(
United States
Environmental Protection
Agency
Office of Research and
Development
Washington, D.C. 20460
EPA/600/R-00/087
September 2000
&EPA
Testing Landscape
Indicators for Stream
Condition Related to
Pesticides and Nutrients:
Landscape Indicators for
Pesticides Study for
Mid-Atlantic Coastal Streams
(LIPS-MACS)
200
200
Coastal Lowlands
•I Mid Coastal Plain, mixed sediments
•I Mid Coastal Plain, fine sediments
[ ""I Mid Coastal Plain, Sand + Overlying Gravel
Deeply Dissected Sand + Overlying Gravel
Inner Coastal Plain
Alluvium + Estuarine Valleys
027LEB01 COV * 10/24/00
-------
Testing Landscape Indicators for Stream
Condition Related to
Pesticides and Nutrients:
Landscape Indicators for Pesticides Study
for Mid-Atlantic Coastal Streams
(LIPS-MACS)
Ann M. Pitchford1
Judith M. Denver2
Anthony R. Olsen3
Scott W. Ator2
Susan Cormier4
Maliha S. Nash1
Megan H. Mehaffey1
'U.S. Environmental Protection Agency
National Exposure Research Laboratory
Environmental Sciences Division
P.O. Box 93478
Las Vegas, NV 89193-3478
2U.S. Geological Survey
Baltimore District Office
8987 Yellow Brick Road
Baltimore, MD 2 1237
3 U.S. Environmental Protection Agency
National Health and Environmental Effects Research Laboratory
Western Ecology Division
2 1 1 1 S.E. Marine Science Drive
Corvallis, OR 97333-4902
4U.S. Environmental Protection Agency
National Exposure Research Laboratory
Ecological Exposure Research Division
26 West Martin Luther King Drive
Cincinnati, OH 45268
September 2000
-------
NOTICE
The U.S. Environmental Protection Agency (EPA), through its Office of Research and
Development (ORD), funded and performed the research described here. This manuscript has
been subject to external and EPA peer review and approved for publication. Mention of trade
names or commercial products does not constitute endorsement or recommendation by the EPA
for use.
11
-------
PROJECT SUMMARY
Title: Testing Landscape Indicators for Stream Condition Related to Pesticides and Nutrients:
Landscape Indicators for Pesticides Study for Mid-Atlantic Coastal Streams (LIPS-MACS)
Principal Investigators: Ann Pitchford (EPA) and Judy Denver (USGS)
Project Goals:
This project is the first study in a long term, national research program, the Landscape
Indicators for Pesticides Study (LIPS). The project is being conducted in the Mid-Atlantic
Coastal Streams (MACS); the U.S. Geological Survey (USGS) is collaborating in the study
through the National Water Quality Assessment (NAWQA) program. The main goal of the
project is to develop landscape indicator models, also termed "landscape indicators," for
pesticides, nutrients, and toxic chemicals in stream water and sediments. Landscape indicator
model development involves the statistical comparison of physical or biological data
characterizing streams (e.g., nutrient, pesticide, or toxic chemical concentrations, or biotic
community composition and abundance), with corresponding spatial information for the stream
and its valley. Besides surficial landscape features such as land cover, slope, and stream features,
this study will include data on soils and hydrogeologic conditions in the analyses.
Approach/Methods:
With the experience gained from evaluating existing data, this study was designed to
obtain collocated water quality, bed sediments, physical habitat, and benthic macroinvertebrate
samples for first-order watersheds for a variety of geologic, hydrologic, and landscape settings,
grouped by hydrogeologic conditions. The hydrogeologic conditions have been synthesized into
a generalized framework of information on physiography, bulk texture of surficial sediments,
topography, and subcropping geology. Seven units have been delineated within the Mid-Atlantic
Coastal Plain (MACP). Each has relatively consistent, natural processes which are expected to
govern the interchange of chemicals between surface and ground waters. Watersheds will be
chosen to provide gradients in developed versus undeveloped land cover types. The field study
will take place during the spring, providing a one-time-only "snapshot" of streams across the
entire area. Water samples will be collected under conditions which represent shallow ground
water contributions to the streams. Measurements proposed include pesticides, pesticide
metabolites, nutrients, and major ions for stream water; physical habitat surrounding the stream
at the sampling point; benthos community composition and abundance; and pesticides, mercury,
arsenic, and PCBs in bed sediments. These data and indices based on these data will be the
dependent variables in the landscape indicator models to be developed using independent
variables such as land cover, topography, soil type, geologic and hydrologic characteristics,
population density, length of roads in watersheds, and mean distance between roads and streams.
The hydrogeologic framework unit will be evaluated as an explanatory variable in the landscape
indicator models. In addition, the differences in results among the hydrogeologic framework
units will be used to evaluate the hypotheses underlying the delineation of the units. Project
resources are leveraged with support from the USGS' NAWQA program and other smaller
projects within the same geographic area.
111
-------
Significance of Research:
In areas with substantial agriculture, industry, or urban development, pesticides and
nutrients, industrial chemicals, pharmaceuticals, and other chemicals can dramatically affect
water quality and biota in streams. The landscape setting, i.e., the location of a stream within its
valley, and the relative proportions of land uses combined with the topography and related
physical features, is expected to be a significant factor in assessing a watershed's condition in
relation to these stressors. Landscape indicators can characterize the landscape setting by
statistically combining and summarizing relevant spatial data. Since measurements are not
possible in every watershed because of cost and practical constraints, these landscape indicators
may offer a means to efficiently estimate the condition of streams with respect to pesticides,
nutrients, and other chemicals in the MACP.
IV
-------
TABLE OF CONTENTS
NOTICE ii
PROJECT SUMMARY iii
PROJECT NARRATIVE 1
INTRODUCTION 1
Purpose of Study 1
Rationale for Study 2
Participants 5
Organization of Research Plan 6
PROJECT OVERVIEW 7
General 7
Objectives 10
Hypotheses 10
Unique Features 12
DESCRIPTION OF STUDY AREA 13
LITERATURE REVIEW 15
Pesticide and Fertilizer Use in the Mid-Atlantic Coastal Plain 15
Pesticides and Nitrates Measured in Streams in the Mid-Atlantic
Coastal Plain 19
Benthic Macroinvertebrates as the Ecological Endpoint... 21
Hydrogeologic Framework 25
Landscape Indicator Models 29
FIELD STUDY 34
Statistical Design 34
Logistics/Methods 39
DATA ANALYSIS 44
Overview 44
Landscape Indicator Models 48
Hydrogeologic Framework 50
Data Management 50
HYDROLOGIC AND MULTIMEDIA MODELING 51
-------
LANDSCAPE INDICATOR APPLICATIONS 53
rT
Potential Applications •"
CT
Stakeholders and Outreach •"
QUALITY ASSURANCE 54
SCHEDULE AND MILESTONES 57
POTENTIAL FOR REDUCING UNCERTAINTY 59
PERFORMANCE MEASURES 60
ANTICIPATED RESULTS/PRODUCTS 61
LITERATURE CITED 62
APPENDIX A. U.S. Geological Survey Pesticide Schedule 2001 75
APPENDDC B. Target Analytes for Sediment Analyses 78
APPENDIX C. U.S. Geological Survey Schedule 2701 80
APPENDIX D. U.S. Geological Survey Schedule 2702 80
APPENDIX E. Benthic Macroinvertebrate Indices 81
APPENDIX F. Physical Habitat Metrics 82
APPENDLX G. Spatial Databases 83
VI
-------
LIST OF TABLES
Table 1. Pesticide and Fertilizer Usage for Corn in North Carolina based on
Surveys Conducted in 1997 and 1998 (NASS, 1999) 16
Table 2. Pesticide and Fertilizer Usage for Soybeans in North Carolina based on
Surveys Conducted in 1997 and 1998 (NASS, 1999) 17
Table 3. Pesticide and Fertilizer Usage for Cotton in North Carolina (NASS, 1999) 18
Table 4. Methods Used in State Sampling Programs for Benthic Macroinvertebrates 24 .
Table 5. Hydrogeologic Framework Description and Hypotheses 28
Table 6. Summary of Types and Numbers of Sites 35
Table 7. Parameters Measured at all Sites 40
Table 8. Activities and Time Estimates for Work at Sampling Sites 42
Table 9. Timing of First-order Watershed Sampling Effort 43
Table 10. Parameters Measured or Calculated for Each Site or Watershed 45
Table 11. Landscape Metrics and Dependent Variables for Analysis 49
Table 12. Models Under Consideration 51
Table 13. Detailed List of Milestones by Fiscal Year 57
Table Al. USGS Pesticide Schedule 2001 (complete list) and the Estimated Amount of the
Active Ingredient Applied in Mid-Atlantic (Gianessi and Puffer, 1990 & 1992a,b) 76
Table Bl. Analytes for Sediments 78
Table Cl. U.S. Geological Survey Major Ions Schedule 2701 80
Table Dl. U.S. Geological Survey Nutrients Schedule 2702 80
Table E1. List of Benthic Macroinvertebrate Indices (primarily from Bode et al., 1996) 81
Table Fl. Calculated Reach-Level Physical Habitat Metrics (after Kaufman et al., 1999) 82
Table Gl. Spatial Databases 83
VII
-------
LIST OF FIGURES
Figure 1. Steps in the landscapes approach 3
Figure 2. Conceptual framework for LIPS-MACS 8
Figure 3. Mid-Atlantic Coastal Plain Study Area 14
Figure 4. Hydrogeologic framework for the Mid-Atlantic Coastal Plain 27
Figure 5. The CART decision process (after Moore et al., 1991) 33
Figure 6. Example of first-order streams and watersheds 37
Vlll
-------
TESTING LANDSCAPE INDICATORS FOR STREAM CONDITION
RELATED TO PESTICIDES AND NUTRIENTS:
LANDSCAPE INDICATORS FOR PESTICIDES STUDY
IN MID-ATLANTIC COASTAL STREAMS
INTRODUCTION
PURPOSE OF STUDY
This research plan for the Landscape Indicators for Pesticides Study ~ Mid-Atlantic
Coastal Streams (LIPS-MACS) describes the rationale and approach of developing a research
project to evaluate statistical landscape indicator models for freshwater streams in the Mid-
Atlantic Coastal Plain. This study is the first in a series of studies which will develop landscape
indicator models for pesticides and toxic chemicals hi selected areas, nationwide. These models,
often termed "landscape indicators," will be developed for pesticides and nutrients hi stream
water and persistent organic pollutants, mercury, and arsenic in sediments.
In the statistical analysis, certain landscape characteristics, termed metrics, will be
compared with dependent variables. Typical metrics include percent agricultural land cover,
presence and extent of riparian zones, soil texture and permeability, percent agriculture on steep
slopes, and soil credibility. Typical dependent variables include the corresponding data
characterizing streams, either biologically or chemically, hi addition to these traditionally used
landscape metrics, the LIPS-MACS study will include hydrogeologic parameters as additional
landscape metrics in the evaluation process. The streams will be characterized with chemical
analyses of both stream water collected during baseflow conditions and sediments, and by
measurements of benthic macroinvertebrates and physical habitat. Base flow conditions
represent shallow ground water contributions to the streams and will provide a longer-scale, time
integrated response, than characterizing stormflow, for example. The chemical analyses will
include pesticides, nutrients, and major ions hi stream water and historically used chlorinated
pesticides, polychlorinated biphenyls (PCBs), and mercury and arsenic in stream sediments.
This study is intended to be consistent with several U.S. Environmental Protection
Agency (EPA) approaches and guidelines including the Landscapes Approach (Jones et al.,
2000); EPA's Guidelines for Ecological Risk Assessment (U.S. EPA 1998), EPA's Evaluation
Guidelines for Ecological Indicators (Jackson et al., 1999), and the pesticide regulatory
perspective.
-------
RATIONALE FOR STUDY
Landscapes Approach
There is a growing interest among Federal agencies, States, and the public to evaluate
environmental conditions at community, watershed, regional, and national scales. At the same
time, the relatively high cost of collecting environmental data has limited the implementation of
regional- and national-scale monitoring programs. However, alternatives to and adaptations of
the traditional monitoring approach are possible using high resolution remotely sensed data and
derivative products now available. Termed the "landscapes approach," this alternative applies a
combination of concepts from landscape ecology, hydrology, and geography in conjunction with
remotely sensed and other spatial data and geographic information system technology to the
environmental assessment process (Jones et al., 2000, O'Neill et al., 1997). The landscapes
approach relies on
analysis of spatially explicit patterns (maps) of ecological characteristics (e.g.,
riparian zones near streams) to interpret ecological conditions;
concepts from the field of landscape ecology, relating changes in landscape
patterns to changes in ecological processes;
an ecological hierarchy theory that analyzes the consequences of landscape change
on ecosystems at multiple scales;
• digital maps of biophysical characteristics and human use to interpret landscape
patterns relative to ecological condition; and
inclusion of humans as part of the environment.
These characteristics distinguish the landscapes approach from the more traditional field or site-
based monitoring programs. We hypothesize that the science of landscape ecology and related
disciplines is integral to the assessment of the vulnerability and sustainability of ecosystem
processes and functions.
The focus of EPA's landscapes approach is on aquatic resources because the EPA has
primary responsibility in assuring their protection and restoration. However, the landscapes
approach process evaluates many aspects of the terrestrial environment because these attributes
are intricately linked to ecological and hydrological processes that influence aquatic resource
conditions, as predicted from ecological hierarchy theory (O'Neill et al., 1986). Because
regional-scale environmental factors and many local-scale factors are beyond human control,
stream management efforts involve minimizing land use impacts that influence stream habitat
(Richards et al., 1993). An understanding of both the aquatic resources and the terrestrial
environment are important to understanding the role pesticides play in the environment.
-------
Data Collection
Dependent
, Variables
Landscape
s* A<-'** • " " "
Landscape
Indicator
Applications
Hot Spot
Identification
Area-Wide
Assessment
"What If
Scenarios
Figure 1. Steps in the landscapes approach.
The basic steps of the landscapes approach are summarized in Figure 1. The collection
and synthesis of "landscape metrics," data characterizing specific spatial aspects of a watershed
or area of interest, are the first steps in the process. Next, data for spatially and temporally
comparable dependent variables are obtained, either from historical studies, or from new field
studies. The landscape metrics are then ranked statistically for their importance in explaining the
variance of dependent variables such as nutrient concentrations in streams (Jones et al., in press),
or for LIPS-MACS, pesticide concentrations in stream water, benthic macro invertebrate
community composition and abundance, or toxic chemicals and metals in stream bed sediments.
The statistical landscape indicator models are based on multivariate combinations of landscape
metrics. The best indicator models are those with high predictive power, i.e., those which
explain the largest amount of variance.
Once the landscape indicators have been developed and evaluated, a number of potential
applications are possible. The landscape indicators can be used to classify geographic areas in a
consistent, quantitative manner, for example, identifying relative ecological vulnerabilities.
Thus, the indicators become a useful tool in deciding where to invest monitoring resources or in
-------
making other management decisions. By factoring the relative vulnerabilities into the decision
process, the landscapes approach allows adaptation of a sampling design to focus on areas at
highest risk. Because of its flexibility and applicability at multiple scales, the landscapes
approach is widely recognized as the only cost-effective method to assess the potential impacts ol
complex natural and anthropogenic forces on the structure and function of ecological resources at
various temporal and geographical scales. The implementation of the landscapes approach has
begun only recently with the availability of high-resolution, remotely sensed data and the
computer technology to manage these data (Jones et al., 1997). This study will be our first
implementation of the landscapes approach for pesticides and toxic substances.
Guidelines for Ecological Risk Assessment
The Guidelines for Ecological Risk Assessment (U.S. EPA, 1998) describes basic
elements for evaluating scientific information on the adverse effects of stressors on the
environment. It is intended to be used as guidance for ecological risk assessments performed by
the EPA. The three major phases of the framework are 1) problem formulation, 2) analysis, and
3) risk characterization. Because the landscapes approach provides tools that can characterize the
geographic setting during initial problem definition, identify potential "hot spot" areas for more
intensive evaluation, facilitate consistent comparisons across geographic areas, and assist in
evaluating "what if scenarios, it can play a multifaceted role in the risk assessment process
(Graham et al., 1991, Hunsaker et al., 1990). After the landscape indicator models for pesticides
and toxic chemicals are developed, the intent of LIPS-MACS is to apply the landscape indicators
to provide examples of how these other aspects of the landscapes approach might be
implemented for pesticides and toxics in streams.
Evaluation Guidelines for Ecological Indicators
EPA's evaluation guidelines for ecological indicators have been designed to encompass a
wide variety of measurement types and assessment situations and are intended to be used for all
EPA indicator development efforts. The 15 guidelines fall into four phases: conceptual
foundation, feasibility of implementation, response variability, and interpretation and utility.
Collectively, they provide a comprehensive, recognized framework and process for
demonstrating indicator performance. The topics to be considered in the development of
indicators include: 1) relevance to the assessment; 2) relevance to ecological function; 3) data
collection methods; 4) logistics; 5) information management; 6) quality assurance; 7) monetary
costs; 8) estimation of measurement error; 9) temporal variability-within season; 10) temporal
variabiliry-across years; 11) spatial variability; 12) discriminatory ability; 13) optimization to
meet data quality objectives; 14) assessment thresholds; and 15) linkage to management action
(Jackson et al., 1999). These topics have been considered in the conceptual formulation of the
landscapes approach and landscape indicator development in general. LIPS-MACS will provide
data and an opportunity to address many of these topics specifically for landscape indicators for
pesticides and toxic substances.
-------
Pesticide Regulatory Perspective
The United States and numerous other countries derive many benefits from manufactured
chemicals, including improved health, food production, and quality of life. At the same time,
these chemicals can cause serious problems for health of ecosystems. The challenge to society is
to wield these chemicals wisely (Calow, 1998). The term "pesticides" is an umbrella concept for
a wide range of chemical substances that can be used to control weeds, insects, and other pests.
The active ingredients of pesticides are often combined into proprietary mixtures by
manufacturers. The properties of the chemicals differ greatly; some are water soluble, some
volatilize, some are adsorbed on soil particles, and some biodegrade rapidly. Unlike many
manufactured chemicals, pesticides (including herbicides, insecticides, nematicides, and
fungicides) are released directly into the environment and widely used in agricultural and urban
areas and in water bodies, forests, and transportation corridors in the U.S. To assess product
safety and evaluate potential risks to human and ecosystem health, and in accord with its
statutory responsibilities, the EPA conducts a registration and evaluation process before any
pesticide can be used. Pesticide- and toxic-substance-related research within the EPA supports
this process by providing state-of-the-science measurements, methods, and models for
development of ecological effects, exposure, and risk assessment protocols and guidelines, and it
provides the scientific basis for credible ecological assessments and evaluations of the impacts of
environmental stressors. Within this context, the Landscape Indicators for Pesticides Study is
focused on improving assessments of the condition of streams and other water bodies with regard
to pesticides, associated nutrients, and toxic chemicals at regional and sub regional scales.
The landscape indicator models developed in this project are expected to be useful to the
U.S. Environmental Protection Agency's (EPA) Office of Prevention, Pesticides, and Toxic
Substances (OPPTS); Office of Water (OW), and Regional Offices; and also State and local
agencies with responsibilities for developing Total Maximum Daily Loads (TMDLs) or concerns
about how water resources are affected by pesticides or toxic substances.
PARTICIPANTS
This study is a collaborative effort by EPA's National Exposure Research Laboratory
(NERL) and USGS' Water Resources Division, Maryland-Delaware-District-of-Columbia
District Office working with the National Water Quality Assessment (NAWQA) program.
Discussions will also take place with the U.S. Department of Agriculture regarding their
involvement in the study, particularly in providing pesticide application rate information.
Within EPA, the NERL Environmental Sciences Division in Las Vegas, Nevada is the
lead for the study; other NERL participants include the Ecological Exposure Research Division
in Cincinnati, .Ohio, and the Ecosystems Research Division in Athens, Georgia. Other EPA
participants include the Western Ecology Division of the National Health and Environmental
Effects Laboratory in Corvallis, Oregon, and the Subsurface Protection'and Remediation
Division of the National Risk Management Laboratory, in Ada, Oklahoma. EPA Regions 2, 3,
-------
and 4 and the Chesapeake Bay Program Office also are involved.
ORGANIZATION OF RESEARCH PLAN
The remainder of this research plan describes the details of LIPS-MACS. Separate
sections provide a project overview, description of the study area, and literature reviews for key
topics. Additional sections describe the field study, data analysis, hydrologic and multimedia
modeling, landscape indicator applications, quality assurance (QA), milestones/ schedule,
performance measures, potential for reducing uncertainty, and anticipated results and products.
-------
PROJECT OVERVIEW
GENERAL
This overview identifies the key features of the project before delving into the details and
specifics. Project elements at the broadest level flow from the hypotheses, to database
development, to the final landscape indicators, and application of these indicators in decision
making (Figure 2). In formulating this project and developing this plan, there are two major
activities that are complete:
• Hypotheses Development: A number of hypotheses concerning pesticide and
toxic chemical behavior, hydrogeologic characteristics, and landscape indicators
are developed and are presented later in this section. These hypotheses are based
on several general research objectives related to assessing condition of streams in
the mid-Atlantic region of the U.S. and are driven by our understanding of
agricultural land use and farming practices and urban pesticide practices in the
area. The overarching issue is the risk to the aquatic environment from pesticides,
nutrients, and toxic chemicals and metals.
• Project Design: The central component of this project is a comprehensive, geo-
referenced database which will facilitate statistical analyses for landscape
indicators and testing of the hypotheses. When these major objectives are
completed, the database will continue to play a crucial role, supporting
applications of the landscape indicators to answer "what if type questions. This
effort also may entail some process-based modeling. A major activity of the
study, especially in terms of field sampling and laboratory analyses, is the
development of this database. It will consist of data on water quality, stream bed
sediments, benthic macroinvertebrates, physical habitat, landscape features, and
pesticide loadings measured at, and geo-referenced to, all the study sites in the
Mid-Atlantic Coastal Plain. These data categories are those considered pertinent
during the development of the hypotheses and design of the analyses. The
literature review and the evaluation of existing data are significant guides to the
selection of the database components. Both existing and new data, and both
spatial and point-based monitoring data, will be collected and included in the
database.
To implement this study, there are two major activities:
Existing Data Acquisition: Existing water quality, hydrologic, and biotic data
were gathered to review for planning purposes as mentioned in the Project
Formulation, above. Much of the data to be used is from existing USGS
programs such as NAWQA. The data for interpreting the behavior of the
-------
HYPOTHESES/
DESIGN
EXISTING/NEW
FIELD/SPATIAL
DATA
LANDSCAPE
INDICATOR
MODELS
HYDROLOGIC/
MULTI-MEDIA
MODELING
NpSCAPE
INDICATORS
'- '^T^r*--—
LANDSCAPE"
^DICATOR
'PLICATIONS^
Figure 2. Conceptual framework for LIPS-MACS. Each rectangle corresponds with actions,
while the gray ovals represent products. Note that since stakeholder involvement extends
throughout the entire project, it is not shown explicitly. The arrows are either single-headed,
implying a one-way progression, or double-headed, implying feedback and iteration.
land use/land cover imagery and related information will be compiled for the
landscape analysis portion of the study and will be an essential component of the
database. The remotely sensed landscape data and related spatial data (such as
population) will encompass the entire study area, "wall-to-wall."
New Data Acquisition: New measures of stream water quality, bed sediments,
and of stream benthic and physical habitat conditions, from a one-time-only field
study, will be included in the database and used in the statistical analyses for
landscape indicator development. These data will provide a "snapshot" of spring
conditions using a consistent sampling design and established sampling and
-------
analytical methods. It is important to note that the same portfolio of measures
will be obtained for all data categories at all the study sites. In addition to these
field data, some process-modeling activities are expected to provide additional
information for input to the statistical analyses. There will be no new aerial
imagery developed specifically for this study.
The most important aspect of the study is the analysis effort which contains four major activities:
Landscape Indicator Models: Traditional multivariate linear regressions will be
applied to examine relationships between the data categories in the development
of the landscape indicator models. The landscape metrics and hydrogeologic
parameters will be the independent variables and stream water quality, sediment
concentrations of toxic chemicals, and benthic condition data will be the
dependent variables. In addition, Classification And Regression Tree (CART)
analysis will be used. Landscape metrics, hydrogeologic parameters, and stream
habitat characteristics will be the independent variables and the aquatic
parameters the dependent variables. In a related series of analyses using
comparative statistics, the generalized framework of seven hydrogeologic units
will be evaluated for its contribution to our understanding of the natural processes
which are expected to govern the interchange of chemicals between surface and
ground waters.
Hydrologic/Multimedia Modeling: This approach represents the application of
existing, physically- and chemically-based process models to typical settings
within the study area. Several hydrologic, pesticide fate and multimedia models
will be used as needed to improve our conceptual understanding of the physical
and chemical processes involved in the landscape. The model capabilities will
include compartmental distribution, hydrologic flow, and fate and transport. We
may use some data derived from the hydrologic and multimedia modeling in the
landscape indicator modeling. Alternatively, the landscape modeling results may
suggest scenarios for the hydrologic and multimedia modeling.
Final Landscape Indicators: These indicators consist of the best models
developed in the statistical analyses (see above) for use as indicators. Sensitivity
analysis results; numbers and types of variables to be used; areas of applicability
within the study area; and estimates of error will be considered. The landscape
indicator model error estimates will rely on a randomly chosen subset of data,
withheld from the initial analysis. A minimum detection level for the indicator
, models will be identified.
Landscape Indicator Applications: Once the landscape indicators are selected,
then it is possible to apply them for a number of different purposes: to identify
-------
relative "hot spots" in a region; to perform area-wide assessments; and to try out
"what if scenarios. For example, historic land cover data could be used as an
input to the indicator and the older results compared with the more recent results.
Alternatively, a projection of future land cover could be used as an input and the
results compared with current status.
Finally, it is our intention to involve Program Offices, Regional Offices, and other stakeholders
in the project to the degree they are interested. We will identify this group early and keep them
informed during the process of selecting the sampling and modeling sites, as results become
available. We will work extensively with this group in deciding how to present the results of this
study, e.g., workshops, reports, news releases.
OBJECTIVES
The objectives of this study are to
1) develop landscape indicators for pesticides and toxic substances in stream water
and bed sediments; and
2) demonstrate the application of the final landscape indicators for the Mid-Atlantic
Coastal Plain
A key assumption in this approach is that first-order streams and their valleys represent
the best scale for investigating landscape effects on streams because of the proximity of the
stream to the landscape features, shorter residence times for ground water prior to discharge to
surface water, and simplicity of the spatial land use patterns encountered.
HYPOTHESES
Two basic sets of hypotheses will be tested in this study, and organized according to the
objectives listed above. The hypotheses are expressed in general terms, but are meant to apply to
individual chemicals and groups of chemicals with similarities in properties and use. The
hypotheses for objective 1 are:
H1.1 Concentrations of pesticides and agrochemicals in stream water and bed
sediments, and concentrations of toxic substances in bed sediments are related to
. landscape metrics.
HI. 1 .a Concentrations of pesticides and agrochemicals in stream water
and bed sediments are related to the amounts of pesticides applied
to the land.
10
-------
H1.1 .b Concentrations of toxic substances in bed sediments above a
background threshold due to atmospheric deposition are
proportional to the amount of urban development in the watershed.
HI .2 Landscape metrics are related to underlying hydrogeologic variables.
HI .2.a Farmland is located in well-drained or artificially drained areas.
HI .3 Concentrations of pesticides and agrochemicals in stream water and bed
sediments and concentrations of toxic substances in bed sediments are related to
hydrogeologic variables.
HI .3.a Pesticide and agrochemical concentrations in stream water are
related to the soil sand content in the watershed.
HI .3.b Pesticide and agrochemical concentrations are related to underlying
geologic formations.
HI .3.c Pesticide and agrochemical concentrations in sediment are related
to the soil sand content in the watershed and the amount of clay
and organic material in the sediment.
HI .3.d Toxic metal concentrations in sediment are related to their content
in the underlying geologic formations and atmospheric deposition.
HI .4 Concentrations of pesticides in stream water and bed sediments are related to a
combination of landscape metrics and underlying hydrogeologic variables.
HI .4.a The relative importance of landscape metrics and hydrogeologic
variables will be difficult to separate statistically because the two
are interrelated.
A key assumption for the hypotheses below is that stream biotic condition is related to benthic
macroinvertebrate community composition and abundance for first-order streams.
HI .5 Benthic macroinvertebrate community data are related to stream and bed sediment
concentrations of pesticides, nutrients, and toxic substances, and physical habitat.
H. 1.5.a Benthic macroinvertebrate community data are related to landscape
metrics.
11
-------
The hypotheses for objective 2 are:
H2.1 Landscape indicator models for pesticides and toxic substances are applicable to
all the first-order, freshwater streams in the Mid-Atlantic Coastal Plain.
H2.2 Stream biotic condition can be estimated using landscape metrics and
hydrogeologic variables.
H2.3. Landscape indicator models for stream biotic condition are applicable to all the
first-order, freshwater streams in the Mid-Atlantic Coastal Plain (1990s data).
H2.4 Landscape indicator models for stream biotic condition demonstrate poorer
conditions during the 1990s compared to similar analyses during the 1970s.
More detailed hypotheses for the hydrogeologic framework are described later (Table 1).
UNIQUE FEATURES
The following are unique features of this study:
Testing landscape indicator concepts for pesticide and toxic chemical
impacts on streams;
• Incorporating geologic and hydrologic data into the landscape indicator
model development process;
• Choosing first-order streams and watersheds for landscape analysis;
Incorporating a hydrogeologic framework into the sampling design to
minimize hydrologic variability;
Combining a gradient study sampling design (based on percent developed
landcover) for landscape indicator development with a probability
sampling design for characterizing hydrogeologic framework areas; and
Characterizing pesticide metabolite concentrations for a large population
of streams: the freshwater streams of the Mid-Atlantic Coastal Plain.
12
-------
DESCRIPTION OF STUDY AREA
The Mid-Atlantic Coastal Plain is a physiographic region known for its rich farmlands,
forests, marshes, and swamps. It extends from southern New Jersey to North Carolina. The
eastern parts of four states; New Jersey, Maryland, Virginia, and North Carolina, the District of
Columbia, and all of Delaware are included within the Coastal Plain (Figure 3). The western
limit of the Coastal Plain is identified by the fall line, the location where waterfalls or rapids
occur in rivers flowing to the Atlantic. The rapids originate at the boundary where the higher and
relatively older, harder rocks transition to the lower, softer, and flatter sediments of the Coastal
Plain. Fall-line cities on the edge of the Coastal Plain include Baltimore, Maryland; Washington,
D.C.; Richmond, Virginia; and Raleigh, North Carolina. Land use/land cover data for the coastal
plain show that urban, commercial, or residential designations comprise 9 percent of the area;
agriculture 30 percent; forest 40 percent; wetland 20 percent; and other 1 percent. The Coastal
Plain area is encompassed by three ecoregions (North Atlantic Coast, Chesapeake Bay Lowlands,
Mid-Atlantic Coastal Plain) (Omernik, 1995) and three biotic communities (Northeastern
Deciduous Forest; Southeastern Deciduous Forest and Evergreen Forest; and Southeastern
Swamp and Riparian Forest) (Brown et al., 1998).
All the rivers of the Coastal Plain drain into the Atlantic Ocean. Major rivers include,
from north to south on the Delmarva Peninsula, the Chester, Choptank, Nanticoke, and
Pocomoke. On the west side of Chesapeake Bay, the rivers include the Susquehanna, Patuxent,
Potomac, Rappahannock, York, and James, while in North Carolina the rivers include the
Chowan, Tar, Neuse, New, and Cape Fear. Rivers are an important source of water supply to
cities such as Baltimore, Washington, D.C., Richmond, and Raleigh, but many of the people on
the Coastal Plain depend on ground water.
The Coastal Plain is well suited to agriculture; nearly all of the coastal plain is flat or
gently sloped, although some areas have relief of 30 meters or more. Elevations across the
Coastal Plain range from sea level along the coast to 230 meters (750 feet) on the western edge in
North Carolina. Based on an analysis of a 30-meter digital elevation map, more than 75 percent
of the Coastal Plain has elevations less than 40 meters (130 feet). The Coastal Plain is underlain
by semiconsolidated to unconsolidated sediments that consist of silt, clay, sand, with some gravel
and lignite (Trapp and Horn, 1997). In general terms, soils in the Coastal Plain include humus-
laden loams near the coast, with sandy loams and clay more toward the west. Much of the land
suitable for agriculture is farmed although suburbanization is encroaching on the farm lands.
Agricultural products include chickens, dairy products, com, soybeans, vegetables, and tobacco.
The climate in the Coastal Plain is humid and temperate. Average annual precipitation
ranges from 132 cm (52 in) per year in the southern coastal portion of North Carolina to
approximately 101 cm (40 in) per year in southern New Jersey, Northern Delaware, and west of
the Chesapeake Bay. The growing season ranges from 200 days in New Jersey to 275 days in
North Carolina.
13
-------
jersey
,\a\A/are
100 0 100 200 Kilometers
States and MAIA boundaries
Hydrogeologic Framework Boundaries
Figure 3. Mid-Atlantic Coastal Plain Study Area. State and the Mid-Atlantic Integrated
Assessment (MAIA) boundaries are shown. The hydrogeologic framework is shaded with
hydrologic framework units shown in outline only.
14
-------
LITERATURE REVIEW
PESTICIDE AND FERTILIZER USE IN THE MID-ATLANTIC COASTAL PLAIN
According to the Federal Insecticide, Fungicide, and Rodenticide Act, pesticides are
defined as any substance or mixture of substances intended for preventing, destroying, repelling
or mitigating any pest, and any substance or mixture of substances intended for use as a plant
regulator, defoliant, or desiccant. Pesticides addressed in this plan include herbicides for weeds
and insecticides for insects, mites, and nematodes. Pesticides in streams can have deleterious
effects on aquatic life, as well as being a potential source of exposure for humans who may use
the water. Fertilizers are defined as the primary plant nutrients: nitrogen, phosphate, and potash.
An overabundance of nutrients in streams can promote algal growth, depress oxygen
concentrations, and increase turbidity.
An understanding of pesticide and fertilizer use in agricultural and urban areas is critical
to our study because these land use categories are the main source of the chemicals we will
measure in the streams. Since stream samples are collected during base flow, the concentrations
of pesticides and nutrients measured will represent the integrated result of short- and long-term
flow paths of the shallow groundwater to the stream. The values measured will represent
averages of usage over the past several years, rather than the year immediately past, or the most
recent application. It would be ideal to have a decade of pesticide and fertilizer use information
for each farm, roadside, and forest in each watershed we are studying, but this is not possible. In
many cases, this type information is not available, or not saved. Even if it were saved, it would
be considered proprietary business information. Information gathered for specific farms by
federal government surveys is protected by confidentiality regulations. Finally, application
estimates for household, urban area, highway right-of-way, and commercial forest uses of
pesticides have proven difficult to find. Efforts to estimate these parameters will continue. The
types of information that are available are summarized below.
Data on pesticide and fertilizer use varies widely in types of data available from state to
state. Some data are available for all states from the National Agricultural Statistical Service
(NASS), as part of their 5-year Census of Agriculture reports. These reports contain information
on the number of acres for which broad categories of agricultural chemicals were used. These
data are summarized at the state level, and by county (NASS, 1997). Additional reports are
prepared annually by NASS for states in the "top-producer" category for selected crops and focus
on specific topics, which often include pesticide and fertilizer use for selected crop types, at the
state level. These reports provide application rates by specific active ingredient in pounds per
acre for the states selected for the survey. Finally, some states prepare their own reports on
pesticide use on an annual basis and these provide detailed summaries at the county level
(Maryland Department of Agriculture, 1999).
15
-------
In agriculture, pesticide and fertilizer use is determined by the type of crop planted and
this in turn depends on the climate, soil, and microclimate of a particular farm field. Common
crops in the five states containing the Coastal Plain include soybeans, corn for grain, fruits, nuts
and berries (in New Jersey), tobacco, vegetables, sweet corn, melons, cotton, and wheat. For
production of soybeans harvested for beans, North Carolina and Maryland are ranked 17th and
19th respectively among states nationwide. For production of corn harvested for grain or seed,
North Carolina is ranked 16th. For tobacco production, North Carolina and Virginia are ranked
first and fourth nationwide. For cotton production, North Carolina is ranked 7th nationwide.
Each crop type has a recommended pesticide application profile which is adapted by the
farmer or commercial applicator to the specific conditions of the individual field. Factors which .
affect the use of pesticides include the condition of the field, whether conventional agriculture or
no-till techniques are being used, what crops were previously grown, and the type of pest
present. One application per year is typical for commonly used pesticides for corn, soybeans, and
cotton (see Tables 1, 2, and 3). These data are based on surveys conducted in 1998 in North
Carolina; other states in the Coastal Plain were not included in this survey.
Table 1. Pesticide and Fertilizer Usage for Corn in North Carolina based on Surveys Conducted
in 1997 and 1998 (NASS, 1999).
Herbicide or
Insecticide/
Fertilizer
2,4-D
Alachlor
Atrazine
Glyphosate
Metolachlor
Paraquat
Simazine
Chlorpyrifos
Terbufos
Nitrogen
Phosphate
Potash
Area Applied
(percent)
27
22
88
14
44
8
2
8
21
98
92
91
Applications
(number)
1.0
1.0
1.0
1.0
1.0
1.4
1.0
1.0
1.0
2.0
1.1
1.0
Rate per
Application
(pounds/acre)
0.40
1.90
1.02
0.64
1.30
0.48
1.25
1.17
1.14
61
48
96
Rate per crop
year
(pounds/acre)
0.40
1.95
1.02
0.64
1.30
0.71
1.25
1.17
1.14
125
54
97
Total Applied
(1,000 pounds)
92
373
774
77
498
50
16
81
201
105,100
42,200
76,100
16
-------
Table 2. Pesticide and Fertilizer Usage for Soybeans in North Carolina based on Surveys
Conducted in 1997 and 1998 (NASS, 1999).
Herbicide or
Insecticide/
Fertilizer
Chlorimuron
-ethyl
Flumetsulam
Glyphosate
Imazaquin
Metolachlor
Nitrogen
Phosphate
Potash
Area Applied
(percent)
12
6
59
1
9
36
34
39
Applications
(number)
1.3
1.0
1.2
1.0
1.0
1.0
1.0
1.0
Rate per
Application
(pounds/acre)
0.02
0.07
0.85
0.06
2.30
23
38
83
Rate per crop
year
(pounds/acre)
0.02
0.07
1.07
0.06
2.30
24
38
83
Total Applied
(1,000 pounds)
5
7
932
1
322
12,400
19,400
47,300
17
-------
Tahle 1 Pestimrie anH Fertilizer Usage for Cotton in North Carolina (NASS, 1999).
Herbicide or
Insecticide/
Fertilizer
Clomazone
Cyanazine
Fluometuron
Glyphosate
MSMA
Pendimethalin
Prometryn
Pyrithiobac-
sodium
Trifluralin
Aldicarb
Cyfluthrin
Disulfoton
Lambda-
cyhalothrin
Phorate
PCNB
Cacodylic acid
Cyclanilide
Dimethipin
Ethephon
Mepiquat
chloride
Paraquat
Thidiazuron
Tribufos
Nitrogen
Phosphate
Potash
Area Applied
(percent)
5
9
41
65
33
28
18
5
20
28
16
3
70
25
6
12
42
3
58
23
3
31
28
98
90
93
Applications
(number)
1.0
1.0
1.0
1.6
1.1
1.0
1.2
1.0
1.0
1.0
1.9
1.0
2.5
1.0
1.0
1.0
1.0
1.0
1.0
1.3
1.1
1.0
1.0
1.9
1.1
1.1
Rate per
Application
(pounds/acre)
0.48
1.01
0.91
0.73
0.93
0.76
0.54
0.06
0.55
0.69
0.04
0.49
0.03
0.8
1.02
1.37
0.13
0.30
1.08
0.02
0.36
0.27
1.02
46
52
96
Rate per crop
year
(pounds/acre)
0.48
1.09
0.91
1.21
1.04
0.76
0.67
0.07
0.55
0.69
0.08
0.49
0.07
0.8
1.02
1.42
0.13
0.30
1.09
0.03
0.40
0.27
1.02
87
55 '
108
Total Applied
(1,000 pounds)
18
69
267
556
241
154
87
2
80
138
9
10
36
140
202
124
40
7
451
5
8
59
202
60,200
35,000
71,600
18
-------
PESTICIDES AND NITRATES MEASURED IN STREAMS IN THE MID-ATLANTIC
COASTAL PLAIN
Pesticides and nitrates are routinely detected in some streams in the Mid-Atlantic Coastal
Plain. Pesticides have been measured in streams and wells in the Mid-Atlantic mainly through
the NAWQA program and some state programs; summaries of this information are available
(Ferrari et al., 1997; Ator and Ferrari, 1997; Zappia and Fisher, 1997; Shedlock et al., 1999).
Nitrates have been measured at more sites than pesticides, and summaries of these data are also
available (for example, McFarland, 1995). Finally, river basin summaries, which discuss
pesticides and nitrates and many other aspects of water quality, are available for all of the
NAWQA study sites for example, the Lower Susquehanna and the Potomac, respectively
(Lindsey et al., 1998; Ator et al., 1998). Some of the details of these studies are discussed further
below. These data suggest that detectable concentrations of pesticides and nitrates will be
measured at many of the study sites.
Chronic low levels of pesticides and much of the nitrate measured in Coastal Plain
streams are attributed to ground-water discharge, i.e., ground water supplying the stream during
base flow conditions (Barbash et al., 1996; Bachman et al., 1998; Shedlock et al., 1999). Base
flow refers to the water that enters the stream from a persistent, slowly varying source (typically
ground water) and maintains stream flow between storms (Dingman, 1994). Nitrate
concentrations in Mid-Atlantic streams commonly exceed 0.15 mg/L as N, a level considered by
the Chesapeake Bay Program to contribute to eutrophication in estuaries. Nitrate concentrations
occasionally exceed 10 mg/L as N, the Federal maximum contaminant level for drinking water
(U.S. EPA, 1994a). Pesticides are present year-round in some streams of the Mid-Atlantic
Region in both urban and rural areas (Ferrari et al., 1997). Concentrations of most compounds
are typically highest during the spring and summer when there are sharp increases shortly after
application with relatively rapid declines in pesticide concentrations to near or below detection
for the remainder of the year (Larson et al., 1997). Chronic low levels of common pesticides are
attributed to ground-water discharge (Hallberg, 1987; Barbash et al., 1996; Shedlock et al.,
1999). Higher levels are commonly related to runoff shortly after application periods and
commonly occur in the spring and summer (Larson et al., 1997; Ferrari et al., 1997). Pesticide
concentrations in Mid-Atlantic streams commonly increase with increasing stream flow (Ferrari
et al., 1997). Herbicides are detectable in streams in many settings, but concentrations are
generally higher in agricultural areas. Insecticide concentrations are typically highest in streams
draining urban watersheds; however, data from such areas in the Coastal Plain are limited.
Detection of pesticides during late winter and early spring base-flow conditions should
represent the contribution of ground-water sources of pesticides to surface water in the absence
of recent pesticide application. The baseflow conditions should be more representative of the
time the water is in contact with the soil than other flow conditions, for example, when storm
flow is present. Ground water is a major source of water to streams in the Mid-Atlantic Coastal
Plain. The upper aquifer is shallow is relatively fast moving. The estimated median percentage
of stream flow derived from ground water is more than 60 percent for the Coastal Plain part of
19
-------
the Chesapeake Bay drainage (Bachman et al., 1998). Ground water provides more than 90
percent of stream flow in parts of the New Jersey Coastal Plain (Stackelburg and Ayers, 1994);
between 40 and 85 percent of stream flow in parts of the Coastal Plain in North Carolina and
southern Virginia (McMahon and Lloyd, 1995); and from 37 to 81 percent in the Delmarva
Peninsula (Gushing et al., 1973).
Only a few pesticides have a strong potential to be delivered to surface water from ground
water in appreciable quantities (Barbash and Resek, 1996). These include compounds such as
the commonly used triazine and acetanilide herbicides that have moderate to high water
solubility and stability and relatively low soil-sorption coefficients. Several of these compounds,
including some metabolites of atrazine, have been detected in streams of the Mid-Atlantic
Coastal Plain during winter base flow. Simazine, metolachlor, alachlor, desethyl atrazine, and
deisopropyl atrazine were detected in base flow samples collected from January through
March 1992 from a small stream on the Delmarva Peninsula located in an agricultural area with
well-drained soils. Pesticide compounds were typically undetectable in samples collected during
the same time period in two other small streams on the Delmarva Peninsula located in more
poorly drained agricultural areas. The minimum laboratory reporting level for pesticides in the
1992 Delmarva samples was 0.05 micrograms per liter (ug/L); however, it is significantly higher
than the 0.001 ug/L reporting level that has been used by the NAWQA program since 1993. We
expect that with the lower reporting limits now in effect, pesticides will be detected more
frequently than before.
Using current analytical techniques, pesticides are detectable in stream samples from a
variety of Coastal Plain land-use settings. Surface water samples collected in two small Coastal
Plain watersheds in North Carolina during December through Marchl993 and 1994 typically
contained atrazine, metolachlor, and alachlor; simazine and diazinon also were detected.
Concentrations of these compounds ranged from below the method detection limit of 0.001 to
0.19 ug/L. The sampled streams drain mostly mixed agricultural and forested watersheds.
Atrazine, desethyl atrazine, metolachlor, simazine, and alachlor were detected in samples
collected between January and March 1997 from Great Egg Harbor River, which drains a
developing urban watershed in New Jersey.
Concentrations of pesticides in surface water seldom exceed maximum contaminant
levels or lifetime health advisory limits for those compounds that have them (Ferrari et al., 1997).
In addition to parent compounds, metabolites of the common herbicide atrazine have been
commonly detected in surface water throughout the Mid-Atlantic Region (Shedlock et al., 1999;
Ferrari et al., 1997). Based on research from other areas where similar pesticides are applied
(Kolpin et al., 1998), there is reason to suspect that metabolites of other commonly used
pesticides, specifically the acetanilides, metolachlor and alachlor, would also be commonly
detected in surface waters of the Mid-Atlantic Coastal Plain (Kalkoff et al., 1998; Phillips et al.,
1999a; Phillips et al., 1999b).
Nitrate data provide additional insights on subsurface processes'affecting stream
concentrations. The importance of understanding subsurface conditions in interpreting the
20
-------
processes affecting surface-water quality has been demonstrated in several studies of riparian
zone function in the Mid-Atlantic Region. Lower concentrations of nitrate in surface water than
in upgradient ground water flowing toward surface-water discharge areas have been attributed to
processes of uptake by riparian zone vegetation and denitrification (Lowrance et al., 1984;
Correll et al., 1992; Osborne and Kovacic, 1993). Several recent studies have shown that
differences between ground- and surface-water chemistry are dependent on a variety of other
factors, as well, that relate to the opportunity for ground water to reach surface water (Bohlke and
Denver, 1995; Phillips and Bachman, 1996; Speiran, 1996; Staverand Brinsfield, 1996).
Factors affecting concentrations of nitrate in streams include aquifer thickness, the
chemical environment of the aquifer, the length of ground-water flow paths, the predominance of
different land uses in a watershed, and changes in land use and chemical application rates over
time. In a study of a small, well-drained agricultural watershed with a forested riparian buffer
overlying a relatively thick surficial aquifer, Bohlke and Denver (1995) found that the lower
concentrations of nitrate measured in surface water than in ground water were related to the lag
time between nitrogen application and ground-water discharge, e.g.,.older ground water dates
back to the time before fertilizers were used. Concentrations of nitrate in surface water resulted
from mixing of younger, high-nitrate ground water from short flow paths with older, low-nitrate
ground water from long flow paths, rather than denitrification in the aquifer sediments of the
riparian zone. Typical residence times for ground water in these surficial watersheds range
approximately from 1 to 40 years (Shedlock et al., 1999). Denitrification has been observed in
saturated aquifer sediments upgradient of riparian zones or at depth beneath riparian root zones
in other Coastal Plain settings (Bohlke and Denver, 1995; Bohlke et al., 1996; Speiran, 1996).
Staver and Brinsfield (1996) measured relatively stable concentrations of nitrate and reported
little evidence of denitrification or uptake by riparian vegetation in ground water discharging to
surface water in a sub-estuary of the Chesapeake Bay. Phillips and Bachman (1996)
demonstrated relations between base-flow stream chemistry and percentage of agricultural land
use, soil characteristics, topography, and geology in well-drained and poorly drained stream
basins. They found that in poorly drained basins, base-flow nitrate concentrations can be
decreased if ground water discharging to streams is subject to anoxic conditions. These data
show the importance of understanding the subsurface characteristics when interpreting the stream
conditions.
BENTHIC MACROINVERTEBRATES AS THE ECOLOGICAL ENDPOINT
Characteristics of stream biota (algae, invertebrates, fish) have been used for many years
to distinguish the degree and extent of human impacts on streams (Karr and Chu, 1999). We
have chosen benthic macroinvertebrates for characterizing aquatic condition for the streams in
this study for several reasons:
• The first- and second-order streams to be sampled are small and benthos will be
present, while fish will be few in abundance and diversity (Paller, 1994).
21
-------
Macroinvertebrates play an important functional role in the stream ecosystem: as
a food resource for demersal fish, as a link between lower and higher trophic
levels, and frequently as the first step in bioaccumulation of pollutants in the food
chain.
Macroinvertebrates effectively monitor environmental conditions; they tend to be
stationary or highly localized, and thus respond to the cumulative impacts of
environmental perturbations over time. Benthic macroinvertebrate characteristics
and indices have been successfully related to environmental factors for many
locations in the U.S. and elsewhere, such as Wisconsin (Hilsenhoff, 1987), Idaho
(Richards and Minshall, 1992), Virginia (Clements et al., 1992), Washington State
(Cuffney et al., 1997), and New Zealand (Quinn and Hickey, 1990). The effects
of contaminant stress may include a reduction in abundance and number of
sensitive species, or a simultaneous increase in the proportion of pollution tolerant
or opportunistic species (Wiederholm, 1984). Cuffney et al., (1997) identified
agriculture as the primary factor causing degradation of biological communities in
the Columbia River basin.
Benthic macroinvertebrates are in wide use as biological endpoints for stream
condition. Macroinvertebrates are easy to sample and standardized methods for
sampling and taxonomic analysis have been developed by Federal agencies and
many states (Plafkin et al., 1989; Cuffney et al., 1993; Kerans and Karr, 1994;
Bode et al., 1996; Lazorchak et al., 1998; Stribling et al., 1998; Karr and Chu,
1999; Barbour et al., 1999). Statewide studies performed within the Coastal Plain
include assessments for Delaware, Maryland, New Jersey, and North Carolina
(Maxted and Dickey, 1990; Klauda et al., 1998; Kurtz et al., 1996; and Kennen,
1999; Lenat, 1993; respectively).
Many factors can affect benthic macroinvertebrates; specific effects on benthos of
pesticides applied at commonly used rates in first-order streams are rarely
investigated (Schulz and Liess, 1999). In the case of herbicides, the potential
impact is on the invertebrate food supply. In the case of insecticides, the potential
impact is directly on the aquatic organisms. Results from a study on the
insecticide lindane have shown that short-term but high contamination has greater
effects on the aquatic fauna than long-term but low contamination with the same
exposure (Abel, 1980). The high concentration conditions often occur when
rainfall follows a pesticide application, resulting in overland flow into the stream.
These high concentration conditions for pesticides (both insecticides and
herbicides) are well documented for the Mid-Atlantic region in the late spring and
early summer (Ferrari et al., 1997). Cuffney et al., (1984) found a shift in
invertebrate species with the application of methoxychlor and reduced total
invertebrate biomass. Schulz and Liess (1999) found that insecticide
contamination has a strong negative effect on the aquatic macroinvertebrate
community.
22
-------
Macroinvertebrate community assemblages in streams are the result of many influences
operating over a hierarchy of scales. Patterns in the distribution of invertebrates at the ecoregion
scale are influenced by regional-scale natural factors such as climate, altitude, and geology
(Corkum and Ciborowski, 1988). The Coastal Plain region selected for this study is relatively
homogeneous with respect to these large scale factors (e.g., mid-latitude wet climate,
sedimentary geology, elevations less than 100 m). This means that landscape- and local-scale
influences become the distinguishing factors between sites. At the landscape scale, upstream
land use and environmental factors explain some site-to-site variability (Klein, 1979; Corkum,
1992; Sweeney, 1993; Krug, 1993; Tate and Heiny, 1995; Richards et al. 1996; Johnson et al.,
1997). Other factors which are important at this scale include stream size, gradient and flow
regime, bed stability, nutrient enrichment, riparian zone characteristics, and food supply (Quinn
and Hickey, 1990). Some of these factors depend on present and previous land use. Finally,
unique characteristics of the sampling site can also introduce site-to-site variability due to
differences in substrate extent and particle size, food availability, current velocity, pH, dissolved
oxygen concentration, and temperature (Quinn and Hickey, 1990). Some of this variability can
be minimized by using a standardized sampling protocol and compo.siting multiple samples from
each site (keeping riffle and pool composites separate). In addition, in LIPS-MACS, the
sampling location will be characterized using a quantitative physical-habitat assessment process
(Lazorchak et al., 1998). The intent is to minimize regional and local scale impacts to focus on
landscape-scale impacts.
Physical-habitat characterization data are essential to interpreting the benthos data
because some of the differences observed in benthos composition and abundance are due to
habitat variability. Physical-habitat data include stream dimensions, substrate qualities, gradient,
habitat complexity and cover, riparian vegetation cover and structure, some anthropogenic
disturbances, and stream-riparian interactions (Kaufmann, 1993). Anthropogenic alterations of
riparian areas and stream channels, drainage of wetlands, grazing, agricultural practices, and
modifications of stream banks, such as revetments or development, generally act to reduce the
complexity of aquatic habitat and result in a loss of species and ecosystem degradation
(Lazorchak et al., 1998). Noting and recording these features when a site is visited are essential
to understanding the benthic survey results. A more detailed description of the stream sampling
and characterization activities is provided in objective 3 in the Technical Approach section.
Macroinvertebrate populations can differ greatly between years depending on variations
in weather and flow regime (Gaspers and Heckman, 1981). Three recent events have probably
affected the benthic macroinvertebrate populations over widespread areas within the Mid-
Atlantic Coastal Plain. These events are the drought during the summer of 1999 and the flooding
associated with tropical storms/ hurricanes Dennis and Floyd in August and September 1999.
Under drought conditions, streams that would normally be flowing year-round have been dry
during the summer and early fall. The dry conditions have the potential to severely stress aquatic
organisms, depending on the timing of the drought compared to the timing of their life cycles
(e.g., stonefly and mayfly nymphs) or their dependence on flowing water (e.g., mollusks). To
estimate the impact of the drought, benthos reference sites will be sampled during the study.
These data will be compared to data from previous years collected by state monitoring programs
23
-------
for the same location. The hurricanes have had a different effect: the associated flooding has
resulted in major habitat modifications in some areas, especially in North Carolina (U.S. Federal
Emergency Management Agency, 1999). We are in the process of determining to what extent the
sampling sites are within the area of severe flooding. (Tidal streams, the area of greatest impact,
are not part of this study.) Also in this case, the use of reference sites may provide insights about
the biological impacts of the extreme variations in weather and stream flow.
Long-term viability of a macroinvertebrate population depends on potential for
recolonization after catastrophe. Williams and Hynes (1976) cite drift from upstream as the most
important mechanism of recolonization. First-order streams have been selected for sampling in
this study. In areas where insecticides are used near first-order streams, depending on the
specific location of the application, there may not be live organisms available upstream for
recolonization by drift. The other mechanism for recolonization is oviposition by adults from
other streams or migrating upstream. When insecticide use continues over a period of years near
all streams in an area, the preconditions for reestablishment of species become adverse. In the
Mid-Atlantic, insecticide concentrations tend to be higher in streams draining urban areas
compared to agricultural areas (Ferrari and Ator, 1997) and this may result in different
macroinvertebrate populations compared to agricultural areas. Because replenishment from
upstream is compromised and nearby streams in developed areas are receiying similar insecticide
applications, we expect that the benthic macroinvertebrate community composition and
abundance in first-order streams may be significantly reduced, even in cases when physical
habitat is conducive to healthy benthic populations.
We have learned that the states follow their own protocols for sampling with the
exception of Delaware and Virginia which use the Mid-Atlantic Coastal Streams (MACS)
Workgroup method for low gradient, nontidal streams (EPA, 1997; see Table 4). This may result
in some differences when our data are compared to the historical reference-site data.
Table 4. Methods Used in State Sampling Programs for Benthic Macroinvertebrates
Method
Identification
Level
Mesh Size
(Urn)
North
Carolina
own
manual
Genus,
species
to be added
New Jersey
own manual
Family
600
Delaware
MACS
manual
Genus,
species
600
Maryland
own manual
Genus
600
Virginia
MACS
manual
Family
600
24
-------
HYDROGEOLOGIC FRAMEWORK
Geologic and hydrologic characteristics, which have not always been considered directly
in previous investigations of landscape indicators, could be particularly important in a coastal
plain setting where surficial sediments are commonly permeable and much of the transport of
contaminants to surface water takes place in the subsurface. The geologic material present at the
earth's surface provides important controls on the shape of the landscape, the formation of soils,
the flow of water, and the chemical environment that water encounters as it moves through the
hydrologic system, hi many cases, surficial landscape coverages, such as ones available for soils,
topography, or land cover, are used as a proxy to represent subsurface conditions. However,
processes that occur beneath the land surface in the shallow ground-water system, cannot be
included in interpretations of landscape indicators and their relationship to water quality patterns
without additional information. This information is provided by interpretation of the geologic
characteristics of a region and their effect on hydrologic flow paths and geochemical reactions
within the soil and aquifer materials. In some cases, geologic factors may exert a primary control
on the transport or transformation of a particular water quality-related constituent, such as by a
geochemical reaction. In other cases, a surficial landscape variable such as the percentage of a
particular land use may be the primary control. It is useful to consider as many of the potential
variables as possible, so meaningful relationships between landscape indicators and water quality
can be quantified, at least to the extent needed to recommend appropriate management strategies.
To properly address the hydrogeologic variables for this project, a digital map of surficial
geology is needed. The underlying geology of the Coastal Plain is an important variable so we
want to use a classification system which describes the important features consistently for our
purposes. Some possible options are explained below.
The hydrogeology of the older, deeper, geologic units of the Coastal Plain has been
mapped previously. These and related studies are described below:
• For Washington, D.C., to the north to Boston, the U.S. Geological Survey
produced a map of engineering geology for the Department of Transportation in
1967.
• Brown et al. (1972) produced a 3-dimensional map of the Coastal Plain from
North Carolina through Long Island using data from more than 2,200 wells (the
first such model for the Mid-Atlantic Coastal Plain). Their purpose was to define
the geometry and internal permeability distribution of each "mappable
chronostratigraphic unit" (they mapped 17 plus the basement surface). The
youngest unit was undifferentiated "post-Miocene."
The Regional Aquifer System Analysis (RASA) program produced a series of
. reports detailing the hydrogeologic framework of the entire Coastal Plain in the
1980s. That program was mainly concerned with mapping and defining major
regional aquifers, so they concentrated mostly on the older confined units and did
not map the surficial units.
25
-------
Winner and Coble (1996) developed a hydrogeologic framework for the North
Carolina coastal plain.
A more recent regional study that treats the Mid-Atlantic Coastal Plain as one unit
identified heterogeneities in the physical setting and land use within the Coastal
Plain that were important in explaining the variations in ground-water quality
(Ator and Ferrari, 1997).
The Delmarva NAWQA study divided the Coastal Plain into seven subareas,
referred to as hydrogeomorphic regions (HGMRs) that define different hydrologic
settings. Each HGMR had a characteristic set of geologic and geomorphic
features, drainage patterns, soils, and land use patterns. They were successfully
used to look at differences in regional water-quality patterns and to transfer results
from local-scale networks within the HGMRs to the regional-scale analysis
(Shedlocketal., 1993,1999).
Although useful, these mapping efforts either do not address the surficial (mostly Miocene and
younger) units which are important to stream flow, or they do not provide consistent coverage
over the entire study area.
To solve these problems, a hydrogeologic framework for the Coastal Plain was developed
recently by USGS ( see Figure 4, Table 5). Based on a regionally consistent map of surficial
geology, and information on landform and geologic setting, it combines these primary natural
factors affecting the flow and quality of near-surface ground water and small streams into one
digital map (Denver and Ator, U.S. Geological Survey, Dover, DE, pers. commun., 1999).
Seven areas are identified in which the occurrence and movement of chemicals into shallow
ground water and streams are controlled by a relatively consistent set of natural processes. The
framework will be combined with other spatial data, such as soils, topography, and subcropping
geology, to represent the basic physical setting of the Mid-Atlantic Coastal Plain for the
landscape indicator analysis process. The areas delineated by the framework are being used to
stratify the selection of sampling sites for this study, to minimize hydrogeologic variability.
26
-------
Hydrogeologic Framework
Mid Atlantic Coastal Plain
Framewrkshp
| Coastal Lowlands
Md Coastal Plain, mixed sediments
Md Coastal Rain, fine sediments
Md Coastal Rain, Sand + Cvertying Gravel I
Deeply Dissected Sand + Cvertying Gravel
Inner Coastal Rain
Alluvium + Estuarine Valleys
N
W
200 0 200 400 Kilometers
Figure 4. Hydrogeologic framework for the Mid-Atlantic Coastal Plain
27
-------
Table 5. Hydrogeologic Framework Description and Hypotheses
Framework Area
Description
Hypothesized Potential for
Pesticide Mobility
Coastal Lowlands
Low-relief platform of the Outer
Coastal Plain and margins of the
major estuaries. Area is very flat
and low lying, with poorly
developed stream drainage and
numerous tidal wetlands. Streams
are low gradient and largely tidal.
Sediments are primarily fine-
grained. Soils of swamps and
marshes contain abundant organic
matter. Soil types reflect chronic
poor drainage and poor oxidation.
High potential for pesticides to be
bound by fine-grained sediments
and organic matter in poorly
drained soils. There may be some
transport of pesticides through
sandy surficial sediments into
ground and surface waters in areas
where pesticides are applied.
Pesticides may run off into
drainage ditches.
Middle Coastal Plain, Mixed
Sediment Texture
Broad platform of the Middle
Coastal Plain. Land surface is
moderately dissected by streams;
local relief ranges from 25 to 30
feet. Sediment texture varies
laterally and vertically and sizes are
mixed, ranging from coarse sands
to clays and silts.
Occurrence and concentrations of
pesticides will vary widely in
association with variations of
sediment type and land use
distribution.
Middle Coastal Plain, Fine
Sediments
Dissected inner portion of Middle
Coastal Plain with predominately
fine-grained sediments at land
surface. Local relief ranges from
20 to 60 feet.
The potential for pesticides to
infiltrate into ground water is low
because of confined conditions.
Pesticides may be transported to
surface water in overland runoff
from areas where they are applied.
Middle Coastal Plain, Sands
With Overlying Gravels
Inner Middle Coastal Plain; the
original broad flat upland surface
has not been completely dissected
by developing stream networks.
Local relief is less than 100 feet.
Greater incision occurs near major
tributaries that cut across the
Middle Coastal Plain.
The potential for pesticides to be
transported to ground water is
relatively high. Pesticide transport
will be affected by variability in
land use and soil characteristics.
The presence of organic matter in
stream beds may limit transport
from ground water to surface water
hi some areas.
28
-------
Framework Area
Description
Hypothesized Potential for
Pesticide Mobility
Middle Coastal Plain, Deeply
Dissected, Sands with
Overlying Gravels
Deeply dissected innermost Coastal
Plain, adjacent to Fall Line,
including sand and gravel caps on
adjacent Piedmont hills. Local
relief ranges from 100 to 150 feet.
Sediments are dominated by coarse
fluvial sands and gravels overlying
marine sands or saprolite of
crystalline rock. Surficial units are
completely incised and there is no
connectivity between upland
surfaces on adjacent interfluves.
The potential for pesticides to be
transported to ground water is
relatively high. Pesticide transport
will be affected by variability in
land use and soil characteristics.
Most pesticide occurrence in
surface water will be associated
with runoff because of stream
incision int confined aquifers and
confining beds.
Inner Coastal Plain
Outcrop and subcrop belt of lower
Tertiary and Cretaceous
formations, deeply weathered
where exposed, with 250 to 300
feet of relief. Some units are
leached and oxidized to depth of
tens of feet. There is widely
contrasting variability in the
permeability and geochemistry of
units. These contrasts affect
aquifer recharge and water quality
characteristics. The landscape is
deeply dissected and streams
typically cut into underlying units.
The potential for pesticides to be
transported into ground water is
moderate in areas with sandy
surficial sediments because of
loamy soils. Pesticide transport
will be affected by variability in
land use. Most pesticide
occurrence in surface water will be
associated with runoff because of
stream incision into confined
aquifers and confining beds.
Alluvial and Estuarine
Valleys
Incised valleys of major rivers that
cut across the Coastal Plain.
Deeper parts of the valleys are
filled by coarse-grained alluvial
sediments. Upper portion of
sequence is typically composed of
fine-grained, organic-rich
sediments. Valleys in North
Carolina are broader with greater
volumes of alluvial fill than are
valleys to the north that drain to the
Chesapeake and Delaware bays,
which are more deeply incised.
The presence of fine-grained
sediments, organic matter and
shallow water table on valley
terraces will limit pesticide
mobility. In areas with sandy
surficial sediments, pesticides may
be present in ground water.
Overland transport to surface water
will be limited by flat topography.
LANDSCAPE INDICATOR MODELS
Ecological indicators are defined by the EPA as measurable characteristics of the
environment, both abiotic and biotic, that can provide quantitative information on ecological
resources (Barber, 1994; Jackson et al., 1999). In this plan, we are using the term in the inclusive
29
-------
sense; it is intended to encompass physical, chemical, and biotic indicators. Indicators can be
classified into either the "condition" or "stressor" categories according to their purpose. In the
broadest usage, an ecological indicator may be based on a single measure or a statistical
combination of measures, or it may be an index based on multiple measures. Landscape
indicators are a particular category of ecological indicators that are determined for a predefined
area, which can be geographic, biogeographic (watershed, ecoregion) or political (State and
county boundaries, Federal regions). They are usually based on remotely sensed data or other
geographic information, and like ecological indicators, they can be. based on a single measure or
a combination of measures. The landscape indicator development and testing approach used in
this project has evolved from the general approach to landscape indicators for the Environmental
Monitoring and Assessment Program (EMAP) landscape monitoring and assessment research
(U.S. EPA, 1994b; Kepner et al., 1995; Jones et al., 1997). Landscape indicator analysis has a
number of unique features:
ability to look past artificial boundaries and fit specific areas into a larger natural
context;
• coverage of 100 percent of selected area, consistent with available data;
adjustability of resolution of results, from fine to coarse scales;
ability to test applicability of concepts from hierarchy theory; and
• ability to evaluate the importance of landscape features especially spatial pattern
and adjacency metrics to stream conditions.
Because of the potential confusion between landscape models and the more complex
landscape indicators based on multiple measures (i.e., soil erosivity based on the Universal Soil
Loss Equation), we will use the term "landscape metrics" to refer to landscape indicators which
are used as independent variables in the landscape indicator models to be developed. A
landscape metric typically is based on one spatial measure or aspect; examples include
population density, human use index (proportion of watershed with urban or agricultural land
use), road density, and proportion of watershed with crops on steep slopes. A landscape
indicator model combines these metrics to predict a dependent variable; for example, predicting
nutrient concentration from land use/land cover information.
The most commonly used metrics involve percentages of land cover/land use (Jones et
al., 1997), but a large number of indicators have been developed spanning landscape ecology,
soil erosion, and wildlife management (Riitters et al., 1992; Jones et al., 1996). Research on the
relationship of land use to stream water quality has largely focused on inorganic nutrients
(Omernik et al., 1981; Osborne and Wiley, 1988; Hunsaker and Levine, 1995; Tufford et al.,
1998; Cronan et al., 1999) rather than on organic chemicals such as pesticides. This is largely
due to the lack of sufficient pesticide data, hence this study. The finest resolution of current
landscape data is typically 30 m x 30 m, but new data will soon be available with 10 m and even
30
-------
1 m resolution. However, results for indicators are often aggregated and reported for much larger
areas, and identifying the appropriate scale(s) for a given indicator is important to its proper
application (Carlile et al., 1989; O'Neill et al., 1991; O'Neill et al., 1996; Johnson, 1994; Keitt et
al., 1997). The importance of the landscape setting and human influence in understanding
benthic macroinvertebrate populations has been noted by Fore et al. (1996), May et al. (1997),
Wang et al. (1997), Karr and Chu (2000) and others, which lends support to the usefulness of the
landscape indicator approach in this context.
The general approach for developing landscape indicators is to assemble a large database
of landscape metrics (independent variables) hypothesized to be important factors contributing to
the variability of the conditions measured (dependent variables). The goal is to remove
redundant metrics so the remaining ones are as independent as possible. A "weight of evidence"
approach based on statistical tests is used to determine which independent variables explain the
most variation in the dependent variables. One of the key precepts is that space can be traded for
time within areas that are similar. In a traditional experiment, indicators would be tested over
time in replicates of one or perhaps several locations, to provide a gradient of conditions for the
same setting. Trading space for tune assumes that looking at many locations within an ecoregion
at one time provides snapshots of many different stages of an environmental situation. It has the
disadvantage that initial conditions are not established and the areas may not be undergoing
similar or parallel processes. The many locations will include responses to factors other than the
ones of interest. However, this is the most practical approach, given the time scale of Federal
careers compared to landscape evolution! For this study, chemical concentrations and biological
condition are the conditions to be measured. Statistical techniques will be used to identify
promising multivariate and hierarchical relationships. The result of this analysis will be the
landscape indicator models, which relate a specific dependent variable to the independent
variables. Multiple regression will be a primary statistical tool used for the landscape indicator
models which take the general form below:
dependent variable = c0 + £ c ( * x(,
where x( is an independent variable, and c0 and Cj are constants. Dependent variables include
ecological condition as expressed by indices for benthic macroinvertebrates (different indices
will be tested); physical habitat; and concentrations of pesticides, pesticide metabolites, nutrients,
and major ions in streams. Independent variables include land use/land cover, topography, soil
type, geologic and hydrologic characteristics, population density, metrics for roads in watersheds,
pattern metrics for land use, and riparian zone characteristics. The hydrogeologic framework unit
will be evaluated as an explanatory variable in the landscape indicator models. Amount of
variability explained, both overall and by individual independent variables, will be used to
evaluate the success of the model and the relative importance of the independent variables. In a
study by Hunsaker and Levine (1995), for inorganic nutrients and conductivity in streams with
the entire watershed as the source area, variance explained ranged from 53 to 86 percent for total
nitrogen and total phosphorus, respectively. Another study (Jones et al.', in press) had similar
results.
31
-------
Besides traditional multiple regression, one of the tools we are planning to use in our data
analysis approach for landscape indicators is Classification And Regression Tree (CART)
analysis (Breiman et al., 1984). CART offers a number of advantages for the analysis of
environmental data; these data often have missing values, interactions between variables, non-
normal distributions, large numbers of and different types of variables, high variability, and high
dimensionality. CART is robust with respect to these problems because it makes no assumptions
about the data distribution and is relatively insensitive to outliers (Brieman et al., 1984). When a
variable is missing at a certain sample location, CART can use a surrogate variable.
Environmental data often have location-specific relationships between variables; the CART
technique can apply different decision rules for each location (Moore et al., 1991) or for each
subset (Walker, 1990). CART is similar to stepwise regression in that it can deal with on-off
type variables such as a pesticide used in one area and not in another.
CART is a heuristic technique which develops a hierarchical structure of rules by
grouping observations into classes. It searches for a set of questions that is most efficient at
discriminating between classes. The rule that provides the largest increase in class purity forms
the first splitting rule of the decision tree (root node). The original data are split into two
descendant nodes based on this rule (see Figure 5). Then the process is repeated iteratively to all
subsequent nodes and their descendants until the tree has attained maximum complexity. Each
splitting rule is fit onto the decision tree as a branch node (Moore et al., 1991). When a splitting
rule is applied, data for which the answer is "Yes" are assigned to the left branch, while
remaining data points are assigned to the right branch. The leaves of the tree are called terminal
nodes (Efron and Tibshirani, 1991). Although rules have been developed to control the size of
trees, a common practice is to construct a large tree and prune it working from the smallest
subsamples toward the larger classifications. CART applications often use cross-validation to
determine the best tree size, although many methods have been used (Sifneos et al., in
preparation). This method is similar to using a test sample and works by dividing the data into
10 groups of equal size, creating a tree with 9/10 of the data and assessing the misclassification
rate for the remaining group of data. Each group of data is tested against the remaining 9/10 in
turn, and the total misclassification is computed for all 10 runs. The best tree is the one giving
the lowest misclassification rate (Efron and Tibshirani, 1991; Clark and Pregibon, 1992).
Because the CART decision tree process is continually dividing data into smaller and
smaller subsamples, the number of locations sampled, or observations, provides an inherent
limit on the size and performance of the tree. Using a decision tree approach with a sample of
128 observations is considered minimally adequate for applying the technique (Miller, 1994).
When data are binary (e.g., presence or absence of species), the classification aspect of CART is
used. When data are continuous (e.g., concentration of chemicals), the regression capability of
CART is used. Example applications of CART to ecological problems include modeling
distributions of kangaroos in relation to climate (Walker, 1990); predicting vegetation
distributions (Moore et al., 1991); explaining spatial factors related to bird biodiversity
(O'Connor et al., 1996); and predicting species richness in fishes (Rathert et al., 1999). In this
study, measures of stream ecological condition determined from benthic and stream water quality
data will be the dependent variables, and physical habitat characterization and landscape metrics
will be the independent variables.
32
-------
«Q
Root Node
ifx = k
1
Yes
r
Class
Assignment :
A
No
Split
Rul
1
Class
Assignment :
B :
r
tinq J:: (-—
e 2 ^^
1
Class *|
C ^
Group of
Samples
I
Group of
Samples
Group of
Samples
Figure 5. The CART decision process (after Moore et al., 1991). A CART analysis is prepared
for each dependent variable of interest, for example, pesticide concentrations in stream water.
Referring to the schematic framework, variables such as percent forest or percent developed land
use/land cover may appear at the highest (root node) level as explanatory variables. Landscape
factors may not be important below a certain threshold, and sites showing little response to
landscape variables could be assigned to Class A. For sites with land use/land cover above the
threshold, the variable at the next level of importance would appear in Splitting Rule 2. Possible
variables for this branch node include the amount of riparian zone along the stream within the
watershed, or the amount of clay in the soil of the watershed, resulting in the assignment of some
sites to Class B and the rest to Class C. It is likely that additional splitting rules would be
developed for some of these classes, extending the diagram further to include variables such as
ecoregion, density of roads, and agriculture on steep slopes.
33
-------
FIELD STUDY
STATISTICAL DESIGN
A particular challenge for the LIPS-MACS study design is to provide the necessary data
for landscape indicator model development and validation, while at the same time providing
adequate data to characterize regional hydrogeologic conditions with known bias. For landscape
indicator model development, stream data for watersheds spanning a broad range of categories of
land use/land cover are necessary. Because of the focus on pesticides, agricultural and urban
land use are of particular interest. For the hydrogeologic framework unit characterization, stream
data which are representative of the hydrogeologic framework unit with a good spatial
distribution are important. The indicators being developed in this study will not be appropriate
for addressing some kinds of questions, for example, assessing very small areas. The goal is to
develop a consistent and comprehensive look at the entire region, and there are tradeoffs between
the level of detail and the size of the area that can be considered. The spatial design (see Table 6)
consists of a one-time survey:
175 sites representative of first-order streams (Figure 6) in the Coastal Plain;
3 nested sets of 5 sites each for a total of 15 sites; and
• 7 benthos reference sites.
In addition, three temporal sites will be sampled for a year. The 175 sites provide the basic data
set for landscape indicator model development and evaluating-the hydrogeologic framework.
Subsets drawn from this larger set and held separately, will provide one form of landscape
indicator model validation. A second type of validation data will be provided by the nested sites.
These sites will enable a limited comparison of results for smaller watersheds nested within a
larger one. The benthos reference sites are high quality, near-pristine sites which are part of the
states' ongoing biological monitoring programs. The historical data available for these sites will
be useful in interpreting our data. We have obtained site information from each of the states and
will select a total of seven sites (one per hydrogeologic framework unit). These reference sites
will be characterized physically and chemically in the same manner as the framework unit sites.
We will establish three temporal sites to evaluate the temporal variability of pesticide
concentrations; each will have a full year of record (sampled biweekly in the spring, and monthly
thereafter). Stream flow and pesticide concentration data will be available for comparison to the
one-time survey being made across the entire Mid-Atlantic Coastal Plain. These sites will be
selected from among those currently being sampled as part of other USGS or EPA programs to
allow for maximum use of resources; our cost will be to pay for the pesticide analyses. Early
results from this sampling should help identify the types and concentrations of pesticides to be
expected from the regional sampling. Current choices for these sites are Chesterville Branch, an
agricultural stream on the Delmarva Peninsula in Maryland; Western Branch, which drains a
34
-------
mixed agriculture-forest-urban watershed near Washington, D.C.; and Lizzie Site, an agricultural
stream in the Contentnea River drainage in North Carolina. Data from these watersheds will
allow for a better understanding of the processes affecting the movement of pesticides at a finer
scale and will help show how regional sampling results relate to local areas. These sites will be
used in the case studies for the hydrologic and multimedia modeling.
Table 6. Summary of Types and Numbers of Sites
Type of Site
Framework unit
Nested
Temporal
Benthos reference
TOTAL
Purpose
Unbiased description of stream
water quality within framework
units, weighted to provide a
gradient over "developed" land
use
Multiple smaller watersheds,
nested within a larger one, to
evaluate scale relationships
Understand seasonal patterns in
specific framework units
Historically unimpaired sites, to
ensure suitable data for
computing benthic
macroinvertebrate metrics
Comment
25 per hydrogeologic
framework unit x 7 units
3 nests with 5 sites each
one site in each of 3
hydrogeologic framework
units
one per hydrogeologic
framework unit
_
Total Number
of Sites
175
15
3
7
200 (approx.)
The statistical process for the first-order watershed site selection is described in two
stages below. In the first stage, we will establish the population of first-order streams and
associated watersheds from which to select the sample. First-order streams will be identified
using the Reach File 3 "start reach" codes (U.S. EPA, 1994c). Euclidean watersheds will be
determined by using the Reach File stream coverage and determining Thiessen polygons, the
boundaries of the polygons being formed by the perpendicular bisectors of the lines joining
adjacent stream segments (Chow et al., 1988). The resulting shape approximates the watershed
boundary. These Euclidean watersheds will be the "first cut" which provides our sampling frame
of first-order watersheds. The sampling sites will be selected randomly from this set. Other
alternatives for developing the watershed boundaries were considered, for example, basing them
on digital elevations, but this is not practical because there are more than 10,000 first-order
watersheds that make up the Coastal Plain. It is also not practical to use flow data since these are
not available for most of these streams. Once the 200 hundred sampling sites are selected and
visited by the sampling crews, we will recompute the watershed boundaries using digital
elevations and actual sampling points. In addition, the watershed boundaries will be evaluated
manually using topographic maps for the sites. This will ensure that the boundaries are as
accurate as possible for the landscape indicator analyses.
35
-------
Next, each watershed's land use/land cover composition will be determined by using land
use/land cover data from the Multi-Resolution Landscape Characteristics (MRLC) Consortium.
These land use/land cover data are derived from 30-meter resolution Landsat Thematic Mapper
data, and classified into 15 land use/land cover classes (Vogelmann et al., 1998). We will define
developed land use/land cover as all agricultural plus all urban, residential, and commercial land
use/land cover categories. The goal is to select sampling sites so that roughly equal numbers of
sites occur in each of 5 percentage categories of developed land use/land cover (0-19 percent, 20-
39 percent,... and 80-100 percent). However, the actual distribution of sites within these
categories shows that on average, there are more watersheds in the less developed rather than the
more developed categories. Using a random selection from this population would result in too
few sites in the upper categories of developed land use/land cover for developing the landscape
indicator statistics.
To provide a more uniform distribution for the sampling sites, we will weight the
selection probability of the first-order watersheds by the frequency of occurrence of developed
land cover. The weighting process increases the probability that rare conditions will be included
in the sample, thus ensuring that an adequate gradient will be available for the landscape
indicator development. This can be accomplished conceptually by representing each watershed
as a line segment having a unit length and placing these segments end-to-end, working
methodically through all the possible watersheds in the study area. A simple random selection
process would use a random start, and move along the line of segments at a fixed interval which
selects the correct number of sites. The fixed interval is the total number of sites to be selected
divided by the length of the total number of sites. If the line segment is of unit length, then this
will be the same as the number of sites. The weighted random selection process is performed by
adjusting the length of each segment for each watershed, by dividing the line segment length of 1
unit by the frequency of occurrence of the land use/land cover class for the watershed. The
frequency of occurrence values are expressed as values within the range from 0 to 1. Since all
the values are less than one, dividing by the frequency of occurrence has the effect of lengthening
the segment. The segments for the rare conditions will be lengthened significantly, making them
more likely to be selected, while the segments with the more frequently occurring conditions will
be lengthened minimally in comparison, making them less likely to be selected. This technique
was demonstrated for EMAP streams by Herlihy et al. (2000). The actual process is somewhat
more complex than the conceptual approach just described, and that is described next.
36
-------
First-Order Watershed Boundary]
First-Order
Stream
jSegment
First-Order
Watershed
Pour Point for
Nested Watershed
Figure 6. Example of first-order streams and watersheds. Unlabeled dots mark the sampling
points (also called pour points) for first-order watersheds. These first-order watersheds are
typical of the 175 sampling sites in the main part of the study; however, none of the actual sites
are likely to be adjacent to each other as shown here. The dark outer boundary marks the edge of
a larger watershed with a "nest" of first-order watersheds contained within. Nested samples will
be collected at the pour points for the three first-order watersheds nested within the larger one
and at the pour point for the larger watershed.
37
-------
Starting with the key ideas above as a guide, and working in the broader context of Olsen
et al., 1999, we will follow the approach of Stevens and Olsen (1999). They rely on a method
(Madow, 1949) for combining systematic and random sampling to sample without replacement,
wiih the probability that an item is included in the sample proportional to an arbitrary weight for
each item. This method involves calculating the cumulative total weight for the items arranged
in some order, and then drawing a systematic sample with a random start using a fixed length
sampling interval along the cumulative weight totals. The steps in the Stevens and Olsen (1999)
procedure are provided verbatim in italics below:
1. Overlay the population domain with an area grid, choose a random location in a cell
designated as the origin cell, and then translate the entire grid so that the origin cell is centered
on the random point.
2. Link each population element to its covering grid cell, and assign grid cell i an
inclusion probability Pi equal to the expected number of samples in its associated portion of the
population. The p( may vary from cell to cell and may be zero [if it contains no item], but cannot
exceed 1. (Ifanypt exceeds 1, then a finer grid is required.) Then arrange the grid cells in
hierarchically randomized order.
3. Draw a sample of grid cells from the randomized list, using Madow's (1949)
technique, which guarantees that cell i is included in the sample with probability pr (Since p, is
the target number of samples in cell i, ^ pi = n, where n is the target sample size and the sum
i
is over the number of grid cells [or items].
4. For each selected grid cell, pick one sample point at random from its associated
population elements, recognizing any differential weighting among such elements.
Variance estimation is performed by estimating pairwise inclusion densities by ignoring the
spatial dependencies among the sample point locations and assuming an independent random
sample design. Then the Horvitz-Thompson theorem for continuous populations yields a
variance estimator (Cordy, 1993 and Stevens, 1997). If the population has spatial structure,
then the resulting estimator will be conservative.
The advantages of this technique as it relates to the LIPS-MACS study are:
• it guarantees that the sample is well spread-out over the extent of the resource
because the hierarchical randomization (Step 2 above) results in a random order
that nevertheless preserves some spatial relationships;
• the design has enormous flexibility to accommodate design constraints; i.e.,
weights may be specified on a regional (e.g., by state boundaries) or elemental
basis (e.g., stream order, or watershed land cover characteristic); and
38
-------
• a variance estimator is available and conservative.
The flexibility is the key element of this design approach. It enables two design objectives to be
addressed using systematic random sampling. Thus, each first-order watershed will have a
known probability of being included. This has a number of advantages; stratification techniques
can be applied to incorporate known characteristics of the population and unequal probability
sampling can be applied to ensure rare conditions are included. This capability will allow the
sampling of the more rare land-use patterns as necessary for the indicator testing, while enabling
the use of these data for other purposes such as evaluating the hydrogeologic framework and
characterizing the Coastal Plain.
LOGISTICS/METHODS
The field study effort has eight basic activities which are discussed further below:
• preliminary visit to sampling sites,
• collection of water samples,
• collection of benthic macroinvertebrate samples,
• performance of physical habitat assessment,
• collection of sediment samples,
• laboratory analyses of samples for pesticides, pesticide metabolites, inorganic
nutrients, major ions, and identification of benthic macroinvertebrates,
• compilation of data into databases, including all laboratory analyses and physical
habitat assessment and rapid visual assessment data, and
• ongoing quality assurance review of activities with both internal and external
audits.
With the exception of the preliminary site visits, and the temporal sites, all the sampling
activities will be conducted in the winter/spring sampling period. The choice of the winter/
spring sampling period is discussed later in this section. Laboratory analyses will follow
immediately and continue for up to 9 months, depending on the parameter. Preliminary site visits
will be conducted by the USGS during the fall and winter. These visits will confirm presence of
the stream, arrange permission with the landowner, identify a convenient access route, and
provide an initial assessment of the actual watershed size compared to the computer-drawn size.
Practical considerations such as convenient lodging, shipping facilities, and health care facilities
will be noted. A process for replacing sites that cannot be used is part of the study design.
To ensure consistency not only within this study but also to facilitate further use in other
39
-------
studies, we will use NAWQA procedures for stream sample collection and analysis and
Environmental Monitoring and Assessment Program (EMAP) methods for collecting sediments
and assessing benthos and physical habitat (see Table 7). On-site measurements and stream
chemistry will provide basic information about the stream and its setting. USGS Pesticide
Schedule 2001 contains both urban and agricultural use pesticides and these compounds have
been successfully identified in urban areas in this region (Ferrari et al., 1998). Pesticide
metabolite concentrations are of interest to the Office of Pesticide Programs at EPA. Pesticide
metabolites data may provide insights on transit times, degree of degradation, flow systems and
transport of chemicals. Benthic macroinvertebrates were chosen because of the lack offish in
these small streams. Sediments are being analyzed for the persistent chlorinated pesticides and
PCBs. Mercury is included because of its potential for long range airborne transport and
deposition; and arsenic is included as an indicator of poultry waste.
Table 7. Parameters Measured at All Sites
Activity
Water sampling and
on-site chemistry
Water sample analysis
(laboratory)
Benthic
macroinvertebrate
sampling
Benthic sample
analysis (laboratory)
Physical habitat
assessment
Sediment sampling
Sediment analyses
Conducted by
USGS
USGS
EPA NERL
contractor
EPA NERL
contractor
EPA NERL
contractor
EPA NERL
contractor
EPA NERL
contractor
Parameters
DO, temperature, pH,
stream discharge,
dissolved alkalinity,
specific conductance
pesticide schedule
2001, major ions
schedule 2701,
nutrients schedule
2702, pesticide
metabolites
pool, riffle settings;
community
composition, and
abundance
300 count organism
identification to
genus and species
Thalweg profile,
woody debris tally,
channel and riparian
characterization
Composite sample
pesticides, PCBs,
mercury, arsenic,
gradation, organic
content
Reference
Shelton, 1994
Shelton, 1994; Zaugg et al.,
1995; and Hosteller and
Thurman, 1999. For
analytes, see Appendices A,
C, and D.
Lazorchak et al., 1998
Klemm and Lazorchak,
1994
Lazorchak et al., 1998;
Kaufmann et al., 1999
Lazorchak et al., 1998
Wesselman and Carr, 2000.
See Appendix B.
40
-------
Water samples will be collected, handled, and analyzed using procedures developed by
the USGS for the NAWQA program (Shelton, 1994). Depth-integrated water samples will be
collected from equal-width increments of a stream cross section using Teflon® or stainless steel
equipment. Samples for pesticide analysis will be passed through a nominal 0.7 micron glass-
fiber filter and collected in baked amber glass bottles. Samples for nutrients and selected major
ions will be passed through a 0.45 micron filter. Samples will be chilled and shipped overnight
to the USGS National Water-Quality Laboratory (NWQL) in Denver, Colorado, and the USGS
Laboratory in Lawrence, Kansas, for analysis of major ions, nutrients, and pesticides, and
pesticide metabolites, respectively. All sampling equipment and supplies will be cleaned
between sites with mild soap and methanol.
Benthic macroinvertebrate collection and 300-count analysis will follow the EMAP
protocol (Lazorchak et al., 1998; Klemm and Lazorchak, 1994). Physical habitat characterization
will be conducted following the EMAP protocol (Kaufmann and Robison, 1998), except for
elements characterizing fish habitat which will not be assessed. Sediment samples will be
collected from each of the 11 benthos transects. Samples will be collected from the top 2 cm of
surficial sediment in depositional areas, using a plastic spoon. The sediment will be stored in 50
mL centrifuge tubes, one tube for each transect. Samples will be kept chilled but not frozen until
delivery to the laboratory. The laboratory analyses will follow modified EMAP protocol to
include additional analytes. The list of target analytes is provided in Appendix B. Additional
parameters will also be noted during this phase; it will include performing an accuracy
assessment for the MRLC land use/land cover designation for the area surrounding the site.
Photographs will be taken of the site in four directions. Adjacent crop types, pesticide
applications, and rills/gullies will be noted.
The field study will be conducted by USGS and contractor staff traveling separately to
each site. The sampling location will be marked, identified with global positioning system
coordinates in compliance with EPA's locational data policy, and maps will be provided so both
crews sample from the same location. Considering crew activities, the expectation is that the
water sampling crews can sample two to three sites daily, while the benthos and physical habitat
crews will sample one or two sites daily depending on the proximity of the sites (see Table 8).
This is nominally an 8-week sampling effort. Issues that are being considered include safety;
having substitutes available in case a regular crew member becomes sick; and having additional
crews available to keep the sampling on schedule.
The timing of the first-order watershed scale sampling was determined by identifying the
best months based on a number of criteria (see Table 9). We will attempt to follow the transition
from winter to spring as warmer temperatures advance from North Carolina to New Jersey. The
months of late February, March, and April, with some variation allowed for weather conditions,
are the best months for this sampling effort. The colder stream temperatures are important for
minimizing biological activity and chemical reactions in the stream water. The timing is also
chosen to occur before pesticides are applied to avoid effects of the initial pesticide surge, which
occurs during the first storm after the pesticide is applied. In general, herbicides are used early in
the planting season, while insecticides are used later when the crop is more mature. The
41
-------
insecticides have the potential to reduce populations of benthic macroinvertebrates, and this
timing is a possible source of variability in the data from site to site. We will consider using
additional crews to shorten the time to cover all sites. Stream samples will be collected under
base flow conditions and should mostly represent contributions from near-surface ground water
which is most directly affected by land uses and other surficial activities. During these months,
concentrations of chemical constituents in the water will be least affected by biological activity.
This timing fits within the windows of acceptability for pesticide applications and benthic
macroinvertebrate sampling.
Table 8. Activities and Time Estimates for Work at Sampling Sites
Water
Sampling
Crew
Totals
Benthos Crew
Totals
Activity
Travel to site
Verify site; establish sampling reach
Collect water chemistry samples;
measure stream discharge;
paperwork
Travel to site
Collect and process benthos
Characterize physical habitat
(modified procedure)
Sample tracking and packing
Group
3 persons
3 people, 3-5
hours
2 persons
2 people, 6- 8
hours
Estimated Time Required per
Site Visit
1 hour or more
1 hour
1-2 hours
2-3 sites per day
1 hour or more
2 hours
2 hours
1 hour
1 or 2 sites per day
42
-------
Table 9. Timing of First-order Watershed Sampling Effort
Activity
Characterization of stream base flow
Minimal biological activity
Collection of benthos samples
First pesticide application
Sample collection time interval
Best months
January through April
December through March
March, April, May (Plafkin et al, 1989; Klauda et al.,
1998)
March 1 to May 1 (Stribling et al., 1998)
late April, May
Late February, March, April
43
-------
DATA ANALYSIS
OVERVIEW
Data analysis includes all the systematic uses of the data and other information to address
the LIPS-MACS hypotheses. Many of the analysis activities will be conducted in parallel. Some
of the analysis activities are required to generate intermediate results necessary for understanding
the data, while others provide direct results. A wide variety of data will be collected or
calculated and summarized for each stream site or corresponding watershed (see Table 10). The
overall data analysis approach which we will follow is described below:
Descriptive statistics and maps will be developed for all the major categories of
data to achieve familiarity with the data and add an additional level of data quality
assurance and validation. Computation of benthic macroinvertebrate indices and
physical habitat metrics are included here.
Association statistics also will be developed for the major categories of data.
These analyses are similar to descriptive analyses except that more than one
parameter is considered at a time. Like descriptive analyses, association analyses
promote data familiarity and they are an important step in data quality assurance
and validation.
Study period representativeness analyses will assess how representative the field
study results are to other periods of time (other years). We will rely on existing
data for these analyses, including stream data from USGS, weather data from the
National Weather Service, and benthos data from the states of Maryland and
North Carolina. If the study period is found to be significantly unusual compared
to typical years or long-term composite conditions, the results will be interpreted
in this light.
Landscape indicator model analyses will rely on multiple regression techniques to
develop predictions for the individual dependent variables as a function of the
landscape metrics. This will be discussed further below.
Multivariate analyses provide a top-down approach to organize many variables
into a smaller number of unique groups. Classification and Regression Tree
analysis will identify rules for grouping data into classes, for example, levels of
human influence or hydrogeologic framework types. Taken together, these
analyses will provide insights on the best approach for applying the landscape
indicators. For example, the spatial applicability for each model will be
evaluated, and the contributions of the hydrogeologic variables will be identified.
This will also be discussed further below.
44
-------
• Case study analyses will rely on hydrologic, pesticide fate and transport, and
multimedia modeling at selected sites to help us articulate conceptual models and
the relative importance of factors such as differing soil and hydrogeologic
conditions. These case studies will start with a few sites and the simpler models
and progress toward more complex models and additional sites. The models
include a multimedia box model, a ground-water flow model, and pesticide fate
and transport models for soil. This work will be conducted in parallel with the
landscape indicator model development. The case studies and modeling are
discussed in the next section.
Table 10. Parameters Measured or Calculated for Each Site or Watershed
Stream
water
chemistry
On-site
-DO,
-tempera-
ture
-pH,
-stream
discharge
-dissolved
alkalinity
-specific
conduc-
tance
Laboratory
-major ions
USGS
schedule
2701
-pesticides
USGS
schedule
2001
-nutrients
USGS
schedule
2702
-pesticide
metabolites
-see
Appendices
A, C, and D
Bed
sediment
data
On-site
-composite of
1 1 samples
per site
Laboratory
-Aldrin
-Chlordane
-DDD
-DDE
-DDT
-Dieldrin
-Endosulfan
-Endrin
-Hepta-chlor
-Hepta-chlor
epoxide
-additional
chlorinated
pesticides
-PCB
Congeners,
-As, Hg
-see
Appendix B
Benthic
macro-
invertebrate
On-site
-9 samples per
site,
combine into
pool and
riffle
composites
Laboratory
-300 count
organism
identification
to genus and
species for
pool and
riffle
composites
-community
composition
-community
abundance
-various
indices
-see
Appendix E
Physical
habitat
data
On-site
-thalweg
profile
-woody
debris tally
-channel
characteri-
zation
-riparian
characteri-
zation
-compass
bearings
between
stations
-see
Appendix F
Rapid
habitat
assessment
On-site
Riffle/run:
-in stream
fish cover
-epifaunal
substrate
-embedded-
ness
-velocity/
depth
-channel
alteration
-sediment
deposition
-frequency
of riffles
-channel
flow status
-condition of
banks
-bank
vegetative
protection
-grazing/
other
pressure
-riparian
vegetation
width
(Pool/glide
similar)
Existing
landscape
data
Databases
-stream
hydrography
-digital
elevation
-soil data
-land use/
land cover
-roads
-county
boundaries
-population
-precipitation
-see
Appendix G
'
Pesticide
loading
data
Databases
estimates:
apportion to
agricultural
land use by
-zip code
-county
45
-------
Multiple comparison analysis techniques will utilize both new and available
stream and ground-water monitoring data to test a) whether the hydrogeologic
framework units are significantly different from each other; and b) whether a
different grouping would be more optimal. This is discussed further below.
Validation analyses will use reserved subsets of the field data, the nested site
data; and existing data from previous ground water and surface water studies to
evaluate the performance of the landscape indicator models. The reserved subsets
will not be used in the landscape indicator model development process, ensuring
that the model development is independent from the validation. Sensitivity
analyses will also be performed on landscape model parameters.
Reconciliation of results begins informally midway through the analysis effort and
will be completed at the end of that phase. Sharing of results will encourage
critical review, comparison with other results, and method refinements.
Consideration of the validation results is part of this process. Ultimately,
judgements of the technical credibility of the results will be made, and these
findings will designate the final landscape indicators.
• Journal article and reports summarizing the above results will be produced
throughout the analysis process, where appropriate.
Additional considerations in the preparation and analysis of the data are described below:
Landscape metrics will be calculated for the actual sites and watersheds sampled,
using the water sampling coordinates measured-by the field crew as the pour
point. Watersheds will be delineated using digital elevation data or hand drawn if
necessary. We expect that a low percentage of the watersheds will require hand
drawing. The spatial data types we expect to use, along with the data resolution,
and sources are listed in Appendix G.
For field study data, preparation includes traditional quality assurance and internal
consistency checks. For existing data, preparation includes review of the methods
and quality assurance information for the study that produced the data to identify
data of questionable quality.
• Concentrations of pesticides in stream water and bed sediments will be tracked as
individual compounds, as groups of compounds with similarities in properties and
use, and as totals of herbicides and insecticides. These totals will be an initial
indication of agricultural versus urban development. Major ions, including
nutrients in stream water, will be treated individually, as will toxic compounds
and arsenic and mercury in sediments. Base flow rates, stream size and
temperature will also be considered in the analyses of the data.
46
-------
• Indices for the macroinvertebrate data and physical habitat data will be computed;
proposed benthic macroinvertebrate indices and physical habitat metrics are listed
in Appendices E and F. Reference site data will be incorporated into these
analyses, according to the process used to apply each metric, and will also be used
for qualitative comparisons for other metrics. Each reference site will be
characterized in the same manner as the framework unit sites to better explain the
data observed. We intend to compare the reference sites with the framework sites,
and examine the differences in benthic community composition and abundance
between the developed and undeveloped watersheds. Data from the nested sites
and other sites in the Coastal Plain will be used to look at scale issues.
• Ecological condition, as determined by macroinvertebrate and physical habitat
data, will be compared to the data from the benthos reference sites that have been
selected because they represent excellent conditions. The combination of the
study sampling sites and the reference sites will give an indication of the relative
range and variability of the macroinvertebrate responses over the study area. If
unusually large spring storms occur in isolated portions of the study area, as
indicated by the rainfall maps, that will increase the variability. Potential impacts
of the 1999 hurricane season (Hurricane Floyd) and the drought of 1999 on
benthic macroinvertebrate community composition and abundance will be
considered in this analysis. This will be accomplished by obtaining hurricane
flooding information from the Federal Emergency Management Agency. We will
compare sampling sites with the flood locations from the previous year. This will
enable us to identify sites which were severely affected by these storms, compared
to those which were not. We also will have 4-kilometer resolution precipitation
maps based on radar data available through the-National Weather Service. This
will enable us to compute rainfall received by a site for the previous 9 months of
time. Drought has been an issue in some areas of the Coastal Plain for the past
several years. The rainfall data will be useful in evaluating the extent of drought
conditions at the sampling sites.
The performance of the study design will be evaluated to determine if enough
samples were collected, or if too many samples were collected, given the
variability measured.
These results will support the analyses that follow.
47
-------
LANDSCAPE INDICATOR MODELS
Model Development
The Landscape indicator analysis will rely heavily on multiple regressions and the CART
technique described earlier. We expect to use SAS®, and CART® software. Various
combinations of metrics will be tested for their success in explaining the variability encountered
in the data. Many of the metrics are correlated with each other and care will be taken to use
independent metrics (Riitters et al., 1995; Griffith and Amrhein, 1997). The best landscape
indicator model will result from an iterative process of selecting statistically significant variables
which are also physically and biologically meaningful. Each analyte will be treated individually
and also grouped into totals, classes, and groups. We will conduct a preliminary analysis (pair
wise correlations etc.) to determine preferred groups and minimize interdependencies. This will
be followed by step wise regression. With the chemicals we will be looking at canonical
correlations, and for landscape metrics, we will apply factor analysis.
Landscape indicator model relationships and hierarchical relationships using CART will
be developed using data from the 175 sites which are representative of about 10,000 first-order,
freshwater streams and their watersheds within the Mid-Atlantic Coastal Plain in New Jersey,
Maryland, Delaware, Virginia, and North Carolina. The list of independent and dependent
variables will be expanded and revised as we evaluate the performance of the landscape pattern
metrics (see Table 11). Some variables, such as cropping pattern, are of interest but not available
to us because of the scale and date (1992) of our land use/land cover data. We are treating
agriculture as a bulk property. We will apply the method of Luther and Haitjema (1998) to
estimate the mean ground water flow path and mean residence times for each watershed.
For the nested sites, watershed delineations and landscape characterization data will be
prepared for each watershed pour point sampled within the nested series of streams. Since this
data set is small, the data will be grouped by stream order and flow, and qualitative data analysis
will be used to evaluate the results. These data will also be used for the model validation below.
The nested watersheds will give us a way to consider how spatial scale affects our results. Our
first-order watershed design will help us to understand processes in small scale watersheds while
the data for the nested watersheds will help us to understand how the watersheds fit together.
These data are not intended for the landscape modeling effort.
48
-------
Table 11. Landscape Metrics and Dependent Variables for Analysis
Dependent variables:
pH in water
dissolved oxygen in water
specific conductance in water
dissolved alkalinity in water
pesticides in water (Appendix A)
pesticides in sediments (Appendix B)
major ions in water (Appendix C)
nutrients in water (Appendix D)
benthic macroinvertebrates (Appendix E)
physical habitat (Appendix F)
pesticide metabolites in water
Landscape metrics:
watershed area
percentage of watershed in agricultural land use (MRLC data)
percentage of contiguous agricultural land use
percentage of watershed in urban land use (MRLC data)
population density
percentage of impervious surfaces (estimated from land use)
length and width of riparian buffer zones
road-to-stream distance
percentage of agriculture on steep slopes
gradient of stream
Universal Soil Loss Equation soil credibility
soil texture, permeability
hydrogeologic unit code
depth to ground water
mean length of ground-water flow path (Luther and Haitjema, 1998)
mean residence time of ground water (Luther and Haitjema, 1998)
Model Validation/ Sensitivity Analysis
Model validation involves the calculation of landscape indicator values for watersheds
which have associated pesticides and nutrient data for stream water and pesticide, PCB, arsenic
and mercury for bed sediments. The calculated values will be compared to the actual values and
percent differences will be determined. Some of the data used for validation will be subsets
from the field data, which will not be used in the landscape indicator development process; this
includes data from the 175 framework unit sites, as well as the nested sites. We have tentatively
identified approximately 25 surface water sites and approximately 50 ground- water sites from
49
-------
previous studies which have pesticides and nutrient data. We will also evaluate these studies and
perform comparisons with these data.
Sensitivity analysis will be performed by making incremental changes in the regression
coefficients while holding the other coefficients constant. We may also test changes to ratios of
coefficients. Some regression models may perform better in certain hydrogeologic framework
units, compared to others. This hypothesis will be evaluated by comparing the goodness-of-fit
statistics for the whole area to those for each hydrogeologic framework unit. The effect of using
higher resolution Soil Survey Geographic (SSURGO) data instead of State Soil Geographic
(STATSGO) data at sites where both are available will be investigated.
HYDROGEOLOGIC FRAMEWORK
We will rely on both existing and new data for surface water and ground water in the
evaluation of the hydrogeologic framework. Existing water-quality data will be used to
determine if the regional aspects of the framework accurately represent factors that affect
regional water quality. Results from local-scale, analyses of structure and function of the stream
biota and habitat and the physical and chemical processes will be compared to the processes
hypothesized as affecting water quality in different subregions of the framework. Available data
will then be used to statistically evaluate the relevance of the framework to the description of
actual water-quality conditions. This effort will include ground-water and surface-water data
available from the NAWQA program and other data that are comparable in quality and study
design. The statistics to be used will largely be determined by the data and are likely to include
nonparametric tests and an analysis of variance to identify differences in chemical concentrations
or other indicators among framework regions. If the data are too heavily censored for this (data
below method detection limits which is common for pesticides), contingency tables could be
used (with a consequent loss of power). We may also use correlations or regressions when
comparing continuous variables. Hydrologic applications of these statistics are covered in
general in Helsel and Hirsch (1992). Some examples of analysis-of-variance-type tests in
environmental science include Blomquist et al. (1996); Ator and Denis (1997); and Ator and
Ferrari (1997). For heavily censored data, probit or logistic regression can be used (Eckhardt and
Stackelberg, 1995; Tesoriero and Voss, 1997; Liu et al.,1996). Blomquist et al. (1996) used
parametric regressions. Ator and Denis (1997) used correlations and two-way ANOVA. Ator
and Ferrari (1997) used contingency tables because the pesticide data were heavily censored.
DATA MANAGEMENT
Data storage and retrieval will be accomplished initially with Microsoft Access®, Arc
Info®, and Arc View® computer software. Field data (water, macroinvertebrate, and sediment
data) will be available to study participants via a USGS-operated website, while the large spatial
files will be shared via compact disk. Eventually the data from the study will be incorporated
into EPA's Environmental Information Management System. A website describing the study and
providing status reports and updates will be available to the public.
50
-------
HYDROLOGIC AND MULTIMEDIA MODELING
Hydrologic, pesticide fate, and multimedia process models can provide additional
insights about watersheds within the study area. By approaching selected sites from a case study
point of view, we hope to gain added conceptual understanding of the physical and chemical
processes involved. The temporal sites will be modeled first because of the more extensive data
available. Starting with a few sites and simple models, we will expand the effort to include sites
representative of all the hydrogeologic framework units. Because pesticide application rates are
such an important factor, we are pursuing acquisition of specific pesticide application rate data
for some of these sites. We may use some of the results derived from the hydrologic and
multimedia modeling in the landscape indicator modeling. Alternatively, the landscape indicator
modeling results may suggest scenarios and case studies for the hydrologic and multimedia ,
modeling.
Many models are being considered to help us investigate the most significant processes as
our data analysis develops (see Table 12). A key factor is data needed for the models. We are
searching for the best available input data for these models and expect use values from the
literature, for example, for atmospheric deposition and irrigation use. The modeling effort will
be built progressively, starting with the simpler cases and models, and then moving to the more
complex.
Table 12. Models Under Consideration
Model
Mend-Tox™
SESOIL
PRZM-3
MODFLOW, MODPATH
GFLOW
MT3D
SPARROW
Purpose
multimedia compartmental
estimates of pesticide
concentrations over time
long term-fate and migration of
pollutants in vadose zone
pesticide degradation and
transformation; vertical leaching in
crop root zone; run off from
different land cover zones
ground-water flow (finite
difference model), advective flow
ground-water flow (analytic
element model) better advective
flow
transport and transformations
spatially referenced regression
model, estimates source and fate of
contaminants in streams
Reference
Cohen, 1986; Onishi et al., 1990
Bonazountas et al., 1997
Mullins et al., 1993
Harbaugh and McDonald, 1996
Kelson and Haitjema, 1994;
Haitjema, 1995
Zheng, 1992
Smith' etal., 1997
51
-------
Mend-Tox™ is a multimedia model which estimates the distribution of organic chemicals
for seven compartments in the environment: air, atmospheric aerosols, surface water, suspended
solids in water, aquatic organisms, soil, and vegetation. Mend-Tox™ identifies significant and
insignificant transport and exposure pathways, and it can estimate potential persistence of
chemicals in the environment. Besides being useful in a case study, Mend-Tox™ may be useful
in the landscape indicator application effort, for answering "What if?" type questions.
The hydrologic and pesticide fate and transport models (SESOIL, PRZM-2, MT3D with
MODFLOW, and GFLOW) may be useful in understanding the relative behaviors of the
pesticides in soils, surface water, and ground water. Further evaluation of their requirements and
outputs will be needed before we decide which one(s) to use for the estimation process. Initial
GFLOW modeling will be conducted by U.S. EPA NERL-Athens staff for two temporal sites in
the Coastal Plain: the Chesterville Branch site in Maryland and the Lizzie site in the Contentnea
River drainage in North Carolina. This modeling effort will share its results with this study and
another research program at Athens. The modeling for the Chesterville Branch site depends on
data collected by a study funded by the U.S. EPA's NRML-Ada and being shared with NERL-
Athens and NERL-Las Vegas. The data being collected with NERL-Las Vegas funding are being
shared with both of the other facilities.
The Spatially Referenced Regressions on Watershed Attributes (SPARROW) model
(Smith et al., 1997) may be used as an adjunct to the landscape indicator evaluation process to
help relate landscape characteristics to water quality. This statistical model reduces common
problems associated with relating surface water quality to watershed landscapes, including
sparseness of sampled locations, spatial bias in the sampling network, and drainage basin
heterogeneity. To account for natural long-term hydrologic variation, SPARROW is typically
used with estimated long-term average contaminant loads at sampling sites to estimate
cumulative downstream loads based on watershed characteristics. A SPARROW model for
nutrients in nontidal portions of the Chesapeake Bay Watershed has been developed (Preston and
Brakebill, 1999) and the addition of the Delaware River Basin to this model is under
consideration. However, surface-water stations with long-term, historical water-quality data are
relatively scarce in the Mid-Atlantic Coastal Plain, particularly for pesticides (Ferrari et al.,
1997). Initially, SPARROW model results which relate landscapes to stream quality in parts of
the Mid-Atlantic Coastal Plain and overlap our study sites will be compared to the landscape
indicator model results. The winter base flow data we collect may be useful for creating a
SPARROW model to predict landscape effects on streams, although SPARROW has yet to be
used in this way. This possibility will be investigated further and pursued if warranted. It may
be possible to corroborate the landscape indicator model analysis with the SPARROW model
results for selected, nested watersheds. The application of SPARROW for estimating pesticide
concentrations in streams is a topic of current interest for the EPA Office of Pesticide Programs.
52
-------
LANDSCAPE INDICATOR APPLICATIONS
POTENTIAL APPLICATIONS
Once the landscape indicator models are validated and the final selection of indicators is
complete, it will be possible to use the results to gain a broader perspective. The landscape
indicator models can be applied to the entire set of approximately 10,000 first-order, freshwater
watersheds in the Coastal Plain to predict their status for selected dependent variables. Land use
change scenarios can be applied to model the sensitivity of the watersheds and identify and rank
the most vulnerable on a relative basis. The results can be displayed on detailed maps. New
monitoring designs for different purposes (identifying most pristine areas, most vulnerable areas,
or areas most suitable for restoration) can be demonstrated and can be evaluated using the
landscape indicators and the underlying data bases. Watersheds with similar issues can be
identified, so they can be treated as a group, such as for Total Maximum Daily Load
development. Specific questions, and priorities will be addressed in conjunction with the
stakeholders.
STAKEHOLDERS AND OUTREACH
One of the principles of the Landscape Ecology Branch (Jones et al, 2000) is to involve
stakeholders early and throughout a project. Thus, meetings with representatives from Regions
2, 3, and 4, and the Office of Pesticide Programs and the Office of Water have already occurred
in FY99 and FYOO. An overview of the project has also been presented to representatives from
state agencies within Region 3. The intent is to continue to find key individuals with interest in
our activities, and build these relationships, involving these individuals in developing the
landscape indicator applications. These individuals would bring different perspectives and needs
into the project for consideration and often contribute substantial expertise gained from working
in the area over a number of years. The decision points where stakeholder involvement is desired
include
• selection of benthos reference sites;
• selection of case study sites;
review of the hydrogeologic framework results;
review of the landscape indicator results; and
• development of scenarios for landscape indicator applications.
Stakeholder contributions will help us to focus the study and meet their needs. We expect to
work closely with the appropriate EPA Regional Offices to develop a pesticide indicator atlas for
the Mid-Atlantic Coastal Streams, to provide public workshops to disseminate the results, and to
publicize the availability of the reports, journal articles, and data which result from this study.
53
-------
QUALITY ASSURANCE
The Mid-Atlantic Coastal Stream Study relies almost entirely on existing protocols and
procedure manuals for its field and laboratory activities. Two large sampling programs, EMAP
and NAWQA, have standardized many sampling and laboratory procedures, and we are taking
advantage of their work. Table 7 (page 40) lists the methods and quality assurance manuals
being followed in the study. Not only is there a substantial cost savings in using existing
protocols, there is also a payoff in comparability of data, which enhances its value for the long
term. Some quality assurance highlights for the study are listed below.
Water sampling and on-site analysis of water samples will be conducted following
USGS NAWQA sampling protocols. This includes independent audits of
sampling crew activities.
Macroinvertebrate sampling will follow EMAP QA protocols.
Physical habitat assessment will follow EMAP QA protocol, with modifications.
• A methods, quality assurance, and safety workshop will be conducted for the
macroinvertebrate and physical habitat field crews the week before spring
sampling starts.
All the laboratory water sample analyses for pesticides, major ions, and nutrients
will be performed by the USGS National Water Quality Laboratory (NWQL).
The NWQL uses quality-control data to measure and monitor bias and variability
in analytical methods (Pirkey and Glodt, 1998). Three levels of quality control at
the NWQL include continuing analyses of method performance, data review and
blind sample programs, and participation in inter-laboratory performance
evaluation studies.
• NWQL method performance is evaluated with results from quality control
samples included with each batch of environmental samples. Quality control
samples for inorganic analyses include blanks, standard reference materials, and
laboratory replicates. Surrogate compounds and laboratory reagent blanks and
spikes are used to monitor analyses for organic compounds.
NWQL monitors method performance throughout the laboratory and over long
periods using blind samples (Ludtke and Woodworth, 1997) and data reviews.
Inorganic data are reviewed with logic checks such as cation-anion balances. Field
and laboratory values and filtered and unfiltered values are also compared. For
organic analyses, long-term data from the first level of quality control are
analyzed to compute method control limits and acceptance criteria.
NWQL participates in multiple inter-laboratory studies with the U.S. EPA, the
54
-------
National Oceanic and Atmospheric Administration (NOAA), and the National
Water Research Institute of Canada (Glodt and Pirkey, 1998).
Data management will follow the EPA guidelines and databases and metadata will
meet the Federal Geographic Data Committee requirements. Data will be
accessible through EPA's Environmental Information Management System.
Modeling follows the EPA guidelines on use of existing models, with
documentation of assumptions, parameter values and sources, boundary and initial
conditions, validation and calibration of the model, and output. It includes
periodic testing of the model with a standard data set and comparison to known
results.
The staff is trained in the use of software (for example, Arc View, Arc/INFO,
SAS, CART, Mend-Tox), maintains their expertise through frequent use, and is
available to mentor others in the use of these software packages.
Data interpretation will be reviewed within the project group as results become
available. When possible, results will be computed two or more ways and the
results will be compared.
The decision analysis tools will follow the EPA guidelines for use of existing
software and for developing new software. When complete, they will be
accessible through a website which will follow EPA website requirements.
This plan will be externally peer reviewed. Interim and final results of the study
will also be peer reviewed as part of the normal journal article submission
process.
Stakeholder input is planned at several stages of the study; this effort is described
in detail in the section "Landscape Indicator Applications."
Quality assurance reports are due from each of the major contributors after the
data collection and sample analysis is complete (USGS, Benthos Sampling and
Benthos Analysis contractors).
Audits will be performed by the EPA as scheduled by the Principal Investigator
and the Project Quality Assurance coordinator. Sampling procedure audits will be
conducted once for each sampling crew during the field season to assess
compliance with sampling protocols. A benthos laboratory sample audit will be
conducted to assess compliance with standard laboratory procedures. The
chemistry laboratories will not be visited because they participate in existing
round-robin and other quality assurance activities, such as those mentioned for the
NWQL, above.
55
-------
A central project file exists at NERL/ESD-Las Vegas, maintained by the Principal
Investigator, with copies of the documentation described above. Project
management records and budget information are also available.
56
-------
SCHEDULE AND MILESTONES
A schedule and milestones for the study, including a Government Performance Results
Act (GPRA) deliverable, have been developed for the years FY98-FY03 (see Table 13). The
table provides a quick overview of the activities and responsibility for completing them.
Table 13. Detailed List of Milestones by Fiscal Year
Year
FY98
FY99
FYOO
Event
Analyzed existing pesticide data for Mid-Atlantic
Reviewed literature
Initiated USGS interagency agreement
Complete hydro-geologic framework
Begin analysis of hydrogeologic framework using existing data
Augment GIS coverages
Characterize watershed support areas for first- and second-
order streams
Begin statistical sampling design and select sites
Select ground water-surface water models
Initiate data and QA management
Initiate arrangements for benthos sampling
Identify scientific collaborators
Contact stakeholders
Complete statistical sampling design and site selection
Involve stakeholders in site selection where practical
Collect and analyze water samples
Collect and analyze benthos samples
Complete analysis of framework performance using existing
data
Prepare database for water sample, benthos, and physical
habitat data; derive interpretive measures and include in
database; prepare metadata
Initiate modeling (except for SPARROW)
Responsibility
EPA, USGS
EPA
EPA
USGS
USGS
EPA
EPA, USGS
USGS, EPA
EPA, USGS
EPA
EPA
EPA
EPA
USGS, EPA
EPA
USGS
EPA
USGS
EPA, USGS
EPA.
57
-------
FY01
FY02
FY03:
Evaluate hydrogeologic framework with new data
Evaluate landscape indicators with new data
Develop Mid-Atlantic Coastal Stream assessment
Complete modeling
Conduct SPARROW modeling
Prototype of decision analysis tool
Share results with stakeholders
GPRA Deliverable: Condition of streams and ground water to
with respect to pesticides and nutrients: development of
landscape indicators for the Mid-Atlantic Coastal Plain, due
9/01
Make data available on Internet
Deliverable: Landscape characterization of first order
watersheds in the Mid-Atlantic Coastal Plain; Journal Article,
9/02
EPA Report: Landscape Atlas for Pesticides and Nutrients in
Mid-Atlantic Coastal Freshwater Streams, (due 9/03)
USGS
EPA
EPA, USGS
EPA
USGS
EPA
EPA, USGS
EPA
EPA
EPA
EPA
58
-------
POTENTIAL FOR REDUCING UNCERTAINTY
Over the long term, the landscape indicators generated from this research are expected to
contribute to assessments of condition, vulnerability, and risk to aquatic ecosystems due to
pesticides and toxic substances at multiple scales across the U.S.
For the Mid-Atlantic Coastal Plain, this research should result in a validated
hydrogeologic framework which will provide a context for the nutrients, pesticides, and toxic
substance results. The combination of the framework and the pesticides and toxic substance
indicators can be used to develop monitoring designs, identify similar watersheds, and estimate
water quality for these parameters. It should also lead to an improved understanding of how
landscape indicators for pesticides and toxic substances vary within hydrogeologic classes,
versus how these parameters vary between these classes. From this information, we should be
able to predict condition and relative ranking of stream segments. This will provide us with a
complete case study showing how the elements of the pesticide and toxic substance data,
landscape and water quality parameters, and ultimately landscape indicators can be integrated in
a regional-scale water quality assessment. These results are expected to be useful to EPA's
Office of Prevention, Pesticides, and Toxic Substances (OPPTS); Office of Water (OW), and
Regional Offices; and also state and local agencies with responsibilities for managing water
resources for pesticides and toxic substances.
The development of the hydrogeologic framework contributes to the following High and
Medium Priority needs identified in "TMDL Scientific Needs: A Regional and Office of Water
Assessment (March 18,1998):
High Priority Monitoring and Assessment Technical Support Needs: Monitoring
designs for identifying impacted water bodies.
• High Priority Modeling Research Needs: Development of watershed similarity
indices to extrapolate loading rates of key stressors.
Medium Priority Monitoring and Assessment Research Needs: Development of
extrapolation techniques to estimate water quality condition in nonmonitored
segments.
Medium Priority Data and GIS Technical Support Needs: Aquatic resource data:
water column physical/chemical data linked to hydrographic coverages.
59
-------
PERFORMANCE MEASURES
The EPA, to better account for the success of its actions, has developed a cascading set of
goals, objectives, subobjectives, milestones, measures, tasks, and products in compliance with
the Government Performance Results Act (GPRA). There are currently 10 longer-term goals for
the EPA under the GPRA. Goal 8, "Provide sound science to improve the understanding of
environmental risk, and develop and implement innovative approaches for current and future
environmental problems," serves as the foundation, or core of the Office of Research and
Development's (ORD) Ecological Research Program. The specific objective associated with
ORD's ecoresearch under this "Sound Science" goal is to provide the scientific understanding to
measure, model, maintain, or restore at multiple scales the integrity and sustainability of
ecosystems now, and in the future.
In addition, the ORD's "Ecological Research Strategy" identifies major objectives, sub-
objectives and products associated with its core research program areas of:
Ecosystem monitoring research
• Ecological processes and modeling research
• Ecological risk assessment research, and
• Ecosystem risk management restoration research
Shorter-term accounting of success is accomplished by establishing and monitoring the
response to the annual performance goals (APGs) and measures (APMs) under GPRA and
progress toward completion of any additional critical research products identified in the ORD's
"Ecological Research Strategy" and its subsequent updates. These goals and measures provide
the "why" and the "what" of our research tasks and projects. This document, as a technical
research plan, addresses not only the "why" and the "what" but also the "how" ~ the approach to
providing products that satisfy the specific performance goals associated with this activity.
This research project supports, Goal 2 (Water), Goal 4 (Preventing Pollution and
Reducing Risk) and Goal 8 (Sound Science). Specific annual performance goals and measures
are listed in the next section.
60
-------
ANTICIPATED RESULTS/PRODUCTS
FYOO:
Research Plan: Testing Landscape Indicators for Stream Condition Related to Pesticides
and Nutrients: Landscape Indicators for Pesticides Study for Mid-Atlantic Coastal
Streams (Due 9/00)
FY01:
GPRA Deliverable: Journal Article: Condition of streams and shallow ground water with
respect to pesticides and nutrients: development of landscape indicators for the Mid-
Atlantic Coastal Plain. This product provides the scientific basis for the use of landscape
indicators which can identify a) similar watersheds and b) streams impacted by nutrients
and pesticides, (due 9/01)
FY02:
Journal Article: Landscape characterization of first order watersheds in the Mid-Atlantic
Coastal Plain, (due 9/02)
FY03:
EPA Report: Landscape Atlas for Pesticides and Nutrients in Mid-Atlantic Coastal
Freshwater Streams, (due 9/03)
61
-------
LITERATURE CITED
Abel, P.D., 1980. Toxicity of hexachlorocyclohexane (Lindane) to Gammaruspulex; mortality in
relation to concentration and duration of exposure. Freshwater Biology, 10, 251-259.
Ator, S.W., and J.M. Denis, 1997. Relation of nitrogen and phosphorus in ground water to land
use in four subunits of the Potomac River Basin. USGS WRIR 96-4268, 26 p.
Ator, S.W., and M.J. Ferrari, 1997. Nitrate and selected pesticides in ground water of the Mid-
Atlantic Region. Water Resources Investigation Report 97-4139, 8 p.
Ator, S.W., J.D. Blomquist, J.W. Brakebill, J.M. Denis, M.J. Ferrari, C.V. Miller, and H. Zappia,
1998. Water quality in the Potomac River Basin, Maryland, Pennsylvania, Virginia, West
Virginia, and the District of Columbia. 1992-96, USGS Circular 1166.
Bachman, L.J., B. Lindsey, J.W. Brakebill, and D.S. Powars, 1998. Ground-water discharge and
base-flow nitrogen loads of non-tidal streams, and their relation to a hydrogeomorphic
classification of the Chesapeake Bay watershed, Middle Atlantic Coast. U.S. Geological Survey
Water-Resources Investigations Report 98-4059, 71 p.
Barbash, J.E, and E.A. Resek, 1996. Pesticides in ground water: distribution, trends, and
governing factors. Chelsea, Michigan, Ann Arbor Press, Inc., 588 p.
Barber, M.C. ed., 1994. Environmental Monitoring and Assessment Program: Indicator
Development Strategy, EPA/620/R-94/022, Athens, GA. U.S. Environmental Protection
Agency, Office of Research and Development, Environmental Research Laboratory.
Barbour, M.T., J. Gerritsen, B.D. Snyder, and J.B. Stribling. 1999. Rapid Bioassessment
Protocols for Use in Streams and Wadeable Rivers: Periphyton, Benthic
Macroinvertebrates and Fish, Second Edition. EPA 841-B-99-002. U.S. Environmental
Protection Agency; Office of Water; Washington, D.C.
Battaglin, W.A., and D.A. Goolsby, 1994. Spatial data in geographic information system format
on agricultural chemical use, land use and cropping practices in the United States. USGS, Water
Resources Investigations Report 94-4176, http://water.usgs.gov/pubs/bat/batOQO.html
Blomquist, J.D., G.T. Fisher, J.M. Denis, J.W. Brakebill, and W.H. Werkheiser, 1996. Water-
quality assessment of the Potomac River Basin ~ Basin description and analysis of available
nutrient data, 1970-90. USGS Water Resources Investigations Report, 95-4221, 88 p.
Bode, R.W., M.A. Novak, and L.E. Abele, 1996. Quality assurance work plan for biological
stream monitoring in New York State. New York State Department of Environmental
Conservation, Albany, NY, 89 p.
62
-------
Bohlke, J.K., and J.M. Denver, 1995. Combined use of ground water dating, chemical, and
isotopic analyses to resolve the history and fate of nitrate contamination in two agricultural
watersheds, Atlantic Coastal Plain, Maryland, Water. Resources Research, Vol. 31, p. 2319-
2339.
Bohlke, J.K., J.M. Denver, R.B. Wanty, and P.B. McMahon, 1996. Some hydrologic controls on
distribution of nitrate on agricultural watersheds. Spring AGU meeting, 1996 Supplement, Vol.
77.
Bonazountas, M., D. Hetrick, and P. Kostecki, 1997. SESOIL in Environmental Fate and Risk
Modeling, Amherst Scientific Publishers, Amherst, MA, 686 p.
Brieman L., J.H. Friedman, R.A. Oshen, and C.J. Stone, 1984. Classification and Regression
Trees, Chapman & Hall, New York.
Brown, P.M., J.A. Miller, P.M. Swain, 1972. Structural and stratigraphic framework, and spatial
distribution of permeability in the Atlantic Coastal Plain, North Carolina to New Jersey, U.S.
Geological Survey Professional Paper 796, 79 p.
Brown, D.E., F. Reichenbacher, and S.E. Franson, 1998. A Classification System of North
American Biotic communities. University of Utah Press, Salt Lake City, UT, 152 p.
Calow, P. 1998. Introduction: the chemical challenge. Environmental Monitoring and
Assessment, 53: 391-394.
Carlile, D.W., J.R. Skalski, J.E. Batker, J.M. Thomas, and V.I.- Cullinan, 1989. Determination of
ecological scale. Landscape Ecology, 2(4), p. 203-214.
Caspers, H., and C.W. Heckman, 1981. Ecology of orchard ditches along the freshwater section
of the Elbe Estuary. Arch. Hydrobiol. Supple. 43, 347-486.
Chow, V.T., D.R. Maidment, and L.W. Mays, 1988. Applied Hydrology, McGraw-Hill, Inc.,
New York, 572 p.
Clark, L. A., and D. Pregibon, 1992. Tree Based Models, in: Statistical Models in S.
T.J. Hastie, and J.M. Chambers, (Eds.), Wadsworth & Brooks California, 608 p.
Clements, W.H., D.S. Cherry, and J.H. Van Hassel, 1992. Assessment of the impact of heavy
metals on benthic communities at the Clinch River (Virginia): evaluation of an index of
community sensitivity. Can. J. Fish. Aquat. Sci. 49: 1686-1694.
Cohen, Y. ed., 1986. "Intermedia transport modeling in multimedia systems," in Pollutants in a
Multimedia Environment, Plenum Press, New York.
Cordy, C., 1993. An extension of the Horvitz-Thompson theorem to point sampling from a
continuous universe. Probability and Statistics Letters, 18, 353-362.
63
-------
Corkum, L.D., and J.J.H. Ciborowski, 1988. Use of alternative classifications in studying broad-
scale distributional patterns of lotic invertebrates. Journal of North American Benthological
Society, 7: 167-179.
Corkum, L.E., 1992. Relationships between density of macroinvertebrates and detritus in rivers.
Arch. Hydrobiol., 125: 149-166.
Correll, D.L., I.E. Jordan, and D.E. Weller 1992. Nutrient flux in a landscape: effects of coastal
land use and terrestrial community mosaic on nutrient Transport to coastal waters. Estuaries,
Vol. 15, No. 4. p. 431-442.
Cronan, C.S., J.T. Piampiano, and H.H. Patterson, 1999. Influence of land use and hydrology on
exports of carbon and nitrogen in a Maine river basin. J. Environ. Qual., 28: 953-961.
Cuffhey, T.F., B.F. Wallace, and J.R. Webster, 1984. Pesticide manipulation of a headwater
stream: invertebrate responses and their significance for ecosystem processes. Freshwater
Invertebrate Biology, Vol. 3, No. 4, 153-171.
Cuffhey, T.F., M.E. Gurtz, and M.E. Meador, 1993. Methods for collecting benthic invertebrate
samples as part of the National Water-Quality Assessment Program. U.S. Geological Survey
Open-File Report 93-406, 66 p.
Cuffney, T.F., M.R. Meador, S.D. Porter, and M.E. Gurtz, 1997. Distribution offish, benthic
invertebrate, and algal communities in relation to physical and chemical conditions, Yakima
River Basin, Washington, 1990. U.S. Geological Survey Water Resources Investigations Report
96-4280, 94 p.
Gushing, E.M., I.H. Kantrowitz, and K.R. Taylor, 1973. Water resources of the Delmarva
Peninsula. U.S. Geological Survey Professional Paper 822, 58 p.
Denver, J.M., and S.W. Ator, 1999, personal comm.
Dingman, S.L., 1994, Physical Hydrology, Prentice Hall, Upper Saddle River, New Jersey, 575 p.
Eckhardt, D.A.V., P.E. and Stackelberg, 1995. Relation of ground-water quality to land use on
Long Island, NY. Ground Water, vol. 33, no. 6, p. 1019-1033.
Efron, B., and R. Tibshirani, 1991. Statistical data analysis in the computer age, Science, Vol.
253, p. 390-395.
Ferrari, M.J., S.W. Ator, J.D. Blomquist, and J.E. Dysart, 1997. Pesticides in surface water of
the Mid-Atlantic region. U.S. Geological Survey Water Resources Investigations Report 97-
4280, 12 p.
Fore, L.S., J.R. Karr, and R.W. Wisseman, 1996. Assessing invertebrate responses to human
activities: evaluating alternative approaches. J. N. American Benthological Society, 15(2), p.
64
-------
212-231.
Gianessi, L.P., and C.A. Puffer, 1990 (revised 1991). Herbicide use in the United States.
Resources for the Future, Quality of the Environment Division, Washington, D.C., 128 p.
Gianessi, L.P., and C.A. Puffer, 1992a. Fungicide use in the United States. Resources for the
Future, Washington, D.C.
Gianessi, L.P., and C.A. Puffer, 1992b. Insecticide use in the United States. Resources for the
Future, Washington, D.C.
Glodt, S.R., and K.D. Pirkey, 1998. Participation in performance-evaluation studies by U.S.
Geological Survey National Water Quality Laboratory: U.S. Geological Survey Fact Sheet FS-
023-98, 6 p.
Graham, R.L., C.T. Hunsaker, R.V. O'Neill, and B.L. Jackson, 199L Ecological risk assessment
at the regional scale. Ecol. Appl. 1: 196-206.
Griffith, D.A., and C.G., Amrhein, 1997. Multivariate Statistical Analysis for Geographers,
Prentice Hall, Upper Saddle River, New Jersey, 345 p.
Haitjema, H.M., 1995. Analytic Element Modeling of Ground-water Flow. Academic Press,
New York.
Hallberg, G.R., 1987. Agricultural chemicals in ground water: extent and implications,
American Journal of Alternative Agriculture, Vol. 11, No. 1, p. 3-15.
Harbaugh, A.W. and M.G. McDonald, 1996. Users documentation for MODFLOW-96, an
update to the U.S. Geological Survey modular finite-difference ground-water flow model, OF 96-
0485, 56 p.
Helsel, D.R., and R.M. Hirsch, 1992. Statistical Methods in Water Resources, Elsevier,
Amsterdam, Netherlands, 522 p.
Herlihy, A.T., D.P. Larsen, S.G. Paulsen, N.S. Urquhart, B.J. Rosenbaum, 2000. Designing a
spatially balanced, randomized site selection process for regional stream surveys: the EMAP
Mid-Atlantic Pilot Study, Environmental Monitoring and Assessment, 63 (1): 95-113.
Hilsenhoff, W.L., 1987. An improved biotic index of organic stream pollution. The Great Lakes
Entomologist, Vol. 20, p. 31-39.
Horn, R.C., and W.M. Grayman, 1993. Water-Quality modeling with EPA Reach File System.
Journal of Water Resources Planning and Management, Vol. 119, No. 2, p. 262-274.
Hostetler, K.A., and E.M. Thurman, 1999. Determination of chloroacetanilide herbicide
metabolites in water using high-performance liquid chromatography-diode array detection and
65
-------
high-performance liquid chromatography/mass spectrometry. Morganwalp, D.W., and H.T.
Buxton, Eds., U.S. Geological Survey Toxic Substances Hydrology Program. Proceedings of the
technical meeting, Charleston, South Carolina, March 8-12, 1999, Volume 2, Contamination of
hydrologic systems and related ecosystems. U.S. Geological Survey Water Resources
Investigation Report 99-4018B.
Hunsaker, C. T., R. L. Graham, G. W. Suter, II, R. V. O'Neill, L. W. Barthouse, and R. H.
Gardner, 1990. Assessing ecological risk on a regional scale. Environmental Management 14:
325-332.
Hunsaker, C.T., and D.A. Lavine, 1995. Hierarchical approaches to the study of water quality in
rivers. Bioscience 45: 193-203.
Jackson, L.E., J.C. Kurtz, and W.S. Fisher, 1999. Evaluation guidelines for ecological
indicators. EPA/620/R-99/005, U.S. Environmental Protection Agency, Office of Research and
Development, Research Triangle Park, NC 27711,
http://www.epa.gov/emap/html/pubs/docs/resdocs/ecol_ind.pdf
Jones, B., J. Walker, K.H. Riitters, J.D. Wickham, and C. Nicoll, 1996. Indicators of landscape
integrity, in J. Walker and D.J. Reuter, (Eds.), Indicators of Catchment Health: a technical
perspective. CSIRO, Melbourne, p. 155-168.
Jones, K.B., K.H. Riitters, J.D. Wickham, R.D. Tankersley Jr., R.V. O'Neill, D.J. Chaloud, E.R.
Smith, and A.C. Neale, 1997. An ecological assessment of the United States Mid-Atlantic
region: A landscape atlas. U.S. Environmental Protection Agency, Office of Research and
Development, Washington, D.C., EPA/600/R-97/130.
Jones, K.B, L.R. Williams, A.M. Pitchford, T.E. Slonecker, J.D. Wickham, R.V. O'Neill, D.
Garofalo, K.H. Riitters, W.G. Kepner, and Goodman, I.A., 2000. A national assessment of
landscape change and impacts to aquatic resources, a 10-year research strategy for the landscape
sciences program. U.S. Environmental Protection Agency, EPA/600/R-00/001.
Jones, K.B., A.C. Neale, M.S. Nash, R.D. VanRemortel, J.D. Wickham, K.H. Riitters, and R.V.
O'Neill, in press. Predicting nutrient and sediment loadings to streams from landscape metrics:
A multiple watershed study from the United States Mid-Atlantic Region, Landscape Ecology.
Johnson, A.R., 1994. Spatiotemporal hierarchies in ecological theory and modeling. GIS and
Environmental Modeling, p. 451-456.
Johnson, L.B., C. Richards, G.E. Host, and J.W. Arthur, 1997. Landscape influences on water
chemistry in Midwestern stream ecosystems. Freshwater Biology, 37: 193-207.
Kalkoff, S.J., D.W. Kolpin, E.M. Thurman, I. Ferrer, and D. Barcelo, 1998. Degradation of
chloroacetanilide herbicides: the prevalence of sulfonic and oxanilic acid metabolites in Iowa
ground waters and surface waters. Environmental Science and Technology, Vol. 32, p. 1738-
1740.
66
-------
Karr, J.R., and E.W. Chu, 1999. Restoring Life in Running Waters, Better Biological
Monitoring, Island Press, Washington, D.C. 206 p.
Karr, J.R., and E.W. Chu, 2000. Sustaining Living Rivers, Hydrobiologia, 422/423, p. 1-14.
Kaufrnann, P.R. (Ed.), 1993. Physical Habitat, p. 59-69 in R.M. Hughes (Ed.), Stream Indicator
and Design Workshop, EPA/600/R-93/138, U.S. Environmental Protection Agency, Corvallis,
Oregon.
Kaufrnann, P.R., and E.G. Robison, 1998. Physical habitat characterization, in Lazorchak, J.M,
Klemm, D.J., and Peck, D.V. (Eds.), Environmental Monitoring and Assessment Program -
Surface Waters: Field Operations and Methods for Measuring the Ecological Condition of
Wadeable Streams. EPA/620/R-94/004F, U.S. Environmental Protection Agency, Washington,
D.C.
Kaufrnann, P.R., P. Levine, E.G. Robison, C. Seeliger, and D.V. Peck, 1999. Quantifying
physical habitat in wadeable streams. EPA/620/R-99/003, U.S. Environmental Protection
Agency, Washington, D.C.
Keitt, T.H., D.L. Urban, and B.T. Milne, 1997. Detecting critical scales in fragmented
landscapes. Conservation Ecology [online] 1(1): 4. Available from the Internet, URL:
http://www.consecol.org/voll/issl/art4
Kelson, V.A., and H.M. Haitjema, 1994. GFLOW Users Manual. Haitjema Software,
Indianapolis, IN.
Kennen, J.G., 1999. Relation of macroinvertebrate community inpairmentto chatchment
characteristics in New Jersey streams. Journal of American Water Resources Association, 35:4,
p. 939-955.
Kepner,W.G., K.B. Jones, D.J. Chaloud, J.D. Wickham, K.H. Riitters, R.V. O'Neill, 1995. Mid-
Atlantic landscape indicators project plan, Environmental Monitoring and Assessment Program.
EPA620/R-95/003, U.S. Environmental Protection Agency, Washington, D.C.
Kerans, B.L. and J.R. Karr, 1994. A benthic index of biotic integrity (B-IBI) for rivers of the
Tennessee Valley, Ecological Applications, 4(4), p. 768-785.
Klauda, R., P. Kazyak, S. Stranko, M. Southerland, N. Roth, and J. Chaillou, 1998. Maryland
biological stream survey: a state agency program to assess the impact of anthropogenic stresses
on stream habitat quality and biota. Environmental Monitoring and Assessment, 51: 299-316.
Klein, R.D., 1979. Urbanization and stream quality impairment. Water Resources Bulletin,
15(4): 948-963.
Klemm, D.J. and J.M. Lazorchak, 1994. EMAP 1994 Pilot Laboratory Methods Manual for
Streams, EPA/620/R-94/003, U.S. Environmental Protection Agency, Washington, D.C.
67
-------
Kolpin, D.W., E.M. Thurman, and S.M. Linhard, 1998. The environmental occurrence of
herbicides: the importance of degradates in ground water. Archives of Environmental
Contamination and Toxicology, 35, 385-390.
Krug, A., 1993. Drainage history and land use pattern of a Swedish river system - their
importance for understanding nitrogen and phosphorus load. Hydrobiologia 251: 285-296.
Kurtz, J, B. V. Kurtz, T. Poretti, D. Miller, C. Bryson, C. Lawless, and J. Sell, 1996. Ambient
monitoring network: Atlantic coastal drainage basin. Water Monitoring Management, New
Jersey Department of Environmental Protection, Trenton, New Jersey, 15 p.
Larson, S.J., P.D. Capel, and M.S. Majewski, 1997. Pesticides in surface waters: distribution,
trends, and governing factors. Chelsea, Michigan, Ann Arbor Press, Inc., 373 p.
Lazorchak, J.M., A.T. Herlihy, H.R. Preston, and D.J. Klemm, 1998. Introduction, in Lazorchak,
J.M., D.J. Klemm, and D.V. Peck, (Eds.), Environmental Monitoring and Assessment Program-
surface waters: field operations and methods for measuring the ecological condition of wadeable
streams. EPA/620/R-94/004F, U.S. Environmental Protection Agency, Washington, D.C.
Lenat, D., 1993. A biotic index for the southeastern United States: derivation and list of
tolerance values, with criteria for assigning water quality ratings. J. N. Am. Benthol. Soc.l2(3):
279-290.
Lindsey, B.D., K.J. Breen, M.D. Bilger, and R.A. Brightbill, 1998. Water quality in the lower
Susquehanna river basin, Pennsylvania and Maryland, 1992-95. USGS Circular 1168.
Liu, Shiping, S.T. Yen, and D.W. Kolpin, 1996. Atrazine concentrations in near-surface aquifers
A censored regression approach. Journal of Environmental Quality, Vol. 25, p. 992-999.
Lowrance, R., R. Todd, J. Fail, Jr., O. Hendrickson, Jr., R. Leonard, and L. Asmussen, 1984.
Riparian forests as nutrient filters in agricultural watersheds. BioScience 34(6): 374-377.
Ludtke, A., and M. Woodworth, 1997. USGS blind sample project - Monitoring and evaluating
laboratory analytical quality. U.S. Geological Survey Fact Sheet FS-136-97, 2 p.
Luther, K.H., and H.M. Haitjema, 1998. Numerical experiments on the residence time
distributions of heterogeneous ground watersheds. Journal of Hydrology, 207, 1-17.
Madow, W.G., 1949. On the theory of systematic sampling II. Annals of Mathematical Statistics
20: 333-354.
Maxted, J.R., and E.L. Dickey, 1990. Invertebrate community of coastal stream habitats.
Technical Report 1-01, Delaware DNREC, Dover, DE.
May, C.W., R.R. Horner, J.R. Karr, B.W. Mar, E.B. Welch, 1997. Effects of urbanization on
small streams in the Puget Sound lowland ecoregion, Watershed Protection Techniques, Vol. 2,
68
-------
No. 4, 6/97,pp483-493.
Maryland Department of Agriculture, 1999, Maryland Pesticide Statistics for 1997 MDA 256-
99.
McFarland, E.R., 1995. Ground-Water flow, geochemistry, and effects of agricultural practices
on nitrogen transport at study sites in the Piedmont and coastal plain physiographic provinces,
Patuxent River Basin, Maryland. USGS Open File Report 94-507.
McMahon, G., and O.B. Lloyd, 1995. Water-quality assessment of the Albemarle-Pamlico
Drainage Basin, North Carolina and Virginia - Environmental setting and water-quality issues.
U.S. Geological Survey Open-File Report 95-136, 72 p.
Miller, T. W., 1994. Model selection in tree-structure regression. Proceedings of the ASA
Statistical Computing Section, American Statistical Association, p. 158-163.
Moore, D.M, B.G. Lees, and S.M. Davey, 1991. A new method to predicting vegetation
distributions using decision tree analysis in a geographic information system. Environmental
Management, 5(1)59-71.
Mullins, J.A., R.F. Carsel, J.E. Scarbough, and A.M. Ivery, 1993. PRZM-2, A model for
predicting pesticide fate in the crop root and unsaturated soil zones: Users manual for release 2.
U.S. EPA 600/R-93/046.
NASS, 1997, 1997 Census of Agriculture. AC97, available at http://www.nass.usda.gov/census/
NASS, 1999, Agricultural Chemical Usage 1998 Field Crops Summary, available at
http://usda.mannlib.cornell.edu/reports/nassr/other/pcu-bb/agch0599.txt
Novak, M.A., and R.W. Bode, 1992. Percent model affinity: a new measure of
macroinvertebrate community composition. J.N. Am. Benthol. Soc., 11(1): 80-85.
O'Connor, R.J., M.T. Jones, D. White, C. Hunsaker, T. Loveland, B. Jones, and E. Preston,
1996. Spatial partitioning of environmental correlates of avian biodiversity in the conterminous
United States. Biodiversity Letters, Vol. 3, p. 97-110.
Olsen, A.R., J. Sedransk, D. Edwards, C.A. Gotway, W. Liggett, S. Rathbun, K.H. Reckhow, and
L.J. Young, 1999. Statistical issues for monitoring ecological and natural resources in the United
States. Environmental Monitoring and Assessment, Vol. 54, p. 1-45.
Omeraik, J.M., A.R. Aberaathy, and L.M. Male, 1981. Stream nutrient levels and proximity of
agricultural and forest lands to streams: some relationships. J. Soil Water Conserv. 36: 227-331.
Omernik, J.M., 1995. Ecoregions of the conterminous United States. Annals of the Association
of American Geographers Vol. 77, p. 118-125.
69
-------
O'Neill, R.V., D.L. DeAngelis, J.B. Waide, and T.F.H. Allen, 1986. A hierarchical concept of
ecosystems. Princeton University Press, Princeton, NJ.
O'Neill, R.V., S.J. Turner, V.I. Cullinan, D.P. Coffin, T. Cook, W. Conley, J. Brunt, J.M.
Thomas, M.R. Conley, J. Gosz, 1991. Multiple landscape scales: an inter site comparison.
Landscape Ecology, 5(3): 137-144.
O'Neill, R.V., C.T. Hunsaker, S.P. Timmins, B.L. Jackson, K.B. Jones, K.H. Riitters, and J.D.
Wickham, 1996. Scale problems in reporting landscape pattern at the regional scale. Landscape
Ecology, 11(3): 169-180.
O'Neill, R.V., C.T. Hunsaker, K. B. Jones, K.H. Riitters, J.D. Wickham, P.M. Schwartz, LA.
Goodman, B.L. Jackson, and W.S. Baillargeon, 1997. Monitoring environmental quality at the
landscape scale, Bioscience 47(8): 513-519.
Onishi, Y., L. Shuyler, and Y. Cohen, 1990. Multimedia modeling of toxic substances, in
Proceedings of International Symposium on Water quality Modeling of Agricultural and Non-
Point Sources, Part II. DeCoursey, D.G., Ed., U.S. Department of Agriculture, ARS-81, pp 479-
509.
Osborne, L.L., and M.J. Wiley, 1988. Empirical relationships between land use/cover and stream
water quality in an agricultural watershed. Journal of Environmental Management, 26: 9-27.
Osborne, L.L., and D.A. Kovacic, 1993. Riparian vegetated buffer strips in water-quality
restoration and stream management. Freshwater Biology, 29, 243-258.
Paller, M.H., 1994. Relationships between fish assemblage structure and stream order in south
Carolina coastal plain streams. Transactions of the American Fisheries Society, Vol. 123, p. 150-
161.
Phillips, P.J., and L.J. Bachman, 1996. Hydrologic landscapes on the Delmarva Peninsula-Part
1: Drainage basin type and base-flow chemistry. Journal of the American Water Resources
Association, Vol. 32, p. 767-778.
Phillips, P.J., G.R. Wall, E.M. Thurman, D.A. Eckhardt, J. vanHoesen, 1999a. Metolachlor and
its metabolites in tile drain and stream runoff in the Canajoharie Creek watershed.
Environmental Science and Technology, 33(20), p. 3531-3537.
Phillips, P.J., D.A. Eckhardt, E.M. Thurman, and S.A. Terraciano, 1999b. Ratios of metolachlor
metabolites in ground water, tile drainage discharge, and surface water in selected areas of New
York State. Morganwalp, D.W., and H.T. Buxton, eds, U.S. Geological Survey, Water
Resources Investigations Report 99-4018B, p.383-394.
Pirkey, S.R., and K.D. Glodt, 1998. Quality control at the U.S. Geological Survey National
Water Quality Laboratory. U.S. Geological Survey Fact Sheet FS-026-98, 4 p.
70
-------
Plafkin, J.L., M.T. Harbour, K.D. Porter, S.K. Gross, and R.M. Hughes, 1989. Rapid assessment
protocols for use in streams and rivers: benthic macroinvertebrates and fish. EPA/444/4-89-001,
U.S. Environmental Protection Agency, Washington, D.C.
Preston, S. D., and J. W. Brakebill, 1999. Application of spatially referenced regression
modeling for the evaluation of total nitrogen loading in the Chesapeake Bay watershed. U. S.
Geological Survey Water-Resources Investigations Report, 99-4054,12 p.
Quinn, J.M., and C.W. Hickey, 1990. Characterisation and classification of benthic invertebrate
communities in 88 New Zealand rivers in relation to environmental factors. New Zealand
Journal of Marine and Freshwater Research, 24: 387-409.
Rathert, D., D. White, J.C. Sifneos, and R.M. Hughes, 1999. Environmental associations of
species richness in Oregon freshwater fishes. Journal of Biogeography, 26(2): 257-273.
Richards, C., and G.W. Minshall, 1992. Spatial and temporal trends in stream macroinvertebrate
species assemblages: the influence of catchment disturbance. Hydrobiologia, 241: 173-184.
Richards, C., G.E. Host, and J.W. Arthur, 1993. Identification of predominant environmental
factors structuring stream macroinvertebrate communities within a large agricultural catchment.
Freshwater Biology, 29: 285-294.
Riitters, K.H., B.E. Law, R.C. Kucera, A.L. Gallant, R.L. DeVelice, and CJ. Palmer, 1992. A
selection of forest condition indicators for monitoring. Environmental Monitoring and
Assessment, 20: 21-33.
Riitters, K.H., R.V. O'Neill, C.T. Hunsaker, J.D. Wickham, D.H. Yankee, S.P. Timmins, K.B.
Jones, B.L. Jackson, 1995. A factor analysis of landscape pattern and structure metrics.
Landscape Ecology, 10: 23-39.
Richards, C., L.B. Johnson, G.E. Host, 1996. Landscape-scale influences on stream habitats
and biota, Canadian Journal of Fisheries and Aquatic Sciences 53(Suppl. 1): p. 295-311.
Schulz, R., and M. Leiss, 1999. A field study of the effects of agriculturally derived insecticide
input on stream macroinvertebrate dynamics. Aquatic Toxicology, 46,155-176.
Shedlock, R.J., P.A. Hamilton, J.M. Denver, PJ. Phillips, 1993. Multiscale approach to regional
ground-water-quality assessment of the Delmarva Penninsula. Chapter 23 in Regional Ground-
Water Quality, W.M. Alley, Ed., Van Nostrand Reinhold, New York.
Shedlock, R.J., J.M. Denver, M.H. Hayes, P.A. Hamilton, M.T. Koterba, L.J. Bachman, P.J.
Phillips, and W.S.L. Banks, 1999. Ground-water assessment of the Delmarva Peninsula,
Delaware, Maryland, and Virginia: Results of investigations, 1987-1991. Water-Supply Paper
2355-A, 42 p.
Shelton, L.R., 1994. Field guide for collecting and processing stream-water samples for the
71
-------
National Water-Quality Assessment program. U.S. Geological Survey Open-File Report 94-455,
44 p.
Sifneos, J. C., D. White, N.S. Urquhart, D. Schafer, (in preparation). A Comparison of Pruning
Methods for Regression Trees: Evidence From Simulation and Published Studies.
Smith, R.A., G.E. Schwarz, and R.B. Alexander, 1997. Regional interpretation of water-quality
monitoring data. Water Resources Research, Vol. 33, No. 12, p. 2781-2798.
Speiran, O.K., 1996. Geohydrology and geochemistry near coastal ground-water-discharge areas
of the Eastern Shore, Virginia. U.S. Geological Survey Water-Supply Paper 2479, 73 p.
Stackelburg, P.E., and M.A. Ayers, 1994. National Water-Quality Assessment Program—Long
Island-New Jersey coastal drainages. U.S. Geological Survey NAWQA Fact Sheet FS 94-012.
Staver, K.W., and R.B. Brinsfield, 1996. Seepage of ground water nitrate from a riparian
agroecosystem into the Wye River Estuary. Estuaries, Vol. 19, p. 359-370.
Stevens, D.L. Jr., 1997. Variable density grid-based sampling designs for continuous spatial
populations. Environmetrics, 8: 167-195.
Stevens, D.L., and A.R. Olsen, 1999. Spatially restricted surveys over time for aquatic resources.
Journal of Agricultural, Biological, and Environmental Statistics, Vol. 4, No. 4, p. 415-428.
Stribling, J.B., B.K. Jessup, J.S. White, D. Boward, M. Hurd, 1998. Development of a benthic
index of biotic integrity for Maryland streams. Maryland Department of Natural Resources,
Annapolis, MD, CBWP-EA-98-3.
Sweeney, B.W., 1993. Effects of stream side vegetation on macroinvertebrate communities of
White Clay Creek in eastern North America. Proceedings of The Academy of Natural Sciences
of Philadelphia, 144: 291-340.
Tate, C.M., and J.S. Heiny, 1995. The ordination of benthic invertebrate communities in the
South Platte River Basin in relation to environmental factors. Freshwater Biology, 33: 439-454.
Tesoriero, A.J., and F.D. Voss 1997. Predicting the probability of elevated nitrate concentrations
in the Puget Sound Basin - Implications for aquifer susceptibility and vulnerability. Ground
Water, vol. 35, no. 6., pp 1029-1039.
Trapp, H. Jr., and M.A. Horn, 1997. Ground Water Atlas of the United States, Segment 11,
Hydrologic Investigations Atlas 730-L, U.S. Geological Survey, Reston, Virginia.
Tufford, D.L., H.N. McKellar, and J.R. Hussey, 1998. In-stream non-point source nutrient
prediction with land-use proximity and seasonality. J. Environ. Qual., 27:100-111.
U.S. Environmental Protection Agency, 1994a. National primary drinking-water standards. U.S.
72
-------
Environmental Protection Agency Fact Sheet 810-F-94-001,2 p.
U.S. Environmental Protection Agency, 1994b. Landscape monitoring and assessment research
plan. EPA 620/R-94/009.
U.S. Environmental Protection Agency, 1994c. EPA Reach File Version 3.0 Alpha release (RF3-
Alpha) technical reference. Office of Wetlands, Oceans, and Watersheds, Office of Water.
U.S. Environmental Protection Agency, 1997. Field and laboratory methods for macro-
invertebrate and habitat assessment of low gradient, nontidal streams. Mid-Atlantic Coastal
Streams Workgroup, Environmental Services Division, Region 3, Wheeling, WV; 23 pages with
appendices.
U.S. Environmental Protection Agency, 1998. Guidelines for Ecological Risk Assessment.
EPA/630/R-95/002F.
U.S. Federal Emergency Management Agency, 1999, Declared Counties Map, Remote Sensing
Damage, DR 1292, North Carolina, http://www.gismaps.fema.gov/1999pages/drl292.htm
U.S. Geological Survey, 1999. The quality of our nation's waters-nutrients and pesticides. U.S.
Geological Survey Circular 1225, 82 p.
Vogelmann, I.E., T.L. Sohl, P.V. Campbell, and D.M. Shaw, 1998. Regional land cover
characterization using Landsat and ancillary sources. Environmental Monitoring and
Assessment, 51: 415-428.
Walker, P. A., 1990. Modeling wildlife distributions using a geographic information system:
kangaroos in relation to climate. Journal of Biogeography, Vol. 17, p. 279-289.
Wang, L., J. Lyons, P. Kanehl, R. Gatti, 1997. Influences of watershed land use on habitat
quality and biotic integrity in Wisconsin streams, Fisheries, Vol. 22, No. 6, 6/97, p. 6-12.
Wesselman, R. and R. Carr, 2000, personal, comm.
Wiederholm, T., 1984. Responses of aquatic insects to environmental pollution, in The Ecology
of Aquatic Insects, Praeger, New York, p. 508-557.
Williams, D.D., and H.B.N. Hynes, 1976. The recolonization mechanisms of stream benthos.
Oikos 27, 265-272.
Winner, M.D., and R.W. Coble, 1996. Hydrogeologic framework of the North Carolina coastal
plain. U.S. Geological Survey Professional Paper 1401-1,106 p.
Zappia, H., and G.T. Fisher, 1997. Water quality assessment of the Potomac River Basin:
Analysis of Available Data 1972-90. USGS, WRIR 97-4051.
73
-------
Zaugg, S.D., M.W. Sandstrom, S.G. Smith, and K.M. Fehlberg, 1995. Methods of Analysis by
the U.S. Geological Survey National Water Quality Laboratory-Determination of Pesticides in
Water by C-18 Solid-Phase Extraction and Capillary-Column Gas Chromatography/Mass
Spectrometry with Selected-Ion Monitoring. U.S. Geological Survey Open-File Report 95-181.
Zheng, C., 1992. MT3D version 1.8, Documentation, and User Guide. S.S. Papadopulos and
Associates, Bethesda, MD.
74
-------
APPENDIX A
U.S. Geological Survey Pesticide Schedule 2001
The following information is excerpted from Zaugg et al., 1995. This method is suitable
for the determination of low-level concentrations (in micrograms per liter and nanograms per
liter) of pesticides and pesticide metabolites in natural-water samples. The method is applicable
to pesticides and metabolites that are (1) efficiently partitioned from the water phase into an
octadecyl (C-18) organic phase that is chemically bonded to a solid inorganic matrix, and (2)
sufficiently volatile and thermally stable for gas chromatography. Suspended paniculate matter
is removed from the samples by filtration, so this method is suitable only for dissolved-phase
pesticides and metabolites... The method was developed in response to the request for a broad .
spectrum pesticide method for use in determining their occurrence and distribution as monitored
by the NA WQA program. Pesticides were selected initially because of their widespread use in
the United States, according to information in ...Gianessi and Puffer, 1990, 1992a, and 1992b,
and their compatibility with the general analytical plan. Other criteria included published
studies of pesticide fate and occurrence of metabolites, responses from NAWQA Study Unit
personnel regarding pesticides of local significance, and U.S. EPA health advisories. Finally
restrictions in the analytical software on the number of ions scanned for specific time intervals
limited the number of pesticides chosen to about 50.
The calibration range is equivalent to concentrations from 0.001 to 4.0 ug/L for most
pesticides. Widely and abundantly used corn herbicides-atrazine, metolachlor, cyanazine, and
alachlor-have upper concentration limits of 20 ug/L Method detection limit (MDL) is defined
as the minimum concentration of a substance that can be identified, measured, and reported with
99-percent confidence that the compound concentration is greater than zero. The MDL is
compound dependent and dependent on sample matrix and instrument performance and other
operational sources of variation. For the listed pesticides, MDLs vary from 0.001 to 0.018 ug/L.
Analytical results are not censored at the MDL; if a pesticide meets the detection criteria
(retention time and mass spectra compared to that of a reference standard) the result is
calculated and reported.
Summary of method: The samples are filtered at the collection site using glass-fiber
filters with 0.7-umpore diameter to remove suspendedparticulate matter...Filtered water
samples are pumped through disposable polypropylene SPE columns containing porous silica
coated with an octadecyl (C-18) phase that is chemically bonded to the surface of the silica. The
SPE columns are dried using a gentle stream of carbon dioxide or nitrogen to remove residual
water. The adsorbed pesticides and metabolites then are removed from the SPE columns by
elution with hexane-isopropanol (3:1). The eluant is further evaporated using a gentle stream of
nitrogen. Extracts of the eluant are analyzed by a capillary-column GC/MS operated in the SIM
mode.
75
-------
Table Al. USGS Pesticide Schedule 2001 (complete list) and the Estimated Amount of Active
Ingredient Applied in the Mid-Atlantic (Gianessi and Puffer, 1990 and 1992 a,b).
USGS Pesticide
Schedule 2001
2,6,-Diethylaniline
Acetochlor
Alachlor
Atrazine
Azinphos-methyl
Benfluralin
Butylate
Carbaryl
Carbofuran
Chlorpyrifos
Cyanazine
Dacthal
Deethylatrazine
Diazinon
Diazinon-dlO (sur.)
Dieldrin
Disulfoton
EPTC
Ethalfluralin
Ethoprofos
Fonophos
Lindane
Linuron
Malathion
Metolachlor
Metribuzin
Detected in Ground Water1
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Detected in Surface
Water2
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Estimated Active
Ingredient Applied in
Mid-Atlantic Region
(Ibs/yr) 3
3,630,000
4,900,000
1,260,000
453,000
992,000
2,340,000
1,090,000
1,070,000
485,000
4,270,000
76
-------
USGS Pesticide
Schedule 2001
Molinate
Napropamide
Parathion
Parathion-methyl
Pebulate
Pendimethalin
Phorate
Prometon
Propachlor
Propanil
Propargite
Proyzamide
Simazine
Tebuthiuron
Terbacil
Terbufos
Terbuthylazine (sur.)
Thiobencarb
Tri-allate
Trifluralin
alpha-HCH
alpha-HCH-d6 (sur.)
cis-Permithrin
p,p'-DDE
Detected in Ground Water1
•
•
•
•
•
•
•
•
•
Detected in Surface
Water2
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Estimated Active
Ingredient Applied
cont. (Ibs/yr) 3
975,000
730,000
'Ator, S.W. and Ferrari, M.J., 1997.
2Ferrari, M.J., et al., 1997.
3Gianessi and Puffer, 1990,1992 a,b.
77
-------
APPENDIX B
Target Analytes for Sediment Analyses
Table Bl. Analytes for Sediments
Analyte (CAS Number)
Pesticides
Aldrin (309-00-2)
Chlordane-cis (5103-71-9)
Chlordane-trans (5103-74-2)
2,4'-DDD (53-19-0)
2,4'-DDD (72-54-8)
2,4'-DDE (3424-82-6)
4,4'-DDE (72-55-9)
2,4'-DDT (789-02-6)
4,4'-DDT (50-29-3)
Dieldrin (60-57-1)
Endosulfan 1 (959-98-8)
Endosulfan 11 (33213-65-9)
Endrin (72-20-8)
Heptachlor (76-44-8)
Heptachlor Epoxide (1024-57-3)
Hexachlorobenzene (118-74-1)
Hexachlorocyclohexane [Gamma-BHC/Lindane] (58-89-9)
Mirex (2385-85-5)
trans-Nonachlor (3765-80-5)
cis-Nonachlor (5103-73-1)
Oxychlordane (27304-13-8)
Poly Chlorinated Biphenyl (PCB) Congeners
2,4-Dichlorobiphenyl, #8 (34883-43-7)
2,2',5-Trichlorobiphenyl, #18 (37680-65-2)
2,4,4'-Trichlorobiphenyl, #28 (7012-37-5)
2,2',5,5'-Tetrachloroblphenyl, #52 (35693-99-3)
2,2',3,5'-Tetrachloroblphenyl, #44 (41464-39-5)
2,3',4,4'-Tetrachloroblphenyl, #66 (32598-10-0)
2,2',4,5,5'-Pentachlorobephenyl, #101 (37680-73-2)
2,3',4,4',5'-Pentachlorobephenyl, #118 (31508-00-6)
2,2',4,4',5,5'-Hexachlorobiphenyl, #153 (35065-27-1)
2,3,3',4,4'-Pentachlorobiphenyl, #105 (32598-14-4)
2,2',3,4,4',5-Hexachlorobiphenyl, #138 (35065-28-2)
2,2',3,4',5,5',6-Heptachlorobiphenyl, #187 (52663-68-0)
2,21,3,3',4,4'-Hexachlorobiphenyl, #128 (38380-07-3)
78
-------
APPENDIX B
Target Analytes for Sediment Samples (continued)
2,2l,3,4,4l,5,5l-Heptachlorobiphenyl, #180 (35065-29-3)
2,2',3,3',4,4',5-Heptachlorobiphenyl, #170 (35065-30-6)
2,2',3,3',4,4',5,6-Octachlorobiphenyl, #195 (52663-78-2)
2,2',3,3',4,4',5,5',6-Nonschlorobiphenyl, #206 (40186-72-9)
Decachlorobiphenyl, #209 (2051-24-3)
3,3',4,4' Tetrachlorobiphenyl, #77* (32598-13-3)
3,3',4,4',5 Pentachlorobiphenyl, #126*
3,3',4,4',5,5' Hexachlorobiphenyl, #169* (32775-16-6)
Metals
Arsenic (7440-38-2)
Mercury (7439-97-6)
Additional Measurements
Percent Moisture
Size distribution
Organic matter content
79
-------
APPENDIX C
U.S. Geological Survey Schedule 2701
Major Ions
Table Cl. U.S. Geological Survey Major Ions Schedule 2701
Calcium, dissolved (mg/L as Ca)
Silica, dissolved (mg/L as SiO2)
Chloride, dissolved (mg/L as Cl)
Iron, dissolved (mg/L as Fe)
pH, wH, laboratory, standard units
Sodium, dissolved (mg/L as Na)
Potassium, dissolved (mg/L as K)
Manganese, dissolved (ug/L as Mn)
Specific conductance (microsiemens/cm)
Magnesium, dissolved (mg/L as Mg)
Sulfate, dissolved (mg/L as SO4)
Residue, dissolved 180c (mg/L)
Fluoride, dissolved (mg/L as F)
APPENDIX D
U.S. Geological Survey Schedule 2702
Nutrients
Table Dl. U.S. Geological Survey Nutrients Schedule 2702
Phosphorus, dissolved (mg/L as P)
Nitrogen (ammonia + organic) (mg/L as N)
Phosphorus, total (mg/L as P)
Phosphorus, ortho (mg/L as P)
Nitrogen (amn & organic) (mg/L as N)
Nitrogen, nitrite (mg/L as N)
Nitrogen, ammonia (mg/L as N)
NO2 + NO3, dissolved (mg/L as N)
80
-------
APPENDIX E
Benthic Macroinvertebrate Indices
Table El. List of Benthic Macroinvertebrate Indices (primarily from Bode et al., 1996)
COMMUNITY
INDICES
DESCRIPTION
Species richness (taxa
number)
This is the total number of species or taxa found in the sample. High species
richness values are associated with clean-water conditions.
EPT richness
EPT denotes the total number of species of mayflies (Ephemeroptera), stoneflies
(Plecoptera), and caddisflies (Trichoptera) found in a 100-organism subsample.
These are considered to be mostly clean-water organisms, and their presence
generally is correlated with good water quality.
Hilsenhoff Biotic index
The Hilsenhoff Biotic Index (HBI) is calculated by multiplying the number of
individuals of each species by its assigned tolerance value, summing these products,
and dividing by the total number of individuals. On a 0-10 scale, tolerance values
range from intolerant (0) to tolerant (10). Tolerance values are listed in a species list
for Wisconsin, developed by Hilsenhoff (1987). High HBI values are indicative of
organic (sewage) pollution, while low values are indicative of clean-water
conditions.
Percent model affinity
This is a measure of similarity to a model, nonimpacted community based on percent
abundance hi seven major groups (Novak and Bode, 1992). Percentage similarity as
calculated in Washington (1984) is used to measure similarity to idealized kick
sample or Ponar sample communities.
Species diversity
Species diversity is a value that combines species richness and community balance
(evenness). Shannon-Wiener diversity values are calculated using the formula in
Weber (1973). High species diversity values usually indicate diverse, well-balanced
communities, while low values indicate stress or impact.
Dominance
Dominance is a simple measure of community balance or evenness of the distribution
of individuals among the species. Simple dominance is the percent contribution of
the most numerous species. Dominance-3 is the combined percent contribution of
the three most numerous species. High dominance values indicate unbalanced
communities strongly dominated by one or more very numerous species.
NCBI
The North Carolina Biotic Index (NCBI) (Lenat, 1993) is similar to the HBI, with
tolerance values developed for North Carolina stream invertebrates and seasonal
factors to correct data to mean summer values. Stream size does not have a large
effect on NCBI.
Maryland B-IBI
The Benthic Index of Biotic Integrity for Maryland coastal plain streams (Stribling et
al., 1998) includes taxa number, EPT taxa number, percent Ephemeroptera, percent
Chironomidae that are Tanytarsini, percent clinger taxa, percent scrapers, and Beck's
Biotic Index.
10-Metric B-IBI
This B-IBI was proposed by Karr and Chu (1999) based on examples for the
Tennessee Valley, Puget Sound, southwestern Oregon, north central Oregon,
northwestern Wyoming, and Japan, and was designed to detect human impact.
81
-------
APPENDIX F
Physical Habitat Metrics
Table Fl. Calculated Reach-Level Physical Habitat Metrics (after Kaufman et al., 1999)
Depth: mean and standard deviation (SD)
Wetted width: mean and standard deviation (SD)
Width:Depth ratio
Width-Depth product
Habitat class: percent of reach in each class
Reach aggregate and individual residual pool metrics
Reach slope: mean and SD
Reach sinuosity from backsighted bearings
Substrate size: percentage by class, mean and SD of size; median, lower and upper quartiles (Q,
Q3), and interquartile range of size class; Log10 of geometric mean diameter
Bankfull width: mean and SD
Bankfull height: mean and SD
Incision height: mean and SD
Bank angle: mean, SD, Q[ and Q3, and interquartile range
Undercut distance: mean, SD, Q! and Q3, and interquartile range
Large woody debris size classes: counts and volumes
Canopy densiometer values (mid-channel): mean and SD
Canopy densiometer values (bank): mean and SD
Riparian vegetation cover metrics
Riparian vegetation presence metrics
Riparian vegetation type: proportion of reach with each type
Presence of human influences: proximity weighted
82
-------
APPENDIX G
Spatial Databases
Table Gl. Spatial Databases
Coverage
Stream Hydrography
Hydrogeologic Framework
Digital elevation
STATSGO soil data
SSURGO soil data
Soils 5 soil data
MRLC Land use/land cover
Roads, Railroads, Pipelines
County and State boundaries
Population (census blocks)
Agrochemical Application Data; and
confined animal feeding operations
(CAFOs)
Agrochemical Application Data (1989)
NEXRAD precipitation
Scale/ Resolution
1:100,000
1:1,000,000
30 m, 90 m
1:1,000,000
1:100,000
point locations
30m
1:100,000
1:100,000
1:24,000
County and possibly zip code polygons
for agrochemicals; exact coordinates of
CAFOs
County scale
4 km grid, 24-hour cumulative total
Source
U.S. EPA, Office of Water, RF3 files
U.S. Geological Survey, Water
Resources Division, Maryland-
Delaware-District of Columbia District
Office
U.S. Geological Survey, EROS Data
Center
U.S. Department of Agriculture, NRCS
U.S. Department of Agriculture, NRCS
U.S. Department of Agriculture, NRCS
U.S. Geological Survey, EROS Data
Center
U.S. Geological Survey, Digital Line
Graph files
U.S. EPA, NERL-RTP
U.S. Bureau of Census
U.S. EPA, ORD, NERL; NASS;
individual states
U.S. Geological Survey (Battaglin and
Goolsby, 1994)
National Oceanic and Atmospheric
Administration
A wide variety of spatial databases will be acquired for this study (Table Gl). These
coverages will be acquired, edge-matched as needed and "clipped" to the Coastal Plain boundary.
The most critical of the spatial coverages is the stream hydrography data because the selection of
watersheds for sampling is based on the overall delineation of streams in the study area. We plan
to use Reach File 3 (RF3) data at a scale of 1:100,000 (U.S. EPA 1994; Horn and Grayman,
1993) as our basis for identifying streams to be sampled. The RF3 data provide stream segment
locations at the best available resolution, covering the entire study area at a consistent scale. It
has codes which designate the segments according to type of segment which is useful in the
selection process. Soil particle size distributions, permeability, and depth to water table are
available from the National Resource Conservation Service's STATe Soil GeOgraphic database
(STATSGO, available for the entire area); and from their higher resolution Soil SURvey
GeOgraphic database (SSURGO, available for approximately half the area). These data will be
used in both the landscape indicator analysis and the hydrologic modeling. The land use/land
cover data are from the Multi-Resolution Landscape Characteristics (MRLC) Consortium. These
83
-------
data were derived from 30-meter resolution Landsat Thematic Mapper (TM) data, and classified
into 15 land use/land cover classes (Vogelmann et al., 1998). Agrochemical coverages include
both fertilizer and pesticide application data and will be compiled from several sources. One
source is from existing published data, such as the 1997 Census of Agriculture. These data are
compiled as pounds of active ingredient applied by county (for certain states only). Total acres in
which pesticides were applied are also available by county. These data will be entered onto a
spreadsheet, identified by Federal Information Processing Standard (FIPS) county code, and
attributed to the agricultural portion of the county polygon coverage. Another source of these
data at a finer resolution is by special request to the state offices of the National Agricultural
Statistical Survey (NASS). Through arrangements with their staff, the application rate data are
assembled according to zip code (confidentiality is preserved by not reporting data for zip codes
containing only one or two farms). We are currently having discussions with NASS regarding
the application of this approach for the Coastal Plain states. We are also considering the
possibility of a direct request for application rate information for selected locations, to the
American Crop Producers Association. Gianessi and Puffer, 1990, 1992a, and 1992b have
identified the twenty most commonly used pesticides in the Mid-Atlantic. We will compile
physical and biological data for these chemicals from readily available sources as needed for the
modeling effort; these data include type of compound (triazine; carbamate, etc.), physical
constants (solubility in water, Henry's Law constant, etc.), type of crops treated; application
method, wildlife lethal dose information, and toxicological information. Additional data for use
in the hydrologic models will also be compiled as needed. Once the data sets are complete, they
will be assembled onto a compact for distribution among the participants. The data will also be
available for viewing at an EPA intranet website. It is not our intention to become distributors
for these data except in their role as part of the final products of the study.
84
------- |