United States ' Office of EPA/600/3-91/022
Environmental Protection Research and Development February f99f
Agency Washington, DC 20460
&EPA • Surface Waters
Monitoring and Research
Strategy - Fiscal Year 1991
Environmental Monitoring and
M Assessment Program
!sj2g"'s"s"~ "'"*•="
-------
-------
EPA/600/3-91/022
February 1991
ENVIRONMENTAL MONITORING
AND ASSESSMENT PROGRAM
(EMAP)
SURFACE WATERS MONITORING
AND RESEARCH STRATEGY
FISCAL YEAR 1991
Environmental Research Laboratory
Office of Research and Development
United States Environmental Protection Agency
Corvallis, Oregon 97333
Environmental Monitoring Systems Laboratory
Office of Research and Development
United States Environmental Protection Agency
Las Vegas, Nevada 89153
Environmental Monitoring Systems Laboratory
Office of Research and Development
United States Environmental Protection Agency
Cincinnati, Ohio 45219
February 1991
Printed on Recycled Paper
-------
EMAP-SURFACE WATERS MONITORING AND RESEARCH STRATEGY
FISCAL YEAR 1991
February 1991
By:
Steven G. Paulsen3, David P. Larsenb, Philip R. Kaufmann0, Thomas R. Whittierd, John R. Baker6,
David V. Peck6, Joan McGue6, Robert M. Hughesd, Dennis McMulien, Don Stevens ,
John L Stoddardd, James Lazorchak9, Wesley Kinneyh, Anthony R. Selle , Randy Hjortd
With contributions from:
W. Scott Overton', Jim Pollard6, Daniel Heggemh, Gary Collins9, Marilyn Morrisond,
Colleen Birch Johnsond, Sandra Thieled, Nita Tallent-Halsell6,
Kit Peres6, Susan J. Christie6, and Janet Mellod
The research described in this report has been funded by the U.S. Environmental Protection
Agency. This document has been prepared at the EPA Environmental Research laboratory in
Corvallis, Oregon, through cooperative agreements CR814701 with University of Nevada-Las
Vegas, CR815168 with Utah State University, and CR815422 with Oregon State University, and
contract numbers 68-C8-0006 with ManTech Environmental Technologies, Inc., and 68-03-3249
with Lockheed Engineering and Sciences Company. The report has been subjected to the
Agency's peer and administrative review and approved for publication. Mention of trade names
or commercial products does not constitute endorsement or recommendation for use.
? Environmental Research Center, University of Nevada-Las Vegas, NV.
U.S. EPA Environmental Research Laboratory, Corvallis, OR.
~. Utah State University, c/o U.S. EPA Environmental Research Laboratory, Corvallis, OR.
ManTech Environmental Technologies, Inc., U.S. EPA Environmental Research Laboratory, Corvallis, OR.
? Lockheed Engineering and Sciences Company, Environmental Monitoring Systems Laboratory, Las Vegas, NV.
Technology Applications, Inc., Environmental Monitoring Systems Laboratory, Cincinnati, OH.
? U.S. EPA Environmental Monitoring Systems Laboratory, Cincinnati, OH.
. U.S. EPA Environmental Monitoring Systems Laboratory, Las Vegas, NV.
Oregon State University, Corvallis, OR.
-------
TABLE OF CONTENTS
Section
Page
List of Illustrations .... i .................... i viii
List of Tables .'. . . . xi
Executive Summary xiii
1.0 INTRODUCTION 1
1.1 Content and Organization of the Research Plan 1
1.2 An Overview of EMAP . 2
1.3 Legislative Mandate for Surface Water Protection 3
1.4 Current Efforts to Monitor Surface Waters 4
1.4.1 State 305(b) Water Quality Monitoring 4
1.4.2 National Contaminant Biomonitoring Program 5
1.4.3, National Stream Quality Accounting Network 5
1.4.4 Hydrologic Benchmark Network ......... . . !;. . . . .... 5
1.5 Unanswered Questions for Surface Waters ./. . . ... . . .... . .... . . . . 5
1.5.1 Societally Important Surface Water Values ...... ...:. !.... .... '.'": 6
1.5.2 Hazards to Aquatic Systems 8
1.6 Specific Objectives of EMAP-Surface Waters 8
1.7 Focus and Purpose of the Research Plan 8
2.0 APPROACH AND RATIONALE . . . ... .....:. .. . .... ......... .... .: 9
2.1 Overview of the EMAP-Surface Waters Approach 9
2.1.1 Design 13
2.1.2 Indicator Approach 17
2.2 Data Quality 20
2.2.1 The Data Quality Hierarchy 24
2.2.2 The Role of DQOs in EMAP 25
2.2.3 DQOs in the Surface Water Component • • • •, 25
2.3 Reporting ......... .'. . ... 28
2.4 EMAP:Surface Waters Program Limitations ^ ..... 28
3.0 MONITORING NETWORK DESIGN ........;... ... ...... . . . . . . . . .:.'. \ . . . . . . . 29
3.1 Populations, Sample Units, Frames, and Other Sampling Concepts . .;'.'': ........... 29
3.1.1 Surface Water Populations: Defining Lakes and Streams 29
3.1.2 Subpopulations ..".'......-.;. .......;...... ........... so
3.1.3 Sample Units 31
3.1.4 Frame Development 32
3.1.4.1 Lake Frame 32
3.1.4.2 Stream Frame 35
3.2 The Two-stage Sample: Tiers 1 and 2 36
3.2.1 Tier 1 Resources ......,...;......... 36
3.2.1.1 Lakes ^... .I .....,,.. 36
3.2.1.2 Streams . ; .......;..... 37
3.2.1.3 Tier 1 Association Rules . . ; ............... 41
3.2.1.4 Adequacy of the Grid for Tier 1 Sample Selection . . . . : . • . . ... . . . 41
3.2.2 Tier 2 Sampling .... .... . . . . 41
3.2.2.1 Tier 2 Site Selection Example . .".'... . ... ..', . . . 42
3.2.2.2 Mechanics of Tier 2 Site Selection (Design Option 2) 44
in
-------
3.3 Higher Grid Densities 45
3.4 Summary 50
4.0 INDICATOR DEVELOPMENT AND EVALUATION 55
4.1 Conceptual Framework for Indicators of Condition 55
4.1.1 Response Indicators 56
4.1.2 Exposure and Habitat Indicators 56
4.1.3 Stressor Indicators 59
4.2 Establishing Unimpaired (Nominal) Condition 59
4.3 Response Indicator Selection 60
4.3.1 Lake Indicators 61
4.3.2 Stream Indicators : 65
4.4 Exposure and Habitat Indicators 66
4.5 Stressor Indicators 70
4.6 Strategy for Indicator Development and Implementation 71
5.0 POPULATION ESTIMATION AND ANALYTICAL APPROACHES 77
5.1 Estimating Population Characteristics i ........ 77
5.1.1 Population Extrapolation from EMAP Sample Data 77
5.1.2 Tier 1 Resource Estimates . . 79
5.1.3 Tier 2 Estimates and Descriptions 80
5.1.4 Analytical Approaches for Detecting Population Differences,
Changes, and Trends ,; 81
5.2 Components of Variance in Regional Population Sampling 85
5.2.1 Variance Model 85
5.2.2 Indicator Variance Magnitudes Estimated from Existing Data . . 86
5.2.2.1 Alternatives for Comparing Variances 86
5.2.2.2 Coefficients of Variation 91
5.2.2.3 Relative Variances 91
5.3 Design Considerations for Estimating Population Status 92
5.3.1 Effects of Index Variability on Estimates of Population Status . 92
5.3.1.1 Theoretical Basis 92
5.3.1.2 Simulated Effects of Index Variability 95
5.3.2 Acceptable Levels of Index Variability for Status Estimates 97
5.3.3 Required Indicator Sampling Intensity for Status Descriptions 99
5.4 Examples of Effects of Indicator Variability on Population Descriptions and Trend
Detection ... 104
5.4.1 Eastern Lake Survey 104
5.4.2 Minnesota Pollution Control Agency 104
5.4.3 Vermont Department of Environmental Conservation 104
5.4.4 Ohio Environmental Protection Agency 104
5.4.5 Descriptions of Status 105
5.4.6 Differences and Changes in Regional Subpopulations 107
5.4.7 Regional Trends 107
6.0 FIELD SAMPLING DESIGN 113
6.1 Index Concept 114
6.1.1 General Design Objectives of Index Sampling 114
6.1.2 Applications of the Index Concept in Synoptic Assessments 114
6.1.2.1 National Stream Survey (NSS) 114
6.1.2.2 National Lake Survey (NLS) 115
6.2 Temporal Considerations for Field Sampling 116
IV
-------
6.3 Index Sampling Design for Lakes and Reservoirs
6.3.1 Temporal Variability ;'.'. 116
6.3.2 Summer Lake/Reservoir Index Period: The Proposal '.'.'.'.'.'.'.'.'.'.'.'.'. 116
6.3.3 Spatial Variability within Lakes and Reservoirs ........ • .'. 119
6.3.4 Spatial Sampling Index for Lakes and Reservoirs: The Proposal 120
6.4 Index Sampling Design for Streams and Rivers .'.[', 120
6.4.1 Temporal Variability 120
6.4.2 Summer Baseflow Stream/River Index Period: The Proposal ...'.'. '.'.'. 121
6.4.3 Spatial Variability within Streams/Rivers . ..' .... 122
6.4.3.1 Upstream-downstream Variations ; . 122
6.4.3.2 Other Stream Habitat Variations ! 123
6.4.4 Spatial Sampling Index for Streams and Rivers: The Proposal 123
6.5 Conclusions ......;... 124
7.0 LOGISTICS APPROACH 125
7.1 Logistics Implementation Components 125
7.2 Logistics Issues 127
7.2.1 Staffing '.'.'.'.'.'.','.'.'.'.'.'. 127
7.2.2 Access , 127
7.2.3 Data Confidentiality . . 128
7.3 Field Operations Scenario: Lakes 128
7.3.1 General Logistics Scenario ; 128
7.3.2 Daily Activities Scenario 129
7.4 Field Operations Scenario: Streams 132
7.5 Organizational Structure 132
8.0 QUALITY ASSURANCE PROGRAM .... ........... 135
8.1 Data Quality Requirements 136
8.1.1 Precision and Bias 135
8.1.2 Comparability 139
8.1.3 Completeness 139
8.1.4 Representativeness 139
8.1.5 Tolerable Background Levels . 139
8.1.6 Method Detection Limits for chemical Constituents 140
8.2 Organization and Staffing Requirements 140
8.3 Quality Assurance Documentation 140
8.4 Quality Control Guidelines 141
8.4.1 Biological Measurements 144
8.4.2 Chemical Measurements 144
8.4.3 Habitat Quality and Site Characterization Measurements 146
8.4.4 Archival of Samples and Specimens 147
8.5 Data Review, Verification, and Validation 147
8.6 Assessment of Data Quality 148
8.7 Quality Assurance Reporting 149
9.0 INFORMATION MANAGEMENT . . . . 151
9.1 Definition of Information Management . . 151
9.1.1 Overview of EMAP-Surface Waters Information Management 151
9.1.2 Objectives of Surface Waters Information Management .,. 151
9.2 Initial System Concept 152
9.2.1 User Requirements 152
9.2.1.1 Issues 152
-------
9.2.1.2 Data Categories ... 152
9.2.1.3 Identified User Groups , .... >, ,,.......;.. 152
9.2.1.4 Data User Requirements ;...•„., . .. . . . 152
9.2.2 System Management and Functional Requirements .......... . . „, . .... ,.> . 153
9.2.3 Data and Information Flow ...,....,..„...:.„,. 154
9.3 Organization of EMAP-Surface Waters Information Management Program . . „ 154
9.3.1 Surface Water Information Manager ;,-.•.....,-..., 15,4
9.3.2 Surface Waters Information Center (SWIG) 154
9.4 Operational Specifications 154
9.4.1 Project Management System 157
9.4.1.1 Communications 157
9.4.1.2 Sample Tracking Information 157
9.4.2 Data and Sample Collection 158
9.4.3 Processing and Storage of Indicator Data 158
9.4.4 Data Access and Transfer 158
9.4.5 Data Analysis and Reporting 158
9.4.6 Data Documentation, Access, and Archival 158
9.4.7 Geographic Information System (GIS) 159
9.4.8 Integration 159
9.4.9 Existing Information/Data 159
9.5 QA/QC for Information Management Activities 159
9.5.1 Field Data Collection 159
9.5.2 Data Transfer 160
9.5.3 Sample Tracking 160
9.5.4 Data Entry/Verification 160
9.5.5 Data Validation 160
9.5.6 Data Analysis 160
9.5.7 Archive and Backup 161
9.5.8 Configuration Management (CM) 161
9.6 Resources: Overview 161
9.7 Implementation Plan 161
10.0 COORDINATION 163
10.1 Within EMAP 163
10.1.1 Tier 3 Activities in Surface Waters 163
10.1.2 Across EMAP Elements 163
10.2 Within EPA (Office of Water and Regions) 164
10.3 Other Federal Agencies 165
10.3.1 U.S. Fish and Wildlife Service 165
10.3.2 USGS National Water Quality Assessment Program 165
10.4 State Agencies 166
10.5 Research Organizations 166
10.6 Conclusions 167
11.0 EXPECTED OUTPUTS AND TIMELINES 169
11.1 Reporting 169
11.1.1 Data Evaluation Reports 169
11.1.2 Internal Program Reports 169
11.2 Timelines 170
VI
-------
12.0 FISCAL YEAR 1991 FIELD AND ANALYSIS ACTIVITIES 171
12.1 General EMAP-Surface Waters .. 171
12.2 TIME Project . . . . 171
12.3 Data Analysis Activities . .. .., .....:..... 172
12.4 Research Needs 172
REFERENCES
173
vii
-------
LIST OF ILLUSTRATIONS
Figure
2-1
2-2
2-3
2-4
2-5
2-6
2-7
2-8
2-9
2-10
2-11
2-12
2-13
3-1
3-2
3-3
3-4
3-53
3-5b
3-5C
3-5d
3-6
Page
U.S. Environmental Protection Agency regions .......... ................... . . ; . . . 10
Aggregations of Omernik's ecoregions .............. . ..... ................ ". ... 11
Concept of a four-tiered approach in EMAP ....... ............................. 12
The baseline grid (not randomized) for North America containing about 12,600 points
in the conterminous United States ............. . ............................. 15
The landscape characterization hexagons are one-sixteenth of the total area and are
centered on the sampling points ................ ......... • > ..... • • • • ..... .... 16
i
Enhancement factors for increasing the base grid density .......................... 16
Spatially Interpenetrating samples on a 4-year rotating cycle ............ ............ 18
General approach for identifying indicators .................. ............ • ....... 21
Indicator approach for EMAP-Surface Waters showing candidate indicators and the
top-down approach to problem identification and diagnosis of probable cause
21
Cumulative frequency distribution for Index of Biotic Integrity in a region showing
how data might be displayed '. • 22
Long-term assessment objectives for EMAP-Surface Waters > 22
The three phases of the data quality objective process • 22
Quality objective hierarchy in the data quality objective process 24
Example of the lake frame • 33
Example of the stream frame, with nodes identifying stream segments that are the
stream sample units • •
34
Cumulative size distribution for lakes in the Northeast ..' 38
Map of the spatial distribution of lakes in the Northeast 43
Hexagon capture rates for lakes (1-250 ha) in the Northeast 46
Areas that group similar numbers of lakes for Tier 2 selection 47
The structure of the EMAP grid, illustrating the fourfold decomposition that will
be the basis of the interpenetrating sample design
48
Selected hexagons containing one or more lakes that will constitute the Tier 2 sample 49
Acid sensitive regions of interest in the Northeast : 51
viii
-------
3-7 Areas of the Northeast requiring threefold augmentation of the grid to supply enough
sample sites for the project 53
4-1 The conceptual process for proceeding from measurements to indicators to assessment
of condition 57
4-2 The process of proceeding from measurements on fish assemblages to indicators such as
Index of Biotic Integrity (IBI) or Index of Well Being (IWB) 58
4-3 The indicator development process, showing the objectives, methods, and evaluation
techniques used in each phase 74
5-1 Frequency distributions are plotted for any identifiable population 78
5-2 Example of estimated distribution plots with upper confidence bounds, generated both
as numbers (upper plot) and area (lower plot)' 82
5-3 An example of comparing population distributions using the chi-square test 84
5-4 Convolution of a standard normal distribution with normally distributed index
variability 94
5-5 Simulated effect of index measurement variance on survey results observed from
sampling 100 lakes from a hypothetical asymmetrically distributed population with a
true mean of 25 and an SDunit of 50 96
5-6 Absolute (a) and relative (b) bias in population percentiles for ratio of index
variance to among-sampiing unit variance 98
5-7 Nomogram for determining sample allocation between replications within sample unit
and adding new sample units in a region based on ratio of cost for replications and
new sites and the variability within the index period and among sample units 100
5-8 Cumulative distribution functions (CDFs) describe the status of regional populations
of lakes and streams, as this figure illustrates for total phosphorus in the Northeast,
based on results of the Eastern Lake Survey 105
5-9 Cumulative distribution functions (CDFs) of the Index of Biotic Integrity for Ohio
streams, and chlorophylls for Minnesota illustrate the status of two indicators having
relatively large index variation relative to population variation 106
5-10 Cumulative distribution functions (CDFs) of the Index of Biotic Integrity for Ohio
streams illustrating situations in which regional subpopulations are quite similar
(Region 3 versus Region 5) and situations in which clear differences among regional
subpopulations occur (Region 1 versus Region 5) 108
5-11 Cumulative distribution functions (CDFs) showing the similarity in the Secchi disk
transparency status of lakes in Vermont measured during two index periods of the same
year, July and August 1985 (upper panel); median values are identified mid-graph. The
lower panel shows another way of displaying data when sites are paired 109
5-12 Cumulative distribution functions (CDFs) using the Index of Biotic Integrity for Ohio
streams show year-to-year differences in the status of small streams 110
ix
-------
5-13 Cumulative distribution functions (CDFs) taken from the Vermont lakes data set on
trophic condition showing comparisons made on paired sites: (a) total phosphorus
during March, comparing 1978 and 1979, and (b) Secchi disk transparency during July,
comparing 1987 and 1988 ... .... . 111
5-14 Superimposed yearly cumulative distribution functions (CDFs) show the confusing
"spaghetti" pattern in multiple overlays of CDFs, based on nine years of data on
Secchi disk transparency in Vermont lakes (upper panel). Results obtained from
yearly CDFs can be more clearly presented to illustrate trends, or lack of trends,
by selecting yearly values representing different quartiles (lower panel) 112
6-1 Seasons of lowest stream flow in the conterminous United States ... > 117
7-1 Flowchart of potential EMAP-Surface Waters daily field activity 131
8-1 Hierarchical model of variance components of importance to EMAP-Surface Waters ...... 138
8-2 Stages in the data acquisition and management process for EMAP-Surface Waters where
data quality can be controlled or assessed ;........... 143
9-1 Data and Information flow for the EMAP-Surface Waters Information Center 155
9-2 Concept of the EMAP-Surface Waters Information Center in relation to the EMAP
Information Center 156
-------
LIST OF TABLES
Table "• Page
2-1 " Sample Fact Sheet and Logic Statement for EMAP-Surface Waters ... 26
3-1 Estimates of Stream Reach Length Based on Segments Represented on Three
1:100,000-scale Topographic Maps ... . ... .... . .;. . . . . . . . . . . . . . ........ ... .... 35
3-2 Summary of the Size Distribution of Lakes in the Northeast (EPA Regions 1 and 2) .". ..... 39
3-3 Summary of Average Capture Rate of Perennial Stream Segments, Lakes, and Reservoirs
(Sample Units per Hexagon) in 40-km2 Hexagons for the Humid (Eastern) and Arid
(Western) Regions of the United States 42
3-4 Estimates of Geographic Regions of Interest in the Northeast Relative to Acidic
Deposition and Number of Tier 1 Hexagons, Number of Lakes, and Number of Hexagons
with Lakes 52
3-5
4-1
4-2
4-3
4-4
Estimates of Tier 1 Coverage of Subpopulations of Interest for Acidic Deposition
in the Northeast 52
Candidate Indicators for Inland Surface Waters 59
Evaluation of Some Candidate Indicators for Inland Surface Waters by EMAP Selection
Criteria
62
Suggested Physical Habitat Measures for Lakes 67
Proposed Chemical Measurements Contributing to EMAP Geochemical Habitat Quality
Indices
69
4-5 Linkages Between Potential Endpoints, Measurements, Metrics, and Response
Indicators for EMAP-Surface Waters 72
4-6 Linkages Between Potential Endpoints, Measurements, and Exposure Indicators
for EMAP-Surface Waters 73
5-1 Tier 1 Inventory Estimates and Estimated Variance of Estimates are Provided from the
Tier 1 Sample for All Resources and Classes of Resources 80
5-2 Coefficients of Variation (CV) of Surface Water Indicators - A Number of Examples
from Lake and Reservoir Studies 87
5-3 Coefficients of Variation (CV) of Surface Water Indicators - A Number of Examples
from Stream and River Studies 88
5-4 Components of Indicator Variability as a Percentage of Among-Sample Unit Variance
(S2unit) - Examples from Lake and Reservoir Studies 89
5-5 Components of Indicator Variability as a Percentage of Among-Sample Unit Variance
(S2) - Examples from Stream and River Studies 90
5-6 Bias of Apparent F(x) as a Function of d2, Assessed for the Normal Distribution
Function with Homogeneous Normal Measurement Error 95
xi
-------
5-7 Rough Categories of Index Measurement Precision (Sum of Analytical, Temporal,
Spatial Components) Relative to Regional Variance for Possible EMAP Indicators 101
I _ _ .
5-8 Rough Costs of Index Measurement Replication Relative to Costs of Additional
Sites (Sample Lakes or Streams) 102
6-1 Time Scales and Sources of Variability in EMAP Indicator Characteristics - Some
Examples 113
7-1 EMAP Logistical Elements for Implementation of Surface Waters Monitoring Programs .... 125
7-2 Proposed General Field Measurement Variables for Lakes ....... . . ..... 128
8-1 Criteria for Selection of Appropriate Sampling and Analytical (or Measurement)
Methodology 137
8-2 Responsibilities of the EMAP QA Coordinator and Surface Waters QA Officer . . 141,
8-3 Quality Assurance Related Documentation of EMAP-Surface Waters 142
8-4 Quality Control Activities Associated with Biological Measurements 145
8-5 Quality Control Activities Associated with Chemical Measurements . . ,... 146
11-1 Proposed EMAP-Surface Waters Implementation Schedule . ; 170
xii
-------
EXECUTIVE SUMMARY
EMAP-SURFACE WATERS MONITORING RESEARCH STRATEGY - FISCAL YEAR 1991
THE ENVIRONMENTAL MONITORING AND
ASSESSMENT PROGRAM
The Environmental Monitoring and Assessment
Program (EMAP) was initiated in 1988 to provide
improved information on the current status of
ecological resources in the United States and on
long-term trends in their condition. Seven broad
resource categories have been defined within
EMAP: near-coastal waters, wetlands, inland
surface waters, the Great Lakes, forests, arid
lands, and agroecosystems. In addition, seven
coordination and integration functions have been
established to assist the groups working in the
seven resource areas and to ensure consistency:
1. Monitoring design
2. Development of indicators of ecological
condition
3. Landscape characterization
4. Quality assurance and quality control
5. Field sampling logistics
6. information management
7. Integration and assessment
This document describes the rationale, objectivesj
and primary elements of the EMAP-Surface
Waters program, which has been designed to
assess the condition of U.S. lakes and streams.
Separate research plans are being prepared for
the other EMAP resource groups, as well as for
each of the coordination and integration func-
tions. As an integrated multi-resource program,
the success of EMAP, and of EMAP-Surface
Waters, will depend on the close cooperation and
coordination among these various program com-
ponents.
OBJECTIVES OF EMAP-SURFACE WATERS
The overall goal of EMAP-Surface Waters is to
provide a quantitative assessment of the current
status of and long-term trends in surface water
condition on regional and national scales. The
specific, long-term objectives of EMAP-Surface
Waters are:
Estimate the current extent (location,
number, surface area or length) of lakes
and streams, on regional and national
scales, with known confidence.
Estimate the current status, changes, and
trends in indicators of the'conditidn of the
nation's lakes and streams, on regional and
national scales, with known confidence.
Monitor indicators of pollutant exposure
and habitat condition within lakes and
streams and seek associations between
human-induced stressors and ecological
condition that identify possible causes of
adverse effects.
Publish annual statistical summaries on the
extent and the status of indicators of eco-
logical condition of lakes and streams, arid
periodic interpretive reports on the status
and trends indicators of ecological condi-
tion of lakes and streams to the EPA
Administrator and the public.
OVERVIEW OF THE EMAP-SURFACE
WATERS APPROACH
This research plan presents the rationale, objec-
tives, approach, and plan for establishing the
monitoring program that will be required to
assess the status and trends in ecological con-
dition of U.S. surface waters. It also includes a
description of the EMAP component that will spe-
cifically address the response of surface waters to
acidic deposition [the TIME project (Temporally
Integrated Monitoring of Ecosystems)].
EMAP-Surface Waters is primarily a top-down
program being designed to evaluate the health or
condition of surface waters with respect to
endpoints of concern. Given multiple societal
concerns and values, EMAP-Surface Waters
proposes to evaluate the condition of lakes and
streams with respect to each of these issues:
XIII
-------
• Biological Integrity
• Trophic condition
• FIshability
The design and Indicator strategies described in
this report are Intended to be adequate for
exploring a variety of questions on national and
broad regional scales rather than to answer site-
specific questions. In this context, EMAP-Surface
Waters will provide information with which we can
determine (1) what proportions of our lakes,
streams, and rivers are degrading or improving,
where, and at what rate, (2) what the likely causes
are of adverse effects, and (3) whether or not
systems are responding as expected to control
and mitigation programs. Information derived
from the program will help in setting national
priorities.
COORDINATION WITH OTHER MONITORING
PROGRAMS
EMAP-Surface Waters is being designed and
funded by the U.S. Environmental Protection
Agency's (EPA) Office of Research and Develop-
ment (ORD). However, the states, other offices
and regions within EPA, and other federal agen-
cies [e.g., U.S. Fish and Wildlife Service (USFWS);
U.S. Geological Survey (USGS)] have contributed
to Its development and will participate in the
collection and use of EMAP data. This coordina-
tion will avoid duplicative monitoring efforts,
facilitate the exchange of data, and increase the
expertise available for refining the program design
and Interpreting the monitoring results. EMAP is
not Intended as a substitute for other monitoring
and research efforts, but instead will provide a
framework for Integrating existing and new data.
EMAP'S HIERARCHICAL DESIGN
Ultimately, EMAP will Involve four tiers, or types of
activities related to monitoring and assessing
ecological condition (Figure 1):
Tier 1: Landscape characterization, to
determine the distribution and extent
(numbers and area) of ecological
resources In the United States.
• Tier 2: Field sampling, to include a suite
of biological, chemical, and physical
measurement to determine the conditions
of a subset of Tier 1 sites.
• Tier 3: More intensive sampling to focus
on special subpopulations of concern or to
obtain more detailed diagnostic informa-
, tion.
• Tier 4: Ecological research,, to comple-
ment the monitoring data collected in Tiers
1,2, and 3.
This document addresses activities primarily at
Tiers 1 and 2. The Tier 3 activity, the TIME
project, is discussed briefly throughout this report,
as it pertains to modifications of specific appli-
cations of the general approach.
Tier 1 Sampling Frame
To achieve its objectives, EMAP-Surface Waters
will use standardized sampling methods and an
unbiased probability-based sampling design to
monitor lakes and streams over broad geographic
areas and for multiple decades. The proposed
design strategy is based on a permanent national
sampling framework consisting of a hexagonal
plate containing a triangular grid of approximately
12,600 points placed randomly over the conter-
minous United States (Figure 2); a similar array is
available for Alaska and Hawaii. A defined area
(40-km2 hexagon) around each point will be char-
acterized using existing maps, aerial photography,
satellite imagery, and existing databases of
landscape information. These 40-krrf hexagons
describe an area sample representing one-
sixteenth of the area of the United States and
provide the basis for the Tier 1 estimates of
surface water extent and distribution.
The characterization carried out on the 40-krn2
hexagons will provide an inventory of all aquatic
resources within them. Thus it will provide a one-
sixteenth sample of the extent, abundance, and
distribution of any aquatic resource that can be
identified during the landscape characterization.
The complete landscape description at Tier 1
ensures that all ecological resources are included
by definition of the monitoring coverage. •
xiv
-------
0)
en
£
0)
o
O
Tier 4
Detailed Diagnostics \ Tier 3
Subpopulation Interests^
Estimates of Condition
and Exposure
Status and Trends
Tier 2
Landscape Characterization
Estimates of Extent and Landuse
Tier 1
Figure 1. Concept of a four-tiered approach in EMAP. Spatial coverage is maximized in lower tiers;
temporal coverage increases at the higher tiers.
Tier 2 Sampling Frame
A subsample of the Tier 1 sample, selected by
probability methods and stratified as needed to
ensure an adequate sample size for each surface
water class and region of interest, will be visited
for detailed site characterization of indicators of
condition of the resource. Data obtained from
these field samples will constitute the information
to be used for reporting on regional status and
trends in ecological condition and exposure. This
second stage of the double sample will constitute
Tier 2 of the design.
The Tier 2 sample will be implemented for
resources of current concern. If a need to
monitor an additional class of resource is
identified, it will be necessary to return to the Tier
1 level, identify the extent and location of that
resource, and select appropriate sites for field
sampling. However, the existence of the
landscape characterization provides a ready-made
frame for sample selection, so that a field
program can be mounted with minimal prepara-
tion. This provides the potential for quick
response to new or emerging issues of ecological
condition.
The Tier 2 sites will be sampled on a four-year
cycle, that is, one-fourth of the sites in a region
will be visited each year. A small subset of sites
will be visited annually during the first decade of
the program to estimate year-to-year variation and
to identify any important trends early. By the fifth
year, all sites will have been sampled and a sec-
ond cycle will begin, using the same subsets of
resource units. The sample sites will be parti-
tioned so that the basic systematic triangular grid
(at one-fourth the density) will be retained in each
annual subsample, to maintain a nearly uniform
spatial distribution of sites each year. Analyses of
surface water status will be reported annually, as
four-year running averages over the four inter-
penetrating sample subsets. A few new sites will
be selected each year, and others will be
dropped, to minimize any bias that might be
xv
-------
Figure 2. The baseline grid (not randomized) for North America, containing about 12,600 points in
the conterminous United States. Spacing between points is about 27 kilometers.
xvi
-------
introduced by the knowledge that certain sites are
EMAP sites, resulting in differential treatment of
those sites.
Flexibility of the EMAP Design
An important characteristic of the EMAP design is
its flexibility. The results from Tiers 1 and 2 can
be summarized according to any subpopulation
(e.g., surface water class) or spatial partitioning
(e.g., region) of interest. The basic EMAP grid
can easily be enhanced for greater sampling den-
sity in regions or for lake or stream classes of
particular concern. The landscape characteriza-
tion at Tier 1 provides a ready-made frame for
sample selection, so that new or supplemental
field programs can be implemented quickly in
response to new or emerging issues. The outputs
from the EMAP-Surface Waters monitoring net-
work can be analyzed and expressed in a variety
of ways, to address a diversity of policy-relevant
questions. The EMAP design will provide informa-
tion on specific indicators measured during a
specific index period, as a "snapshot" of the
overall condition of a system.
INDICATOR APPROACH
In the EMAP-Surface Waters approach, ecological
risk assessment will focus on the ecosystem
attributes or endpoints of concern. Two types of
endpoints are used in risk assessment:
1. Assessment endpoints: Formal expres-
sions of the actual environmental values
that are to be protected. They should have
unambiguous operational definitions, have
social or biological relevance, and be
amenable to prediction or measurement
(e.g., probability of a >10% reduction in
game fish production).
2. Measurement endpoints: The quantitative
results of the measurements taken to char-
acterize assessment endpoints. Measure-
ment endpoints must correspond to or be
predictive of assessment endpoints (e.g.,
the fraction of streams that exceed the
LCoo for a toxicant to largemouth bass).
The term "indicator" has been adopted within
EMAP to refer to the specific environmental char-
acteristics to be measured or quantified through
field sampling, remote sensing, or compiling of
existing data. Four types of indicators are being
selected and developed for EMAP-Surface
Waters:
1. Response indicator: A characteristic of
the environment measured to provide evi-
dence of the biological condition of a
resource at the organism, population, com-
munity, or ecosystem level or organization
(e.g., trophic state index, fish assemblage).
2. Exposure indicator: A characteristic of
the environment measured to provide evi-
dence of the occurrence or magnitude of
contact with a physical, chemical, or bio-
logical stressor (e.g., nutrient concen-
trations, tissue residues, toxicity tests).
3. Habitat indicator: A physical, chemical, or
biological attribute measured to character-
ize the condition necessary to support an
organism, population, community, or eco-
system in the absence of pollutants (e.g.,
availability of snags, substrate of stream
bottom, vegetation type and extent).
4. Stressor indicator: A characteristic
measured to quantify a natural process, an
environmental hazard, or a management
action that causes changes in exposure
and habitat (e.g., land use).
Assessment endpoints by definition must relate to
the environmental values identified for each
resource category in EMAP. Therefore, the ideal
measurement endpoints for EMAP are response
indicators. EMAP will collect data on response,
exposure, and habitat indicators at its field
sampling sites. Data on land use and land cover
will be collected as part of the landscape
characterization activity within EMAP. Additional
stressor indicators will be assembled from other
sources.
Selection of assessment endpoints, and the sub-
sequent selection of indicators (Figure 3), requires
consideration of public values, policies, and cur-
rent threats to surface waters, and an under-
standing of environmental stresses and their
impact on the chemical and physical habitat, and
hence on the biota. The endpoints, and thus the
indicators representing them, must motivate
change in policy when poor conditions are found.
Figure 4 provides a general overview of EMAP-
Surface Waters indicators, relating the selected
xvii
-------
Biotic
Indicators
of Condition
Biological
Communities
(processes and
interactions altered
due to exposure to
modified chemical,
physical, and biological
habitats)
Endpoints -
of Concern
Drive Policy
Chemical Habitat
Physical Habitat
Biological Habitat
(alteration of these
habitats due to stresses)
Policy
Directed
Toward
Impacting
Anthropogenic
Stresses
Context of these
decisions may
be resource class
specific or
regionally specific
Figure 3. General approach for identifying indicators. It requires (1) consideration of the public
values or endpoints (which drive policy) and the potential stresses (at whBch policy is
directed), and (2) knowledge of the interactions between chemical, physical, and bio-
logical processes.
ENDPOINTS
IMPACTS
RESULTS
Trophic State
Rshnbillty
Biotic Integrity
ORGANISMS
Fish
Macro-
Invertebrates
Phytoplankton/
Periphyton
Sedimentary
Diatoms
Semloquatlc
Vertebrates
Eutrophlcatlon
Acidification
Contamination
Habitat alteration
Physical Habitat
Index
Water Quality
Toxicity Bloassays
Nutrient Loadings Landuse/Landcover
Contaminant
Loadings
Water quality
degradation
Physical habitat
deterioration
Decrease or
extirpation of
native species
Atmospheric deposition/
emissions
Chemical application
estimates
Flow/stage records
Stocking and harvesting
records
Direction of impact
Direction of diagnosis
Figure 4. Indicator approach for EMAP-Surface Waters showing candidate indicators and the top-
down approach to problem identification and diagnosis of probable cause.
xviii
-------
Table 1. Candidate Indicators for Inland
Surface Waters
Response
Trophic state index
Sedimentary diatom assemblage
Zooplankton assemblage
Macroinvertebrate assemblage
Fish assemblage
Gross external anomalies
Semiaquatic wildlife assemblage (birds)
Exposure/Habitat
Physical habitat quaiity
Chemical habitat quality
Sediment toxicity
Chemical contaminants in fish
Chemical contaminants in sediments
Biomarkers
Stressor
Land use and land cover
Human and livestock population density
Atmospheric emission and deposition
Use of chemicals
Pollutant loadings
Flow and channel modification
Lake level and shoreline modification
Introduced species
Stocking and harvesting records
endpoints, response indicators, exposure indi-
cators, and stressors. Table 1 lists indicators
being considered for use in EMAP-Surface
Waters.
ASSESSING SURFACE WATER CONDITION
Surface Water Populations
To answer questions about the condition of lakes
or streams, it is necessary to specify explicitly
what constitutes the group under consideration.
Target populations can be regionally defined,
such as lakes in the Northeast, or defined on the
basis of certain attributes, such as lakes larger
than 20 ha. One of the general goals of EMAP is
to include all ecosystems; therefore at the
broadest level, the target population for EMAP-
Surface Waters is all lakes and streams in the
United States, excluding the Great Lakes, for
which a separate EMAP component has been
developed.
Within EMAP, subpopulations are the classes of
resource types about which we wish to make
statements of condition and trends. Tier 1
resources are one kind of subpopulation; they
function as strata in the EMAP design. Other
kinds of subpopulations are groups of lakes or
streams that are not strata, but classes of interest.
They may be classes within a particular Tier 1
resource, or classes that include more than one
Tier 1 resource. Subpopulations may, but need
not be, hierarchical. There is great flexibility to
establish a variety of subpopulations on which to
focus the estimation of condition and trends.
Subpopulations serve two major purposes: (1) to
increase the precision about estimates of condi-
tion and trends and (2) to target groups on which
to report.
Establishing Unimpaired (Nominal) Condition
The assessment of ecosystem condition (or
health) requires both (1) the determination of
certain criteria considered indicative of a healthy
sustainable resource and (2) the absence of
known stressors and detectable symptoms of
ecosystem stress. The challenge for EMAP-
Surface Waters is to conduct such an assessment
using the types of information and measurements
that can be collected within the constraints of the
EMAP design.
EMAP has adopted the terms nominal and sub-
nominal to refer to healthy and unhealthy condi-
tions, respectively. Classification of a lake or
stream as nominal or subnominal will rely not on
any single indicator, but on the full set of
monitored response, exposure, habitat, and
stressor indicators. Specific approaches for
dealing with apparent inconsistencies in indicator
signals, or for formally combining indicators into
a joint index of surface water condition, will be
explored as part of the EMAP-Surface Waters
indicator development process.
The process of selecting and implementing the
four types of indicators cannot be separated from
the problem of how the data collected and the
resultant indicators will be used to make state-
ments about lake and stream conditions. The
indicators will show that some of the waterbodies
are impaired (subnominal, unhealthy). The task is
to establish the criteria that define impaired
xix
-------
systems, that Is, to determine that below a certain
Index or metric score, a lake or stream is con-
sidered to be in unacceptable condition, relative
to a particular endpolnt of concern. Some of the
approaches for establishing these criteria include
selecting and assessing reference sites, using
historical data and/or pristine sites, using
ecological models, and basing the criteria on the
empirical distribution of indicator values.
EMAP-Surface Waters will interpret ecological
condition based on a combination of these
approaches, also relying heavily on a series of
regional reference sites and knowledge of historic
conditlonsasdescribed through paleolimnological
reconstructions, expert opinions, and previous
biological data collections.
POPULATION ESTIMATION AND
ANALYTICAL APPROACHES
Indicator measurements taken in the field and in
the laboratory will be used to index the status and
characteristics of each sample waterbody. EMAP
sample lakes and streams will be chosen as a
probability sample from explicitly defined popula-
tions of surface waters, thus information from
these sample sites can be used to infer charac-
teristics of the population from which the sites
were drawn. Each sample waterbody will have a
weighting factor inversely proportional to its
sample inclusion probability. For example, if
there were a 1 In 10 chance of selecting a par-
ticular lake for field sampling, that lake would
represent 10 lakes in a statistical sense. These
sample site weightings will be used to extrapolate
from Index conditions In the set of sample sites to
Index conditions In the population statistically
represented by those sites.
Tier 1 Resource Estimates
There are several kinds of Tier 1 estimates,
depending on the data on which they are based
and the nature of the populations and the selec-
tion process. For example, resource inventories
are provided by the landscape descriptions and
other data on the 40-knf characterization hexa-
gons established on the grid points. These Tier 1
estimates will typically be of the areal extent of
surveyed resources, by whatever classes have
been Identified during the Tier 1 classification.
Estimates of total surface area of lakes could be
provided by size class and geographic region, as
could estimates of the numbers of lakes. Esti-
mating the numbers of any class of lakes or
streams is possible, since waterbodies can be
unambiguously represented by points.
Tier 2 Resource Estimates
The Tier 2 sample consists of a subset of surface
waters on which the suite of indicators will be
measured. Measurements of additional variables
may also be obtained from maps, remote
imagery, and other sources for use in higher
resolution classification of surface waters. Tier 2
estimates will be of the areal extent of classes
defined by the Tier 2 information, and of the
numbers of resource units in those populations.
The Tier 2 sample will lend itself to more complete
population descriptions in terms of the indicator
variables that have been measured on the
samples of population units.
Analytical Approaches for Detecting
Population Differences, Changes, and Trends
An important feature of EMAP-Surface Waters will
be the description of differences among sub-
populations and changes in subpopulations over
time. For example, it may be important to illus-
trate or describe differences among lakes influ-
enced by different land use patterns or practices,
or to determine whether the condition of different
types of lakes differs (e.g., drainage versus seep-
age lakes, or lakes versus reservoirs). During the
early years, it will be important to determine
whether there are detectable year-to-year differ-
ences within subpopulations of lakes, especially if
changes in the magnitude of stressors are occur-
ring or are anticipated. As more data are
obtained, it will become feasible to detect trends
in a variety of subpopulations.
An assortment of techniques will be used in the
examination of EMAP data for change and trend.
An extensive range of standard statistical tech-
niques is applicable. Linear model techniques,
such as analysis of variance and regression will
be used with various assumptions regarding spa-
tial and temporal variance components, statistical
independence, explanatory variables, nature of
trend, and nature of change. Nonparametric
alternatives, based on ranks or signs of differ-
ences, will also be used.
xx
-------
FIELD SAMPLING DESIGN
The basic design is to sample all Tier 2 sites once
in the late summer period. Water and sediment
samples from lakes and reservoirs will be
collected in a mid-lake location approximately at
the deepest point of the lake. Biological samples
will be collected from representative major habitat
types within lakes. Physical habitat indices will be
used to describe the lake and its riparian zone as
a whole. In streams and rivers, the sampling
location will be a reach, 30 channel-widths long,
located midway between the mapped upstream
and downstream confluences defining the stream
segment that is the sample unit. Water samples
will be taken at a single mid-channel location in
the middle of this reach, and biological samples
will be systematically sampled in representative
habitat types within the reach. Physical habitat
characteristics will be assessed over the entire 30-
channel-width section and its riparian zone.
INDEX CONCEPT
There are both temporal and spatial aspects of
index measurements. The optimal field sampling
design depends upon the objectives of monitor-
ing, as well as on the variability of the indicators
to be measured. The overall data quality objec-
tives of EMAP require that indicator measure-
ments be relatively stable, but responsive to year-
to-year changes in biotic stress and exposure.
In the temporal context, an ideal index period is
a time during which the values of response and
exposure indicators are relatively stable. It must
be a time period when indicator biota are present
and measurable. To facilitate diagnosis of prob-
able causes of subnominal indicator values, it is
advantageous for the index time period to be a
season in which indicator organisms are exposed
to maximum environmental stress.
In a spatial context, an index value should provide
in a single measurement, or an integration of a
number of measurements, information adequate
to represent each sampling unit in classification,
regional extrapolations, or trend analysis. If the
sampling unit is heterogeneous, the index meas-
urement should, through multiple sampling points,
sufficiently incorporate the spatial variability of
indicator measurements within the sampling unit
(lake or stream). Alternatively, the measurements
might be made at locations which themselves
integrate the ecological stresses impinging on the
sampling unit as a whole. For example, pelagic
fish and mid-lake water quality might reflect
benthic biogeochemical processes, chemical
inputs from streams, and littoral spawning,
rearing, and feeding areas.
VARIABILITY
Temporal Variability
Temporal variability in the surface water charac-
teristics to be measured by EMAP indicators
occurs over a range of time scales. Variability
over decades, years, and seasons, and within
seasons, may affect the precision and inter-
pretation of status and trends in the condition of
lakes and streams. Some sources of variability
are cyclical, and might be considered "noise";
others may be trends that EMAP should quantify.
Rivers and streams are generally more persistent
than lakes on a geological time scale and their
long-term characteristics are probably more
stable. However, flowing waters have much
smaller hydrologic retention times than lakes; so
the amount and character of water in their chan-
nels more closely reflects fluctuations in the
sources and sinks for water, energy, and chemical
substances. As a result, flowing waters often
exhibit greater seasonal, within-season, and diel
variation than do most lakes, for those character-
istics that depend on the amount and flowpath of
hydrologic inputs.
Spatial Variability
Spatial variability in lakes and reservoirs occurs in
both vertical and horizontal dimensions. Vertical
variability in temperate lakes occurs primarily
because of temperature stratification. Horizontal
variation in lake characteristics occurs because of
a wide variety of factors, including lake size,
morphometry, and substrate, and the location and
magnitude of inlets and outlets. For streams,
stream discharge, gradient, channel width, water
depth, substrate, and many other physical charac-
teristics change in a predictable fashion along the
course of a stream from its headwaters to its
mouth. Similarly, geology, soils, vegetation, and
the influence of groundwater on streamflows
change markedly from stream headwaters to the
mouths of rivers.
xxi
-------
The basic design, sampling surface waters once
In the late summer season, should be adequate
for meeting EMAP-Surface Waters goals in most
regions, If some systematic spatial replication
within sample sites is employed. However, the
precision of EMAP indicator measurements will be
evaluated through temporal replication on subsets
of 20 to 30 samples lakes and streams in each
region.
LOGISTICS APPROACH
Implementing the EMAP-Surface Waters program
will require comprehensive logistics planning. A
logistics plan must be developed before the start
of field activities to assure that program goals will
be met. Regional logistics plans will be updated
annually. The logistics plan will address 15
elements:
1. Overview of logistical activities
2. Staffing
3. Communications
4. Sampling schedule
5. Site access
6. Reconnaissance
7. Waste disposal plan
8. Safety plan
9. Procurement and inventory control
10. Training
11. Field operations
12. Laboratory operations
13. Data management activities
14. Quality assurance
15. Logistics review/recommendations
Field activities will be start in 1991 with a lake
pilot program In the Northeast. Additional regions
will be phased into the program in each of the fol-
lowing years, as funding permits. The TIME
project will also begin In the Northeast in FY91.
ORGANIZATIONAL STRUCTURE
Coordinating the logistics activities of 50 or more
surface water teams across the nation will be very
difficult. Regionalizing these logistics activities
Into various centers will be the most effective
mechanism for conducting EMAP field operations.
The EPA regional offices can provide the initial
structure for these logistics centers. The EPA
regions have firsthand knowledge of the environ-
mental conditions within their respective regions
and will have a major role within E:MAP, part of
which could be in logistics. The EPA regions
represent the Agency's primary contact with the
states, and can work with the states at the pro-
gram level. Securing cooperation from the states
for EMAP is essential because of requirements
regarding collection permits and access permis-
sion. The EPA regions and states also have
highly experienced field personnal, and their
participation in EMAP field activities would be
extremely beneficial. As key personnel directing
field team activities year after year, they will
provide the program with critical continuity.
The long-term success of EMAP depends on the
development of an interagency program with
common goals for the monitoring of the eco-
logical condition of the environment. Surface
water monitoring alone could involve numerous
agencies within the Department of the Interior, the
Department of Agriculture, and the Department of
Defense. As EMAP evolves into an interagency
program, agreements between agencies will have
to be established to define responsibilities.
QUALITY ASSURANCE
Quality assurance must be an integral part of any
program that intends to produce useful informa-
tion. Consistent with this precept, and with EPA's
policy of ensuring that all environmental data are
of known and documented quality, EMAP-Surface
Waters will include a comprehensive quality'
assurance/quality control (QA/QC) program/
Major elements of the program will Include:
1. Developing and documenting standard
operating procedures (e.g., methods
manuals).
2. Conducting staff training.
3. Maintaining suitable facilities and equip-
ment.
4. Using QC samples to validate analytical
data and the methods used to collect data.
5. Conducting external audits.
6. Performing extensive data verification and
validation checks on the database
management system.
xxii
-------
The development of data quality objectives will
provide the framework for balancing the tradeoffs
between the quality of data needed to make
sound decisions and EMAP constraints and costs.
INFORMATION MANAGEMENT AND
REPORTING
To be of maximum use, data must be transformed
into useful information as quickly as possible.
Therefore the goal of EMAP-Surface Waters is to
produce annual statistical summaries of the moni-
toring results for the preceding year within nine
months after data collection has been completed.
These reports will provide summaries of response,
exposure, and habitat indicators for the regions
sampled, but with minimal interpretation.
Interpretive reports will be published for the
Congress, interested scientists, and decision
makers after each four-year sampling cycle is
completed. Special scientific reports and peer-
reviewed papers also will be published period-
ically, to address particular topics of interest.
A computerized information management system
(IMS) will be established to ensure that EMAP-
Surface Waters data are made available in a
timely manner to other EMAP resource groups, to
other local, state, and federal organizations, and
to academic institutions. The system must be
capable of drawing from historical databases
already in place for similar environmental
monitoring efforts, and it must be compatible
across the IMSs for all EMAP resource groups.
RESEARCH NEEDS
The EMAP-Surface Waters program must be
accompanied by an aggressive research program.
The range of researchable issues is quite long.
Key issues can be divided into four categories:
1. Indicators of condition
2. Diagnostic indicators
3. Approaches for defining nominal and
subnominal
4. Statistical approaches
Extensive work is required to refine our selection
of indicators of condition. Existing indicators
need regional refinement and new indicators are
required to fill in gaps. Additional work needs to
be done to improve our diagnostic capabilities
with biological indicators.
Although the decision on what is an acceptable
condition for a water body is ultimately a societal
one, EMAP-Surface Waters is obligated to provide
creative approaches to addressing this issue. The
process of gathering scientists, policy analysts,
managers, and decision makers together to tackle
this problem will offer many challenges.
Many statistical challenges exist in addressing
national assessments such as those proposed.
The process of combining multiple indicators into
a single statement of condition will require an
extensive statistical effort. The task of doing trend
analysis on population distributions is not simple.
And finally, the effectiveness of regulatory activi-
ties must be assessed. Through the EMAP-
Surface Waters program, we hope to evaluate
changes and trends in response indicators and
then associate them with changes or trends in
exposure or stressor information. This area must
be addressed soon.
SUMMARY
EMAP is not intended to describe all components
of an ecosystem or resource type, explain how
systems function, address specific stressors to
lakes and streams (e.g., acidification; chemical
contamination), or replace compliance monitoring.
The program will, however, provide information
about the condition of specific indicators at a
specific index period as a "snapshot" of the
overall condition of a system. EMAP-Surface
Waters will provide an ongoing monitoring frame-
work within which the condition of lakes and
streams can be assessed on a broad scale, so
that the magnitude and geographical location of
various problems can be assessed and mitigation
and research priorities can be set more objec-
tively. It will be possible to add new variables or
make regional modifications so that new issues
can be more quickly evaluated.
The task at hand is to start a scientifically sound,
policy-relevant program that is solid enough to
supply useful information, yet flexible enough to-
incorporate changes as our understanding grows
and the program develops. The indicators
proposed thus far are a beginning. . .,.-'.•-
xxiii
-------
xxiv
-------
"I am always doing what I cannot yet do in order that I might learn how to do it."
Vincent van Gogh
1.0 INTRODUCTION
Vincent van Gogh was certainly not noted for his
ecological or statistical contributions. Yet he is
considered by many to have been a genius in his
field. His thoughts on improving his own skills
are insightful. This quotation from van Gogh cap-
tures an interesting perspective: often, the only
way to really learn how to do something is to
start. The situation is certainly far from that
extreme with respect to ecological monitoring, but
it is a useful perspective to consider.
As you read this research plan, you will find that
we cannot currently reach all the objectives we
have set. We are painfully aware of this. But we
can implement aspects of the program, providing
policy-relevant information and filling portions of
the existing information gaps. As we implement
these portions, we will learn and the program will
improve. By implementing aspects of the pro-
gram, we will more clearly define the key features
that are missing and that should be given high
priority on a research agenda. If the program
does not stimulate or is not accompanied by an
aggressive research agenda, its full potential will
not be realized.
Over the next few years, we will be building a
database from which refinements to the program
can grow. The task at hand is to start a
scientifically sound, policy-relevant program that
is solid enough to supply useful information, yet
flexible enough to incorporate changes as the
program and our understanding develop.
In this document, we describe our vision of what
is needed to evaluate the ecological condition of
our nation's surface waters. We ask your assis-
tance in evaluating whether the vision is sound,
the approach rational, and the strategy adequate
to provide for continual improvement in the
program.
1.1 CONTENT AND ORGANIZATION OF THE
RESEARCH PLAN
This plan is organized into 12 sections. Section
1 is a general introduction to the Environmental
Monitoring and Assessment Program (EMAP) and
the Surface Waters qomponent, and their focus
and objectives, followed by a summary of the
approach for meeting these objectives. Each
successive section provides additional levels of
detail about specific aspects of the approach
such as design, indicators, or information
management. Sections of the Research Plan are:
• Section 2, Approach and Rationale: a brief
overview of all aspects of EMAP-Surface
Waters (EMAP-SW),
• Section 3, Monitoring Network Design; a
detailed description of the sampling design,
and site selection,
• Section 4, Indicator Development Strategy:
a list of the selection of indicators, their
intended use, and the process of continual
improvement in achieving descriptions of
ecological condition for surface waters,
• Section 5, Estimation and Analysis: an
overview of the procedures and issues
related to data analysis for status, trends,
and associations between indicators.
• Section 6, Field Sampling Design: a dis-
cussion of index periods and issues per-
tinent to sampling the lakes and stream
reaches within this framework.
• Section 7, Logistics; an outline of the
issues related to conducting a field pro-
gram of this type and issues that require
resolution.
• Section 8, Quality Assurance: a summary
of the approach and procedures that will
be implemented to ensure that the quality
of the data collected is high enough to
meet program objectives.
• Section 9, Information Management: an
explanation of data management issues
and a description of the procedures for
ensuring that users will be provided with
collected data quickly and efficiently.
-------
• Section 10, Coordination: an
explanation of the approach for integrating
Information within EMAP-Surface Waters,
with other components of the program,
and with programs outside EMAP.
• Section 11, Expected Outputs and Time-
lines: a description of the reports and
summaries expected from EMAP-Surface
Waters and an overview of proposed time-
lines for implementing EMAP-Surface
Waters.
• Section 12, Fiscal Year 1991 Field and
Analysis Activities: an outline of the
specific objectives for the field activities
and data analysis activities planned for the
coming year.
1.2 AN OVERVIEW OF EMAP
The need to establish baseline environmental
conditions against which future changes can be
documented with confidence has grown more
acute with the increasing complexity, scale, and
social importance of environmental problems
such as global atmospheric change, acidic depo-
sition, sustainability, and the loss of biological
diversity. It is therefore critical for us to develop
and implement monitoring programs that can per-
mit quantitative, scientific assessments of the
complex effects of pollutants and other stressors
on ecosystems.
In 1988, the U.S. Environmental Protection
Agency's (EPA) Science Advisory Board recom-
mended implementing a program within the EPA
to monitor ecological status and trends and to
develop innovative methods for anticipating emer-
ging environmental problems before they reach
crisis proportions. More recently, the EPA has
been urged to establish a program to assess
whether the nation's efforts to protect the
environment are producing the expected results
in maintaining and Improving environmental qual-
ity. In response to the need for better assess-
ments of the condition of the nation's ecological
resources, the EPA's Office of Research and
Development began planning the Environmental
Monitoring and Assessment Program (EMAP).
EMAP Is a strategic approach designed to meet
the growing need to identify and quantify the
extent, magnitude, and location of degradation or
Improvement in environmental condition. When
fully implemented, EMAP will answer critical
questions for policy- and decision-makers and the
public, such as: What is the current extent of our
ecological resources? For example, how many
hectares are there of estuaries, lakes, forests,
deserts, wetlands, and grasslands, and how many
kilometers of streams? How are these resources
distributed geographically, and is tlhis changing?
What percentage of the resources appears to be
adversely affected by pollutants or other human-
induced environmental stress, and in which
regions are the problems most severe or wide-
spread? Which resources are degrading, where,
and at what rates? Are adversely affected eco-
systems improving in response to control and
mitigation programs?
In order to answer these questions, an integrated
monitoring network with the following objectives
will be implemented within EMAP:
• Estimate current status, extent, changes,
and trends in indicators of the condition of
the nation's ecological resources on a
regional basis with known confidence.
• Monitor indicators of pollutant exposure
and habitat condition, and seek associa-
tions between human-induced stresses and
ecological condition that identify possible
causes of adverse effects.
• Publish annual statistical summaries and
periodic interpretive reports on status and
trends to the EPA Administrator and the
public.
Assessments of whether the condition of the
nation's ecological resources is improving or
degrading require that data be collected over
large geographic and temporal scales. For this
reason, comparability of data among geographic
regions and over extended time periiads is critical.
Meeting this need simply by aggregating data
from many individual, local, and short-term
networks has proven ineffective. There is also a
need for extrapolating the findings from the sites
sampled to the populations of lake and stream
resources as a whole. EMAP networks therefore
will be statistically designed to allow extrapolation
to the entire population, to provide unbiased esti-
mates of status and trends in indicators of eco-
logical condition, with known uncertainty, and to
determine associations among indicators of con-
dition, exposure, and stress, all with quantifiable
-------
confidence limits. This will be accomplished to
focus on national and regional scales over
periods of years to decades. These character-
istics distinguish EMAP from other current
monitoring efforts.
EMAP is being designed around six primary
activities:
• Strategic evaluation, testing, and
development of indicators of ecological
condition, pollutant exposure, and habitat
condition, and protocols for collecting data
on these indicators.
• Design and evaluation of a comprehensive
and versatile integrated monitoring frame-
work.
• Nationwide characterization of the extent
and location of ecological resources.
• Demonstration studies and implementation
of integrated sampling designs.
• Development of data handling and quality
assurance, as well as spatial analysis and
statistical procedures, for efficient analysis
and reporting on status and trends data.
• Assessments of the probable causes of
environmental conditions and trends.
Seven ecological resource groups form the basic
structure of EMAP: agroecosystems, arid lands,
forests, Great Lakes, near-coastal systems, inland
surface waters, and wetlands. Of course, eco-
logical resources could be split in a variety of
ways, but our goal is to ensure that the groups
encompass all ecological resources. Six activities
cross the entire program and play a key
coordination role in its success: monitoring
systems design, development and evaluation of
indicators of ecological condition, quality
assurance, logistics, information management,
and integration and assessment. Two compo-
nents function both as ecological resources and
as cross-cutting coordination activities: air/
deposition and landscape characterization. This
document describes the development of the
Surface Waters component of EMAP.
1.3 LEGISLATIVE MANDATE FOR SURFACE
WATER PROTECTION
Although 75% of the earth's surface is covered by
water, only 0.02% of the volume is contained in
rivers and lakes (Ehrlich et al. 1977). In its
1990-91 report, the World Resources Institute
(1990) identified the quantity and quality of fresh
water, for human and other biological uses, as
constituting a growing concern worldwide. In the
conterminous United States, inland surface waters
comprise less than 2% of the area (Geraghty et al.
1973), a small proportion that belies their
importance for our society. For example:
• Inland surface waters provide 79% of the
water used daily in the United States
(Ehrlich et al. 1977).
• Water use in the 7 southwestern states
exceeds runoff 9 out of 10 years (Ehrlich et
.al. 1977).
• Swimming and fishing rank first and
second, respectively, among all outdoor
participatory sports (USDI 1989).
• 31.5 million people fished inland waters in
1988; they spent $329.8 million (USD!
1989).
Understanding the importance of surface waters
and expressing concern about their extensive
degradation, Congress in 1972 established the
protection of surface waters as a priority with the
passage of the Federal Water Pollution Control
Act (P.L. 92-500). The primary objective of P.L.
92-500 was to restore and maintain the physical,
chemical, and biological integrity of the nation's
waters [Section 101 (a)]. An interim goal was to
provide for the protection and propagation offish,
shellfish, and wildlife and for recreation in and on
the water [Section 101(a)(1)]. Section 105(d)(3)
requires the EPA to conduct, on a priority basis,
an accelerated effort to develop and apply
improved methods of measuring the effects of
pollutants on the chemical, physical, and bio-
logical integrity of water. Section 304(a)(1) states
that the EPA shall develop and publish criteria on
the effects of pollutants on biological community
diversity, productivity, stability, and eutrophication.
Section 305(b) mandates biennial reports that
assess the extent to which all waters provide for
the protection and propagation of a balanced
community of aquatic life.
-------
Since 1972, the Act has been further streng-
thened. The Water Quality Standards Regulation
(U.S. EPA 1983) requires that states designate
aquatic life uses consistent with the goals of the
Act, establish criteria for standards to protect
those uses, and develop an antidegradation policy
protective of waters exceeding the criteria. The
Water Quality Act of 1987 amends P.L. 92-500
and emphasizes ambient standards and assess-
ments as the driving forces behind further pollu-
tion abatement. Section 303(c)(2)(B) allows
states to adopt criteria based on biomonitoring (a
key EMAP component). Section 304(1)(A)
requires listing of waters not expected to attain
protection and propagation of balanced biological
communities. Section 304(m)(2)(g) requires the
EPA to study the effectiveness of applying best
available pollution controls for protecting
balanced communities. Section 314(a) requires
trophic classification of all publicly owned lakes
and an assessment of the status and trends of
water quality in those lakes. Finally, Section 319
mandates Identification of waters that cannot
protect balanced aquatic communities without
nonpolnt source pollution controls.
The Federal Water Pollution Control Act and
subsequent amendments in 1977,1981, and 1987
[collectively called the Clean Water Act (CWA)]
not only mandated the protection of surface
waters but established, by law, the requirement
for several monitoring activities. Monitoring for
permit limit compliance is established under
section 308. The CWA also provides for short-
and long-term monitoring under Title I-Research
and related Programs, section 106(e)(1). The
EPA is authorized under this section to fund
states to monitor, compile, and analyze data on
the quality of navigable waters, including
biological monitoring and classification according
to trophic status. These data are to be annually
updated and Included in the biennial reports
required under section 305(b) of the CWA.
The amendments to the Clean Air Act, passed in
the 1990 session of Congress, require that the
effects of acidic deposition on surface waters be
monitored, in order to assess surface water
response to decreases in emissions. Because
this mandate Is more specific than the general
EPA mandate to monitor the condition of surface
waters, and because the amendments require an
assessment in a relatively short period of time (by
January 1,2003), EMAP will include a component
designed specifically to address the question of
aquatic response to acidic deposition (the TIME
project).
Other legislation requires assessment ,of
environmental risk to aquatic communities as well.
Of particular importance are stressors with
potential regional impacts that fall under the
Federal Lands Policy and Management Act
(FLPMA), the National Environmental Policy Act
(NEPA), the Resource Conservation and Recovery
Act (RCRA), and the Federal Insecticide,
Fungicide and Rodenticide Act (FIFRA), In
addition, the Endangered Species; Act mandates
assessments and protection of rare and threat-
ened species and the National Forest Manage-
ment Act of 1976 requires conservation of animal
diversity.
1.4 CURRENT EFFORTS TO MONITOR
SURFACE WATERS
Given these legal mandates to protect and restore
surface water quality, what efforts currently exist
to document the conditions of surface waters? A
brief description follows of four current programs
that document some aspect of surface water con-
dition: monitoring by the States in support of
Section 305(b) requirements (conducted by States
and reported through the EPA), the Fish and Wild-
life Service National Contaminant Biomonitoring
Program (NCBP), the U.S. Geological Survey's
National Stream Quality Accounting Network
(NASQAN), and the U.S. Geological Survey's
Hydrologic Benchmark Network. This section is
not intended to be a complete description of
these programs or a complete listing of every
monitoring program in existence, but rather a
brief discussion of four national programs of
relevance to EMAP. A new stream monitoring
program, the U.S. Geological Suivey's National
Water Quality Assessment Program (NAWQA), is
described in Section 10 as a program with which
EMAP needs to interact closely.
1.4.1 State 305(b) Water Quality Monitoring
As described in Section 1.3, the States are
currently required to monitor the quality of surface
waters in support of section 305(b) of the Clean
Water Act and report these findings to the EPA.
For its part, the EPA attempts to establish
consistency among these efforts by preparing
guidelines for the States in reporting their
information. Use of consistent measures and
approach is promoted and based on the degree
-------
to which a waterbody is in compliance with the
State water quality standards established for that
waterbody. Water quality standards within the
States consist of the objective for water quality
expressed as "beneficial use," plus numeric and
narrative criteria designed to ensure maintenance
of the beneficial use. The degree to which bene-
ficial use is supported is ranked by four cate-
gories: fully supporting, fully supporting but
threatened, partially supporting, and not sup-
porting. The States have considerable discretion
in determining exactly what variables are meas-
ured, which methods are used, how many and
which waterbodies are assessed, and how decis-
ions about the degree of use support are made.
The biennial report submitted by the EPA to
Congress (U.S. EPA 1990d) uses the State reports
to summarize the total river miles or lake acres in
each of the four support categories, the major
causes of use impairment, the source of pollution
in those waters not fully supporting their uses,
and the number of waters affected by toxic
pollutants.
1.4.2 National Contaminant Biomonitoring
Program
The National Contaminant Biomonitoring Program
(NCBP) is maintained by the U.S. Fish and
Wildlife Service to document temporal and
geographic trends in concentrations of environ-
mental contaminants that may threaten fish and
wildlife resources. Originated as part of the
multiagency National Pesticide Monitoring Pro-
gram, the NCBP has been in operation since
1967. The fishery component of the program
analyzes tissue from samples collected at over
112 stations at key points in major rivers
throughout the United States and in the Great
Lakes. Whole fish samples of 2 bottom-feeding
species and 1 representative predatory species
are analyzed for 22 organic chemical compounds
and 7 metals (Schmitt et al. 1985, Lowe et al.
1985). The river stations were selected in
conjunction with the USGS NASQAN stations.
Collection of samples from all stations takes place
during the fall of two successive years.
1.4.3 National Stream Quality Accounting
Network
In 1971, the U.S. Geological Survey began the
National Stream Quality Accounting Network
(NASQAN) with the objectives of (1) accounting
for the quantity and quality of water moving within
and from the United States, (2) depicting area
variability, (3) detecting changes in stream quality,
and (4) laying the groundwork for future assess-
ments of changes in stream quality. In 1986,
these objectives were modified: (1) assess water
quality trends nationally, ,and to the extent
possible, relate those trends that are detected to
upstream land and water use, (2) account for the
mass transport of selected constituents from the
continent, (3) account for the transport of
selected constituents to estuaries and/or other
significant water bodies (e.g., the Great Lakes)
that may be considered critical or otherwise of
national significance, and (4) account for the
quality of water crossing international boundaries
(in a small number of cases). Of the 412 sites on
major U.S., rivers, 58% are monitored bimonthly
and 42 quarterly. Variables measured are pre-
dominantly major anions and cations, selected
metals, suspended sediments, dissolved oxygen,
and fecal coliform and streptococcal bacteria.
Periphyton, phytoplankton, organic carbon, and
pesticide measurements were all discontinued
between 1979 and 1981; Several descriptions of
trends in water quality variables have resulted
from this program (Smith et al. 1987).
1.4.4 Hydrologic Benchmark Network
This network, established in 1964, is intended to
document natural changes in hydrologic charac-
teristics, provide better understanding of the
hydrologic structure of natural basins, and provide
a comparative basis for studying the effects of
humanity on the hydrologic environment. The 58
stations located across the country are minimally
impacted by humans. More than 50 physical and
chemical measurements are made at fixed sam-
pling intervals. Samples are collected quarterly at
73% of the stations, bimonthly at 23% ,of the
stations, and monthly at 4% of the stations.
1.5 UNANSWERED QUESTIONS
SURFACE WATERS
FOR
The mandate to protect and restore the condition
of surface waters has been outlined in Sections
1.3 and 1.4, along with current efforts to monitor
this progress. What then do we know about the
condition of our lakes, streams, and rivers from a
national perspective? In evaluating the NASQAN
data for 1974-1981, Smith et al. (1987) determined
that widespread decreases in fecal bacteria and
lead concentrations had occurred, but that con-
centrations of nitrate, chloride, arsenic, and
-------
cadmium had increased over broad regions.
Their general conclusions were that point source
controls had been effective, although nonpoint
source contaminants were on the rise, and that
more detailed work was necessary to evaluate the
policy Implications of the apparent changes.
Results from the NCBP (Schmitt et al. 1985, Lowe
et at. 1985) show no significant trends nationally
In the geometric means of metal concentrations in
fish tissue. However, small but significant
decreases in DDT were observed, PCB residues
remained widespread but showed declines, con-
centrations of dieldrin did not change, and
chlordane concentrations were slightly lower and
less widespread. The impact of the reported
changes on aquatic biota is not well documented.
The EPA, In summarizing the 1986 State 305(b)
reports, estimated that designated beneficial uses
were supported in 74% of the river miles and 73%
of the lake acres assessed (U.S. EPA 1987C). In
summarizing the 1988 reports (U.S. EPA 1990e),
the EPA estimated that 70% of the river miles and
74% of the lake acres assessed were reported to
support designated uses. One might legitimately
argue that a difference between 74% and 70% is
unimportant. The issue, however, is that not only
do we not know if the differences are actual
and/or meaningful, but we currently have no
mechanism for determining this.
In either case, these numbers present a not too
gloomy picture of the condition of lakes and
streams. They stand in stark contrast, however,
to other available information:
• 20% of the native fishes of the western
United States are reported to have become
extinct or endangered (Miller 1989).
• 32% of the Colorado River basin native
fishes are reportedly extinct, endangered,
or threatened (Carlson and Mirth 1989).
• Since 1910, annual Columbia River salmon
and steelhead runs reportedly have
declined by 75-85%, or 7-14 million fish
(Ebel et al. 1989).
» The Missouri River commercial fish harvest
is reported to have been reduced by over
80% since 1945 (Hesse et al. 1989).
« Since 1850, 67% of the fish species from
the Illinois River and 44% from the Maumee
River have declined or disappeared; 70% to
84% of these species were lost from small
or medium-sized streams (Karr et al. 1985).
• Since 1933, the, Tennessee River system
has lost 20% of its mollusk Species
(Williams et al. 1989) and 46% of the
remaining species are endangered or seri-
ously depleted throughout their ranges
. (Jenkinson 1981).
• 38 states reported fish consumption clo-
sures, restrictions, or advisories in 1985
(Moyer, 1986). ,
We possess a great deal of information about the
water quality of our aquatic resources, but we
have little consistent understanding of overall eco-
logical conditions or changes in conditions. Our
understanding of the reliability or level of uncer-
tainty surrounding the information is minimal.
There is some sense that many of the most
severe problems have been relieved, but we may
in fact have lost ground to the pressures of
increased human impact and less obvious non-
point sources of pollution or habitat alteration.
Apparently, we lack a quantitative framework from
which we can produce reliable estimates of the
condition of lakes and streams in the United
States, especially in the areas of nonpoint sources
of pollution and habitat alteration. This concern
is also reflected in the EPA's recent Science
Advisory Board report (SAB 19SO), which lists
physical habitat alteration and biological diversity
arnong the major ecological risks requiring the
attention of the EPA. The,National Research
Council (1990) reached similar conclusions in
their recent review of the USGS NAWQA Program
(see Section 10.0): we genuinely need a long-
term national assessment of quality on a broad
spatial scale in the United States. Brown and
Roughgarden (1990) also reflected this concern in
their call for a National Institute of Ecology with a
U.S. Ecological Survey as a prime component.
1.5.1 Societally Important Surface Water
Values
To be effective, the information from a national
monitoring program must prompt action when
required. This means that the information
produced must be related to perceptions of
aquatic health and represent issues and values of
concern and importance to the public, aquatic
scientists, and decision makers. In ecological risk
-------
assessment, these perceptions are called end-
points. Suter (1990) and Hunsaker et'al. (1990)
defined ecological endpoints as the environmental
entity of concern and the descriptor or quality of
the entity. Such endpoints represent concepts
that are societally important but that tend to be
nebulous or abstract and do not lend themselves
to direct measurement.
Historically, for surface waters, these endpoints
have been expressed as designated uses, which
have included habitat for aquatic life, fishing,
swimming, navigation, and drinking water supply.
Some of these uses depend directly on ecological
condition (e.g., habitat for aquatic life, fishing).
Others are less directly dependent on ecological
condition (e.g., drinking water, navigation),
although attainment may be indirectly associated
with the consequences of degraded ecological
condition. Also, as currently expressed in the
Clean Water Act, physical, chemical, and bio-
logical integrity embody a societal concern and
desire that surface waters be unimpaired and
healthy. In particular, increasing attention is
drawn to the biological condition of ecosys-
tems~or biological integrity-not explicitly defined
by legislation but defined by Karr and Dudley
(1981) as "a balanced, integrated, adaptive com-
munity of organisms having a species composi-
tion, diversity, and functional organization
comparable to that of natural habitats in the
region." This is similar to one current concept of
biological diversity defined by Noss (1990) as the
diversity of life and its processes which has
functional, structural, and compositional attributes
at the genetic, species, community, and eco-
region levels of organization. This concept is
much broader than simply a rare and endangered
species issue.
In the early stages of EMAP-Surface Waters devel-
opment, our plan was to address the ecological
condition of lakes and streams as a single end-
point of concern: biological integrity. After
extensive consideration and discussion with
others, it became apparent that not all of society's
ecological values related to surface waters fall
neatly under the biological integrity umbrella. In
fact, many aquatic systems are managed in a
manner supported by the public as an improve-
ment in ecological condition, but which reduces
biotic integrity. One of these practices is the
management of lakes and streams to increase
game fish, usually by replacing naturally occurring
species and stocks with exotic species or hatch-
ery stocks. Lakes are also fertilized to increase
fish production. Thus, attaining the societal value
of fishability often conflicts with the other societal
values of biotic integrity and clear water.
There is also considerable public interest in the
trophic condition (often expressed as clarity or
productivity) of surface waters, and in particular,
lakes, to the extent that assessing the trophic
condition of lakes appears as part of the Clean
Water Act, Section 314. The Clean Lakes Pro-
gram evolved to support restoration of lakes
undergoing culturaleutrophication, and the North
American Lake Management Society (NALMS)
formed to deal primarily with management of
trophic condition. The desire for lakes With clear
water is often in direct conflict with the desire for
lakes that support a productive fishery or a
diverse fish and wildlife fauna.
Given multiple societal concerns ,or values,
EMAP-Surface Waters proposes to evaluate the
condition of lakes and streams with respect to
each of these issues: . ;
• Biological integrity <: • ' ;
• Trophic condition x
• Fishability
There are a few issues to be considered relating
to these three endpoints of concern. First, as
already mentioned, it is likely that for individual
waterbodies our assessments of these endpoints
will conflict (e.g., a lake may have an excellent
fishery but be in undesirable biotic and trophic
condition). In this case, it is important to
remember that EMAP will not focus on individual
waterbodies but will make its assessments at
regional and .national scales. However, the
regional proportion of lakes and streams with
conflicting assessments will be of interest. We
must also develop regional expectations for biotic
integrity, trophic state, and fishability, because
these vary among regions for natural reasons,
such as ecoregional and zoogeographic differ-
ences. Second, some lakes (and streams) are
naturally fishless or eutrophic. This means EMAP-
Surface Waters will need (1) to make this deter-
mination for each waterbody, (2) to evaluate these
waterbodies as separate subpopulations (see
Section 3), and (3) to estimate the regional pro-
portion of waterbodies naturally occurring in this
condition. Third, fishability is a value-laden
endpoint, which complicates its assessment (what
is good fishing for one is not necessarily good
-------
fishing for all). In any case, our primary focus will
be on biottc integrity.
1.5.2 Hazards to Aquatic Systems
We have classified current hazards to inland
surface waters into nine somewhat overlapping
categories. Some of these reflect the effects of
land use or pollutants, others are related directly
to resource management:
1. Eutrophication (anthropogenically induced
nutrient enrichment)
2. Chemical Contamination (toxics)
3. Atmospheric pollutants (e.g., acidic
deposition)
4. Habitat alteration (physical structure,
substrate, riparian vegetation)
5. Row modification
6. Thermal alteration
7. Introduced species
8. Harvest imbalance (overstocking or
overharvesting)
9. Global climate change
One objective of the program is to relate any
degraded condition of aquatic resources to these
various hazards or combination of hazards, and
any improved condition of aquatic resources to
reduction In these hazards. However, diagnosing
problems to this level of resolution is a very long-
term objective.
1.6 SPECIFIC OBJECTIVES OF EMAP-
SURFACE WATERS
Objectives for EMAP-Surface Waters parallel the
program objectives. More specifically they are to:
• Estimate the current extent (location,
number, surface area or length) of lakes
and streams, on regional and national
scales, with known confidence.
• Estimate the current status, changes, and
trends in indicators of the condition of the
nation's lakes and streams, on regional and
national scales, with known confidence.
• Monitor Indicators of pollutant exposure
and habitat condition within lakes and
streams and seek associations between
human-Induced stressors and ecological
condition that identify possible causes of
adverse effects.
• Publish annual statistical summaries on the
extent and the status .of indicators of
ecological condition of lakes and streams,
and periodic interpretive reports on the
status and trends indicators of ecological
condition of lakes and streams to the EPA
Administrator and the public.
1.7 FOCUS AND PURPOSE OF THE
RESEARCH PLAN
This Research Plan presents the rationale, objec-
tives, approach, and plan for establishing a
monitoring program designed to assess the status
and trends in ecological condition of the nation's
lakes and streams. It also includes descriptions,
where appropriate, of the EMAP component (the
TIME project) that will specifically address the
response of surface waters to acidic deposition.
The current document is not intended to be a final
plan for establishing these monitoring programs
but rather part of a continually improving process.
It is our immediate objective to inform potential
EMAP data users of the approach proposed for
describing and monitoring the condition of lakes
and streams, and to elicit input and advice from
the scientific, management, and regulatory com-
munities that will improve the program. We
recognize from the outset that no one agency or
group has the expertise to develop a program of
this magnitude independently. We fully acknow-
ledge that we currently lack many of the tools
needed to reach the full potential of EMAP.
However, we firmly believe that we can begin the
process and by so beginning we can make
strides toward improvement. The proposed
design, we believe, provides sufficient flexibility to
allow modifications as we learn from pilot and
demonstration programs. The indicators we have
proposed are a beginning. They are not com-
plete, but they will provide useful information on
the condition of aquatic resources. An aggressive
research program for developing improved
indicators, as well as other tools, is urgently
needed for monitoring in general, and we hope
that EMAP can provide some of the stimulus for
the development of such a program. In several
sections of this research plan, we present the
major issues and problems associated with that
topic and current options for addressing the
issue. Early emphasis has been on developing a
lakes program, with a stream and river program
to follow. In summary, we welcome all advice
and input for this program.
-------
2.0 APPROACH AND RATIONALE
Section 2 is an overview of the EMAP approach
and highlights those aspects that make it unique.
This section also includes an overview of the
technical approach for meeting the goals and
objectives previously outlined for EMAP-Surface
Waters. The limitations of EMAP are also
presented. What is needed to accomplish the
EMAP-Surface Water objectives? What is the
fundamental concept of EMAP? These subjects
are introduced in the overview and explained in
detail in the remaining sections of the plan.
2.1 OVERVIEW OF THE EMAP-SURFACE
WATERS APPROACH
What sets the EMAP-Surface Waters approach
apart from existing monitoring efforts? The
fundamental distinctions occur in the areas of
design and indicator strategy. This is primarily a
top-down program being designed to evaluate
health or condition of surface waters with respect
to endpoints of concern and then to statistically
extrapolate those findings to the ecological
resource populations of interest. Our discussion
of this difference is not intended to criticize
existing programs but rather to acknowledge that
they were designed for different objectives and as
such cannot effectively be used to meet the broad
regional and national scale objectives of EMAP.
The strategy for EMAP employs the following
features that allow estimation, with known
confidence, of indicators of the ecological
condition of regional surface water populations:
• Explicit definition of surface water target
populations and their sampling units, and
identification of an explicit frame for listing
or otherwise identifying all the potential
sampling units within each target
population (Section 3).
• Probability sample site selection from the
population frame. A uniform grid and clus-
tered sampling will be used to obtain a
randomized, systematic sample of surface
waters with a geographical distribution
reflecting that of the population (Section 3).
• Representation of ecological conditions in
sample lakes and streams using ecological
indicators within an index concept (Section
4.
• Uniform sampling and analytical
methods for a suite of response,
exposure, and stressor indicator
measurements (Sections 6 and 8).
• A documented program of rigorous quality
control, quality assurance, and quality
assessment (Section 8).
These features, coupled with a sound under-
standing of the ecological function of lakes and
streams, comprise an effective program capable
of providing increasingly valuable information
about the condition of, and our protection for,
ecological resources.
EMAP-Surface Waters will estimate the condition
of lakes, reservoirs, streams, and rivers on a
national scale as well as on relatively broad
regional scales. Data from the program will allow
estimation of the spatial extent and geographic
distribution of various classes of surface waters.
Additionally, it will estimate the current status of
and future changes or trends in indicators of
ecological condition. The design and indicator
strategies described in this section are intended
to be adequate for exploring a variety of ques-
tions on national and broad regional scales rather
than to answer specific questions that might be
asked. In this context, EMAP-Surface Waters will
provide information with which we can determine
(1) what proportions of our lakes, streams, and
rivers are degrading or improving, where, and at
what rate, (2) what the likely causes are of
adverse effects, and (3) whether or not systems
are responding as expected to control and miti-
gation programs. Information derived from the
program will help in setting national priorities. For
example, in order to set research and mitigation
priorities for streams and rivers on a national
scale, we must be able to determine the relative
importance or magnitude of toxic contaminants
compared with habitat alteration problems as they
affect stream condition. Given this focus, the size
of the United States, and the number of types of
ecological resources that should be addressed,
EMAP-Surface Waters must confine its general
activities to regions no smaller than the EPA
regions or the aggregate ecoregions of Omernik
(Figures 2-1 and 2-2) until cooperative efforts with
other federal agencies and States permit a
sampling density with greater spatial resolution.
-------
U)
I
$
G
I
I
1
(U
E
UJ
w
u:
10
-------
0 -O
fi g
i£
O.T5
it
II
II
= i
.0 O
* O)
n C
I*
o n
'5i CO
S-2
ID n
£ g
£ .2
w ~
Q
(A
.2 o)
o> ,2
111
|S»
E = 2
811
ll|
< O
! '£
o
D1
n
CVJ
£
I
11
-------
A comprehension of the four-tiered approach to
meeting EMAP-Surface Water objectives is essen-
tial for reading this document and understanding
EMAP as a whole. Figure 2-3 illustrates the basic
levels of the four-tiered approach. A tier is a level
and type of activity related to monitoring and
assessing ecological condition. Most of the
discussion presented in this document focuses on
Tiers 1 and 2, but two additional tiers are
important for long-term planning.
Tier 1 of the four-tiered approach is the broadest
level and the one in which estimates of the extent
of a resource and its geographical distribution are
generated. It involves the characterization of
existing resources and their pattern of use on the
landscape. For EMAP-Surface Waters, this means
that during Tier 1 we will generate the number
and surface area of lakes, as well as their geo-
graphical distribution. For streams, we will
develop estimates of the number of stream
reaches and total miles of stream. The initial
source of this information will be the USGS
1:100,000-scale Digital Line Graphs (DLGs). The
estimates will be periodically updated using new
remote imagery (satellite imagery and aerial
photography). In some instances, Tier 1 activities
may also provide information about the condition
of these resources, if the data can be derived
from various remote sensors.
Tier 2 activities involve field sampling to determine
the conditions of a subset of Tier 1 sites. A suite
of biological, chemical, and physical measure-
ments will be obtained from each of these sites
and the information aggregated to make state-
ments about the regional condition of that
resource type rather than statements about the
specific condition of any particular site.
Tier 1 and 2 activities will allow the successful
completion of the primary objectives of EMAP and
will be the priority in early funding. However,
other information needs will require an additional,
but complementary, level of effort: at Tiers 3 and
4. The need for two types of information will drive
o
0)
S.
1
o
"5
CO
D)
C
'w
s
Tier 4
Detailed Diagnostics y Tier 3
Subpopulatlon Interests
Estimates of Condition
and Exposure
Status and Trends
Landscape Characterization
Estimates of Extent and Landuse
Tier 2
Tier 1
Figure 2-3. Concept of a four-tiered approach in EMAP. Spatial coverage is maximized! in lower tiers;
temporal coverage increases at the higher tiers.
12
-------
Tier 3 activities within EMAP: more explicit
regional diagnostic information, and information
about status, trends, and diagnostics for more
specific subpopulations of interest, such as playa
lakes or saline lakes.
During Tier 1 and 2 activities, if a resource
exhibits poor conditions throughout a region, but
the likely cause cannot be determined from the
available data, then more detailed work in that
area could be implemented to identify the source
of the problem. The problems resulting from
selenium toxicity, most notably at Kesterson
Wildlife Refuge but also widely observed
throughout the West, provide an example of the
type of situation that might warrant movement to
Tier 3. If the EMAP-Surface Waters approach
were applied to this example, Tier 2 activities
might identify widespread deaths and deformities
in waterfowl, but with unknown cause. The
Kesterson Wildlife Refuge would be the focus of
concern, but given the apparent extent of the
problem throughout the western United States, a
more detailed study of regions might be under-
taken and the link to selenium toxicity identified.
This kind of study could be done under a Tier 3
concept within EMAP. More extensive spatial
sampling and perhaps intensive temporal samp-
ling might be required. A probability sampling
approach might still form the basis of the
monitoring effort.
A second use of Tier 3 might be to evaluate con-
cerns that exist nationwide but are restricted to a
specific subpopulation not adequately covered
using the standard EMAP grid density. It is
important, before embarking on this type of Tier
3 activity, to have an understanding of the size of
this subpopulation in relation to the resource as a
whole. This activity should also be accompanied
by funding above the base program so that it
does not impact the Tier 1 and 2 efforts. The
current acidic deposition scenario is an excellent
example of this situation. The initial questions of
status and extent (Tiers 1 and 2) for aquatic
effects of acidic deposition were addressed during
the National Surface Water Survey (NSWS), con-
ducted as part of the National Acid Precipitation
Assessment Program (NAPAP). Changes in emis-
sions have now been proposed under the renewal
of the Clean Air Act. There is a need for a
program that evaluates how effectively these
emission reductions improve conditions in
affected aquatic systems. High-interest areas,
such as the Adirondacks, Southern Blue Ridge,
and the Sierras, are too small for the Tier 1 EMAP
grid density to evaluate adequately; interpreting
any regional changes in aquatic conditions in
these areas will require a few well-defined sites
that receive more temporally intensive monitoring
than that provided by the annual regional visits of
Tiers 1 and 2.
Tier 4 is a research component that complements
Tier 1,2, and 3 activities. It is important to think
of Tier 4 as an activity rather than a specific
place. It is the activity of ecological research that
is critical, rather than a particular place. Initially,
EMAP will not establish new research sites, but
will rely on existing programs to play this role.
The National Science Foundation's Long Term
Ecological Research (LTER) Program maintains a
network of 17 sites across the United States. The
Department of Energy has a Research Park net-
work. Additional sites are maintained by other
federal agencies and university researchers.
These are all sites of active research on the
functional and structural aspects of ecosystems.
These networks provide (1) important information
on system characteristics, which will be useful in
developing and testing indicators of ecological
condition, (2) detailed process information char-
acterizing aspects of systems not amenable to
general monitoring, and (3) a link between spati-
ally extensive and temporally intensive data, to
increase the interpretation of EMAP results. Tier
4 should generate input that modifies activities at
Tiers 1, 2, 3 and should in turn be influenced to
evaluate hypotheses generated from results of
Tier 1, 2, and 3 activities. This interaction
between Tier 4 and Tier 1, 2, and 3 activities
should ensure that activities of the first three tiers
continue to be grounded in sound ecological prin-
ciples. Although EMAP will not initially establish
its own Tier 4 sites, we propose that some of the
existing research sites be included in the Tier 2
sample. As EMAP proceeds, gaps in the existing
Tier 4 networks may be identified and filled.
This document addresses activities primarily at
Tiers 1 and 2. The Tier 3 activity, the TIME
Project, is discussed briefly throughout this
document, as it pertains to modifications or
specific applications of the general approach, and
more completely in Stoddard (1990).
2.1.1 Design
In trying to describe resources such as lakes or
stream reaches for the nation, we could either
13
-------
census all the systems or sample a subset and
extrapolate from that subset to the entire popu-
lation of interest. Given the number of lakes and
stream reaches, the census approach is imprac-
tical. Thus an approach for selecting and samp-
ling a subset is needed. We have chosen to
select this subset on a "probability1 basis; that is,
our subset Is selected in such a way that the
probability of Including any particular lake or
stream reach can be calculated. Information from
the sampled systems can be weighted by the
Inverse of the Inclusion probability to calculate
unbiased regional estimates.
Meeting the EMAP-Surface Water objectives out-
lined in Section 1 requires developing a design
that is capable of:
« Estimating, with known uncertainty, the
status and health of any regionally defined
resource.
• Describing baseline data leading to
rigorous detection and description of
trends in status and health of regionally
defined resources.
• Identifying associations among attributes,
both within and among resources, to
establish possible causes of impaired
condition.
• Quickly responding to new issues and
questions.
Important requirements and features of the design
Include:
• Explicit definition of target populations and
their sampling units.
• Explicit definition of a frame for listing or
otherwise representing all the potential
sampling units within each target popula-
tion.
• Use of probability samples on well-defined
sampling frames to estimate population
attributes rigorously through randomization.
• Rexibility to accommodate a variety of
resource types and a variety of problems,
some of which have not yet been specified.
• Hierarchical structure permitting sampling
at a coarser or finer level of resolution than
the general grid density, giving flexibility at
global, national, regional, or local scales.
• Ability to focus on subpopulations of
potentially greater interest, such as specific
types of lakes rather than all lakes.
• Ability to quantify statistical uncertainty and
sources of statistical variability for
populations and subpopulations of interest.
The proposed design strategy is a randomized,
systematic sampling of sites selected in associa-
tion with a uniform spatial grid. This sampling
strategy employs a permanent national sampling
framework consisting of a hexagonal plate con-
taining a triangular grid of approximately 12,600
points placed randomly over the conterminous
United States (Figure 2-4); a similar array is
available for Alaska and Hawaii. A defined area
(40-km2 hexagon) around each point will be char-
acterized using existing maps, satellite imagery,
aerial photography, and existing databases of
landscape information (Figure 2-5}. These data
will be periodically updated, most likely at 10-year
intervals.
These 40-km2 hexagons describe a sample area
one-sixteenth of the area of the United States, and
provide the basis for regional estimates of the
areal extent of resources and numbers, and their
changes over time. The collection of resource
units contained within these 40-km2 hexagons for
any explicitly defined subpopulation (e.g., lakes
between 10 and 2,000 ha) constitutes a Tier 1
sample. From data collected at Tier 1 (e.g.,
landscape descriptions), classification and further
subpopulation development can be conducted.
The sample of resources reflects the true proper-
ties of that resource. This follows from the rigor-
ous prescription of the protocol.
The triangular nature of the grid points also per-
mits us to increase or decrease the density of the
grid three-, four-, or sevenfold, as well as by
multiples of these values (Figure 2-6). This
flexibility allows for sampling at densities
appropriate for the resource, whether it be
common or scarce, and at the scale of resolution
desired. The design is flexible enough for sample
selection at a resolution appropriate for use by
individual States should they desire.
14
-------
Figure 2-4. The baseline grid (not randomized) for North America containing about 12,600 points in
the conterminous United States. Spacing between points is about 27 kilometers.
15
-------
Figure 2-5. The landscape characterization hexagons are one-sixteenth of the total area and are
centered on the sampling points. The randomly positioned sampling grid occupies a
common but randomly selected position in each of the base tessellation hexagons.
Figure 2-6. Enhancement factors for increasing the base grid density. Enhancement will be made
only in the sample grid, not in the base grid.
16
-------
The characterization carried out on the 40-km
hexagons will provide an inventory of all aquatic
resources within them. Thus it will provide a
one-sixteenth sample of the extent, abundance,
and distribution of any aquatic resource that can
be identified during the landscape characteri-
zation. A subsample of this Tier 1 sample,
selected by probability methods, will be visited for
detailed site characterization of indicators of
condition of the resource. Data obtained from
these field samples will constitute the information
to be used for reporting on regional status and
trends in ecological condition and exposure. This
second stage of the double sample will constitute
Tier 2 of the design.
The complete landscape description at Tier 1
ensures that no ecological resource is excluded
by definition of the monitoring coverage, so that
there are no orphan resources. This does not
mean that every resource of potential interest is
necessarily monitored at the Tier 2 or field level.
The Tier 2 sample is implemented for resources of
current concern. If a need to monitor an addi-
tional class of resource is identified, it will be
necessary to return to the Tier 1 level, identify the
extent and location of that resource, and select
appropriate sites for field sampling. However, the
existence of the landscape characterization pro-
vides a ready-made frame for sample selection,
so that a field program can be mounted with
minimal preparation. This provides the potential
for quick response to new or emerging issues of
ecological condition.
An important aspect of the EMAP design is the
temporal and spatial interpenetrating nature of site
characterization and field visits. Whereas the
sampling grid consists of 12,600 points distributed
across the conterminous United States, only one-
fourth of these will be considered each year and
a subset of those will be sampled. Thus, over a
four-year cycle, each grid point will be consid-
ered. After four years, the cycle will be repeated,
revisiting grid points and sites selected during the
first four-year cycle. Figure 2-7 illustrates the
temporal and spatial distribution of the interpene-
trating approach. During the early years of the
program, a subset of sites will be visited each
year to evaluate year-to-year variability and to
identify important trends that might be detected
early. A few new sites will be selected each year
to estimate any bias that might be introduced by
knowledge that certain sites are EMAP sites,
resulting in differential treatment of those sites or
surrounding areas (see discussion in Section 6).
2.1.2 Indicator Approach
Traditionally, environmental monitoring programs
have focused on individual chemical or biological
species at the local scale. The concentration or
response of these individual species was
assumed to be equivalent to poor or good
environmental condition. Individual chemicals
and organisms, however, do not exist in isolation,
but rather interact with other physical, chemical,
and biological factors to produce ecosystem
response. In addition, we have learned over the
last two decades that continuing, persistent, and
cumulative pollution is occurring not only at the
local scale but also on regional, continental, and
global scales. In some instances, control and
regulatory programs at the local scale have
aggravated or contributed to problems at the
regional scale (e.g., acidic deposition). The
complexity of environmental processes and the
concept of regional and national scale effects
require a new approach to environmental moni-
toring, both in terms of what we measure and
where we measure it.
Instead of focusing on problems that receive the
greatest media or public attention, we need to
focus on problems that pose the greatest risk to
the environment. In Unfinished Business: A
Comparative Assessment of Environmental Prob-
lems (U.S. EPA 1987b), the EPA considered the
highest potential risks to ecosystems to be global
warming and stratospheric ozone depletion, fol-
lowed by regional problems of habitat alteration,
nonpoint source pollution, and risks from criteria
air pollutants. Because environmental problems
are becoming increasingly complex (i.e., cumula-
tive effects from multiple pollutants at multiple
scales), traditional monitoring approaches and
indicators alone are not enough to assess eco-
logical condition. EMAP is being designed with a
"top-down" approach in a risk assessment frame-
work. This top-down approach puts the primary
focus on the endpoints of concern rather than on
the environmental perturbations or stressors.
Risk assessment Is the formal framework adopted
by EPA to estimate the risks to human health and
welfare arising from various hazards and
associated mitigation strategies (Yosie 1987,
Deisler 1988). Although initially applied by EPA in
1-7
-------
Year
Cycle 1 234 56 7 8 9 10 11 12 13 14 15 16 17 18 19 20
1 +• -H -h -{- ' -K
2 A A A A A
3 * ' ' * * * ,*'
4 D D D -• D ' D
Temporal distribution
+ A. * + A + A + A + A + A
* n * n * n *•" b- *- n * n
A-..+ A + A+ A +.'. A • + A + •
n . *• n * n * n * n. * n *.
+ A ,+ A -HA + A ''..-I- A •+. A
* n •*• n '* n * 'n * n •* n
Spatial distribution
|+year1 ^ year2 % years D year4 |
Legend ;
Figure 2-7. Spatially interpenetrating samples on a 4-year rotating cycle.
the human carcinogen arena, the risk assessment
approach is now being considered for evaluating
ecological risks at various spatial scales
(Bascietto et ai. 1990, Hunsaker et ai. 1990).
Ecological risk assessment procedures estimate
the effects of anthropogenic activities on eco-
logical resources and allow the significance of
those effects to be interpreted with quantified
uncertainty estimates. The ideal output of the risk
assessment process Is an estimated probability
that an event of a certain magnitude will occur
(e.g., a 90% probability of loss of fish populations
In 20% of the lakes in a region; Suter 1990). Key
components of risk assessment include (1) selec-
tion of endpoints, (2) qualitative and quantitative
descriptions of the sources of the hazard (e.g.,
locations and emission levels for pollutant
sources), (3) identification and description of the
reference environment within which effects occur
or are expected to occur, (4) estimaltion or meas-
urement of spatiotemporal patterns of exposure,
and (5) quantification of the relationship between
exposure in the modified environment and effects
on biota (Hunsaker and Carpenter 1990). The
adaptation of the standard risk assessment para-
digm to accommodate ecological ristk at different
time and space scales has been proposed (Suter
1990, Hunsaker et al 1990, Warren-Hicks et al.
1989) and is currently under active discussion
within EPA (Bascietto et al. 1990).
Although the sequence of activities in traditional
risk assessment moves from source to exposure
assessment and then to effects assessment, an
epidemiological or top-down approach to eco-
logical risk assessment is also appropriate (Suter
1990, Fava et al. 1987). In effects-driven risk
assessments, we observe an effect or response,
18
-------
then work from this response toward identification
of the hazard or stressors by using exposure
information. This top-down or effects-driven
assessment is under development for EMAP.
Ecological risk assessments focus on the valued
ecosystem attributes or endpoints of concern.
Suter (1990a) recognized two types of endpoints
that are used in risk assessments. Assessment
endpoints are formal expressions of the actual
environmental value that is to be protected, they
should have unambiguous operational definitions,
have social or biological relevance, and be amen-
able to prediction or measurement (e.g., a
decrease in biodiversity, probability of a >10%
reduction in game fish production, or the
proportion of waterfowl killed within a region of
pesticide use). The quantitative results of the
measurements taken to characterize assessment
endpoints are termed measurement endpoints.
Measurement endpoints must correspond to or be
predictive of assessment endpoints. Examples of
measurement endpoints include the percentage of
trees that exhibit visual symptoms of foliar
damage due to oxidants, or the fraction of
streams that exceed the LCgQ for a toxicant to
largemouth bass.
Assessment endpoints by definition must relate to
the environmental values identified for each
resource category in EMAP. Therefore, the
measurement endpoints for EMAP are ideally
response indicators. Exposure and habitat indi-
cators are not our first choice as measurement
endpoints because they do not integrate the
effects of multiple pollutants or the effects of
unknown hazards. Stressor indicators are farthest
removed from the effects of interest and therefore
are also poor measurement endpoints. Defini-
tions of these four indicator types are:
• Response Indicator: A characteristic of
the environment measured to provide, evi-
dence of the biological condition of a
resource at the organism, population,
community, or.ecosystem level of organi-
zation (e.g, trophic index, fish assem-
blages).
• Exposure Indicator: A characteristic of
the environment measured to provide evi-
dence of the occurrence or magnitude of
contact with a .physjcal, chemical, or bio-
logical stressor (e.g., nutrient concen-
trations, tissue residues, toxicity tests).
• Habitat Indicator: A physical, chemical, or
biological attribute measured to character-
ize the condition necessary to support an
organism, population, community, or eco-
system in the absence of pollutants (e.g.,
availability of snags, substrate of stream
bottom, vegetation type and extent).
• Stressor Indicator: A characteristic
measured to quantify a natural process, an
environmental hazard, or a management
action that effects changes in exposure
and habitat (e.g., land use).
EMAP will collect data on response, exposure,
and habitat indicators at its field sampling sites.
Data on land use and land cover will be collected
as part of the landscape characterization activity
within EMAP. Additional stressor indicators will
be assembled from other sources. Within EMAP,
indicators are identified through the development
of conceptual models of ecosystems. These
models may be based primarily on how current
and anticipated stresses affect ecosystems, or
from a perspective of the structural, functional,
and recuperative features of "healthy" ecosystems.
Specific indicators for EMAP-Surface Waters are
discussed in Section 4; they should be seen as a
candidate list reflecting current progress but
requiring refinement. The suite of indicators
selected must be used in a coordinated manner
to provide a comprehensive picture of surface
water condition.
Selection and development of indicators for
EMAP-Surface Waters will be a long-term process.
Some indicators are ready for implementation on
a national basis, others have been used effectively
in certain regions and require modification for new
regions, and still others have been used only in
isolated instances and require demonstration of
their regional and national applicability. A multi-
phase process has been identified to guide the
current selection of indicators and their future
development;
• Phase ,1 - identification of issues (environ-
mental values and apparent .stressors) and
assessment endpoints. .
• Phase 2 - development of a set of, candi-
date indicators linked to the identified
endpoints and responsive to expected
stressors. ',",".'.'..'.'.',
1.9
-------
• Phase 3 - screening of the candidate indi-
cators based on a set of indicator evalua-
tion criteria, selecting as research indi-
cators those that appear to fulfill key
requirements, rejecting those that clearly
don't, and holding In a state of evaluation
candidate Indicators that may, in the near
future, advance to research status.
• Phase 4 - quantitative testing and evalua-
tion of the expected performance of
research indicators on regional scales, to
identify the subset of developmental indi-
cators suitable for regional demonstration
projects.
• Phase 5 - implementation of a core set of
Indicators In a full national EMAP program
with annual sampling and statistical sum-
maries, and periodic reevaluation of indi-
cators.
Selection of assessment endpoints, and the sub-
sequent selection of indicators (Figure 2-8),
requires consideration of public values, policies,
and current threats to surface waters, and an
understanding of biological and chemical proc-
esses. Figure 2-8 suggests that selection of biotic
Indicators of condition is driven by our under-
standing of biological communities, their inter-
actions, and the Impact of various stresses on the
chemical and physical habitats in which the
organisms exist. Selection of indicators is equally
Influenced by the endpoints or values of concern.
These endpoints, and thus the indicators repre-
senting them, must be sufficient to motivate
change In policy when poor conditions are found.
The exposure/habitat and stressor indicators
must be selected based on our current under-
standing of environmental stresses and their
Impact on the chemical and physical habitat and
hence on the biota.
As discussed in Section 1, primary consideration
In EMAP-Surface Waters will be given to those
ecological characteristics for which we manage
surface waters, specifically:
• Biological Integrity
• Trophic condition
• FIshabillty
Figure 2-9 provides a general overview of EMAP-
Surface Waters Indicators, relating the selected
endpoints, response indicators, exposure indi-
cators, and stressors. Figure 2-10 shows how the
results for one indicator (Index of Biotic Integrity)
might be displayed in a cumulative frequency
distribution (CDF). The Y-axis shows the cumu-
lative percentage of the stream miles, and the
X-axis shows the IBI scores. In this instance, an
IBI of 40 was delineated as the acceptable score.
Forty-two percent of the stream miles had a score
of 40% or lower. We could transfer the data into
a bar chart showing 58% of the population in
nominal or acceptable condition and 42% in
subnominal condition.
In addition to knowing what proportion of the
surface waters are in good or poor (nominal/
subnominal) condition, it is important to determine
the likely causes of these conditions. In EMAP,
this diagnosis will be achieved primarily by using
the exposure, habitat, and stressor indicators as
defined earlier. In limited situations, the response
indicators themselves may shed some light on
likely causes of current conditions or trends.
Statistical associations between the occurrence of
poor conditions as defined by response indicators
and values for the exposure, habitat, and stressor
indicators will be used to infer the most likely
categories of probable cause. Correlative anal-
yses of these types cannot prove causality but will
narrow the range of likely explanations for
observed regional patterns. Ultimately, we would
like to be in a position to display the results for
each of the three endpoints with a nominal/sub-
nominal split, and partitioning of the subnominal
sites to the probable cause for poor conditions
(Figure 2-11).
2.2 DATA QUALITY
An essential complement to the careful design
and indicator strategies outlined earlier in this
section is consideration of data quality. Pro-
duction and assurance of quality data should be
an integral part of any program that intends to
produce useful information. EMAP is committed
to producing data useful for its intended pur-
poses. To meet this objective, it isi necessary to
have a process for identifying the quality of data
needed. The following discussion represents a
process that has just begun for EMAP-Surface
Waters but that should be ongoing during the
early stages of implementation.
Data quality objectives (DQOs) are statements of
the level of uncertainty a decision maker is willing
to accept in results derived from environmental
20
-------
Biotic
Indicators
of Condition
Biological
Communities
(processes and
interactions altered
due to exposure to
modified chemical,
physical, and biological
habitats)
Endpoints ••-
of Concern
Drive Policy
Chemical Habitat
Physical Habitat
Biological Habitat
(alteration of these
habitats due to stresses)
Policy
Directed
Toward
Impacting
Anthropogenic
Stresses
Context of these
decisions may
be resource class
specific or
regionally specific
Figure 2-8. General approach for identifying indicators. It requires consideration of the public values
or endpoints (which drive policy), the potential stresses (at which policy is generally
directed), and the current understanding of interactions between chemical, physical, and
biological processes.
ENDPOINTS
IMPACTS
RESULTS
Trophic State
Fishability
Biotic Integrity
ORGANISMS
Fish
Macro-
invertebrates
Phytoplankton/
Periphyton
Sedimentary
Diatoms
Eutrophication
Acidification
Contamination
Habitat alteration
Physical Habitat
Index
Water Quality
Toxicity Bloassays
Nutrient Loadings
Contaminant
Loadings
Water quality
degradation
Physical habitat
deterioration
Semiaquatlc
Vertebrates
Decrease or
extirpation of
native species
Landuse/Landcover
, Atmospheric deposition/
emissions
Chemical application
estimates
Flow/stage records
Stocking and harvesting
records
Direction of impact
Direction of diagnosis
Figure 2-9. Indicator approach for EMAP-Surface Waters showing candidate indicators and the top-
down approach to problem identification and diagnosis of probable cause.
21
-------
100%
75%-
50% -
25%
44,800 km
30
IBI Score
Threshold
Figure 2-10. Cumulative frequency distribution for Index of Biotic Integrity in a region showing
how data might be displayed. With a threshold value of 40 identifying systems in
poor or good condition, the data can be transferred to a simple bar chart displaying
the proportion of the stream miles in nominal or subnominal condition.
Si
-1
x 5
i m
Nominal
Unknown
Habitat Alteration
Atmospheric
Non-Point Source
Point Source
Trophic
Condition
Flshablllty
Biotic
Integrity
Figure 2-11. Long-term assessment objectives for EMAP-Surface Waters.
22
-------
data. The DQOs are generated in a multi-stage
process (Figure 2-12). Stage I involves defining
the major questions or problems of concern. At
this stage, the focus is on the decision maker or
data user. Information needs are identified, based
on an understanding of how the environmental
data to be collected will be used. Also, resource
and time constraints are identified and the
consequences of Type I and Type II errors are
examined.
A Type I error results in a false positive (e.g., data
indicate adverse ecological effects when no such
effects exist), whereas a Type II error results in a
false negative (e.g., data indicate no adverse
ecological effects when they actually exist).
Stage II involves defining the information needed
to answer the questions or make the decisions
identified in Stage I. This process includes
developing pertinent subordinate questions that
may need to be answered jn order to fully
address the problem. Also, at this stage, the
population of interest should be clearly defined.
There should also be an identification of specific
design constraints, i.e., the desired confidence in
the results. Finally, Stage II should examine
existing data and confirm the need for new data,
in cases where data do not exist to provide the
necessary information.
Stage III involves determining a scientific
approach to data collection and the data quality
requirements for that approach. This process
includes considering as many approaches to col-
lecting the necessary data as possible, along with
considering the levels of data quality required to
meet the constraints specified in Stage I and
Stage II. This stage should also serve to identify
research activities needed to meet the require-
ments of Stages I and II.
— »
' <
*' " <
STAGE I: PROBLEM QF CONCERN
• What is the purpose of the environmental data?
. What are the resources and time constraints?
. What are the consequences of Type I and Type II
errors?
1
STAGE II: INFORMATION NEEDS
• What is the population of interest? • •'••
• What level of confidence must attend results?
• Does pertinent, usable data currently exist? •
What new data are needed?
.•; - , i,,- .• ,
STAGE III: SCIENTIFIC APPROACH
• What approaches to data collection are available?
''•"'.... ."•'.' •' •"
commensurate, with 'Stage1 IL requirements?
• What R&D activities are needed to meet Stage II ,
requirements?/ > />;, -V'-vVV-- .
i
i
Figure 2-12. The three phases of the data quality objective process.
23
-------
Throughout the process, there is continuous
feedback and communication at all levels
Involving the decision makers, the scientists
Involved In Identifying specific information needs,
and those Individuals collecting the data neces-
sary to provide this information. EMAP is com-
mitted to the DQO process as a means of optim-
izing the allocation of limited resources and
assuring that data collected in the program
provides the Information needed to meet program
goals.
2.2.1 The Data Quality Hierarchy
Data quality exists at several levels (Figure 2-13).
Measurement data quality includes attributes such
as precision, accuracy, representativeness, com-
parability, and completeness associated with the
measurement of environmental variables. At the
next level, it also includes the uncertainty
associated with the methods used to assimilate
these measurement data into an assessment (i.e.,
provide information from the data). For EMAP,
this can be perceived as "indicator quality," since
the indicators are the tools used to provide infor-
mation about specific aspects of ecosystem con-
dition. Factors affecting quality at this level
comprise not only measurement data quality, but
also sampling design and statistical data analysis.
At the next level, these indicators are aggregated
into an overall assessment of system condition.
The uncertainty associated with each indicator
must be included in an estimate of the overall
uncertainty of this aggregate assessment. This
uncertainty will then be compared to ecosystem-
level quality objectives to assure that data
collection and interpretation {activities are
consistent with program objectives in terms of the
quality of the information provided. Finally, EMAP
intends to integrate informaition across
ecosystems in order to make regional-scale
assessments of ecological condition. Again, the
uncertainty associated with each component of
the evaluation (i.e., individual ecosystem
SOCIETAL/ENVIRONMENTAL
ENDPOINTS
Cross-Ecosystem Assessments
ECOSYSTEM QO \ Aggregate Indicators
Indicators of
Ecological Interest
Baseline Data
Figure 2-13. Quality objective hierarchy in the data quality objective process.
24
-------
assessments) must be incorporated into an esti-
mate of overall uncertainty for the assessment
and compared to cross-ecosystem quality objec-
tives. The DQO process requires that sources of
variability be identified at each level and that all
relevant sources be considered in generating
estimates of uncertainty at any level of the
hierarchy.
2.2.2 The Role of DQOs in EMAP
The EMAP mission provides the Stage I input to
initiate the DQO process. The EPA perceived a
need within the Agency and in other client groups
for information regarding the current extent of
various ecological resources (i.e., how much of
each resource exists), the current status or
condition of those resources, and some indication
of trends in extent and condition over time. The
need for environmental information has been
stated in very qualitative terms at this level.
At this point in the process, the tools necessary to
measure extent and define condition within each
of these ecosystems are being developed. This,
in turn, will allow policy and decision makers to
articulate the requirements for data quality in
more quantitative terms. The program is now in
Stage II, with extensive feedback to Stage I, and
the process may require several iterations.
In Stage II, each resource group within EMAP
must develop a series of indicators that, in
aggregate, allow for an overall assessment of
ecosystem condition. Quantitative "logic state-
ments" must be developed describing the data to
be collected for each indicator and the way in
which the data will be used to provide information
on system condition. These statements should
include critical values above which the system is
in an acceptable or marginally impacted condition
(i.e., nominal) and below which the system is
significantly impacted (i.e., subnominal). This
critical value must be scientifically defensible.
Where possible, these statements should also
relate each indicator to endpoints of societal
interest or concern, so the ramifications of
changes in system condition can be understood
and appreciated by a variety of client groups.
In addition to developing these logic statements,
a series of error constraints must be developed
that identify all known sources of error or uncer-
tainty associated with the indicator. These will
include measurement error (the difference
between sample values and in situ true values).
Measurement error can be further divided into
analytical error (error associated with the meas-
urement process) and error from other sources
such as sample design, collection, handling,
storage, and preservation. Sampling error is a
function of natural spatial and temporal variability
and sampling design. Wherever possible, esti-
mates should be provided for each source of
uncertainty. In this way, factors that contribute
significantly to the overall variability of the
indicator are identified and the effectiveness of
various options in resource allocation can be
evaluated. For example, if spatial variability is the
major factor in the overall uncertainty associated
with an indicator, and measurement error is small
by comparison, it may be judicious to use a less
precise and less costly method of analysis and
invest more resources in increasing the sampling
density within a region to reduce the overall
uncertainty in the data.
Early in the program, individual indicators will be
used to make discreet assessments of condition.
Tools for making aggregate ecosystem assess-
ments and cross-ecosystem assessments will be
developed over time. The DQO process should
provide the framework for this development,
assuring that assessment tools at all levels
provide information of sufficient quality to meet
program objectives.
2.2.3 DQOs in the Surface Water Component
In order to assimilate the information necessary to
evaluate indicators of potential interest to the
Surface Water project, a questionnaire was devel-
oped and sent to researchers with expertise in the
use of each indicator. Upon completion, these
questionnaires will become "fact sheets" that pro-
vide the initial input for the development of indi-
cator quality objectives. Table 2-1 is an example
of a fact sheet for the sediment toxicity test. For
some indicators, data exist that allow quantitative
estimates to be developed for each source of
uncertainty identified for the indicator. Historical
databases for these indicators are currently being
evaluated by the surface waters resource group.
However, for many indicators of potential interest
to the Surface Water project, no data of the
appropriate scale or methodology exist from
which to generate such estimates. Therefore,
data from monitoring activities performed in 1991
will be used to develop quantitative logic state-
ments and error constraints for these indicators.
25
-------
Table 2-1. Sample Fact Sheet and Logic Statement for EMAP-Surface Waters
EMAP-SURFACE WATERS INITIATIVE INDICATOR FACT SHEET
INDICATOR: Lake Sediment Toxicity
1. What Is the problem or environmental concern that this indicator will address?
Are the bottom sediments of lakes toxic to aquatic life?
2. How does this problem relate to societal interests or concerns?
If bottom sediments are toxic to aquatic life, they may also be a hazard to humans. These sediments
also provide critical habitats for many species necessary to maintain the aquatic food chain. As a
result, sediments that are toxic to these species will adversely affect the natural balance of the
ecosystem, commercial and recreational fisheries, and the aesthetics of lake environments.
3. What Information is needed to address the problem?
A method of measuring sediment toxicity.
Sublssues:
a. A definition of what is to be considered toxic (acute versus chronic, degree of response, etc.).
b. One or more indicator test organisms that, once exposed to sediment, exhibit an adverse
response (e.g., mortality, impaired growth, reduced reproduction, etc.) if the sediment is toxic.
c. A proper laboratory test design to reflect in-field toxicity.
4. What data will be collected in the field? By what method or methods?
Collection of sediment samples.
Sublssues:
a. Sampling device (core, grab, etc.)
b. Number of samples per lake
c. Depth of sample
5. What laboratory analyses will be performed? By what method or methods?
Laboratory bioassays.
Sublssues:
a. Choice of species
b. Type of exposure (solid-phase, pore water, elutriate)
c. Acute versus chronic
d. Chronic endpoints
e. Test duration
(Continued)
26
-------
Table 2-1. Sample Fact Sheet and Logic Statement for EMAP-Surface Waters (Continued)
Proposal:
EMAP-Surface Waters will perform 10-day acute solid-phase bioassays with Hyalella sp. amphipods
and 7-day solid-phase chronic assays with Ceriodaphnia and fathead minnow larvae. Overlying,
clean water (reconstituted laboratory water) will-be used in all tests. The endpoints for Ceriodaphnia
tests are mortality and reproduction; the endpoints for the fathead minnow test are mortality and
growth. . : ; , ; ..-.-•-
6. How will the field/laboratory data be used to provide the information necessary to address the
problem? Be specific.
Data from laboratory bioassays will be analyzed statistically to determine if field-collected sediment
samples produce mortality significantly greater than control, (control sediment, to be defined). Such
samples will be categorized as acutely toxic., -Samples for which chronic endpoints show responses
significantly different from controls will be categorized as chronically toxic. By inference, these
sediments should fail to support similarly sensitive species of comparable life-stage in the field and
would result in population and community level impacts in these systems.
7. What are the sources of variability for the indicator? Where possible, provide estimates of the
variability associated with each source.
Examples of sources of variability: : = ; ,; -••..-..• ,.,.•.,
Measurement -
Sampling -
Analytical
Other (sample collection, handling, etc.)
Spatial -..-.-••.
Temporal
Measurement variability can be due to differences in the health of test organisms, to variability
associated with multiple operators, or to variability in control sediment or laboratory water between
laboratories or over time. For the Ceriodaphnia and fathead minnow tests, single laboratory
precision is approximately 30-35%. Amphipod bioassay data will be assessed to .determine the
variability associated with this method. Nonmeasurement sampling variability is unknown at this time.
Sampling: Representativeness of sediment sample, that is, how well does a sediment sample
represent all sediment in a given lake? .-• ;• , ,• .
Test representativeness, that is, how well does the laboratory bioassay assess the
toxicity of the sediment in the field?
DRAFT LOGIC STATEMENT FOR SEDIMENT TOXICITY INDICATOR
What number and percentage of lakes in a given region of the United States have sediments toxic to
aquatic life, as demonstrated by effects observed relative to controls in laboratory bioassays?
27
-------
2.3 REPORTING
EMAP-Surface Waters will produce four types of
reports:
1. Annual statistical summary reports
2. Periodic interpretive reports
3. Special interest reports
4. Scientific articles
To be of maximum use, the data must be trans-
formed Into useful information as quickly as
possible. The objective of EMAP-Surface Waters
Is to publish summaries of the preceding field
season surveys within nine months of data collec-
tion. These reports will provide summaries of
response, exposure, and habitat indicator data for
resources and regions sampled, but will provide
minimal interpretation.
Interpretive reports will be published following
completion of four-year sampling cycles. These
reports will attempt to summarize indicator results
for the preceding four years and will integrate
Information from the suite of response, exposure,
habitat, and stressor indicators to determine the
regional and national status of lakes and streams.
Trends will be evaluated and the likely causes for
current conditions and changes in condition will
be assessed.
Special Issue reports will be used to present data
pertinent to specific issues such as changes in
surface water chemistry resulting from
Implementation of the Clean Air Act.
2.4 EMAP-SURFACE WATERS PROGRAM
LIMITATIONS
It is important to describe the limitations of EMAP-
Surface Waters, as well as what the program will
attempt to do. The program is not intended to
describe all components of an ecosystem or
resource type. It will not describe how systems
function. It will, however, provide information
about the condition of specific indicators at a
specific index period as a "snapshot" of the
overall condition of a system.
EMAP has not been optimized to address specific
stressors to surface waters such as acidification
or chemical contamination. Studies addressing
specific stressors are envisioned as the third tier
of EMAP. Currently, the only Tier 3 activity
planned for EMAP is the TIME project, which will
modify the EMAP-Surface Waters design to
address the response of surface waters to
changes in acidic deposition. EMAP is not
intended to be compliance monitoring and will not
replace the need for those activities. In general,
EMAP is intended to provide a common sampling
frame within which to assess the condition of
lakes and streams on a broad scale so that the
magnitude and geographical locaition of various
problems can be assessed and mitigation and
research priorities can be set more objectively.
The monitoring program is not intended to be
truly anticipatory, but rather to provide an
ongoing monitoring framework within which new
variables can be added or regional modifications
can be made so that the magnitude and extent of
new issues can be more quickly evaluated.
28
-------
3.0 MONITORING NETWORK DESIGN
The broad EMAP objectives and the more specific
surface water components of those objectives
have been discussed in previous sections.
Meeting these objectives requires a design
capable of:
• Estimating, with known uncertainty, the
status and health of any regionally defined
resource.
"•" Describing baseline data leading to rigor-
ous detection and description of trends in
status and health of regionally defined
resources.
• Identifying associations among attributes,
both within and among resources, to
establish possible causes of impaired
condition.
• Responding quickly to new issues and
questions.
Important requirements and features of the design
described in this section include:
« Explicit definition of target populations and
their sampling units.
• Explicit definition of a frame for listing or
otherwise representing all potential samp-
ling units within each target population.
« Use of probability samples on well- defined
sampling frames to rigorously estimate
population attributesthrough randomization
and use of probability methods for sample
unit selection.
• Flexibility to accommodate a variety of
resource types and a variety of problems,
some of which have not yet been specified.
• Hierarchical structure that permits sampling
at a coarser or finer level of resolution than
the general grid density, giving flexibility at
global, national, regional, or local scales.
• Ability to focus on subpopulations of
potentially greater interest (e.g., specific
types of lakes rather than all lakes).
A general overview of the EMAP design was pre-
sented in Section 2.0; greater detail about the
design can be found in Overton et al. (1990).
This section describes how the general EMAP
design will be used in the surface water compo-
nent of EMAP. It includes sections on specifi-
cation of explicit populations of lakes and
streams, sample units, frame development, and
association rules by which lake and stream
sample units will be selected for fielcl visits.
This section should give the reader a clear picture
of how sample units will be selected lo meet the
design objectives of EMAP. Some decisions have
been made regarding pieces of the design puzzle;
for other pieces, various options under consider-
ation will be presented, some only briefly. Our
initial attention is on lakes, particularly lakes in the
Northeast, the area of the country where EMAP-
Surface Waters will conduct a pilot study.
Examples from this area are used to illustrate
some of the decisions, procedures, and unan-
swered questions.
Consideration of streams is less well developed.
The present schedule calls for a stream pilot
study in FY92 that will primarily address questions
about identifying indicators for arid streams of the
west. Thus the directions and decisions regard-
ing streams are more tentative and general.
3.1 POPULATIONS, SAMPLE UNITS,
FRAMES, AND OTHER SAMPLING
CONCEPTS
3.1.1 Surface Water Populations: Defining
Lakes and Streams
To answer questions about the condition of lakes
or streams, it is necessary to specify explicitly
what constitutes the group under consideration.
This target population of lakes or streams consti-
tutes the body about which estimates of condition
will be made. Target populations can be region-
ally defined, such as lakes in the Northeast, or
defined on the basis of certain attributes, such as
lakes larger than 20 hectares. However, one of
the general goals of EMAP is to include all eco-
systems, so that there are no "orphan" systems.
Therefore, at the broadest level, the target
population for EMAP-Surface Waters is all lakes
and streams in the United States (Alaska and
29
-------
Hawaii Included), excluding the Great Lakes, for
which a separate component has been devel-
oped.
What constitutes a lake or stream? Cowardin et
al. (1979) developed a classification system for
wellands and deepwater habitats of the United
States. Deepwater habitats are "permanently
flooded lands lying below the deepwater boun-
dary of wetlands" and wetlands are "lands tran-
sitional between terrestrial and aquatic systems
where the water table is usually at or near the
surface or the land is covered by shallow water."
This classification includes definitions of deep-
water lacustrine (lake-like) and riverine habitats
that serve as a starting point for explicit definitions
of lakes and streams.
However, several modifications to the Cowardin
definitions are necessary, primarily to distinguish
what we generally consider lakes and streams
from wetlands. In the Cowardin classification
system, a deepwater habitat is > 2 m in depth
and has a surface area of at least 8 ha. This is
used as an approximate break between lacustrine
(lake) and palustrine (marsh, swamp) habitats.
Thus, our definition of a lake will correspond with
Cowardin's deepwater lacustrine habitat for lacus-
trine waterbodies > 8 ha. To cover lakes < 8 ha,
the combination of deepwater and lack of emer-
gent vegetation will be used to separate lakes
from wetland habitats. If the depth is > 2 m, the
waterbody will be considered a lake. If the depth
is < 2 m, the spatial extent of emergent vegeta-
tion will be used; if coverage is < 30% of the
waterbody area, the habitat will be classified as a
lake. One hectare will form the lower limit of
habitats considered as lakes.
An additional element of the distinction between
lakes and wetlands is the shifting of lakes to
wetlands due to natural filling, anthropogenic
filling, or climatic variations. Wetlands may shift
to lakes as a function of climatic variation or
direct management activities (e.g., impoundments
or dredging). Tabulation of the shifts from lakes
to wetlands and back will identify the long-term
trends In resource shifts. Shifts may be obser-
vable as part of the periodic landscape character-
ization, or as a result of repeated site visits.
For streams, the primary modification of the
Cowardin classification will be to include shallow
as well as deepwater riverine habitats. As
described later, the resource unit of interest will
be stream channels. Since stream traces as
shown on maps will be used to identify stream
units, it will not be possible to determine without
a field visit whether the channel represented by
the stream trace actually contains water, particu-
larly during the proposed low flow index periods.
There are several possible states (ephemeral,
intermittent, flowing) for the stream channel;
tabulation of the state of stream channels relative
to these possibilities will be necessary.
In practical terms, maps will be u&sd to represent
the populations of interest and to identify and
select sample units (lakes and streams) (see
Section 3.1.4 on frame development). The maps
chosen will contain most of the water bodies
meeting the above definitions, but they will con-
tain errors of misclassification, omission, or
location. In order to estimate the extent of lake
and stream resources of different types, it will be
necessary to estimate the error rates of the maps.
Both landscape characterization (including a wet-
lands inventory component) and site visits will be
used to determine if a mapped representation is
an accurate depiction of the desired resource
type. The probability-based design and site selec-
tion procedures permit inferences about map
accuracy and estimation of the error in the map
population estimates.
3.1.2 Subpopulations
The role of classification has been considered for
surface waters. Other resource groups have
chosen to classify for a variety of reasons,
including: '
• Flexibility of design, allowing the design to
be modified for specific classes (e.g.,
sampling each class with a different
approach).
• Sampling allocation, to prevent over-
sampling or undersampling of classes or to
increase the precision of estimates about
particular classes.
• Logistical and reporting convenience.
• Acknowledgment of natural regional
classes among resource types.
For example, EMAP-Forests identifies particular
forest types that can be delineated regionally.
EMAP-Near Coastal defines biogeographic prov-
30
-------
inces, and within each province, estuarine types.
EMAP-Wetlands defines approximately 16 wet-
lands classes for which estimates can be made.
Within EMAP, subpopulations are the classes of
resource types about which we wish to make
statements of condition and trends. Tier 1
resources are one kind of subpopulation; they
function as strata in the EMAP design. Other
kinds of subpopulations are groups of lakes or
streams that are not strata, but are classes of
interest. These may be classes within a particular
Tier 1 resource, or classes that include more than
one Tier 1 resource. Subpopulations may, but
need not, be hierarchical. There is great flexibility
to establish a variety of subpopulations on which
to focus the estimation of condition and trends.
Subpopulations serve two major purposes: (1) to
increase the precision about estimates of condi-
tion and trends, and (2) to target groups on which
to report.
For surface waters, a variety of classifications
have been discussed; in fact, there are so many
ways in which surface waters could be classified,
with substantive arguments for each, that we
decided to make classification a part of the Tier 1
and Tier 2 evaluations. The flexibility of the
design permits examination of candidate subpop-
ulations to determine whether sufficient sample
sizes could be obtained to make reasonable esti-
mates of condition. If the examination indicates
an insufficient sample size, the grid density can
be increased to meet the desired sample size (if
costs of the increased sampling can be met).
This same flexibility permits combining sites into
subpopulations after Tier 2 sampling (post-strati-
fication) to make statements about the condition
for different subpopulations.
Some examples illustrate possible surface water
subpopulations: size classes of lakes or streams,
reservoirs versus lakes; warmwater versus cold-
water lakes, small seepage versus drainage lakes,
alpine lakes, high-elevation streams, streams in
EPA Region III, lakes in the glaciated Upper
Midwest, lakes with acid neutralizing capacity
(ANC) < 50 jueq/L, and lakes in agricultural or
forested land. The flexibility of the EMAP design
allows the specification of subpopulations at the
Tier 1 level from landscape information or other
available data, or at the Tier 2 level after collection
and processing of field data.
Initial stages of the program will focus on the
most general subpopulations, such as EPA
regions (desirable by many as reporting units),
and ecological regions or lake or stream regions
to describe ecological conditions across relatively
homogeneous areas. We are also considering
size classes, to prevent oversampling the numer-
ous small lakes and streams (see Section 3.2.2.1).
Subpopulations will also play a large role in the
design of the TIME component of EMAP. In this
case, subpopulation groups will be created early
in the collection of field data, in order to assure
that the required density of sites exists in each
subpopulation. As described in Section 3.3,
regional trend detection for subpopulations of
acid sensitive lakes requires that approximately 15
(or more) sites be located in each of the regional
subpopulations. It is anticipated that TIME
subpopulations will be formed on the basis of
sampled information. In particular, chemical
characteristics of lakes in the Northeast pilot will
be used to describe subpopulations of lakes that
can be expected to respond similarly to changes
in the levels of acidic deposition that result from
emissions reductions.
The potential utility of other kinds of subpopula-
tions will be evaluated during the early phases of
EMAP-Surface Waters. Although a large number
of potential subpopulations can be envisioned, we
will be limited by the overall budget for routine
sampling.
3.1.3 Sample Units
Sample units are the individual lakes or stream
segments of interest and populations constitute
the sum of all possible sample units. Identifying
each sample unit uniquely is a necessary step
toward site selection, to prevent duplicate
selection by a single identification.
Individual lakes and reservoirs that meet the
criteria defining lakes will be sample units.
Operationally, the sample units will be those that
occur on maps chosen to represent the popula-
tions of interest. Since maps will be used to
select sites, and some of the criteria for defining
lakes are not evident from the maps, field charac-
terization will be used to identify waterbodies that
are not lakes (do not meet our modified Cowardin
criteria). A map error can be estimated by tabu-
lating the numbers of non-lakes and making
population inferences.
31
-------
For streams, sample units will be segments of the
stream network defined by the confluences or
upstream termini of stream channels appearing on
the selection maps. Segments could be defined
based on morphometric properties such as chan-
nel type, gradient, or change in stream substrate.
The logistics of implementing these types of
criteria are excessive, however, relative to the
possible improvement over using confluence
rules. Maps will be used to select sites, and
estimates of map error will be made via site visits,
reconnaissance, or characterization.
Each sample unit will be uniquely defined by a
node. As described later, the USGS Digital Line
Graphs (DLGs; 1:100,000 map scale) will be used
to create the lake frame. These digital maps
conveniently contain label points for each lake,
generally located at the center of the lake. Each
label point uniquely identifies a lake in this digital
database. Thus, for lakes, the label point will be
the unique node (e.g., Figure 3-1).
For streams, we are presently planning to identify
stream segments by locating nodes at the conflu-
ences and upstream termini of stream traces indi-
cated on maps, as well as at the outlets of lakes
and reservoirs. Each node would uniquely define
a downstream segment (Figure 3-2). Another
option would be to locate nodes at the midpoint
of a stream segment identified by confluences
and termini. This option would have advantages
for the few instances in which streams bifurcate.
In this case, each node would uniquely specify
the segment within which it was embedded. The
midpoint alternative would be relatively easy to
implement if DLGs are used, but more difficult
and time consuming if midpoints are located
manually and need to be precise.
These stream segment definitions create some
very short and some extremely long segments. A
quick estimate based on sampling from three
1:100,000-scale maps is presented in Table 3-1.
We tentatively suggest a minimum stream seg-
ment length of 500 m and a maximum of 5 km,
but we have not evaluated the consequences of
such delimiters. These are arbitrarily selected and
intended to place bounds on the range of lengths
of stream segments treated as sample units. We
solicit arguments proposing objective criteria for
minimum and maximum stream segment lengths.
3.1.4 Frame Development
In addition to identifying the target population and
the sample units, an essential component of a
survey design is to specify the frame, The frame
consists of the list of sample units that make up
a population, or a representation of that list, such
as a map. Explicit list frames that identify each
potential sample unit are often constructed. The
National Lake Survey used list frames for various
parts of the country, constructed by digitizing the
location of lakes identified on maps (e.g.,
1:250,000-scale USGS topographic maps for the
•Eastern Lake Survey; Linthurst et al. 1988).
Sometimes it is inefficient to construct a list frame
because the population is so large. In these
cases, an alternative is to create a map that repre-
sents the target population. Sites are selected
from the map, with well-defined rules. This was
the route chosen for the National Stream Survey
(Messer et al. 1986). Maps represented the target
population of stream segments in selected
regions. A systematic rectangular grid of points,
superimposed on the map, combined with a site
selection rule, was used to select stream seg-
ments for characterization and subsampling. In
this case, the map represented the target popu-
lation, as a virtual list frame, without explicitly
listing each sample unit.
3.1.4.1 Lake Frame
For lakes, it is feasible to create a list frame from
which sample units can be randomly drawn. The
USGS 1:100,000 series topographic maps contain
an inventory of lakes that includes lakes < 1 ha
across the conterminous United States. Digital
versions of these maps are available (referred to
as digital line graphs, or DLGs); they contain
several layers, including one of hydrography
(stream and lake traces). The advantages of
using DLGs for lake selection include automatic
processing with attendant decrease in human
errors, knowledge of locational accuracy (USGS
specifications), and lake areas included as part of
the file.
We developed a method of extracting the lake
coverages from this DLG series and creating a
point coverage of all lakes represented in the DLG
series, each with a unique identifier. This cover-
32
-------
Lake or Reservoir
Tier 1 Lake
Lake Nodes
T i e r.' .1 Sample
Figure 3-1. Example of the lake frame. Each lake is identified by a unique node. Lake nodes
occurring within hexagons identify the Tier 1 sample.
33
-------
Intermittent
Perennial
« Reach Nodes
ier 1 sample
Figure 3-2. Example of the stream frame, with nodes identifying stream segments that are the stream
sample units. Tier 1 stream segments, located by their unique nodes, are highlighted.
Each node uniquely identifies a downstream segment.
34
-------
Table 3-1. Estimates of Stream Reach Length Based on Segments Represented on Three 1:100,000-
scale Topographic Maps
Segment length (km)
Map
Mean
Std. dev.
Max.
Min.
Tacoma, WA
Hendersonville, NC
Steamboat Springs, CO
2.4
1.2
2.5
2.2
1.2
2.0
10.1
7.2
9.5
0.1
0.1
0.5
50
158
46
age could be used as a list frame from which to
draw the sample lakes; however, to be consistent
with the EMAP grid design, we will superimpose
the grid on this lake frame and select lakes
occurring within the 40-km2 hexagons using one
of the association rules described in Section
3.2.1.3.
3.1.4.2 Stream Frame
We are less definite about the frame materials for
streams. We will have to use maps as the frame
for selecting sample units, because constructing
list frames is too costly and is inconsistent with
the use of EMAP's grid design. There are no com-
pelling reasons to use a list frame for streams.
Several options for the choice of maps are under
consideration. Leading candidates are the USGS
1:100,000 and 1:250,000 topographic series. Site
selection should occur from a relatively consistent
set of maps across the region of interest.
To our knowledge, the USGS series of topo-
graphic maps at various scales is the only rela-
tively consistent set of maps that could be used
for selecting streams. Several map scales are
available. The 1:24,000 (7 1/2 minute) series are
too detailed for EMAP, and the manual labor
involved in locating grid points and 40-km hexa-
gons on thousands of 7 1 /2 minute maps would
be overwhelming. These maps could be used to
verify and characterize sites selected from the
other maps. Other options are outlined in the fol-
lowing paragraphs.
3.1.4.2.1 USGS 1:100,000 Topographic Series
Paper Maps and Digital Line Graphs
The paper maps are intended to cover the entire
conterminous United States, but not all are
available in topographic version. Some are
available as planimetric maps. Stream and lake
representation (detail) varies from place to place,
but the coverage is more detailed (contains more
streams and lakes) than that of the 1:250,000-
scale maps. Disadvantages of using the paper
maps include manual site selection, variation in
hydrographic resolution from place to place
(though less severe than on 1:250,000-scale
maps), and the possibility of too much detail at
the small stream size.
The digital line graph (DLG) version of these
maps consists of several layers, including land
ownership and political boundaries, transporta-
tion, and hydrology. The database is large,
requiring massive storage capacity. About 1,800
1:100,000-scale maps cover the conterminous
United States. The database contains 43 giga-
bytes, requiring 241 tapes for storage. The
hydrography coverage alone requires five
discpaks.
For streams, this digital database might be used,
but with some modifications. We will have to
address the massive storage requirements of this
system.
3.1.4.2.2 U.S. EPA's River Reach File
The River Reach File (RRF) is essentially an
enumeration of all lakes and river reaches at a
1:500,000 scale. It is being upgraded to
incorporate the USGS 1:100,000 DLGs. Proto-
35
-------
types have been completed; complete national
conversion Is underway, funded in stages by EPA
regions and the Office of Water. The version
based on the DLG Is called RF3. Conversion of
the DLGs into the River Reach File includes edge
matching, so traces from one map to another can
be tracked electronically. It may be feasible to
conduct Tier 1 site selection automatically. One
great advantage of a functional River Reach File
Is that the file Is tied to several EPA databases
that characterize the reaches in different ways.
We are actively Investigating the use of RF3 as a
lakes and streams frame.
3.1.4.2.3 1:250,000-scale USGS Topographic
Series Maps
These paper maps cover the entire conterminous
United States. Stream and lake representation
(detail) varies from place to place. Because these
are available only as paper maps, their use would
require manual site selection. The main disadvan-
tages Include manual site selection, variation in
hydrographic resolution from place to place, and
removal of detail so that this series does not
contain as many of the smaller streams as the
more detailed series. None of these disadvan-
tages Is insurmountable. The 1:250,000 series
maps were used quite successfully for site selec-
tion for the Eastern Lake Survey (Landers et al.
1988) and National Stream Survey (Messer et al.
1986) conducted as part of the National Acidic
Precipitation Assessment Program.
The choice of map representation will depend
ultimately upon whether the maps represent the
resource of interest adequately, the ease of use of
the maps, and the relationship of the map data-
bases to other databases. It would be ideal if the
RF3 could be used as an EMAP frame, so evalua-
tion of its use is of high priority.
3.2 THE TWO-STAGE SAMPLE: TIERS 1
AND 2
3,2.1 Tier 1 Resources
In EMAP, the first stage in the two-stage sampling
process Is called a Tier 1 sample, and a Tier 1
resource Is a first-stage stratum. The function of
designating a resource as a Tier 1 resource is to
Insure that an adequate sample size is generated
to make population estimates. Thus resources
that are of particular interest, or disinterest, or
rare, or top spatially limited to get adequate
representation may be designated as Tier 1
resources. Other examples of Tier 1 resources
include any resource that has the potential for
precluding the selection of others of greater or
equal interest, or a resource that requires special
treatment.
Tier 1 resources are strata on which Tier 2
samples are drawn in EMAP's two-stage sampling
process. Initial designation of Tier 1 resources
does not preclude future revision as necessary to
focus on particular subpopulatipns inadequately
represented during the initial phases.
Lakes and streams will constitute separate Tier 1
resources. For each, several size classes are
under consideration as a further distinction of Tier
1 resources. The possible need to distinguish
several size classes as Tier 1 resources is driven
by the characteristic of the EMAP design that
selects sites in proportion to their abundance. In
fact, any nonstratified random selection process
would select sample units in proportion to their
abundance. Since there are substantially greater
numbers of small lakes and streams than large,
small systems could preclude the selection of
larger systems. One way of balancing the selec-
tion is to identify size classes as separate Tier 1
resources and allocate samples according to
desired sample sizes for each stratum (see
Section 3.2.2.1).
Also, the grid design is inefficient for selecting the
relatively large, unique systems. It is preferable to
list these systems separately and to sample them
from a list frame, or to census them. This special
interest group will contain lakes larger than 2,000
ha (half the size of the 40-km2 hexagons) and the
largest rivers in the United States. We have not
identified cutoff sizes yet; the subject is under
discussion.
3.2.1.1 Lakes
The distinction between lakes and reseiivoirs is a
conventional way to stratify; however, lakes and
reservoirs will be considered together at the Tier
1 resource level. Reservoirs can be described as
a separate subpopulation, if necessary, for finer
scale descriptions. There are several reasons for
combining lakes and reservoirs at the Tier 1 level:
• In many areas, lakes and reservoirs func-
tion similarly, and display similar kinds of
36
-------
problems, so distinguishing between them
provides little additional insight.
• Similar indicators of condition will be used
for lakes and reservoirs.
• In some instances natural lakes are artifici-
ally controlled, making them a transition
between lakes and reservoirs.
• It is often difficult to distinguish between
lakes and reservoirs on the basis of
mapped information, particularly for the
smaller impoundments.
• There tends to be an inverse relationship
between the occurrence of reservoirs and
natural lakes; in lake-rich regions, relatively
few reservoirs occur; in lake-poor regions,
a greater density of reservoirs occurs. Site
selection based on the grid design should
reflect the relative distribution and abun-
dance of lakes and reservoirs. Sampling
density can be modified as necessary by
increasing or decreasing the density of the
basic grid (Overton et al. 1990). However,
see Thornton et al. (1990) for discussion of
differences between lakes and reservoirs.
In any case, the natural lake/reservoir status of
any site selected for field visit will be recorded;
data summaries can be created for each group
and the condition of each subpopulation can be
described. If sufficient interest or justification
develops to consider reservoirs as a separate Tier
1 resource, the flexibility of the design will
accommodate this.
As noted earlier, lake size is being considered for
allocation of lakes into different Tier 1 resources.
The largest lakes (e.g., those > 2,000 ha) consti-
tute a separate group, because they are infre-
quently captured by the grid and increasing the
grid density is an inefficient way to select sparsely
distributed lakes. They can be identified and
listed as part of the Tier 1 sample selection proc-
ess and will constitute a separate Tier 1 resource.
The remainder of the lakes can be divided into
size classes, primarily as a device to assure ade-
quate selection of larger lakes. Each size class
will constitute a separate Tier 1 resource. A
recently completed summary of the size distri-
bution of lakes in EPA Regions 1 and 2 (North-
east) indicates that of the -18,000 lakes > 1 ha,
70% are < 10 ha, supporting the need to
establish size criteria as Tier 1 resource
delineators (Figure 3-3 and Table 3-2).
As already mentioned, lake size is used to identify
separate Tier 1 resources to prevent overwhelm-
ing the sample selection with small lakes. Strati-
fication by size classes permits the selection of
lakes such that inclusion probabilities within strata
are equal, a desirable property for minimizing the
variance of some kinds of population estimation
for that stratum and for simplifying the calculation
of population statistics. See Section 5 for a
discussion of inclusion probabilities and weights.
It is likely that stratification will produce vastly
different inclusion probabilities among strata, so
we will face the need to combine sample units
with different inclusion probabilities if, as is likely,
subpopulations are developed across the size
strata. Hence, it may be more desirable to
develop other methods to balance the selection of
large and small lakes, in which case all lakes will
be considered as a single stratum.
3.2.1.2 Streams
In order to represent the broad range of stream
sizes, from headwater mountain creeks to rivers
as large as the Mississippi or Columbia, it is
necessary to consider several possible Tier 1
resources based on size. The discussion begins
with a description of the two ends of the size
spectrum, then moves on to the intermediate
sizes. Five size classes of streams are suggested,
from Class I (smallest) through Class V (largest);
each class could constitute a separate Tier 1
resource. At the present time, only very general
size criteria are proposed. There is no need to
make explicit decisions about streams yet, so we
present the broad issues, with some discussion
about how these issues are being resolved. Part
of our activity over the next year will be to refine
our definitions of stream size classes and our
approaches for considering the broad spectrum
of stream sizes.
The characteristic of "size" for streams has no
single measure that is satisfactory for all
purposes. The closest measure of stream size is
probably a measure of discharge, which might be
average annual discharge, low flow discharge, or
some other expression of the hydrograph. A dis-
advantage of using stream flow is the general lack
of information on stream flow at the Tier 1 level.
37
-------
o o
NOIiaOdOHd
38
-------
Table 3-2. Summary of the Size Distribution of Lakes in the Northeast (EPA Regions 1 and 2)a
Part A
Size class (ha)
Tier 1 Sample
DLG Inventory
Number of lakes Total area (ha)
Number of lakes Total area (ha)
1-5
5-10
10-15
15-20
20-100
mn-9sn
662
219
94
58
143
41
1,600
1,500
1,150
1,000
6,000
5,900
10,745
3,565
1,476
917
2,665
548
27,200
25,300
18,000
15,800
115,500
82,500
Part B
Size class (ha)
250-500
500-1,000
1,000-2,000
> 2,000
DLG Inventory
Number of lakes
227
112
62
59
Total area (ha)
78,300
77,900
84,600
442,300
a Part A describes the Tier 1 sample and the DLG inventory for lakes < 250 ha in surface area. Part B describes the DLG inventory
for lakes > 250 ha. The DLG database hydrography is described in the text.
Other measures of size or surrogates for size
include watershed area (calibrated regionally),
stream length, stream order, channel dimensions,
or wetted usable area. The last measure pertains
to the area of the stream channel that aquatic
organisms use as habitat, usually measured
during a low flow period.
3.2.1.2.1 Class I
These are the smallest streams of potential
interest; they are generally represented on
1:100,000- or 1:250,000-scale maps as first or
second order (Strahler) perennial or intermittent
streams and have a high potential for being dry or
intermittent during the proposed EMAP sampling/
index period. They can contain subsurface flow
or enduring pools during parts of the year. In
much of the arid and semi-arid West, these
streams generally do not contain flowing water,
yet the stream channels function as an important
ecological resource. Therefore for this class, the
proposed resource unit is a channel segment
rather than a stream segment. Three general
types of channels can be described:
• Dry: These channels function primarily to
discharge water during episodic storm
events; they retain an insufficient amount of
surface or subsurface water to support
hydrophytic vegetation in the channel or
riparian zone; the riparian zone is inhabited
by upland vegetation up to the stream
bank and sometimes in the stream
channel.
• Generally dry: These channels often con-
tain no flowing water, but the channel
(including bank) substrate contains enough
water to support hydrophytic vegetation in
the riparian zone. Particularly in the west,
these ribbons of green in an otherwise dry
landscape act as corridors linking different
landscape types and are sites for locally
high biodiversity. They generally do not
contain traditional stream communities of
39
-------
macroinvertebrates and fish, except those
with enduring pools. Nonetheless, they are
an Important aquatic ecosystem.
• Perennial flow: These channels generally
contain flowing water and regionally typical
stream fish and macroinvertebrate com-
munities, as well as hydrophytic riparian
vegetation. They are the conventional
streams that we generally discuss.
The above distinctions might not be so important
In the humid regions of the country, but for vast
areas of the West, where flowing water is limited,
and where the flowing water that does exist is
often diverted from the stream channels, it is
important to focus on the channel itself. One
possible indicator of the gain or loss of stream
resources In arid regions might be the proportion
of systems that change from one of the above
classes to another. These shifts are apparent
during wet/dry climatic cycles.
The EMAP grid is efficient at selecting these
stream segments because their proportion and
density is high. Our expectation is that on the
average, three to six first and second order
segments will be captured per hexagon in humid
regions of the country, but on the average, only
one segment per two hexagons can be expected
In dry regions. The EMAP grid density appears
capable of selecting a sufficient number of these
segments for EMAP estimation.
3.2.1.2.2 Class V
These are the largest rivers; they stand out by
their uniqueness and should receive special atten-
tion. Examples are the Columbia, the Mississippi,
and the Missouri. The USGS has summarized the
largest rivers In the United States based on three
criteria of size: discharge at the mouth, water-
shed area, and length. It might be necessary to
treat these systems apart from the other EMAP
systems because each one could be tracked and
described separately, perhaps one-fourth of them
each four years.
3.2.1.2.3 Class IV
We usually call this group of streams rivers, but
they are smaller than the Class V rivers. They are
sparsely distributed relative to the EMAP grid,
thus they are Infrequently captured. Creating an
adequate sample could pose a problem. It can
also be argued that these rivers should be treated
as an extensive resource, included with the Class
V rivers as a single Tier 1 resource to be sampled
by probability methods other than the grid.
One option for identifying rivers that fall into this
class is based on an estimation of the stream
density necessary to meet a minimum sample
size. Lower stream densities would result in
sample sizes too low for adequate estimation of
condition. An alternative is to enumerate and
map this class of rivers based on criteria similar to
those used by the USGS to identify the largest
rivers in the nation. An inventory could be
conducted for each USGS hydrographic basin.
For Class IV and V rivers, a stream's length, from
a point well below its headwaters to its mouth,
would be considered the resource unit. The point
at which a Class V or IV river begins could be
located such that the watershed area above it is
similar in size to stream watersheds in the region
selected from the EMAP grid. Thus the river
would not be sampled up to its headwaters, but
to some point substantially downstream. A list of
these rivers and a map identifying the origin and
terminus of each river resource unit can be
created.
3.2.1.2.4 Classes II and III
These streams fall between the small Class I
streams and the Class IV rivers. They are
approximated by third to fifth order on USGS
1:100,000- and 1:250,000-scale maps. It might be
reasonable to treat this group of streams as a
single class. We expect that an adequate sample
size can be obtained in humid regions via the
EMAP grid where, on the average, one to two
segments per hexagon occur. In dry regions, the
capture rate is about 1 segment in 10 hexagons,
so it may be necessary to increase the grid
density to acquire a large enough sample.
3.2.1.2.5 Other
Artificially created diversions, canals, and other
waterways that do not follow the natural drainage
pattern of streams could be important resources
in some areas where most of the flow of water is
through these artificial systems. These areas and
systems will be considered on a case-by-case
basis as stream implementation proceeds.
40
-------
3.2.1.3 Tier 1 Association Rules
Association rules guide the selection of sites from
the frame. They are required because the sys-
tematic grid points of the EMAP design do not fall
on lake or stream nodes. Two options have been
considered for lake and stream sample selection.
One is to choose the nearest lake or stream, via
its .unique node, to the grid point within each
40-km2 hexagon. The second is to create a list of
lakes or streams that occur within the hexagon
and then select one of these for each Tier 1
resource, via a probability sample.
Given that the two options for selecting the Tier 2
sample do not result in appreciable differences in
precision, there are several reasons for preferring
the list approach over the nearest lake or stream
approach. First, the list approach conforms to
the standard sampling techniques of double sam-
ple and cluster sample, and thus has well known
statistical properties. Second, the list approach is
easier to implement. Finally, and perhaps most
important, the full set of sample units in each
hexagon would be carried into the second stage
of the two-stage selection process, and so will be
available for modification of the sample at a future
time.
3.2.1.4 Adequacy of the Grid for Tier 1
Sample Selection
To gain familiarity with the grid design and to
evaluate how many lakes and stream segments
fall within the 40-km2 hexagons, we sampled 59
1:100,000-scale USGS topographic maps in dif-
ferent parts of the country. We used a mylar
overlay, with the 40-km2 hexagons outlined at the
Tier 1 density, and counted lakes and stream
segments that fell within the hexagons. Table 3-3
summarizes the capture rates and the frequencies
with which no lakes or streams appeared within
the hexagons. For this study we selected only
nodes associated with perennial stream traces
without distinguishing size classes.
Although the table summarizes only our results for
broad geographic regions, we also evaluated the
capture rates at the level of Omernik's aggregate
ecoregions (Figure 2-2). For example, in the
Middle Atlantic Coastal Plain ecoregion (2 maps,
18 hexagons), an average of 10.4 streams were
captured per hexagon and all hexagons captured
streams. At the other extreme, in the Subhumid
Agricultural Plains ecoregion, Northern Section (4
maps, 36 hexagons), no streams were captured
(on average) and 97% of the hexagons were with-
out stream segments. In areas where stream
density is fairly high, from 2 to 7 stream segments
were captured per hexagon, with approximately
20% of the hexagons without stream segments.
In areas where stream density is low, from 0 to
0.4 stream segments were captured per hexagon,
with more than 75% of the hexagons without
segments.
For lakes, regional variation was less severe. On
the average, one or two lakes occurred per hexa-
gon, regardless of the region sampled, except for
the driest regions. In some cases, more than 10
lakes were present in a hexagon. About 50% of
the hexagons had no lakes, but with regional vari-
ation. The lowest no capture rate (about 30%)
was in the glaciated regions of the Upper Midwest
and the Northeast. The highest no capture rates
occurred in the xeric regions (60-90%). Reser-
voirs were sparsely distributed. The no capture
rate was about 80-90%.
We concluded that for small systems, the capture
rate seems adequate. In some areas where lake
or stream density is low, it will be necessary to
increase the grid density, if it is deemed important
to make subpopulation estimates in these areas.
Because of the flexibility of the grid design,
changes in density can be accomplished as par-
ticular regions are implemented for sampling.
3.2.2 Tier 2 Sampling
Physical, chemical, and biological measurements
(Section 4) made at Tier 2 sites, plus landscape
descriptions, will complete the information needed
to estimate condition and to classify subpopula-
tions. Measurements to be taken at Tier 2 sites
are covered in Section 4. This section describes
the process by which Tier 2 sites are selected
from the candidate Tier 1 sites. Where to sample
within Tier 2 sites (index sites) and when to
sample (index periods) are addressed in Section
6.
A Tier 2 subsample will be drawn from the
collection of sites comprising the Tier 1 sample.
The size of the Tier 2 sample will be determined
through a combination of the following criteria:
• Importance of estimating resource status in
a particular region, relative to the avail-
ability of funds to conduct the field work.
41
-------
Region
Perennial stream segments
Lakes
Reservoirs
Hurnld/Subhumld
average (36 maps)
% no capture
5.2
19%
•1.7
48%
0.3
85%
Arfd/Sernl-arld
average (23 maps)
% no capture
0.3
84%
1.1
63%
0.4
75%
Data taken from 59 USGS 1:100,000-scale topographic (T) or planimetric
(P) maps; nine hexagons per map were sampled.
* Expected variation of lake or stream types
within the region. , ; , ^ -, V
• Expected variation of indicator values"
within the region. •.
• Quantity of the resource in particular
regions; if many lakes or streams occur in
a region, it may useful to increase the
sampling effort to characterize the many
possible subpopulations.
An operational guideline is that there should be 50
to 100 Tier 2 sites per subpopulation over the
four-year cycle in order to make adequate sub-
population estimates. Larger samples do not
increase precision enough to warrant the
increased costs (Linthurst et al. 1986).
3.2.2.1 Tier 2 Site Selection Example
EMAP-Surface Waters plans to conduct a pilot
survey of lakes in the northeastern United States
(covering EPA Regions 1 and 2). The following
example illustrates the site selection process as it
is being applied to this pilot project.
The design report (Overton et al. 1990) describes
two options for selecting Tier 2 sites from the Tier
1 candidate sites. Discussed with reference to
lakes, the options are:
1: 'OPTION 1: Within each 40-km2 hexagon
- that contains one or more lakes, select a
lake using one of the rules of association;
this reduced Tier 1 sample would be sub-
sampled to yield the Tier 2 sample.
2. OPTION 2: Select a subset of the 40-km2
hexagons, then select a lake from each of
the selected 40-km2 hexagons using one of
the association rules.
Option 2 has advantages for lake and stream site
selection and will be used to illustrate the Tier 2
site selection process for EPA Regions 1 and 2.
USGS 1:100,000 Digital Line Graphs were used in
the lake Tier 1/Tier 2 site selection process.
These digital coverages contain almost all lakes
represented on USGS 1:100,000-scale topo-
graphic maps; lakes < 1 ha are included as part
of this series. A map of the spatial distribution
(Figure 3-4) and the size distribution by area
(Table 3-2 and Figure 3-3) illustrate the nature of
this data base. A great majority of the lakes are
< 20 ha, creating a situation in which small lakes
would dominate any sample unless strata (Tier 1
42
-------
Figure 3-4. Map of the spatial distribution of lakes in the Northeast. Size class represented Is 1-250
ha in area.
43
-------
resource designation) based on size were
constructed.
The Northeast accounts for only 4.4% of the con-
terminous United States. If samples were allo-
cated in proportion to area of a region, -35
samples per year would be allocated to the North-
east, based on the assumption that we can afford
to sample 800 lakes (Tier 2) nationally when
EMAP-Surface Waters is fully operational.
Because of the greater density of lakes in some
regions of the country, it might be appropriate to
allocate a disproportionate amount of the samp-
ling effort to these areas, to make refined esti-
mates of condition of subpopulations. For the
pilot, we estimate we can afford to sample—100
lakes; this number will be used for planning.
We have tentatively selected three Tier 1 resource
groups based on lake size (area): small (1-20 ha),
large (20-2,000 ha), and off grid (> 2,000 ha). We
are continuing to discuss the boundary between
small and large lakes; it will lie in the range of 10
to 20 ha. This illustration uses the Tier 1 resource
"small lakes".
The following process describes one way to
select a Tier 2 sample that meets EMAP criteria
for population estimation. We are currently evalu-
ating a second process and will use the pilot of
FY91 to evaluate the two. We use a Tier 2 sam-
ple of 40 each for the small and large lakes, with
20 for special purposes (e.g., lakes > 2,000 ha or
reference lakes). Tier 1 sample sizes are: 910
lakes > 1 ha and < 20 ha, and 188 lakes > 20
and < 2,000 ha. There are 49 lakes > 2,000 ha
in the Northeast. Over a four-year cycle, 160 Tier
2 samples would be selected for small and large
lakes. At the present time, we think it is important
to oversample, understanding that we can cut
back the sample size in future years without
Interfering with the estimation process signifi-
cantly. Nearly all the Tier 1 sample of large lakes
would be selected. This could argue for grid
augmentation for this size group.
3.2.2.2 Mechanics of Tier 2 Site Selection
(Design Option 2)
Options 1 and 2 are both two-stage sampling
designs, but Option 2 follows the standard two-
stage format more closely than Option 1. Either
option can be used with any of the association
rules, and Option 2 with the "mini-list random
selection" is exactly the standard two-stage
design format. Option 2 adds some flexibility in
resampling if the initial selection results in
nontarget lakes.
The steps in the Tier 1 /Tier 2 sites selection
process are:
• Intersect the complete array of 40-km?
hexagons with the lake coverage to extract
those lakes .contained within hexagons in
the region of interest. This Tier .1 database
contains numbers of lakes per hexagon
and lake areas, so identifying small, large,
and off-grid lakes per hexagon is straight-
forward.
• Map the location of each hexagon with its
associated number of lakes (including 0) in
each Tier 1 resource category, for exam-
ple, a map of the small (1 to <2.0 ha) lake
captures.
• Select a subset of hexagons with the fol-
lowing constraints:
- Probability of selection is proportional
to the number of lakes contained within
each hexagon.
- Spatial distribution of lakes is ade-
quately preserved.
- One lake is chosen randomly from each
of the selected hexagons.
The result is a set of Tier 2 lakes whose weights
(or inclusion probabilities) are approximately
equal, and whose spatial distribution represents
the spatial distribution of the universe of lakes.
The procedure for meeting these constraints is to
delineate areas containing approximately equal
total number of lakes. The areas should be spati-
ally compact. The process by which the delinea-
tion is accomplished is statistically irrelevant; it
can be done subjectively, arbitrarily, or by some
objective protocol. There may be some advan-
tage to a subjective grouping using some criteria
of similarity (e.g., land use, ecological subregion)
other than proximity.
The number of lakes used to define these areas is
determined from the desired Tier 2 sample size
and a constraint that two hexagons be chosen
44
-------
within each area, on the average. The number of
lakes is calculated as:
2 * (Tier 1 sample size)
(Tier 2 sample size)
Hexagons are selected within each of the areas
such that the probability of hexagon selection is
proportional to the number of lakes in the hexa-
gon. Then a lake is selected randomly from each
chosen hexagon.
A specific example of these steps is illustrated for
small lakes in the sequence of panels of Figure 3-
5, with a Tier 1 sample size of 1,217 and a Tier 2
sample size of 80. Figure 3-5a illustrates the
hexagon capture rate for this size class. Areas
are delineated such that each area contains
approximately 30 lakes: 2 * (1,217/80), as in
Figure 3-5b. These areas were arbitrarily
selected, subject to the constraint that each area
is compact and that approximately 30 lakes are
contained in each area.
For this exercise, we assumed that each hexagon
with more than eight lakes was counted as if eight
lakes were present. .There has been some discus-
sion about the need to limit the number of sample
units per hexagon; the present illustration
assumes a particular value. Upon conclusion of
the discussions, the actual site selection will
follow specific agreed-upon rules.
Hexagons wererchosen from each of these areas,
generally two hexagons per area with probability
proportional to number of lakes per hexagon.
This was done by conceptually stringing the hexa-
gons along a line, each hexagon represented by
a line segment proportional to the number of
lakes within the hexagon. All hexagons within
each area are strung together sequentially, then
the next area is joined. With a random start,
hexagons were selected systematically by a
pointer that landed on the line every 15.2 units.
This pointer distance is the ratio of the Tier 1 Tier
2 sample size. The result of this process is the
selection of hexagons in proportion to the number
of lakes they contain; hexagons with a greater
number of lakes have a higher probability of
selection.
Then one lake is selected randomly per selected
hexagon. The result of these steps constitutes
the Tier 2 sample (four year). The year 1 sample
is obtained by choosing selected lakes in year 1
hexagons; year 1 hexagons are, identified as part
of the interpenetrating structure of the EMAP grid
(e.g., Figure 3-5c). Locations of Tier 2 hexagons
are illustrated in Figure 3-5d.
This general process will be used for lakes and
streams throughout the country, as the frame
materials are developed; the outcome will be the
list of Tier 2 lakes and streams selected for field
visitation. This Tier 2 subsample meets the EMAP
requirements of random selection and spatial
representation. , .
3.3 HIGHER GRID DENSITIES
Increasing the grid density is the preferred
method for increasing the sample size as needed
for any subpopulation. For lakes and streams it
appears that local high-interest areas may require
more spatially intense sampling than is provided
for by the standard grid, density. These areas will
be defined as the need arises; they are likely to
occur in arid regions of the country, or in areas ,
where a particular subpopulation of high interest
is sparsely distributed.
It will be necessary to use an iterative,process to
select the Tier 2 sample. The first step is to
define the desired Tier 2 sample size, then deter-
mine whether an adequate number, of sites is
available from the grid. If not, the grid density will
be increased until the desired sample size is.
achieved. The flexibility of the design makes this
feasible.
A specific case of the need for higher grid density
sampling occurs in the design of the TIME portion
of EMAP-Surface Waters. The main emphasis in
TIME is the detection of trends, rather than
descriptions of current status; good descriptions
of the acid sensitive status of current populations
of lakes and streams are already available from
the NSWS surveys (Linthurst etal. 1986, Landers
et al. 1987, Kaufmann et al. 1988). In order to
detect regional trends in a policy-relevant time
span (as dictated by the Clean Air Act, this time
span is about 10 years), and because of the vari-
ability inherent in acid-sensitive lakes and
streams, we determined that stratification of sites
will be necessary. Stratifying sites into sub-
populations narrows the various components of
variance that affect our ability to detect trends.
Simulations of trends suggest that approximately
15 sites will be needed per subpopulation, in
order to meet the requirements for trend
45
-------
Figure 3-5a. Hexagon capture rates for lakes (1-250 ha) in the Northeast.
46
-------
Figure 3-5b. Areas that group similar numbers of lakes for Tier 2 selection. Number of lakes per
area is 2 * Tier 1 sample size/tier 2 sample size.
47
-------
Cycle
1
2
3
4
Year
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
+ • + + * +•••' +'
A A A A A
% *K *'." * ^
. n . • n n — n n
Temporal distribution
-f A + A .-••+- A + A + A + A
* n * , D ';* n * ,D *'" n * n
A + A -f.'•; A + „. A + A +. A +i
D #'••• -p-:'-:'*;" D,'-"*" P1'.* D #•' IH '" *
+ A -f A -h A + A + A + A
* D * D * n * n * n # n
Spatial distribution
yean A year2 ^6 years D year4
Legend
Figure 3-Ec. The structure of the EMAP grid, illustrating the fourfold decomposition mat will be the
basis of the interpenetrating sample design.
48
-------
Figure 3-5d. Selected hexagons containing one or more lakes that will constitute the
sample. One lake per hexagon will be selected randomly within the hexagons, except
for those marked with an "x," from which two lakes will be sampled.
49
-------
magnitude and length of time to detect a trend
(see TIME conceptual plan; Stoddard 1990).
Based on Information gathered in the Eastern
Lake Survey, we estimate that approximately 10
subpopulatlons will be necessary to adequately
describe the types of lakes found in the North-
east, with 15 samples per year required. A total
of 150 Tier 2 lakes will therefore be necessary to
detect trends In the region.
Figure 3-6 Indicates the approximate geographic
regions that are sensitive to acidic deposition in
the Northeast (Stoddard 1990). Table 3-4 dis-
plays the number of Tier 1 hexagons and lakes
within these regions. These figures indicate that
the total number of lakes is sufficient in Tier 1 to
supply the needed coverage for TIME. Samples
for TIME, however, need to be .evenly spread over
the subpopulatlons of Interest, many of which are
not clearly mapable at this point. ;
Estimates of the coverage of subpopulatlons of
Interest was generated based on frequency of
occurrence of these subpopulatlons in the
Adlrondacks and extrapolated to the Northeast.
Table 3-5 presents these estimates. The results
suggest that the final Tier 2 density will be
sufficient for the coastal area only and that Tier 1
density will be acceptable for the subpopulation of
drainage lakes with moderate base cations in
Vermont and New Hampshire. A threefold aug-
mentation of the grid will be necessciry to select
the required density of sites in the regions where
the subpopulations are most likely to be found
(i.e., the Adirondacks, and the high-elevation
portions of Vermont and New Hampshire, Figure
3-7).
3.4 SUMMARY
This section describes how the general EMAP
design is applicable to the Surface Water com-
ponent of EMAP. It explains how the populations
of interest are specified and how the sample units
for lakes and streams are selected. It describes
frame development, rules for site selection, and
the Tier 1 /Tier 2 site selection process. As part of
the discussion, several pilot studies are presented
to illustrate how the EMAP design will be used for
surface waters,
.Our conclusion is that in general the grid design
will be adequate for selecting much of the surface
water resource, but there are some systems (the
larger, rare lakes and streams) and some regions
where off-grid site selection will be necessary.
50
-------
ACID-SENSITIVE AQUATIC AREiSXO? Tp/PETiEAST
t 3* U
i iiiinii niiimin
Figure 3-6. Acid sensitive regions of interest in the Northeast.
51
-------
Table 3-4. Estimates of Geographic Regions of Interest in the Northeast Relative to Acidic
Deposition and Number of Tier 1 Hexagons, Number of Lakes, and Number of Hexagons
with Lakes
Subreglon
Adirondacks
Southwest Adirondacks
Northeast Coastal Zone
High interest areas in
Vermont and New Hampshire
Mtd-Northeast
Number of
Hexagons
41
13
94
20
153
Number of
Lakes
131
38
204
73
509
Number of Hexagons
with Lakes
36
11
66
18
152
Table 3-5. Estimates of Tier 1 Coverage of Subpopulations of Interest for Acidic Deposition in
the Northeast
Subpopulation
Number of Tier 1 Grid Sites
Coastal New England
Low base cation, low DOC drainage lakes in the Adirondacks
Low base cation, high DOC drainage lakes in the Adirondacks
Moderate base cation, low DOC drainage lakes in the Adirondacks
Moderate base cation, high DOC drainage lakes in the Adirondacks
Mounded, low DOC seepage lakes
Mounded, high DOC seepage lakes
Low base cation, low DOC drainage lakes in Vermont and New Hampshire
Low base cation, high DOC drainage lakes in Vermont and New Hampshire
Moderate base cation drainage lakes in Vermont and New Hampshire
66
8
5
2
3
2
2
3
3
38
52
-------
Figure 3-7. Areas of the Northeast requiring threefold augmentation of the grid to supply enough
sample sites for the project.
53
-------
-------
4.0 INDICATOR DEVELOPMENT AND EVALUATION
Section 4 addresses the issues of selecting,
evaluating, and using a set of measurements to
assess the ecological condition of surface waters.
This complex process is constrained by the goals
of EMAP-Surface Waters (to make regional and
national assessments of the ecological condition
of lakes and streams over decades) and by the
realities of actually performing field sampling at
these scales.
This section discusses a wide range of difficult
topics, including what organisms and chemical
and physical features to target, when and where,
within a waterbody, to sample, how to convert the
measurements (data) into meaningful assess-
ments of ecological condition and how to relate
the measurements to issues of concern to society
and science. There are no simple solutions for
these issues that will satisfy every need. The
material presented here is a beginning, with many
aspects needing further substantial effort.
This section presents, for surface waters, an
overview of EMAP indicators (4.1), criteria for
determining whether a waterbody is impaired
(4.2), ecological response indicators chosen for
surface waters (4.3), exposure/habitat indicators
(4.4), stressor indicators (4.5), and a strategy for
the continuing process of selecting, developing,
evaluating, and using indicators (4.6). The sub-
sections on response indicators for lakes (4.3.1)
and for streams (4.3.2) include synopses of the
rationale for their inclusion, and issues related to
sampling and making assessments from the data.
Before any field work is performed, operations
manuals (laboratory and field methods, quality
assurance, field implementation) will be prepared.
4.1 CONCEPTUAL FRAMEWORK FOR
INDICATORS OF CONDITION
Assessments of surface water condition must
relate to the values and problems of present or
potential concern to the public and to scientists.
These issues are referred to as endpoints of
concern, and are reflected in the ways lakes and
streams are managed. For EMAP-Surface Waters,
these values or endpoints are trophic state,
fishability, and biotic integrity, as described in
Section 1.5. Thus, our assessments of condition
will be made in terms of each of these endpoints,
rather than the single entity of surface water
condition.
Neither ecological health (as a single entity
analogous to human health) nor the status of the
endpoints of concern can be assessed directly
(Karr 1981, Rapport 1989, Schaeffer et al. 1988).
Health assessments are initiated by making a
series of measurements. Yet, the raw data from
these measurements, by themselves, are not suf-
ficient. These data must be evaluated within the
framework of ecosystem theory and known haz-
ards to surfaces waters (Section 1.5), and con-
verted into indicators of condition.
The selection of indicators and data collection
methods is constrained by the broad-scale per-
spective of EMAP. We will use an index sampling
approach, which is a sampling regime that targets
one or a very few sample times and locations
within each waterbody, limiting effort on individual
units to maximize regional coverage with available
resources. The index sampling concept and the
choices of when and where in a waterbody to
sample are explained in greater detail in Section
6. Index sampling measurements are intended to
be single "snapshots" of conditions to accompany
snapshots from other waterbodies within the pop-
ulation of interest. The interest is in the collection
of pictures of the population of lakes or stream
reaches, not the individual snapshot.
The snapshot characterizations derived from
index sampling have been criticized because the
data are inadequate for addressing ecological
processes, for describing seasonal or diel varia-
bility, for understanding individual lakes or
streams, and for detecting ecologically important
events such as episodic acidification, peak algal
densities, or sensitive life stages. It is certainly
true that an index sample is not adequate for
these purposes, but that is not the objective of
the program. Other investigators are addressing
these issues in research programs designed spe-
cifically for those purposes. Duplication of their
efforts with a new program would be counter-
productive.
In a program that uses multiple indicators, it is not
possible to choose an index time and location
optimal for all indicators and all possible stresses.
However, it is necessary that the sampling times,
55
-------
locations, and protocols be adequate for expres-
sing ecological conditions. For example, in lakes,
a summer epilimnetic sample may be enough to
characterize water chemistry and trophic condi-
tion, but would miss low pH associated with
snowmelt, or spring phytoplankton blooms in
lakes.
As discussed in Section 2.1.2, EMAP has four
types of indicators of ecological condition:
response, exposure, habitat, and stressor. It is
Important to understand that these categories
provide a conceptual framework, acting as a tool
to guide the process of selecting, evaluating, and
implementing the actual measurements used to
assess ecological condition. These categories are
not meant to be rigid boxes in which each meas-
urement type may serve only one function.
4.1.1 Response Indicators
Response indicators are derived from measure-
ments that describe the biotic condition of organ-
Isms, populations, communities, or other compo-
nents or processes of the aquatic ecosystem as
they relate to the endpoints of concern (fishability,
trophic condition, biotic integrity). Response
indicators are the focal points upon which the
health of surface water systems will be deter-
mined. The measurements should be amenable
to quantifying the integrated response of the eco-
logical resources to individual or multiple
stressors. Response Indicators should clearly
relate to aspects of the environment valued by the
public and the scientific community, such as over-
enriched or brown waters, biodiversity, ecosystem
sustainabiiity, and wilderness recreational value.
Response indicators may also be chosen to iden-
tify emerging problems. Exposure and habitat
indicators can also be used to define waterbody
types and to describe expected conditions.
Figure 4-1 illustrates the conceptual process of
condition assessment using response indicators.
The rectangles represent types of information.
The diamonds indicate processes of information
refinement. The raw data from a collection of
measurements must be evaluated within the eco-
logical context (waterbody type and size, season,
geographic location, etc.) that defines what is
expected for similar waterbodies. These evalua-
tions produce scores for metrics (standards of
measurement). The metric scores can be further
aggregated into indices. This process can range
from simply summing unweighted metric scores
to including metric scores in complex models.
Both metric scores and index scores; are used as
response indicators in the process of deriving an
overall statement of condition. The complexity of
information (amount of detail) is reduced at each
step of the process.
Figure 4-2 presents a simplified example of this
process using fish assemblages. The measure-
ments are the fish species and their abundances.
These data are evaluated in their ecological con-
text (e.g. warmwater streams in the Midwest,
draining watersheds of < 100 km2). Among the
metrics that could be scored from these data are
species richness and proportion of exotic species.
These metrics must be calibrated by the number
of species and the proportion of exotic species
expected in unimpaired warmwater streams of
that size in the Midwest. The individual metric
scores can be used as indicators of biotic integ-
rity and fishability in the overall assessment of
those endpoints. These scores can also be incor-
porated into an index, such as the Index of Biotic
Integrity, which in turn becomes part of the final
assessment process. Using both metric scores
and index scores preserves more oi: the detail of
the raw data.
4.1.2 Exposure and Habitat Indicators
Exposure and habitat indicators serve a diagnos-
tic function when measured in conjunction with
response indicators. They are used to identify the
likely causes of impaired conditions as detected
by the response indicators. The exposure and
habitat indicators are characteristics of the
aquatic environment that give evidence of the
occurrence and/or magnitude of a response indi-
cator's contact with a physical, chemical, or
biological impact.
Historically, we have evaluated primarily water
quality (i.e., measures of physical and chemical
characteristics of surface waters). These assess-
ments have relied largely on what we have called
exposure indicators.
EMAP-Surface Waters plans to use measures
such as sediment toxicity, chemical contaminants
in fish, and general water chemistry (chemical
habitat) as exposure indicators. Habitat indicators
are attributes that describe the physical, chemical,
and biological conditions of the environment.
They are extremely useful for classifying water
56
-------
Measurement
(Raw Data)
Indicators
Evaluation
& Calibration
Figure 4-1. The conceptual process for proceeding from measurements to indicators to assessment
of condition.
57
-------
Indicators
Fish Species
Counts & Weights
Expectations
or warm water stream
<100 km 2
Scores meet
Expectations for
Small Midwestern
Warm Water
(reams?.
Figure 4-2. The process of proceeding from measurements on fish assemblages to indicators such
as Index of Biotic Integrity (IBI) or Index of Well Being (IWB).
58
-------
body types and for providing natural limits on
system behavior. The chemical and physical
habitat indicators can be used to model condi-
tions that promote healthy ecological situations,
as well as indicate probable causes of poor
condition. These indicators are discussed for
lakes and streams in Section 4.4.
4.1.3 Stressor Indicators
Stressor indicators describe characteristics,
processes, or activities that give rise to exposure
(impact) within aquatic ecosystems. They can
be thought of as characterizing the sources of
exposure. Some examples include land use and
land cover, pesticide application rates within the
watershed, livestock management activities,
human population densities, export of pollutants,
flow and channel modification, presence of intro-
duced species, and stocking and harvest
records. Stressor indicators may include natural
processes, such as climate cycles! However, the
primary intent of Stressor indicators is to charac-
terize the activities over which humans have con-
trol and that can create deleterious exposures.
Stressor indicators proposed for EMAP-Surface
Waters are summarized in Section 4.5.
The primary role of exposure, habitat, and stres-
sor indicators within EMAP is to identify the
probable causes for impaired conditions. Thus,
the assessment process for these indicators dif-
fers from that envisioned for response indicators.
In addition, exposure and habitat indicators are
important in defining expected conditions, so
that nominal/subnominal conditions can be set.
Except for the final assessment of ecological
condition, many of the steps in Figure 4-1 may
be used to assess these diagnostic indicators.
After metric and index scores have been deter-
mined, further analyses will be required to relate
these indicators to the behavior of the response
indicators. Correlation of unacceptable condi-
tions in response indicators with various expo-
sure, habitat, and stressor indicators is the first
level of this analysis. Further diagnosis will
undoubtedly be required. This is an area with
many research opportunities. It is also worth
noting that this diagnosis stage will be most
fruitful after EMAP has successfully used the
response indicators to characterize the condition .
of lakes and streams.
Table 4-1 presents indicators currently being
considered for use in EMAP-Surface Waters.
Table 4-1. Candidate Indicators for Inland
Surface Waters
Response
Trophic state index
Sedimentary diatom assemblage
Zooplankton assemblage
Macroinvertebrate assemblage
Fish assemblage
Gross external anomalies
Semiaquatic wildlife assemblage (birds)
Exposure/Habitat
Physical habitat quality
Chemical habitat quality
Sediment toxicity
Chemical contaminants in fish
Chemical contaminants in sediments
Biomarkers
Stressor
Land use and land cover
Human and livestock population density
Atmospheric emission and deposition
Use of chemicals .
Pollutant loadings
Flow and channel modification
Lake level and shoreline modification
Introduced species
Stocking and harvesting records
4.2 ESTABLISHING UNIMPAIRED
(NOMINAL) CONDITION
The process of selecting and implementing all
four types of indicators cannot be separated
from the problem of how the data collected and
the resultant indicators will be used to make
statements about lake and stream conditions.
The indicators will show that some of the water-
bodies are impaired (subnominal, unhealthy).
The task is to establish the criteria that define
impaired systems, that is, to determine that
below a certain index or metric score, a lake or
stream is considered to be in unacceptable con-
dition, relative to a particular endpoint of
concern. Some of the approaches for establish-
ing these criteria include selecting and assessing
reference sites, using historical data and/or
pristine sites, using ecological models, and
basing the criteria on the empirical distribution of
indicator values.
59
-------
EMAP-Surface Waters will interpret ecological
condition based on a combination of these
approaches, but will rely heavily on a series of
regional reference sites (Hughes et al. 1986,
Hughes 1989) and knowledge of historic condi-
tions as described through paleoiimnological
reconstructions, expert opinions, and previous
biological collections. This approach will allow us
to Integrate professional judgement, an under-
standing of historical or pristine conditions, and
knowledge of current ecosystem research to sel-
ect the least disturbed but typical sites of a
region. The response indicator values, and the
statistical variability found in the reference sites
can be used to develop the expected or accep-
table conditions for the region. These sites can
also be monitored to help detect natural spatial
and temporal variation. We expect that the water-
bodies selected for Tier 2 sampling (Section 3.2.2)
will represent the overall range of conditions for
each region and that very few of the best sites will
be selected. Purposely selecting additional
unimpaired sites will ensure a robust model of
unimpaired conditions. It will also be useful to
examine the range of stressors impinging on the
Tier 2 waterbodies, and to purposely select addi-
tional systems to ensure that the model of
severely Impaired systems is well defined.
Another option for determining whether a system
Is Impaired Is to use one or more ecological
models. Models may be based on historical data
from field collections for a key variable or set of
variables, on complex ecosystem studies, or on
laboratory toxicity tests. Empirical models that
Identify relationships, such as maximum fish
species richness as a function of lake area or
stream size, are an effective way to establish
criteria. Habitat classification models should be
useful for assessing exposure and habitat indica-
tors. However, key exposure and habitat vari-
ables differ considerably across the nation. For
example, stream ecosystem research stresses
physical habitat in the Pacific Northwest, flow
regime and habitat in the Intermountain West, and
sediment In the Midwest. The fact that models
tend to be stressor or species specific and/or
applicable to limited geographic areas will con-
strain their usefulness at first. However, as EMAP
progresses, models should become increasingly
Important for refining criteria.
Selecting an Impaired/unimpaired 'boundary
based on the empirical distribution of indicator
data from Tier 2 sampling sites has been sug-
gested from time to time. For example, the
median value, or the 25th percentile, or some
obvious breakpoint in the distribution of values
could be selected as a criterion. There are
problems with using this type of purely empirical
approach. First, distributions of most indicators
are continuous (without clear breakpoints).
Second, given the extent and frequency of diffuse
pollution, there is a high probability that even the
highest quality grid sites may be at least partially
impaired.
North America's biomes and ecoregions vary con-
siderably, and they offer only a coarse classifi-
cation of waterbodies. All indicators do not follow
the same regional boundaries, because indicators
are controlled by different limiting factors. At a
minimum, though, we should expect different indi-
cator scores in different regions because of differ-
ences in habitats and species. The combination
of reference sites, regional expert opinion, and
historical data offers a means of evaluating
indicator scores against regional potentials or
standards. These then can be combined into
national scores.
We do not expect to answer any of the questions
quickly. No matter which methods are used to
estimate the impaired or unimpaired condition,
there will still be considerable natural variability
across a region and among lake types and sizes.
Determination of the boundaries will therefore
have to involve some degree of judgment. The
question thus becomes, "With what level of
confidence do we want to avoid making a type 1
or type 2 error and which is of greater concern?"
In many respects, the definition of acceptable
condition is a social not a scientific one. Our
intent in EMAP is not to defend any specific value
as the one and only value but rather to lead the
discussion on how to objectively establish this
criterion. EMAP will produce data that can pro-
vide the foundation for the assessment and take
part in the technical discussion on establishing
these boundaries. We believe that we also have
some obligation to make an initial decision as to
where the boundaries might be. At that point,
however, many competing social, economic,
scientific, and political issues must be considered.
4.3 RESPONSE INDICATOR SELECTION
The process of selecting indicators has been
complex, governed by EMAP's emphasis on
ecological response and on broad-scale assess-
60
-------
merits. It has been guided by consultations With
other professionals in aquatic ecology and by
examination of available databases and the scien-
tific and management literature. Much of the
effort has been directed to response indicators,
which address the first of the EMAP objectives.
The discussions here are not intended to be a
review of existing literature but rather a summary
of our plans to use various taxa and indicators.
There is a general feeling that our ability to
measure and interpret biological data lags behind
our ability to measure and interpret chemical data.
In some respects this is true. However, our ability
to interpret the meaning of chemical data in terms
of its impact on the biota is somewhat limited. If
the biota are the concern, it is prudent to take
direct measurements, even if our ability to
interpret them is in its early stages.
Some indicators are considered "core," in that
established sampling and evaluation methodol-
ogies exist. Others are developmental indicators,
depending on the amount of research or testing
of sampling and/or evaluation methods that is
needed.
EMAP-Surface Waters has several general criteria
for evaluating the usefulness of response indi-
cators. Indicators should be biological and
should incorporate elements of ecosystem struc-
ture and function. The selected indicators should
correlate with changes in other unmonitored bio-
logical components. They must have clear con-
nections with the three endpoints of concern, and
must be responsive to a broad array of potential
stressors. An ideal indicator is applicable in a
broad range of surface water types across the
nation. The response indicators must be sensitive
to varying levels of stress, but not produce
excessive false alarms. This includes sensitivity to
important episodes that occur outside the samp-
ling period. Table 4-2 presents the indicators
being considered for EMAP-Surface Waters and
indicators that have been considered but not
selected.
Additional criteria apply to all types of indicators.
Useful indicators are cost effective, providing
considerable information in a limited amount of
sampling time. They should be easily imple-
mented and provide reproducible results with low
sampling variability.
4.3.1 Lake Indicators
This subsection describes the response indicators
chosen for lakes, and discusses some of the vari-
ety of concerns that affect their implementation in
the Tier 2 sampling (Section 3.2.2). Iri the first
year in each region, part of the pilot program will
be dedicated to addressing these concerns.
The proposed response indicators for lakes,
discussed in the following paragraphs, are:
• Trophic state '..'.-... . ,
• Fish assemblages
• Macrobenthos assemblages
• Zooplankton assemblages
• Sedimentary diatom assemblages
• Semiaquatic wildlife assemblages
Trophic state indices directly address the trophic
state issues of water clarity, noxious algal.blooms,
and anaerobic conditions in lakes. It will not be
possible to completely characterize the trophic
state of many lakes with single mid-summer
samples. That kind of information requires
sampling during spring or fall turnover, as well as
during peak algal blooms. However, for EMAP's
purposes, an adequate characterization can be
achieved with a single mid-lake epilimnetic
summer sample.
Trophic state will be assessed from Secchi disk
transparency measures and concentrations of
chlorophyll-a, total phosphorus, and total nitro-
gen. These data will be converted to one or more
indices, such as Carlson's Trophic State Index
(Carlson 1977, Shapiro 1979, Walker 1984, and
Brezqnik 1984). Information on total suspended
solids, dissolved organic carbon (DOC), and
macrophytes will be used to evaluate norialgal
turbidity and production and eventually to modify
current indices.
Fish assemblages will be used to address two of
the endpoints of concern: biotic integrity and
fishability. The overall assemblage structure
(species, abundances, and weight, length, and
general condition of individuals) will be used to
evaluate biotic integrity; the sports fish portion of
the assemblage will be used to address fishability.
In most regions and lake types, fish assemblages
are relatively stable over the summer index
period.
61
-------
I
S
E
(3
Q
E
II
E
O3
JE
ii
X
•c
0
•I
O
o
g
•5
W
%
J
1
1
O
t
3
CO
•o
I
ro
H
42
(A
Q
If
'•a
o
•O
tS
*?
0
Q
w
•s
c
1
UJ
Table 4-2.
1
S
g
~OJ
0 J
O LU
^0)
"S w
*CI CO
•§ -2
isS
g
£. UJ
-1
t»s
CL
O
oc<
°1
^2 rn
JO ffl
O ID
Indicator
XX X 2 XXX -i -J X
_ _ ^ 2 2 — ' — ' 2 — '
.
xx x 2 x -i x -i x 2
X2 2 2 x x x xx 2
22 2 2 x x 2 -" -> x
xx x xx -i 2 2 x x
2 x 2 xx x -i -i -" x •
2x x x x x x xx x
xxxxxx2xx2
2x 2 2x x -J -i -• x
X22X2-JX-1-JX
"B-JXX-XX-I-J-J-JX
"S
C Q)
M 73 — ^ « •*-*
S *— ^ -*- — *?% *S co
82 1^= la fi Is 1 I'i «i 1
1 Ii i }i ii ii fill H ii it
;
1
'^
o
W £
S 1
05 ."D
§, .. 0
''" rQ
; O "to
•o ' S-S-
S o —
ci CL . rt §
0 to c _
*3 » 0) ,-Q -C
"g" | * I § o 8
"g t. .c1 "o "c w "
" w ~ ~n E -° S
^ o o ^ g' ro "«
iilifll
l-Siilll
lllsili
JG (T ^2 X u3'ac J£2
H- Ol ^ - ._ •— . ^ —
E o • „
Si "* .C
If 1 • -2
o ]ro o
c t
1 • I-. S
| i; I
S >, o
^ s-. s
c to
2 ' i' ^
0) 0
£ € o
O -g> _
15 « -s oi -i
0) .52 e— O
.-£
{/) m ;i CO r-
« . ig 0 -E §
8 .2 « § g E
O *2 nj c ••-• o
& 5^1 S '•§.
•- S « g ra i fc
W S -n S -C >, C
a) .2 ii ^* t- w c
O) o) s; to o v^_ nj
c S» = m ^ E
5 2 § 3 -a
O O O) W 0} >
m'clQca)§!S
0 8 J2 1 IT ffl 0 .E
(0 .d O T3
-------
|
IK
n
.2
"S
C
1?
M
E
n
£
O)
'!
£
n
•c
i
O
§
1
1
ff\
_
<
5
u
*
«
i
1
g
V)
*
JS
e
i
M
o
1S
1
0)
•§
iW
S
^2 '
c.
3
to
S5*
e a?
9e Q)
Table 4-2. Evaluatic
(Continu
a
2
- (/
a
fH
L_
"o
^C
C
i
-1
<§u=
"75 c
g 0
•n EQ
£ C8
.« 0
X Q
&
3
.g
— IJJ
.a
c
o
II
O> Q.
Q) O
EC <
M
•n
Correlate
Indicator Unmeasurei
2-J-i
X — I — J — i _ I i_j«j i i
25-i -J _i_i_j _i_i_i
x 2 x x xxx x x-i
2X2 -" 2X55 5
5xx 2 xx-i -i-J -J
X -I X -J -JX-l -i -I -1
_i_j_i i _i_i_i -j — i —i
5222 x 5 -" ->->->
_,_•_, _, _,_,_. -j-j
2X2 -" _.2x xx x
1 I
°~ 1
o5 c c a
Is! Il
"O g* O -2 P
sif^ |i
t5 f1 5 § -01 •*-
•' 2 ™ >. 'E -£ '£
™ a> "m 3 C T3
.0 f c g o S i
§• i_ C O ^ W
° ;3 T3 | ^ I
^ ° I £ £ S ~w
o
I s I
3 C -
L_ tQ f?
S 2 *-
"o ^» -
I illl
g .<« ^ c 5 §
O J2 eo c *- o
Q. 5 -JS 5 '-5
"• -*-1 S T3 -^ Q.
a Correlates with changes in
components.
b Is applicable on a regional
c Integrates temporal and sp
Relates unambiguously am
a relevant exposure or hab
e Can be quantified by syno
in a cost-effective manner.
63
-------
Sampling will generally be done with sonar and
overnight sets of gill nets in deep waters, and with
traps and seining and/or electrofishing in shallow
areas. All these methods have possible restric-
tions. Seines are not effective in weedy, rocky, or
snag-filled areas. Gill nets may not be allowed in
some lakes due to fish mortality. Sonar data do
not indicate species, traps are difficult to set on
dropoffs, and electrofishing requires a boat and
trailer. There is a need for research to assess
problems of variability and effectiveness with
these methods in the various regions and lake
types and with various fish assemblages. How-
ever, we lack sufficient funds to use ail gear in all
lakes sampled in this summer's (1991) pilot. We
will determine the minimum gear requirements for
use In all lakes and test the other types of gear in
a subset of lakes.
Currently, there are no indices of biotic integrity
for fish assemblages in lakes comparable to
Karr's IBI (Karr et al. 1986), which was developed
for warmwater streams, or for fishability (fishing
quality) in lakes, although the importance of fish
in lakes is well documented (Northcote 1988,
Tonn 1990). Such indices are research tasks for
EMAP-Surface Waters. Biotic integrity will be
assessed by metrics such as species richness,
percent sensitive species, percent exotic species,
age/size structure, rates of external anomalies,
and rates and types of species replacement. All
these metrics must be evaluated within the con-
text offish assemblage type (Tonn and Magnuson
1982).
Fishing quality is an endpoint of concern for
lakes. Fishability will be assessed by metrics
such as presence, relative abundance, and age/
size structure of game fish populations, addres-
sing the questions: Are there any large game
fish? Are populations reproducing? The pres-
ence of external anomalies, as well as consump-
tion criteria violations, address concerns about
perceptions of edibility and the health hazards of
eating fish. We also plan to use a questionnaire,
such as that in Plafkin et al. (1989), to assess
fishing quality.
Benthic macroinvertebrate assemblage struc-
ture will be used to evaluate biotic integrity.
Macrolnvertebrates play a significant role in the
food web, between primary producers and fish,
and are important indicators of biotic integrity in
their own right. Macroinvertebrates are a par-
ticularly critical class of organisms to monitor in
those systems that naturally have very few or rio
fish. This group has been used historically as a
biomonitoring tool. Sampling will probably be
done with a combination of several methods,
such as sweep and kick nets, handpicking from
natural substrates, Ekman or Ponar dredges, and
sediment corers. Research is needed on collec-
tion method variability to develop protocols to
assure consistent, repeatable sampling.
Currently, there are no indices of biotic integrity
for macroinvertebrate assemblages in lakes. Con-
siderable research is needed to develop these
indices, to modify indices that are in use for
streams, and to determine the sensitivity of
species to various stressors. Candidate indices
and metrics include: taxa richness, number and
proportion of Ephemeroptera and Trichoptera
taxa, a series of ratios of certain taxonomic
groups (e.g., Chironomidae, scrapers, shredders)
to the total, and assemblage level indices such as
the Biotic Index (Hilsenhof 1988).
Zooplankton assemblages will be evaluated to
assess biotic integrity primarily, and fish food
abundance to a lesser degree. Ziooplankton per-
form the same role as macrbinvertebrates, but in
the water column, and they are relatively easy to
sample. Our major concern with zooplankton is
their species richness ancl what the proportional
abundance of various taxa can tell us about a
lake's quality. Zooplankton are sensitive to fish
predation and fish introductions, as well as to
trophic state and dominant phytoplankton, and
they are the dominant animals in most lakes in
terms of numbers and biomass.
Because we are interested mostly in rotifers,
cladocerans, and copepods, we \vill take an index
sample by vertically towing a fine mesh Wisconsin
net at the deepest point of the lake. Multiple tows
will be taken at a subset of lakes, and some lakes
will be sampled each month during the pilot to
assess spatial and temporal variability.
Data interpretation will center on species richness,
relative abundance of generally tolerant or intol-
erant taxa, and feeding guilds. Zooplankton are
typically not very speciose, with 4-15 species per
tow in most small lakes (Lind 1974, Ruttner 1963).
Sedimentary diatom assemblage data will be
used to evaluate the extent and time course of
environmental and biological change (biotic
integrity). The sensitivity of individual species to
64
-------
changes in water chemistry can be matched with
fossils preserved in deep sediments to track
temporal changes in such factors as pH, metals,
and nutrient levels. Samples collected from the
top of the lake sediments integrate the diatom
assemblages over the full year. The entire sedi-
ment core preserves the lake's history.
Sample collection, preparation, and analysis
methods are fairly well developed for lakes. A
single core of bottom sediments provides suffici-
ent material to develop diatom profiles for the
entire lake. We will examine diatom assemblages
from the top of the core and from 50-100 cm
deep in the core. Recent developments in statis-
tical analysis of diatom data make it possible to
calibrate top assemblages with current condition,
and then to reconstruct historic conditions from
deep assemblages. The use of diatom analyses
in paleoreconstructions has been growing quickly.
Diatoms have been used effectively to evaluate
trends in acidification, eutrophication, and general
pollution conditions (Agebeti and Dickman 1989,
Charles et al. 1990, Lowe 1974, Stoermer et al.
1985).
Semiaquatic wildlife assemblages may be used
to further assess biotic integrity. This indicator
includes a variety of birds, mammals, amphibians,
and reptiles that spend a substantial portion of
their lives in or near lakes. Many of these species
are important recreationally (for hunting or as
"watchable" wildlife), and as a group linking lakes
with their watersheds. This indicator is probably
the least developed of those being evaluated, but
it has considerable potential for future work.
An example of this kind of indicator is the U.S.
Fish and Wildlife Service's (USFWS) Breeding Bird
Survey (Bobbins et al.,.1986). The methodology
could be modified for lakeside and/or canoe
observations. These data could be assessed
using a modification of the Brooks et al. (in press)
response guild system. However, the optimal
index period for birds is in the spring, which does
not coincide with the mid-summer sampling
period currently planned. Considerable research
is needed to develop and assess protocols for
surveying this component of lake ecosystems.
We propose to conduct this research at a subset
of 20-30 lakes this summer (1991).
4.3.2 Stream Indicators
This subsection describes the response indicators
chosen for streams and discusses some of the
variety of concerns that affect their implementa-
tion in the Tier 2 sampling. Many questions and
topics discussed here are the same as those for
lakes, but there are considerable differences in
the details. In the first year in each region, the
pilot program will be dedicated to addressing
these concerns.
Fish assemblages will be used to address biotic
integrity and fishability. As in lakes, fish assem-
blages are relatively stable during the late summer
index period and interannually (Matthews 1986,
1990, Ross et al. 1985, Karr et al. 1986). How-
ever, Grossman et al. (1990) found high inter-
annual coefficients of variation among stream fish
species, and Matthews (1990) found high spatial
variation in dominant stream fishes. The overall
assemblage structure (species, abundances, and
weight, length, and condition of individuals) will
be used to evaluate biotic integrity; the sports fish
portion of the assemblage will be used to address
fishability. In most regions, streams will be
sampled by electrofishing and/or seining. Elec-
trofishing is not effective in streams with very low
conductivity, in very turbid or colored systems,
and with some species. Seines are not effective
in weedy, rocky, or snag-filled areas. There is a
need for research to assess the variability and
effectiveness of these methods in the various
regions and stream types.
The Index of Biotic Integrity (IBI) for stream fish
assemblages has proven useful in the Midwest
(Karr et al. 1986, Steedman 1988), and has been
modified for use in other regions (Miller et al.
1988, Oberdorff and Hughes, in press). A
research task for EMAP is to modify and evaluate
the IBI in other regions where the expected
species diversity is great enough to use the IBI
framework. In regions and systems that naturally
have few fish species, biotic integrity will be
assessed by metrics such as percent sensitive
species, percent exotic species, age/size struc-
ture, rates of external anomalies, and rates and
types of species replacement.
As it is for lakes, fishing quality is an endpoint of
concern in some kinds of streams (e.g., trout
streams). Until a "Fishing Quality Index" can be
developed, fishability will be assessed by metrics
such as game fish species present, relative abun-
65
-------
dance of game fish, age/size structure of game
fish populations (Are there any large game fish?
Are populations reproducing?), presence of
external anomalies (Do the fish look edible?), and
consumption criteria violations (Are the fish
actually safe to eat?).
Benthic macroinvertebrate assemblage struc-
ture will be used to evaluate biotic integrity.
Macroinvertebrates play a significant role in the
food web, between primary producers and fish,
and are important indicators of bfotic integrity.
The usefulness of macroinvertebrates increases in
streams that naturally have very few or no fish.
This group has been used extensively as a bio-
monitoring tool, to assess the effects of point
sources of pollution as well as nonpoint sources
(Plafkin et al. 1989, Fiske 1988, Metcalfe 1989).
Sampling will probably be done with a combina-
tion of sweep and kick nets, and handpicking
from natural substrates. Research is needed on
collection method variability to develop protocols
to assure consistent, repeatable sampling.
There are a number of indices of biotic integrity
for macroinvertebrate assemblages in streams
(Lenat 1988). However, considerable research is
needed to further evaluate these indices and to
determine species sensitivity to various stressors
in many regions. Candidate indices and metrics
include: taxa richness, number and proportion of
Ephemeroptera/Plecoptera/Trichoptera (EPT)
taxa, a series of ratios of certain taxonomic
groups (e.g., EPT, Chironomidae) to the total,
foraging guilds (percent shredders, grazers,
collectors) as suggested by Cummins and Klug
(1979), and assemblage level indices such as the
Biotic Index (Hilsenhof 1987).
Semiaquatic wildlife assemblages may also be
able to augment biotic integrity assessments for
streams. For example, Brooks et al. (in press)
found bird guilds to be more sensitive indicators
than fish or benthos of landscape deterioration in
Pennsylvania streams. In the long term, it may be
possible to expand the Breeding Bird Survey to
include transects that coincide with EM AP-Surface
Water sampling units.
4.4 EXPOSURE AND HABITAT INDICATORS
As described in Sections 2.1 and 4.1.2, exposure
and habitat indicators are intended to serve a
diagnostic function when measured in conjunction
with response indicators. That is, when response
indicators demonstrate impaired ecological condi-
tions, exposure and habitat indicators will be used
to identify the probable (proximal, immediate)
causes of that condition. In addition, exposure
and habitat indicators are important in defining
expected conditions, so that nominal/subnominal
conditions can be set. Most water quality moni-
toring now relies on exposure indicator measure-
ments to define and detect impairment.
As with response indicators, the broad scale of
EMAP, along with its objectives, directed the
selection of exposure and habitat indicators. For
EMAP, it is not necessary (or practical) to com-
pletely characterize the water chemistry or sub-
strate of each waterbody. Instead, the choice of
methods (and targeted data) is based on the
physical and chemical attributes thait most directly
and extensively affect the response indicators
(and the endpoints of concern). For example,
pesticides are less likely to occur in detectable
amounts in mountain lakes than in agricultural
streams. Thus, pesticides may be targeted in the
latter but not the former. These indicators (and
data) also need to be amenable to the index
sampling concept. That is, they should provide
robust characterizations of chemical and physical
conditions without a need for targeting short
periods of maximum concentrations (greatest
stress conditions).
This section summarizes the five exposure and
habitat indicators for both lake and stream
systems. Conceptually, there is little difference
between lakes and streams relative to these indi-
cators, but there are some obvious differences in
the details of designing, implementing, and ana-
lyzing exposure and habitat indicators in each.
Physical habitat quality, or the size, shape, and
hydrology of the stream channel or lake basin,
characterizes the physical conditions that provide
the setting for different natural biological and
chemical conditions. Alterations are likely to
impact the ecological condition. Thus, physical
habitat information is needed to fully interpret the
response indicator data (Gorman and Karr 1978,
Platts et al. 1983). For example, if the physical
habitat score is high but the biological scores are
low, then a chemical impact is probably causing
the low response indicator score. On the other
hand, if the physical habitat is severely degraded,
the biological component will be impaired no
matter how good the chemical habitat is (Plafkin
66
-------
et al. 1989). Research is needed to define these
relationships.
In general, physical habitat has four components:
waterbody size and persistence, and habitat type
and complexity (Table 4-3). Size may be a com-
bination of surface area, maximum and average
depth, and volume of water. For streams, volume
is a function of cross section area and gradient.
Persistence is a measure of the stability of the
vplume, of water (e.g., lake level fluctuations,
frequency and regularity of stream floods or
droughts). Habitat type refers to a combination of
water depth and velocity and substrate type,
Some habitat types for lakes are profundal (deep),
littoral (shallow, along the shore), and vegetated
or not. Some stream habitat types are riffles,
pools, and backwaters. Substrate type refers to
either the bottom material (sand, mud, cobble) or
to objects in the water on which organisms live
(vegetation, logs, stumps, tree roots). Habitat
complexity refers to the number and proportion of
the different kinds of water depth and velocity,
habitat, and substrate types occurring in the
waterbody.
Some of this information will be obtained from
maps and existing data (e.g., waterbody size).
The remainder will be collected during Tier 2
sampling. Generally, these data will be collected
by recording observations made along one or
more transects of the lake or stream. A research
task for this indicator is to develop a set of
measures that adequately characterizes each
waterbody, and that provides consistent, quan-
titative, repeatable data.
Table 4-3. Suggested Physical Habitat Measures for Lakes
Lake Size and Persistence Index
• Lake surface area (1:24,000-scale maps)
• Lake depth at subjective deepest point (field survey)
• Lake lever fluctuation
- Annual percent change in maximum depth (field shoreline survey)
- Annual percent change in lake area (field shoreline survey)
- Level control structures
• Lake hydrologic residence time (Tr = [est. volume]/[runoff x topographic watershed area])
Lake Habitat Complexity Index
• Littoral dominance
- Percent littoral surface area (field sonar survey)
- Temperature and dissolved oxygen stratification
• Bottom habitat complexity
- Bathymetric maps
- Coefficient of variation in lake depth along two cross-lake transects (field sonar survey)
- Profundal substrate
- Littoral substrate
- Macrophytes
• Lake shoreline complexify
- Shoreline development (1:24,000-scale maps)
- Shoreline development (field estimate)
- % artificial shoreline (several types)
- Shoreline vegetation
67
-------
Chemical habitat quality is the suite of variables
that indicates (1) the general chemical climate
surrounding aquatic organisms and (2) possible
water column stresses on the biota (Hynes 1974,
Warren 1974). Much variability in aquatic species
composition and abundance not explained by
physical habitat is likely to be associated with
natural and anthropogenic differences in water
quality. Levels of major ions, pH, temperature,
and dissolved oxygen often control the chemical
behavior and biotic stresses caused by other
chemical constituents (e.g., toxins, nutrients,
metals). Quantitative description of the water
quality is a necessary part of the diagnosis of
proximal causes of impairment, and facilitates an
evaluation of the linkages between human activi-
ties and biotic responses.
EMAP-Surface Waters will examine four general
attributes of water quality: geochemical type
(Ionic strength), acid-base status (sources of
acidification), nutrient status, and redox status
(possible toxic conditions) (Table 4-4). There is
some overlap among the chemical species col-
lected to assess these categories. The geochem-
ical type variables (e.g. Na, K, Si, Mg, Ca, SO4,
NO.,, CI) indicate the association between water
quality and soil weathering processes, and some
anthropogenic disturbances. Acid-base status
includes measures of hydrogen ion activity and
buffering capacity. The former controls the
solubility and toxicity of many chemicals, the
latter describes the degree to which the water-
body resists pH changes. Nutrient status
addresses the supply of chemical compounds
that often limit the growth of algae and macro-
phytes. The redox status of waters (assessed by
dissolved oxygen, pH, and temperature) is a
major control on the solubility, mobility and
toxicity of many chemicals, including nitrogen and
toxic heavy metals. Additional clues to anthro-
pogenic disturbance (e.g., road salting, sewage
disposal, irrigation) may be assessed with other
combinations of chemical measurements.
For lakes, water temperature (profile), dissolved
oxygen (profile), conductance, pH, and Secchi
disk transparency will be measured mid-lake. Epi-
limnetlc (0.5 m) and near bottom (for lakes low in
dissolved oxygen) water samples will be collected
for laboratory analyses, using National Surface
Water Survey methods. For streams, grab sam-
ples will be taken in the approximate deepest
water.
Sediment toxicity tests (Nebekeret al. 1984) will
address possible biological exposure to toxic
materials that have accumulated in sediments,
which potentially affect biotic integrity and
fishability. Toxicity and toxic materials act as a
kind of endpoint of concern for the public, parti-
ally because of human health concerns (which are
not directly within EMAP's scope). EMAP is not
designed to pinpoint toxic hotspots, but it should
be able to characterize, on a regional basis,
whether and how much toxics are impacting
biotic integrity and fishability.
Toxicity may be demonstrated by any of several
adverse responses shown by test organisms (e.g.,
mortality, impaired growth, reduced reproduction),
when compared with controls. In addition, there
are several possible methodologic choices, such
as which test species to use, the type of exposure
(solid phase, pore water, elutriate), acute versus
chronic, and test duration. Other research ques-
tions relate to the variability of laboratory
operation, the representativeness of the collected
sediments, and how to define control sediments.
EMAP-Surface Waters will evaluate 10-day acute
solid-phase bioassays with Hyalella amphipods
and 7-day solid-phase chronic assays with
Ceriodaphnia and Pimephales larvae. Overlying,
clean water will be used in all tests. The
measurement endpoints for Ceriodaphnia tests
are mortality and reproduction, and for Pime-
phales larvae, mortality and growth. We will be
unable to conduct sediment toxicity tests at all
pilot sites; however, we plan to do them for a
subset of 40-50 sites.
Measurements of chemical contaminants in
fish (Schmitt and Brumbaugh 1990, Schmitt et al.
1990) will provide direct assessments of the expo-
sure to chemicals that fish have experienced, and
the potential health hazard for humans as well as
for other biota.
Numerous research questions still need to be
answered for this indicator. Several of these
questions involve the risk assessment process,
such as which chemicals to target, whether to use
whole fish or fillets, which species to use (game
species or not), and which health criteria to use
(there are large differences in criterion levels
among regulatory agencies). The answers to
some of the questions will influence other issues,
such as what reproductive state to target (some
68
-------
Table 4-4. Proposed Chemical Measurements Contributing to EMAP Geochemical Habitat Quality
Indices
Measurement
Temperature
Dissolved oxygen
Transparency (Secchi depth)
Conductance
TSS (unfiltered)
Turbidity
pH (field)
pH (closed-system)
pH (air-equilibrated)
ANC (Gran)
DIG
DOC
Color
Na
K, Mg, Ca
S04
Cl, NO3, Si
Total N (unfiltered)
NH4
Ttoal P (unfiltered)
Altm (closed headspace)
Aljm (closed headspace)
Total Al
Chlorophyll-a
Geochemical type
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Acid-base
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Nutrient
X
X
X
X
X
X
X
X
X
X
X
X
Disturbance Rec
X
X X
X
X
X
X
X
X
X
X
X
X
X
X
X
chemicals accumulate in the gonads), which
chemical analyses to use, and whether to archive
samples.
Throughout this activity, we will maintain close
coordination with the U.S. Fish and Wildlife
Service National Contaminant Biomonitoring
program. The proposed analytes are: Aldrin,
alpha-BHC, Chlordane, Dacthal, ODD, DDEE,
DDT, Dieldrin, Endrin, Heptachlor, Heptachlor
epoxide, Hexachlorobenzene, Lindane, Mirex,
Nonachlor, Oxychlordane, PCBs, Toxaphene,
arsenic, cadmium, copper, lead, mercury,
selenium, and zinc.
A set of biomarkers taken from fish blood, gill,
and liver tissues will be evaluated to diagnose
early signs of stress. We are particularly
concerned with screening for short-lived chem-
icals, such as carbonates, pyrethroids, and
parathion, that are difficult to detect by other
means. These indicators are also being examined
for their ability to give early warning signals of
ecosystem stress. A third reason for using bio-
markers is to provide an organism level indicator
to complement the ecosystem (trophic state) and
assemblage level response indicators.
69
-------
We propose to evaluate biomarkers at a subset of
lakes because of limited funds and because the
sensitivity of biomarkers in moderately impaired
and minimally impacted lakes remains question-
able. A second research objective is to compare
blornarker sensitivity and responses with those of
the more typically used assemblage, ecosystem,
and exposure indicators.
4.5 STRESSOR INDICATORS
As described in Sections 2.1 and 4.1.3, stressor
Indicators will be used, along with the exposure
and habitat indicators, to further diagnose the
probable causes of impaired conditions. Pre-
sumably, the stressor indicators will address the
human actions causing impairment ratherthan the
proximal causes. For example, pesticide applica-
tion is a source of stress that shows up as
chemical contaminants or toxic sediments (expo-
sure Indicators) and impaired biological assem-
blages (response indicators). This section
describes these indicators only from the broadest
view, because the highest priority for development
and implementation goes to the response indica-
tors first, then to the exposure and habitat
Indicators, and finally to the stressor indicators.
Few of the data needed to assess the stressor
indicators will be obtained from Tier 2 field
sampling. Instead, these data will be collected
from a variety of existing data sources, including
maps and management and regulatory agency
reports and databases. A major source of data
for the stressor indicators will come from the
landscape characterization of the 40-km2 hexa-
gons made for Tier 1. One issue for the surface
water component of EMAP is whether the hexa-
gon characterization is adequate, or whether the
complete watersheds for the Tier 2 waterbodies
should be characterized.
Land use and land cover characterizations will
be the most general stressor indicators. Land use
will describe the most prevalent types of human
Induced stress (Omernik 1977, Smart et al. 1985).
For example, agricultural land use tends to be
associated with channelization, increased nutrient
and sediment loading, increased pesticide levels,
and decreased width of riparian woodlands, lead-
ing to less stream shading, more erosion, and
less in-stream cover. Urban land use is associ-
ated with increased toxic materials, nutrient
loading, lake shoreline alteration, and channel
modifications. Areas with mining activities tend to
have increased heavy metals and decreased pH.
Land cover integrates human activity (land use)
and natural vegetation potentials, such as climate
and soil type. Remote sensing and available
maps are extremely useful sources of this type of
information.
Human and livestock population density will
describe the general level or intensity of humanr
induced activity. That is, greater numbers of
people generally mean more kinds of stress.
Livestock density is important in areas of low
human density, because a concentration of cattle,
hogs, or poultry in feedlots or,farms usually
means greater nutrient loading. Where livestock
graze, there is high potential for riparian
vegetation removal, bank trampling, erosion, and
nutrient loading (Platts 1979).
Pollutant loadings can be assessed by analyzing
current pollution discharge permits, both municiT
pal and industrial. These will further describe the
kinds and intensity of nutrient, chemical, and
thermal stresses. •?•
Flow and channel modifications as well as lake-
level and shoreline alterations can be assessed
from landscape characterizations, regulatory
agency records (e.g., Army Corps of Engineers),
and habitat evaluations during Tier 2 sampling.
Normally, channelization removes a great deal of
the habitat diversity from streams, as does
dredging in lakes. Flow modification often
stresses aquatic ecosystems ; by (disrupting the
natural patterns of flood and low flow, as well as
modifying temperature regimes (Karr et al. 1983,
Simpson et al. 1982).
. • ' ~ - X' ' ',
Stocking, harvesting and species introduction
records provide information about the biological.
stresses in aquatic ecosystems. Management
practices generally intended to enhance fishing
quality often degrade the biotic integrity of surface
waters. Stocking activities also mean that the
waterbody cannot maintain the desired harvest
levels. Introduced species, intentional or other-
wise, have considerable potential for decreasing
biotic integrity (Miller et al. 1989).
In examining all the indicators, we will first
interpret the raw data to assess status, detect
trends, and correlate probable causes. This
information will be useful, but some researchers
have found that guilds, indices, and proportionate
abundances are less variable and more interpret-
70
-------
able than presence/absence, absolute abun-
dance, population abundance or biomass, or
individual physical and chemical variables (Rahel
in press, Plafkin et al. 1989, Karr et al. 1986,
Land res 1983, Carlson 1977). Of course indices
can camouflage or distort natural phenomena
(Jackson et al. 1990, Boyle etal. 1990), especially
when they are not grounded in ecological theory
or concepts, when they are simple mathematical
manipulations, or when an index developed for
one region or use is applied without adjustment to
a much different region or use. Regardless of the
indices used, the raw database will be maintained
to facilitate multivariate statistical analysis, data
exchange, and later reinterpretations as our
ecological insights grow.
Tables 4-5 and 4-6 present potential linkages
between endpoints, measurements, and response
and exposure indicators, respectively, providing
an overview of the indicator approach for EMAP-
Surface Waters. We hope ultimately to link the
stressor indicators with the ultimate causes of
environmental degradation, resource overcon-
sumption, human overpopulation, and an incom-
pletely developed sense of environmental ethics
(Ehrlich et al. 1977, Nash 1989).
4.6 STRATEGY FOR INDICATOR DEVEL-
OPMENT AND IMPLEMENTATION
Given the spatial, temporal, and ecological enor-
mity and complexity of EMAP-Surface Waters, it
is essential to have a basic strategy that defines
a process, and a set of goals and criteria by
which to evaluate whether the steps have been
successfully completed. Each phase in the gen-
eral strategy can be expanded to include as much
detail as needed to assure achievement of its
goals. The strategy described here has been
modified from the overall EMAP Indicator Strategy
(Knapp et al., 1991).
The indicator development strategy does not pro-
pose a linear process (Figure 4-3). Instead, there
may be considerable branching off and feeding
back to previous steps. Some indicators may
progress relatively quickly through to full imple-
mentation; others may stall or progress slowly for
many years. There are six basic phases:
1. Identify environmental values and apparent
stressors.
2. Develop a set of candidate indicators that
are linked to endpoints of concern and
responsive to expected stressors.
3. Screen candidate indicators to select
research indicators with reasonably
well-established databases, methods, and
responsiveness.
4. Quantitatively evaluate expected regional
scale performance of research indicators to
identify developmental indicators for
regional demonstrations.
5. Demonstrate developmental indicators on
a regional scale, using the sampling frame,
methods, and data analyses intended for
full (core) implementation.
6. Implement core indicators with annual
sampling and data analyses and periodic
reevaluation of indicators.
The first two phases are meant to generate ideas
for endpoints and indicators. These two phases
encourage broad-scale, lateral thinking, with a
focus on breadth rather than depth of coverage,
and may be revisited at any time (Figure 4-3).
Essentially, EMAP-Surface Waters has completed
the first pass through phases 1 and 2, using a
series of workshops, informal literature reviews,
and exploratory data analyses.
Phases 3 to 5 are oriented toward critical evalu-
ation and integrative filtering of the candidate
indicators down to a defensible, practical set of
core indicators. Where phases 1 and 2 are inclu-
sive, these three phases focus on excluding indi-
cators that are currently not feasible within EMAP
and on documenting the value of those selected.
Each of these phases uses its own detailed
research plan, sets of evaluation criteria, and peer
reviews of the decisions made.
In practice, EMAP-Surface Waters will need to
develop individual phase 3 to 6 strategies for
lakes and streams on a regional basis, because
(1) the surface waters component is being imple-
mented in several regional steps (i.e., the first
year will be a pilot study of lakes in the Northeast,
with other regions and streams added sequentially
over five years), and (2) there are major differ-
ences among waterbody types, sampling
71
-------
CO
o
I
2 S.
saE^^s^o
jo|(0 CO CO ? i- £ CO o T- CO Y~ N-
""""AAAVVOACVAAA
Is* IO T~ I^« •>, f*»« IO
CDCM ACD"^ CO N.
o? iro^^co^io
T- T— T- CO AT-OOi™C\I
: *—s. -— s-i, <-» ••— — •—
. Q T— %«* O CT •*- /?s
V V A CC-3< V V
AAAVl— Ah--— • CEO
1OO+CO CO ,1O
CM T- T- CO to IO CO T- CJ
AA, AVr'-AVV. V
&
•g
Illllll
J= S51
8-.2
ge-S
•s
ents
CO O
•si
|Is
iljllii
2 o •§ 5 o •* •"
g-i
.W T3
(/f .
.is-
2 g
v>
c
o
S-i-
if!
Ill
*
"= IS
II
.2' co
CD CD
s y
E!l82§.E-g;
i? l^iEfif •
co
S'
CO
w
• -.
To o
-iflpl 1
-w
-
CD
S)
g"
.§
Jtlilllilill
i.fll.S.Ell-B^l^
i Jg co co
.
OQX
c
o
1
"g
x:
B
o
21
JS
cu
fish and
T3
CD
CO
CO
o
•D
c
CO
CO
CO
s
I
£
CO
p
•o
c
CO
croinv
ton
JS
Q.
o
s
CO
l -s
S "a
£•
s
g
CO
o
o
I
£>
ed from regi
Det
edi
Includ
72
-------
CD
C
I
§
o
O
.g)
"o
o
!!
2 2
fl
in
S
o
v>
•o
a>
1 o
s -5
»•=:
•o o
1
D.
•
-
a
o
1
c
o
O
<
o
XI
o
x:
CD
.0
.
Q.
O
u
S t;
s
'o
CO w
"g 2
11
(0 C
1
Q.
CD
O
O
CD
S
§
si
1*"
if §
I
£
co
c
o
•s
eg
2
CD
1
i
T3
ClT C
'S CO
Q. C
C CO
°J d.
J. CO
p
CO =6
Of'fr
0 «
8
iT-a- -5
8
I
3 ,
CO
•O
c:
co
Q.
CO
I
.s
•a
£•
5
g
il
j "ro '
:^
I
o
o
o"
o
W
S^i
^zi
O to" <»
s«3 »
-" t3)^
5
CO
•o
8-.
0
sedim
•o
•e
T 0>'7B "ft
0.5^ #5
5 g
E
TJ
i
.a
|
T3
O
i-
mortality
Si
a) -5; o)
Is^-s
•£8
•o .s
* I J ^
l||l
|| ||
lrf = S
^is-g
illl
o — •§ ^
.'7. CO CO CO
CO OJ - -
cu IS c c
IB 2 -S -S
—j CL CO CO
as £ ~?
ts-e.-^1
g1?!
> CD ^0
= ]?.£ 1
CO CD c -1
C CO -O
S xi c
"' *_* ^ ni
•g > .5 is
« 3^=5
5"5-l i
£ «o -£. c
a. <" af .s
CD g £ -
||l|
« 13 0-15
'* E ^ §
y <" oS c
iri-i
5-88*
•5
O
I
I
1
I
1
I
£
CD
XI
•s,
£
O)
I1 i
I
co
i
I
i
o
£ .2
i
J' I
=5
I I
•is
xi ^
x: xi
"5 •= "3;
•5 ?!S
5|1
8cS
•« = "?
1=3 i
o i- jc
S.Stg
i- c o
» g -S
ts si
73
-------
Phase 1
Phase 2
Phase 3
Phase 4
Phase 5
Phase 6
IDENTIFY
ISSUES/ASSESSMENT ENDPOINTS
Objectives .' Methods
Develop indicators
linked to endpoints
- Qualitative evaluation
Expert Knowledge
Literature Review
Conceptual Models
CANDIDATE INDICATORS
Expert Knowledge
Literature Review
Conceptual Models
Prioritize based
on criteria
- Qualitative/Quantitative
evaluation
• reject, suspend, or
proceed
RESEARCH INDICATORS
Evaluate expected
performance
• quantitative testing
and evaluation
Analysis of Existing Data
Simulations
Pilot Tests
Example Assessments
Conceptual Models
DEVELOPMENTAL INDICATORS
Evaluate actual
I performance on a
/ regional scale
I - build infrastructure
I - demonstrate utility
I - assess logistics
^ CORE INDICATORS
Regional Demonstration
Projects
Regional Statistical
Summaries
Evaluation •
Workshops
Criteria
Criteria
Peer Review
Criteria
Peer Review
Criteria at
Regional Scale
Peer Review
Agency Review ot
Summary
Implement regional
and
National monitoring
- periodic reevaluatlcn
EMAP Data Analysis Feedback from
Correlate Old Indicators with Peers and Agencies
Proposed Replacements Peer Review
Assess Promising
Candidate Indicators
Revisit Assessment Endpoints
Figure 4-3. The indicator development process, showing the objectives, methods, and evaluation
techniques used in each phase.
74
-------
methods for lakes and for streams, current level
of development of indicators in various regions,
and regional importance of the end points of
concern. An overview of a generalized plan for
phases 3 to 6 is presented in the following para-
graphs.
Phase 3 (select research indicators) relies on
literature reviews and expert knowledge from
professional scientists and water resource
managers who have experience with a region's
lakes or streams. Study of available species
distributions and water quality, sampling
methodologies, and interpretive techniques is
emphasized. The results of this phase should
be (1) a list of research indicators for the region
and waterbody type, along with reasons for their
selection, and (2) a proposed set of tasks for
the next phase. EMAP-Surface Waters has
completed the first pass through this phase for
lakes in the Northeast.
Phase 4 (select developmental indicators)
increases the level of detail of the information
collected for phase 3, concentrating on ques-
tions of spatial and temporal variability, data
interpretability, trend and association objectives,
and proposed field and laboratory methods.
During this phase we will acquire and analyze
existing databases, simulate the minimum
detectable^ trends (given the expected variability
of the data), and determine preferred index
periods and locations. A mock assessment
should be developed to evaluate the utility of
the complete suite of indicators. In some cases
there should be a pilot test of indicators,
especially in the first years of EMAP.
Phase 5 (identify core indicators) uses a
regional demonstration project to address
issues posed by the selection criteria and
identified by the earlier phases. This phase
should demonstrate whether the sampling and
analysis protocols are adequate for generating
regionally interpretable information.
In EMAP-Surface Waters, phases 4 and 5 are
not necessarily sequential. All indicators sel-
ected have been field tested to some degree
somewhere. Remaining issues include variabil-
ity and regional applicability (evaluated by
demonstration projects), as well as cost,
redundancy, and methodological questions best
evaluated in pilot studies. This summer (1991),
we propose to conduct a demonstration on
about 60 lakes and a pilot on a subset (20-30)
of these lakes.
Phase 6 is the full implementation of the core
indicators at regional, and eventually national,
scales. This phase must balance continuity of
methods, to maximize trend detection, with a
set of procedures for continually improving the
indicators. EMAP-Surface Waters must be open
to other sources of information, along with the
development and implementation of core indica-
tors. Historical data and information from local
experts should continually be sought and used
to help interpret impairment and change.
We must make an effort to include citizen moni-
toring of'more waters, or of some waters more
frequently. For example, Secchi depth and
benthic macroinvertebrate assemblages (U.S.
EPA 1990c, Cummins and Wilzbach 1985) offer
easily applied and reliable measures of condi-
tion. Most importantly, we must develop formal
agreements or informal mechanisms to incor-
porate data from other federal agencies (e.g.,
The USGS National Water Quality Assessment
Program, the USFWS Breeding Bird Survey, and
the National Contaminant Biomonitoring Pro-
gram) and from State museums and water qual-
ity and fish and wildlife agencies.
75
-------
-------
5.0 POPULATION ESTIMATION AND ANALYTICAL APPROACHES
Indicator measurements taken in the field and in
the laboratory will be used to index the status and
characteristics of each sample waterbody
(Section 4.0). Because EMAP sample lakes and
streams will be chosen as a probability sample
from explicitly defined populations of surface
waters, information from these sample sites can
be used to infer characteristics of the population
from which the sites were drawn. Each sample
waterbody will have a weighting factor inversely
proportional to its sample inclusion probability.
For example, if there were a 1 in 10 chance of
selecting a particular lake for field sampling, that
lake would represent 10 lakes in a statistical
sense. These sample site weightings will be used
to extrapolate from index conditions in the set of
sample sites to index conditions in the population
statistically represented by those sites.
This section describes the type of data analyses
planned and some of the approaches we will
employ for quantitatively estimating population
status, changes, and trends and evaluating
uncertainty in those estimates. Section 5.1
describes the basis for population description and
extrapolation, in which the main source of uncer-
tainty derives from the selection of sample sites to
represent population characteristics. Analytical
approaches for detecting population differences
and trends are outlined in the last part of Section
5.1. An additional source of uncertainty in
population status and trend analysis results from
variability in our indicator measurements. We
organize the variability of indicator measurements
into a conceptual model of variance (5.2.1) and
quantify the major components of variability in a
number of example indicators (5.2.2). We then
evaluate the effect of indicator variability on the
robustness of population status descriptions (5.3).
Finally, we provide examples of population status,
change, difference, and trend analyses (5.4), illus-
trating the effects of indicator variability and some
design and analytical options for accommodating
the degree of variability we may encounter in
EMAP-Surface Waters.
5.1 ESTIMATING POPULATION CHARAC-
TERISTICS
5.1.1 Population Extrapolation from EMAP
Sample Data
Data collected by the sampling effort will be
analyzed to estimate, with known confidence, the
current status, extent, changes, and trends in the
condition of the nation's surface waters. In con-
junction with data from other monitoring sources,
the data will be examined to uncover correlative
associations between indicators of stress and indi-
cators of condition. Status and extent will be
portrayed through the use of descriptive statistics
and visual displays of spatial pattern. Descriptors
of status will include the usual descriptive statis-
tics as well as estimates of the proportion of the
population in various categories. The kinds of
analytic output to be generated will be dictated by
the specific objectives of the various aspects of
EMAP, with some constraint imposed by the
design. Estimates of such proportions are facili-
tated by the cumulative distribution function
(CDF), and the CDF will be a prime focus of the
analyses.
Cumulative distribution functions, such as the one
shown for lake pH in Figure 5-1, represent the
collective variation of an attribute over a set of
units such as lakes. The CDF shows the esti-
mated proportion of lakes in the population
having a value of the variable (in this case pH)
less than any specified value. The confidence
bound is a 95% one-sided confidence bound on
the estimated number, scaled by the estimated
total number of lakes. Examination of curves
such as Figure 5-1, constructed from response
indicator survey data, will provide estimates of the
"number, area, or length of aquatic resources at or
below certain condition values for that indicator,
and will constitute some of the primary analytical
outputs of the EMAP program.
It is necessary that the properties of estimates
and analyses be adequate for the EMAP objec-
tives and that any constraint be acceptable in this
respect. The algorithms of estimation and anal-
ysis are specific for the design and oriented to the
nature of the needed output. Descriptive statistics
(for example, mean, median, standard deviation)
and the CDF will be generated using the Horvitz-
77
-------
cq
o
CD
o
u-
CM
CD
O
CD
Figure 5-1. Frequency distributions are plotted for any identifiable population. This plot is for the
variable pH in Region 1 of the Eastern Lake Survey; The confidence bound has been
suppressed above the figure border.
Thompson (HT) estimation formulae (Horvitz and
Thompson, 1952). These formulae permit estima-
tion from any probability-based sampling design,
and require only specification of the inclusion
probabilities. The inclusion probabilities are
quantifiable for any probability sample, and the
strict requirement that EMAP designs be proba-
bility designs has provided the assurance that
these formulae can be used.
Inclusion probabilities for EMAP sample units are
of two kinds, first and second order. First order
inclusion probabilities are simply the probabilities
with which the individual sampling units are
included in the sample. They must be known for
each selected unit. They are generated and
archived at the time of sample selection and will
be designated by the symbol n.{, referring to the
tth sampling unit. Second order, or pairwise,
Inclusion probabilities are the probabilities with
which two specific sampling units are included in
the sample. They are designated as n«, with
obvious extension of the notation, referring to the
probability of simultaneously including units i and
j. Certain design features are required to deter-
mine 7r,j, for example, stratum or cluster identifi-
cation, and sample size. That information will
also be retained as part of the individual data
record, so that storage and use of the data are
kept uncomplicated.
Estimation formulae are simplified by use of
weights rather than inclusion probabilities, where
w --
w,- _
In practice, it is also appropriate to store the
weights rather than the inclusion probabilities as
part of the data set. The HT formulae are then
expressed as:
=
' ieS '
V[f J = Zy^Wj-l) + Z
ieS • ieS jei
(1)
(2)
where y is any attribute, and Ty is the total of that
attribute over any specific identified population.
The summation is restricted to the specific set of
.units, S, in the sample or any subset of the sam-
ple defined by a specific population. Estimates
78
-------
over subpopulations are thus provided by sub-
setting the sample.
In some instances, interest will focus on totals; in
other cases, there will be more interest in means
or percentiles. An ^estimate of the mean is
obtained by dividing f by the N, number of units
in the population. If N is not known, an estimate
is provided by setting y=1 and applying (1). Esti-
mators of proportions are obtained using indicator
functions. For example, to estimate the propor-
tion of the population with attribute A, set Vj = 1,
if sample unit i has attribute A, and set y( = 0
otherwise, and apply (1). Thenfy/N provides the
desired estimator.
Basic population parameters will be estimated by
some form of (1), and so will be strict HT esti-
mators. The variance estimator (2) is unbiased if
all pairwise inclusion probabilities are positive and
known exactly. However^ in many cases, design;
features will require approximations in calculating
joint inclusion probabilities. The approximation
developed by Overton (1987) will be used in most
instances. An heuristic derivation of the joint
inclusion probability for units i and j can be given
as follows. The Tier 2 sample will be selected
using a method similar to Madow's (1949)
method, in which the sample units are given an
order (which may be random, but need not be).
Let
N
WT =
wTk = Zw,
A random number, r, between 0 and wT/n is
drawn, and unit k is selected if wT,k-1
-------
represented by points. Point representation |s
necessary in order to unambiguously count the
number of such units in the 40-km hexagons.
Given such a count, nri over the n grid points,
Nf « 16Znfl
estimates the total number of units in the defined
class, with variance again estimated by (2) with Wj
» 16 and w,j from (4).
Estimates representing resource inventory will
then be summarized in a table, similar to the one
presented In Table 5-1.
5.1.3 Tier 2 Estimates and Descriptions
The Tier 2 sample consists of a subset of surface
waters on which the suite of indicators will be
measured. Measurements of additional variables
may also be obtained from maps, remote
imagery, and other sources for use in higher
resolution classification of surface waters. Tier 2
estimates will be of the areal extent of classes
defined on the Tier 2 information, and of the
numbers of resource units in those populations.
A table similar to Table 5-1 can be generated for
Tier 2 estimates. In addition, the Tier 2 sample
lends itself to more complete population descrip-
tion In terms of the Indicator variables that have
been measured on the samples of population
units.
Estimating formulae for Tier 2 will follow the
general forms (1) and (2), but the inclusion
probabilities are somewhat more complex, repre-
senting the sampling process at both tiers. When
possible, selection at Tier 2 will be made in a
manner that cancels out the variable inclusion
probabilities that were generated at Tier 1, but
additional variable probability will be generated by
the use of stratification at Tier 2 selection. There
will be some restriction on the ability to remove
Tier 1 variable probability, and the general for-
mulae will be used throughout.
In addition to inventory estimates, a general char-;.
acterjzation of these populations will be by the
estimated CDF for the variable, y, of interest.
Indicator variables will play such a role. In this, it
is necessary to consider only a particular range of
the variable y as defining a subpopulation, and
estimate the number of lakes, for example, in the
population that are in this subpopulation. This is
repeated for all values of y in the sample, and the
estimated CDF plotted as in Figure 5-1. Such dis-
tributions are generated for a variety of y-vari-
ables, such as numbers (frequency distributions),
areas (areal distributions), lengths, or other
attributes. For example, it may be useful to,
generate the distribution of stream miles on the
variable, y, indicating the miles of stream that
have a specific attribute.
Table 5-1. Tier 11nventory Estimates and Estimated Variance of Estimates are Provided from the Tier
1 Sample for All Resources and Classes of Resources
Estimated area by class
Estimated number of units
Resource
A
B
C
This can be repeated for any geographic subdivision or other spatial partitioning of the United States, or for any population defined
In any manner.
80
-------
The page of descriptive statistics that was pre-
scribed (Overton 1985) for the National Lake
Survey (NLS), and also used for the National
Stream Survey (NSSj, will also serve EMAP
descriptions (see Figure 5-2). Quantiles are
interpolated from the fitted distribution function.
In the NLS, confidence bounds were defined for
the number of units in the class in question, and
presented in scaled form.
Extension to EMAP demands greater versatility,
and we recommend that the distributions be
generated both as numbers and as proportions,
and that the confidence bounds be generated
accordingly. Then a single page of description
might report only one kind of distribution, but in
both forms. To keep the two forms distinct, we
recommend that the scale of the distribution of
numbers be labeled as numbers, but that the
plots be of fixed dimension to facilitate compari-
son. Confidence bounds on the distribution of
numbers should continue to be one-sided, as in
the NLS (Overton 1985), and ascending and
descending analyses should be provided, depen-
ding on the variable. The ascending analyses
provide upper bounds on the numbers of
resource units having values of the variable below
a particular value, and descending analyses
provide upper bounds on the numbers of units
having values above a particular value. Con-
fidence bounds on proportions of numbers of
units will usually be appropriately provided by
exact binomial bounds, and these are to be two-
sided, with descending analyses unnecessary.
Binomial bounds can be used only when the vari-
able probabilities have been eliminated by the
selection process at Tier 2; otherwise, it may be
necessary to use ratio variances for these con-
fidence bounds. Combined strata will involve
means of binomials, and will also not be suitable
to exact binomial bounds.
Distributions of area or length will follow the direct
estimation methodology (Overton 1985) but pro-
portions of areas and lengths must be treated as
ratios, and so will require special treatment in
generating confidence bounds.
5.1.4 Analytical Approaches for Detecting
Population Differences, Changes and
Trends
An important feature of EMAP-Surface Waters will
be the description of differences among sub-
populations and changes in subpopulations over
time. For example, it may be important to illus-
trate or describe differences among lakes influ-
enced by different land use patterns or practices,
or to determine whether the condition of different
types of lakes differs (e.g., drainage versus seep-
age lakes, or lakes versus reservoirs). Also,
during the early years, it will be important to
determine whether there are detectable year-to-
year differences within subpopulations of lakes,
especially if changes in the magnitude of stres-
sors are occurring or are anticipated. As more
data are obtained, it will become feasible to
detect trends in a variety of subpopulations.
An assortment of techniques will be used in the
examination of EMAP data for change and trend.
An extensive range'of standard statistical tech-
niques is applicable. Linear model techniques,
such as analysis of variance and regression will
be used with various assumptions regarding
spatial and temporal variance components, statis-
tical independence, explanatory variables, nature
of trend, and nature of change. Nohparametric
alternatives, based on ranks or signs of differ-
ences, will also be used (e.g., Loftis et al. 1989).
Factors that influence the ability to detect differ-
ences, changes, and trends include:
• Inherent variation !
• Sample size
• Level of significance
• Level of power
• Size of the differences to be detected
• Temporal and spatial autocorrelation
Inherent variation refers to the natural variation
that occurs within and among systems, as well as
the variation introduced during the measurement
process, as described in Section 5.2. Sample
sizes of 50 to 100 are expected to be adequate
for most subpopulatioh descriptions. Relatively
little reduction in uncertainty occurs with
increased sample sizes above 50 to 100 (Linthurst
et al. 1986). The level of significance (a) refers to
the probability of correctly detecting a change
when a change is present. In most studies; a is
set at the 0.05 level, implying that if a change has
occurred, we would expect to detect it 95% of the
time. It may be appropriate to relax this con-
straint somewhat for EMAP purposes.
81
-------
VARIABLE: ANC AS UEQ/L FOR LAKESs.2000 HA
Xcl: 0.0 Xc2: 50.0 Xpl: 200.0
16 FEB 1986
Population Size (N): 1483 SE(N): 57.5
Lake Area (A): 72412 SE(A): 12013
Sample Size: 163
1.0
o.B-
o.e-
F(X)
o.*-
0.2-
0.0-
1.0'
-100.0 0.0 100.0 200.0 300.0 400-0 500.0 600.0 700.0 800,0 900.0 '000.0
0.6-
G(X)
0.*-
0.2-
0.0-
-100.0 0.0 100.0 200.0 300.0 *0
-------
An important consideration in the selection of.,
techniques for change and trend detection will be
the sensitivity of the tests, or, in statistical terms,
the power of the tests. Statistical power is the
probability that a test of an hypothesis will be
rejected given that the hypothesis is false. Gener-
ally, power increases with increasing departure
from the hypothesis, that is, a large change is
more likely to be detected than a small change.
A statistical test of a particular hypothesis can be
characterized by specifying the size of the test
(probability of rejecting a true hypothesis) and its
power function (probability of rejecting a false
hypothesis as a function of degree of departure
from the hypothesis). Where choices among
alternative methods of change and trend detec-
tion exist, the more powerful test will be preferred.
Power will also be used to evaluate the adequacy
of the design, and of the sample sizes in various
subpopulations. The evaluation will be in terms of
describing the probability of detecting a change
or trend of a given magnitude. If the magnitude
of change that is detectable with, say 80% proba-
bility, is unacceptably large for some subpopula-
tion, then the sample size for that subpopulation
may have to be increased. Provisions for doing
so are incorporated in the general EMAP design.
One routine method for determining detectable
differences among subpopulations is a compari-
son of mean values with a Z-statistic. A general
equation that can be used to estimate MDC
(minimum detectable change) is:
MDC = (Za/2+ZJ * s * [2(1-r)/n]1/2 * [1,t
where: Za/2 and Z^ are tabulated "Z" values for
various levels of significance (a) and
power (/3)
s = sample standard deviation
r = temporal autocorrelation
n = sample size
p = spatial autocorrelation
for a two-sided test for differences; if a one-sided
test is sufficient, Za can be used.
However, EMAP will also use techniques that do
not focus solely on change in mean value or
central tendency. Some changes might be of
great .interest but would not have large effects on
mean values, such as changes in variance. For
example, one sign that an ecosystem is stressed
is an increase in variability. Since the CDF will be
the primary tool for summarizing and presenting
population descriptions; analyses of change and
trend will utilize the CDF, or some property
thereof. An example is the examination of the
quantiles for differences. The test is carried out
by using the CDF of the pooled subpopulations to
define classification criteria (e.g., quintiles of the
combined population) and doing a chi-squared
.test of homogeneity. Figure 5-3 illustrates this
approach. Such a test is sensitive to changes in
location as well as changes in scale or shape of
the distribution. The test is applied in the con-
ventional manner if both samples have approxi-
mately uniform inclusionprobability. Extension is
needed if there is appreciable variation in inclu-
sion probability in either sample.
Comparison of subpopulations also provides a
general technique for multivariate comparison. A
population can be classified on several variables
or attributes, and the CDF of the classes com-
pared by the chi-squared test prescribed above.
In the case of repeat observations of the same
population units, a more powerful test can be
used. Again, a procedure based on a chi-squared
test and sensitive to changes in location, scale,
and shape can be used. The test was used by
Overton (1989) to examine repeat measurements
of lake chemistry for change. The first step in the
, procedure is a sign test of the differences of the
two observations. Then the combined distribution
is" used to determine median and quintile values.
,Chi-squared tests of counts from each sample in
the class intervals defined by combined popula-
. tion percentiles are used to test for changes in
scale and shape. An important aspect of EMAP's
interpenetrating design is that it will achieve its full
potential power to detect change only after repeat
visits to all sites, that is, after two complete
cycles.
Repeat visits to a site permit a paired analysis that
essentially eliminates the population variation
component.' Thus, ability to detect change will
increase greatly in years 5 through 8 of the samp-
ling. Similarly, the power to detect a persistent
trend will continue to increase as more years of
data become available.
83
-------
OLD ANC CLUSTER II
FALL 1986 VS. FALL 1984
THE TOTAL NUMBER OF POINTS PLOTTED; 53
NUMBER OF POINTS ON DIAGONAL LINE: 0
CHI-SQUARE. 0.01886793
1 OF CHI-SQUARE CRfTICAL VALUE . 3.84
27
ADDmONAL POINTS WfTH SUM EQUAL TO MEDIAN: 1
CHI-SQUARE FOR INDEPENDENCE. 4.923037
1 DF CHI-SQUARE CRmCAL VALUE . 3.84
ADDITIONAL POINTS WITH SUM EQUAL TO A QUARTILE: 0
CHI-SQUARE FOR INDEPENDENCE . 10.1S38S
3 DF CHI-SQUARE CHrTCAL VALUE . 7.81473
80 100 120 140 160
Figure 5-3. An example of comparing population distributions using the chi-square test. ANC values
are for samples taken in fall of 1984 and 1986 from lakes during Eastern Lake Survey -
Phase I and Phase II. The expectation is for points to fall on the 1:1 line iff no difference
exists. Specific portions of the distributions can be evaluated by dividing the analysis by
the median, quartiles, or other percentiles.
84
-------
The development of techniques for trend detec-
tion for regional subpdpulations is at a relatively
primitive state and much will be done over the
next several years. Loftis et al. (1989) evaluated
several techniques for trend detection in individual
waterbodies. They recommend two nonpara-
metric tests to use with seasonal sampling: an
analysis of covariance on ranks and a seasonal
Kendall test. They briefly addressed extending
these tests to detecting trends in regional mean
values, but that issue remains unresolved.
Trend detection methods that are specific to the
interpenetrating design structure of EMAP are
being developed. In particular, methods that
account for temporal correlation and its effect on
the power of tests of trend are needed.
5.2 COMPONENTS OF VARIANCE IN
REGIONAL POPULATION SAMPLING
5.2.1 Variance Model
For any characteristic of interest to be sampled
from a surface water population, uncertainty in
population estimates may result from a number of
sources, ranging from sample selection to analyt-
ical error. A useful approach that provides the
information necessary for optimizing effort at
various levels of sampling is to consider the
magnitudes of variation in the environmental char-
acteristic over a range of spatial and temporal
scales. In doing this, it is important to distinguish
between the uncertainty in estimating population
statistics (e.g., CDFs, medians, quartiles) and the
actual magnitude of variation in the environmental
characteristic itself.
For example, consider an indicator variable such
as fish species richness in a population of lakes
within the Northeast region. We are interested in
describing the regional variation and trends in the
true species richness within these lakes. If we
had perfect knowledge of the true species rich-
ness in each lake within the Northeast, and
species richness did not vary within the lakes, we
could describe with virtual certainty (by a census)
the distribution of species richness values in the
regional population of lakes. If, rather than taking
a census of the lakes, we look only at a subset (a
"sample") of the total number of lakes, there is
uncertainty in our estimates of the true population
statistics introduced by the sampling process.
Section 5.1 describes the procedures for quantify-
ing the uncertainty in estimating statistical
parameters of the true population distributions
(CDF's, medians, quartiles, etc.) that results from
selecting a sample of lakes, rather than taking a
census of the population. Those confidence
bounds do not explicitly incorporate uncertainty in
measuring the indicator values in each individual
sample lake.
Unlike the idealized situation; in the previous
paragraph, species richness in each of our
sample lakes might actually vary over annual,
seasonal, and short-term time scales. We can
circumvent this problem somewhat by targeting
an index time and position in each lake, but even
these index values of species richness may vary
within this temporal and spatial window and may
vary cyclically from year to year. To complicate
the matter, our measurements of species richness
may vary even if true species richness in each
lake remains invariant. Our measurements are
not perfect because of variations in fish sampling
efficiency, gear selectivity, variations in the
location and behavior of fish at different times,
and uncertainties in taxonomic identification.
We may describe variation in an index measure-
ment (e.g., fish species richness or Secchi disk
transparency measured at a defined window of
space and time) as being derived from a number.
of sources:
• True differences in index values among
surface water sample units (lakes or stream
segments) across a region: e.g., fish
species richness differences in lakes or
streams across the Northeast.
• Among-year temporal variation in each
surface waterbody in the absence of
trends: e.g., normal fluctuations in fish
species richness between summer index
periods of different years occurring within
single lakes or stream segments.
• Index sampling variation within each
waterbody, including:
- a within-season temporal component:
e.g., variation in fish species richness
within a late summer index time period,
and •...-•-.•:
, - a spatial component: e.g., spatial varia-
tion in species richness at a given time
but in different potential index locations
85
-------
within a mapped stream segment or
lake identified as a sample unit. A
subtle but important distinction is that
this is not necessarily the spatial
variation of the index value as
measured in all possible locations in the
lake or stream segment, but only those
locations that might be sampled within
the index definition. For some
variables, this might include the whole
lake, for others only positions within the
profundal zone, for example.
• Measurement variation (error): Here we
are defining measurement error as the
variation in the estimation of an indicator
(e.g., species richness) that might -be
observed if the measurement were repli-
cated at a single time and place (the
"single place" may be much larger than a
single point). Measurement error derives
from variation in sample collection tech-
nique, handling, analysis, or species iden-
tification.
Application of the Index concept eliminates the
need to include variability at times and places
outside of the Index sampling window or location.
Consequently, seasonal variation outside the
index time period and spatial variation outside the
defined index location are not included in our
variance modelling. Those variances, however,
have great bearing on the selection of the indi-
cator measurements and the chosen definition of
the index time and location. The spatial and
temporal variation not included in the model is
important In subsequently limiting the interpre-
tation of population statistics describing index
conditions In a region. A clear example of this is
found In the NSS (see subsection 6.1.2.1), in
which the index sampling window specifically
excluded episodic variability in stream chemistry
associated with storm flows. Interpretation of the
biological relevance of population distributions of
index chemistry benefitted by subsequent consid-
eration of the episodic characteristics of streams
as they relate to measured index chemistry
(Kaufmann et al. 1988, 1989, Eshleman 1988,
Baker etal. 1990).
Based on the components described, a general-
ized hierarchical model of variance in index values
within a population of surface waters may be
proposed. For an Indicator variable, y:
y =
•••+ eunit . -•. ;.
e annual
+ e index-temporal
+ €
s index-spatial,
measurement
where each of the component deviations (errors)
are normally distributed with expected means of
zero and respective variances of o2unit, o2^,
-------
Table 5-2. Coefficients of Variation (CV) of Surface Water Indicators - A Number of Examples from
Lake and Reservoir Studies
Variable3
Secchi
Secchi
Secchi
Chi a
Chi a
Chi a
TP
TP
TP
TP
S04
SOBC
ANC
pH (fall)
DI-pH (core)
DI-pH (sur-
ficial core)
Study6
Vermont
Minnesota
ELS-H
Vermont
Minnesota
Knowlton
Vermont
Minnesota
Knowlton
ELS-II
ELS-II
ELS-ll
ELS-II
ELS-II
PIRLA-II
PIRLA-II
cvm
9.2%
na
na
30%
na
12%
13%
na
10%
2:3°
3.3°
14°
0.10°
0.04°
0.04°
CVfcrt
12%
29%
16%
26%
49%
.- 45%
23%
41%
27%
21%
9.0%
3.3%
12%
' 0.06
time-
integ.d
time-
integ.d
cvin.sp
na
na
ince
na
na
na
na
na
na
inc
ince
ince
ince
ince
0.10
0.21
cvy,
12%
17%
27%
44%
20%
26%
25%
13%
12%
19%,
26%
0.13
na
na
. cvunit
43%
66%
47% ...
72%
- 87%
76%
62%
81%
103%
.• '259%' ..--
.42%
.: 72% ..-:,•
144%^
0.76
0.76f
0.76f
Mean
4.9 m
, 1.7m
3.7m
4.3 /ig/L
17.5 pg/L
33.0 fjg/L
24.8 fjg/L
62 jug/L
63-fjg/L
8.7,/ig/L
124/zeq/L
374/zeq/L
61 /^eq/L
.5.9
5.9f
5.9f
Secohi = Secchi disk depth, Chi a = chlorophyll-a, TP = total phosphorus,'S04, = sulfate, SOBG = sum of base cations, DI-pH
" diatom-inferred pH; for pH variables, we report SD rather than CV; 100 SD/grand mean; for pH variables, we report SD rather
thanCV.
Vermont lakes [n=17 (TP); n=63 (Chi); n=63 (Secchi)] data from Eric Smeltzer (pers. comm.); Minnesota lakes [n=69 (TP);
n=57 (Chi); n=69 (Secchi)] data from Steve Heiskary (pers. comm.); ELS-II data (n = 145) from Herlihy et al. (1991); Knowlton
data are as reported by Knowlton et al. (1984) for eastern lakes and reservoirs (n = 186); PIRLA-II data for ELS-II Adirondack lakes
from Dixit et al. (1990). ' , . , .
0 These measurement CVs were independently calculated and were not included in the nested analysis of variance.
Diatom cores integrate conditions over the whole year, effectively eliminating within-year temporal variability.
8 Spatial variance component CV\,.S for mid-lake sampling is included within the temporal component CV^.j .*
f
For diatom-inferred pH, we have calculated these means on the basis of the Adirondack lakes population from ELS-II.
87
-------
Table 5-3. Coefficients of Variation (CV)a of Surface Water Indicators - A Number of Examples from
Stream and River Studies
Variable6
IBI-stream
IBI-river
Sp.-rlch
stream
Sp.-rlch
river
Turb.
T-dP
N03
SO4
Cond.
ANC
pH
a 100 SD/grand
Study0
Ohio
Ohio
Ohio
Ohio
NSS
NSS
NSS
NSS
NSS
NSS
NSS
cvm
na
na
na
na
9.7%d
4.2%d
3.5%d
1.0%d
0.56%d
1.2%d
0.035d
cvin.
12%
15
16%
19%
41%
54%
13%
3.5%
3.5%
1.5%
0.43
mean; for pH variables, we report SD rather than
b IB! = Index of Biotic Integrity (Karr et al.
CVin-sP
na
na
na
na
46%
89%
14%
4.8%
6.4%
2.3% '
0.69
CV.
CVVr
10%
13%
17%
24% .
na
na
na
na
na
na
na
1986) in wadable streams and rivers sampled by boat;
cvunit
37%
28%
50%
27%
127*
232
196"
119%
86%
95%
0.93
Sp.-rich
Mean
32 units
30 units
14 species
15 species
3.8 NTU
59/jg/L
66 i*eq/L
282 /ieq/L
115/jS/cm
830jueq/L
6.6 pH units
= total number of fish
species in wadable stream and large .river sampled sites; Turb. = turbidity; T-dP = total dissolved phosphorus; NOj and SQ,
are filtered samples; Cond. = specific conductance; ANC = acid neutralizing capacity by Gran titration; pH = pH determined
In closed-headspace sample.
c Ohio = Ohio Department of Environment stream and river data provided by Chris Yoder and Ed Rankin (pers. comm.); NSS
- National Stream Survey data on 500 Mid-Atlantic and Southeast streams [Kaufmann et al. 1988 On press)].
These measurement CVs were independently calculated from NSS measurement variances reported by Cougan et al. (1988)
and population means In column 8-a procedure not within nested analysis of overall variance.
88
-------
Table 5-4. Components of Indicator Variability as a Percentage of Among-Sample Unit Variance
(S unit) - Examples from Lake and Reservoir Studies
Variable3
Secchi
Secchi
Secchi
Ln-Chl a
Ln-Chla
Log10-Chl a
Ln-TP
Ln-TP
Log10-TP
TP
S04
SOBC
ANC
pH (fall)
DI-pH (core)
DI-pH (sur-
ficial core)
Studyb
Vermont
Minnesota
ELS-II
Vermont
Minnesota
Knowlton
Vermont
Minnesota
Knowlton
ELS-II
ELS-II
ELS-II
ELS-II
ELS-II
PIRLA-II
PIRLA-II
S nVS2unit
4.6%
na
na
17%
na
0.14%
4.4%
na
0.058%
na
0.30°
0.22°
0.9,1°
1.7°
0.3°
0.3°
8 in-t/S2unit
7.2%
19%
12%
13%
32%
1.8%
14%
26%
0.42%
0.67%
4.5%
0.21
0.73
5.3
time-integ.
time-integ.d
S in-s/s2unit
na
na
ince
na
na
na
na
na
na
ince
ince
ince
ince
ince
1.7
7.5
"VM-i
7.5%
7%
na
14
25
0.37
18
9.5%
0.098%
na
8.2%
7.2%
3.4
13
na
na
SDunit
(2.1 m)
(1.1 m)
(1.8 m)
-25 vg/L
-15 fjg/L
-50 fjg/L
-65 pg/L
(23 fjg/L)
(53 /^eq/L)
(268 /ieq/L)
(87 peq/L)
(0.77 pH unit)
(0.77 pH unit)f
(0.77 pH unit)f
Secohi = Secchi disk depth, Chi a = chlorophyll-a, TP = total phosphorus, SO4 = sulfate, SOBC = sum of base cations, DI-pH
= diatom-inferred pH.
Vermont lakes [n=17 (TP); n=63 (Chi); n=63 (Secchi)] data from Eric Smeltzer (pers. comm.); Minnesota lakes [n=69 (TP);
n=57(Chl); n=69 (Secchi)] data from Steve Heiskary (pers. comm.); ELS-II data (n = 145) from Herlihyetal. (in press); Knowlton
data are as reported by Knowlton et al. (1984) for eastern lakes and reservoirs (n = 186); PIRLA-II data for ELS-II Adirondack lakes
from Dixit et al. (1990). "
These measurement variances were independently calculated and were not included in the nested analysis of variance.
Diatom cores integrate conditions over the whole year, effectively eliminating within-year temporal variability.
Spatial variance component &ln_s for mid-lake sampling is included within the temporal component &-m_t.
For diatom-inferred pH, we have calculated these SD's on the basis of the Adirondack lakes population from ELS-II.
89
-------
Table 5-5. Components of Indicator Variability3 as a Percentage of Among-Sample Unit Variance (S2)
— Examples from Stream and River Studies
Var!ableb
IBI-stream
IBI-river
Sp.-rieh
stream
Sp.-rich
river
Turbidity
T-dP
N03
S04
Conductance
ANC
PH
a 100 SD/grand
Study0
Ohio
Ohio
Ohio
Ohio
NSS
NSS
NSS
NSS
NSS
NSS
NSS
mean; for pH variables,
S m/s2unit
na
na
na
na
0.59d
0.03d
0.03d
0.0075d
0.0042d
0.01 7d
0.001 4d
we report SD
8 in-t/S2unit
11
29
11
50
11
5.4
0.44
0.086
0.16
0.025
0.26
rather than CV.
IB) = Index of Biotic Intearity (Karr et al. 1986) In wadable streams and
S in-s/S2unit
na
na
na
na
13
15
0.52
0.16
0.55
0.060
0.54
rivers sampled by
S vr/S2unit
7.4
20
12
75
na
na
na
na
na
na
na
boat; Sp.-rich =
SDUnit
(12 unit)
(8.3 unit)
(7.1 species)
(4.1 species)
(4.87 NTU)
(138 M9/L)
(130 /ieq/L)
(337 Aieq/L)
(99 fiS/cm)
(790 jueq/L)
(0.93 unit)
total number of fish
species In wadable stream and large river sampled sites; Turb. = turbidity; T-dP = total dissolved phosphorus; NC^ and SO4
are filtered samples; Cond. = specific conductance; ANC = acid neutralizing capacity by Gran titration; pH = pH determined
In closed-headspace sample.
Ohio » Ohio Department of Environment stream and river data provided by Chris Yoder and Ed Rankin (pers. comm.); NSS
* National Stream Survey data on 500 Mid-Atlantic and Southeast streams [Kaufmann et al. 1988 (in press)].
These measurement variance ratios were calculated using independently derived NSS measurement variances reported by
Cougan et al. (1988) divided by #unlt-a procedure not within the nested analysis of overall variance.
90
-------
Some of the data are from probability sampling ,
[e.g., National Surface Water Survey (NSWS)],
others are not. The purpose of these tables is to
provide a somewhat representative range of indi-
cator precision arid regional heterogeneity upon
which we can base preliminary decisions regard-
ing the necessary temporal and spatial frequency
of sampling efforts. In order to express variance
Information on comparable grounds, we provide
the information first in terms of coefficients of
variation (100 X SD/grand mean) for indicator
data in lakes (Table 5-2) and streams (Table 5-3).
Using coefficients of variation, the magnitudes of
the variance components can be compared with
the general level of the indicator values in the
sample or population of surface waters. The
second pair of tables (5-4 and 5-5) describes the
same information, but in terms of relative
variances (the variance components are
expressed as percentages of the among-sample
unit variance). The among-sample unit variance,
S unjt, not shown in the tables, would have a,
va|ue of 100% in every row. Rather than pro-
viding S2unjt, we provide its square root, SDunjt,
expressed in the appropriate indicator measure-
ment units, to facilitate visualizing the magnitudes
of variance. The component variances were cal-
culated using a nested analysis of variance, so
the component variances add up to the total
population variance (i.e., sample unit variance
does not include that amount due to index tem-
poral and spatial differences).
5.2.2.2 Coefficients of Variation
Coefficients of variation (CV) among lakes and
streams in a region tend to vary from 28% to
232% of the mean, with biological variables
tending to be less variable than chemical vari-
ables (Tables 5-2 and 5-3). This finding may
simply reflect the generally smaller size (and
presumably more homogeneous characteristics)
of regions examined in the biological studies. It
may also reflect the natural constraints on species
pools, or the intentional constraints placed on the
range of indices such as IBI.
With one exception, CV's for among-year variabil-
ity range between 10% and 26%. CV's for spatial
variation within the index location were available
only for NSS and ELS-II data. For NSS data,
Cvin-sp varied from 2% to 5% for the higher pre-
cision chemical variables at the bottom of the
table, and from 14% to 46% for "noisier" variables
such as nitrate, phosphorus, and turbidity. Spatial
variability of mid-lake profundai samples in the
ELS-II is included in CVjn_t, which is a measure of
temporal variability within the index time period
and ranges from 3.3% for the sum of base cations
(SOBC) to 21% for total phosphorus. Quick
examination of the phosphorus mean shows that
levels were extremely low in the ELS-II lakes--
elevating the CV's for that variable. Within-index
period temporal CV's were mostly between 2%
and 10% for the precise chemical variables and
from 12% to 20% for most of the biological vari-
ables. The highest CV's for index temporal varia-
bility were chlorophyll-a and variables such as
phosphorus and stream turbidity, which are bio-
logically active or responsive to stream flow.
Measurement CV's were below well below 3% for
all the precise group of chemical variables, with
the exception of ANC in ELS-II. The CV for this
variable is inflated by the fact that ANC can take
negative values, and also that it does not vary
widely within the ELS-II region. .Coefficients of
measurement variation for other chemical vari-
ables are mostly under 13%. Measurement CV's
for chlorophyll-a ranged from 12% to 30%, and
probably account for a moderate fraction of the
high value (49%) observed within season in the
Minnesota data, which in our data set did not
enable calculation of measurement variability.
5.2.2.3 Relative Variances
Variance components expressed as a percentage
of among-sample unit variance in a region (Tables
5-4 and 5-5) are useful because they allow cal-
culation of the degree of distortion and uncer-
tainty that these components cause in the
descriptions of population means and distribu-
tions (see Section 5.3). To sum the percent
relative variances across rows in the relative
variance tables, set S2unit (the variance across
sample units in the region) at 100% and add it to
the other variance components. The total vari-
ance is then always a percentage greater than
100% that expresses the combined amount of
index measurement variance in excess of true dif-
ferences among sample lakes and streams across
the region. For example, a variance sum of 120%
means that index measurement variances equal to
20% of S2unit contribute to the total observed
variance observed in the regional sample.
Among-year variability is less than 10% of
among-sample unit variance for most indicators
other than phosphorus, chlorophyll-a, and the
91
-------
stream biological measures; with the exception of
small stream IBI (column 6 of Tables 5-4 and 5-5).
The river fish measures (20% to 75%) and Minne-
sota chlorophyll (25%) have the greatest between-
year variance (20% to 75%) among the studies we
examined. It seems likely that a large proportion
of what Is Interpreted here as biological variability
may simply result from differences in the location
of mobile fish populations and the capture effici-
ency of sampling gear. Alternately, the high vari-
ance of river fish measurement components rela-
tive to among-river differences might also reflect
a narrow range of meaningful differences in "con-
dition" across the regional set of large Ohio rivers
sampled.
Spatial variance within the profundal zone of
northeastern lakes sampled by the ELS-II was
Included as part of the within-fall index period
variance, and for all the variables presented in
Table 5-4, those combined variances were less
than about 5% of the variation among lakes.
Diatom-Inferred pH from surficial cores was more
variable over the lake bottom (7.5%) than was pH
for deeper cores (1.7%). Among the example
stream studies, Information on the spatial com-
ponent of variance in streams was available only
for physico-chemical data from the National
Stream Survey (Table 5-5). These upstream/
downstream differences were surprisingly small,
amounting to less than 1% of the among-stream
variance except for phosphorus (15%) and tur-
bidity (13%). It is likely that the within-sample unit
spatial component is a large part of the variance
for biological measurements (e.g., fish, macro-
Invertebrates) in streams, rivers, and lakes). Strat-
ification or targeting of specific physical habitat
types will probably be necessary to reduce the
effect of large spatial variability. This problem is
discussed in Sections 6.3 and 6.4 and, with physi-
cal habitat characterization, is a major area of
productive pilot research that is likely to result in
major gains in the precision of biological indices.
Temporal variability of most chemical variables
within the index sampling period is less than 5%
of among-lake or stream variability across the
regions examined in Tables 5-4 and 5-5. The
chemical constituents that are biologically active
and sensitive to stream discharge, such as
phosphorus and turbidity, appear in most cases
to be more variable (5% to 26%). Chlorophyll
variability differs substantially among regions, a
result that may reflect our failure to consistently
Isolate observations from an appropriate index
sampling period in our analysis of some of the
data sets. Secchi disk transparency and small
stream fish measurements (species richness and
IBI) have moderately low within-index temporal
variability in some regions (7% to 111%). Variabil-
ity of large river fish species richness and: IBI
(50% and 29%) was much higher than that for
smaller streams, and probably reflects spatial
habitat sampling variations, capture efficiency,
and fish population mobility rather than true
variation in condition of the fish community within
the summer index time period.
With one exception, the measurement component
of variability is less than 2% of the among-unit
variance for all of the chemical variables shown in
' Tables 5-4 and 5-5. In most cases it is extremely
small, less than 0.1%. The exception is total
phosphorus in Vermont lakes (4.4%), which was
somewhat higher in relative terms because the
concentrations involved were quite low (popula-
tion mean 25 fig/L) and the among-lake variability
was also low (SD = 15 A«g/L). Measurement vari-
ability for Secchi disk transparency (field duplicate
measurements) in Vermont lakes (4.6%) was
approximately the same as that for total phos-
phorus. The measurement variability of chloro-
phyll analyses in Vermont lakes (17%) was higher
than that reported by Knowlton et al. (1984)
(Table 5-5), probably reflecting the relatively low
concentrations and among-lake variation in the
Vermont lakes.
5.3 DESIGN CONSIDERATIONS FOR
ESTIMATING POPULATION! STATUS
5.3.1 Effects of Index Variability on Estimates
of Population Status
5.3.1.1 Theoretical Basis
Section 5.1 describes the algorithms for extrapo-
lating population means and their standard errors
from index values at surface water sample sites.
These extrapolation procedures do not separately
account for variance in the index values within
individual surface water sampling units. Rather,
they estimate the total sampling variance in the
population. If large imprecision exists in the
estimation of the index values for each sample
unit, we may be misled into believing that there is
a great deal of regional heterogeneity among
surface waters, when in fact a large portion of the
population variance is due to index measurement
imprecision rather than true differences among
92
-------
sample sites across a region. In this context, we
consider index measurement error to be in the
broadest sense-to include all components of
variability besides that due to the regional
variation among sample units themselves. Define
for this discussion:
Total population sampling variance then becomes:
°tot = "unit + °index'
and the variance of the population mean is:
o2.
tot =
unit
mean
"unit
'unit
where nunit is the number of sample units and
n is the number of temporal (or spatial) index
measurement replicates. It is clear from this
expression that if determining the population
mean is the sole objective, greater precision can
always be gained by increasing the number of
sites, rather than the number of replicates, no
matter what the magnitude of index variability.
Cost differences come into play to determine the
most economical allocation of sampling effort
between sample units and replications (see Sec-
tion 5.3.3). Furthermore, without some replica-
tion, even if it is not systematic (i.e., across all
sample sites), we will know what the population
variance is, but we will not know whether it is due
to the imprecision of index determinations on
each site or to true differences among sites within
a region.
In many cases in EMAP, the population mean val-
ues of indicators will not be of greatest interest in
describing the status of a resource. EMAP will
often rely more heavily on the shape of the
regional population distributions and other stat-
istics, such as percentiles and estimates of the
percentage of surface waters with indicator values
above or below nominal or subnominal values.
Overton (1989) assessed the effects of index
measurement variation (o2^^) on estimated dis-
tribution functions describing status, as provided
by EPA's National Surface Water Survey (Linthurst
et al. 1986, Landers et al. 1987, Kaufmann et al.
1988, in press). As does the EMAP design, the
surveys employed probability sampling from well-
defined populations of waterbodies, selected from
well-defined and explicit frames. Rigorous
inference is permitted by these probability
designs. Ordinarily, in sample surveys, the effect
of index measurement variation that has an
expectation of zero is not critical, as such error
contributes no bias to the usual estimates of
population means or totals, and the contribution
of such error to the variance of such estimates is
ordinarily low. However, the distributions of the
biological or chemical variables in surveys such
as the NSWS and EMAP can be biased by sub-
stantial variance in the measured indicator values
for each sample waterbody.
In the context of sample surveys, cumulative dis-
tribution functions (CDF), such as the one shown
for lake pH in Figure 5-1, represent the collective
variation of an attribute over a set of units such as
lakes. Examination of such CDF curves, con-
structed from response indicator survey data, will
provide estimates of the number, area, or length
of aquatic resources at or below certain condition
values for that indicator, and will constitute some
of the primary analytical outputs of the EMAP pro-
gram. Index measurement variance increases the
collective variation in the data, leading to bias in
estimates of the number, area, or length of water-
bodies having values of response indicators
above or below specified criteria values.
The case of a normal population with normally
distributed index measurement variation" is well
known and serves to identify the nature of the
bias problem and establish an intuitive base for
assessing its magnitude. The following discussion
and development within this section borrows
heavily from Overton (1989).
"*-'!.', • ,• -
The curves in Figure 5-4 illustrate the general
pattern of distortion of the true population
distribution (CDF Curve E), as successively
greater magnitudes of index measurement vari-
ance are added (convoluted) to the distribution
curves D, C, B, and A. The convolution of the
normal population distribution with index meas-
urement error distributions of progressively
greater variance leads to a family of normal
distributions with successively greater total
variance (Overton, 1989).
Overton (1989) examined proportional error
models and several asymmetric distributions and
concluded that their behaviors are in essence
similar to that of near-symmetric distributions with
homogeneous normal measurement error. The
models for more asymmetric distributions still
93
-------
-3
Figure 5-4. Convolution of a Standard Normal Distribution with normally distributed index variability,
where d2 = (o2unit+o2index)/o2unit (figure from Overton, 1989).
show maximal bias in the neighborhood of ±1
SD, but maximal bias increases as asymmetry
increases.
The similarity in the behavior of the different
distribution and error models examined by
Overton (1989) allows an assessment of accep-
table index measurement variance to be made in
terms of the normal population with homogene-
ous error. Table 5-6 shows bias of CDFs resulting
from index measurement variance over a range
that we might encounter in the selection of
indicators and the design of a field sampling
program. Maximum bias occurs in the neigh-
borhood of ± 1 SD from the population mean, and
relative bias continues to increase toward the
lower tail of the CDF. For larger index meas-
urement variances, maximum bias will be farther
from the population mean.
When index measurement variance is half the
magnitude of the population variance, the
estimated population distribution is biased
considerably (relative biases of +0.31 and +1.25
times the true percentages at Z values of ± 1 SD
and ±2 SD, respectively). For values of index
measurement variance less than 50% of the vari-
ance among sample units, maximum bias for
symmetric populations is still close to ±1 SD.
Consideration of bias at different CDF values of
F(x) also identifies the bias at the Z = ±1 SD
values to be the most important. In the lower end
of the distributions shown in Figure 5-4, the
population fractions that have index values below
-2 SD are so small that they may be of little
concern, and the bias may therefore be of little
importance in general interpretations. Toward the
mean, bias decreases to zero; so that assessing
bias at Z = ±1 SD insures that it is negligible
near the mean. At +1 SD, bias again increases
to its maximal value, but relative bias is small, so
it is clearly the lower end of the CDF that
dominates our concern. Thus Overton (1989)
emphasized a focus on the bias at -1 SD as the
most crucial, not only representing the worst
case, but also the values of greatest interest, and
the values at which the maximal bias has the
greatest importance.
94
-------
Table 5-6. Bias of Apparent F(x) as a Function of d , Assessed for the Normal Distribution Function
with Homogeneous Normal Measurement Error3
Bias
At ± 1 SD
Rel. Bias
At ±2 SD
Bias
Rel. Bias
1.001
1.01b
1.05
1.10°
1.20
1.50
2.00
a
o
d -
0.0001
0.0012
0.0059
0.0115
0.0220
0.0485
0.081 1
°unit + "index
•
^unit
0.0008
0.0076
0.0372
0.0726
0.1387
0.3054
0.5111
from Overton (1989).
0.000
0.0005
0.0027
0.0055
0.0112
0.0285
0.0559
0.0024
0.0238
0.1200
0.2424
0.4921
1.2521
2.457
Design standard suggested by Overton (1989).
allowable limit suggested by Overton (1989).
For normally distributed populations with normally
distributed measurement errors, Overton (1989)
concluded:
• Bias is maximal between 1 SD and 2 SD
from the population mean; for small relative
magnitudes of index measurement
variance, maximum bias occurs close to
± 1 SD.
• Bias is zero at the mean; for small relative
index measurement variance, bias is neg-
ligible near the mean and in the tails.
« Relative bias is large only for very small
(highly negative) values of Z (where Z is a
multiple of the population standard devia-
tion).
5.3.1.2 Simulated Effects of Index Variability
In addition to this theoretical analysis, we also
used simulations to assess the effect of a broad
range in magnitudes of index variance (Figure
5-5). These simulations represent a sampling of
100 lakes from a hypothetical asymmetrically
distributed population with a true mean of 25 and
an SDunit of 50 (among-lake variance of 2500).
We used Monte Carlo techniques and a propor-
tional error model that would be appropriate for
many variables with asymmetrical distributions
and multiplicative error structures that are of
interest in EMAP. The hypothetical distributions
were randomly sampled to yield the mean
observed sample distributions (solid lines). The
crosses depict the range within which 90% of the
simulated sample distributions fell.
The six observed distributions (a through f in
Figure 5-5) all have the same among-lake mean
and variance, with the identical "true" frequency
distribution represented in each figure by a
broken line. Index measurement variance
increases from 1% of among-lake variance in
Figure 5-5a to 100% of among-lake variance in
Figure 5-5f. To clarify, in case f, half the observed
total variance is due to index measurement
variability and the other half is from differences
among lakes across the region. When index
measurement error is only 1 % of the magnitude of
among-lake variance (Figure 5-5a), the observed
95
-------
100
50
0
a)
1%
100
0 20 40 60 80
20 40 60 80
§100
100
50
25%
C
8
CD
100-
20 40 60 80 0 20 40 60 80
100
50
0
0 20 40 60 80 0 20 40 60 80
Indicator Measurement
Figure 5-5. Simulated effect of index measurement variance on survey results observed from
sampling 100 lakes from a hypothetical asymmetrically distributed population with a true
mean of 25 and an SDunjt of 50. The true distribution is shown by the dotted line.
Crosses show 90% range of simulated sample distribution. Index measurement variance
increases from 1% to 100% of among-lake variance in (a) through (f).
96
-------
distribution is almost identical to the true dis-
tribution. A positive bias in lower percentiles
becomes progressively greater as index measure-
ment variance becomes larger. This bias results
in an overestimation of the number of lakes below
some named value of the variable along the
X-axis.
Estimates for the percentiles Q5, Q10, Q25, Q^,
Q75, and Qgg have insignificant absolute bias
when index measurement variance is 10%, but
acquire progressively greater positive biases as
index measurement variance increases (Figure
5-6a). These absolute biases in percentiles do
not exceed 10% of the total population number
until index measurement variance rivals that of
among-lake variance. However, Figure 5-6b
shows that when index measurement variance
exceeds 25% of among-lake variance, errors in
the 5th percentile can exceed 50% (i.e., saying
that 5% of the population is < X, when the true
percentile is really 2.5%). With the same indicator
measurement variance (25%), bias in estimating
the population median is still quite small (5%) in
relative terms (we say that 50% of the lakes are
below some value, when the true value is only
47.5%). Once index measurement is 50% or
100% of among-lake variance, biases are quite
large (Figures 5-6a and 5-6b). In fact the true
population percentages lie entirely outside the
90% range of the simulated sample distributions
for values of the indicator in the region of the 1 0th
and 25th percentiles (Figures 5-5e and 5-5f).
Figures 5-6a and 5-6b show that the absolute and
relative magnitudes of bias in population esti-
mates can be quite divergent, and they have dif-
ferent implications for interpretation. It is perhaps
necessary to clarify by illustration the meanings of
absolute and relative bias. Imagine that we have
sampled 100 lakes in a regional population, with
single observations, using a very "noisy" indicator
that has an index measurement variance 50% of
that derived from true differences in lake condition
across the region. From this survey, we state that
5% of the lakes are "impaired" and 25% are
"threatened." Figure 5-5e shows our observed
frequency distribution, our estimate of the true
population distribution. Figure 5-6a shows that in
fact the estimate of 5% impaired is overestimated
by about 3.5%; really only 1.5% of the population
is impaired. In a relative sense (Figure 5-6b), this
is an overestimate of about 70% (i.e., 3.5%/5.0%).
5.3.2 Acceptable Levels of Index Variability for
Status Estimates
Overton (1989) suggested that for the normal dis-
tribution, a reasonable upper bound of index
measurement error yielding acceptable bias over
most of the estimated population distribution is
10% of o2unjt. For this level of index variability,
the maximum bias is approximately 0.115, and
critical relative bias is approximately 0.07 times
the estimated population proportion (Table 5-6).
At values of index measurement variance < 1% of
sample unit variance, maximum bias (0.001) and
relative bias (<0.01) are very small, and Overton
(1989) suggested this degree of relative precision
in index measurement might be viewed as a
design standard.
It should be noted that acceptable bias is ulti-
mately a function of the precision required in any
given population condition statement. Information
from even very imprecise indicators may be very
useful. As a hypothetical example, suppose that
the Z value of -2.0 in Figure 5-4 is "subnominal".
The true population CDF (Curve E) shows that 2%
of the population is subnominal. An indicator
without impressive precision (e.g., index meas-
urement variance = 20% of among-lake variance)
estimates that about 3% of the population is
subnominal. A very "noisy" indicator, with
measurement variance equal to the among-lake
variance, overestimates the population percentage
of subnominal waters by over 3 times (7%), but
still allows the policy-relevant statement that
< 10% of surface waters are subnominal. Some
caution is in order, however, for if the populations
and their variances are extremely asymmetrical,
even the mildly precise indicator (indicator meas-
urement variance = 20%) could overestimate the
subnominal proportion by about 4 times (12%),
not allowing even the rough assessment that
< 10% of the population is subnominal. In the
asymmetrical distribution, index measurement
variance must be < 10% to yield the correct
rough conclusion that < 10% of the population is
subnominal.
In light of the illustrations in the previous
paragraph, the index measurement variance cri-
teria values of 1% (design standard) and 10%
(upper bound of acceptability) recommended by
Overton (1989) appear to be reasonably robust
guidelines for design of indicators and field
sampling efforts, particularly if analysis of trends
97
-------
<-} 20
g<
&
•a
10 •
0>
8P
Absolute Bias in Percentage Estimates
-------
is desired. For population status estimates alone,
however, these guidelines may be rather stringent,
particularly if population variances are approxi-
mately normal in distribution. Considerably
noisier indicators will probably be useful in these
applications, depending both upon the degree of
symmetry in their error distributions and on the
precision and resolution of population descrip-
tions desired. The differences in biases regarding
statements near the tails of CDF's based on
indicators with different variance distributions
(e.g., normal versus asymmetrical) underscore the
importance of obtaining Information regarding
indicator variability during EMAP pilot studies.
Such studies may also be a necessary compo-
nent of ongoing EMAP. ,
In general, it appears likely that if index meas-
urement variance is greater than about 50% of
variance among sample units (e.g., lakes), and we
believe that the regional population of lakes spans
a meaningful range of the indicator variable, then
the index measurement may be too imprecise to
answer with confidence some questions posed
concerning the proportions of the population
within stated ranges of ecological condition
defined by the indicator values. In such cases,
we must either refrain from making high-resolution
statements, seek a more precise indicator, or
achieve greater precision in the indicator by better
targetting, by replicating samples from each
waterbody in space or time, by taking more
spatially extensive samples within the
waterbodies, or by deconvolution of population
distribution estimates after the fact.
5.3.3 Required Indicator Sampling Intensity for
Status Descriptions
An examination of the relative magnitudes of
variance components in regional data sets can be
used to assess optimal allocation of sampling
effort among sample units (sites), within-index
period replicate measurements, and duplicate
measurements to reduce the biases in estimated
population distributions. Sampling intensity must
be adequate for robust calculations of index
values that do not, by their uncertainty, distort
population distribution functions or preclude
timely detection of relevant trends. The strategy
for evaluating the allocation of sampling effort
between replicates and additional sample sites
across a region is similar, whether the goal is
minimizing the error in the estimated population
mean or minimizing bias and uncertainty in
descriptions of the population distribution. Both
cases involve a comparison of the magnitude of
variation for individual site index values with
variation among all sites across a region of
interest.
It is clear from the expression for o2mean (page
5-17) that, if the cost of index sample replications
is equal to that for adding sample units, the
degree of precision in estimating the mean gained
for each incremental increase in total sample
number is always greater for increases in nunit
than for n . This is true even if index variation
(^index) 's 9reater tnan tne regional variation
among sample units such as lakes (o2^). The
advantage of replicating index measurements
within sample sites comes rapidly into play if the
cost of such replicates is substantially lower than
that for additional sites and if c?-mdex is large
relative to o2^.
The optimum allocation of sampling effort
between numbers of measurement replications
(n ) and sample sites (nunjt) for minimizing the
standard error of the population mean, adapted
from Snedecor and Cochran (1980, p. 453) is:
"unit \
"unlt\
unit
We calculated optimum sample allocations for a
reasonable range of cost and variance ratios that
we might expect in measuring response and
exposure indicators in EMAP-Surface Waters
(Figure 5-7). These calculations show that, for
cases in which costs of index measurement repli-
cation are roughly equal to the cost of adding
new sites, there is no advantage in replication.
This conclusion holds no matter what the variance
ratio of sample units to index measurement. We
caution that this conclusion pertains to minimizing
error in estimation of the population mean. Repli-
cation of index measurements at sites (though not
necessarily all sites) is certainly useful in revealing
the portion of the variance observed over a region
that is simply due to index measurement error,
rather than to true differences among sample
units (e.g., among lakes).
99
-------
•a
3
c
1
5-
4-
1-
SAMPLING OPTIMIZATION
Index Measurement Replication
Inexpensive replicates
Cost (replicate)/
Cost (unit)
.,,,,•"•"" 1/25
...•• 1/10
Costly replicates
0.0
02
0.4
0.6
~08~
—I—
1.0
H h
+
+
High
BCD E
Indicator Precision Ranges
Low
Figure 5-7. Nomogram for determining sample allocation between replications within sample unit and
adding new sample units in a region based on ratio of cost for replications and new sites
and the variability within the index period and amdng sample units.
Replication of index measurements within indi-
vidual water bodies becomes more advantageous
than adding sample sites in reducing the error of
the estimated population mean when the ratio of
replication cost to new site cost is greater than
the ratio of index measurement variance to
among-site variance [(crep/cunjt) >
)]' as illustrated in Figure 5-7. For
example, if index measurement variance is one-
fifth that among sample sites (x axis value of 0.2
in the figure), systematic replication is not justified
unless the cost of each replicate index measure-
ment is less than one-fifth the cost of adding an
addition sample unit (third curve up from the
bottom in the figure). In most cases, replicate
costs (e.g., an additional visit during the index
period) will probably cost more than one-fifth that
for a new site, unless the replicates are spatial or
analytical replicates on the same lake or stream
during a single site visit. If index measurement
variance is unknown (the case for most EMAP
variables) or is expected to be large, replication
on at least a subsample of lakes (or streams) is
necessary to determine the portion of the meas-
ured regional variation in surface water condition
that derives simply from measurement variation.
This information might allow a recalculation of
regional means, variances and frequency distribu-
tions that would be observed if replicated index
measurements at each waterbody had been
averaged (this recalculation and correction pro-
cedure is termed "deconvolution").
To apply Figure 5-7 to decisions about field
sampling design in EMAP, we need information
100
-------
about the ranges of indicator variability within
sample units and across regions and the relative
cost of replication compared with additional sam-
ples. This information is in Tables 5-7 and 5-8.
Table 5-7 (based on Tables 5-4 and 5-5) shows
rough categories of relative precision in indicator
measurements that correspond with ranges A
through E in Figure 5-7. At the far left of the
x-axis (range A), index measurement variance is
almost insignificant relative to the range of
variation among different surface waters across a
region. At the other extreme, range E extends to
conditions where the magnitude of index meas-
urement variance rivals or equals that observed in
sample units across a region. If the range of an
indicator variable measured across a region
describes a meaningful difference in surface water
condition, then indicators with variance charac-
teristics in the A range can be considered 'Very
precise" indicators, and those in the E range can
be considered "noisy." These indicator precision
categories must be interpreted in light of the
meaningful ness of the range of surface water
condition across the population used as a basis
for variance comparisons. For example, in cases
where a population of surface waters is homo-
geneous, index measurement variation (though
perhaps small in absolute terms) may be large
compared with the range of indicator measure-
ments observed among different waters in the
region. In such cases, the fact that the indicator
is noisy relative to the range of variation in
condition in the population of surface waters
simply means that condition within individual
Table 5-7. Rough Categories of Index Measurement Precision (Sum of Analytical, Temporal, Spatial
Components) Relative to Regional Variance for Possible EMAP Indicators3
Group
A Very high precision
B High precision
C Medium precision
D Low precision
0-5% NSS: SO4, Cond., ANC, pH, NO3
ELS-II: ANC
PIRLA-II: DI-pH (deep core)
5-15% ELS-II: SOBC, SO4
PIRLA-II: DI-pH (surficial cores)
15-25% Ohio (streams): fish species richness, IBI
ELS-II: pH, Secchi depth
NSS: T-dP, turbidity
Minnesota, Vermont: Secchi depth
25-50% Ohio (large rivers): IBI
Vermont: Chi a, TP
Minnesota: TP
E Noisy
>50%
Ohio (large rivers): fish species richness
Minnesota: Chi a
Table based on summing columns 3 to 6 in Tables 5-4 and 5-5; information on all index measurement variance compartments
was not available for many variables (e.g., inter-annual variance for NSS and within-reach variance for Ohio IBI). Data sources
are cited in Tables 5-4 and 5-5.
See also Rgure 5-7.
101
-------
Table 5-8. Rough Costs of Index Measurement Replication Relative to Costs of Additional Sites
(Sample Lakes or Streams)8 , f
Category
"unit
''replicate
Type of Measurement (Examples)
Inexpensive
10-25
Moderate
5-10
Costly
Very costly
1-2
Duplicates of routine physical, chemical variables
(Secchi depth, Chi a, D.O., SO4)
Spatial replicates of routine physical, chemical variables —• same
visit, same waterbody, but different location
(Secchi depth, Chi a, D.O., SO4) - :• .;••>"
Spatial replicates of complex analysis on same visit, same
waterbody, but different location
(paleo cores, tnacroinvertebrate and fish assemblage, habitat
measurements)
Temporal replication of simple measurement or collection of easily
obtained sampled
(Volunteer measures of Secchi depth, or recovery of sediment trap
by returning field crew)
Temporal replicates (repeat visits) •
(Secchi depth, Chi a, D.O., SO4, IBI, physical habitat variables,
chemical habitat variables)
Temporal replicates of very costly analyses
Relative costs are crudely estimated assuming a total cost of $5,000 per site/visit, $300 per site cost of full suite analytical
duplicate of routine chemistry variables, somewhat higher for paleoecological analysis of sediment cores, arid maoroinvertebrate,
and fish assemblage sampling/analysis.
surface waters is nearly as variable as the
condition of different surface waters across a
region. The indicator may be noisy in the
homogeneous surface water population but
adequate to discern differences in surface water
condition In a more heterogeneous region.
Many of the chemical habitat variables and the
diatom-inferred pH measurements from deep
cores are in the "very high precision" categories
(A), in which index measurement variance is less
than 5% of sample unit variation. We caution that
inter-annual variability is not available for the NSS
chemical variables, but if ELS-II data are any indi-
cation, that component might add another 3% to
8% to the index measurement variability and
would move those variables down to group B.
The "high precision" group (B), has index
measurement variance between 5% and 15% of
among-sample unit variance. Group B includes
diatom-inferred pH from surficial cores, and
chemical variables that are not very active
biologically, such .as the sum of base cations
(SOBC) and lakewater SO4. The "medium pre-
cision" group (C) includes many biologically
active and flow sensitive variables such as
phosphorus, turbidity, and. Secchi disk trans-
parency. The medium precision group also
includes the determination offish species richness
and IBI in small streams. Chlorophyll-a,
phosphorus in some regions, and large river IBI
determinations are of relatively lower precision
102
-------
(Group D), with index measurement variances
between 25% and 50% of among-site variance in
the data sets we examined. Among the indicator
measurements we examined, only Minnesota lake
phosphorus and fish species richness in large
rivers had combined index measurement variance
components summing to more than 50% of
among-site differences.
The relative cost categories in Table 5-8 are very
crudely estimated, and assume that the cost of
adding a new site is $3,000 to $5,000. Observe
that most of the precise indicators, such as the
chemical variables (group A in Table 5-7), are also
the least expensive to replicate, as long as the
replication takes place during the same sample
site visit. For most of these, even spatial repli-
cation during the same sample visit is probably
not justified, except in very large lakes and
streams (but see discussion about littoral and
backwater habitats in 6.3.3 and 6.4.3). Spatial
replication appears to be advantageous in cases
where the indicator is of moderate to low pre-
cision (groups C, D, E), AND the major source of
indicator measurement variation either is ana-
lytical (or methodological) or is due to spatial
heterogeneity within the sample lake or stream.
Likely examples of this case are IBI or fish
species richness in lakes and streams.
The general conclusion of the analysis discussed
over the last several paragraphs is that inex-
pensive, easily obtained duplicates or replicates
are desirable at all sites for high variance (noisy)
indicator measurements. Replicates on a sub-
sample of sites are advantageous in cases where
the cost of replication is high relative to the cost
of adding new sites in a region. Optimum alloca-
tion of sampling activities among samples over a
region and spatial, temporal, and analytical
replicates within single lakes or stream segments
depends on accurate information concerning vari-
ances and costs at various levels of sampling.
Pilot research activities will include efforts to
quantify the relative costs and relative magnitudes
of variance components.
Compared to procedures for optimally allocating
sampling effort for obtaining precision in esti-
mated population means, the procedures for
obtaining adequate precision and accuracy in
estimated distributions of indicator values in
surface waters are much more difficult to eval-
uate. Once evaluated, however, they are in some
ways more straightforward to describe. Section
5.3.1 developed the theoretical basis for
evaluating the effect of index measurement vari-
ance on population distributions observed in
regional surveys.
If the variance of an indicator is excessive, it can
be cut in half by replicating the sample and using
the average of the two measurements as the
index value for each waterbody. The effect of
variance of this refined indicator is then evaluated
in the manner described in Sections 5.3.1 and
5.3.2, but the indicator measurement variance is
reduced because the index value for each sample
site is now the average of replicate measure-
ments.
The variance tables (5-4 and 5-5) show that com-
bined index measurement variances for most of
the chemical exposure and diatom-inference indi-
cators examined are less than 10% of among-site
variance (many are less than 5%), probably pre-
cise enough to justify single measurements for
describing status. The more precise of the fish
assemblage indicators and the flow-responsive
and biologically active chemical species have
combined index measurement variances of about
10% to 30%. At the high variance end of this
range, it would be advantageous to replicate
spatial sampling within waterbodies to reduce
measurement error. This would reduce bias in
estimation of true population distributions. For
some indicators, such as fish assemblage, vari-
ance might more effectively be reduced by
increasing the spatial extent of fish collection
within each lake or stream reach. When com-
bined index measurement variance exceeds about
30% of the among-site variance in a region, as
occurs with fish species richness and IBI in large
rivers, index measurement variance must be
reduced if we need to make fine resolution state-
ments about the condition of the regional popula-
tion. By fine resolution, we mean, for example,
statements about the fifth or tenth percentiles
rather than simply the median. Wherever pos-
sible, it will be more efficient and economical to
increase spatial replication within sample water-
bodies, rather than to increase the temporal fre-
quency of sampling within the index season.
However, knowledge of the component sources
of indicator variability is necessary (from literature
or pilot studies). This information will allow
evaluation of where in the sampling hierarchy
(analytical duplicates, spatial replicates, temporal
replicates, etc.) increases in sample sizes are
likely to result in greater indicator precision.
103
-------
5.4 EXAMPLES OF EFFECTS OF INDICA-
TOR VARIABILITY ON POPULATION
DESCRIPTIONS AND TREND DETEC-
TION
The previous sections describe approaches that
can be used to characterize regional populations,
differences among subpopulations, and changes
and trends within subpopulations. Earlier sections
also cover the importance of the relative magni-
tude of Index variation versus population variation
in making these estimates. In this section, we
illustrate some of these descriptions with exam-
ples selected from data sets similar to the kinds of
data we expect to obtain from EMAP, that is, data
sets covering reasonably large areas and includ-
ing repeat visits over a period of several years.
These data sets were among those used in Sec-
tion 5.2 to quantify and illustrate the relative
magnitudes of components of variation. A brief
description follows of the data sets from which
the illustrations are drawn. Only one case, the
Eastern Lake Survey, involved a probability design
from which estimates of regional condition can be
derived. In the other studies, lakes or streams
were selected for other specific purposes.
5.4.1 Eastern Lake Survey
As part of the National Acid Precipitation Assess-
ment Program, a variety of chemical attributes
were measured in lakes > 4 ha in the northeast-
ern United States. Lake populations were iden-
tified, target populations were specified, and
probability samples were drawn for field visitation.
The bulk of the sample lakes were visited once
during fall overturn in 1984; a subset of lakes was
visited three times during the subsequent year,
during spring, summer, and fall, to estimate the
magnitude and influence of seasonal variability on
estimates derived during a single index period.
For our illustrations, we use total phosphorus
sampled In these lakes. Data are from Linthurst
et al. (1986) and Herlihy et al. (In press).
5.4.2 Minnesota Pollution Control Agency
The Minnesota Pollution Control Agency conducts
routine lake monitoring throughout the state to
estimate trophic condition. Total phosphorus,
chlorophyll-a, and Secchi disk transparency are
measured in selected lakes several times from
ice-out in spring to late summer. Over 1,000
lakes have been monitored in the past 10 years,
with consistent methods. Only a small subset of
the lakes was visited repeatedly in different years.
These data can be retrieved from the EPA's
STORET system. We selected a subset of lakes
that were sampled repeatedly among years from
the database of over 1,000 lakes. These data
were obtained in cooperation with Steve Heiskary
and Bruce Wilson of the Minnesota Pollution
Control Agency.
5.4.3 Vermont Department of Environmental
Conservation
The Vermont Department of Environmental Con-
servation samples lakes throughout Vermont,
targeting trophic condition and visiting lakes
several times during the year. Lakeis are sampled
during spring overturn for total phosphorus and
during the open water seasons for Secchi disk
transparency and chlorophyll-a. A citizen's
monitoring program gathers additional Secchi
disk readings at numerous lakes to supplement
the Department's monitoring program. Many of
the lakes were sampled repeatedly for seven to
nine years for total phosphorus and Secchi disk
transparency; thus this database is useful for illus-
trating the kinds of trend detection techniques
EMAP will be using initially. Eric Smeltzer of the
Vermont Department of Environmental Conserva-
tion made the data available to us.
5.4.4 Ohio Environmental Protection Agency
The Ohio Environmental Protection Agency con-
ducts routine stream sampling for a variety of
physical, chemical, and biological attributes.
Most of the monitoring program Is devoted to
identifying and characterizing impacts and Impair-
ments associated with point source discharges,
but some sampling has been devoted to charac-
terizing minimally impacted systems. In the illus-
trations below, we have drawn from the Agency's
extensive fish assemblage data collection efforts.
As part of their routine fish collection program,
streams are visited two to three times during the
summer and sampled by electrofishing methods
over 100 to 300 meter sections of the streams.
During the past 10 to 15 years, several thousand
sites have been sampled, so there is an extensive
database from which to draw illustrative material.
Relatively few of the sites have been visited over
a period of years, so the primary purpose of these
data is to illustrate differences among regional
subpopulations. The Agency aggregates the raw
fish collections records into indices that charac-
104
-------
terize each site each time the site is sampled.
Our illustrations use one of the indices, the Index
of Biotic Integrity, modified from the one devel-
oped by Karr et al. (1986). Chris Yoder has made
various parts of this database available for our
exploratory use.
5.4.5 Descriptions of Status
As discussed earlier, one of the initial outputs of
EMAP-Surface Waters will be descriptions of the
status of regional populations and subpopulations
of indicators of interest. Cumulative distribution
functions will form the basis of these descriptions,
with summaries of various population parameters
included. An example of this type of summary is
given by the regional description of total phos-
phorus in lakes in the Northeast derived from the
Eastern Lake Survey (Figure 5-8). In this exam-
ple, confidence bounds on population status can
be established because the survey was based on
probability methods. As in the surface water sur-
veys, only the upper bound is shown. A summary
attached to the figure gives selected population
parameters. During the initial phases of EMAP,
we expect much of the output to be this type of
descriptive summary of various populations and
subpopulations.
Part of the Eastern Lake Survey program docu-
. mented the minor contribution to variance that
was due to index variability relative to population
Minimum: 0
25th Pet: 4.1
Median: 7.5
75th Pet: 12.9
Maximum: 376.4
Mean: 14.03
0.0 <| i i i i i i i i i i
o 10
20 30
Total Phosphorus
40
50
Figure 5-8. Cumulative distribution functions (CDFs) describe the status of regional populations of
lakes and streams, as this figure illustrates for total phosphorus in the Northeast, based
on results of the Eastern Lake Survey. The population of lakes in the Northeast is split
into two parts to show regional subpopulations with clearly differing status. Upper 95%
confidence bounds on each distribution are included.
105
-------
variation, so distortion of the CDF was not an
issue. For some of the EMAP variables, distortion
will be of Importance. Figure 5-9 is an example
drawn from the Maine database, showing a CDF
derived from data collected in August 1985.
Chlorophyll-a is one of the variables for which we
expect index variation to be an important compo-
nent of total variation and therefore to contribute
some distortion to the display. The illustration
shows the CDF derived from the data collected
and the estimated true CDf with the distortion
removed by deconvolution. Observe that, like the
simulations shown in Section 5.3, distortion is
greatest toward the ends of the distribution. A
similar example is given for the Index of Biotic
Integrity in Ohio, a variable that is less "noisy"
than chlorophyll-a is likely to be.
5.4.6 Differences and Changes in Regional
Subpopulations
During its initial phases, EMAP-Surface Waters will
identify possible subpopulations of interest for
targetting further specialized investigations. It will
be useful to determine whether characteristics of
the proposed subpopulations differ from each
other. As more data are collected, there will be
increasing Interest in evaluating the significance of
temporal changes within subpopulations. Most
statistical methods test the significance of
differences in selected population parameters
such as means or medians; EMAP-Surface Waters
will evaluate differences over the entire dis-
tributions (CDFs). Differences and changes might
be more readily detected by looking at the entire
distribution rather than some specific part.
Comparisons will rely on chi-square tests on pro-
portions of the distributions that fall in different
quartiles or qulntiles. The expectation is that,
under the scenario of no difference, proportions
expected in each category will be equal; if differ-
ences should be present, they could be detected
by shifts in the proportions within each quartile or
quintile. The general procedure for paired tests is
shown In Figure 5-3; for unpaired tests, individual
subpopulations are scored relative to the average
CDF of the combined subpopulations. The follow-
ing examples summarize the chi-square tests,
indicating the probability that the calculated chi-
square score would be exceeded by chance.
High probability values indicate that real differ-
ences among subpopulations are highly likely.
The initial set of illustrations covers differences
among subpopulations. We expect large differ-
ences among various candidate subpopulations,
and these will be evident graphically. As we
begin to target particular kinds of subpopulations
at a finer scale, the role of chi-square tests will
become more important. As an example of large
differences in subpopulations, the Eastern Lake
Survey total phosphorus data for the Northeast
can be split into two distinct regional subpopula-
tions, illustrated by clear differences between the
CDFs in Figure 5-8. The northern subpopulation
of lakes occurs primarily in forested landscapes in
which soils are generally nutrient poor; the south-
ern subpopulation occurs in land with substantial
agricultural development and urbanization.
Another example, taken from the Ohio Environ-
mental Protection Agency database, compares
two ecoregional subpopulations, Regions 1 and 5
in the Huron Erie Lake Plain and the Eastern Corn
Belt Plains (Figure 5-10). Again, observe the
substantial differences. On the other hand, other
cases will produce subpopulations that are not as
distinguishable (Figure 5-10).
Temporal differences in populations or subpopula-
tions are likely to be more difficult to distinguish
or detect than major regional differences. Our
illustrations include two types of time intervals:
within-season comparisons and between-year
comparisons. Since index sampling will be the
basis for much of the descriptions, it is of interest
to determine whether differences among potential
index periods occur. For some of the databases,
we can compare different possible index periods.
For example, the Vermont trophic measurements
taken in July and August can be compared as dif-
ferent index periods (Figure 5-11). Extending this
type of analysis among years produces compari-
sons of data collected during the same index
period but in different years. We can illustrate
two types of comparisons EMAP will conduct.
The first will be during the early years when
relatively few sites will be revisited and com-
parisons will be based on independent samples
(Figure 5-12). The second will be after the first
four-year cycle, when comparisons can be made
with repeat visits (Figure 5-13).
106
-------
DEGONVOLUT
POPULATION
10
20
30
40
50
60
IB!
1.0
0.8 H
O
Q_
§0.6
Q.
> 0.4
S\"
O
0.2
0.0
DECONVOLUT
POPULATION
0
Log of Chlorophyll -a
Figure 5-9. Cumulative distribution functions (CDFs) of the Index of Biotic Integrity for Ohio streams,
and chlorophyll-a for Minnesota illustrate the status of two indicators having relatively
large index variation relative to population variation. The distorted CDFs can be decon-
volved to present a truer picture of the CDF of the regional population.
107
-------
1.0-
LU
§2 0.8-
|0.6-
S°-4
DL
LU 0.2-
0.0-
ECOREGION
i i i i i i i i
TTII||||I|||||
O
10
20
REQION 3
REGION 5
30
10
11
18
13
19
12
19
12
13
IB
IBI
40
50
CHI 3.88
PHOB O.4S
60
LLI
LU
Q_
Si
|
O
1.0-
0.8-
0.6-
0.4-
0.2-
0.0-
ECOREGION
TT
10
20
REQION 1
REGION B
13
7
8
12
8
12
S
13
O
20
I | I I I I ! I I I
30 40
IBI
CHI 20.23
PROB O
50
60
Figure 5-10. Cumulative distribution functions (CDFs) of the Index of Biotic Integrity for Ohio
streams illustrating situations in which regional subpopulations are quite similar
(Region 3 versus Region 5) and situations in which clear differences among regional
subpopulations occur (Region 1 versus Region 5). Chi-square summaries indicate
differences in subpopulations if the probability is low (e.g., 0.10).
108
-------
HI H r\
IVE'PERCENTAGI
o o o
:p». b) oo c
i ,- , . i , i , i , , i i
AUG
""":" JULY
3 0.2-
S
o o.o-
0
8
12
Secchi (m)
12-
0
8
12
JULY SEGCHI
Figure 5-11. Cumulative distribution functions (CDFs) showing the similarity in the Secchi disk
transparency status of lakes in Vermont measured during two index periods of the
same year, July and August 1985 (upper panel); median values are identified mid-
graph. The lower panel shows another way of displaying data when sites are paired;
the data set contains several measurements made during each month; for the com-
parisons, one value was selected randomly for each month for each lake. Chi-square
tests indicate no difference in these two index periods, in both cases.
109
-------
30
40
50
60
IBI
1982
1985
14
21
26
&
24
11
S.7
8
18
17"
CHI 15.18
PROB O
Figure 5-12. Cumulative distribution functions (CDFs) using the Index of Biotic Integrity for Ohio
streams show year-to-year differences in the status of small streams. In this case,
sample sizes are quite large, so relatively small differences can be detected among
streams with independent samples. In this illustration, the difference in the median
value was 3 units (28 for 1985 and 31 for 1982).
5.4.7 Regional Trends
Although we have not worked out all the methods
for trend detection, we can use some of the data
to illustrate the nature of the distributions over
time. Vermont's database consists of data for
many lakes that were sampled each year for up to
nine years, from 1981 to 1988. When seven sep-
arate CDFs for Secchi disk transparency are over-
laid, they display a "spaghetti" pattern (Figure
5-14, upper panel). Other ways of displaying
these data are more useful for illustrating trends
(or lack thereof), such as trends in particular
quartiles (Figure 5-14, lower panel), or in the
proportion of lakes that exceeds (or is less than)
a particular critical value, such as an impairment
criterion. "-•
110
-------
o-
0
1978 TOTP
12-
0
0
8
12
1987 SECCHI
Figure 5-13. Cumulative distribution functions (GDFs) taken from the Vermont lakes data set on
trophic condition showing comparisons made on paired sites: (a) total phosphorus
during March, comparing 1978 and 1979, and (b) Secchi disk transparency during July,
comparing 1987 and 1988. In one case, the test indicates high probability of differ-
ences between the two years (total phosphorus), but in the other, differences are not
detectable.
111
-------
23456789 10 11
Secchi (m)
7-
111 11 i i i i • i i i rrn i i i MM i i M
1981
1982
1983
1984
1985
1986
1987
Figure 5-14. Superimposed yearly cumulative distribution functions (CDFs) show the confusing
"spaghetti" pattern in multiple overlays of CDFs, based on nine years of data on Secchi
disk transparency in Vermont lakes (upper panel). Results obtained from yearly CDFs
can be more clearly presented to illustrate trends, or lack of trends, by selecting yearly
values representing different quartiles (lower panel). The dashed lines represent a 2%
per year change in Secchi disk transparency, for comparison with the trend graphs.
112
-------
6.0 FIELD SAMPLING DESIGN
This section describes the design approach for
gathering field information and for obtaining
chemical and biological samples from surface
water sites selected for field visits in Tier II of
EMAP. We explain where and when field meas-
urements will be made on these lake and stream
sample units. The surface water sample units to
be visited are lakes, reservoirs, and stream
segments within several size classes as described
in Section 3.0. The actual measurements to be
made and the chemical/biological samples to be
collected are described in Section 4.0.
We should view the data quality objectives
(DQOs) for the field sampling design in the
context of (1) the DQOs for the spatial design
(site selection) and (2) the choice of EMAP indi-
cators. For total quality assurance (QA), the DQO
of Tier 1 sample site selection (Section 3.0) is to
provide a set of field sampling sites representative
of the populations of surface waters within the
regions of interest. The DQOs of indicator design
are to provide practical measures that accurately
and precisely measure the ecological condition,
exposures, and stresses at these sampling sites.
The field sampling design described in this sec-
tion defines, within each sample unit (lake or
stream segment), the specific places and times at
which these indicator measurements will be
made. The overall DQOs of field sampling design
are to provide adequate representation of the
sites in order to classify the condition of the sites
and make robust estimates of surface water
status and trends on national and regional levels.
Biological, chemical, and physical characteristics
of surface waters vary both temporally and spati-
ally. The degree to which sampling must incor-
porate this variability depends upon how robust
the indicator measurements are and what informa-
tion is required of them. The field sampling
design is not intended to provide adequate samp-
ling frequency and coverage to describe this vari-
ability at each individual site, unless changes in
this variability are a major ecosystem response to
environmental stress. Ideally, indicator measure-
ments might be made at single well-chosen points
in space and time within each lake and stream.
However, temporal or spatial variability in the
indicator measurements at individual sites may
prevent robust site condition classification and
thus regional status descriptions. In such cases,
several courses may be taken:
• Take more samples and express the index
value for each water body as the mean of
these replicate measurements (e.g., sample
/ ' more frequently if temporal- variability is
high; sample more locations within the lake
, or stream if spatial heterogeneity is high).
• Target the sampling to a more specific time
and location. Temporal sampling may be
targeted more to the time of greatest bio-
logical stress, or it may be targeted to a
more stable time period that can be associ-
ated with stressful conditions by calibra-
tion. Spatial sampling within heterogene-
ous lakes or stream's may be stratified to
collect separate sets of information from
different habitat types (e.g., littoral versus
profundal, riffle versus pool), Similarly,
sampling may be aimed specifically at one
habitat type of greatest interest, or one that
is most influenced by the condition of sur-
rounding habitats.
• Deconvolute, or correct bias in population
distributions calculated from the sample
data, based on quantification of the magni-
tude of index measurement variance in a
subset of sample sites. Deconvolutio'n can
be used to compensate for index measure-
ment error in population distributions, but
no analogous procedure can be used to
extract trends from excessively "noisy"
index data.
Our basic design is to sample all surface waters
once in the late summer period. Water and sedi-
ment samples from lakes and reservoirs will be
collected in a mid-lake location judged to approx-
imate the deepest point of the lake. Biological
samples will be collected from representative
major habitat types within lakes. Physical habitat
indices will be used to describe the lake and its
riparian zone as a whole (Section 4.0). In streams
and rivers, the sampling location will be a reach,
30 channel-widths long, located midway between
the mapped upstream and downstream conflu-
ences defining the stream segment that is the
sample unit. Water samples will be taken at a
single, mid-channel location in the middle of this
113
-------
reach, and biological samples will be system-
atically sampled In representative habitat types
within the reach. Physical habitat characteristics
will be assessed over the entire 30-channeI-width
section and its riparian zone.
6.1 INDEX CONCEPT
6.1.1 General Design Objectives of Index
Sampling
Many broad-scale surveys and trend monitoring
networks have employed indices to characterize
sample units, limiting sampling effort on individual
units in order to maximize regional coverage with
available personnel and financial resources. The
Index values obtained in such studies are
Intended as single values of each exposure or
response Indicator variable that are robust
enough to represent each sample lake and stream
for the purposes of classifying its condition,
tracking trends, and extrapolating these sample
data to regional and national scale descriptions.
There are both temporal and spatial aspects of
index measurement, and the optimal field samp-
ling design depends upon the objectives of
monitoring, as well as the variability of the
Indicators to be measured.
The overall DQOs of EMAP require that indicator
measurements be relatively stable, but responsive
to year-to-year changes in biotic stress and
exposure. The relationships among stressor,
exposure, and response indicators have been dis-
cussed In Sections 2.0 and 4.0, as has the inter-
pretation of ecological condition based upon
response Indicator measurements. As they apply
to the field sampling design, EMAP's overall
DQOs require that the Indicator measurements be
taken at locations and times that balance often
opposing requirements for Indicator stability,
spatial representativeness, and relevance to biotic
stresses.
In the temporal context, an ideal index period is
a time during which the values of response and
exposure Indicators are relatively stable. It must
be a time period when indicator biota are present
and measurable. To facilitate diagnosis of prob-
able causes of subnominal indicator values, it is
advantageous for the Index time period to also be
a season In which indicator organisms are
exposed to maximum environmental stress. In
some cases the periods of maximal environmental
stress to indicator organisms may be very short-
lived or episodic. In these cases it is advan-
tageous to choose an index period for which con-
ditions may be related in a predictable way to
conditions during the most stressful times, so that
exposure might be inferred by calibration from
other temporally intensive sampling.
In the spatial context, an index value should
provide in a single measurement, or an integration
of a number of measurements, information ade-
quate to represent each sampling unit in classifi-
cation, regional extrapolations, or trend analysis.
If the sampling unit is heterogeneous, the index
measurement should, through multiple sampling
points, sufficiently incorporate the spatial
variability of indicator measurements within the
sampling unit (lake or stream). Alternatively, the
measurements might be made at locations which
themselves integrate the ecological stresses
impinging on the sampling unit as a whole. For
example, pelagic fish and mid-lake water quality
might reflect benthic biogeochemical processes,
chemical inputs from streams, and littoral
spawning, rearing, and feeding areas.
6.1.2 Applications of the Index Concept in
Synoptic Assessments
Several examples are useful for illustrating the
application of the temporal and spatial aspects of
the index concept.
6.1.2.1 National Stream Survey (MSS)
The objective of the EPA's National Stream
Survey (NSS) (Kaufmann et al. 1988, in press)
and other acid rain related stream suiveys (Knapp
et al. 1988) was to index conditions of biological
exposure to low pH, high aluminum, and other
chemical conditions related to stream acidification
associated with acidic deposition. Streams were
sampled during "baseflow" (not storrn flows) dur-
ing the spring season after snowmelt but before
leafout. Chemistry during this index time period
was shown to be quite stable in southeastern
streams (Messer et al. 1986, 1988), and popula-
tions could be adequately described based on
single spring baseflow measurements in sample
streams. In the mid-Atlantic region, where spring
season chemical variability was greater
(Kaufmann et al. 1988, Baker et ail. in press),
stream sampling sites were characterized by
averaging two spring baseflow measurements.
114
-------
At each sampling visit, the NSS measured charac-
teristics at the upstream and downstream ends of
each stream segment (sample unit). Water chem-
istry samples were taken at mid-depth and mid-
stream, ignoring spatial heterogeneity that might
be present in relatively small volumes of water
(relative to seasonal runoff) outside:-the main
channel flow. On the average, the distance
between measurement locations (i.e., the distance
between confluences on 1:250,000-scale maps)
was 3 km. In the low ANG streams of greatest
interest in the NSS, the median downstream
chemical change was +5 /zeq/L ANC and +0.06
pH units per km of .stream length. To incorporate
these characteristic downstream trends, chemical
index values measured at each end of the sample
stream segments were used to estimate, by inter-
polation, the chemical characteristics of the whole
segment. , ,
Chemical conditions more stable than spring
baseflow might have been observed during the
summer baseflow period in NSS streams. The
intent of the NSS index, however, was not only to
minimize within-season and episodic variability,
but also to maximize the probability of sampling
chemical conditions potentially limiting for aquatic
organisms. For a spring index stream sampling
period to be biologically relevant, sensitive life
stages of aquatic biota must also be present
during the sampling period. Spring is the season
when acid-sensitive swim-up'fry of many sport
fish are present in streams within these regions,
although fry of some trout (Salmo gairdneri, S.
trutta, Salvelinusfontinalis) populations may also
be present at other times of the year. Prolonged
periods of low pH and ANG are most likely to be
observed in the winter and spring seasons in
streams within the Northeast, .Mid-Atlantic, and
Southeast (Colquhoun etal. 1981, Murdoch 1986,
Witt and Barker 1986- Olem 1986, Nix et al. 1986,
Ford et al. 1986, Kaufmann et al, 1988);
The NSS effectively indexed the chronic exposure
of aquatic biota to acid stresses. However, the
most stressful times of exposure for fish fry are
probably not during spring baseflows, but rather
during episodic pH depressions that occur during
spring storm flows. It is not practical to sample
storm episodes on a synoptic scale (Messer et al.
1987), but episodic conditions can be inferred to
some extent from spring baseflow index chemis-
tries using hydrolpgic models that also take into
account the chemistry of precipitation (e.g.,
Eshleman 1988, Gerritsen et al. 1989).
In contrast to biologically relevant chemistry (in
the context of acidic deposition), general biotic
condition in response to acid stresses need not
necessarily be measured during the actual time of
maximum stress. The presence and abundance
of nonanadromous fish species .have in most
eases been measured .during the traditional
summer field season; Although fish presence and
absence must be interpreted in relation to other
natural and anthropogenic stresses, this biological
information can be compared with spring season
pH and. then related to .modelled, episodic condi-
tions (e.g., Baker et al. .in press,-Morgan and
Janicki in press).. ..--. :
6.1.2.2 National Lake Survey (NLS)
The EPA's Eastern and Western Lake Surveys
(Linthurst, et ah 1986, Landers, et al.. 1987)
employed a single mid-lake sample taken at a
depth of 1.5 nrvduring fall turnover to represent
the nonlittoral lake water volume.: The relatively
well-mixed fall turnover period was targetted to
minimize spatial (particularly vertical) variability.
Although sampling might also have been under-
taken during spring mixing following ice-put, the
fall period had the advantages of (1) relatively
stable water chemistry over a sampling window of
several weeks,. (2) ease of access, and (3) safety.
Unlike the strategy used in the NSS, the National
Lake, Survey (NLS) did,not directly target the
season of maximal acid stress, As in, streams,
spring pH minima in lakes often coincide with the
presence of acid-sensitive developmental stages
of:manyfis,h species. However, the robust chem-
ical classifications of lakes based .on fall sampling
can be related by calibration to the more biologic-
ally relevant acid stresses that occur during snow-
melt (Eshleman 1988), or in the spring season
(Baker et al. in press, Herlihy et al. in prep.). In
addition, fish species presence and abundance
measured in traditional mid-summer surveys have
effectively been related to fall index chemistry. As
for streams, the influence of acid stresses must
be evaluated within the context of .other natural
and anthropogenic stresses such as fishing har-
vest and physical habitat quality changes.
115
-------
6.2 TEMPORAL CONSIDERATIONS FOR
FIELD SAMPLING
Sections 2.0 and 4.0 discuss the rationale for
measuring EMAP-Surface Water response and
exposure indicators In the late summer season.
An EMAP design goal Is to target a time period of
high stress to aquatic biota. This design goal
facilitates the diagnosis of probable cause of
ecological impairment. The index period must
also be a time when robust measurements of
ecological condition can be made.
For a number of biotic stresses, late summer is
generally a time of high stress due to many
coincident factors. In most of the United States,
streamflows are at a minimum during the late
summer and fall (Figure 6-1). The major excep-
tions are the Rocky Mountafn region, much of
Alaska, and parts of the Great Plains, where
almost all precipitation is locked up in the form of
snow until the spring thaw. Lowest lake and
reservoir levels tend, with some lag, to corre-
spond with the season of lowest stream runoff.
Furthermore, the highest stream and lake water
temperatures tend to coincide with these seasons
of lowest runoff In much of the United States.
Evapotransplratlon and irrigation withdrawals from
surface waters are also at a high during the late
summer, as are the pollutant inputs from irrigation
returns and municipal waste disposal. Low flows
and lake levels minimize the dilution of point and
nonpolnt source pollutant inputs, and high tem-
peratures increase metabolic oxygen demands,
while at the same time decreasing the solubility of
dissolved oxygen. High water temperatures and
resulting rapid metabolic rates increase the expo-
sure of aquatic organisms to toxins and increase
the toxic effects of many substances. The poten-
tial for algal blooms and hypolimnetic deoxygen-
atlon Increases during the summer period of ther-
mal stratification in eutrophic lakes and reservoirs.
These numerous stress factors converge to make
late summer to early fall the period of maximal
stress to aquatic biota in many surface waters.
Late summer/early fall stress is common and
widespread in surface waters, but this season is
by no means the only period of important biotic
aquatic stresses In the United States. Examples
of Important exceptions are in the areas of sur-
face water acidification, lake eutrophication, and
stream physical habitat alteration.
f- ..,;'• ; i
Spring is the season in which biotic stresses due
to surface water acidification from deposition are
typically greatest. Surface waters in areas most
affected by acidic deposition are generally most
dilute during spring and provide less ANC than
during summer and fall. Winter may be the time
of critical stress in eutrophic lakes that freeze,
because biochemical degradation off organic mat-
ter may result in low dissolved oxygen concentra-
tions under the ice that are limiting to fish. Winter
is also the critical season of stress in many small
high-elevation lakes that freeze to the bottom; in
fact, most of these lakes do not support naturally
reproducing fish populations. Removal of riparian
vegetation from small northern and subarctic
streams can result in fish habitat loss due to the
formation of anchor ice as heat loss by radiation
is increased in these streams, making winter the
season- of maximal stress for many fish. Similarly,
channel and riparian zone activities that reduce
physical complexity in streams may reduce slack-
water habitat during winter and spring high flows,
making these high-flow seasons the periods of
maximal stress for fish. For example, removal or
depletion of large woody debris in Pacific coastal
streams has resulted in reduction of winter cover
from high water velocities for anadromous and
resident salmonid fishes.
6.3 INDEX SAMPLING DESIGN FOR LAKES
AND RESERVOIRS
6.3.1 Temporal Variability
Temporal variability in the lake and reservoir
characteristics to be measured by EEMAP indica-
tors occurs over a range of time scales. Varia-
bility over decades, years, seasons, and within
seasons may affect the precision and interpreta-
tion of status and trends in condition of lakes and
reservoirs (Sections 5.3 and 5.4). Table 6-1 lists
examples of time scales of variability and some
common sources of variability. 6-1. Some of
these sources are cyclical, and might be consid-
ered "noise"; others may be trends that EMAP will
want to quantify.
6.3.2 Summer Lake/Reservoir Index Period:
The Proposal
Based upon the information discussed in Sections
6.2, 4.0, and 5.0, we propose a mid- to late sum-
mer index period (generally from July to mid-
September) for sampling lakes and reservoirs in
most of the United States.
116
-------
i
15
0)
O
D
w
w
£
117
-------
Table 6-1. Time Scales and Sources of Variability in EMAP Indicator Characteristics - Some Examples
Time Scale
Source of Variability
Years to decades
Within/Among seasons
Centuries and longer Tectonic activity that changes physiography
Continental glaciation
Climatic change
Changes in species pool resulting from migration, extinction, evolution
Residual effects of "resetting" by fire, volcanism
Cyclic.,climatic variation (e.g., ENSO)
Land use changes
Residual effects of "resetting" by fire, floods, mass movements,
volcanism
Hydrologic inputs, floods, evapotranspiration
• Lake/Reservoir stratification
• Seasonal cycles of plant production and die-off, with 'associated
chemical uptake and release
• Competitive, synergistic interactions among phytoplankton, macro-
phytes, and other groups
• Grazing, secondary consumption :
? Life history, developmental stages, macroinvertebrate emergence, fish
migration
Die! • Wind, wave action, temperature, sunlight, seiches
• Plant production, uptake, release
• Mobility of organisms (active and passive)
We plan pilot studies for each of the major
regions before or at the time of full EMAP imple-
mentation In those regions. Available information
on Indicator variability suggests that, for most of
our proposed Indicator measurements, one visit
to each site within the summer index period will
probably be adequate to describe regional status.
Seasonal sampling and wlthln-season replication
in a subset of approximately 20 to 30 of the first
year's sites within each region will be aimed at
testing the relevance, robustness, and practicality
of the proposed Index period and sampling fre-
quency. The questions that will be re-evaluated
with regard to our choice of Index sampling time
are as follows:
• Relevance
• Is the season a period of stressful con-
ditions for biota?
- Are sensitive indicator species present
in sufficient number to be collected
effectively?
Robustness: Is temporal variability suf-
ficiently small to yield acceptable accuracy
and precision in regional extrapolations of
surface water condition?
Practicality
- Is the jndex sampling period long
enough to allow sampling crews to
complete planned site visits?
- Is index period weather suitable for safe
sampling?
- Can field crews find, reach, and gain
access to the sampling sites?
118
-------
- Are enough willing, qualified field tech-
nicians available during the index time
period?
Our answer to all the above questions is a quali-
fied "yes" for single site visits during the summer
index period for lakes and reservoirs in the North-
east, where the first EMAP regional pilot will be
undertaken. Pilot activities will test the effec-
tiveness and feasibility of that choice. Except for
the arid West, where a spring period may be
advantageous for flowing waters, the summer
period is our likely choice in the other regions of
the country.
6.3.3 Spatial Variability within Lakes and
Reservoirs
Spatial variability in lakes and reservoirs can be in
both vertical and horizontal dimensions. Vertical
variability in temperate lakes occurs primarily
because of temperature stratification. Summer
stratification occurs with warming of surfiace
layers of water, whereas less marked and less
stable winter stratification occurs when water at
temperatures < 4°C lies above warmer water (this
is usually under ice). Strong summer stratification
severely limits vertical mixing, so epilimnetic arid
hypolimnetic processes become largely separ-
ated. Shallow and deep waters take on different
physical, chemical, and biological characteristics
that reflect the different light levels, atmospheric
gas exchange, hydrologic inputs, and biological
activities in upper and lower layers. Phosphorus
and heavy metal concentrations in hypolimnetic
waters may greatly exceed those in the epilimnion
if organic decomposition consumes oxygen and
allows release of these substances from bottom
sediments under reducing conditions. Stratifica-
tion is usually most pronounced in lakes or reser-
voirs that are deep in relation to their surface area
or length dimensions. Additionally, lakes experi-
encing only gentle winds due to their climate or
topographic setting tend to become more
strongly stratified than exposed lakes in windy
regions. The depth of thermal stratification can
be affected by turbidity. Turbid waters may
stratify more intensely than clear waters because
solar energy is more rapidly attenuated in the
surface waters.
Phytoplankton species and abundance may Vary
markedly at different depths, even in waters that
are riot thermally stratified, these differences are
the result of differential growth rates, sinking
rates, and mobility at different water depths and
temperatures and among different planktonic
species. Chlorophyll concentrations reflect the
differences in algal biomass.
Horizontal : variation in lake characteristics
measured by EMAP indicators occurs because of
a wide variety of factors, including lake size,
rriorphometry, and substrate, and the location and
magnitude of inlets and outlets. The spatial
variability of EMAP indicator measurements dir-
ectly associated with water column characteristics
(e.g., chlorophyll, temperature, water chemistry)
may tend to increase with lake size, particularly if
shallow depths and complex morphometry
impede lateral mixing. For example, water clarity
and phytppiankton production rates often vary
markedly at different locations in large lakes or
reservoirs with numerous, relatively isolated arms
that receive different hydrologic and chemical
inputs. Similarly, depth-integrated samples taken
at near-shore locations in lakes with gradually
sloping bottoms will show a greater influence of
littoral epilirrinetic processes than those which
remain deep very close to shore.
In contrast to measurements of chemistry or
phytoplankton in the water column, horizontal
variability in measurements of fish or benthic
macroinvertebrate assemblages are not as
strongly dependent upon the degree of lateral
mixing. Rather, they are more strongly dependent
upon the spatial heterogeneity of substrate, the
water depth, and the distribution Of macrophytes
and other elerhents that provide habitat structure
and cover. Fish, being highly rnobile, may move
from profuridal to littoral habitats to find food arid
cover at different times of the day and year, The
mobility of fish wiH often be interpreted as tem-
poral variability if the spatial scale of sarhpling is
riot large in comparison with the range of short-
term migrations. Different developmental stages
of the same fish species may be found in very dif-
ferent types of habitat within the same lake
because the stages have different requirements
and preferences for water depth, temperature,
chemistry, food, and cover.
119
-------
6.3.4 Spatial Sampling Index for Lakes and
Reservoirs: The Proposal
Spatial aspects of our proposal for measuring
EMAP Indicators differ according to the specific
type of Indicator measurement:
• Fish assemblage: Fish samples will be
collected from profundal and major littoral
habitats (beach, rock, rubble, emergent
macrophytes, submergent macrophytes).
Samples will remain separate during pilot
studies. Pilot research will be used to
determine how to incorporate fish assem-
blage data from different habitat types into
metrics that Index whole-lake biointegrity.
• Benthlc macroinvertebrate assemblage:
Macrolnvertebrates will be collected from
profundal and major littoral habitats as will
fish. Replicate benthic samples (n=5) from
the same habitat type will be analyzed sep-
arately for a subset of 20 to 30 lakes in
each region in order to evaluate sampling
variation. Replicate samples in the remain-
der may be combined before counting and
Identification. As for fish assemblage data,
pilot research will be used to determine
how to Incorporate macroinvertebrate
assemblage data from different habitat
types into whole-lake biointegrity indi-
cators.
• Sedimentary diatom assemblage: Sedi-
ment cores and surficial sediments for
analysis of algal remains will be collected
from profundal habitats only. Replicate
sediment cores (n=3) will be analyzed sep-
arately in a subset of 20 to 30 lakes in
each region to assess spatial variability.
For the remaining lakes, a single mid-lake
core will represent the lake.
• Trophic state: Water samples will be col-
lected at one mid-lake location. Seech!
disk transparency will be replicated at two
to four additional sites where profundal
sediments are sampled.
• Physical habitat: Measurements as out-
lined In Section 4.0 are designed to
describe the entire lake, including
profundal, littoral and riparian areas.
Additional characterization activities will
describe topographic .drainage basin of
lake from maps, aerial photographs, and
other studies.
Chemical habitat: Single waiter samples
will be collected mid-lake at 1.5 m depth.
Profiles of dissolved oxygen, specific con-
ductance, pH, and temperature will be
measured at the same location. We will
assess index variability in a subset of 20 to
30 lakes in each region. We will examine
spatial variability by taking multiple profiles
in different parts of the lake. We will
assess temporal variability in these lakes
by taking mid-lake samples at Iwo different
times in the index period. V/e will treat
large lakes (surface area > 2,000 ha) dif-
ferently, obtaining samples from a number
of subjectively chosen representative pro-
fundal locations (e.g., separate arms of a
large reservoir).
6.4 INDEX SAMPLING DESIGN FOR
STREAMS AND RIVERS
6.4.1 Temporal Variability
Streams and rivers, like lakes and reservoirs,
exhibit temporal variability in the characteristics to
be measured by EMAP indicators. Over the long-
est time scales (years, decades, centuries, and
longer), the sources of this variability are much
the same as those presented as eiKamples for
lakes. Rivers and streams are generally more
persistent than lakes on a geological time scale
and their long-term characteristics are probably
more stable. However, these flowing waters have
much smaller hydrologic retention times than
lakes, so the amount and character of water in
their channels more closely reflects fluctuations in
the sources and sinks for water, energy, and
chemical substances. As a result, flowing waters
often tend to exhibit greater seasonal, within-
season, and diel variation than do most lakes, for
those characteristics that depend on the amount
and flowpath of hydrologic inputs. Important
exceptions to this pattern are shallow, ephemeral
lakes, which, like other ephemeral waters, exhibit
extremes in physical, chemical, and biological
variability.
On time scales shorter than a year, characteristics
of flowing water ecosystems measured by EMAP
indicators respond to seasonal patterns in precipi-
tation, runoff, water temperature, watershed
120
-------
activities, riparian cover, riparian organic inputs,
periphyton productivity, fish migration, and
macroinvertebrate life histories. Virtually all
characteristics of streams and rivers change to
some extent with discharge. Short-term changes
in discharge may cause marked changes in
stream water chemistry. The chemical changes
and increases in water velocity accompanying
spates may cause increases in the downstream
movement of macroinvertebrates (drift), and may
cause fish to move downstream or into lower
velocity backwaters or tributaries. Similarly,
increases and decreases in discharge can cause
marked changes in the extent and quality of
physical habitat in a stream or river channel.
Channel and bank scour and deposition from
extreme flows can cause catastrophic change in
macroinvertebrate, fish, and periphyton popula-
tions, and may greatly change the morphometry
and flow characteristics of the channel when
discharges subside. These periodic "resetting"
mechanisms are necessary for the long-term
maintenance of,many ecosystems, despite their
immediate detrimental effects. In regional
aggregations, effects associated with the natural
or altered disturbance regime become "noise,"
and should, in fact be sampled and characterized
as part of the natural variability in regional surface
water population descriptions.
6.4.2 Summer Baseflow Stream/River Index
Period: The Proposal
Based upon available information discussed in
Sections 4.0 and 5.0, we propose a mid- to late
summer index period for sampling streams and
rivers in most of the United States. This period is
generally from July to mid-September. As for
lakes, we plan pilot studies in each of the major
regions before or at the time of full EMAP imple-
mentation in those regions. Available information
on indicator variability suggests that, for most of
our proposed indicator measurements, one visit
to each site during baseflow within the summer
index period will probably be adequate to
describe regional status. Seasonal sampling and
within-season replication in a subset of approxi-
mately 20 to 30 of the first year's sites within each
region will be aimed at testing the relevance,
robustness, and practicality of the proposed index
period and sampling frequency. The following
questions, which will be re-evaluated with regard
to our choice of index sampling time, are almost
the same as those for lakes and reservoirs:
« Relevance
- Is the season a period of stressful
conditions for biota?
- Are sensitive indicator species present
in sufficient number to be collected
effectively?
- Are important flowing waters dry during
the summer and better sampled during
a wetter season?
• Robustness
- Is temporal variability small enough to
yield acceptable accuracy and precision
in regional extrapolations of surface
water condition?
- Are stream and river flows and physical
habitat quality and extent stable enough
to provide a reasonable physical tem-
plate for biota within the summer index
period?
• Practicality
- Is the index sampling period long
enough to allow sampling crews to
complete planned site visits?
- Is index period weather and stream flow
suitable for safe sampling?
- Can field crews find, reach, and gain
access to the surface water sampling
sites?
- Are enough willing, qualified field tech-
nicians available during the index time
period?
As for lakes, our answer to all the above ques-
tions is again a qualified "yes" for single site visits
during the summer index period for streams and
rivers in the Northeast, where the first EMAP
regional pilot will be undertaken on lakes. Pilot
activities in each region will test the effectiveness
and feasibility of the chosen index season. For
example, spring sampling will be evaluated for
streams in the arid West and for coastal streams
with anadromous fish populations. Pilot studies
will be undertaken in each region to re-evaluate
121
-------
the summer period as our likely choice in the
Northeast and other regions of the country.
Within the summer Index season, stream chemis-
try may change considerably in some regions.
This effect might be minimized by choosing rela-
tively stable low flows of summer, but the pro-
found effects of summer plant and microbial
growth cycles will have a large effect on the low
flows of summer.
6.4.3 Spatial Variability within Streams/Rivers
6.4.3.1 Upstream-downstream Variations
Stream discharge, gradient, channel width, water
depth, substrate, and many other physical charac-
teristics change In a predictable fashion along the
course of a stream from its headwaters to its
mouth (Strahler 1957, Leopold et al. 1964). Simi-
larly, geology, soils, vegetation, and the influence
of groundwater on streamflows change markedly
from stream headwaters to the mouths of rivers.
Recognizing these physical and chemical pat-
terns, Vannote et al. (1980) proposed the Stream
Continuum Hypothesis, stating that the structure
and function of biological communities in streams
should reflect these upstream-downstream pat-
terns of variation. The reach segment design
proposed for sampling small and mid-sized
streams in EMAP was chosen for its efficiency in
capturing broad-scale regional variations in
stream characteristics. It is not the ideal design
for describing downstream changes within individ-
ual streams. However, examination of longitudinal
(downstream) variation in streams is incorporated
Into the EMAP stream sampling design through
stream size stratification (see Section 3.2).
Though It is unlikely that EMAP will be able to
provide complete, detailed descriptions of the
downstream trends in any but the largest river
systems, major patterns of upstream-downstream
variation will be incorporated into the estimates of
regional condition. In addition, specific subsets of
stream size can be compared to detect those
upstream-downstream trends that are generally
consistent among river systems in a given region.
The population sampling units for streams and
rivers in EMAP are segments between mapped
confluences. A similar design used in the NSS
showed fairly consistent qualitative patterns of
change in physical and chemical characteristics
between upstream and downstream sampling
points an average of 3 km apart (see subsection
5.2.2 and Kaufmann et al., 1988). The NSS
sample units were segments (reaches) depicted
on 1:250,000-scale USGS topographic maps.
Though differences between upstream and down-
stream sample sites were often large, they were
small in relation to differences among streams.
The imprecision resulting from sampling only one
point on each segment would cause only a small
distortion of regional population descriptions of
stream chemistry.
Despite the fact that NSS results showed that
upstream-downstream variation on stream seg-
ments was relatively insignificant compared to
regional variation (Table 5-5 and subsection
5.2.2.3), interpreters of NSS data found it
necessary to use both upstream and downstream
chemical information in an assessment of the
acid-base status of streams. The MSS experience
has bearing on the design for EMAP. In contrast
to the EMAP target resource, the NSS targeted a
narrower range of stream size (only streams with
watershed areas < 155 km2, but large enough to
appear on 1:250,000-scale maps). In addition, the
NSS showed that, outside of streams acidic
because of mine drainage or natural organic
acids, acidic conditions in streams were over-
whelmingly concentrated at the upstream portions
of the stream segments examined. The biggest
disadvantage of sampling only downstream ends
of NSS segments would have been the complete
avoidance of most of the acidic stream resource--
rather than simply poor precision in approxi-
mating stream segment characteristics by one
point measurement. The NSS needed upstream
as well as downstream measurements on sample
stream segments in order to extend the scope of
inference to streams at the small end of its
targeted range. A further benefit, of course, was
that relatively precise estimates of the extent of
acidic stream length could be made. Because
EMAP targets a broader size range of streams
than NSS and employs stream size stratification
in its design, there is less of a disadvantage in
sampling one point on each stream segment~at
least for chemistry. Furthermore, the average
length and the variation in length of sample
stream segments in EMAP could be reduced,
compared with NSS, if EMAP used finer resolution
1:100,000-scale maps as a basis for its sampling
frame.
In contrast with streamwater chemistry, stream
channel morphology and therefore biological
assemblages may show cyclical spatial patterns
122
-------
between mid-segment and confluence areas at
the ends of segments. Confluences are often hot
, spots of biological activity because of tributary
inputs of sediment and organic debris. We will
need to investigate these effects of within-
segment position in EMAP pilot research activi-
ties.
6.4.3.2 Other Stream Habitat Variations
We anticipate that a major source of variation in
EMAP fish, macroinvertebrate, and periphyton
assemblage measurements will be associated with
physical habitat differences within each sample
stream or river sample segment (reach). Channel
features such as riffles, pools, and backwaters
provide a wide variety of water velocities, depths,
and substrates favoring different biotic assem-
blages. These habitats also differ in their ability to
physically and biochemically retain dissolved and
paniculate organic matter and other nutrients.
Riparian interactions differ among channel habi-
tats and along the course of stream channels.
Stream ecologists frequently report that macro-
invertebrate and periphyton assemblages differ
more between contrasting habitats (e.g., riffles
versus pools) than among similar habitats in
streams across often large regions. For this
reason it will be necessary for EMAP's field design
to stratify sampling within individual stream
sample units, or sampling could specifically target
one habitat type. Riffle habitats are often sampled
in synoptic macroinvertebrate surveys because
they often contain numerous and diverse assem-
blages of species, and because they can be sam-
pled relatively easily using consistent methods
from stream to stream.
Single habitat sampling would be an acceptable
design for EMAP if the program were interested
mainly in ecological changes associated with
exposure to changes in water chemistry, temper-
ature, and other characteristics that do not
necessarily change the relative abundance of
habitat types. To the contrary, EMAP plans to
monitor ecosystem changes caused by changes
in physical habitat. For this reason, initial field
activities will separately sample major physical
habitat types within each stream sample site.
This pilot research will investigate whether
assemblages within particular habitat types may
perform the function of overall condition
indicators in reflecting changes that occur in the
relative abundance of other physical habitat types
or the condition of their biota.
6.4.4 Spatial Sampling Index for Streams and
Rivers: The Proposal
We propose to represent stream and river seg-
ments by sampling a mid-segment location.
Logistical field trials and pilot sampling efforts will
be undertaken to determine whether, for practical
reasons, measurements should instead be taken
at the downstream end of the sample stream seg-
ment (i.e., just upstream of the confluence that
defines the downstream end of the mapped seg-
ment), it is often very difficult to find and then
access mid-segment positions on small streams
with confidence.
Chemical habitat measurements and their .associ-
ated water samples on each sample stream seg-
ment will be collected at a single well-mixed
location in mid-channel.
For physical habitat measurements and biological
sampling, a field sampling reach 30 channel-
widths long will be defined midway between the
upstream and downstream ends of each sample
stream segment. For example, a field sampling
reach approximately 100 m long would be identi-
fied on a typical second-order stream channel
segment 3 km long that averages about 3 m wide.
Similarly, the field sampling reach, within a 5-km
mapped segment of a major river 50 m wide
would be 1,500 km long. Adjusting the field
sampling reach length to the size of the stream
allows sampling of a segment long enough to
incorporate the morphometric and biological
variability within the mapped stream segment.
Physical habitat measurements will evaluate
stream channel morphometry, substrate, hydraulic
retention, habitat complexity, and riparian
interactions over the full length of the field
sampling reach. Biological samples will be
collected separately from major habitat types
within the same portion of the stream segment.
For three rough ranges of stream size, the types
of habitats to be sampled are as follows:
• Small (wadable) streams
- Main channel riffles
- Main channel pools
- Peripheral backwaters
- Side channels
123
-------
• Large streams (nonwadable midstream)
- Main channel riffle accessed by wading
from bank
- Main channel pool accessed by wading
from bank
- Peripheral backwaters
- Side channels
• Rivers
- Natural cutbank, or periphery of pool
- Natural bank on depositional side of
bend, or periphery of riffle
- Bank revetments
- Navigable sloughs, backwaters, and
side channels
- WadaWe backwaters
Sample collections for fish assemblages within
major habitat types will remain separate during
pilot studies. For macroinvertebrates and peri-
phyton, sampling will be replicated (n=5) within
each habitat type. These replicate benthic
samples may be combined before analysis for
most streams, but will be analyzed separately to
evaluate heterogeneity within habitat type for a
subset of 20 to 30 streams within each region.
Pilot research will be used to determine how to
Incorporate fish, macroinvertebrate, and peri-
phyton assemblage data from these different
habitat types Into metrics that index whole-stream
condition, diversity, and biointegrity.
6.5 CONCLUSIONS
The field sampling design described in Sections
6.3 and 6.4 defines the specific places and times
at which EMAP indicator measurements will be
made within each sample unit (lake, reservoir,
stream segment, river segment). In earlier sec-
tions of this report, we define the index concept
(4.0 and 6.1) and qualitatively and quantitatively
explore the accuracy, precision, and bias of indi-
vidual site measurements in a number of synoptic
applications of this concept (5.2.2). Using statis-
tical theory and examining simulations in the con-
text of EMAP goals, we estimate the effect of indi-
cator measurement on the expected precision
and utility of estimates of population status (5.3 to
5.5).
Our basic design for estimating population status
is to sample a statistical subset of surface waters
(Section 3.0) once in the late summer season.
Preliminary data discussed in Section 6.2 suggest
that for most EMAP indicators this sampling fre-
quency should be adequate to meet the goals of
EMAP in most regions, if some systematic spatial
replication within sample sites isi employed.
However, the precision of EMAP indicator meas-
urements will be evaluated through temporal repli-
cation on subsets of 20 to 30 sample lakes and
streams in each region.
Physical habitat measurements and resulting
indices of physical habitat quality will be designed
to be representative of each sample lake and
stream and its associated riparian zone.
Water and sediment samples from lakes and
reservoirs will be collected from a mid-lake
location judged to approximate the deepest point
of the lake. In rivers and streams, water samples
will be taken at a single well-mixed mid-channel
location in the middle of each mapped stream
segment. Pilot study spatial replications within
sample units will re-evaluate the suitability of
these mid-lake and mid-reach index locations.
Fish, macroinvertebrates, and periphyton will be
collected separately from major habitats within
each lake or stream. Macroinvertebrate and peri-
phyton sampling will also be replicated within
habitat types. These benthic replicates will be
analyzed separately for a subset of sample sites,
and the remainder may be combined before iden-
tification and counting. Pilot research will be
used to re-assess the adequacy of temporal and
spatial sampling. Pilot research will also be used
to incorporate surficial sediment diatom informa-
tion, and habitat-specific information on fish,
macroinvertebrate, and periphyton assemblages
into metrics that index whole-stream or whole-lake
condition, diversity, and biointegrity. The
challenge is to develop biological metrics that
quantitatively index the changes in communities
resulting from changes in habitat availability, as
well as other exposures and stresses.
124
-------
7.0 LOGISTICS APPROACH
This section describes the logistics approach for
implementing EMAP-Surface Waters. It includes
a summary of requirements for the logistics plans
(Section 7.1), a discussion of the major logistics
issues (Section 7.2), a field operations scenario
for lakes and streams (Sections 7.3 and 7.4), and
a proposed organization structure (Section 7.5).
7.1 LOGISTICS IMPLEMENTATION COMPO-
NENTS
Implementing the EMAP Surface Waters program
will require detailed, comprehensive logistics
planning. Logistics considerations include coor-
dination and oversight of all implementation
support activities (access permission, procure-
ment, etc.) and the actual data collection
activities. A logistics plan must be developed
before the start of field activities to assure that
program goals will be met. Regional logistics
plans will be updated annually. The logistics plan
will address all, elements given in Table 7-1 as
specified by U.S. EPA (1990).
ELEMENT 1. Overview of Logistics Activities:
Summarize the types of activities required to
complete the project. Maintain a timeline or Gantt
chart showing all critical path milestones (e.g.,
project design, indicator selection, site selection,
access permission, reconnaissance, procurement,
methods selection, development of standard oper-
ating procedures, and resolution of specific qual-
ity assurance [QA] issues). Show required deliv-
erable products such as plans, manuals, and
reports. Provide logistics budget summaries.
ELEMENT 2. Staffing and Personnel Require-
ments: Describe the number of personnel and
the organizational structure necessary to accom-
plish project objectives. Define who is respon-
sible for staffing and interagency and teaming
mechanisms. Consider work schedules to deter-
mine whether extra positions should be created or
whether existing personnel should work overtime.
Create a contingency plan for replacing staff
members when necessary. Identify key personnel
and provide plans for retaining them.
ELEMENTS. Communications: Address commu-
nications among field crews, laboratory crews,
and supervisory personnel, and between EMAP
participants and any local organizations who
should be informed of EMAP field activities.
Include plans for tracking samples, data, crews,
and equipment and supplies. Discuss how field
crews should interact with the public or with the
media. Explain how approved changes in stan-
dard operating procedures will be documented
and communicated for implementation.
Table 7-1. EMAP Logistical Elements for
Implementation of Surface Waters
Monitoring Programs
1. Overview of logistical activities
2. Staffing
3. Communications
4. Sampling schedule
5. Site access
6. Reconnaissance
7. Waste disposal plan
8. Safety plan
9. Procurement and inventory control
10. Training .
11. Field operations
12. Laboratory operations
13. Data management activities
14. Quality assurance
15. Logistics review/recommendations
ELEMENT 4. Sampling Schedule: Based on proj-
ect, indicator, and statistical design or other
program requirements, devise an efficient sched-
ule for field activities. Consider geographical
sampling windows within geographical areas and
other factors such as climate and site access
constraints.
ELEMENTS. Site Access: Address issues related
to gaining access to sampling sites, including
scientific collection permits, if required. Develop
a list of local contacts to discern property owner-
ship, jurisdiction, and the best site access
methods. Address plans to obtain appropriate
access permission and applicable collection
permits. Consider how to coordinate activities in
the same area of more than one resource task
group. Discuss ways to arrange long-term access
125
-------
rights, track changes in ownership of private sites
and management of public sites, notify owners
and managers before revisiting the sites for future
monitoring, and provide contingency plans in
case of future failure to obtain re-access per-
mission.
ELEMENTS. Reconnaissance: Define criteria for
selecting base operation sites (considering per-
sonnel and technical support requirements),
determining geographical location with respect to
sampling sites, and defining time constraints
imposed by sampling design or climate: Samp-
ling sites identified as having potentially difficult
physical or legal access should be visited during
field reconnaissance. Additional resources
needed for sampling should be identified if the
access problem is due to physical conditions. If
the access problem is legal, one last attempt
should be made to obtain permission to sample.
ELEMENT 7. Waste Disposal Plan: Explain how
chemical and biological wastes will be stored,
transported, and disposed of safely and legally.
Address what permits will be needed for storage,
transport, and disposal of wastes.
ELEMENT 8. Safety Plan: Discuss how emer-
gency situations will be evaluated and handled.
Determine emergency contact personnel and
what emergency services will be available in the
field. Explain what procedures will be used to
initiate search and rescue operations. List the
training or other preventive measures required to
conduct field operations safely. Indicate how this
field safety plan will be developed in conjunction
with laboratory, processing, and materials
handling safety plans.
ELEMENTS. Procurement and Inventory Control:
Identify equipment, supply, inventory control and
resupply, and services requirements of the field
program and the processes by which they will be
acquired and maintained. Determine where back-
up equipment will be stored and how sites will be
resupplied. Consider shipping regulations,
especially for chemical and biological materials.
Determine what analytical or other services will be
needed and the best mechanisms for acquiring
them. A procurement schedule should be pro-
vided for all items.
ELEMENT 10. Training Program: Describe who
will prepare, review, and revise the field training
and operations manual and the procedures for
field measurements, sampling, sample handling,
shipment, data recording, quality control, safety,
waste disposal, and communications. Outline a
schedule for the completion of these items.
Describe training needs and identify who-will
conduct and review training. Address how per-
sonnel will be evaluated to ensure competency.
ELEMENT 11. Field Operations: Indicate :t'h'e
organizations that wijl perform each of the" daily
field activities. Describe how and when the daily
field activities will be performed. Discuss arid
schedule the major events within field operations
(i.e., mobilization, demobilization, and phase
changes in sampling activities). Consider con-
tingencies such as back-up personnel in the event
of sickness. Require real-time evaluation to
identify and resolve problems.
ELEMENT 12. Laboratory Operations: Indicate
what organizations will be responsible for each
type of sample preparation or analysis and for
formulating each laboratory operations manual.
If EPA conducts the activities directly, provide a
development plan for providing appropriate lab-
oratory facilities.
ELEMENT 13. Information Management:
Describe any data management activities that
might be affected directly by field operations.
Establish guidelines for the timely and responsive
transferral of information from field personnel to
data managers. Indicate the groups that will be
responsible for preparing and reviewing field data
forms; provide a schedule for the completion of
these forms. Develop a schedule for completion
of the information management plan by the infor-
mation management group.
ELEMENT 14. Quality Assurance: Describe who
will provide input to the QA plan on field samp-
ling, sample handling and preparation, sample
shipment, .sample disposition, and data manage-
ment. A schedule for completing the QA plan
should be provided to the logistics team and
included in the logistics plan. QA activities should
be coordinated with other resource groups using
similar methods. This effort should identify
common methods and standards when possible.
ELEMENT 15. Logistics Review and Recommen-
dations: For each year of study within each
resource group, summarize logisitics activities.
Discuss how personnel will be debriefed to
identify and resolve problems. Discuss pilot
126
-------
studies and associated methods evaluation
experiments; present logistics data summaries
within the full-scale project.
Field activities will start in 1991 with a lake pilot
program in the Northeast. Additional regions will
be phased into the program in each of the follow-
ing years. We plan to begin a stream pilot in the
West in 1992. The program will be fully Imple-
mented across the nation by 1995 for lakes and
by 1996 for streams. A proposed regional imple-
mentation schedule for EMAP-Surface Waters is
given in Section 11.
7.2 LOGISTICS ISSUES
The -complexity of this program poses a number
of logistics issues; overlooking or ignoring
apparently minor issues or details will eventually
jeopardize the success of the program. These
issues will be addressed fully in each of the
regional logistics plans before the implementation
of field activities. This can be accomplished only
through long-range planning and coordination. A
brief discussion of the major issues (staffing,
access, and data confidentiality) is provided in the
following sections.
7.2.1 Staffing
Due to the kind of field data needed for indicator
evaluation (Section 4.0), field personnel will
require a high degree of expertise. They must
have knowledge of fish and macroinyertebrate
taxonomy, field sampling methods, and sample
handling. Finding personnel with expertise in
these areas will be a major challenge.
Field personnel will be required to undergo an
intensive training program. Training assures that
protocols are understood and strictly adhered to,
and provides consistency across the program.
Training is expensive, and these costs can be
reduced in the long-term only by retaining staff.
Therefore, retaining key personnel during nonfield
seasons is critical to program continuity and cost
effectiveness. We may be able to achieve this by
employing field personnel for other EMAP tasks
(e.g., assessment, integration, reporting), but it
will place additional requirements on recruitment
of personnel.
Various state resource agencies, EPA regional
offices, other federal agencies, and universities
have large pools of experienced personnel. Long-
term agreements with these agencies and institu-
tions to provide key personnel during the field
season may also be a solution to this issue. To
accomplish this, EMAP will have to demonstrate
its utility to the States and other agencies by
providing additional data and information addres-
sing their problems. A concerted effort to inform
these agencies of the goals and objectives of
EMAP, and getting these agencies involved in the
early planning phases of the programs are initial
steps to be taken.
7.2.2 Access
Obtaining access information and permission to
visit sampling sites is a difficult and time-
consuming task. If land is publicly owned,
approval must be obtained from the appropriate
authority and permission may be conditional (e.g.,
upon the use of nonmotorized transport). Contin-
gency plans for these conditions will have to be
developed. If land is owned privately, each
landowner will have to be contacted, and written
access permission will have to be obtained. Col-
lection permits will have to be obtained from the
states regardless of land ownership. These state
collection permits are usually very specific relative
to site location, numbers and kinds of specimens,
and identification of field personnel. As a result,
these permits are difficult and time-consuming to
obtain.
Gaining access permission and knowledge of
access routes will require reconnaissance. The
amount of time devoted to sampling a lake or
stream site is dependent upon the physical
access conditions. Some sites may be accessible
only by foot. Each lake or stream identified as
having difficult access should be visited prior to
sampling to determine how sampling crews and
gear can be transported to the sampling site and
how samples can be transported adequately.
Although it sounds like an overwhelming task,
access permission and all that it implies can be
accomplished. The logistics team for EMAP-
Surface Waters was responsible for carrying out
the National Surface Water Survey (NSWS) of the
National Acid Precipitation Assessment Program
(NAPAP) and is an experienced and accom-
plished team.
127
-------
7.2.3 Data Confidentiality
Data confidentiality Is an Issue of particular
concern to EMAP. Many landowners may be
reluctant to allow access to lakes or streams from
their property because they fear regulatory and
enforcement actions. Access is not a design con-
straint, and any denials by landowners could
affect population estimates (see Section 3.0).
Cooperating agencies within the Department of
Agriculture often conduct field programs under an
agreement of confidentiality with landowners.
EMAP data may have to be aggregated in such a
way that Individuals cannot be identified to assure
landowners and cooperating agencies that site-
specific data will not be used against their inter-
ests. Agreeing to withhold certain information,
however, is In direct conflict with the Freedom of
Information Act and EPA's policy on data'confi-
dentiality. This Issue will have to be resolved
soon. The EPA Office of General Counsel is cur-
rently being consulted in this matter.
7.3 FIELD OPERATION SCENARIO: LAKES
We present the following field operation scenario
to demonstrate that the proposed field activities
are logistically feasible within the allotted time
frame. This scenario is only one of many that
could be developed at this time. It is strictly
hypothetical and does not necessarily include all
proposed Indicator parameters or the order in
which activities would take place when the pro-
gram Is actually implemented. Indicators are
being evaluated and developed for pilot pro-
grams; the actual protocols will be solidified later,
as the program evolves.
7.3.1 General Logistics Scenario
The scenario described here and the associated
support activities are based on experience gained
during the NSWS lake chemistry surveys (Morris
et al., 1986; Bonoff and Groeger, 1987) and the
Upper Peninsula of Michigan lake acidity fish
survey (Baker et al., 1990). The scenario is based
on the parameters listed in Table 7-2 and on the
following 11 assumptions.
1. The index period will be July and August:
a sampling window of about nine weeks.
2. Approximately 800 Tier II sites will be
sampled per year, and it will take 4 years
to sample all Tier II sites.
3. Site selection is completely random and
does not consider site access. ,
4. The majority of lakes sampled will be
smaller than 20 ha (see Section 3.0) and
will be accessible by road. Assumptions
concerning accessibility are based on the
Eastern Lakes Survey (ELS).
5. Distance between sites will be one-fourth
the density of the Tier II sites, or about
150 miles.
6. Small, motorized boats will be the primary
sampling platform.
7. Fishing gear will be set out overnight; two
days will be required at each site,
8. Four-wheel drive vehicles will be used for
site access and each sampling team will
need a second vehicle for logistics
support.
9. Field mobile laboratories will not be used
and there will be a minimum of sample
preparation in the field.
10. Samples requiring immediate laboratory
analysis will be shipped to the appropriate
laboratory by overnight courier the day
after collection.
11. A field crew will consist of five people:
four for field sampling activities and one
, for logistics support.
Table 7-2. Proposed General Field Measure-
ment Variables for Lakes
Fish assemblage and gross pathology
Macroinvertebrate assemblage
Diatom assemblage from sediment cores
Macrophyte cover
Habitat quality
In situ water quality '>
Chlorophyll-a/Suspendecl solids
Nutrients
Sediment Chemistry/Toxicity
128
-------
Based on these assumptions, a field crew of four
people will be required to sample a lake over a
two-day period. Transit time between sites will
require most of one day; with one day off, a field
crew could sample two lakes per week. Larger
lakes (> 20 ha) and lakes with difficult access will
require additional time and/or staff. A total of 40
to 50 field crews will be required to sample all 800
lakes within the index period. Allowance for
downtime due to weather and other factors will
have to be considered in determining the actual
number of field crews. Each field crew will be
responsible for sampling 15 to 20 sites in an area
approximately the size of Oregon. To organize
and coordinate the activities of 40 to 50 field
crews, 5 to 10 regional logistics centers will be
developed across the nation. These regional
logistics centers (see Section 7.4) could possibly
be integrated to support the field activities of the
other EMAP resource groups (Wetlands, Near
Coastal, Forests, Agroecosystems, and Arid
Lands).
7.3.2 Daily Activities Scenario
The five-member field crew will be housed at a
motel within 50 miles of the sampling site. Addi-
tional space will be obtained at the motel or some
other location to store equipment and supplies
and to calibrate instruments. Crews will begin
each morning's activities by calibrating the
instruments and checking to see that all neces-
sary equipment and supplies are loaded into the
vehicle. The four-member sampling crew will then
depart for the lake. The fifth crew member (base
site crew member) will remain at the motel base
site; this individual will provide logistics support
(picking up equipment and supplies, shipping and
tracking samples, transmitting data, contacting
landowners, etc.) and will provide a communica-
tions link with program management. These base
site activities could not be completed by one of
the sampling crew members because of the
demanding schedule, as discussed below. The
base site crew member, on a limited basis, could
also be a backup sampler if someone were sick
or if additional personnel were needed to sample
the lake. Figure 7-1 and the following discussion
summarize daily activities for the sampling crew.
On day one, the field crew will first verify that they
are at the appropriate lake. Lake verification will
be based on landscape features and topographic
maps, and by the Global Positioning System
(GPS) once this system is fully implemented.
After verifying the identification of the lake, a
calibration check will be performed on all instru-
ments (e.g., the in situ water quality analyzer) to
be used that day. The sampling crew will then
split into a two-member boat crew and a two-
member shore crew. The fishing gear (nets
and/or traps) need to be deployed as early as
possible, and the shore crew will start to prepare
the gear and other equipment immediately. At
this time, the boat crew will conduct a habitat
evaluation, which will include sonar transects for
depth and structure and in situ water quality
profiles that will determine thermal stratification
and anoxic condition. This information will be
used to determine where to deploy the fishing
gear. The boat crew will then pick up the fishing
gear from the shore crew and deploy it. This
activity will be followed by the collection of
profundal benthic invertebrates (dredge sample)
and sediment diatoms (core sample) at multiple
locations. During this time, the shore crew will
conduct a shoreline habitat evaluation and collect
and sort littoral benthic invertebrates. After
collecting the profundal benthic samples, the boat
crew will join the shore crew and help process
and preserve samples. During all of these opera-
tions, data will be entered into field data loggers.
These systems will have programmed QA checks
for parameter ranges and completeness. Backup
hardcopy forms and notes will be filled out to
assure data are not lost in the event of data
logger failure. The field crew will check the data
loggers, forms, and sample labels to assure com-
pleteness before returning to the motel base site.
At the motel, the base site crew member will
debrief the sampling crews and check the data
loggers, forms, and sample labels, and the con-
dition of the samples. The preserved benthic
invertebrate and sediment diatom samples will
then be prepared for shipment. During this
period, the sampling crew will prepare equipment
and supplies for the next day. The base site crew
member will then download the data from the field
data loggers onto a portable computer.
On day two, instruments will be calibrated at the
motel and a calibration check will be performed at
the lake as described for day one. On the lake,
the boat crew will first take various in situ meas-
urements (temperature, dissolved oxygen, pH,
conductivity, Secchi depth, etc.) and collect water
and sediment samples at the deepest point in the
lake. These samples will be transferred to the
shore crew to be processed and preserved. After
129
-------
Dayf
Sampling Crew Members
Site Access/Verification
Instalment Calibration Check
(7-8 am)
Boat Crew Members
Habitat Evaluation
• Sonar Transects
• In Situ Temp/DO
Profiles
(8-1 Oam)
Shore Crew Members
Fish Sampling Gear
Preparation (8-9 am)
Sampling Gear
Pickup
Deploy Fishing Gear
(10-1 2pm)
•^
r
Shoreline Habitat Evaluation
Collect Littoral Benthic
Invertebrates
(9-12^
Collect Profundal Benthic
Invertebrate and Diatom
Samples
(12-2pm)
Sort Benthic Samples
Process Sediment Diatom
Cores
(12-3pm)
Sample Transfer
Complete Sample Preparation
Check Data Loggers, Forms, and Sample Labels
Return to Base Site
(4-5 pm)
Base Site Crew Member
Debriefs Sampling Crew
Checks Samples, Labels, and
Forms
Download Data Loggers
Store Samples for Shipping
(5-7 pm)
Sampling Crew Members
Prepare for Day 2
Sampling
(5-6 pm)
Figure 7-1. Flowchart of potential EMAP-Surface Waters daily field activity (page 1 of ?.)„
130
-------
Day 2
Sampling Crew Members
Site Access
Instrument Calibration Check
(7-8 am)
Boat Crew Members
Deepest Point of Lake
• In Situ DO Temperature
Profiles
• Collect Water and Sediment
Samples
(8-1 Oam)
Shore Crew Member
Sample Transfer
Habitat Evaluation
• Macrophyte Mapping
• Blue-Green Algal Cover
(10-12pm)
Filter and Process
Samples
(10-12pm)
Collect Fish
- Minnow Traps
- Retrieve Nets/Traps
- ID, Count, Measure, Weigh Fish
Examine Fish for Gross Pathology
Keep Target Species
- Age & Growth
- Tissue Analyses
(12-2pm)
Process Fish Samples and Prepare Samples
Check Data Loggers, Forms, and Samples
(2-4 pm)
Return to Base Site
(4-5 pm)
Base Crew Members
Debrief Field Crew Members
Check Samples, Labels,
and Forms
Download Data Loggers
Pack Samples for Shipping
(5-7 pm)
Field Crew Members
Check all Equipment and
Supplies
(5-6 pm)
Figure 7-1. FLowchart of potential EMAP-Surface Waters daily field activity (page 2 of 2).
131
-------
completing these activities, the boat and shore
crews will rejoin to retrieve the fish nets and traps
set out the previous day and to seine along the
appropriate beaches. Members of the sampling
crew will also snorkel along the shore of the lake
to determine if fish are present and to identify
additional fish species that may not have been
caught by the sampling gear. Collected fish will
be examined for gross pathology. Target speci-
mens will be held and processed on shore for
age, growth, and tissue analyses. Before leaving
the lake, the field crew will check the data
loggers, forms, and sample labels for complete-
ness as described for day one.
At the motel, the base site crew member will
debrief the sampling crews and again check the
data loggers, forms, and labels, and the condition
of the samples, the base site crew member will
then download the field data loggers and pack
the samples for shipping, as described for day
one.
On day three, the field crew will move to a new
motel base site close to the next site to be
sampled. The base site member will transmit data
electronically via a phone modem to the task
group Information center. After completing this
activity, the base site member will take the
samples to the overnight courier and pick up
additional supplies and equipment. The specific
location of the courier will have been predeter-
mined during reconnaissance and verified through
daily communications with program management.
At the new base site, the sampling crew will go
through a complete maintenance check of all the
equipment.
7.4 FIELD OPERATIONS SCENARIO:
STREAMS
Planning, efforts for streams have not been con-
ducted at the same level as lakes because of a
later Implementation schedule; therefore, we do
not present a detailed field operations scenario
here. A general scenario is presented, based on
experience from the NSWS stream surveys
(Knapp et al., 1987; Hagley et al., 1989) and on
the following six assumptions.
1. Most of the parameters given in Table 7-2,
except Chlorophyll-a, will be measured in
streams.
2. Assumptions 1, 2, 3, 4, 7, 8, and 10
presented in Section 7.3.1 are valid for
streams.
3. The majority of sites will not be immedi-
ately adjacent to a road but will be within a
few hundred meters of a road. Assump-
tions concerning accessibility are based on
the NSWS National Stream Survey (NSS).
4. The majority of sites will be first, second, or
possibly third order streams. A different
sampling frame will be used for fourth and
fifth order streams.
5. Only one daywill.be required to sample a
site.
6. A field crew will consist of four to five
people.
Based on these assumptions, a field crew will
sample three sites a week, with a transit time of
one day between sites. A total of 25 to 30 field
crews will be required to sample all 800 streams.
Moving base sites will be used, as described for
lakes. Again, 5 to 10 regional logistics centers
will have to be developed across the nation to
organize and coordinate the activities of the field
crews. These centers will be identical to those
developed for lakes and possibly to those of the
other EMAP resource groups.
7.5 ORGANIZATIONAL STRUCTURE
Coordinating the logistics activities (staffing,
training, deployment, tracking, etc.) of 50 or more
Surface Water teams across the nation will be
very difficult. As mentioned in sections 7.3 and
7.4, regionalizing these logistics activities into
various centers will be the most effective mech-
anism for conducting EMAP field operations. The
EPA regional offices can provide the initial struc-
ture for these logistics centers. The EPA regions
have firsthand knowledge of the environmental
conditions within their respective regions and will
have a major role within EMAP, part of which
could be in logistics. The EPA regions represent
the Agency's primary contact with the states, and
are working with the states at the program level.
Securing cooperation from the states for EMAP is
essential because of requirements regarding
collection permits and access permission. The
EPA regions can provide the crucial link in obtain-
ing state cooperation for EMAP. The EPA regions
132
-------
and states also have highly experienced field
personnel, and their participation in EMAP field
operations would be extremely beneficial. As key
personnel directing field team activities year after
year, they will provide the program with the con-
tinuity critical to EMAP field activities.
Surface Waters is only one of seven EMAP
resource groups that will be conducting annual
field monitoring surveys. Other resource groups
are Wetlands, Great Lakes, Near Coastal, Agro-
ecosystem, Forest, and Arid Lands. Although the
various resource groups will not necessarily be
sampling at the same location or at the same
time, logistics efforts should be integrated as
much as possible to reduce costs. Shared
regional logistics centers with permanent ware-
house facilities and staff will aid in this integration.
The long-term success of EMAP is dependent on
the development of an interagency program with
common goals for the monitoring of the ecolog-
ical condition of the environment. Surface water
monitoring alone could involve numerous agen-
cies within the Department of the Interior (e.g.,
U.S. Geological Survey, U.S. Fish and Wildlife
Service, National Park Service, U.S. Bureau of
Reclamation), the Department of Agriculture (e.g.,
U.S. Forest Service), and the Department of
Defense (e.g., U.S. Army Corps of Engineers). As
EMAP evolves into an interagency program,
agreements between agencies will have to be
established to define responsibilities. As with the
EPA regions and states, these agencies have
highly experienced field personnel, and it is
anticipated that personnel from these agencies
will participate in both field activities and the
regional logistic centers.
The Boise Interagency Fire Center, which may
serve as a model for future EMAP regional logistic
centers, is the national logistical support center
responsible for coordinating and dispatching the
closest suitable personnel, equipment, and aircraft
for wild fires that exceed the capabilities of local
and regional resources of land management
agencies. This center is the result of an inter-
agency program, with agreements between the
U.S. Bureau of Land Management, the U.S.
Bureau of Indian Affairs, the U.S. Forest Service,
the National Park Service, the National Weather
Service, and the U.S. Fish and Wildlife Service.
The U.S. Bureau of Land Management manages
the land, facilities, and changes as a key facet of
the center and is host to. the other five agencies.
The objectives of the Boise Interagency Fire
Center are:
• Interagency programs and services are
developed through coordination and coop-
eration.
• Effective use of interagency programs and
services by cooperating agencies.
• Equitable cost sharing of interagency
programs and services.
Similar arrangements need to be considered for
interagency EMAP logistics centers.
133
-------
-------
8.0 QUALITY ASSURANCE PROGRAM
A comprehensive quality assurance (QA) program
will be implemented to ensure that all required
data collected within EMAP can be used to
achieve the objectives of the program. Quality
assurance programs are mandated by the EPA for
all environmental data acquisition activities it
sponsors or in which it participates (Stanley and
Verner 1985). The EMAP Quality Assurance Pro-
gram Plan (Einhaus et al. 1990) documents the
overall policies, organization objectives, and
functional responsibilities designed to achieve
data quality goals for EMAP activities. The QA
program for EMAP-Surface Waters will be
designed to ensure that the type, amount, and
quality of data collected will be in accordance
with the data quality objectives (Section 2.0)
established for the program. The design of the
overall QA program for the Surface Waters com-
ponent must consider two major constraints: a
broad range in scales of implementation (spatial
and temporal) of data collection activities, and the
participation of a potentially large number of
individuals and organizations in data collection
activities.
The major focus of data collection (and thus QA)
will be the regional-scale sampling efforts
conducted annually to monitor and assess status
and trends in ecological condition. In addition,
special studies on smaller areas (regional, sub-
regional, or local) will be conducted to address
specific questions (e.g., acidification or eutro-
phication), to investigate possible cause and
effect relationships between ecological compo-
nents (or processes) and changes in ecosystem
condition, or to develop and evaluate appropriate
ecological indicators. Each of these levels of
implementation will require QA programs that
differ in focus and intensity. The QA programs
should be consistent with each other, however,
and with EMAP's overall approach to QA. Such
consistency will increase the level of confidence
with which data user compare data from different
though related programs within EMAP-Surface
Waters and use data from different EMAP
components.
We expect the participation of federal and state
agencies, contractors, private consultants,
production-mode analytical laboratories, and
scientists from universities or other research
institutions. Over the long lifetime of the program,
the list of participants will undoubtedly change.
Groups will have different levels of expertise in the
principles and practices of QA. Existing moni-
toring programs considered for integration into
the EMAP framework may have QA requirements
that are initially incompatible with those
established for the Surface Waters component, or
EMAP as a whole. Differences in sampling and
analytical methodology, whether among partici-
pating groups, among regions, or as a result of
new technologies over the life of EMAP, must be
monitored and assessed in order to quantify and
minimize their impact on the interpretation of the
observed status and trends in ecological condi-
tion.
The overall QA program implemented for EMAP-
Surface Waters will be based on a philosophy of
guidance, assistance, and commitment to
improvement. It will not be based on the
"enforcement" approach required for regulatory or
compliance monitoring programs. In developing
information of known quality, consistent with data
quality objectives, emphasis will be on con-
sistency in implementation, quality control (QC),
prompt corrective action, and continuous
improvement. This approach will identify and
correct problems as soon as possible, to mini-
mize their impact on data quality. The QA staff
within the Surface Waters resource group will
receive training in carrying out their responsi-
bilities. Such training should include, but is not
limited to, improving quality and productivity,
preparing and reviewing QA-related plans and
reports, and conducting and reporting audit
activities. The QA staff will provide guidance,
training, technical support, and tools (e.g.,
performance audit materials, QA documentation)
to all participants to implement QA programs that
are consistent with the data quality requirements
of the Surface Waters component, and EMAP as
a whole. Although the emphasis of the QA
program is to provide guidance and support, the
management and QA staff must also have the
means to deal with instances of poor perfor-
mance, when necessary, to avoid compromising
data quality.
The following subsections outline the general
approaches, conceptual rationale, and guidelines
proposed for designing and implementing the
overall QA program for EMAP-Surface Waters.
135
-------
Section 8.1 presents aspects of data quality
requirements for individual variables, metrics, and
Indicators. Section 8.2 outlines the organizational
structure and staffing requirements of the QA
program. Then Section 8.3 describes the docu-
mentation that will be required as part of the QA
program. Section 8.4 presents the guidelines and
proposed approaches for incorporating QC pro-
cedures Into the data acquisition process.
Section 8.5 explains the proposed procedures
and guidance for the 'review, verification, and
validation of data. Section 8.6 details the
assessment of data quality. Finally, Section 8.7
summarizes reporting activities and products
related to the QA program.
8.1 DATA QUALITY REQUIREMENTS
In all data collection activities, data quality
requirements will be specified in five areas:
precision, bias, comparability, completeness, and
representativeness (Stanley and Verner 1985,
Smith etal. 1988). In addition (when appropriate),
minimum tolerable background levels of chemical
constituents will be established. These levels
represent the maximum concentrations of con-
stituents that can be contributed by the sample
collection and measurement process. Method
detection limits will also be specified, monitored,
and assessed (Section 8.1.6).
Ideally, data quality requirements will be based on
the overall data quality objectives (Section 2). In
some cases, the requirements will be qualitative,
In others, quantitative. Data quality requirements
will also be reviewed periodically throughout the
program, and revised as necessary in response to
Improved capability, additional knowledge, or
technological or resource limitations.
Data quality requirements, constrained by sam-
pling, measurement, or logistical considerations,
will determine the appropriate methodology.
Criteria for Initial selection of appropriate chemical
and biological methodologies are presented in
Table 8-1. These criteria were chosen primarily
by means of the approach advocated by Hunt
and Wilson (1986), with the addition of criteria
relevant to the design of a data management
system (Section 9).
8.1.1 Precision and Bias
Precision and bias are estimates of random and
systematic error in a measurement process
(Kirchner 1983, Hunt and Wilson 1986). Collec-
tively, they provide an estimate of the total error
or uncertainty associated with an individual
measurement, or set of measurements. In theory,
random and systematic errors can be determined
at any point in a collection and measurement
process. Figure 8-1 summarizes error compo-
nents of interest for EMAP-Surface Waters
(described in Section 4.0). Estimates of the
various error components will be determined
primarily by replicate sampling; such sampling
can be modified to address and control major
sources of variability. The statistical design and
sampling plan should act to minimize systematic
errors in all components except measurement
error (o2meas). Systematic errors in these
components will be minimized by using docu-
mented methodologies and standardized proce-
dures, and evaluated using samples of known
composition that can be subjected to the entire
collection and measurement process. Variance
components of the collection and imeasurement
process (e.g, among analytical laboratories or
among individuals identifying biological speci-
mens), although of less interest to data inter-
pretation, should be estimated periodically so that
QA efforts can be allocated to control major
sources of error.
The precision and bias requirements will be used
to define criteria for monitoring collection and
measurement activities and for maintaining them
in a state of statistical control (i.e., the distribution
of individual measurements have a stable and
predictable distribution over time; Taylor 1987).
Estimates of precision and bias are also neces-
sary to evaluate the other three data quality
indicators (comparability, completeness, and
representativeness).
In general, data from one or more measurements
of variables will be combined (and possibly
transformed or categorized) into metrics; one or
more metrics will be incorporated into an indi-
cator, and one or more indicators will be used to
provide an estimate of the ecological health of a
population of lakes or streams. An example of
the relationship between measurements, metrics,
and indicators, and endpoints appears in Figure
4-1. Random and systematic errors associated
136
-------
Table 8-1. Criteria for Selection of Appropriate Sampling and Analytical (or Measurement)
Methodology8
Criteria
Comments
Range of values of interest
Lowest value of interest (values below
this will probably have uncertainties
> 100%); or, in case of trace-level
analyses, the required limit of
detection.
Maximumtolerable measurementerror
(random and systematic)
Standard reference on which method
is based
Required frequency of sampling or
collection
Collection, analytical, or measurement
constraints
Sampling handling considerations or
measurement conditions
Reporting requirements
Data reporting time
May need to be tailored for different regions or different projects.
Chemical: Used to determine appropriate limit of detection.
Biological: May help define the minimum effort required in obtaining data (e.g.,
at least 100 organisms in a benthic sample are needed for determining species
composition and relative abundances).
Defined on basis of sample collection and measurement only).
Additional error components of interest can be defined in terms of short-term or
long-term (e.g., within-cJay or within-batch versus among-day, among-batch, or
among-group).
Should be expressed in both absolute and relative terms:
Absolute: M = lowest value of interest
Relative: M = some specified percentage of true or most probable value
"Knot value" = Absolute/Relative; represents value at which
absolute error equals relative error.
Describe any required modifications.
Requirements based on sampling plan; used to estimate variance components
of interest.
Site selection criteria, special equipment requirements, use of hazardous
reagents, etc,
Appropriate holding times (e.g., APHA 1989)
Operational: May be based on data reporting time requirement.
Maximum: If greater than operational, point at which sample is no longer
considered representative of conditions at time of collection.
. Containers/preservation techniques
Type of variable (numeric, coded, character, categorical, etc.).
Reporting Units (mg/L, jteq/L, number of individuals, etc.).
Number of significant figures desired, and maximum number of decimal places.
In addition to being used to select method of appropriate sensitivity, information
is needed to design data base, also for collection forms (hard copy or electronic).
Time period between collection and incorporation of validated data into data base
for use in analysis and reporting.
Based on Hunt and Wilson (1986).
137
-------
pop.
Total Population
Variance
02un'rt
CF2total (Total Variance)
<32pop. = Sampling Variance of Population (Extrapolation of sample to population)
Function of sample size, sample variance
a2unit = Sample Variance
Due to differences in index values among sites within a region
cr2corr * Variance due to Correlations Among Sites
Spatial or temporal patterns of sample sites within a region
O2yr = Variance in index value at a site among index periods (years)
O2index = Variance within an index period .
Spatial (e.g., among potential sampling locations during a single visit)
• Temporal (different visits within an index period)
a2meas. - Variance in measurement within a single index sample
Sample collection, processing, and analysis
"Worst-case" estimate would include different teams, different
laboratories, different dates of analysis
Figure 8-1. Hierarchical model of variance components of importance to EMAP-Surface Waters.
138
-------
with these various "composite" values will have to
be evaluated regarding potential impact on data
quality.
8.1.2 Comparability
Comparability is defined as "the confidence with
which one data set can be compared to another"
(Stanley and Verner 1985, Smith et al. 1988). For
EMAP-Surface Waters, there are three levels of
comparability:
1. Short-term: within a single sampling period
(EMAP's basic reporting unit).
2. Intermediate-term: over a complete
sampling cycle (4-5 yr).
3. Long-term: among sampling cycles.
Data from these assessments are needed to com-
pare and interpret surface water data only.
Comparability can be maximized through the use
of standardized methodologies and similar QA
programs; it can be evaluated quantitatively
through estimates of precision and bias, and
through carefully designed method comparison
studies.
We also need to be cognizant of data compara-
bility external to EMAP-Surface Waters. EMAP-
Surface Waters data should be comparable with:
• Data collected by other EMAP ecosystems
groups (e.g., agroecosystems, forests, arid
wetlands).
« Other environmental data sets (e.g., depo-
sition data from NADP or other networks),
and data from existing monitoring pro-
grams being incorporated or integrated
into EMAP-Surface Waters.
• Data that will be collected in the future,
whether as part of EMAP itself, or other
monitoring efforts that may develop in the
future.
In these cases, comparability will need to be
evaluated with respect to the available QA and
QC data. The degree of comparability required
will depend on the intended use of the data (trend
detection, associative analyses, etc.).
8.1.3 Completeness
For EMAP-Surface Waters, the requirements for
completeness will be based on the amount of
data required to make conclusions pertinent to
the program (or project-specific) objectives. For
example, in order to make a conclusive statement
(with an acceptable level of confidence about the
condition of a subpopulation of lakes (or streams)
with respect to biotic integrity, valid data for
indicators related to biotic integrity from a
minimum number of sampled lakes (or streams)
will be required. This minimum number may differ
for other indicators (such as trophic state).
Completeness may also need to be evaluated at
the level of individual samples or measurements,
For example, if an expected taxonomic group is
rnissing from a sample, any metrics or indicators
that included this taxon would be suspect, and
the sampling site could not be utilized in esti-
mating ecological condition with respect to bio-
logical integrity. In the case of chemical samples,
valid data may be available for 99% of the sam-
ples collected in a region, but if the missing or
invalid 1% of the data all come from a group of
sampling sites of interest (e.Q., seepage lakes),
assessments of condition may be compromised.
8.1.4 Representativeness
Representativeness is defined as "the degree to
which the data accurately and precisely represent
a characteristic of a population parameter, a
variation of a property, a process characteristic,
or an operational condition" (Stanley and Verner
1985; Smith et al. 1988). Representativeness can
be affected by problems in any or all of the other
indicators of data quality, as well as by issues
such as the location of a sampling site, the time
of sampling, and the statistical selection of
sampling sites. More specific to a QA program is
the representativeness of samples or procedures
used to control and assess data quality as com-
pared to the range of conditions being sampled.
8.1.5 Tolerable Background Levels
Background is operationally defined as the
amount of contamination due to collection,
handling, processing, and measurement. It is
most relevant to those chemical constituents
present in the environment in very low Concen-
trations. Requirements for tolerable background
limits will be based on the lowest concentration of
interest that is required to assure that repre-
139
-------
sentath/eness or completeness requirements are
not compromised. Careful adherence to sample
collection, handling, and processing protocols will
minimize background levels. Blank samples of
various types will be used to provide estimates of
background levels.
8.1.6 Method Detection Limits for Chemical
Constituents
For chemical constituents, the method detection
limit (MDL) will be defined, monitored, evaluated,
and reported. The MDL is important in evaluating
data quality, providing the data user with an
Indication of the lowest concentrations that were
determined with confidence. The MDL also
assists In determining the capabilities of analytical
laboratories with respect to the analytical
requirements for EMAP-Surface Waters.
For EMAP-Surface Waters, the MDL is defined as
the lowest concentration of a constituent that an
analytical process can distinguish from zero
concentration, with a single measurement, with
99% confidence. Based on the MDL for each
constituent, acceptance criteria will be estab-
lished. Single measurements of a low-concen-
tration standard will be compared against the
acceptance criteria as an indication the MDL is
being achieved for a particular batch of samples.
Repeated measurements of the low-concentration
standard will be used to evaluate the achieved
MDL (including among-batch variability) against
the required MDL value. The achieved MDL will
be estimated as a one-sided 99% confidence
Interval for the set of low-concentration standard
measurements. The achieved MDL values will be
reported with all data quality summaries
associated with EMAP-Surface Waters.
8.2 ORGANIZATION AND STAFFING
REQUIREMENTS
The EMAP QA coordinator, located at EMSL-
Clnclnnatf, has overall responsibility for
Implementing consistent and adequate QA pro-
grams within EMAP as a whole. The QA officer is
responsible for designing and implementing the
QA program for the Surface Waters component.
Table 8-2 summarizes the responsibilities of the
QA coordinator and the Surface Waters QA offi-
cer. The QA officer will be assisted by one or
more coordinators In implementing the large-scale
annual sampling operations. Special projects of
regional scale will have a designated QA repre-
sentative to oversee the QA program for the
project. We anticipate that each localized,
intensive monitoring program will have a desig-
nated on-site QA representative, who will be
supported by the QA officer and regional QA
representatives.
8.3 QUALITY ASSURANCE DOCUMENTATION
Before implementation of field sampling opera-
tions, a number of different documents will be
prepared (or existing documents utilized) as part
of the QA program (Table 8-3).
Primary guidance for implementing the QA pro-
gram will be provided by the EMAP Quality Assur-
ance Program Plan (Einhaus et al. 1990). The
policies, organization, objectives, and functional
activities that pertain specifically to the QA
program for the Surface Waters component will
be detailed in a QA project plan. The Surface
Waters QA project plan will be used as guidance
in preparing QA project plans for special studies,
regional or local in focus. In general, the QA
project plan for any specific data collection
activity will detail the QC and data quality assess-
ment activities (summarized in the following sub-
sections) that will be used to ensure achievement
of required data quality. Additional QA documen-
tation appropriate for specific data collection
activities may include guidance documents or QA
program plans for other federal or state agencies
and facilities, provided they meet or exceed the
requirements set forth in the EMAP QA program
plan and the Surface Waters QA project plan.
Documentation pertaining to the EMAP-Surface
Waters QA program will be reviewed periodically,
and revised as necessary to reflect changes
based on previous performance, or other modifi-
cations to either the QA program or to EMAP in
general. Changes in various aspects of the QA
program should also be incorporated into revision
of standard operating procedures related to
sample collection and measurement.
The general QA project plan for EMAP-Surface
Waters activities will be reviewed by repre-
sentatives of the major participants (e.g., EPA
laboratories) before submission for formal
approval. The existing QA program plan for
EMAP (Einhaus et al. 1990) specifies signature
approval by the Surface Waters QA officer, the
Surface Waters technical director, the associate
140
-------
Table 8-2.
Responsibilities of the EMAP QA Coordinator and Surface Waters QA Officer
QA Coordinator:
• Prepare and revise EMAP QA program plan, and assist in implementation of the overall QA program.
• -Provide guidance to EMAP resource groups in developing data quality objectives (DQOs) and
auditing programs, and for preparing QA project plans and standard operating procedures.
• -.Review and approve EMAP QA documents and reports, including QA project plans, DQO documents,
and resource group QA annual reports and workplans.
• Review QA programs implemented by individual resource groups. ,
• Prepare an overall QA annual report and workplan for EMAP.
• Coordinate EMAP QA-related activities with EPA Agency QA personnel.
Surface Waters QA Officer:
• Prepare the QA project plan and QA annual report and workplan for the Surface Waters resource
group.
• Coordinate QA-related activities within the Surface Waters resource group.
• Facilitate the development of DQOs and methods selection for the Surface Waters resource group.
• Prepare or coordinate the preparation of data quality assessment reports for inclusion into EMAP
Surface Water reports and data bases.
• Conduct audits of data acquisition activities within the Surface Waters resource,group.
• Assisting in the preparation of QA project plans for special research projects conductecl or sponsored
by the Surface Waters resource group.
director for Inland Aquatic Systems, and the
EMAP QA coordinator.
8.4 QUALITY CONTROL GUIDELINES
Quality control is applicable to ail stages of a data
acquisition process, from design through samp-
ling and analysis, data management, and inter-
pretative reporting. For the Surface Waters
component, this process will be similar to that
illustrated in Figure 8-2. Each stage in the
process represents a point at which quality
control measures can be implemented (if neces-
sary or desirable to monitor those aspects that
are most subject to error or inconsistency).
Those stages conducted after the commencement
of field operations also represent points where
assessments of data quality (Section 8.7) can be
made. In some cases, such assessments are
necessary for monitoring sources of error to
optimally allocate control measures among points
in the process where they are most needed.
General activities that will maximize the success
of a QC program include:
• Documentation of procedures related to
design, sampling, measurement, informa-
tion management, data analysis, reporting,
and QA (Section 8.3).
141
-------
Table 8-3. Quality Assurance Related Documentation of EMAP-Surface Waters
• EMAP Quality Assurance Program Plan ;
Describes philosophy and QA policies of EMAP and provides guidance for designing and
Implementing QA programs within EMAP
• Surface Waters Quality Assurance Project Plans
Detail the QC and assessment activities that will be used in the QA program for Surface Waters
• Field Operations Manuals .
Standard operating procedures for sample collection, handling, and processing, collection of field
data, and data management activities (including QA and QC procedures). Also describe other
logistical procedures (e.g., sample shipping, waste disposal, communications, safety) conducted in
the field.
• Analytical Methods Manuals
Standard operating procedures for sample analysis (including QA/QC procedures).
• Quality Assurance Project Plans from EMAP Support Groups
- Landscape characterization • ,
- Information management
• Other QA program plans and appropriate QA project plans for other participating groups (e.g.,
agencies, laboratories, principal investigators). .
• Standardized training programs to ensure
a minimal level of competency in all
aspects of the project.
• A program of calibration and preventative
maintenance for all sampling and analytical
equipment and instrumentation.
• Periodic site visits by knowledgeable
members of the QA or management staffs
to ensure that sampling and measurement
activities are being conducted appropri-
ately, to recommend corrective actions as
necessary, and to assist on-site personnel
with addressing QA-related issues.
Where appropriate, collection and measurement
processes will be monitored through-the use of
frequent QC checks using samples of known
composition, or through replicate measurements.
Control charts will be maintained whenever
possible. Use of these tools allows for rapid
identification and resolution of problems related to
sample collection or measurement, and provides
documentation that the process is being main-
tained in a state of statistical control. Specific
examples of QC activities are provided in the fol-
lowing subsections.
It is important to recognize that the utility of QC
measurements will be constrained, especially in
the field, by the relatively brief index period each
year (about 2.5 months) and by the turnaround
time between collection of samples and subse-
quent analysis (especially for complex analyses
such as organic compounds or fish tissue anal-
yses). Resampling a site because of subsequent
identification of a problem will not be feasible in
most cases (although sample reanalysis will be,
•depending on the amount of sample remaining
and any holding time considerations). It thus
becomes critical to minimize the potential of
142
-------
Statistical Sampling Design
1 T—
Sample Site Selection
(Maps)
Sampling Plan
(Index Sampling
Requirements)
Index Sampling Location
at a Sampling Site
(Field Crews)
i
I
Sample Collection
I
Field Data Collection
and Recording
Sample Handling
and Transport
I
Data Entry, Review,
and Verification
Data Transfer
Sample Proc
(e.g., Sorting Sp
Aliquoting, Pres
t
Sample Trs
to Labora
f
!o?rfnt« ^ Sampling Tracking
SSf — *- l*™al°"
mster
tory
Sample Preparation
and Analysis
^_^_ i
Sample Archival
t
Analytical Data ^
(incl. QA/QC)
f
Data
Verification
t
Data Entry
and Review
Summaries and
Interpretive
Reports
Figure 8-2. Stages in the data acquisition and management process for EMAP-Surface Waters where
data quality can be controlled or assessed. Double-lined boxes represent stages
conducted before implementation of field sampling operations.
143
-------
sampling-related errors through careful attention
to standardized procedures.
8.4.1 Biological Measurements
Quality control activities applicable to the
collection, Identification, and associated attributes
of fish, benthos, diatoms and other biological
samples are summarized in Table 8-4. For col-
lection offish samples, the activities proposed are
based on procedures previously utilized in lakes
for the Biologically Relevant Chemistry Project
(Cuslmano et al. 1990, Baker et ai. 1990) and
other projects, and for streams, the Episodic
Response Project (Peck et al. 1988) and the EPA
Rapid Bioassessment protocols (Plafkin et al.
1989), and in guidelines presently being
developed for use in Canadian monitoring
programs (Hart 1990). For benthos, many of the
activities utilized In streams (Peck et al. 1988,
Plafkin et al. 1989) are Identical (or easily
modified) for lake benthlc sampling. For sediment
core sampling (for diatom assemblages) QC pro-
cedures have been developed and evaluated as
part of the Paleoecological Investigation of Recent
Lake Acidification (PIRLA) project (Charles and
Smol in press). For toxlcity tests, QC procedures
have been developed as part of standardized test
methods (e.g., EPA or ASTM); these can be modi-
fied as necessary to meet the needs of the EMAP-
Surface Waters component.
When possible, some type of control criteria will
be established to ensure an adequate sampling
effort has been conducted at each site to collect
a representative index sample. In cases where
this Is not feasible, some type of replicate
sampling, repeated measurement, or additional
effort sampling program will be conducted at a
subset of sites to provide an estimate of sampling
efficiency or precision. Such estimates can sub-
sequently be used to develop control criteria as
the program continues. Repeated sampling and
measurement strategies will be designed in con-
Junction with the sample replication scenarios
presented In Section 4.0. Repeated or indepen-
dent checks on sample processing and taxo-
nomlc Identifications will be conducted on a
subset of samples collected. Reference collec-
tions of biological specimens will be developed
and maintained by participating groups during the
course of the program. Such collections will be
eventually be archived at a permanent collection
facility (e.g., a museum) for future use (Section
8.4.4).
Several issues pertaining to QC of biological
measurements within EMAP-Surface Waters
remain, including developing efficient techniques
to identify, assess, and monitor systematic errors
associated with sample collection, especially
fishes in lakes. As presented in Section 5.4,
different types of sampling gear are required in
lakes to obtain a representative sample of the fish
assemblage. Additional sampling programs to
estimate and control biases in collection will be
costly and complex to conduct for all the different
types of gear proposed. We must also assure
that sampling and associated QA activities do not
result in a reduction in population sizes more
severe than would result from any potential
environmental stressor. Testing and reporting
conditions for bioassay tests will need to be
precisely specified, in terms of test organisms,
holding conditions, reference conditions, and
reporting procedures, to ensure consistency in
performance among all participating laboratories.
Changes in taxonomic relationships and nomen-
clature will need to be tracked as new studies are
published that revise taxonomic relationships or
nomenclature, or report on new species.
8.4.2 Chemical Measurements
Quality control activities for chemical measure-
ments are well documented (e.g., Hunt and
Wilson 1986, Taylor 1988). Table 8-5 summarizes
these activities. Quality control procedures for
dilute surface water samples have been used for
large-scale studies during the EPA's Aquatic
Effects Research Program (e.g., Silverstein et al.
1988 for lakes, Cougan et al. 1988 for streams),
and refined for smaller scale operations for the
Episodic Response Project (Peck et al. 1988).
Specialized collection and handling procedures
may be required for certain types of water
samples (e.g., those being analyzed for organic
constituents) to minimize contamination and,
prevent changes in composition between collec-
tion and analysis. Use of field blank samples will
monitor possible contamination during collection
and analysis.
In the laboratory, appropriate types of control
samples, and control charts, will be used to
monitor and evaluate statistical control of the
analytical process. For inorganic analyses, at
least one check standard, at a concentration near
the middle of the calibration range, will be
analyzed periodically with routine samples.
144
-------
Table 8-4. Quality Control Activities Associated with Biological Measurements
Fish
Benthos
Sediment Cores
(Diatom assemblages) Bioassays
Established criteria for
sampling effort; number
of individuals, number
of taxa, or sampling
effort (bias, compara-
bility)
Repeated sampling by
different crew (precision,
comparability)
Proper preservation and
labeling of samples in
field (completeness)
Independent checks on
counts, weights, and
identification; calibration
checks of balances and
measuring devices
(completeness,
precision, bias)
Use of standard taxo-
nomic keys and nomen-
clature (bias, compara-
bility)
Collection of voucher
specimens, curation of
reference collections
(bias, comparability)
Established precision
levels for length-weight
regression equations
(precision, bias)
Replicate age determin-
ations from scales or
other structures (pre-
cision)
Proper preservation,
labeling of samples in
field (completeness)
Establish minimum
number of individuals
required (precision,
bias)
Independent check on
sorting, counting, and
identification (precision,
completeness)
Use of standard taxo-
nomic keys and nomen-
clature (bias, compara-
bility)
Collection of voucher
specimens, curation of
reference collections
(bias, comparability)
Collect replicate cores
(precision)
Section cores at lake.
shore (bias, compara-
bility) . ' . '
Proper preservation,
labeling of core samples
(completeness)
Independent checks on
counting and identifica-
tion (precision, com- ,
pleteness)
Preparation of photo-
graphic plates of differ-
ent taxa, maintenance of
reference collection
(bias, comparability)
Use of standard taxo-
nomic keys and nomen-
clature (bias, compara-
bility)
Monitor performance of
instrumentation used for
dating core layers using
blanks and internal stan-
dards (precision, bias)
Establish minimum
number of test organ-
isms per trial (compara-
bility)
Establish minimum
number of replicate
determinations required
•for a valid test (com-
pleteness, precision).
Establish restart criteria
based on mortality in,
control test, or evidence
of container effect (com-
pleteness, compara-
bility)
Establish performance,
criteria for test results,
e.g., minimum number •
of young required for
valid reproductive end-
point (precision, bias,
completeness, compara-
bility) '
Use of control charts to
monitor replicate pre-
cision, response to refer-
ence toxicants (preci-
sion, bias)
Attributes of data quality that are addressed for each activity are listed in parentheses.
Additional standards may be necessary to deter-
mine detection limits for analytes present in low
concentrations. For organic analyses, internal
standards may not be available; matrix spikes or
duplicate analyses on a subset of routine samples
will be required to monitor random and system-
atic errors.
When possible, standard reference materials
(SRM) or certified reference materials (CRM) will
be used periodically as nonblind samples to assist
laboratories in maintaining statistical control.
Such materials will be of most use for chemical
analyses of sediment samples, for organic anal-
yses, and for analyses of compounds 'in fish
tissue. Such materials are currently being used
145
-------
Table 8-5. Quality Control Activities Associated with Chemical Measurements
Field
• Calibration checks of instrumentation using independent standards of similar composition to sample;
use of control charts to monitor performance (precision, bias, comparability).
• Preventative maintenance program for all equipment and instrumentation (completeness).
• Standardized procedures for collecting, handling, and processing samples; procedures for minimizing
the potential of contamination during collection, handling, or processing (comparability).
• Proper preservation and labeling of samples (completeness).
• On-site review of recorded data and other information (comparability).
• Periodic use of field blank samples and audit material to check for effects of collection and processing
(background levels, bias).
Laboratory
• Standardized procedures for preparation, calibration, and analysis (comparability).
• Preventative maintenance program for analytical instrumentation (completeness).
• Routine use of control samples (blanks, check standards, matrix spikes, etc.) and control charts to
monitor statistical control of analytical process (precision, bias, method detection limit).
• Periodic use of reference material (standard reference materials, certified reference materials) or other
sample of known composition as internal standards to check for systematic errors in analysis
(precision, bias).
• Review of analytical data (and associated QC sample data) immediately after analysis and before
entry Into data base (completeness).
by the Near Coastal component of EMAP, and
this program should provide information related to
feasibility, cost, and preliminary performance data
that can be used in the QA program for Surface
Waters. It would be advantageous to subject
such reference samples to the entire collection
and measurement process, rather than just to the
analytical phase. This would assist in monitoring
potential errors (random and systematic) associ-
ated with sample collection and field processing.
The feasibility of implementing such an approach
as a QC tool will be investigated as part of the QA
program for EMAP-Surface Waters.
8.4.3 Habitat Quality and Site Characterization
Measurements
Some site characterization data will be collected
by the EMAP Landscape Characterization
resource group, using remote sensing techniques,
The Landscape Characterization group will docu-
ment QC activities associated with the landscape
characterization measurements in a separate QA
plan. For measurements collected directly as part
of the EMAP-Surface Waters effort, the most
critical QC activities, once standardized methods
are implemented, will be the development and use
of standardized codes and categories. For
146
-------
measurements collected from maps, an indepen-
dent check of the measurements, conducted peri-
odically by a "second person or group, would
serve to detect and correct errors on a timely
basis. For data collected during a site visit,
prope'r calibration of instruments (e.g., calibration
of an electronic depth finder against a calibrated
sounding line), and repeated measurements by a
second person on a subset of sites will be used
as the primary means of minimizing errors. Such
repeated measurements also will provide esti-
mates of the magnitude of measurement errors.
It will be desirable to implement methods to
monitor for systematic errors in collecting these
data (whether they result from a particular method
or from different crews using the method).
8.4.4 Archival of Samples and Specimens
Archival activities will involve samples for chemical
analyses, the curation of biological specimens
(fish, invertebrates), slides of diatoms from sedi-
ment cores, and validated data bases. For chem-
ical samples, samples of water and sediment will
generally be archived during a particular index
period, in case some type of reanalysis is war-
ranted. Such samples (or a subset) may be pre-
served and archived for longer periods to permit
future analyses of constituents other than those
initially determined. An example might include
more detailed analyses of samples when the
results of bioassay experiments indicate possible
toxicity. Such long-term storage may be feasible
for some inorganic constituents, but may not be
feasible for organic compounds. Samples of fish
tissue may be placed in a permanent specimen
bank, such as that established ,by the National
Institute of Standards and Technology (NIST) for
possible analyses in the future.
Voucher specimens of fish and invertebrates will
be collected as part of the routine QC program
(Table 8-4). Periodically, such specimens will be
placed into a permanent collection. Possible
options for curation include the establishment of
a specimen banking and curation system specif-
ically for EMAP, or to make arrangements with
regional facilities (e.g., national museums,
university museums, or state biological survey
agencies) to incorporate specimens collected as
part of EMAP into permanent collections. The
location of archived specimens would be reported
in the appropriate summary or interpretative
reports.
The existing EMAP QA program plan instructs all
participants in the program to retain raw data and
associated documentation (e.g., notebooks,
instrument output records) for seven years, with
permission for disposal required from the program
director. Validated data bases associated with
EMAP-Surface Waters are anticipated to be
archived permanently as part of the overall EMAP
information management program. :
8.5 DATA REVIEW, VERIFICATION, AND
VALIDATION
This aspect of the QA program overlaps with the
QA program that will be established for the
information management program (Section 9).
Quality assurance for Surface Waters information
management can be considered operationally as
two separate elements. One element, data
review, verification, and validation, discussed
here, involves ensuring the accuracy of all infor-
mation that is ultimately entered into or accom-
panies a data base. The second component,
which is also addressed in Section 9.0, involves
maintaining the security and integrity of validated
data bases once when have been archived and
made available for use in data analysis, reporting,
or distribution to users outside the Surface Waters
component, or outside EMAP. Of primary con-
cern here is the prevention of deletions, altera-
tions, or irr'etrievability Of information stored in
data bases.
The general approach in minimizing data-related
errors before archival will be to emphasize the
review of information at the point of collection or
measurement as soon as possible after the sam-
ple or measurement has been obtained. Where
feasible, data recording will be done electron-
ically, with standardized recording forms being
used as backups. In the field, data logging
devices (hand held computers that display
screens similar to rrianualfield forms) are being
considered for use, arid are currently being tested
by other EMAP task groups. Use of data logging
devices would reduce the time required for data
entry. These devices can automatically check for
erroneous data as they are entered (e.g., range
checks on numeric data, misspellings, or invalid
codes). Similar types of devices may also be
utilized for laboratory measurements.
The review process will be automated to the
extent possible/ but not to the; exclusion of a
manual review by qualified and knowledgeable
147
-------
persons. In order to complete the review, verifi-
cation, and validation process as quickly as
possible, a substantial investment in resources
and personnel required for data entry and data
review and verification will be required.
Data review will Initially involve a check of raw
data (e.g., what is recorded on data forms) before
entry Into an electronic data base. In the field,
forms should be reviewed by a second person
before leaving a site. Another review should
occur immediately after data entry by comparing
the entered data to the raw data. Data from QC
samples or measurements, and review of control
charts, can be used to determine if reanalysis (or
remeasurement) Is required before data are
entered.
The data verification process, which is largely
automated, checks that entered values are correct
and, when possible, internally consistent within a
sample or set of measurements. Examples of
verification procedures include range checks,
checks for duplicate entries, frequency checks of
coded variables to Identify inappropriate codes,
and format checks to ensure data have been
entered in the correct format. Data from QC
check samples, blanks, or replicate samples, or
redundant measurements for critical parameters
(e.g., field versus laboratory pH), can be used to
determine if there are problems with sample col-
lection or measurement. Internal consistency
checks Include Ion balance and conductivity
calculations for inorganic chemical constituents.
For biological samples and measurements, inter-
nal consistency checks include summing species
proportions In samples to ensure they do not total
to more than 100%, checks for missing taxa,
evidence of "container effects" in bioassay experi-
ments, and the taxonomic accuracy of species
Identifications.
Once the entered data are verified as accurate,
they are validated by examination against regional
expectations to Identify and explain outlier
samples or sites. Validation may involve compar-
ison with historical data, or the use of association
and multh/arlate analyses.
8.6 ASSESSMENT OF DATA QUALITY
Data quality assessment for EMAP-Surface waters
will occur at three levels: (1) within a region (i.e.,
within a year's sampling), (2) among regions (i.e.,
within a four-year sampling cycle), and (3) among
sampling cycles. Qualitative .assessments/.will',
include documenting methods, using a sampling
design that ensures unbiased and representative"
samples, and visiting sites to ensure consistency
among participating groups. Quantitative assess-
ment will attempt to estimate errors associated
with sample collection and measurement that are
important in interpreting indicators or in opti-
mizing the QA program through time (adjusting;,
the effort and intensity of QC to areas where it is
needed the most).
The primary means of assessing error and uncer-
tainty will be through carefully designed perfor-
mance evaluation studies. These studies will test
null hypotheses related to data quality require1
ments for random and systematic errors. The
design will be based on consideration of Type I
and Type II errors and Will attempt to provide
estimates of (1) total measurement error (Figure
8-1) for use in data interpretation activities, and
(2) important components of variance within
measurement error that can be used to determine
which steps in the collection and measurement
process require more, or less, QC emphasis. The
sample sizes and frequency of measurement will
be optimized to provide the necessary answers in
the required reporting period.
Performance evaluation studies could be con-
ducted using performance audit samples for
chemical analyses, reference samples for bio-
logical measurements, and "round-robin" studies
using natural samples (either chemical or bio-
logical). For water column chemistry, appropriate
performance audit materials are available for most
inorganic constituents. Depending on the con-
stituent, appropriate materials for organic
analyses may or rriay not be currently available.
Materials for sediment chemistry and fish tissue
chemistry may be limited in their availability and
appropriateness.
As mentioned previously, an important issue in
the program is the impact that using different
methodologies, or modifying or changing meth-
odologies over time, will have on data interpre-
tation, particularly in the detection of trends in
ecological condition. The QA program for EMAP-
Surface Waters will provide standard guidelines
for implementing a new methodology in a region
or at a specific site. Performance evaluation
studies will provide some information on methods
comparability, but comparability studies should be
a more intensive effort designed to test specific
148
-------
hypotheses related to the comparability with a
previous methodology. Such a comparability
study must be conducted, evaluated, and
approved before new or modified methods can be
implemented.
8.7 QUALITY ASSURANCE REPORTING
In addition to the documentation described in
Section 8.3, other types of reports will be pro-
duced periodically as part of the QA program,
including (1) summary reports of site visits, (2)
performance evaluation (or method comparisons)
summaries, and (3) assessments of data quality.
Summary reports of site visits will serve to identify
and track issues and subsequent corrective
actions, and provide information to update other
QA documentation.. The results of performance
evaluation studies will be reviewed and returned
to participants within a short time (within 30 days)
after submission. Evaluation summaries of QA-
related data and other appropriate information
concerning data quality will be prepared and
included in the appropriate EMAP-Surface Waters
reports (Section 12).
In addition, the EMAP QA program plan requires
the Surface Waters resource group to prepare a
QA annual report and workplan (Table 3-2) each
fiscal year. This document will be submitted to
the EMAP QA coordinator. The QA annual report
and workplan summarizes:
• The status of all data acquisition activities
• The status of DQO development
• QA activities (e.g., QA project plans and
standard operating procedures)
• A summary of current data quality,
including auditing activities '. ,
• QA resources utilized and recommenda-
tions for modifications in the QA program
• Planned QA-related activities, for the
upcoming fiscal year ,.
The EMAP QA coordinator will combine,the QA
annual reports and workplans from each resource
group into an overall QA annual report and
workplan for EMAP. ".'.'"-
149
-------
-------
9.0 INFORMATION MANAGEMENT
As do all research programs, EMAP-Surface
Waters depends upon data: data generated both
within and external to the program. Integration of
the data is the key to determining the current
status, extent, changes, and trends in the con-
dition of our nation's lakes and streams on
regional and national scales. Proper data man-
agement will be a critical factor in the success of
the EMAP-Surface Waters Program. Section 9
describes the rationale, approach, and objectives,
and the plan for ensuring the effective manage-
ment of this essential resource.
9.1 DEFINITION OF
MANAGEMENT
INFORMATION
Information is data that has been manipulated so
that it is useful. In other words, information must
have value, or it is still data.
INFORMATION = f(DATA, PROCESSING)
Information management is the efficient manage-
ment of data and information through all stages
and processes. The information Management
Program must provide for the effective collection,
processing, storage, cataloging, retrieval, dis-
semination and reporting of all pertinent data and
information. All these activities must be consid-
ered as part of any comprehensive information
management effort.
An information management system (IMS) must
support all these data and information activities
through the combination of data bases and com-
puter programs that perform the various functions
and operations on the data.
9.1.1 Overview of EMAP-Surface Waters Infor-
mation Management
EMAP-Surface Waters will collect a large volume
of data. As the sampling program is expanded,
the quantity of data collected will increase expo-
nentially. The ability of EMAP-Surface Waters to
manage and disseminate the large amounts of
information collected will have a major influence
on the success of the program.
An established goal of EMAP-Surface Waters is to
publish statistical summaries within nine months
after completion of data collection. The timely
transformation of data into information will
therefore be a key to success. A computerized
IMS will be required to ensure that EMAP-Surface
Waters data are made available in a timely
manner.
The IMS must be made available at various levels
to other EMAP resource groups, other local, state,
and federal organizations, and to academic insti-
tutions The system must also be capable of
drawing from historical data bases already in
place for similar environmental monitoring efforts,
and it must be compatible across all EMAP
Resource Group information management
systems.
Although EMAP-Surface Waters information man-
agement encompasses a variety of activities, all
are directed toward a single fundamental objec-
tive: timely and cost-effective access to all data
generated or utilized by the program. The Infor-
mation Management Program must provide for
the effective collection, processing, storage,
cataloging, retrieval, and dissemination of all
pertinent data. All these activities must be con-
sidered as part of any comprehensive information
management efforts.
9.1.2 Objectives of Surface Waters Information
Management
The overall objectives of EMAP-Surface Waters
information Management are to:
• Design an IMS that is responsive to user
requirements within and outside EMAP-
Surface Waters.
• Provide state-of-the-art information
management technology within the guide-
lines of the Office of Information and
Resource Management (OIRM).
• Facilitate the wide use of EMAP-Surface
Waters data.
• Produce statistical summary reports within
nine months after completion of data col-
lection.
• Ensure integration with present and his-
torical data.
151
-------
• Develop a flexible IMS that can
adapt to the programs changing
needs.
9.2 INITIAL SYSTEM CONCEPT
9.2.1 User Requirements
9.2.1.1 Issues
An important Issue for the EMAP-Surface Waters
Information Program Plan is adequate security, to
ensure the integrity of the information. We must
assure that EMAP-Surface Waters data bases are
protected from damage caused by mismanage-
ment, computer viruses, unauthorized access, or
hardware/software failures.
The confidentiality of data is also a very important
issue. EMAP-Surface Waters may access and
integrate data from other agencies that require
controlled access and confidentiality agreements.
EMAP-Surface Waters, EPA, and inter-agency pol-
icies on data confidentiality must be addressed in
the maintenance and distribution of data. These
Issues are discussed at length by Franson (1990).
9.2.1.2 Data Categories
Completed data files will be grouped into the
categories In the following list. Access to the data
files will vary according the identified user group
(subsection 9.2.1.3).
• Level A (Raw) - Data direct from field or
analytical laboratory collection.
• Level B (Verified) - Raw data files that have
completed the verification analysis defined
In quality assurance (QA) plans.
• Level C (Validated) - Verified data files that
have completed the validation analysis.
• Level D (Enhanced) - Validated data files in
which missing values have been filled in
Using established statistical procedures.
9.2.1.3 Identified User Groups
The Information generated by EMAP-Surface
Waters will be used by the following groups:
• Decision makers at all levels of government
who set environmental policy.
• Resource managers and regulators who
require an objective basis for allocating
resources and prioritizing actions, especi-
ally those focusing on the protection and
enhancement of ecological resources.
• Researchers working on the development
and understanding of ecological indicators
and processes. ,
• Those interested in evaluating the effective-
ness of the nation's environmental policies
for protecting and enhancing ecological
resources (e.g., the EPA Administrator and
Senior Management staff, Congress, and
the public).
9.2.1.4 Data User Requirements
Users and their requirements can be separated
into six general categories.
1. EMAP-Surface Waters Core Research Group:
Includes those individuals and groups charged
with designing, implementing, and interpreting
the data from the field sampling programs.
Requirements: This group needs to have
access to a comprehensive data set including:
• Project management information
• Sample tracking
• Shipment tracking
• Raw data files
• Quality assurance/Quality control (QA/QC)
reports
• Field logs
• Logistics
• Summary reports
• Maps
• Verified and validated data sets
This group also requires access to the data on
a time basis as close to real time as possible.
Raw data used by this group will not have
been through QA. This group needs access
to all data described in the other categories.
2. EMAP-Surface Waters team: Includes individ-
uals and groups who are involved in the
EMAP-Surface Waters effort but not in the
day-to-day field operations. The team includes
outside participating agencies, logistical
support personnel, EMAP-Surface Waters
152
-------
QA/QC personnel, program reviewers, and
EPA Headquarters personnel.
Requirements: This group requires access to
summary information regarding logistics and
project management as well as validated data.
They require access only to those files that
have been validated and verified. They do not
require real-time access, nor do they need to
have access to a comprehensive data set.
3. EMAP Program: Includes all researchers dir-
ectly involved in the design, implementation,
and analyses of the national EMAP program.
This group includes members of other task
groups, members of the Synthesis and
Integration Team, and personnel in other
agencies directly involved in EMAP.
Requirements: This group requires final sum-
maries regarding logistics and project
management. They require access only to
those files that have been validated and veri-
fied. They do not require real-time access, nor
do they need to have access to a comprehen-
sive data set. They need data in a context
that can be integrated with data from other
disciplines. Document summaries with inter-
pretation and graphic outputs will be most
useful.
4. Other federal and state agencies involved in
environmental monitoring programs: This
group includes the EPA regions, other EPA
offices, state agencies, universities, and other
scientific research organizations.
Requirements: This group requires access
only to validated and verified data. They do
not require any data concerning logistics and
project management. They need summarized
characterization data for each station sampled.
They also need access to an index of available
data (both EMAP and historical). They require
access only to those files that have been
validated and verified. They need data in a
form that can be integrated with data from
other disciplines. Document summaries with
interpretation and graphic outputs will be most
useful.
5. Legislators and Environmental Managers.
Requirements: This group needs summarized
and interpreted data. They do not require any
data regarding logistics or project manage-
ment. This group of users will be best served
by published reports, maps, and an on-line
summary system.
6. The General Public.
Requirements: This group needs summarized
and interpreted data. They do not require any
data concerning logistics or project manage-
ment. This group of users requires published
reports, maps, and an on-line summary
system.
9.2.2 System Management and Functional
Requirements
Information management has different functions
within three levels of EMAP-Surface Waters opera-
tion: Resource Group projects, the Resource
Group Program; and overall EMAP. To support
the needs of these levels, the EMAP-Surface
Waters IMS must provide the following products
and services.
• Resource Group Projects
- Field data collection (data loggers, bar
code readers)
- Sample tracking
- Analytical laboratory data collection
- QA/QC analysis and reporting
- Data transfer from other agencies
- configuration management
• Resource Group Program
- Access to existing data bases
- Data integration and analysis
- Data base management
- Presentation and reporting
• Overall EMAP
Data base transfer/access
Data base catalog
Archival/Backup of data
Integration of data from
resource groups
Training and support
multiple
At each of the levels above, programs will be
designed, developed, and tested to meet the
needs identified. Standardization of components
153
-------
will facilitate QA activities. Resource group
standards will be developed by the EMAP Infor-
mation Center (EIC) with input and review from
the Information Management Committee (IMC).
Where appropriate, standards will be implemented
as defined.
9.2.3 Data and Information Flow
Figure 9-1 shows the overall flow of information
and data through the Information Management
System.
9.3 ORGANIZATION OF EMAP-SURFACE
WATERS INFORMATION MANAGEMENT
PROGRAM
The following structure has been developed for
the EMAP-Surface Waters Information Manage-
ment Program to respond to user and functional
requirements.
9.3.1 Surface Water Information Manager
The EMAP-Surface Waters Information Manager is
responsible for planning, coordinating, and
facilitating Information management activities
within the group. The Information Manager
reports to the Surface Water Technical Director,
and is responsible for ensuring that data and
information collected by EMAP-Surface Waters
are properly captured, stored, and transmitted
within EMAP-Surface Waters and EMAP in gen-
eral. The Information Manager works closely
with the Technical Director and technical staff to
assure that information management adequately
supports EMAP-Surface Waters in achieving its
scientific objectives. The Information Manager
accomplishes these tasks by establishing and
directing a Surface Waters Information Center
(SWIG), described in Section 9.3.2.
The Information Manager, as an executive mem-
ber of the Information Management Committee
(IMC), serves as the liaison between EMAP-
Surface Waters and the Information Management
Director (IMD), representing the needs of
EMAP-Surface Waters scientists and managers.
9.3.2 Surface Waters Information Center
(SWIC)
The Surface Waters Information Center (SWIC),
managed by the Surface Water Information
Manager, is the focal point for planning,
coordinating, and implementing information
management within EMAP-Surfeice Waters. The
SWIC is one of the key elements to successful
information management. The SWIC is com-
posed of personnel (e.g., Information Manager,
Data Base Administrator, programmers, data
clerks) and hardware/software resources that
support the EMAP-Surface Waters Information
Management Program in the planning, implemen-
tation, and operation of the information man-
agement activities.
Figure 9-2 shows the overall structure of the
Surface Water Information Center.
The following items describe some of the key
personnel and their roles in supporting the SWIC:
• Information Manager (see Section 9.3.1)
• Data Base Administrator: This person is
responsible for assembling, documenting,
and administering the EMAP-Surface
Waters data bases.
• Programmer: The programmer provides
the programming support required for the
information management activities. He or
she will develop systems necessary for
data collection, tracking, and dissemination
(data entry screens, reports, sample track-
ing, analysis, etc.).
• Data Clerk/Librarian: The Data Clerk/
Librarian is the person who helps develop
the data bases and the EMAP-Surface
Waters data catalog and data dictionary.
• Technical Support for Computer Services:
The technical support person will in charge
of the installation and maintenance of all
hardware, software, and communications.
This will not require a full-time person.
9.4 OPERATIONAL SPECIFICATIONS
The IMS must have the flexibility needed to
handle the array of data types resulting from
sample collection and processing. It must also
support a variety of analysis, presentation, and
reporting activities. For the 1991 Demonstration
Project, the Statistical Analysis System (SAS) will
be used for data management. SAS will also be
used for most statistical analyses. SAS has been
selected as the data management system
154
-------
155
-------
EMAP
INFORMATION
CENTER
AND
CENTRAL DATA
CATALOG
Surface Water
Information
Center
Figure 9-2. Concept of the EMAP-Surface Water Information Center in relation to the EMAP
Information Center.
156
-------
because no relational data base system is avail-
able to EPA through current contracting mech-
anisms. When a relational data management
system becomes available to EPA through the
Office of Information Resource Management
(OIRM), the EMAP-Surface Waters data manage-
ment system will be converted to the selected
relational data base system, allowing users more
flexible and efficient access to EMAP-Surface
Waters data. After the conversion to a relational
data base system, SAS will continue to be used
as a principal data analysis tool.
To meet the requirements of the user groups, the
EMAP-Surface Waters IMS will consist of the
following major components for operations: proj-
ect management; data and sample collection;
processing and storage of indicator data; data
access and transfer; data analysis and reporting;
data documentation, access, and archival;
geographical information system (GIS); and
integration of data.
9.4.1 Project Management System
The project management component of the IMS
requires frequent and accurate status reports
about field collection and laboratory processing
activities. The project management component of
the IMS is used for this purpose. This component
has two major elements: (1) a communications
system for rapidly transferring information
between field crews, processing laboratories, and
the SWIG, and (2) a sample tracking system for
monitoring the status of sampling events, sample
shipments, and status of analyses on a real-time
basis. These two elements of the Project
Management Information System are discussed
below.
9.4.1.1 Communications
Field crews and processing laboratories will use
standard formats to submit data to the SWIG in
established time frames. Reliable communica-
tions software and programs will facilitate this
information exchange. For example, programs
will be developed to automatically log remote
computers into the central processing center and
to perform file transfers into predetermined
directories. Initial processing of the data will be
accomplished automatically. When processing is
complete, the SWIG will be notified and requested
to acknowledge that it is aware the data are ready
for additional processing.
Field crews will access data through the SWIG
communications link. These data will include
logistical information, such as the locations of
equipment, personnel, and motels, and site-
specific information about sampling locations.
9.4.1.2 Sample Tracking Information
The sample tracking system will track samples
from initial collection through completion of all
analyses and/or processing. To accomplish this,
each sampling event and sample type will be
assigned a unique identification number. These
numbers will be entered into the IMS before data
collection. Sample numbers will be bar coded to
facilitate data entry by the field crews (if funding
permits). Information to be entered for each sam-
ple in the sample tracking system available for
retrieval and review includes:
• Sampling site name.
• The time the sample was collected, includ-
ing date, hour, and duration of sampling
effort.
• Type of sample (e.g., grab samples to be
processed for benthic species composition
and biomass, fish tissue sample to be
processed for contaminant concentrations).
• Identification of the individual/team that
collected the sample.
• A list of the analyses and processing activi-
ties planned for that sample, and the status
of those analyses and activities (e.g.,
collection completed, analyses completed).
• Directions to files containing raw data
generated for each sample.
• Directions to textual files containing
descriptive information about the sampling
event (e.g., field team comments).
When samples are transferred from field crews to
analytical laboratories, a record of the exchange
will be entered into the sample tracking system,
both upon release and upon receipt of the
materials. The identity and disposition of any
sample can then be established by checking the
sample status in the IMS. The status of all
analyses and results will be available through the
sample tracking system. When all processing for
157
-------
a sample Is complete, a "flag" will be set, and
users will be notified automatically when logging
onto the VAX computer. The GIS system will be
linked to the sample tracking system to display
the status of sampling activities.
9.4.2 Data and Sample Collection
Field data will be entered directly into a portable
field data recorder. These data include direct field
measurements, as well as site information (e.g.,
latitude, longitude, state, sample ID). The data
will be automatically verified as described in the
QA section of this chapter. Field data will be
submitted to the SWIG In established time frames,
and In approved formats.
Sample numbers will be assigned to each sample
collected to ensure that all samples in the plan
are properly Identified and tracked. These num-
bers will be assigned by the field sampling
coordinator and entered into the IMS before the
sampling event. All sample labels will be pro-
duced before the sampling event.
9.4.3 Processing and Storage of Indicator Data
Quality assurance will be performed on all data
received by the SWIG, using the procedures
described In Sections 8.0 and 9.5, then the data
will be converted into SAS data sets. The data
sets will be stored In data libraries by indicator
type. Following Initial data processing, the
required data analyses will be performed and
summary data bases produced. The IMS will
maintain data and relevant analytical results in
both raw and summarized form. This will elim-
inate costly, redundant analyses.
9.4.4 Data Access and Transfer
The EPA VAX network will be the primary means
of access to data in the IMS. All data base
design work and documentation, including the
code libraries, data dictionary, standard operating
procedures for data handling, and GIS standards
and base coverages will be available over this
network. Users who do not have access to the
VAX network will be provided with direct dial
access to the SWIG, as appropriate. Ultimately,
the data, reports, and findings of EMAP-Surface
Waters will be important to many other groups
within EMAP, to the scientific community, and to
the general public. Access authorization will be
established under the direction of the EMAP-
Surface Waters Technical Director.
Historical data sets or data collected by other
organizations that may be important, though not
likely to be used regularly, will be documented
and processed; QA will be conducted. Access
through the IMS to confidential data, or data for
which the quality is suspect or cannot be
determined, will be limited.
All data made available for general use will be in
read-only format, allowing users to access the
data without compromising the integrity of the
data base. Requests for copies of or access to
data in the IMS will be submitted to a qualified
person. A schedule will be developed for pro-
viding access to these data. The release sched-
ule will depend on the availability of personnel to
process the data, as well as the urgency of the
request.
9.4.5 Data Analysis and Reporting
Analysis will be done only on data that have
passed QA scrutiny. Qualified scientists will
develop programs for the analyses. The IMS
does not perform assessments of environmental
health, but assists the Resource Group in per-
forming data analyses and data presentations.
Data exchange interfaces will be developed
between the IMS, GIS, and other tools for data
analyses. All results will be displayable shortly,
after they are incorporated into the IMS.
9.4.6 Data Documentation, Access, and
Archival
All data received by the SWIG will be converted
into SAS data sets in the VAX system. Complete
documentation of all data bases stored in the
SWIG is of paramount importance. The Data Set
Index (DSI) will be the principal data information
source and will include a catalogue listing all
available data, modes of access, and quality of
the data. A data library contained within the DSI
will provide users with important information
about the contents of each data set. A Central
Data Dictionary will document information on
standards that have been developed for data sets -
generated by EMAP and external data sets. The
standards will include field names, formats,
documentation, acceptable ranges, and codes. A
data base backup system will be developed for
158
-------
complete and rapid recovery of all information in
the event of a data base failure.
9.4.7 Geographic Information System (GIS)
A major requirement of SWIG capabilities will be
to create maps and perform geographically based
analyses; therefore, the data generated for
EMAP-Surface Waters will be geo-referenced.
Spatial analyses will be accomplished using
ARC-INFO, a GIS that is used throughout EPA.
ARC-INFO is also used by most of the federal and
state agencies participating in the EMAP program.
ARC-INFO is a powerful tool which includes
extensive analytical capabilities and interfaces
with a number of other major software products,
including SAS and ERDAS (a common software
tool for processing data collected by satellites).
ARC-INFO is not user friendly; therefore, user
friendly interfaces for routine data analysis and
display will be developed by GIS team analysts.
EMAP-Surface Waters data analysts will work with
other data management groups within EMAP and
other agencies to develop standards and cover-
ages for GIS applications. Standards will be
developed for data accuracy, naming conven-
tions, and documenting and archiving completed
maps.
9.4.8 Integration
All data integrated into the IMS will be converted,
if necessary, to EMAP-Surface Waters formats
and standards.
9.4.9 Existing Information/Data
Existing data bases constitute an important
source of information. They will be used to
develop indicators, design the sampling program,
and interpret the results of EMAP-Surface Waters
surveys. Pertinent data bases will be included in
the Data Set Index. Data bases used frequently
will be converted to SAS data sets.
9.5 QA/QC FOR INFORMATION MANAGE-
MENT ACTIVITIES
Another important QA issue associated with a
monitoring program the size of EMAP-Surface
Waters is ensuring and maintaining the integrity of
the large number of values that eventually will be
entered into the data management system (NRC
1990, Risser and Treworty 1986, Packard et al.
1989). The usefulness of this system is in part
determined by the quality of the information it
contains. The scientific importance of this project
requires the utmost confidence in the validity of
the final data base. QA for information manage-
ment includes those measures that ensure that
information entered into the system is not
corrupted.
Data become corrupted in two general ways: (1)
incorrect information and (2) missing, incomplete,
or nonretrievable information (Kanciruk et al.
1986). Examples of errors of the first type include
typographical errors, incorrect species identifi-
cation, and inaccurate instrument calibration.
Although all errors cannot be completely elimin-
ated through data management protocols, the
potential for including incorrect information in the
data base can be reduced. The second type of
error is the omission of important information
relating to legitimate data values (Kanciruk et al.
1986). Such pieces of information, called "data
qualifiers" assist in the correct interpretation of
data values. Qualifying information will be antici-
pated and a structured system for recording and
retrieval of this information will be developed. The
system will be designed with flexibility to allow the
inclusion of unanticipated qualifiers.
QA and information management personnel Will
work in conjunction with other EMAP-Surface
Waters personnel to prevent the corruption of
data and to ensure the integrity of the data
through all phases of activity (collection through
dissemination and reporting). The features
described in Section 9.5.1 through 9.5.8 will
provide the foundation for the management and
QA of all data through these phases. Automating
as much of the standardized QA/QC process as
possible will be the ultimate goal.
9.5.1 Field Data Collection
A systematic numbering system will be developed
for unique identification of individual samples,
sampling events, stations, shipments, equipment,
and diskettes. Whenever possible, sample con-
tainers, equipment, and diskettes will be pre-
labelled to eliminate confusion in the field. The
pre-labeling will reduce the number of incorrect or
poorly affixed labels. If funding permits, bar code
readers will be used to facilitate accurate sample
number entry and identification in the field.
Standard operating procedures (SOPs) will docu-
ment the use of the field computer systems (data
159
-------
loggers) and the data entry procedures. Data
loggers are hand held computers that display
screens similar to manual field forms; they will
automatically check for errors in data being
entered (range checks on numeric data, biological
species code verification). Explicit contingency
plans using field forms will also be made, in the
event that the field system fails. If field forms
must be used, qualified persons will conduct
manual verification before the data are entered
into the system.
At periodic Intervals, field crews will back up data
collected In the field. Procedures for these field
back-ups will be defined and crews trained in their
use. Data will not be purged from the field com-
puters until the Information has been transferred
and backed up.
9.5.2 Data Transfer
To minimize the errors associated with entry and
transcription of data from one medium to another,
data will be captured electronically when possible.
The communications protocols used to transfer
data electronically will have mechanisms by which
the completeness and accuracy of the transfer
can be checked. If a file cannot be verified upon
receipt of the data (i.e., incorrect number of
bytes), a new file transfer will be requested.
9.5.3 Sample Tracking
The tracking and handling of samples collected in
the field will be critical for QA of the resulting
data. Sample tracking systems that provide
accurate Information on the status and location of
the samples will be developed.
9.5.4 Data Entry/Verification
As stated previously, data will be captured elec-
tronically in order to minimize the errors associ-
ated with entry and transcription of data from one
medium to another. However, at certain points of
data acquisition, it will be necessary to enter
these data manually (i.e., results of data analysis).
When manual entry is required, the data entry
process will Include an accuracy check in which
data will be entered twice or a 100% visual verifi-
cation will be performed. The approved method
will depend on a cost/benefit analysis. In many
Instances, the use of bar code labels should elim-
inate the need for manual entry of routine information.
The following examples show the types of checks
that will be performed if manual entry is required:
• Range checks on numeric data: Field and
laboratory numeric data and taxonomic
species codes must be checked for accep-
tability using established criteria. Data that
fall outside the acceptable range will be
flagged and reviewed by QA personnel.
• Certain fields will contain coded informa-
tion. When codes are used, they will be
compared with the codes established by
the scientific personnel and information
management to assure compliance-
• Data that have been flagged will be
reported. QA personnel will review this
report and release data that have passed
the QA check for addition to the data base.
All errors must be corrected before the
data can be entered.
9.5.5 Data Validation
All discrepancies identified by the computer will
be documented in hard copy. These discrepancy
logs will be saved as part of the EMAP-Surface
Waters archive. Data will be flagged as described
in Section 9.5.4. All identified discrepancies
should be brought to the attention of a qualified
QA person, who will determine the appropriate
corrective action to be taken. Data will not be
transferred to the data base until all discrepancies
have been resolved by this person. A record of
all additions will be kept in hard copy.
9.5.6 Data Analysis
Laboratories will be required to collect data and
transmit it to a main repository. These data will
include sample tracking information, results, and
QA/QC information. Each set of data transmitted
or received will include a full set of QA/QC data
as specified by the EMAP-Surface Waters QA/QC
plan.
Standard formats will be developed for data sub-
mitted electronically by the laboratories. A
specific format will be developed for each file type
within each discipline. Files that do not meet
required specifications will be deleted and data
will be requested in the proper format.
160
-------
9.5.7 Archive and Backup
Data generated, processed, and incorporated by
EMAP-Surface Waters will be stored and archived
on redundant systems. This will ensure that if one
system is destroyed or incapacitated, information
management personnel will be able to reconstruct
the data bases. Procedures will be developed to
archive the data, monitor the process, and
recover the data if necessary.
Several backup copies of the data (all levels) and
of the programs used for processing the data will
be maintained. Backups of the entire system will
be maintained off site, using designated system
backup procedures.
9.5.8 Configuration Management (CM)
Methods must be developed for keeping accurate
records of the who, what, when, and why of any
modification to the IMS. Whenever possible, a
software configuration development and manage-
ment package should be used, for example, a
Computer Aided Software Engineering (CASE)
tool.
Configuration management (CM) is the process of
controlling the system's organization, construc-
tion, and maintenance. Although CM and QA are
separate but complementary disciplines, both are
required to ensure system integrity.
Five major sets of activities are involved in CM:
• Configuration identification - definition and
identification of items subject to configura-
tion control.
• Configuration control - evaluation, coordi-
nation, and approval or disapproval of pro-
posed changes to controlled items.
• Configuration status accounting - recording
and monitoring of changes to controlled
items.
• Data management - maintenance of official
correspondence records, CM records, and
controlled documentation.
• Configuration auditing - verification that
controlled items are what their documenta-
tion states they are and that they meet
their assigned requirements.
A CM program appropriate for EMAP-Surface
Waters will be developed and documented.
9.6 RESOURCES: OVERVIEW
The availability of information management
resources is vital to the success of EMAP-Surface
Waters. These resources are made available to
EMAP-Surface Waters through authorized access,
existing capabilities, or the procurement and
acquisition of new resources. Existing resources
and funding for new resources is limited; there-
fore, all steps must be taken to assure cost-
effective use of all resources. EMAP-Surface
Water's approach is to make use of existing and
planned resources to the extent possible, and to
acquire additional resources only where needed.
Existing resources include current EPA staff,
existing national and site contracts, EPA standard
hardware and software, institutional resources
(e.g., universities), other agencies' resources, and
other contractor resources.
EMAP-Surface Waters will work with EPA's Office
of Information Resources Management (OIRM)
and the Office of Administration and Resources
Management (OARM) to assure that the EMAP-
Surface Waters Information Management Program
is properly planned and coordinated with the
Agency's IRM Program and that EMAP-Surface
Waters resource requirements are conveyed
through the appropriate channels. EMAP-Surface
Waters will make extensive use of the automated
data processing (ADP) resources available at the
Office of Research and Development Laboratories.
The Information Manager will work with the ADP
coordinators and laboratory directors to plan and
obtain approval for the utilization of these
resources by EMAP-Surface Waters.
EMAP-Surface Waters Information Management
will identify an ADP coordinator at each location
within the program. Each ADP coordinator is an
existing local resource whose responsibilities
include assisting the Information Manager in
development of the EMAP-Surface Water's ADP
resources and ADP plans. These individuals are
key figures in assuring that SWIG is fully sup-
ported by the ADP capabilities.
9.7 IMPLEMENTATION PLAN
The EMAP-Surface Waters IMS will be imple-
mented in several phases over a four-year period.
The SWIG will become operational when a full-
161
-------
time Information Manager assumes control, basic
Information management systems are in place,
and Initial data collection activities have begun.
In FY91, Information Management will expand
areas of development and operation for the large-
scale pilot. Activities will focus on staffing and
start-up of the SW1C. An Information Manage-
ment Program Plan, and a System and Data Base
Design Document will be generated. Initial
prototypes of the various systems and data bases
identified in the System and Data Base Design
Document will be completed before the 1991 field
season.
In FY92, Information Management will complete
development of its core systems. External data
will be integrated with EMAP-Surface Waters data.
Prototype systems for electronic data access, the
Data Set Index, and the Central Data. Catalogue
system will be designed and developed.
162
-------
10.0 COORDINATION
This section focuses on coordination, among
components within EMAP and between EMAP-
Surface Waters and other state and federal
programs.
In developing a complex program such as EMAP,
a wjde range of issues must be acknowledged
and addressed throughout the early phases. The
coordination of activities and analyses must occur
at multiple levels both within the program and
across other programs. For example, within the
Surface Water component of EMAP, there is a
need to integrate activities in the lake and stream
components with one another and to integrate the
Tier 1 and 2 activities with those that are
occurring or might occur at Tiers 3 and 4 (see
Section 2 for a description of the four-tier
concept). Equally important is the need to inte-
grate information from the various ecological
resource categories within EMAP. Outside EMAP
itself is a host of programs that, although they
cannot adequately address the objectives of
EMAP by themselves, can complement the infor-
mation from EMAP to provide a clearer view of
current status and trends in indicators of
condition of surface waters and diagnosis of
conditions.
10.1 WITHIN EMAP
10.1.1 Tier 3 Activities in Surface Waters
The primary focus of EMAP-Surface Waters during
early phases must be on providing estimates of
the extent and status of current conditions at
broad regional scales. By doing so, we should be
able to more effectively identify priority regions
where significant proportions of the resource are
in poor condition, and thus in need of attention,
or where particularly sensitive subpopulations are
at risk. We have described this as Tier 3 level
activity. As described in Section 2.0, Tier 3
activities may be of several types: (1) more
detailed diagnostic work to determine cause-effect
linkages and remediation strategies or (2) follow-
on monitoring to evaluate the effectiveness of a
specific legislative or regulatory activity by
monitoring early warning or particularly sensitive
subpopulations. Depending on the exact nature
of these activities, they should complement the
design of EMAP-Surface Waters and use indicator
methodologies that provide comparable and con-
sistent results.
As explained throughout this plan, a Tier 3 activity
is being designed within EMAP-Surface Waters to
evaluate the effectiveness of the expected revi-
sions to the Clean Air Act. The Temporally Inte-
grated Monitoring of Ecosystems (TIME) project
is intended to provide a Tier 3 activity that will
elucidate, through monitoring, the changes and
trends in chemical conditions in acid-sensitive
surface waters of the United States that result
from changes in acidic deposition. Specific goals
are to:
• Monitor current status and trends in chem-
ical indicators of acidification in acid-
sensitive regions of the United States.
• Relate changes in acidic deposition to
changes in surface water conditions.
• Assess the effectiveness of the Clean Air
Act emissions reductions in improving the
acid/base status of surface waters.
the TIME project must be able to make regional
estimates of the changes in acid-base status,
detect regional trends in a policy-relevant time
span, and relate regional changes in surface
water condition to regional changes in acidic
deposition. The project design is national in
scope but regional in implementation (Stoddard
1990). The design, implementation, and analysis
of the TIME project is being conducted to maxi-
mize the use of existing resources and coordinate
all aspects (e.g., quality assurance (QA) and data
base management).
As new special projects emerge over the lifetime
of EMAP-Surface Waters, we will make similar
efforts to enhance the comparability and utility of
the information generated.
10.1.2 Across EMAP Elements
An important aspect of EMAP is the inclusion of
all ecological resources within the program. From
the national perspective, this provides the
opportunity to evaluate the relationships between
conditions and problems in each resource cate-
gory, their impacts on one another, and the
163
-------
potential to more effectively evaluate compre-
hensive ecological resource management strate-
gies. The achievement of these potentials
requires extensive coordination in the selection of
indicators, methodologies, and design.
Within EMAP, the focus to date has been the
design and evaluation of programs to identify
status, trends, and probable cause of conditions
within each ecological resource group. Discus-
sions on enhancing the integrative aspect of the
program have been gradually increasing and will
be the focus of the Integration and Assessment
Team. Coordination between ecological resource
groups has been facilitated by the coordination
, teams within EMAP (i.e., statistics and design,
indicator development, integration and assess-
ment, information management, QA, and logis-
tics). The activities sponsored by these teams
currently provide the framework to ensure our
future ability to more fully integrate the
information from each aspect of the EMAP.
Workshops have been held by the indicator devel-
opment coordinator to facilitate discussion
between the ecological resource groups. Effec-
tive communications will ensure that all groups
take into account, as they develop their programs,
the advantages of exchanging information with
other groups. For example, measurement of
nitrate In headwater streams by the Surface
Waters group could help identify nutrient leakage
for the Forest and Agroecosystem groups.
Coordination is also taking place as the specific
resource categories (e.g., wetlands from lakes)
are clearly defined and all ecological resources
are considered. For example, riparian habitats
might fall under wetlands, surface waters, or
individual terrestrial groups (e.g., Arid Lands,
Forests, Agroecosystems). The Arid Lands, Wet-
lands, and Surface Waters teams are now plan-
ning a joint pilot to evaluate how riparian habitat
in the southwest can best be evaluated within the
framework of EMAP. Similar coordination will be
required between EMAP-Surface Waters, Wet-
lands, Forests, and Agroecosystems to address
riparian habitats in other regions of the country.
The Integration and Assessment team, in conjunc-
tion with the Statistics and Design team, and the
ecological resource groups will begin efforts
during FY91 to evaluate design alternatives that
might maximize the ability to integrate information
from various ecological resource groups. These
efforts can be .coupled with the design and pilot
activities of each resource group to evaluate
long-term options for the program.
10.2 WITHIN EPA (OFFICE OF WATER AND
REGIONS)
As described in Section 1.0, the U.S. EPA is
responsible for providing the states with guidance
in fulfilling their reporting requirements under
Section 305(b) of the Clean Water Act. The EPA
then integrates the information provided by the
states and produces a report for the Administrator
and Congress. The difficulty faced by the Office
of Water is in reporting national status and trends.
The flexibility allowed the states in the choice of
methods, designation of beneficial uses and cri-
teria, and selection of monitoring sites for evalu-
ation makes the reporting of national status and
trends an all but impossible task. The host of
compliance-oriented issues for which the Office of
Water and the regions are also responsible has
made their charge quite extensive and given them
little time to address the issue of national and
regional status and trends. We have begun work-
ing with the Office of Water and the EPA regional
offices to ensure that we effectively provide the
national status and trends information they
require. We have been working with the biocri-
teria development activities within the Office of
Water to ensure comparable, or at least comple-
mentary, approaches between EMAP-Surface
Waters and the guidance given the states.
One current limitation concerns spatial coverage.
As presently envisioned and budgeted, EMAP-
Surface Waters will be able to provide resolution
only at a scale of 7 to 10 regions over the conter-
minous United States, rather than resolution at a
state level. We recognize that the program will
increase in value to the individual states, as well
as to the Office of Water, if state-level resolution
can be achieved. We have begun discussions to
determine the various options available for achiev-
ing state-level resolution.
We hope that, in addition to providing information
potentially useful to the Office of Water and to the
regions, EPA regional offices will act as the coor-
dination links to the states. The regional offices
are familiar with the individual states within their
boundaries, understand the state organizational
structures, and know the individuals within various
state agencies through whom we will need to
work. The regions possess much of the expertise
164
-------
needed to coordinate and implement the pro-
gram. The regions also provide the nationwide
infrastructure necessary to the success of a
long-term effort such as EMAP.
In short, not only do we view the Office of Water
and the U.S. EPA regional offices as clients for
EMAP information, but we view them as important
to the implementation and long-term success of
the program.
10.3 OTHER FEDERAL AGENCIES
Almost every federal agency that has a mandate
or jurisdiction over some natural resource has an
interest in surface water resources. These agen-
cies include, but are not limited to, the Agriculture
Research Service, Bureau of Indian Affairs, Bureau
of Land Management, Bureau of Reclamation,
U.S. Fish and Wildlife Service (USFWS), National
Science Foundation, National Park Service, U.S.
Army Corps of Engineers, U.S. Forest Service,
and U.S. Geological Survey (USGS). We firmly
believe that EMAP should be a multi-agency pro-
gram. The concept of the program is important,
rather than its location within any particular
agency. If aspects of the program objectives are
being met through activities in other agencies,
then there is no need to duplicate those aspects
within EMAP.
Although we will need to coordinate with all the
federal agencies mentioned above, we directed
our early efforts toward the ones that have active
surface water monitoring programs, but no
planned interaction with other EMAP resource
groups. This led to early interaction with USFWS
and USGS. Programs maintained by each of
these agencies were discussed in Section 1.0. In
Sections 10.3.1 and 10.3.2, we describe potential
scenarios for interaction. Discussions have been
taking place between EMAP-Surface Waters and
both these agencies, but no firm commitments
have been achieved.
10:3.1 U.S. Fish and Wildlife Service
As described in Section 1.0, the USFWS maintains
the National Contaminant Biomonitoring Program
(NCBP). We have discussed with the USFWS
their participation in EMAP-Surface Waters in a
variety of contexts. We would like to see the
concept of NCBP expanded to include the full
range of lakes, reservoirs, streams, and rivers that
will be sampled under EMAP-Surface Waters.
This would not replace the current NCBP, but
would function in addition to it. As a major
federal agency responsible for biological
resources, the USFWS is currently addressing
monitoring both on and off the National Wildlife
Refuges. Their concept appears to parallel that of
EMAP in proposing to use information on ambient
biotic assemblages, bioassay or toxicity tests,
tissue residue data, and physical habitat data. As
USFWS proceeds with these plans, we would like
to work with them to ensure that their efforts and
ours complement one another. The USFWS has
extensive experience and expertise in dealing with
aquatic biota from which EMAP-Surface Waters
could benefit. The rigorous design and spatial
coverage proposed by EMAP-Surface Waters
could provide a beneficial context within which to
evaluate the results of the National Wildlife Refuge
monitoring efforts and the NCBP information.
Cooperation and coordination of USFWS activities
and EMAP activities would benefit more than just
EMAP-Surface Waters. The USFWS maintains the
National Wetlands Inventory, which has begun
discussions and interaction on a cooperative
program with EMAP-Wetlands. Wildlife popula-
tions have been viewed as potentially valuable
indicators of ecological resource condition for all
of the terrestrial EMAP groups. We have pro-
posed pursuing joint programs in most EMAP
resource groups so that new efforts truly demon-
strate added value and complement existing
activities to the maximum extent possible.
10.3.2 USGS National Water Quality Assess-
ment Program
In addition to maintaining the NASQAN and
Benchmark networks, the USGS recently began
development of the National Water Quality
Assessment (NAWQA) program (Hirsch et al.
1988), which has been reviewed by the National
Research Council (1990). The goals of this
program are to:
• Provide a nationally consistent description
of current water quality conditions for a
large part of the nation's water resources.
• define long-term trends (or lack of trends)
in water quality.
• Identify, describe, and explain, as possible,
the major factors that affect observed water
quality conditions and trends.
165
-------
In many respects, EMAP-Surface Waters and
NAWQA sound similar in scope and intent. Both
are Intended to address aquatic resources on
regional and national scales over long periods of
time. But each program has a different focus and
will provide different information. An obvious
difference is that NAWQA will focus on streams
and rivers and EMAP-Surface Waters will focus on
lakes, reservoirs, streams, and rivers. NAWQA is
focusing on the component of aquatic resources
that accounts for approximately 60% of the sur-
face water withdrawal use and supplies roughly
60% of the human population served by surface
water supplies. EMAP-Surface Waters has
selected a national probability design so that the
sample Information can be extrapolated with
known confidence to the entire surface water
resource. NAWQA's goals are to describe water
quality and to understand, within a hydrologic
basin, those factors that influence it. EMAP-
Surface Waters will focus on describing ecological
condition, aspects of the chemical and physical
habitat, and land use patterns that may affect
ecological condition.
We believe that each of these programs has its
own strengths as well as weaknesses. USGS has
a mandate to monitor water quality, as does EPA.
USGS has an excellent reputation in the quality of
science it conducts, its experience in addressing
water quality and hydrology issues, and its ability
to commit to long-term programs. The Office of
Research and Development within EPA is gaining
a reputation for developing creative and innova-
tive approaches to address aquatic resource
issues and for designing programs that provide
information needed to answer policy related
questions. We firmly believe that the public and
Congress will be best served if EMAP and
NAWQA can be designed and conducted in a
manner that will maximize the complementary
nature of their information. We have intentionally
begun the EMAP-Surface Waters activities with
lakes rather than streams to provide more time for
EPA and USGS to develop creative ways to
ensure that NAWQA and EMAP-Surface Waters
do not duplicate efforts but work in a
complementary fashion. The two groups have
discussed these issues over the past 18 months
and will continue doing so until resolution.
10.4 STATE AGENCIES
Interaction with state agencies is important to the
success of EMAP. We believe that the states may
provide the most effective mechanism for imple-
menting the field aspects of the program and a
means of institutionalizing the activity. However;
this cannot be done if it simply adds yet ariother
project that the states are expected to carry out
but does not give them the resources or support
needed to do the job. Many states have the
required expertise, or access to it. EMAP will
need to coordinate interaction with the EPA
regions who already have extensive interaction
with state personnel. Participation in the program
will certainly be of more interest to the states if we
can intensify the activity to provide state-level
resolution of information. We need to find crea-
tive approaches for satisfying the objectives of the
program while supplying information of value to
the states. These interactions have just begun in
EPA Regions 1 and 2, the northeastern United
States, and the FY91 activities will serve as a pilot
for how we conduct interaction with the states in
the future.
10.5 RESEARCH ORGANIZATIONS
As described in Section 2.0, research organiza-
tions such as universities and state and federal
research centers are important groups with which
EMAP-Surface Waters needs to interact. The
research that will allow EMAP to reach its poten-
tial will come from these groups. Interaction with
these groups during the developmental stages of
EMAP will provide a mechanism for ensuring that
the program is grounded in sound scientific prin-
ciples. Existing field sites can provide locations
for testing and developing indicators of ecological
condition and for understanding the relationships
between our indicators of response and indicators
of exposure, habitat and stress. Existing sites,
such as the Long-Term Ecological Research
(LTER) sites funded by the National Science
Foundation (Franklin et al. 1990, Magnuson 1990,
Swanson and Sparks 1990), not only provide
excellent sites for testing, developing, and
understanding indicators, but also, if included as
part of Tier 2 sampling, may provide information
extremely useful for interpreting and understand-
ing the changes or trends in indicator values that
are observed. EMAP, on the other hand, can pro-
vide useful information that will allow research
organizations to evaluate the context of their
findings in terms of the portions of the resource
their sites might represent and perhaps allow
them to extrapolate their findings to a broader
area. Once again, this is a situation in which
programs certainly have different objectives but
166
-------
can complement one another in significant ways
if an early effort is made to communicate. The
interaction at this level should by no means be
limited to the LTER program. Important eco-
logical research is occurring at many other loca-
tions and settings, and a mechanism must be
developed over the next year to take advantage
of the potential benefits which can be derived
from interaction.
10.6 CONCLUSIONS
The types of interaction and coordination needed
to create a successful program that best serves
the public interest have been presented. During
the next year, EMAP-Surface Waters will pursue
these coordination efforts and seek to establish
mechanisms to facilitate this coordination. We
envision the need for a scientific advisory panel to
ensure the sound footing of the program and an
interagency coordination panel to facilitate the
interaction and coordination needed among par-
ticipating state and federal agencies.
167
-------
-------
11.0 EXPECTED OUTPUTS AND TIMELINES
Section 11.0 describes the types of products and
outputs expected from EMAP-Surface Waters,
along with expected timelines for field activities.
The actual timing of these activities depends upon
the available funding, but we present a proposed
implementation schedule for discussion purposes.
11.1 REPORTING
11.1.1 Data Evaluation Reports
A key element of EMAP-Surface Waters is the
production of information from the collected data
in a timely manner and the dissemination of that
information to users. Four types of reports are
currently envisioned:
• Annual statistical summary
• Periodic interpretive reports
• Special interest reports
• Scientific articles
To be of maximum use, data must be transformed
into useful information as quickly as possible.
The objective of EMAP-Surface Waters is to pub-
lish summaries of the preceding field season's
surveys within nine months of data collection.
These reports will provide summaries of response,
exposure, and habitat indicators for resources
and regions sampled, but will provide minimal
interpretation. Rigorous statistical evaluation of
trends and associations between indicators will
not be made. These reports are intended simply
to get the information out to interested parties as
quickly as possible. An example of the Annual
Statistical Summary is presented in Paulsen et al.
(1990).
Interpretive reports will be published following
completion of four-year sampling cycles. These
reports will attempt to summarize indicator results
for the preceding four years and integrate infor-
mation from the suite of response, exposure,
habitat, and stressor indicators to determine
regional and national status of lakes and streams.
Trends will be evaluated and the likely causes for
current conditions and changes in condition
assessed. The later will be accomplished by anal-
ysis of the associations between response indi-
cators and the exposure/habitat indicators. This
diagnosis of probable cause will be further eval-
uated using information available for various
stressor indicators. A detailed example of an
interpretive report and the likely approach for
analyses will be developed during the upcoming
year.
Special issue reports will be used to present data
pertaining to specific issues, such as changes in
surface water chemistry resulting from implemen-
tation of the Clean Air Act or other special issue
projects within EMAP-Surface Waters. Special
interest analyses of particular indicators may also
be published in this type of report. General
reports for a broad spectrum of the public will be
produced as part of this series.
Investigators within EMAP, and from programs
outside EMAP, will be actively encouraged to pro-
duce analyses of the EMAP-Surface Waters
results for publication in the peer review literature.
One of the best methods of evaluating the quality
of data produced in EMAP-Surface Waters will be
its use in both policy-oriented and science-
oriented publications.
11.1.2 Internal Program Reports
A series of reports and plans will be required
before the program can commence field activity.
These include, but are not limited to:
• Implementation plans
• Methods manuals
• Quality assurance plans
• Information management plans
• Field operations manuals
• Quality assessment reports
Each of these plans should be considered a
"living" document. This implies that they will be
updated before field implementation each year, as
a result of the evaluation of the previous year's
activities. The Implementation Plan will contain
the objectives and outline of the field activity for
that season. This will include a listing and loca-
tion of all sites to be visited along with indicators
to be collected. Additional details for all aspects
of that season's activities, such as quality assur-
ance, information management, or field opera-
tions, can then be found in updated program
plans and manuals. An assessment of the quality
of data collected will be required following each
field season.
169
-------
11.2 TIMELINES
Table 11-1 presents the timelines proposed for the
implementation of EMAP-Surface Waters. Pilots
are activities designed to test various aspects of
the system, such as Indicator evaluation or samp-
ling and logistic constraints. Site selection for
pilot activities need not be constrained to the
EMAP-Surface Waters sampling grid. Demonstra-
tions are activities in which parts of the system
continue to be tested. - A key aspect ,of the
demonstration Is the use of the sampling grid to
provide policy-relevant information for some sub-
set of the indicators on a population of aquatic
resources of interest. The demonstration may
incorporate additional sampling to evaluate indi-
cator performance or examine components of
variability for indicators of interest.
The timeline proposed in Table 11 -1 represents an
extremely optimistic time frame. It assumes ade-
quate and timely funding for acquiring and train-
ing the staff needed for program planning, imple-
mentation and data analysis.
Table 11-1. Proposed EMAP-Surface Waters Implementation Schedule3
FY91
FY92
FY93
FY94
FY95
FY96
Lake pilot/
demo;
Regions 1, 2
Lake demo;
Regions 1,2
Lake pilot/
demo;
Regions 3,4,&
Stream pilot;
Regions 8,9
Lake demo;
Regions 1,2
Lake demo;
Regions 3,4,5
• -
Lake pilot/
demo;
Regions 6,7,8
Stream demo;
Regions 8,9,10
Stream pilot
Regions 6,7
Lakes;
Regions 1,2
Lake demo;
Regions 3,4,5
Lake demo;
Regions 6,7,8
Lake pilot/
demo;
Regions 9,10
Stream demo;
Regions 8,9,10
Stream demo;
Regions 6,7
Stream pilot;
Regions 3,4
Lakes;
Regions 1,2
Lakes;
Regions 3,4,5
Lake demo;
Regions 6,7,8
Lake demo;
Regions 9,10
Streams;
Regions 8,9,10
Stream demo;
. Regions 6,7
Stream demo;
Regions 3,4
Stream pilot;
Regions 1,2,5
Lakes;
Regions 1,2
Lakes;
Regions 3,4,5
Lakes;
Regions 6,7,8
Lake demo;
Regions 9,10
Streams;
Regions 8,9,10
Streams;
Regions 6,7
Stream demo;
Regions 3,4
Stream demo;
Regions 1 ,2,5
Regions refers to the EPA regions.
Tho laka pilot and demonstration in 1992 are scheduled for EPA Regions 3, 4, and 5. If funding constraints limit the scope of
activities, Region 5 will be the highest priority. This will ensure that lake pilots and demonstrations are completed in the
northern tier of lake states in succeeding years.
170
-------
12.0 FISCAL YEAR 1991 FIELD AND ANALYSIS ACTIVITIES
"It is common sense to take a method and try it. If it fails, admit it frankly and try another. But
above all, try something."
;- Franklin D. Roosevelt
As a first step in accomplishing the objectives of
EMAP-Surface Waters, a field pilot will be imple-
mented in the lakes of the Northeast, EPA
Regions 1 and 2, in 1991. The goal of this activity
is to demonstrate in the Northeast Region the
utility and value of the EMAP-Surface Waters
sampling design and indicator approach, and at
the same time, to collect the information neces-
sary to improve and develop a technically sound
and cost-effective program. Specific objectives
include the following:
• Evaluate the adequacy of the proposed
Tier 2 sample allocation scheme and
collect information to evaluate alternative
, designs.
• Evaluate the reliability, variability and
sensitivity of the proposed response
indicators over a range of environmental
conditions.
• Refine field sampling methodologies.
• Identify and resolve unexpected logistical
problems associated with sampling lakes in
the Northeast.
• Develop and evaluate data handling, quality
assurance, and statistical procedures for
efficient analysis and reporting.
• Produce, on a pilot basis, regional esti-
mates of the condition of lakes in the
Northeast based on a limited suite of
response, exposure, and habitat indicators.
The Temporally Integrated Monitoring of Eco-
systems (TIME) project will also begin in the
Northeast in FY91. Specific objectives associated
with this activity include:
• Use field information to refine the sub-
populations of concern relative to acidic
deposition in the Northeast.
Evaluate the appropriateness of the
proposed grid augmentation for addressing
issues related to the Clean Air Act.
Refine the selection of regional long-term
monitoring sites to represent subpopula-
tions of concern.
Produce regional estimates of lakewater
quality with respect to acidic deposition.;
12.1 GENERAL EMAP-SURFACE WATERS
In order to meet the objects just outlined, the
EMAP-Surface Waters pilot will take place in EPA
Regions 1 and 2, the northeastern United States.
Approximately 60 lakes will be sampled during
this effort. Of these sites, approximately 40 will
be selected by procedures outlined in Section 3.0.
Ten percent of these will be revisited in order to
evaluate index period variability. Indicators
outlined in Section 5.0 will be collected using the
late summer index period.
An additional 20 sites will be sampled to evaluate
indicator performance. Ten highly perturbed
sites, 10 minimally perturbed sites and 10 ran-
domly selected sites will be chosen and .sampled
using the protocols outlined for the regional
demonstration activity. A pilot implementation to
be conducted in spring 1991 will detail the pro-
posed activities.
12.2 TIME PROJECT
The TIME project will be implemented in the
Northeast in 1991 as well. This activity will
consist of continued monitoring of the approxi-
mately 30 LTM sites described in Stoddard (1990),
and an additional 85 sites in the subpopulations
of interest. The regional sampling in 1991 will
help us to (1) confirm the level of augmentation of
the standard grid that will be needed for under-
represented areas and (2) begin to generate new
estimates of regional condition that will be used to
help evaluate the Clean Air Act.
171
-------
12.3 DATA ANALYSIS ACTIVITIES
Many of the data analyses that were begun in
1990 will continue during 1991. The primary effort
will be to develop an approach for examining the
associations between response indicators and
exposure/habitat indicators. The statistical
approaches selected will aid us in resolving
outstanding design issues such as (1) whether to
use the hexagon or watershed characterization
and (2) the density needed to achieve the desired
level of uncertainty in population estimates.
Continued evaluation of existing data will help to
refine aspects of the design such as the level of
change and trend detectable with the variability
that exists in our indicators. We will pursue
increased Interaction with members of the aca-
demic community, the states, and other federal
agencies to Identify data sets and analysis
approaches that might better allow us to reach
our objectives.
12.4 RESEARCH NEEDS
As stated at the outset, EMAP must be accom-
panied by an aggressive research program. The
range of researchable issues is quite long. Key
Issues can be divided into four categories: (1)
Indicators of condition, (2) diagnostic indicators,
(3) approaches for defining nominal/subnominal,
and (4) statistical approaches.
Extensive work Is required to refine our selection
and choice of indicators of condition. Existing
indicators need regional refinement and new indi-
cators are required to fill the gaps. Additional
work needs to be done to improve our diagnostic
capabilities with biological indicators.
The field of diagnostic indicators of exposure and
habitat offers many challenges and potential
benefits. Biomarkers appear to be a potentially
valuable tool in evaluating both condition and
exposure.
Although the decision on what is an acceptable
condition is ultimately a societal one, we have an
obligation to provide creative approaches to
addressing this issue. The process of gathering
scientists, policy analysts, managers, and decision
makers together to tackle this problem will offer
many challenges.
In addressing national assessments such as those
proposed, many statistical challenges exist. The
process of developing indicators from many
metrics is a difficult task but has been accom-
plished successfully. However, the process of
combining multiple indicators into a single state-
ment of condition will require extensive statistical
as well as ecological insight and effort. Trend
analysis at a single site is somewhat straight-
forward but the task of doing trend analysis on
population distributions is not. This effort will
need attention early in the program. In assessing
the effectiveness of regulatory activities, we would
like to evaluate changes and trends in response
indicators and then associate them with changes
or trends in exposure or stressor information.
Regional analysis of associations among indica-
tors and associations of trends in indicators is an
area that must be addressed soon.
172
-------
REFERENCES
Agbeti, M., and M. Dickman. 1989. Use of lake fossil diatom assemblages to determine historical
changes in trophic status. Can. J. Fish.Aquat. Sci. 46:1013-1021.
Baker, J.P., D.P. Bernard, S.W. Christensen, M.J. Sale, and others. 1990a. Biological Effects of
Changes in Surface Water Acid-base Chemistry, SOS/T Report 13 in Acidic Deposition: State of
Science and Technology. National Acid Precipitation Assessment Program, 722 Jackson Place,
Washington, D.C.
Baker,,J.P., S.A. Gherini, S.W. Christensen, C.T. Driscoll, J. Gallagher, R.K. Munson, R.M. Newton, K.H.
Reckhow, and C.L. Schofield. 1990b. Adirondack Lakes Survey: An Interpretative Analysis of Fish
Communities and Water Chemistry, 1984-1987. Adirondack Lakes Survey Corporation, Ray Brook,
NY,
Baker, LA., J.M. Eilers, R.B.. Cook, P.R. Kaufmann, and AT. Herlihy. In press. Current water chemistry
and biogeochemical processes in case study regions: intercomparisons and synthesis. Chapter 18
jn D.F. Charles, ed. Acidic Deposition and Aquatic Ecosystems: Regional Case Studies. Springer-
Verlag, New York.
Baker, LA, P.R. Kaufmann, AT. Herlihy, and J.M. Eilers. 1990. Current Status of Surface Water
Acid-Base Chemistry, SOS/T Report 9, in Acidic Deposition: State of Science and Technology,
National Acid Precipitation Assessment Program, 722 Jackson Place, Washington, D.C. 1990.
Battarbee, R.W. 1986. Diatom analysis. Pages 527-570 in B.E. Berglund, ed. Handbook of Holocene
Paleoecology and Paleohydrology. John Wiley and Sons, New York.
Bascietto, J., D. Hinckley, J. Plafkin, and M. Slimak. 1990. Ecotoxicity and ecological risk assessment:
Regulatory applications at EPA. Environ. Sci. Technol. 24(1):10-15.
Bonoff, M.B. and A.W. Groeger. 1987. National Surface Water Survey, Western Lakes Survey - Phase I,
Field Operations Report. EPA 600/8-87/018. U.S. Environmental Protection Agency, Environmental
Monitoring Systems Laboratory, Las Vegas, Nevada. 32 pp.
Boyle, T.P., G.M. Smillie, J.C. Anderson, and D.R. Beeson. 1990. A sensitivity analysis of nine diversity
and seven similarity indices. J. Water Pollut. Control Fed. 62:749-762.
Brezonik, P.L 1984. Trophic state indices: rationale for multivariate approaches. Pages 441-445 in
Lake and Reservoir Management. EPA/440/5-84/001. U.S. Environmental Protection Agency,
Washington, D.C.
Brooks, R.P., D.E. Arnold, E.D. Bellis, C.S. Keener, and M.J. Croonquist. In Press. A methodology for
biological monitoring of cumulative impacts on wetland, stream and riparian components of
watersheds. Pages (in press). J.A. Kusler, and G. Brooks, eds. Proc. Internat. Symp.: Wetlands and
river corridor management. Assoc. Wetland Managers, Inc., Berne, N.Y.
Brown, J.H., and J. Roughgarden. 1990. Ecology for a Changing Earth. Bull. Ecol. Soc. Amer.
71:173-188.
Carlson, R.E. 1977. A trophic state index for lakes. Limnol. Oceanogr. 22:361-369.
Carlson, C.A., and RT. Muth. 1989. The Colorado River: Lifeline of the American Southwest. Pages
220-239 in D.P. Dodge, ed. Proceedings of the International Large Rivers Symposium. Can. Spec.
Publ. Fish. Aquat. Sci. 106.
173
-------
Charles, D.F. and J.P Smol. In press. The PIRLA II Project: Regional assessment of lake acidification
trends. Verh. Internal Verein. Limnol. • • .
Charles, D.F., M.W. Binford, E.T. Furlong, F.A. Kites, M.J. Mitchell, S.A. Norton, F. Oldfield, M.J.
Paterson, J.P. Smol, AJ. Uutala, J.R.,White, D.R. Whitehead, and R.J.,Wise. 1990. Paleoecological.
Investigation of recent lake acidification in the Adirondack Mountains, N.Y, J: Paleolimnol. 3:195-241.
Church, M.R., K.W. Thornton, P.W. Shaffer, D.L Stevens, B.P. Rochelle, G.R. Holdren, M.G. Johnson,
J.J. Lee, R.S. Turner, D.L. Cassell, DA Lammers, W.G. Campbell, C.I. Liff, C.C. Brandt, LH. Liegel,,
G.D. Bishop, D.C. Mortenson, S.M. Pierson, and D.D. Schmoyer. 1989, Future effects of long-term
sulfur deposition on surface water chemistry in the Northeast and Southern Blue Ridge Province.
EPA-600/3-89/061. Washington, DC: U.S. Environmental Protection Agency.
Colquhoun, J.R., J. Symula, M. Pfeiffer, and J. Feurer. 1981. Preliminary Report of Stream Sampling for
Acidification Studies. Technical Report 812, New York State Department of Environmental
Conservation, Albany, N.Y. ,. . ,
Cougan, K.A., D.W. Sutton, D.V. Peck, V.J. Miller, and J.E. Pollard. 1988. National Stream Survey
-Phase I: Quality Assurance Report. EPA 600/4-88/018. U.S. Environmental Protection Agency, Las
Vegas, Nevada. 205 pp.
Courtemanch, D.L., S.P. Davies, and E.B. Laverty. 1989. Incorporation of. biological information in water
quality planning. Environ. Manage. 13:35-41.
Courtemanch, D.L., and S.P. Davies. 1988. Implementation of biological standards and criteria in
Maine's water classification law. Pages 4-9 in.T.P. Simon, LL Hoist, and LJ. Shepard, eds. .
Proceedings of the First National Workshop on Biological Criteria. EPA/905/9-89/003. U.S.
Environmental Protection Agency, Chicago, IL.
Cowardin, LM., V. Carter, F.C. Golet, and E.T, LaRoe. 1979. Classification of Wetlands and Deepwater
Habitats of the United States. FWS/OBS-79/31. USDI Fish and Wildlife Service. 103 pp.
Cummins, K.W., and M.J. Klug. 1979. Feeding ecology of stream invertebrates. Ann. Rev. Ecol. Syst.
10:147-172.
Cummins, K.W., and M.S. Wilzbach. 1985. Field Procedures for Analysis of Functional Feeding Groups
of Stream Macroinvertebrates. Contribution 1611. Appalachian Environmental Laboratory, University
of Maryland, College Park MD.
Cuslmano, R.F., J.P. Baker, W.J. Warren-Hicks, V. Lesser, W.W. Taylor, M.C. Fabrizio, D.B. Hayes, and
B.P. Baldigo. 1990. Fish Communities in Lakes in Subregion 2B (Upper Peninsula of Michigan) in
Relation to Lake Acidity. EPA/600/3-89/021,, U.S. Environmental Protection Agency, Washington,
D.C.
Delsler, P.P., Jr. 1988. The risk management-risk assessment interface. Environ. Sci. Technol.
220:15-19.
Dlxlt, S.S., J.P. Smol, D.S. Anderson, and R.B. Davis. 1990. Utility of scaled chrysophytes for inferring
lakewater pH In northern New England lakes. J. Paleolimnol. 3:269-286.
Ebel, W.J., C.D. Becker, J.W. Mullan, and H.L Raymond. 1989. The Columbia River - towards holistic
understanding. Pages 205-219 in DP. Dodge, ed. Proceedings of the International Large River
Symposium. Can. Spec. Publ. Fish. Aquat. Sci. 106.
174
-------
Ecology Committee. 1980. Goals and criteria for design of a biological monitoring system. Science
Advisory Board. U.S. Environmental Protection Agency, Washington, D.C. 53 pp.
Einhaus, R.L, D.M. McMullen, R.L Graves, and P.M. Friedman. 1990. Environmental Monitoring and
Assessment Program Quality Assurance Program Plan. Revision 0. U.S. Environmental Protection
Agency, Cincinnati, OH. 123 pp.
Ehrlich, P.R., A.H. Ehrlich, and J.P. Holdren. 1977. Ecoscience: Population, Resources, Environment.
W.H. Freeman, Sanfrancisco, CA. 1051pp.
Eshleman, K.N. 1988. Predicting regional episodic acidification of surface waters using empirical
techniques. Water Resour. Res. 24:1118-1126.
Fava, J.A., W.J. Adams, R.J. Larson, G.W. Dickson, K.L. Dickson, and W.E. Bishop. 1987; Research
Priorities in Environmental Risk Assessment. Society of Environmental Toxicblogy and Chemistry.
Washington, D.C.
Federal Register 48 (217) :51400-51413.
Fjerdingstad, E. 1965. Some remarks on a new saprobic system. Pages 232-235 in Biological Prob-
lems in Water Pollution. Public Health Service Publication No. 999-wp-25. Third Seminar. U.S.
Public Health Service, Cincinnati!, OH.
Fiske, S. 1988. The use of biosurvey data in the regulation of permitted rtonpoint discharges in Vermont
streams. Pages 67-74 in T.P. Simon, L.L Hoist, and LJ. Shepard, eds. Proceedings of the First
National Workshop on Biological Criteria. EPA/905/9-89/003. U.S. Environmental Protection
Agency, Chicago, IL
Ford, J., A.D. Newell, P.L Brezonik, C.T. Driscoll, J.S. Kahl, and K.E. Webster. 1989. In preparation.
Trends in Surface Water Chemistry for Low Alkalinity Lakes of the Northeastern United States:
Results from the EPA long-term Monitoring Project. U.S. Environmental Protection Agency, Corvallis,
OR.
Franson, S. 1990. EMAP Data Confidentiality: Issues and Suggestions. Internal Report. U.S.
Environmental Protection Agency, Las Vegas, Nevada.
Geraghty, J.J., D.W. Miller, F. Van Der Leedeh, and F.L Troise. 1973. Water Atlas of the United States.
Water Information Center, Inc., Port Washington, NY. 122 plates.
Gerritsen, J., J. Dietz, H.T. Wilson, Jr., and A.J. Janicki. 1989. Prediction of Episodic Acidification in
Maryland Coastal Plain Streams. Maryland Dept. of Natural Resources. Annapolis, MD.
Gorman, O.T., and J.R. Karr. 1978. Habitat structure and stream fish communities. Ecology
59:507-515.
Grossman, G.D., J.F. Dowd, and M. Crawford. 1990. Assemblage stability in stream fishes: a review.
Environ. Manage. 14:661-671.
Hart, D.R. 1990. QA Guidelines for Biology in Aquatic Environmental Protection. Prepared for
Environment Canada, Conservation and Protection, Inland Waters Directorate, Burlington, Ontario.
Hagley, C.A., C.L Mayer, and R. Hoenicke. 1989. National Stream Survey - Phase I: Field Operations
Report. EPA 600/4-88/023. U.S. Environmental Protection Agency, Environmental Monitoring
Systems Laboratory, Las Vegas, NV. 34 pp.
175
-------
Heriihy, A.T., D.H. Landers, R.F. Cusimano, W.S. Overton, P.J. Wigington, Jr., A.K. Pollack, and T.E.
Mitchell-Hall. 1991. Temporal Variability in Lakewater Chemistry in the Northeastern United States:
Results of Phase II of the Eastern Lakes Survey. EPA/ . U.S. Environmental Protection Agency,
Washington, D.G.
Hesse, LW., J.C. Schrnulbach, J.M. Carr, K.D. Keenlyne, D.G. Unkenholz, J.W. Robinson, and G.E.
Mesti. 1989. Missouri River fishery resources in relation to past, present, and future stresses. Pages
352-371 In D.P. Dodge, ed. Proceedings of the International Large River Symposium. Can. Spec.
Publ. Fish. Aquat. Sci. 106.
Hllsenhoff, W.L 1987. An improved biotic index of organic stream pollution. Gr. Lakes Entomol.
20:31-39.
Hilsenhoff, W.L 1988. Rapid field assessment of organic pollution with a family level biotic index. J.
North Am. Benthol. Soc. 7:65-68.
Hirsch, R.M., W.M. Alley, and W.G. Wilber. 1988. Concepts For a National Water-Quality Assessment
Program. U.S. Geological Survey Circular 1021. 42pp.
Horvltz, D.G., and D.J Thompson. 1952. A generalization of sampling without replacement from a finite
universe. J. Am. Statist Assoc. 47:663-685.
Hughes, R.M. 1989. Ecoregional biological criteria. Pages 147-151 in Water Quality Standards for the
21st Century. U.S. Environmental Protection Agency, Office of Water, Washington, D.C.
Hughes, R.M., and S.G. Paulsen. 1990. Indicator strategy for inland surface waters. Pages 4.1-4.20 in
C.T. Hunsaker and D.E. Carpenter, eds. Environmental Monitoring and Assessment Program Eco-
logical Indicators. EPA/600/3-90/060. U.S. Environmental Protection Agency, Research Triangle
Park, NC.
Hughes, R.M., D.P. Larsen, and J.M. Omernik. 1986. Regional reference sites: a method for assessing
stream potentials. Environ. Manage. 10:629-635.
Hughes, R.M., C.M. Rohm, T.R. Whittier, and D.P. Larsen.. 1990. A regional framework for establishing
recovery criteria. Environ. Manage. 14:673-684.
Hughes, R.M., S.A. Thiele, D. McMuIlen, J. Lazorchek, S.G. Paulsen, J.P. Baker, D. Peck, J. Pollard, D.P.
Larsen, S. Hedthe, T.R. Whittier, P.R. Kaufmann, and J.R. Karr. 1990. Appendix B: indicator fact
sheets for Inland surface waters. Pages B.1-B.32 in C.T. Hunsaker and D.E. Carpenter, eels. Eco-
logical Indicator Report for the Environmental Monitoring and Assessment Program. U.S. Environ-
mental Protection Agency, Research Triangle Park, NC.
Hunsaker, C.T., and D.E. Carpenter, eds. Ecological Indicator Report for the Environmental Monitoring
and Assessment Program. U.S. Environmental Protection Agency, Research Triangle Park, NC.
Hunsaker, C.T., R.L Graham, G.W. Suter II, R.V. O'Neill, LW. Barnthouse, and R.H. Gardner. 1990.
Assessing ecological risk on a regional scale. Environ. Manage. 14(3):325-332.
Hunt, D.T.E., and A.L Wilson. 1986. The Chemical Analysis of Water. Second Edition. Royal Society
of Chemistry, London, England. 683 pp.
Hynes, H.B.N. 1974. The Blolgoy of Polluted Waters. University of Toronto Press, Toronto, Ontario.
Jackson, DA, H.H. Harvey, and KM. Somers. 1990. Ratios in aquatic sciences: statistical short-
comings with mean depth and the morphoedaphic index. Can. J. Fish. Aquat. Sci. 47:1788-1795.
176
-------
Jenkinson, J.J. 1981. Endangered or threatened aquatic mollusks of the Tennessee River system. Bull.
Am. Malacolog. Union. 1981:43-45.
Karr, J.R. 1981. Assessment of biotic integrity using fish communities. Fisheries 6:21-27:
Karr, J.R., and D.R. Dudley. 1981. Ecological perspective on water quality goals. Environ. Manage.
5:55-68.
Karr, J.R., LA. Toth, and G.D. Garman. 1983. Habitat Preservation for Midwest Stream Fishes-
Principles and Guidelines. EPA/600/3-83/006. U.S. Environmental Protection Agency, Corvallis,
OR. . -• - •• - . ••• ;?'••-...:-.•:
Karr, J.R., L.A. Toth, and D.R. Dudley. 1985. Fish communities of midwesterh rivers: A history of
degradation. BioScience 35:90-95.
Karr, J.R., D.D. Fausch, P.L Angermeier, P.R. Yant, and I.J. Schlosser. 1986. Assessing biological
integrity in running waters: a method and its rationale. Special Publication 5. Illinois Natural History
Survey, Champaign, IL 28 pp.
Kaufmann, P.R., A.T. Herlihy, J.W. Elwood, M.E. Mitch, W.S. Overton, M.J. Sale, J.J. Messer, K.A.
Cougan, D.V. Peck, K.H. Reckhow, A.J. Kinney, S.J. Christie, D.D. Brown, C.A. Hagley, and HI.
Jager. 1988. Chemical Characteristics of Streams in the Mid-Atlantic and Southeastern United
States. Volume I: Population Descriptions and Physico-Chemioal Relationships.
EPA/600/3-88/021 a. U.S. Environmental Protection Agency, Washington, D.C. 397pp.
Kaufmann, P.R., A.T. Herlihy, M.E. Mitch, and W.S. Overton. in press. Stream chemistry in the eastern
United States: 1. Synoptic survey design, acid-base status and regional patterns. Water Resour.
Res. , - .'•• , ••'•••• . .... •. • • •-•''•. •
Kirchner, C.J. 1983. Quality control in water analyses. Environ. Sci. Techno!. 17:174A-181A.
Knapp, C.M., C.L Mayer, D.V. Peck, J.R. Baker, and G.J. Filbin. 1987. National Surface Water Survey,
National Stream Survey (Phase I - Pilot Survey) Field Operations Report. EPA 600/8-87/019. U.S.'
Environmental Protection Agency, Environmental Monitoring Systems Laboratory, Las Vegas, NV.
110 pp.
Knapp, C.M., W.P. Saunders, Jr., D.G. Heimbuch, H.S. Greening, and G.J. Filbin. 1988. Maryland
Synoptic Stream Chemistry Survey: Estimating the Number of Distribution of Streams AffectecI by or
at Risk from Acidification. Prepared for the Maryland Department of Natural Resources, Power Plant
Research Program. Report No. AD-88-2.
Knapp, C.M., D.R. Marmorek, J.P. Baker, K.W. Thornton, J.M. Klopatek, and D.P. Charles. 1991. The
Indicator Development Strategy for the Environmental Monitoring and Assessment Program. U.S.
EPA Environmental Research Laboratory, Corvallis, OR.
Knowlton, M.F., M.V. Hoyer, and J.R. Jones. 1984. Sources of variability in phosphorus and chlorophyll
and their effects on use of lakes survey data. Water Resour. Bull. 20:397-407.
Landers, D.H., J.M. Eilers, D.F. Brakke, W.S. Overton, P.E. Kellar, M.E. Silverstein, R.D. Schonbrod, R.E.
Crowe, R.A. Linthurst, J.M. Omernik, S.A. Teague, and E.P. Meier. 1987. Characteristics of takes In
the Western United States. Volume I. Populations Descriptions and Physico-chemical Relationships.
EPA/600/3-86/054a. U.S. Environmental Protection Agency, Washington, D.C. 176pp. ,
Landres, P.B. 1983. Use of the guild concept in environmental impact assessment. Environ, Manage.
7:393-398. ; <
177
-------
Lee, D.S., C.R. Gilbert, C.H. Hocutt, R.E. Jenkins, D.E. McAllister and J.R. Stauffer, Jr. 1980. Atlas of
North American Freshwater Fishes. North Carolina State Museum of Natural History Raleigh NC
854pp.
Lenat, D.R. 1988. Water quality assessment of streams using a qualitative collection method for benthic
macroinvertebrates. J. North Am. Benthol. Soc. 7:222-233.
Leopold, LB., G.M. Wolman, and J.P. Miller. 1964. Fluvial Processes in Geomorphology. W.H
Freeman and Co., San Francisco, CA. 522 pp.
Lind, O.T. 1974. Handbook of Common Methods in Limnology. Mosby. Saint Louis, MO. 154pp.
LInthurst, R.A., D.H. Landers, J.M. Eilers, D.F. Brakke, W.S. Overton, E.P. Meier, and R.E. Crowe. 1986.
Characteristics of Lakes in the Eastern United States. Volume I: Population Descriptions and
Physico-chemical Relationships. EPA-600/4-86/007a. U.S. Environmental Protection Agency
Washington, D.C.
Loftis, J.C., R.C. Ward, R.D. Phillips, and C.H. Taylor. 1989. An Evaluation of Trend Detection
Techniques for Use In Water Quality Monitoring Programs. EPA/600/3-89/037. US. Environmental
Protection Agency, Corvallis, OR.
Lowe, R.L 1974. Environmental Requirements and Pollution Tolerance of Freshwater Diatoms.
EPA/670/4-74/005. U.S. Environmental Protection Agency, Cincinnati, OH. 333 pp.
Lowe, T.P., T.W. May, W.G. Brumbaugh, and DA Kane. 1985. National Contaminant Biomonitorlng
Program: Concentrations of Seven Elements in Freshwater Fish, 1978-1981.
Madow, W.G. 1949. On the theory of systematic sampling II. Ann. Math. Stat. 20:333-354.
Matthews, W.J. 1986. Fish faunal structure in an Ozark stream: stability, persistence, and a
catastrophic flood. Cope/a 1986:388-397.
Matthews, W.J. 1990. Spatial and temporal variation in fishes of riffle habitats: a comparison of
analytical approaches for the Roanoke River. Am. Midi. Nat. 124:31-45.
Messer, J.J., C.W. Ariss, J.R. Baker, S.K. Drouse, K.N. Eshleman, P.R. Kaufmann, R.A. Linthurst J M
Omernik, W.S. Overton, M.J. Sale, R.D. Schonbrod. S.M. Stambaugh, and J.R. Tuschall, Jr. 'l986
National Stream Survey Phase I - Pilot Survey. EPA/600/4-86/026. U.S. Environmental Protection
Agency, Washington, D.C. 179 pp.
Messer, J.J., C.W. Arris, J.R. Baker, S.K. Drouse, K.N. Eshleman, A.J. Kinney, W.S. Overton, M.J. Sale,
and R.D. Schonbrod. 1988. Stream chemistry in the Southern Blue Ridge: feasibility of a regional
synoptic sampling approach. Water Resour. Bull. 24:821-829.
Metcalf, J.L 1989. Biological water quality assessments of running waters based on macroinvertebrate
communities: history and present status in Europe. Environ. Pollut. 60:101-139.
Miller, D.L, P.M. Leonard, R.M. Hughes, J.R. Karr, P.B. Moyle, LH. Schrader, B.A. Thompson, R A
Daniels, K.D. Fausch, G.A. Fitzhugh, J.R. Gammon, D.B. Halliwell, P.L Angermeier, and D.J. Orth.
1988. Regional applications of an index of biotlc integrity for use in water resource management
Fisheries 13:12-20.
Miller, R.R., J.D. Williams, and J.E. Williams. 1989. Extinctions of North American fishes during the oast
century. Fisheries 14:22-38.
178
-------
Morgan, R.P.,.and AJ. Janicki. In press. Evaluation of Relationships between Sensitivity to
Acidification, Watershed Characteristics, and Fish Distributions in Western Maryland. Maryland
Department of Natural Resources, Annapolis, MD.
Morris, F.A., D.V. Pecki M.B. Bonoff, K.J. Cabbie, and S.L. Pierett. 1986. National Surface Water Survey,
Eastern Lake Survey (Phase I - Synpotic Chemistry) Field Operations Report. EPA 600/4-86/010.
U.S. Environmental Protection Agency, Environmental Monitoring Systems Laboratory, Las Vegas,
NV. 46pp. ; .-..-.
Moyer, S. 1986. Aquatic Contaminants: A Threat to the Sport Fishing Industry. Sport Fishing Institute,
Washington, D.C. 165 pp.
Murdoch, P.S. 1986. Chemical Input-Output Budgets and Stream Chemistry Dynamics during a
Two-year Period in Biscuit Brook, Catskill Mountains, New York. U.S. Geological Survey Water
Resources Investigations Report 86 (Draft). Prepared in cooperation with U.S. Environmental
Protection Agency.
Nash, R.R. 1989. The Rights of Nature: A History of Environmental Ethics. University of Wisconsin
Press, Madison, Wl. 290pp.
National Research Council. 1990. A Review of the USGS National Water Quality Assessment Pilot
Program. National Academy Press, Washington, D.C. ,
Nebeker.AV., M.A. Cairns, J.H. Gakstatter, K.W. Malueg, G.S. Schuytema, and D.F. Krawczyk. 1984.
Biological methods for determining toxicity of contaminated freshwater, sediments to invertebrates.
Environ. Toxicol. Chem. 3:617-630.
Nelson, M.K., C.G. Ingersoll, and F.J. Dwyer. 1989. New Standard Guide for Conducting Solid-phase
Toxicity Tests with Freshwater Invertebrates. Draft #5. ASTM Committee E47 on Biological Effects
and Environmental Fate, Subcommittee E47.03 on Sediment Toxicity. American Society for .Testing
and Materials, Philadelphia, PA.
Nix, J.F., K.W. Thornton, D.E. Ford, and J.T. Majcolm. 1986. Storm Event Sampling of Southwestern
Arkansas Streams Susceptible to Acid Precipitation. U.S. Environmental Protection Agency
Cooperative Agreement 811863-01-0 (Draft). <
Northcote, T.G. 1988. Fish in the structure and function of freshwater ecosystems: A "top-down" view.
Can. J. Aquat. Sci. 45:361-379. • , . ,
Noss, R.F. 1990. Indicators for monitoring biodiversity:, A hierarchical approach. Conserv. Biol.
4:355-365.
Oberdorff, T., and R.M. Hughes. In press. Modification and application of the Index of Biotic Integrity to
the Seine River, France. Hydrobiologia
Ohio EPA. 1988. Biological Criteria for the Protection of Aquatic Life. Division of Water Quality
Monitoring and Assessment, Columbus, OH. 384 pp.
Olem, H. 1986. Episodic Changes in Stream Water Quality in Five Watersheds in the Southern Blue
Ridge Province. Interim Report. Chattanooga, TN.
Omernik,J.M. 1977. Nonpoint Source Stream Nutrient Level Relationships: A Nationwide Study.
EPA/600/3077/105. U.S. EPA Environmental Research Laboratory, Corvallis, OR.
179
-------
Overton, W.S. 1985. Analysis Plan for the Eastern Lake Survey. Working draft, March 1985. Technical
Report 113. Department of Statistics, Oregon State University, Corvallis, OR.
Overton, W.S. 1987. Phase II Analysis Plan, National Lake Survey. Working draft, April 1987. Technical
Report 115. Department of Statistics, Oregon State University, Corvallis, OR.
Overton, W.S. 1989. Effects of Measurement and Other Extraneous Errors on Estimated Distribution
Functions In the National Surface Water Surveys. Department of Statistics Technical Report 129.
Oregon State University, Corvallis, OR.
Overton, W.S., and S.V. Stehman. 1987. An Empirical Investigation of Sampling and Other Errors in the
National Stream Survey: Analysis of a Replicated Sample of Streams. Technical Report 119.
Department of Statistics, Oregon State University, Corvallis, OR.
Overton, W.S., D. White, and D.L Stevens. 1990. Design Report for EMAP. Environmental Monitoring
and Assessment Program Part I. Draft. Department of Statistics, Oregon State University, Corvallis,
OR.
Peck, D.V., J.M. Engels, K.M. Howe, and J.E. Pollard. 1988. Aquatic Effects Research Program:
Episodic Response Project Integrated Quality Assurance Plan. Revision!. EPA 600/X-88/274. U.S.
Environmental Protection Agency, Las Vegas, NV.
Plafkln, J.L, M.T. Barbour, K.D. Porter, S.K. Gross, and R.M. Hughes. 1989. Rapid Bioassessment
Protocols for Use in Streams and Rivers: Benthic Macroinvertebrates and Fish. EPA 444/4-89/001.
U.S. Environmental Protection Agency, Washington, D.C. 172pp.
Platts, W.S. 1979. Livestock grazing and riparian/stream ecosystems - an overview. Pages 39-45 in
O.B. Cope, ed. Proceedings of the Forum - Grazing and Riparian/Stream Ecosystems. Trout
Unlimited, Denver, CO.
Platts, W.S., W.F. Megahan, and G.W. Minshall. 1983. Methods for Evaluating Stream, Riparian, and
Blotic Conditions. General Technical Report INT-138. U.S. Department of Agriculture, U.S. Forest
Service, Ogden, UT.
Rahel, FJ. In press. The hierarchical nature of community persistence: a problem of scale. Am. Nat.
Rapport, DJ. 1989. What constitutes ecosystem health? Perspect. Biol. Med. 33:120-132.
Robblns, C.S., D. Bystrak, and P.H. Geissler. 1986. The Breeding Bird Survey: Its First Fifteen Years,
1965-1979. Resource Publication 157. United States Department of the Interior, Fish and Wildlife
Service, Washington, D.C. 196 pp.
Ross, S.T., WJ. Matthews, and A.A. Echelle. 1985. Persistence of stream fish assemblages: effects of
environmental change. Am. Nat. 126:24-40.
Ruttner, F. 1963. Fundamentals of Limnology. Translated by D.G. Frey and F.E.J. Frey. University of
Toronto Press, Toronto, Ontario. 295 pp.
Schaeffer, D.J., E.E. Herricks, and H.W. Kerster. 1988. Ecosystem health: 1. Measuring ecosystem
health. Environ. Manage. 12:445-455.
Schtndler, D.W. 1987. Detecting ecosystem responses to anthropogenic stress. Can; J. Fish. Aquat.
Scf. 44:6-25.
180
-------
Schmitt, C.J., and W.G. Brumbaugh. 1990. National Contaminant Biomonitoring Program: concentra-
tions of arsenic, cadmium, copper, lead, mercury, selenium, and zinc in U.S. freshwater fish,
1976-1984. Arch. Environ. Contam. Toxicol. 19:731-747.
Schmitt, C.J., J.L Zajicek, and M.A. Ribick. 1985. National pesticide monitoring program: Residues of
organochlorine chemicals in freshwater fish, 1980-81. Arch. Environ. Contam. Toxicol. 14:225-260.
Schmitt, C.J., J.L. Zajicek, and P.H. Peterman. 1990. National Contaminant Biomonitoring Program:
residues of organochlorine chemicals in U.S. freshwater fish, 1976-1984. Arch. Environ. Contam,
Toxicol. 19:748-782.
Shapiro, J. 1979. The current status of lake trophic indices: a review. Pages 53-99 in I.E. Maloriey,
ed. Lake and Reservoir Classification Systems. EPA/600/3-79/074. U.S. Environmental Protection
Agency, Washington, D.C.
Silverstein, M.E., M.L Faber, S.K. Grouse and I.E. Mitchell-Hall. 1987. Western Lake Survey-phase I:
Quality Assurance Report. EPA/600/4-87/037. U.S. Environmental Protection Agency, Las Vegas,
NV.
Simpson, P.W., J.R. Newman, M.A. Keirn, R..M. Matter, and P.A. Guthrie. 1982. Manual of Stream
Channelization Impacts on Fish and Wildlife. FWS/OBS-82/24. U.S. Fish and Wildlife Service,
Washington, D.C.
Smart, M.M., J.R. Jones, and J.L. Sebaugh. 1985. Stream-watershed relations in the Missouri Ozark
Plateau Province. J. Environ. Qua/. 14:77-82. , :
Smith, F., S. Kulkarni, LE, Myers, and M.J. Messner. 1988, Evaluating and presenting quality assurance
sampling data. Pages 157-168 in L.H. Keith, ed. Principles of Environmental Sampling. American ;
Chemical Society, Washington, D.C. .
Smith, R.A., R.B. Alexander, M. G. Wolman. 1987. Water-quality trends in the nation's rivers. Science
235:1607-1615.
Snedecor, G.W., and W.G. Cochran. 1967. Statistical Methods. Sixth Edition. The Iowa State
University Press, Ames, IA. 507pp.
Stanley, T.W., and S.S Verner. 1985. The U.S. Environmental Protection Agency's quality assurance
program. Pages 12-19 in J.K. Taylor and T.W. Stanley, eds. Quality Assurance for Environmental
Measurements. ASTM STP 867. American Society for Testing and Materials, Philadelphia, PA.
Steedman, R.J. 1988. Modification and assessment of an index of biotic integrity to quantify stream
quality in southern Ontario. Can. J. Fish. Aquat. Sci. 45:492-501.
Stehman, S.V., and W.S. Overton. 1987a. Estimating the variance of the Horvitz-Thompson estimator in
variable probability, systematic samples. Proceedings of the Section on Survey Research Methods of
the American Statistical Association.
Stehman, S.V., and W.S. Overton. 1987b. An Empirical Investigation of the Variance Estimation Method-
ology Prescribed for the National Stream Survey: Simulated Sampling from Stream Data Sets. ,
Technical Report 118. Department of Statistics, Oregon State University, Corvallis, OR.
Stevens, D.L, W.S. Overton, and A. Herstrom. 1987. Spatial representation of National Lake Survey /
data. Manuscript.
181
-------
Stoddard, J. 1990. Environmental Monitoring and Assessment Program: Plan for Converting the >
NAPAP Aquatic Effects Long-Term Monitoring (LTM) Project to the Temporally Integrated Monitoring
of Ecosystems (TIME) Project. Internal Report. U.S. EPA Environmental Research Laboratory,
Corvallls, OR. ,, •.,, .
Stoermer, E.F., J.A. Wolin, C.L Schelske, and D.J. Conley. 1985. An assessment of ecological changes
during the recent history of Lake Ontario based on siliceous algal microfossils preserved in the
sediments. J, Phycol. 21:257-276. :• ••',.'•'
Strahler, A.N. 1957. Quantitative analysis of watershed geomorphology. Trans. Am. Geophys. Union
38:913-920.
Study Group on Environmental Monitoring. 1977. Environmental Monitoring. National Research
Council, National Academy of Sciences. Washington, D.C. 173 pp.
Strter, G.W. 1990. Endpoints for regional ecological risk assessments. Environ. Manage. 14:9-23.
Taylor, J.K. 1988. Quality Assurance of Chemical Measurements. Lewis Publishers, Chelsea, Ml.
328 pp.
Thornton, K.W., B.L Kimmel, and F.E. Payne. 1990. Reservoir Limnology. Ecological Perspectives.
John Wiley and Sons, New York. 256 pp.
Tonn, W.M. 1990. Climate change and fish communities: a conceptual framework. Trans. Am. Fish.
Soc. 119:337-352,
Tonn, W.M., and J.J. Magnuson. 1982. Patterns in the species composition and richness oi: fish assem-
blages In northern Wisconsin lakes. Eco/ogy63:1149-1166.
Uutala, A.J. 1990. Chaoborus (Diptera:Chaoboridae) mandibles - paleolimnological indicators of the
historical status of fish populations in acid-sensitive lakes. J. Paleolimnol. 4:in press.
U.S. Department of Interior. 1989. News release. July 17. U.S. Department of the Interior, Washington,
D.C. . ,
U.S. EPA. 1984a. Monitoring Strategy. Office of Water, Washington, D.C. 132pp.
• 1984b. Waterbody Surveys and Assessments for Conducting Use Attainability Analyses.
Volume III: Lake Systems. Technical Support Manual. Office of Water Regulations and Standards,
Washington, D.C. ,
1987a. Surface Water Monitoring: A Framework for Change. Office of Water and Office
of Policy, Planning, and Evaluation, Washington, D.C. 57 pp.
_. 1987b. Unfinished Business: A Comparative Assessment of Environmental Problems.
Office of Policy Analysis and Office of Policy, Planning, and Evaluation, Washington, D.C. 100 pp.
• 1987c. National Water Quality Inventory. 1986 report to Congress. EPA/440/4-87/008.
Office of Water, Washington, D.C. 184pp.
. 1988. Report of the National Workshop on Instream Biological Monitoring and Criteria.
Office of Water Regulations and Standards, Washington, D.C. 34pp.
. 1989a. Nonpolnt Sources Agenda for the Future. Office of Water, Washington, D.C.
31 pp.
182
-------
_. 1989b. Short-term Methods for Estimating the Chronic Toxicity of Effluents and
Receiving Waters to Freshwater Organisms. EPA/600/4-89/001: U.S. EPA Environmental Monitoring
Systems Laboratory, Cincinnati, OH. '
, Region IV. 1989. Region IV Workshop on Biomonitoring and Biocriteria. Water Quality
Management Branch and Ecological Support Branch, Atlanta, GA. 26 pp.
• 1990a. Environmental Monitoring and Assessment Program: Guidelines for Preparing
Logistics Plans. EPA/600/X-90/161. U.S. Environmental Protection Agency, Environmental
Monitoring Systems Laboratory, Las Vegas, NV. 31 pp.
. 1990b. Environmental Monitoring and Assessment Program: Quality Assurance
Program Plan. Revision 1. U.S. Environmental Protection Agency, Cincinnati, Ohio.
_. 1990c. Volunteer Water Monitoring: A Guide for State Managers. EPA/440/4-90/010.
Office of Water, Washington, D.C. 78 pp.
. 1990d. National Water Quality Inventory. 1988 Report to Congress. EPA/440/4-90/003.
Office of Water, Washington, D.C. 187pp.
. 1990e. Reducing Risk: Setting Priorities and Strategies for Environmental Protection.
SAB-EC-90-021. U.S. Environmental Protection Agency, Science Advisory Board, Washington, D.C.
U.S. General Accounting Office (GAO). 1981. Better Monitoring Techniques are Needed to Assess the
Quality of Rivers and Streams. Vol. I. CED-81:30. Washington, D.C.
_. 1988. Protecting Human Health and the Environment through Improved Management.
GAO/RCED-88-101. Washington, D.C. 246 pp.
Vannote, R.L, G.W. Minshall, K.W. Cummins, J.R. Sedell, and C.E. Gushing. 1980. The river continuum
concept. Can. J. Fish.Aquat.Sci. 37:130-137.
Walker, I.R. 1987. Chironomidae (Diptera) in paleoecology. 'Quat. Sci. Rev. 6:29-40.
Walker, W.W., Jr. 1984. Trophic state indices in reservoirs. Pages 435-440 in Lake and Reservoir
Management. EPA/440/5-84/001. U.S. Environmental Protection Agency, Washington, D.C.
Warren, C.E. 1974. Biology and Water Pollution Control. W.B. Saunders, Philadelphia, PA.
Warren, C.E., W.J. Liss, and M.J. Wevers. 1987. Artificial Streams for Ecosystem Toxicity Studies.
EA-5079. Electric Power Research Institute, Palo Alto, CA.
Warren-Hicks, W., B.R. Parkhurst, and S.S. Baker, Jr., eds: 1989. Ecological Assessment of Hazardous
Waste Sites: A Field and Laboratory Reference Document. EPA/600/3-89/013. U.S. Environmental
Protection Agency, Office of Research and Development, Washington, D.C. 251 pp plus appendix.
Williams, J.E., J.E. Johnson, D.A. Hendrickson, S. Contreras-Balderas, J.D. Williams, M. Navarro-
Mendoza, D.A. McAllister, and J.E. Deacon. 1989. Fishes of North America endangered, threatened,
or of special concern: 1989. Fisheries 14:2-20.
Witt, E.G., and J.L. Barker. 1986. Stream Chemistry Response during Episodes of Acidic Rainfall and
Snowmelt Runoff on Laurel Hill, Somerset County, Pennsylvania, November 1983 to July 1985. U.S.
Geological Survey Open File Report 86 (Draft). Prepared in cooperation with U.S. Environmental
Protection Agency.
183
-------
World Resources Institute. 1990. World Resources 1990-91: A Report by the World Resources
Institute. Oxford University Press, New York.
Yosle, R.F. 1987. EPA's risk assessment culture. Environ. Sci. Technol. 21(6):526-531<
*U.S. GOVERNMENT PRINTING OFFICE:! 991 -5t8 -1 8 7/256 H2
184
-------
-------
to
75
=• o>
(A
0)
8
m
1
o
1
cb
to
to
I 2^ "
•88
If!
:r o o
til
~2l
> 0) U
3" O tfl
Q) D- (D
a?3
3 O -,
5 o S
5 ^ =
•"* QJ -.
il
CD O
o. m
o. o
_. m
2 m
D •<
So
3-
3-g"
° o
> me
3 y
(C Q)
E »
oj «
------- |