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 «
                                                                                 
-------