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.
-------
ABSTRACT
This document provides guidance for designing water quality monitoring projects and
selecting monitoringparameters. Although the focus is on forest management and streams in the
Pacific Northwest and Alaska, a broader perspective is taken, and much of the information is
more widely applicable.
Part I reviews the regulatory mechanisms for nonpoint source pollution and defines seven
types of monitoring. A step-by-step process for developing monitoring projects is presented.
Because monitoring is a sampling procedure, study design and statistical analysis are explicitly
addressed. The selection of monitoringparameters is defined as afunction of the designated uses,
management activities, sampling frequency, monitoring costs, access, and the physical envi-
ronment. Approximately 30 parameters are rated with regard to these controlling factors. A
qualitative combination of these rankings yields recommended monitoring parameters for
various management activities. This parameter selection process has been incorporated into an
interactive PC-based expert system called PASSSFA.
Part n is a technical review of the parameters, which are grouped into six categories: physical
and chemical constituents, flow, sediment, channel characteristics, riparian, and aquatic organisms.
The review of each parameter is organized into seven sub-sections: definition, relation to
designated uses, response to management activities, measurement concepts, standards, current
uses, and assessment.
-------
-------
CONTENTS
Page
List of Illustrations vii
Preface viii
Glossary xi
About the Authors xii
Executive Summary xiii
PART I. DEVELOPING A MONITORING PROJECT
1 INTRODUCTION
1.1 Purpose of the Guidelines 3
1.2 Organization and Use of the Guidelines 4
1.3 Types of Monitoring 6
1.4 Legal Background 8
2 CONTEXT AND STRUCTURE OF MONITORING PROJECTS
2.1 Legal Context of Water Quality Monitoring Efforts 14
2.2 Structure of a Water Quality Monitoring Project 18
3 STATISTICAL CONSIDERATIONS IN WATER QUALITY MONITORING
3.1 Relevance of Statistics to Water Quality Monitoring 22
3.2 Statistical Design in Water Quality Monitoring 23
3.2.1 General Design and Replication 23
3.2.2 Benefits of a Proper Statistical Design 26
3.2.3 Design Problems and Constraints 26
3.3 Principles of Sampling 28
3.4 Principles of Statistical Testing 29
3.4.1 Assumptions and Distributions 29
3.4.2 Statistical Compromises 31
4 PRINCIPLES OF DEVELOPING A MONITORING PLAN AND SELECTING
THE MONITORING PARAMETERS
4.1 Purpose of Monitoring 36
4.2 Designated Uses of Water 38
4.3 Type of Management Activity 40
4.4 Frequency of Monitoring 40
4.5 Cost of Monitoring 42
4.6 Access to Monitoring Sites 44
4.7 Availability of Existing Data 45
4.8 Physical Environment 45
4.8.1 Ecoregion Concept 45
4.8.2 Climatic Considerations 47
4.8.3 Land Form 47
4.8.4 Geology and Soils 48
4.8.5 Summary 48
5 PARAMETER RECOMMENDATIONS AND INTERACTIONS
5.1 Recommended Parameters 49
5.1.1 Forest harvest 52
5.1.2 Road Building and Maintenance 53
5.1.3 Forest Fertilization 54
5.1.4 Application of Herbicides and Pesticides 54
5.1.5 Grazing 54
5.1.6 Dispersed Recreation 55
5.1.7 Developed Recreation/Rural Populations 55
5.1.8 Placer Mining/Sand and Gravel Extraction 55
5.1.9 Hardrock Mining 56
5.1.10 Wildfire and Prescribed Burning 56
-------
Page
5.2 Expert System 59
5.3 Parameter Selection and Interactions 60
REFERENCES: PARTI 66
PART H. REVIEW OF MONITORING PARAMETERS
1 INTRODUCTION
1.1 Purpose and Use of Part H 71
1.2 Selection and Organization of the Parameters in Part II 72
2 PHYSICAL AND CHEMICAL CONSTITUENTS
Introduction 73
2.1 Temperature 73
2.2 pH 76
2.3 Conductivity 78
2.4 Dissolved Oxygen 80
2.5 Nutrients 83
2.5.1 Nitrogen 83
2.5.2 Phosphorus 86
2.6 Herbicides and Pesticides : 89
3 CHANGES IN FLOW
Introduction 92
3.1 Increases in the Size of PeakFlows 92
3.2 Changes in Low Flows 95
3.3 Water Yield 96
4 SEDIMENT
Introduction 98
4.1 Suspended Sediment 98
4.2 Turbidity 102
4.3 Bedload 105
5 CHANNEL CHARACTERISTICS
Introduction 109
5.1 Channel Cross-section 109
5.2 Channel Width/width-depth Ratios 111
5.3 Pool Parameters 113
5.4 Thalweg Profile 115
5.5 HabitatUnits 116
5.6 Bed Material 118
5.6.1 Particle-Size Distribution 118
5.6.2 Embeddedness 121
5.6.3 Surface vs. Subsurface Particle Size Distributions 125
5.7 Large Woody Debris 127
5.8 Bank Stability 130
6 RIPARIAN MONITORING
Introduction 133
6.1 Riparian Canopy Opening 133
6.2 Riparian Vegetation 135
7 AQUATIC ORGANISMS
Introduction 140
7.1 Bacteria 140
7.2 Aquatic Flora 143
7.3 Macroinvertebrates 147
1A Fish 151
REFERENCES: PARTE 156
VI
-------
LIST OF ILLUSTRATIONS
Box Page
1. How to use this document 5
2. Case Study: Bull Run Watershed, Oregon 9
3. Case Study: South Fork Salmon River, Idaho 19
4. Case Study: Design of a Monitoring Project on the Snohomish River, Washington 25
5. Ways to reduce the cost of a monitoring project 37
6. Case Study: Sediment Monitoring in the Little North Fork of the Clearwater River, Idaho 53
7. Case Study: Silver Fire Recovery Project, Siskiyou National Forest 57
Figure*
1. The three levels of the antidegradation policy 11
2. Ecological integrity as a function of chemical, physical, and biological integrity 12
3. Flow diagram for monitoring and controlling nonpoint sources of pollution 15
4. Development of a monitoring project 20
5. Schematic representation of the trade-offs among level of significance, power, and
variability for two normally distributed populations 33
6. Maximum allowable coefficient of variation to detect change 35
7. Ecoregions of Idaho, Oregon, and Washington 46
8. Schematic representation of the three main embeddedness measurements 120
Table*
1. General characteristics of monitoring types 8
2. Effects of water quality parameters on the major designated uses of water 39
3. Sensitivity of the water quality monitoring parameters to management activities 41
4. Frequency and cost of data or sample collection and analysis by monitoring parameter 43
5. Usefulness of the different parameters to monitor the effects of management activities
on streams in forested areas in the Pacific Northwest and Alaska 50
6. Interrelationships among the different water quality monitoring parameters 62
7. Biologic effects of various dissolved oxygen concentrations on salmonids,
non-salmonid fishes, and aquatic invertebrates 79
8. Equilibrium concentration of un-ionized ammonia in mg L-1 as a function of
temperature and pH 83
9. Classification of bed material by particle size , 118
10. Rating system for evaluating stream channel condition 130
11. Requirements for number of trees to be left along streams in western and eastern
Washington following timber harvest 136
*For clarity and simplicity, the legends listed below do not correspond exactly to the legend hi the listed figure or table. The intent
is to provide a general description and quick reference to the figures and tables in the text.
Vll
-------
PREFACE
Over the last few decades, the forest industry has come under increasing scrutiny, and this is particularly true
in thePacific Northwest. Since World War II the economy has been moving away from primary industries to service
and manaufacturing, and there has been a rapid shift to an urban-based population. This, together with a greater
awareness of environmental issues, has led to increasing public concern over the adverse effects of forest land
management on water quality and stream condition. Much of the public attention has been directed towards the
dramatic decline in the number of salmonid fishes in many of the major river basins, and the resultant economic,
cultural, and legal implications. Considerable concern also has been expressed over the impact of land management
activities on the other designated uses of water, such as domestic water supply and recreation, and the other values
of water bodies that may not be recognized, such as the health of aquatic and riparian ecosystems. The current trend
is clearly towards increasingly stringentregulation of forest practices, and there are no signs that public concern will
abate in the future.
Passage of theNational En vironmental Policy Act (NEPA) in 1969 provided a means for the regulatory agencies
and the public to openly evaluate the potential environmental impacts of major management actions and participate
in the federal planning process. However, there often is not a comparable, clearly defined process by which the
public and regulatory agencies can evaluate the effects of management activities on the environment. This is
particularly true in the forestry arena, as nonpoint source pollution is controlled primarily by the formulation and
adoption of BestManagementPractices (BMPs). Effective BMP evaluation can be done only by directly monitoring
the effects of management activities on the designated uses of the water bodies of concern. Nearly 20 years of
experience has shown that the protection of streams through BMPs is an iterative process, and state water quality
agencies, together with the U.S.Environmental Protection Agency (EPA), havebeen given the primary responsiblity
for overseeing the adoption, implementation, and evaluation of the management practices needed to adequately
protect water quality. Clearly this mandate can be carried out only if there is an effective means to monitor the effects
of forestry activities on aquatic ecosystems and the designated uses of water.
An ideal parameter for monitoring the impacts of a land management activity such as forestry should
• be highly sensitive (responsive) to the management action(s),
• have low spatial and temporal variability,
• be accurate, precise, and easy to measure, and
• be directly related to the designated uses of the water body.
This ideal parameter should then be monitored in the context of a project which will (1) provide useful feedback
to the managers, (2) directly link management activities to the status of the designated uses both on-site and
downstream, (3) allow statistical inferences to be made to larger populations, and (4) allow quantitative estimates
of risk and uncertainty. Since such ideal parameters do not exist and monitoring projects rarely are able to fulfill
all these objectives, Region 10 of the U.S. EPA, which includes Washington, Oregon, Idaho, and Alaska, proposed
a year-long project to develop guidelines for monitoring the effects of forestry-related activities on streams. The
present document stems from that initial proposal, and it addresses both the design of monitoring projects and the
selection of parameters for monitoring nonpoint sources of pollution in forested areas.
These Guidelinesfollow a tradition of concern atEPAfor the aquatic effects of management activities in forested
areas. Earlier publications addressed road construction (EPA, 1975), timber harvest (EPA, 1976), the application
of forest chemicals (EPA, 1977), an evaluation of nonpoint silvicultural sources (EPA, 1980), and effectiveness of
nonpoint controls (EPA, 1988). Our hope is that this document will stimulate further analysis and progress in the
design and execution of water quality monitoring projects.
The field of water quality monitoring in forested areas is still young, and the preparation of these Guidelines
made apparent the relative paucity of published information on the results of monitoring projects and many of the
parameters evaluated in this document. Even "unsuccessful" monitoring projects can help direct future monitoring
efforts if the results are properly evaluated and disseminated.
These Guidelines represent our best effort to define the key elements that lead to a successful monitoring
project. By direct interviews, literature reviews, and the generous assistance of numerous experts, we have
attempted to summarize the state of the art and anticipate future developments. We fully recognize that monitoring
nonpoint source pollution in forested areas is, like management, an iterative process. We hope that these Guidelines
vm
-------
will provide the basis for continued improvements, as the future of our streams in forested areas will largely depend
upon the quality of our monitoring efforts.
EPA, 1975. Logging road and protection of water quality. U.S. Environmental Protection Agency, Region 10,
Water Division, EPA 910/9-75-007. Seattle, WA. 312 p.
EPA, 1976. Forest harvest, residue treatment, reforestation and protection of water quality. U.S. Environmental
Protection Agency, Region 10, Water Division, EPA 910/9-76-020. Seattle, WA. 273 p.
EPA, 1977. Silvicultural chemicals and protection of water quality. U.S. Environmental Protection Agency,
Region 10, Water Division, EPA 910/9-77-036. Seattle, WA. 224 p.
EPA, 1980. An approach to water resources evaluation of non-point silvicultural sources (A procedural
handbook). U.S. Environmental Protection Agency, Environ. Res. Lab., EPA-600/8-80-012. Athens, GA.
836 p.
EPA, 1988. Effectiveness of agricultural and silvicultural nonpoint source controls. U.S. Environmental Protection
Agency, Water Div., Region 10, EPA 910/9-88-210. Seattle, WA. 181 p. + app.
IX
-------
ACKNOWLEDGMENTS
TTiedevelcpnentandexecutionofa]rojectsuchasth^
Generally projects ace most vulnerable during the early stages of formulation, and in this case most of the initial impetus came
outof theRegion lOoffice of theU.S.EnvironmentalProtectionAgency(EPA)in Seattle. Tom Wilson, Chief, Officeof Water
Planning, and Elbert Moore, Nonpoint Source Coordinator, both played key roles in developing the conceptual basis for the
projectand garnering support Their shepherdingrole continued as the project was funded and brought to completion, and we
deeply appreciate their efforts and especially their patience.
A Project Steering Committee was formed at the Center for Streamside Studies (CSS) at the University of Washington to
oversee theformulationofaworkplan, the hiringofacocmiinatOT.andtoassistwim me t^parationof the document Members
of the Project Steering Committee included Dr. RobertNaiman, Director of CSS; Dr. Ken Raedeke, CSS; Dr. Dennis Hair, a
research scientist for the U.S. Forest Service based at the University of Washington; Dr. Loveday Conquest, Center for
QuantitativeStudies at the University of Washington; Mr. SteveRalph, Coordinator for the Timber/Fisheries/Wfldlife(TFW)
AmbientMonitoringProgram basedatCSS; and the three aumoreofmeGz«oriald,botnattheForestServicePacificNorthwestRegional Office. Mr. GinoLuchetti, King County, Washington, and
Mr.BnK#McC&mmOT,MtHoodNationalFcH^teachcontributedidea^
other government agencies, and several universities were called upon for their comments, and we hope that these individuals
will see that their suggestions have helped to strengthen the Guidelines.
Afinal set of thanks must go to graphics designer Ms. AprilRichardsonand the ResearchPublications Editor at the School
of Fisheries, Mr. Marcus Duke. April combined our various and diverse ideas into a coherent cover design, while Marcus
transformedroughtextandmeresketehesoffiguresandt^
this document might never have become a finished publication.
Despite the contributions of all these individuals to the Guidelines, final responsiblity for its content and remaining
shortcomings rests with us. We emphasize that in preparing this document we viewed it not as an ultimate truth, but as a set
of guidelines which mustbeevaluatedandapplied with discretion. Itis our hope that the Guidelines willprove useful to a wide
audience. However, even if the Guidelines do no more than stimulate further thought and a more critical design and execution
of water quality monitoring projects, we will have achieved a major portion of our objectives.
JUNE 1991
LEE H. MACDONALD
ALAN W. SMART
ROBERT C. WISSMAR
-------
GLOSSARY
bankfull stage
bankfull discharge
BMP
BOD
cm
CSS
dbh
DO
EPA
ft
g
ha
hr
IBI
ISI
IWB
L
LC-50
LWD
m
M.
mg
ml
mm
min
NPDES
ppb
ppm
PCE
RAPID
RBP
reach
sec
TMDL
less than
greater than
the water surface elevation of a stream flowing at channel capacity
discharge at bankfull stage
Best Management Practice
biological oxygen demand
centimeter
Center for Streatnside Studies
diameter breast height
dissolved oxygen
U.S. Environmental Protection Agency
foot/feet
gram
hectare
hour
index of biotic integrity
interstitial space index
index of well being
liter
lethal concentration, 50%; the concentration of a toxin or pollutant
that kills half of the organisms in a test population per unit time
large woody debris
meter
micro (Ifr6)
milligram
milliliter
millimeter
minute
National Pollutant Discharge Elimination System
parts per billion
parts per million
percent cobble embededdness
Rapid Aerial Photographic Inventory of Disturbance
Rapid Bioassessment Protocol
a continuous portion of a stream between two designated points
second
total maximum daily load
XI
-------
ABOUT THE AUTHORS
Lee H. MacDonald
Associate Professor, Watershed Science Program
Department of Earth Resources
Colorado State University
Fort Collins, CO 80523
(303) 491-6109
Dr. MacDonald recently joined the faculty at Colorado State University as associate professor of land use
hydrology. He takes a process-based approach to analyze the effects of land use activities on wetlands and the
quantity, quality, and timing of streamflow. Dr. MacDonald has extensive experience both as a consultant and with
the United Nations. Prior to joining the faculty at CSU, Dr. MacDonald was at the Center for Streamside Studies
at the University of Washington. Dr. MacDonald's formal training includes an interdisciplinary undergraduate
degree from Stanford University (1974), a M.S. in resource ecology from the University of Michigan (1981), and
a Ph.D. from the Department of Forestry and Resource Management at U.C. Berkeley (1989).
Alan Smart
Water Division
U.S. Environmental Protection Agency, Region 10
1200 Sixth St., Seattle, WA
(206) 553-2579
Mr. Alan Smart is a USDA Forest Service hydrologist assigned to the Environmental Protection Agency in
Seattle, where he provides expertise relating forestry activities to water quality protection. He has worked on the
Siskiyou National Forest in southwestern Oregon planning timber sales and burn rehabilitation projects, and served
as forest hydrologist on the Santa Fe National Forest Prior to his work with the Forest Service, he was a surface
water surveillance hydrologist with the U.S. Geological Survey in Oklahoma. Mr. Smart received a B .S. degree in
forest management (1972) at the University of Montana, and did graduate work in forest hydrology at Oregon State
University (1972-76).
Robert C. Wissmar
Professor, Center for Streamside Studies & Fisheries Research Institute
University of Washington
Seattle, WA 98195
(206) 543-7467
Dr. Robert Wissmar is an aquatic ecologist with the Center for Streamside Studies in Fisheries and Wildlife
(CSS) and the Fisheries Research Institute at the University of Washington. He is involved with teaching and
research programs that examine the influences between riparian and aquatic ecosystems. His research interests
include stream habitat-riparian forest interactions across riverine landscapes; carbon and nutrient cycling; and
juvenile fish-prey interactions and habitat requirements. His education includes: B.S. (Zoology), University of
Utah, 1965; M.S. (Zoology), University of Idaho, 1968; and aPh.D. (Zoology), University of Idaho, 1972.
Xll
-------
EXECUTIVE SUMMARY
1. The purpose of these Guidelines is to assist people in developing water quality monitoring plans. This includes
the design of monitoring projects and the selection of monitoring parameters. The rationale for the Guidelines
is that nonpoint sources of pollution more commonly limit the designated uses of water in forested areas than
point sources. Effective monitoring projects are essential to determine trends, evaluate control efforts, and
assess the impact of management activities on the designated uses of water.
2. The scope of the Guidelines is limited to forested areas in Washington, Oregon, Idaho, and Alaska (Region
10 of the U.S. Environmental Protection Agency). This helps reduce the variability in the range of conditions
to be monitored, and the number of management activities that must be evaluated. Although the focus is on
the effects of forestry and forestry-related activities on streams, other management activities that often occur
in forested areas (e.g., grazing, mining, and recreation) also are discussed because they directly affect water
quality in forested areas, and the effects of these other activities generally cannot be monitored independently
from forest management activities. Similarly, the Guidelines focus on streams and do not directly address
monitoring procedures in lakes, reservoirs, and other downstream areas. However, the Guidelines explicitly
recognize that upstream changes in water quality can affect downstream designated uses, and this must be
considered when formulating a monitoring plan and selecting monitoring parameters.
3. For these Guidelines water quality is defined in the broadest possible sense. Hence the monitoring parameters
include not only the traditional physical and chemical constituents of water, but also those parameters which
directly affect the designated uses of water. A total of 30 parameters or groups of parameters are evaluated
and reviewed, and these are as follows: physical and chemical constituents (temperature, pH, conductivity,
dissolved oxygen, nitrogen, phosphorus, herbicides and pesticides); flow characteristics (size of peak flows,
amount of low flow, water yield); sediment (suspended sediment, turbidity, bedload); channel characteristics
(cross-section, width and width/depth ratio, pool parameters, thalweg profile, habitat units, bed material
particle size, embeddedness, surface vs. subsurface bed material particle size, large woody debris, bank
stability); riparian characteristics (riparian canopy opening, riparian vegetation); and aquatic organisms
(bacteria, algae, macroinvertebrates, and fish).
4. Seven types of monitoring are defined—trend, baseline, implementation, effectiveness, project, validation,
and compliance monitoring. Because of the focus on instream, channel, and riparian parameters, the
Guidelines generally are less applicable to implementation monitoring and some types of effectiveness
monitoring.
5. The legal background for water quality monitoring is reviewed. Two key roles for water quality monitoring
are to determine if the designated uses for aparticular water body are being impaired, and whether water quality
standards are being met. Answers to these questions often determine the type and intensity of monitoring
activities. Regular feedback of the monitoring results through well-defined feedback loops is an essential
component of any monitoring project. The design and execution of monitoring projects must be considered
an iterative process, as the process of data collection and analysis inevitably will have implications for the
frequency, location, and type of measurements.
6. The most important step in developing a monitoring plan is to clearly define the objectives. A clear and detailed
set of objectives will help preclude unrealistic expectations and greatly facilitate the design of a monitoring
plan. A pilot project can prove extremely useful and cost-effective when there is some uncertainty about the
type and location of monitoring activities.
7. The statistical considerations of water quality monitoring are very important since water quality monitoring
is a process of sampling selected parameters in space and over time. Replication of samples, treatments, and
controls is essential if any statistical inferences or generalizations are to be made. The use of statistics permits
quantitative estimates of risk, error, and uncertainty.
8. Common designs and sampling procedures are briefly reviewed with regard to water quality monitoring.
Examples and case studies are used to illustrate some of the salient points. Key problems discussed include
overlapping management activities, determination of cause-and-effect, separation of natural and anthropo-
genic causes, and the potential time lag between management activities and changes in the parameter being
monitored. The trade-offs between sample size, sample variability, level of significance, power (probability
of detecting a real difference), and minimum detectable effect are explicitly reviewed and illustrated.
xiu
-------
9. The selection of monitoring parameters is presented as a function of the designated uses of the water body being
monitored, the type of management activities, and the cost of monitoring. Monitoring costs are broken into
the frequency of sampling, the range of flow conditions needing sampling, data collection time, equipment
costs, and analytic costs. All 30 parameters evaluated in the Guidelines are qualitatively ranked with regard
to each controlling factor in a series of tables. Access to the proposed monitoring sites, the availability of
existing data, and the physical environment (e.g., climate, land form, geology, and soils) are other important
factors that cannot be qualitatively ranked, but which influence the relative value of the different monitoring
parameters.
10. Chapter 5 of Part I integrates these qualitative rankings to provide an overall evaluation of the usefulness of
each parameter for the 10 different management activities considered in the Guidelines. The rationale for the
relative rankings presented in Table 5 is briefly discussed for each management activity.
11. Theresultsindicatethatthechoiceof monitoring parameters is rarely clear. For monitoring forestmanagement
activities, most of the traditional physical and chemical parameters have only limited usefulness because of
theirrelativeinsensitivity, their high cost of monitoring, or both. Theparameters related to channel characteristics
are promising because of their relative sensitivity and low measurement costs. Aquatic organisms—
particularly invertebrates—also have some specific features that may prove useful for monitoring, but more
work is needed before they can be widely utilized in the Pacific Northwest and Alaska.
12. A limitation of the procedure for selecting the most useful parameters is that each parameter is considered
independently. A final table evaluates the magnitude of the interrelationships between each possible pair of
parameters, and this helps to identify those parameters that may be overlapping or redundant with regard to
monitoring particular management activities.
13. The second part of the Guidelines is a technical review of each of the parameters evaluated in Part I, and this
is designed to provide an overview of each parameter and to serve as a reference section. For each parameter
there are seven sub-sections: (1) definition, (2) relation to designated uses (i.e., how changes in the parameter
affect the designated uses of water), (3) effect of management activities on the parameter, (4) measurement
concepts, (5) standards, (6) current uses, and (7) assessment. The assessment section is designed as an overall,
qualitative summary of the parameter as it relates to water quality monitoring—particularly for forestry
activities—in streams in the Pacific Northwest and Alaska.
14. The technical review of each parameter provides a summary of the relevant literature, but it is not a
comprehensive review or an operational manual. However, the numerous literature citations allow rapid
identification of sources for more detailed information, including field measurement techniques and analytic
procedures.
15. The parameter selection procedure presented in Part I has been incorporated into a PC-based expert system
calledPASSSFA (PArameter Selection System for Streams in Forested Areas). The executable version of the
expert system allows users to quickly identify appropriate monitoring parameters through an interactive series
of questions and answers. The confidence level assigned to each recommended parameter provides a relative
indication of the likely usefulness of that parameter givenaparticularsetofmanagementactivities, designated
uses, and monitoring constraints. A "what if function allows the user to quickly alter his or her response to
a particular question and then generate a revised list of recommended monitoring parameters.
xiv
-------
PARTI
DEVELOPING A MONITORING PROJECT
-------
-------
1. INTRODUCTION
1.1 PURPOSE OF THE GUIDELINES
The purpose of this document is to assist land use
managers and their technical staff in developing water
quality monitoring plans for forested areas in Washington,
Oregon, Idaho, and Alaska. The document focuses on the
design of monitoring projects and the selection of the para-
meters to be monitored given (1) the designated and existing
water uses, (2) the type and intensity of management activi-
ties, (3) the environmental setting, and (4) the monitoring
objectives and constraints. Even though the discussion and
examples are directed towards forest managementactivities
in the Pacific Northwest and Alaska, at least the monitoring
principles and the hierarchy of decision-making should be
broadly applicable both to other management activities and
to non-forested environments. The Guidelines—particu-
larly the technical reviews of individual parameters in Part
II—are not intended to provide a step-by-step guide to field
procedures and analytic techniques, as this information is
readily available from the references cited in the text
TherationalefordevelopmgmeseGzttWe/wzesstemsfrom
the increasing emphasis on controlling nonpoint sources of
water pollution, and the recognition that monitoring is an
essential component of any water pollution control pro-
gram. The emphasis on nonpoint sources is due to the
realization that nonpoint sources are the major cause of
water quality impairment in rivers and lakes in the U.S.
(EPA, 1989a). Better control of nonpoint sources is neces-
sary if the broad objectives of the Clean Water Act are to be
achieved.
The conceptual and methodological problems associ-
ated with assessing and monitoring nonpoint sources are
quite distinct from those associated with point sources. For
point sources the quantity and type of pollution usually can
be measured prior to its release. Comparable data cannot
easily be collected for nonpoint sources, so the assessment
and monitoring of nonpoint source pollution must rely on
data collected in the receiving waters. This greatly compli-
cates monitoring, as the pollution is diluted and it may be
difficult to separate management effects from natural pro-
cesses. Furthermore, many of the parameters and measure-
ment techniques used to characterize point source effects
cannot be applied to nonpoint sources. These difficulties are
particularly prominent for forest management activities, as
the resultant pollution often represents a change in an
existing value rather than the introduction of entirely new
pollutants. Hence the two primary objectives of this docu-
ment are (1) to provide guidelines for developing effective
nonpoint source monitoring plans, and (2) to review the
parameters that are or mightbeusefulformonitoringnonpoint
sources of pollution in forested areas.
Throughout this document the term water quality is
used in the broadest possible sense. This means that water
quality includes not only the traditional physical and chemi-
cal constituents such as pH, temperature, and discharge, but
also those parameters that affect the existing and designated
uses of a water body. In many parts of the Pacific Northwest,
for example, maintenance of salmonid fisheries is an impor-
tantdesignateduse. The need toprotectthis use necessitates
concern over other water quality parameters such as the
amount of large woody debris, the number and size of pools,
the density of the riparian canopy, and the particle size of the
bed material. Although these are not normally included in
water quality monitoring projects, each of these parameters
is relatively sensitive to certain forest management activi-
ties and directly related to habitat quality (Section 4.3). We
expect thatfutureprograms to evaluate and control nonpoint
source pollution will have to explicitly acknowledge this
broader definition of water quality, and that some of these
parameters may prove more useful or important than the
traditional chemical and physical constituents of water
quality. Hence these Guidelines evaluate and review nearly
thirty different parameters pertaining to the major desig-
nated uses of water in the Pacific Northwest and Alaska.
-------
Parti
Many of the parameters, such as habitat types and aquatic
macroinvertebrates, actually incorporate a large number of
specific measurements, but were grouped as a single pa-
rameter for practical reasons.
The restriction of this document to nonpoint source
pollution in forested areas means that the management activi-
ties of primary concern are forest harvest, road construction
and maintenance, forest fertilization, pest and weed control,
andrecreau'on. Other aspects of forestmanagement, such as
mechanical site preparation and intermediate stand treat-
ments, are not explicitly considered, as their impact on
water quality is conceptually similar to the impact of forest
harvest androadbuilding, and typically lesser in magnitude.
Grazing, concentrated recreation, mining, and rural
settlements are other management activities that commonly
occur in forested areas and which can adversely affect water
quality. Although these other management activities are
outside the scope of this document, it is often inefficient or
impossible to monitor only the impact of forestry activities,
andnotsimultaneouslyconsidermeeffectsofminmg.grazing,
recreational developments, or fire. Hence the selection of
parameters to monitor each of these other management
activities is briefly discussed in Chapters 4 and 5. To the
extent possible, the parameter reviews in Part II also ac-
knowledge the importance of these other management ac-
tivities. Relatively littleattention is devoted to the effects of
rural communities and recreational developments on water
quality, as extensive literature already exists on this topic.
Mining impacts also are not discussed in detailbecause they
are so variable with regard to the minerals being extracted,
the method of operation, and the processing techniques. In
general the review of specific parameters in Part II should
prove helpful in developing monitoring projects that ex-
plicitly consider and evaluate one or more of these other
management activities.
The Guidelines are designed to be most applicable to
perennial streams and small rivers. In larger river systems
water quality usually is controlled by agricultural, indus-
trial, and municipal wastes, and it becomes very difficult to
distinguish the impact of forest management activities.
Furthermore, some of the parameters discussed in the
Guidelines (e.g., habitat types) cannot be easily applied to
large rivers.
The Guidelines also do not address the design of water
quality monitoring projects for lakes, as lakes are so distinct
in terms of their physical and biological characteristics.
Most lake monitoring projects rely on the collection and
analysis of water samples, while a much broader range of
monitoring parameters can be used in streams and rivers.
Although someof the parameters discussed in the Guidelines
are relevant to water quality monitoring in lakes, the
Guidelines do not explicitly address their potential use in
lacustrine environments.
However, the Guidelines explicitly consider the need to
protect water quality in downstream lakes and reservoirs, as
this can be an important designated use of water from
forested areas. Concern over downstream lake water quality
may affect the design of a stream monitoring project by
requiring additional parameters to be monitored, or by
further constraining the allowable change in certain pa-
rameters. The Guidelines explicitly identify such situations,
and suggest which parameters are most appropriate for
protecting lake water quality.
Finally, the Guidelines consider only those measure-
ments which can be made either in or immediately adjacent
to the stream channel. Observations on upslope areas often
are essential to understanding the cause of changes observed
in the stream channel, and they also may exhibit a higher
sensitivity to management actions than inchannel measure-
ments. Nevertheless, upslope measurements represent a
completely different set of monitoring techniques that are
not addressed in the present document.
1.2 ORGANIZATION AND USE OF THE
GUIDELINES
The Guidelines are divided into two parts. Partlpresents
the background and principles of developing a water quality
monitoring plan for nonpoint source pollution in forested
areas. It includes a discussion of the factors that should be
considered in developing a monitoring plan, and a set of
tables summarizing the sensitivity, cost, and usefulness of
the various monitoring parameters vis-a-vis the designated
uses of water and the management activity to be monitored.
Part II presents individual technical reviews of the monitor-
ing parameters that have been or might be used to monitor
water quality in forested areas.
A user's guide to this document is provided in Box 1.
This indicates that those who wish only to learn more about
a certain parameter should go directly to the relevant section
in Part H.
Those who have a clearly defined monitoring objective,
but are uncertain about the parameters to be monitored, can
use the five tables in Chapters 4 and 5 of Part I as a
qualitative guide to the most appropriate parameter(s).
Table 2 (page 39) rates each parameter with regard to its
effects on different designated uses (i.e., how does a change
in parameter X limit different designated uses). Table 3
(page 41) rates the sensitivity of the parameters to different
management activities (i.e., how likely is parameter X to
change as a result of various management activities). Table
4 (page 43) provides a general indication of the "typical"
frequency and sampling cost associated with each param-
eter. Table 5 (page 50) integrates the data from Tables 2-4
to provide a qualitative evaluation of the usefulness of each
parameter for monitoring different management activities.
Tables 6A and 6B (page 62) summarize the interactions
among the parameters; these tables can be used to identify
-------
CHAPTER 1. INTRODUCTION
Box 1. How TO USE THIS DOCUMENT
Guidance on specific aspects of developing a monitoring plan can be found as follows:
Defining objectives
Identifying types of monitoring
Understanding the legal framework for monitoring
Sampling, statistical design, and statistical trade-offs
Factors which constrain and help define a monitoring plan
Recommended monitoring parameters
Parameter Interactions
Part I, Section 4.1
Part I, Section 1.3
Part 1, Sections 1.4 and 2.1
Part I, Chapter 3
Part I, Chapter 4
Part I, Section 5.1
Part I, Section 5,3
A quick and generalized evaluation of the monitoring parameters with regard to the following topics cart
be found in the table as indicated:
Effect on designated uses
Sensitivity to management activities
Frequency and cost of sampling
Recommended use for monitoring different
management activities
Interactions {i.e., the effect of one parameter
on another parameter)
Table 2, p. 39
Table 3, p. 41
Table 4, p. 43
Table 5, p. 50-51
Tables 6A and 6B, p. 62-65
Information on the expert system can be found in Part I, Section 5.2.
More information about specific monitoring parameters can be found in the appropriate section of
Part II.
Physical and chemical constituents
Discharge parameters
Sediment parameters
Channel characteristics
Riparian condition
Aquatic organisms
p. 71
p. 90
p,96
p. 107
p. 131
p. 138
closely related parameters that might be redundant or nec-
essary to evaluate the cause of an observed change.
It should be recognized that these tables represent a
generalized, qualitative evaluation that will not apply in all
cases. The diversity of environments and processes in the
Pacific Northwest and Alaska, together with the variation in
management, precludes any absolute evaluation. Hence the
tables should be regarded as an initial guide, and the values
will need to be adjusted according to local knowledge and
experience. The tables should also be interpreted in the
context of the text in Chapters 4 and 5, as these two chapters
outline the principles behind the qualitative rankings and
provide more detail on how the rankings might be adjusted
under different circumstances.
Sections 1.3 through 3.4 of Part I will be useful to those
who are uncertain about the structure and concepts that must
be considered when developing a monitoring plan. Again
the diversity of objectives, environments, and management
activities means that the emphasis in the Guidelines is on
basic principles, and the user will need to apply these to their
particular situation. In Section 1.3 the various types of
monitoring are described, while in Section 1.4 the legal back-
ground for monitoring nonpoint sources is provided.
Chapter 2 places the types
of monitoring into the legal
framework described in Sec-
tion 1.4, and then outlines the
basic process of developing
water quality monitoringplans.
Chapter 3 discusses the
statistical considerations in
monitoring. It points out that
monitoring is inherently a
sampling procedure. This
means that questions of statis-
tical design and analysis must
beexpliciflyrecognized. Again
this chapter does not provide
step-by-step procedures, but
emphasizes the understanding
and application of basic prin-
ciples.
Chapter 4 reviews the fac-
tors that must be considered in
developing a monitoring plan.
Formulation of specific objec-
tives is seen as the single most
important step. The remaining
sections of Chapter 4 discuss
the other important consider-
ations involved in formulating
a monitoring plan. These in-
clude the designated uses of
water, type of management ac-
tivity, frequency and cost of monitoring, availability of
existing data, and the physical environment. Although the
discussion and examples focus on land management activi-
ties in forested areas in the Pacific Northwest and Alaska,
the principles and hierarchy can be applied both to other
management activities and non-forested environments.
ChapterSqualitativelyrankstheparameterswithregard
to their usefulness for monitoring the water quality effects
of land use actions in forested areas. These recommenda-
tions represent an integration of the information in Chapters
1-4 and in Part n. Again it should be emphasized that these
rankings are not absolute, but a relative evaluation of which
parameters should be most generally applicable. The sec-
ond part of Chapter 5 discusses the interactions among the
various parameters, and Tables 6A and 6B show how each
parameter affects all of theother parameters discussed in the
Guidelines. This information helps identify those closely
related parameters which (1) might be another source of
change for the parameters being monitored, or (2) might
serve as a surrogate for the parameter of interest.
Part II of this document can be viewed as a reference
section for each of the monitoring parameters discussed in
Part I and ranked in Tables 2-6. The individual parameters
are grouped into five classes as follows: water column
-------
Parti
measurements (temperature, pH, conductivity, dissolved
oxygen, nutrients, herbicides, and pesticides); discharge
parameters (size of peak flows, low flows, and water yield);
sediment parameters (suspended sediment, turbidity, and
bedload); channel characteristics (channel cross-section,
width and width-depth ratio, pool parameters, thalweg
profile, habitat type, bed material size, embeddedness, large
woody debris, and bank stability); riparian condition (ripar-
ian canopy opening, riparian vegetation); and biological
components (bacteria, algae, invertebrates, and fish).
For convenience the discussion of each parameter is
divided into seven sub-sections:
1. definition,
2. relation to designated uses,
3. response to management activities,
4. measurement concepts,
5. standards,
6. current uses, and
7. assessment
The assessment section is a qualitative evaluation and
summary, and it can be read separately if desired. The
extensive references at the end of Part II direct the reader to
more detailed sources of information on each parameter.
In summary, the Guidelines are designed to help guide
the development of a monitoring plan, with particular em-
phasis on the selection of the parameters to be monitored. It
must be recognized, however, that the Guidelines are sub-
ject to several limitations. First, the tables are a qualitative
evaluation based on a combination of experience and pub-
lished data. We have tried to integrate the views of many
experts, but there will always be some divergence of opin-
ions. Second, it is not possible to develop a set of guidelines
which will apply in all environments for all conditions. Any
divergence between the Guidelines and one's individual
views should be used to stimulate further discussion and a
critical reassessment Ultimately, however, local knowl-
edge and experience should take precedence over any gen-
eralized guidelines. Third, the discussion and matrices are
based on current knowledge. In many cases data on the
sensitivity and variability of a parameter are not available,
or are known only for a particular environment. As more
data are accumulated, our opinion as to relative usefulness
ofaparametermaychange.Finally.measurementtechniques
are evolving, and this will affect the ease of measurement,
the inherent variability, and the sensitivity to detect change.
1.3 TYPES OF MONITORING
The term "monitor" is defined as to watch or check.
Although it is not an explicit part of the definition, the term
monitoring suggests a series of observations over time. This
repetition of measurements over time for the purpose of
detecting change distinguishes monitoring from inventory
and assessment While both inventories and assessments
can based on a single measurement or observation, they also
can incorporate a series of observations to obtain a better
estimate of aparticular parameter. For example, the number
of species of fish in a particular reach might be counted as
partof an inventory offish species, andseveral counts might
be made in ordertoobtainamoreaccurateestimate. Similarly,
maximum daily water temperature might be measured sev-
eral times over the course of a summer to assess whether
summer temperatures might be an important limitation to
the quality of fish habitat under the existing conditions.
However, if water temperatures are measured over several
years to determine the effect of upstream management ac-
tivities or climatic variations, this is clearly monitoring. The
overlap in the definitions of assessment, inventory, and
monitoring means that in some cases the primary distin-
guishing feature of monitoring will be the intent to assess
change rather than the number or type of measurements.
Often an assessment or inventory serves as the first step
towards establishing a monitoring project. Knowledge of
the spatial and temporal variability is essential to develop-
ing an efficient monitoring plan (Chapter 3). To the extent
that inventory and assessment techniques overlap with
monitoring procedures, these Guidelines can help with the
conceptual problems of deciding what, where, and how to
inventory or assess water quality.
A number of federal and state agencies have defined the
different types of monitoring carried out by their particular
organization (e.g., Potyondy, 1980; Solomon, 1989). Un-
fortunately these definitions are not consistent, and this has
often resulted in semantic confusion. In most cases a clear
statement of the purpose of the monitoring will be the best
method of defining the type of monitoring, and it then is
simply a matter of attaching a mutually agreeable label to
that particular type of monitoring. For the purposes of this
document, the following types of monitoring are defined:
1. Trend monitoring. In view of the definition of moni-
toring, this term is redundant. Use of the adjective
"trend" implies that measurements will be made at
regular, well-spaced time intervals in order to deter-
mine the long-term trend in a particular parameter.
Typically the observations are not taken specifically
to evaluate management practices (as in type 4),
management activities (as in type 5), water quality
models (as in type 6), or water quality standards (as in
type 7), although trend data may be utilized for one or
all of these other purposes.
2. Baseline monitoring. Baseline monitoring is used to
characterize existing water quality conditions, and to
establish a data base for planning or future compari-
sons. The intent of baseline monitoring is to capture
much of the temporal variability of the constituent(s)
of interest but there is no explicit end point at which
continued baseline monitoring becomes trend moni-
toring. Those who prefer the terms "inventory moni-
toring" and "assessment monitoring" often define
-------
CHAPTER 1. INTRODUCTION
them such that they are essentially synonymous with
baseline monitoring. Others use baseline monitoring
to refer to long-term trend monitoring on major streams
(e.g., Potyondy, 1980).
3. Implementation monitoring. This type of monitoring
assesses whether activities were carried out asplanned.
The most common use of implementation monitoring
is to determine whether Best Management Practices
(BMPs) were implemented as specified in an environ-
mental assessment, environmental impact statement,
other planning document, or contract Typically this
is carried out as an administrative review and does not
involve any water quality measurements. Implemen-
tation monitoring is one of the few terms which has a
relatively widespread and consistent definition. Many
believe that implementation monitoring is the most
cost-effective means to reduce nonpoint source pollu-
tion because it provides immediate feedback to the
managers on whether the BMP process is being car-
ried out as intended (Section 1.4). On its own, how-
ever, implementation monitoring cannot directly link
management activities to water quality, as no water
quality measurements are being made.
4. Effectiveness monitoring. While implementation
monitoring is used to assess whether a particular activity
was carried out as planned, effectiveness monitoring
is used to evaluate whether the specified activities had
the desired effect (Solomon, 1989). Confusion arises
over whether effectiveness monitoring should be lim-
ited to evaluating individual BMPs, or whether it also
can be used to evaluate the total effect of an entire set
of practices. The problem with this broader definition
is that the distinction between effectiveness monitor-
ing and other terms, such as project or compliance
monitoring, becomes blurred.
To minimize confusion within this document, ef-
fectiveness monitoring will be used in the narrow
sense of evaluating individual management practices,
particularly BMPs (Section 1.4). Monitoring the
effectiveness of individual BMPs, such as the spacing
of water bars on skid trails, is an important part of the
overall process of controlling nonpoint source pollu-
tion (Sections 1.4 and Chapter 2). However, in most
cases the monitoring of individual BMPs is quite
different from monitoring to determine whether the
cumulative effect of all the BMPs results in adequate
waterquality protection. Evaluating individual BMPs
may require detailed and specialized measurements
best made at the siteof, or immediately adjacent to, the
management practice. Thus effectiveness monitoring
often occursoutsideof the stream channel andriparian
area, even though the objectiveof aparticularpractice
is intended to protect the designated uses of a water
body. In contrast, monitoring the overall effective-
ness of BMPs usually is done in the stream channel,
and it may be difficult to relate these measurements to
the effectiveness of individual BMPs.
5. Project monitoring. This type of monitoring assesses
the impact of a particular activity or project, such as a
timber sale or construction of a skirun on water quality.
Often this assessment is done by comparing data taken
upstream and downstream of the particular project,
although in some cases, such as a fish habitat im-
provement project, the comparison may be on a before
and after basis. Because such comparisons may, in
part, indicate the overall effectiveness of the BMPs
and other mitigation measures associated with the
project, some agencies consider project monitoring to
be a subset of effectiveness monitoring. Again the
problem is that water quality is a function of more than
the effectiveness of the BMPs associated with the
project.
6. Validation monitoring. Since the issue of validating
water quality standards is beyond the scope of this
document, validation monitoring in these Guidelines
is discussed primarily with regard to the quantitative
evaluation of a proposed water quality model to pre-
dict a particular water quality parameter. In keeping
with the basic principles of modeling (e.g., James and
Burges, 1982), the data set used for validation should
be different from the data set used to construct and
calibrate the model. This separation helps ensure that
the validation data willprovide an unbiased evaluation
of the overall performance of the model. The intensity
and type of sampling for validation monitoring should
be consistent with the output of the model being
validated.
7. Compliance monitoring. This is the monitoring used
to determine whether specified water-quality criteria
are being met The criteria can be numerical or de-
scriptive. Usually the regulations associated with
individual criterion specify the location, frequency,
and method of measurement.
It should be emphasized that these seven types of
monitoring are not mutually exclusive. Often the distinc-
tion between them is determined more by the purpose of
monitoring than by the type and intensity of measurements.
Regular sampling of coliform bacteria to meet health stan-
dards, for example, will produce data that also can be used
to indicate long-term trends. Table 1 is a broad classifica-
tion of monitoring types according to the parameters being
measured, the frequency of monitoring, the duration of
monitoring, and the intensity of data analysis. At this point
no consensus exists on the definitions of monitoring types,
and this, together with the proliferation of monitoring ter-
minology, means that each monitoring plan should explic-
itly define the monitoring terminology being used.
These Guidelines are not equally applicable to all seven
monitoring types as defined above. Most of the parameters
used for trend, baseline, and project monitoring are explic-
-------
Parti
Table 1. General characteristics of monitoring types.
Type of
monitoring
Trend
Baseline
Implementation
Effectiveness
Project
Validation
Compliance
Number and
type of water
quality parameters
Usually water column
Variable
None
Near activity
Variable
Few
Few
Frequency of
measurements
Low
Low
Variable
Medium to high
Medium to high
High
Variable
Duration of
monitoring
Long
Short to medium
Duration of project
Usually short to medium
>Project duration
Usually medium to long
Dependent on project
Intensity of
data analysis
Low to moderate
Low to moderate
Low
Medium
Medium
High
Moderate to high
itly considered, and the general discussion on developing a
monitoring plan is directly relevant. On the other hand,
implementation monitoringgenerally does not involve water
quality measurements, and so the Guidelines are less ap-
plicable. Effectiveness monitoring of individual BMPs also
may use different parameters than the ones discussed in
these Guidelines.
Since validation monitoring is used to evaluate model
accuracy, the parameters to be measured are defined by the
model output Usually these will correspond to some of the
monitoring parameters reviewed in this document, but this
may not necessarily be the case. Similarly, the parameters
and procedures for compliance monitoring usually are spe-
cified by the regulating agency. Some standards are written
in qualitative language, and in such cases several different
procedures might be used to assess a broadly defined stan-
dard such as "biological integrity." Nevertheless, most of
the constituents incorporated in compliance monitoring
projects in forested areas are included in these Guidelines.
Most water quality monitoring projects will involve
more than one of the types of monitoring defined above.
The integration of several monitoring types into one project
usually is due to multiple objectives. As suggested previ-
ously, distinct objectives attained through different types of
monitoring do not necessarily require distinct and indepen-
dent data collection efforts. If the monitoring objectives are
clearly specified, one usually finds considerable overlap in
terms of the data needs, and recognition of this can result in
considerable cost savings.
Box 2—the first of five case studies presented in the
Guidelines—is an overview of ongoing water quality moni-
toring efforts in the Bull Run watershed near Portland,
Oregon. This particular project has been subjected to con-
siderable scrutiny by several parties with diverse interests,
and it recently underwent a thorough technical review
(Aumen et al., 1989). Although it can be argued that the
monitoring efforts on the Bull Run watershed are relatively
unique in terms of their cost and intensity, the ongoing
revisions in the monitoring project have much broader
implications. Of particular interest is the reallocation of
monitoring effort from effectiveness monitoring to imple-
mentation monitoring.
1.4. LEGAL BACKGROUND
The different types of monitoring have evolved partly in
response to the changing objectives and legal requirements
for water quality monitoring. Passage of the Federal Water
Quality Act of 1965 led to the widespread adoption of
instream water quality standards. This stimulated state and
local agencies to initiate more intensive monitoring programs,
but these were oriented more towards meeting the legal
requirements than facilitating management decisions
(Sanders and Ward, 1979).
In 1972 the Federal Water Pollution Control Act estab-
lished a regulatory system for point sources of water pollu-
tion. This added a permit system and self-monitoring of
effluent discharge to existing instream monitoring efforts.
A national goal that all waters should be fishable and
swimmable was established.
Section 208 of the 1972 law recognized that nonpoint
sources could adversely affect water quality and should be
controlled. States were required to prepare plans for con-
trolling nonpoint sources, although implementation was
voluntary (Hohenstein, 1987). The primary mechanism for
regulating nonpoint sources is by adopting and implementing
BMPs (Best Management Practices). In general terms
BMPs are defined as those practices, or combination of
practices, that are practical and effective in preventing or
reducing pollution from nonpoint sources to levels compat-
ible with water quality goals (Lynch and Corbett, 1990).
The current EPA definition of BMPs is as follows:
Methods, measures or practices selected by an
agency to meet its nonpoint control needs. BMPs
include but are not limited to structural and non-
structural controls and operation and maintenance
procedures. BMPs can be applied before, during
and after pollution-producing activities to reduce or
-------
Box 2. CASE STUDY: BULL RUN WATERSHED, OREGON1
The 275-km8 Bull Run watershed lies within the Mt. Hood National Forest about 50 km east of Portland on the
west side of the Cascades. By law the principal management objective is to provide "pure, clear, raw, and potable"
water for the Portland metropolitan area. The water is of exceptionally high quality and is chlorinated but not filtered
before it enters into the municipal supply system. This means that water quality is of utmost concern, and the Bull
Run watershed has the most intensive water quality monitoring program of any municipal watershed in the United
States. In order for management to protect water quality, public access is restricted and activities such as timber
harvest are carefully controlled.
The three monitoring objectives are as follows: (1) to assess the effects of management activities on water
quality, (2) to monitor compliance with raw water quality standards, and (3) to investigate physical processes in order
to improve the predictability of watershed response to management activities and climatic events. The types of
monitoring conducted to meet these objectives are implementation, effectiveness, validation, and trend monitoring.
Implementation monitoring is the process of ensuring that the site-specific management requirements were
carried out as planned. Effectiveness monitoring evaluates whether the management practices, including BMPs,
adequately protected water quality. As defined by the Bull Run project, effectiveness monitoring consists of inventory
monitoring and water quality sampling. Inventory monitoring measures those instream, riparian, and upslope
characteristics that can affect water quality. Examples include large woody debris, stream shading due to the riparian
canopy, and revegetation rates on exposed ground.
Water quality and discharge are measured at five key stations. Four of these are located at the mouths of the
major tributaries to the water supply reservoir, and the fifth is located at the intake to the diversion facility. Another
14source-search stations are used to monitorthe largerstreams and the main reservoir. Samples from these stations
help determine the source of any high value recorded at a key station. Short-term monitoring stations are set up as
needed to monitorthe water quality effects of management activities such as salvage logging, prescribed burning,
and road construction and maintenance. These project monitoring stations are preferably located upstream and
downstream of the management activity, although sometimes a paired-watershed design is utilized.
Water quality standards have been established at the key stations for 39 variables. Discharge, temperature, and
conductivity are measured continuously, while pH, color, turbidity, conductivity, suspended sediment, and several
bacteriological indicators are measured at least weekly. The water quality standards are based on historical data,
and a series of parametric and non-parametric statistical tests are used to determine when a particular value exceeds
the standards. Any deviation must be evaluated and explained through a systematic procedure involving the source-
search stations, additional investigations, and administrative procedures. Most deviations stem from the fact that the
relatively short water quality record does not adequately reflect the more extreme climatic events.
A recent technical review provides an excellent case study of the design and operation of this intensive water
quality monitoring program (Aumen et al., 1989). Specific recommendations that may be more widely applicable
include the following:
. increased emphasis on bn-sfte monitoring during the course of the management activity, as this is the most
effective method for minimizing adverse impacts of management activities on water quality;
. elimination of the source-search stations because they had not proven useful in detecting the cause of
deviations, and they absorbed a large proportion of the monitoring resources;
. fewer, but more intensive, project monitoring sites;
. increased reservoir monitoring to detect subtle long-term effects;
. a shift away from sampling at equal time intervals to flow-based sampling with special emphasis on storm
events;
. more emphasis on data interpretation and understanding watershed processes and functions; and
, altering procedures to allow a more flexible response to often meaningless deviations from the raw water quality
standards.
The attached graph indicates how these recommendations have resulted in a reallocation of monitoring effort
among the different types of monitoring being used on the Bull Run watershed. Previously about 75% of the effort
was directed towards effectiveness monitoring, whereas the revised project places much more emphasis on
implementation and validation monitoring.
'Sources: Aumen et al., 1989; McCammon, 1989.
(Continued on p. 10)
-------
Parti
Current
10% **• 10%
Modified
2%
Monitoring types
20%
78%
40%
| implementation
|H| Effectiveness
[g| Validation
H Trend
Box 2—cont. The reallocafon of monitoring effort in the Bull Run monitoring project.
eliminate the introduction of pollutants into receiv-
ing waters (CFR, 1990).
SomespecificexamplesofBMPsincludetheappropriate
placement of waterbars on skid trails, stabilization and
treatment of cut and fill slopes, and seasonal restrictions on
road use and log skidding. Certain BMPs may be certified
as approved practices for controlling nonpoint source pol-
lution by state water quality control agencies. In the case of
forestry, Alaska, Idaho, Oregon, and Washington all have
Forest Practice Acts which include specific BMPs in the
associated rules and regulations. Each state water quality
agency then certifies these BMPs as appropriate nonpoint
source controls under the authority of the Clean Water Act
If water quality objectives are not met despite the proper
implementation of BMPs, states can de-certify BMPs and
require more stringent measures to minimize adverse effects
on water quality.
The state water quality agencies also may certify BMPs
for those federal agencies which are conducting land-dis-
turbing activities. In general, these agency-derived BMPs
must meet or exceed the relevant state requirements speci-
fied under legislation such as a state Forest Practices Act
Often state certification of BMPs submitted by a federal
agency leads to a delegation of responsibility to that federal
agency to protect and restore those water bodies under its
jurisdiction. Such responsibility is subject to approval and
review by the state water quality agency, and this in turn is
dependentupontheongoingmonitoringprogram conducted
by the federal agency.
Asecondmechanismtocontrolnonpointsourcepollution
is through water quality standards. Water quality standards
are legal requirements combining the designated uses of
water with the numerical criteria necessary to protect those
uses (see S ection 4.2 for further discussion of the designated
uses of water). Water quality criteria are either numeric
limits or narrative descriptions of water quality. The objec-
tive of establishing specific criteria is to protect human
health and aquatic life, as well as the designated uses of
water (EPA, 1988). EPA has recommended specific criteria
for nearly one hundred water quality parameters, including
bacteria, color, nutrients, metals, and a wide range of chemi-
cals (EPA, 1986). Most of these criteria are expressed in
numeric terms and have been adopted by the states in order
to protect human health and aquatic life.
In thecaseofnonpointpollution due toforestry activities,
the use of water quality standards is hindered by the limited
number of criteria which are applicable. Of the nearly one
hundred criteria set out in EPA's Quality Criteria for Water
(EPA, 1986), forest management activities are likely to
affect only a few—dissolved oxygen, temperature, turbid-
ity, suspended solids, and perhaps nitrate-nitrogen. Yet
forestry activities are known to affect numerous other pa-
rameters which do not have EPA-specified criteria, but
which have great significance for aquatic and riparian
ecosystems (Salo and Cundy, 1987; Raedeke, 1988). Ex-
amples of these other parameters include the amount and
type of large woody debris in stream channels, the quality of
the streambed material for spawning, the amount of pool
habitat, and the type and density of riparian vegetation (Part
II, Chapter 5 and Section 6.2).
The primary role of water quality standards in regulat-
ing nonpoint source pollution has been affirmed by court
decisions stating that water quality standards must be used
to evaluate whether BMPs are effective (Hohenstein, 1987).
Again monitoring is the basis for determining if existing
water quality meets the relevant standard(s), and whether
BMPs are effective.
The third mechanism for regulating water quality in
forested areas is the antidegradation policy. This policy has
three tiers or levels of protection (Fig. 1), and some form of
monitoring is required for the successful implementation of
-------
CHAPTER 1. INTRODUCTION
Maintain and
protect water quality
for Outstanding
Resource Waters
Maintain and protect high quality waters
Maintain and protect existing uses and water quality
figure 1. The three levels of the antidegradation policy for protecting existing uses and maintaining or improving existing water quality
(EPA, 1988).
each tier. Basically EPA requires the states to place water
bodies into one of these three tiers, with each tier having a
differentlevel of protection against the degradation of water
quality. The lowest tier or level of protection requires that
existing instream uses be fully supported (i.e., water quality
is not limiting the existing uses). Monitoring for this
purpose could vary from a periodic qualitative assessment
of recreational suitability to quantitative measurements of
parameters such as salmonid egg survival or intergravel
dissolved oxygen.
The second tier or middle level of protection applies to
those waters which have a water quality level higher than
that necessary to support recreation and the propagation of
fish and other wildlife. In these water bodies a lowering of
water quality is allowed only if: (1) the state determines that
this is necessary to accommodate important economic or
social development, and (2) the decision is based on a
review and comment process by the public and state and
local agencies. However, the allowed degradation cannot
adversely affect the existing uses as specified by the first tier
of the antidegradation policy. Monitoring for this purpose
may need to be more quantitative and statistically based as
discussed in Chapter 3.
The third and highest tier of the antidegradation policy
applies tothosewatersdesignated by thestates as Outstanding
Resource Waters. Such designation means than no degra-
dation in water quality is allowed. Typically Outstanding
Resource Waters are the highest quality waters in the state,
and they have some special characteristics to justify this
high level of protection. Water bodies in state or national
parks, or water bodies which are part of the National Wild
and Scenic River System, most often are considered for
designation. Again monitoring at this third tier may require
quantitative measurements and an appropriate statistical
design to detect a specified level of change.
Each of these three main mechanisms for controlling
nonpoint source pollution—BMPs, water quality standards,
and antidegradation—were further enhanced by the 1987
amendments to the Clean Water Act In particular, Section
319 required the states to identify those water bodies which
do notmeetwater quality standards due to nonpointpollution.
Each state then must develop a program to improve water
quality to the point that water quality standards are met.
These programs must be approved by EPA, although the
actual implementation is carried out by the state agency
responsible for water quality.
For those water bodies that still do not meet water
quality standards, despite the implementation ofpointsource
controls and effective BMPs, a load allocation process may
be initiated (bottom of Fig. 3, p. 15). Currently each state
establishes its own priority list for initiating the load allo-
cation process according to the value of the designated uses
and the risk of damage to those uses.
This load allocation process is discussed in Section 2.1,
but basically it is a relatively data-intensive procedure which
may require detailed monitoring. The first step is to identify
the constituents) of concern, and determine the frequency
and timing of water quality violations. The loading capacity
of the water body for each constituent that violates water
quality standards must be quantitatively assessed, and then
a safety factor is subtracted from this loading capacity.
Further subtraction of the contribution of natural sources
yields the Total Maximum Daily Load (TMDL). This is the
amount of pollution which can be contributed by anthro-
pogenic activities, and it is allocated among all the point
sources ("wasteloadallocation")andnonpoint sources ("load
-------
Parti
Biological
integrity
Figure 2. Ecological integrity is attainable when chemical, physical, and biological integrity occur simultaneously (EPA, 1990).
allocation"). In equation form,
Loading capacity - safety factor - contribution from
natural sources = TMDL
TMDL = wasteload allocation + load allocation
Todatetheloadallocationprocess has been developedfor
only a few water bodies in the Pacific Northwest, although
it undoubtedly will be more widely applied in the future.
In summary, a variety of mechanisms have been devel-
oped to control nonpoint source pollution, and there will be
continuing adjustments and additions to these regulatory
tools in the future. To a certain extent it is this variety of
objectives and approaches which has led to the proliferation
of monitoring types and the confusion over monitoring
terminology (Section 1.3). For example, compliance
monitoring usually refers to the monitoring associated with
meeting numerical water quality criteria and the limits
specified in point source discharge permits. Implementa-
tion and effectiveness monitoring often are associated with
the process of implementing and evaluating BMPs, but
effectiveness monitoring also could apply to the evaluation
of specific pollution control programs. Trend monitoring is
necessaryfor Unsuccessful application of the antidegradation
policy.
In late 1985 EPA recognized the changing needs in
water quality monitoring and initiated a study of its surface
water quality monitoring efforts. Several specific needs
were identified (EPA, 1987), and these still are defining
some of the current directions in water quality monitoring.
The first need was to develop and use biological moni-
toring techniques as well as the traditional physical and
chemical water quality parameters. As shown conceptually
in Figure 2, the biological integrity is one component of the
ecological integrity, and biological monitoring is needed to
evaluate the biological integrity. Hence the emphasis on
developing biological criteria stems from the need for im-
proved techniques to evaluate the condition of water bodies,
as well as the need to more directly relate water quality criteria
to designated uses (EPA, 1990). EPA is now working with
the states to develop narrative biological criteria (EPA,
1990), and numerical biological criteria are likely to be
developed subsequently.
The Rapid Bioassessment Protocols for aquatic mac-
roinvertebrates and fish (Part II, Sections 6.2 and 6.3, re-
spectively) are expected to serve as the prototype techniques
for assessingand defining the biological integrity of streams.
However, the establishment of biological criteria for streams
in thePacificNorthwestand Alaska is likely to be an extended
A second recommendation of the EPA monitoring study
was that water quality monitoring programs should aim to
demonstrate the results of water pollution control efforts
-------
CHAPTER 1. INTRODUCTION
(EPA, 1987). Such data are needed to create a feedback loop
for guiding future management decisions, and to justify the
resources being spent on water pollution control efforts.
This need to link water pollution control efforts and water
body condition has important implications for the selection
ofmonitoringparametersandthetypeofmonitoringprojects
being undertaken. In particular, this objective suggests a
move away from the traditional fixed-site monitoring sta-
tions in downstream locations, as water quality changes at
these downstream sites tend to be smaller in magnitude and
more difficult to relate to management actions. Locations
higher in the watershed, where the monitoring data can be
more easily linked to specific management actions, are
more likely to be emphasized in the future. The need to
document change may also stimulate a shift away from the
traditional physical and chemical parameters to parameters
that are more sensitive to management activities and which
can be directly related to the designated uses. An objective
of these Guidelines is to facilitate this change in emphasis.
A third challenge identified in the EPA review was to
identify and characterize pollutants from nonpoint sources
(EPA, 1987). The selection and review of monitoring para-
meters in these Guidelines can be considered as one com-
ponent of EPA's response to this challenge.
-------
2. CONTEXT AND STRUCTURE
OF MONITORING PROJECTS
The previous Chapter defined seven types of water
quality monitoring and discussed the development of non-
point water pollution control programs in the U.S. The
purpose of this Chapter is to show how the different types of
water quality monitoring projects fit into the general legal
framework described in Section 1.4. This functional de-
scription of water quality monitoring will then be used as the
basis for outlining the overall structure of water quality
monitoring projects.
Chapters 3, 4, and 5 expand upon the broad outlines
presented in this chapter. Chapter 3 reviews the principles
of statistical design and sampling as they apply to water
quality monitoring. In Chapter 4 more specific guidance is
provided on developing water quality monitoring projects.
This includes detailed discussions of driving factors such as
the specification of monitoring objectives, the designated
uses of the water body to be monitored, the type of man-
agement activities being carried out, and the physical envi-
ronment. Chapter 5 integrates the information from Parts I
and n of the Guidelines in order to suggest parameters for
monitoring theeffects of forestry activities on streams in the
Pacific Northwest and Alaska.
2.1 LEGAL CONTEXT OF WATER QUALITY
MONITORING EFFORTS
The most important aspects of the laws regulating
nonpoint sources of water pollution (Section 1.4) can be
summarized as follows. First, there is a broad mandate to
ensure that the designated uses of water are protected, and
to make all the nation's waters fishable and swimmable.
Second, Section 208 of the 1972 Water Pollution Control
Act required every state to establish effective Best Man-
agement Practices (BMPs) to control nonpoint source pol-
lution. The establishment and revision of BMPs for various
management activities is an ongoing, iterative process.
Third, Section 303(d) of the 1972 Water Quality Act re-
quired the states to list those water bodies that have desig-
nated uses impaired by water quality. The states now have
begun a process of establishing relative priorities among
these water bodies, developing plans for controlling the
specific types of pollution limiting the designated uses, and
implementing these plans as funds permit. To date the
primary emphasis has been on controlling point sources
through the wasteload allocation process, but for many
water bodies there is a need to incorporate load allocations
for nonpoint sources.
Fourth, the states recently have been encouraged to
develop plans to protect or restore those water bodies that
are impaired or threatened by nonpoint sources of pollution.
Funds to implement these plans have become available from
EPA undertheauthority of Section 319 of the Water Quality
Act of 1987. The fifth and final mechanism to address
nonpoint sources of pollution is the antidegradation policy,
which is designed to protect water quality when the water
quality already is higher than existing standards. It also is
being implemented on a state-by-state basis.
To a large extent the success of each of these policies
and pollution control mechanisms is dependent on moni-
toring. Water quality monitoring data and information on
the designated uses are needed to determine which policy is
to be applied, the relative priorities for action, and the devel-
opment of specific plans for remediation. High quality moni-
toring data are increasingly needed to evaluate the effec-
tiveness of pollution control efforts and thereby justify the
expenditure of public funds. There also will be an increasing
need for water quality monitoring data to compare the differ-
ent procedures for controlling nonpoint source pollution.
All these different objectives and data needs have helped
spur the observed proliferation in monitoring types and
terminology (Section 1.3). A schematic overview of how
water quality standards, BMPs, and the load allocation
process are used to control nonpoint sources of pollution is
presented in Figure 3. Figure 3 also shows how the major
-------
Initial
Inventory/Assessment
compliance
monitoring
feedback loop
Are designated uses protected AND
are water quality standards being met?
Trend and
compliance
monitoring
feedback loop
Implementation
monitoring
feedback loop
Effectiveness
monitoring
feedback loop
Collect imple-
mentation data
i
Collect imple-
mentation data
Water quality
data collection
I
implementation
monitoring
feedback loop
Collect BMP
effectiveness data:
are BMPs effective?
1
N°
Yes
Collect water quality
data: are standards
being met?
Load
Allocation
Process
No
£
Yes
Study timing and
frequency of non-
compliance
Develop and refine model
for problem constituents
Validation
monitoring
feedback loop
I
Management decisions
Collect data
on modeled
parameters
Figure 3. Flow diagram for monitoring and controlling nonpoint sources of pollution.
-------
Parti
monitoring types defined in Section 1.3 fit into these three
components of the current regulatory structure.
As indicated in Figure 3, the first step in developing a
plan to control nonpoint source pollution is to determine
whether water quality is (1) limiting the designated uses for
that water body, and (2) meeting water quality standards.
Suchan evaluation necessitates an initial set of waterquality
measurements (baseline monitoring). If the designated uses
arenot impaired andthestandardsarebeingmet,theprocedures
shown on the left-hand side of Figure 3 are applied. Basi-
cally this involves the routine application of BMPs and
regular water quality measurements, with monitoring being
an essential component of both of these activities.
The effective application of BMPs requires regular
implementation monitoring (i.e., determining that the BMPs
were applied as planned). This information must be fed
back to managers in order for them to assess whether the
BMP planning and implementation process is working.
This implementation monitoring feedback loop (Fig. 3) is a
crucial link in helping to ensure that BMPs are properly
integrated into ongoing management activities, and gener-
ally is regarded as one of the most cost-effective means for
controlling nonpoint source pollution.
Similarly, continued water quality monitoring is re-
quired to ensure that (1) the existing and designated uses of
water continue to be unimpaired, (2) the applicable water
quality standards continue to be met, and (3) there is no
degradation of water quality. In theory these three goals all
fall under the umbrella of meeting water quality standards,
but in practice these goals often must be considered sepa-
rately. According to the definitions in Section 1.3, a com-
bination of trend and compliance monitoring is needed to
achieve these goals. The trend and compliance monitoring
feedback loop on the left-hand side of Figure 3 emphasizes
that these data must be evaluated on a continuing basis, and
a degradation in water quality probably will force a change
intheproceduresbeingusedtoUrnitnonpointsourcepollution.
If the initial assessment of water quality indicates that
the designated uses are impaired, or that the standards are
not being met, the process on the right-hand side of Figure
3 is followed. Again the first management action is to
prescribe and refine BMPs, as this is the primary means to
protect water quality from nonpoint source pollution (Sec-
tion 1.4). The implementation of BMPs must be regu-
larly monitored to ensure that the observed water quality
problems arenotjustaresultofsubstandardfieldoperations.
Note that this implementation monitoring is not neces-
sarily limited to internal reviews. Many states have con-
ducted extensive project reviews to assess the actual appli-
cation of forest practice rules and BMPs. In most of these
reviews, an interdisciplinary review team has conducted on-
site evaluations of selected projects. Typically no measure-
ments of water quality are made, although the field review
team may qualitatively evaluate management impacts on
stream channels. On the basis of these field reviews, the
interdisciplinary team makes recommendations regarding
specific management practices and suggests procedures for
ensuring that these are fully implemented. Such qualitative
field reviews often are considered to be the most cost-ef-
fective means for protecting water quality, as current BMPs
are believed to adequately protect water bodies from rapid
and obvious degradation, and these reviews directly address
the problem of implementation (e.g., Idaho Dept. of Health
and Welfare, 1989;NCASI, 1988).
The second type of monitoring shown on the right-hand
sideofFigure3,continuingtrendandcompUancemonitoring,
is similar to the trend and compliance monitoring discussed
for unimpaired water bodies. Such monitoring provides the
data needed to determine if (1) water quality standards are
continuing tobe violated, (2) thedesignateduses areimpaired,
and (3) water quality is improving. Again these data must
be regularly analyzed and evaluated to determine what ad-
ditional control measures should be undertaken. The water
quality data also can help indicate the effectiveness of
BMPs in protecting water quality, and for this reason some
agenciesregard the trend and compliance monitoring shown
in Figure 3 as another type of effectiveness monitoring.
Continuing water quality problems often trigger a more
intensive review of BMP effectiveness. As mentioned in
Section 1.3, several approaches can be taken to BMP ef-
fectiveness monitoring. The simplest is a qualitative field
inspection, which can be done individually (e.g., observing
road drainage problems during storm events) or as part of a
formal review team. The review team process is similar to
that already discussed for implementation monitoring, but
for BMP effectiveness monitoring the review team also
must attempt to qualitatively assess whether proper imple-
mentation of BMPs adequately protected the water bodies
of concern.
As noted, water quality monitoring is a secondary,
broad-scale method of evaluating BMP effectiveness. By
definition this approach relies on inchannel measurements,
although additional upslope observations are needed to deter-
mine the cause(s) of any observed change in water quality.
The emphasis on inchannel measurements means that these
GztfWefc'nescanbeusedtohelpformulateplansformonitoring
the overall effectiveness of BMPs.
In some cases it may be difficult to determine the precise
cause of a particular water quality problem. Since both
water quality monitoring and the interdisciplinary review
team approach tend to assess the overall effectiveness of
BMPs, a third mode of BMP effectiveness monitoring—
evaluating individual BMPs—has been used to obtain the
necessary rigor. This mode usually involves detailed field
measurements on replicated sites. For many BMPs, mea-
surements must be made outside of the stream channel, as
these will have the necessary sensitivity to the practice
being evaluated, and be less subject to confounding factors
than instream measurements. Although evaluating indi-
vidual BMPs may be a more costly approach than monitor-
-------
CHAPTER 2. CONTEXT AND STRUCTURE
ing inchannel parameters, these more specific measure-
ments generate the detailed information needed to modify
or promulgate specific practices.
Typically several iterative cycles are needed to establish
and refine BMPs, and this is the basis of the BMP monitor-
ing feedback loop and the concept of adaptive management.
In 1972 the Clean Water Act recognized that in some
situations water quality standards would not be met, despite
the regulation of point sources and the application of BMPs.
In these cases the degradation of the designated uses is
presumed to be due to the cumulative impact of the various
sources. Control of the critical pollutant(s) then must be
addressed through the load allocation process summarized
in Section 1.4 and shown at the bottom of Figure 3.
The first step in the wasteload allocation process is to
determine how often and when water quality criteria are
being violated. These monitoring data are then combined
with land use, point source, and watershed information to
develop a water quality model for the constituents of con-
cern. This model provides the technical basis for formulat-
ing the managementdecisions needed to reduce thepollutant
load and bring the water body into compliance with the
applicable water quality standards. The entire process is
very data intensive, as it requires (1) identification of the
pollutant sources; (2) determination of the times, frequency,
and magnitudes of violation; (3) calibration and validation
of the water quali ty model; and (4) continued monitoring to
determine if management actions are effective. The diffi-
culty in applying the wasteload allocation process (e.g., Ice,
1990) is one reason why this approach has been imple-
mented for only a few water bodies in the Pacific Northwest
and Alaska.
This review of the regulatory context of water quality
monitoring illustrates the role of most of the major types of
monitoring (baseline, implementation, effectiveness, trend,
compliance, and validation) defined in Section 1.3 and the
critical need for feedback loops. The presence of these
different types of monitoring does not necessarily mean that
there should be six distinct data sets. In many cases data
from one type of monitoring can be utilized for other
purposes. For example, the data needed for trend and com-
pliance monitoring also can be used to evaluate the overall
effectiveness of BMPs, or perhaps for validation monitor-
ing. This overlap is why the monitoring types defined in
Section 1.3 are distinguished more by their specific objec-
tives than the particular monitoring technique or datacollec-
tion methods. The fact that a single monitoring activity can
serveseveral purposes also suggests thatacarefullydesigned
monitoring plan can substantially reduce the data collection
costs.
Although effective feedback loops are critically impor-
tant to the design and execution of monitoring projects, it is
remarkable how often water quality data are collected but
notanalyzed, or are utilized in a relatively superficial manner.
If the data are not being regularly analyzed, those monitor-
ing efforts by definition are only fulfilling an administrative
requirement or political need. In such cases the resources
being directed towards monitoring are being inefficiently
utilized, and the data being collected have essentially no
value because they are not being transformed into useful
information for land managers, scientists, or the public
regulatory agencies.
This suggests a failure in either defining the objectives
(Section4.1), or in designing and implementing the feedback
loops. Feedback loops need to be explicitly incorporated
into monitoring plans, and this needs to be done in several
ways and at several different levels. For example, the resource
specialist has to ensure from the start that time is allocated
for analyzing and evaluating the data. Specialists and man-
agers should recognize that data analysis actually can be a
cost-effective process, as it can improve (1) the location and
timing of data collection, (2) the choice of monitoring
parameters, and (3) the appropriateness of the monitoring
objectives. Rapid data analysis also can provide early
feedback to adjust BMPs if additional protection is needed,
or conversely identify situations where fewer controls will
suffice to protect water quality.
On a different level, the establishment of an implemen-
tation monitoring project is an excellent means to bring
together managers, operational personnel, and resource
specialists. In all likelihood this team approach will facili-
tate the development of the other feedback loops as each
member becomes more involved in the monitoring process
and works for its success.
Although feedback loops are essential to developing an
effective monitoring plan, in many cases it may be best to
first conduct a pilot monitoring project. As discussed in
Chapter 3, a pilot project provides much of the initial data
needed to define a monitoring plan that is efficient in terms
of its design and sampling procedure. A pilot monitoring
project also allows time for personnel to become familiar
with sampling devices and analytical equipment, thus im-
proving the reliability of subsequent data. A pilot project
also provides a set of test data for analysis and evaluation,
which helps clarify the linkage between the water quality
measurements and the monitoring objectives, hi short, a test
project forces one to go through each stage of developing
and implementing a monitoring plan, but without a long-
term commitment of resources. All too often a monitoring
project, once established, takes on a life of its own and is
difficult to modify even though it may not be meeting the
original objectives. A pilot project is far easier to modify
because it is conducted on a trial basis.
In almost every case the development of a water quality
monitoring plan should be considered as an iterative pro-
cess. It is unrealistic to expect that the monitoring param-
eters, sampling locations, sampling frequency, and mea-
surement techniques will be optimal from the beginning.
Each watershed and each monitoring project is different, and
in the absence of a priori information on the statistical dis-
-------
Part!
tribution of the parameters in time and space, no monitoring
program will be optimal. Also, our knowledge of monitor-
ing parameters and techniques will continue to change.
Thus one should expect to refine the monitoring program
over time as the data is collected and analyzed. On the other
hand, a change in parameters or techniques could well
preclude any statistical comparisons with earlier data—
another reason why a pilot project should be conducted
before a long-term monitoring project is initiated.
The case study of the South Fork of the Salmon River
illustrates how a long-term monitoring project has adapted
asnceds changed andadditionalinformationbecameavailable.
The common theme over the nearly 25 years of monitoring
has been to focus on those parameters that provide specific,
quantitative information on the limiting factors (spawning
andrearinghabitat)forthedesignateduseofgreatestconcern
(salmonid fisheries) at least cost (Box 3).
2.2 STRUCTURE OF A WATER QUALITY
MONITORING PROJECT
Many of the key steps for defining and implementing a
waterquah'ty monitoringprojecthavebeen identified through
the discussion in Section 2.1. Although the definition of the
specific steps in developing a monitoring project tends to
vary according to the author of the guidelines and the par-
ticular monitoring situation (e.g., Boynton, 1972), the key
steps'are as follows:
• propose—together with the managers—the general
objectives;
• define the approximate budget and personnel
constraints;
• review existing data;
• determinemonitoringparameters.samplinglocations,
sampling procedures, and analytic techniques;
• evaluate hypothetical or real data;
• reassess monitoringobjectives and compatibility with
existing resources;
• initiate monitoring activities on a pilot basis;
• analyze and evaluate data;
* reassessmonitoringobjectivesandcompatibilitywith
existing resources;
• modify monitoring project as necessary;
• continue monitoring;
• prepare regular reports and recommendations.
Figure4 is aschematic representation of these key steps,
and it also indicates 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
asdistinctand sequential as indicatedinFigure4. Decisions
made at each step often have repercussions for the entire
monitoring project, and sometimes this may force a reas-
sessment of previous steps. For example, preliminary
identification of the possible sampling locations may neces-
sitate a review of the budget constraints or the 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 4 is one approach to optimize the process
of developing a monitoring project.
The first step is to identify the general objectives, and
this is best done by the managers in consultation with the
technical staff. Once the general objectives have been deter-
mined, the approximate personnel andbudgetary constraints
must be specified in order to ensure that the subsequent
monitoring 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 em-
ployed. If past data are unavailable, change 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 participation of both the managers and the
technical staff in order to ensure that the specific objectives
are technically and financially feasible. The importance of
this interaction is often overlooked, and a failure in com-
munication 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
technical specialistdoesnotUstentothemanager.thespecialist
may design a monitoring project that will not provide the
necessary guidance for management decisions. Input from
both the managers and the specialists is needed to balance
the need for more data and the cost of acquiring that data.
Both sides also must be explicitly aware of the risks and
uncertainties associated with monitoring in a highly variable
environment (Section 3.2.3).
Often the technical specialist will need to take the lead
roleinformulatingthespecificobjectivesbecause the specialist
will be more familiar with previous monitoring efforts and
the likelyimpactsofmanagementactivities on water quality
and aquatic resources. Formulation of the specific objectives
also requires some knowledge of the fluvial systems to be
monitored (Section4.8) and the likely impactof management
activities (Section 5.1).
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, moni-
toring parameters, sampling locations, sampling frequency,
and data analysis techniques (Section 4.1).
Once the specific objectives have been formulated, the
next step is to select the parameters to be measured (Chapter
5) and set out the protocols for collecting data and analyzing
-------
Box 3. CASE STUDY: SOUTH FORK SALMON RIVER, IDAHO
The South Fork Salmon River Is located in the mountains of Idaho in the west, centra) part of the state. The 1290-
mile2 watershed is characterized by steep slopes and shallow, coarse-textured, granitic soils that are extremely
erodible. Almost all of the watershed is National Forest land. From 1945 to 1965, intensive logging and the associated
road construction and maintenance accelerated erosion in the watershed. Extreme climatic events in 1964 and 1965
flushed much of the eroded material downstream. The deposition of the sand-sized material in the main channels
resulted in extensive channel aggradation, Widespread concern about valuable salmon and steelhead fisheries led
to a moratorium on logging activities, a broad-scale watershed rehabilitation program, and a long-term program to
monitor spawning and rearing habitat conditions.
The initial monitoring program began in 1966 in the upper portions of the watershed, where most of the past logging
activity had occurred. In each of the four major spawning areas, a series of 10-20 channel cross-sections were
established to monitor long-term trends in the particle size distribution of the streambed surface. The particle size
distribution of subsurface sediments was evaluated by taking samples at 20 randomly selected sites in each spawning
area using a McNeil core sampler. Sets of five cross-sections spaced at 1 -mile intervals were usedto monitor changes
in the particle size distribution of the streambed surface in rearing areas along the main channel, A series of 24 surveyed
channel cross-sections evaluated changes in bed elevation over time: 6 of these were located in key spawning and
rearing areas, with the remainder in representative pool and riffle areas. Additional monitoring data included: large-
scale aerial photo surveys at about 5-year intervals, fixed photo points to document changes in streambed surface
conditions, surveys of the depth of accumulated sand in key pools, snorkel surveys of juvenile fish populations in key
rearing areas, and occasional water samples for chemical analysis to evaluate heavy metal outputs from mining
activities in the headwaters of the basin.
By 1978 monitoring data showed that most of the accumulated sand deposits had been flushed out of the system.
As a result, a new land management plan was developed that allowed for a cautious continuation of logging activities
in the South Fork. However, all logging was contingent upon the continued improvement offish habitat conditions as
documented by an expanded set of monitoring activities. This monitoring was a three-phase effort to: (1) evaluate on-
site effects of specific management activities; (2) assess changes in the tributary streams as a result of the renewed
logging activities; and (3) monitor changes in aquatic habitat conditions in the main stem of the South Fork of the Salmon
River. A technical monitoring committee was established to recommend procedures, evaluate results, and advise land
managers about trends in habitat quality and river conditions. The committee membership included representatives
from the two National Forests responsible for managing the watershed, the U.S. Forest Service Intermountain
Research Station, the U.S. Fish and Wildlife Service, the Idaho Department of Fish and Game, and the Idaho Wildlife
Federation.
On-site data collection was designed to assure that logging and road construction activities were performing as
planned. Areas of concentrated erosion, such as road prism failures and drainage failures, were measured to
document total soil loss. At selected sites in zero-order (headwater) basins, the amount of downslope soil movement
was estimated by the use of detention basins.
Tributary channel monitoring has included measurements of embeddedness in disturbed and undisturbed basins.
Some instantaneous discharge and sediment measurements have been made during high flow (spring snowmelt)
periods. In the mainstream the previous monitoring efforts have been continued, with additional data being collected
on embeddedness and the macroinvertebrate populations.
The technical monitoring committee has continued to evaluate the monitoring results and make recommendations
for change as necessary. For example, additional tributary embeddedness data were collected when an initial survey
suggested that the degree of embeddedness was directly proportional to the extent of road construction in the
watershed. Changes in the mainstream monitoring efforts have included modifying the macroinvertebrate sampling
locations and altering the location and sampling frequency of the surveyed channel cross-sections. Monitoring costs
have been an important constraint, and these have been taken into account in the deliberations of the technical
monitoring committee. In developing and evaluating the monitoring program, the committee also considered natural
perturbations, including an oxbow breach in the South Fork of the Salmon, which caused extensive changes in channel
morphology, wildfires, and localized floods within specific tributaries.
Sources: W. F. Megahan, Intermountain, Forest and Range Experiment Station, U.S. Forest Service, Boise, Idaho; Torquemadaand
Plaits (1988).
-------
Propose general
objectives
Define personnel and budgetary constraints
Review existing data
Identify specific
objectives
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 4. Development of a monitoring project.
-------
CHAPTER 2. CONTEXT AND STRUCTURE
field samples. Provision for outside analyses and repetitive
samples is needed for quality assurance and quality control.
The frequency, duration, and location of measurements
(Chapter 3) will be determined by the objectives and the
decisions with regard to the trade-offs among sample size,
variability, risk, and uncertainty (Section 3.4.2).
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 procedures
and attainable objectives. Problems at this stage may ne-
cessitate 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. Delaying the
final cost estimates until this step is unusual, but the advan-
tages are (1) the managers already have bought into the
monitoring project by helping to define it, and this makes it
easier to obtain thenecessary 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 4, the balancing of monitoring
needs and budgetary constraints should be a two-step, itera-
tive 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 ob-
jectives. A final synthesis occurs when the monitoring plan
has been fully conceptualized.
If at this stage the proposed monitoring plan substan-
tially exceeds the available resources, it may be necessary to
revise the monitoring objectives. Alternatively, a smaller
reduction in cost might be achieved by reducing the number
of sampling sites, reducing the number of parameters to be
monitored, or reducing the frequency of sampling (see Box
5, p. 37, for suggestions on how to reduce the cost of a
monitoring project). The danger of adjusting sampling
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 moni-
toring 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 provides more impetus to the rapid analysis of
field data, and subsequent modification of the monitoring
plan. All too often the monitoring plan is considered as a
final, fixed document, and then there is not as much incentive
to analyze the data as it is 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 improved, 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 special-
ists.
As shown in Figure 4, 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 struc-
ture or objectives of the overall monitoring project. Con-
tinued monitoring will then lead to the accumulation of data
that must be checked, stored, and analyzed. A description
of these steps is beyond the scope of this document, but data
storage andretrieval is another key aspect of monitoring that
is often neglected in the planning phase.
The final step in Figure 4 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 pre-
sented, the information will be much more widely dissemi-
nated, and this will reflect favorably on those responsible for
the monitoring project. More importantly, the data are more
likely to be evaluated by the 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.
-------
3. STATISTICAL CONSIDERATIONS
IN WATER QUALITY MONITORING
3.1 RELEVANCE OF STATISTICS TO
WATER QUALITY MONITORING
Statistics are an inherent component of nearly all water
quality monitoring programs. In most cases a precise for-
mulation of the monitoring objectives (Section 4.1) results
in a question that is best answered on a statistical basis. For
example, 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 either 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.
Other common monitoring objectives include the
characterization of a parameter (baseline monitoring), de-
termination of trends (trend monitoring), evaluation of
models and standards (validation monitoring), and assess-
ment with regard to a set standard (compliance monitoring).
Each of these requires collecting and analyzing data. Statis-
ticsprovidesthescientific basis and procedures forstudying
numerical data and making inferences about a population
based on a sample of that population (Mendenhall, 1971;
Sokal and Rohlf, 1981).
By its very nature, water quality monitoring is a sam-
pling procedure. It simply is not possible to make continu-
ous 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 character-
ize a parameter with a specified degree of uncertainty?
• How many samples are likely to be needed to deter-
mine 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 objectives, 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 therole
of statistics in water quality monitoring, or past experiences
in which the application of statistics led to unexpected con-
flict or uncertainty. Statistics can make a strong positive
contribution to water quality monitoring programs by:
. providing an overall design for collecting and analyz-
ing data;
• facilitating the precise specification of objectives,
including an explicit recognition of the uncertainty
and potential errors;
• providing a quantitative means to optimize the loca-
tion and times of sampling, and thereby reduce costs;
• providing a rigorous set of procedures for analyzing
the data collected in a water quality monitoring pro-
gram; and
. providing a quantitative basis for making inferences
about the characteristics and response of the popula-
tions being sampled.
To take full advantage of these potential benefits, those
responsible for preparing monitoring plans should consult
with a statistician both early and often. Too often a statis-
tician is consulted after the data have been collected, and the
statistician's tools are unable to salvage inconsistent
or unreplicated data.
-------
CHAPTER 3. STATISTICAL CONSIDERATIONS
These five contributions imply that statistical proce-
dures 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; statisti-
cal applications is a mechanical activity." Those respon-
sible for a water quality monitoring program still must
decide how much uncertainty can be tolerated and balance
the relative risks and costs associated with different types of
errors (Section 3.4.2). The managers 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 prin-
ciples which must be considered in developing a water
quality monitoring program. Theoverall goal is todemystify
the role of statistics and statisticians in water quality moni-
toring programs. The specific objectives are to: (1) explain
how statistical considerations should be taken into account
in designing and implementing water quality monitoring
programs; and (2) explicitly discuss the trade-offs between
sample size, inherent variability, level of significance, sta-
tistical power, and the minimum detectable effect.
Specific guidance on the selection and use of statistical
tests is not addressed in this document, as a number of texts
provide a much more extensive review of experimental
design and data analysis (e.g., Gilbert, 1987; Green, 1979;
Sanders et al., 1987; and Zar, 1984). The books by Gilbert
and Sanders are particularly noteworthy because they focus
on the statistical methods for monitoring environmental
pollution and the design of water quality monitoring net-
works, respectively. A series of papers by the U.S. Forest
Service provides a particularly clear and simple explanation
of the statistical aspects of water quality monitoring in
forested areas (Ponce, 1980a,b), but these may not be as
readily available. Two books that focus on nonparametric
statistics are recommended: the classic, easily understood
text by Siegel (1956) and the recent, more rigorous treatment
by Daniel (1990).
3.2 STATISTICAL DESIGN IN WATER
QUALITY MONITORING
3.2.1 GENERAL DESIGN AND REPLICATION
The overall design of a monitoring project is largely
determined by the monitoring objectives (Section 4.1) and
closely tied to the type of monitoring (Section 1.3). In many
cases the design of the monitoring plan will determine the
statistical procedures used to analyze the 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
sampling sites. In an idealized, simple experiment, the
experimental units would be randomly selected and half as-
signed 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 representa-
tive samples of much larger populations. Repeated mea-
surements on the experimental units generate the data used
to describe the sampled populations, and to draw inferences
about the larger population from which the experimental
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 in-
crease the use of nonparametric 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 illus-
trates severalkey elements 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 designated experimental units, and this
might be timber harvest, forest fertilization, or gravel ex-
traction. 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 changedovertimeataparticularsite. 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 experi-
mental unit, and the data collected are a sample of all possible
measurements in time. In some cases the onset of a man-
agement activity can be used to separate the data into two
groups (i.e., before and after), and one can test for signifi-
-------
Parti
cant change over time by comparing the means and vari-
ances 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, discharge, or other variables.
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 com-
plicating 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
techniques, and hereferences Harned etal. (1981) for analyz-
ing discharge-related parameters and Montgomery and
Reckhow (1984) for analyzing serially correlated data.
A completely different approach is to model the time
series sequence using the techniques developed by Box and
Jenkins (1976). This requires a minimum of 50-100 data
points collected at approximately equal time intervals, does
not allow for missing data, and is more complex than the
techniques mentioned previously (Gilbert, 1987; Mont-
gomery and Johnson, 1976).
The first and most common design to evaluate changes
over time is to monitor a single site. This approach is useful
to detect seasonal or other trends.butabasic problem is that
statistical inferences cannot be made either about the cause
of any observed change at the monitoring site or about the
cause of similar changes observed at other sites. Data from
other sites are necessary for making inferences about other
locations (Hurlbert, 1984).
The paired-site approach is the second design which
often is used to evaluate change. In this design two sites are
monitored, and a statistical relationship between the sites is
established for theparameter(s) of interest. After this initial
calibration period, one site is subjected to a treatment (e.g.,
timberharvest), and the other is leftas acontrol. A significant
change in the statistical relationship between the 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
treatmenteffectfrom 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
parameter being measured. An estimate of the variability is
necessary to make any statistically based inference about
the cause of an observed difference between the treated and
control sites. Since in most cases sites are not replicated,
claims of cause and effect must be based on other informa-
tion and not statistical testing (Hurlbert, 1984). Ideally data
should be collected to document the processes responsible
for the observed change.
Multiple pairs of treated and control sites, although
costly, usually result in a high sensitivity to change. Both
the control and treated sites are subject to the same extrane-
ous factors, so theexclusion of these factors greatly increases
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
particularprojectoractivity. However, the known differences
in water quality and stream characteristics between upstream
and downstream locations (e.g., Naiman et al., 1991) means
thatapre-projectcalibrationperiodisessentialforunreplicated
sites. As in the paired-site approach, the absence of multiple
treated (downstream) andcontrol (upstream) sites means that
theinferenceofcause-and-effectmustbebasedonquah'tative
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 statistics 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 clearcutting or road building.
Typically the experimental units and treatments) are al-
ready 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
access 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
between treatments in one or more blocks, regardless of the
variation among the different blocks. Thus 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 compari-
sons represent two of the simplest forms of a block design.
The combination of a treated watershed and a control water-
shed form one, unreplicated block. Additional paired wa-
tersheds 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 statistical inferences can be
made regarding (1) the cause of any observed differences,
and (2) the likely result of a similar treatment on other
-------
CHAPTER 3. STATISTICAL CONSIDERATIONS
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 down-
stream site usually 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
toreph'catepairswithoutanytreatmentorproject, and compare
the upstream-downstream differences between these
untreated pairs to the differences for the pairs where there
actually is a management activity. Alternatively, a rela-
tionship between each upstream and downstream location
could be established during a calibration period, and a
change in this relationship following management activities
wouldindicateatreatmenteffectratherthanasite difference.
In practice it is often assumed that natural variability
overwhelms any consistent site effect between the upstream
and downstream locations. In this case, a calibration period
mightnot 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 case study in Box 4 summarizes the objectives and
corresponding statistical design for a water quality monitor-
ing project in the Snohomish River basin (Washington).
Although this particular project is focusing on the effects of
commercial agriculture, the principles guiding the timing
and location of sampling are equally applicable to forestry
and the other nonpoint sources of pollution discussed in
these Guidelines.
The problem of separating site differences from treat-
ment differences is particularly acute for many of the
channel parameters reviewed in Part II, Sections 5.1-5.6.
Channel cross-sections, pool parameters, and bed material
particle size are all sensitive to environmental factors such
as the local geology and landfbrms (Section 4.8), and they
may exhibit considerable variation over relatively short
distances. As discussed in Section 3.3., theproblem of spatial
variation can at least be alleviated by carefully identifying
the sites to be monitored, and monitoring prior to initiating
management activities. This availability of pre-project data
is critical to inferring management effects, and also influ-
ences the choice of statistical test(s).
As indicated earlier, more complicated designs have
been developed to analyze multiple factors and the inter-
actions between them. The primary problem associated
with these designs is that the number of blocks usually is a
product of the number of differentlevels for each factor, and
the desire to examine several factors at once rapidly leads to
a large number of experimental 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., conifer-
ous, deciduous, or no tree cover) increases the number of
Box 4. CASE STUDY: DESIGN OF A MONITORING PROJECT
ON THE SNOHOMISH RIVER, WASHINGTON
In 1987 a 3-year water qualify monitoring project was initiated on the Snohomish River basin in western
Washington. The overall purpose of the project was to evaluate the effect of commercial agriculture on water quality
in the major tributaries and the main stem of the Snohomish River. The three specific monitoring objectives were to
(1) establish a baseline set of water quality data, (2) determine if there were any trends in water quality over the course
of the study, and (3) determine the effect of commercial agriculture on water quality (Luchetti et al. 1987).
The statistical design is based on a comparison of water quality upstream and downstream from major agricultural
areas. Fifteen paired sites were established, with 10 of these pairs on tributary reaches and 5 pairs located along the
mainstem of the Snohomish River. The pairs varied considerably in terms of the size of the stream being monitored
and the amount of agricultural activity between the upstream and downstream sampling locations. Measured water
quality parameters included turbidity, dissolved oxygen, nitrates, orthophosphates, and fecal coliform.
Sampling was stratified by wet (November-March), dry (March-October), and storm periods. The wet and dry
periods were sampled at 10 and 7 equal time intervals, respectively. Variations in discharge caused some of the wet
and dry period samples to be placed in a different strata than originally intended. In each year three storms were
sampled at 5 or 6 of the paired sites, and 3-5 separate water samples were taken from each site over the course of
the storm.
The design of the monitoring plan allows for statistical testing of the differences between years, seasons, stream
pairs, and location (upstream or downstream). When the monitoring is completed, data will have been collected on
the type and extent of agriculture between each upstream and each downstream location, as well as the application
of Best Management Practices (BMPs). By combining this land use information whh the water quality data, the project
expects to be able to qualitatively and quantitatively document the magnitude and impact of commercial agriculture
on water quality.
Source: G. Luchetti, Surface Water Management Division, King County, Seattle, WA.
-------
Parti
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 interac-
tions, the monitoring budget, and the statistical trade-offs
(Section 3.4.2).
3.2.2 BENEFITS OF A PROPER STATISTICAL
DESIGN
Theimportance of the statistical design can be illustrated
by an example. Suppose a land manager needed to deter-
mine if clearcutting increases the number of landslides. The
staff reviewed a recent set of aerial photos and determined
thatthere were 25 landslides on 5,000 acres of clearcuts, and
50 landslides on 20,000 acres of mature forest. Although the
number of landslides per unit area proportionally was twice
as high on clearcuts as mature forest, it is not possible to
make any generalizations or statistical conclusions because
data on the variability of thenumber of landslides on clearcut
and forested areas was unavailable. Statistical analyses
require multiple measurements 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 im-
portant 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. 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 treatments (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 for each potentially
harvestable forested unhand each clearcut unit. Since these
data represent a sample of a much larger population of
clearcut and potentially harvestable forested areas, statisti-
cal techniques can be used to determine the following:
• the mean and variance of the number of landslides in
each land use type;
• theapproximate shape of the underlying distribution of
the data (e.g., normal, binomial, Poisson, lognormal);
. the significance probability associated with the ob-
served 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 collection. For example, if the staff
thought that the difference in the number of landslides in the
clearcut and forested units was relatively large, measure-
ments mightonly be made onarandomsampleof 10% of the
units. If this sample indicated a statistically significant
difference, measuring the remaining units might be un-
necessary, and a substantial savings in the cost of the study
would be realized.
The potential benefits of an adequate statistical design
are even more apparent if there are several sources of varia-
tion. 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
procedures can be used to separate these factors. Such
information is extremely useful for developing practical
managementprocedures, such asidentifying high-risk areas
or predicting sediment input 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 (Section
3.3). 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 thedata. Continuing with theprevious 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 informa-
tion is lacking, a pilot study can be extremely helpful in
determining an appropriate statistical design, identifying
strata, and estimating sample size.
3.2.3 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 management activities,
• the possible time lag between an action and its effect
of 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 develop-
ment 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 de-
-------
CHAPTER 3. STATISTICAL CONSIDERATIONS
pendence of the parameter 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. Unfortunately
any change in monitoring sites or methodology may pre-
clude comparisons with the earlier data, thus making it
important to conduct a pilot monitoring project before
initiating large-scale or long-term monitoring activities
(Section 2.2). 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 monitoring efforts may be found in the use
of habitat types (Part II, Section 4.5.). 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 pilotprojects might be preferable to determine the
spatial and temporal variability, help identify the 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 activities from natural events.
This is best done by comparing changes in water quality at
unmanaged control sites to changes at sites with manage-
ment activities (Section 3.2.1). The unmanaged control
sites provide an index of change due to some key factors,
such as climate, and the removal of these factors increases
the sensitivity of the monitoring. Problems with this ap-
proach include the additional costs of monitoring the control
sites, and the fact that in many areas it is becoming in-
creasingly difficult to find sites that have not been subjected
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 a
weaker statistical relationship between sites, there is a
declining ability to detect significant changes due to man-
agement activities.
To a certain extent additional observations can substi-
tute for a rigid statistical design. Qualitative or quantitative
data that document the processes linking management ac-
tivities 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 pro-
cedures is the problem of overlapping activities. Relatively
few watersheds are subject to just one managementactivity.
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 inchannel 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 BMPs. This may not be important
if the monitoring objective is simply to determine the
overall condition and trend in water quality. However, the
common objectives of most water quality monitoring pro-
grams 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 measurements. Multivariate tech-
niques offer the potential to resolve the effects of several
independent (causal) variables, but their use in water quality
monitoring is hampered by the large number of variables
needing consideration, and the limited number of monitor-
ing sites that can provide the necessary dependent (effect)
data.
The fourth limitation to using standard statistical pro-
cedures is the potential lag time between a management
activity anditseffecton 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). Simi-
larly, the time lag between the detachment of a soil particle
and its delivery to the stream channel is often substantial
(Swanston et al., 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.
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
-------
Parti
management activities (e.g., Lisle, 1982; Griggs, 1988).
Statistics can be used to evaluate the likelihood of experi-
encing a certain event within a specified time period, and
this information 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 statis-
tical design may not be necessary. A source-search method-
ology 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 obser-
vations, usually in the upstream direction, in order to iden-
tify the source of potential or existing water quality prob-
lems. 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 aroad network during runoff events, for
example, can provide a useful, qualitative review of BMPs,
and indicate where road-related water quality problems are
developing. This type of reconnaissance can be extremely
cost-effective as it facilitates the early identification of
problemswithoutembarkingonacostlymonitoringscheme.
Such activities also offer the potential to resolve adverse
impacts at an early stage, and thereby reduce the costs of
repairs or future mitigation measures.
In short, there is a need to complement any instream
monitoring program with additional observations or mea-
surements. These should aim at (1) providing a direct link
between upslope or riparian management activities and
instream water quality; and (2) enhancing one's under-
standing of watershed processes. 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.
3.3 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 theuniverseofpotentialsamph'ngtimes or locations,
and then select individual times or locations for sampling
according to a random numbers table or any random pro-
cedure. 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 particu-
lar days 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 parameters, 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 may be equally efficient for
time-dependent (e.g., seasonally varying) parameters.
Theadvantages of stratifiedrandom sampling aresimilar
to the advantages of arandomizedblockdesign, 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 variability 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 (Part II, Sections 6.3 and 6.4), while discharge is
often used to establish strata for sampling sediment and the
other physical and chemical constituents of water. Stratifi-
-------
CHAPTER 3. STATISTICAL CONSIDERATIONS
cation 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 according to the position on 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 appli-
cable for specific conductance, pH, or nutrient concentra-
tions. Although the value of stratification can be statisti-
cally tested.aplotof the dataprovides 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 costs of the sampling scheme needed to
adequately characterize each stratum.
For water quality programs thedecisionregardingwhere
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 et al. (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 downstream of the activity. Minimizing the
distance between the upstream (control) and downstream
(treated) sites will help minimize confounding site differ-
ences. Trend monitoring usually is done at stable, accessible
sites on major streams or rivers.
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 monitoring sites and direct the monitoring activities
towards the problem areas.
Monitoring of channel characteristics or water quality
within aparticularbasin orregion may bestbe achieved with
a spatially 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
usedstreamclassificationsystemsat this time. Once thestream
segments have been identified, an additional stratification
into habitat units (Part n, Section 5.5) 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 parameters a general site
description, such as just upstream of a particular tributary,
usually is sufficient. In less turbulent reaches, some param-
eters, 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.
Other parameters, particularly those that pertain to chan-
nel geomorphology, may exhibit a great deal of variation
over a few meters. For these parameters 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, geomorphically determined loca-
tions 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 israndom
(Part II, Section 4.6.2).
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 combined?
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 prin-
ciples apply to these fine-scale questions of sampling as
they do to the larger-scale questions of sample location and
timing discussed above. Composite samples over time or
space can represent a substantial savings 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.
3.4 PRINCIPLES OF STATISTICAL TESTING
3.4.1 ASSUMPTIONS AND DISTRIBUTIONS
Most of the better-known statistical tests have been
developed for scalar data that follow a bell-shaped or "nor-
mal" distribution. Data with these attributes normally are
analyzed with parametric statistics. Nonparametric statistics
are applied to data which have an unknown population dis-
tribution, or that are ranked or categorized rather than
-------
Parti
measured. For normally distributed data, nonparametric
statistics are less efficient than parametric statistics (i.e.,
they aremore likely to yieldfalseconclusions)(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 distribution.
The advantage of nonparametric statistics for water quality
monitoring is that they require fewer assumptions about the
underlying distribution of the data. The focus of the follow-
ing section is on parametric statistics, but most of the
principles also apply to nonparametric statistics.
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 inde-
pendent, and
4. the errors in the data are randomly distributed.
Each of these assumptions can be tested, but only major
violations can be positively identified. Although these as-
sumptions may not be strictly true in many cases, the
relevant question is whether a violation of these assump-
tions 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 thesefour assumptions for parametric statis-
tics in the context of water quality monitoring.
1. The first step in determining if the distribution of a
data setis normal is to plotiL Data can be plotted over
time, against a controlling variable such as discharge,
or as a frequency distribution. A plot of the raw data
is important to visualize the distribution, as this allows
a quick and qualitative check for patterns, extreme
values, and obvious errors. Often a frequency distri-
bution of water quality data show 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 thedatashouldalwaysbeplotted
again, as this provides a familiarity with the data and
an intuitive check on the transformation 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 transformation and normal-
ization 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 distribution is to deter-
mine whether 2/3 of the data falls within one standard
deviation of the mean, 95% of the datafalls within two
standard deviations 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. Random sampling was discussed briefly in
Section 3.3, and it is discussed in the statistics texts
cited previously.
3. The third assumption of spatial and temporal indepen-
dence 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 day (i.e., they are serially correlated).
One could, however, randomly sample from a popu-
lation 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 (Section 3.2). 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 pre-
cipitation data for a particular day from a series of
gages. However, statistical analysis of daily precipi-
tation 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 procedures do not con-
tain systematic errors. This is done by randomly
assigning treatments to the experimental units, by
random sampling, and by careful attention to mea-
surement procedures (Sokal and Rohlf, 1981). Sys-
tematic errors can be removed if the cause can be
identified, and the removal of systematic errors is a
key procedure in Hankin and Reeves' (1988) method-
ology 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 system-
atic errors is a particularly important consideration
when analyzing trend data from a single site. Quality
assurance and quality control techniques (e.g., EPA,
1983) are an important means for reducing the pos-
-------
CHAPTER 3. STATISTICAL CONSIDERATIONS
sibility of non-random errors. Plots of the data and
residuals can help identify unusual trends, and non-
parametric tests can be used to assess whether the
errors are significantly non-random.
3.4.2 STATISTICAL COMPROMISES
Probably the most commonly asked question in water
quality monitoring is whether significant change has oc-
curred. The use of inferential statistics requires that this
question bemoreproperlyphrasedas"Canweconclude 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 Section 3.4.1, then
our ability toanswer this question depends on five interacting
factors:
1. sample size,
2. variability,
3. level of significance,
4. power(i.e.,theprobabilityofdetectingadifference
when one exists), and
5. minimum detectable effect.
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 specialistand the manager mustrealize
that there is no perfect solution in statistics, and an explicit
recognition of these trade-offs is a necessary stage in de-
signing a statistically-sound monitoring plan. The follow-
ing 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 para-
metric statistics, butthesame principles apply to all statistical
tests using both parametric and nonparametric statistics.
Sample size. The relationship between sample size, ac-
curacy, and uncertainty 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
parameter. (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 be-
cause less uncertainty is associated with the estimated
population means. Unfortunately this increased ability to
detecta difference between two populations is a logarithmic
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 increases 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. Typically 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 a 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 accept-
able, it usually is statistically advantageous to increase the
sample size.
A variety of statistical techniques can be used to esti-
mate the appropriate sample size given the statistical objec-
tives and the known variability of the parameter. Often
information on the variability of a parameter can be ob-
tained 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 objec-
tives and techniques (Chapter 2 and Section 3.2.2). 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 uncertainty of the population estimates. This
estimate is based on the minimum detectable effect for the
t-test (seep. 33 and Fig. 6) (L. Conquest, Univ. Washington,
pers. comm.).
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 standardized measure of the rela-
tive 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 tech-
niques and sampling methodology. Errors in measurement
can spring from a wide variety of sources, and these include
equipment problems, actual measurement problems, tran-
scription problems, and inaccurate data entry. Reducing
-------
Parti
these errors can be very time-consuming, but it is a neces-
sary part of the quality assurance and quality control aspects
of any monitoring project.
Theothereffectivemeans to reduce the variability of the
data is to modify thesampling scheme. If someofthefactors
causing the variability can be identified, then the data can be
more efficiently stratified. A series of statistical 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 objec-
tives. 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 downstream 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 involve-
ment of the technical staff. This has advantages for inter-
preting the monitoring results, and for designing future
monitoring plans.
Level of significance. The level of significance refersto
the probability that an apparently significant difference is
notreal but simply due to chance. This is the a value listed
in statistical tables and shown graphically in Figure 5.
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) indicates a
higher level of confidence that the difference is real, and a
lower probability that it was due to chance. The most
common ways of obtaining a stronger level of significance
are to: (1) increase the sample size, and (2) reduce the
variability by altering the measurement techniques or sam-
pling design.
Figure 5A 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 5B
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 5C is similar to Figure 5B, except that
the variability of populations A and B have been reduced
(i.e., the estimated means of the two populations have less
uncertainty). 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 ais setatO.05.but
for many water quality applications a higher a level (or
weaker level of significance) may be more appropriate. One
justification for a higher a is that an objective of most moni-
toring 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 have occurred, andrestoration 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
occurred. 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 - P).
Figure 5D illustrates that a weaker level of significance
will increase the power. If there is considerable overlap
between the two populations, a substantial increase in
power can result from a relatively small increase (weaken-
ing) in the level of significance.
An example of a situation where a weaker level of
significance might be appropriate is monitoring for bacterial
contamination in 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 than to miss a contamination episode.
Other reasons for adopting a weaker level of significance
include the high variability of water quality parameters, and
the costs associated with increasing the sample size in order
to achieve a stronger level of significance.
One should also keep in mind the distinction between
the statistical level of significance and the level of signifi-
cancerelevanttothedesignatedusesof water. A25%decline
in outmigrating salmonids may not be statistically signifi-
cant because of the large interannual variability, but a loss
of 25% of the outmigrating salmonids due to poor water
quality or habitat deterioration isaserious impairment of the
designated use for fisheries. Conversely, a level of signi-
ficance 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 understand-
ing 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 difference when in fact there is a difference
(Mendenhall, 1971). This probability is usually designated
as 1 - P. When comparing two sample means, the quantity
P is commonly known as Type II error. Type II error can be
-------
CHAPTER 3. STATISTICAL CONSIDERATIONS
A.
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
B.
Population A
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
Figure 5. 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.
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 signifi-
cance, it exhibits a similar response. An increase in sample
size usually 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 popula-
tions. This enhances the possibility of Type II error and
correspondingly reduces the power.
Minimum detectable effect (MOEX 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 vari-
ability, level of significance, and statistical power (Green,
1989). Any decision regarding the desired minimum detect-
able 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? Al-
though in some cases the answer might be that no change is
-------
Parti
c.
Population A
a = 0.05
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 Fig. 5B.
Population B
D.
Population A
A slightly weaker level of significance
substantially increases the power for
the same population means and
variances as in Fig. 5B.
Population B
Figure 5—cont
acceptable, this isnotastau'sticallyacceptableanswer because
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, quan-
titative objectives for the monitoring plan and, by implica-
tion, 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 concern
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).
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 if 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 background,
and then to evaluate the numerical change in a sometimes
highly variable parameter. For example, some states permit
-------
CHAPTER 3. STATISTICAL CONSIDERATIONS
c
o
I
1
•5
soo
250
.2 200
o
1
o 150
.2
.0
(0
E
3
1
50
150%
125%
100%
75%
50%
25%
15%
0
10 15 20
Sample size (per group)
25
30
Figure 6. Maximum allowable coefficient of variation to detect changes ranging from 15 to 150%. Rgure assumes a two-sample West
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 of L. Conquest, Center for Quantitative Studies, University of Washington).
forest harvest activities to increase turbidity by only 10 or
20% above background. In view of the temporal variability
associated with turbidity, this standardbecomes very difficult
to enforce (Part II, Section 3.2).
Figure 6 graphically shows the percent change detect-
able according to sample size and the coefficient of varia-
tion. This assumes a constant significance 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 6. In very rough terms, Figure
6 indicates that the treatment effect, expressed as percent of
the mean, will have to be somewhat larger than the coeffi-
cient 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 corre-
spondingly low coefficient of variation, and this has impor-
tant implications for selecting monitoring parameters
(Chapters 4-5).
-------
4. PRINCIPLES OF DEVELOPING A MONITORING PLAN
AND SELECTING THE MONITORING PARAMETERS
This chapter discusses the key factors that influence the
development of a monitoring plan and the selection of
monitoring parameters. Probably the most crucial of these
factors is the formulation of specific monitoring objectives,
and this is the topic of Section 4.1. Section 4.2 defines the
designated uses of water, and Table 2 qualitatively assesses
the sensitivity of the different designated uses to changes in
each of the monitoring parameters reviewed in these
Guidelines. This is followed by an assessment of the effects
of various management activities on each of the monitoring
parameters (Section 4.3 and Table 3). Sections 4.3 and 4.4
evaluate the parameters with regard to the frequency of
measurement and the cost of monitoring. The cost of
monitoring is broken into separate categories, including the
frequency of measurement, data or sample collection time,
equipment costs, and analytic costs (Table 4). Box 5 (page
37) summarizes the ways in which the cost of a particular
monitoring project might be reduced.
In many forested areas access can be a serious constraint
to the frequency andlocation of sampling, and this is discussed
in Section 4.6. The importance of existing data is reviewed
in Section4.7, and there is abrief discussion of how monitor-
ing projects might evolve as a result of accumulating data.
The final section of Chapter 4 briefly analyzes why it is so
important to understand the physical features and processes
of a watershed when designing a monitoring project.
4,1 PURPOSE OF MONITORING
The most important step in formulating a water quality
monitoring project is the initial specification of the objec-
tives. As discussed previously, the monitoring objectives
often are the primary means for distinguishing among the
seven different types of monitoring defined in Section 1.3.
Identifying the objective(s) and type of monitoring then has
implications for the type, intensity, and scale of measure-
ments (e.g., Table 1, page 8). Thus a very precise formula-
tion of the monitoring objective(s) should lead to an effi-
cient and effective water quality monitoring project. Vague
or unrealistic objectives are likely to result in monitoring
that collects unnecessary data and ultimately is unable to
answer the pertinent management questions.
Careful formulation of the objectives is essential also
because it precludes unrealistic expectations. Sometimes
technical specialists will exaggerate the importance or ca-
pabilities of a water quality monitoring project in order to
justify funding. However, water quality data often are am-
biguous, and even the best statistical test carries a certain
level of risk (Chapter 3). In the absence of a quick or definitive
answer, the land manager or decision-maker can become
disillusioned with the amount of resources required to sustain
a water quality monitoring project, and this may result in the
monitoring project being reduced in scope or even abol-
ished. This not only inhibits the accumulation of long-term
databutalsoreducesthecredibility of the technical specialist.
The resulting series of incomplete or discontinuous moni-
toring projects are self-defeating, as the data may not permit
(1) the separation of management effects from natural
variability; and (2) an understanding of the effects of rare
events, such as a 5- or 10-year storm, on water quality and
channel morphology.
The importance of properly formulating the monitoring
objective can be illustrated by an example. -A typical concern
of forest managers, regulators, and fishery scientists is
whether forest management activities are adversely affect-
ing the fish in watershed X. This general question provides
no indication as to whether the concern is directed towards
trophy-sized trout or the biological integrity of the fish
populations. A more specific identification of the designated
uses and perceived adverse effect(s) is needed to determine
whether the monitoring should focus on the number offish,
species diversity, total biomass, productivity, or condition.
-------
CHAPTER 4. PRINCIPLES OF DEVELOPING AND SELECTING
Box 5: WAYS TO REDUCE THE COST OF A MONITORING PROJECT
I. INSTITUTIONAL STRATEG;ES
1. What other agencies are currently collecting streamflow or water quality
data (e.g., municipal water companies, fish hatcheries, hydropower facili-
ties, National Park Service, etc.)? Cooperation with other agencies will
reduce your costs.
2. What data have already been collected by other monitoring projects, or by
other agencies or researchers? Historical data can be used to extend your
records and help interpret your results. Water quality data from other
projects or comparable sites can improve your sampling efficiency.
3. Whocanhelp? Considerhavingmembersofthepublicconductsomeofthe
water quality monitoring or sample collection. This may require additional
quality control measures, but should help improve communication and
understanding.
4. What are all your monitoring needs, and how might these be combined?
Consider how to integrate different monitoring activities. For example, data
from project monitoring also might be used for effectiveness, implementa-
tion, and baseline monitoring.
5. Can your objectives be made more specific? Narrower objectives will
usually reduce the scope of a monitoring project by eliminating extraneous
data.
II- METHODOLOGICAL
6,
7.
8.
9.
Which inexpensive parameters can provide data relevant to the objectives?
Some parameters are much less expensive in terms of their sampling
frequency and cost of data collection and analysis. Can these provide the
necessary data?
How much statistical rigor is required in the results? A reduction in the level
of significance, for example, often can be justified in water quality studies,
and this can reduce the frequency of sampling.
Is your sampling scheme efficient? Proper stratification of your sampling
locations and sampling times can reduce the number of samples needed to
characterize a given parameter with a specified level of uncertainty. A
review of the monitoring project with a statistician can help ensure that your
monitoring design is efficient and effective for your objectives.
Can a qualitative monitoring project achieve the primary goals? Often a
visual inspection during critical periods can indicate whether a particular
activity is generating adverse impacts. However, forthis effort to becredible
it may be necessary to involve representatives from the regulatory agencies
or the concerned public.
Under most circumstances a more precise set of objec-
tives can be defined prior to initiating a monitoring project
through discussions and qualitative investigations. If the
general question is defined as the effects of forest harvest
activities on the fish populations on watershed X, one might
first identify the fish species present and the relative impor-
tance of the beneficial uses associated with those fish
populations. Once the key species and beneficial uses have
been identified, an experienced fish biologist should be able
to qualitatively suggest what factor(s) might be limiting to
the various species (e.g., spawning habitat, winter or sum-
mer rearing habitat, or food availability). This assessment
then helps the manager to identify those monitoring param-
eters most closely correlated
with the limiting factor(s) for
the fish population(s) of inter-
est The objective of the moni-
toring program might then be
refined to a more specific ques-
tion such as: Do the forest man-
agement activities in watershed
X adversely affect the winter
rearinghabitatforcoho salmon?
This clarification of the moni-
toring objective immediately
begins to suggest that certain
habitat parameters, such as pool
characteristics and large woody
debris, might be more useful or
easier to monitor than the actual
population of over-wintering
coho.
A further narrowing of the
monitoring objectives can oc-
cur by identifying the specific
managementactivities thatcould
affect the designated uses of
concern. In the above example,
the question is what activities
might reduce the availability of
winter rearing habitat for coho
salmon. If the road network in
Watershed X is stable and well
established, the monitoring
might be directed towards
evaluating the impact of forest
harvest activities. In such cases
information on the layout of the
harvest units can be crucial to
determining whether theimpacts
on riparian zones and stream
channels willbedirector indirect
A harvest unit located well
upslope, for example, would not
be expected to directly affect
the amountorrecruitment of large woody debris.butit could
contribute sediment, which might reduce pool volumes.
Harvest units adjacent to the stream channel could more
directly influence the amount of cover and structure for
over-wintering coho. Such spatial information can be very
helpful in identifying which parameters should be included
in a particular monitoring program (e.g., large woody debris
or pool characteristics), which stream reaches should be
monitored, and what type of upslope information needs to
be collected in order to demonstrate a link between man-
agement activities and changes in water quality.
As noted earlier, the identification of the monitoring
parameters has direct implications for the frequency and
-------
Parti
intensity of sampling. If, as in this example, both large
woody debris and residual pool depth are likely to be affected
by management activities and are limiting coho salmon
populations, then these are the most appropriate monitoring
parameters, and annual measurements are likely to suffice.
On the other hand, if turbidity had been identified as a
limitingfactor.morefrequentmeasurementsoverarangeof
flow conditions would be needed. Often a reduction in the
anticipated frequency, of measurements due to a shift in
parameters will permit more sites to be sampled or allow
other parameters to be monitored at each site.
This process of specifying the objectives usually will
require more time and effort than simply initiating mea-
surements of a standard water quality parameter such as
turbidity or suspended sediment Nevertheless, the poten-
tial savhigsinmonitoringeffort,andimprovementin project
results, usually makes this front-end investment extremely
worthwhile.
Another example of a typical but unworkable objective
is "determine the effect of recreational activities on water
quality." Again this provides relatively little guidance re-
garding the parameters to be measured or the frequency and
locationof sampling. Amoreprecise definition of theactivities
potentially affecting water quality is needed to develop an
efficient monitoring project After further discussion and
investigation (e.g., the application of Tables 2-5 in these
Guidelines), the objective might be refined to "determine
the effect of overnight camping on the bacteriological
quality of streams draining the XYZ Wilderness Area."
This would yield a monitoring project that would focus on
one or two of the bacterial parameters (Part II, Section 7.1),
and measurements would be limited to a few sites during
peak recreational use.
In somecases a clarification of the objectives mightlead
to the conclusion that a water quality monitoring project is
not necessary. Implementation monitoring typically is an
administrative review and does not involve water quality
measurements. Effectiveness monitoring also may not
require any inchannel measurements if it is evaluating a
Best Management Practice (BMP), which is normally ap-
plied away from the stream channel. For example, if the
monitoring objective is to determine if water bar spacing on
skid trails is adequate to protect aquatic resources, the best
approach wouldbe to measure the sedimentandmnoff from
anumberofskidtrailswithadifferentspacing of water bars.
Measurements of suspended sediment concentrations in the
streamchannelwouldhaveamuchlower sensitivity because
they integrate numerous other factors (e.g., bank stability,
sediment storage in the channel, etc.) and are less sensitive
to the management practice being evaluated.
A useful procedure to assess theadequacy of aproposed
monitoring objective is to create a hypothetical data set
consistentwiththedesignofthemonitoringproject. If these
hypothetical data can be analyzed in a way that meets the
monitoring objective, then the monitoring plan can be
considered satisfactory. However, if the hypothetical data
are insufficient or ambiguous, then it will be necessary to
reviewthemonitoringobjective(s). The simpleactof creating
adata set and then directly relating the data to the objective(s)
can be a very powerful tool for refining a monitoring plan.
msummary,clearlyspecifyingthemonitoringobjective(s)
is the single most important step in a developing a monitor-
ing project All too often the focus is on collecting data
without due regard to the purpose for which the data is being
collected. Often the monitoring parameters are selected be-
cause they are known and familiar, rather than because they
are the most efficient or appropriate. Once a monitoring
protocol is established, institutional inertia sometimes re-
sults in its continuation regardless of whether the monitor-
ing objectives are being met.
4.2 DESIGNATED USES OF WATER
The rationale for public regulation of water quali ty is to
protect the existing and designated uses of water (Section
1.4). Although the specific designated uses vary from state
to state, they generally include agricultural use, industrial
use, public water supplies, recreational use, and the propaga-
tionof fishandwildlife(EPA, 1988). Eachstateisresponsible
for determining which designated use(s) should be applied
to the water bodies within that state. The designation of uses
is important because it determines the water quality criteria
that will be applied to that water body (EPA, 1988). Desig-
nation of coldwater fisheries as an existing or attainable use,
for example, results in much more stringent water quality
criteria than if industrial water supply were the only des-
ignated use. Establishing fish and wildlife propagation as a
designated use is particularly useful in that it helps protect
water bodies on the basis of their intrinsic value rather than
relying solely on human uses, and it often leads to a large
number of relatively stringent criteria. The general goal of
the Clean Water Act of 1972—to support and propagate
aquatic life—means that all states have a narrative water
quality standard to protect the "biologic integrity" of the
aquatic ecosystem. This, together with the anti-degradation
policy, provides a relatively comprehensive framework for
protecting water quality in forested areas (see Section 1.4).
Identifying the designated uses is another key step in
developing a water quality monitoring project. Streams
used for domestic water supplies, for example, probably
should be monitored for bacteriological contamination, but
this may be unnecessary if the only designated use is for the
propagation of fish and wildlife. Conversely, pool param-
eters and bed material particle size may be very important if
cold-water fisheries are a designated use, but are not nearly
as relevant if the only designated use is for hydropower or
public water supplies.
Table 2 presents a qualitative evaluation of the relation-
ship between the monitoring parameters discussed in these
-------
Table 2. 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 parameters
Domestic
Water quality water
parameters supply
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
Channel cross-sections
Channel 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
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
3
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
-
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
-
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
3
1
1
1
-------
Parti
Guidelines and the most common designated uses of water.
These relative values cannot be assumed to apply under all
conditions, but they provide an initial indication of which
parameters are likely to be most directly related to that
particular designated use most of the time.
The relationships suggested in Table 2 also do not mean
thattheseparametersarethemostappropriateformonitoring.
A close relationship between a parameter and a designated
use indicates that a particular parameter should be consid-
ered for inclusion in a monitoring project, but other factors
must also be evaluated. Some of these other factors include
the relative sensitivity of a parameter to both management
activities andenvironmentalfactors; the type of management
activities being carried out; the ease of measurement; the
spatial and temporal variability of the parameter; the envi-
ronmental setting; and the scale of the monitoring project
Although not all of these factors can be fully defined for
each parameter, each of these is discussed in the following
sections and in conjunction with the review of each param-
eter (Part II).
4.3 TYPE OF MANAGEMENT ACTIVITY
The type of management activity is another important
consideration in developing a water quality monitoring
plan. Each of the monitoring parameters discussed in these
guidelines has a different sensitivity to human activities.
Stream channel morphology, for example, is unlikely to be
affected by forest fertilization, but may be relatively sensi-
tive to grazing, road building, or road maintenance. Sum-
mer low flows might be increased by forest harvest, but are
relatively insensitive to most other management activities
except perhaps grazing.
Table 3 presents an empirical evaluation of the sensitiv-
ity of themonitoringparameters discussed in ihes&Guidelines
to a variety of management activities in forested areas in the
Pacific Northwest and Alaska. The focus is on forest man-
agement activities, which are separated into forest harvest,
road building and maintenance, forest fertilization, and the
application of herbicides and pesticides. Other manage-
ment activities in Table 3 include grazing, dispersed rec-
reation, developed recreation/small communities, placer
mining/sand and gravel extraction, and hardrock mining.
Although these latter activities are outside the scope of this
document, they have been included in Table 3 because they
so often occur within the same watershed as forestry-related
activities and their effect on streams must be considered
when developing a water quality monitoring plan.
The values presented in Table 3 represent a qualitative,
generalized assessment of the relative sensitivities of the
parameters reviewed in Part II to individual management
activities. Clearly there can be a great deal of variability
with regard to the absolute impact of a given management
activity, but the relative rankings should be consistent. The
suggested values also will vary according to a number of
other factors, the most important of which probably is the
environmental setting. Streams in a bedrock environment
or a steep, V-shaped valley, for example, will be much less
likely to experience changes in channel width because of
grazing or road building than streams in an alluvial setting.
The sensitivity of the parameters to aparticular management
activity also may vary according to stream size. These
factors are discussed in more detail in Section 4.8.
Thepurposeof TableS is to help select those parameters
worthy of further consideration in developing a water qual-
ity monitoring plan. An efficient monitoring plan should
focus on those parameters that are most sensitive to past and
plannedmanagementactivities. Again, however, sensitivity
to a particular management activity is not sufficient reason
for inclusion in a monitoring project One must also consider
factors such as the costs of measuring the different param-
eters, and whether the natural spatial and temporal variabil-
ity is likely to mask the effects of management. Road
building activities, for example, are likely to affect peak
flows, suspended sediment concentrations, and stream
channel morphology, but these parameters vary greatly in
terms of their ease of measurement and the time period
needed to detect significant change. The following sections
will identify and discuss these considerations in more detail.
4.4 FREQUENCY OF MONITORING
An important constraint in developing a monitoring
plan is the anticipated cost of obtaining the necessary data.
In this section the cost of acquiring data is analyzed in terms
of the typical frequency of sampling and the range of flow
conditions that need to be sampled. Section 4.5 considers
the time required to obtain a sample, the equipment required
to obtain a sample, and the cost of analyzing the sample or
field data. All of these factors must be evaluated before one
can estimate the cost of acquiring data on a particular
monitoring parameter.
As discussed in Chapter 3, the sampling frequency is a
function of the statistical objectives of the monitoring project.
Any change in the desired accuracy or reliability of the
results directly affects the sample size and the choice of
parameters. All theparameters discussed in these Guidelines
also are subject to spatial and temporal variability, and this
again affects their relative precision and ability to detect
change.
A monitoring project that is attempting to detect a
relatively small change with a high degree of certainty will
be more costly than a monitoring program with a lower
standard for identifying a statistically-significant change.
More measurements will increase the precision and hence
the ability to detect change (Section 3.4.2), but the marginal
cost and benefit of each additional measurement will vary
according to the parameter.
-------
Table 3. Sensitivity of the water quality monitoring parameters to management activities, assuming average management practices:
1 = directly affected and highly sensitive; 2 = moderately affected and somewhat sensitive; 3 = indirectly affected and not
very sensitive; 4 = largely unaffected.
Sensitivity of monitoring parameters to management activity
Forest manaqement activities
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
Channel cross-sections
Channel width/width-
depth ratio
Pool parameters
Longitudinal or
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
Harvest
1-2
3
3
3
2
2
2
4
1-2
1
1
1-3
1-3
1-3
2
2
2
2
2
2
2
2
1
2
1-3
1-3
4
1
1
2
Road build-
ing and
maintenance
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
2
3
4
3
1
1
Other management activities
Fertilizers
4
3
3
2
3
1
1
4
4
3
3
3
3
3
4
4
4
4
4
4
4
4
4
3
3
3
4
2
3
3
Applications
Herbicides
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
Pesticides
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
4
2
3
Placer
mining3
2
3
3
3
2
3
3
4
4
4
4
1-2
1-2
1
1
1
1
1
1
1
1
1
3
2
2
3
4
1
1
2
Hardrock
mining
3
1
1
3
2-3
3
3
4
3
3
3
1-3
1-3
3
3
3
3
3
3
3
3
3
4
3
4
4
3
3
2
3
Grazing
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
2
Recreation
4
4
4
4
3
3
3
3
3
4
4
3
3
4
4
4
4
4
4
4
4
4
4
3
3
3
1
3
3
1
aPlacer mining also includes sand and gravel extraction.
-------
Parti
In the first column of Table4, themonitoringparameters
are grouped according to the typical frequency and timing of
measurements. Parameters that need to be measured only
annually, seasonally, or more frequently over a relatively
short time period (e.g., daily for 2 weeks in mid-summer)
are rated as having a low sampling frequency. These
include most of the geomorphic and riparian parameters, as
well as theforest chemicals such as herbicides andpesticides.
Those parameters rated as having a high frequency of
monitoring, such as the sediment parameters, must either
be measured over all flow conditions or be intensively
monitored over a series of high flow events.
The frequency of sampling for mostof the water column
parameters cannot be easily defined because of the large
range of monitoring objectives. Low flow, baseline, or
trend data might be obtained with relatively few measure-
ments, while monitoring total nutrient loads (e.g., to protect
downstream oligotrophic lakes) requires much more fre-
quent sampling.
The second column in Table 4 indicates the flows over
which sampling should be carried out in order to properly
characterize theparameterof interest Formanyparameters,
such as those relating to channel characteristics and riparian
conditions, the measurements can be made whenever it is
practical and safe. Other parameters must be measured at
high flows, and this can be an important constraint when
access to the sampling site is difficult (Section 4.6), or when
there is no bridge or other structure from which samples can
be safely taken.
The channel and riparian parameters generally have the
lowest measurement frequency and are the least restrictive
with regard to the timing of measurements. This is due to the
fact that they are—with the exception of habitat types—not
flow-dependent Although large discharge events can have
a major effect, the parameters listed under channel charac-
teristics usually are monitored on an annual basis.
In contrast, the three sediment parameters—turbidity,
suspended sediment, and bedload—are highly dependent
on discharge. Since virtually all of the sediment transport
occursduringhighflowevents.andtherecan be considerable
variation in sediment transport within a given storm event,
frequent sampling is needed during high discharge events
(Part II, Chapter 4). A similar logic applies to the monitoring
of changes in water yield and the size of peak flows (Part n,
Chapters).
Conductivity and the bacteriological parameters also
are correlated with discharge, but they generally vary less
than the sediment parameters. The relatively consistent
inverse relationship between conductivity and discharge
means that fewer samples are needed to determine con-
ductivity as compared to bacterial concentrations. Bacterial
contamination is more variable both within and among
storm events.
The other biologic parameters—algae, invertebrates,
and fish—exhibit seasonal variation. The optimal time and
frequency of sampling will vary with location and objective,
but at a minimum the invertebrate populations should be
sampled in the spring and fall (EPA, 1989b), and the resident
fish population in winter and summer (Part II, Chapter 7).
The water column parameters exhibit considerable
variation with regard to the desirable frequency of monitor-
ing. Herbicides and pesticides are present only as a result of
man's activities, and this simplifies the process of establish-
ing a baseline or background level. The primary monitoring
objective for herbicides and pesticides is to assess the
inadvertent delivery of these chemicals into the aquatic
ecosystem. Typically the highest concentrations occur
immediately after application, and monitoring efforts are
directed towards this relatively short time period. By
predicting the average travel time from the application area
to the monitoring site, an efficient sampling scheme can be
developed. In most states only 4-8 samples are required to
monitor an aerial application of herbicides or pesticides; the
adequacy of this sample size is discussed in Part II, Section
2.6. The more persistent and mobile chemicals may have a
secondary peak associated with the first runoff event, and
this may require a second sampling period (similar to the
sampling design for forest fertilization).
The frequency of sampling for nitrogen and phosphorus
will depend upon the purpose of the monitoring and the type
of management activities. A more intensive monitoring
program may be required for forest fertilization than for
herbicides and pesticides because there are several different
pathways by which nitrogen and phosphorous might reach
the stream channel. As was the case for herbicides and
pesticides, there is an initial peak due to the directapplication
of fertilizer into the aquatic environment, and possibly a
second peak in conjunction with the first runoff event
following application. Detecting this second peak requires
access to the sampling site on short notice, or frequent
sampling using an automated sampler. The cost of monitor-
ing for this second peak can be greatly reduced by ran-
domly analyzing only a small proportion of all the samples
collected between the initial peak and the first runoff event
The sampling frequencies for dissolved oxygen, pH,
and temperature are more complex because they often
fluctuate both daily and seasonally. To obtain meaningful
data, a monitoring project must either sample over this
entire range or determine the most critical period and then
consistently sample at this time. This means that an initial
period of intensive sampling may be needed to determine
the most sensitive period(s) for a particular parameter at a
sampling site, after which the monitoring can be limited to
that particular time.
4.5 COST OF MONITORING
Other key factors in assessing the cost of a monitoring
project include the amount of staff time, funds, expertise,
-------
Table 4. Frequency and cost of data or sample collection by monitoring parameters. L = low; M = medium; H = high; V = variable;
NA = not applicable.
Parameter
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
Channel cross-sections
Channel 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
Typical
frequency
L-M
L-M
M
L-M
M
L-H
L-H
L
H
M
H
H
H
H
L
L
L
L
L
L
L
L
L
L
L
L
M-H
L-M
L-M
L-H
Flow conditions
for sampling
L
L
All
L
V
V
V
L-M
H
L
All
H
H
H
L
L
L
L
L
L
L
L
L
L
NA
NA
All
L
L-M
L
Collection
time
L
L
L
L
L
L
L
L
M-H
M-H
M-H
L-M
L
M
M
M
M
M
M
M
H
H
M
L-M
L-M
L-M
L
M
L-M
H
Equipment
costs
L
L
L
L-M
M-H
L
L
L
M-H
M-H
H
L
M
M
M
L
L-M
M
L
L
L
M
L
L
L-M
L
L
L-M
L-M
M-H
Analysis
costs
L
L
L
L
L
M
M
H
H
L-H
H
M
L
M
M
L
L-M
M
M
M
L
M-H
L
L
L-M
L
M
H
M-H
M
-------
Parti
and equipment needed to make and interpret an individual
measurement. The monitoring parameters evaluated in
these Guidelines exhibit a wide variation in terms of their
ease of measurement and in the equipment required. For
many parameters a simultaneous discharge measurement is
needed to properly interpret the data. Certain parameters
also require more expertise to collect and analyze the field
samples or data, and to interprettheresults. Hence the value
of a parameter for a particular monitoring project depends
on the availability of staff time, expertise, equipment, and
expendable funds for outside analyses.
The last three columns in Table 4 provide a qualitative
ranking of the parameters with regard to the time needed to
collect a sample, the equipment needed to collect a sample,
and the costs of analyzing the sample or the raw data. For
some parameters a range of techniques or measurements
could be used, and the table is based on the techniques most
commonly used for monitoring streams in forested areas.
Associated costs, such as the need to house and maintain
equipment, or improve access to the monitoring site, can be
significant but are not included in Table 4 because these
costs vary greatly.
For most of the physical and chemical constituents of
water, the process of collecting the sample is relatively
straightforward. However, for certain parameters, such as
phosphorus, considerable care needs to be taken to avoid
sample contamination. Equipment and analysis costs for
the water column parameters vary from virtually nil in the
case of temperature to more than $100 per sample for a
commercial analysis of pesticide or herbicideconcentrations.
The relative ease and low cost of measuring temperature,
pH, and conductivity partly explains why these parameters
are included in most monitoring programs.
Monitoring changes in water yield and the size of peak
flows can be expensive because of the need to establish and
maintain one or more stream gaging stations. A very long
monitoring period will be needed to detect changes in the
size of the peak flows with a long recurrence interval,. In
contrast, it may be possible to monitor low flows in a well-
controlled reach without establishing a continuously re-
cording gaging station.
Theprocessoftakingandanalyzingsuspended sediment
and turbidity samples is notparticularly difficult or expensive.
Accurate bedload samples are more difficult to obtain. The
mainproblems associated with measuring all three sediment
parameters are (1) the need to sample intensively during
highflows,(2)theneedtosimultaneouslymeasuredischarge,
and (3) the difficulty of safely sampling small streams
during high flow events. This last problem means that
sampling locations are limited largely to bridges or other
structures for all but the smallest (e.g., first-order) streams.
Bedload sampling carries the additional problem of how to
sample coarse (>5-10 cm) sediment
Most of the channel morphology characteristics are
more time-consuming to measure and analyze. Surveying
equipment is often required. As discussed in the previous
section, a major advantage of these parameters is that they
usually are measured on an annual basis.
The sampling frequency associated with the riparian
parameters also is relatively low in most cases. Time and
equipment needs can be characterized as low to moderate in
most cases, although they can vary considerably depending
upon the actual techniques being used.
Considerable variation can also occur in the time, equip-
ment, and expertise needed to monitor algae, invertebrates,
and fish. Generally thecostofobtainingasampleismoderate-
to-high, and considerable expertise may be required to ana-
lyze samples of periphy ton or invertebrates. The bacterio-
logical parameters more closely resemble some of the water
column parameters in that relatively little time is needed to
collect the sample, but the analysis is moderately expensive.
The sampling frequency may be relatively high depending
upon the intensity of recreational use and the likelihood of
contamination.
The differences in sampling and analytic costs shown in
Table4 are important in determining the spatial and temporal
intensity of sampling and in selecting the appropriate pa-
rameter for monitoring. Many public land management
agencies have seasonal staff who can be utilized for moni-
toring purposes, and a minimum of funds for outside analy-
ses or equipment. This predisposes the monitoring program
towards those parameters (e.g., channel characteristics) that
can be measured annually and which do not incur high
analytic costs. Private companies are often characterized as
having fewer staff available for monitoring and more flex-
ibility in contracting for outside analyses. These consider-
ations must be recognized as a possible constraint to the
developmentof an optimal waterqualitymonitoringprojecL
4.6 ACCESS TO MONITORING SITES
The ease of access to a monitoring site, particularly
during storm events, can be a controlling factor in selecting
the parameters to be monitored. As shown in the second
column of Table 4, several parameters must be measured
during high flow events. If access to the sampling site is not
possible during high flow events, or there are no structures
from which measurements can be made, this precludes the
use of those parameters. Many of the other parameters are
relatively independent of discharge and can be measured at
the most convenient time, such as during summer low flows.
To the extent that a monitoring program is based on this
latter category of parameters, access is not as important a
criterion. However, ease of access can greatly affect the cost
of a monitoring project, as transportation time is often the
most expensive component.
Automatic water sampling devices may be able to
alleviate the problem of access and sampling during high
flow events, but they cannot eliminate it. Most automatic
-------
CHAPTER 4. PRINCIPLES OF DEVELOPING AND SELECTING
samplers are unable to take more than 25-30 samples. Thus
automated samplers can greatly reduce the number of trips
needed to service a sampling site, but they cannot be left to
take samples over an entire season. In most cases they are
used to sample at specified time intervals, but particularly in
smaller watersheds the sampling frequency may not ad-
equately represent specific storm events. Intake conditions
for automated samplers will vary during a storm event, and
this can be a serious limitation to the accuracy of suspended
sediment measurements (Part II, Chapter 4). Another po-
tential problem with automated, samplers is that their limited
pumping capacity complicates their use in rivers or streams
with more than a 15-ft change in water level or stage.
Finally, automatedsamplers can only be used for parameters
that are relatively stable over time. Hence they may be
suitable for suspended sediment, but they cannotbe used for
parameters such as dissolved oxygen or pH.
As discussed in Part n, Chapter 4, a sampling scheme
based on equal volumes of discharge can greatly improve
the quality of the results. This requires coupling the auto-
matic sampling device to a microprocessor and a continu-
ously recording discharge measurement device (Thomas,
1985). Such sampling schemes are particularly helpful for
calculating total fluxes of nutrients or sediment. Total flux
data are most likely to be needed in the wasteload allocation
process (Chapter 2) and for estimating the total nutrient
loading for downstream areas.
4.7 AVAILABILITY OF EXISTING DATA
Another important consideration in developing a water
quality monitoring plan is the amount of existing data. A
major shortcoming of many monitoring projects is that they
are initiated subsequent to management activities. This
means the background or undisturbed value of a particular
parameter must be extrapolated from a comparable, undis-
turbed site, which in many cases is difficult, and one can
never completelyresolvernequestion as tothecomparability
of the sites (Section 3.2.3).
The presence of an existing data set also is critical for
putting an observed change in context. If a 50% change
occurs in summer low flows or total flux of suspended
sediment, is this a significant change that can be ascribed to
a particular management activity? Determining a manage-
ment-induced change requires one or more of the following:
(1) a demonstration that it lies outside the normal range for
mat variable, (2) a demonstrable shift in values as compared
to an adjacent basin, or (3) physical data that link a particular
practice to an observed change in the stream channel. An
analysis of change either at a particular site (e.g., trend
monitoring) or in the relationship between treated and control
sites (e.g., project monitoring) requires pre-disturbance data.
In cases when baseline or pre-disturbance data are
inadequate, the third approach—collecting additional physi-
cal data to link management effects to changes in water
quality—must be used. Often this will require measure-
ments closer to the areas where the management activities
are taking place (i.e., farther upstream or out of the stream
channel). The intent is to directly observe the management-
induced changes in erosion, runoff, or other processes, and
relate these to changes in one or more of the instream
parameters. Of course data making this linkage are useful
even when pre-disturbance or long-term data sets are
available, but they are particularly necessary when no data
exist prior to management. As noted in the introduction
(Section 1.1), upslope measurements and monitoring tech-
niques are not discussed in these Guidelines even though
some watershed information is essential to interpreting any
instream water quality data.
As data accumulate there is an increasing capability to
evaluate fluctuations and discern change. This can be a
strongargumentforcontinuinganexistingmonitoringproject,
even though the project may not necessarily be optimal in
terms of parameter selection or sampling location. Often the
best means to alter an existing project is to begin monitoring
the additional desired sites or parameters while still continu-
ing with the existing project. As data accumulate on the
additional parameters or from additional sites, it may be
possible to statistically relate these data to the original sites
orparameters. Someoftheparametersorsamplinglocations
in the original monitoring project then can be eliminated.
Clearly any alteration of a monitoring project will carry
some cost in terms of the statistical reliability of the results,
and this must be weighed against the potential benefits.
4.8 PHYSICAL ENVIRONMENT
4.8.1 ECOREGION CONCEPT
The physical environment must also be taken into
account when selectingwaterqualitymonitoringparameters
and designing a monitoring program. Forested areas in
Washington, Oregon, Idaho, and Alaska exhibit consider-
able variability with regard to their climate, hydrology,
geology, landforms, and soils. This variation is reflected in
stream channel morphology, water chemistry, runoff pat-
terns, aquatic flora and fauna, and the riparian ecosystem.
This implies that the parameters discussed in these guide-
lines also will vary in terms of (1) the values that can be
expected under undisturbed conditions, (2) their sensitivity
to different management practices, and (3) their usefulness
in detecting a change in water quality.
Resource managers have long attempted to group or
classify areas with similar conditions. The purpose of
classification is to justify the extrapolation of site-specific
data to other areas, and to define a region for comparing data
from different sites (e.g., identify strata for sampling).
-------
Parti
Numerous aquatic classification systems have been pro-
posed, with each based on a specific set of physical charac-
teristics and each having a somewhat different purpose
(e.g., Bailey, 1976; Brussock et al., 1985; Hawkes, 1975;
SCS, 1981; USGS, 1982). At present the ecoregion classi-
fication of Omemik (1987) is being applied by EPA to the
continental U.S.
Ecoregions aredefined as areas of relative homogeneity
in ecological systems or in relationships between organisms
and their environments (Crowley, 1967; Omernikand Gal-
lant, 1986). Omernik (1987) identified 76 different
ecoregions in the continental U.S. based on land surface
form, potential natural vegetation, land use, and soils. The
map by Omernik and Gallant (1986) indicates that 14
ecoregions are represented in Washington, Oregon, and
Idaho. Eight of these are both extensive in area and have
forests as theirpotential natural vegetation (Fig. 7). Ecoregions
have not yet been defined or mapped for Alaska.
Some of the specific applications envisaged for the
ecoregion concept include: (1) comparing similarities and
differences within and among ecoregions; (2) helping to
establish water quality standards in line with regional pat-
terns of tolerance and resilience to human impacts; (3)
helping to locate monitoring, demonstration, or reference
sites; (4) facilitating extrapolation from site-specific stud-
ies; and (5) predicting the effects of changes in land use and
pollution control efforts (Omernikand Gallant, 1986). These
applications are largely untested although several studies
State boundary
Ecoregion boundary
Figure 7. Ecoregionsof Idaho, Oregon, and Washington. Numbers correspond to original numbersas designated byOmernikand Gallant,
1986. Asterisks indicate ecoregions with forests as their potential natural vegetation.
* 1. Coast Range 12. Snake River Basin/High Desert
* 2. Puget Lowland 13.
* 3. Willamette Valley *15.
• 4. Cascades 16.
* 5. Sierra Nevada *17.
* 9. Eastern Cascades Slopes and Foothills
10. Columbia Basin
*11. Blue Mountains
Northern Basin and Range
Northern Rockies
Montana Valley and Foothill Prairies
Middle Rockies
18. Wyoming Basin
19. Wasatch and Uinta Mountains
-------
CHAPTER 4. PRINCIPLES OF DEVELOPING AND SELECTING
have found a correspondence between ecoregions and spa-
tial patterns in water chemistry and fish distributions in
small streams in Arkansas and Ohio, respectively (Rohm et
al, 1987; Larsen et al., 1986).
A recent study of 49 small streams in Oregon used
statistical techniques to group physical habitat data, water
quality data, and data on fish, invertebrate, and periphyton
assemblages (Whittier et al., 1988). The greatest differ-
ences were between montane and nonmontane regions.
Differences among the montane ecoregions were character-
ized as subtle. None of the data sets were able to distinguish
all of the ecoregions included in the study, and the identified
clusters were not consistent across data sets. Nevertheless,
the results were regarded as providing support for the
geographic classification of streams (Whittier et al., 1988).
4.8.2 CLIMATIC CONSIDERATIONS
Climate is a key driving force for runoff patterns and
most of the erosion and sediment transport processes oper-
ating in a watershed. Different climatic regimes, when
coupled with differences in factors such as the vegetation
and geology, result in very different landscapes and land-
scape dynamics. This is the basic premise behind classifi-
cation systems such as the ecoregion concept It is impor-
tant,however,nottorelyonthemappedecoregion boundaries
as thebasis for parameter selection anddata comparisons, as
ecoregions are generalized concepts which may not be ap-
plicable to specific sites. A preferred approach is to evaluate
how the climatic regime affects the relative rates and im-
portance of individual processes, andusethis understanding
to devise an appropriate water quality monitoring project.
Even though these Guidelines focus on monitoring pa-
rameters in the stream channel and riparian zone, a broader
perspective is needed to interpret the data. The delivery of
water and sediment to the stream channel is controlled by
upslope processes, and climatic factors play a large role in
determining the relative importance of the different pro-
cesses. Climate plays an equally important role in defining
some of the critical instream processes, such as the sediment
transport capacity. Unfortunately direct relationships be-
tween climate and specific parameters, such as bed material
particle size or width-depth ratios, usually cannot be drawn
because of the influence of other factors such as stream
gradient, stream size, and geology. Nevertheless, climatic
conditions must be taken into account when designing a
water quality monitoring project, and some of the more
important considerations are discussed below.
First, the variability in precipitation is inversely propor-
tional to average annual precipitation. Drier areas, such as
eastern Oregon, eastern Washington, and southern Idaho,
typically have more year-to-year variability in discharge,
and for rain-dominated climates this implies that drier areas
will have more temporal variability in sediment concen-
trations and sediment transport rates. Second, areas with
more rainfall, such as the Olympic Peninsula, tend to have
lower concentrations of nutrients and other dissolved ions.
Third, higher discharges result in greater dilution, and this
may make changes in concentration due to management
activities more difficult to detect In areas with a high annual
discharge, a small change in nutrient concentration could
result in a very different total flux, and this could be critical
for downstream oligotrophic lakes.
The variation in the intensity and amount of precipita-
tion is a key factor in determining the types and rates of
sediment input into the stream channel. A variety of sediment
production and delivery processes have been observed in
the Western Cascades (e.g., Swanson et al., 1987), but there
is little quantitative data on their relative importance. Mass
failures are believed to be a dominant source of sediment for
streams in steep, forested lands, and these usually are
triggered by extreme storm events.
The type of precipitation can also be important. Forest
harvest can significantly increase the size of the larger peak
flows in areas subject to rain-on-snow events, and forest
harvest can also increase the peak runoff rates during spring
snowmelt (e.g., Troendle and King, 1985). In contrast,
forest harvest in rain-dominated areas may not increase the
larger peak flows (e.g., Wright et al., 1990). Hence knowl-
edge of the climatic regime and the causes of flood events is
essential to predicting the effect of forest management
activities on peak flows of varying return intervals (Part n,
Section 3.1). Such predictions are needed to determine if
one should attempt to monitor changes in the size of peak
flows.
A change in the size of peak flows also has implications
for the stability of the banks, bed material, and large woody
debris. Each of these channel features then has implications
for the biotic component.
Other aspects of the climatic regime should influence
the selection of monitoring parameters. In areas with a cool
climate, for example, it may not be necessary to monitor
changes in water temperatures due to forest harvest.
Climate also has a series of indirecteffects on the various
water quality monitoring parameters. These are too exten-
sive to be detailed here, but they include effects on weath-
ering rates, erosion rates and hence slope steepness, vegeta-
tion type, and productivity of the aquatic biota. The most
important of these are discussed in the sections reviewing
the individual parameters (Part IT).
4.8.3 LAND FORM
Land form refers to the shape of the terrain and the
pattern of the drainage system, and it is an important
consideration when developing a monitoring project. In
particular, the steepness of the sideslopes is a major factor
in determining which runoff and erosion processes are
-------
Parti
likely to dominate, and in determining the rate at which
water and sediment will be delivered to the stream channel.
Stream gradient is one of the most important factors for de-
termining the rate at which water, sediment, and large
woody debris are moved downstream. Both the movement
of material into the stream channel and the movement of
material downstream directly affect monitoring parameters
such as bank stability, channel morphology, turbidity, bed
material particle size, and dissolved oxygen.
Areas with steep terrain generally will have more dy-
namic, high-energy streams. Steeperbasins tend to produce
more sediment (Swanson et al., 1987), with mass wasting
being of primary importance. In these basins there typically
is less of a lag between the production of sediment and its
delivery into the stream system. These characteristics sug-
gest that the steeper the basin, the greater the response to
management activities (Swanson et al., 1987). This larger
response to management activities does not necessarily
make it easier to detect change through water quality moni-
toring. In steep basins the fine sediment will not be stored
in the stream channel, and this could limit the ability of some
of the channel parameters, such as embeddedness or bed
material particle size, to indicate change. High gradient
streams also are more likely to have cut down into bedrock,
and this restricts the amount of change that could be ex-
pected in parameters such as thalweg profile, habitat types,
and channel cross-sections.
Steep land forms generally are more susceptible to
extreme events such as landslides and debris flows. Recent
studies in the western Cascades suggest that sediment input
and channel morphology often are dominated by relatively
rare events (L. Benda, Univ. Washington, pers. comm.). If
the basic pattern is one of severe disruption followed by a
long period of recovery, it may be difficult for monitoring
projects to detect the superimposed impact of management
activities (e.g., Lisle, 1982). In areas of steep terrain it may
bepreferable to monitor upslope characteristics, such as the
frequency of landslides in cut areas or along roads, rather
than relying on instream measurements.
4.8.4 GEOLOGY AND SOILS
Geology is another important factor in determining
landforms, stream characteristics, and soil types. The per-
meability, depth, and porosity of the soil and bedrock are
critical to characterizing the runoff processes in a water-
shed. Soil androcktypesaffectthetypeoferosionalprocesses
and the rate of sediment delivery. Geologic features can
control stream gradient and channel morphology.
An understanding of these geologic considerations is
necessary for the proper design of monitoring projects.
Bedrock channels are unlikely to show much change in
channel morphology or habitat types as a result of manage-
ment activities. Turbidity may not be as useful a parameter
in areas dominated by coarse-grained sediment Con-
versely, bedload measurements may be difficult or rela-
tively unimportant in streams with beds comprised of silt or
other fine particles. Bedrock outcrops may control stream
gradient in a particular reach, and this will reduce the
sensitivity of the stream to deposition or erosion.
Background levels of nutrients and dissolved ions also
are highly dependent on bedrock type and soil depth. The
texture, depth, and permeability of the soil will influence the
proportion of fertilizers, pesticides, and herbicides leached
into the stream channel immediately after application and
during the first runoff event Coarse-grained soils tend to
have higher permeabilities and less ability to capture and
hold dissolved ions.
Soil texture is a major factor in assessing its susceptibil-
ity to compaction and surface erosion. Relative soil and
bedrock permeabilities, among other factors, strongly influ-
ence hillslope drainage. High soil water contents and excess
pore pressures are critical contributing factors to mass fail-
ures. Areas with frequent mass failures are difficult to moni-
tor because these events tend to overwhelm the changes due
to forest management activities.
4.8.5 SUMMARY
The physical factors discussed in the previous sections
are only some of the more important considerations that
must be taken into account when developing a water quality
monitoring plan. The point is that one must be aware of the
dominant physical and biological processes operating in the
upslope areas and in the stream channel, and use this infor-
mation to identify those parameters that are more likely to
be affected by management and less likely to be subjected
to extreme fluctuations by extraneous events. This knowl-
edge is also necessary to relate changes observed in the
stream channel to management activities.
Probability also plays a strong role in water quality
monitoring. A monitoring plan implicitly assumes average
conditions, and extreme events such as a 100-year flood, a
largedebris flow, or a volcanic eruption can disrupteven the
most carefully designed monitoring project. The technical
specialist designing the monitoring project must qualita-
tively assess the probabilities of different natural distur-
bances and structure the design of the project accordingly.
As discussed in Chapters 2 and 3, any monitoring project
must have (1) a procedure for regular data analysis and
interpretation, and (2) flexibility to adapt as new informa-
tion is acquired. If just these two components are actively
incorporated into the monitoring project, then there is an
excellent chance that the project will be successful.
-------
5. PARAMETER RECOMMENDATIONS
AND INTERACTIONS
5.1 RECOMMENDED PARAMETERS
The preceding chapters have shown that the selection
and use of monitoring parameters depends upon a wide
range of factors. The key factors were identified and
discussed in Chapter 4, and the choice of parameters can be
summarized by the notation:
Selected parameter(s) = function (objectives,
designated uses, management activities, cost, and
environmental setting).
Of these five controlling factors, the specific monitor-
ing objectives were singled out as the most important factor
in determining the parameters to be monitored. Specifying
the objectives largely controls the frequency and location of
sampling (Section 4.1), as well as the level of certainty and
accuracy desired in the monitoring program (Chapter 3). It
should be noted that the level of certainty and accuracy
specified in the objectives also affects both the choice of
parameters and the cost and location of sampling; this is just
one example of the important interactions among the five
key factors identified above.
The sensitivity of the designated uses of water to changes
in the water quality parameters is discussed in Section 4.2
and qualitatively evaluated in Table 2. Table 3 ranks the
sensitivity of the water quality monitoring parameters re-
viewed in these Guidelines to the most prominent human
activities in forested areas in the Pacific Northwest and
Alaska. The intent of these two tables is to provide an initial
screen for identifying those parameters most important for
protecting a specific designated use (Table 2), and those
parameters most likely to be affected by a particular man-
agement activity (Table 3).
Other key considerations in developing a water quality
monitoring program are monitoring costs, the availability of
existing data, access to the monitoring sites, and the physical
environmentthe monitoring site. Table4providesageneral
rating of both the sampling frequency needed to evaluate
each parameter, and the flow conditions under which a
particular parameter typically needs to be sampled. The
second part of Table 4 indicates the relative cost of measur-
ing a parameter, and this was broken down into three
components: the time needed to collecta sample or fielddata,
the equipment needed to collect a sample or field data, and
the cost of analysis. Again this information is intended to
serve as a screen or filter for selecting the most appropriate
monitoring parameter for a particular situation.
Some of the factors that affect the selection of monitor-
ing parameters cannot be summarized in tabular form. The
physical environment affects both the spatial and temporal
variability of a parameter, and the sensitivity of different
parameters to specific management activities (Section 4.8).
Hence consideration of the physical environment must be
done on a case-by-case basis, and this requires a sensitivity
to, and knowledge of, watershed processes.
The importance of these case-specific factors means
that these Guidelines cannot specify which monitoring pa-
rameters are most appropriate under all conditions. Nev-
ertheless a combination of Tables 2-4 and the distilled
experience of those involved in this project provides a
qualitative indication as to the monitoring parameters most
likely to be useful most of the time.
Table 5 integrates all this information into a qualitative
ranking of the usefulness of each parameter evaluated in
these Guidelines to monitor the water quality effects com-
monly associated with various management activities. For
the purposes of this table "usefulness" is based on the (1)
sensitivity of a parameter to the specified management
activity; (2) importance of a parameter with regard to the
designated uses characteristic of forested areas in the Pacific
Northwest and Alaska; and (3) typical costs of measurement
and data analysis, including consideration of the sampling
frequency and time needed to detect change. A quantitative
-------
s.i
•- =
cB S>
"S CD
if
.o s
8«
•-s E
s j
2 o
S|
il
m
25
S =
° •*•
75 ir
jg CD
0 f= •
CO fc
55 H ;
S <>>
•3 =-
sSf
"o S rt
S>^E
» SS
©
o I1
11
^- ^
CD f2>
o .E
« c
-
-
CD 0
02
gs.
•o c
O -Q
CO O
5 £
'a
o
C
^^i
OC 2 c
"5 "to
^3 E
Parameters
eo •* eo
CMCOWCOCOCMCMCOCO
CMCOCOOJCOCM-r-' jo
co S <2 =
e|o|f
--m IS
c c^S.31
CO (0
-------
ble 5— cont
£
CO
iS
1
CD
E
CD
|
CO
E
"o
"o
CD
k.
0
E
82
CD
ID
E
(0
g.
•5
CO
CO
CD
_c
3
"CD
co
co
CD
•|=
CO
•s
CD
Q)
O)
CO
CO
E
CD
.c:
5
1
->
't3
CO
'c
CD
CD
9
c
10
1
o
-g
s
CO
a:
CD
CO
Q.
T3 >
||
<3) O
Q 2
•§
2
CD
Q.
.CO
CO
.O
0
"o.
Q.
«t
•o
•oS
CO 0>
t3C ^
'5
_Q
D)
"c
"E
I1
"c
"E
CO
c
.0
Q.
8.
E
.0
to
£
o
2
'a
0
CD
T5
'o
to
CD
a.
co
2
2
CD
2
CD
N
•c
i£
8
CO
O
•E
'co
E
ti)
i
CO
z:
irameters
CD
Q.
CM • CM CO CO^CO COCO '*COCMCO
CO COCO -*COCO COW T- CO CM •*
T'T'^r ^r^-w 'TCM T-ww1*
CM CMCM CO-^-i- CMi- i-CMCMCM
t -t t t ^ W CMCM •* CM CO W
CO CMCM CMCOCM CMCM ^t-^CMCO
"c
YCO CMCO COCMCM CMCM ^tCOCMCO
j
w
1 •§ -s
1 I 1 1 ^ _l 1
ill SiiL
OXCQ JjCQ ITO>
-------
Parti
procedure was not used in preparing Table 5 because each
factor integrates several sub-factors, and not all the sub-
factors are amenable to a numerical ranking. In the follow-
ing sections the results and rationale are summarized for
each of the major management activities listed in Table 5.
5.1.1 FOREST HARVEST
The cutting and yarding of trees affects streamflow and
runoff patterns, disturbs the soil and exposes it to erosion; it
also can decrease slope stability and alter the inputs of
organic material and light into the stream system. This
rangeofpotential effects suggests thatnumerous parameters
could be used to monitor the water quality impacts of forest
harvest. However, each potential parameter has a unique set
of advantages and disadvantages.
As discussed in Chapter 2 of Part n, forest harvest is
known to alter water chemistry, but these changes generally
are not large enough to limit the designated uses of water. A
further disadvantage of monitoring water chemistry is that
observations need to be made over a range of flow condi-
tions, Intergravel dissolved oxygen (DO) is likely to be one
of the most useful of the chemical and physical components
of water if fine sediment is a concern.
A large number of paired-watershed experiments have
shown thatforestharvest usually increases total wateryield,
increases the size of the smaller peak flows, and increases
summer low flows. Detecting these hydrologic changes
requires several years of data and—with the exception of
summer low flows—considerable effort. Assuming that
large areas of the catchment are not compacted (e.g., no
more than 15%), the potentially most significant change is
in the size of peak flows in areas subject to rain-on-snow
events. However, change in the size of the larger peak flows
(e.g., events greater than the mean annual flood) areprecisely
those changes thatare most difficult to detect because of the
need to measure and compare infrequent peak flows.
Absolute changes in the rate of sediment transport are
difficult to measure because of the need to intensively
sample high flow events and the difficulty of obtaining
accurate results. This makes trend or validation monitoring
a difficult objective. Suspended sediment and turbidity
monitoring may be more successful if done on a compara-
tive rather than absolute basis. A typical example is a
comparison of turbidity levels upstream and downstream of
a particular activity (e.g., project monitoring). To be sta-
tistically valid, such comparisons must be replicated and
include either a pre-disturbance calibration period or a
comparison to other paired sites thatare not treated (Section
3.2). Intergravel DO may be a very useful surrogate for fine
sediment as it is often easier to measure.
Most channel characteristics are easier to monitor be-
cause measurements typically are made only on an annual
basis. They also can be directly related to one of the most
important and restrictive designated uses, namely habitat
quality for coldwater fisheries. These channel parameters
have two primary limitations: (1) They can be highly
variable within a given reach, and (2) their relative response
to forest harvest and natural events in different environ-
ments is still difficult to predict and distinguish. The case
study from the Little North Fork of the Clearwater River in
Idaho indicates that cobble embeddedness was selected as
the best parameter for monitoring the effects of forest
harvest and fire on fish habitat (Box 6).
Stream temperature and the riparian vegetation can be
very useful monitoring parameters if forest harvest extends
into the riparian zone. Both of these are relatively sensitive
and easy to measure. Evaluation of the ripariancanopy opening
using a RAPID-type technique (Grant, 1988) can be very
useful to quickly assess current condition over a large area,
but it is not advocated as a monitoring technique because of
its lower sensitivity and the lag period between management
activities and observed change (Part II, Section 6.1).
Of the biologic parameters, the macroinvertebrate com-
munity probably offers the greatest promise for monitoring.
Periphy ton also might be effective for monitoring purposes,
but samples are more difficult to collect and analyze. Cer-
tain fish species can be very sensitive to forest harvest
activities, but difficulties in measurement techniques and
the presence of confounding factors may make it difficult to
directly link management activities to the particular fish
parameters) being monitored.
In summary, temperature and riparian vegetation are
likely to be among the most useful monitoring parameters if
the forest harvest activities are near enough to the stream
channel to affect stream shading and the input of organic
materials. The usefulness of turbidity or suspended sediment
will depend on the objectives of the monitoring. Intergravel
DO and some of the channel characteristics should be
considered as indirect indicators of changes in upslope
erosion, sediment transport, and runoff. Invertebrate moni-
toring can serve as the means to link the physical changes
(temperature, turbidity, and channel morphology) to the
biological integrity and designated uses of the stream. Less
frequent measurements of large woody debris are useful to
evaluate trends in this component of fish habitat and to help
assess the adequacy of the silvicultural prescriptions to
maintain the input of large woody debris to streams.
In general the effects of forest harvest are proportional
to the percent of the area disturbed and the percent of the
vegetation removed. Other factors that can ameliorate or
exacerbate the effects on water quality include the yarding
system employed (e.g., skyline, tractor, or helicopter), the
location of the harvest units with regard to ephemeral and
permanent stream channels, the pattern of harvest within the
catchment, the sensitivity to mass failures, and the particular
climatic conditions following harvest
Clearcutting normally has a more severe impact than
selection cutting per unit area, as all the trees are being
removed. However, selection cutting has to disturb a larger
-------
CHAPTER 5. RECOMMENDATIONS AND INTERACTIONS
Box 6*. CASE STUDY; SEDIMENT MONITORING IN THE LITTLE NORTH FORK
OF THE CLEARWATER RIVER, IDAHO
The Little North Fork of the Clearwater River is a 30-mile fong watershed of mixed federal, state, and private forest
lands in northern Idaho. Although much of the watershed has been subjected to extensive road building and logging
over the past several years, some sub-drainages are still pristine roadless areas.
Fish species in the Little North fork are predominantly west slope cutthroat trout and native rainbow trout, with some
Dolly Varden and eastern brook trout. During the fall kokanee trout from Dworshak Reservoir spawn in the Little North
Fork and its tributaries. Though fish population data are scarce in this region, there has been growing concern that
populations are decreasing and species composition is shifting to more sediment-tolerant species.
Much of the watershed has a high erosion hazard due ,to steep slopes, unstable granitic soils, and a tendency for
mass-wasting. The two primary geologic materials in the basin are the Idaho Batholith (a large igneous intrusion) and
a Pre-Cambrian metamorphic mica schist.
Before this project little information was available on the sediment conditions of streams in the watershed. Some
streams appearto have large quantities of sediment, but there is no documentation of stream sediment conditions prior
to the recent management activities. The Idaho Division of Environmental Quality (DEQ) saw a need to determine
baseline (i.e., pristine) sediment levels for the tributaries and main stem of the Little North Fork of the Clearwater River.
This baseline data was to be complemented by an evaluation of sediment quantities in various tributary watersheds
with differing amounts of erosive soils, burned areas, and forest management activities.
Cobble embeddedness was chosen as the best parameter to quantify instream sediment. During 1988 and 1989,
cobble embeddedness was measured in 22 stream reaches, and the average number of hoop samples per reach was
18. Measured cobble embeddedness varied from 22-93%.
In cooperation with DEQ, the Idaho Department of Fish and Game began inventorying habitat conditions and fish
populations in the Little North Fork and its tributaries in 1990. The resulting data will be used to evaluate the relationship
between fish and sediment as measured by cobble embeddedness and habitat conditions.
Idaho DEQ is now inventorying the percentage of erosive soils, acres harvested, miles of road, and burned acres
within each drainage. These factors then will be related to the cobble embeddedness data. In future years trends in
cobble embeddedness will be monitored throughout the Little North Fork Clearwater River.
Source: Jack Skills, Division of Environmental Quality, Idaho Department of Health and Welfare, Coeur d'Alene, Idaho,
area in order to obtain an equivalent volume of wood.
Commercial and precommercial thinning generally have
the least impact. The similar effects of these different forest
harvest operations suggest that the parameter recommenda-
tions apply equally well to clearcutting, selective harvest,
and thinning, even though the relative magnitude of the
water quality impacts may differ.
Site preparation is not explicitly considered in Table 5
because it usually follows immediately after forest harvest,
and its water quality impacts are both highly variable and
difficult to separate from the water quality impacts of forest
harvest Extensive brush raking or other mechanical soil
disturbance can result in site prep having a more detrimental
effect on water quality than the actual cutting and removal
of trees. Because site prep usually has agreatereffecton soil
disturbance than vegetation removal, sediment-related pa-
rameters generally will be of primary importance.
5.1.2 ROAD BUILDING AND MAINTENANCE
The primary concern associated with road building and
maintenance is the increased rate of erosion. Often this is
bestmonitored in thesmaller ephemeral channels thatdirectly
drain the road prism because the relative effects are much
greater than in the higher-order downstream channels.
Measurements of erosion and sedimentation can be direct
(e.g., suspended sediment, thalweg profile) or indirect (e.g.,
intergravel DO). Other potential water quality problems
due to road building and maintenance include higher con-
ductivities attributed to road salting, the runoff of fertilizer
and herbicide residues from cut and fill slopes, and the input
of sand applied to improve winter traction into stream
systems.
The selection of parameters to monitor sediment from
road building and maintenance presents the same Hobson's
choice as described for monitoring sediment from forest
harvest activities. The channel morphology parameters
represent simple techniques that are still being proven.
Monitoring of turbidity and suspended sediment concen-
trations requires relatively intensive storm sampling.
Intergravel DO may not pro vide a direct link to management
actions. The riparian parameters are more appropriate for
evaluating long-term change than short-term water quality
effects. Aquatic macroinvertebrates probably are the best
choice among the biological parameters, but they may not
be as sensitive as some of the physical parameters.
-------
Parti
These considerations suggest that a project to monitor
the water quality effects of road construction and road
maintenance might best rely on some of the channel charac-
teristics and direct observations on or adjacent to the road
network during storms. With regard to the individual
channel characteristics, sediment deposition might be
monitored by measuringresidual pool depth, pool volumes,
or longitudinal profiles. A change in the balance between
sediment inputs and sediment transport could be inferred by
monitoring changes in the bed material particle size. Mea-
surement of deposition and scour, and changes in the bed
material particle size, should provide a reasonable assess-
ment of whether and how adverse effects might be occur-
ring. Turbidity or suspended sediment observations on the
water draining from the road network during storm periods
could provide direct evidence for any observed changes in
thelarger stream channels. Monitoring aquatic invertebrates
might provide useful supplemental data and help establish
the link between sediment and aquatic organisms.
5.1.3 FOREST FERTILIZATION
Selecting parameters to monitor the application of fores t
chemicals is much simpler because of the very specific
impact of the chemicals on water quality. Most forest
fertilization programs in the Pacific Northwest apply only
organic nitrogen or urea (Gessel et al., 1979). Monitoring
the effects of forest fertilization on water quality is best
achieved by taking water samples after the application of
fertilizers and then during the first runoff event following
fertilization. These samples should be analyzed for the
major forms of nitrogen. Discharge data is required if
concern exists over the total flux of nitrogen as well as the
maximum concentration. Total flux is essential when there
is a desire to minimize eutrophication. Simultaneous tem-
perature measurements may be helpful as an indicator of the
rate of biological activity during the period of increased
nutrient availability due to forest fertilization. Suspended
sedimentdata may beusefulbecauseadsorbed nutrients can
be a substantial component of the total nutrient budget
The indirect and biological effects of fertilization can be
best monitored by measuring the algal community (e.g.,
biomass, cholorophyll-a, or growth). In slower-flowing
streams it may be possible to detect substantial changes in
the algal community by measuring daily fluctuations in pH.
5.1.4 APPLICATION OF HERBICIDES AND
PESTICIDES
The most efficient and direct approach for monitoring
the water quality effects of herbicide and pesticide appli-
cations is by taking and analyzing water samples. Although
the analytic costs may be relatively high, in most cases only
a few samples need to be taken immediately after applica-
tion.
The potentially high analytic costs can be reduced in
several ways. One approach is to first analyze a composite
sample, and then analyze individual samples only if the first
testindicatessignificantcontamination. A second approach
is to mix a dye tracer with the chemical and use this to
indicate which samples should be analyzed. Finally, trace
enrichment cartridges can be used to estimate the total flux
over the period of sampling (NCASI, 1984). For the more
persistent and mobile chemicals, additional sampling dur-
ing the first runoff event may be necessary.
Again temperature measurements may be helpful as an
indicator of chemical reaction rates and biological activity
during the period of monitoring. Discharge measurements
are needed if total losses to the aquatic system are being
estimated. Suspended sediment data are necessary if the
adsorbed component is of concern.
For herbicides, an alternative approach is to monitor
changes in the canopy cover or some other aspect of the
riparian vegetation. Generally such measurements will be
a less sensitive indicator because the riparian vegetation will
respond only to acute doses of herbicides. On the other
hand, monitoring the riparian vegetation avoids the problem
of capturing a transient peak concentration in the stream,
and changes in the riparian vegetation can be directly linked
to several designated uses.
Similarly the effect of pesticide applications on water
quality can be monitored on a coarse scale by sampling the
aquatic invertebrate populations. A disadvantage of using
the riparian vegetation to monitor herbicides, and aquatic
invertebrates to monitor pesticides, is the ubiquitous prob-
lem that an observed change may be due to other factors.
5.1.5 GRAZING
Of all the management activities considered in Table 5,
grazing has the widestrangeof water quality effects. Grazing
can affect water quality by changing the pattern and timing
of runoff, and increase sediment loads by removing the
vegetative cover and trampling the streambanks. Animal
wastes can directly impair water quality through bacterial
contamination and increasing nutrient levels. This range of
effects means that almost all of the parameters discussed in
these Guidelines could be used for monitoring grazing
impacts on water quality.
The choice of monitoring parameters in a particular
situation will depend on the designated uses as well as the
intensity and pattern of grazing. Bacterial contamination is
important if domestic water supply or recreation is a desig-
nated use. Nutrients will be more critical if eutrophication
of downstream water bodies is a concern. Bank stability is
less likely to be a useful monitoring parameter if theriparian
areas are fenced off or if there are sufficient watering points
away from the stream channel.
In general, however, livestock tend to congregate in
riparian areas. In these cases bank stability and the riparian
-------
CHAPTER 5. RECOMMENDATIONS AND INTERACTIONS
vegetation are two parameters that can be quickly assessed
and are directly affected. These two parameters may not
necessarily be the most sensitive to grazing effects over an
entire watershed, but they are often among the first to be
affected because grazing tends to be concentrated in the
riparian zone. Theotherchannelcharacteristicthatmightbe
particularly useful to monitor grazing impacts is thechannel
cross-section, as this can provide data on bank slope, chan-
nel aggradation, and stream widening.
Either phosphorus or nitrogen can be used to monitor
the additional nutrient inputs due to grazing. Phosphorus
may be preferred to nitrogen as it generally is more respon-
sive to grazing, and the background levels of phosphorus in
forested areas typically are very low. Compliance monitor-
ing can be relatively simple and inexpensive because ac-
companying discharge data are not needed, and sampling
can be directed towards those times when concentrations are
most likely to exceed the water quality criteria (e.g., summer
low flows). If the total nutrientload is of concern, monitoring
costs will be greatly increased because sampling will have
to be done over the entire range of flows, and continuous
discharge data will be required.
In the absence of other management activities, it might
be possible to evaluate the effects of increased nutrient
levels by monitoring algae or chlorophyll-a. Use of either
of these parameters presumes that the algae are nutrient-
limited, which may not be the case in shaded, forested
streams (Part II, Section 6.2).
5.1.6 DISPERSED RECREATION
The primary means by which dispersed recreation can
adversely affect water quality is through the inadequate
disposalof human wastes. Since inmostcasestherecreational
users are using the same water for drinking and perhaps
bathing, bacteriological monitoring usually will be of pri-
mary importance.
The inadequate disposal of human waste also will
increase nutrient inputs, but in most streams this effect will
be relativelyminoranddifficultto detect. However, nutrients
can accumulate in lakes, and regular monitoring of phos-
phorus and nitrogen may be needed to evaluate the impact
of dispersed recreation on oligotrophic lakes, particularly in
the alpine zone. The focus of recreational use on lake shores
and stream banks means that in heavily used or fragile areas,
riparian vegetation and bank condition shouldbemonitored.
5.1.7 DEVELOPED RECREATION/RURAL
POPULATIONS
Water quality impacts of developed recreation sites,
such as ski resorts, and rural populations involve a mix of
activities that lie largely outside the scope of this document.
Some of the concerns, such as the input of nutrients and the
use of herbicides and pesticides, are analogous to the forest
management activities considered earlier. Other concerns,
such as the quality of runoff from streets and other paved
areas, are not included in these Guidelines and are not con-
sistent with the monitoring parameters reviewed here. Nev-
ertheless, Table 5 does identify those parameters mostlikely
to be useful for monitoring some of the water quality
impacts expected from developed recreation areas and rural
settlements.
For example, the potential pollution due to wastewater
and septic tanks is best monitored by bacteriological and
nutrient parameters. Storm sampling and discharge mea-
surements generally will be necessary because elevated
levels of bacteria and nutrients are most likely to occur
during storm events. Specific conductance might be useful
to indicate the amount of wastewater being dischargedif the
conductance of the receiving waters is both known and
relatively low.
Local use of herbicides and pesticides can be expected,
but the scale of application suggests that neither of these is
likely to significantly impair water quality. The irregular
useofsuchchemicals suggests thattissueanalysisor sediment
sampling may be a better method to estimate herbicide and
pesticide loadings than the sporadic and costly analysis of
water samples.
Almost any rural settlement and associated human
activities will result in an increased sediment load. The need
for monitoring this effect should be based on a field as-
sessment of the potential for increased erosion and consid-
eration of the downstream designated uses. If there is aneed
for sediment monitoring, it usually will be analogous to
project monitoring and involve a comparison of measure-
ments upstream and downstream of the settlement or rec-
reation site.
Rural settlements and developed recreation sites often
result in a modification of the stream channel by localized
clearing of the riparian vegetation and the construction of
roads, culverts, etc. This leads to a variety of effects on
stream temperature, channel morphology, and many other
parameters. Again the monitoring of such effects should be
done only after the need is clearly identified through a
qualitative evaluation of water quality and stream condi-
tion, and with due consideration for the designated uses.
5.1.8 PLACER MINING/SAND AND GRAVEL
EXTRACTION
These activities are outside the scope of this document,
but they have been included because they sometimes occur
in forested environments in the Pacific Northwest and
Alaska, and they can confound the effects of forest man-
agement activities on water quality. The primary concerns
with regard to placer mining are the release of fine sediment
and the destabilization of the stream channel. Direct
monitoringmcludesmethreesedimentparameters(turbidity,
suspended sediment, and bedload), as well as many of the
-------
Part!
channel characteristics. Indirect monitoring can be done
with other parameters such as intergravel DO or some of the
biological parameters—particularly invertebrates or
coldwater fishes. Often turbidity and suspended sediment
data will complement rather than duplicate data on the
channel characteristics. The precise channel characteristic
of greatest sensitivity and utility will depend on the size of
the sedimentbeing released and the transport capacity of the
stream. Generally a combination of bed material particle
size and an aggradation indicator (e.g., pool depth, channel
cross-sections, or longitudinal profile) is likely to provide
the most useful information that can be directly linked to
placer mining and the designated uses. Changes in channel
morphology can trigger secondary effects on the size of the
riparian canopy opening and the riparian vegetation. Al-
though these latter two parameters may be useful for broad-
scale assessments, generally they will be less useful for
monitoring because they are secondary effects and therefore
less sensitive to change.
The release of suspended sediment is also a concern for
sand and gravel extraction, and this suggests that similar
monitoring guidelines should apply. Since removal of large
amounts of sand and gravel may destabilize the stream
channel by altering the sediment load, some monitoring of
the channel characteristics is essential. The precise param-
eters to be monitored will depend on factors such as the
stream gradient and the extent to which the channel is
constrained or incised.
5.1.9 HARDROCK MINING
Again this activity is outside the general scope of these
Guidelines, but a brief discussion is pertinent because
hardrock mining often occurs in forested areas. The effects
of hardrock mining on water quality may also confound or
complicate water quality monitoring projects focusing on
forest management activities. Although historic mining
activities often have resulted in excessive sediment inputs,
hardrock mining should not alter the sediment input or
channelmorphologyprovided proper managementpractices
are used. Typically the greatest impact of current hardrock
miningactivitiesisonstreamchemistry. Thepreciseparameters
that will be most affected depend on the type of mine, the
chemical characteristics of the rock being mined, and the
extraction process being used. This uncertainty is
acknowledged in Table 5, but generally it is important to
monitor at least pH and conductivity. Any change in these
parameters should then trigger a more extensive evaluation
of stream water chemistry.
Since most water chemistry parameters are difficult to
continuously monitor, it is important to regularly sample
aquatic macroinvertebrates or other organisms with a life
span of at least several months. This monitoring will ensure
detection of sudden, toxic releases that might otherwise not
be detected by periodic water sampling. Flow measure-
ments may also be necessary to complement the chemical
data and ensure that the mining operation is not withdrawing
excessive amounts of water.
5.1.10 WILDFIRE AND PRESCRIBED
BURNING
Wildfires and prescribed burning were not included in
Table 5 because their effects on water quality are both
diverse and variable. Removal of the vegetative canopy will
reduce transpiration and tend to increase water yield, espe-
cially during the growing season and immediately after-
wards. Loss of the canopy cover along streams can increase
peak summer water temperatures (Wright, 1978) and make
temperature an important monitoring parameter. Generally
both nitrogen and phosphorus inputs into the aquatic system
will increase after a fire, and this is due primarily to the
disruption of the terrestrial nutrient cycles (Tiedemann et
al., 1978; Wright, 1978).
The effects of fire on sediment yield vary according to
the frequency and intensity of fires, the steepness of the
hillslope and drainage network, and the extent to which the
vegetation controls the movement and storage of sediment
(Swanson, 1978). Fire can greatly increase surface erosion
by temporarily creating a hydrophobic soil layer (e.g.,
Dyrness, 1976; Megahan and Molitor, 1975). Fire also
increases surface erosion and sediment delivery rates by
removing the litter layer and organic debris that traps
sediment both on hillslopes and in the stream channel. The
magnitude of these effects will depend on the geomorphic
sensitivity of the landscape, and this is largely a function of
slope steepness (Swanson, 1978).
These data suggest that nitrogen and phosphorus should
each be monitored when downstream eutrophication is a
concern. In such cases continuous discharge data must also
be collected. The need to monitor stream temperature will
depend on the extent to which the riparian vegetation was
removed by fire. In steep lands substantial increases in
sediment yield can be expected, and the selection of
monitoring parameters will depend largely on which ero-
sional processes were altered by the fire. An increase in
surface erosion may transport only the fine particles (e.g.,
sand-sized or smaller), and inchannel measurements should
focus on turbidity or possibly suspended sediment. If
coarser materials are being delivered to the stream channel
by debris flows or other mass movements, one may wish to
monitor some of the channel characteristics such as channel
cross-section, width-depth ratios, or thalweg profiles.
Complementary data on the amount of large woody debris
in the stream channel should be collected if the fire burned
into the riparian zone (the zone of future recruitment) or
consumed some of the large woody debris within the active
channel.
The Silver Fire case study (Box 7) is one example of
how a broad range of monitoring activities were integrated
-------
Box 7. CASE STUDY: SILVER FIRE RECOVERY PROJECT,
SISKIYOU NATIONAL FOREST
In 1987 the second largest fir© in Oregon's history burned 96,500 acres of the Siskiyou National Forest,
Approximately 56% of the burned area, or 53,600 acres, was located within the Kalmiopsis Wilderness. In the non-
wilderness portion, trees containing an estimated 262 million board feet of lumber were killed by the fire and were
potentially available for salvage. However, most of this area had no road access, and prior to the fire the environmental
community had actively sought designation of the roadless areas as wilderness or national park.
The burned area also included more than 30 miles of the Illinois River. This is one of the most remote Whitewater
rivers in the continental U.S., and ft is designated as a Wild and Scenic River. Underthe antidegradation policy (Section
1.4), no deterioration of water quality can occur that will interfere with, or be injurious to, the designated uses.
Following the recommendations of an interdisciplinary project team, nearly a dozen timber salvage sales and
related recovery projects were conducted over a 2-year period. The recovery projects were aimed at (1) salvaging as
much of the burned timber as possible; (2) minimizing erosion from the fire and harvest activities; (3) protecting the
designated uses, Including salmon and steelhead habitat; and (4) restoring wildlife habitat. Specific project activities
Included 11 miles of new road construction, helicopter and limited cable harvest of fire-killed trees, reforestation of all
harvested and sensitive areas within 2 miles of a road, aerial seeding of about 6,000 acres of the most intensly burned
area with a mixture of annual grasses, installing checkdams on sensitive drainages, contour felling of trees on intensely
burned sites, and installing fish habitat improvement structures in selected areas using native materials.
Concern over the effects of the salvage sales and the effectiveness of the recovery projects resulted in a series of
monitoring activities. These were developed with the assistance of various governmental agencies, researchers, and
, public groups. Monitoring activities were developed to answer four basic questions: (1) were the protective measures
implemented as planned; (2) were the recovery projects and mitigation efforts effective in preventing additional damage
to the resources in the burned area; (3) were the assumptions used to predict the effects of the fire and subsequent
limber salvage valid; and (4) what is the status of the designated uses of water after timber harvest and fire recovery?
Fourteen of the 30 monitoring activities focus on water quality, fish habitat, and other water-related resources, and
these are summarized in the table. Specific objectives for the monitoring activities were based on the designated uses
considered to be at risk. For example, steelhead and salmon fisheries were identified as important designated uses;
pool and summer rearing habitat were identified as the most important limiting factors tothese uses. Thus the monitoring
efforts aredirected towards parameters such as maximum summer water temperatures, pool volumes, stream shading,
and fish cover. In some of the more productive reaches, a series of cross-sections are being used to assess changes
in channel aggradation/degradation, bed material particle size, and habitat types. Landslides were identified as the
largest potential source of sediment, and the number, size, and location of major landslides are being monitored by the
periodic analysis of aerial photos.
Sampling locations were established at the mouth of the main tributaries entering the Illinois River and on the Illinois
River itself. Sampling locations along the Illinois River were selected according to availability of access and proximity
to both key fish habitat areas and salvage activities. Sampling times were selected to maximize the likelihood of
detecting management-induced effects. As indicated in the table, the duration and frequency of sampling varies by
monitoring activity.
Monitoring efforts within the Silver Fire Recovery Project were designed to balance instream measurements related
to water quality and the designated uses with upslope implementation and effectiveness monitoring. In general, the
individual monitoring activities will not be able to demonstrate the effects of specific management activities on the
various designated uses. However, the combined data should be able to assess BMP effectiveness and identify
important changes in the designated uses and their respective cause(s). Annual reports will summarize the data
collected, evaluate conditions in the burned area, and assess the effects of management activities. This information
will be available to other governmental agencies and the public.
Source: Siskiyou National Forest, P.O. Box 440, Grants Pass, Oregon 97526.
-------
—
8
cc
2
fe
i
£
••1
1
-1
••§
i
*
s
*rt
t/
Box 7— cont Summary
£
5
g
ro
(5
"5
1
TO
§
|
•S
to
$
^
'S
^
g
1
c
O
E
O
I
«.
O
o>
CO
N.
to
in
^
CO
CM
z
•>
111
2
1
n
O)
•c
t
c
c
2
co in N- co co •'t en •'t in1* "* N fc SS
0 '«•»-
iini ' • ' ' ' ' ' '"COCO
i-
i-inCM T- ' ' CM CM T- T- CM ' • 5S
CM W
T-WCM T- « CM 'cM-r-1 ' * ' St
CM P>
tJ-Ot-~ T- CO CM CO"* CM'r- COCMCOO
CM CM T- O>
X • X ' ' ' ' ' ' ' X . •• X
x • x ' ' ' xxx* x "x
XXX ' ' ' '' XX'X'
, , , x x x i.iiti.
E'
to
•x. 8 T3 f I -&|f§
I-BJB -slg | il f -s-sll
if! in 1 11 t fill
•Bci§ .groa.s'DL "c ^2 £J "•• w .If
"tl S o-ll Urn 5'i 2 l|ll ^
8.^2 -R "g -g *? ^)"S .s S 1 "a.1? o ^ "tf
•^>o> ro°g-ig g C Q) 5 OT ^~ > — ~ H ftt
O 55 Q O O> Q < OUJCL < OUJ UJ,^J
!£*
1
.S
H
z
ectiveness, V = validatipn,
blowcfawn event ; V;
* §>
H I1
Ul |
» w
1 i
i , I
11 S
¥ ^' "
K cf
^" ^ ..»!
-------
CHAPTER 5. RECOMMENDATIONS AND INTERACTIONS
to address a variety of management concerns and objec-
tives. This case study is particularly interesting because it
explicitly spelled out the different frequencies and duration
of each component of the monitoring program.
5.2 EXPERT SYSTEM
Table 5 and the accompanying discussion provide one
simple means for assessing which parameter(s) might be
most useful for a particular monitoring project However,
more than one management activity may need to be moni-
tored, more than one designated use may be adversely
affected, or more time, funds, or equipment may be avail-
able than is implicitly assumed in Table 5. Table 5 also
assumes that measurements could be made during high flow
events, and this assumption was made to keep Table 5 from
becoming too unwieldy.
To overcome these limitations and make better use of
the information contained in the Guidelines, an interactive
expert system has been developed. This is called PASSSFA
(PArameter Selection System for Streams in Forested Ar-
eas), and it uses a relatively inexpensive, commercially-
available expert system shell called VP-Expert (Paperback
Software, 1989).
In functional terms, an expert system is a computer
program that reasons about a problem in much the same
way, and with about the same performance, as specialists
(Waterman, 1986). Typically expert systems are designed
to help people think through difficult problems and provide
suggestions about what to do without taking over every
aspect of the task. The ultimate goal is to allow people
knowledgeable in the subject area, but with less training and
experience than an expert, to achieve nearly the same level
of performance as an expert with regard to a certain set of
decisions. Unlike traditional computer programs that use
set algorithms, expert systems use rules based on reasoning
or judgment (Buchanan, 1989). Expert systems can and do
incorporate model simulations, but these simply provide
additional input to the decision-making process.
PASSSFA relies on over 300 rules that were developed
from the information contained in Chapter 4 and Section
5.1. These rules specify the response of parameters or
groups of parameters to the controlling factors (manage-
ment activities, designated uses, frequency of sampling,
time required to collect a sample, equipment costs, analysis
costs, and access during high flows). From the user's
response to these questions, PASSSFA generates a list of
suggested monitoring parameters. PASSSFA runs on IBM
or IBM-compatible personal computers with at least 384K
of RAM and DOS version 2.xx or later. PASSSFA uses
video display screens with CGA, EGA, VGA, or a Hercules
monochrome adapter. A color monitor is preferred.
Running the program does not require a user's guide, as
the first four screens provide the necessary introductory
information. The program then prompts the user with a
series of questions, with the number of questions varying
according to the responses received. When sufficient in-
formation has been obtained to select from among all the
parameters reviewed in these Guidelines, those parameters
that meet the user's needs and criteria are displayed along
with a confidence factor.
The confidence factors indicate on a scale of 0 to 100 the
relative certainty that a particular parameter will be useful.
The values assigned are based on experience and judgment,
and they have no statistical meaning and cannot be quanti-
tatively analyzed. Because PASSSFA selects only those
parameters that meet the specified criteria, the confidence
factors associated with the parameters displayed at the end
of a consultation range from a minimum of 65 to a maximum
of 100. In most cases the estimated confidence factors do
not exceed 85 or 90, as one is rarely certain that a particular
parameter will be useful under all possible conditions. It
also is possible that no parameter will meet the user's
criteria; in this case the final screen displays the message
"None available given constraints." The user must then
relax one of the criteria (e.g., indicating other designated
uses for the water body in question, increasing the amount
of time or funds available for monitoring, or allowing access
during high flows). A "What if function in the program
allows the user to modify one response without repeating
the entire consultation, and this is particularly useful when
the initial result is a null set of parameters.
The advantages of referring to PASSSFA in conjunc-
tion with these Guidelines are as follows: (1) PASSSFA
lists an estimated confidence factor with each recommended
parameter, (2) multiple managementactivities and designated
uses can be evaluated at one time, (3) the explicit inclusion
of a range of possible responses in the rule base yields a set
of recommended parameters more directly tailored to the
user's needs and responsibilities, and (4) the effects of a
change in one controlling factor can quickly be evaluated.
Some of the disadvantages include the following: (1) an
inability to easily print out the results of a particular con-
sultation, and (2) the generation of unrealisticalry high
confidence factors whenmorethanonemanagementactivity
or designated use is selected. Both of these disadvantages
result from the particular software shell used to construct
PASSSFA.
The problem of generating unrealisticalry high confi-
dence factors appears only when more than one designated
use or management activity is selected. In such cases a
parameter may be selected by more than one rule. Although
parameters selected by several rules will be listed only once
at the endoftheconsultation,theconfidence value associated
with that parameter is altered. Specifically, the confidence
factor is calculated by adding the confidence factors of the
individual rules that are satisfied (expressed as decimals
rather than on a scale of 0 to 100), and then subtracting the
product (again calculated using decimal values). For ex-
-------
Part!
ample, if the same parameter was selected by two different
rules with confidence factors of 70 and 80 (0.70 and 0.80),
respectively, the combined confidence factor in decimals
would be 0.70 + 0.80 - (0.80 x 0.70), or 0.94. In PASSSFA
this parameter would then be displayed at the end of the
consultation with a confidence factor of 94.
In the case of water quality monitoring, the simple
selection of a parameter by two different rules does not
necessarily mean that particular parameter is more likely to
be useful. Local conditions and professional judgment still
havetobeapplied.Morerealisticestimatesofflieconfidence
factor can be obtained only by repeating the consultation
with only one management activity and one designated use,
and recording the confidence values obtained in each case.
The "what if function of the expert system shell facilitates
this type of repeated consultation.
Cost considerations precluded the inclusion of
PASSSFA with each copy of the Guidelines, but copies of
the expert system can be obtained by sending a blank,
formatted diskette with at least 225K-bytes of avail-
able space to the Seattle office of the U.S. Environ-
mental Protection Agency at:
U.S. EPA, Region 10
NPS Section, WD-139
1200 Sixth Ave.
Seattle, WA 98101
5.3 PARAMETER SELECTION AND
INTERACTIONS
The discussion in Section 5.1 illustrates several impor-
tantpoints regarding the selection of water quality monitor-
ing parameters for forest management activities. First, in
most cases the choice of monitoring parameters is not easy
or clear-cut. Road building and maintenance and forest
harvest are two common management activities where the
most direct and sensitive water quality monitoring param-
eters are also the most difficult and costly to measure. The
choice of parameters is much clearer with regard to monitor-
ing fertilizer, herbicide, and pesticide applications, but
these activities are less frequent and generally do not pose
a chronic threat to the designated uses of water. The absence
of well-defined monitoring parameters for forest manage-
ment activities should not be surprising since the uncer-
tainty regarding what to monitor was a principal rationale
for the preparation of these Guidelines.
A second general conclusion that emerges from Table 5
is that the traditional physical and chemical water column
parameters have limited usefulness for monitoring most
management activities in forested areas. As indicated in
Table 3, the primary water quality effects of road building
and maintenance and forest harvest stem from the increased
sediment load and the reduction in the riparian vegetation.
These changes can adversely impact most of the usual
physical and chemical water quality constituents, but with
the exception of temperature these effects are generally
small or indirect. Suspended sediment and turbidity are two
physical water column parameters that can be used to
directly monitor changes in the amount of fine sediment, but
in most cases intensive monitoring is needed during storm
events in order to obtain useful data.
Third, the parameters listed under channel characteris-
tics should be considered in any monitoring project having
changes in sediment, flow, or riparian vegetation as a
possible concern. These parameters have the advantage of
being easier to measure and integrating the effects of all the
individual storm events. Their primary disadvantages are
(1) the lack of long-term data to evaluate their usefulness for
water quality monitoring, and (2) the difficulty of relating
observed changes to specific management activities.
Fourth, one ormore of the biologic parameters is ranked
as at least moderately useful as a monitoring technique for
almostall of the various managementactivities. The advan-
tage of the biologic parameters is that they can be directly
related to the designated uses of water, they often are quite
sensitive to management impacts, the sampling frequency is
low-to-moderate, and the costof sampling and data analysis
also can be considered moderate. It may not always be clear
whatspecific biological parameter shouldbe monitored, but
EPA's current emphasis on developing techniques and
criteria for biological monitoring appears to be well founded.
The danger is that some of these techniques will be adopted
before they can be fully validatedforthedifferentecoregions.
These four generalizations are less applicable to the
other management activities of mining, grazing and recre-
ation. Themostappropriateparametersformonitoringsand
and gravel extraction, placer mining, and recreation (either
dispersed or developed) are relatively clear-cut. It is more
difficult to generalize about the choice of parameters for
monitoring hardrock mining because there is so much varia-
tion in the extraction and processing of the ore.
The potential impacts of grazing on stream systems are
so extensive that a wide variety of water quality parameters
could be utilized. Like grazing, wildf ires can affect a variety
of stream andchannel parameters, and the selection of param-
eters will depend on the physical environment, the desig-
nated uses, and the intensity and location of the fire relative
to the stream channel.
An important limitation of Table 5 is that each param-
eter is considered independently. However, many of the
parameters are closely related. Turbidity, for example, is
often used as a surrogate for suspended sediment concentra-
tion. Channel cross-sections, width-depth ratios, thalweg
profiles, and pool parameters all are responsive to changes
in the balance between discharge and sediment concentra-
tions. These types of interrelationships mean that not all
parameters that are rated highly in Table 5 should be used.
-------
CHAPTER 5. RECOMMENDATIONS AND INTERACTIONS
The best approach is to identify the parameters of most
interest from Tables 2-5, and then rely on the technical
review of the individual parameters in Part II to help make
the final choice.
Tables 6A and 6B provide one way to overcome the
problem of consideringeachparameter independently. These
qualitatively rank the effect of change in one parameter on
all the other parameters reviewed in these Guidelines. The
intent of these tables is that once one possible monitoring
parameter has been identified, the tables can be used to
assess which other parameters are most likely to respond to
changes in that parameter. Closely related parameters can
be identified, and this information should help to eliminate
any duplication of monitoring effort and maximize the
independence of the parameters selected for monitoring.
Proper use of the tables requires careful attention to the
cause-and-effect ordering of the interactions. For example,
the width-depth ratio can greatly affect water temperature,
but water temperature has only a very tenuous effect on the
width-depth ratio.
AsecondlimitationofTableSisthatitdoesnotexplicitly
link the variousparanietersthatmightbecombined to produce
an optimal monitoring program. Bed material particle size
and thalweg profile, for example, might be a powerful
combination to evaluate sediment-related changes in allu-
vial channels. The particular combination most appropriate
for a given situation can be determined only through an
understanding of how management activities affect the
various processes operating in a watershed. Hence the
Guidelines cannot provide specific solutions, but the com-
bination of Tables 5,6, andPartll should allow an informed
decision to be made.
Similarly, Table 5 does not necessarily indicate which
combination of parameters is best suited to demonstrate the
effect of a particular management activity on the designated
uses of a water body. In the absence of a well-replicated
study, or associated data on the causal processes, a simple
observation of trends for the one or two "most useful"
parameters will not rigorously demonstrate the cause of an
observed trend. Table 6 does provide some indication of
cause-and-effect linkages by indicating how one parameter
will affect each of the other monitoring parameters re-
viewed in these Guidelines. However, if water quality is to
improve, monitoring results must feed back into manage-
ment practices, and this can only be done effectively when
the cause of an observed change in water quality can be
identified.
Determining cause-and-effect often is further compli-
cated by the presence of multiple management activities
upstream of the monitoring site. For this reason it is
important to understand why aparticularparameter is being
recommended in Table 5 and how it is, or is not, linked both
to the management activity and the designated uses. This
information is explicitly set forth in the discussion of each
parameter in Part II, and both the manager and the monitor-
ing specialist must be aware of the strength of these link-
ages.
There is also an inherent limitation as to what can be
learned from monitoring activities that are limited to the
stream channel and adjacent riparian areas. Inchannel data
are essential for determining trends in water quality, evalu-
ating whether the designated uses are being impaired, and
assessing whether the applicable water quality standards are
beingmet. For practical reasonstheseGMttfe/yieshadtofocus
on those parameters and monitoring techniques that can be
utilized in the stream channel and adjacent riparian zone.
However, a full understanding and interpretation of these
data requires a broader watershed perspective. Knowledge
of the type and location of management activities and
upslope processes is necessary if the monitoring results are
to provide effective feedback to management and enhance
our understanding of watershed behavior.
Ultimately it is the value of this feedback to manage-
ment thatdetermineswhetheraparticularmonitoringproject
is successful. Similarly, the internal feedback loops largely
determine whether the data collection and analysis efforts
within a particular monitoring project will meet the project
objectives. At least for the foreseeable future, the develop-
ment and operation of water quality monitoring projects will
remain an iterative process. The range of controlling factors
and natural conditions simply precludes the development of
an optimal water quality monitoring project on the first try.
These Guidelines are one means to facilitate or advance this
iterative process for water quality monitoring projects in
forested areas.
-------
a
a
a
0
1
"i
a
QJ •
E CD
S 75
£5 "55
s =
^—* >%
S II
8 ^*"
"§ ^oj
•si
g-t
•— CO
w 4=
al
CO ^
2. If
£ f
"o co
Q> .S
js .E
"•"* .2
T3 ^«
9? "o
S|
12 1
?o *^~
<3 "O
•QJ g
E
m CJ
ti
•c ^.
O ^3*
1>.I
75
a '"*
§v>
co II
f"fc _T
12
S2 w
li
Table 6A.lnterrela
hand side of the ta
'c
CD
E
T3
CD
CO
1
CO
(D
D
C
re
0
c
E
3
O
O
TO
O
CO
b
1
~J
To
\u
Q.
CO
CD
"to
co
CD
'o
1
CD
x
Q_
Z
m
0
Q
§
ts
3
|
"
Q.
d.
E
1-
Parameters
CM't^i-'^CM-*'*'*-* -<1-CM-* COCO) i- i- i- T- i-
COCO-r-CMCOCMCMCOCO CO1^--* T- | ^ 'S' •* •* ^ •*
COCO-«-CMC\ICJC\JCOCO CO-'t'l' |-r-CO CO COCOCOCO
CJCOt-CMCJCOCOCaCVI III T-T-T- CM ,r-T--sf^-
CMCO-r-i-T-COCO'*-
-------
Table 6A— cont.
I
E
73
CO
S
_o
•*-
tC
CO
CD
O
C
CO
O
C
1
1
CD
'5
^
73
CO
O
73
CD
CD
73
"-S
f_
73
CD
73
C
CD
CL
CO
CD
E>
CO
o
to
Q
$
~*
CO
CD
Q.
co
73
to
CD
Q.
73
)o
CD
Q-
CO
O
a
O
Q
73
1
I
Q.
Q.
1
Parameters
* — i — ^ — ^ i — ^ — ^ — ^~ i — ^ — i —
COCOCO^CO CO'* CMCVJi-t-
T-i--r-cOt- -^ i- 1-CVICMCM
COCOCO-*-* •*« CMCMCNl"
•^•^T-CJT- T-CM CMCOCO^?
coco-*coco coco -
-------
T3
1
S
"<5
a .
ll
E CD
11
If
"o it
"at "S
:5 "co
to "J5
•^ g-
E tp
CO Q>
S.-S-
II
0 O
Jjlf
S ^
•E ii
CO CO
CO •-
If
QJ kX
|s
QJ "^^
— =; >s
*^§
g .S3
II
o *^
""§ <5
^j ^5
cxi c
05 71
t^J
•^"
*£^ 5^
g* ^
if
CtJ «y—
co "co
Cl>
J~ >.
§
" II
«S i—
< ..
CO Cg
25 -^
jCO S
"o £
.§ ^
to •§
?^ "25
Table 6B. Contil
on the left-hand
c
*c
re
o
DC
8
E?
c
CD
0
2
n
"o
"5
c
re
.c
en
E
.2
CO
0
.Q
1o
^
1
en
>
>*
|-
S
J£
TO
CD
8
1
CO
1
.0
"S
m
to
•g
CO
X
D>
•(S
T3
'5
c
.c
0
„___,
C
1
o
U
U.
c/
_Q
n
X.
a
a
—
2
n
<
as
•c
n
CD
c
f
,±
**
£
°CZ
CD
&
£
"ra
1/5
i
3
•§
1
3
CO
i
1
CD
.N
CO
.^
cz
=>
^
1
CO
8
a.
Q.
CD
T
.C
"S
S
0
CD
U
Parameters
^-^-^COCO^r,-^^ -• ^.** ^-^-^ ^- <***•*•
^" ^ ^ W CO ^ ^' •*'* ' ^f ^J" 'sf •^J'^'^ ^ '^•'^•'S''^
CO i- CO i- i- CM CM t t •*•*•* •*•*'* •* <*•*•*•*
•*•*•* • „,
1 £ fe- 8 . , , | 1 llfle
1 | 1 |&||j ||| ||tl Ililflll
1 liloaJl'il ^1|| "st"!!! .1 ^ 1 .1 i 8 1 1
^11— Q.OQ — ZQ-IECl. LLJ Q. _J > CO) CO)— CO OiOOSCLH-I
-------
to
1
'c
CO
E
o
_o
*
c
to
CO
Q.
DC
1
•c
t
cc
ct
"a
c
a
6
u.
CO
S
to
^
a>
g
"^*
m
•0 .N
m
™
tO K
1'i
n:
P
rt
a> a)
3 S
1 9
Q.
H ~~
»^.
.c
^
'i
c
co
c
o
a.
1
|1
o a
tfi
Ji
c
o
arameters
o.
COCO CO-*1* •*•* 'tCOCMl
-*Tf - ^ ^^ f* COT-I-
^.^ ^ ^^ ^^ ^ |^_
^^.^.^.^ ^.^J. CO-r-CM
CO CO CO T-"T- T- | CO CM T- T-
't't'^'CMCO i — CO T- T— CM
1-1- T-CM| 1-1- 'tcocMT-
1—CM CM |i— COCM Tf^fCMi—
1—1— 1 '^tTf 't'^)' •* CM i— i—
1
.
i— j T— XI* ^J" •^t ^" * CO • i~ i— T—
|i- -r- -*T- -"t-* -
-------
REFERENCES: PARTI
APHA, 1989. Standardmethods for the examination of water and
wastewater, 17th ed. Am. Public Health Assoc. Washington,
D.C. 1460 p.
Avcrctt, 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.
Denver, CO. 16 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.
Bailey, R.G., 1976. Ecoregions of the United States. Map (scale
1:7,500,000). USDA For. Serv., Intermount. Reg. Ogden,
UT.
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. Hydrol. 55:3-23.
Box, G.E.P., and G.M. Jenkins, 1976. Time Series Analysis:
Forecasting and Control, 2nd ed. Holden-Day, San Fran-
cisco.
Brussock, P.P., A.V. Brown, and J.C. Dixon, 1985. Channel form
and stream ecosystem models. Wat. Resourc. Bull. 21:859-
866.
Buchanan, E.G., 1989. An overview of expert systems. Pages
129-153 in J.D. Meindl (ed.), Brief Lessons in High Tech-
nology. Stanford Alumni Assoc., Stanford, CA.
CFR, 1990. Code of Federal Regulations, Title 40, Part 130,
Section 2, Office of the Federal Register. Nat Arch. Rec.
Admin., Washington, D.C.
Crowley, J.M., 1967. Biogeography. Canad. Geographer 11:312-
326.
Cupp, C.E., 1989. Streamcorridorclassificationforforestedlands
of Washington. Hosey and Assoc. Bellevue, WA. 46 p.
Daniel, W.W., 1990. Applied Nonparametric Statistics. PWS-
Kent Publ. Co. Boston, MA. 635 p.
Dymcss, C.T., 1976. Effects of wildfire on soil wettability in the
High Cascades of Oregon. USDAFor. Serv. Res. Pap. PNW-
202. Portland, OR. 18 p.
EPA, 1983. Guidelines and specifications for preparing Quality
Assurance Program plans. U.S. Environmental Protection
Agency, Off. Res. Develop., EPA-600/8-83-024. Washing-
ton, D.C. 22 p.
EPA, 1986. Quality criteria for water: 1986. U.S. Environmental
Protection Agency, Office of Water Regulations and Stan-
dards. Washington, D.C.
EPA, 1987. Surface water monitoring: a framework for change.
U.S. Environmental Protection Agency, Office of Water.
Washington, D.C. 41 p.
EPA, 1988. Introduction to waterquality standards. U.S.Environ-
mental Protection Agency, Office of Water Regulations and
Standards, EPA-440/5-88-089. Washington, D.C. 29 p.
EPA, 1989a. Nonpoint sources—Agenda for the future. U.S.
Environmental Protection Agency, Off. Water, WH-556. 31
p. Washington, D.C.
EPA, 1989b. Rapid bioassessmentprqtocolsforuseinstreams and
rivers: benthic macroinvertebrates and fish. U.S. Environ-
mental Protection Agency, Office of Water, 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.
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 Resource, Univ. Washington. Seattle,
WA.
Gilbert, R.O., 1987. Statistical methods for environmental pollu-
tion monitoring. Van Nostrand Reinhold. New York, NY.
Grant, G.E., 1988. The RAPID technique: a new method for
evaluating downstream effects of forest practices on riparian
zones. USAD For. Serv. Gen. Tech. Rep. PNW-220. 36 p.
Gray, D.H., and W.F. Megahan,,1981. Forest vegetation removal
and slope stability in the Idaho batholith. USDA For. Serv.,
Ihtermount. For. Range Exper. Sta. Res. Pap. INT-271. 23 p.
Green, R.H., 1989. Power analysis and practical strategies for
environmental monitoring. Environ. Res. 50:195-205.
Green, R.H., 1979. Sampling Design and Statistical Methods for
Environmental Biologists. John Wiley & Sons, New York,
NY.
Greeson, P.E., et al. (eds.), 1977. Methods for collection and
analysis of biological and microbiological samples: tech-
niques of waterresources investigations. Geological Survey,
U.S. Dept. of the Interior, Washington, D.C., 332 p.
Griggs, G., 1988. Impact of the January 1982 flood in Santa Cruz
-------
REFERENCES
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. USGS Prof. Pap. 1434. Washington,
D.C.
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. U.S. 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
Water-Resources Investigations of the United States Geo-
logical Survey. Applications of Hydraulics, Book 3. U.S.
GPO. Washington, D.C. 59 p.
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.
Hamed, D.A., C.C. Daniel m, and J.K. Crawford, 1981. Methods
of discharge compensation as an aid to the evaluation of water
quality trends. Wat. Resourc. Res. 17:1389-1400.
Hawkes, H.A., 1975. River zonation and classification. Pages
312-374 in B.A. Whitton (ed.), River Ecology. Blackwell,
London.
Hohenstein, W.G., 1987. Forestry and the Water Quality Act of
1987. J. Forestry 85(5):5-8.
Hurlbert,S.H., 1984. Pseudoreplicationandthedesignof ecological
field experiments. Ecol. Monogr. 54:187-211.
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.
Corvallis,OR. lip.
Idaho Dept. Health and Welfare, 1988. Forest practices water
quality audit. Div. Environ. Qua!., Water Qual. Bur. Final
Rep. Boise, ID. 24 p. + appendices.
James, L.D., and S.J. Surges, 1982. Selection, calibration, and
testing of hydrologic models. Chapter 11 in Hydrologic
Modeling of Small Watersheds. Am. Soc. Agric. Engineers.
533 p.
Kunkle, S., W.S. Johnson, andM. Flora, 1987. Monitoring stream
water for land-use impacts. Nat. Park Serv., Wat. Resourc.
Div. Ft. Collins, CO.
Larsen, D.P., J.M. Omernik, R.M. Hughes, CM. Rohm, T.R.
Whittier, A.J. Kinney, A.L. Gallant, and D.R. Dudley, 1986.
The correspondence between spatial patterns in fish assem-
blages in Ohio streams and aquatic ecoregions. Environ.
Mgmt. 10:815-828.
Lisle, T.E., 1982. Effects of aggradation and degradation on riffle-
pool morphology in natural gravel channels, Northwestern
California. Water Resour. Res. 18(6): 1643-1651.
Luchetti, G., K. Thomburgh, and C. Rawson, 1987. A water
quality monitoring program for major agricultural areas of
the Snohomish River basin. Tulalip Tech. Rep. Snohomish
Co., WA.
Lynch, J. A., andE.S. Corbett, 1990. Evaluation of bestmanagement
practices for controlling nonpoint pollution from silvicultural
operations. Wat. Resourc. Bull. 26(l):41-52.
McCammon, B.P., 1989. Monitoring in the Bull Run watershed.
Columbia Gorge Ranger Distr. Troutdale, OR. 8 p.
Megahan, W.F., and D.C. Molitor, 1975. Erosional effects of
wildfire and logging in Idaho. Pages 423-444 in Irrig. Proc.
and Drainage. Amer. Soc. Civil Eng.
Mendenhall, W., 1971. Introduction to probability and statistics.
Duxbury Press. Belmont, CA. 466 p.
Montgomery, D.C., and L.A. Johnson, 1976. Forecasting and
Time Series Analysis. McGraw-Hill. New York, NY.
Montgomery, R.H., and K.H. Peckhow, 1984. Techniques for
detecting trends in lake water quality. Wat. Resourc. Bull.
20:43-52.
NCASI (Nat. Council of the Paper Industry for Air and Stream
Improvement), 1988. Procedures for assessing the effective-
ness of Best Management Practices in protecting water and
stream quality associated with managed forests. Tech. Bull.
No. 538. New York, NY. 23 p. + appendices.
Omernik, J.M., 1987. Ecoregions of the conterminous United
States. Ann. Assoc. Am. Geograph. 77(1): 118-125.
Omernik, J.M., and A.L. Gallant, 1986. Ecoregions of the Pacific
Northwest. Map (scale 1:7,500,000). U.S. Environmental
Protection Agency EPA/600/3-86/033. Corvallis, OR.
125 p.
Paperback Software, 1989. VP-Expert: Rule-based expert system
development tool. Paperback Software Internal., Berkeley,
CA.
Ponce, S.L., 1980a. Water quality monitoring programs. USDA
For. Serv. Tech. Pap. WSDG-TP-00002. Fort Collins, CO.
66p.
Ponce, S.L., 1980b. Statistical methods commonly used in water
quality data analysis. USDA For. Serv. Tech. Pap. WSDG-
TP-00001. Fort Collins, CO. 136 p.
Potyondy, J.P., 1980. Technical guide for preparing water quality
monitoring plans. USDA For. Serv., Intermount. Reg. 43 p.
Raedeke, K.J. (ed.), 1988. Streamside management: riparian
wildlife and forestry interactions. Contribution No. 59, Inst.
of Forest Resources, Univ. of Washington. Seattle, WA.
Rohm, C.M., J.W. Giese, and C.C. Bennett, 1987. Evaluation of
an aquatic ecoregion classification of streams in Arkansas.
FreshwaL Ecol. 4:127-140.
Rosgen, D.L., 1985. A stream classification system. Pages91-95
in R.R. Johnson, C.D. Zeibell, D.R. Patton, P.P. Ffblliott, and
R.H. Hamre (eds)., Riparian Ecosystems and their Manage-
ment: Reconciling Conflicting Uses. USDA For. Serv. Gen.
Tech. Rep. RM-120. Fort Collins, CO.
Salo, E.O., and T.W. Cundy (eds.), 1987. Streamside Manage-
ment: Forestry and Fishery Interactions. Contr. no. 57, Inst.
Forest Resources, Univ. Washington. Seattle, WA. 471 p.
Sanders, T.G., and R.C. Ward, 1979. Relating stream standards to
regulatory water quality monitoring practices. Pages 339-
348 in L.G. Everett and K.D. Schmidt (eds.), Establishment
of Water Quality Monitoring Programs, Proc. of Symposium,
San Francisco, CA. June 12-14,1978. Amer. Wat. Resourc.
Assoc. Minneapolis, MN. 370 p.
Sanders, T.G., R.C. Ward, J.C. Loftis, T.D. Steele, D.D. Adrian,
and V. Yevjevich, 1987. Design of networks for monitoring
water quality, 2nd ed. Water Resource Publ. Littleton, CO.
328 p.
SCS (Soil Conservation Service), 1981. Land resource regions
andmajorlandresource areas of theUnitedStates. Agricultural
Handbook296. Map (scale 1:7,500,000). U.S. GPO. Wash-
ington, D.C. 156 p.
Siegel, S., 1956. Nonparametric Statistics for the Behavioral
Sciences. McGraw-Hill. New York, NY. 312 p.
Sokal, R.R., and F.J. Rohlf, 1981. Biometry. W.H. Freeman and
Co. San Francisco, CA. 859 p.
Solomon, R., 1989. Implementing nonpoint source control: Should
BMPs equal standards? Pages 155-162«D.D.Hook,andR.
Lea (eds.), Proceedings of the symposium: The Forested
Wetlands of the Southern United States, Orlando, FL. July
-------
Parti
12-14,1988. USDAFor.Serv.Gen.Tech.Rep.SE-50. 168p.
Swanson, F.J., 1978. Fire and geomorphic processes. Pages 401-
420 in Fire Regimes and Ecosystem Properties. USDA For.
Serv. Gen. Tech. Rep. WO-26.
Swanson, F.J., L.E. Benda, S.H. Duncan, G.E. Grant, W.F.
Megahan, LJM. 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, List. Forest Resources, Univ.
Washington, Seattle.
Swanston,FJ., 1969. Mass wasting in coastal Alaska. USDA For.
Serv., Pac. Northw. For. Range Exper. Sta., Res. Pap. PNW-
83. 15 p.
Swanston, FJ., RJ. Janda, T. Dunne, and D.N. Swanston, 1982.
Sediment budgets and routing in forested drainage basins.
USDA For. Serv., Pac. Northw. For. Range Exper. Sta.; Gen.
Tech. Rep. PNW-141. 23 p.
Tiedemann, A.R., J.D. Helvey, andT.D. Anderson, 1978. Stream
chemistry and watershed nutrient economy following wild-
fire and fertilization in eastern Washington. J. Environ. Qual.
7:580-588.
Thomas, R.B., 1985. Measuring suspended sediment in small
mountain streams. USDA For. Serv. Gen. Tech. Rep. PSW-
83. 9p.
Torquemada, R.J., and W.S. Platts. 1988. A comparison of
sediment monitoring techniques of potential use in sediment/
fish population relationships. USDAFor. Serv., Litermount.
Sta., Progr. Rep. I to Idaho Fish and Game Department. 55 p.
Troendle, C.A:, andR.M. King, 1985. The effect of timber harvest
on the Fool Creek watershed, 30 years later. Wat. Resourc.
Res. 21(12):1915-1922.
USGS (U.S. Geological Survey), 1982. Hydrologic unit map of
the United States. Map (scale 1:5,000,000). U.S. GPO.
Washington, D.C.
Waterman, D.A., 1986. A guide to expert systems. Addison-
Wesley, Reading, MA.
Whittier, T.R., R.M. Hughes, and D.P. Larsen, 1988. Correspon-
dence between ecoregions and spatial patterns in stream
ecosystems inOregon. Can. J. Fish. Aquat. Sci. 45:1264-1278.
Wright, K.A., K.H. Sendek, R.M. Rice, and R.B. Thomas, 1990.
Logging effects onstreamflow: storm runoff at Caspar Creek
in northwestern California. Wat. Resourc. Res. 26(7):1657-
1667.
Wright, H.E., Jr., 1978. 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.
Zar, J.H., 1984. Biostatistical Analysis. Prentice-Hall. Englewood
Cliffs, NJ.
-------
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 selectmultiple
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 selectionprocess,
asecondpurposeofPartnistosummarizecurrentknowledge
on each parameter. This should help the reader understand
the rationale, possibilities, and constraints for monitoring
each of theSO 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, qualitativeevaluation of thepotentialroleof the
parameterin 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 some of 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 or riparian 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 lOdifferentparameters
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
that appear to have considerable potential for monitoring the
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
the typeofpollution,theratesof chemical andphysicalprocesses
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 managementactivi-
ties—including forest fertilization—can increase nitrogen
orphosphorus 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 sediment or 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 waterquah'tymonitoringprojects.
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 theprimary 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 temperatures reflect both 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 peak daily 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) and Berman and Quinn (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. Forexample, attemperaturesbelow about5°C,
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°C increase in temperature (Eastman, 1970). The equilib-
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 etal., 1987).
These temperature increases are due almost entirely to
theadditionalinputof 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
consideredif instantaneous ratherthan 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 organismsrespondmore 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 summer and 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 becauseof 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 cool-water 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 otherchanges due to opening up theforestcanopy 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/[Hf] = -
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 notdirectly 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 ICh7 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 (COz). Com-
pared to most other atmospheric gases, carbon dioxide is
relatively soluble, and in solution it combines with water to
form carbonic acid (^COs). 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 plays 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
salmonideggproduction 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 thatpH 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 theeffects 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 managementactivities can indirectly affectpH 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 CO2 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). The variation in mining and extraction
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
pH can 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 suitable only for rough estimates (APHA,
1976). Usually pH is measured electronically with a pair 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 to be recorded, as theequilibrium 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 thenearestO.l pH unit (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 aircanbeusedto 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 COa 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
EPAhas setapHrange 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 criteria necessary to protect freshwa-
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 for pH 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-
eries, 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.
MonitoringpH 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 a non-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 unit is ^mhos/cm.where Wpmhos is equal to 1 mhos.
Conductivity may also be reported in millisiemens/meter,
with 1 millisiemen/m equal to 0.1 jimhos/cm (APHA, 1976).
Pure water not in contact with the atmosphere has a
conductivity of approximately 0.05 jamhos/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
UnitedStates has aconductivity of 2 to42jimhos/cm (Hem,
1970). The range for potable water in the U.S. is 30 to 1500
|imhos/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 at least as 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
{jmhos/cm was well below the range of 50-120 pmhos
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; Harr andFredriksen,
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 anddischarge 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 theelectrical 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 thebicarbon-
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 et al., 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 isonesituationwhereconductivitymonitoring
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 L-1. 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
maynot be 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). Subsequentresearchhas 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-
iThe 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: Q,=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
intergravelDO 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), but these
do not relate changes in intergravel DO to management
activities. Hence the cause-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 I/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 most directly 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 Lr1 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 be 8.0 and 9.5 mg L-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
intergravel 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 mgL-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
susceptibility of a redd 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 theamount 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 particulate 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 particulate
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) first to 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) togaseousN2if 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
begins 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 canalter 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-1 (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 inNeedle 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 left to decompose (Harr and Fredriksen, 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 losses of
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. Particulate 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. Mostof these areon 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 timelag 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 the aquatic ecosystem was estimated at 0.5%
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 considerednonpomtsourcesandrequire relatively
intensive wet season monitoring to determine their effect on
water quality. Municipal wastewater plants are point sources
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
ammoniaincludesboth ionized (NH4+) and un-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 mg L"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
dissolvedandparticulate 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 high nitrite 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). These 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 (Part I, 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 attenuatedor completely obscured as a result of increased
Table 8. Equilibrium concentration of un-ionized ammonia in
mg L1 as a function of temperature and pH.
Temperature (°C)
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
-------
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-
gest that relatively intensive monitoring for urea, total
ammonia, and nitrate for approximately 4 days usually will
detect the peak nitrogen concentration due to the direct
application of fertilizer into the fluvial system. Simulta-
neous discharge data should also be collected to permit an
estimate of the total amount of fertilizer delivered directly
into the stream. High losses suggest a need to modify the
method and conditions of fertilizer applications.
Monitoringplansshouldrecognizethatasecondincrease
in nitrate concentrations often occurs during the first runoff
events following fertilization. Although this latter peak also
is unlikely to violate water quality standards, the higher
flows mean that the majority of the nitrogen loss can
occur at this time (Norris et al., 1983). This second set of
samples will indicate the overall ability of the terrestrial
nitrogen cycle to scavenge and fix additional nitrogen
inputs.
The reality of operational forest fertilization programs
is that proper application procedures—specifically the use
of pellets and buffer strips—generally shouldpreventaquatic
nitrogen levels from exceeding national standards (e.g.,
Cline, 1973). Often the increase in aquatic nitrogen will be
large in relative terms, but small in absolute terms. This
increase may enhance aquatic productivity for a couple of
years without any apparent adverse impacts to fast-flowing
streams. More careful monitoring is needed when the water
is used for domestic consumption within a relatively short
distance from the fertilized area, and when slow-flushing
ponds or lakes are present downstream. Generally the
monitoring station(s) should be as close to the application
area as possible, as this will minimize both the dilution
effect and biotic uptake.
Mandatory monitoringandstricter water quality criteria
may be appropriate for basins that drain into oligotrophic
lakes. In thesecases there is along-term, cumulative hazard
to lake water quality. This danger is best addressed by setting
stricter standards for fertilizer applications and minimizing
erosion.
Very intensive monitoring may be required in those
basins designated as impaired under Section 319 of the
Clean Water Act. In these basins a water quality model and
a load allocation process may need to be developed for the
water quality constituents limiting the designated uses of
water (Part I, Section 1.4). An assimilative capacity also
must be determined and allocated between point and non-
point sources (Ice, 1990). If nitrogen is one of the constitu-
entsofconcern, themonitoringprogram will have to evaluate
the timing and frequency of standard violations, the relative
contribution of each source, natural background levels, the
validity of the water quality model, and changes due to
management actions.
Assessment
Almost any forest management activity will affect some
aspect of the nitrogen cycle. Logging, fire, and forest fer-
tilization can substantially increase nitrogen concentrations
in streams. However, in most cases the absolute amount of
nitrogen which enters into streams is small because most
forest streams have such low background levels of nitrogen
compounds. The observed increases in aquatic nitrogen
caused by logging and fire typically represent only a very
small fraction of the total nitrogen capital on site (e.g.,
Sollins and McCorison, 1981).
The forest management activity with the greatest poten-
tial for increasing nitrogen concentrations in forest streams
is forest fertilization. Aerial applications of nitrogenous
fertilizers such as urea generally deliver some nitrogen
directly into the drainage system. When efforts are made to
limit drift and avoid direct application into streams and
lakes, water quality standards should not be exceeded (e.g.,
Cline, 1973; Fredriksen et al., 1975; Bisson, 1982). This
suggests that a limited amount of water quality monitoring
should suffice in most cases.
In evaluating the effects of an increase in nitrogen con-
centrations in streams, one must consider all the transforma-
tions and processes of the aquatic nitrogen cycle. For ex-
ample, an increase in organic nitrogen will notaffectprimary
productivity as much as an increase in inorganic nitrogen. In
some streams an increase in plant-available nitrogen will
have little biologic effect because primary productivity is
limited by other factors. In many cases a small or moderate
increase in nitrogen due to forest activities can be locally
beneficial by increasing primary productivity.
Determination of acceptable increases in stream nitro-
gen concentrations also requires consideration of the down-
stream uses of water. Streams that flow into oligotrophic
lakes and streams used for water supply purposes may
require more frequent monitoring and more stringent stan-
dards. Continuous discharge measurements are essential
for any monitoring that involves calculating total load,
while simultaneous discharge measurements are needed to
properly interpret any instantaneous measurements of nitro-
gen concentrations.
2.5.2 PHOSPHORUS
Definition
In natural waters phosphorus can be separated into two
fractions, dissolved and paniculate (Leonard et al., 1979).
Dissolved phosphorus is found almost exclusively in the
-------
CHAPTER 2. PHYSICAL AND CHEMICAL CONSTITUENTS
form of phosphate ions (PO4-3), and these bind readily with
other chemicals (Hem, 1970). There are three main classes
of phosphate compounds—orthophosphates, condensed
phosphates, and organically bound phosphates. Each of
these can occur in soluble forms (i.e., dissolved phosphorus)
or be bound to paniculate matter (APHA, 1976). In general,
only orthophosphates are readily available for biotic uptake.
The common analytic tests are total dissolved phosphate,
which includes all three classes of phosphate compounds,
and dissolved soluble and plant-available phosphate, which
is essentially equivalent to orthophosphate.
Since dissolved phosphorus can be readily transported to
the drainage network by surface and subsurface flow, the
forms and concentrations of dissolved phosphorus in aquatic
systems are directly related to the terrestrial phosphorus
cycle. Like nitrogen, most of the phosphorus in terrestrial
ecosystems is insoluble and thus immobile. Losses of soluble
phosphorus due to leaching in Douglas-fir and silver fir
forests represent an insignificant portion of the total phos-
phorus on the site (Cole, 1979). Annual phosphorus budgets
for four forested coastal watersheds range from a net gain of
0.1 kg ha-1 yr1 to a net loss of 0.3 kg ha-1 yr1 (Feller and
Kimmins, 1979). Subalpine watersheds subject to heavy
spring snowmelt and periodically intense rainfall may not
conserve phosphorus (Leonard et al., 1979).
Any unbound phosphate ions that enter into streams and
lakes, or which are released by microbial decomposition,
are readily taken up by aquatic plants and microorganisms.
The rapid biological uptake and the ease of chemical bond-
ing explain why phosphate concentrations in natural waters
generally are very low (Hynes, 1970; Hem, 1970; APHA,
1976). Mean annual phosphorus concentrations in small
forest streams on the west slope of the Cascades typically
are less than 0.06 mg L-1 (Brown et al., 1973; Feller and
Kimmins, 1979; Hair and Fredriksen, 1988; Martin and
Hair, 1989).
Paniculate inorganic phosphorus is mineral in origin
and enters the stream channel primarily by soil erosion and
sediment transport. Particulate organic phosphorus comes
from a variety of sources and can enter the stream channel
through fluvial transport or direct deposition. In slow-
moving streams and lakes, there may be a net loss of
phosphorus through the settling of sediment and organic
material (EPA, 1986b).
Phosphorus is not considered to be limiting to forest
growth in the Pacific Northwest and is rarely applied as
fertilizer (Gessel et al., 1979). Forest soils have a high
capacity to bind dissolved phosphorus through a variety of
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.
Relation to Designated Uses
Phosphorus is an essential nutrient for plant growth.
However, an increase in plant-available phosphorus may
not necessarily increase primary production, as other fac-
tors may be limiting (e.g., Scrivener, 1988). In small forest
streams light is often the key limiting factor (Gregory et al.,
1987). In contrast, larger streams and lakes are light-
saturated, and in theseaquatic systems nutrients tend to limit
primary production.
In aquatic ecosystems phosphorus is usually the limit-
ing nutrient (e.g., Mohaupt, 1986). A general rule of thumb
is mat the optimal nitrogen to phosphorus ratio for primary
production is 16:1; a lower ratio suggests that nitrogen is
limiting, while a higher ratio indicates that phosphorus is
limiting. A survey of streams in western Washington
indicated that phosphorus was more likely to limit primary
production in glacial streams and in streams draining gra-
nitic areas, while nitrogen was more often limiting in
streams draining volcanic landforms (Thut and Haydu,
1971). As discussed in Section 7.2, an increase in primary
production normally will stimulate invertebrate and fish
production.
The desirability of an increase in primary production
depends on the local and downstream uses of water. In
many forested areas an increase in stream production might
be considered beneficial. Often, however, these streams
flow into lakes and rivers that are extensively used for
fishing and recreation. Increased primary production in
downstream receiving waters due to nutrient enrichment
(eutrophication) can impair certain designated uses. Ad-
verse effects can include changes in water chemistry, re-
ductions in dissolved oxygen levels, less recreational use,
and a decline in esthetic values.
Naturally occurring phosphate concentrations present
few problems with regard to acute or chronic toxicity.
Laboratory experiments have shown that fish can accumu-
late elemental phosphorous from aqueous concentrations as
low as 1 jag L-1, but elemental phosphorous is not present
under natural conditions in forested areas (EPA, 1986b).
Response to Management Activities
Studies in the Pacific Northwest indicate that forest
management activities are unlikely to substantially increase
phosphate concentrations in aquatic ecosystems. Data from
small catchment studies in Oregon showed thatclearcutting
and burning had no effect on phosphate concentrations
(Brown et al., 1973; Hair 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
Other studies have suggested that phosphate losses
increase after wildfire, but the increases are relatively small
in both relative and absolute terms (Wright, 1981). Wide-
spread application of phosphate fertilizers could increase
aqueous phosphate concentrations, but phosphate fertil-
izers are rarely used in the Pacific Northwest (Gessel et al.,
1979).
Human waste can contribute significant amounts of
phosphorus to aquatic ecosystems. Dispersed recreation
and septic tanks are two common nonpoint sources. Many
rural communities are unable to afford the tertiary sewage
treatment necessary to remove the primary nutrients and
protect water bodies from euttophication. Livestock and
other animalsrepresentanotherpotentiallyimportantsource
of phosphate contamination. The chemistry of phosphorus
is such that most of the phosphorus entering into aquatic
ecosystems will be in the form of orthophosphates and
either sorbed onto soil particles or incorporated into organic
compounds. Thus soil erosion can be a primary source of
phosphorus, whereas nitrogen usually reaches aquatic sys-
tems in dissolved forms (Mohaupt, 1986).
Measurement Concepts
The procedures for analyzing the various forms of
phosphorus are discussed in considerable detail in standard
references (e.g., APHA, 1989;Stednick, 1991). Acompari-
son of inorganic paniculate and dissolved phosphorus can
help determine whether the primary source of phosphorus
is due to erosion or solute transport. Dissolved phosphorus
is the focus of most monitoring programs, and is often
separated into soluble and plant-available phosphate (i.e.,
orthophosphate), and total phosphate. Dissolved organic
phosphate can be estimated by subtracting soluble and
plant-available phosphate from total phosphate.
Since the analytical techniques are well defined and
most forest streams are considered to be well mixed, the
primary measurement problems are (1) determining the
timing, location and intensity of sampling; and (2) avoiding
contamination. Determining the type and intensity of mea-
surements largely depends on the monitoring objectives;
this is discussed in detail in Part I, S ection 4. Contamination
is a serious problem because the natural concentrations of
phosphorus in forested streams are <0.02 mg L-1 (EPA,
1986b; Kunkle et al., 1987).
If the primary monitoring concern is the total load of
phosphorus, both particulate and dissolved phosphorus
should be sampled. More intensive sampling will be needed
during high flows, which is when most of the particulate
phosphorus will pass by the monitoring station. Continuous
discharge data will be needed to calculate the total load
(concentration times discharge). Itmaybepossible to establish
a quantitative relationship between particulate phosphorus
and turbidity, and this could substantially reduce the costs of
monitoring particulate phosphorus.
Analysis of the effects of a particular managementactiv-
ity generally will entail more limited sampling at carefully
selected locations. For both project and trend monitoring,
an analysis of total phosphate often will suffice. Simulta-
neous discharge data are needed to help interpret the tempo-
ral variability in concentration and optimize sampling.
Standards
A national water quality criteria of 0.10 jig L-1 of
elemental phosphorus has been established for marine and
estuarine waters. No standard has been set for phosphate
concentrations in freshwater because the threat of eutrophi-
cation is so location-specific (EPA, 1986b). Some of the
specific factors thatmay affect the desirable phosphate level
include the extent to which phosphate is actually limiting
primary production, the technological feasibility of phos-
phate control, and the beneficial uses of the water body.
Although specific standards have not been established,
EPA (1986b) has made some general recommendations
regarding the maximum concentration of phosphorus in
streams and lakes. To prevent euttophication, total phos-
phates as phosphorus (PO4-3-P) should not exceed 0.025 mg
L-1 for any lakeorreservoir. (Total phosphates asphosphorus
refers to the mass of phosphorus atoms per liter; phosphorus
atoms represent only 32.6% of the total mass of phosphates
per liter.) Where streams enter into reservoirs or lakes, the
concentration of total phosphates as phosphorus should not
exceedO.OSOmgL-1. Mackenthun (1973) suggested thattotal
phosphorus concentrations shouldnotexceedO.lOmgL-1 for
streams that do not flow into reservoirs or lakes. Concentra-
tions of PO4'3-P above 0.10 mg L-1 may interfere with the
coagulation process in water treatment plants (EPA, 1986b).
Current Uses
Phosphate concentrations are not a regular part of most
water quality monitoring programs in forested areas. This
omission is due to the very low background levels of phos-
phate in forest streams in Alaska and the Pacific Northwest,
and the relatively minor impact of most forest management
activities on phosphorus concentrations. However, phos-
phorus often is the limiting nutrient in aquatic ecosystems,
and this may make it the primary constituent of interest in
some water bodies. Currently the load allocation process
(Part I, Section 1.4) has been initiated for phosphorus in
both the Tualatin (Oregon) and Spokane River basins. The
need to quantify all natural and anthropogenic sources
means that phosphorus concentrations in streams emanat-
ing from forested areas also must be monitored.
Assessment
Phosphate monitoring in forested areas is most likely to
be necessary when eutrophication threatens downstream
-------
CHAPTER 2. PHYSICAL AND CHEMICAL CONSTITUENTS
rivers, lakes, or reservoirs. Although participate sources
may need to be evaluated, it is much easier to monitor total
phosphorus only in downstream receiving waters most
likely to be affected by eutrophication. Changes in dis-
solved phosphorus concentration usually areeasier to detect
in lakes and reservoirs because the concentrations are less
variable over time (i.e., it is much less dependent on dis-
charge), and changes in phosphorus concentrations tend to
accumulate over time. This accumulation of phosphorus is
easier to detect than smaller, incremental increases in phos-
phorus inputs.
Morefrequentmonitoringofphosphatesinforeststreams
may be necessary if phosphate fertilization becomes a com-
mon practice. Under these circumstances the guidelines for
monitoring would be similar to the guidelines suggested for
monitoring nitrogen inputs from forest fertilization (Section
2.5.1).
Although phosphate monitoring may not be a necessary
component for most water quality monitoring programs in
forested areas, additional data on phosphate concentrations
can be useful. Information on the normal range of values in
different ecoregions and the temporal variation within
streams will be helpful if monitoring phosphorus loads
becomes necessary. Similarly, long-term data on phospho-
rus inputs, nitrogen inputs, and surrogates for free-floating
algal biomass (e.g., chlorophyll a) could help determine the
limiting factors for plant growth in larger water bodies, and
hence the likely response of a particular water body to a
projected change in light, nutrients, or other factors.
2.6 HERBICIDES AND PESTICIDES
Definition
Pesticides are chemicals used to control undesired plant
or animal species (i.e., pests). Chemicals for controlling
unwanted plants are called herbicides. Chemicals for con-
trolling animal pests commonly are called pesticides, even
though this is not consistent with the strict definition given
above. Other terms, such as insecticides, are specific to
particular groups of animals. Since herbicides may have
quite different effects on water quality and aquatic ecosys-
tems than the chemicals used for controlling animal pests,
there is a need to distinguish between these two broad
classes of pest-control chemicals. These Guidelines follow
the more common and practical approach of using the term
pesticide only for those chemicals which are directed against
animal pests.
Herbicides and pesticides can be applied directly by
ground-based methods or sprayed from aircraft. Since most
of the commonly used herbicides and pesticides affect a
broad range of organisms, their use has engendered consid-
erable controversy.
Most of this concern has focused on the aerial applica-
tion of herbicides and pesticides, as it is very difficult to
prevent some spray from being applied in or near the stream
channels. In addition to the contamination problem posed
by overspray and drift, herbicides and pesticides can be
transported from the point of application to the aquatic
ecosystem by leaching, volatilization, and erosion. In most
cases the risk of contamination from these three transport
processes is lower than risk of contamination from direct
overspraying and drift (EPA, 1977).
The susceptibility of a chemical to leaching depends on
its solubility in water and its tendency to adhere to soil
particles. Another pathway for these chemicals to reach the
aquatic ecosystem is by absorption onto soil particles and
subsequent erosion. The dependence of these transport
mechanisms on the movement of water means that the
potential for contaminating streams and lakes is closely tied
to the amount of precipitation, runoff, and erosion following
application. High runoff events shortly after application
generally pose the greatest risk for loss from the terrestrial
environment to theaquatic ecosystem (e.g.,Reynolds, 1989).
The tendency to use less persistent herbicides (i.e., ones
which break down more rapidly) and the relatively low
levels of pesticide application suggest that transport from
the terrestrial to the aquatic ecosystem is rarely a problem
except for a few chemicals such as picloram and atrazine
(Fredriksen et al., 1975; NCASI, 1984a).
Relation to Designated Uses
Although herbicide and pesticide contamination can
adversely effect several designated uses, the protection of
domestic water supplies has the highest priority. For many
of the more common herbicides, EPA has recommended a
maximum mean concentration for any 24-hr period. These
maxima have been derived from combining toxicity tests on
aquatic organisms with a safety factor (EPA, 1977). Since
most silvicultural chemicals have a half-life of no more than
a few weeks, the potential for bioaccumulation is relatively
low and the 24-hr standard is adequate. However, herbi-
cides, pesticides, and their intermediate breakdown products
areallsubjecttoadsorbtionbysoilorfinesedimentparticles,
and this can affect the relative persistence, toxicity, and
biologic uptake of these chemicals.
The possibility of sublethal effects on aquatic organ-
isms cannot be excluded, but the intermittent application,
short duration of exposure, and relatively low concentra-
tions all suggest that these effects generally are small or
insignificant. A detailed study of the effects of glyphosphate
on a stream in coastal British Columbia, for example,
confirmed short-term avoidance and stress by coho salmon
and two species of invertebrates in an oversprayed tributary;
long-term effects were deemed negligible (Reynolds et al.,
1989).
-------
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-leavedspecies, 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 and 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
thatlosses 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.).
Anotherapproach to assess oversprayanddriftis 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 quali tative 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 difference
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 maximarange 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 releasedinto theaquatic
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 significantiy 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 sprayingunder 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 reporting forms
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
waterquality parameters. Nevertheless, forestharvest, road
building, and other management activities can result in
substantial changes in the volume and timing of runoff, and
this has long been asourceof 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.
Flowparameterswere included in \heGuidelines because
of their potential sensitivity to forestmanagementactivities,
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 andrain-on-snow events. Bothof these lattercauses
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 the distribution 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
etal., 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 et al., 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 constructinga stilling well and establishing a stage-
discharge relationship. This is relatively expensive and
requires acontinuinginput of 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 first few storms following
the growing season, when the higher soilmoisture carryover
results in a greater proportion of runoff. Major floods, such
as those with a return interval of 50 years or more, should not
beasgreatiyaffectedbyforestmanagementactivities.asthe
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 sno wpack (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
magnitudeof any temperature increasedue 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 mostsmallcatchment studies in thePacific Northwest
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-
nent of 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 notexperiencesuch 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 changes in 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 aperfectrelationship 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 thesedata 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
waterformumcipal oragricultural 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
Northwestand 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 (Linsley et al., 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 small 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. ThecasestudyoftheSouthForkoftheSalmon
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-1
(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 otherfactorsarenotlimiting(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 andmixing
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 insuspended
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.
-------
Part II
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 sterns 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 cover; and surface erosion from burned areas or
areas disturbedby site preparation activities (Swanson et al.,
1987). Surface erosion can include both fluvial detachment
and transport as well as dry ravelandsurfacecreep(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 sedimentin 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 finesediments(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
increase in suspended sediment are as follows: (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
obtaininga 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 onapresumed relationship 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 sedimentconcentrations for aspecified
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
duringperiods 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 theconcepts
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.
Suspendedsediment 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 "setfleable 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-
mentmakeitoneofthemorecommonlyusedparametersfor
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 sampler is 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 inaparticularbasin. However,themultipleprocesses
involved in sediment storage and delivery preclude the use
of suspended sediment concentrations as a quantitative
measure of specific hillslope 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
have the greatest geomorphic impact (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
orabsorbed 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.
Turbidity instreamsisusuallyduetothepresenceofsuspended
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).
These properties 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
watertechnically is safe to drink. However, suspendedmatter
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. Arecentreview
concluded that the ability of salmonids to find and capture
food is impaired at turbidities in the range of 25-70 MTU
(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 thatan increase in turbidity can lead
to an increase in surface water temperature and a resultant
decline in therateof 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, the same activities that generate largeamounts
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 et al., 1988). Road
reconstruction duringthe summer of 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 there is a strong relationshipbetween 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
sedimentis nonlinearon an arithmetic plot. Generally about
80% of the variability in suspended sediment concentra-
tions can be explained by simultaneous turbidity measure-
ments. Detailed analysesof 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 meother hand, three watersheds in theOregon Coast
Range showed significant differences in the relationship
between suspended sediment and 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 the stream 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 inhibitor restrict algal growth, butprotocols 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 higherdischarges (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
errorshasimportantimplicationsbothfordetectingincreases
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
currentEPAregulationsrequire turbidity inmunicipal water
supplies not to exceed 1 MTU (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
commonlyusedformonitoringnonpointsources 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 canbetransportedeither 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 asimple 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
(Section 5.6.1; 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
sourcesof fish habitat. Atvery 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 the amountofmicrohabitatavailableforjuvenile
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 (Everest etal., 1987). The deposition
of fine sedimentreduces the habitat space for young fish and
aquatic macroinvertebrates (Sections 5.6.1, 7.3, and 7.4;
Everest etal., 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
arespentin the U.S.each year to remove sedimentdeposited
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
cutand fill areas (e.g., Megahan, 1978), and altering drainage
patterns in ways that tend to increase the numberof 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 evaluatingtheimpactof 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, erodibje 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
transportrates due to forestmanagement activities, and then
relate these changes 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 notallow separation ofbedloadand suspended
load.
The difficulty of accurately measuring bedload has led
to the development of numerous equations to predict bedload
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 largest runoff 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
differentcross-sectionsduringmoderately high flows. These
measurements may indicate the flow at which the bed
material begins to move, and this provides a useful 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 cannotbemadeexceptin cases whereadownstream
trap is available that can be surveyed on a regular basis. The
high year-to-year variationin 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 most field investigations are conducted
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,
whenseparated by particle sizeandcombined with discharge
data, may help this interpretation process. Bedload data
rarelyprovide 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-
-------
Part II
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. Measurements of habitat type(e.g.,pools, riffles,
etc.) were pioneered by fish biologists and are used to
evaluate the quality of fish 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 amongthesemonitoring
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 directionof flow. Cross-sectional dataareobtainedby
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 is akey parameter in most hydraulic
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 (thedeepestportion 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. Thiscanhavebeneficialoradverseeffects depending
upon theinitial conditions and thedesignateduse(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
sedimentloadsfrompreviousmanagementactivities.channel
incision may represent a return to "natural" conditions and
an improvement in habitat quality and channel capacity
(e.g., Megahan et al., 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 detectchange 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
to meet the statistical requirements (Platts etal., 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 netsedimenttransportwithinaparticular
reach, and evaluate changes in channel morphology (e.g.,
width-depth ratio, bank slope, andbankfull 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
havebeencollectedbuthavenotbeen 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 arelatively 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 as bed material particle size 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. Adecreasein depth tends to reduce the number
of pools (Beschta and Plaits, 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
erosion andacorrespondingincrease in sedimentinputs into
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
sedimentdeliveredto 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 widthor the width-
depth ratio can be used as an indicator of a change in the
relativebalancebetween the sediment load 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
consistent channel form such as straight riffle reaches, 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.
Determiningwidthanddepthatastandard 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 and are relativelyinsensitive 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 monitoringprojectinto the contextof
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-term measurements at various
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 aconcave 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
113
-------
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).
In larger streams with deeperpools, directmeasurements are
considerably more difficult. Also, a series of conceptual
problems inmakingpoolmeasurements must beconsidered
before embarking on a classification or monitoringprogram.
First, it may be difficult to determine exactly what
constitutes apool. 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 measuringpool 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
Most surveys 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 lowormoderategradientstreams
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 particle 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 at any 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 profUe and 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 gradientas determined by athalwegprofile
is an important criterion for classifying streams. The
channel gradientalso 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 Platts, 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
difficulttopredictthe precise effect ofachangeinonefactor
on the change in other factors. Beschta and Platts (1986)
suggest that stream channel morphology is affected more by
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. Surveyingequipmentis needed to obtain sufficient
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 II
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
thalwegprofile. 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.
Thesamestudiesthatsupport 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 channel characteristics. 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 thechannelmorphology 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 recent stream 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
et al., 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, dammedpools become
larger, while low gradient riffles and scour pools may become
glides. Habitat unit surveys must becarriedoutat 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 withagravelsubstrate.forexample.probablywillhave
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 shiftin the relative
proportions of habitat types has occurred, and any change can
be related to both cause (i.e., the physical processes causing
thechange)andeffect(changeinvalueof fisheries resources).
Response to Management Activities
Therelativelyrecentdevelopmentofhabitatunitsurveys
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 managementimpact(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 theareaofeachhabitatunit, 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 wherever possible 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.
Useofhabitatunitsformonitoringenvironmental change
hasnot been extensively tested becauseofthepaucityof long-
term data. Extensive stream surveys that estimate or measure
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 to determine what level of change is acceptable and
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, GiffordPinchotNafl.
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 medianparticle 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 d16 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
thebed 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
acloserelationshjpbetween 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
median bed material particle size. Forestharvest,roadbuilding
andmaintenance,andplacerminingalltendtoincreaseerosion
and sediment delivery rates (S wanson et al., 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 using liquid COi or liquid nitrogen. The
frozen sample is then thawed and sieved in order to obtain
the particle size distribution. One major advantage 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
particle diameter in mm. Use of the phi index normalizes the
particle-size distributions so they can be analyzed using
parametricstatisticsandplotteddirectly on arithmetic graph
paper (Wolman, 1954).
The selection of the sampling technique should be
determined by the objectives of the sampling. 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 indicate a change in the sediment load (Dietrich
et 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, andstream
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
Very large boulders 4,096 - 2,048
Large boulders 2,048 -1,024
Medium boulders 1,024 - 512
Small boulders 512-256
Large cobbles 256 -128
Small cobbles 128-64
Very coarse gravel
Coarse gravel
Medium gravel
Fine gravel
Very fine gravel
Very coarse sand
Coarse sand
Medium sand
Fine sand
Very fine sand
Coarse silt
Medium silt
Fine silt
Very fine silt
Coarse clay
Medium clay
Fine clay
Very fine clay
64-32
32-16
16-8
8-4
4-2
2.0-1.0
1.0-0.5
0.50 - 0.25
0.250-
0.125
0.125-
0.062
0.062 - 0.031
0.031 -0.016
0.016-0.008
0.008 - 0.004
0.004 - 0.0020
0.0020-0.0010
0.0010-0.0005
0.0005 - 0.00024
16-80
80-40
40-20
20-10
10-5
5-2.5
2.5-1.3
1.3-0.6
0.6 - 0.3
0.3-0.16
0.16-
0.08
0.08-
0.04
0.04-
0.02
0.02-
0.01
0.01 -
0.005
0.005 -
' 0.0025
-12-(-11)
-11-(-10)
-10-(-9)
-9-(-8)
-8 - (-7)
-7-(-6)
-6 - (-5)
-5 - (-4)
-4-(-3)
-3 - (-2)
-1 -(0)
0-1
1 -2
2-3
3-4
4-5
5-6
6-7
7-8
8-9
9-10
10-11
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 parameters) 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 d50 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 amongecoregions,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 ah*
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
from acoaf se-bedded stream. Hence samplingimmediately
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
materialparticlesize, for example, might be done on 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
facilitatecross-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). Themethod
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 = ZDf/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 Df (note: Df = 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%. Salmonid density also declined with an embeddedness
of 50% or more. It was inferred that an increase in embedded-
ness wouldreduce winter habitat, with the preciserelationship
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 Dc from each hoop are
summed, and a percent cobble embeddedness (PCE) for
each hoop is calculated from the formula:
PCE = EDe/ZDt
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
(Part I, 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.
fl23
-------
Part II
In developing a monitoring plan using embeddedness,
the objectives will dictate whether hoops or individual
particles shouldbe 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 + (I-HAIE
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 theparticles 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 Platts (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 thereladonshipbetweenpercentcobbleembeddedness
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.unpublishedstudiesbyfederalandstateagencies
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 beingmeasuredinanumberof
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-WildUfestreamsurveyprogram. 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 proveusef ul only in certain environments
and stream types. Mostof the workon 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 sedimenttransport 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). Adimensionlessratio, q*, has been defined as the
estimatedbedload transport rate 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 et al., 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
The 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 etal. (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
clearcutting andbroadcastburning—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.
Along-term study in theSouthForkof 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 apparentstateofequilibrium had beenreached, 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, theaccumulation 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-
ticles were still accumulating in thedeeper layers (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 theparticle-size distribution of
the surface layer, but this method cannot be used for the
subsurface layers. McNeil samplers do not allow 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 etal., 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 d50
(Platts et al., 1983), comparisons generally should incorpo-
rate some measure of the variation in the particle-size
distribution, such as the d84 and the dj6 (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 suggest 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-surface particle 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 created by 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.
Surface 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 couldbe 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),andlargewoody 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 theeffects 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 etal., 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 atmoving 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 caused6% 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 sediment in 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 one of the mostimportant sources of habitatand
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
co ver and suitable habitat can be found over a wide range of
flow and climatic conditions. LWD also may allow a finer
partitioningof Unavailable habitat. Pools formed by 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 et al., 1987).
Several potentially detrimental effects are associated
withLWD 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 et al., 1987).
Typically the average piece size also declines becauseof 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, the capacity of a system 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 arecentreview of techniques
to measure and map LWD. Selecting a 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 estimates
by measuring a subsample of the pieces (Hankin andReeves,
1988). Biomass of LWD also can be estimated with tech-
niques derived from inventories of forest residues (Van
Wagner, 1968). This procedure inventories all LWD inter-
sectedbyaseriesof cross-sections across the stream (Froelich
etal., 1972;Lammel, 1972), and is most applicable when the
minimum piece size is relatively small. Special procedures
or categories may be needed for measuring debris jams,
standing trees, and snags within the stream channel (Platts
etal., 1987).
Information on channel features associated with LWD
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 andWard, 1987). Surveys of habitat types orchannel
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 being considered include defining 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, or removal of standing timber from 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 ofLWD.
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
timber 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 at the time of themanagementactivity. 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° (Plaits 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
natural events thatdisruptthequasi-equilibrium of thestream,
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 are discussed in Chapters 3,4 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
impacton 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
evaluatestream channel condition. This uses4-6 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.
Asimilarprocedurecanbeusedtodeterminestreambank
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
from thestream. Again theactualpercentageofthestreambank
that has been altered is estimated to yield a quantitative
rating of streambank soil alteration. Platts et al. (1983,
1987) found mat this technique had wider confidence inter-
vals than the streambank stability rating, but the accuracy
-------
Part II
Table 10. Parameters and range of values used for evaluating
stream channel condition (Pfankuch, 1978).
Channel location
Parameter
Range of values
Upper bank
Sideslope 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,
pers. 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 el
al., 1987), (2) inability to identify specific causes of bank
instability (Platts et al. ,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 that a 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
amountofsedimentandnutrientsreachingthestream 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 widthofthe 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
theirpurpose, 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
dominantupslope species ratherthantheriparian 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
highlyinteractiveresponseexists between changes inchannel
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 die 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 directradiation to reach the stream andraise
peak summer water temperatures. Less shading also will
result in greater temperature fluctuations on both a seasonal
and a daily basis (Section 2.1). Areduction 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-
creasein invertebrateand fish production. However, increased
primary productivity may be offset by decreased inputs of
detrital food subsidies, leaves, and other organic material
from theriparianzone. The netbalancebetween the increased
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 the riparian 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
element in channel morphology, sediment transport, and
quality of aquatic habitat (Bisson et al., 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) determine thecontinuity of openreaches
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. S uggested procedures
toanalyzeand 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 the U.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 overa 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
during such amonitoringperiodcannotbeplaced in 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-typeapproach
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-
port a prevalence of vegetation typically adapted for a life in
-------
Partll
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;
Gresswcll et al., 1989). A recent bibliography on riparian
research and managementlisted 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 typesof 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 theirroots can provide substantial protec-
tion during high flows. Grassy banks may provide a more
complete cover, butthey 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.root densities, chemicalproperties of the soil, 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 offish 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
zonebecause 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 greatestthreat 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 have developed
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 the amount 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 Stubbendieck,
1986;Huschetal., 1982; Tueller, 1988) shouldbeconsulted
for more details on vegetation monitoring techniques.
Standards
In the 1970s the forestpractice rules for riparian areas in
Oregon and Washington were designed to maintain ad-
equate shadeandminimizetheintroduction 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 under the Timber-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 areafunctions and management (Raedeke, 1988;
Gresswelletal., 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 15m
Representative of stand
Representative of stand
2to1
Minimum size of
leave trees
Representative of stand
Representative of stand
30 cm or next largest
available0
Rumber of trees/
305 m each side
Gravel/ Boulder/
cobble3 bedrock
50 trees 25 trees
1 00 trees 50 trees
75 trees 25 trees
3 water less than 2 m
8m
to1
15 cm or next largest
available
25 trees 25 trees
•Gravel and cobble 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.
=Or next largest available requires that the next largest trees to
those specified in the rule be left standing when those available
are smaller than the sizes specified. Ponds or lakes which are type
1, 2, or 3 waters shall have the same leave tree requirements as
boulder/bedrock streams.
For wildlife habitat within the riparian management zone, leave an
average of 12 undisturbed and uncut wildlife trees per hectare at the
ratio of 1 deciduous tree to 1 conifer tree equal in size to the largest
existing trees of those species within the zone. Where the 1 to 1 ratio
is notpossible, then substitute either species present. Forty percent
or more of the leave trees shall be live and undamaged on completion
of harvest. Wildlife trees shall be left in clumps, whenever possible.
Eastern Washington
(A) Leave all trees 30.5 cm or less in diameter breast height (dbh).
(B) Leave all snags within the riparian management zone that do
not violate the state safety regulations.
(C) Leave 40 live conifer trees/hectare between 30,5 cm dbh and
50.8 cm dbh distributed by size as representative of the stand.
(D) Leave 7 conifer tree/hectare 50.8 cm dbh or larger.
(E) Leave the 5 largest deciduous trees/hectare 40.6 cm dbh or
larger. Where these deciduous trees do not exist, and where
5 snags/hectare 50.8 cm in dbh orlarger do notexist, substitute
12 conifer trees/hectare 50.8 cm in dbh or larger. If conifer
trees of 50.8 cm dbh or larger do not exist within the riparian
zone, then substitute the 5 largest conifer trees/hectare.
(F) Leave 7 deciduous trees between 30.5 cm and 40.6 cm dbh
where they exist.
(G) On streams with a boulder/bedrock bed, the minimum leave
tree requirement shall be 185 trees/hectare 10.2 cm dbh or
larger.
(H) On streams with a gravel/cobble (less than 25.4 cm diameter)
bed, the minimum leave tree requirement shall be 335 trees/
hectare 10.2 cm dbh or larger.
(I) On lakes and ponds the minimum leave tree requirement shall
be 185 trees/hectare 10.2 cm dbh or larger.
-------
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 largeepisodic 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 stabili ty 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 canbeusedtodefinetheconditionsinthatparticular
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 bedone 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 have different die-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 notaparticularlyaccurateorconsistentindicator
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
for up to48 hr to determine if gas bubbles form; gas formation
is regarded as a sign that coliform bacteriaare 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). For this reason at least
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
thrcereplicates doubles 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. Thefilter
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 themembranefiltertest.
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 heterotrophic 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 fresh water bathing, the geometric mean value of at
least five samples equally spaced over a 30-day period
should not exceed 126 E. coli/lOQ 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 costlyandnot 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 a particular significance, and the use of 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
frequent transport. 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 benthic communities (Pry fogle
andLowe, 1979). Epiphyticalgaereferstoattachedmicroalgae
(e.g. diatoms) that grow on the surface of macrophytes
(Cattaneo and Kalff, 1980).
-------
Part II
Relation to Designated Uses
Benthic 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; Power etal., 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, the importance of autotrophic production increases.
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 et al., 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). Developmentof anaerobicconditions
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 littledefinitive 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
(Stockner and Shortreed, 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 primaryproducers 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 theproductivity and composition of theaquatic
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 et al., 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 etal., 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 et al., 1979).
The values andlimitations of speciesrichnessanddiversity
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 and4%
(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 nutrientavailability (Clark etal., 1979;
Hudon and Legendre, 1987).
-------
Part II
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 the
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
biologicalintegrity of waterbodies. 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 butalsocanbe related
to a designated use, such as recreation. In streams, however,
benthic algae production and biomass probably are the most
useful of all theaquatic 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 et al., 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 ORGANISMS
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
structureand functions of aquatic ecosystems shouldimprove
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 et al.,
1983), while APHA (1989) defines macroinvertebrates as
those invertebrates retained on aU.S. StandardNo. 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 periphy ton (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 degreeof 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). This means 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 afterpartial or complete removal of theriparian
canopy (e.g., Hansmann and Phinney, 1973; Murphy et al.,
1981). However, no increase was found in Carnation Creek
in coastal British Columbia, wherephosphorus was found to
be the limiting factor (Stockner and Shortreed, 1988).
Logging-induced increases in aquatic insects have been
observed in northern California (Erman et al., 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 (Newboldetal., 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, functional feedinggroup 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 et al., 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 serniquantitative methods
is thatresults are onapersamplebasis rather than per unitarea
(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 are quicker 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, mostprojects should use 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
disadvantages of artificial substrates are discussed in Greeson
etal. (1977), Rosenberg andResh (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 S urber
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-
ingtheyearbecauseof natural lifecycle processes (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, butwhen this is notpossibletheoptimal 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
typeofinvertebratecommunity observed. The most diverse
invertebrate communities generally occur in riffie/run
habitats with gravel and cobble bottoms (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). Itis therefore beneficial to use more than one
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 the community balance.
Communities dominated by just a few taxa indicate 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 in North Carolina. In Oregon, species richness
showed good correlation with trout populations from high
-------
Part II
desertstreams (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 importantfortheproper interpretation of species richness
data. EPAhaspublishedpollution tolerance information on
mostmajoraquatic 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 (Platts 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 in use benthic macroinvertebrates to assess
water quality impairment; protocols IV and V use fish. RBP
I relies upon the qualitative abundance of different
macroinvertebrate 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 particulate 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 particulate 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 III, 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
the Clean 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 five ecoregions within the state. These indices have been
successfully 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. Qual.,
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.
The Rapid Bioassessment Protocols (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 a particular 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 thesections 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 spe-
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), Everest et al. (1985), Reiser and Bjornn (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 forest management on streams. Studies in coastal Oregon
(Hall etal., 1987), southwestern British Columbia (Hartman
etal., 1987; Chamberlin, 1988), southeastern Alaska (Gib-
bons etal., 1987),andtheOlympicPeninsula in Washington
(Cederholm and Reid, 1987) investigated the response of
the fish populations to different types and intensities of
forcstmanagementpractices. An important, unify ing result
of these studies is that forest management can affect a wide
variety of physical and biological parameters, including
temperature, bedmaterial,primaryproductivity,peakrunoff,
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
forestharvestor other managementactivities. 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 offish (e.g.,
Hartman et al., 1987).
Most of the links between management activities,
physical and biological change, and effects on fish are
discussed in the context of 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 offish. 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
can countthe number of outmigrating juveniles (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 et al., 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, snorkelingpermits a true estimate
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 beobtainedbycountingspawningnests(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), large rivers (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 water quality (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 obtained by 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 et al., 1988).
The IBI is the basis for EPA's Rapid Bioassessment
Protocol (REP) 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).
153
-------
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 offish 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,arecoveryplanmustbepreparedfor 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-
fect habitat quality could be precluded or severely 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, the 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
theknownorexpectedmanagementimpacts.onecan 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
establish a cause-and-eff ectrelationship. Data on the streams
or lakes of concern also should be compared to data from
unimpaired sites.
The IBI shows promise as a technique to evaluate stream
condition from a variety of measurements offish populations,
trophic structure, and species composition (Karr, 1981; Miller
etal., 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). At this time most of
the work on biolo-gical assessment and 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
ahigh public profile. Often the moststringentconstraints 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 mat 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 pas t conditions,
but it also is a disadvantage because the number offish 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 si2e 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
forstockmanagementandotherpurposes,andtheavailability
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., andK. 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 insouthwestOregon. 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 FJP.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.
Angermcier,P.L.,andJ.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. Standard methods for the examination of water and
wastewater. American Public Health Association. Washing-
ton, D.C. 1193 p.
APHA, 1980. Standard methods for the examination of water and
wastewater, 15th ed. American Public Health Association.
Washington, D.C. 1134 p.
APHA, 1989. Standardmethodsfortheexaminationofwaterand
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 atWagon 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. USDAFor. Serv.,
Gen. Tech. Rep. RM-28. Fort Collins, CO. 45 p.
Berman, C.H. andT.P. Quinn, 1990.Theeffectof elevatedholding
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 in the Oregon Coast
Range. Wat. Resourc. Res. 14:1011-1016.
Beschta, R.L., 1979. Debris removal and its effects on sedimen-
tationin an Oregon CoastRange 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:
fisheriesandforestryinteractions. Pages 191-232inE.O. Salo
and T.W. Cundy (eds.), Streamside Management: Forestry
and Fishery Interactions. Contr. no. 57, Ihst. 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
Bevan, 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 Ecol.
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 in forested streams in the Pacific Northwest:
past, present and future. Pages 143-190 in E.O.SaloandT.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., Inc. 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, Inc. 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
atgagingstations. U.S.G.S., U.S.Dep.Interior,Book3,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.,andR.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 inreceiving 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. Oceanpgr. 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.,andK.P.McLeod, 1987. Developmentof 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.
Goodell, 1975. The evaluation of initial changes in peak
streamflow following logging of a watershed on the west
coast of Canada. Lit. Hydrol. Sci. Publ. 117:475-486.
Cherry, J., and R.L. Beschta, 1989. Coarse woody debris and
channel geomorphology: a flume study. 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 in R.E., Gresswell, et al. (eds.),
Riparian Resource Management. U.S. Bureau of LandMan-
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, EJ. Gilroy, RM. 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 SP. Gessel et al. (eds.),
ForcstFertilizationConference,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.Tutry, 1979. Juvenile salmonid field trapping
manual. Dcpt 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. Lit. Pac.
Sal. Fish. Comm. Bull. 18. 71 p.
Coutant, C.C., 1969. Behavior of a sonic-tagged chinook salmon
and steelhead 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.
Culp, J.M., 1988. The effect of streambank clearcutting on the
benthic 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
CreekSteeringCommittee,c/oPac.Biol.Sta. Nanaimo,B.C.
Cupp, C.E., 1989. Stream corridor 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.
Deuscn, 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.
S wanston (eds.), Sediment Budgets and Routing in Forested
Drainage Basins. USDAFor. Serv., Pac. Northw. For. Range
Exper. Sta., Gen. Tech. Rep. PNW-14L
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, Lie. 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. Quality criteria for water: 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^t40/5-88-089. 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.
SedeU, T.N. Nickelson, J.M. Johnson, and G.N. Haugen,
1985. Salmonids. 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. Kimmins, 1979. Chemical characteristics of
small streams near Haney in Southwestern British Columbia.
Wat. Resourc. Res. 15(2):247-258.
Ferguson, R.I., 1986. Riverloadsunderestimatedbyratingcurves.
Wat. Resourc. Res. 22(l):74-76.
Foster, I.D.L., J.A. Dealing, 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 hi
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), ForestLandUses 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. Bender 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, andM.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 and T.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 Watershed Management, 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
LakeTahoe: the first fiveyearsof the Lake Tahoelnteragency
Monitoring Program. Tahoe Research Group, Institute of
Ecology, Univ. California. Davis, CA. 12 p.
Grant, G.E., 1986. Downstream effects of timber harvestactivities
on the channel and valley floor morphology of western
Cascade streams. Ph.D. dissertation, John Hopkins Univ.
364 p.
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.B. Kauffman, 1989. Nutrient cycling at the
land-water interface: the importance of the riparian- zone.
Pages 61-68 tnR.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.L
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 and T.W.
Cundy (eds.), StreamsideManagement: Forestry andFishery
Interactions. Contr. no. 57, List. 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. Billings,MT. 193p.
Grette, G.B., 1985. The role of large organic debris in juvenile
salmonidrearing 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. U.S. 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,
andlanduse in theSalmonRiver Drainage, Idaho. Pages 181-
182 in R.E. Gresswell et al. (eds.), Riparian Resource Man-
agement U.S. Bureau Land MgmL 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 /nT.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: aretrospective. Pages 399-416 in 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.
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
periphyton in coastal streams of Oregon. Ecology 54:194-
199.
Harmon, M.E., J.F. Franklin, F.J. Swanson, and others, 1986.
Ecology of coarse woody debris in temperate ecosystems.
Pages 133-202 in A. Macfadyen, and E.D. Ford (eds.), Ad-
vancesinEcologicalResearch: VolumelS. 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
streamflows. Paper presented at Soc. Am. Foresters 1989
Annual Convention. Spokane, WA. 24 p.
Hair, 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.,andR.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 instorm
hydrographs after road building and clear-cutting in the
Oregon Coast Range. Wat. Resourc. Res. 11(3): 436-444.
Harris, T., and T. 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, LJB., and L. Powell, 1987. Some
effects of different Streamside treatments on physical condi-
tions and fish population processes in Carnation Creek, a
coastalrainforeststreaminBritishColumbia. 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. Technical review of sediment criteria. 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, E.J., andW. 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. 91:420r
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, andT.W.Beers, 1982. Forest mensuration.
John Wiley & Sons. New York, NY. 402 p.
Hutchinson, G.E., 1975. Limnological botany: Volume IH of a
treatise on limnology. John Wiley & Sons. New York, NY.
Hynes, HJ8.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.
Corvallis, OR. lip.
Ice, G.G., 1991. Dissolved oxygen and woody debris: detecting
sensitiveforeststreams. Paperpresented at ASCEGas 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. Assessment of biotic integrity using 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 K.J. Raedeke (ed.),
Streamside Management: Riparian Wildlife and Forestry
Interaction. Cont. no. 59, Inst. Forest Resources, Univ. of
Washington. Seattle, WA.
Keeton, W.T., 1967. Biological science. W.W. Norton and Co.,
Inc. 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 steelheadpopulations in Lagunitas Creek, MarineCounty,
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. Eos 70(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
qualitativecollectionmethodforbenthicmacroinvertebrates.
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, and J.L.H. Paulhus, 1982. Hydrol-
ogy for engineers. McGraw-Hill, San Francisco. 508 p.
Lisle, T.E., 1982. Effects of aggradation and degradation on riffle-
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(1): 18-33.
Lotspeich, F.D., and F.H. Everest, 1981. A new method for
reporting and interpreting textural composition of spawning
gravel. USDAFor. 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. Verb. Int. Verein. Theor. Angew. Limnol.
19:1610-1616.
MacDonald, L.H., 1989. Cumulativewatershedeffects: the impli-
cations of scale. Eos 70(43): 1114-1115.
Mackenthun,, KM., 1973. Toward a cleaner aquatic environ-
ment U.S. Environmental Protection Agency. Washington,
D.C.
Mars ton, 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. 19:35^3.
Mason, W.T., C.I. Weber, P.A. Lewis, and B.C. Julian, 1973.
Factors affecting the performance of basket and
macroinvertebrate 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 IdsutcK). Can. J. Fish.
Aquat.Sci. 46:1551-1557.
McNeil, W.J., 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.
Mcgahan, W.F., 1978. Erosion processes on steep granitic road
fills in central Idaho. Soil Sci. Soc. America J. 42(2):350-
357.
Mcgahan, W.F., 1982. Channel sediment storage behind obstruc-
tions in forested drainage basins draining the granitic bedrock
of the Idaho Batholith. Pages 114-121 in F.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.
Mcgahan, W.F., and C.C. Bonn, 1989. Progressive, long-term
slope failure following road construction and logging on
noncohesive,graniticsoilsoftheIdahoBatholith. Pages 501-
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.
Mcgahan, 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. 7:410-432.
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.andD.J.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. •Structural and temporal
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 aflatland
river system and optimization of sampling. Pages 297-304 in
Monitoring to Detect Changes in WaterQuality 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 andManage-
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
forestandrangelandmanagementonanadromousfishhabitat
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, BJ3., Jr., M. Afzal, andW.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., andK.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.C., 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 amillion. 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 and streambanks with respect to grazing in
Northeastern Utah. Pages 73-81 j«R.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, andC.K. Graham,
1989. Changes in salmon spawning and rearing habitat from
increased delivery of fine sediment to the South Fork S almon
River, Idaho. Trans. Am. Fish. Soc. 118:274-283.
Porterfield, G., 1972. Computation of fluvial-sediment discharge.
Chapter C3/n 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. BoiseNational Forest cobble embeddedness
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 j'wT.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,
W.J.Matthews,P.B.Moyle,B.Statzner,I.R.WaisDeBadgen,
1988. Biotic and abiotic controls inriver and stream commu-
nities. J. North Amer. Benthological Soc. 7:456-479.
Pryfogle, A.A., and R.L. Lowe, 1979. Sampling and 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, K.J. (ed.), 1988. Streamside management: riparian
wildlife and forestry interactions. Contribution No. 59, Inst.
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 Silviculture for wildlife and fish: a tinie 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. and T. 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
Surf ace 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, and D.L. Johnson (eds.), Fisheries Techniques. Am.
Fish. Soc. Bethesda, MD.
Reynolds, R.C., and N.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, L.B. 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. Pages490-498 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,
Dubuque, Iowa. 603 p.
Rodgers, J.H., Jr., K.L.Dickson, andJ. Cairns, 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.
Rosgen, D.L., 1985. A stream classification system. Pages 91-95
in 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.
Rothachcr, J., 1970. Increases in water yield following clear-cut
logginginthePacificNorthwest 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. Grant. In press. Downstream effects of
timber harvesting on channel morphology in Elk RiyerBasin,
Oregon. J. Environ. Qual. 20.
Salo, E.O., and T.W. Cundy (eds.), 1987. Streamside Manage-
ment: Forestry and Fishery Interactions. Contr. no. 57, Inst.
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
Precambrian lake basin in central Canada. J. Fish. Res. Board
Can. 33:2526-2543.
Schumm, S. A., 1960. The shape of alluvial channels inrelation 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 16 years 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
Northweststreamhabitatmanagementworkshop. 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 wtW.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. Effectsof chronic turbidity onfeeding, growth
andsocial 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. Tech. 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.O. Salo andT.W. Cundy (eds.), Streamside
Management: Forestry and Fishery Interactions. Contr.no.
57, Inst. 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. Thome, 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. Centre, 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.),Streamside Management: Forestry andFishery
Interactions. Contr. no. 57; Inst. Forest Resources, Univ.
Washington. Seattle, WA.
Tassone,B.L., 1988. Sedimentloadsfrom 1973 to 198408HB048
Carnation Creek at the mouth, British Columbia. Session 2:
Sediment and gravel. 7«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 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. 9 p.
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/
fish population relationships. USDAFor. Serv., Litermount.
Sta., Progress Report I to Idaho Fish Game Dep. 55 p.
Triska, F.J., F.C. Kennedy, R.J. Avanzino, andB.N. Reffly. 1983.
Effect of simulated canopy cover on regulation of nitrate
uptake and primary production by natural periphyton com-
munities. Pages 129-159 i'«T.D. Fontaine, and S.M. Bartell
(eds.), Dynamics of Lotic Environments. Arm Arbor Science
Publishers, Inc. Ann Arbor, ML
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., and R.M. King, 1985. The effect of 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. Dept. 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.,
Ihtermount. Sta. Boise, ID. 505 p.
VanHaveren,B., 1988. Areevaluation 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 and how? Page 207-215 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, G.P., 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. Eos70(43):1106.
Winget, R.N., and F.A. Mangum, 1979. Biotic condition index:
integrated biological, physical, and chemical stream param-
eters for management USDA For. Serv., Intermount Reg.
Ogden, UT. 51 p.
Wissmar, R.C., M.D. Lilley and M. DeAngelis, 1987. Nitrous
oxiderclease from aerobic riverine sediments. J. Freshwater
Ecol.4(2):209-218.
Wissmar, R.C., DM. McKnight, and C.N.Dahm, 1990. Contribu-
tion of organic acids to alkalinity in lakes within theMountSt.
Helens blast zone. Limnol. Oceanogr. 35:535-542.
Wissmar, R.C., andFJ. Swanson, 1990. Landscape disturbances
and lotic ecotones. Pages 65-89 in R. J. Naiman and H. De-
camps (eds.), The Ecology and Management of Aquatic-
Terrestrial 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 onstreamflow: 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 in spawning beds. Trans.
Am. Fish. Soc. 104(1):111-121.
Zand, SM., 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.
------- |