspended sediment can adversely affect several other
designated uses of water. High concentrations of suspended
sediment can damage turbines in hydroelectric plants. Sus-
pended matter reduces the value of water for esthetic pur-
poses. For example, it is unacceptable in municipal water
supplies for esthetic reasons; moreover, it reduces the effi-
cacy of normal treatment procedures (EPA, 1986b).
Suspended sediment will settle out in still or slow-
moving waters, and this can result in clogged irrigation
canals and reduced reservoir storage capacity. In some
cases, however, the deposition of suspended sediment can
be regarded as beneficial. For example, deposition during
high flow events provides additional nutrients and soil
materials. This regular deposition is a major reason why
alluvial valleys often are among the most productive and
fertile farmlands.
-------
Partli
Effects of Management Activities
Forest management activities can affect the amount of
suspended sediment in streams by altering both the erosion
rate and the rate of transport into the stream channel. The
range of management activities, and the number of erosion
and transport processes, have resulted in an extensive lit-
erature on the relationship between forest management and
sediment yield. However, recent changes in forest man-
agement practices may make it impossible to directly ex-
trapolate from previous studies, even if they wereconducted
in a comparable environment (Everest et al., 1987). The
following paragraphs provide a brief summary rather than a
comprehensive overview.
Most comprehensive studies of the effects of forest
management have found road-building and road mainte-
nance to be a primary source of sediment (e.g., Brown and
Krygier, 1971; Megahan and Kidd, 1972). This sediment
can be eroded from the road surface (e.g., Reid and Dunne,
1984), from road fills (e.g., Megahan, 1978), or from slope
failures associated with road construction and drainage
(e.g., Duncan et al., 1987; Megahan and Bohn, 1989). In
most cases there is a sharp increase in sediment yield associ-
ated with road-building activities, and a rapid decline as
roads stabilize (e.g., Beschta, 1978). Increased sediment
yields tend to be more persistent if the erosion stems from
slope failures or surface runoff associated with continued
heavy traffic.
Forest harvest can increase sediment yields by a variety
of processes: surface erosion from landings, skid trails, and
other compacted areas; slope failures triggered by removal
of the tree coven and surface erosion from burned areas or
areas disturbed by site preparation activities (Swanson etal.,
1987). Surface erosion can include both fluvial detachment
and transport as well as dry ravel and surface creep (Swanson
et al., 1987). Historic practices of disturbing the stream
channel and removing large woody debris also have been
shown to increase the amount of fine sediment in the stream
channel (Bilby, 1981; Megahan, 1982). Removal of, or a
reduction in, the riparian vegetation is a another mechanism
by which forest management activities can increase the
amount of fine sediments (e.g., Platts, 1981). Grazing often
exacerbates the effect of reducing the vegetative cover by
simultaneously trampling the vegetation, compacting the
soil, and trampling the streambanks (Gifford, 1981). ,
In some cases management activities may have no sta-
tistically significant effect on suspended sediment concen-
trations. Some of the key factors controlling the actual
incrcaseinsuspendedsedimentareasfollows: (1) the intensity
of disturbance, (2) the areal extent of disturbance, (3) the
proximity of the disturbance to the channel system, and (4)
the storm events experienced during the periods when the
site is most sensitive to erosion and mass movements
(Everestetal., 1987; Swanson etal., 1987). The high natural
variability of suspended sediment often makes it difficult to
detect a statistically significant increase in suspended sedi-
ment from well-planned and properly executed forest har-
vest operations.
Measurement Concepts
Suspended sediment concentrations are determined by
obtaining a water sample, drying or filtering the sample, and
then weighing the residual sediment. Concentrations are
typically expressed in milligrams per liter (mg L-1), and this
usually is equivalent to parts per million (ppm) because 1L
of water has a mass of approximately 1 million milligrams.
As sediment concentrations increase, however, the density
of water exceeds 1000 g L-1, and this causes an increasing
divergence between milligrams per liter and parts per mil-
lion.
The primary problem with measuring suspended sedi-
ment is how to sample in time and space. Estimates of the
total amount of suspended sediment over time often are
based on apresumedrelationship between the concentration
of suspended sediment and stream discharge, but this is by
no means constant or reliable (e.g., Ferguson, 1986). For
example, suspended sediment concentrations for a specified
storm event typically are much higher after a dry period than
after an earlier, but recent, storm. Often suspended sediment
concentrations are higher during periods of increasing dis-
charge (i.e., the rising limb of the hydrograph) and lower
during periods of decreasing discharge (i.e., the falling limb
of the hydrograph). However, detailed studies indicate that
this is not always the case (e.g., Rieger and Olive, 1986;
Williams, 1989a). Walling and Webb (1982) discuss how
the physical processes of sediment production and yield
need to be taken into account to better predict sediment yield
and thereby reduce the apparent variability of suspended
sediment concentrations.
Suspended sediment concentrations can show consid-
erable spatial variability. The increase in suspended sedi-
ment concentration with depth is well known (e.g., Guy,
1970), but the size and concentration of suspended sediment
also can vary according to local turbulence and velocity.
Thomas (1985) provides adetailed discussion of the concepts
and methods of measuring suspended sediment in small
mountain streams.
The concentration of suspended sediment also is highly
sensitive to the method of sampling. Any sampler disrupts
the flow lines, and this can bias the sample. Orifice size,
length of the intake nozzle relative to the sampler, and the
percent of the sample bottle filled all can influence the
accuracy of the sample. The hydraulic requirements of
suspended sediment samplers generally preclude sampling
within 10 cm or so of the stream bottom (Guy and Norman,
1970), and this limits the accuracy of any attempt to obtain
an absolute estimate of suspended sediment flux.
Suspended sediment samplers can be separated into two
basic types—point-integrated and depth-integrated. Point-
-------
CHAPTER 4. SEDIMENT
integrated samplers take one sample from a particular depth,
whereas depth-integrated samplers allow one to sample
continuously as they are raised and lowered (Guy and
Norman, 1970). Since sediment flux or sediment load is of
interest in most monitoring programs, a depth-integrating
sampler is preferred. New, open-frame samplers allow the
use of larger, wide-mouthed plastic bottles instead of the
traditional pint milk bottles.
For logistical reasons most monitoring programs are
using automated pump samplers. The primary limitation of
these is that the intake nozzle cannot be positioned so it will
sample correctly under all conditions; thus each pump
sampler is measuring something different (Thomas, 1985).
For estimates of suspended sediment transport, or for cor-
rect comparisons between stations, data from the pump
sampler must be adjusted according to a site-specific rela-
tionship between a depth-integrated sampler and the pump
sampler (Thomas, 1985).
Calculating the sediment load or sediment flux requires
continuous discharge measurements. Porterfield (1972)
provides detailed information on the procedures to obtain
fluvial sediment discharge data, and provides a series of
plots illustrating the variation in suspended sediment con-
centration over individual runoff events. Recent work by
Cohn et al. (1989) and Walling and Webb (1982) illustrates
the difficulties of accurately predicting suspended sediment
concentrations from discharge data.
Most sampling schemes take individual or composite
samples at regular time intervals (e.g., daily). Since high
flow events are relatively rare, a sampling system based on
equal time intervals will result in a large number of samples
at relatively low flows, when suspended sediment concen-
trations are low, and very few samples at high flows, which
is when most of the suspended sediment transport takes
place. This is both inefficient and results in a high level of
uncertainty with regard to the total sediment load. A stage-
activated system can greatly increase sampling efficiency
by sampling only the higher flows-
Thomas (1985) suggests linking a microprocessor to a
stream gage recorder and an automated sediment sampler in
order to sample on a volume basis. This increases the number
of high flow samples and reduces the number of low-flow
samples, with a significant improvement in both efficiency
and accuracy. While such systems illustrate the potential for
improved sampling procedures, they may be too costly for
most monitoring applications.
Standards
Water quality standards usually are set in turbidity units
rather than the concentration of suspended sediment. The
general criteria established by EPA is that "settleable and
suspended solids should not reduce the depth of the com-
pensation point for photosynthetic activity by more than 10
percent from the seasonally established norm for aquatic
life" (EPA, 1986b).
Current Uses
The importance and intuitive appeal of suspended sedi-
ment make it one of the more commonly used parameters for
water quality monitoring. However, in most cases discharge
also must be measured at the same time. In general, sampling
should also focus on the high discharge events when the
majority of suspended sediment is being transported. The
unpredictable and short-term nature of most high runoff
events suggests that if an automatic sampleris being used to
take samples at constant time intervals, it may be best to take
relatively frequent samples, and then discard those that do
not correspond to any runoff event Continuous discharge
data are needed to interpret the suspended sediment data and
estimate sediment loads and fluxes.
Simultaneous discharge measurements may not be nec-
essary if the monitoring objectives are relatively limited.
For example, construction of a bridge during summer
baseflow periods may be monitored by comparing upstream
and downstream suspended sediment concentrations. Such
measurements will provide some indication of the effects of
the management activity on suspended sediment concen-
trations, but in the absence of discharge data there will be
no data on the total amount of sediment released by the
project, or how the total load might compare to the total
suspended sediment load during different storm events.
As discussed in Chapter 3 of Part I, the rigorous assess-
ment of management impacts on suspended sediment re-
quires data from replicated treated and untreated sites.
Ideally data collected over time are used to determine the
changes due to management, while data from matched sites
are necessary to account for changes in the frequency and
intensity of runoff events during the monitoring period. The
tremendous temporal variability in suspended sediment
concentrations suggests that paired (i.e., treated and un-
treated) sites are necessary to detect even relatively large
changes. This is the approach taken in paired-catchment
studies (e.g., Brown and Krygier, 1971), but the statistical
conclusions from paired-catchment studies usually are
limited by the lack of replication of treated and control sites
(Part I, Section 3.2). Other studies have limited their ability
to detect change by simply monitoring suspended sediment
atone location over time (e.g., Tassone, 1988).
If sufficient suspended sediment is available, it may be
helpful to occasionally conduct particle-size analyses to
more accurately understand the implications for the aquatic
ecosystem. At a typical density of 2.65 g/cm3,1 mg L"1 of
suspended sediment can represent 90 particles of very fine
sand, 90,000 particles of medium silt, or 90,000,000 par-
ticles of fine clay. In each case, the suspended sediment
concentration is identical, but the relative effects on turbid-
-------
Part II
ity, gravel permeability, and bed material particle size
will be very different.
Assessment
Suspended sediment is a very useful indicator of active
erosion in aparticularbasin. However, the multiple processes
involved in sediment storage and delivery preclude the use
of suspended sediment concentrations as a quantitative
measure of specific hUlslope and channel processes. On the
other hand, suspended sediment concentrations are very
sensitive to landscape disturbance, and its conceptual sim-
plicity gives it broad appeal.
The primary problem with using suspended sedimentas
a monitoring tool is its inherent variability. Representative
samples are difficult to obtain, and suspended sediment con-
centrations vary tremendously over time and space. Thus
it is often difficult to determine if there has been a sig-
nificant increase in suspended sediment, and whether an
observed increase is due to management activities or natural
causes. These problems are exacerbated as one moves
farther downstream because the impact of individual man-
agement activities is diluted and the amount of suspended
sediment from other sources becomes larger.
Suspended sediment can and should be included in a
monitoring plan provided it is recognized a priori that (1)
identifying an increase in suspended sediment due to forest
management requires several years of background data
from the basin or site where management will occur and a
similar set of data from comparable, unmanaged site(s); and
(2) calculating suspended sediment fluxes and loads results
in an inherent uncertainty of at least 25-50%.
Suspended sediment also is just one component of the
overall sediment budget Changes in bedload generally
havethegreatestgeomorphicimpact(Section4.3), but these
may or may not be correlated with suspended sediment
(Williams, 1989b). Turbidity (Section 4.2) is highly cor-
related with suspended sediment, but this relationship must
be determined for each basin and usually each site. As
indicated above, the adverse impact of suspended sediment
also is a function of the size distribution of the suspended
particles.
4.2 TURBIDITY
Definition
Turbidity refers to the amount of light that is scattered
or absorbed by a fluid (APHA, 1980). Hence turbidity is an
optical property of the fluid (Hach, 1972), and an increasing
turbidity is visually described as an increase in cloudiness.
Turbidityinstreamsisusuallyduetothepresenceofsuspended
particles of silt and clay, but other materials such as finely
divided organic matter, colored organic compounds, plank-
ton, and microorganisms can contribute to the turbidity
value of a particular water sample. Since relative propor-
tion, size, weight, and refractive properties of these materi-
als varies considerably, a correlation of turbidity with the
weight concentration of suspended matter cannot be as-
sumed (APHA, 1980).
Prior to about 1970 turbidity was measured primarily in
Jackson turbidity units (JTU). Jackson turbidity units are
determined by slowly increasing the depth of water in a clear
cylinder until a candle flame placed under the bottom of the
cylinder disappears into a uniform glow (Hach, 1972).
Several problems are associated with JTUs: (1) usable range
is 25 JTUs and greater; (2) turbidity due to dark-colored
particles cannot be measured as too much light is absorbed;
and (3) very fine particles are not measured (APHA, 1980).
These problems have led to the widespread replacement of
Jackson's candle turbidimeter with photoelectric turbi-
dimeters.
Photoelectric turbidimeters measure turbidity in neph-
elometric turbidity units (NTU); they are able to accurately
measure much lower levels of turbidity, and measurements
generally are not affected by particle color (Hach, 1972).
Theseproperties make photoelectric turbidimeters andNTU
units the preferred method for measuring turbidity in streams.
The differences in measurement techniques mean that there
is no standard conversion between Jackson turbidity units
and nephelometric turbidity units (APHA, 1980).
Relation to Designated Uses
Turbidity is an important parameter of drinking water
for both aesthetic and practical reasons. A strong public
reaction can be expected to a turbid water supply, even if the
water technically is safe to drink. However, suspended matter
provides areas where microorganisms may not come into
contact with chlorine disinfectants, so high turbidity levels
may limit the efficacy of normal treatment procedures
(EPA, 1986b). Small rural communities may not be able to
afford the additional treatment costs necessitated by an
increase in the turbidity of their basic water supply (Harvey,
1989).
Turbidity also has a direct detrimental effect on the
recreational and aesthetic use of water. The more turbid the
water, the less desirable it becomes for swimming and other
water contact sports (EPA, 1986b). In many forested areas
tourism and recreation are important components of the
local economy, and increased turbidity could adversely
affect the attractiveness of a water body for fishing, boating,
swimming, or other water-related activities.
Most of the biological effects of turbidity are due to the
reduced penetration of light in turbid Waters. Less light
penetration decreases primary productivity, with periphy-
ton and attached algae being most severely affected. De-
clines in primary productivity can adversely affect the
-------
CHAPTER 4. SEDIMENT
productivity of higher trophic levels (Section 7.2; Gregory
etal., 1987).
High turbidity levels adversely affect the feeding and
growth of salmonids and other fish species. A recentreview
concluded that the ability of salmonids to find and capture
food is impaired at turbidities in the range of 25-70 NTU
(Lloyd et al., 1987). Other studies indicate that growth is
reduced and gill tissue is damaged after 5-10 days of exposure
to water with a turbidity of 25 NTU (Sigler, 1980; Sigler et
al., 1984). At 50 NTU some species of salmonids are
displaced (Sigler, 1980; Harvey, 1989).
As in the case of suspended sediment, the relationship
between turbidity and water temperature is not well known.
The increased absorption may or may not be balanced by an
increased reflectance. EPA's Quality Criteria for Water
(EPA, 1986b) indicates that an increase in turbidity can lead
to an increase in surface water temperature and a resultant
decline in the rateof mixing (NAS, 1974). Reduced mixing
could trigger a series of adverse effects due to the lower
concentration of dissolved oxygen in the unmixed deeper
portions of rivers and lakes (Section 2.4; EPA, 1986b).
Although this effect is unlikely to occur in the turbulent
streams characteristic of most of the Pacific Northwest and
Alaska, the 'increased tendency towards stratification in
turbid waters could be significant in reservoirs, lakes, and
other downstream areas. Higher turbidity levels also could
reduce the solar heating of the streambed materials, but the
high absorbtion of solar radiation in water means that this is
applicable only in waters less than about 10 cm deep.
Effects of Management Activities
Most studies of the effects of management activities on
streams have measured suspended sediment rather than
turbidity, as suspended sediment concentrations are not
dependent upon the types of materials in suspension. Sus-
pended sediment also has the advantage of being in units
that can be converted to total flux over time and then related
to other components of the sediment budget (e.g., erosion
processes, inchannel sediment storage, and bedload trans-
port). Hence the effects of management activities on turbid-
ity generally have to be inferred from the relatively
numerous studies that have monitored suspended sediment
concentrations. Extrapolation from these studies is usually
possible because of the relationship between the concen-
tration of suspended sediment and turbidity.
In general, thesameactivities that generate large amounts
of suspended sediment will more or less proportionally
increase turbidity. However, in watersheds with coarse
soils (i.e., little clay or silt), erosion and sediment yield
rates can be relatively high while turbidity levels show only
a moderate increase. Conversely, watersheds which prima-
rily have clay or clay-like sediment sources could have
consistently high turbidity levels but only moderate concen-
trations of suspended sediment; this is reportedly the case
for some of the basalt watersheds in Idaho (J. Skille, Idaho
Dept. of Health and Welfare, Coeur d'Alene, pers. comm.).
One of the few studies that used both turbidity and
suspended sediment to evaluate the effects of road recon-
struction and timber harvest was conducted on the east side
of the Cascades in Washington (Fowler etal., 1988). Road
reconstruction duringthesummerof 1979 increased turbidity
levels (in NTUs) by a factor of 25 and suspended-sediment
concentrations by a factor of nearly 50. During the follow-
ing summer suspended sediment concentrations were el-
evated by about 50% as compared to the upstream control
site, while there was less than a 15% increase in turbidity. In
the third post-treatment year, both suspended sediment and
turbidity concentrations were lower at the downstream site
than at the upstream control site. Timber harvest activities
using a longspan skyline system and variable-width riparian
zones had no detectable effect on suspended sediment or
turbidity (Fowler etal., 1988). These results suggest that, at
least for the above watershed (which was described as
having sandy to loamy soils), suspended sediment concen-
trations appeared to be more sensitive to disturbance than
turbidity.
Measurement Concepts
Turbidity measurements are subject to the same consid-
erations as measurements of suspended sediment (Brown,
1983) because the most common cause of turbidity in forest
streams is suspended sediment. With turbidity, however,
there is an additional source of variation due to the different
substances that can cause an increase in turbidity. At a
particular site, for example, high turbidity levels might be
due largely to organic acids at one point in time, while at
another time the turbidity might be due primarily to silts and
clays from earthflows or bank erosion. This variation in the
sources of turbidity complicates comparisons between sites.
Typically thereisastrongrelationshipbetween turbidity
and discharge. As in the case of suspended sediment, this
relationship will vary by site, within storms (i.e., whether
discharge is increasing or decreasing), and between storms.
The relative ease of measuring turbidity as compared to
suspended sediment has led to a number of studies seeking
to predict suspended sediment from turbidity (e.g., Kunkle
and Comer, 1971; Beschta and Jackson, 1980). These indi-
cate that the relationship between turbidity and suspended
sediment is nonlinear on an arithmetic plot. Generally about
80% of the variability in suspended sediment concentra-
tions can be explained by simultaneous turbidity measure-
ments. Detailed analyses of data from three sites in Vermont
(Kunkle and Comer, 1971) and the five key monitoring
stations on the Bull Run municipal watershed near Portland
(Aumen et al., 1989) indicated that a single relationship
could be used to predict suspended sediment concentrations
at each group of closely-related sites.
-------
Part II
On the other hand, three watersheds in the Oregon Coast
Range showed significant differences in the relationship
between suspended sedimentand turbidity (Beschta, 1980).
There also were significant differences in the suspended
sediment-turbidity relationship for different storms at each
site. Nevertheless, the pooled turbidity data for each water-
shed still could account for nearly 80% of the variability in
suspended sediment concentrations on that watershed
(Brown, 1983).
Turbidity tends to be less sensitive to the sampling
location within astream than suspended sediment, as turbid-
ity is primarily a function of the smaller particles (silts,
clays, and colloids). Hence the materials causing turbidity
tend to be more evenly distributed within the water column
and across thestream cross-section, and grab samples usually
are considered to be sufficiently representative. It is recom-
mended that samples be analyzed for turbidity within 24
hours (APHA, 1980), as algal growth can cause an increase
in turbidity. In forested areas it is often assumed that water
temperature and water quality (e.g., paucity of nutrients)
will inhibit or restrict algal growth, but protocols for sample
collection and storage should consider this possibility.
Sediment flocculation also can cause turbidity values to
change over time.
The variability in turbidity among sites and over time
generally makes it quite difficult to determine a natural or
background level for any specified level of discharge. The
natural variation is almost always greater than 10% about
the mean for any given discharge, and the variation tends to
increase with higher discharges (Brown, 1983). Uncertainty
due to instrument differences and analytical errors also
amount to approximately 10% (APHA, 1980). Thecombined
uncertainty due to natural variability and measurement
errors hasimportantimplications both for detecting increases
in turbidity due to forest harvest and other management
activities, and for enforcing relatively narrow turbidity
standards.
Standards
Turbidity standards can be either relative or absolute.
Drinking water standards usually are in absolute terms, and
currentEPAregulations require turbidity in municipal water
supplies not to exceed 1NTU (EPA, 1986b).
Relative turbidity standards have been established in
some states. California, for example, specifies that a timber
harvest cannot increase turbidity by more than 20% above
background. Alaska and Washington allow an increase of
5 NTU for domestic water supplies when the background
turbidity is less than 50 NTU, and no more than a 10%
increase in turbidity when the background level is greater
than 50 NTU (Harvey, 1989). The general criteria for the
protection of freshwater fish and other aquatic life is that the
depth of the photosynthetic compensation point should not
be reduced by more than 10% from the seasonally estab-
lished norm for aquatic life (EPA, 1986b). As suggested
above, the basic problem with enforcing these standards is
that background levels are seldom defined and difficult to
determine. This suggests that only continuing major viola-
tions can be unambiguously identified.
Current Uses
Probably the most common use of turbidity measure-
ments is to monitor the quality of domestic water supplies.
More frequent sampling is required as the measured turbid-
ity approaches or exceeds the 1 NTU standard.
Turbidity often is used to monitor the effects of a
specific management activity (project monitoring). Typi-
cally this involves a comparison of measurements taken
upstream (control) and downstream (treated) of a particular
project, such as the construction of a bridge, with the
presumption that any increase in turbidity is due to that
activity. This procedure is particularly effective during low
flow periods when the background turbidity is both low and
consistent. Assessing the effects during storm periods is
considerably more difficult (i.e., less sensitive).
Turbidity measurements provide an indication of the
amount of suspended material in the water, but the precise
relationship between turbidity and the mass of suspended
material depends on the size and type of suspended particles.
This relationship must be established for each stream or
sampling location, and simultaneous measurements of
suspended sediment and turbidity must be made over the
full range of expected discharges. In some cases a single
relationship may apply at several sites, but this must be
based on a careful statistical analysis of the data from each
site. The relationship between suspended sediment and tur-
bidity cannot be assumed to be stable over time, as changes
in sediment sources or transport processes may alter the
relative balance between suspended sediment and turbidity.
The relative ease of measuring turbidity means that it is
commonly usedformonitoringnonpointsources of sediment.
Suspended sediment tends to be measured in more detailed
studies, or when there is a need to estimate sediment loads
(e.g., to calibrate or validate a sediment yield model). If an
additional uncertainty of ±25% is acceptable, turbidity can
be used to estimate suspended sediment concentrations.
Estimation of the suspended sediment load requires con-
tinuous discharge measurements.
Assessment
Turbidity is relatively quick and easy to measure. Sus-
pended sediment usually is the primary source of turbidity
in forest streams in the Pacific Northwest and Alaska.
Simultaneous measurements of suspended sediment and
turbidity generally result in a relationship that can predict
about 80% of the variation in suspended sediment concen-
trations from measured turbidity values. Thus turbidity can
-------
CHAPTER 4. SEDIMENT
be used as a surrogate for suspended sediment concentra-
tions. The relative ease of measuring turbidity means that
qualitative field observations and synoptic sampling can be
used to identify specific sediment sources (source-search
methodology discussed in Part I, Section 3.2.3).
Turbidity is regarded by many as being the single most
sensitive measure of the effects of land use on streams. This
is due partly to the fact that relatively small amounts of
sediment can cause a large change in turbidity, and partly to
the estimated accuracy of turbidity measurements (ap-
proximately+10%) (APHA, 1980; Brown, 1983). Although
the variation in turbidity with discharge generally is greater
than 10% (Brown, 1983), both the accuracy and variability
of turbidity measurements compare favorably with the other
sediment parameters (suspended sediment and bedload) as
well as the channel characteristics (Chapter 5).
The disadvantages of turbidity are twofold. First, the
relationship with suspended sediment must be determined
for each site, even though some studies have shown that
several sites with similar physical characteristics may have
identical relationships. Second, turbidity is highly variable.
As in the case of suspended sediment (Section 4.1), turbidity
varies according to the discharge; the occurrence of spo-
radic events such as debris flows, landslides, or the break-
down of log jams; the timing of the sample relative to the
season of the year, the time since the last runoff event; and the
timing within a storm hydrograph. The range and nonlinear
'nature of these variations make it very difficult to establish
and enforce a narrowly defined turbidity standard for storm
events. Narrow turbidity standards are much easier to develop
and apply during low flow periods when background levels
are consistently low (e.g., a comparison of turbidity levels
upstream and downstream of a bridge construction site).
Turbidity measurements are particularly effective in the
case of project monitoring (e.g., samples are taken upstream
and downstream of a particular management activity).
4.3 BEDLOAD
Definition
Bedload is the material transported downstream by
sliding, rolling, or bouncing along the channel bottom
(Ritter, 1978). Typically particles > 1.0 mm in diameter are
transported as bedload, while particles <0.1 mm in diameter
are transported as suspended load. Particles between 0.1
and 1.0mm in diameter can be transported either as suspended
loader asbedloaddependingon the local hydraulic conditions
(Everest et al., 1987). Thus even at a single site a particle
may be transported as bedload or suspended load depending
on the discharge and other hydraulic factors.
Bed material load, a term often confused with bedload,
is the transport of particles of a grain size normally found in
the stream bed (Linsley et al., 1982). Thus a stream bed
comprised primarily of silt and clay particles will have most
of its bed material load transported as suspended sediment,
while the bed material load of a coarse-bedded stream (e.g.,
gravels and cobbles) will be transported almost entirely as
bedload.
Relation to Designated Uses
Bedload is an important component of the total sedi-
ment load of a stream. The proportion of the sediment load
transported as bedload varies considerably and cannot be
characterized by a simple relationship to suspended sediment
load or to discharge (Williams, 1989b).
The amount and size of the bed material, in conjunction
with the discharge, slope, and geology, largely determine
the overall type and shape of the channel. Wide, shallow
channels are characteristic of streams transporting coarse
bedload in unconstrained alluvial valleys (Ritter, 1978). As
discussed in Sections 5.1-5.2, streams with a high width-
depth ratio are more likely to experience high water tem-
peratures that may be detrimental to coldwater fisheries.
Streams with coarse bedload tend to have a lower sinuosity
than streams that have fine particles as their bed material
(Section5.6.1;Schumm, 1960). Streams with high volumes
of bedload and erodible banks often are braided, and the
rapid changes in channel location characteristic of braided
streams result in continuing high erosion and sediment
transport rates. The unstable channels in braided reaches
provide relatively poor habitat for salmonids, and the large
amounts of sediment transported downstream from braided
reaches can adversely affect reservoir storage capacity and
other designated uses such as fisheries and irrigation.
Large amounts of easily transported bedload tend to fill
in pools andreduce the larger-scale features that are important
sources offish habitat. At very high flows, however, the pools
may be scoured (e.g., Campbell and Sidle, 1985).
The type and amount of bedload is very important in
determining theamountofmicrohabitatavailableforjuvenile
fish and macroinvertebrates (Section 5.6.1). In general,
coarser material provides more habitat space, whereas fine
sediments tend to fill up the interstitial spaces between
larger particles. Fine sediment is usually defined as par-
ticles <0.83 mm in diameter, but some studies have used
values of up to 6.4 mm (Everestetal., 1987). The deposition
of fine sediment reduces the habitat space for young fish and
aquatic macroinvertebrates (Sections 5.6.1, 7.3, and 7.4;
Everest et al., 1987).
The deposition of these finer bedload materials (e.g.,
sand-sized particles) also has been shown to adversely
affect gravel permeability and the suitability of the gravel
for spawning salmonids (e.g., Everest et al., 1987; Lisle,
1989). A lower permeability usually reduces the concentra-
tion of intergravel dissolved oxygen (Section 2.4), and this
can be directly related to salmonid spawning success, and
-------
Part II
the number and diversity of aquatic invertebrates (Chapman
and McLeod, 1987).
As suggested above, the deposition of bedload has an
adverse effect on reservoir capacity and can clog up irriga-
tion and shipping channels. Hundreds of millions of dollars
are spent in the U.S. each year to remove sediment deposited
behind dams and in the lower reaches of rivers and estuaries.
Effects of Management Activities
The effect of forest management activities on the avail-
ability and transport of bedload has been shown to range
from severe (e.g., Megahan et al., 1980) to no significant
difference (Moring, 1975; Sheridan etal., 1984). Part of the
observed variation in effects is due to the type and intensity
of management. In southwest Oregon, for example,
clearcutting was found to approximately double the bedload
yield as compared to a control watershed, while patch and
selection cuts had no apparent effect (Adams and Stack,
1989). The range of erosion and sediment transport pro-
cesses operating in the Pacific Northwest and Alaska is
another reason why widely different results should be ex-
pected from different studies, and why simple generaliza-
tions cannot be made about the effects of management
activities on bedload (Swanson et al., 1987).
As noted in Sections 4.1 and 4.2, forest harvest can
increase erosion rates by generating overland flow on com-
pacted areas, increasing the number of slope failures (e.g.,
Ice, 1985; Megahan and Bohn, 1989), and increasing the
rate of dry ravel and soil creep (e.g., Ziemer, 1984). Al-
terations in the amount of large woody debris (LWD) in the
stream channels will alter the sediment storage capacity in
the stream channel (Section 5.7; Megahan, 1982). Removal
of LWD, or a reduced rate of recruitment of LWD into the
stream channel, can result in an apparent increase in sedi-
ment yield at the mouth of the basin (Megahan, 1982), even
though there may be no net change in the rate of sediment
delivered to the stream channel from upslope.
Road construction and road maintenance can increase
the amount of bedload by creating areas prone to surface
runoff (Reid and Dunne, 1984), altering slope stabilities in
cut and fill areas (e.g., Megahan, 1978), and altering drainage
patterns in ways thattend to increase thenumberof landslides
and debris flows (e.g., Megahan et al., 1978; Megahan and
Bohn, 1989). Similarly, grazing can increase the amount of
overland flow and decrease bank stability (Section 5.8;
Gifford, 1981). Sand and gravel extraction within the
stream channel will alter the channel hydraulics and prob-
ably cause a short-term increase in bedload transport until
the stream re-establishes a stable channel. Longer-term
effects of sand and gravel extraction are difficult to predict
The material eroded or detached by these different hill-
slope erosional processes must then be delivered to the
stream channel and transported by the stream before it can
be measured as bedload. Often significant amounts of
material can be stored in the channel (Dietrich et al., 1982).
In streams draining the Idaho batholith, for example, 15
times more sediment was stored in the channel than was
delivered out of the basin on an annual basis (Megahan,
1982). When evaluating the impactof managementactivities
on bedload, one must also consider whether the material is
composed of silt- and clay-sized particles, which probably
will be transported as suspended sediment, or coarser par-
ticles, which will be transported as bedload.
Extensive studies on the South Fork of the Salmon River
in Idaho have attempted to link the effects of forest man-
agement and road building to an increase in bedload and the
quality of fish habitat. In this basin the combination of
management activities, credible soils, and severe storms has
resulted in extensive sedimentation. The large amounts of
bedload reduced pool depths and literally buried many of
the prime salmonid spawning and rearing areas with sand
(Megahan, 1980; Box 3, page 19). In'other parts of the
Pacific Northwest, studies have documented increased
amounts of fine sediment in the bed material in response to
forest harvest and road-building (Section 5.6.1; Cederholm
etal., 1981; Scrivener, 1988). However, very few published
studies have attempted to monitor changes in bedload
transport rates due to forest management activities, and then
relate thesechanges to the designated uses of the water body
being monitored. The paucity of such studies has strong
implications with regard to the relative utility of monitoring
bedload transport rates.
Measurement Concepts
The measurement of bedload must be regarded as diffi-
cult. Sampling devices disturb the flow in the vicinity of the
sampler, and this biases the sample (Guy and Norman,
1970; Emmett, 1980). The most common bedload sam-
pling device, the Helley-Smith sampler, consists of a flared
rectangular orifice with an attached mesh bag. The sampler
is placed on the stream bottom with the opening facing
upstream for a specified time, and the sediment caught in the
mesh bag is dried and weighed to get a transport rate in mass
per unit time per unit stream width (Helley and Smith,
1971). The most commonly used design has a 76-mm (3.0-
inch) square opening and a mesh size for the sample bag of
about 0.25 mm. This has been reported to have a catch
efficiency of about 1.0 for particles from 0.5-16 mm in
diameter (Emmett, 1980). Sampling of larger bedload
particles requires a larger sampler, and the catch efficiency
is less well known.
Bedload transport rates vary across the stream cross-
section, so representative samples should be taken at regular
intervals across the stream (Emmett, 1980). Numerous
studies, however, have shown that bedload moves in irregu-
lar sheets or waves (e.g., Beschta, 1981; Reid and Frostick,
1986). This can be due to migrating dunes or bedforms, and
to unpredictable events, such as the breakup of a stream
-------
CHAPTER 4. SEDIMENT
armor layer, the release of sediment stored behind channel
obstructions (e.g., Megahan 1982), and sudden inputs of
sediment (Swanson et al., 1987; Sidle, 1988). Several
studies indicate that bedload transport tends to be higher on
the falling limb of the hydrograph (i.e., declining discharge)
than on the rising limb of the hydrograph (increasing dis-
charge). This is the, reverse of the usual hysteresis effect for
suspended sediment, and it is attributed to the initial resis-
tance of the surface armor layer to entrainment (Reid et al.,
1985; Lisle, 1989). Thus bedload samples taken from the
same location at constant flow can be expected to vary
greatly over relatively short time periods, and sampling
should be conductedover several sedimenttransport"cycles"
(Emmett, 1980). Under normal field conditions an accuracy
of no better than 50-100% can be assumed (e.g., Lisle,
1989). The temporal distribution of bedload transport in any
given year will vary according to the size of the bed material
and the flow regime of the stream in question, but in most
streams the majority of bedload will be transported only
during the two or three largest flows in a particular year.
In some cases the long-term sediment transport rate can
be estimated by measuring the amount of sediment that
accumulates in a lake or other sediment trap (Foster et al.,
1990). Such estimates may need to be adjusted by the trap
efficiency, which is a function of the residence time of the
inflowing water (Barfield et al., 1981). This procedure
generally does not allow separation of bedload and suspended
load.
The difficulty of accurately measuring bedload has led
to the development of numerous equations to predictbedload
transport. However, these equations are seldom able to
predict observed transport rates over the entire particle-size
range found in natural streams (e.g., Reid and Frostick,
1986). Flume data are often used to develop and validate
these transport equations, but they then have to be ex-
trapolated to field conditions.
Standards
No standards have been established or proposed for
bedload, and this is probably due to the difficulty of mea-
suring and evaluating bedload transport.
Current Uses
In some monitoring projects a set of bedload samples is
taken during selected field visits. Bedload transport, however,
is highly dependent on stream discharge and is less frequent,
than suspended sediment transport. Thus virtually all of the
annual bedload transport occurs during peak snowmelt or a
few of the largestrunoff events. While this greatly shortens
the period of sampling, these few events must be intensively
sampled if annual bedload transport is to be estimated.
Williams (1989a, 1989b) found no consistent relationship
between bedload, discharge, and suspended load, and con-
cluded that the concept of a constant bedload proportion is
not generally valid.
In streams that are not heavily armored, there may be
some value in occasionally measuring bedload transport at
different cross-sections during moderately high flows. These
measurements may indicate the flow at which the bed
material begins to move, and this provides auseful check on
theoretical estimates based on depth, velocity, and particle
size. Occasional field measurements also will help to
understand therelativebalancebetween suspended sediment
load and bedload for the sampled discharges. This in turn
may help indicate the relative sensitivity of the stream
system to different types of sediment inputs.
Relatively few monitoring studies can afford to inten-
sively sample bedload. As a result, estimates of bedload
transport cannotbemadeexcept in cases whereadownstream
trap is available that can be surveyed on a regular basis. The
high year-to-year variation in bedload transport, particularly
for coarser bed materials (e.g., Sidle, 1988), suggests that a
relatively long-term record is needed to obtain a reliable
estimate of bedload transport rates. Less frequent samples
are useful only as crude indicators and generally should be
interpreted qualitatively rather than quantitatively.
Assessment
One can argue that some data are always better than no
data, but in the case of bedload it is questionable whether a
limited amount of quantitative data has any real value for
estimating bedload transport (Williams, 1989b). Unless the
stream is intensively sampled during high flow events, or a
trap exists where sediment accumulations can be periodi-
cally measured, annual bedload transport should not be
estimated.
As suggested above, occasional bedload samples may
be helpful in gaining a qualitative assessment of stream
behavior. Typically mostfieldinvestigationsareconducted
during low flow periods and require a series of geomorpho-
logic clues to develop an appreciation for stream condition,
sediment transport capacity, and sensitivity to increased
sediment load. Bedload samples during high flow events,
when separatedbyparticlesizeandcombined with discharge
data, may help this interpretation process. Bedload data
rarely provide unique, quantitative information for anything
more than very crude model verification or forest planning,
but they can provide some additional insight into stream
behavior.
Another problem with measuring bedload transport
rates is the difficulty of directly relating specific bedload
transport rates to adverse effects on the various designated
uses such as salmonid habitat. Often the effects of bedload
on.the designated uses can be more directly assessed by
monitoring parameters such as cobble embeddedness, re-
sidual pool depths, pool-riffle ratios, or cross-section pro-
files. Although these parameters all have their own draw-
-------
Parti!
backs, and it may be difficult to link observed changes to
specific management activities, at least such measurements
can be directly related to many of the important designated
uses of forest streams. Most of the channel characteristic
parameters also have the advantage of being considerably
easier to measure than bedload transport, and these are the
subject of the following chapter.
-------
5. CHANNEL CHARACTERISTICS
INTRODUCTION
The parameters reviewed in this chapter relate to the
shape of the stream channel, the structural features within
the stream channel, and the stability of the stream banks.
These channel characteristics can be monitored on different
spatial scales and from different perspectives. For example,
bed material particle size and embeddedness evaluate the
surface of the streambed on a scale of a few centimeters,
whereas a thalweg profile evaluates the topography of the
deepest part of the streambed on a scale of tens or hundreds
of meters. Measurementsofhabitattype(e.g.,pools, riffles,
etc.) were pioneered by fish biologists and are used to
evaluate the quality offish habitat, but these measurements
are functionally related to the parameters that might be used
by fluvial geomorphologists (e.g., residual pool depth or
the number of debris dams caused by large woody debris).
Most of the characteristics of stream channels that
might be used for monitoring are controlled by the same
basic set of interacting factors. Among the most important
of these are the amount and size of sediment, the duration
and size of peak flows, slope of the valley bottom, valley
bottom width, steepness of the sideslopes, and the local
geology. Some of these factors can be considered constant
for a given site, while the factors that do vary (discharge and
sediment) are relatively difficult to monitor (Chapters 3 and
4). Stream channel characteristics may be advantageous for
monitoring because their temporal variability is relatively
low, and direct links can be made between observed changes
and some key designated uses such as coldwater fisheries.
The importance of these controlling factors suggests
that many of the channel characteristics will have a similar
response to management activities. Some of the parameters
which are most closely related include channel cross-sec-
tions (Section 5.1) and channel width/width-depth ratio
(Section 5.2); pool parameters (Section 5.3) and thalweg
profile (Section 5.4); and the three parameters relating to
bed material (particle size, embeddedness, and surface vs.
subsurface bed material particle size; Section 5.5). In most
cases it is not necessary to monitor each of these closely
relatedparameters, and the selection among these monitoring
parameters will depend upon the particular combination of
management activities, designated uses, and site-specific
conditions. General recommendations are difficult because
relatively few studies have used channel characteristics as
the primary parameters for monitoring management im-
pacts on streams.
The relatively low temporal variability of channel char-
acteristics must be balanced against (1) the potentially large
spatial variability, and (2) the problem of separating man-
induced changes from changes due to natural events. Proper
statistical design can help alleviate both of these consider-
ations, and the much lower frequency of sampling will allow
more sites or more parameters to be measured. In many
cases a combination of several channel parameters may be
the best approach to evaluate and understand observed
changes in the stream channel.
5.1 CHANNEL CROSS-SECTION
Definition
A channel cross-section is a topographic profile of the
stream banks and stream bed along a transect perpendicular
to the direction of flow. Cross-sectional data are obtained by
measuring distance and surface elevations along the desig-
nated transect or cross-section. The endpoints of the cross-
section are arbitrary, but they should extend at least above
the estimated bankfull stage and preferably beyond the
current floodplain. If change over time is to be monitored,
the elevation data must be related to a permanent bench-
mark.
-------
Part II
Cross-section data are needed to calculate discharge
using any of the velocity-area methods (Buchanan and
Somers, 1969). Cross-sections often are used as the sam-
pling transect for other instream parameters such as bed
material particle size (Section 5.6.1), embeddedness (Sec-
tion 5.6.2), and the type and amount of large woody debris
(Section 5.7).
A series of cross-sections referenced to a single bench-
mark is useful to determine the precise slope of the stream
channel. Channel slope isakeyparameterin mosthydraulic
calculations and for stream classification.
Relation to Designated Uses
A cross-section of the channel and adjacent floodplain
is one of the key pieces of information necessary to predict
the velocity and water surface elevation during high flow
events. Such predictions are needed for a variety of engi-
neering and management purposes, including structural
design, estimation of flood heights, and the stability of
channel protection measures.
For these types of engineering purposes, cross-section
data typically are collected at a single point in time. At best
such data can provide only a qualitative indication of chan-
nel condition.
Monitoring of changes in the channel cross-section can
provide important insights into channel stability, bank sta-
bility, and the relative balance between sediment (particu-
larly bedload) and discharge (Beschta and Platts, 1986).
Widening of the stream channel, filling in of the channel
thalweg (the deepestportion of the channel), increasing bed
elevation (i.e., channel aggradation), and declining cross-
sectional area all indicate an excess of sediment Net depo-
sition of sediment usually results in more extreme stream
temperatures, a decrease in the amount and quality of fish
cover, a change in the quality of the spawning habitat, a
possible reduction in habitat space for algae and macroin-
vertebrates, increased bank erosion, and an increased like-
lihood of flooding (Section 4.1).
Channel incision or bed erosion (channel degradation)
usually indicates a reduction in coarse sediment inputs or an
increase in sediment transport capacity due to higher peak
flows. This can havebeneficial or adverse effects depending
upon theinitialconditionsandthedesignateduse(s). Channel
incision will lead to bank steepening and bank instability,
and this will increase the sediment load. Bank instability
also will lead to a toppling of the riparian woody vegetation
immediately adjacent to the stream channel, which can trig-
ger a series of secondary effects (Section 6.2). On the other
hand, if the channel already has been subjected to increased
sedimcntloadsfrompreviousmanagementactivities.channel
incision may represent a return to "natural" conditions and
an improvement in habitat quality and channel capacity
(e.g.f Megahan etal., 1980).
Response to Management Activities
The shape and area of the channel cross-section can
change in response to a variety of management activities.
Management can alter the size or frequency of peak flows
(Section 3.1) and the sediment load (Chapter 4), and these
are likely to affect the shape and area of the channel cross-
section. A decline in bank slope may be due to grazing
impacts. Rapid infilling and an increase in the width-depth
ratio suggests an excess of coarse sediment. Erosion at the
toe of the bank may lead to a slumping of the oversteepened
bank, and these changes can be quantified by systematically
monitoring selected cross-sections.
In each case additional information on management
activities and natural events should be collected. For
example, the cause of infilling could be either several years
of below-average rainfall or an upstream landslide. In
northern California and parts of the Northwest, an apparent
downcutting in certain stream channels is actually part of
the long-term recovery from the large sediment deposits
associated with the extreme 1964 flood (Lisle, 1982).
Measurement Concepts
Cross-sections are surveyed by establishing a line per-
pendicular to a stream and measuring bed surface elevations
either at regular intervals or at pronounced changes in slope.
If the cross-section is notperpendicular to the stream channel
and flow direction, errors will accumulate in the estimates
of cross-sectional area and discharge. Cross-section data
always should be plotted for error-checking and improved
visualization of channel form.
Typically a cross-section is measured by a two-person
crew with surveying equipment. However, one person can
survey a cross-section by stretching a tape across the stream
and then measuring the height of the tape above the ground
surface. Some investigators have found this latter technique
to be more efficient (Platts et al., 1983).
Often a series of cross-sections are necessary to charac-
terize a stream reach, establish transects for sampling other
parameters, and provide quantitative data for statistical
analysis. Groups or clusters of cross-sections can be located
by random sampling, stratified random sampling, or sys-
tematic placement around random samples (Part I, Chapter
3). Stratified random sampling can be effective in reducing
variability, decreasing sample size, or increasing the ability
to detect change if the strata are properly chosen and the user
has some prior information on the types and variability of
the strata. Either of the two random sampling techniques are
acceptable provided the number of samples is large enough
tomeetthestatisticalrequirements(Plattsetal., 1983). Data
from cross-sections can be grouped by habitat type (Section
5.5) to determine general trends.
-------
CHAPTER 5. CHANNEL CHARACTERISTICS
Standards
No standards for channel cross-sections have been
established or proposed.
Current Uses
The most common reason for collecting cross-section
data is to calculate discharge using the standard velocity-
area technique (Buchanan and Somers, 1969). Data from
multiple cross-sections are used to evaluate fish habitat
conditions, estimate net sediment transport within aparticular
reach, and evaluate changes in channel morphology (e.g.,
width-depth ratio, bank slope, and bankf ull depth). Certain
other parameters, such as bed material particle size and
embeddedness, can be properly interpreted only if they are
referenced to a particular location along a thalweg profile or
channel cross-section.
Cross-section data have been an important component
of monitoring the South Fork of the Salmon River (Box 3,
page 19) and the Silver Fire Recovery Project (Box 7, page
57). In many other monitoring projects, cross-section data
have been collectedbuthave not been analyzed to determine
the specific changes occurring over time. The ready avail-
ability of computer software programs and digitizing tables
means that comparative analyses can be done more quickly
than in the past. Reference cross-sections are being estab-
lished by the Timber-Fish-Wildlife Ambient Monitoring
Program in Washington and by the U.S. Forest Service.
Assessment
Stream cross-sections provide a quick and useful visu-
alization of the stream channel. Repeated measurements of
the same cross-section is a relatively simple means to monitor
changes in the stream channel. Sampling locations for other
monitoring parameters often are established on the basis of
reference cross-sections.
The sensitivity of a cross-section to change is highly
dependent on a variety of site factors. Bedrock can limit
scour or lateral migration. In steeper reaches, where the
stream has a high sediment transport capacity, there may be
no net deposition despite an increase in sediment load.
Conversely, in downstream alluvial reaches the channel
cross-section may be relatively responsive to changes in
both the sediment load and the size of peak flows. This
suggests that a series of cross-sections may be needed to
assess the overall patterns of channel change within a
catchment.
The primary problem with monitoring cross-sections is
that it may be very difficult to determine the cause of an
observed change. A channel cross-section represents an
integrated response to natural events, the physical environ-
ment, and management impacts. Separation of these factors
requires several different approaches. First, cross-sections
should be monitored over a relatively long time period, as
short-term changes resulting from unusual climatic events
can mask a quite different overall trend. .Second, data on
other parameters, such asbed material particlesize or riparian
vegetation, are necessary to fully characterize and understand
any observed changes in channel morphology. Finally, the
data on channel cross-sections must be put in the context of
a broader watershed assessment, and this should include
data on the type and location of management activities,
watershed characteristics, and the historical climate.
In summary, cross-section data are most useful if com-
bined with other monitoring parameters. Cross-section data
alone may be difficult to relate directly to the designated
uses of the water body of concern. A determination of
channel aggradation or degradation, for example, may per-
mit inferences to be made about certain designated uses
such as wildlife or fisheries, but are not a direct measure of
these uses and may not indicate the cause of an observed
change. On the other hand, channel cross-sections are rela-
tively easy and inexpensive to measure, particularly in
smaller streams. Thus a combination of channel cross-
section data with other parameters more closely linked to
the key designated uses (e.g., spawning habitat) can provide
the basis for a relatively powerful and inexpensive monitor-
ing procedure.
5.2 CHANNEL WIDTH/WIDTH-DEPTH
RATIOS
Definition
Sediment accumulation in the stream channel reduces
stream depth. To maintain the same channel capacity, there
usually is a corresponding increase in stream width. These
interrelated changes provide the basis for two geomorphic
parameters that can be used for monitoring purposes—
stream width and the width-depth ratio.
Both stream width and stream depth have to be defined
with regard to a certain discharge. This discharge can be
specified in absolute terms (e.g., 30 cubic feet per second),
in geomorphic terms (e.g., bankfull), or in terms of recur-
rence interval (e.g., a 5-year event). Because streams almost
always are several times wider than they are deep, a small
change in depth can greatly affect the width-depth ratio.
One must also specify whether the depth is the average
depth for the cross-section or the maximum (thalweg)
depth.
An alternative to measuring channel width is to monitor
the width of the riparian canopy opening, and this approach
is reviewed in Section 6.1.
-------
Part II
Relation to Designated Uses
A decrease in channel depth and an increase in channel
width can have major adverse effects on the biological
community. A decrease in depth tends to reduce the number
of pools (Beschta and Platts, 1986), and this will reduce
certain types of fish habitat. An increase in stream width
will lead to an increase in net solar radiation and higher
summer water temperatures (Beschta et al., 1987). The
combination of shallower pools and increased solar radia-
tion can greatly affect the suitability of the stream for
coldwater fisheries. An increase in stream width and an
increase in light penetration is likely to increase primary
production, although this may be partly offset by a reduced
input of organic debris into the aquatic ecosystem from the
riparian zone (Gregory et al., 1987).
An increase in channel width is achieved through bank
erosionandacorrespondingincreaseinsedimentinputsinto
the stream channel. An increase in bank erosion is particu-
larly important because the sediment is delivered directly
into the stream channel (Section 5.8). The adverse effects
of an increased sediment load were reviewed in Chapter 4.
An increase in the riparian canopy opening due to an
increase in stream width can have a series of adverse
biological effects. Such an increase is likely to reduce the
amount of riparian vegetation, and this will reduce the
ability of the riparian zone to capture nutrients and sediment
(Section 6.2). The riparian zone is also a major source for
large woody debris, an important element in pool formation
and habitat diversity in most forested streams in the Pacific
Northwest and Alaska (Section 5.7).
Response to Management Activities
Forest harvest, road building, road maintenance, and
other management activities often increase the amount of
sediment delivered to the stream channel. Usually an increase
in coarse sediment will lead to an accumulation of sediment
in the deeper parts of the stream channel. If the runoff
remains unchanged, an unconstrained stream generally re-
sponds by increasing its width (e.g., Lisle, 1982; Grant,
1988). Although the magnitude of this increase in width will
be affected by the valley shape and the bank materials, Lisle
(1982) observed increases in width even in constrained,
non-alluvial materials. Thus changes in width or the width-
depth ratio can be used as an indicator of a change in the
relative balancebetweenthesedimentload and thesediment
transport capacity.
Grant (1988) noted that an increase in channel width
also could result from an increase in the size of peak flows.
As shown in Section 3.1, increases in the size of peak flows
due to forest harvest generally are small except in areas
subject to rain-on-snow events. This additional mechanism
for channel widening does not preclude the use of channel
width as a monitoring technique, but it does suggest that
additional data are required to understand the cause of any
observed changes. Harvest of the riparian vegetation also
can decrease bank and channel stability and thereby initiate
a cycle of bank erosion and channel widening (Section 6.2).
Measurement Concepts
The determination of channel width and channel depth
is problematical because both parameters are flow-depen-
dent. Depth tends to increase with flow more rapidly than
width (Dunne and Leopold, 1978; Leopold and Maddock,
1953), but this relationship may not be constant at a given
cross-section. A stream with a wide, flat floodplain, for
example, will experience a sudden increase in width when
the flow overtops the banks and spreads across the flood-
plain. Thus the monitoring of changes in width and depth
should be done at specified discharges and locations. A
geomorphically based discharge, such as active channel
width or bankfull width, is most commonly used but may be
relatively subjective. The resulting uncertainty must be
taken into account when drawing inferences from the data.
Cross-section location will affect the width-depth ratio
and, as noted in Section 5.1, the sensitivity to change. For
example, stream width and width-depth ratios are likely to
differ across riffles, sharp bends, and pools. This variation
can be minimized by measuring widths and depths at a
consistentchannel form such as straightrifflereaches, using
average depth rather than maximum depth, or by using
average values obtained from several different cross-sec-
tions.
The sensitivity of stream width and width/depth ratios
to management impacts and natural events will vary with
stream type and location. A bedrock stream in a steep, V-
shaped valley will not alter its width in response to an
increase in sediment load as easily as a stream in a wide
valley with unconsolidated alluvial sediments. Channel
shape is also affected by the relative proportions and abso-
lute amounts of bedload and suspended load (e.g., Schumm,
1960). Streams with cohesive materials tend to have nar-
row, deep channels, while streams in a sandy or other non-
cohesive substrate tend to be wide and shallow.
Standards
No standards have been set or proposed for changes in
stream width or width-depth ratios.
Current Uses
Although a considerable amount of cross-section data
can be obtained from gaging stations, stream inventories,
and other studies, channel width has not been extensively
used as a monitoring technique. Powell (1988) documented
the increase in stream width that occurred in both the careful
and the intense logging treatments on Carnation Creek in
-------
coastal British Columbia. Channel width and depth data
also have been collected in conjunction with the intensive,
long-term monitoring effort on the South Fork of the Salmon
River (Box 3, page 19; Torquemada and Platts, 1987).
Present efforts by agencies such as the U.S. Forest
Service to inventory fish habitat and stream channel condition
should generate a large amount of stream width and width-
depth data. It remains to be seen how well these particular
parameters can define stream condition and monitor man-
agement impacts.
Assessment
On-the-ground measurements of channel widths and
width-depth ratios have the potential of being relatively
sensitive indicators of changes in the size of peak flows and
sediment yields. Channel width and width-depth ratio can
be related to the value of streams for fish and recreation.
Defining channel width and depth in the field is not a
trivial problem. For this reason it is best to monitor channel
width at a series of cross-sections. Use of geomorphic
indicators such as bankfull width or active channel width
must be done with great care, as these tend to be subjective
and a major runoff event can alter the channel cross-section
and make identification of bankfull features questionable.
Determining width and depth at a standard discharge may be
logistically difficult unless it is done at an existing gaging
station. The problem with using gaging stations as monitor-
inglocationsisthattheyusuallyareplacedatgeomorphically
stable locations andarerelativelyinsensitive to management-
related changes in channel form.
Measuring channel width or width/depth ratios 'also
suffers from the same basic limitation as any other instream
measure—namely, that it does not provide any information
on the cause of an observed change. Hence monitoring data
must be combined with information on management ac-
tivities, storm events, and sediment sources (e.g., roads,
debris flows, landslides, or abreakdown of debris dams). As
noted earlier, one also has to put the changes observed from
arelatively short-term monitoring projectintothecontextof
larger changes such as extreme floods or major sediment
inputs. Only with this additional information can the effects
of forest management begin to be deciphered.
Finally, the magnitude and rate of change in channel
width and width-depth ratio will depend on factors such as
the slope of the stream, the shape of the valley bottom, the
bank and bed materials, and the recent flood history. Al-
though this may make it difficult to establish specific
standards, it should not mask general trends. These consid-
erationsalso indicate thatlong-termmeasurementsatvarious
locations within the watershed are needed for adequate
monitoring.
CHAPTER 5. CHANNEL CHARACTERISTICS
5.3 POOL PARAMETERS
Definition
Pools can be defined as sections of the stream channel
that have a concave profile along the longitudinal axis of the
stream, or as areas of the stream channel that would contain
water even if there were no flow. This means that the
maximum depth of pools is deeper than the average thalweg
depth, and water velocities at low flows often are lower than
the average velocity. Pools are an important component of
the aquatic habitat, and they can be classified and measured
in several different ways.
Pools usually are classified by the process that created
the pool (e.g., undercut bank, debris dam, beaver dam,
plunge pool, etc.). This classification is useful for evaluat-
ing the abundance and type of fish habitat (Bisson et al.,
1982), although the various categories of pools and other
habitat types have not been standardized (Section 5.5;
Platts, 1983). Nevertheless, the number and type of pools in
a particular reach could be enumerated, and changes over
time could be monitored.
More commonly the depth, residual depth, volume, or
area of pools are measured, and these measurements can be
used as monitoring parameters. Pool depth can be either
average depth or maximum depth. Residual pool depth
refers to the depth of the pool below the downstream lip of
the pool (i.e,, the depth of the water which would be trapped
in the pool if there was no discharge) (Lisle, 1987). Pool
area refers to the total surface area of the pool. Both pool
depth and pool area will vary with discharge, whereas
residual pool depth is not discharge-dependent
Relation to Designated Uses
Pools are an important morphological feature in stream
channels and an essential type of fish habitat. In general, a
variety of pool types are needed to provide the range of
habitat needed by different species and age classes of fish.
Slow-moving dammed or backwater pools may be neces-
sary for salmonid survival under harsh winter conditions.
Deep undercut pools may provide protection from high
temperatures. Young fish may require shallow, low-quality
pools to avoid predation. Particularly in smaller streams,
pools provide the majority of the summer rearing habitat
(Beschta and Platts, 1986). Pools also may be important
sites for recreational activities such as fishing and swim-
ming.
Response to Management Activities
Those pools characterized by low flow velocities (e.g.,
backwater or dammed pools) are particularly susceptible to
infilling with sediment. Hence the depth, area, or volume of
-------
Part II
these pools can serve as a relatively sensitive indicator of
changes in the coarse sediment load due to forest harvest,
road building and maintenance, mining, or other manage-
ment activities. On the South Fork of the Salmon River
logging and road maintenance caused an influx of sand-
sized material that filled in many of the prime salmonid
spawning and rearing areas (Megahan et al., 1980).
Changes in pool area, pool volume, or residual pool
depth also can be caused by changes in the features that
create pools. Thus a reduction in the input of large woody
debris may lead to a reduction in the number and size of
pools (Section 5.7). Similarly, a change in the size or
frequency of peak flows will alter the ability of the stream
to transport coarse sediment, and this may alter pool mea-
surements.
The total area, depth, or frequency of pools may not
always be a reliable indicator of adverse management ef-
fects. Streams immediately downstream of active glaciers,
for example, usually are braided and have little or no pool
areas. Landslides, debris flows, and other mass movements
typically result in a loss of pool area and volume, and these
pulsed inputs of sediment may or may not be triggered by
management activities (Swanson et al., 1987).
Measurement Concepts
Pool depth, pool area, and pool volume are all direct
physical measurements, and they are relatively simple to
make in small streams. Recent publications have encour-
aged the use of visually estimating the width, depth, or area
of pools within a stream reach, and then adjusting these
visual estimates for any systematic bias by measuring a
certain percentage of the pools (Hankin and Reeves, 1988).
Inlargerstreamswithdeeperpools.directmeasurementsare
considerably more difficult. Also, a series of conceptual
problems in makingpool measurements must be considered
before embarking on a classification or monitoring program.
First, it may be difficult to determine exactly what
constitutes a pool. Large, still pools are easy to classify, but
the change from pools to runs or glides is one point on a
continuum. Platts et al. (1983) found a consistent observer
bias when measuring pool areas along stream cross-sections.
This consistent bias resulted in a relatively narrow 95%
confidence interval for the data (±10%), but poor year-to-
year accuracy and precision.
A second problem associated with pool measurements
is that pool depth, pool area and pool volume are all flow-
dependent. An increase in stage will increase the value of
these parameters. Although this may not be a problem in
streams with a consistent summer baseflow, it does mean
that stage or water depth must be recorded and taken into
account when analyzing the data. The advantage of residual
pool depth is that it is independent of discharge (Lisle,
1987).
Similarly, the classification of pools and other habitat
types is stage-dependent, but this fact is often ignored
(Section 5.5). At higher flows a pool may become a run, or
a pocket water may become a riffle. Hence any summary
statistics on pool-riffle ratios or the frequency of pool types
also must consider the discharge at the time the data were
collected. For this reason comparisons between surveys
must be done with extreme caution.
Standards
No standards for any pool parameters have been estab-
lished or proposed.
Current Uses
Mostsurveys offish habitat or stream channel condition
have utilized some measure of pool area, length, depth, or
volume. Many of these surveys also identify the primary
cause of each pool. These data are then used to generate
summary statistics on the pool-riffle ratio, pool area, or pool
volume per unit length of stream channel. The expectation
is that subsequent surveys should be able to determine
whether substantial shifts have occurred in these values.
Alternatively, one could monitor changes in individual
pools, but this approach assumes that the pool-forming
structure is constant in time. Studies of woody debris in
streams indicate that the larger pieces are relatively stable
(Sedell et al., 1988), but it would be prudent to monitor at
least several pools of as many different types as possible.
Pool parameters probably are most useful in alluvial
channels. Studies of stream channel development follow-
ing the Mount St. Helens eruption indicate that in many
reaches a riffle-pool geometry developed after only a couple
of years (Meyer and Martinson, 1989). This suggests that
pools could be used for monitoring even under relatively
high sediment loads. Pool parameters are unlikely to be
useful in bedrock channels that are regularly scoured by
high flows.
Assessment
In many streams, pool parameters have considerable
potential for monitoring. Decreases in pool depth or pool
volume may be relatively sensitive indicators of logging-
induced changes in the coarse sediment load or the size of
peak flows. Since pool parameters have not been exten-
sively monitored in the past, there is little documentation to
guide the selection of a particular parameter. Residual pool
depth does have the advantage of being independent of
discharge. Residual pool depth also may be the most
sensitive pool parameter, as an increase in coarse sediment
is likely to first affect pool depth. Monitoring pool param-
eters will be most useful in low or moderate gradient streams
in alluvial valleys (Everest et al., 1987).
-------
CHAPTER 5. CHANNEL CHARACTERISTICS
To be useful, any monitoring of pool parameters should
be combined with data on the pool-forming features. Log-
ging in or near the riparian zone, for example, may alter the
type and amount of large woody debris in the stream
channel, and this will directly affect the number and size of
pools. This suggests that the sample size should be large
enough to allow for random changes in pool-creating struc-
tures, or the pools should be stratified by pool type.
Pool measurements are most likely to be useful when
combined with discharge and other morphological data.
Bed material pardcle size (Section 5.6) can be an extremely
useful parameter in interpreting the cause and significance
of a change in pool depth or pool volume. Flood history and
local discharge data are important because large storms can
reduce the size or number of pools, and this effect must be
distinguished from forest management activities. Addi-
tional long-term data are needed to better assess the value of
pool parameters for monitoring, but pool parameters prom-
ise some significant conceptual and practical advantages in
monitoring forest activities.
5.4 THALWEG PROFILE
Definition
The "thalweg" is defined as the deepest part of the
stream channel atany given cross-section. A thalweg profile
refers to the topographic variation of the thalweg along the
stream axis (i.e., in the upstream-downstream direction).
This can be measured with regard to the water surface or
surveyed against a fixed elevation. A survey of the thalweg
with regard to a benchmark elevation also can be referred to
as a longitudinal profile. Sometimes, however, a longitudi-
nal profile can refer to a profile along the streambank or
water surface. Thus thalweg profUeand longitudinal profile
often are synonymous, but this may not always be the case.
Elevation data from a surveyed thalweg profile can be
used to calculate an average channel gradient. Thalweg
profile data show the variation in bed structure (e.g., pools,
riffles, etc.) along the surveyed reach. In particular, a
thalweg profile can accurately delineate pools along the
main channel and be used to determine residual pool depth
(Section 5.3). Both a cross-section (Section 5.1) and a
thalweg profile can provide data on the overall degradation/
aggradation of the stream channel, but only a thalweg
profile can provide quantitative information on the structure
and gradients along the stream axis. The length of the
thalweg profile also can be compared with the length of the
valley floor to yield the thalweg sinuosity. In most cases the
thalweg sinuosity will be similar to the channel sinuosity.
Relation to Designated Uses
The average gradient as determined by a thalweg profile
is an important criterion for classifying streams. The
channel gradient also is needed for a wide variety of hydrau-
lic calculations and models, including water surface profiles
and sediment transport capacity. Local gradients are im-
portant for estimating shear stress and small-scale'hydrau-
lic behavior.
Thalweg profiles provide detailed and unambiguous
data on pool depth and pool length. These pool parameters
can be directly related to fish habitat value (e.g., Bisson et
al., 1982). Changes in flow velocities and stream depths due
to changes in the bed profile will affect the number and type
of aquatic organisms. An estimate of channel sinuosity is
useful for stream classification (e.g., Rosgen, 1985; Cupp,
1989), and for helping to evaluate one of the ways in which
energy is dissipated in streams (e.g., Schumm, 1977).
Effects of Management Activities
Changes in sediment load or peak runoff can affect the
overall elevation of the thalweg profile through aggradation
or degradation, and alter the structure and habitat types
along the profile (Beschta and Plaits, 1986). More specifi-
cally, an increased sediment load can affect local gradients
by filling in pools and by reducing the gradient within steep
riffles (Sullivan et al., 1987). As discussed in Section 5.3,
pool infilling can be a relatively sensitive indicator of adverse
management impacts. A decline in sediment tends to result
in channel incision, and this has been observed downstream
of newly built dams (e.g., Shen and Lu, 1983; Bradley and
Smith, 1984) and after a moratorium on timber harvest
(Megahanetal.,1980).
A change in the size of peak flows also can be expected
to affect the thalweg profile by altering the sediment trans-
port capacity. An increase in peak flows will tend to
increase the stream channel width and depth (e.g., Schumm,
1977), but the interactions among bed material transport,
bank erosion, sediment inputs, and discharge often make it
difficult to predict the precise effect of a change in one factor
on the change in other factors. Beschta and Plaits (1986)
suggest that stream channel morphology is affected moreby
management-induced changes in sediment than manage-
ment-induced changes in flow.
Measurement Concepts
A thalweg profile is a relatively simple monitoring
technique, and it is relatively inexpensive to obtain in small
streams. Surveyingequipmentisneededtoobtainsufficient
accuracy. In areas with dense riparian vegetation, the task
becomes more difficult because of the problems associated
with obtaining a clear line of sight. Surveying a thalweg
profile can be difficult on bends, as the thalweg usually
-------
Part I!
coincides with the area of greatest velocity and may lie
beneath an undercut bank. In larger streams a boat and other
equipment may be needed to accurately locate and measure
the thalweg profile.
No standard length exists for a thalweg profile, but a
general rule of thumb is that it should extend for approxi-
mately 20-30 channel widths or 2-3 meander segments. In
general it should include at least several distinct pools, but
the exact location and length will depend upon the objec-
tives of the monitoring and the expected changes in stream
channel morphology.
The length of a thalweg profile, when divided by the
equivalent length of the valley floor, yields the thalweg
sinuosity. The thalweg sinuosity will be similar to, or may
slightly overestimate, the channel sinuosity. For short pro-
files it may be possible to directly measure the valley floor
length. Longer thalweg profiles should start and stop at
easily defined locations such as bridges so that the valley
bottom length can be measured from topographic maps.
Thalweg profiles longer than 2-3 meander lengths should be
used if an accurate estimate of sinuosity is needed.
Standards
At present no standards or regulations exist regarding a
thalweg profile. The state of Idaho, however, is considering
the use of thalweg profiles and residual pool depths to
monitor sediment production.
Current Uses
In the past thalweg profiles have been measured prima-
rily in the context of research on stream hydraulics and fish
habitat. Relatively little long-term monitoring data are
available. Nevertheless, surveyed thalweg profiles are at-
tracting increasing interest because of their relative sensi-
tivity to increased sediment inputs, and their ability to
quantitatively and unambiguously assess changes in stream
channel morphology.
The samestudies that support the useof pool parameters
as indicators of management effects also can be used to
support the use of thalweg profiles. As with any monitoring
technique, thalweg profiles are subject to the problem of
separating man-induced impacts from natural changes.
However, a thalweg profile may have some advantage in
that it relies on detailed measurements in a particular loca-
tion. This enables one to separate individual changes in the
stream profile—e.g., the breakdown of a particular debris
dam—from the general trend.
Assessment
Thalweg profiles are a specific technique for assessing
certain types of changes in stream channel morphology over
time. A thalweg profile is complementary to channel cross-
sections in that it evaluates changes along the length of a
reach, and it offers a possibly more rigorous approach to
monitoring the frequency, depth, and length of pools. On the
other hand,athalwegprofilecannotprovideasmuchdetail on
all the different habitat types which are of concern to fisheries
biologists (e.g., pocket water, runs, etc.) and which might
occur along a typical thalweg profile. Thalweg profiles also
can yield data on sinuosity and gradient; both of these are
useful for classifying streams and a variety of other purposes.
The disadvantages of thalweg profiles are similar to the
other parameters used to monitor channelcharacteristics. One
major problem is how to link an observed change in the stream
channel with a particular management activity. This problem
is particularly acute for the channel morphology parameters, as
their values are the integrated result of a large number of
interactingprocesses. This is why a combination of parameters
may be needed to properly evaluate the changes due to man-
agement activities and determine the possible cause(s).
Another disadvantage is the problem of setting a thresh-
old or standard for allowable change. In the case of thalweg
profiles, one should not just look at an overall change in the
gradient, but attempt to interpret all of the smaller changes
in bed slope and pool size. Both qualitative and quantitative
evaluations may be needed, as streams vary greatly in their
sensitivity and response to management impacts (Sullivan
etal., 1987). The more recentstream classification schemes
(e.g., Cupp, 1989; Frissell, 1987; Rosgen, 1985) may help
to interpret thalweg profile data by stratifying the data
according to stream type. This will facilitate a comparison
among streams, and thereby help to determine the expected
range of variability for a particular type of stream.
5.5 HABITAT UNITS
Definition
Most stream reaches in forested areas of the Pacific
Northwest encompass a variety of channel features that
include different types of riffles and pools. Each of these
features provides different habitat values for different fish
species at various life history stages. These channel features
are referred to as channel units, habitat types, or habitat
units. The term habitat unit is used here because it empha-
sizes the ecological importance of these channel features,
and it implies an analysis on a unit-by-unit basis. Habitat
type refers to the basic classification system used to delin-
eate individual channel or habitat units.
Over the last few years, the identification and measure-
ment of habitat units have become important tools for
quantifying fish habitat and identifying limiting factors for
fish populations (e.g., Bisson et al., 1982; Hankin and
Reeves, 1988). Observations of change in individual
habitat units, the relative abundance of different units, or the
-------
CHAPTER 5. CHANNEL CHARACTERISTICS
sequence of units, represent a quite different use of the
methodology and a class of monitoring techniques that
currently are under extensive investigation.
Physical parameters used to separate habitat units in-
clude channel slope, depth, bed material, roughness, and
flow velocity. Since each of these parameters is continu-
ously variable rather than discrete, the designation of habitat
boundaries is somewhat arbitrary. Different studies have
used different classification criteria, although most typi-
cally distinguish about five major habitat types (e.g., Platts
etal., 1983; Ralph, 1989). Many researchers subdivide the
two basic categories of pools and riffles into different sub-
types (e.g., plunge pools, lateral scour pools, backwater
pools, low-gradient riffles, rapids, and cascades).
Both the size and the classification of individual habitat
units are flow dependent; that is, they increase or decrease in
area and volume, and even the classification of individual
habitat units may change with a change in discharge. The
effect of a change in flow is not consistent among habitat
types: for example, as flow increases, dammed pools become
larger, while low gradientrifflesandscourpools may become
glides. Habitat unitsurveysmustbe carried outat similar flow
conditions in order to be comparable (Platts et al., 1983).
Data on the frequency and size of individual habitat
units can be used to determine the relative proportion of
each habitat type within a stream reach. Ratios or indices
of habitat abundance can then be constructed. The pool-
riffle ratio is by far the most common of these, and this has
been used extensively by fish habitat managers to assess the
need for habitat rehabilitation.
Relation to Designated Uses
Habitat composition provides the basis for a relatively
direct link between the physical processes governing stream
morphology and the suitability of the stream for fish repro-
duction and growth. The spatial distribution and abundance
of different habitat units are critical to the relative success of
different fish species. Streams that have a high proportion of
riffles with agravel substrate, for example, probably will have
few large obstructions and an abundance of coarse sedi-
ment. The relative paucity of rearing habitat in such streams
is likely to limit the population of some fish species or life
stages while perhaps favoring others. Models used to predict
habitat value require data on the frequency and abundance of
different habitat types (e.g., Bovee, 1982). Any change in
the flow regime or in the distribution of habitat units can be
expected to alter the suitability of the stream for different fish
species and the overall fish community dynamics.
An inventory of habitat types can provide an overall
summary of both channel morphology and habitat complex-
ity. Repeated surveys can show whether a shift in the relative
proportions of habitat types has occurred, and any change can
be related to both cause (i.e., the physical processes causing
the change) and effect (change in valueof fisheries resources).
Response to Management Activities
The relatively recentdevelopment of habitat unitsurveys
means that very few data are available on how the overall
distribution of habitat types changes in response to man-
agement activities. However, existing knowledge of sedi-
ment transport and other stream processes can be used to
predict how particular habitat units might change given a
specific mariagementimpact(e.g.,Lisle, 1982; Sidle, 1988).
For example, in all but the steepest streams an increase in
coarse sediment would be expected to reduce the area and
volume of pools and increase the percentage of stream area
occupied by riffles (Section 4.3). Similarly, a reduction in
large woody debris removes important pool-forming ele-
ments, and this should increase the area of riffles and
decrease the number and size of pools (Section 5.7). In !
stable reaches with abundant fine sediment, an increase in
the size of peak flows may lead to an increase in scour pools
or glides, and a corresponding decrease in riffles.
Table 6 in Part I qualitatively ranks the sensitivity of
habitat units to each of the other monitoring parameters
discussed in these Guidelines. By combining this informa-
tion with Table 3 in Part I (sensitivity of the parameters to
particular managementactivities),one can determine which
management activities are most likely to affect habitat units.
Predicting the precise type, rate, and magnitude of change
will depend on the stream reach being evaluated and the
local knowledge of stream processes.
Measurement Concepts
Quantitative habitat data are obtained by identifying
and measuring individual habitat units within a designated
stream reach (e.g., Bovee, 1982). The typical procedure is
for a two-person crew to walk a stream channel, with one
person measuring individual habitat units while the other
person records the data. Hankin (1984) recommended that
stratified sampling be utilized to increase efficiency and
reduce error. This concept has led to the procedure of
visually estimating the area of each habitat unit, and measur-
ing a systematic sample of each habitat type to develop a
correction factor for the visual estimates (Hankin andReeves,
1988). This procedure has been widely adopted in the
Pacific Northwest, as it allows an experienced two-person
crew to inventory approximately 1-3 miles of stream channel
per day (G. Reeves, U.S.F.S. Pac. Northw. Res. Sta.,
Corvallis, pers. comm.). Generally the data are used only to
generate summary statistics, and changes in individual
channel units, or in the sequence of units, are not evaluated.
Use of stratified sampling does not resolve the basic
problem of how to classify habitat types. Some investiga-
tors identify only riffles and pools because other habitat
types, such as glides, runs, and pocket waters, cannot be
systematicallyidentified(Plattsetal., 1983). Others (Bisson
et al., 1982) have employed habitat classification systems
-------
Part II
that include subsets of the major habitat types (riffles and
pools) because this more detailed classification system may
provide more insight into the suitability of the stream for
different fish species.
As noted earlier, habitat composition varies with dis-
charge, and this must be considered when undertaking
stream surveys. Observers should be given similar training
in order to ensure consistency. Repetitive surveys should be
conducted by the same people whereverpossible in order to
eliminate any bias between surveyors. If specific habitat
units are being monitored, particular care must be given to
defining the boundaries between adjacent habitat units, as
demarcation errors will reduce the accuracy of the proce-
dures and hence the ability to detect change (Platts et al.,
1983).
At this point there are little or no data to indicate
whether it is best to monitor individual habitat units or to
utilize summary statistics for a stream reach. Some re-
searchers posit that changes in the sequence of habitat units
may be one of the most sensitive and revealing monitoring
techniques that can be derived from habitat unit surveys.
Standards
Currently there are no regulations or standards for habitat
composition. In some National Forests pool-riffle ratios are
being monitored, and a decline in this ratio is considered an
adverse management effect Often a pool-riffle ratio of 1:1
is considered optimal, but the limited literature suggests that
this is highly variable among streams and fish species, and
should not be utilized as a standard (Platts et al., 1983).
Current Uses
An inventory of habitat units usually is conducted to
assess the suitability of the stream for fishery resources.
Unfortunately, "ideal" conditions are difficult to define and
are likely to vary widely according to the fish species of
interest, the flow regime, and other environmental factors.
Hence we may be able to identify stream reaches that have
clearly been impacted by land management activities and
offer poor quality habitat for salmonids, but it may not be
possible to clearly rank streams classified as "acceptable."
Thus one benefit of conducting habitat surveys will be a
better understanding of the existing variability of habitat
units among streams. To the extent that fish census data are
available, and other factors such as fishing pressure can be
accounted for, it should be possible to better define "ideal"
habitat conditions.
Use of habitat units for monitoring environmental change
hasnot been extensively tested becauseofthepaucityof long-
term data. Extensive stream surveys that estimate ormeasure
each habitat unit only recently have been initiated in Wash-
ington, Oregon, and Idaho by agencies such as the U.S.
Forest Service. Much of the data have not yet been analyzed,
but the results are expected to document a large amount of
variability in undisturbed streams. Subsequent surveys will
be needed todetermine what level of changeisacceptableand
how to distinguish changes due to land management activities
from changes due to natural causes. A few repeat surveys
have at least indicated that survey data are consistent (S.
Ralph, Univ. of Washington; D. Bates, GiffordPinchotNati.
Forest; and G. Luchetti, King County, WA, pers. comm.).
Assessment
Habitat unit surveys provide a useful, quantitative char-
acterization of stream channels. At this point, however, our
ability to classify and measure habitat units probably exceeds
our capability to interpret the results. This should change as
comparative data become available and the results of indi-
vidual surveys are linked to land management activities. As
with other geomorphic parameters, it may prove difficult to
separate land use effects from the effects of natural events.
Habitat unit surveys may be relatively insensitive to
land use practices. A small amount of sediment, for ex-
ample, might significantly alter the bed material (Section
5.6) or residual pool depth (Section 5.3), but might not alter
the size of, or ratios among, different habitat units. We
should expect that different habitat units will exhibit differ-
ences both in their sensitivity to change, and in their recovery
rate once change does occur. More experience is needed to
determine if it is better, for example, to directly monitor pool
parameters (Section 5.3) or large woody debris (Section 5.7)
rather than habitat units. In view of this uncertainty, current
efforts to conduct large-scale habitat unit surveys must be
viewed with some concern.
In summary, habitat unit surveys are important to im-
prove our knowledge of the relationship between aquatic
life, fish production, and stream channel morphology. By
then linking habitat data to land use activities and climatic
events, we can better define optimal conditions and suscep-
tibility to change. At present, however, we do not have the
experience or data to fully assess the potential of habitat unit
surveys as a monitoring technique.
5.6 BED MATERIAL
5.6.1 PARTICLE-SIZE DISTRIBUTION
Definition
The composition of the material along the stream bed is
a very important feature of stream channels. The most com-
mon method to characterize the bed material is to classify it
by particle size. By taking a sufficiently large sample, one
can construct a plot of particle size versus frequency in
percent.
-------
CHAPTER 5. CHANNEL CHARACTERISTICS
Different points in the particle-size distribution are used
to provide a simple characterization of the bed material.
Common variables include the median particle size (d50) and
dg4, which is the particle diameter equal to or larger than
84% of the particles (clasts) on the channel bottom. The d84
and dj6 are used to describe the variability of the particle-
size distribution around the mean because they are each one
standard deviation away from the mean when the data are
transformed onto a logarithmic scale.
Another approach to evaluating the bed material is
simply to estimate or measure the percent of the bed surface
covered by fine particles. The size limit for fine particles
will vary by location and purpose of the monitoring, but
usually ranges between 2 and 8 mm in diameter. This
approach implicitly assumes that fine sediment is of pri-
mary concern, and it is not necessary to determine the size
distribution of the coarser bed materials.
Chapman and McLeod (1987) conclude that the fredle
index shows some promise as a measure of gravel suitability
for salmonid spawning in the Northern Rockies. The fredle
index is defined as dg/sg, where dg is the geometric mean
particle size, and sg is the geometric standard deviation
(Lotspeich and Everest, 1981).
Relation to Designated Uses
The particle size of the bed material directly affects the
flow resistance in the channel, the stability of the bed, and
the amount of aquatic habitat (Beschta and Platts, 1986).
Because the flow resistance is one part of the overall energy
loss in streams, the mean particle size can be related to the
other factors that control energy loss in streams such as the
stream gradient (Hack, 1957) and the sinuosity.
Although a direct relationship exists between the size of
the bed material and the stability of the bed, other factors such
as the slope, depth, local turbulence, and bank characteristics
will affect whether a particular particle will be moved. The
frequency of bedload transport is of critical importance for
fish spawning and the other organisms utilizing the .stream
bottom for cover, foraging, or as a substrate.
The size of the bed material also controls the amount and
type of habitat for small fish and invertebrates. If the bed is
composed solely of fine materials, the spaces between par-
ticles are too small for many organisms. Coarser materials
provide a variety of small niches important for small fish—
especially juvenile salmonids—and benthic invertebrates.
Coarser materials also have more interflow through the bed,
effectively expanding the suitable habitat for benthic inverte-
brates and other organisms down into the stream bed, and
facilitating salmonid reproduction. Platts et al. (1979) found
acloserelationshipbetween geometric mean particle size and
gravel permeability. Hence a decrease in the median particle
size of bed material will decrease the permeability of the bed
material, and this will tend to decrease intergravel dissolved
oxygen (DO) concentrations. Even a small decline in inter-
gravel DO can severely affect the survival of salmonid eggs,
alevins, and invertebrates (Section 2.4).
Effects of Management Activities
One of the most common and probably the most damag-
ing effect of forest management activities is to decrease the
medianbedmaterialparticlesize. Forestharvest,roadbuilding
and maintenance,andplacerminingall tend to increase erosion
and sediment delivery rates (Swansonetal., 1987). Most of
the material reaching the stream channel as a result of human
activities will be sand-sized or smaller. The deposition of this
material in the stream channel then has a series of adverse
effects (Chapter 4; Everest et al., 1987).
There is some evidence that an increased deposition of
fine materials may be partially self-perpetuating. In some
cases the onset of bedload transport is delayed when the
interstitial spaces are filled with fine sediment (Reid et al.,
1985). A reduced frequency of bedload transport then
provides more opportunity for the deposition of fine par-
ticles and fewer opportunities for fines to be washed out
during high flows (Beschta and Jackson, 1979).
Measurement Concepts
The characterization of bed material has been the subject
of considerable study. Pebble counts are used to develop a
particle size distribution for the bed surface material, while
bulk samplers are used to determine the particle size distri-
bution in the surface or subsurface. The selection of a
measurement technique depends on the time and equipment
available, as well as on the objectives of the sampling.
Pebble counts are a systematic method of sampling the
material on the surface of the stream bed (Wolman, 1954).
Typically a grid or transect is established, and the sizes of
100 or more particles are tabulated to establish a frequency
distribution. Since each sampled particle represents a
portion of the bed surface, the frequency distribution repre-
sents the percent of the stream bed covered by particles of a
certain size, and not the percent by volume or weight.
Particles smaller than 2-4 mm are difficult to measure in the
field and may be classified only as fines (Wolman, 1954).
Other studies estimate the size of fine particles by feel or
comparison to reference samples. Pebble counts are simple
and rapid, but there may be some bias against selecting very
small or very large particles.
A second approach to determining the particle-size
distribution of the bed material is by obtaining and sieving
bulk samples. A McNeil sampler is the most common means
to obtain a bulk sample. The McNeil sampler is a metal,
tube-shaped device that is driven into the streambed to the
desired sampling depth. Coarse material within the sample
tube is extracted by hand. By capping the tube when
extracting the corer most of the fine sediments are retained
(McNeil and Ahnell, 1964; Platts et al., 1983). The other
-------
Part II
major technique to obtain a bulk sample is to freeze a sample
of the bed material usingliquid COa or liquid nitrogen. The
frozen sample is then thawed and sieved in order to obtain
the particlesize distribution. One majoradvantage of frozen
cores is that they retain the vertical structure in the sample,
thereby permitting comparisons between particle-size dis-
tributions at different depths (Section 5.6.3). Platts et al.
(1983) discuss both these techniques in detail and conclude
that (1) neither the McNeil sampler nor the freeze core
technique is adequate when substrate particles larger than
about 25 cm are present, and (2) neither takes a completely
representative sample.
One difficulty with evaluating the extensive literature
on bed material particle size is the variation in the systems
used to classify particle sizes. Some investigators have used
many size classes, while others have used as few as six size
classes (Platts et al., 1983; Chapman and McLeod, 1987).
Each size class can be associated with a specific term (e.g.,
sand, gravel, cobbles, boulders), but these terms are not
necessarily consistent (Platts et al., 1983). The most com-
mon classification system in the U.S. is presented in Table
9. A classification commonly used in the scientific litera-
ture is the phi index, where phi = -Iog2 d, with d being the
particlediameterin mm. Useofthephi index normalizes the
particle-size distributions so they can be analyzed using
parametric statistics and plotted directly on arithmetic graph
paper (Wolman, 1954).
The selection of the sampling technique should be
determinedbytheobjectivesofthesampling. Characteriza-
tion of the bed material can be done most easily by using
Wolman pebble counts or by measuring the percent of the
bed surface covered by fines. McNeil core samples and
freeze cores both are useful in assessing the suitability of the
substrate as spawning gravel. Freeze cores can be used to
determine the variation in the particle-size distribution with
depth. Comparisons between the surface and subsurface
samples may indicatea change in thesedimentload (Dietrich
ct al., 1989; Section 5.6.3).
Standards
Currently there are no existing or proposed standards
for bed material particle size. The state of Idaho has been
considering the use of percent of fines on the bed surface as
a criterion, but this was rejected because the percent of fines
on the bed surface could not be directly linked to specific
designated uses of water (Harvey, 1988).
Current Uses
Bed material particle size has been used extensively in
research, stream classification, stream inventories, and stream
monitoring. Some monitoring projects have successfully
used visual estimates or photographic comparisons to esti-
mate particle size or percent fines (e.g., Megahan et al.,
Table 9. Classification of bed material by particle size (adapted
from Platts etal. 1983).
Size range
Class name Millimeters Inches 4>*
Very large boulders 4,096-2,048 16-80 -12 -(-11)
Large boulders 2,048 - 1 ,024 80-40 -1 1 - (-1 0)
Medium boulders 1,024-512 40-20 -10 -(-9)
Small boulders 512-256 20-10 -9 - (-8)
Large cobbles 256-128 10-5 -8 - (-7)
Small cobbles 128-64 5-2.5 -7 -(-6)
Very coarse gravel 64 - 32 2.5 - 1 .3 -6 - (-5)
Coarse gravel 32 - 1 6 1 .3 - 0.6 -5 - (-4)
Medium gravel 16-8 0.6-0.3 -4 - (-3)
Fine gravel 8-4 0.3-0.16 -3 - (-2)
Very fine gravel 4-2 0.1 6 - -2 - (-1 )
0.08
Very coarse sand 2.0 -1.0 0.08 - -1 - (0)
0.04
Coarse sand 1 .0 - 0.5 0.04 - 0-1
0.02
Medium sand 0.50 - 0.25 0.02 - 1-2
0.01
Fine sand 0.250 - 0.01 - 2-3
0.125 0.005
Very fine sand 0.125- 0.005- 3-4
0.062 0.0025
Coarse silt 0.062-0.031 - 4-5
Medium silt 0.031-0.016 - 5-6
Fine silt 0.016-0.008 - 6-7
Very fine silt 0.008-0.004 - 7-8
Coarse clay 0.004 - 0.0020 - 8-9
Medium clay 0.0020-0.0010 - 9-10
Fine clay 0.0010-0.0005 - 10-11
Very fine clay 0.0005-0.00024 - 11-12
*phi.
1980). Generally visual techniques are less sensitive and
less reliable than the more systematic and quantitative
sampling methods (Chapman and McLeod, 1987).
Both pebble counts and McNeil core samples have been
used extensively by the U.S . Forest Service to inventory and
monitor stream condition, but the resulting data remain
largely unpublished. Long-term studies on the effective-
ness of bed material particle size as a monitoring technique
are surprisingly scarce, although a number of studies have
investigated the effect of logging on bed material particle
size with varying results (e.g., Platts and Megahan, 1975;
Megahan et al., 1980; Sheridan et al., 1984; Scrivener,
1988). Probably much of this variation in results is due to
the different geologies and stream characteristics. Bed
material particle size is probably less appropriate as a
monitoring technique in areas where clays and silts pre-
dominate, or in very steep gradient streams.
-------
CHAPTER 5. CHANNEL CHARACTERISTICS
Assessment
Bed material particle size may have considerable prom-
ise for monitoring purposes as it appears to be relatively
sensitive to changing sediment loads (e.g., Megahan et al.,
1980; Platts et al., 1989). Additional effort is needed to
more precisely define the parameter(s) to be monitored, to
strengthen the link between bed surface particle size and
various designated uses, and to determine the environments
in which a bed material parameter is most useful.
The selection of a bed material monitoring parameter
should consider whether a complete particle size distribu-
tion is needed, or whether a single number, such as the dso or
percent fines, will suffice. Chapman and McLeod (1987)
suggest that geometric mean particle size and percent of the
bed surface covered by fines should both be used to define
habitat quality.
Sampling locations also need to be clearly defined. An
ideal sampling location has a high sensitivity to manage-
ment impacts and minimal response to natural events. Since
these two criteria are likely to be in conflict, detailed studies
are needed to determine the most appropriate sampling
location(s) within a stream channel. Some studies suggest
that percent fines should be evaluated within the egg pock-
ets of salmonid fishes, as these have the lowest variability
and the most direct link to a designated use (spawning
success of coldwater fishes) (Chapman and McLeod, 1987).
Chapman and McLeod (1987) reviewed the linkages
between bed material particle size and quality of fish habi-
tat. Large amounts of fine sediment clearly are detrimental
to salmonid reproduction and rearing, but quantitative rela-
tionships at lower levels of fine sediment are more difficult
to establish (Everest et al., 1987). These quantitative rela-
tionshipsalso are likely to vary among ecoregions, suggesting
a need for varying standards or criteria.
In some areas, bed material particle size may not be a
useful monitoring parameter. Steep headwater streams,
streams with a clay substrate, and low-gradient rivers all
may exhibit little change in their bed material particle-size
distribution despite a changing sediment load.
The timing of sampling also may affect the results. At
high flows the finer particles tend to be flushed or washed
fromacoarse-beddedstream. Hence sampling immediately
after a high flow may indicate a coarser streambed surface
than sampling after a relatively quiescent period (Adams
and Beschta, 1980).
These constraints in using bed material particle size for
monitoring may be alleviated by combining particle size
data with other channel parameters. Monitoring of bed
material particle size, for example.might be doneon selected
cross-sections or in selected pools and riffles within a
thalweg profile. This would permit changes in bed material
to be more directly linked to deposition or scour, as well as
to changes in the quality and amount of fish habitat Moni-
toring bed material particle size within cross-sections or a
thalweg profile also simplifies the problem of identifying
sampling sites. In general, a combination of techniques will
facilitate cross-verification and our understanding of stream
response to management activities.
5.6.2 EMBEDDEDNESS
Definition
In streams with a large amount of fine sediment, the
coarser particles tend to become surrounded or partially
buried by the fine sediment. As shown in Figure 8A,
embeddedness quantitatively measures the extent to which
larger particles are embedded or buried by fine sediment.
The measure was first used to quantify stream sedimenta-
tion in the 1970s and early 1980s (Klamt, 1976; Kelly and
Dettman, 1980). Since then the method has undergone a
series of modifications and has been used as an indicator of
the quality of over-wintering juvenile salmonid habitat
(Munther and Frank, 1986; Burns and Edwards, 1987;
TorquemadaandPlatts, 1988;Potyondy, 1988). The method
and its application continue to be improved and standard-
ized by researchers in Idaho (Skille and King, 1989) and
Montana (Kramer, 1989).
Currently variation exists in the suggested minimum
and maximum size of rocks to be measured and in the specific
feature being measured. Most researchers define the tech-
nique as cobble embeddedness, even though measurements
typically are made on all rocks with a primary axis between
4.5 cm (very coarse gravel) and 30 cm (small boulders).
Torquemada and Platts (1988) modified the method to
measure rocks as small as 1.0 cm, and the inclusion of these
smaller particles led them to use the term embeddedness
rather than cobble embeddedness.
The difficulty in measuring cobble embeddedness and
the high variability of individual measurements have stimu-
lated research into a series of related measurements. One
alternative is to measure the height of the rocks above the
bed surface, and this is termed "total free space" (Fig. 8B).
Conceptually this is similar to bed roughness, and it is an
indicator of the area protected from the current. Such areas
are important fish rearing and macroinvertebrate habitat
This measurement also has been termed "living space" by
Skille and King (1989) and "interstitial space" by Kramer
(1989).
To reduce the variability associated with measurements
from individual particles, Kramer (1989) suggested that the
total free space from all particles within a specified sample
area (typically a 60-cm diameter circle) be summed and then
divided by the area sampled. This was termed the "intersti-
tial space index" (ISI), where
ISI = SDf/Area.
-------
A.
Fine
sediment
Water
column
Plane of
embeddedness
B.
Fine
sediment
Plane of
embeddedness
c.
Free matrix particles
Plane of
embeddedness
Rgure 8. Schematic representation of the three main embeddedness measurements—embeddedness, free space, and free matrix
particles. Dm represents the length of the primary axis. A. Embeddedness for a single particle is equal to De/Dt. B. Free space for
a single particle is equal to D| (note: D| = Dt - De). C. Free matrix particles. (Adapted from Burns and Edwards, 1985.)
-------
CHAPTER 5. CHANNEL CHARACTERISTICS
An average ISI can be determined for each sampled
stream reach. ISI appears to be more sensitive to change
than cobble embeddedness, and it also is more directly re-
lated to the designated use of streams for fisheries.
A third embeddedness measure (Fig. 8C) is the percent
of free matrix particles. Free matrix particles are defined as
those rocks (typically 4.5-30 cm along the primary axis)
having zero embeddedness (Fig. 8C; Burns and Edwards,
1985). Percent free matrix is calculated by dividing the
number of free matrix particles by the total number of
similarly sized particles within the sampled area. Percent
free matrix particles correlates closely with percent embed-
dedness (Burns andEdwards, 1985; TorquemadaandPlatts,
1988; Munther and Frank, 1986; Potyondy, 1988).
Relation to Designated Uses
Cobble embeddedness has both biological and physical
significance. Biologically, areas with a high embeddedness
have very little space for invertebrates or juvenile fish to
hide or seek protection from the current. The accumulation
of fines also fills in the spaces between larger particles, and
this limits the interstitial habitat. Similarly, the reduction in
surface area associated with increasing embeddedness (de-
creasing total free space) limits the attachment area for
periphyton.
Chapman and McLeod's (1987) review noted lower
aquatic insect densities when embeddedness exceeded 65-
75%. Salmoniddensityalsodeclinedwithanembeddedness
of 50% or more. It was inferred that an increase in embedded-
ness wouldreduce winter habitat, with the precise relationship
varying according to the fish species and fish population
density.
The physical effects of embeddedness are similar to the
effects of a decrease in bed material particle size discussed
in Section 5.6.1. Increasing embeddedness decreases chan-
nel roughness, and the resulting reduced bed friction losses
will have repercussions on the stream hydraulics and overall
channel morphology. Total free space is closely related to
bed roughness and may be proportional to Manning's "n."
The fine particles associated with increasing embed-
dedness adversely affect gravel permeability and inter-
gravel dissolved oxygen. Chapman and McLeod (1987)
note that an abundance of fine particles in the interstices of
the bed may delay the onset of bed movement during high
flows, and this in turn could facilitate the accumulation of
fine particles.
Response to Management Activities
The use of cobble embeddedness for water quality
monitoring presumes that increasing embeddedness re-
flects an increased input of fine sediments to the stream
channel. Measurements of embeddedness on 19 tributaries to
the South Fork Salmon River in Idaho indicated that streams
in heavily roaded and logged watersheds had a significantly
higher cobble embeddedness than undisturbed watersheds
(Burns and Edwards, 1985). No differences were found
between undisturbed and partially disturbed watersheds.
In 1986 embeddedness was sampled on 120 streams in
the BoiseNational Forest (Potyondy, 1988). No statistically
significant differences in mean embeddedness were found
between developed and partially developed watersheds.
Nevertheless, the study concluded that there is a relation-
ship between mean embeddedness and sediment-producing
activities, but both natural and management-induced factors
are important in determining embeddedness levels.
Studies on the Payette National Forest in Idaho com-
pared embeddedness levels in watersheds with different
degrees of mining activities (Burns and Ries, 1989). The
study concluded that at least 5 consecutive years of data are
needed to evaluate trends in embeddedness.
Measurement Concepts
The basic procedure for measuring embeddedness is to
select a particle, remove it from the streambed while retain-
ing its spatial orientation, and then measure both its total
height (Dt) and embedded height (De) perpendicular to
the streambed surface (Fig. 8A). Percent embeddedness is
calculated for each particle until at least 100 particles are
measured. Individual embeddedness values are averaged to
yield a mean embeddedness value.
The technique as modified by Skille and King (1989)
uses 60-cm diameter hoops as the basic sample units. The
total height (Dt) and embedded height (Dc) are measured
for each particle which meets the specified size criterion.
The individual values of Dt and De from each hoop are
summed, and a percent cobble embeddedness (PCE) for
each hoop is calculated from the formula:
An average of the PCE values from all the hoop samples
yields the percent cobble embeddedness for the sampled
reach.
The number of hoops needed to characterize a site de-
pends on the variability among hoop samples and the desired
level of precision. A general rule is that one reach requires
approximately 20 hoops (approximately 500-700 particles)
and may require up to 1 full day for a two-person field crew
to complete.
The use of hoops rather than individual particles as the
basic sampling unit substantially increases the number of
particles that must be measured, but reduces the variability
among sample units. This makes it easier to detect change
(Parti, Section 3.4.2) and results in an embeddedness value
that more closely represents the condition of the stream
reach. The earlier technique of using individual particles as
the sample unit may be more applicable within one habitat
type where the variability is likely to be lower.
-------
Part II
In developing a monitoring plan using embeddedness,
the objectives will dictate whether hoops or individual
particles should be the sample unit. To characterize a stream
reach with different habitat types, Skille and King (1989)
suggest three randomly spaced hoop samples along cross-
sectional transects placed two stream widths apart.
The embeddedness value for a randomly placed hoop
should be adjusted if the hoop incorporates a substantial
area of fine sediments with no exposed rocks (Torquemada
andPlatts, 1988). Failure to correct for the area occupied by
fines will cause embeddedness to be underestimated. The
corrected value is known as the weighted embeddedness,
and it is defined as:
WE =
HA x 100 + fl-HA->B
100
where WE = percent weighted embeddedness,
HA = percent of hoop area occupied by fines,
and
E = percent embeddedness.
Skille and King (1989) suggest that the weighted value
should be used if more than 10% of the surface area within
the hoop is occupied by fine sediment.
The size of the particles and the diameter of the hoop can
be adjusted according to the type of stream. Most recent
studies have used hoops 60 cm in diameter and measured all
particles with a primary axis of 4.5-30.0 cm. Fines are
usually defined as particles less than 6.4 mm (0.25 inches)
in diameter. These particle and hoop sizes are believed
appropriate for streams up to 20 feet wide and with a
gradient of up to 3% (Skille and King, 1989). Torquemada
and Plaits (1988) modified the method for use in smaller
streams by reducing the hoop diameter to 30 cm and de-
creasing the minimum rock size to 1.0 cm.
The time required to evaluate embeddedness can be
substantially reduced by measuring the height of free matrix
particles and counting the remaining embedded particles.
Since therelationshipbetweenpercentcobbleembeddedness
and percent free matrix particles may vary according to
stream order, geology, climate, etc., inferences about per-
cent embeddedness cannot be made from free matrix data
until the interrelationship has been defined for that site.
If the monitoring objective is to evaluate changes in the
deposition of fine sediments, the interstitial space index
(ISI) may be the preferred embeddedness parameter. Both
the ISI and percent embeddedness can be calculated from
one set of field measurements.
Standards
The State of Idaho Water Quality Bureau currently is
proposing a cobble embeddedness criterion. This specifies
that cobble embeddedness in fry overwintering habitat
should not exceed natural baseline levels at the 95% confi-
dence level. Baseline levels of cobble embeddedness are to
be determined in similar watersheds that are unaffected by
nonpoint sediment sources (Harvey, 1989).
Current Uses
Ongoing, unpublished studies by federalandstateagencies
are measuring embeddedness as one means to assess the
effects of land management activities on streams. Use of the
revised measurement techniques and more intensive sampling
should allow a better evaluation of the usefulness of embed-
dedness to monitor the effects of management activities.
Currently embeddedness is being measured in a number of
National Forests, particularly in Idaho and Montana. Embed-
dedness also is part of the Forest Practices BMP Effectiveness
Monitoring Program in Idaho. In Washington four classes of
embeddedness are being visually estimated in the Timber-
Fish-Wildlife stream survey program. These field applications
will help evaluate the methodology for measuring embed-
dedness and determine its usefulness for assessing the effects
of past and present management activities.
Assessment
Current research and monitoring efforts should help
clarify the links between embeddedness, other characteris-
tics of the stream channel, and fisheries. Measurement of
one or more embeddedness parameters (percent cobble em-
beddedness, total free space, or percent free matrix par-
ticles) probably will proveuseful only in certain environments
andstream types. Mostof the work on embeddedness has been
conducted in granitic basins in Idaho, and embeddedness
may not be as appropriate in basins where most of the
anthropogenically induced sediment load is comprised of
silts and clays. Similarly, embeddedness may not be a
useful monitoring parameter in high-energy, steep gradient
channels where deposition of fine particles is unlikely. Low
gradient downstream reaches may lack the coarse particles
needed to measure embeddedness.
The strong interest in embeddedness as a monitoring
parameter is due to the recognition that sediment often is the
most important pollutant from forest management activities
in the Pacific Northwest and Alaska. Hence there is a great
need for reliable methods to evaluate sediment inputs and
the resultant effects on the designated uses of the water.
Embeddedness has shown promise, but the immediate need
for a monitoring technique has resulted in widespread use
and adaptation before cobble embeddedness could be ad-
equately field-tested and validated. Users should be aware
that the various embeddedness techniques are likely to
undergo further changes and improvements, and this could
severely limit the comparability of data collected over time.
-------
CHAPTER 5. CHANNEL CHARACTERISTICS
5.6.3 SURFACE vs. SUBSURFACE
PARTICLE SIZE DISTRIBUTIONS
Definition
The bed material in alluvial stream channels consists of
mixed grain sizes. Often the surface of the bed is coarser
than the underlying material. This armoring or pavement
has been attributed to a settling of the smaller particles down
into the bed during active transport (Parker and Klingeman,
1982), and selective transport of finer particles when the
larger particles are immobile (Sutherland, 1987). Surface
coarsening has been observed downstream of dams when
bedload was eliminated (Shen and Lu, 1983; Bradley and
Smith, 1984).
An alternative to this "equal mobility" explanation for
the armoring of gravel-bedded streams and rivers is that the
armoring is a result of the sediment supply being less than
the sediment transport capacity (e.g., Kinerson and Dietrich,
1989). If one assumes that the subsurface particle size
distribution is similar to the particle size distribution of the
bedload (e.g., Parker et al., 1982) and that the banks are
relatively resistant to erosion, then the difference between
the surface and subsurface particle size distribution should
be quantitatively linked to the sediment supply (Dietrich et
al., 1989). A dimensionless ratio, q*, has been defined as the
estimated bedload transportrate for the median grain size on
the bed surface divided by the estimated bedload transport
rate for the median grain size of the subsurface material
(Dietrich etal., 1989).
Under this hypothesis streams with a high sediment load
and no surface coarsening should have a high q*, while
streams with a low sediment load should have a well-
developed coarse surface layer and a low q*. With an
increased sediment load, streams that initially had a low q*
would experience a fining of the bed surface material. With
a higher q*, relatively little of an increased sediment load
could be accommodated by a fining of the bed surface, and
the stream would be more subject to aggradation, pool fill-
ing, and overall channel instability. An increased sediment
supply also would lead to a greater proportion of the stream
bed being occupied by finer materials (Kinerson, 1990).
Relation to Designated Uses
Th,e effects of an increase in the sediment supply, and
the corresponding fining of the bed surface relative to the
subsurface, have been discussed in Chapter 4 and in Sec-
tions 5.6.1-5.6.2. Briefly, an increase in fine sediment will
decrease the permeability of the bed material in alluvial
channels, which will decrease intergravel DO (Section 2.2)
and degrade spawning habitat. A predominance of fine
sediment decreases macroinvertebrate biomass and diver-
sity (Chapman and McLeod, 1987; Everest et al., 1987).
Mean particle size in the bed material is inversely correlated
with habitat suitability for aquatic insects and fish (Chapman
and McLeod, 1987). By reducing pool depth and pool
volume, sediment deposition reduces the suitability of a
stream for adult fish (Section 5.3). Increasing embedded-
ness and surface fines reduce winter carrying capacity for
salmonids in the northern Rockies (Section 5.6.2). Com-
prehensive reviews of the effects of sediment on aquatic
organisms are presented in Chapman and McLeod (1987)
and Everest et al. (1987). Scrivener (1988) summarizes the
forest management-sediment-fisheries interactions for the
Carnation Creek study in coastal British Columbia.
Effect of Management Activities
The impact of forest management on sediment produc-
tion is discussed in Chapter 4. Swanson et al. (1987) and
Everest et al. (1987) both provide excellent overviews of
natural sediment production rates, the processes governing
the input of sediment into streams, the impact of sediment
on aquatic ecosystems, and the extent to which forest
management activities are likely to increase sediment pro-
duction rates. Swanson et al. (1987) conclude that mass
failures are the dominant source of sediment, but the pro-
cesses that deliver sediment to the stream channel are more
variable. Forest management activities—particularly road
building, poor road maintenance, and the combination of
clearcuttingandbroadcast burning—usually have the great-
est effect on sediment yields. Steeper basins appear to be
more sensitive to management impacts, and evaluating man-
agement impacts is complicated by the random occurrence
and potential impact of large storm events (Swanson et al.,
1987). Everest et al. (1987) note that while the felling and
bucking of trees can have minimal impact on fine sediment
production and yield, roads, tractor logging, and ground-
disturbing site preparation activities tend to have a much
larger impact.
A long-term study in the South Fork of the Salmon River
in Idaho showed that for the first 10 years after a logging
moratorium was imposed the percent of fines (<4.75 mm in
diameter) declined relatively rapidly in both the surface and
subsurface layers (Box 3, page 19; Platts et al., 1989). This
was followed by a period of less rapid decline, and from
about 1981 to 1985 there was a small increase in percent
fines. Surface fines were removed more rapidly than sub-
surface fines because they were more exposed to the shear
stress imposed by the flowing water. On the other hand, once
an apparentstateof equilibrium had been reached, the percent
of fines in the surface layer remained at approximately half
the concentration found in the subsurface layer. Consider-
able variation was found between monitoring sites, and this
was partly attributed to differences between low-gradient
spawning areas and higher-gradient rearing habitats (Platts
etal., 1989).
-------
Part II
The Carnation Creek study in coastal British Columbia
monitored changes in particle size distribution in the top (0-
15 cm) and bottom (15-30 cm) layers of bed material over
a 13-year period. Within and below the area of intense
streamside logging, the accumulation and cleansing of fines
was highly responsive to both the input of sediment and the
occurrence of runoff events. Chronic sedimentation resulted
in fines penetrating deeper into the streambed, and these
deeper layers were much slower to recover because scour to
these depths was much less frequent. Hence the annual rate
of change in the particle-size distribution declined with
increasing particle size and increasing depth. Significantly,
8 years after the intensive logging treatment the changes in
gravel composition were still accelerating, and fine par-
ticlesweresdllaccumulatinginthedeeperlayers(Scrivener,
1988).
Measurement Concepts
Different techniques can be used to sample the surface
and subsurface bed material (Section 5.6.1). Particle-size
distributions for the bed surface can be obtained by pebble
counts, McNeil samplers, or freeze cores. Pebble counts
allow rapid determination of the particle-size distribution of
the surface layer, but this method cannot be used for the
subsurface layers. McNeil samplers do notallow separation
of material by depth, and this limits their use to situations
where separate samples can be taken from the surface and
subsurface layers. Freeze cores sample both the surface and
subsurface layers, and they preserve the spatial structure of
the sample. However, freeze cores are difficult to obtain in
the field, and—like McNeil samplers—they are limited in
terms of the maximum particle size that can be sampled
(Platts et al., 1983; Section 5.6.1).
Data on the particle-size distribution in the surface and
subsurface layer can be analyzed in several different ways.
The simplest method is to compare the median (dso) particle
size of the surface and subsurface materials. Since quite
different particle-size distributions can have a similar dso
(Platts etal., 1983), comparisons generally should incorpo-
rate some measure of the variation in the particle-size
distribution, such as the d84 and the d16 (where d is diameter,
and the number is the percent of particles that are smaller
than the specified percentage). In cases where the particle
size distribution of the surface and subsurface layers is
known, one should consider developing a statistical mea-
sure of the differences between the two distributions.
Standards
No standards for the relationship between surface and
subsurface particle-size distributions have been established
or proposed.
Current Uses
Values of q* have been determined for a series of flume
experiments (Dietrich et al., 1989) and a number of streams
in California with a widely varying sediment supply
(Kinerson and Dietrich, 1989). The data collected to date
shows that rivers and streams with a high sediment supply
generally lack a coarse surface layer and have a q* close to
1.0. Considerable local variation occurred within- stream
reaches. In sediment-rich streams, for example, areas with
an armor layer and a low q* could be found immediately
downstream of debris jams and other obstructions which
functioned as sediment traps (Kinerson and Dietrich, 1989).
Chapman and McLeod (1987) also noted large differences
in particle-size distributions between salmonid egg pockets
and immediately adjacent areas. This instream variability
should be minimized by selecting relatively straight, fea-
tureless reaches with little form roughness.
Some studies on the infiltration of fine sediment into
gravel layers or redds suggest that further work is needed
before the difference in particle-size distribution or q* can
be adopted as a monitoring technique. Beschta and Jackson
(1979) showed that the relative size differences between
coarse and fine bed material can greatly affect the behavior
of fine particles. When sands with a median particle size of
0.5 mm were added to a clean gravel bed with a median
particle size of 15 mm, the sand was trapped in the intersti-
tial spaces within the uppermost top 10 cm. Reducing the
median diameter of the sand to 0.2 mm allowed the sand to
filter down through the gravel and the interstitial voids were
filled from the bottom of the flume upwards. At Carnation
Creek the fine (sand-sized) particles intruded into the gravel
a few centimeters below the depth of scour, and they were
not winnowed out until a subsequent event scoured to that
depth. These results supgest that monitoring the bed mate-
rial particle size in the surface layer may be best for evalu-
ating short-term changes, but a comparison of the surface
and sub-surfaceparticle size distributions provides a longer-
term perspective on the amount and type of sediment load.
Some of the complexities of the interactions between fine
sediment and alluvial streambeds were recently reviewed
by Jobson and Carey (1989).
Empirical support for the use of q* or a similar measure
can be derived from field observations of salmonid redds.
Chapman and McLeod (1987) cite several studies in which
it was observed that a seal of fine particles formed over the
clean gravels createdby the spawning female. In these cases
the deposition of fine sediments also may be affected by the
special hydraulics associated with the redd.
Assessment
The relationships between sediment supply, sediment
transport capacity, and the surface and subsurface particle-
size distribution are in a state of active investigation. Both
-------
CHAPTER 5. CHANNEL CHARACTERISTICS
theoretical considerations and preliminary field data sug-
gest that differences between the surface and subsurface
particle-size distributions can be used to monitor changes in
sediment supply. However, characterizing the subsurface
particle-size distribution is not a simple task, and this may
ultimately limit the usefulness of q* or an associated mea-
sure as a monitoring technique.
The primary advantage of comparing surface and sub-
surface particle-size distributions is that this appears to
provide an immediate assessment of the sediment transport
capacity in relation to the sediment supply. The use of q*
normalizes the surface conditions against the particle-size
distribution and predicted bedload transport capacity of the
subsurface layer. This yields a single index for evaluating
current conditions and comparing different streams. How-
ever, if one is concerned solely with changes over time and
has time-trend data available, the surface particle-size dis-
tribution could serve as the primary monitoring parameter.
S urface and subsurface comparisons may not be neces-
sary if predictions could be made of the bed surface particle-
size distribution for streams in the absence of management
activities. In other words, if the undisturbed particle-size
distribution can be predicted from channel characteristics or
by comparisons to other streams, the actual bed surface
particle-size distribution could be compared to the predicted
distribution in order to evaluate stream condition.
In summary, a variety of studies suggest a direct rela-
tionship between an increase in sediment supply due to land
use and a change in the surface particle-size distribution. In
most cases, however, there already have been some adverse
land use impacts, and no data are available on the pre-
disturbance particle-size distribution of the bed surface.
Under these circumstances a comparison of the surface and
subsurface particle-size distributions may yield a quantita-
tive measure of the sediment supply relative to the sediment
transport capacity. The variability of such a measure within
a particular stream reach, and the complexity of sediment
transport in alluvial channels, mean that additional work
will be needed before q* or a similar measure can be adopted
as a standard for monitoring management impacts.
5.7 LARGE WOODY DEBRIS
Definition
Large pieces of wood in streams have been referred to by
a variety of names including large organic debris (LOD),
coarse woody debris(CWD),andlarge woody debris (LWD).
The type and size of material included in this designation
has varied according to the objectives of the person measur-
ing the debris. Studies on the energetics of stream systems
have included material as small as 2.5 cm in diameter as LWD
(e.g., Harmon etal., 1986). However, studies of the effects of
woody debris on channel morphology typically use a much
larger minimum size for LWD—usually 10 cm in diameter
and2min length (Sedell etal., 1988; Bilby and Ward, 1987).
The amount of LWD in stream channels depends on a
variety of factors. Stream size is an important determinant,
with smaller streams usually containing more wood than
larger systems (Swanson et al., 1982; Bilby and Ward,
1987). Riparian tree density is positively related to LWD
amount in streams in eastern Washington (Bilby and
Wasserman, 1989). Bed characteristics .also have, been
shown to influence LWD amount, as streams with boulder
or bedrock substrates typically contain only about half the
LWD compared to streams with finer substrates (Bilby and
Wasserman, 1989). Catastrophic events, such as major
windstorms or landslides, also have a major impact on the
amount and location of LWD in some stream channels
(Keller and Swanson, 1979; Bisson et al., 1987).
Stream size plays a major role in determining the size of
LWD in stream channels as well as the amount of LWD.
Generally, the average size (diameter, length, or volume) of
LWD in a stream channel increases with increasing stream
size (Bilby and Ward, 1987). This increase is caused by the
increased capacity of larger channels to move material
downstream. Thus, in larger channels, smaller wood is
selectively flushed from the system or deposited on the
floodplains, leaving only the larger pieces. This causes a
decrease in the amount of LWD, but an increase in average
piece size. Pieces of wood with a low probability of being
moved by the stream are most important in influencing
channel morphology (Bilby and Ward, 1987). In general,
pieces one-half the channel width in length or longer are
regarded as being relatively stable (Bisson et al., 1987).
Relation to Designated Uses
LWD influences stream systems and their biota in a
number of ways. Large wood has a major impact on channel
form in smaller streams (Sullivan etal., 1987). The location
and orientation of LWD can influence channel meandering
and bank stability (Swanson andLienkaemper, 1978; Cherry
and Beschta, 1989). LWD tends to cause both a greater
variability in channel width and an increase in average
channel width (Keller and Swanson, 1979). LWD also
forms and stabilizes gravel bars (Lisle, 1986).
LWD is often the most important structural agent form-
ing pools in small streams. Bilby (1984b) reported that over
80% of the pools in a small stream in southwest Washington
were associated with wood. Similarly, Rainville et al.
(1985) found that 80% of the pools in a series of small
streams in the Idaho Panhandle were wood-associated.
While the relative importance of LWD in pool formation
decreases with increasing channel width, wood in large
rivers forms pools along the channel margins or in second-
ary channels, and these pools may be very important for fish
populations (Bisson et al., 1987).
-------
Part II
Another way in which wood affects channel shape is by
forming waterfalls. Waterfalls form plunge pools and also
influence sediment transport in streams.- The greater the
proportion of the drop in elevation of a stream caused by
waterfalls the less efficient the system is at moving sediment
downslope (Heede, 1972). The proportion of channel drop
accounted for by summing the heights of LWD-caused
waterfalls ranged from 30 to 80% in streams in the western
Oregon Cascades (Keller and Swanson, 1979). In streams
in the Oregon CoastRange, wood caused 6% of the total fall
(Marston, 1982). In western Washington the proportion of
elevation drop caused by LWD was found to decrease with
increasing stream size. LWD accounted for >15% of the
elevation drop in stream channels <10 m wide, but <5% of,
the elevation drop in channels 10-20 m wide.
LWD also influences sediment transport in streams by
forming depositional sites. Wood wasresponsiblefor storing
half the sedimentin several small streams in Idaho (Megahan
and Nowlin, 1976), The importance of wood in retaining
sediment in small streams has been demonstrated by the
release of very large amounts of material after removal or
disturbance of LWD (Baker, 1979; Beschta, 1979).
LWD also can provide storage sites for leaves, twigs,
and other organic material. In small streams in forested
areas, this fine organic material can provide the bulk of the
energy and materials entering into the aquatic food web. In
the absence of LWD, much of the terrestrial organic matter
entering the stream is flushed rapidly downstream with little
opportunity for the biota to utilize this material (Bilby and
Likens, 1980).
LWD is oneof themostimportantsources of habitat and
cover for fish populations in streams. Most of the work
documenting this function of LWD has been done on
salmonids in the Pacific Northwest (Sedell et al., 1984;
Bisson et al., 1987; Sedell et al., 1988). Generally there
appears to be a direct relationship between the amount of
LWD and salmonid production; no known data indicate an
upper end to this relationship (Bisson et al., 1987). One of
the key functions of LWD with regard to fish production is
to increase habitat complexity, and this helps ensure that
cover and suitable habitat can be found over a wide range of
flow and climatic conditions. LWD also may allow a finer
partitioning of the available habitat. Pools formedby LWD,
for example, are favored habitat by certain species and age
groups of salmonids (Bisson et al., 1982). More complex
wood structures, such as rootwads or small debris jams,
attract more fish than single logs (Sedell et al., 1984;
McMahon and Hartman, 1989). In a number of experi-
ments, wood removal has been demonstrated to reduce fish
population densities (Lestelle, 1978; Bryant, 1983; Dolloff,
1986; Elliott, 1986; Bisson etal., 1987).
Several potentially detrimental effects are associated
with LWD in streams. Historically, massive wood accumu-
lations on larger rivers impeded navigation. Most of these
accumulations were removed around the turn of the century
(Sedell and Luchessa, 1982). In most large rivers today,
wood is found primarily along the channel margins or in off-
channel areas (Bisson et al., 1987) and therefore poses little
hazard to navigation.
Movement of wood during high flow events may dam-
age structures located in or near streams. Large woody
debris may also increase flood damage by partially blocking
the channel during high flow events (e.g., Griggs, 1988).
Often the risk of damage is exacerbated by ill-advised
development in floodplain areas and changes in the hy-
drologic regime due to changes in land use. Most flood-
routing models ignore the potential for LWD to influence
water movement through a drainage system, even though
this can greatly restrict channel capacity (P. Williams, P.
Williams & Assoc., Ltd., San Francisco, CA, pers. comm.).
Large wood accumulations may form blockages to the
passage of anadromous fishes. For many years this was
perceived as a serious problem, and wood was removed
from channels to prevent the formation of blockages. How-
ever, many LWD accumulations which appear to be
blockages at low flows are passable at higher discharges. In
addition, these blockages normally occur in steeper chan-
nels where spawning and rearing habitat for anadromous
fish is limited. Historical estimates suggest only 5-20% of
available anadromous fish habitat was inaccessible because
of debris blockages (Sedell et al., 1984).
While LWD may contain some compounds toxic to
stream biota, under most conditions leaching of these mate-
rials occurs at a very slow rate. This keeps concentrations
well below toxic levels (Bisson etal., 1987). Similarly.LWD
is seldom a cause for low dissolved oxygen concentrations in
stream water. Wood is relatively resistant to decomposition,
and LWD has a low surface area to volume ratio. Taken
together, these two factors result in LWD having a low
biochemical oxygen demand (Bisson et al., 1987).
Response to Management Activities
Historically, the amount of LWD in streams has been
reduced as a result of several management practices. Wood
in larger river systems was removed to improve navigation
and reduce flooding hazards, at the turn of the century
(Sedell and Luchessa, 1982). Extensive clearing of wood
from smaller streams was conducted through the early
1980s to reduce bank and bed scour and provide upstream
passage for anadromous fish (Bilby, 1984b; Sedell et al.,
1988). After channel clearing, much of the residual debris
is unstable and is flushed from the stream channel, further
reducing the amount of LWD (Bilby, 1984b).
The practice having the most widespread influence on
LWD in Pacific Northwest streams has been the harvest of
trees from riparian areas. Although the amount of LWD in
streams may increase immediately after harvest owing to
the introduction of logging slash, much of this material is
rapidly decomposed or flushed from the system by high
-------
CHAPTER 5. CHANNEL CHARACTERISTICS
flows. Harvest of the larger trees in the riparian zone
removes the primary source of LWD, and this results in a
gradual decrease in wood over time as inchannel material
decomposes or is moved downstream (Swanson and
Lienkaemper, 1978; Grette, 1985; Bisson efal., 1987).
Typically the average piece size also declines because of the
introduction of smaller pieces of wood and the relatively
small size of the LWD contributed by the second-growth
riparian vegetation (Sedell et al., 1988). Recent research
indicates that the decrease in LWD following removal of
riparian vegetation may occur much more rapidly than
previously thought (B. Bilby, Weyerhaeuser Co., pers.
comm.).
The length of time needed for riparian areas to produce
LWD after harvest depends upon the size of the stream.
Measurable contributions of wood from second-growth ri-
parian areas did not occur until 60 years after harvest for
third-order channels on the Olympic Peninsula (Grette,
1985). Bilby and Wasserman (1989) indicate that it takes
longer than 70 years for streamside vegetation to provide
stable material to streams wider than 15 m in southwestern
Washington. Thus larger streams are likely to be deficient
in LWD for a longer period of time after timber harvest than
smaller streams.
A decline in the amount and average size of LWD in
streams following timber harvest leads to a reduction in
waterfalls, a decrease in pool frequency and size, and a
decrease in the amount of sediment and finer organic matter
retained by LWD (Bilby, 1984b; B. Bilby, Weyerhaeuser
Co., pers. comm.). However, in some instances an increase
in pool frequency has been associated with a decrease in
LWD due to the replacement of large pools by numerous
small pools (McDonald and Keller, 1983).
Relatively little is known about the importance of up-
stream source areas for maintaining LWD in larger rivers. If
upstream areas are an important source of LWD in down-
stream areas, any reduction in LWD in these smaller up-
stream channels could have important off-site impacts.
However.thecapacityofasystem to transport wood increases
in a downstream direction. This suggests that on-site
recruitment from the riparian area is the most likely source
of stable LWD in larger rivers.
Measurement Concepts
Platts et al. (1987) provide arecent review of techniques
to measure and map LWD. Selectinga methodology depends
upon the objectives of the monitoring. Measurement
techniques vary widely in terms of effort required, and they
range from a simple enumeration of pieces to a detailed
description of the characteristics and location of each piece.
More detailed descriptions might include measuring the
size of each piece, mapping the associated channel charac-
teristics, and noting the location and orientation of each
piece relative to a permanent benchmark. An alternative
procedure for monitoring the stability of LWD is to tag and
relocate each piece on an annual or storm basis.
Various criteria have been employed to delineate those
pieces of wood to be included in an LWD survey. Most
surveys include only those pieces which extend below the
waterline at bankfull discharge and exceed some minimum
dimensions. Surveys measuring the biomass of organic
material in stream systems will use a smaller minimum size
than studies of LWD influences of channel morphology or
fish habitat (Harmon et al., 1986).
The cost of monitoring LWD increases considerably if
volume or biomass estimates are needed, as this requires at
least the length and diameter of each piece. Length mea-
surements may include the entire piece or just that portion
extending below the bankfull channel. Diameter may be
measured at the mid-point of the piece or by averaging the
diameter at both ends. Probably the most efficient proce-
dure to determine volume or biomass is to visually estimate
the length and diameter and then correct the visual estimate,
by measuring a subsample of the pieces (Hankin and Reeves
1988). Biomass of LWD also can be estimated with tech
niques derived from inventories of forest residues (Va
Wagner, 1968). This procedure inventories all LWD intei
sected by a series of cross-sections across the stream (Froelic
etal., 1972;Lammel, 1972),andismostapplicablewhenth
minimum piece size is relatively small. Special procedure
or categories may be needed for measuring debris jams
standing trees, and snags within the stream channel (Platt
etal., 1987).
Information on channel features associated with LWE
is sometimes collected during surveys. Data may include
the following:
• type of habitat unit or channel feature
• surface area, or volume of wood-associated pools
• surface area or volume of sediment stored behind
LWD
• number and heights of waterfalls; and
• volume or biomass of fine organic matter
(Bisson et al., 1982, 1987; Platts et al., 1983, 1987;
Bilby and Ward, 1987). Surveys of habitat types or channel
features also may include data on the presence or absence of
LWD (e.g., Ralph, 1989).
Standards
Standards for LWD in streams have not been estab-
lished for any state, although an attempt was made in the
development of Washington's forest practice regulations to
maintain wood levels at those seen in old-growth stands
(Bilby and Wasserman, 1989). However, LWD amounts
and characteristics vary as a function of stream size, veg-
etation type, and other factors, thus inhibiting the estab-
lishment of strict numerical standards.
Most Pacific Northwest states have established Best
Management Practices (BMPs) to control the adverse effects
-------
Part II
of forest management on stream channels and riparian areas.
The most recent revisions of these BMPs have incorporated
provisions for retaining LWD in streams and ensuring a
continuing supply from the riparian area. Approaches cur-
rently in use or beingconsideredincludedefining strips along
the stream in which no harvest is permitted (e.g., Alaska),
establishing specific numbers of trees to be left along the
stream (e.g., Washington), or establishing a minimum basal
area which must be retained along the stream (e.g., Oregon).
Generally, the regulations applying to larger, fish-bear-
ing waters are more stringent than those used on smaller
streams. On larger streams the disturbance of inchannel
debris, orremoval of standing timberfrom the riparian area,
is generally prohibited or restricted. Thus LWD in the
channel is protected, slash introduction during timber har-
vest is reduced, and the future source of LWD for the
channel is retained.
Although some states have developed regulations to
restrict forest management activities near smaller streams,
frequently slash is introduced to these smaller channels
during timber harvest In cases where the amount of slash
entering the channel is considered to pose a threat to down-
stream resources, cleaning of the channel may be required.
Factors considered in deciding whether or not to remove a
piece of wood from the channel include the size of the
woody debris and the extent to which it is embedded in the
streambank or channel.
Current Uses
Recentprograms to inventory stream condition and fish
habitat on forest lands usually include some measurements
of LWD. Generally the LWD measurements focus on the
number and size of LWD pieces and their association with
various channel features. As most of these programs are of
recent origin, relatively little of the resulting data have been
used to develop management prescriptions (Bilby and
Wasserman, 1989).
When possible, comparable surveys should be conducted
on similar, unmanaged streams. For example, upstream wil-
derness areas can provide reference data on the natural
loading.recruitmentrate.anddownstream transport of LWD.
Such comparisons of logged and unlogged reaches can
provide insights into management impacts on LWD. How-
ever, the long residence time of LWD in streams suggests
that the ultimate impact of forest harvest on amounts,
characteristics, and functions of LWD may not be evident
for years or decades.
Assessment
Large woody debris (LWD) performs a variety of func-
tions critical to the maintenance of productive fish habitatin
stream systems. Various management activities, including
limber harvest, alter the amount and characteristics of LWD
in Pacific Northwest streams; therefore, monitoring activi-
ties evaluating stream conditions on forest lands should
incorporate measurements of LWD. This need to monitor
LWD is increasingly recognized, but monitoring programs
with a LWD component are only now being established.
The types of measurements which should be taken will
depend upon the objectives of the specific monitoring
project, but should include, as a minimum, wood abundance
and piece size.
Logging and fish habitat improvement projects are the
two activities most likely to alter the amount of large woody
debris in stream channels. On-site measurements of wood
frequency and piece size can be a relatively sensitive indi-
cator of management impacts. In downstream locations
changes in the LWD size and frequency usually occur more
slowly and may not be easily detectable.
The long time required for a tree to mature and enter into
the stream channel suggests that one should monitor the
vegetation in the riparian zone and plan for future recruit-
ment (Section 5.2). Hence long-term monitoring of large
woody debris in stream channels is needed to fully assess the
adequacy of present practices, whereas a simple inventory
may suffice for evaluating conditions with regard to fish
habitat, channel morphology, and sediment storage.
The extensive changes in forest practice regulations
over the last twenty years means that long-term trends in
LWD must be evaluated in the context of the regulations in
force atthetimeofthe management activity. Hence the data
from long-term monitoring projects may not be directly
applicable to current practices, but they can provide some
guidance to the formulation of future regulations.
5.8 BANK STABILITY
Definition
Stream and river banks control limit the lateral move-
ment of water. Typically the bank areas can be identified by
a change in substrate and a break in slope between the
channel bottom and the stream banks. In many streams the
slope of the bank exceeds 45° (Platts et al., 1987).
Bank stability is a rather imprecise term that refers to the
propensity of the stream bank to change in form or location
over time. In alluvial channels the stream and river banks
tend towards a dynamic equilibrium with the discharge and
sediment load. The bank material, vegetation type, and
vegetation density also affect the stability and form of the
streambanks (Platts, 1984). Change in any one of these
factors is likely to be reflected in the size and shape of the
stream channel, including the banks (Chapter 5).
Even in undisturbed streams some bank instability
usually occurs. In valleys with a defined floodplain there is
often lateral migration through bank erosion and point bar
accretion (e.g., Leopold et al., 1964; Ritter, 1978). In V-
shaped valleys there is less opportunity for lateral migra-
-------
CHAPTER 5. CHANNEL CHARACTERISTICS
tion, and bank instability may stem from the input and
eventual removal of obstructions emanating from fallen
trees, landslides, or debris flows.
A higher incidence of bank instability can be initiated by
naturalevents that disruptthequasi-equih'briumofthe stream,
or by human disturbance. Extreme floods, wildfires, and land-
slides are three examples of short-term disturbances likely to
affect channel form and bank stability. Climatic and tectonic
change are two long-term processes that affect discharge,
sediment load, and channel stability, but the time scale of
these changes is well beyond the range of current water
quality monitoring efforts. The ways in which human activi-
ties alter the discharge, sediment load, and streamside veg-
etation cover arediscussed in ChaptersS, 4 and 6, respectively.
Relation to Designated Uses
Bank stability can be an important indicator of water-
shed condition and can directly affect several designated
uses. Unstable banks contribute sediment to the stream
channel by slumps and surface erosion. Because all the
material from an eroding streambank is delivered directly
into the stream channel, the adverse impact of bank instabil-
ity can be much greater than the adverse effects of a
comparable area of eroding hillslope.
Although in some cases the erosion of one bank will be
matched by deposition on the opposite bank, streambank
erosion caused by management activities generally will
increase stream width. The corresponding increase in stream
surface area allows more direct solar radiation to reach the
stream surface, and this will raise maximum summer water
temperatures (Sections 2.1,5.2). In most cases an eroding
streambank will provide little or no cover for fish.
Actively eroding streambanks also support little or no
riparian vegetation, and the loss of this vegetation adversely
affects a wide range of wildlife species (Raedeke, 1988),
reduces available forage for domestic livestock, and re-
duces the long-term input of organic matter into the aquatic
ecosystem. Both the increase in summer water temperatures
and the loss offish cover along an eroding streambank will
be exacerbated by the reduction in riparian cover.
Response to Management Activities
The management activity that probably has the greatest
impact on streambank stability is grazing (e.g., Platts, 1981).
A reduction in the timing and intensity of grazing in the
riparian zone often results in a decrease in channel cross-
section, an increase in channel depth, and an increase in
vegetation along the channel banks. All these changes sug-
gest an increase in streambank stability, a reduction in
sediment inputs into the stream channel, and an increase in
the density of the riparian vegetation.
Increasingly stringent regulations have greatly reduced
the direct adverse effects of forest management activities on
those streams that have fish, are used for domestic water
supply, or otherwise are granted a high level of protection.
Small headwater streams and ephemeral channels generally
do not have the same level of protection, and this can result
in forest harvest and other management activities having a
direct, adverse impact on bank stability. A large number of
management activities can indirectly affect bank stability,
as any change in the size of the larger (channel-forming) flows
or in the size and flux of sediment is likely to alter channel
morphology and hence bank stability (Sections 5.1-5.5).
Measurement Concepts
Standard procedures to evaluate bank stability have not
been developed. Many stream monitoring programs focus
on bank instability rather than bank stability, as eroding
streambanks are often easier to identify and measure. Dif-
ferent monitoring programs have developed a variety of
procedures to evaluate bank stability, and these range from
qualitative, visual estimates to detailed measurements of
each bank failure.
Perhaps the most widely used procedure related to bank
stability is the method developed by Pfankuch (1978) to
evaluate stream channel condition. Thisuses 4-6 parameters
to evaluate the condition of the upper stream banks, the
lower stream banks, and the channel bottom. These pa-
rameters are empirically weighted, and many of them are
directly related to bank stability (Table 10). Summing the
scores for all 15 parameters yields an overall rating for the
stream channel (Pfankuch, 1978). Its use in the Pacific
Northwest is sometimes criticized because it regards large
woody debris as a destabilizing factor. Such comments do
not demonstrate that the general method is faulty, but
suggest that alterations in the parameters and scoring are
needed as the technique is transferred to other areas and we
gain an improved understanding of fluvial geomorphology.
A simpler procedure focusing solely on streambank
stability is described in Platts et al. (1983, 1987). This
technique assigns the bank along a specified cross-section to
one of four stability classes according to the percentage of
the bank covered by vegetation and rocks, and the size class
of the rock material. The estimated percentage of the bank
protected against fluvial erosion by rocks and vegetation
provides a numerical rating of streambank stability.
A similarprocedure can be used to determine streambank
soil alteration (Platts et al., 1983; 1987). In this case the
observer must visualize the appearance of the streambank
under optimal conditions. The site is then assigned to one of
five soil alteration classes according to the percentage of the
streambank that has been broken down, eroded, or cut back
fromthestream. Again theactualpercentageofthestreambank
that has been altered is estimated to yield a quantitative
rating of streambank soil alteration. Platts et al. (1983,
1987) found that this technique had wider confidence inter-
vals than the streambank stability rating, but the accuracy
-------
Part II
Tabte 10. Parameters and range of values used for evaluating
stream channel condition (Pfankuch, 1978).
Channel location
Parameter
Range of values
Upper bank
Sidestope gradient 0-8
Mass wasting potential 0-12
Debris jam potential 0-8
Vegetative cover 0-12
Lower bank
Channel capacity 0-4
Bank rock content 0-8
Obstructions and flow deflectors 0-8
Bank cutting 0-16
Sediment deposition 0-16
Channel bottom
Angularity of bed particles 0-4
Brightness of bed particles 0-4
Consolidation of bed particles 0-8
Stability and size of bed particles 0-16
Amount of scour and deposition 0-24
Aquatic vegetation 0-4
of both procedures could be rated as no better than fair.
Errors were reduced when the rating was based on specific
cross-sections rather than along a designated stream reach.
Standards
No standards for bank stability have been established or
proposed.
Current Uses
Pfankuch's (1978) channel condition and stability pro-
cedure has been widely used by the U.S. Forest Service.
Other monitoring programs have also taken elements from
this rating system and incorporated them into their own
stream evaluation forms (e.g., Ralph, 1989; G. Luchetti,
pcrs. comm., King County, WA). Although the selection
and weighting of the parameters have never been rigorously
tested, the wide use of this procedure suggests a certain level
of acceptance. One advantage is its accessibility to people
with relatively little technical training, and it seems to
provide relatively consistent results (Pfankuch, 1978). The
arbitrary selection and weighting of parameters means that
it should be modified according to local needs and experi-
ence, but this is rarely done.
Assessment
Streambank stability is an easily assessed parameter
that can be used to indicate whether a particular stream has
been disrupted from a quasi-equilibrium state. This disrup-
tion could be due to natural causes, or alterations in
discharge, sediment load or vegetative cover caused by
management actions (e.g., urbanization, grazing, forest
harvest). Some of the major limitations to the use of bank
stability include (1) lack of accuracy and precision (Platts et
al., 1987), (2) inability to identify specific causes of bank
instability (Platts etal., 1987), (3) varying sensitivity among
stream reaches, and (4) difficulty of separating natural
causes and management impacts.
The lack of accuracy and precision is partly a function
of the techniques being used. The visual estimation tech-
niques described by Platts et al. (1983, 1987) are likely to
have greater uncertainty than the multi-parameter approach
of Pfankuch (1978). One cannot conclude thata change in
bank stability has occurred until the observed change sig-
nificantly exceeds the error in the rating system, but this
error is rarely recognized.
The cause of bank instability may be difficult to deter-
mine, particularly when there is more than one factor.
Grazing has the most direct and obvious impact on bank
stability (Platts, 1981), and this may mask other manage-
ment impacts. Discharge and sediment yield tend to be
controlled by upslope processes, and so the linkage to bank
stability may not be immediately obvious.
Bank stability may be most useful as a quick indicator
of a shift in the equilibrium of the stream system. An
observed increase in bank instability should then trigger
more intensive investigations. By combining an inventory
of management activities with specific measurements of
other parameters such as the bed material particle size, it is
usually possible to determine the primary cause(s) of the
observed disequilibrium. Often, however, bank instability
may not be the most sensitive indicator of disturbance.
Changes in the suspended sediment load, for example, may
not immediately trigger bank instability, but could still
have a detrimental effect on spawning success. Similarly,
grazing impacts are likely to be expressed through the
riparian vegetation before they lead to bank instability.
Nevertheless, the ease of evaluating bank stability sug-
gests that it can play an important role, particularly when
budgets for assessment and monitoring are severely limited.
-------
6. RIPARIAN MONITORING
INTRODUCTION
Characteristics of the riparian zone are rarely consid-
ered as water quality parameters, yet the riparian zone
directly affects many of the designated uses of water. As
noted in Sections 2.1 and 5.8, the type and amount of
riparian vegetation is an important controlling factor for
stream temperatures and bank erosion, and both tempera-
ture and bank erosion can be directly related to the quality
of fish habitat The riparian zone also plays a key role in
defining channel morphology and creating fish rearing
habitat through the input of large woody debris. Finally, the
riparian zone is believed to be important in controlling the
amountof sedimentand nutrients reaching thestream channel
from upslope sources.
Over the past 25 years, several major studies have
documented the effects of forest harvest in the riparian zone
on streams and water quality. The results of these studies
have led to more stringent regulation of forest management
activities adjacent to certain classes of streams (e.g., peren-
nial streams a designated use of with coldwater fisheries or
domestic water supply). The documented effects of man-
agement activities on the stability and vegetation of riparian
zones, and the established linkages between the riparian
zone and various designated uses, provide the rationale for
including two riparian parameters in the Guidelines.
The first parameter is the width of the riparian canopy
opening. Changes in the width of the riparian canopy opening
generally result from changes in the balance between
sediment and discharge. Hence the width of the riparian
canopy opening may be a useful parameter for quickly
determining historical trends in stream condition over large
areas using aerial photographs.
The second parameter—riparian vegetation—is much
more broadly defined. A variety of measurements can be
made regarding the type and condition of the riparian
vegetation, and these measurements may differ widely in
their purpose, the amount of effort required, their sensitivity
to different management activities, and their relation to the
designated uses. The point is that the riparian vegetation and
the width of the riparian canopy opening are important
components of stream condition, and they can be useful
parameters for monitoring the effects of management ac-
tivities on streams.
6.1 RIPARIAN CANOPY OPENING
Definition
The riparian canopy opening refers to the gap between
the canopy of the riparian vegetation on opposite banks of
a stream or river. Often small streams are completely
shaded by woody vegetation and hence have no riparian
canopy opening in their undisturbed state. In steep, narrow,
V-shaped valleys, considerable shading can result from the
dominant upslope species rather than theriparian vegetation.
In lower-gradient and higher-order streams, the stream
channel by definition is wider and there commonly is a gap
or opening between the parallel strands of the riparian
vegetation. Streams with an alluvial valley floor tend to
have more extensive and complex stands of riparian vegeta-
tion that develop in response to periodic flooding and high
water tables.
These riparian and upslope forests that shade undis-
turbed stream channels can be altered by both natural
disturbances (e.g., landslides, debris flows, and stream
channel erosion) and forest management activities. Often a
highly interactiveresponseexistsbetween changes in channel
morphology and changes in the riparian forest (Wissmar
andSwanson, 1990). For example, channel or bank erosion
often changes the size and location of the stream channels,
which results in a corresponding loss of the streamside
-------
Part II
vegetation and an increase in the width of the riparian
canopy opening.
Monitoring of the riparian canopy opening offers a
relatively rapid means of assessing the influences of a
variety of management activities on both the streamside
vegetation and the stream channel. Identification of the
source areas and quantitative mapping of the changes in the
riparian canopy opening over time can help determine the
primary cause(s) of adverse change (Grant, 1988).
Relation to Designated Uses
An increase in the width of the riparian canopy opening
will allow more direct radiation to reach the stream and raise
peak summer water temperatures. Less shading also will
result in greater temperature fluctuations on both a seasonal
and a daily basis (Section 2.1). A reduction in canopy cover
may increase the amount of reradiated long-wave radiation,
thereby allowing more heat loss at night. Heat loss can be
crucial to the icing up and formation of anchor ice in colder
environments (Beschta et al., 1987).
In light-limited forest streams, an increase in the width of
the riparian canopy opening can increase primary production
(Gregory et al., 1987). This may induce a corresponding in-
crcasein invertebrate andfish production. However, increased
primary productivity may be offset by decreased inputs of
dctrital food subsidies, leaves, and other organic material
from theriparian zone. Thenetbalancebetweentheincreased
primary production and the decreased detrital inputs will
depend on the size of the stream and the presence or absence
of other limiting factors, such as plant-available nutrients.
Changes in the size and structure of the riparian canopy
will adversely affect a wide range of animal species depen-
dent on riparian habitats (Deusen and Adams, 1989). A
reduction in the width of the riparian zone may reduce the
purported ability of theriparian zone to trap excess nutrients
and sediments coming from upslope (Green and Kaufmann,
1989; Section 6.2). An increase in the riparian canopy
opening is likely to reduce the long-term delivery of large
woody debris (LWD) into the stream channel (Grant, 1988).
In many forested streams LWD is an extremely important
clement in channel morphology, sediment transport, and
quality of aquatic habitat (Bisson etal., 1987; Section 5.7).
Response to Management Activities
Changes in the size of the riparian canopy opening can
result from a variety of interacting fluvial and geomorphic
processes. Probably the most common cause is an increase
in coarse sediment. This can increase channel width
through bank erosion (Section 5.2), with a corresponding
loss of the riparian vegetation. Recolonization of the
enlarged streambed by riparian species will proceed slowly
at best until the source of the excess sediment is removed, or
the excess sediment in the channel is stored or transported
out of the stream reach.
Grant (1988) noted that an increase in channel width and
the riparian canopy opening also can result from an increase
in the size of peak flows. As noted in Section 3.1, peak flow
increases from forest activities usually are small or are lim-
ited to the smaller, more frequent storms. The major excep-
tion is in areas subject to rain-on-snow events; in these
environments forest harvest can substantially increase the
size of the larger peak flows (Section 3.1). In general, peak
flows probably are less likely to enlarge the size of the
riparian canopy opening and initiate channel morphological
changes than increases in the amount of coarse sediment
(i.e., bedload). Other possible causes of fluvial disturbance
that can increase the riparian canopy opening include debris
flows, extreme discharge events, entrainment and transport
of large woody debris in flood plain areas, and increased
lateral migration of stream channels.
Measurement Concepts
A detailed procedure for measuring and analyzing
changes in the riparian canopy opening has been published
as the RAPID (Rapid Aerial Photographic Inventory of
Disturbance) technique (Grant, 1988). This requires a
historical sequence of aerial photographs on a scale of at
least 1:24,000. The basic approach is to (1) identify "initia-
tion sites" where the increase in riparian canopy opening
begins; (2) determine the spatial links between the initiation
sites and downstream increases in the width of the riparian
canopy opening; (3)determinethecontinuityofopenreaches
along the stream; and (4) measure the width of the riparian
canopy opening and note the condition of the surrounding
forest at 100- to 300-m intervals on each set of photos.
These data are mapped onto drainage network maps at the
same scale as the aerial photographs. Suggested procedures
to analyze and summarize the quantitative data are presented
by Grant (1988). Adaptations to the RAPID technique may
be needed according to the specific vegetation, topography,
geology, and climate in the basin under study.
Initiation sites are identified and mapped in order to
elucidate the cause(s) of an increase in the riparian canopy
opening. For example, a sudden, continuous increase in the
width of the riparian canopy opening might be traced to a
landslide or debris flow, whereas a more gradual increase
in the width suggests a more dispersed source of sediment or
an increase in the size of peak flows.
In many cases visual observations of the aerial photos
will provide an indication of current condition, and riparian
canopy opening measurements on successive aerial photos
can demonstrate if adverse changes have occurred. The
advantage of the full RAPID-type approach is that historical
and current riparian and channel conditions can be quanti-
fied. This facilitates an understanding of the possible
cause(s) of an observed change, an assessment of the signifi-
cance of change, and the prediction of future trends.
-------
CHAPTER 6. RIPARIAN MONITORING
As suggested above, the sensitivity of the riparian
canopy opening to forest management activities will vary
with stream type, location, and geological setting. For
example, bedrock streams in steep, V-shaped valleys usu-
ally show little alteration in stream channel width in re-
sponse to increased sediment load. Streams in wide valleys
with unconsolidated alluvial sediments are likely to be
much more sensitive to changes in flow and sediment flux
(e.g., Lisle, 1982).
Standards
No standards have been established or proposed to
regulate changes in the riparian canopy opening. However,
most states have established Best Management Practices to
restrict the removal of trees along fish-bearing streams or
streams used for domestic water supply (Section 6.2). Leav-
ing trees in or immediately adjacent to the stream channel
helps maintain channel and bank stability, and therefore
reduces the potential for an increase in the width of the
riparian canopy opening.
Current Uses
Several studies have evaluated the concept of using the
riparian canopy opening for monitoring. Grant (1986)
showed a relationship between riparian canopy opening and
area harvested for eight basins in or near the Middle Fork of
the Willamette River, Oregon. A sequence of photographs
for the Breitenbush River (Oregon) was used to document
changes in the riparian canopy opening, and these changes
were related to salvage logging and the December 1964
flood (Grant, 1988). In Western Washington the RAPID
technique was used to identify the major disturbance events
and changes in the riparian canopy, but it was not possible
to directly relate disturbance events to downstream changes
(Jeanette Smith, University of Washington, Seattle, pers.
comm.).
A study of the Elk River Basin in southwest Oregon
used the RAPID technique to document changes in the
riparian canopy opening and relate these changes to timber
harvest activities and large storm events (Ryan and Grant, in
press). Upstream and downstream changes were not well
synchronized, and this made it difficult to infer causal
relationships. Currently, developmental projects designed
to assess change in the riparian canopy have been initiated
by theU.S. Forest Service and Washington State Department
of Natural Resources.
Assessment
Determination of riparian canopy opening from aerial
photographs appears to have considerable promise for
quickly assessing stream condition and adverse manage-
ment effects over a large area. These data can help assess the
impact of past events and guide future activities. Monitor-
ing of the riparian canopy opening is relatively unique
because it uses an historic sequence of aerial photos as the
primary data source. Such photos are available for most of
the productive timberlands in the western U.S., and this
permits change to be assessed over a period of several
decades.
In contrast, most current water quality monitoring pro-
grams have only a few years of data. Any change observed
duringsuchamonitoringperiodcannotbeplacedin historical
context, and this severely limits our ability to evaluate the
significance of the observed change. Long-term data on the
riparian canopy opening may provide some of the needed
historical context
On the other hand, the RAPID-type approach is not as
sensitive to change as ground-based measurements. Unless
unusually detailed aerial photos are available, an increase in
the riparian canopy opening cannot be detected until a
substantial increase in stream width has occurred. By this
time much of the original banks and vegetation will have
been lost, and some designated uses will have been im-
paired. In some cases it may be difficult to relate changes in
the riparian canopy opening and the width of the stream
channel to the potential causal factors such as landslides,
forest harvest, extreme floods, or debris flows.
In summary, RAPID-type techniques have consider-
able potential for assessing change on a relatively broad
temporal and spatial scale. This can help direct ground-
based monitoring projects to the most critical locations, and
provide a very useful context for the shorter-term data
collected from such projects. However, as stream invento-
ries or monitoring projects are initiated over larger areas and
the data record is extended in time (Minshall et al., 1989;
Gresswelletal., 1989), the need foraRAPID-type approach
may gradually decline.
6.2 RIPARIAN VEGETATION
Definition
Riparian vegetation has been defined as "Vegetation
growing on or near the banks of a stream or other body of
water on soils that exhibit some wetness characteristics
during some portion of the growing season" (AFS, un-
dated). Other authors have specified that the soil should be
saturated within the rooting depth of the plants for at least
some portion of the growing season (Platts et al., 1983;
Minshall etal., 1989).
These definitions suggest that riparian areas are a par-
ticular type of wetland. Wetlands have been defined by EPA
as "Those areas that are inundated or saturated by surface or
groundwater at a frequency and duration sufficient to sup-
portaprevalence of vegetation typically adapted for a life in
-------
Part II
saturated soil conditions." A unified set of criteria for de-
lineating wetlands has recently been adopted by several
federal agencies with wetland responsibilities (U.S. Army
Corps of Engineers, 1989). In most ecoregions one can find
a wide variety of vegetation types on the streambanks and
flood plains, including coniferous and deciduous trees,
grasses, shrubs, forbs, ferns, and mosses.
Relation to Designated Uses
Riparian vegetation, and the exploitation of this vegeta-
tion, affects most of the designated uses of water through a
variety of different processes. Many of these interactions
have been discussed in other sections, and an extensive
literature is available on the interactions between riparian
zones and aquatic ecosystems (e.g., Raedeke, 1988;
Gresswell et al., 1989). A recent bibliography on riparian
research and management listed over 3,500 references (Van
Deventer, 1990).
Some of the most important biological and physical
effects of riparian vegetation on the designated uses of water
are as follows: (1) providing organic material that can be
used as food sources for aquatic organisms (Sections 7.3-
7.4); (2) supplying large woody debris that alters sediment
storage, influences channel morphology, and enhances fish
production (Section 5.7); (3) shading the stream and reduc-
ing temperature fluctuations (Section 2.1); (4) reducing
bank erosion (Section 5.8); and (5) providing habitat and
cover for both aquatic and terrestrial organisms. Social
benefits include streamside esthetics.
The relative importance of these different functions is
heavily influenced by vegetation type. Deciduous trees
provide large amounts of leaves and other organic material,
which are generally higher in nitrogen than coniferous
debris, and thus more readily broken down by invertebrates
(Bilby, 1988). More rapid breakdown leads to more rapid
utilization and higher productivity.
On the other hand, coniferous trees are the most impor-
tant source of large woody debris in most parts of the Pacific
Northwest and Alaska (Section 5.7). Coniferous branches,
boles and root wads tend to be larger than their deciduous
equivalents, and this increases both their stability within the
stream channel and the diversity of aquatic habitats, par-
ticularly at high flows (Sedell et al., 1984; Bisson et al.,
1987). Coniferous wood does not decay as rapidly as alder
and most other deciduous species, and this also contributes
to channel and habitat stability (Sedell et al., 1988).
Both coniferous and deciduous species are effective in
shading the stream and thereby reducing peak summer
temperatures. Streams with little or no vegetative canopy
may have lower winter minima and be more susceptible to
the formation of anchor ice (Platts, 1984).
All types of vegetation can be effective in reducing bank
erosion, although they differ in the type of protection (e.g.,
Hackley, 1989; Platts and Nelson, 1989). Large trees and
root wads can divert or deflect the flow in small or moderate-
sized streams, and their rootscan provide substantial protec-
tion during high flows. Grassy banks may provide a more
complete cover, but they may not be as resistant to undercut-
ting or abrasion.
Few studies have been done on the filtering and buffer-
ing capacities of riparian vegetation in forested zones (Green
and Kauffman, 1989). In most undisturbed forest ecosys-
tems the nutrient and sediment yields are so low that the
filtering capacity of the riparian zone is not a key concern.
In agricultural areas, however, nutrient exports are impor-
tant and the riparian zone has been shown to be a sink for
sediment as well as nitrogen, phosphorous, calcium, mag-
nesium, and potassium sulfate (Lowrance et al., 1984;
Lowrance et al., 1986; Green and Kauffman, 1989). The
influence of different vegetation types on sediment and
nutrient yields, and in some situations water yield, is com-
plicated by differences in other factors such as the preva-
lence of overland flow, height of the water table, rooting
depth.rootdensities.chemicalpropertiesofthesoil, nitrogen-
fixing ability of the plants, and seasonal growth patterns.
Various types of riparian vegetation provide different
types of habitat (Raedeke, 1988). Species such as otters,
beavers, deer, and bald eagles all have different habitat
needs and are more or less dependent on riparian vegetation.
Hence management of the riparian zone will depend in part
on the selected wildlife and fisheries objectives. The
uncertainty and subjective nature of habitat evaluations are
illustrated by the observation that streams bordered by brush
had a higher standing crop of fish than streams bordered by
trees, yet the U.S. Forest Service usually assigns a higher
habitat value to a tree cover (Platts et al., 1983).
The importance of the riparian vegetation to the adja-
cent aquatic ecosystem diminishes in the downstream di-
rection because of the increase in discharge and stream size
(Bilby, 1988). In small streams the riparian vegetation may
be the dominant source of organic matter, while in larger
streams instream primary production tends to dominate
(Hynes, 1970). Removal or alteration of the riparian veg-
etation in a single reach can significantly alter temperature
and water quality in low discharge, narrow streams, but the
impact of a comparable change is likely to be undetectable
in large streams or rivers (Bilby, 1988).
Response to Management Activities
The abundance of moisture makes the riparian zone
exceptionally diverse and productive (Kauffman, 1988).
This higher productivity often results in a more intensive
exploitation of riparian resources. In many areas the largest
trees are in the floodplains and alluvial valleys, and the
riparian zones have been more heavily logged because the
trees were readily accessible and could be floated down-
stream. Grazing pressure usually is higher in the riparian
zone because there typically is more shade, surface water for
-------
CHAPTER 6. RIPARIAN MONITORING
drinking, and more succulent vegetation (Platts, 1981).
Riparian areas also tend to be the focus of recreational
activities such as camping and fishing (Kauffman, 1988).
Several researchers have argued that livestock repre-
sents the single greatest threat to trout and wildlife habitat in
the western U.S. (e.g., Behnke and Zarn, 1976; Platts,
1981). The inherent conflicts between livestock and fish
make management of the riparian zone a more intractable
problem in range lands than in forest lands.
The functions of the riparian zone described in the
previous section provide the basis for predicting the effects
of different management activities. Any reduction in the
riparian canopy cover, for example, can affect stream tem-
peratures, organic matter inputs, bank stability, and so on. A
reduction in the riparian cover can result directly from
management (e.g., harvesting trees in the riparian zone,
grazing), or indirectly as a result of changes in the size and
amount of sediment and discharge. As noted in Section 6.1,
there are strong interactions between changes in the riparian
vegetation and changes in stream channel morphology. The
effects of management activities are reviewed in the sec-
tions on flow (Chapter 3), sediment (Chapter 4), channel
characteristics (Chapter 5), and riparian canopy opening
(Section 6.1).
Measurement Concepts
Although riparian vegetation affects many aquatic habi-
tat and water quality parameters, generally it is more effec-
tive to monitor these other parameters directly rather than
monitoring the riparian vegetation. Estimates of cover or
rearing habitat for juvenile salmonids, for example, focuses
on the type and abundance of cover in the stream, and not the
potential cover, such as dead branches and snags, which
may fall into the stream. Similarly, water quality parameters
such as nitrate, conductivity, and turbidity are measured
directly, and the influence of the riparian vegetation is
difficult to assess. A notable exception is the increase in
water temperature caused by removal of the riparian canopy.
In short reaches with negligible groundwater flow, the in-
crease in summer maximum temperatures is a direct func-
tion of the additional exposure of the stream surface to
incoming solar radiation, and this effect can be predicted
(Beschta et al., 1987; Section 2.1).
It follows that, with the exception of temperature, any
precise measurement or characterization of the riparian veg-
etation provides an accuracy which cannot be translated into
a more precise assessment of water quality or the impair-
ment of designated uses. Thus relatively simple techniques
that are repeatable over long time periods usually provide
the best approach to monitor the condition of the riparian
vegetation, and to evaluate the likely effects of the riparian
vegetation on water quality (Platts et al., 1989).
An extensive literature on vegetation sampling is avail-
able, and the techniques for forests (e.g., Husch etal., 1982),
shrublands, and grasslands (e.g.,:Cook and Stubbendieck,
1986; Tueller, 1988) can be applied as appropriate to the
riparian zone. More often than not, however, stream inven-
tory and water quality monitoring programs havedeveloped
ad hoc techniques for monitoring the riparian vegetation
according to their particular objectives and conditions. The
choice of qualitative or quantitative methods is determined
by the parameter being measured, the anticipated use of the
data, and the cost of collecting that data.
Some of the most commonly measured parameters
include vegetation type, vegetation cover, and vegetation
density. Vegetation type is usually a qualitative categoriza-
tion which can be as simple as tree, shrub, grass or bare (e.g.,
Platts et al., 1983). More commonly the vegetation type is
based on the dominant overstory species or specified plant
communities (e.g., Platts and Nelson, 1989).
Vegetation cover usually refers to the downward pro-
jection of the canopy onto the ground surface (Husch et al.,
1982). Visual estimation techniques can be used to provide
a quick, qualitative measure (Platts et al., 1987). Quantita-
tive measurements usually rely on point- or line-intercept
methods.
Forest cover density can be assessed by measuring light
intensity or by using a spherical densiometer (Lemmon,
1957). The latter uses a point sampling technique to deter-
mine theamount of clear sky in the hemisphere centered over
the observer.
Data on stream shading can be obtained by several
different methods. Sampling procedures for the spherical
densiometer in large and small streams are discussed in
Platts et al. (1987). Stream surface shading can be deter-
mined by measuring the height, density, crown width, and
offset of the riparian vegetation. The Solar Pathfinder™ is
a much simpler technique which directly maps the extent of
shading on the specified day (Platts et al., 1987). Each of
these techniques produces data useful for assessing changes
in the riparian canopy over time, or for predicting the effect
of riparian canopy removal on stream temperatures.
Vegetation density refers to the number of plants per
unit area. For practical reasons density is most useful in
forestry. It can be measured on either fixed- or variable-
sized plots, and foresters often combine size and density
data to obtain estimates of basal area or volume (Husch et
al., 1982). Density, species, and size class data can be
combined with growth and mortality data to estimate the
future recruitment of large woody debris (e.g., Bilby and
Wasserman, 1989).
Changes in the riparian vegetation due to grazing,
logging, or other management activities can be assessed by
each of these techniques. Cover, density, and biomass are
more likely to reflect short-term management impacts than
vegetation type. More frequent monitoring will be required
in grazed areas owing to the rapid seasonal changes in forage
availability and consumption. Platts et al. (1987) suggest a
simple procedure to rate vegetation use in herbaceous areas.
-------
Part II
Grazing strategies in riparian areas are discussed by Platts
(1989), while recent books (e.g., Cook and Stubbehdieck,
1986; Husch etal., 1982; Tueller, 1988) should be consulted
for more details on vegetation monitoring techniques.
Standards
In the 1970s theforestpracticerulesforriparian areas in
Oregon and Washington were designed to maintain ad-
equate shade and minimize the introduction of sediment and
forest chemicals (Bilby and Wasserman, 1989). Currently
the rules for riparian areas are in a state of flux as a result of
increased concern over the future recruitment of large
woody debris into stream channels (Section 5.7). Present
forest practice rules for Idaho require that 75% of the
existing shade be left along Class I (fish-bearing) streams.
In Washington the forest practice rules were substantially
modifiedin 1988 undertheTimber-Fish-Wildlifeagreement,
and the number of leave trees that currently are required
along streams in eastern and western Washington is pre-
sented in Table 11. The complexity of the leave tree re-
quirements in Table 11 illustrates the difficulty of trying to
account for the diversity of natural systems in environmen-
tal regulations.
None of the state forest practice rules include any
standards relating to grazing in the riparian zone, as that is
outside their legal mandate. Land management agencies
with substantial grazing lands have established utilization
standards and Best Management Practices intended to pro-
tect the designated uses of water. The adequacy and evalu-
ation of these is outside the scope of this document, but a
recent publication by Minshall et al. (1989) and the pro-
ceedings of two recent conferences provide a good overview
of riparian area functions and management (Raedeke, 1988;
Gresswell et al., 1989).
Current Uses
The primary objectives of monitoring the riparian veg-
etation in forested areas are to maintain adequate shade,
Table 11. Type and number of trees required to be left along streams in western and eastern Washington after timber harvest (Bilby and
Wasserman, 1989.)
RMZ
Water type and maximum Ratio of conifer
average widthb width to deciduous
1 and 2 waters 23 m and over 30 m
1 and 2 waters under 23 m 23 m
3 water 2 m and over 1 5 m
Representative of stand
Representative of stand
2to1
Minimum size of
leave trees
Representative of stand
Representative of stand
30 cm or next largest
available0
Number of trees/
305 m each side
Gravel/ Boulder/
cobble3 bedrock
50 trees 25 trees
100 trees 50 trees
75 trees 25 trees
3 water less than 2 m
8 m
to1
15 cm or next largest
available
25 trees 25 trees
•Gravel and cobbie streambeds are composed predominately of
material <25 cm in diameter.
^Washington water typing system is based on domestic water use,
fish use, and size of streams. A detailed description of the criteria
may be found in Washington Forest Practices Rules and Regula-
tions.
-------
CHAPTER 6. RIPARIAN MONITORING
indirectly assess bank stability, and ensure an adequate
supply of coarse woody debris. The first two are relatively
straightforward, but the latter requires knowledge of tree
growth rates, recruitment rates, the durability and stability
of woody debris within a given stream, and specification of
the desired amount of woody debris.
Tree growth rates for most areas are adequately known,
and the durability and stability of large woody debris has
been studied in a number of different streams (Section 5.7).
Specification of the desired amount of woody debris is a
political question. The last factor, recruitment rates, is the
least known. Estimating recruitment rates is difficult be-
cause root wads, tree trunks, and large branches can enter
into the stream channel by a variety of processes that vary
both in magnitude and frequency (Bissonetal., 1987; Sedell
etal., 1988). Small-scale, relatively frequent inputs include
windthrow, natural mortality of trees along the stream
channel, and bank erosion leading to the toppling of trees
into the stream channel. Three more episodic mechanisms
for delivering large quantities of woody debris to streams
and rivers are debris flows, avalanches and landslides
(Swanson and Lienkaemper, 1978).
The potential for each of these delivery mechanisms
will vary with reach and catchment. In wide alluvial valleys
the potential for episodic inputs is relatively small, and the
future recruitment of large woody debris should focus on
those trees which may fall directly into the stream channel.
Streams with a rapidly migrating channel may have a wider
recruitment zone.
The potential for large episodic inputs is greater in steep,
unstable terrain, and areas subject to heavy falls of snow or
rain. Large episodic events greatly expand the potential
source area for large woody debris, but the frequency of
these events is relatively low. Little or no data exists on the
relative importance of the different processes for delivering
large woody debris to the stream system in different
catchments.
The methodology of Bilby and Wasserman (1989) pro-
vided a technical base for the Forest Practice Rules in Wa-
shington, but the procedures were not intended as a monitor-
ing technique. Until better data is available on recruitment
rates, monitoring to protect the designated uses of water will
have to rely on measurements of large woody debris in
streams and an assumed recruitment rate from trees immedi-
ately adjacent to the stream channel. An inherent limitation
of these procedures is the long time frame needed to study the
recruitment and stability of large woody debris in stream
channels (Section 5.7). The management and silviculture of
riparian zones is a primary focus of Oregon's COPE (Coastal
Oregon Productivity Enhancement) program.
Assessment
Riparian vegetation is of critical importance to water
quality because of its proximity to, and interactions with,
aquatic ecosystems. In small streams the riparian vegeta-
tion usually is the largest source of organic material and
hence a critical source of detritus for aquatic food webs. The
riparian zone is also the primary source of large woody
debris (Section 5.7). The amount of shade cast by riparian
vegetation is an important factor in determining maximum
stream temperatures and may also influence winter minima.
Low overhanging riparian vegetation provides cover for
salmonids and other fish (Platts etal., 1983). A reduction in
the riparian vegetation through overgrazing, logging, or
intensive recreational use can lead to bank erosion and
instability. Bank erosion can have a disproportionate effect
on water quality and the designated uses of water because
the sediment is delivered directly into the stream channel
(Section 5.8).
Monitoring of the riparian vegetation is another means
of assessing management impacts in the riparian zone and
evaluating whether certain designated uses are impaired.
However, riparian vegetation cannot be used as a direct
indicator of water quality except in the case of stream
temperatures. For this reason most water quality monitor-
ing programs use relatively simple, qualitative indicators to
assess the type, density, and cover of the riparian vegetation.
Detailed quantitative monitoring is most appropriate for
1. assessing stream shading and predicting the thermal
effects of changes in the riparian canopy,
2. predicting the size and future recruitment of large
organic debris;
3. measuring the amount of cover for fisheries, and
4. assessing bank stability and bank erosion as a function
of vegetative cover.
Only the first of these cases can be classified as a
traditional water quality parameter, even though the other
three have clear linkages to water quality and some desig-
nated uses of water.
Management goals and the type of vegetation will largely
determine the type of monitoring. In cool forested areas that
are not heavily grazed, the emphasis should be on main-
taining a healthy riparian canopy and ensuring adequate
future inputs of large organic debris. In warmer areas stream
surface shading is more likely to be the primary concern, and
measurements of the riparian canopy can guide the intensity
of management in the riparian zone. If grazing is the primary
use, the emphasis should be on regular monitoring of bank
stability (Section 5.8) and vegetative cover.
-------
7. AQUATIC ORGANISMS
INTRODUCTION
Aquatic organisms can be very useful for monitoring
because they effectively integrate a large number of habitat
characteristics. In other words, if the habitat requirements
of a particular organism are known, the presence of that
organism can be used todefinetheconditionsinthatparticular
water body. Furthermore, those conditions can be assumed
to have been met for the life span of the organism being
monitored. Thus aquatic organisms have the great advan-
tage of allowing inferences to be made regarding past con-
ditions, which may allow sampling to be done less frequently
than is usually necessary for the parameters considered in
Chapter 2 (physical and chemical constituents) and Chapter
4 (sediment).
In this chapter aquatic organisms have been grouped
into four parameters—bacteria, algae, invertebrates, and
fish. Bacterial monitoring is the most straightforward and
typically involves estimating the numbers of up to four
types of bacteria. Algae, invertebrates, and fish are far more
complex, as one can make any number of measurements
relating to the numbers of organisms, species composition,
and productivity. These different measurements are not
considered separately for several reasons.
First, we did not wish to duplicate the considerable
amount of information already available on the use of these
organisms for monitoring. Second, there is no consensus
about which measurements should be made, and in many
cases the choice will depend on the purpose of the monitor-
ing. Third, the use of aquatic organisms for monitoring is
undergoing rapid change as different states attempt to estab-
lish biological criteria for water quality. Fourth, the tre-
mendous variability in aquatic ecosystems has made it
difficult to establish rigorous and sensitive monitoring pro-
cedures. Finally, the use offish for monitoring purposes is
often hindered by the problem of separating extraneous
factors, such as fishing pressure, from the effects of
management activities.
In general, aquatic organisms have considerable poten-
tial for monitoring changes in water quality. Aquatic
invertebrates are particularly promising because of their
diversity, sensitivity to habitat change, relative ease of
identification, and they are subject to fewer extraneous
controlling factors. The complexity and diversity of aquatic
ecosystems means that the sections on algae, invertebrates,
and fish should be considered as a general overview rather
than an in-depth review.
7.1 BACTERIA
Definition
A wide variety of diseases are spread by aquatic micro-
organisms. These include bacterial diseases (e.g.,
Legionnaire's disease, cholera, typhoid, and gastrointesti-
nal illness), viral diseases (e.g., polio, hepatitis, and gastroi-
ntestinal illness), and parasitic diseases (amoebic dysen-
tery, flukes, and giardiasis). Many of these diseases are
rarely found in the U.S., and the analytic procedures for
detecting many of these organisms are time consuming and
costly. For these reasons most drinking and recreational
waters are routinely tested only for certain bacteria which
have been correlated with human health risk. If the average
concentration of these bacteria falls below the designated
standard, it is assumed that the water is safe for that use and
that there are no other pathogenic bacteria that represent a
significant hazard to human health (APHA, 1989).
The four groups of bacteria most commonly used for
water quality monitoring are total coliforms, fecal colif-
orms, fecal streptococci, and enterococci. The total colif-
orms (TC) group includes a wide range of aerobic and
facultatively anaerobic bacteria. Their ability to ferment
-------
CHAPTER 7. AQUATIC ORGANISMS
lactose and produce gas helps define the group and also is
the basis for one of the primary testing methods. Many
coliform bacteria are non-pathogenic and are not associated
with human waste.
Fecal coliform (FC) bacteria are mostly those coliform
bacteria which are present in the gut and feces of warm-
blooded animals. The primary species in this group are
Escherichia coli and Klebsiella species. They are distin-
guished by their ability to produce gas from lactose at a
temperature of 44.5±0.2°C. Generally they are less able to
survive in natural waters than non-fecal coliform bacteria.
Fecal streptococci (FS) also are found in the intestines
of man and animals, but in animals FS is usually more
common than FC. This observation has led to efforts to use
the FC/FS ratio to determine whether contamination is due
to man, animals, or a mixture of the two; however, a number
of restrictions on the use of the FC/FS ratio exit (EPA,
1978). One problem is thatFS andFC havedifferentdie-off
rates in natural waters, so the FC/FS ratio is useful only for
the first 24 hr after contamination has occurred. The more
limited ability of some FS species to survive in natural
waters indicates that FS concentrations should not be the
sole test of bacterial contamination.
The enterococcus group of bacteria is part of the larger
FS group. These bacteria are of particular interest for
monitoring recreational waters because they appear to be a
better indicator of the risk of gastrointestinal illness than
TC, FC, or FS (Vasconcelos and Anthony, 1985).
Relation to Designated Uses
The concentration of TC bacteria has long been used as
the primary criterion for the sanitary condition of domestic
water supplies. Experience has repeatedly demonstrated a
positive correlation between the TC count and the incidence
of gastrointestinal disease. However, many of the TC
bacteria do not have a direct effect on human health and are
found outside of animal intestines and feces. This means
that TC are not a particularly accurate or consistent indicator
of the actual health risk.
FC, FS, and enterococci are more specialized groups of
bacteria than TC. Their more restricted habitat means that
their concentration can be more directly linked to sanitary
water quality and human health risks. FC have been con-
sidered a better indicator of water quality than total coli-
forms for over 2 decades. More recent evidence indicates
that enterococci concentrations are most closely correlated
with gastroenteritis among swimmers (DuFour, 1982).
Escherichia coli was the next best indicator, while the
broader group of FC were a relatively poor indicator of
health risk. Both FC and FS, although less tolerant of the
aquatic environment than most other types of coliform
bacteria, can survive for several days in fresh water.
The public generally is aware of the significance of
coliform bacteria in indicating water quality. Severe ad-
verse public reaction can be expected if recreational or
domestic waters do not meet bacteriological standards.
Response to Management Activities
High counts of TC, FC, orFS usually are associated with
inadequate sewage treatment, poorly functioning septic
tank drainfields, or high concentrations of animals. In
forested areas, high levels of coliform bacteria usually will
be associated with inadequate waste disposal by recre-
ational users, the presence of livestock or other animals in
the stream channel or riparian zone, and poorly maintained
septic systems. Since each of these is a relatively dispersed
source, and the soil is an excellent filtration medium, bacte-
rial contamination can be greatly reduced simply by locat-
ing these activities away from the stream or lake boundary
(Kunkle et al., 1987). However, septic systems may not
function effectively in cold climates or in certain soil types.
Measurement Concepts
The two most common methods for measuring TC, FC,
and FS are the multiple-tube fermentation technique and the
membrane filter technique. The multiple-tube fermentation
technique places varying amounts of the sampled water in
tubes containing a growth media. These tubes are incubated
forupto48hr to determine if gas bubbles form; gas formation
is regarded as a sign that coliform bacteria are present in that
sample ("presumptive test")- Tubes testing positive may be
subjected to additional procedures to confirm the presence
of coliform bacteria ("confirmed test" and "completed test";
APHA, 1989).
The two main problems with the multiple-tube fermen-
tation test are as follows: (1) an individual tube indicates
only whether coliform bacteria are present or absent in that
particular sample, and (2) the false positive rate for a single
tube is 13% (Federal Register, 1989). Forthis reason atleast
five replicate tubes at several different dilutions now are
required to obtain a reliable estimate of bacterial concentra-
tion (APHA, 1989).
Replication is needed to provide a higher level of con-
fidence in the results. If the true concentration of coliform
bacteria is one organism per ml, for example, there is a 37%
chance that a well-mixed, 1-ml sample will not have any
coliform bacteria (APHA, 1976). If five tubes, each with a
1-ml sample, are tested, there is a less than 1% chance that
all five tubes will yield a negative result. Replications also
assist in making a quantitative estimate of the coliform
concentrations, as a properly diluted sample will have a
mixture of positive and negative results. For this reason
replications generally are required when testing for colif-
orm bacteria (Federal Register, 1989).
Different dilutions are needed to obtain a quantitative
estimate of bacterial concentrations. Ideally the range of
dilutions will span the range of results from nearly all
-------
Part II
positive to nearly all negative. Thereplications and range of
dilutions allow a statistically based estimate of the coliform
concentration for that particular sample, and this is known
as the Most Probable Number (MPN). The binary nature of
the procedure (i.e., each tube is either positive or negative)
results in a relatively large confidence interval around the
MPN. For five replicates at three dilutions, the 95% confi-
dence interval usually spans a factor of 10. The use of only
threereplicatesdoubles or triples thesizeofthe95%confidence
interval, and this is why the minimum number of replicates
has been raised to five tubes (APHA, 1989).
The second analytic procedure is the membrane filter
technique. In this method different volumes of the sample
water arepassedthroughaspecial0.45-micron filter. The filter
and retained bacteria are placed on a selective growth me-
dium and incubated for 24 hr. At the end of this period, the
actively growing, closed coliform colonies are identified
and counted. For best results the quantity of water filtered
should yield about 50 colonies for TC, and between 20 and
60 colonies for FC. No more than 200 bacterial colonies
Should be present on one membrane filter (APHA, 1989).
There are several advantages to themembrane filter test.
First, the filtering procedure allows for larger volumes of
water and, hence, more accurate testing of less polluted
waters. Second, the results are more precise and have a
lower confidence interval because each sample yields a
quantitative result. Third, the procedure yields results within
24 hr, although additional testing may be necessary for
further verification. A major disadvantage is that the test
can be hindered by high concentrations of either suspended
solids or non-coliform bacteria (APHA, 1989; Federal
Register, 1989).
In mid-1989 EPA approved a third analytic procedure
for total coliforms in finished drinking water, the MMO-
MUG test. Conceptually this is similar to the fermentation
tube technique, except that the end result in the MMO-MUG
test is a change in color rather than the production of gas.
The MMO-MUG test may prove more convenient because
the incubation period is only 24 hr, and it is not affected by
large numbers of heterolrophic bacteria. MMO-MUG tubes
with the growth medium are commercially available.
Standards
The drinking water criteria for TC is zero with some
allowance for an occasional positive test.
For freshwater bathing, the geometric mean value of at
least five samples equally spaced over a 30-day period
should not exceed 126 E. coli/lQQ ml, or 33 enterococci/
100 ml (EPA, 1986b). The maximum value for any single
sample is determined by the intensity of recreational use and
the site-specific standard deviation of the logarithmic val-
ues. Thus the allowable maximum for a single sample will
be higher in areas which are infrequently used for bathing,
and higher in areas which are subject to more variability in
bacterial counts. This approach means the standards are
based partly on the relative health risk rather than an abso-
lute standard.
Current Uses
Bacteriological testing is regularly carried out to ensure
the safety of domestic water supplies and to protect public
health in recreational areas. Most states have adopted FC
as the primary standard for bacterial contamination in rec-
reational waters and test accordingly. As indicated above,
the standards include both single sample maxima and a 30-
day average. FC are preferred over TC because they are a
more specialized group of bacteria and a more direct indica-
tor of fecal contamination and public health hazard. Use of
enterococci for monitoring recreational waters is becoming
more common because this test is more sensitive and pro-
vides a better estimate of the human health risk.
TC, FC, and FS concentrations often vary widely over
relatively short time periods. For this reason any monitoring
or assessment program should analyze a series of samples
before coming to a conclusion about the bacterial quality of
the water. In most cases, the maximum concentration of
coliform bacteria will occur in conjunction with high runoff
events, which wash more coliforms into streams and lakes.
Common sources are manure from animals and bypass
water from small community sewage treatment plants. In
still waters used for bathing, the maximum concentration of
coliform bacteria may occur during warm-weather periods
when there is intensive use.
Very specific tests can be performed to identify the
different species of coliform bacteria, and this information
can help identify the source of the-contamination. Since
these tests are relatively costly and not widely available, the
source(s) of contamination usually is identified by estab-
lishing a more intensive sampling program keyed to land
use. Use of the FC/FS ratio is cautioned because of the
different mortality rates and sources of these two groups in
natural waters.
Assessment
Bacterial contamination is the only water quality
monitoring parameter discussed in these Guidelines that
has little effect on aquatic organisms, but is very significant
to human use. Bacterial contamination in forested areas can
result from a variety of sources, including dispersed and
developed recreation, wild and domestic animal popula-
tions, and human settlements.
The use of bacterial parameters to monitor water quality
for drinking and bathing is based more on correlations than
a direct causal link. Historically, total coliforms have been
used as the primary bacterial indicator of human health risk;
however, over the last 20 years, three more specialized
groups of bacteria have been increasingly utilized for water
-------
quality monitoring because they show a better correlation
with human health risk. The three groups for which proce-
dures (APHA, 1989) and standards (EPA, 1986b) have been
adopted are fecal coliforms, fecal streptococci, and entero-
cocci. Each of the four groups currently used in water quality
monitoring has aparticular significance, and the useof more
than one group may be beneficial in some cases.
Total coliforms are useful for assessing contamination
of finished drinking water because they are the largest and
most diverse group, and any bacterial contamination of
drinking water is considered unacceptable. Fecal coliforms
are a better indicator of contamination in natural waters, and
this is largely due to their more specialized nature. Fecal
streptococci are similar to fecal coliforms, but more com-
mon in animals than fecal coliforms. In some cases the ratio
of fecal coliforms to fecal streptococci can provide some
insight on the source(s) of contamination, but this ratio must
be applied with caution (EPA, 1978). Streptococci are the
most recent addition to the family of bacterial parameters,
and this appears to be the best indicator of contamination for
recreational waters because it is both more sensitive and
more directly correlated with human health risk (e.g.,
Vasconcelos and Anthony, 1985).
Bacterial counts tend to be highly variable over time,
and the standards for drinking and bathing explicitly recog-
nize this variability. The standards for bathing waters also
recognize that the link between bacterial counts and human
health is indirect. Thus waters used infrequently for bathing
have less stringent standards than waters at designated
bathing beaches.
This means that the type and frequency of monitoring
for bacterial contamination should depend on the beneficial
use. More intensive monitoring is appropriate in areas
which provide domestic water supplies, or in areas which
have heavy recreational use. In these cases any sign of
contamination is likely to require an immediate manage-
ment response and public notification. On the other hand,
the high variability of bacterial counts means that any single
test is of questionable value, and this is particularly true for
total coliform.
The above considerations suggest that one should err on
the side of caution when designing a bacteriological moni-
toring program and analyzing the resulting data. The
relatively high cost of not detecting contamination and the
relatively low cost of analyzing individual samples mean
that monitoring should be more regular and intensive than
for most of the other monitoring parameters discussed in
these Guidelines. For the same reasons any statistical analy-
sis might use a larger alpha value (i.e., a greater likelihood
that the results are due to chance) in exchange for more
power (i.e., a greater likelihood of finding contamination
when it is present) (Part I, Section 3.4.2).
CHAPTER 7. AQUATIC ORGANISMS
7.2 AQUATIC FLORA
Definition
The flora responsible for primary production in aquatic
environments can be classified taxonomically, function-
ally, or morphologically. In classical plant taxonomy, the
primary groups of aquatic plants are the algae, vascular
macrophytes, and mosses. In most streams and lakes in
forested areas, the bulk of the primary productivity is due to
algae (Hynes, 1970).
Aquatic ecologists often use a functional classification
with three primary categories: (1) free-floating or plank-
tonic forms, (2) plants attached to the substrate, and (3)
plants rooted into the substrate (Weitzel, 1979). The rela-
tive importance of these three categories is determined
largely by the physical features of the habitat. Free-floating ,
plants, for example, are significant only in still waters or
large rivers where there is sufficient time for them to build
up their populations. Rooted aquatic plants are rarely found
in areas where the bed material is coarse or subject to
frequenttransport. Attachedplants—mainly benthic algae—
are most important in gravel-bedded headwater streams.
Some streams may receive free-floating plants washed in
from lakes or backwater areas (Hynes, 1970).
Morphologic classification systems for aquatic flora
can be simpler than the taxonomic and functional ap-
proaches. The usual distinction is between microflora and
macroflora, but these are arbitrary size classes, and in -the
initial growth stages macroflora species can be part of the
microflora (Hynes, 1970).
Most studies of aquatic flora have concluded that the
attached plant community is best suited to water quality
monitoring (Weitzel, 1979). Two terms are commonly used
to refer to the attached flora—Aufwuchs and periphyton.
Although some authors consider these synonymous,
Aufwuchs—a German term meaning attached growth-
refers to all organisms growing on or attached to a substrate,
and this includes heterotrophic organisms such as bacteria,
bryozoa, and sponges, as well as small mobile organisms
(e.g., protozoans and insect larvae) living within the mat
(Power etal. 1988; Ruttner, 1953; Wotton 1988). Periphy-
ton often has a slightly narrower definition—aquatic flora
growing on submerged substrates—and this may or may not
be limited to the microflora (Cattaneo 1987; Hutchinson,
1975; Odum, 1971; Weitzel, 1979). In forested streams in
the Pacific Northwest, the attached algal communities are
commonly referred to as benthic or epibenthic algae (Hudon
and Legendre 1987). Diatoms usually are the most impor-
tantand diverse algal group in benthiccommunities(Pryfogle
andLowe, 1979). Epiphyticalgaerefers toattachedmicroalgae
(e.g. diatoms) that grow on the surface of macrophytes
(Cattaneo and Kalff, 1980).
-------
Part II
Relation to Designated Uses
Bcnthic algae can be the dominant group of primary
producers (photosynthetic organisms) in stream ecosys-
tems (Hynes, 1970; Wetzel, 1983). Mats of attached algae
form rich assemblages of plant, bacteria, and animal species,
all of which are important components of the overall food
web(Weitzel,1979;Poweretal., 1988). In small headwater
streams, the contribution of organic matter by benthic algae
may be outweighed by inputs of organic matter from ripar-
ian and forest vegetation. With increasing stream size,
however.theimportanceofautotrophicproductionincreases.
Increased benthic algal production is linked to increased
production of benthic invertebrates and fish (Gregory et al.,
1987).
In lakes and downstream portions of slow-flowing
rivers, all three functional plant groups—free-floating, at-
tached, and rooted—can affect the designated uses of water
and be ecologically important habitats (Power etal., 1988).
High levels of free-floating plants, for example, will impair
the clarity of the water and may have adverse esthetic effects.
Aquatic macrophytes can adversely impact recreational
uses such as swimming and boating, and also degrade the
esthetic value.
Ecologically, an increase in primary production can
increase the production of invertebrates and fish in streams.
However, nocturnal respiration can cause oxygen depletion
in waters with high primary production and low reaeration
rates. Even relatively small reductions in dissolved oxygen
can have adverse effects on both invertebrate and fish com-
munities(Section2.4). Developmentofanaerobicconditions
will alter a wide range of chemical equilibria, and may
mobilize certain chemical pollutants as well as generate
noxious odors.
High primary production also can lower the concentra-
tion of nitrogen and phosphorus because of the rapid uptake
of nitrate, ammonium, and phosphate by algae and other
aquatic plants (Section 2.5). Aquatic plants can influence
the color, taste, and odor of water (APHA, 1976).
Response to Management Activities
Numerous studies have related organic pollution to
specific aquatic plants or plant community parameters.
Relatively little definitive data are available on the effects of
forest management activities on aquatic plants. Specific
activities that might be expected to affect aquatic plants
include herbicide applications, opening up of the riparian
canopy, increased stream temperature, increased nutrient
concentrations, and sedimentation.
Aerial herbicide applications may adversely affect pri-
mary productivity, but this is highly dependent upon the
protective measures taken. The use of buffer strips, appro-
priate application technology, and good weather conditions
can greatly reduce the amount of herbicide reaching the
stream channel. Sullivan etal. (1981) found no toxic effects
on stream and pond benthic algae following the application
of a herbicide (Roundup) in coastal Oregon. In the coastal
Carnation Creek watershed in British Columbia, Holtby and
Baillie (1989) observed a decline in benthic algal standing
crop and biomass accumulation in the first month after a
glyphosphate application. In both studies the large temporal
and spatial variability in algal growth and abundance made
it difficult to determine the effect of the herbicides.
Partial or complete removal of the riparian canopy will
increase direct solar radiation, and this may increase benthic
algal growth. In headwater streams of the Cascades, pri-
mary production is proportional to sunlight at low light
intensities. At 20% of full sunlight, the benthic algal com-
munities are photosynthetically saturated, and additional
sunlight may not enhance production (Gregory et al., 1987).
The temperature increases associated with forest har-
vest and sedimentation (Section 2.1) affect primary pro-
duction andrespiration. In general, an increase in temperature
will increase the rate of respiration more rapidly than the
rate of photosynthesis, so an increase in temperature de-
creases net primary production (Gregory et al., 1987). In
most cases the effects of a change in temperature cannot be
detected, as a laboratory study showed that primary pro-
duction increased by only 30% following a 10°C increase in
temperature (Gregory et al., 1987). High light intensities
appear to favor filamentous green algae, and this may explain
the observed increase in abundance following clearcutting
(StocknerandShortreed, 1988).
As discussed in Section 2.5, a variety of forest manage-
ment activities can increase the availability of nitrogen and
phosphorous, and this has been demonstrated to stimulate
primary production (e.g., Gregory, 1980; Stockner and
Shortreed, 1978; Triska et al., 1983). Increased stream
productivity, due to increased nutrient output from water-
sheds following harvest, typically lasts only a few years
(Gregory et al., 1987; Vitousek et al., 1979). The rapid
uptakeofnutrients by primary producers means that increases
in production may be quite localized (e.g., Holtby and
Baillie, 1989).
Increased sedimentation can reduce primary production
by reducing the area of suitable substrate and by reducing
the depth of light penetration. The most damaging sediment
is sand-sized particles, as they are easily mobilized and do
not provide an adequate surface for colonization (Hynes,
1970). Increased bedload may increase primary production
by increasing stream width and temperature (Section 4.3).
An increase in silt- and clay-sized particles will tend to
decrease primary production by reducing the amount of
light within the water column and coating the stream bed
(Section 4.2).
This discussion indicates that forest management ac-
tivities affect the productivity and composition of the aquatic
flora in different ways by a variety of processes. The net
effect will depend on the relative balance and interactions
-------
CHAPTER 7. AQUATIC ORGANISMS
among these effects. In most cases the net change in the
aquatic flora can be linked to some of the designated uses of
water. Usually, however, the spatial and temporal variabil-
ity in the aquatic flora will preclude the definitive detection
of management effects in streams, and hence the impact on
designated uses can only be assumed. For example, benthic
algae in many streams undergo dynamic cycles of growth,
senescence, decay, and export. Although information is
available about some factors that influence algal biomass at
a site, little is understood about the effects of factors such as
floods or algal grazing by aquatic organisms.
Measurement Concepts
Of all the aquatic plants, algae have long been the most
widely used indicator of water quality and stream condition
(Hynes, 1966; APHA, 1976; Weitzel, 1979). Some ad-
vantages of using algae include the following:
1. Their presence and growth integrate numerous physi-
cal factors;
2. their relatively short life cycle makes them useful
indicators of short-term impacts;
3. they are sensitive to certain pollutants, such as herbi-
cides and excessive inputs of nutrients, which may not
affect other organisms; ,
4. sampling can be easy and inexpensive depending on
the situation; and
5. relatively standard methods exist for evaluating the
structural and functional characteristics of algal com-
munities (EPA, 1989).
Disadvantages to the use of algae and other aquatic plants
are as follows:
1. They are highly variable with location (Pryfogle and
Lowe, 1979);
2. they are highly sensitive to small changes in current
velocity, substrate type, and other physical factors
(Weitzel etal., 1979);
3. considerable expertise and time are needed to identify
both attached and free-floating microflora species;
and
4. the use of qualitative information, such as the presence
or absence of particular species, may be invalid or
appropriate only on a very coarse scale (Weitzel,
1979; Weitzel et al., 1979).
Both species and community parameters have been
used to characterize aquatic plants and monitor water qual-
ity. The simplest technique is to use selected species as
indicators of water quality. This assumes that the habitat
requirements of a particular species are known, that the
habitat requirements are relatively constant, and that pres-
ence or absence is solely a function of water quality. Lists
of species associated with organic pollution have been
developed and used to distinguish up to nine different zones
of pollution (APHA, 1976; Weitzel et al., 1979). The
indicator species approach is limited in that it allows only a
qualitative assessment of stream condition from specific
pollutants, and it has been widely criticized (Patrick, 1973;
Pielou, 1975; Platts et al., 1983).
Community parameters can be divided into structural
characteristics, such as species richness, diversity, or bio-
mass, and functional characteristics, such as productivity
(Odum, 197 l;Rodgers etal., 1979). For benthic algae, these
parameters can be measured from natural or artificial sub-
strates.
Artificial substrates generally are accepted as being
comparable to natural substrates (Weitzel et al., 1979), and
their use eliminates the variability due to substrate type.
Several studies have shown that the variation between
replicates for parameters such as chlorophyll-a and ash-free
dry weight typically is 20-25% (Weitzel etal., 1979). Much
larger differences were found between artificial substrates
placed in apparently similar locations, and this was ascribed
to small differences in current velocity and solar radiation
(Weitzel etal., 1979).
The values andlimitations of species richnessand diversity
data are discussed in conjunction with benthic invertebrates
(Section 7.3), as they have been studied more intensively
than the aquatic flora. Patrick (1973) asserts that diatoms
are well suited to monitor pollution, but her methodology
requires counts of 5,000-8,000 individuals per site. Other
studies have used smaller counts, and the available data
suggest that at least 500 organisms are needed to estimate
the species distribution (Weitzel, 1979). Normal procedures
for fixing diatoms leave only the frustules (shell), and this
precludes the separation of live and dead diatoms. Inclusion
of dead diatoms in estimates of community parameters can
substantially bias the results (Owen et al., 1979). Other
floristic groups, such as macrophytes or phytoplankton,
usually are too location-specific or too rare to use for
estimating community parameters.
Biomass refers to the organic matter content per unit
area or volume, and this is sometimes incorporated in
monitoring programs. A correlation between water quality
and biomass is difficult to establish because so many other
factors, such as light, nutrients, and grazing intensity, may
be limiting (Weitzel etal., 1979). Another problem with the
use of biomass data is that up to 80% of the dry weight of
benthic algal communities is composed of sediment, diatom
frustules, and other inorganic matter that accumulates in the
algal mat For this reason biomass estimates should always
be based on the ash-free dry weight (Weitzel et al., 1979).
Chlorophyll-a is often used as a surrogate for biomass.
Typically the amount of chlorophyll-a is 1-2% of the ash-
free dry weight, but values can range between 0.15 and 4%
(APHA, 1976; Weitzel et al., 1979). Factors affecting the
concentration of chlorophyll-a include the age and physi-
ological state of the organism, amount of dead biomass
present, community composition, and abiotic factors such
as light intensity and nutrient availability (Clark et al., 1979;
Hudon and Legendre, 1987).
-------
Parti!
The Autotrophic Index is theratio of ash-free dry weight
to chlorophyll-a. A value less than 50-100 indicates that
virtually all the periphytic organisms are algae that are
actively photosynthesizing (autotrophs), and that there are
few organisms utilizing organic matter and pollutants (het-
erotrophs). Values higher than 100-200 indicate that a
substantial proportion of the biomass is composed of or-
ganisms that are not photosynthesizing (APHA, 1976;
Weitzel, 1979). However, ratios of 200-400 for actively
growing filamentous assemblages have been observed un-
der laboratory conditions (S. Gregory, Oregon State Univ.,
pers. comm.). Hence the Autotrophic Index is potentially a
useful ratio but may have limited applicability when die
primary pollutants are not rich in organic matter.
The primary metabolic processes of aquatic plants are
primary production (photosynthesis) and respiration. Nei-
ther of these is easily measured, particularly in stream
systems where the flow of water is of critical importance
(Weitzel, 1979; Wetzel, 1983). In most cases an index that
approximates production can be obtained by measuring the
accumulation of organic material (e.g., biomass) on artifi-
cial substrates over a period of 1-2 weeks. Other factors
besides water quality may affect production, respiration,
and the net rate of biomass accumulation; these include
grazing, sloughing, scour, colonization, and deposition.
Hence a high turnover rate (primary production divided by
biomass) can result in a low rate of biomass accumulation
but a high rate of primary production. In the absence of this
type of detailed information, it is difficult to relate water
quality to either algal production or biomass.
Standards
No specific standards have been established or pro-
posed for aquatic plant communities, although an objective
of the Clean Water Act is to restore and maintain the
biological integrity of water bodies. More specific biological
criteria are now being developed by the states (Part I,
Section 1.4; EPA, 1988b; EPA, 1990).
Current Uses
The use of aquatic plants other than benthic algae for
monitoring water quality may be more appropriate in lakes.
In lakes, both free-floating plants and aquatic macrophytes
may be directly linked to specific designated uses. Thus an
observed increase in algal biomass or production can not
only indicate a change in water quality but also can be related
to a designated use, such as recreation. In streams, however,
benthic algae production and biomass probably are the most
useful of all the aquatic flora parameters to monitor changes
in water quality. In both streams and lakes, the two main
problems with monitoring aquatic plants are (1) detecting a
statistically significant change in the face of large spatial
and temporal variability, and (2) relating any observed
change to specific management activities.
The first problem is a sampling problem. It can be
addressed by specifying the trade-offs between sampling
costs, the risk of an erroneous result, and the probability of
obtaining the true answer. A small pilot study is often
needed to adequately evaluate these trade-offs (Part I,
Chapter 3). Long-term data are necessary to determine if an
observed change is either part of a larger trend or within the
range of previous changes.
The second problem may be more difficult. In most
cases a variety of additional data (e.g., nutrient concentra-
tions and incoming solar radiation) will be needed to deter-
mine the cause of observed change. An increase in biomass
or chlorophyll-a, for example, could be caused by an in-
crease in nutrient levels, warmer temperatures, or a reduc-
tion in grazing. Data on management activities within the
watershed usually are necessary to determine the likely
cause(s). In most cases the results will not be definitive, and
some extrapolation or assumptions will have to be made.
Assessment
Benthic algae and attached algae on large macrophytic
plants (epiphytic algae) can dominate primary production in
many streams and rivers and provide the main source of
organic matter. Attached algae provide both food and habitat
for a wide range of invertebrates, and these invertebrates are
an important source of food for salmonids and other fish
(Power etal, 1988).
In lakes free-floating plants and macrophytes may be of
primary importance. Species composition, biomass, and
productivity of aquatic plants have been used to indicate up '
to seven different levels of lake eutrophication. Such detailed
determinations usually are based on the identification of
large numbers of diatoms, and this generally precludes their
use in most monitoring projects. These procedures also may
be of limited applicability in forested areas because there
typically is very little eutrophication, and the applicability
of the techniques to oligotrophic systems has not yet been
established.
Attempts to relate forest management activities to the
composition and growth of benthic algae have met with
limited success. The variability associated with replicated
artificial substrates (glass slides) within a sampler is 20-
25%. Differences between samplers placed in "similar"
environments are much greater, and this severely limits
one's ability to detect statistically significant change over
time or space. Although aquatic plants can be directly
linked to several designated uses, it usually is better to
measure the causative factors (e.g., changes in temperature,
riparian canopy opening, or bed material particle size)
rather than the resulting change in benthic algae or other
aquatic plants.
Aquatic plants are more likely to affect the designated
uses of water in lakes than in streams. In both stream and
-------
CHAPTER 7. AQUATIC ORGAN/SMS
lake ecosystems, some algal indicator such as chlorophyll-a
is generally the most appropriate monitoring technique. The
collection of presence/absence, species richness, and spe-
cies diversity data all require a trained taxonomist and may
require the identification of a large number of microorgan-
isms. The cost and difficulty of carrying out such a program
has led most people to use some indicator such as the con-
centration of chlorophyll-a. This is a useful approximation of
algal abundance, but it is not sensitive to small changes. Ar-
tificial substrates are unlikely to provide greater sensitivity,
and their use is advantageous only if other parameters, such
as ash-free dry weight or species composition, are to be ob-
tained from the samples. The area covered by aquatic mac-
rophytes might be another useful indicator of river or lake
conditions. No single technique is optimal under all situa-
tions, and additional data will be needed to identify the most
likely cause(s) of an observed change in the aquatic plants.
Even though the value of aquatic plants for water quality
monitoring may be limited, any data will increase our
understanding of the aquatic system. A measurement of
instream primary production, for example, may provide
some indication of the likelyresponseofthealgal community
to nonpoint source pollutants like nutrients and sediments,
even though we may not be able to directly measure this re-
sponse. Such information also could help indicate the relative
balance between primary production and terrestrial organic
inputs, and this information could help guide riparian zone
management. In general, an increased understanding of the
structure and functions of aquatic ecosystems should improve
both management effectiveness and the protection of aquatic
resources.
7.3 MACROINVERTEBRATES
Definition
Macroinvertebrates are animals without backbones that
are large enough to be seen with the naked eye. The lower
size limit is arbitrary. The U.S. Geological Survey has
adopted a mesh size of 0.21 mm as the most suitable for
sampling macroinvertebrates in flowing waters (Platts etal.,
1983), while APHA (1989) defines macroinvertebrates as
those invertebrates retained on a U.S. Standard No. 30 sieve
(0.595 mm openings).
A wide variety of taxonomic groups are found in fresh-
water environments, and these include annelids, crusta-
ceans, flatworms, mollusks, and insects. Benthic macroin-
vertebrates, which live on the stream bottom, are the group
most amenable to systematic study. Most research has
focused on aquatic insects, as these are the most common
and diverse macroinvertebrates in forested areas. It follows
that most freshwater monitoring programs have been di-
rected towards benthic aquatic insects, and these organisms
will be the primary focus of this section.
Relation to Designated Uses
Macroinvertebrates play several major roles in aquatic
ecosystems. They graze on periphyton (attached algae) and
feed on the terrestrial organic material that falls into the
stream. Other invertebrates act as predators and filter
feeders. Macroinvertebrates are a major food source for
most fish species in forested areas (Gregory et al., 1987).
Much of the ecological importance of macroinvertebrates
stems from their position as an intermediate trophic level
between microorganisms and fish (Hynes, 1970).
Benthic macroinvertebrates have several characteris-
tics which make them potentially useful as indicators of
water quality. First, many macroinvertebrates have either
limited migration patterns or a sessile mode of life, and this
makes them well suited for assessing site-specific impacts.
Second, their life spans of several months to a few years
allow them to be used as indicators of past environmental
conditions (Platts et al., 1983). Third, benthic macroin-
vertebrates are abundant in most streams. Fourth, sampling
is relatively easy and inexpensive in terms of time and
equipment (EPA, 1989). Finally, the sensitivity of aquatic
insects to habitat and water quality changes often make
them more effective indicators of stream impairment than
chemical measurements (EPA, 1990). In Ohio, for ex-
ample, 36% of impaired stream segments detected with
biosurveys could not be detected using chemical criteria
alone (Ohio EPA, 1988).
Disadvantages of monitoring macroinvertebrates in-
clude arelatively high degree of variability within or between
sites (Minshall and Andrews, 1973), local or regional varia-
tions in the sensitivity of a given organism to stress (Winget
and Mangum, 1979), the need for specialized taxonomic
expertise, and the cost of processing (sorting and identifying
invertebrates) samples containing numerous organisms.
Much of the variability between samples is due to the highly
heterogeneous distribution of macroinvertebrates with depth,
current speed, and substratum (Platts et al., 1983; Morin,
1985). Thismeans that sampling locations mustbe carefully
selected and that sampling usually should be stratified by
habitat type (Part I, Section 3.3).
Most monitoring techniques require macroinvertebrate
identification to genus or species. Interpretation of the
results requires knowledge of the habitat requirements of
the identified taxa and familiarity with the typical macroin-
vertebrate community in the study area. In some sampling
techniques, considerable effort may be needed to separate
organisms from the substrate. The difficulties associated
with the separation, identification, and enumeration of taxa
may produce inadequate sampling programs (Jackson and
Resh, 1988).
-------
Part II
Response to Management Activities
The effects of forest activities on macroinvertebrate
communities vary. Increases in the riparian canopy opening
or the amount of organic material in the streams generally
enhance aquatic insect populations. An increase in fine
sediment usually has the opposite effect (Gregory et al.,
1987; Section 3.1). Removing the riparian canopy de-
creases the input of terrestrial organic material and the
number of detritivores. However, this decline often is over-
whelmed by the corresponding increase in primary pro-
duction and herbivorous insects (Gregory et al., 1987).
Several studies have documented an increase in primary
productivity after partial or complete removal of the riparian
canopy (e.g., Hansmann and Phinney, 1973; Murphy et al.,
1981). However, no increase was found in Carnation Creek
in coastal British Columbia, where phosphorus was found to
be the limiting factor (Stockner and Shortreed, 1988).
Logging-induced increases in aquatic insects have been
observed in northern California (Erman etal., 1977) and the
Oregon Cascades (Murphy et al., 1981). While logging
activities may increase total abundance, species diversity is
usually reduced (Gregory et al., 1987).
Invertebrate communities also are affected by manage-
ment practices on forest lands. Buffer strips 30 m wide
appeared to protect invertebrate communities from logging-
induced changes (Newbold etal., 1980), but buffer strips 10
m wide still resulted in a decrease in detrital inputs and
macroinvertebrate densities (Gulp, 1988). The net effect of
logging on aquatic macroinvertebrates depends on the rela-
tive balance among all the controlling factors.
Measurement Concepts
A variety of sampling and data analysis techniques can
be used to monitor macroinvertebrate communities. Some
of the more common parameters include presence or ab-
sence data,functionalfeedinggroup analysis, and community
parameters. Sample collection techniques can be equally
varied, ranging from the placement of uncolonized sub-
strates to kick nets, drift nets, and fixed-area substrate
samples.
Sampling Techniques. Sampling techniques for mac-
roinvertebrate can be classified as qualitative, semiquanti-
tative, or quantitative (Plaits etal., 1983). Qualitative tech-
niques rely on indicator species or an evaluation of selected
functional or taxonomic groups. Generally the samples for
qualitative evaluation are not collected on the basis of a
specified area or collection effort, and this severely limits
any numerical analyses.
Sampling procedures that use uniform substrates or a
specified amount of collection effort (e.g., a 3-hour drift net
sample, or 50 sweeps with a dip net) are termed semi-
quantitative techniques (Platts et al., 1983). Data from these
samples can be used for qualitative purposes, such as the
presence or absence of particular taxa, or for estimating
population characteristics such as diversity, total numbers, or
biomass. The primary limitation of semiquantitative methods
is that results are on a per sample basis rather than per unit area
(Platts etal., 1983).
Quantitative techniques involve complete sampling in a
specified area. The resulting density data are on an absolute
basis (e.g., number of organisms per unit area), and this
allows a comparison of populations over time or space. Data
collected using quantitative techniques can be used to es-
timate productivity as well as population characteristics.
Although qualitative techniques typically arequicker and
easier than semiquantitative or quantitative procedures, they
yield less specific information. This generally makes quali-
tative techniques less sensitive and less reliable. Since a
similar level of expertise is needed to analyze the samples
and interpret the results, mostprojectsshoulduse semiquanti-
tative or quantitative sampling methods (Platts et al., 1983).
This range of sampling procedures indicates that a wide
variety of sampling techniques have been developed to
accommodate varying study objectives and locations. The
composition of the substrate, water depth, and current
velocity largely determines the most appropriate technique.
The most common methods include various types of nets,
substrate sampling techniques, and the placement and sub-
sequent retrieval of artificial substrates (Greeson et al.,
1977). Each technique has a different set of errors and bias,
making comparisons of data from different sampling tech-
niques difficult (Platts et al., 1983)." For this reason moni-
toring studies should select and utilize one of the better-
known techniques and apply this as widely as possible to
ensure comparable data.
Artificial substrate samplers are useful in large rivers or
wherever natural substrates cannot be effectively sampled
(EPA, 1989). The most common artificial substrate tech-
niques make use of multiplate (Hester and Dendy, 1962) or
basket (Mason et al., 1973) samplers. Multiplate samplers
are a set of stacked plates that are left in a stream or lake for
a period of at least several weeks and then retrieved for
analysis. Basket samplers are similar in principle, but utilize
rocks as the substrate for colonization. Advantages and
disadvantagesof artificial substrates are discussed in Greeson
etal. (1977), Rosenberg and Resh (1982), andEPA'sRapid
Bioassessment Protocols (EPA, 1989). The most common
criticism is that they do not provide a representative sample
of the natural community. Major advantages include lower
sample variability, and elimination of substrate differences
between sample sites.
Drift nets are used to sample macroinvertebrates that
have been dislodged or are migrating, and typically they are
left in place for at least several hours. However, the nets can
become clogged if they are not regularly cleared, and this
will reduce the number of organisms captured in the nets.
Drift net data are expressed as numbers and biomass of
organisms per unit discharge (APHA, 1989).
-------
CHAPTER 7. AQUATIC ORGANISMS
Dip nets are used to qualitatively collect organisms
associated with backwater areas, nearshore areas, and de-
posits of organic debris. Collection techniques can be spe-
cified by area and effort in order to obtain semiquantitative
data. In deep waters and in areas with fine substrates, a
variety of grab samplers, such as Eckman or Peterson
dredges, may prove most effective.
In small forested streams, Surber (Surber, 1937) and
modified Hess (Waters and Knapp, 1961; Jacobi, 1978)
samplers are most often used for quantitative sampling
(Platts et al., 1983). Both of these samplers utilize a frame
to delineate a specific area of stream bottom and a net to
capture the benthic fauna as the substrate is disturbed to a
depth of 5 or 10 cm. The primary difference is that the
modified Hess sampler uses a closed frame, while the Surber
sampler relies on the current to carry dislodged organisms
into the attached net The mesh size of the net must be large
enough to allow the free flow of water and fine sediments,
but small enough to capture most of the benthic inverte-
brates. APHA (1989) suggests a mesh size of 0.595 mm, but
in forest streams with little or no fine sediments a smaller
mesh size may be preferable. For qualitative or
semiquantitative samples, a kick net typically is used. Kick
nets can be made by attaching a fine meshed screen between
two rods. The net is held vertically in the stream while the
substrate immediately upstream is disturbed. The current
then carries the dislodged organisms into the net By speci-
fying the area and effort sampled, semiquantitative data can
be obtained (Platts et al., 1983).
Sampling methods must take into account the time of
year, number of samples per site, and habitat to be sampled.
Significant changes in invertebrate populations occur dur-
ing the year because of natural life cycleprocesses (Minshall
and Andrews, 1973). To account for these changes, sam-
plingprograms must define which season(s) will be sampled
and maintain this sampling period throughout the life of the
study. Collecting samples in more than one season is
preferable, but when this is notpossible the optimal sampling
season is the period when most macroinvertebrates are both
large enough to be retained during sieving and sorting, and
identifiable with the most confidence (EPA, 1989). In
Region 10 this is typically late winter and early spring.
However, sampling effectiveness is reduced during or just
after periods of high water. This suggests that the optimal
sampling time in streams with snowmelt runoff will be just
prior to spring snowmelt, while rain-dominated streams
should be sampled after winter storms when the flow regime
is relatively stable.
The number of samples that should be collected at each
site is a function of the size of the site to be sampled and the
and variability between replicate samples. Quantitative
methods generally require more samples per site than semi-
quantitative methods because of the greater variability in
invertebrate densities compared to relative abundances
(APHA, 1976). In general, quantitative methods will re-
quire at least 5-10 samples per site in order to detect sta-
tistically significant differences (Platts et al., 1983).
The habitat selected for sampling will greatly affect the
type of invertebrate community observed. The most diverse
invertebrate communities generally occur in riffle/run
habitats withgravelandcobblebottoms(EPA, 1989). Since
areas with the greatest diversity will provide the most
sensitive indicators to environmental changes, riffle/run
habitats are usually preferred for sampling when they are
available. Sampling methods developed in North Carolina
take qualitative samples from five microhabitats (riffles,
macrophytes, logs, sand, and leaf packs) from each site to
document invertebrate populations (Lenat, 1988).
Data Analysis. A variety of community and population
indices can be used to characterize benthic macroinverte-
brates, although the choice will be somewhat constrained
by the particular sampling technique used to collect the
sample. One useful approach is to divide benthic aquatic
insects into functional feeding groups such as shredders,
collectors, scrapers, and predators (Cummins, 1973).
Changes in the relative abundance of the different func-
tional feeding groups can indicate habitat change. For
example, an increase in the number of scrapers as compared
to shredders suggests an increase in theproduction of attached
algae due to a reduction in the riparian canopy or an increase
in stream width. Considerable care is needed in the separa-
tion of organisms, as closely related species can fall into
different functional feeding groups. Platts et al. (1983)
conclude that this approach shows promise, but still must be
regarded as experimental. They recommend that the func-
tional feeding group approach be used in conjunction with
more conventional community analysis techniques.
Some of the more commonly used community param-
eters include abundance, species richness, diversity indices,
and biotic indices. Each of these parameters considers only
a part of the overall invertebrate population characteristics,
and each has certain drawbacks in terms of representing the
complex assemblage of organisms present at any given site
(Elliott, 1977). Itisthereforebeneficialtousemorethanone
community measure for assessing invertebrate populations.
Abundance can be expressed in absolute terms as the
number of individuals per unit area present, or in relative
terms as a percentage of total numbers. The absolute abun-
dance is a useful indicator of the overall productivity at a
site. Relative abundance values, such as percent contribu-
tion of the dominant taxon, indicate thecommunity balance.
Communities dominated by justafew taxaindicate environ-
mental stress (EPA, 1989).
Species richness generally refers to the total number of
taxa present. The total number of taxa in specific orders
(e.g., total number of mayflies, stoneflies, and caddisflies)
also is a useful indicator (EPA, 1989). Lenat (1988) ob-
served a high correlation between species richness and
water quality inNorth Carolina. In Oregon, species richness
showed good correlation with trout populations from high
-------
Part II
desert streams (JoeFurnish, ValeBLMDistrict,pers. comm.)
In some instances, however, moderate degradation may
allow new species to colonize a site while not excluding less
tolerant species (Gregory etal., 1987). Under these circum-
stances species richness will be maximized, and a signifi-
cant decline will not occur until habitat degradation begins
to eliminates the less tolerant species. Hence knowledge of
the tolerance ranges of different taxa to different pollutants
is important for the proper interpretation of species richness
data. EPA has published pollution tolerance information on
most majoraquatic insectorders (e.g., Harris andLawrence,
1978; Hubbard and Peters, 1978).
Diversity indices combine species richness and relative
abundance. A variety of indices have been developed, with
the Shannon-Wiener index probably being the most com-
mon (Plaits et al., 1983). The use of diversity indices for
detecting environmental stress has been criticized because
they:
1. do not incorporate any trophic community structure,
2. exhibit considerable variation even in undisturbed
sites,
3. may be insensitive to disturbance, and
4. are insensitive to the ecological differences between
sites (e.g.,Pielou, 1975; Zand, 1976).
Various biotic indices have been developed to capture
more of the complexities of natural populations. The Biotic
Condition Index (BCI) incorporates stream habitat, water
quality, and environmental tolerances of aquatic insects
(Winget and Mangum, 1979). Tolerances have been esti-
mated or determined for several hundred aquatic insects.
The BCI is based on the mean tolerance of the aquatic
insects predicted for a site divided by the actual mean
tolerances of the aquatic insects found on the site. This
method has been used extensively by the Forest Service and
the Bureau of Land Management in the Western U.S.
In an effort to provide state governments with a cost-
effective integrated biological index, EPA developed five
Rapid BioassessmentProtocols(RBP) (EPA, 1989). Proto-
cols I, II, and III use benthic macroinvertebrates to assess
water quality impairment; protocols IV and V use fish. RBP
I relies upon the qualitative abundance of different
maeroinvertebrate taxa and professional judgment to deter-
mine whether water quality is impaired or unimpaired. It
was designed as a quick method to screen different sites
(EPA, 1989).
Rapid Bioassessment Protocol II (RBP II) is a more
intensive and systematic procedure intended to distinguish
among three categories of water quality (non-impaired,
moderately impaired, and severely impaired). Separate
collections of macroinvertebrates are obtained from riffle/
run areas and coarse paniculate organic matter. To reduce
sample processing time, a 100-organism subsample is ran-
domly sorted from the composited riffle/run samples. Each
organism in this subsample is classified to the lowest taxo-
nomic unit (order, family or genus) and functionally by
feeding group. Larger subsamples (200 or 300 organisms)
can be sorted, but they have not been shown to increase the
sensitivity of the procedure (EPA, 1989). The macroin-
vertebrates collected from coarse paniculate organic matter
are classified as shredders or non-shredders. From these
data eight community, population, and functional feeding
group parameters are calculated. These are combined to
yield a single evaluation of "biotic integrity," and this is
compared to the biotic integrity of a comparable, unimpaired
site ("reference station") (EPA, 1989). The particular
combination and valuation of parameters in RBP II were
developed from a single field study in North Carolina (EPA,
1989), although several of the individual parameters have
been derived from previous studies.
RBP HI, a more detailed protocol for benthic macroin-
vertebrates, is very similar to RBP II, but requires identifi-
cation to the genus or species level. The more precise
valuation of the eight metrics allows four levels of impair-
ment (severe, moderate, slight, and no impairment) to be
distinguished. Again validation is based on a field study in
North Carolina and the use of similar procedures in other
studies (EPA, 1989).
Before the Rapid Bioassessment Protocols are imple-
mented in EPA's Region 10, further study is recommended
to determine if:
1. riffle/run habitats adequately characterize the "bio-
logical integrity" of a stream reach and accurately
determine impairment,.
2. a subsample of 100 macroinvertebrates is sufficient to
characterize the riffle/run community,
3. classification to the family level is sufficient in RBP
II,
4. the pollution tolerance data developed for species in
other areas are applicable to Region 10, and
5. the selection and combination of the possible metrics
used to obtain the biotic integrity are relevant and
appropriate in all cases.
Once these methodological questions have been answered,
the different protocols must be validated in the different
ecoregions of the Pacific Northwest and Alaska.
Standards
The principal objectives of the Clean Water Act are "to
restore and maintain the chemical, physical and biological
integrity of the Nation's waters" (Section 101). Current
water quality programs focus on chemical integrity and, to
a lesser degree, on physical integrity (EPA, 1990). It is
becoming apparent, however, that chemical criteria do not
always protect biological integrity, even though the water
quality criteria for parameters such as pH and dissolved
oxygen are based in part on the sensitivity of aquatic mac-
roinvertebrates (Part I, Section 1.4; EPA, 1986b). The
inadequacy of chemical and physical criteria to protect
biological integrity is particularly true for nonpoint source
-------
CHAPTER 7. AQUATIC ORGANISMS
pollution and habitat degradation. To achieve the goals of
theClean Water Actandprotectinstream biological integrity,
EPA is requiring the incorporation of narrative biological
criteria into state water quality standards (Part I, Section 1.4;
EPA, 1990).
Current Uses
Fifteen states are now developing biological assessment
programs to support future development of biological crite-
ria (EPA, 1990). Some state programs use biological
monitoring to evaluate stream impairment, but are not
developing specific biological standards. Other states are
refining sampling and evaluation methods so that biological
standards can be implemented in the future.
Four states currently have biological criteria that are
used to enforce water quality standards: Arkansas, North
Carolina, Maine, and Ohio. In all four states the biological
criteria are based, at least in part, on macroinvertebrate
community characteristics (EPA, 1990). Ohio has the most
comprehensive biological criteria. Biological indices have
been developed for fish and macroinvertebrates for each of
the fiveecoregions within the state. These indices havebeen
successfuUy incorporated into the State water quality stan-
dards.
In Region 10, Oregon has been using macroinvertebrate
assessments to determine stream impairment below point
source discharges (R. Hafele, Oregon Dep. Environ. Qua!.,
pers. comm.). Biological assessment methods for monitor-
ing impairment due to nonpoint pollution sources are now
being developed in Oregon and Idaho (T. Maret, Idaho Dep.
Environ. QuaL.pers. comm.). In Oregon macroinvertebrate
assessments are an important component of the nonpoint
source monitoringprogram, and Washington is studying the
use of macroinvertebrates for monitoring and evaluation.
Assessment
Aquatic insects display several characteristics which
make them potentially useful for monitoring purposes.
They are relatively sensitive to change, abundant in aquatic
ecosystems, and can be directly linked to an important
designated use (fisheries). Their use in monitoring has been
limited by the difficulties in defining appropriate param-
eters to measure, the level of expertise required to analyze
macroinvertebrate collections, and the difficulty in obtain-
ing representative samples.
TheRapidBioassessmentProtocols (EPA, 1989) are an
important step towards establishing sampling procedures
and measurement parameters for assessing water quality
using macroinvertebrates. Additional work will be needed
to establish and verify these assessment procedures for the
different ecoregions. Currently the applicability and reli-
ability of the methodology is being studied in several
watersheds in Oregon and Washington (R. Hafele, Oregon
Dep. Environ. Qual., pers. comm.).
An important limitation of the Rapid Bioassessment
Protocols is that they were not designed for quantitative
water quality monitoring. The original'intent was to de-
velop inexpensive screening tools, and the maximum
resolution of the current protocols is four qualitative levels
of water quality (EPA, 1989). Quantitative field data may
allow additional inferences to be made.
In summary, aquatic macroinvertebrate monitoring is a
useful tool for evaluating general water quality condition
and the extent to which designated uses are impaired or
supported. Biological measurements often are less expen-
sive then detailed chemical analyses, as a trained entomolo-
gist can use aquatic insect data to infer a great deal about the
site under consideration. To be most effective and reliable,
however, biological studies need to be integrated into a
monitoring plan that includes both physical and chemical
evaluations.
7.4 FISH
Definition
Both resident and anadromous fish communities are
found in many of the streams and lakes in forested areas in
EPA's Region 10. Twelve salmonid species are commonly
found in the forested watersheds of the Western U.S., and
these generally are regarded as the most valuable sport and
commercial species (Everest, 1987). All of the salmonid
species have life stages that are directly affected by manage-
ment activities and natural disturbances in watersheds. For
some water quality parameters, salmonid spawning and
rearing is the most restrictive designated use.
Fish are a useful surrogate or integrator of a variety of
physical and biological factors. Some of the factors neces-
sary to sustain or restore aparticular fish population include
the following:
1. adequate streamflow (i.e., water depth and habitat
space), . :
2. sufficient spawning habitat,
3. sufficient rearing habitat, .
4. appropriate food sources at different life stages, and
5. proper environmental conditions (particularly tem-
perature, dissolved oxygen, and turbidity).
For anadromous fish there must also be an absence of
migration barriers.
The use of fish for monitoring presents many parallels
to the sections on algae (Section 6.2) and macroinvertebrates
(Section 6.3). Monitoring can be based on the presence or
absence of particular species, numbers of a particular sper
cies, or community parameters such as productivity, den-
sity, and diversity (e.g., Hendricks et al., 1980). The
conceptual advantages and disadvantages of these different
-------
Part II
parameters are briefly discussed in the following sections,
as are the specific techniques which pertain to the use offish
for water quality monitoring.
Relation to Designated Uses
Fisheries are a very important designated use in fresh,
estuarine and salt waters. Sport and commercial fishing—
primarily of salmonid species—are each worth hundreds of
millions of dollars. In many rural areas sport and commer-
cial fishing are major components of the local economy.
Fish also have important economic, cultural, and subsis-
tence values for many native Americans.
Ecologically fish are important because they represent
the higher trophic levels in streams and lakes. Although fish
are the primary predators of macroinvertebrates, their role
in the food web varies by species and age. At certain times
fish are an important food source for terrestrial fauna such
as bears, raptors, and raccoons. Because fish are high in the
aquatic food web, they can serve as excellent indicators of
the overall physical, chemical, and biological condition of
streams.
Salmonids and other large species usually have consid-
erable public appeal. A decline in, or loss of, these species
willgenerateconsiderableadversepublicreaction. Spawning
areas, fish ladders, and falls with actively jumping fish may
be popular public attractions.
Salmonid species generally have the most stringent
habitat requirements. Summaries of habitat requirements
for different salmonid species can be found in Everest
(1987),Everestetal. (1985), Reiser andBjornn (1979), and
fisheries reference books. Most monitoring activities have
focused on salmonids because of their economic impor-
tance, strict habitat requirements, and the fact that their
habitat requirements generally are better known than most
other fish species.
Response to Management Activities
Concern over the response of resident and anadromous
fish populations to forest management activities has been a
major stimulus to long-term, detailed studies on the effects
of forestmanagernenton streams. Studies in coastal Oregon
(Hall et al., 1987), southwestern British Columbia (Hartman
elal., 1987; Chamberlin, 1988), southeastern Alaska (Gib-
bons etal.,1987),andtheOlympic Peninsula in Washington
(Ccderholm and Reid, 1987) investigated the response of
the fish populations to different types and intensities of
forestmanagementpractices. An important, unifying result
of these studies is that forest management can affect a wide
variety of physical and biological parameters, including
temperature,bedmaterial,prirnaryproductivity,peak runoff,
low flows, and macroinvertebrate populations. Each of
these changes will in turn have a series of effects on fish
reproduction, rearing, and growth. The magnitude of these
effects will vary by species and age class. In some cases
adverse effects on one species may benefit another species.
The complexities of these interacting physical and bio-
logical effects makes it very difficult to predict the effects of
forest harvester othermanagementactivities. The adoption
of increasingly stringent BMPs and forest harvest regula-
tions, particularly in the riparian zone, means that the simple
characterizations applied in the past may no longer be
appropriate. The detailed, long-term forestry-fisheries
studies cited above have demonstrated the need to evaluate
impacts by species and life cycle stage, and not rely on
single, broad measures such as the total number of fish (e.g.,
Hartman elal., 1987).
Most of the links between management activities,
physical and biological change, and effects on fish are
discussed in the contextof the individual monitoring param-
eters such as temperature (Section 2.1), turbidity (Section
4.2), bed material (Section 5.6), and large woody debris
(Section 5.7). An excellent review of forestry-fisheries
interactions can be found in Salo and Cundy (1987).
Measurement Concepts
A wide variety of techniques have been used to assess
changes in the number and condition of fish. In many cases
the links between the fisheries measurements, water quality,
and management actions are tenuous. Since a complete
review is beyond the scope of this document, only the most
common and appropriate monitoring techniques are dis-
cussed in this section. Edited volumes by Alabaster (1977)
and Hocutt and Stauffer (1980) provide good overviews of
biomonitoring, while many of the fisheries techniques are
discussed in Nielsen and Johnson (1983).
Fish population counts or estimates probably are the
most common parameter. For anadromous fish, counts are
most often made of the number of fish returning to spawn or
the number of fish carcasses following spawning. One also
cancountthenumberofoutmigratingjuveniles(e.g.,smolts)
from a particular stream or river, but this requires the use and
regular maintenance of traps, nets, or weirs. Species which
rear for many months in streams, such as coho, are much
easier to count than species which outmigrate after emer-
gence and rear in estuaries, such as chum or pink salmon.
Counts of outmigrating young provide a more specific
indication of spawning and rearing habitat productivity than
counts of resident fish or returning adults.
Transient or resident populations within a stream reach
can be counted by a variety of means (Platts et al., 1983).
Electrofishing is the most common field technique (EPA,
1989), and this is discussed in detail by Reynolds (1983).
Electrofishing has the advantage of being relatively accu-
rate and efficient. It is particularly useful in areas which are
turbid or have numerous obstructions such as aquatic veg-
etation, woody debris, or undercut banks. Voltage, pulse,
and frequency adjustments are necessary for the following:
-------
CHAPTER 7. AQUATIC ORGANISMS
1. to reduce size selectivity,
2. to ensure efficient sampling in different-sized streams
with varying water quality, and
3. to minimize fish mortality.
The accuracy of population estimates can be improved by
making multiple passes with the electroshocker and remov-
ing the shocked fish after each pass. Some species can be
grouped together for total population estimates, while other
species with a different probability of capture must be esti-
mated separately (Platts et al., 1983). Electrofishing allows
for the collection of length and weight data, and this can be
used to evaluate condition and population structure.
Other methods to capture fish and estimate population
size include toxicants and explosives (Platts et al., 1983).
While these may allow more accurate population estimates,
they kill or alter the populations being counted and now are
rarely used.
Direct observation by snorkeling is an increasingly
common technique. It is particularly useful in streams with
low conductivity and in remote areas. Again there will be
variation in the accuracy of the technique by species. Trout
and salmon are more likely to hold their territory and be
counted, whereas darters and sculpins tend to be more
secretive during the day (Platts etal., 1983). Snorkeling can
be combined with habitat surveys to provide estimates of
species density and species composition for different habi-
tat types (Hankin and Reeves, 1988). Population estimates
obtained through snorkeling can be improved by
electrofishing in a subsample of the snorkeled habitats
(Hankin and Reeves, 1988). In small or steep-gradient
streams, direct observations may be limited to pools and
glides. The difficulty of obtaining accurate underwater
counts means that most surveys provide only an index of the
true population. Thus comparisons over time and space can
be made only when the counting procedures and conditions
arecomparable. In general, snorkelingpermitsatrueestimate
of fish populations only for certain species under particu-
larly favorable conditions using a carefully executed survey
(Platts etal., 1983).
For anadromous species, accurate counts of the return-
ing adult fish and departing smolts can be obtained by
placing nets or weir traps on the stream of interest. These
capture all migrating fish, but complete counts may require
several months. To prevent mortality the captured fish must
be regularly removed, and individuals often are counted and
weighed at this time. Conlin and Tutty (1979) provide a
useful field guide to trapping juvenile salmonids.
An estimate of the number of spawning salmonid pairs
can be obtainedby counting spawning nests (redds). Ground-
based counts are usually more accurate and less costly, and
they are the only appropriate technique for smaller forested
streams. Aerial surveys may be preferable on larger rivers,
but these are usually less accurate (Bevan, 1961). The
timing of the redd count is critical because early counts may
exclude late-spawning fish, while late counts may underes-
timate redd numbers because of the decreasing ability to
distinguish contemporary redds over time. Redd counts are
much more difficult for species or runs that spawn in lakes
(e.g.,sockeye),largerivers(e.g.,chinook),or glacial streams
(e.g., spring chinook and sturgeon).
Emergence traps are used to estimate the number of
juveniles emerging from a single redd. Emergence success
or percent survival through emergence is estimated from an
assumed egg count for each species. Low emergence
numbers are most often ascribed to infiltrating sediment, but
other causes, such as temperature, disease, and predation,
must also be considered. Another approach being used in
Idaho on a trial basis is to place egg baskets with known
numbers of eyed eggs in artificial redds, and use emergence
traps to obtain percent emergence.
Species presence or absence, species richness, and
diversity indices all have been used as relative or qualitative
indicators of waterquality (e.g., Warren, 1971; Cairns etal.,
1973; Langford and Howells, 1977). The limitations of
these parameters have already been discussed (Sections 6.2
and 6.3) and are briefly reviewed for fish in lotic environ-
ments in Hendricks et al. (1980). In evaluating these
measurements consideration must be given to the biogeo-
graphic region, season of measurement, and stream size
(Karr, 1981). Generally fewer species of fish occur in
undisturbed streams and lakes in the Pacific Northwest than
in the Midwest or Southeast, and this hampers the use of
diversity or richness measures as indicators of water quality.
However, the number of native species may be a sensitive
measure of the deterioration of pools and other habitat types
(Miller etal., 1988).
Over the last decade there have been several attempts to
develop more comprehensive and meaningful measures of
fish communities. The index of well being (IWB) incorpo-
rates two diversity and two abundance estimates with ap-
proximately equal weight (Gammon, 1980). The index of
biotic integrity (IBI) is obtainedby weighting and summing
12 individual measures (metrics) (Karr, 1981). The metrics
were selected on the basis of experience in the Midwest, and
they include parameters such as the total number of species,
the number of species tolerant and intolerant of poor water
quality, several trophic measures, and several indicators of
condition (Karr, 1981; Angermeier and Karr, 1986). While
this has been widely adopted in the East and Midwest,
substantial alterations must be made in order to apply it to
sites in Washington, Oregon, Idaho, and Alaska (Hughes
and Gammon, 1987; Miller etal., 1988).
The IBI is the basis for EPA's Rapid Bioassessment
Protocol (RBP) V. As with RBPII and RBP III (Section
7.3), the habitat quality and IBI for the site under study is
compared to the habitat quality and IBI for an unimpaired
reference station. Concurrent collection of water quality
data is also recommended (EPA, 1989). RBP V is designed
to distinguish five levels of water quality impairment (EPA,
1989).
-------
Part II
EPA's rapid bioassessment techniques also include a
protocol, RBPIV, for quickly assessing the general condi-
tion and trend of a particular stream reach. The assessment
is based on the completion of a questionnaire by a qualified
fish biologist familiar with the stream reach under study.
Although data from the questionnaire is qualitative, it is one
way to identify reaches needing further study (EPA, 1989).
Standards
At present there are no specific standards or criteria for
fish populations or community parameters. However, fish
do represent an important designated use in many streams
and lakes, and the broad objective of point and nonpoint
source water pollution control programs is to protect all
designated uses. Hence there is a general standard to protect
and maintain naturalpopulations of fish in unimpaired streams
and to restore fish communities in streams adversely af-
fected by management. Over the next several years, these
general standards will be formalized, as EPA is requiring the
addition of narrative biological criteria to state water quality
standards. The continuing application and refinement of
narrative criteria is intended to lead to quantitative biologi-
cal criteria within a few years (EPA, 1990).
Recently several species of salmonid fishes have been
proposed for listing as threatened or endangered by the U.S.
National Marine Fisheries Service. Under the Endangered
Species Act, a recovery plan mus the prepared for each listed
species. This recovery plan could require strict habitat
protection measures as well as the direct protection of the
designated species. If deemed necessary under the recovery
plan, land management activities that might adversely af-
fccthabitatquality could be precluded orseverely curtailed.
Current Uses
Some advantages of using fish for monitoring water
quality are as follows:
1. their mobility and relatively long life span allows
them to indicate broad-scale and long-term habitat
conditions,
2. their higher trophic position means that they can be
used as an integrator of changes in the lower trophic
levels,
3. they are relatively easy to collect and identify in the
field, and
4. the habitat requirements of many species are rela-
tively well known (EPA, 1989).
Disadvantages include the following:
1. the difficulty of obtaining a representative sample or
an accurate estimate of the population,
2. the variety of extraneous factors that can affect fish
populations during different life history stages (e.g.,
fishingpressure.predation,disease) (Hellawell, 1977;
Hocutt, 1981), and
3. the mobility and limited residence time of anadro-
mous species in freshwater.
The simple presence or absence of a particular fish
species may not be a particularly useful monitoring tech-
nique unless we know that it utilized the stream in the past
The mobility and adaptability of fish can result in a few
individuals being found even under extremely adverse con-
ditions. For example, die Mt. St Helens eruption caused the
upper part of the Toutle River in Western Washington to
have lethal summer temperatures, unsuitable spawning
gravels, virtually no cover, highly turbid water, and high
winter flushing flows. Yet a few anadromous fish were able
to spawn and produce offspring that successfully completed
their freshwater stage. Such examples suggest that habitat
change sufficient to cause complete loss of a species has to
occur on such a scale that monitoring becomes merely a
confirmation of the obvious.
In many cases a field examination by a fisheries biolo-
gist will permit identification of the key habitat variables for
the species of interest By combining this information with
theknown orexpected management impacts, one can develop
a series of hypotheses or questions that will point to specific
monitoring technique(s).
This implies a need to identify the causes and effects
being ascribed to forest activities, and designing the moni-
toring program accordingly. A carefully documented de-
cline in fish populations, for example, will only provide the
information that a particular population is declining; for a
remedial management program to be effective, more spe-
cific information is required.
In most cases it will be desirable to monitor each link in
the postulated cause-and-effect chain. Concern over fine
sediment as a limitation on spawning success, for example,
indicates the need to do the following:
1. identify fine sediment sources (perhaps only on a
qualitative or reconnaissance basis),
2. monitor changes in bed material particle size or
embeddedness (Section 4.6), and
3. evaluate spawning success.
Data on each of these three components are needed to
establishacause-and-effectrelationship. Data on the streams
or lakes of concern also should be compared to data from
unimpaired sites.
The IB I shows promise as a technique to evaluate stream
condition from a variety of measurements offish populations,
trophic structure, and species composition (Karr, 1981; Miller
et al., 1988). Application to the Willamette River in Oregon
required modification of both the scoring system and 5 of
the 12 metrics. The IBI more closely corresponded to changes
in water quality and substrate than the simpler index of well
being (Hughes and Gammon, 1987). Further work is needed
to determine the most appropriate metrics and scoring sys-
tems for the IBI by ecoregion, size of stream, and type of
pollution (Hughes and Gammon, 1987). Atthistimemostof
the work on biolo-gical assessmentand the IBI is focusing on
-------
CHAPTER 7. AQUATIC ORGANISMS
macroinvertebrates rather than fish, and this presumably is
due to the fact that invertebrates are less mobile, more
numerous, more diverse, have a shorter life cycle, and are not
as subject to extraneous factors such as fishing pressure.
Assessment
In the Pacific Northwest fish represent an important
designated use of most waters. Fish populations can be
economically and culturally important, and they often have
a high public profile. Often the most stringent constraints on
water quality stem from the need to protect coldwater
fisheries. The relative absence of certain species from a
suitable water body can be a quick and important indication
of serious impairment. However, the quantitative monitor-
ing of fish populations, although of critical importance for
fisheries management, often is of limited or uncertain value
for water quality monitoring.
The limited value of fish for monitoring stems from
their mobility, multi-year life span, ecological role, and the
numerous extraneous factors that can affect their popula-
tion. High mobility means that it is difficult to obtain an
accurate population estimate, and this limits the likelihood
of detecting a statistically significant change. Their multi-
year life span may be an advantage in that the number offish
in a certain age group or size class integrates past conditions,
but it also is a disadvantage because the number of fish may
not provide useful data on current conditions. The position
offish at the top of the food web means that they are affected
by any fluctuation at other trophic levels, and this may make
it difficult to identify the cause of an observed change.
Similarly, interspecific competition is often very important,
and this may require an entire set of species to be monitored
rather than a single population. Predation is particularly
important for alevins and juveniles.
Finally, fish populations can be affected by a wide range
of factors unrelated to forest activities, and these greatly
complicate any postulated links between fish populations and
management Fishing pressure, disease, hatchery releases,
flow conditions, and other factors can affect anadromous and
resident fish populations. Anadromous fish populations also
are a function of growth and survival rates in the ocean. The
inability to accurately estimate the marine mortality of a
particular run or population makes it very difficult to relate the
returning run size to basin-wide water quality.
Given the numerous factors affecting fish populations
and our knowledge of the habitat requirements of many of
the most important fish species, it often will be most cost-
effective to directly monitor selected habitat parameters and
then assume that these will affect fish populations. In
many cases, however, fish population data will be needed
for stockmanagementandotherpurposes, and unavailability
of such data must be considered when designing a water
quality monitoring program.
-------
REFERENCES: PART II
Adams, J.M., and R.L. Beschta, 1980. Gravel bed composition in
Oregon coastal streams. Can. J. Fish. Aquat. Sci. 37:1514-
1521.
Adams, T..N., and K. Sullivan, 1988. The physics of forest stream
heating: A simple model. Weyerhaeuser Technical Report.
Adams, P.W., and W.R. Stack, 1989. Streamwater quality after
logging in southwest Oregon. USDAFor. Serv., Proj. CompleL
Rep. (Suppl. no. PNW 87-400).
AFS, undated. Aquatic habitat inventory: glossary and standard
methods. Habitat Inventory Committee, Western Division,
American Fisheries Society. 24 p.
Alabaster, J.S. (ed.), 1977. Biological monitoring of inland
fisheries. Applied Science Publishers, London. 226 p.
Alderdicc, D.F., and F.P.J. Velsen, 1978. Relation between
temperature and incubation time for eggs of chinook salmon
(Oncorhynchustshawytscha). J. Fish. Res. Board Can. 35:69-
75.
Anderson, H.W., 1963. Managing California's snow zone lands
for water. USDA Forest Service Res. Pap. PSW-6. 28 p.
Angermeier, P.L., and J.R. Karr, 1986. Applying an index of biotic
integrity based on stream-fish communities: considerations
in sampling and interpretation. N. Am. J. Fish. Man. 6:418-
429.
APHA, 1976. Standardmethodsfortheexaminationofwaterand
wastcwatcr. AmericanPublicHealthAssociation. Washing-
ton, D.C. 1193 p.
APHA, 1980. Standard methods for the examination of water and
wastcwatcr, 15th ed. American Public Health Association.
Washington, D.C. 1134 p.
APHA, 1989. Standard methods for the examination of water and
wastewater, 17th ed. American Public Health Association.
Washington, D.C. 1460 p. -
Aumen, N.G., TJ. Grizzard, and R.H. Hawkins, 1989. Water
quality monitoring in the Bull Run watershed, Oregon. Task
Force Final Rep. to City of Portland, Oregon, Bureau of
Waterworks. 108 p.
Baker, C.O., 1979. The impacts of logjam removal on fish
populations and stream habitat in western Oregon. Ph.D.
thesis, Oregon State University. Corvallis, OR. 86 p.
Barfield, B.J., R.C. Warner, and C.T. Haan, 1981. Applied
hydrology and sedimentology for disturbed areas. Oklahoma
Technical Press. Stillwater, OK. 603 p.
Bates, C.G., and A.J. Henry, 1928. Forest and stream-flow
experiment at Wagon Wheel Gap, Colorado. Monthly Weather
Rev., Suppl. no. 3. U.S. Dep. Agric. Weather Bur., Wash-
ington, D.C. 79 p.
Behnke, R.J., and M. Zarn, 1976. Biology and management of
threatened and endangered western trouts. USDA For. Serv.,
Gen. Tech. Rep. RM-28. Fort Collins, CO. 45 p.
Berman,C.H.andT.P.Quinn,1990.Theeffectofelevatedholding
temperatures on adult spring chinook salmon reproductive
success. Report to TFW Cooperative Monitoring, Evaluation
and Research Committee. March 1990.
Berris, S.N., and R.D. Harr, 1987. Comparative snow accumula-
tion and melt during rainfall in forested and clear-cut plots in
the Western Cascades of Oregon. Water Resour. Res.
23(1): 135-142.
Beschta, R.L., 1978. Long-term patterns of sediment production
following road construction and logging hi the Oregon Coast
Range. Wat. Resourc. Res. 14:1011-1016.
Beschta, R.L., 1979. Debris removal and its effects on sedimen-
tation in an Oregon Coast Range stream. Northw. Sci. 53:71-
77.
Beschta, R.L., 1981. Patterns of sediment and organic matter
transport in Oregon Coast Range streams. Pages 179-188 in
Erosion and Sediment Transport in Pacific Rim Steeplands.
IAHS Publ. No. 132.
Beschta, R.L., R.E. Bilby, G.W. Brown, L.B. Holtby, and T.D.
Hofstra, 1987. Stream temperature and aquatic habitat:
fisheries and forestry interactions. Pages 191-232»«E.O.Salo
and T.W. Cundy (eds.), Streamside Management: Forestry
and Fishery Interactions. Contr. no. 57, List. Forest Re-
sources, Univ. Washington. Seattle, WA.
Beschta, R.L., and W.L. Jackson, 1979. The intrusion of fine
sediments into a stable gravel bed. J. Fish. Res. Board Can.
36(2):204-210.
Beschta, R.L., and W.L. Jackson, 1980. Turbidity and suspended
sediment relationship. Pages 271-282 in Proceedings,
Symposium on Watershed Management. Amer. Soc. Civ.
Engineers.
Beschta, R.L., and W.S. Platts, 1986. Morphological features of
smallstreams: significance and function. Wat. Resour. Bull.
22(3):369-379.
-------
REFERENCES
Sevan, D.E., 1961. Variability in aerial counts of spawning
salmon. J. Fish. Res. Board Can. 18:337-348.
Bilby, R.E., 1981. Role of organic debris dams in regulating the
export of dissolved and paniculate matter from a forested
watershed. Ecology 62:1234-1243.
Bilby, R.E., 1984a. Characteristics and frequency of cool-water
areas in a western Washington stream. J. Freshwater Ecpl.
2:593-602.
Bilby, R.E., 1984b. Post-logging removal of woody debris affects
stream channel stability. J. Forestry 82:609-613.
Bilby, R.E., 1988. Interactions between aquatic and terrestrial
systems. Pages 13-30 in K.J. Raedeke (ed.), Streamside
management: riparian wildlife and forestry interaction. Con-
tribution No. 59, Institute of Forest Resources, University of
Washington. Seattle, WA.
Bilby and Ward, 1987. Changes in large organic debris character-
istics and function with increasing stream size in western
Washington. Weyerhaeuser Co. Tech. Rep.
Bilby, R.E., and G.E. Likens, 1980. Importance of organic debris
dams in the structure and function of stream ecosystems.
Ecology 61:1107-1113.
Bilby, R.E., and L.J. Wasserman, 1989. Forest practices and
riparian management in Washington State: data based regu-
lation development. Pages 87-94 in Gresswell, R.E., et al.
(eds.), Riparian Resource Management, U.S. Bureau of Land
Management. Billings, MT.
Bisson, P.A., 1982. Annual fertilization and water quality—final
report. Weyerhaeuser Company, Tech. Rep. 050-5411-02.
Centralia, WA. 32 p.
Bisson, P.A., 1988. Final report: 1988 Weyerhaeuser forest
fertilization/water quality monitoring studies: Forks Creek,
Ryderwood, Silver Lake. Weyerhaeuser Company Technol-
ogy Center. Tacoma, WA. 26 p.
Bisson, P.A., J.L. Nielsen, R.A. Palmason, and L.E. Grove, 1982.
A system of naming habitat types hi small streams, with
examples of habitat utilization by salmonids during low
streamflows. Pages 62-73 in Acquisition and Utilization of
Aquatic Habitat Inventory Information, Western Division.
Am. Fish. Soc.
Bisson, P.A., M.D. Bryant, C. A. Dolloff, G.B. Grette, R.A. House,
M.L. Murphy, K V. Koski, and J.R. Sedell, 1987. Large
woody debris hi forested streams in the Pacific Northwest:
past, present and future. Pages 143-190 in E.O. Salo andT.W.
Cundy (eds.), StreamsideManagement: Forestry and Fishery
Interactions. Contr. no. 57, List Forest Resources, Univ.
. Washington. Seattle, WA.
Booth, D.B., 1990. Stream-channel incision following drainage-
basin urbanization. Wat. Resourc. Bull. 26(3):407-417.
Bosch, J.M., and J.D. Hewlett, 1982. A review of catchment
experiments to determine the effect of vegetation changes on
water yield and evapotranspiration. J. Hydro. 55:3-23.
Bothwell, M.L., and J.G. Stockner, 1980. Influence of second-
arily-treated kraft mill effluent on the accumulation rate of
attached algae in experimental continuous-flow troughs. Can
J. Fish. Aquat. Sci. 37:248-254.
Bovee, K.B., 1982. A guide to stream habitat analysis using the
instream flow incremental methodology. Instream Flow
Information Paper 12. U.S. Fish Wildl. Serv., Office of
Biological Services. 248 p.
Brady.N.C., 1974. The nature and properties of soils. MacMillan
Publishing Co., tic. New York, NY. 639 p.
Bradley, C., and D.G. Smith, 1984. Meandering channel response
to altered flow regime: Milk River, Alberta and Montana.
Wat Resourc. Res. 20(12):1913-1920.
Brown, G.W., and J.T. Krygier, 1971. Clearcut logging and
sediment production in the Oregon Coast Range. Wat.
Resourc. Res. 7:1189-1198.
Brown, G.W., 1983. Forestry and Water Quality. Oregon State
Univ. Book.Stores, Lie. Corvallis, OR. 142 p.
Brown, D.J.A., 1988. Effect of atmospheric nitrogen deposition
on the surface chemistry and the implications for fisheries.
Environ. Pollut 54:275-284.
Brown, G.W., A.R. Gahler, and R.B. Marston, 1973. Nutrient
losses after clear-cut logging and slash burning in the Oregon
Coast Range. Wat. Resourc. Res. 9(5):1450-1453.
Bryant, M.D., 1983. The role and management of woody debris in
west coast salmonid nursery streams. N. Am. J. Fish. Mgmt.
3:322-330.
Buchanan, T.J., andW.P. Somers, 1969. Discharge measurements
at gaging stations. U.S.G.S., U.S. Dep.Interior.BookS, Chap.
A8, Applications of Hydraulics, TWRI. U.S. GPO. 65 p.
Burns, D.C., and R.D. Ries. 1989. Embeddedness of salmonid
habitat of selected streams on the Payette National Forest,
1987-1988. For the Payette National Forest. 36 p.
Burns, D.C., and R.E. Edwards. 1985. Embeddedness of salmonid
habitat of selected streams on the Payette National Forest.
For the Payette National Forest. McCall, ID. 30 p.
Burns, D.C., and R.E. Edwards, 1987. Embeddedness of salmonid
habitat of selected streams on the Payette National Forest,
1985-1986. U.S. Forest Service, Payette National Forest.
McCall, ID. 20 p.
Busch, D.E., 1978. Successional changes associated with benthic
assemblages in experimental streams. Ph.D. thesis, Oregon
State University. Corvallis, OR. 91 p.
Cairns, J., Jr., K.L. Dickson, and G. Lanza, 1973. Rapid biological
monitoring system for determining aquatic community struc-
ture in receiving systems. Pages 148-163 in J. Cairns, Jr., and
K.L. Dickson (eds.), Biological Methods for the Assessment
of Water Quality. Amer. Soc. Test Mater., ASTM Special
Technical Publication 528. Philadelphia, PA.
Campbell, A.J. and R.C. Sidle, 1985. Bedload transport in a pool-
riffle sequence of a coastal Alaska stream. Wat. Resourc.
Bull. 21(4):579-590.
Cattaneo, A., 1987. Periphy ton on lakes of different trophy. Can.
J. Fish. Aquat Sci. 44:296-303.
Cattaneo, A., and J. Kalff, 1980. The relative contribution of
aquatic macrophytes and their epiphytes to the production of
macrophyte beds. Limnol. Oceanogr. 25:280-289.
Cederholm, C.J., L.M. Reid, and E.O. Salo, 1981. Cumulative
effects of logging road sediment on salmonid populations in
the Clearwater River, Jefferson County, Washington. Pages
39-74 in Proceedings from the Conference, Salmon-Spawn-
ing Gravel: A Renewable Resource in the Pacific Northwest.
State of Washington Wat. Res. Ctr., Rep. 39. Pullman, WA.
Cederholm, C.J., and L.M. Reid, 1987. Impacts of forest manage-
ment on coho salmon (Oncorhynchus kisutch) populations of
the Clearwater River, Washington: aprojectsummary. Pages
373-398 in E.O. Salo and T.W. Cundy (eds.), Streamside
Management: Forestry and Fishery Interactions. Contr.no.
57, List. Forest Resources, Univ. Washington. Seattle, WA.
Chamberlin, T.W. (ed.), 1988. Proceedings of the Workshop:
Applying 15 Years of Carnation Creek Results, Nanaimo,
British Columbia. Jan. 13-15,1987. Carnation Creek Steer-
ing Committee, c/o Pac. Biol. Sta., Nanaimo, B.C. 239 p.
Chang, H.H., 1988. Fluvial Processes in River Engineering. J.
Wiley & Sons.
-------
Part II
Chapman, D.W., and K.P.McLeod, 1987. Development of criteria
for fine sediment in the Northern Rockies ecoregion. U.S.
Environmental Protection Agency, Water Div., 910/9-87-
162. Seattle, WA. 279 p.
Cheng, J.D., T.A. Black, J. DeVries, R.P. Willington, and B.C.
Goodcll, 1975. The evaluation of initial changes in peak
streamflow following logging of a watershed on the west
coast of Canada. Int. Hydrol. Sci. Publ. 117:475-486.
Cherry, J., and R.L. Beschta, 1989. Coarse woody debris and
channelgeomorphology: aflumestudy. Wat.Resourc.Bull.
25(6):1031-1036.
Clark, J.R., K.L. Dickson, and J. Cairns, Jr., 1979. Estimating
Aufwuchs biomass. Pages 116-141 in R.L. Weitzel (ed.),
Methods and Measurements of Periphyton Communities: A
Review. Amer. Soc. Test Mater., ASTM Spec. Tech. Publ.
690. Philadelphia, PA.
Clifton, C., 1989. Effects of vegetation and land use on channel
morphology. Pages 121-129 inR.E., Gresswell, et al. (eds.),
Riparian Resource Management. U.S. Bureau of Land Man-
agement. Billings, MT
Cline, C., 1973. The effects of forest fertilization on the Tahuya
River, Kitsap Peninsula, Washington. Washington State
Dept. Ecology. 55 p.
Cohn, T.A., L.W. DeLong, E.J. Gilroy, R.M. Hirsch, and D.K.
Wells, 1989. Estimating constituent loads. Wat. Resourc.
Res. 25(5):937-942.
Cole, D.W., 1979. Mineral cycling in the forest ecosystems of the
Pacific Northwest. Pages 29-36 in S.Pi'Gessel et al. (eds.),
ForestFertilizationConference,CollegeofForestResources,
Univ. Washington. Seattle, WA.
Combs.B.D., 1965. Effects of temperature on the development of
salmon eggs. Prog. Fish-Cult. 27:134-137.
Conlin,K.,andB.D.Tutty, 1979. Juvenile salmonid field trapping
manual. Dept, of Fisheries and Oceans, Fisheries andMarine
ScrviceManuscriptReport#1530. Vancouver.B.C., Canada.
136 p.
Cook, C.W., and J, Stubbendieck (eds.), 1986. Range research:
basic problems and techniques. Soc. Range Mgmt. Denver,
CO. 317 p.
Cooper, A.C., 1965. Theeffectof transported stream sediments on
survival of sockeyeandpinksalmoneggs andalevin. Int. Pac.
Sal. Fish. Comm. Bull. 18. 71 p.
Coutant, C.C., 1969. Behavior of a sonic-tagged chinook salmon
and stcelhead trout migrating past Hanford thermal dis-
charges. Pages 21-26 in Biological effects of thermal dis-
charges: annual progress report for 1968. Battelle Memorial
Institute Pacific Northwest Laboratories, AEC Research and
Development, Rep. BNWL-1050. Richland, WA.
Gulp, J.M., 1988. The effect of streambank clearcutting on the
bcnthic invertebrates of Carnation Creek, British Columbia.
Pages 75-80 in T.W. Chamberlin (ed.), Proceedings of the
Workshop: Applying 15 Years of Carnation Creek Results,
Nanaimo, British Columbia. Jan. 13-15, 1987. Carnation
CreckSteering Committee, c/o Pac.Biol. Sta. Nanaimo, B.C.
Cupp,C.E., 1989. Streamcorridor classification for forested lands
of Washington. Hosey and Assoc. Bellevue, WA. 46 p.
Cummins, K.W., 1973. Trophic relations of aquatic insects.
Annu. Rev. Entomol. 18:183-206.
Dcusen, M.S. and P.W. Adams, 1989. Riparian areas: fish and
wildlife havens. Woodland Fish & Wildl., June. 8 p.
Dietrich, W.E., T.D. Dunne, N.F. Humphrey, and L.M. Reid,
1982. Construction of sediment budgets for drainage basins.
Pages 5-23 in F.J. Swanson, R.J. Janda, T.Dunne, and D.N.
Swanston (eds.), Sediment Budgets and Routing in Forested
Drainage Basins. USDAFor. Serv., Pac. Northw. For. Range
Exper. Sta., Gen. Tech. Rep. PNW-141.
Dietrich, W.E., J.W. Kirchner, H. Ikeda and F. Iseya, 1989.
Sediment supply and the development of the coarse surface
layer in gravel-bedded rivers. Nature 340:215-217.
Doelle, H.W., 1975. Bacterial metabolism. Academic Press,
738 p.
Dolloff, C.A., 1986. Effects of stream cleaning on juvenile coho
salmon and Dolly Varden in southeast Alaska. Trans. Am.
Fish. Soc. 115:743-755.
DuFour, A.P., 1982. Fresh recreational water quality and swim-
ming-associated illness. Second Nat. Symp. on Wastewater
Disinfection. Orlando, FL.
Duncan, H., J.W. Ward, and R.J. Anderson, 1987. A method for
assessing landslide potential as an aid in forest road place-
ment. Northw. Sci. 61(3).
Dunne, T., and L.B. Leopold, 1978. Water in environmen-
tal planning. W.H. Freeman and Co. San Francisco, CA.
818 p.
Eastman, R.H., 1970. General chemistry: experiment and theory.
Holt, Rinehart, and Winston, Inc. San Francisco, CA.
Elliott, J.M., 1977. Some methods for the statistical analysis of
. samples of benthic invertebrates. Freshwater Biol. Assoc.,
Sci. Publ. No. 25. Ambleside, England. 156 p.
Elliott, S.T., 1986. Reduction of a Dolly Varden population and
macrobenthos after removal of logging debris. Trans. Am.
Fish. Soc. 115:392-400.
Emmett, W.M., 1980. A field calibration of sediment-trapping
characteristics of the Helley-Smith bedload sampler. U.S.
Geological Survey, Prof. Pap. 1139. 44 p.
EPA, 1977. Silvicultural chemicals and protection of water
quality. U.S. Environmental Protection Agency. Seattle,
WA. 224 p.
EPA, 1978. Microbiological methods for monitoring the environ-
ment. U.S. Environmental Protection Agency, EPA-600/8-
78-017. Cincinnati, OH.
EPA, 1986a. Ambient water quality criteria for dissolved oxygen.
U.S. Environmental Protection Agency, Off. Water Regula-
tions and Standards. Washington, D.C. 46 p.
EPA, 1986b. Qualitycriteriaforwater: 1986. U.S. Environmental
Protection Agency, Off. Water Regulations and Standards.
Washington, D.C.
EPA, 1987. Surface water monitoring: a framework for change.
U.S. Environmental Protection Agency, Off. Water Regula-
tions and Standards. Washington, D.C. 41 p.
EPA, 1988a. Dissolved oxygen: water quality standards criteria
summaries. U.S. Environmental Protection Agency, Off.
Water Regulations and Standards. Washington, D.C. 34 p.
EPA, 1988b. Introduction to water quality standards. U.S.
Environmental Protection Agency, Off. Water Regulations
and Standards, EPA-440/5-88-089. Washington, D.C. 29 p.
EPA, 1989. Rapid bioassessment protocols for use in streams and
rivers: benthic macroinvertebrates and fish. U.S. Environ-
mental Protection Agency, Off. Water Regulations and Stan-
dards, EPA/444/4-89-001. Washington, D.C.
EPA, 1990. Biological criteria. U.S. Environmental Protection
Agency, Off. Water Regulations and Standards, EPA-440/5-
90-004. Washington, D.C. 41 p.
Erman, D.C., J.D. Newbold, and K.B. Roby, 1977. Evaluation of
streamside bufferstrips for protecting aquatic organisms.
California Water Resources Center Contribution no. 165.
Univ. Calif. Davis, CA, 448 p.
-------
REFERENCES
Everest, F.H., 1987. Salmonids of Western forested watersheds.
Pages 3-8 in E.O. Salo and T.W. Cundy (eds.), Streamside
Management: Forestry and Fishery Interactions. Contr.no.
57, Inst. Forest Resources, Univ. Washington. Seattle, WA.
Everest, F.H., N.B. Armantrout, S.M. Keller, W.D. Parante, J.R.
Sedell, T.N. Nickelson, JJvl. Johnson, and G.N. Haugen,
1985. Sahnonids. Pages 199-230 in E.R. Brown (ed.),
Management of wildlife and fish habitats in forests of West-
ern Oregon and Washington. USDA Forest Service, PNW
Region. Portland, OR.
Everest, F.H., R.L. Beschta, J.C. Scrivener, KV. Koski, J.R.
Sedell, and C.J. Cederholm, 1987. Fine sediment and salmo-
nid production: a paradox. Pages 98-142 in E.O. Salo and
T.W. Cundy (eds.), Streamside Management: Forestry and
Fishery Interactions. Contr. no. 57, Inst. Forest Resources,
Univ. Washington. Seattle, WA.
Federal Register, 1989. National primary drinking water regula-
tions: analytic techniques; coliform bacteria. Federal Regis-
ter 54(135): 29998-30002.
Feller, M.C., and J.P. Kimrnins, 1979. Chemical characteristics of
small streams near Haney in SouthwesternBritishColumbia.
Wat. Resourc. Res. 15(2):247-258.
Ferguson, R.I., 1986. River loads underestimated by rating curves.
Wat. Resourc. Res. 22(l):74-76.
Foster, I.D.L., J.A. Bearing, R. Grew, and K. Orend, 1990. The
sedimentary data base: an appraisal of lake and reservoir
sediment based studies of sediment yield. Pages 19-43 in
D.E. Walling, A. Yair, and S. Berkowicz (eds.), Erosion,
Transport and Deposition Processes, Proceedings of the
Jerusalem Workshop. March-April 1987. IAHS Publ. no.
189.
Fowler, W.B., T.D. Anderson, and J.D. Helvey, 1988. Changes in
water quality and climate after forest harvest in central
Washington State. USDA For. Serv., Pac. Northw. Res. Sta.
Res. Pap. PNW-RP-388. 12 p.
Fredriksen, R.L., 1971. Comparative chemical water quality—
natural and disturbed streams following logging and slash
burning. Pages 125-137 in J.T. Krygier and J.D. Hall (Di-
rectors), Forest Land Uses and Stream Environment. Oregon
State Univ. Corvallis, OR.
Fredriksen, R.L., D.G. Moore, and L.A. Norris, 1975. The impact
of timber harvest, fertilization and herbicide treatments on
streamwater quality in western Oregon and Washington.
Pages 283-313 in B. Bernier and C,H. Winget (eds.), Forest
Soils and Forest Land Management. Laval University Press.
Quebec, Canada.
Frissell, C.A., W.J. Liss, C.E. Warren, aridM.D. Hurley, 1986. A
hierarchical framework for stream habitat classification:
viewing streams in a watershed context. Environ. Mgmt
10(2): 199-214.
Froelich, H., D. McGreer, and J.R. Sedell. 1972. Natural debris
within the stream environment. U.S. Internal. Biol. Prog.,
Coniferous Forest Biome Rep. no. 96. 10 p.
Gammon, J.R., 1980. The use of community parameters derived
from electrofishing catches of river fish as indicators of
environmental quality. Pages 335-363 in Seminar on Water
• Quality Management Trade-offs. U.S. Environmental Pro-
tection Agency EPA 905/9-80-009. Washington, D.C.
Garde, R.J., and K.G. Ranga Raju, 1985. Mechanics of sediment
transportation and alluvial stream problems. Wiley Eastern
Ltd. New Delhi. 618 p.
Gessel, S.P., B.C. Steinbrenner, andR.E. Miller, 1979. Response
of Northwest forests to elements other than nitrogen. Pages
29-36 in S.P. Gessel et al. (eds.), Forest Fertilization Confer-
ence. College of Forest Resources, University of Washing-
ton.
Gibbons, D.R., W.R. Meehan, K.V. Koski, andT.R. Merrell, Jr.,
1987. History of fisheries and forestry interactions in South-
eastern Alaska. Pages 297-329 in E.O. Salo andT.W. Cundy
(eds.), Streamside Management: Forestry and Fishery Inter-
actions. Contr. no. 57, Inst. Forest Resources, Univ. Wash-
ington. Seattle, WA.
Gifford, G.F., 1981. Watershed responses to grazing manage-
ment Pages 147-159 in D.M. Baumgartner (ed.), Interior
West WatershedManagement, Proceedings of a Symposium,
Spokane, WA. April 8-10,1980. Publ. by WSU Coop. Ext.
Pullman, WA.
Goldman, C.R., and E.R. Byron, 1986. Changing water quality at
Lake Tahoe: the first fiveyearsof the Lake Tahoelhteragency
Monitoring Program. Tahoe Research Group, Institute of
Ecology, Univ. California. Davis, CA. 12 p.
Grant, G.E., 1986. Downstream effects of timber harvest activities
on the channel and valley floor morphology of western
Cascade streams. Ph.D. dissertation, John Hopkins Univ.
364p.
Grant, G.E., 1987. Assessing effects of peak flow increases on
stream channels—a rational approach. Pages 142-149 in
Proceedings of the California Watershed Management Con-
ference, Rep. no. 11, Wildland Resources Center, Univ.
California. Berkeley, CA.
Grant, G.E., 1988. The RAPID technique: a new method for
evaluating downstream effects of forest practices on riparian
zones. USDA Forest Service, Gen. Tech. Rep. PNW-220.
36 p.
Green, D.M., and J.E. Kauffman, 1989. Nutrient cycling at the
land-water interface: the importance of the riparian zone.
Pages 61-68 i'nR.E. Gresswelletal. (eds.), Riparian Resource
Management. U.S. Bureau Land Mgmt. Billings, MT.
Greeson, P.E., et al. (eds.), 1977. Methods for collection and
analysis of biological and microbiological samples: tech-
niques of water resources investigations. U.S. Geological
Survey. Washington, D.C. 332 p.
Gregory, S.V., 1980. Effects of light, nutrients, and grazing on
periphyton communities in streams. Ph.D. thesis, Oregon
State University. Corvallis, OR. 151 p.
Gregory, S.V., G.A. Lamberti, D.C. Erman, KV. Koski, M.L.
Murphy, and J.R. Sedell, 1987. Influence of forest practices
on aquatic production. Pages 233-255 in E.O. Salo andT.W.
Cundy (eds.), Streamside Management: Forestry and Fishery
Interactions. Contr. no. 57, IhsL Forest Resources, Univ.
Washington. Seattle, WA.
Gresswell, R.E., B.A. Barton, and J.L. Kershner, eds., 1989.
Practical approach to riparian resource management: an edu-
cational workshop. Billing, Montana, May 1989. U.S. Bur.
LandMgmt.BLM-MT-PT-89-001-4351. Bfflings,MT. 193p.
Grette, G.B., 1985. The role of large organic debris in juvenile
salmonid rearing habitat in small streams. M.S. thesis, Univ.
Washington. 105 p. ;
Grier, C.C., 1975. Wildfire effects on nutrient distribution and
leaching in a coniferous ecosystem. Can. J. For. Res. 5:599-
607.
Griggs,G., 1988. Impact of the January 1982 flood in Santa Cruz
County. Pages 205-228 in Landslides, Floods, and Marine
Effects of the Storm of Jan. 3-5,1982, in the San Francisco
Bay Region, California. U.S. Geol. Surv., Prof. Pap. 1434.
Washington, D.C.
-------
Part II
Guy, H.P., 1970. Fluvial sediment concepts. Chapter Cl in
Techniques of Water-Resources Investigations of the United
States Geological Survey. Applications of Hydraulics, Book
3. US. GPO. Washington, D.C. 55 p.
Guy, H.P., and V.W. Norman, 1970. Field methods for measure-
ment of fluvial sediment. Chapter C2 in Techniques of Wa-
ter-Resources Investigations of the United States Geological
Survey. Applications of Hydraulics, Book 3. U.S. GPO.
Washington, D.C. 59 p.
Hach, C.C., 1972. Understanding turbidity measurement Indus-
trial Wat. Engineer.:18-22.
Hack, J.T., 1957. Studies of longitudinal profiles in Virginia and
Maryland, U.S. Geol. Survey, Prof. Paper 294-B:45-97.
Hackley, P.R., 1989. Riparian vegetation, streambank stability,
and land useintheSalmonRiver Drainage, Idaho. Pages 181-
182 in R.E. Grcsswell et al. (eds.). Riparian Resource Man-
agement U.S. Bureau Land Mgmt Billings, MT.
Hall, J.D., and R.L. Lantz, 1969. Effects of logging on the habitat
of coho salmon and cutthroat trout in coastal streams. Pages
355-375 irtT.G. Northcote (ed.), Symposium on Salmon and
Trout in Streams. H.R. MacMillan Lectures in Fisheries,
Univ. British Columbia.
Hall, J.D., G.W. Brown, and R.L. Lantz, 1987. The Alsea
watershed study: a retrospective. Pages 399-416 in E.O. Salo
and T.W. Cundy (eds.). Streams id e Management: Forestry
and Fishery Interactions. Contr. no. 57, List. Forest Re-
sources, Univ. Washington. Seattle, WA.
Hankin, D.G., 1984. Multistage sampling designs in fisheries
research: applications in small streams. Can. J. Fish. Aquat.
Sci. 41:1575-1591.
Hankin, D.G., and G.H. Reeves, 1988. Estimating total fish
abundance and total habitat area in small streams based on
visual estimation methods. Can. J. Fish Aquat Sci. 45:834-
844.
Hansmann, E.W., and H.K. Phinney, 1973. Effects of logging on
pcriphyton in coastal streams of Oregon. Ecology 54:194-
199.
Harmon, M.E., J.F. Franklin, FJ. Swanson, and others, 1986.
Ecology of coarse woody debris in temperate ecosystems.
Pages 133-202 in A. Macfadyen, and E.D. Ford (eds.), Ad-
vances in Ecological Research: Volume 15. Academic Press.
New York, NY.
Harr, R.D., 1980. Streamflow after patch logging in small drain-
ages within the Bull Run Municipal Watershed, Oregon.
USDAFor.Serv.Res.Pap.PNW-268. 16 p.
Harr, R.D., 1983. Potential for augmenting water yield through
forest practices in western Washington and western Oregon.
Wat Resourc. Bull. 19(3):383-393.
Harr, R.D., 1989. Cumulative effects of timber harvest on
strcamflows. Paper presented at Soc. Am. Foresters 1989
Annual Convention. Spokane, WA. 24 p.
Harr, R.D., A. Levno, and R. Mersereau, 1982. Streamflow
changes after logging 130-year-old Douglas fir in two small
watersheds. Wat Resourc. Res. 18(3):637-644.
Harr, R.D., and R.L.Fredriksen, 1988. Water quality after logging
small watersheds within the Bull Run Watershed, Oregon.
Wat Resourc. Bull. 24(5):1103-1111.
Harr, R.D., R.L. Fredriksen, and J. Rothacher, 1979. Changes in
Streamflow following timber harvest in Southwestern Or-
egon. USDAFor.Serv.Pap.PNW-249. 22 p.
Harr, R.D., and F.M. McCorison, 1979. Initial effects of clearcut
logging on size and timing of peak flows in a small watershed
in Western Oregon. Wat Resourc. Res. 15(l):90-94.
Harr, R.D..W.C. Harper, and J.T.Krygier, 1975. Changes in storm
hydrographs after road building and clear-cutting in the
Oregon Coast Range. Wat Resourc. Res. 11(3): 436-444.
Harris, T., andT. Lawrence, 1978. Environmental requirements
and pollution tolerance of Tricoptera. U.S. Environmental
Protection Agency, Environmental Support Laboratory, EP A-
600/4-78-063. Cincinnati, OH. 308 p.
Hartman, G., J.C. Scrivener, L.E., and L. Powell, 1987. Some
effects of different streamside treatments on physical condi-
tions and fish population processes in Carnation Creek, a
coastal rain forest stream in British Columbia. Pages 330-372
in E.O. Salo and T.W. Cundy (eds.), Streamside Manage-
ment Forestry and Fishery Interactions. Contr. no. 57, Inst.
Forest Resources, Univ. Washington. Seattle, WA.
Harvey, G.W., 1988. Evaluation of sediment criteria for inclusion
in Idaho water quality standards. Water Quality Bureau, Div.
Environment. Boise, ID. 18 p.
Harvey, G.W., 1989. Technicalreviewofsedimentcriteria. Water
Quality Bureau, Idaho Dept. of Health & Welfare. Boise, ID.
29 p.
Heede, B.H., 1972. Influences of a forest on the hydraulic
geometry of two mountain streams. Wat Resourc. Bull.
8(3):523-530.
Hellawell, J., 1977. Biological surveillance and water quality
monitoring. Pages 69-88 in J.S. Alabaster (ed.). Biological
Monitoring of Inland Fisheries. Applied Science Publishers
Ltd. London.
Helley.EJ., and W. Smith, 1971. Development and calibration of
pressure difference bedload sampler. Open file rep. U.S.
Geological Survey, Wat Res. Div. Menlo Park, CA. 18 p.
Hem, J.D., 1970. Study and interpretation of the chemical charac-
teristics of natural water. U.S. Geological Survey Water-
Supply Paper 1473. Washington, D.C. 363 p.
Hendricks, M.L., C.H. Hocutt, and J.R. Stauffer, Jr., 1980. Moni-
toring of fish in lotic environments. Pages 205-231 in C.H.
Hocutt, and J.R. Stauffer, Jr. (eds.), Biological Monitoring of
Fish. Lexington Books, D.C. Heath and Company. Lexing-
ton, MA.
Hester, F.E., and J.S. Dendy, 1962. A multiplate sampler for
aquatic macroinvertebrates. Trans. Am. Fish. Soc. 91:420-
421.
Hocutt, C.H., 1981. Fish as indicators of biological integrity.
Fisheries 6(6):28-31.
Hocutt, C.H., and J.R. Stauffer, Jr. (eds.), 1980. LexingtonBooks,
D.C. Heath and Co. Lexington, MA. 416 p.
Hoffman, R.J., 1986. A horizontal intergravel pipe for sampling
water quality in salmonid spawning gravel. N. Am. J. Fish
Mgmt. 6:445-448.
Holtby, L.B., 1988. Effects of logging on stream temperatures in
Carnation Creek, British Columbia, and associated impacts
on the coho salmon (Oncorhynchus kisutch). Can J. Fish.
Aquat. Sci. 45:502-515.
Holtby, L.B., and S.J. Baillie, 1989. Effects of the herbicide
roundup (glyphosphate) on periphyton in Carnation Creek,
British Columbia. Pages 224-231 in P.E. Reynolds (ed.),
Proceedings of the Reynolds Creek Herbicide Workshop,
Forest Pest Management Institute, Forestry Canada. Sault
Ste. Marie, Ontario.
Holtby, L.B., T.E. McMahon, and J.C. Scrivener, 1989. Stream
temperatures and inter-annual variability in the emigration
timing of coho salmon (Oncorhynchus kisutch) smolts and
fry and chum salmon (O. keta) fry from Carnation Creek,
British Columbia. Can. J. Fish. Aquat. Sci. 46:1396-1405.
-------
REFERENCES
Hubbard, M., and W. Peters, 1978. Environmental requirements
and pollution tolerance of Ephemeroptera. U.S. Environ-
mental Protection Agency, Environmental Support Labora-
tory, EPA-600/4-78-061. Cincinnati, OH. 461 p.
Hudon, C. and Legendre, P. 1987. The ecological implications of
growth forms in epibenthic diatoms. J. Phycol. 23:434.
Hughes, R.M., and J.R. Gammon, 1987. Longitudinal changes in
fish assemblages and water quality in the Willamette River,
Oregon. Trans. Am. Fish. Soc. 116:196-209.
Husch,B.,C.I. Miller, and T.W. Beers, 1982. Forest mensuration.
John Wiley & Sons. New York, NY. 402 p.
Hutchinson, G.E., 1975. Limnological botany: Volume in of a
treatise on limnology. John Wiley & Sons. New York, NY.
Hynes, H.B.N., 1966. The biology of polluted waters. Liverpool
Univ. Press. 202 p.
Hynes, H.B.N., 1970. The ecology of running waters. Univ.
Toronto Press. 555 p.
Ice, G., 1978. Reaeration in a turbulent stream system. Ph.D.
thesis, Dept. of Forest Engineering, Oregon State Univ.
174 p.
Ice, G.G., 1985. Catalog of landslide inventories for the North-
west. Nat. Council Paper Industry for Air and Stream
Improvement. Tech. Bull. 456. 78 p.
Ice, G.G., 1990. Technical problems associated with the use of
total maximum daily load limits for forest practices. Unpubl.
paper, Nat. Council for Air and Stream Improvement.
CorvaUis, OR. lip.
Ice, G.G., 1991. Dissolved oxygen and woody debris: detecting
sensitive foreststreams. Paper presented at ASCE Gas Transfer,
Water Surface Conference, Minneapolis, MN. 14 p.
Jackson, J.K., and V.H. Resh, 1988. Sequential decision plans in
monitoring benthic macroinvertebrates: cost savings, classi-
fication accuracy, and development of plans. Can. J. Fish.
Aquat. Sci. 45:280-286.
Jacobi, G.Z., 1978. An inexpensive circular sampler for collecting
benthic macroinvertebrates in streams. J. Arch. Hydrobiol.
83(1):126-131.
Jobson, H.E., and W.P. Carey, 1989. Interaction of fine sediment
with alluvial streambeds. Wat Resourc. Res. 25( 1): 135-140.
Karr, J.R., 1981. Assessmentofbioticintegrityusing fish commu-
nities. Fisheries 6(6): 21-27.
Kauffman, J.B., 1988. The status of riparian habitats in Pacific
Northwest forests. Pages 45-55 in KJ. Raedeke (ed.),
Streamside Management: Riparian Wildlife and Forestry
Interaction. Cont no. 59, List. Forest Resources, Univ. of
Washington. Seattle, WA.
Keeton, W.T., 1967. Biological science. W.W. Norton and Co.,
Lie. New York, NY. 955 p.
Keller, E.A., and F.J. Swanson, 1979. Effects of large organic
material on channel form and fluvial processes. Earth Surf.
Proc. 4:361-380.
Keller, H.M., F.,Forster, and P. Weibel, 1986. Factors affecting
stream water quality: results of a 15 year monitoring study in
the Swiss prealps. Pages 215-225 in Monitoring to Detect
Changes in Water Quality Series. IAHS Publ. no. 157.
Kelly, D.W., and D.H. Dettman. 1980. Relationships between
streamflow, rearing habitat, substrate conditions, and juve-
nile steelhead populations in Lagunitas Creek, Marine County,
1979. Unpub. Rep., Marine County Water District. 23 p.
Kinerson, D., 1986. Bed surface response to sediment supply.
M.S. thesis, Univ. California, Berkeley.
Kinerson, D., and W.E. Dietrich, 1989. Bed surface response to
sediment supply. Eos70(43):1121.
Klamt, R.R. 1976. The effects of coarse granitic sand on the
distribution and abundance of salmonids in the central Idaho
batholith. M.S. thesis, Univ. Idaho. Moscow, ID. 85 p.
Kramer, D. 1989. Evaluation and revision of the hoop method for
monitoring embeddedness. Unpub., 13 p.
Kunkle, S.H., and G. H. Comer, 1971. Estimating suspended
sediment concentration in streams by turbidity measure-
ments. J. Soil Wat. Conserv. 26(1):18-20.
Kunkle, S., W.S.Johnson, andM. Flora, 1987. Monitoring stream
water for land-use impacts. Wat. Resourc. Div., Nat. Park
Serv. Ft. Collins, CO.
Lammel,R.F., 1972. Natural debris and logging residue within the
stream environment. M.S. thesis, Oregon State Univ.
Corvallis, OR. 78 p.
Langford, T.E., and G. Howells, 1977. The use of biological
monitoring in the fresh-water environment by the electrical
industry in the UK. Pages 115-124 in J.S. Alabaster (ed.),
Biological Monitoring of Inland Fisheries. Applied Science
Publishers Ltd. London.
Lemmon, P.E., 1957. A new instrument for measuring forest
overstory density. J. For. 55(9):667-669.
Lenat, D.R., 1988. Water quality assessment of streams using a
qualitative collectionmethod for benthicmacro invertebrates.
J. N. Am. Bentholog. Soc. 7:222-233.
Leonard, R.L., L.A. Kaplan, J.F. Elder, R.N. Coats, and C.R.
Goldman, 1979. Nutrient transport in surf ace runoff from a
subalpine watershed, Lake Tahoe Basin, California. Ecol.
Monogr. 49:281-310.
Leopold, L.B., andT.Maddock, Jr., 1953. The hydraulic geometry
of stream channels and some physiographic implications.
U.S. Geol. Survey, Prof. Paper 252.
Leopold, L.B., M.G. Wolman, and J.P. Miller, 1964. Fluvial
processes in geomorphology. W.H. Freeman and Co., San
Francisco. 522 p.
Linsley, R.K., Jr., M.A. Kohler, andJ.L.H. Paulhus, 1982. Hydrol-
ogy for engineers. McGraw-Hill, San Francisco. 508 p.
Lisle, T.E., 1982. Effectsof aggradation and degradation onriffle-
pool morphology in natural gravel channels, Northwestern
California. Wat. Resourc. Res. 18(6):1643-1651.
Lisle, T.E., 1986. Stabilization of a gravel channel by large
streamside obstructions and bedrock bends, Jacoby Creek,
northwestern California. Geol. Soc. Am. Bull. 87:999-1011.
Lisle, T.E., 1987. Using "residual depths" to monitor pool depths
independently of discharge. USDA For. Serv., Res. Note
PSW-394. 4 p.
Lisle, T.E., 1989. Sediment transport and resulting deposition in
spawning gravels, north coastal California. Wat. Resourc.
Res. 25(6):1303-1319.
Lloyd, D.S., J.P. Koenings, and J.D. LaPerriere, 1987. Effects of
turbidity in fresh waters of Alaska. N. Am. J. Fish. Mgmt.
. 7(l):18-33.
Lotspeich, F.D., and F.H. Everest, 1981. A new method for
reporting and interpreting textural composition of spawning
gravel. USDA For. Serv., Pac. Northw. For. Range Exp. Sta.,
Res. Note PNW-369.
Lowrance, R., J.K. Sharpe, and J.M. Sheridan, 1986. Long-term
sediment deposition in the riparian zone of a coastal plain
watershed. J. Soil Wat.Cons.41(4):266-271.
Lowrance, R.R., R.L. Todd, and L.E. Asmussen, 1984. Nutrient
cycling in an agricultural watershed: phreatic movement. J.
Env. Qual. 13:22-27.
Lyford, J.H., Jr., and S.V. Gregory, 1975. The dynamics and
structure of periphyton communities in three Cascade Moun-
-------
Part II
tain streams. Verh. Int. Verein. Theor. Angew. Limnol.
19:1610-1616.
MacDonald, L.H., 1989. Cumulative watershed effects: the impli-
cations of scale. Eos70(43):1114-1115.
Mackenthun,, KM., 1973. Toward a cleaner aquatic environ-
ment. U.S. Environmental Protection Agency. Washington,
D.C.
Marston, R.A., 1982. The geomorphic significance of log steps in
forest streams. Ann. Assoc. Am. Geog. 72:99-108.
Martin, C.W., and R.D. Harr, 1989. Logging of mature Douglas-
fir in western Oregon has little effect on nutrient output
budgets. Can. J. For. Res. 19:35-43.
Mason, W.T., CJ. Weber, P.A. Lewis, and E.G. Julian, 1973.
Factors affecting the performance of basket and
mtcroinvertcbrate multiplate samplers. Freshwater Biol.
3:409-436.
McMahon, T.E. and G.F. Hartman, 1989. Influence of cover
complexity and current velocity on winter habitat use by
juvenile coho salmon (Oncorhynchus tisutch). Can. J. Fish.
Aquat.Sci. 46:1551-1557.
McNeil, WJ., and W.H. Ahnell, 1964. Success'of pink salmon
spawning relative to size of spawning bed materials. U.S.
Fish. Wild!., Spec. Sci. Rep. 469. Washington, D.C. 15 p.
Megahan, W.F., 1978. Erosion processes on steep granitic road
fills in central Idaho. Soil Sci. Soc. America J. 42(2):350-
357.
Megahan, W.F., 1982. Channel sediment storage behind obstruc-
tions in forested drainagebasins draining the granitic bedrock
of the Idaho Batholith. Pages 114-121 wF.J. Swanson, R.J.
Janda, T. Dunne, and D.N. Swanston (eds.), Sediment Bud-
gets and Routing in Forested Drainage Basins. USDA For.
Serv., Pac. Northw. For. Range Exp. Sta., Gen. Tech. Rep.
PNW-141. Portland, OR.
Megahan, W.F., and C.C. Bohn, 1989. Progressive, long-term
slope failure following road construction and logging on
noncohesive,graniticsoilsoftheIdahoBatholith. Pages501-
510 in Headwaters Hydrology (June), Am. Wat. Resourc.
Assoc.
Megahan, W.F. and W.J. Kidd, 1972. Effects of logging and
logging roads on erosion and sediment deposition from steep
terrain. J. For. 70(3):136-141.
Megahan, W.F., and R.A. Nowlin, 1976. Sediment storage in
channels draining small forested watersheds. Pages 4.115-
4.126 in Proceedings, Third Federal Interagency Sedimenta-
tion Conference, Water Resources Council. Washington,
D.C.
Megahan, W.F., N.F. Day, and T.M. Bliss, 1978. Landslide
occurrence in the western and central northern Rocky Moun-
tains physiographic province in Idaho. Pages 116-139 in
Proceedings, FifthNorth American Forest Soils Conference,
Colorado State Univ. Ft. Collins, CO.
Megahan, W.F., W.S. Platts, and B. Kulesza, 1980. Riverbed
improves over time: South Fork Salmon. Pages 380-395 in
Symposium on Watershed Management. Am. Soc. Civ. Eng.
New York, NY.
Meyer, D.F., and H.A. Martinson, 1989. Rates and processes of
channel development and recovery following the 1980 erup-
tion of Mount St. Helens, Washington. Hydrol. Sci. J.
34:115-127.
Meyer, J.L., W.H. Mclowell, T.L. Bott, J.W. Elwood, C. Ishizaki,
J.M. Melack, B.L. Peckarsky, B.J. Peterson, and P.A.
Rublee,.1988. Elemental dynamics in streams. J. N. Am.
Benthol. Soc. 7:410432.
Miller, D.L., P.M. Leonard, R.M. Hughes, J.R. Karr, P.B. Moyle,
L.H. Schrader, B.A. Thompson, R.A. Daniels, K.D. Fausch,
G.A. Fitzhugh, J.R. Gammon, D.B. Halliwell, P.L.
Angermeier.andDJ.Orth, 1988. Regional applications of an
index of biotic integrity for use in water resources manage-
ment. Fisheries 13(5): 12-20.
Minshall,G.W.,andD.A. Andrews, 1973. Structuralandtemporal
variations of the benthic macroinvertebrate community in-
habiting Mink Creek, Idaho, U.S.A., a 3rd order Rocky
Mountain stream. J. Freshwater Ecol. 1:13-26.
Minshall, G.W., S.E. Jensen, and W.S. Platt, 1989. The ecology
of stream and riparian habitats of the Great Basin Region: a
community profile. U.S. Fish. Wildl. Serv., Biol. Rep.
85(7.24). 142 p.
Mohaupt, V., 1986. Nutrient-discharge relationships in a flatland
river system and optimization of sampling. Pages 297-304 in
Monitoring to Detect Changes in Water Quality Series. IAHS
Publ. no. 157.
Morin, A., 1985. Variability of density estimates and the optimi-
zation of sampling programs for stream benthos. Can. J. Fish.
Aquat. Sci. 42:1530-1534.
Moring, J.R., 1975: The Alsea Watershed Study: Effects of
logging on the aquatic resources of three headwater streams
of the Alsea River, Oregon. Fish. Res. Rep. 9. Oregon Dep.
Fish Wildl. Portland, OR. 66 p.
Munther, G., and G. Frank. 1986. 1985 Fisheries habitat and
aquatic environment monitoring reports. U.S. For. Serv.,
Deerlodge, Lolo, and Bitterroot National Forest. U.S. For.
Serv. Missoula,MT.
Murphy, M.L., C.P. Hawkins, andN.H. Anderson, 1981., Effects
of canopy modification and accumulated sediment on stream
communities. Trans. Am. Fish. Soc. 110(4):469-478.
Naiman, R.J., D.G. Lonzarich, T.J. Beechie, and S.C. Ralph, in
press. Stream classification and the assessment of conserva-
tion potential. Conference on the Conservation and Manage-
, ment of Rivers, York, England. September, 1990.
Naiman, R.J., and J.R. Sedell, 1979. Relationships between
metabolic parameters and stream order in Oregon. Can. J.
fish. Aquat. Sci. 37:834-847.
NAS (National Academy of Sciences/National Academy of Engi-
neering), 1974. Water Quality Criteria, 1972. U.S. GPO,
Washington, D.C.
NCASI, 1984a. A guide to monitoring streamwater quality fol-
lowing forestry herbicide concentration. Nat. Council of the
Paper Industry for Air and Stream Improvement, Tech. Bull.
No. 430. New York, NY. 40 p. + appendices.
NCASI, 1984b. A study of trace-enrichment cartridges for use in
sample collection and analysis of silvicultural herbicides.
Nat. Council of the Paper Industry for Air and Stream Im-
provement, Spec. Rep. no. 84-03. 35 p.
Newbold, J.D., D.C. Erman, and K.B. Roby, 1980. Effects of
logging on macroinvertebrates in streams with and without
buffer strips. Can. J. Fish. Aquat. Sci. 37:1076-1085.
Nielsen, L.A., and D.L. Johnson (eds.), 1983. Fisheries Tech-
niques. Am. Fish. Soc. Bethesda, MD. 473 p.
Noggle, C.C., 1978. Behavioral, physiological and lethal effects
of suspended sediment on juvenile salmonids. M.S. thesis,
Univ. Washington. Seattle, WA. 87 p.
Norris, L.A., H.W. Lorz, and S.V. Gregory, 1983. Influence of
forest and rangeland management on anadromous fish habitat
in Western North America. USDA For. Serv., Gen. Tech.
Rep. PNW-149. 95 p.
-------
REFERENCES
Odum,E.P., 1971. Fundamentals of ecology. WilliamP.Saunders,
Philadelphia, PA.
Ohio EPA, 1988. Water quality inventory 305(b). Report Vol. 11,
Div. Water Qual. Monitor. Assess., Columbus, Ohio.
Owen,B.B., Jr., M. Afzal, and W.R. Cody, 1979. Distinguishing
between live and dead diatoms in periphyton communities.
Pages 70-76 in R.L. Weitzel (ed.), Methods and Measure-
ments of Periphyton Communities: A Review. Amer. Soc.
Test. Mater., ASTM Spec. Tech. Publ. 690. Philadelphia,
PA.
Parker, G. and P.C. Klingeman, 1982. On why gravel bed streams
are paved. Wat Resourc. Res. 18(5):1409-1423.
Patrick, R., 1973. Use of algae, especially diatoms, in the assess-
ment of water quality. Pages 76-95 in J. Cairns, Jr., and K.L.
Dickson (eds.), Biological Methods for the Assessment of
Water Quality. Amer. Soc. Test. Mater., ASTM Special
Technical Publication 528. Philadelphia, PA.
Peters, G.B., H.J. Dawson, B.F. Hrutfiord, and R.R. Whitney,
1976. Aqueous leachate from western red cedar: Effects on
some aquatic organisms. J. Fish. Res. Board Can. 33:2703-
2709.
Pfankuch, D., 1978. Stream reach inventory and channel stability
evaluation. USDA For. Serv., Northern Region. Missoula,
MT. 26 p.
Pielou, E.G., 1975. Ecological diversity. J. Wiley & Sons. New
York, NY. 165 p.
Phillips, R.W., R.L. Lantz, E.W. Claire, and J.R. Moring, 1975.
Some effects of gravel mixtures on emergence of coho salmon
and steelhead trout fry. Trans. Am. Fish. Soc. 104(3):461-
466.
Plamondon, A.P., A. Gonzalez, and Y. Thomassin, 1982. Pages
49-70 in Effects of logging on water quality: comparison
between two Quebec sites. Can. Hydrol. Symp.:82.
Fredericton, NB, Canada.
Platts,W.S.,1981. Effects of livestock grazing. USDAFor.Serv.,
Gen. Tech. Rep. PNW-124. Portland, OR. 25 p.
Platts, W.S., 1983. How many stream habitat evaluation systems
do we need?—less than a million. Pages 212-220 in Western
Proceedings: 63rd annual conference of the Western Asso-
ciation of Fish and Wildlife Agencies, Western Association
of Fish and Wildlife Agencies. Boise, Idaho.
Platts, W.S., 1984. Determining and evaluating riparian-stream
enhancement needs and fish response. Pages 181-190 in
Pacific Northwest stream habitat management workshop.
California Cooperative Fishery Research Unit, Humboldt
State Univ. Arcata, CA.
Platts, W.S., 1989. Compatibility of livestock grazing strategies
with fisheries. Pages 103-110 in R.E. Gresswell, et al. (eds.),
Riparian Resource Management. U.S. Bureau of Land Man-
agement. Billings, MT.
Platts, W.S., C. Armour, G.D. Booth, M. Bryant, J.L. Bufford, P.
Cuplin, S. Jensen, G.W. Lienkaemper, G.W. Minshall, S.B.
Monsen, R.C. Kelson, J.R. Sedell, and J.S. Tuhy, 1987.
Methods for evaluating riparian habitats with applications to
management. USDA For. Serv., Gen. Tech. Rep. INT-221.
Ogden,UT. 177 p.
Platts, W.S., and W.F. Megahan, 1975. Time trends in riverbed
sediment composition in salmon and steelhead spawning
areas: SouthForkSalmonRiver.Idaho. Trans.N. Am. Wildl.
Nat. Resourc. Conf. 40:229-239.
Platts,, W.S..W.F. Megahan, and G.W. Minshall, 1983. Methods
for evaluating stream, riparian, and biotic conditions. USDA
For. Serv., Gen. Tech. Rep. INT-183, 71 p.
Platts, W.S., and R.L. Nelson, 1989. Characteristics of riparian
plant communities andstreambanks withrespectto grazing in
Northeastern Utah. Pages 73-81/nR.E. Gresswell, etal.(eds.),
Riparian Resource Management U.S. Bureau of Land Man-
agement Billings, MT.
Platts, W.S., M.A. Shirazi, and D.H. Lewis, 1979. Sediment
. particle sizes used by salmon for spawning with methods for
evaluation. U.S. Environmental Protection Agency, EPA-
600/3-79-043. Washington, D.C.
Platts, W.S., R J. Torquemada, M.L. McHenry, and C.K. Graham,
1989. Changes hi salmon spawning and rearing habitat from
increased delivery of fine sediment to the South Fork Salmon
River, Idaho. Trans. Am. Fish. Soc. 118:274-283.
Porterfield,G., 1972. Computation of fluvial-sediment discharge.
Chapter C3 in Techniques of Water-Resources Investigations
of the United States Geological Survey. Applications of
Hydraulics, Book 3. U.S. GPO. Washington, D.C. 66 p.
Potyondy, J.P. 1988. BoiseNationalForestcobbleembeddedness
baseline inventory: Results and relationship to management
activities. For the Boise National Forest. 37 p.
Powell, L.H., 1988. Stream channel morphology changes since
logging. Pages 16-25 wtT.W. Chamberlin (ed.), Proceedings
of the Workshop: Applying 15 Years of Carnation Creek
Results, Nanaimo, British Columbia. Jan. 13-15, 1987.
Carnation Creek Steering Committee, c/o Pac. Biol. Sta.,
Nanaimo, B.C.
Power, M. E., R.J. Stout, C.E. Gushing, P.P. Harper, F.R. Hauer,
WJ.Matthews, P.B. Moyle, B. Statzner.I.R. Wais DeBadgen,
1988. Biotic and abiotic controls in river and stream commu-
nities. J. North Amer. Benthological Soc. 7:456-479.
Pryfogle,A.A.,andR.L.Lowe,1979. Samplingand interpretation
of epilithic lotic diatom communities. Pages 77-89 in R.L.
Weitzel (ed.), Methods and Measurements of Periphyton
Communities: A Review. Amer. Soc. Test. Mater., ASTM
Spec. Tech. Publ. 690. Philadelphia, PA.
Raedeke, KJ. (ed.), 1988. Streamside management: riparian
wildlife and forestry interactions. Contribution No. 59, List.
of Forest Resources, Univ. of Washington. Seattle, WA.
Rainville, R.P., S.C. Rainville, and E.L. Lider, 1985. Riparian
silvicultural strategies for fish habitat emphasis. Pages 186-
189 in S ilviculture for wildlife and fish: a time for leadership,
Proc., 1985 Wildlife and Fish Ecology Working Group. Soc.
Am. Foresters. Bethesda, MD.
Ralph, S.C.., 1989. Timber/Fish/Wildlife stream ambientmonitor-
ing field manual". Center for Streamside Studies, University
of Washington. Seattle, WA.
Reid, L.M. andT. Dunne, 1984. Sediment production from forest
road surfaces. Wat. Resourc. Res. 20(11):1753-1761.
Reid,I.andL.E.Frostick, 1986. Dynamics of bedload transport in
Turkey Brook, a coarse-grained alluvial channel. Earth
Surface Process. Landforms 11:143-155. Wiley & Sons, Ltd.
Reid, I., L.E. Frostick, and J.T. Layman, 1985. The incidence and
nature of bedload transport during flood flows in coarse-
grained alluvial channels. Earth Surf. Proc. Landforms
10:33-44.
Reiser, D.W., and T.C. Bjomn, 1979. Habitat requirements of
anadromous salmonids. USDA For. Serv., Gen. Tech. Rep.
PNW-96. 54 p.
Reynolds, J.B., 1983. Electrofishing. Pages 147-164 in L.A.
Nielsen, andD.L. Johnson (eds.), Fisheries Techniques. Am.
Fish. Soc. Bethesda, MD.
Reynolds, R.C.,andN.M. Johnson, 1972. Chemical weathering in
the temperate glacial environment of the northern Cascade
-------
Part II
Mountains. Geochim. Cosmochim. Acta 36:537-554.
Reynolds, P.E., ed, 1989. Proceedings of the Carnation Creek
herbicide workshop. Forest Pest Management Institute,
Forestry Canada. Sault Ste. Marie, Ontario. 349 p.
Reynolds, P.E., J.C. Scrivener, LJB. Holtby, and P.D. Kingsbury,
1989. An overview of Carnation Creek herbicide study:
historical perspective, experimental protocols, and spray
operations. Pages 15-26 in P.E. Reynolds (ed.), Proceedings
of the Reynolds Creek Herbicide Workshop, Forest Pest
Management Institute, Forestry Canada. Sault Ste. Marie,
Ontario.
Rieger, W. A., and L J. Olive, 1986. Sediment responses during
storms events in small forested watersheds. Pages 490-498 in
A.H. El-Shaarawi and R.E. Kwiatkowski (eds.), Statistical
Aspects of Water Quality Monitoring. Proc. of Workshop,
Canada Centre for Inland Waters. October 7-10, 1985.
Elsevier.
Ritter, D.F., 1978. Process Geomorphology. Wm.C. Brown,
Dubuquc, Iowa. 603 p.
Ro4gers,J.H.,Jr.,K.L.Dickson,andJ.Caims,Jr., 1979. Areview
and analysis of some methods used to measure functional
aspects of periphyton. Pages 142-167 in R.L. Weitzel (ed.),
Methods and Measurements of Periphyton Communities: A
Review. Amer. Soc. Test Mater., ASTM Spec. Tech. Publ.
690. Philadelphia, PA.
Rosenberg, DM., and V.H. Resh, 1982. The use of artificial
substrates in the study of freshwater benthic macro-
invertebrates. In J. Cairns, Jr. (ed.), Artificial Substrates.
Ann Arbor Science Publishers.Inc. Ann Arbor, MI. 279 p.
Rosgcn, D.L., 1985. A stream classification system. Pages91-95
MI R.R. Johnson, C.D. Zeibell, D.R. Patton, P.P. Ffolliott, and
R.H. Hamre (eds.). Riparian Ecosystems and their Manage-
ment: Reconciling Conflicting Uses. USDA For. Serv., Ben.
Tech. Rep. RM-120. Fort Collins, CO.
Rothacher, J., 1970. Increases in water yield following clear-cut
logging in the Pacific Northwest. Wat.Resourc.Res. 6(2):653-
658.
Ruttncr, F., 1953. Fundamentals of Limnology. University of
Toronto Press. Toronto, Canada.
Ryan, S.E., and G.E. Gran,t In press. Downstream effects of
timber harvesting on channel morphology in Elk River Basin,
Oregon. J. Environ. Qual. 20.
Salo, E.O., and T.W. Cundy (eds.), 1987. Streamside Manage-
ment Forestry and Fishery Interactions. Contr. no. 57, InsL
Forest Resources, Univ. Washington. Seattle, WA. 471 p.
Schindler, D.W., B.W. Newbury, K.G. Beaty, and P. Campbell,
1976. Natural water and chemical budgets for a small
Precambrianlake basin in central Canada. J. Fish. Res. Board
Can. 33:2526-2543.
Schumm, S. A., 1960. The shape of alluvial channels in relation to
sediment type. U.S. Geol. Survey, Prof. Paper 352-B.
Schumm, S.A., 1972. Fluvial geomorphology: channel adjust-
ment and river metamorphosis. Pages 396-417 in S.A.
Schumm (ed.), River Morphology. Benchmark Papers in
Geology. Dowden, Hutchinson & Ross, Inc. Stroudsberg,
PA.
Schumm, S.A., 1977. The fluvial system. Wiley Interscience.
New York, N.Y.
Scrivener, J.C., 1988. Changes in concentration of dissolved ions
during 16years at Carnation Creek, British Columbia. Pages
75-80 in T.W. Chamberlin (ed.), Proceedings of the Work-
shop: Applying 15 Years of Carnation Creek Results,
Nanaimo, British Columbia. Jan. 13-15, 1987. Carnation
Creek Steering Committee, c/o Pac. Biol. Sta., Nanaimo,
B.C.
Sedell, J.R., and K.J. Luchessa, 1982. Using the historical record
as an aid to salmonid habitat enhancement. Pages 210-223 in
N.B. Armantrout (ed.), Acquisition and Utilization of Aquatic
Habitat Inventory Information. Am. Fish. Soc., West. Div.
Portland, OR.
Sedell, J.R., F.J. Swanson, and S.V. Gregory, 1984. Evaluating
fish response to woody debris. Pages 222-245 in Pacific
Northwest stream habitat management workshop. California
Cooperative Fishery Research Unit, Humboldt State Univ.
Arcata, CA.
Sedell, J.R., P.A. Bisson, F.J. Swanson, and S.V. Gregory, 1988.
What we know about large trees that fall into streams and
rivers. In Maser, C., et al. (eds.), From the forest to the sea:
a story of fallen trees. USDA For. Serv., Gen. Tech. Rep.
PNW-GTR-229. 153 p.
Shen, H.W., and J.-Y. Lu, 1983. Development and prediction of
bed armoring. J. Hydraul. Eng. 109(4):611-629.
Sheridan, W.L., M.P. Perensovich, T. Paris, and K Koski, 1984.
Sediment content of streambed gravels in some pink salmon
spawning streams in Alaska. Pages 153-165 inW.R.Meehan,
T.R. Merrell, Jr., and T.A. Hanley (eds.), Fish and Wildlife
Relationships in Old-growth Forests: Proceedings of a Sym-
posium. Am.Inst. Fish. Res. Biologists. Available from J.W.
Reintjes, Rt. 4, Box 85, Morehead City, NC 28557.
Sidle, R.C., 1988. Bed load transport regime of a small forest
stream. Wat. Resourc. Res. 24(2):207-218.
Sigler, J.W., 1980. Effects of chronic turbidity on feeding, growth
and social behavior of steelhead trout andcoho salmon. Ph.D.
dissertation, Univ. Idaho. Moscow, ID. 158 p.
Sigler, J.W., T.C. Bjornn, and F.H. Everest, 1984. Effects of
chronic turbidity on density and growth of steelheads and
coho salmon. Trans. Am. Fish. Soc. 113:142-150.
Skaugset, A.E., 1980. Fine organic debris and dissolved oxygen
in streambed gravels in the Oregon Coast Range. M.S. thesis,
Oregon State Univ. Corvallis, OR.
Skaugset, A., and G. Ice, 1989. Research on dissolved oxygen in
streams. Pages C15-C30 in Abstracts to Presentations at the
1989 West Coast Regional Meeting. Nat'l Council of the
Paper Industry for Air and Stream Improvement. Corvallis,
OR.
Skille, J., and J. King, 1989. Proposed cobble embeddedness
sampling procedure. Unpublished paper available from the
USDA For. Serv., Intermount. Res. Sta. Boise, ID. lip.
Smart, P.L., and I.M.S. Laidlaw, 1977. An evaluation of some
fluorescent dyes for water tracing. Wat. Resourc. Res. 13(1).
Sollins, P., and P.M. McCorison, 1981. Nitrogen and carbon
solution chemistry of an old growth coniferous forest wa-
tershed before and after cutting. Wat. Resourc. Res. 17(5):
1409-1418.
Stednick, J.D., 1991. Wildland Water Quality Sampling and
Analysis. Academic Press, Inc.
Stockner, J.G., and K.R.S. Shortreed, 1978. Enhancement of
autotrophic production by nutrient addition in a coastal
rainforest stream on Vancouver Island. J. Fish. Res. Board
Can. 35:28-34.
Stockner, J.G., and K.S. Shortreed, 1988. The autotrophic com-
munity response to logging in Carnation Creek, British Co-
lumbia: a six year perspective. Pages 75-80 in T.W.
Chamberlin (ed.), Proceedings of the Workshop: Applying
15 Years of Carnation Creek Results, Nanaimo, British
Columbia. Jan. 13-15, 1987. Carnation Creek Steering
-------
REFERENCES
Committee, c/o Pac. Biol. Sta., Nanaimo, B.C.
Stottlemeyer, R., 1987. Natural and anthropic factors as determi-
nants of long-term streamwater chemistry. Pages 86-94 in
C.A. Troendle, M.R. Kaufmann, R.H. Hamre, and R.P.
Winokur (tech. coord.), Proceedings of a Technical Confer-
ence: Management of subalpine Forests: Building on 50
Years of Research, Silver Creek, CO. July 6-9,1987. USDA
For. Serv., Gen. Tdch. Rep. RM-149. 253 p.
Stumm, W., and J.J. Morgan, 1981. Aquatic Chemistry, 2nd ed.
Wiley.
Sullivan, D.S., T.P. Sullivan, and T. Bisalputra, 1981. Effects of
Roundup herbicide on diatom, populations in the aquatic
environment of a coastal forest. Bull. Env. Contamin. Toxic.
26:91-96.
Sullivan, K., T.E. Lisle, C. A. Dolloff, G.E. Grant, and L.M. Reid,
1987. Stream channels: the link between forests and fishes.
Pages 39-97 in E.G. Salo andT.W. Cundy (eds.), Streamside
Management: Forestry and Fishery Interactions. Contr.no.
57, List Forest Resources, Univ. Washington. Seattle, WA.
Surber, E.W., 1937. Rainbow trout and bottom fauna production
in one mile of stream. Trans. Am. Fish. Soc. 66:193-202.
Sutherland, A.J., 1987. Static armor layers by selective erosion.
Pages 243-260 in C.R. Thorne, J.C. Bathurst, and R.D. Hey
(eds.), Sediment Transport in Gravel-bed Rivers. Wiley and
Sons, NY.
Swanson, F.J., S.V. Gregory, J.R. Sedell, and A.G. Campbell,
1982. Land-water interactions: The riparian zone. Pages
267-291 in R.L. Edmonds (ed.), Analysis of Coniferous
Forest Ecosystems in the Western United States. Hutchinson
Ross Publ. Co. Stroudsburg, PA.
Swanson, R.H., and G.R. Hillman, 1977. Predicted increased
water yield after clearcutting in West-Central Alberta. North.
For. Res. Cenlie, Edmonton, Alberta. I.R. NOR-X-198. 40
P-
Swanson, F.J., and G.W. Lienkaemper, 1978. Physical conse-
quences of large organic debris in Pacific Northwest streams.
USDA For. Serv., Gen. Tech. Rep. PNW-69. Portland, OR.
15 p.
Swanson, F.J., L.E. Benda, S.H. Duncan, G.E. Grant, W.F.
Megahan, L.M.Reid, and R.R. Ziemer, 1987. Mass failures
and other processes of sediment production in Pacific North-
west forest landscapes. Pages 9-38 in E.O. Salo and T.W.
Cundy (eds.), StreamsideManagement: Forestry andFishery
Interactions. Contr. no. 57, Inst. Forest Resources, Univ.
Washington. Seattle, WA.
Tassone, B.L., 1988. Sediment loads from 1973 to 1984 08HB048
Carnation Creek at the mouth, British Columbia. Session 2:
Sediment and gravel. In T.W. Charnberlin (ed.), Proceedings
of the Workshop: Applying 15 Years of Carnation Creek
Results, Nanaimo, British Columbia. Jan. 13-15, 1987.
Carnation Creek Steering Committee, c/o Pac. Biol. Sta.,
Nanaimo, B.C.
Thomas, R.B., 1985. Measuring suspended sediment in small
mountain streams. USDA'For. Serv., Gen. Tech. Rep. PSW-
83. 9p. -
Thut, R.N., and E.P. Haydu, 1971. Effects of forest chemicals on
aquatic life. Pages 159-171 in J.T. Krygier, and J.D. Hall
(eds.), Forest Land Uses and Stream environment: Proceed-
ings of a Symposium. Oregon State Univ. Corvallis, OR.
Torquemada, R.J., and W.S. Platts. 1988. A comparison of
sediment monitoring techniques of potential use in sediment/
fishpopulationrelationships. USDA For. Serv., Intermount.
Sta., Progress Report I to Idaho Fish Game Dep. 55 p.
Triska, F.J., F.C. Kennedy, RJ. Avanzino, andB N. Reilly, 1983.
Effect of simulated canopy cover on regulation of nitrate
uptake and primary production by natural periphyton com-
munities. Pages 129-159 wtT.D. Fontaine, and S.M. Bartell
(eds.), Dynamics of Lotic Environments. Ann Arbor Science
Publishers, Inc. Ann Arbor, MI.
Triska, F.S., J.R. Sedell, K. Cromack, S.V. Gregory, and F.M.
McCorison, 1984. Nitrogen budget for a small coniferous
forest stream. Ecol.Monogr. 54:119-140.
Troendle, C.A., 1983. The potential for water yield augmentation
from forest management in the Rocky Mountain region. Wat.
Resourc. Bull. 19(3):359-373.
Troendle, C.A.,andRM. King, 1985. Theeffectof timber harvest
on the Fool Creek watershed, 30 years later. Wat. Resourc.
Res. 21(12):1915-1922.
Tueller, P.T., ed., 1988. Vegetation science applications for
rangeland analysis and management. Kluwer Academic
Publishers. Boston, MA. 642 p.
U.S. Army Corps of Engineers et al., 1989. Federal manual for
identifying and delineating jurisdictional wetlands. U.S.
Army Corps of Engineers, U.S. Environmental Protection
Agency, U.S. Fish & Wildlife Service, and U.S. DepL of
Agriculture Soil Conservation Service. Cooperative Techni-
cal Publication. 76 p.
VanDeventer, J.S., 1990. Abibliographyofriparianresearchand
management: fisheries, wildlife, vegetation, and hydrologic
responses to livestock grazing and other land uses. Draft
manuscript proposed as USDA For. Serv., Gen. Tech. Rep.,
Intermount. Sta. Boise, ID. 505 p.
VanHaveren,B., 1988. A reevaluation of the Wagon Wheel Gap
Forest watershed experiment. For. Sci. 34(1):208-214.
Van Wagner, C.E., 1968. The line intersect method in forest fuel
sampling. For. Sci. 14:20-26.
Vasconcelos, G.J., and N.C. Anthony, 1985. Microbiological
quality of recreational waters in the Pacific Northwest. J.
Wat. Poll. Control Fed. 57(5):366-377.
Vitousek, P.M., J.R. Gosz, C.C. Grier, J.M. Milillo, W. A. Reiners,
and R.L. Todd, 1979. Nitrate losses from disturbed ecosys-
tems. Science 204:469-474.
Walling, D.E., and B.W. Webb, 1982. Sediment availability and
the prediction of storm-period sediment yields. Pages 327-
337 in Recent Developments in the Explanation and Predic-
tion of Erosion and Sediment Yield. Proc. Exeter Sympo-
sium, July 1982. IAHS no. 137.
Warren, C.E., 1971. Biology and water pollution control. W.B.
Saunders Co., Philadelphia, PA, 434 p.
Waters, T.F., and R.J. Knapp, 1961. An improved bottom fauna
sampler. Trans. Am. Fish. Soc. 90:225-226.
Wetzel,R.F., 1975. Limnology. W.B. Sanders Co. Philadelphia,
PA. 743 p.
Wetzel, R.G. 1983. Attached algae-substrate interactions: fact or
myth, and when andhow? Page207-215 in R.G. Wetzel (ed.),
Periphyton in Freshwater Ecosystems. Junk, The Hague.
Weitzel, R.L., 1979. Periphyton measurements and applications.
Pages 3-33 in R.L. Weitzel (ed.), Methods andMeasurements
of Periphyton Communities: A Review. Am. Soc. Test.
Mater., ASTM Spec. Tech. Publ. 690. Philadelphia, PA.
Weitzel, R.L., S.L. Sanocki, and H. Holecek, 1979. Sample
replication of periphyton collected from artificial substrates.
Pages 9-115 in R.L. Weitzel (ed.), Methods and Measure-
ments of Periphyton Communities: A Review. Amer. Soc.
for Testing and Materials Spec. Tech. Publ. 690, Philadel-
phia, PA.
-------
Part II
Williams, OJP., 1989a. Sediment concentration versus water
discharge during single hydrologic events in rivers. J. Hy-
drology 111:89-106.
Williams, G.P., 1989b. Proportion of bedload to total sediment
load in rivers. Eos 70(43): 1106.
Wingct, R.N., and F.A. Mangum, 1979. Biotic condition index:
integrated biological, physical, and chemical stream param-
eters for management USDA For. Serv., Litermount. Reg.
Ogden, UT. 51 p.
Wissmar, R.C., M.D. Lilley and M. DeAngelis, 1987. Nitrous
oxide release from aerobic riverine sediments. J. Freshwater
Ecol.4(2):209-218.
Wissmar,R.C.,D.M.McKnight, andC.N.Dahm, 1990. Contribu-
tionof organic acids to alkalinity in lakes within theMountSt.
Helens blast zone. Limnol. Oceanogr. 35:535-542.
Wissmar, R.C., and F J. Swanson, 1990. Landscape disturbances
and lotic ccotones. Pages 65-89 in R. J. Naiman and H. De-
camps (eds.), The Ecology and Management of Aquatic-
Tenestrial Ecotones. Parthenon Press, London.
Wolman, M.G., 1954. A method of sampling coarse river-bed
material. Trans. Am. Geophys. Union 35(6):951-956.
Wotton, R.S. 1988. Dissolved organic material and trophic dy-
namics. BioScience 38:172-177.
Wright, H.E., Jr., 1981. The role of fire in land/water interactions.
Pages 421-444 in Fire Regimes and Ecosystem Properties.
USDA For. Serv;, Gen. Tech. Rep. WO-26.
Wright, K.A., K.H. Sendek, R.M. Rice, and R.B. Thomas, 1990.
Logging effects on streamflo w: storm runoff at Caspar Creek
in northwestern California. Wat. Resourc. Res. 26(7):1657-
1667.
Wringler, N.H., and J.D. Hall, 1975. Effects of logging on water
temperature and dissolved oxygen hi spawning beds. Trans.
Am. Fish. Soc. 104(1):111-121.
Zand, S.M., 1976. Indexes associated with information theory in
water quality. J. Wat. Poll. Control Fed. 48:2026-2031.
Ziemer, R.R., 1981. Roots and the stability of forested slopes.
IAHS 132:343-357.
Ziemer, R.R., 1984. Response of progressive hillslope deforma-
tion to precipitation. Pages 91-98 in Symposium on the Ef-
fects of Forest Land Use on Erosion and Slope Stability.
Environment and Policy Institute, East-West Center. Univ.
Hawaii, Honolulu.
-------
Federal Recycling Program "m *A Printed on Recycled Papet
-------
------- |