DRAFT
DO NOT CITE OR QUOTE
&EPA
EPA/600/R-10/023A
March 2010
External Review Draft
A Field-based Aquatic Life Benchmark for
Conductivity in Central Appalachian Streams
NOTICE
This information is distributed solely for the purpose of predissemination peer review under
applicable information quality guidelines. It has not been formally disseminated by the U.S.
EPA. It does not represent and should not be construed to represent any Agency determination
or policy.
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20460
-------
DISCLAIMER
This information is distributed solely for the purpose of predissemination peer review
under applicable information quality guidelines. It has not been formally disseminated by the
U.S. EPA. It does not represent and should not be construed to represent any Agency
determination or policy.
Preferred Citation:
U.S. EPA (Environmental Protection Agency). 2010. A Field-based Aquatic Life Benchmark for Conductivity in
Central Appalachian Streams. Office of Research and Development, National Center for Environmental
Assessment, Washington, DC. EPA/600/R-10/023A.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 ii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS
LIST OF TABLES v
LIST OF FIGURES vi
LIST OF ABBREVIATIONS AND ACRONYMS vii
AUTHORS, CONTRIBUTORS, AND REVIEWERS viii
ACKNOWLEDGMENTS xi
EXECUTIVE SUMMARY xii
1. INTRODUCTION 1
1.1. CONDUCTIVITY 1
1.2. APPROACH 2
2. DATA SETS 5
2.1. DATA SET SELECTION 5
2.2. DATA SOURCES 5
2.3. DATA SET CHARACTERISTICS 6
3. METHODS 8
3.1. EXTIRPATION CONCENTRATION DERIVATION 8
3.2. TREATMENT OF POTENTIAL CONFOUNDERS 10
3 .3. DEVELOPING THE SPECIES SENSITIVITY DISTRIBUTION 11
3.4. CONFIDENCE BOUNDS 11
3.5. ESTIMATING BACKGROUND 12
4. RESULTS 13
4.1. EXTIRPATION CONCENTRATIONS 13
4.2. SPECIES SENSITIVITY DISTRIBUTIONS 13
4.3. HAZARDOUS CONCENTRATION VALUES AT THE 5th PERCENTILE 13
4.4. UNCERTAINTY ANALYSIS 13
5. CONSIDERATIONS 15
5.1. SELECTION OF INVERTEBRATE GENERA 15
5.2. SEASONALITY, LIFE HISTORY, AND SAMPLING METHODS 15
5.3. INCLUSION OF REFERENCE SITES 16
5.4. DEFINING THE REGION OF APPLICABILITY 16
5.5. BACKGROUND 17
5.6. INCLUSION OF OTHER TAX A 17
5.7. TREATMENT OF RARE SPECIES 17
5.8. SELECTION OF THE EFFECTS ENDPOINT 18
5.9. USE OF MODELED OR EMPIRICAL DISTRIBUTIONS 18
5.10. TREATMENT OF CAUSATION 19
5.11. TREATMENT OF MIXTURES 19
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 iii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
6. AQUATIC LIFE BENCHMARK 20
REFERENCES 21
APPENDIX A: CAUSAL ASSESSMENT 41
APPENDIX B: CONFOUNDING 66
APPENDIX C: EXTIRPATION CONCENTRATION VALUES FOR
INVERTEBRATES 94
APPENDIX D: GRAPHS OF OBSERVATION PROBABILITIES AND
CUMULATIVE DISTRIBUTION FUNCTIONS FOR EACH GENUS 101
APPENDIX E: VALIDATION OF METHOD USING FIELD DATA TO DERIVE
AMBIENT WATER QUALITY BENCHMARK FOR
CONDUCTIVITY USING KENTUCKY DATA SET 157
APPENDIX F: DATA SOURCES AND METHODS OF LANDUSE/LAND COVER
ANALYSIS USED TO DEVELOP EVIDENCE OF SOURCES OF
HIGH CONDUCTIVITY WATER 170
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 iv DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES
1. Summary statistics of the measured water quality parameters 24
2. Number of samples with reported genera and conductivity meeting our acceptance
criteria for calculating the benchmark value 26
3. Genera excluded from 95th percentile extirpation concentration calculation
because they never occurred at reference sites 26
4. Hazardous concentration at the 5th percentile for invertebrates in Ecoregions 69
and 70 26
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 v DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
1. Data are from Tier III Ecoregions 69 and 70 spanning the states of Ohio,
Pennsylvania, Kentucky, Tennessee, West Virginia, and Maryland 27
2. Box plot showing seasonal variation of conductivity in the reference streams of
Ecoregions 69 and 70 in West Virginia from 1999 to 2006 28
3. Histogram of the frequencies of observed conductivity values in samples from
Ecoregions 69 and 70 from March to October 28
4. Example of a weighted CDF and the associated 95th percentile extirpation
concentration value 29
5. Three typical distributions of observation probabilities 30
6. The species sensitivity distribution for all year 31
7. The cumulative distribution of XC95 values for the 35 most sensitive genera and
the bootstrap-derived means and two-tailed 95% confidence intervals 32
8. Species sensitivity distribution for all year 33
9. Examples of a monthly year-long stream conductivity record in a stream 34
10. Correlation of conductivity values sampled from the same site in spring and
summer 35
11a. Anions 36
lib. Cations 37
11c. Dissolved metals 38
lid. Total metals 39
lie. Other water quality parameters 40
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 vi DRAFT—DO NOT CITE OR QUOTE
-------
CDF
DCX
GLIMPSS
HCX
KDOW
RBP
SSD
TMDL
U.S. EPA
WABbase
WVDEP
WVSCI
XCx
LIST OF ABBREVIATIONS AND ACRONYMS
cumulative distribution function
depletion concentration
genus level index of most probable stream status
hazardous concentration
Kentucky Division of Water
rapid bioassessment protocol
species sensitivity distribution
total maximum daily load
United States Environmental Protection Agency
Watershed Assessment Branch Data Base
West Virginia Department of Environmental Protection
West Virginia Stream Condition Index
extirpation concentration
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 vii DRAFT—DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
AUTHORS
Susan M. Cormier, Ph.D.
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Cincinnati, OH 45268
Glenn W. Suter II, Ph.D.
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Cincinnati, OH 45268
Lester L. Yuan, Ph.D.
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Washington, DC 20460
Lei Zheng, Ph.D.
Tetra Tech, Inc.
Owings Mills, MD 21117
CONTRIBUTORS
R. Hunter Anderson, Ph.D.
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Cincinnati, OH 45268
Jennifer Flippin, M.S.
Tetra Tech, Inc.
Owings Mills, MD 21117
Jeroen Gerritsen, Ph.D.
Tetra Tech, Inc.
Owings Mills, MD 21117
Michael Griffith, Ph.D.
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Cincinnati, OH 45268
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 viii DRAFT—DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
CONTRIBUTORS (continued)
Michael McManus, Ph.D.
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Cincinnati, OH 45268
John Paul, Ph.D.
U.S. Environmental Protection Agency
National Health and Environmental Effects Research Laboratory
Research Triangle Park, NC 27711
Samuel P. Wilkes, M.S.
Tetra Tech, Inc.
Charleston, WV 25301
REVIEWERS
Margaret Passmore, M.S.
U.S. Environmental Protection Agency
Office of Monitoring and Assessment, Freshwater Biology Team
Wheeling, WV 26003
Charles Delos, M.S.
U.S. Environmental Protection Agency
Office of Water, Health and Ecological Criteria Division
Washington, DC 20460
John Van Sickle, Ph.D.
U.S. Environmental Protection Agency
National Health and Environmental Effects Research Laboratory, Western Ecology Division
Corvallis, OR 97333
Chuck Hawkins, Ph.D.
Western Center for Monitoring and Assessment of Freshwater Ecosystems
Department of Watershed Sciences
Utah State University
Logan, UT 84322
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 ix DRAFT—DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
REVIEWERS (continued)
Christopher C. Ingersoll, Ph.D.
U.S. Geological Survey
Columbia Environmental Research Center
4200 New Haven Road
Columbia, MO 65201
Charles A. Menzie, Ph.D.
Exponent
2 West Lane
Severna Park, Ml) 21146
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 x DRAFT—DO NOT CITE OR QUOTE
-------
ACKNOWLEDGMENTS
Greg Pond, Susan B. Norton, Teresa Norberg-King, Peter Husby, Peg Pelletier, Treda
Grayson, Amy Bergdale, Candace Bauer, Brooke Todd, Lana Wood, Heidi Glick, Cris Broyles,
Linda Tackett, Stacey Lewis, Bette Zwayer, and Ruth Durham helped bring this document to
completion by providing comments, essential fact checking, editing, formatting, and other key
activities. Statistical review of the methodology was provided by Paul White, John Fox, and
Leonid Kopylev.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 xi DRAFT—DO NOT CITE OR QUOTE
-------
EXECUTIVE SUMMARY
This report uses field data to derive an aquatic life benchmark for conductivity that may
be applied to waters in the Appalachian Region that are dominated by salts of SO42 and HCO3
at circum-neutral to mildly alkaline pH. This benchmark is intended to protect the aquatic life in
the region. It is derived by a method modeled on the U.S. EPA's standard methodology for
deriving water quality criteria. In particular, the methodology was adapted for use of field data.
Field data were used because sufficient and appropriate laboratory data were not available and
because high quality field data were available to relate conductivity to effects on aquatic life.
This report provides scientific evidence for a conductivity benchmark in a specific region rather
than for the entire United States.
The method used in this report is based on the standard methodology in that it used the
5th percentile of a species sensitivity distribution (SSD) as the benchmark value. SSDs represent
the response of aquatic life as a distribution with respect to exposure. It is implicitly assumed
that if the exposure level is kept below the 5th percentile of the SSD, at least 95% of species will
be protected. Data analysis followed the standard methodology in aggregating species to genera
and using interpolation to estimate the percentile. It differs primarily in that the points in the
SSDs are extirpation concentrations (XCs) rather than median lethal concentrations (LC50s) or
chronic values. The XC is the level of exposure above which a genus is effectively absent from
water bodies in a region. For this benchmark value, the 95th percentile of the distribution of the
probability of occurrence of a genus with respect to conductivity was used as a 95th percentile
extirpation concentration. Hence, this aquatic life benchmark for conductivity is expected to
avoid the local extirpation of 95% of native species (based on the 5th percentile of the SSD) due
to neutral to alkaline effluents containing a mixture of dissolved ions dominated by salts of
SO42 and HCO3 . Because it is not protective of all genera and protects against extirpation
rather than reduction in abundance, this level is not fully protective of rare species or waters
designated by state and federal agencies as exceptional.
This field-based method has several advantages. Because it is based on biological
surveys, it is inherently relevant to the streams where the benchmark may be applied and
represents the actual aquatic life use in these streams. Another advantage is that the method
assesses all life stages and ecological interactions of many species. Further, it represents the
actual exposure conditions for elevated conductivity in the region, the actual temporal variation
in exposure, and the actual mixture of ions that contribute to salinity as measured by
conductivity.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 xii DRAFT—DO NOT CITE OR QUOTE
-------
The disadvantages of field data result from the fact that exposures are not controlled. As
a result, the causal nature of the relationship between conductivity and the associated biological
impairments must be assessed. Also, any variables that are correlated with conductivity or the
biotic response may confound the relationship of biota to conductivity. Assessments of
causation and confounding were performed and are presented in the appendices. They
demonstrate that conductivity is a cause of impairment and the relationship between conductivity
and biological responses apparently is not significantly confounded.
The chronic aquatic life benchmark value for conductivity derived from all-year data
from West Virginia is 300 [j,S/cm. It is applicable to parts of West Virginia and Kentucky. It is
expected to be applicable to the same regions in Ohio, Pennsylvania, Tennessee, and Maryland,
but data from those states have not been analyzed. It may also be appropriate for other nearby
regions such as Ecoregions 67 but has only been validated for use in Ecoregions 68, 69, and 70 at
this time. However, this level may not apply when the relative concentrations of dissolved ions
are not dominated by salts of SO4 2 and HCO3 .
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 xiii DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
1. INTRODUCTION
At the request of U.S. Environmental Protection Agency (U.S. EPA) Regions 3, 4, and 5,
and the Office of Water, the Office of Research and Development has developed an aquatic life
benchmark for conductivity that may be applied in the Appalachian Region associated with
mixtures of ions dominated by salts of SO42 and HCO3 anions at circum-neutral to alkaline pH.
The benchmark is intended to protect the aquatic life in streams and rivers in the region. It is
derived by a method modeled on the U.S. EPA's standard methodology for deriving water
quality criteria (Stephen et al., 1985). In particular, the methodology was adapted for use of field
data. Field data were used because sufficient and appropriate laboratory data were not available
and because high quality field data were available to relate conductivity to effects on aquatic life
in streams and rivers.
1.1. CONDUCTIVITY
Although the elements comprising the common mineral salts such as sodium chloride
(NaCl) are essential nutrients, aquatic organisms are adapted to specific ranges of salinity and
experience toxic effects from excess salinity. Salinity is the property of water that results from
the combined influence of all disassociated mineral salts. The most common contributors to
salinity in surface waters, referred to as matrix ions, are
Cations: Na+, Ca2+, Mg2+, K+
Anions: CI , HC03 , C032 , S042
The salinity of water may be expressed in various ways, but the most common is specific
conductivity. Specific conductivity (henceforth simply conductivity) is the ability of a material
to conduct an electric current measured in microsiemens per centimeter ([j,S/cm) standardized to
25°C. (In this report we use "conductivity" to refer to the measurement and resulting data and
"salinity" to refer to the environmental property that is measured.) Currents are carried by both
cations and anions—but to different degrees depending on charge and mobility. Effectively,
conductivity may be considered an estimate of the ionic strength of a salt solution. A measure
such as conductivity is necessary because the effects of salts are a result of exposure to all of the
ions in the mixture—not to any one individually. Hence, unless an individual ion occurs at a
much higher concentration relative to its toxicity than other ions, the individual ion would not be
the only potential cause, and a criterion for an individual ion could be under-protective. The
ionic composition of mixtures of salts affects its toxicity (Mount et al., 1997). Therefore, this
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 1 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
aquatic life benchmark for conductivity is applicable for streams in the Appalachian Region
where conductivity is dominated by salts of SO42 and HCO3 at circum-neutral to mildly
alkaline pH.
Salinity has numerous sources (Ziegler et al., 2007). Freshwater can become increasingly
salty due to evaporation, which concentrates salts such as occurs with irrigation return waters
(Rengasamy, 2002), or diversions that reduce inflow relative to evaporation (e.g., Pyramid Lake,
Nevada). Intrusion of saltwater occurs when ground water withdrawal exceeds recharge
especially near coastal areas (Bear et al., 1999; Werner, 2009). Freshwater can also become
salty with the additions of brines and wastes (Clark et al., 2001), minerals dissolved from
weathering rocks (Pond, 2004), and runoff from treating pavements for icy conditions
(Environment Canada and Health Canada, 2001; Evans and Frick, 2000).
Exposure of aquatic organisms to salinity is direct. Fish, amphibians, mussels, and
aquatic macroinvertebrates are exposed as they ventilate their gills or other respiratory surfaces
in the course of taking up oxygen. The respiratory surfaces contain specific structures to actively
take up nutrient ions and control the osmotic balance of organisms. However, these structures
may only be able to operate within a range of salinities. For example, some aquatic insects, such
as most Ephemeroptera (mayflies), have evolved in a low salt environment. Because they would
normally lose salt, their cuticle is permeable to the uptake of salt, and they take up salt using
specialized external chloride cells on their gills (Komnick, 1977). Also, some life stages of
animals may be particularly sensitive. For instance, ionic concentrations and transport processes
are essential to regulate membrane permeability during external fertilization of eggs, including
those of fish (Tarin et al., 2000).
1.2. APPROACH
The approach used to derive the benchmark is based on the standard method for the U.S.
Environmental Protection Agency's (U.S. EPA's) published Section 304(a) Ambient Water
Quality Criteria. Those criteria are the 5th percentiles of species sensitivity distributions (SSDs)
based upon laboratory toxicity tests, such that the goal is to protect 95% of the species in an
exposed community (Stephan et al., 1985). SSDs are models of the distribution of exposure
levels at which species respond to a stressor. That is, the most sensitive species responds at
exposure level X\, the second most sensitive species responds atX2, etc. The species ranks are
scaled from 0 to 1 so that they represent cumulative probabilities of responding, and the
probabilities are plotted against the exposure levels (Posthuma et al., 2002). Centiles of the
distribution can be derived using interpolation, parametric regression, or nonparametric
regression.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 2 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
For the conductivity benchmark, the SSDs are derived from field data. There are several
reasons that some pollutants, such as suspended and bedded sediments (U.S. EPA, 2006;
Cormier et al., 2008), and some assessment endpoints do not lend themselves to laboratory
testing. For example, traditional toxicity testing and the resulting criteria derivation procedures
used by EPA do not allow for the study of a pollutant's effects on migration, predation, and other
behaviors or for species interaction. Furthermore, toxicity tests are rarely completed for the most
susceptible species and sensitive life stages, which are difficult to identify or to maintain and test
in the laboratory. The result is that the criteria are derived based on toxicity tests conducted
upon species that can be cultured in a laboratory setting, and these tests do not include a
substantial fraction of the species inhabiting an ecosystem. In sum, SSDs based on laboratory
studies cannot replicate the full range of effects or species interactions that could reasonably be
expected to occur in the environment (Suter et al., 2002).
The choice to use field data to derive benchmarks of any kind poses some challenges.
Because causal relationships in the field are uncontrolled, unreplicated, and unrandomized, they
are subject to random responses and to confounding. Confounding is the appearance of
apparently causal relationships that are due to noncausal correlations. In addition, noncausal
correlations and the inherent noisiness of environmental data can obscure true causal
relationships. The potential for confounding is reduced, as far as possible, by identifying
potential confounding variables, determining their contributions, if any, to the relationships of
interest, and eliminating their influence when possible and as appropriate based on credible and
objective scientific reasoning (see Appendix B). In addition, the evidence for and against salts as
a cause of biological impairment is weighed using causal criteria adapted from epidemiology
(see Appendix A).
Because relationships between conductivity and biological responses appear to vary
among regions and among different mixtures of ions, this benchmark is limited to two
contiguous regions with a particular dominant source of salinity. The regions are Level III 69
(Central Appalachian) and 70 (Western Allegheny Plateau) (see Figure 1) (U.S. EPA, 2007;
Omernik, 1987; Woods et al., 1996). Low salinity rain water, sometimes so low as to not be
accurately measured by conductivity, becomes salty as it interacts with the earth's surface.
Along surface and ground water paths to the ocean, water contacts bare rock. The rock
demineralizes and contributes salts that accumulate. A large surface to volume ratio of
unweathered rock increases dissolution of rock. For the most part, these salts are not degraded
by natural processes but can be diluted by more rain or by less salty tributaries. Drought
increases salt concentrations. Addition of wastes or waste waters also contributes salts. The
prominent sources of salts in Ecoregions 69 and 70 are mine overburden and valley fills from
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 3 DRAFT—DO NOT CITE OR QUOTE
-------
1 large scale surface mining, but they may also come from slurry impoundments, coal refuse fills,
2 or deep mines. Other sources include effluent from waste water treatment facilities and brines
3 from natural gas drilling and coalbed methane production. This benchmark for conductivity
4 applies to waters influenced by current inputs from these sources in Ecoregions 69 and 70 with
5 salts dominated by SC>42+ and HCO3 anions at circum-neutral to mildly alkaline pH.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 4 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
2. DATA SETS
Data are required to develop the benchmark and to support it. This section explains how
the data were selected, describes the data that were used, and explains how the data set was
refined to make it useful for analysis.
2.1. DATA SET SELECTION
The Central Appalachia (69) and Western Allegheny Plateau (70) ecoregions were
selected for development of a benchmark for conductivity because available data were of
sufficient quantity and quality, and because conductivity has been implicated as a cause of
biological impairment in these ecoregions (Pond et al., 2008). These regions were judged to be
similar in terms of water quality, including resident biota and sources of conductivity.
Confidence in the quality of reference sites in West Virginia was relatively high owing to the
extensively forested areas of the region and well-documented process by which West Virginia
Department of Environmental Protection (WVDEP) assigns reference status. They use a tiered
approach. Only tier 1 was used when analyses involved the use of reference sites, thus avoiding
the use of conductivity as a characteristic of reference condition. Nevertheless, conductivity
values from WVDEP's reference sites were low and similar in different years (see Figure 2),
providing evidence that the sites were reasonable reference sites. The 75th percentiles were
below 200 [j,S/cm in most years.
2.2. DATA SOURCES
All data used in this study were taken from the WVDEP's in-house Watershed
Assessment Branch Data Base (WABbase) 1999-2007. The WABbase contains data from
Level III Ecoregions 66, 67, 69, and 70 in West Virginia (see Figure 1) (U.S. EPA, 2000;
Omernik, 1987; Woods et al., 1996). Chemical, physical, and/or biological samples were
collected from 3,286 distinct locations during the sampling years 1999-2007. WVDEP uses a
tiered sampling design collecting measurements from long-term monitoring stations; targeted
sites within watersheds on a rotating basin schedule; probability sites (Smithson, 2007); and sites
chosen to further define impaired stream segments in support of total maximum daily load
(TMDL) development (WVDEP, 2008b). Most sites have been sampled once during an annual
sampling period, but TMDL sites have been sampled monthly for water quality parameters.
Some targeted sites represent least disturbed or reference sites that have been selected by a
combination of screening values and best professional judgment (Bailey, 2009). Water quality,
habitat, watershed characteristics, macroinvertebrate data (both raw data and calculated metrics),
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 5 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
and supporting information are used by the State to develop 305(b) and 303(d) reports to the
U.S. EPA (WVDEP, 2008b). Quality assurance and standard procedures are described by
WVDEP (2006, 2008a). All contracted analyses for chemistry and macroinvertebrate
identification follow WV's internal quality control and quality assurance protocols. This is a
well-documented, regulatory database. We judged the quality assurance to be excellent based on
the database itself, supporting documentation, and experience of EPA Region 3 personnel.
Background information was also obtained from the literature and other sources for the
assessments of causality and confounding (see Appendices A and B). (1) Toxicity test results
were obtained from peer-reviewed literature and from the U.S. EPA's Ecological Toxicology
Database. (2) Information on the effects of dissolved salts on freshwater invertebrates was taken
from standard texts and other physiological reviews. (3) The original data for Table 3 in Pond et
al. (2008) were obtained from the authors to evaluate the relative contribution of different ions in
drainage from valley fills of large scale surface mining. (4) The constituent ions for Marcellus
Shale brine were provided by EPA Region 3 based on analyses by drilling operators.
2.3. DATA SET CHARACTERISTICS
Biological sampling usually occurred once per sampling period (March through October)
with the WVDEP (1996-2007) sampling protocol. Repeat biological samples from the same
location were minimal and not excluded from the data set. They represented approximately 4%
of the sampled sites; therefore, no correction was made for pseudoreplication. Summary
statistics for ion concentration and other parameters for the data set are provided in Table 1. The
benchmark applies to waters with a similar composition.
Data from a sampling event at a site were excluded from calculations if they lacked a
conductivity measurement, for obvious reasons. They were excluded if the samples were
identified as being from a large river (>155 km2), because the assemblages are not comparable
with wadable streams (Flotemersch et al., 2001). They were excluded if the salt mixture was
dominated by CF rather than SO42 (conductivity >1,000 [j,S/cm, SO4 <125 mg/L, and
CP >250 mg/L). Four sites with elevated conductivity, high chloride and low sulfate were
removed in response to concerns that the benchmark might be biased by sites with salts
dominated by Marcellas Shale brines.
Data were excluded from calculations if the organisms were not identified to the genus
level, and a genus was excluded if it was never observed at reference sites or it was observed at
<30 sampling sites. Invertebrate genera, that did not occur at WVDEP tier 1 reference sites
represented 7% of the spring genera and 8% of the summer genera, were excluded from the
SSDs. They were excluded so that the data would be relevant to potentially unimpaired
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 6 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
conditions and so as to not include opportunistic salt-tolerant organisms. The exclusion of
genera that were observed at fewer than 30 sampling locations in the composited ecoregion
ensured reasonable confidence in the evaluation of the relationship between conductivity and the
presence and absence of a genus.
Before identifying the extirpation concentration (e.g., 95th percentile extirpation
concentration [XC95]) for each genus, we removed effects of low pH by excluding sites with a
pH of <6. This prevented potential confounding of conductivity effects by the effects of acid
mine drainage (see Appendix B).
We evaluated the effects of spring benthic invertebrate emergence, temperature, and
different conductivities associated with season by partitioning the data set into spring
(March-June) and summer (July-October) subsets.
In the WABbase, 498 benthic invertebrate genera were identified of which 213 genera
occurred at the 75 reference sites in the two ecoregions (see Table 2). Genera that did not occur
at reference sites were excluded from the SSD (see Table 3). Greater than 90% of genera
observed at reference sites as defined by WVDEP occur in both Ecoregions 69 and 70. This
indicates that the same sensitive genera exist in both ecoregions. Ecoregions 69 and 70 had
304 genera in common. Of the overall 498 genera, 170 occurred at >30 sampling locations in
Ecoregions 69 and 70. Of the genera occurring at >30 sampling sites, 128 genera occurred in
Ecoregions 69 and 129 in Ecoregion 70.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 7 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
3. METHODS
The derivation of the benchmark for conductivity includes three steps: First, the
benchmark values (XCs) for the invertebrate genera were derived. Second, the XC95 values were
used to generate an SSD and the 5th percentile of the distribution, the 5th percentile hazardous
concentration (HC05). (The HC( terminology for concentrations derived from SSDs is not in the
1985 U.S. EPA method, but has become common more recently [Posthuma et al., 2002]).
Finally, background values were estimated for the regions to ensure that the benchmark is not in
the background range. These steps are explained in this section.
Extirpation is defined as the depletion of a population to the point that it is no longer a
viable resource or is unlikely to fulfill its function in the ecosystem (U.S. EPA, 2003). In this
report, extirpation is operationally defined for a genus as the conductivity value below which
95% of the observations of the genus occur and above which only 5% occur. In other words, the
probability is 0.05 that an observation of a genus occurs above its XC95 conductivity value. This
is a chronic endpoint because the field data set reflects exposure over the entire life cycle of the
resident biota. The 95th percentile was selected because it is more stable than the maximum
value, yet still represents the extreme of an organism's tolerance of conductivity.
3.1. EXTIRPATION CONCENTRATION DERIVATION
The XC95 is estimated as the 95th percentile of the cumulative distribution of probabilities
of observing a genus at a site with respect to the concurrently measured conductivity at that site.
The XC95 estimates a conductivity value above which very few, less than 5%, of the observations
of a particular genus are likely to be found.
Observed conductivity values were nonuniformly distributed across a range of possible
values (see Figure 3), and, therefore, we were more likely to observe a genus at certain
conductivity values simply because more samples were collected at those values. To correct for
the uneven sampling frequency, we used weighted cumulative distribution functions to estimate
the XC95 values for each genus. The purpose of weighting is to avoid bias due to uneven
distribution of observations with respect to conductivity by converting the sampling distribution
to one that mimics an even distribution of sample across the gradient of conductivity. It creates a
distribution more like the design of a toxicity test, which is appropriate when developing an
exposure-response relationship. To compute weights for each sample, we first defined
equally-sized bins, each 0.048 log conductivity units wide, that spanned the range of observed
conductivity values. We then calculated the number of samples that occurred within each bin
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 8 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
(see Figure 3). Each sample was then assigned a weight wt = 1 ///,, where //, is the number of
samples in the ith bin.
The value of the weighted cumulative distribution function, F(x), of conductivity values
associated with observations of a particular genus was computed for each unique observed value
of conductivity, x, as follows:
N„ M,
< x and Gy)
F(*)=M 'I u, (1)
2>,£/(G5)
i=i ;=i
where x, is the conductivity value in the/h sample of bin A'/, is the total number of bins, M, is
the number of samples in the ith bin, G,7 is true if the genus of interest was observed in jth sample
of bin and/is an indicator function that equals 1 if the indicated conditions are true, and 0
otherwise. The XC95 value is defined as the conductivity value, x where F(x) = 0.95. Eq. 1 is an
empirical cumulative distribution function, and the output is the proportion of observations of the
genus that occur at a given conductivity or lower. However, the individual observations are
weighted to account for the uneven distribution of observations across the range of
conductivities.
An example of a weighted cumulative distribution function (CDF) is shown in Figure 4
for the mayfly, Drunella. The horizontal dashed red line indicates where F(x) = 0.95, and the
conductivity value at the intersection of this line and the CDF is the XC95 value.
This method for calculating the XC95 will generate a value even if the genus is not
extirpated. For example, the occurrence of Nigronia changes little with increasing conductivity
(see Figure 4). Therefore, it is necessary to identify those values that are actual extirpation
values. We did this by examining plots of probabilities of occurrence, estimated as the
proportion of samples within each bin in which the genus was observed. Examples are shown in
Figure 5. The solid line is provided to help illustrate the association, and its position is
calculated using a nonparametric smoothing spline fit with 3 degrees of freedom. These curves
are not used in the calculation of the XC95. The conductivity at the red, horizontal, dashed line is
the estimated XC95 from the weighted cumulative distribution. The actual XC95 was greater than
(>) the calculated XC95 if the trend was increasing or flat and the seven highest conductivity bins
were not zeros (see Figure 5). For example, the XC95 for Cheumatopsyche (an extremely salt
tolerant genus) is >9,180 [j,S/cm (see Appendices D-l and D-2).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 9 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
3.2. TREATMENT OF POTENTIAL CONFOUNDERS
Potentially confounding variables for the relationship of conductivity with the extirpation
of stream invertebrates were evaluated in several ways—which are described in Appendix B.
Based on the weight of evidence, only low pH was a likely confounder. Because low pH waters
are in violation of existing water quality criteria and because the data set was large, we excluded
sites with pH <6 before identifying the XC95 for each genus.
We evaluated the effects of spring benthic invertebrate emergence, temperature, and
different conductivities associated with season by partitioning the data set into spring
(March-June) and summer (July-October) subsets. However, we found that the SSDs for spring
and all year were similar. Therefore, we used the SSD for the combined spring and summer
samples, thus avoiding the need to apply judgment to define seasons that vary with longitude and
elevation.
To further evaluate the effect of confounders on the HCos, XC95 values for the full year
were determined from a data set from which sites were removed as follows:
• pH of <6 (removes acidity and associated dissolved metals as a cause);
• rapid bioassessment protocol (RBP) score <135 (removes marginal habitat conditions as
a cause). The RBP score is WV's composited index of qualitative measures of habitat
parameters such as bank erosion, stream sinuosity, embeddedness, and cover; and
• fecal coliform >400 colonies/100 mL (removes sources of potential organic enrichment
and potential toxicants from sewage treatment plants, failing septic tanks and livestock as
causes).
XC95 values were recalculated with the trimmed data set and compared. That analysis found that
the HCos (300 (j,S/cm) was similar to the HC05 (297 (j.S/cm) for the data set with only low pH
removed so the data set was not partitioned for either RBP score or fecal coliform when
calculating the benchmark.
Other potential confounders were evaluated, but they were not partitioned from the data
set prior to calculating the XC95 and HC05 values. Rather, we evaluated the potential magnitude
of confounding by determining the degree of correlation of the confounder with conductivity and
with the number of ephemeropteran genera. We also evaluated contingency tables of the
occurrence of any Ephemeroptera at a site with respect to high and low levels of conductivity
and the potential confounder. Ephemeroptera were selected as an effect endpoint that allowed us
to evaluate a greater range of exposures and confounding factors than occurs for individual
genera. The confounding analysis focused on Ephemeroptera, because they are among the most
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 10 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
sensitive genera. Other evidence of confounding was included when appropriate data were
available.
3.3. DEVELOPING THE SPECIES SENSITIVITY DISTRIBUTION
The SSDs are cumulative distribution plots of XC95 values for each genus relative to
conductivity (see Figure 6). The cumulative percentile for each genus P is calculated as
P = RJ(N +1) where R is the rank of the genus and N is the number of genera. Some
salinity-tolerant genera are not extirpated within the observed range of conductivity. So, like
laboratory test endpoints reported as "greater than" values, we retained field data that do not
show the field endpoint effect (extirpation) in the database. In this way, they can be included in
N when calculating the proportions responding, because they fall in the upper portion of the
SSD. The HCos was derived by using interpolation to estimate the percentile between the XC95
values bracketing P = 0.05 (i.e., the 5th percentile of modeled genera). The benchmark is
obtained by rounding the HC05 to two significant figures as directed by Stephen et al. (1985).
Exploratory SSDs were developed using different data sets to evaluate effects of
potentially influential factors. The results of these exploratory analyses and other tests are
discussed in the treatment of confounding factors (see Appendix B).
3.4. CONFIDENCE BOUNDS
The purpose of this analysis is to characterize the statistical uncertainty in the benchmark
value by calculating confidence bounds on the HC05 values. Because the XC95 values were
estimated from field data and then the HC05 values were derived from those XC95 values, we
used a method that generated distributions and confidence bounds in the first step and propagated
the statistical uncertainty of the first step through the second step.
Bootstrap estimates of the XC95 were derived for each genus used in the derivation of the
benchmark by sampling with replacement from the data set used to derive the benchmark
2,145 times (the number of observations in the data set) (Efron and Tibshirani, 1993). For each
bootstrap sample, the XC95 was calculated by the same method applied to the original data (see
Section 3.1). That process was repeated 1,000 times to create a distribution of XC95 values for
each genus. These distributions were used to calculate a two-tailed 95% confidence interval on
the XC95 for each genus. The XC95S from the original data set, the mean XC95S of the bootstrap
distributions, and the confidence intervals are shown for the most sensitive 35 genera in Figure 7.
Uncertainty in the HC05 value was evaluated by generating an HC05 from each of the
1,000 sets of bootstrapped XC95 estimates. The distribution of 1,000 HC05 values was used to
generate a two-tailed 95% confidence bounds on these bootstrap-derived values.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 11 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
3.5. ESTIMATING BACKGROUND
In general, a benchmark should be greater than natural background. The background
conductivities of streams were estimated using reference sites from the WABbase. The
75th percentile of this distribution in the summer index period (August-October), which is the
period of greatest conductivity, is 100 [j,S/cm for Ecoregion 69 and 234 [j,S/cm for Ecoregion 70.
The 75th percentile was selected because sites were among the least disturbed based on best
professional judgment (U.S. EPA, 2000). We also estimated the background conductivity for the
area using only probability samples from the WABbase, which do not rely upon any selection
criteria other than representativeness of a stream order. The 25th percentile was selected because
impaired sites are also included in the random sample (U.S. EPA, 2000). A total of
1,271 probability-based samples were collected from Ecoregions 69 and 70. The background
values, based on the 25th percentile, were 72 [j,S/cm for Ecoregion 69 and 153 [j,S/cm for
Ecoregion 70. The bases for these methods are explained in Section 5.5.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 12 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
4. RESULTS
4.1. EXTIRPATION CONCENTRATIONS
The XC95 values are presented in Appendix C. Values are calculated for all
macroinvertebrate genera that were observed at a minimum of 30 sampling sites in the two
ecoregions. Distributions of occurrence with respect to conductivity are presented for each
genus of macroinvertebrate in Figure D-l and the CDFs used to derive the XC95 values are
presented in Figure D-2.
4.2. SPECIES SENSITIVITY DISTRIBUTIONS
SSDs for invertebrates in spring, summer, and the entire sampling year (March through
October) are derived from XC95 values of 150 genera (see Figure 6). The SSDs do not reach a
horizontal asymptote at 100% of genera because salt-tolerant genera are included in the SSD that
are not extirpated within the observed range of conductivity values. The lower third of the SSD
is shown in Figure 8 for better viewing of the plots near the 5th percentile of genera.
4.3. HAZARDOUS CONCENTRATION VALUES AT THE 5TH PERCENTILE
The hazardous concentration values at the 5th percentile of the SSDs are summarized in
Table 4. The HC05 spring value is lower than the summer value and similar to the full year. The
HC05 for year-long XC95 values are similar to spring values because the spring-only genera have
low XC95 values (see Figure 6). Other seasonal differences result from exclusion of some taxa
due to sample sizes less than 30 in spring or summer or seasonal differences in the ability to
sample some genera (see Section 5.2). Rounding the HC05 for all year of 297 [j,S/cm to two
significant figures yields a benchmark value of 300 [j,S/cm (see Figure 8).
4.4. UNCERTAINTY ANALYSIS
The following HC05S resulted from the bootstrap-derived statistics, a lower confidence
bound of 225 [j,S/cm and an upper confidence bound of 305 [j,S/cm. These confidence bounds
are asymmetrical with respect to the point estimate of 297 [j,S/cm. In general statistical practice,
confidence bounds around estimates are not infrequently asymmetric. In the case of bootstrap
generated estimates as used here, asymmetry occurs because statistical resampling from the
distribution of data generates more realizations that produce values lower than the point estimate
than realizations that produce higher values.
Confidence bounds represent the potential range of HC05 values using the SSD approach,
given the data and the model. Conceptually, these confidence bounds may be thought of
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 13 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
representing the potential range of HCos values that one might obtain by returning to WV and
resampling the streams. The contributors to this uncertainty include measurement variance in
determining conductivity and sampling variance in the locations for monitoring and in collecting
and enumerating organisms. It also includes variance due to differences in stream reaches,
weather, and other random factors.
The confidence bounds do not address potential systematic sources of variance such as
differences between geographic areas or between different organizations performing the
sampling using different protocols. The contributions of those sources of uncertainty (in addition
to the sampling uncertainty) can best be evaluated by comparing results of independent studies.
One estimate of that larger uncertainty is provided by comparing the all-year HCos values
derived from West Virginia and Kentucky data. Even though the data were obtained in different
areas by different agencies using different protocols, the values differ by only 7% (see
Appendix E for details). In addition, the 95% confidence bounds on the HCos values for the two
states overlap, suggesting that the sampling variance (i.e., the uncertainty captured by the
confidence intervals) may be the largest component of total uncertainty. While this result is from
only one comparison of two states, it does provide a reassuring validation of the WV results.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 14 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
5. CONSIDERATIONS
Because of the complexity of field observations, decisions must be made when deriving
field-based benchmark values that are not required when using laboratory data. In the case of
conductivity, additional decisions must be made to address a pollutant that is a mixture and a
naturally occurring constituent of water.
5.1. SELECTION OF INVERTEBRATE GENERA
Selection of genera to model can affect the results. Using the data set of all taxa includes
taxa that may occur due to a competitive advantage in polluted water. Some taxa, such as
Corbicula, are not native to streams in North America. Using only genera found at sites with
minimal disturbance as defined by reference sites somewhat alleviates this problem. The
reference site genera are often linked to state narrative water quality standards; thus, they
represent the aquatic life use that state water quality criteria should be designed to protect.
Furthermore, the importance of losing species that inhabit minimally disturbed sites may be
clearer to decision makers and stakeholders. In this particular case, using all genera including
invasive species would increase HC05 by only 2% in the full year data set.
Genera are also selected for statistical reasons. We restricted genera used in analysis to
those recorded at a minimum of 30 sampling sites to reduce the chance that an apparent
extirpation is due to sampling variance and to increase the likelihood that the models and
exploratory analyses for potential confounding are reasonably strong.
5.2. SEASONALITY, LIFE HISTORY, AND SAMPLING METHODS
The seasonality of life history events such as emergence of aquatic insects can affect the
probability of detecting a species, because eggs and early instars are not captured by the
sampling methods used. As a result, annual insects that emerge in the spring are present but
unlikely to be detected in the summer, when conductivities increase in some streams.
Some invertebrate genera are observed only in the spring probably due to the size of the
early life stages or diapausing eggs in summer months. Because many of these same genera may
be among the most sensitive genera, they could have a strong influence on the HC05. Therefore,
we evaluated the summertime conductivities of the streams in which these genera were found in
springtime. The conductivities of these streams were often stable throughout the year (see
Figure 9), but in some streams that supported sensitive genera in spring, conductivities increased
in the summer. This suggests that some genera might withstand higher conductivities that
coincide with certain parts of a genus's lifecycle; however, these genera clearly do not tolerate
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 15 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
those conductivities in the spring. Lower effects levels were not due to insufficient test range of
conductivities because exposures as high as 5,200 [j,S/cm occurred in the spring samples.
Furthermore, when we partitioned the data into spring and summer periods, the XC values are
lower in the spring than in the summer (see Figures 6 and 8).
The effects of seasonality and life history were evaluated by comparing occurrences of
individual invertebrate genera and XC95 and HC05 values partitioned for season (see Figures 6
and 8). The data set was partitioned into spring and summer based on seasonal patterns of
conductivity at WVDEP reference sites (see Figure 2). The spring season is March through
June. The summer season is July through October. The exposure is characterized by water
quality parameters measured on the same date that a taxon is observed in the stream. Both high
and low conductivity streams are represented in both spring and summer samples. However, the
conductivity in certain streams may be three times greater in the summer than the spring.
We cannot be sure whether the greatest exposures in summer are tolerated by the
spring-emergent genera. However, streams with conductivity <300 [j,S/cm in summer are also
below the benchmark in spring 98% of the time (see Figure 10). For simplicity, we recommend
the year-round value (see Section 6), but seasonal variation should be considered when planning
monitoring of conductivity.
5.3. INCLUSION OF REFERENCE SITES
If high quality (i.e., reference) sites are not included in the data set, effects on sensitive
species will not be incorporated into the benchmark. That is, the lower end of the SSD will be
missing. For example, in a region where all watersheds include tilled agricultural land uses, all
sites are affected by sediment, so a legitimate SSD for sediment should not be derived by this
method in that region. In this case, WVDEP's reference sites were included as well as many
probability sites with >90% forest cover, which are believed to be representative of good- to
high-quality systems.
5.4. DEFINING THE REGION OF APPLICABILITY
If the method described here is applied to a large region, the increased range of
environmental conditions and a greater diversity of anthropogenic disturbances may obscure the
causal relationship. However, if the region is too small, the available data set may be inadequate,
and the resulting benchmark value will have a small range of applicability. In this case, we
chose two adjoining regions that have abundant data, >90% of genera in common, and a
common dominant source of the stressor of concern.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 16 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
5.5. BACKGROUND
For naturally occurring stressors, it would not, in general, be appropriate to derive a
benchmark value that is within the background range. In this case, background conductivities for
Ecoregions 69 and 70 are 100 and 234 [j,S/cm, respectively, using 75th percentiles of reference
sites in West Virginia. Reference sites are sites that are judged to be among the best within a
category, but they are not necessarily pristine or representative of natural background. Some
reference sites have unrecognized disturbances or have recognized disturbances that are less than
most others in their category. Some have extreme values of a stressor because of measurement
error or unusual conditions at the time the sample was taken. For those reasons, when estimating
background concentrations, it is conventional to use only the best 75% of reference values. The
cutoff percentile is based on precedent and on the collective experience of EPA field ecologists
(U.S. EPA, 2000).
The background values based on the 25th percentile range between 72 [j,S/cm for
Ecoregion 69 (n = 617) and 153 [j,S/cm (n = 654) for Ecoregion 70 for probability samples in
West Virginia. Samples from a probability design include all types of waters including impaired
sites. In some regions there are no pristine streams. To characterize the best streams, the
25th percentile is commonly used by EPA field ecologists (U.S. EPA, 2000). None of these
values exceed the HC05 values in Table 4.
5.6. INCLUSION OF OTHER TAXA
Fish were not included because their occurrence is affected by stream size making it
difficult to determine XC95 values. Some of the affected streams naturally have no fish. In
addition, the WABbase data set used to derive the benchmark does not contain data for fish.
Other data sets that do contain fish are not as large and do not contain as great a range of
conductivity values. A separate SSD might be developed for fish, once these technical issues are
resolved. Data for plants and amphibians are not available. Additional findings regarding
mussels could change this analysis if they are found to be more sensitive to conductivity than the
invertebrates used here. Mussels were not represented because genera did not occur in a
minimum of 30 samples. Additional analyses may be necessary to ensure protection of federally
or state listed rare, threatened, or endangered species of fish, amphibians, and mussels.
5.7. TREATMENT OF RARE SPECIES
Species listed by West Virginia Department of Natural Resources (WVDNR, 2007) as
threatened were among the genera observed. Because taxa were identified to genus, we are not
certain if the species are included. Therefore, we recommend that the invertebrate taxa,
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 17 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Alloperla, Pteronarcys, Cordulegaster, Ephemera, and Sweltsa, be identified to species in
subsequent monitoring to evaluate the risk to these threatened taxa. Furthermore, freshwater
mussels were not well represented in the samples perhaps due to the sampling methods. Stephan
et al. (1985) recommend lowering the concentration below the 5th percentile when necessary to
protect threatened, endangered, or otherwise important species.
5.8. SELECTION OF THE EFFECTS ENDPOINT
We have used the extirpation concentration as the effects endpoint, because it is easy to
understand that an adverse effect has occurred when a genus is lost from an ecosystem.
However, for the same reason, it may not be considered protective. An alternative is to use a
depletion concentration (DCX) based on a percent reduction in abundance or capture probability.
Another option is to use only those taxa sensitive to the stressor of concern, thus developing an
SSD for the most relevant taxa. DC values or other more sensitive endpoints may be considered
when managing exceptional resources.
In this study, an invertebrate genus may represent several species, and this approach
identifies the pollutant level that extirpates all species within that genus (i.e., it is the level at
which the least sensitive among them is rarely observed). In a review of extrapolation methods,
Suter (2007) indicated that although species within a genus respond similarly to toxicants,
different species within a genus may have evolved to partition niches afforded by naturally
occurring causal agents such as conductivity. Hence, an apparently salt tolerant genus may
contain both sensitive species and tolerant species. A potential solution would be to use distinct
species. However, this may not be practical because some taxa are very difficult to identify
except as late instars. We chose to follow Stephen et al. (1985) by using genera until such time
that the advantages and disadvantages of using species can be more fully studied.
5.9. USE OF MODELED OR EMPIRICAL DISTRIBUTIONS
When deriving XC and HC values, one might use a percentile of an empirical distribution
or fit a function to the data and calculate the value from the resulting model. Models use all of
the data and, therefore, are resistant to biases associated with any peculiar data at the percentiles
of interest or to uneven distributions of data. However, there is no a priori reason to believe that
these distributions have a prescribed mathematical form, and fitted models may fit the data
poorly at the percentiles of interest. The use of a nonparametric regression method to alleviate
the problem of assuming a particular functional form can result in biologically unlikely forms,
may reduce the potential generality of the model, and is not readily understood. The use of
empirical distribution functions without fitted models eliminates the problems of model selection
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 18 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
and makes the method easier to understand and implement. With respect to SSDs, this issue is
unresolved, and assessors are encouraged to consider the properties of their distributions when
deciding whether to fit or not (Newman et al., 2002; Suter et al., 2002). In this case, data are
abundant, and either the empirical or modeled methods could work well. In the interest of
conceptual and operational simplicity, we identify the XC95 as the conductivity value at which
the empirical cumulative probability is 0.95. Similarly, the HC05 is determined by interpolation
of points on the empirical distributions of XC95 values as described in Stephan et al. (1985).
5.10. TREATMENT OF CAUSATION
Causation should not be an issue in laboratory toxicity tests, but, even with rigorous
treatment of confounders, skeptics will question whether observed field relationships are truly
causal (Kriebel, 2009). Like many epidemiologists, we believe that statistical analysis of
relationships should be supplemented by the consideration of qualitative criteria for causation.
In this case, we used evidence of causal characteristics derived from Hill's considerations
(Cormier et al., 2010) to evaluate the causal relationship of conductivity and extirpation of
organisms (see Appendix A).
5.11. TREATMENT OF MIXTURES
In natural waters, salinity is a result of mixtures of ions. We use conductivity as a
measure of the mixture. However, waters with different mixtures of salts but the same
conductivity may have different toxicities. In this case, the benchmark value was calculated for
a relatively uniform mixture of ions in those streams that exhibit elevated conductivity in the
Appalachian Region associated with salts dominated by SO42 and HCO3 anions at
circum-neutral to mildly alkaline pH. Recent increases in drilling for natural gas may change the
toxicity of salinity in this region, and monitoring should be designed to evaluate differences.
The relative contributions of individual salts from large scale surface coal mining were described
by Pond et al. (2008). Whereas Ca2+, Mg2+, SO42 , and HCO3 are the four most common ions to
drain from surface coal mines, ions of Na+ and CI are the two most common in seawater and
brines from Marcellus Shale drilling operations (Bryant et al., 2002). Because the few sites with
very elevated CF were found to be outliers in the distributions of occurrence, they were deleted
from the data set used to derive the XC95 values. Hence, the use of the benchmark value in other
regions or in waters that are contaminated by other sources such as road salt or irrigation return
waters may not be appropriate. However, for the circum-neutral to alkaline drainage from
surface mines and valley fills, these four primary ions are highly correlated with conductivity
(see Figures lla-e).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 19 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
6. AQUATIC LIFE BENCHMARK
The aquatic life benchmark of 300 [j,S/cm was developed for year-round application.
This level is intended to prevent the extirpation of 95% of invertebrate genera in this region. The
estimated two-tailed 95% lower confidence bound of the HCos point estimate is 225 [j,S/cm and
the upper bound is 305 [j,S/cm.
The aquatic life benchmark has been validated by an independent data set. Application
of the same methodology to data from the State of Kentucky gave a very similar result,
319 [j,S/cm with a lower confidence bound of 180 [j,S/cm and an upper bound of 439 [j,S/cm (see
Appendix E).
The method used to develop the benchmark is an adaptation of the standard method for
deriving water quality criteria for aquatic life (Stephan et al., 1985), so it is supported by
precedent. Because the organisms are exposed throughout their life cycle, this is a chronic value.
The aquatic life benchmark for conductivity is provided as scientific advice for reducing
the increasing loss of aquatic life in the Appalachian Region associated with a mixture of salts
dominated by salts of SO42 and HCO3 anions at circum-neutral pH. The aquatic life
benchmark for conductivity is applicable to parts of West Virginia, which provided the data for
its derivation, and Kentucky, which gave essentially the same result. It may be applicable to
Ohio, Tennessee, Pennsylvania, and Maryland in Ecoregions 68, 69, and 70. (Region 68
[Southwestern Appalachia] does not occur in WV and is not included in the derivation of the
benchmark value, but it is included in the validation data set from Kentucky [see Appendix E]).
The aquatic life benchmark may also be appropriate for other nearby regions. However, this
level may not apply when the relative concentrations of dissolved ions are different (see Table 1
for the ranges of concentrations in the data set used to derive the benchmark value).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 20 DRAFT—DO NOT CITE OR QUOTE
-------
1 REFERENCES
2
3
4 Bailey, J. (2009) Watershed assessment section's 2009 standard operating procedures. West Virginia Department of
5 Environmental Protection, Division of Water and Waste Management, Watershed Branch, Charleston, VA.
6 Bear, JA; Cheng, HD; Sorek, S; et al. (1999) Seawater intrusion in coastal aquifers: concepts, methods and practices.
7 Norwell, MA: Kluwer Academic Publishers.
8 Bryant, G; McPhilliamy, S; Childers, H. (2002) A survey of the water quality of streams in the primary region of
9 mountaintop / valley fill coal mining, October 1999 to January 2001. In: Draft programmatic environmental impact
10 statement on mountaintop mining / valley fills in Appalachia - 2003. Appendix D. U.S. Environmental Protection
11 Agency, Region 3, Philadelphia, PA. Available online at
12 http://www.epa.gOv/Region3/mtntop/pdf/appendices/d/stream-chemistry/MTMVFChemistryPartl .pdf.
13 Clark, ML; Miller, KA; Brooks, MH. (2001) US Geological Survey monitoring of Powder River Basin stream-water
14 quantity and quality. U.S. Geological Survey, Cheyenne, WY. Available online at
15 http://pubs.usgs.gov/wri/wriO 14279/html/report.htm.
16 Cormier SM; Paul, JF; Spehar, RL; et al. (2008) Using field data and weight of evidence to develop water quality
17 criteria. Integr Environ Assess Manag 4(4) :490-504.
18 Cormier, SM; Suter, GW, II; Norton, SB. (2010) Causal characteristics for ecoepidemiology. Hum Ecol Risk
19 Assess 16(1):53—73.
20 Efron, B; Tibshirani, R. (1993) An introduction to the bootstrap. Monographs on statistics and applied
21 probability, 57. Boca Raton, FL: Chapman & Hall/CRC.
22 Environment Canada and Health Canada. (2001) Priority substances list assessment report: road salts. Available
23 online at http://www.ec.gc.ca/Substances/ese/eng/psap/final/roadsalts.cfm.
24 Evans, M; Frick, C. (2000) The effects of road salts on aquatic ecosystems. National Water Research Institute
25 (NWRI), Saskatoon, Saskatchewan NWRI Contribution No 02-308.
26 Flotemersch, JE; Cormier, SM; Autrey, BC. (2001) Comparisons of boating and wading methods used to assess the
27 status of flowing waters. U.S. Environmental Protection Agency, Office of Research and Development, Cincinnati,
28 OH. EPA/600/R-00/108. Available online at http://www.epa.gov/nerleerd/MCD_nocover.pdf.
29 Komnick, H. (1977) Chloride cells and chloride epithelia of aquatic insects. Int Rev Cytol 49:285-328.
30 Kriebel, D. (2009) How much evidence is enough? Conventions on causal inference. Law Contemp Probl
31 72:121-136.
32 Mount, DR; Gulley, DD; Hockett, R; et al. (1997) Statistical models to predict the toxicity of major ions to
33 Ceriodaphnia dubia, Daphnia magna, and Pimephalespromelas (fathead minnows). Environ Toxicol Chem
34 16(10):2009-2019.
3 5 Newman, MC; Ownby, DR; Mezin, LCA; et al. (2002) Species sensitivity distributions in ecological risk
36 assessment: distributional assumptions, alternate bootstrap techniques, and estimation of adequate number of
37 species. In: Posthuma, L; Suter, GW, II; Traas, TP; eds. Species sensitivity distributions in ecotoxicology. Boca
3 8 Raton, FL: Lewis Publishers; pp 119-132.
39 Omernik, JM (1987) Ecoregions of the conterminous United States. Ann Assoc Am Geograph 77:118-125.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 21 DRAFT—DO NOT CITE OR QUOTE
-------
1 Pond, GJ. (2004) Effects of surface mining and residential land use on headwater stream biotic integrity in the
2 eastern Kentucky coalfield region. Kentucky Department of Environmental Protection, Division of Water,
3 Frankfort, KY. Available online at http://www.water.ky.gov/NR/rdonlyres/ED76CE4E-F46A-4509-8937-
4 lA5DA40F3838/0/coal_miningl.pdf.
5 Pond, GJ; Passmore, ME; Borsuk, FA; et al. (2008) Downstream effects of mountaintop coal mining: comparing
6 biological conditions using family- and genus-level macroinvertebrate bioassessment tools. J N Am Benthol Soc
7 27:717-737.
8 Posthuma, L; Suter, GW, II; Traas, TP; eds. (2002) Species sensitivity distributions in ecotoxicology. Boca Raton,
9 FL: Lewis Publishers.
10 Rengasamy, P. (2002) Transient salinity and subsoil constraints to dryland farming in Australian sodic soil: an
11 overview. Aust J Exp Agri 42:351-361.
12 Smithson, J. (2007) West Virginia Stream/River Survey Design 2007—2 111. WV Department of Environmental
13 Protection, Division of Water and Waste Management, Charleston, WV 25304.
14 Stephan, CE; Mount, DI; Hanson, DJ; et al. (1985) Guidelines for deriving numeric National Water Quality Criteria
15 for the protection of aquatic organisms and their uses. U.S. Environmental Protection Agency, Washington, D.C.
16 PB85-227049. Available online at http://www.epa.gov/waterscience/criteria/library/85guidelines.pdf.
17 Suter, GW, II. (2007) Ecological risk assessment. 2nd Edition. Boca Raton, FL: CRC Press.
18 Suter, GW, II; Traas, T; Posthuma, L. (2002) Issues and practices in the derivation and use of species sensitivity
19 distributions. In: Posthuma, GW, Suter, II, Traas, T; eds. Species sensitivity distributions in ecotoxicology, L. Boca
20 Raton: Lewis Publishers; pp 437-474.
21 Tarin, JJ, A Cano, JJTarbin.. (2000) Fertilization in protozoa and metazoan animals: Cellular and molecular
22 aspects. Berlin: Springer-Verlag.
23 U.S. EPA (U.S. Environmental Protection Agency). (2000) Nutrient criteria technical guidance manual: rivers and
24 streams. Office of Water, Washington, DC. EPA/822/B-00/002. Available online at
25 http://www.epa.gov/waterscience/criteria/nutrient/guidance/rivers/rivers-streams-full.pdf.
26 U.S. EPA (Environmental Protection Agency). (2003) Generic Ecological Assessment Endpoints (GEAEs) for
27 ecological risk assessment. Risk Assessment Forum Washington, DC. EPA/630/P-02/004B. Available online at
28 http://oaspub.epa.gov/eims/eimscomm. getfile?p_download_id=429201.
29 U.S. EPA (Environmental Protection Agency). (2006) Framework for developing suspended and bedded sediments
30 water quality criteria. Office of Water, Washington, DC. EPA-822-R-06-001. Available online at
31 http://www.epa.gOv/med/Prods_Pubs/framework_sabs_20060601 .pdf.
32 U.S. EPA (Environmental Protection Agency). (2007) Level III ecoregions of the Continental United States
33 (revision of Omernik 1987). National Health and Environmental Effects Research Laboratory, Corvallis, OR.
34 Available online at ftp://ftp.epa.gov/wed/ecoregions/us/useco.pdf.
3 5 Werner, AD. (2009) A review of seawater intrusion and its management in Australia. Hydoegeol J Published on
36 line http://www.springerlink.com/content/q75147m6j6766470/fulltext.pdf (accessed 10/14/2009).
37 WVDEP (West Virginia Department of Environmental Protection). (2006) Department of Water and Waste
3 8 Management Division of Water and Waste Management quality assurance project plan for watershed branch
39 monitoring activities. WVDEP, Charleston, WV; pp. 279.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 22 DRAFT—DO NOT CITE OR QUOTE
-------
1 WVDEP (West Virginia Department of Environmental Protection). (2008a) West Virginia integrated water quality
2 monitoring and assessment report. West Virginia Department of Environmental Protection, Charleston, WV.
3 Available online at
4 http://www.wvdep.org/Docs/16495_WV_2008_IR_Supplements_Complete_Version_EPA_Approved.pdf.
5 WVDEP (West Virginia Department of Environmental Protection). (2008b) West Virginia Department of
6 Environmental Protection Watershed Assessment Branch 2008 Standard Operating Procedures. Vo. 1 SOPP. 206.
7 WVDNR (West Virginia Department of Natural Resources). (2007) Rare, threatened and endangered animals. West
8 Virginia Natural Heritage Program. February 2007. Available online at
9 http://www.wvdnr.gov/Wildlife/documents/Animals2007.pdf (accessed 10/12/2009).
10 Woods, AJ; Omernik, JM; Brown, DD; et al. (1996) Level III and IV ecoregions of Pennsylvania and the Blue
11 Ridge Mountains, the Ridge and Valley, and the Central Appalachians of Virginia, West Virginia, and Maryland.
12 U.S. Environmental Protection Agency, Office of Research and Development, Corvallis, OR; EPA/600R-96/077.
13 Ziegler, CR; Suter, GW, II; Kefford, BJ. (2007) Candidate Cause: Ionic Strength. Available online at
14 www.epalgov/caddis (accessed 10/13/09).
15
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 23 DRAFT—DO NOT CITE OR QUOTE
-------
Table 1. Summary statistics of the measured water quality parameters
° ^
>3*
to
s
a
to
s
S"
a
a,
Sf
<§>
TO
TO'
*
S
s
to ^
[v Co
0
s
1
a.
2 §-
7? ^
nr\ S
H ©
3 §
2 §
O ^
H r?
<*^
hh Oq
H to
W|
o ^
o
H
W
Units
Min
25th
percentile
Median
75th
percentile
Max
Mean
Valid N
Conductivity
|iS/cm
15.4
153
269.5
576
11,646
490.90
2,145
Hardness
mg/L
0.51
52.04
94.97
196.44
1,491.79
181.73
1,087
Alkalinity
mg/L
0.2
32.05
69.6
120
560
86.57
1,366
S04
mg/L
1
17.5
40
170.8
6,000
179.4
1,365
Ca, total
mg/L
0.002
14
26.2
51.1
430
48.89
1,091
Mg, total
mg/L
0.05
3.94
6.6
14.91
204
14.39
1,089
Chloride
mg/L
1
3
5.53
12.03
1,153
17.93
1,055
TSS
mg/L
1
3
4
6
190
6.5
1,380
Fe, total
mg/L
0.005
0.14
0.27
0.51
110
0.75
1,369
NO2-NO3
mg/L
0.01
0.1
0.21
0.38
30
0.47
1,113
Al, total
mg/L
0.01
0.09
0.12
0.24
12
0.27
1,372
Al, dissolved
mg/L
0.011
0.02
0.05
0.06
0.93
0.055
1,225
Fe, dissolved
mg/L
0.001
0.02
0.05
0.061
31.8
0.147
1,196
Mn, total
mg/L
0.003
0.02
0.042
0.105
7.25
0.145
1,367
Mn, dissolved
mg/L
0.01
0.03
0.07
0.22
1.06
0.16
19
Total phosphate
mg/L
0.01
0.02
0.02
0.03
2.36
0.039
1,116
Se, dissolved
mg/L
0.001
0.001
0.001
0.001
1.26
0.006
290
-------
Table 1. Summary statistics of the measured water quality parameters (continued)
Units
Min
25th
percentile
Median
75th
percentile
Max
Mean
Valid N
Se, total
mg/L
0.0003
0.001
0.001
0.005
1.26
0.006
472
Fecal coliform
counts/100 mL
0.19
40
175
600
250,000
1,515
1,998
DO
mg/L
1.0
8.2
9.2
10.2
18.4
9.2
2,118
pH
standard units
6.02
7.29
7.63
7.97
10.48
7.6
2,145
Catchment area
km2
0.08
3.71
10.47
31.64
153.82
24.86
2,141
TSS = Total suspended solids
Note: K+ and Na+ not measured
-------
Table 2. Number of samples with reported genera and conductivity meeting
our acceptance criteria for calculating the benchmark value. Number of
samples is presented for each month, ecoregion, and database
Month
Region
3
4
5
6
7
8
9
10
Total
69
1
63
188
103
79
267
232
58
987
70
4
186
232
179
194
237
118
8
1,158
Total
5
249
420
282
273
504
350
62
2,145
Table 3. Genera excluded from 95th percentile extirpation concentration
calculation because they never occurred at reference sites
Argia
Baetisca
Calopteryx
Chironomus
Corbicula
Dineutus
Ferrissia
Fossaria
Palpomyia
Paratendipes
Nanocladius
Prostoma
Sphaerium
Stenochironomus
Stictochironomus
Tokunagaia
Tribelos
Tricorythodes
Table 4. Hazardous concentration at the 5th percentile for invertebrates in
Ecoregions 69 and 70
Season
HCos
All year
297 [j,S/cm
Spring
322 [j,S/cm
Summer
479 [j,S/cm
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 26 DRAFT—DO NOT CITE OR QUOTE
-------
Omernik Level III Ecoregions 69 and 70
Legend
| | Ecoregion 69 and 70
Advisory
] States
Figure 1. Data are from Tier III Ecoregions 69 and 70 spanning the states of
Ohio, Pennsylvania, Kentucky, Tennessee, West Virginia, and Maryland.
Data source: State outlines from U.S. EPA Base Map Shapefile, Omernik
Level III Ecoregions from National Atlas (National Atlas.gov) Projection
NAD1983UTM17N.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 27 DRAFT—DO NOT CITE OR QUOTE
-------
n 1 1 1 1 1 1 1 r
12456789 12
Month
Figure 2. Box plot showing seasonal variation of conductivity (jiS/cm) in the
reference streams of Ecoregions 69 and 70 in West Virginia from 1999 to
2006. A total of 97 samples from 70 reference stations were used for this
analysis. The 75th percentiles were below 200 [j,S/cm in most years.
&
a
o
^¦rTl-U
1.5
Tttm fl -i-i ¦
—f—
2.0
—f—
2.5
—T~
3.0
—T~
3.5
-1
4.0
log conductivity
Figure 3. Histogram of the frequencies of observed conductivity values in
samples from Ecoregions 69 and 70 from March to October. More of the
sampled sites were in the midrange than in the extremes.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 28 DRAFT—DO NOT CITE OR QUOTE
-------
Drunella
Niaronia
Figure 4. Example of a weighted CDF and the associated 95th percentile
extirpation concentration value. Each point shows the weighted proportion of
samples with Drunella or Nigronia present at (Fn(x)) the conductivity less than
the indicated conductivity value ([j,S/cm). The XC95 is the conductivity at the
95th percentile of the cumulative distribution function (CDF) (horizontal dashed
line). The CDF was calculated from observations from March through October
(all year; black connected points) from March through June (spring; green
connected points), and from July through October (summer; red connected
points). As there were fewer than 30 observations of Drunella between July and
October, no CDF was developed for the summer index period. In a CDF, genera
that are affected by increasing conductivity (e.g., Drunella) show a steep slope
and asymptote well below the measured range of exposures; whereas, genera
unaffected by increasing conductivity (e.g., Nigronia) have a steady increase over
the entire range of measured exposure and do not reach a perceptible asymptote.
1.5 2.0 2.5 3.0 3.5
1.5 2.0 2.5 3.0 3.5
log conductivity
log conductivity
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 29 DRAFT—DO NOT CITE OR QUOTE
-------
LEPIDOSTOMA
o ^
|
s
5
«s
>5*
ft
Sk
5r
'<&
jg
¦ >.
m s
H S
B
O
2
Sn
H
!?¦
a
o
o
H
°r5r
HH (/Q
H
W|
o ®?
o
H
M
>.
,-t; tp
15 °
ra
_Q
a
I °
ro
e
03
_Q o
o
V
V
V
CCD \
\
lO
O \
\o
0 \
^ DO
\ °
V
O
^¦-.Q Q 0
DIPLOPERLA
a
fN
>S O
5 £
P o
ns
>
i_
k_
4J
Cfl
_Q
O
y odo
crfW. o
1.5 2.0 2.5 3.0
log conductivity
3.5
4.0
~i 1 1 r
1.5 2.0 2.5 3.0 3.5
loq conductivity
Figure 5. Three typical distributions of observation probabilities (March through October). Open circles are the
probabilities of observing the genus within a range of conductivities. Circles at zero probability indicate no individuals
at any sites were found at these conductivities. The line fitted to the probabilities is for visualization. The vertical red
line indicates the XC95. Note that different genera respond differently to increasing salinity. Lepidostoma declines,
Diploperla has an optimum, and Cheumatopsyche increases. The XC95 for genera like Cheumatopsyche are reported as
"greater than" because extirpation did not occur in the measured range.
-------
o
A-
oo
d
ro
!
0
c
0
O)
o
4—
o
c
o
•e
o
CL
o
o
CL
CN
d
o
o
100 200 500 1000 2000 5000 10000
Conductivity (|jS/cm)
Figure 6. The species sensitivity distribution for all year (March through
October [black circles], March through June [green triangles], and July
through October [red +]). More than 100 genera are included. The HCos is the
conductivity at the intercept of the CDF with the horizontal line at the
5th percentile.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 31 DRAFT—DO NOT CITE OR QUOTE
-------
100 158 251 398 631 1000 1585
Conductivity (|jS/cm)
Figure 7. The cumulative distribution of XC95 values for the 35 most
sensitive genera (red circles) and the bootstrap-derived means (blue
x symbol) and two-tailed 95% confidence intervals (whiskers). The
5th percentile is shown by the dashed line.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 32 DRAFT—DO NOT CITE OR QUOTE
-------
o
00
LO
CN
O
CN
O
LO
LO
O
o
o
100
200
297
500
1000
2000
Conductivity (|jS/cm)
Figure 8. Species sensitivity distribution for all year. The dotted horizontal
line is the 5th percentile. The vertical arrow indicates the HCos of 297 [j,S/cm.
Only the lower 50 genera are shown to better discriminate the points in the left
side of the distribution.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 33 DRAFT—DO NOT CITE OR QUOTE
-------
MC-00150-0.2
KNG-00D9&-0
8
Cn <
S
KNL-ul jcT-5.
OUS-O0194-2.04
- . e *
n 1 1 r
IDA 15D 200 25D
Julian day
-1 1 1 1 1—
SO 1CO 15D 2CC 2SU
Julian day
—r
131
Figure 9. Examples of a monthly year-long stream conductivity record in a
stream. Filled circles: events when macroinvertebrates were sampled
concurrently with conductivity; open circles: physical-chemical sampling only.
Some streams had a steady level of conductivity (plots on left); others increased
in summer month (plots on right).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 34 DRAFT—DO NOT CITE OR QUOTE
-------
o
o
o
o
343 P50.7 %)
123 ( 18.2%)
oo
o0<
o
o
o
>
-I—I
o
=3
OO
CD
&-
d>
E
E
=3
tn
207 ( 30.6 %)
o
o
00
4(0.6%)
CN
00
O
10
32
100
316
1000
3162
10000
Spring Conductivity
Figure 10. Correlation of conductivity values sampled from the same site in spring
and summer. When conductivity is <300 (.iS/cm (broken lines) in March thru June, the
conductivity is <300 jiS/cm in the same stream 63% of the time July through October.
When the conductivity is <300 uS/cm in July through October, the conductivity in the
same stream March through June is <300 uS/cm 98% of the time.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 35 DRAFT—DO NOT CITE OR QUOTE
-------
Conduct
Sulfate
Chloride
Figure 11a. Anions. Matrix of scatter plots and absolute Spearman correlation
coefficients between conductivity ([j,S/cm), sulfate (mg/L), and chloride
concentrations (mg/L) in streams of Ecoregions 69 and 70 in West Virginia. All
variables are logarithm transformed. The smooth lines are the locally weighted
scatter plot smoothing (LOWESS) lines (span = 2/3).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 36 DRAFT—DO NOT CITE OR QUOTE
-------
Conduct
0 2 4 6
J I I L
0.95
Hardn
0.93
0.96
Mg
-6 -2
2 4 6
I I I
0.92
0.99
0.91
0.78
to
- ID
0.77
0.7
Ca
0.78
tTVtt
3 4 5 6 7 8 9
n n
i i i
-2 0 2 4
Alkal
0 2 4 6
Figure lib. Cations. Matrix of scatter plots and absolute Spearman correlation
coefficients between conductivity ([j,S/cm), hardness (mg/L), Mg (mg/L), Ca
(mg/L), and alkalinity (mg/L) in the streams of Ecoregions 69 and 70 in West
Virginia. All variables are logarithm transformed. The smooth lines are the
locally weighted scatter plot smoothing (LOWESS) lines (span = 2/3).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 37 DRAFT—DO NOT CITE OR QUOTE
-------
-4 -3 -2 -1 0
_l I I I L
-4 -3 -2 -1 0
J I I L
Conduct
0.64
0.14
0.12
0.08
o o jB
Dis Mn
NA
0.01
0.28
o o
Dis Se
0.21
0.06
o
°oo o
o o
\°%° '
I og o
o o
Dis Al
0.06
I I I I I I I
3 4 5 6 7 8 9
I I 1 I I I I
7 -5 -3 -1
Dis Fe
~i—i—i—i—i—r
-6 -2 0 2 4
Figure 11c. Dissolved metals. Matrix of scatter plots and absolute Spearman
correlation coefficients among conductivity ([j,S/cm) and dissolved metal
concentrations (mg/L) in the streams of Ecoregions 69 and 70 in West Virginia.
All variables are logarithm transformed. The smooth lines represent the locally
weighted scatter plot smoothing (LOWESS) lines (span = 2/3).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 38 DRAFT—DO NOT CITE OR QUOTE
-------
Conduct
6-4-2 0 2
_l I I L
0.35
Mn
~ O ODD
0,08
0,12
-4 -2 0 2 4
J I I I L
0.03
0,57
0,09
0.13
0.28
0.13
CO
- ID
ITTI il
! i ill i i
02 Q
6-4-2 0
Al
3 4 5 6 7 8 9
-4-2 0 2
Figure lid. Total metals. Matrix of scatter plots and absolute Spearman
correlation coefficients between conductivity (jj,S/cm) and total metal
concentrations (mg/L) in the streams of Ecoregions 69 and 70 in West Virginia.
All variables are logarithm transformed. The smooth lines represent the locally
weighted scatter plot smoothing (LOWESS) lines (span = 2/3).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 39 DRAFT—DO NOT CITE OR QUOTE
-------
Conduct
- - Ablp
-MR
-
3 5 7
6 8 10
0.51
PH
0.39
0.31
Temp
-2 4
0.26
0.2
0.29
Fecal
o 15 30
0.26
0.25
0.35
0.12
Watshed
-2 2 4
60 140
0.25
0.07
0.22
0.25
0.1
RBP
0.1B
0.01
0.07
0.12
0.1
0.64
Embed
5 15
_J I L_
0.11
0.14
0.49
0.11
0.14
0.13
DO
0 10 20
0.05
0.06
0.12
0.05
0.03
0.09
0.07
TP
-4 0
I I I
TT M I
-4 -2 0
0.08
0.03
0.08
0.02
0.02
0.22
0.11
0.04
0.15
N023
Figure lie. Other water quality parameters. Matrix of scatter plots and absolute
Spearman correlation coefficients between environmental variables in the streams of
Ecoregions 69 and 70 in West Virginia. The smooth lines are locally weighted scatter plot
smoothing (LOWESS) lines (span = 2/3). Conduct is logarithm transformed specific
conductance ([j,S/cm); Temp is water temperature (°C); RBP is Rapid Bioassessment
(Habitat) Protocol score (possible range from 0 to 200); Fecal is logarithm transformed
fecal coliform bacteria count (per 100 mL water); Watershed is logarithm transformed
watershed area (km2); embeddedness is a parameter score from the Rapid Bioassessment
Protocol (possible range from 0 to 20); DO is dissolved oxygen (mg/L); TP is logarithm
transformed total phosphorus (mg/L); N023 is logarithm-transformed nitrate and nitrite
(mg/L).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 40 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX A
CAUSAL ASSESSMENT
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
ABSTRACT
Because associations in the field are not necessarily causal, this appendix reviews the
evidence that salts are a cause of impairment of aquatic macroinvertebrates in streams in
Ecoregions 69 and 70 of West Virginia. The goal is to establish that salts are a general cause,
not that they cause all impairments, nor that there are no other causes of impairment, nor that
they cause the impairment at any particular site. The evidence is organized in terms of six
characteristics of causation. The inferential approach is to weigh the body of evidence, as is
done in epidemiology. The results are positive; the available evidence indicates that salts, as
measured by conductivity, are a common cause of impairment of aquatic macroinvertebrates in
the region of concern. The following appendix (B) addresses the potential for other variables to
confound the effects of salts.
A.l. INTRODUCTION
To assure that that the association of conductivity with the extirpation of aquatic taxa
reflects a causal relationship, we use epidemiological arguments. The most widely accepted
epidemiological approach was first used to show that smoking causes cancer in humans (Hill,
1965; U.S. DHEW, 1964). It consists of weighing the available evidence on the basis of causal
considerations. As in the case of tobacco smoke, conductivity represents a mixture, and its
effects are not necessarily immediately apparent following exposure. Hill's approach for
establishing a probable causal relationship has been adapted for ecological applications (Fox,
1991; U.S. EPA, 2000; Suter et al., 2002; Cormier et al., 2010). We rely on the same approach
to demonstrate that mixtures of ions that elevate conductivity in streams in the Mountain and
Plateau Regions of Central Appalachia are causing local extirpation of species.
The causal characteristics used in this assessment are described in Cormier et al. (2010)
(see Table A-l). Each causal characteristic is defined and related to Hill's considerations and to
the types of evidence in the Stressor Identification (SI) Guidance (U.S. EPA, 2000) and the
Causal Analysis/Diagnosis Decision Information System (CADDIS) Web site
(http://www.epa.gov/caddis). The SI and CADDIS types of evidence indicate the types of
information that are potentially available to demonstrate characteristics of causation from
Cormier et al. (2010). Hill's considerations are a mixture of types of evidence, sources of
information, and quality of information, but they are included because they are traditional.
A.l.l. Assessment Endpoints
This causal assessment evaluates whether the aqueous salinity, as measured by
conductivity, is capable of causing local extirpation of stream biota in an area of the Central
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 42 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Appalachia including Ecoregions 69 (Central Appalachia) and 70 (Western Alleghany Plateau)
(Woods et al., 1996). These regions include parts of the states of Ohio, Pennsylvania, Maryland,
West Virginia, Kentucky, and Tennessee. The entities of concern are benthic invertebrates,
possibly including rare and threatened species. The effect is local extirpation from streams in
their natural range. Depending on the type of evidence, different biological measurement
endpoints are used. In particular, the number of ephemeropteran genera is used in many of the
quantitative analyses, because most of the sensitive genera are Ephemeroptera and the number of
genera is a good summary of the consequences of extirpation. However, the assessment is of
general causation in the regions of concern, not for any specific genus or location.
A.1.2. Data Sets
The same data sets used in the derivation of the aquatic life benchmark were used in the
causal assessment, particularly the West Virginia Department of Environmental Protection's
(WVDEP's) WABase. In addition, evidence was drawn from the literature involving laboratory
studies, a data set from U.S. Environmental Protection Agency (U.S. EPA) Region 3 described in
Pond et al. (2008a), and geographic information, and related information described in
Appendix F.
A.1.3. Weighting
The evidence is weighted using a system of plus (+) for supporting conductivity as a
cause, minus (-) for weakening and zero (0) for no effect. (Both neutral evidence and
ambiguous evidence have no effect on the inference.) One to three plus or minus symbols are
used to indicate the weight of a piece of evidence.
Note that these scores are for particular pieces of evidence, not for causation as a whole.
For example, a particular study may convincingly demonstrate that a source exists that is
associated with elevated conductivity in the region, so it is scored + + +, but alone it is not
convincing evidence that conductivity causes extirpation of biota.
Any relevant evidence receives a single plus, minus, or zero to register the evidence and
to indicate a decreased or increased support for a causal relationship (see Table A-2). The
+ + + or —
+ + or —
+ or -
0
Convincingly supports or weakens
Strongly supports or weakens
Somewhat supports or weakens
No effect
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 43 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
strength of evidence is considered next. The strength of a relationship is indicated by the
magnitude of a measure of association (e.g., a correlation coefficient) or the number of
relationships that display the causal characteristic. After strength, the other possible unit of
weight is assigned depending on causal characteristic and on the type of evidence. Additional
considerations that may result in a higher score are presented in Table A-3.
A.2. EVIDENCE OF CHARACTERISTICS OF CAUSATION
A.2.1. Co-occurrence
Because causation requires that causal agents interact with unaffected entities; they must
co-occur in space and time. Co-occurrence corresponds to Hill's consistency, Si's
co-occurrence, and CADDIS's co-occurrence in space and time.
A.2.1.1. Correlation of Cause and Effect
In the Watershed Assessment Branch Data Base (WABbase), conductivity and the
number of ephemeropteran genera were moderately correlated (r = -0.63) (see Figure A-l).
This relationship holds even when elevated levels of potential alternative causes (confounders)
are removed (see Figure A-2). In the data set created by Pond et al. (2008a), ephemeropteran
genera and conductivity were highly correlated (r = -0.90).
A.2.1.2. Contingency Table
We constructed a contingency table of the presence of Ephemeroptera at sites near
background conductivity (<200 (j,S/cm) and higher conductivities (>1,500 (j,S/cm) and recorded
the ratio of presence or absence of mayflies (see Table A-4). It shows that mayflies co-occur
with low conductivity but that all mayfly species are absent from more than 3/4 of sites where
conductivity is high. This analysis supplements the correlations by emphasizing the difference
between high and low conductivity sites with respect to a clear endpoint, the absence of all
Ephemeroptera.
We also compared the number of genera at sites with lower conductivities (<200 (j,S/cm)
and higher conductivities (>1,500 (j,S/cm) with and without the co-occurrence of other
parameters that are somewhat correlated with conductivity or are known biological stressors (see
Tables B-l, B-2, B-3, and B-4). Whatever the level of the other parameter, when conductivity
was low, Ephemeroptera were well represented and occurred less often or not at all at high
conductivity. Hence, the potentially confounding agents were not responsible for the observed
co-occurrence of conductivity and biological impairments. Other analyses of potential
confounders are described in Appendix B on confounding.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 44 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
A.2.1.3. Co-occurrence in Paired Watersheds Over Time
Conductivity is shown to increase after the construction of valley fill coal mining
operations, and the number of ephemeropteran genera is low relative to a paired unmined
watershed (see Table A-5).
A.2.1. Summary. In sum, when conductivity is low, the number of genera is high. Even
when other stressors are absent, where conductivity is high, the number of genera is low (see
Figure A-2). The evidence for co-occurrence of conductivity with biological effects is strong,
relevant, consistent, and of high quality and is, therefore, conclusive (see Table A-6).
A.2.2. Preceding Causation
Each causal relationship is a result of a web of preceding cause and effect relationships
that begin with sources and include pathways of transport, transformation, and exposure.
Evidence of sources of a causal agent increases confidence that the causal event actually
occurred and was not a result of a measurement error, chance, or hoax (Bunge, 1979). Although
preceding causation was not recognized by Hill, it corresponds to a type of evidence in the
U.S. EPA's SI and CADDIS process, causal pathway.
A.2.2.1. Complete Source to Cause Pathway
Because exposure to aqueous salts does not require transport or transformation (i.e.,
organisms are directly exposed to salts in water immediately below sources), only evidence of
the occurrence of sources is relevant. Potential sources of increased conductivity in the region
include surface and underground coal mining, effluent from coal preparation plants and
associated slurry impoundments, effluent from coal fly ash impoundments, winter road
maintenance, brines from natural gas and coalbed methane drilling operations, treatment of waste
water, human and animal waste, scrubbers at coal fired electric plants, and demineralization of
crushed rock (Ziegler et al., 2007). The ionic composition of these waters is not uniform (see
Table A-7). In particular, bicarbonate and sulfate are the dominant anions in streams at mined
and unmined sites, but Marcellus Shale brine is almost entirely chloride salts. However, only
four sites were found to have elevated conductivity with high chloride and low sulfate, so shale
brines are rarely the sole dominant source of conductivity (see Section 2.3).
A.2.2.2. Evidence from Literature
High conductivity leachate has been shown to flow from valley fills created during coal
mining operations (Bryant et al., 2002; Merricks et al., 2007). In contrast, conductivity increases
only slightly following clear-cutting and burning. Dissolved mineral loading may be increased
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 45 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
slightly by harvesting but also declines quickly as vegetation re-establishes (Swank and
Douglass, 1977). Golladay (1988) and Arthur et al. (1998) found increases in nitrogen and
phosphorus export in logged catchments in the Appalachians but minor differences in calcium,
potassium, or sulfate concentrations between logged and undisturbed watersheds. Likens et al.
(1970) actually found sulfate concentrations to decrease following clear cutting and experimental
suppression of forest growth by herbicides.
A.2.2.3. Co-occurrence of Sources and Conductivity
Conductivity increases where surface mining operations occur in a watershed and not in
an adjacent unmined watershed (see Table A-5) and are higher overall in mined watersheds with
valley fill than in unmined watersheds (see Table A-7).
A.2.2.4. Characteristic Composition
Correlation and regression analyses suggest that, in Ecoregions 69 and 70, conductivities
above 500 [j,S/cm contain high levels of the ions of Ca2+, HC03 , Mg2+, and S042 (see
Figure 10a~b) which is consistent with large scale surface coal mining and valley fill sources
(Pond et al., 2008a). In contrast, the dominant ions of municipal waste water and of Marcellus
Shale brine are Na+ and CI , which rarely dominate conductivity in those regions (see Section 2.3
and Table A-7). Therefore, the causal assessment relates primarily to mixtures of salts typical of
alkaline coal mine drainage and associated valley fill discharges.
A.2.2.5. Correlation of Conductivity with Sources
Scatter plots of conductivity levels were generated for seven land cover classifications:
open water, agriculture, residential, barren, valley fill, abandoned mine lands, and forested (see
Appendix F for methods). From 2,151 sites in Ecoregion 69D described in the WVDEP
WABbase, 191 <20 km2 watersheds were found for which there were macroinvertebrate samples
identified to the genus level with at least one chemistry sample and TMDL land cover
information. Small <20 km2 subwatersheds were selected to reduce confounding from multiple
sources. These subwatersheds drained into the Coal, Upper Kanawha, Gauley, and New Rivers
(see Figure F-2. Land use and land cover were generated from publically available databases
(see Appendix F). Land use and land cover were arc sine square root transformed to better
depict the upper and lower portions of the distribution. Scatter plots and Spearman rank
correlations of six land use categories and conductivity are shown in Figure A-3.
Although conductivity typically increases with increasing land use (Herlihy et al., 1998),
the densities of agricultural and urban land cover were relatively low, and a clear pattern of
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 46 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
increasing conductivity and increasing land use is not seen. At relatively low urban land use, the
range of conductivity is highly variable (see Figure A-3). This may be caused by unknown mine
drainage, deep mine break-outs, road applications, poor infrastructure condition (e.g., leaking
sewers or combined sewers), or other practices. In contrast, there is a clear pattern of increasing
conductivity as percent area in valley fill increases, and decreasing conductivity with increasing
forest cover (see Figure A-3).
A.2.2. Summary. In sum, large scale surface mining and associated valley fills constitute a
common and long-term source of high conductivity water in this region. The evidence for
this source is abundant and of high quality. Hence, the evidence of preceding causation
leading to high conductivity is conclusive (see Table A-8).
A.2.3. Interaction and Physiological Mechanisms
Causal agents alter affected entities by interacting with them through a physical
mechanism. Evidence that a mechanism of interaction exists for a proposed causal relationship
strengthens the argument for that relationship. This characteristic corresponds to Hill's
plausibility, Si's mechanism, and CADDIS's mechanistically plausible cause.
A.2.3.1. Mechanism of Exposure
Aqueous salts are dissolved ions that are readily available for uptake by aquatic
organisms as they pass over their respiratory surfaces.
A.2.3.2. Mechanism of Effect
The internal fluids of freshwater organisms are saltier than the water in which they live.
As a result, freshwater organisms must use many physical structures and physiological
mechanisms to maintain a balance of water content and ionic content. To maintain the balance
of ions, they excrete hypotonic urine; possess impermeable scales, cuticles or exoskeletons; and
use semipermeable membranes to redistribute ions (Bradley, 2009, Evans, 2008a, b; Wood and
Shuttleworth, 2008; Thorp and Covich, 2001). Other methods of absorption include rectal
pumping of water in Odonates and drinking by Megalopterans and Coleopterans. Anion, cation,
and proton transport include passive, active, uniport, and co-transport (Nelson and Cox, 2005).
Many freshwater invertebrates have chloride cells that actively take up chloride and other ions
through gills (Komnick, 1977; Bradley, 2009). Members of the orders Ephemeroptera,
Plecoptera, and Heteroptera have chloride cells on their body surfaces. Some dipterans and a
few Trichopterans have chloride epithelia, and anal papillae are present on other members of
these orders.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 47 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Numerous specific mechanisms are involved in the toxicity of high-conductivity
solutions; one is discussed here for the sake of illustration. The ion regulation system includes
antiport anion exchange proteins that co-transport CI against the concentration gradient into the
cell simultaneously with HCO3" down the concentration gradient and out of the cell (Nelson and
Cox, 2005; Bradley 2009). If external HCO3 is high, the gradient is not favorable for CI uptake
(Avenet and Lingnon, 1985).
Some physiological processes are especially dependent on proper ionic balance. These
include nerve conduction, muscle contraction, and secretion. Reproduction, including
fertilization, polymerization of egg mass coverings, and embryonic development depend on ionic
balance, graphically illustrated by the swelling of fish eggs upon fertilization (Tarin et al., 2000).
At the organismal level, effects of aqueous salts on aquatic arthropods include mortality (Kefford
et al., 2003, 2005a) and reduced growth, reproduction, and hatching success (Clark et al., 2004a;
Hassell et al., 2006; Kefford and Nugegoda, 2005; Kefford et al., 2004, 2006, 2007; Nielsen et
al., 2003; Brock et al., 2005). These effects strongly suggest that population density can be
reduced over generations of persistent exposure to elevated conductivity (Zalizniak et al., 2007).
A.2.3. Summary. In sum, aquatic organisms are directly exposed to aqueous salts, and the
relative amounts and concentration of salts may exceed the capacity of organisms to regulate
their internal ionic composition. The importance of osmoregulation and ionic homeostasis has
been demonstrated in diverse animal models with results published in the peer-reviewed
literature. The evidence is drawn from a long history of physiological investigations (see
Table A-9).
A.2.4. Specific Alteration
A specific cause induces a specific effect in particular receptors. This alteration is
obscured in many studies by broad definitions of causes and effects, but, when a specific effect
of a cause is characterized, it strengthens the evidence for a causal relationship. If the specific
effect of a cause occurs with no other causes, it can be diagnostic of that cause. This
characteristic corresponds to specificity in Hill's considerations and in the Si's types of evidence,
and to symptoms in CADDIS.
A.2.4.1. Specificity of Genera
In a paper focusing on mayflies, principal component analysis sorted mined and
residential sites from reference sites primarily on the basis of specific conductance and pH
(Pond, 2009). In the same study, a nonmetric multidimensional scaling model strongly
associated Ephemerella, Drunella, Cinygmula, Epeorus, and Ameletus with the low conductivity
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 48 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
reference sites and no mayflies or Caenis, Baetis, Isonychia, and Stenonema with the higher
conductivity sites. The first group has XC95 <600 [j,S/cm, and the second group of tolerant
mayflies had XC95 >729 [j,S/cm.
The derivation of 95th percentile extirpation concentration (XC95) values and species
sensitivity distributions in this document demonstrated that a characteristic set of genera
(primarily Ephemeroptera) were extirpated at relatively low conductivities and others were
resistant. The relative sensitivities are consistent with the findings of Pond et al. (2008a) and
with analyses of data from Kentucky (see Appendix E). This is not meant to suggest that
conductivity is the only possible cause of loss of these genera. Rather, it indicates that the loss of
those genera consistently occurs where conductivity is elevated. If a random set of genera were
lost, it might suggest that various causes were acting that co-occur with elevated conductivity,
but that was not the case.
A.2.4.2. Specificity of Assemblages
Using an independent data set collected in West Virginia, nonmetric multidimensional
scaling was applied to biological metrics, and sites were sorted into distinct ordination space
characterized by low, medium, and highly elevated conductivities associated with surface mines
with valley fill (Pond et al., 2008a).
A.2.4. Summary. In sum, some genera are sensitive to conductivity, and others are not. The
evidence for effects specific to high conductivity is reasonably strong, relevant, consistent,
and of high quality and is, therefore, supportive (see Table A-10).
A.2.5. Sufficiency
For an effect to occur, sufficiently susceptible entities must experience a sufficient
magnitude of exposure. This characteristic corresponds to biological gradient in Hill's
considerations. In SI and CADDIS, multiple types of evidence may demonstrate sufficiency
including stressor-response in the field, laboratory tests of site media, manipulation of exposure
and stressor-response from laboratory studies.
A.2.5.1. Laboratory Tests of Defined Ion Mixtures
Mount et al. (1997) tested the acute lethality of several mixtures of salts to two planktonic
crustaceans (Ceriodaphnia dubia and Daphnia magna) and a fish (Pimephalespromelas). A
mixture of K2SO4 and KHCO3 salts was the most toxic combination of salts tested in the study.
The 48-hour LC50 for Ceriodaphnia with K2SO4 and KHCO3 corresponds to 438 [j,S/cm. The
96-hour LC50 for Pimephales promelas also with K2SO4 and KHCO3 corresponds to
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 49 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
1,082 [j,S/cm. The ion matrix of alkaline mine drainage normally contains little K+, and instead,
Ca2+ and Mg2+ are the dominant cations. Conductivity measurements below some valley fills
were greater than 4,000 [j,S/cm. This study demonstrates that mixtures of specific salts can be
acutely lethal at concentrations corresponding to conductivities measured in the region. The
Mount et al. (1997) model has been used to estimate that salt mixtures in some streams below
valley fills are sufficient to cause acute lethality in Ceriodaphnia (U.S. EPA, 2009). However,
these tests are marginally relevant. The crustaceans are not taxonomically similar to the
invertebrate species that are affected, the 48-hour test durations are far shorter than the life-cycle
exposures in the field, and the effect (acute lethality) is unlikely to be the cause of
population-level effects in the field. Life-cycle effects on local insects are likely to occur at
much lower levels of conductivity (U.S. EPA, 2009). However, these tests do indicate that the
ion mixture could be toxic to common surrogate laboratory organisms used to evaluate toxicity.
A.2.5.2. Laboratory Tests of Mine Discharges
Kennedy et al. (2003, 2004, 2005) tested coal mine discharge waters in Ohio with
Ceriodaphnia dubia and a mayfly (Isonychia bicolof). In 7-day lethality tests, the mayfly was
about three times as sensitive as the crustacean. Lowest observed effect concentrations (LOECs)
for survival of mayflies (mid to late-instars) at 20°C occurred at 1,562, 966, and 987 [j,S/cm in
three tests. These values bracket the Isonychia XC95 of 1,177 [j,S/cm. Ceriodaphnia tests with
simulated effluent containing only major ions showed that the toxicity of this effluent was not
due to heavy metals or Se (Kennedy et al., 2005).
Echols et al. (2009) performed 10-14 day toxicity tests of coal processing effluent from
Virginia with Isonychia bicolor. They obtained LOEC values for survivorship in three tests of
1,508 to 4,101 [j,S/cm. The lower toxicity of these waters may be due to the dominance of
sodium, which has the lowest toxicity of the common cations (Mount et al., 1997). In any case, it
is not surprising that these acute lethality tests yield higher conductivity levels than the Isonychia
XC95 (1,177 (j,S/cm) which is a result of full life-cycle exposures and effects. In particular, in the
final test which yielded the lowest LOEC, survival at the end of the test was approximately 25%
and still declining. The effluent contained no detectable toxic trace metals or metalloids except
selenium (8.5 (J,g/L), so the authors stated that the toxicity was likely due to salinity.
A.2.5.3. Laboratory Tests of Ambient Waters
Waters from below valley fills in the region of concern were tested by Merricks et al.
(2007). Ceriodaphnia dubia LC50 values in 48-hour tests were established for some but not all
waters from Lavender Fork with undiluted concentrations of 2,497-3,050 [j,S/cm. These tests
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 50 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
used relevant mixtures of ions, but the test species, duration and endpoint have low relevance
and are likely to underestimate toxicity in the field.
A.2.5.4. Field Exposure-Response Relationships
As Hill suggested, a biological gradient in the field suggests that the exposures reach
levels that are sufficient to cause effects. Pond et al. (2008a, b) reported that the number of taxa
decreases as conductivity increases or as the amount of surface mining and associated valley fills
increases. Analyses conducted for this report using the WABbase data sets show that as
conductivity increases the total number of genera and the number of ephemeropteran genera
decrease at conductivity levels shown to extirpate sensitive genera (see Figure A-l). This
analysis shows not only the co-occurrence of elevated conductivity and loss of stream biota but
also that there is a regular exposure-response relationship that extends to the lowest observed
concentrations (evidence of sufficiency). The same data set was also modeled by partitioning for
potential confounding parameters. Streams with higher temperatures (>22°C), low pH (<6), poor
habitat (<135) and high fecal coliform (>400 colonies/100 mL) were excluded. The effect of
conductivity was still strong (see Figure A-2). Also, the distributions of individual genera show
that as conductivity increases the occurrence and capture probability decreases for many genera
(see Appendix D).
A.2.5.5. General Knowledge
The susceptibility of organisms and communities is a function of genetic, evolutionary,
developmental, and physiological legacies. Numerous studies have characterized the
disappearance of freshwater taxa with increasing salt concentration (Remane, 1971; Wetzel,
2001). Species native to the Mid-Atlantic Highlands (Ecoregions 69 and 70) have evolved for
very low conductivity water and are expected to decline as salinity increases above background.
Conductivities below 70 [j,S/cm were common in forested areas.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 51 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
A.2.5. Summary. In sum, the available evidence suggests that conductivities in the region of
concern reach levels that are sufficient to cause effects on stream communities (see
Table A-l 1). Most laboratory toxicity tests have been conducted with species, effects and
exposures that have low relevance and sensitivity to salinity. However, they still show that
ambient salinities observed in the regions of concern can cause severe effects. More to the
point, saline coal mine effluents from the region are lethal to a mayfly species at
conductivities similar to its extirpation concentration. Correlations of conductivity and stream
biological metrics confirm that conductivity is strongly associated with gradients of biological
response down to the levels where sensitive genera are extirpated. These relationships are
strong even when other stressors were present. Finally, general studies of the effects on
aquatic organisms of changes in salinity suggest that the observed magnitude of increases in
salinity in these regions is sufficient to cause the extirpation of some species, but the studies
were not conducted at levels as low as those occurring in this region.
A.2.6. Time Order
Logically, a causal event occurs before an effect is observed. Evidence of time order is
provided by changes in the invertebrate assemblages after the introduction of a source that
increased conductivity. This characteristic corresponds to temporality in Hill's considerations
and in the SI types of evidence and to temporal sequence in CADDIS.
We could not obtain conductivity and biological survey data for before and after a valley
fill or other source of saline effluents began operation. Hence, this characteristic of causation is
scored No Evidence (NE).
A.2.7. Evaluation of the Body of Evidence
Conclusions concerning causality are based on the weight of evidence from all types of
evidence that support or weaken all of the characteristics of causation. The property of the body
of evidence was termed coherence by Hill. In SI and CADDIS, it is divided into two
considerations: the consistency of evidence and the coherence of evidence (i.e., the
reasonableness of explanations of any inconsistencies in the evidence).
This causal assessment found that the available evidence supports a causal relationship
between mixtures of matrix ions in streams of Ecoregions 69 and 70 and biological impairments.
That conclusion is based on evidence showing that the relationship of conductivity to the loss of
aquatic genera has the characteristics of causation.
1. Co-occurrence—The loss of genera occurs where conductivity is high even when
potential confounding causes are low but is rare when conductivity is low (+ + +).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 52 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
2. Preceding Causation—Sources of conductivity are present and are shown to increase
stream conductivity in the region (+ + +).
3. Interaction—Aquatic organisms are directly exposed to dissolved salts. Physiological
studies over the last 100 years have documented the many ways that physiological
functions of all organisms are affected by excess salt or the combinations of ions for
which they do not have physiological capacity or mechanisms to regulate (+).
4. Alteration—Some genera and assemblages are affected at sites with higher conductivity
while others are not. These differences are characteristic of high conductivity (+ +).
5. Sufficiency—Increased exposure in both concentration and duration to salt affects
invertebrates based on both field and laboratory analyses (+ + +).
6. Time order—Conductivity increases, and local extirpation occurs after mining permits
are issued, but before and after data are not available (NE).
Other potential causes of the loss of genera in the region include elevated temperatures
associated with loss of shade or increased impervious surfaces, siltation from various land use
activities, low pH from atmospheric deposition and abandoned mines, aluminum toxicity from
abandoned mines, and nutrient enrichment from various sources. Se toxicity has also been
implicated. When these causes are minimized, a relationship between conductivity and mayfly
richness is still evident (see Appendix B).
This causal assessment does not attempt to identify the constituents of the mixture that
account for the effects. Constituents of the mixture in neutral and somewhat alkaline waters that
increase as conductivity increases are all considered as contributing to the local extirpation of
genera in the region of concern. The dominant ions include HC03 , S042 , Ca2+, K+, and Mg2+.
REFERENCES
Arthur, MA; Coltharp, GB; Brown, DL. (1998) Effects of best management practices on forest streamwater quality
in eastern Kentucky. J Amer Water Resour Assoc 34(3):481-495.
Avenet, P; Lignon, JM. (1985) Ionic permeabilities of the gill lamina cuticle of the crayfish, Astacus
leptodactylus (E). J Physiol 363:377-401.
Bradley, TJ. (2009) Animal Osmoregulation. New York:Oxford University Press; pp. 149-151.
Brock, MA; Nielsen, DL; Crossle, K. (2005) Changes in biotic communities developing from freshwater wetland
sediments under experimental salinity and water regimes. Freshwater Biol 50:1376-1390.
Bryant, G; McPhilliamy, S; Childers, H. (2002) A survey of the water quality of streams in the primary region of
mountaintop / valley fill coal mining, October 1999 to January 2001. In: Draft programmatic environmental impact
statement on mountaintop mining / valley fills in Appalachia - 2003. Appendix D. U.S. Environmental Protection
Agency, Region 3, Philadelphia, PA. Available online at
http://www.epa.gOv/Region3/mtntop/pdf/appendices/d/stream-chemistry/MTMVFChemistryPartl.pdf.
Bunge, M. (1979) Causality and modern science. Third revised edition. New York: Dover Publications, Inc.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 53 DRAFT—DO NOT CITE OR QUOTE
-------
1 Clark, TM; Flis, BJ; Remold, SK. (2004a) Differences in the effects of salinity on larval growth and developmental
2 programs of a freshwater and a euryhaline mosquito species (Insecta: Diptera, Culicidae). J Exp Biol
3 207:2289-2295.
4 Cormier, SM; Suter GW, II; Norton, SB. (2010) Causal characteristics for ecoepidemiology. Hum Ecol Risk Assess
5 16(l):53-73.
6 Echols, B; Currie, R; Cherry, DS. (2009) Preliminary results of laboratory toxicity tests with the mayfly, Isonychia
7 bicolor (Ephemeroptera: Isonychiidae) for development as a standard test organism for evaluating streams in the
8 Appalachian coalfields of Virginia and West Virginia. Environ Monitor Assess, published online 04 November
9 2009.
10 Evans, DH. (2008a) Teleost Fish Osmoregulation: What have we learned since August Krogh, Homer Smith, and
11 Ancel Keys? Am J Physiol Regul Integr Comp Physiol 295: R704-R713.
12 Evans, DH. (2008b) Osmotic and Ionic Regulation: Cells and Animals. Boca Raton: CRC Press.
13 Fox, GA. (1991) Practical causal inference for ecoepidemiologists. J Toxicol Environ Health 33(4):359-374.
14 Golladay, SW. (1988) The effects of forest disturbance on stream stability. Dissertation. Virginia Polytechnic
15 Institute and State University, Blacksburg, Va.
16 Hassell, KL; Kefford, BJ; Nugegoda, D. (2006) Sub-lethal and chronic lethal salinity tolerance of three freshwater
17 insects: Cloeon sp. and Centroptilum sp. (Ephemeroptera: Baetidae) and Chironomus sp. (Diptera: Chironomidae).
18 J Exp Biol 209:4024-4032.
19 Herlihy, AT; Stoddard, JL; Johnson, CB. (1998) The relationship between stream chemistry and watershed land
20 cover data in the mid-Atlantic region, U.S. Water, Air, and Soil Pollution 105:377-386.
21 Hill, AB. (1965) The environment and disease: Association or causation. Proceed Royal Soc Med 58:295-300.
22 Kefford, BJ; Papas, PJ; Nugegoda, D. (2003) Relative salinity tolerance of macroinvertebrates from the Barwon
23 River, Victoria, Australia. Mar Freshwater Res 54:755-765.
24
25 Kefford, BJ; Dalton, A; Palmer, CG; Nugegoda, D. (2004) The salinity tolerance of eggs and hatchlings of selected
26 aquatic macroinvertebrates in south-east Australia and South Africa. Hydrobiologia 517:179-192.
27 Kefford, BJ; Nugegoda, D. (2005) No evidence for a critical salinity thresholds for growth and reproduction of the
28 freshwater snail Physa acuta. Environ Pollut 54:755-765.
29 Kefford, BJ; Zalizniak, L; Nugegoda, D. (2006) Growth of the damselfly Ischnura heterosticta is better in saline
30 water than freshwater. Environ Pollut 141:409-419.
31
32 Kefford, BJ; Nugegoda, D; Zalizniak, L; Fields, EF; Hassell, KL. (2007) The salinity tolerance of freshwater
33 macroinvertebrate eggs and hatchlings in comparison to their older life-stages. Aquat Ecol 41:335-348.
34 Kennedy, AJ; Cherry, DS; Currie, RJ. (2003) Field and laboratory assessment of a coal processing effluent in the
35 Leading Creek Watershed, Meigs County, Ohio. ArchEnviron Contam Toxicol 44(3):324-331.
36 Kennedy, AJ; Cherry, DS; Currie, RJ. (2004) Evaluation of ecologically relevant bioassays for a lotic system
37 impacted by acoal-mine effluent, using Isonychia. Environ Monit Assess 95:37-55.
3 8 Kennedy, AJ; Cherry, DS; Zipper, CE. (2005) Evaluation of ionic contribution to the toxicity of a coal-mine effluent
39 using Ceriodaphnia dubia. Arch Environ Contam Toxicol 49:155-162.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 54 DRAFT—DO NOT CITE OR QUOTE
-------
1 Komnick, H. (1977) Chloride cells and chloride epithelia of aquatic insects. Int Rev Cytol 49:285-328.
2 Likens, GE; Bormann, FH; Johnson, NM; et al. (1970) Effects of forest cutting and herbicide treatment on nutrient
3 budgets in the Hubbard Brook watershed-ecosystem. Ecol Monogr 40:23-47.
4 Merricks, TC; Cherry, DS; Zipper, CE; et al. (2007) Coal-mine hollow fill and settling pond influences on
5 headwater streams in southern West Virginia, USA. Environ Monit Assess 129(l-3):359-378.
6 Mount, DR; Gulley, DD; Hockett, R; et al. (1997) Statistical models to predict the toxicity of major ions to
7 Ceriodaphnia dubia, Daphnia magna, and Pimephalespromelas (fathead minnows). Environ Toxicol Chem
8 16(10):2009-2019.
9 Nelson, D; Cox, M. (2005) Lehninger principles of biochemistry. 4th edition. New York: W. H. Freeman & Co.;
10 pp. 395-397.
11 Nielsen, DL; Brock, M; Crossle, K; Harris, K; Healey, M; Jarosinski, I. (2003) The effects of salinity on aquatic
12 plant germination and zooplankton hatching from two wetlands sediments. Freshwater Biol 48:2,214-2,223.
13 Pond, GJ. (2010) Patterns of Ephemeroptera taxa loss in Appalachian headwater streams (Kentucky, USA).
14 Hydrobiologia. 641(1):185—201.
15 Pond, GJ; Passmore, ME; Borsuk, FA; et al. (2008a) Downstream effects of mountaintop coal mining: comparing
16 biological conditions using family- and genus-level macroinvertebrate bioassessment tools. J N Am Benthol Soc
17 27:717-737.
18 Pond, GJ; Bailey, JE; Lowman, B. (2008b). West Virginia GLIMPSS (genus-level index of most probable stream
19 status): abenthic macroinvertebrate index of biotic integrity for West Virginia's wadeable streams. West Virginia
20 Department of Environmental Protection, Division of Water and Waste Management, Watershed Assessment
21 Branch. Charleston, WV.
22 Remane, A. (1971) Ecology of brackish water. In: Remane A; Schlieper, C, eds. Biology of brackish water. New
23 York, NY: John Wiley and Sons.
24 Suter, GW, II; Traas, T; Posthuma, L. (2002) Issues and practices in the derivation and use of species sensitivity
25 distributions. In: L. Posthuma, L; Suter, GW, II; Traas, T; eds. Species sensitivity distributions in ecotoxicology,.
26 Boca Raton, FL: Lewis Publishers, pp 437-474.
27 Swank, WT; Douglass, JE. (1977) Nutrient budgets for undisturbed and manipulated hardwood ecosystems in the
28 mountains of North Carolina. In: Correll, DL; ed. Watershed research in eastern North America: A workshop to
29 compare results. Edgewater, MD: Smithsonian Institution Press; pp 343-363.
30 Tarin, JJ; Cano, C. Tarbin, JJ. (2000) Fertilization in protozoa and metazoan animals: Cellular and molecular
31 aspects. Berlin: Springer-Verlag.
32 Thorp, JH; Covich, AP; editors. (2001) Ecology and classification of North American freshwater invertebrates.
3 3 Academic Press, New York.
34 U.S. DHEW (Department of Health Education and Welfare). (1964) Smoking and Health: Report of the Advisory
3 5 Committee to the Surgeon General., Washington, D.C. Public Health Service Publication 1103.
36 U.S. EPA. (Environmental Protection Agency). (2000) Stressor identification guidance document. Washington, DC.
37 EPA/822/B-00/025.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 55 DRAFT—DO NOT CITE OR QUOTE
-------
1 U.S. EPA (Environmental Protection Agency). (2009) The Effects of Mountaintop Mines and Valley Fills on
2 Aquatic Ecosystems of the Central Appalachian Coalfields. Office of Research and Development, National Center
3 for Environmental Assessment, Washington, DC. EPA/600/R-09/138A.
4 Wetzel, RG. (2001) Limnology. 3rd edition. San Diego, California: Academic Press.
5 Wood, CM; Shuttleworth, TJ. (2008) Cellular and molecular approaches to fish ionic regulation. Vol 14: Fish
6 Physiology. San Diego, CA: Academic Press, Inc.
7 Woods, AJ; Omernik, JM; Brown, DD; et al. (1996) Level III and IV ecoregions of Pennsylvania and the Blue
8 Ridge Mountains, the Ridge and Valley, and the Central Appalachians of Virginia, West Virginia, and Maryland.
9 U.S. Environmental Protection Agency, Office of Research and Development, Corvallis, OR; EPA/600R-96/077.
10 Zalizniak, LB; Kefford, BJ; Nugegoda, N. (2007) Effects of pH on salinity tolerance of selected freshwater
11 invertebrates. AquatEcol 43:135-144.
12 Ziegler, CR; Suter, GW, II; Kefford, BJ. (2007) Candidate Cause: Ionic Strength. Available online at
13 www.epalgov/caddis (accessed 10/13/09).
14
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 56 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-l. Definitions of causal characteristics
Characteristic
Description
Co-occurrence
The cause co-occurs with the unaffected entity in space and time
Preceding causation
Each causal relationship is a result of a larger web of cause and effect
relationships
Time order
The cause precedes the effect
Interaction
The cause physically interacts with the entity in a way that induces the
effect
Alteration
The entity is changed by the interaction with the cause
Sufficiency
The intensity, frequency, and duration of the cause are adequate, and the
entity is susceptible to produce the type and magnitude of the effect
Table A-2. Relationships between qualities of evidence and scores for
weighing evidence
Qualities of the evidence
Score, not to exceed three
minus or three plus
Logical implications
+ o,-
Strength
Increase score
Other qualities
Increase score
Table A-3. Other considerations used to weight the evidence concerning the
influence of potentially confounding variables
Quality of evidence
Alternative outcomes
Directness of cause
Proximate cause, sources, or intermediate causal connections
Specificity
Effect attributable to only one cause or to multiple causes
Relevance to effect
From the case or from other similar situations
Nature of the association
Quantitative or qualitative
Independence of association
Independent or confounded
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 57 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-4. Presence of mayflies contingent on stream conductivity (data
from WABbase)
Mayflies present
Mayflies absent
Total
Near background conductivity
889
35
924
(<200 (j,S/cm)
96.2%
3.8%
High conductivity (>1,500 (j,S/cm)
28
101
129
21.7%
78.3%
Total
917
136
1,053
Table A-5. Temporal increase of conductivity 2 years after permitting of
mining operations
Never mined
Ash Fork
Permit 1994,1996
Boardtree Branch
Permit 1996; Stillhouse
1998
2003
2006/07
1998
2003
2007
1998
2003
2007
[j,S/cm
44a
39b
42b/39a
l,396a
3,015b
3,390a
51 la
3,200b
3,970a
% E
27.23
29.21
1.23
0
# E
6
4
2
0
#P
5
6
0
0
# EPT
20
14
5
3
TT
41
24
20
8
aSingle measurement.
bMean value.
E = Ephemeroptera; P = Plecoptera; T = Trichoptera; TT = total taxa.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 58 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-6. Summary of evidences and scores for co-occurrence
Type of evidence
Evidence
Score
Correlation of cause and
effect
Ephemeroptera were correlated with conductivity in
two studies r = -0.63 (see Figure A-l) and r = -0.90.
This is strong quantitative evidence from multiple
studies.
+ + +
Contingency table
The contingency table (see Table A-4) provides
strong quantitative evidence that high conductivity is
strongly associated with severe effects
(Ephemeroptera absent at >75% of sites).
+ +
Co-occurrence in paired
watersheds over time
24% to 100%) difference (see Table A-5) is large and
quantitative.
+ +
Overall score
Relevant, strong, consistent.
+ + +
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 59 DRAFT—DO NOT CITE OR QUOTE
-------
o
S"4
>3*
§•
£
a
to
s
S"
a
a,
Sf
^s
^s
TO
TO'
*
^S
£
s
On ^
O <*>
0
S
1
a.
2 §-
7? ^
J> ^5
nr\ S
H ©
3 §
2 §
O ^
H r?
<*^
hh Oq
H to
W|
o ^
o
H
W
Table A-7. Total cations and anions measured in water originating from surface mined sites with valley fills,
unmined sites, or Marcellus Shale brine. Individual ions are presented as a fraction of the total cations or anions.
For mined sites, n = 13; unmined sites, n = 7; Marcellus Shale brine, n = 3. Measurements of HCO3 and NC^TST were
not available for Marcellus Shale brine sites.
Mined (Valley Fill)
Unmined
Marcellus Shale Brine
Mean
Median
Range
Mean
Median
Range
Mean
Median
Range
Total Cations (mg/L)
282.4
238.9
72.7-515.2
15.7
15.9
7.0-25.6
23,862.0
21,719.0
8,650.0-41,217.0
Ca
0.48
0.48
0.42-0.55
0.46
0.46
0.37-0.63
0.24
0.23
0.20-0.28
Mg
0.42
0.42
0.28-0.51
0.28
0.27
0.22-0.36
0.02
0.02
0.02-0.02
K
0.04
0.04
0.02-0.05
0.11
0.11
0.06-0.18
0.02
0.01
0.005-0.05
Na
0.06
0.03
0.02-0.25
0.15
0.14
0.06-0.24
0.72
0.70
0.69-0.78
Total Anions (mg/L)
926.8
730.4
228.1-1,734.4
44.7
47.2
21.9-66.5
28,296. la
18,620.8a
14,326.3-51,941.33
bHC03
0.25
0.25
0.06-0.48
0.54
0.57
0.34-0.66
NA
NA
NA
CI
0.0076
0.0042
0.0032-0.0036
0.07
0.06
0.04-0.11
0.999
0.999
0.998-0.999
NO3-N
0.0036
0.0031
0.0013-0.011
0.01
0.01
0.002-0.04
NA
NA
NA
S04
0.73
0.74
0.51-0.93
0.38
0.35
0.29-0.51
0.0013
0.0011
0.0011-0.0016
aTotal anions includes only CI and S042 .
HC. O3 converted from measurement of alkalinity as CaC03.
NA = not applicable due to lack of data.
-------
Table A-8. Summary of evidences and scores for preceding causation
Type of evidence
Evidence
Score
Complete source-to-
cause pathway
Sources are present, and no intermediate steps in the
pathway are required.
+
Correlation of
conductivity with
sources
Figure A-3, r = 0.61. This is moderately strong
quantitative evidence from the case.
+ +
Evidence from literature
Multiple publications link conductivity to sources in
the region and eliminate some other land uses as
sources.
+
Co-occurrence of
sources and conductivity
When valley fills are present, conductivity is 12- to
90-fold greater than at unmined sites (see Tables A-5
and A-7). This is strong quantitative evidence from the
case.
+ +
Characteristic
composition
Ambient mixtures of ions have characteristic
compositions that can be associated with particular
sources. Most sites with elevated conductivities have
compositions characteristic of coal mining with valley
fill. This is relevant but quantitatively weak evidence.
+
Overall score
Relevant, strong, consistent.
+ + +
Table A-9. Summary of evidences and scores for interaction and
physiological mechanism
Type of evidence
Evidence
Score
Mechanism of exposure
Salts readily dissolve in water and interact
directly with aquatic organisms.
+
Mechanism of effect
Many mechanistic studies show that
osmoregulation and homeostasis of specific ions
are sensitive to disruption, particularly in
mayflies.
+
Direct evidence
No studies of ionic compensation are available
for organisms in the region.
NE
Overall score
Relevant but not case-specific.
+
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 61 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-10. Summary of evidence and scores for specific alteration
Type of evidence
Evidence
Score
Specificity of genera
Specific genera are consistently sensitive to
conductivity. This quantitative evidence is
independently confirmed.
+ +
Specificity of
assemblage
A model based on specific biology discriminated
effects of conductivity associated with mining.
+
Overall score for
interaction
Relevant, independently confirmed, and consistent,
but only two types of evidence.
+ +
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 62 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-ll. Summary of evidence and scores for sufficiency
Type of evidence
Evidence
Score
Laboratory tests of
defined ion mixtures
The tests were high quality, but the species and durations
have low relevance for determining the conductivity level
at which effects occur, and the effect levels are supportive
only if assumptions are made about acute/chronic and
intertaxa extrapolations.
0
Laboratory tests of mine
discharges
This evidence is relevant in that it comes from nonacid
mine effluents in the region and includes an
Ephemeropteran; but the ionic mixtures were somewhat
different, the effect was lethality and the durations were
short. The results for one set of tests matched the XC95 for
the test genus, but were higher for the other.
+
Laboratory tests of
ambient waters
These tests showed acute lethality to an apparently
resistant species at high conductivity levels. Its relevance
is too low to support or weaken.
0
Field exposure-response
relationships for
Ephemeroptera
This is strong evidence because it is highly relevant, was
obtained independently in two separate data sets, with
moderate to strong correlations. It is not convincing in
itself because of the potential for confounding, which is
treated in Appendix B.
+ +
Field exposure-response
relationships for genera
As conductivity increases, genera no longer are observed.
+ +
General knowledge
General knowledge indicates that salinity can cause the
loss of species but does not indicate that the salinity levels
observed in this case are sufficient.
0
Overall score
The exposure-response relationships in the field, with
some support from laboratory studies, provide positive
evidence that the conductivity levels observed are
sufficient to cause the associated effects.
+ + +
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 63 DRAFT—DO NOT CITE OR QUOTE
-------
d>
o
o
CD
CD
O
O
ID
O
^r
o
CN
: '
% ,-P?
ooo0 8 °
o
~\ I I I I
32 100 316 1000 10000
Conductivity (nS/cm)
CD
B
Q.
O
g>
E
CO
E> O O
K*astx®
¦tHHEKjO-O O
^di||Hoo ©
^^^¦BtaoCLQ-O -
CO
CD
CD
E
CD
CD
Cl
CO
I I I I I T
32 100 316 1000 10000
Conductivity (M-S/cm)
Figure A-1. As conductivity increases, the number of total and
epheineropteran genera decrease.
Data source: WABase
S_9 • •• • ••• •
' ••• 'vf ••~•••••• •
~ - ... ¦»»¦ n mm* * m* •••
* » « •« «4N«« •• «
• » m •• • *r ¦¦ • «• • • • •«
• •• * • '»>«»-»••• «»• •
• • •*« • • •* mn • • •••• ••
~ • ~ ~ wni «¦ iwim»~
• «**• —mt "-¦¦¦ . ww • *
• • « MM • •••«•»•• •• •
• • ••• -«»" ~~
•• • • • • m_ •
• * t » ~ *
~ • H • •
• •• • •
• •
100 316
Conductivity (pS/cm)
~ • • •
M #• «W • Mi
••• rnmm • • ~~ m
» *m • «m
100 316
Conductivity (M.S/cm)
Figure A-2. As conductivity increases, the number of total and
ephemeropteran genera decrease even when potentially confounding
parameters are minimized. (Excluded: streams with higher temperatures
[>22°C], low pH [<6|, poor habitat [400 colonies/100 ml.|).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 64 DRAFT—DO NOT CITE OR QUOTE
-------
I
a
5
3
>5*
ft
Sk
0
2
1
i,
S f*
J> So
nr\ S
H ^
. . *-*.
o
o
a
<^k
5?
-cst
t t 1 1 1 r
ore am o.io &is ojra cs &»
asrrKFl_pd,_Urtiari
n 1 1 1 1 1~
C.DC DOS 0.10 0.15 0.2D 0.25
asHKjt_pct_/iip-
Figure A-3. Conductivity associated with different land uses in 151 watersheds in Ecoregion 69D. There is a
clear pattern of increasing conductivity as percent area in valley fill increased, but no pattern with other land use. From
left to right, they are (a) mountaintop mining-valley fill, (b) abandoned mine lands, (c) mined, (d) barren, (e) forested,
(f) water, (g) urban/residential, and (h) agricultural. Land use and land cover were arc sine square root transformed to
better depict the upper and lower portions of the distribution.
-------
APPENDIX B
CONFOUNDING
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
ABSTRACT
The purpose of this appendix is to determine which, if any, of the variables that may
co-occur with conductivity alter our ability to model the relationship between conductivity and
occurrence of genera. The point was not to determine whether the confounders are general
causes (i.e., can they cause impairments in the region of concern?).
The appendix addresses its purpose in two ways. First, it supports Appendix A by
demonstrating that none of the potential confounders is responsible for the association between
conductivity and biological effects. Second, it supports the development of the benchmark value
by determining whether the confounders have significant influence on the causal relationship
between salts and macroinvertebrate assemblages. The inference was performed by identifying
potential confounders and then determining the occurrence and strength of ten types of evidence
for confounding for each of them. The effect of confounders was found to be minimal and
manageable. Potential confounding by low pH was minimized by removing sites with pH <6
from the data set when calculating the aquatic life benchmark. The influence of Se could not be
evaluated due to poor data and should be investigated. The signal from conductivity was strong
so that other potential confounders that were not strongly influential could be ignored with
reasonable or greater confidence. We do not argue that these variables have no influence, but
their effects are minimal given the streams that would be affected by the aquatic life benchmark.
B.l. INTRODUCTION
The goal of this analysis is not to eliminate confounding variables. They are natural
variables such as temperature and habitat structure that cannot be literally eliminated like
eliminating smokers in an epidemiological study. Nor is the goal to equate the levels of
confounders to an ideal or pristine level. High conductivity effluents do not enter wilderness
streams. Rather, the streams are subject to some level of disturbance. The goals are (1) to define
a set of streams in which the effects of elevated conductivity can be identified without significant
influence by confounding variables, and (2) to estimate conductivity levels that would protect
against the unacceptable effects of salts in those streams (i.e., typical streams receiving high
conductivity effluents in the region of concern).
Because of those goals and the nature of the data, it is not appropriate to use multivariate
statistics to try to eliminate confounders. Multiple regression methods depend on assumptions of
independence, additivity, and normality that are not met. Propensity scores depend on a
counterfactual assumption (i.e., it assumes that the confounders can be different than they are).
This condition is met in propensity score analyses of epidemiological or econometric studies in
which, for example, a cancer patient could be a smoker or not or might live in a city or not. That
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 67 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
counterfactual assumption is often not met by ecological data. In particular, the alkaline ions
that contribute to elevated conductivity also contribute to raising the pH. Therefore, pH and
conductivity are not mechanistically independent, and counterfactual assumptions cannot be
applied. Attempts to statistically eliminate the influence of pH would artificially reduce the
effects of salinity. However, the epidemiological weight-of-evidence approach used here can
make use of the fact that, once we have eliminated acidic sites, the neutral to moderately alkaline
pH levels that remain are not toxic to stream organisms.
Confounding is a bias in the analysis of causal relationships due to the influence of
extraneous factors (confounders). Confounding occurs when a variable is correlated with both
the cause and its effect. The correlations are usually due to a common source of multiple,
potentially causal agents. However, they may be observed for other reasons (e.g., when one
variable is a by-product of another) or due to chance associations.
Confounding may result in identification of a cause that is in fact a noncausal correlate.
That possibility is commonly addressed by applying Hill's (1965) considerations or some
equivalent set of criteria for causation as in Appendix A. This is done because statistics alone
cannot determine the causal nature of relationships (Pearl, 2000; Stewart-Oaten, 1996).
Confounding can also bias a causal model resulting in uncertainty concerning the actual
magnitude of the effects. A variety of approaches may be used to determine whether
confounders significantly affect the results. They are related to three of the characteristics of
causation used to determine that elevated conductivity is a cause of impairment of stream
communities in Appendix A (co-occurrence, sufficiency, and alteration). We provide a
relatively complete list, but we only used Evidence Types 1, 2, 3, 5, 6, and 8.
1. Co-occurrence of confounder and cause: Confounders are correlated with the cause of
interest. A low correlation coefficient is evidence against the potential confounder.
2. Co-occurrence of confounder and effect: Potential confounders are correlated with the
effect of interest. A low correlation coefficient is evidence against the potential
confounder.
3. Co-occurrence of confounder and cause: Even when the confounder is not correlated
with the cause of interest, it may be influential at extreme levels. A lack of influence at
extreme levels of the cause and the potential confounder is evidence against the potential
confounder.
4. Co-occurrence of confounder and effect: If the frequency of the effect does not diminish
when the potential confounder is not present, the confounder can be discounted in that
subset.
5. Sufficient confounder: The magnitude of the potential confounder (e.g., concentration of
a co-contaminant) may be compared to exposure-response relationships from elsewhere
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 68 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
(e.g., laboratory toxicity tests) to determine if the exposure to the potential confounder is
sufficient. If it is not sufficient that is evidence that it is not acting as a confounder.
6. Sufficient confounder: If the confounder is estimated to be sufficient in a subset of cases,
those cases may be removed from the data set, and the remaining set reanalyzed to
determine the influence of their removal on the results.
7. Sufficient confounder: Multivariate statistical techniques may be used to estimate the
magnitude of confounding or to adjust the causal model for confounding, if their
assumptions hold.
8. Sufficient confounder: If the potential confounder occurs in a sufficiently small
proportion of cases, it can be ignored.
9. Alteration: If a potential confounder has characteristic effects that are distinct from those
of the cause of concern, then the absence of those effects can eliminate the potential
confounder as a concern in either individual cases or the entire data set.
10. Alteration: If the effects are characteristic of the cause of concern and not of the potential
confounder, then the potential confounder can be eliminated as a concern in either
individual cases or the entire data set.
Weighing evidence for confounding differs from weighing evidence for causation. The
causal assessment in Appendix A determines whether dissolved salts are an important cause of
biological impairment in the region. This assessment of confounding takes the result of the
causal assessment as a given and attempts to determine whether any of the known potential
confounders interfere with estimating the effects of conductivity to a significant degree. That
requires a different weighting and weighing method from the one in Appendix A, which would
be used if the goal were to determine whether the potential confounder is itself a cause.
As in Appendix A, the number of ephemeropteran genera is used as a standard metric for
the effects of conductivity, which may or may not be confounded. Because the sensitive genera
are primarily Ephemeroptera and the endpoint effect is extirpation of 5% of genera, this is an
appropriate metric.
B.2. WEIGHTING
The evidence is weighted using a system of plus (+) for supporting the potential
confounder (i.e., the evidence suggests that the potential confounder is actually causing the effect
to a significant degree), minus (-) for weakening the potential confounder (i.e., the evidence
suggests that the potential confounder does not contribute to the effect to a significant degree),
and zero (0) for no effect. One to three plus or minus symbols are used to indicate the weight of
a piece of evidence.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 69 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
+ + + Convincingly supports or weakens
+ + Strongly supports or weakens
+ Somewhat supports or weakens
0 No effect
Any relevant evidence receives a single plus, minus, or zero to register the evidence and
to indicate a decreased or increased potential for confounding (see Table B-l). The strength of
evidence is considered next. Criteria for scoring the strength of evidence are presented below for
the common types. They were developed for transparency and consistency and are based on the
authors' judgments. After strength, the other possible unit of weight is assigned depending on
the type of evidence.
For co-occurrence (Evidence Types 1-4), strength or consistency of the association is the
primary consideration. For comparison for any of the potential confounders, the correlation
coefficient for conductivity and number of ephemeropteran genera is 0.63, a value in the upper
end of the moderate range. Correlations, as measures of co-occurrence, can be scored as in
Table B-2.
These scores are based on conventional expectations for a confounder that is itself a
cause. That is, a potential confounder such as deposited sediment by itself can cause extirpation
of invertebrate genera (independent combined action) or can act in combination with
conductivity to extirpate invertebrate genera (additive or more than additive combined action).
However, sometimes correlations are anomalous. For example, a confounder may actually
decrease effects. Such anomalous results require case-specific interpretation, based on
knowledge of mechanisms and characteristics of the ecosystems being analyzed.
Anomalous results may also result from violation of the expectation that a confounder
should be correlated with both conductivity and the effect. If only one of the correlations is
observed, that result requires additional interpretation. If the potential confounder is correlated
with the effect, but not with conductivity, the result may be due to chance, or to a partitioning of
causation in space. That is, they are independent, because the confounder impairs communities
at different locations than conductivity. This could occur if the potential confounder and
conductivity have different sources. In any case, it is not a confounder of conductivity.
In the contingency tables (Evidence Type 3), the frequency of occurrence of any
Ephemeroptera (i.e., of the failure to extirpate all ephemeropteran genera) is presented for
combinations of high and low levels of conductivity and of the potential confounder. If the
frequency of occurrence is much lower when the confounder is present at high levels, this is
supporting evidence for confounding. Note, the goal here is not to determine the effects of
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 70 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
exceeding a criterion or other benchmark. Rather the goal is to clarify the co-occurrence of
conductivity, confounders, and effects by determining the frequency of effects at each possible
combination of extremely high and low levels of conductivity and the potential confounder. It is
expected that, if a variable is indeed a confounder, its influence on the occurrence of effects
would be seen at an extreme level. This use of contingency tables could reveal influences of
confounders that are obscured when the entire ranges of data are correlated. Therefore, clearly
high and low levels of conductivity and the potential confounder are used in contingency tables.
A potential confounder gets a plus score if its presence at a high level reduces the
probability of occurrence by more than 25% and a minus score if it does not (see Table B-3). It
gets a double plus score if its presence at a high level reduces the probability of occurrence by
more than 75% and a double minus score if it raises it by less than 10%. Any decrease in effects
at high levels of a potential confounder is anomalous and is treated as strong negative evidence.
The evidence concerning sufficiency of the confounder (Evidence Types 5-8) is diverse.
Only Evidence Type 6 was sufficiently common and consistent to develop scoring criteria. For
Evidence Type 6, the primary consideration is the degree of departure of the correlation in the
truncated data set (regarding pH, RBP, and fecal coliform) from the correlation of conductivity
and Ephemeroptera (r = 0.63) in the full data set (see Table B-4).
For alteration, the primary consideration is the degree of specificity of the effects of the
confounder relative to those of the salts. This type of evidence is rare and is scored ad hoc when
it occurs.
Additional considerations that may result in a higher score are presented in Table B-5.
B.3. WEIGHING
After the individual pieces of evidence had been weighted, the body of evidence for a
potential confounder was weighed based on the credibility, diversity, strength, and coherence of
the body of evidence (see Table B-6). The body of evidence, rather than a single piece of
evidence, was considered to determine how strongly these potential confounders might affect the
model.
B.4. POTENTIAL CONFOUNDERS
B.4.1. Habitat Quality
Stream habitat may be modified in reaches that receive high conductivity effluents.
Habitat quality was represented by a qualitative index, the RBP derived by the WVDEP, which
increases as habitat quality increases.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 71 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Although habitat scores were correlated with both conductivity and biological response,
which indicates a potential for confounding, low RBP was judged to have little effect on the
derivation of the 5th percentile hazardous concentration (HC05) for conductivity (see Table B-7
and B-8).
B.4.2. Organic Enrichment
Sources of organic enrichment such as domestic sewage and animal wastes are also
sources of salts that contribute to conductivity. Fecal coliform counts are an indicator of organic
enrichment and the presence of sources that may contain other toxicants such as household
waste. The data show no indication of significant confounding associated with fecal coliform
counts and effects attributed primarily to organic enrichment (see Tables B-9 and B-10).
B.4.3. Nutrients
Nitrogen and phosphorus may also come from sewage and animal wastes or from
fertilizers used in agriculture or mine reclamation. Because neither nutrient was correlated with
conductivity or Ephemeroptera, effects could not be confounded by nutrients when conductivity
increased (see Table B-l 1).
B.4.4. Deposited Sediment
Sources of salts can be associated with erosion and silt that affect stream organisms. A
qualitative measure of embeddedness was evaluated by correlation and by contingency table (see
Table B-13). Embeddedness was judged to have little if any effect on the derivation of the HC0s
for conductivity (see Tables B-12 and B-13).
B.4.5. High pH
The dissolution of limestone and dolomite increases as unweathered surface area of rock
increases. Waters draining crushed limestone and dolomite contain HCO3 which contributes to
higher pH and alkalinity. The HCO3 that raises pH is also a major anion moiety that contributes
to conductivity. Hence, pH directly reflects a major constituent of conductivity (HCO3 ) and
should not be analyzed as a potential confounder. In addition, salts influence hydrogen ion
activity which is measured as pH. In any case, variance in pH was judged to have little effect on
the derivation of the HC05 for conductivity in waters above pH 7 (see Tables B-14 and B-l 5).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 72 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
B.4.6. Low pH
Because low pH from acid mine drainage is known to be an important cause of
impairment where it occurs, it was judged a priori to be a potentially important environmental
variable. That preconception was supported by the evidence summarized here. Therefore, sites
with pH <6 were not used to calculate the XC values. However, Table B-15 suggests that even
below pH 4.5, conductivity is more important than acidity to the occurrence of Ephemeroptera
(see Tables B-15 and B-16). So, although the benchmark applies to waters with neutral or basic
pH, high conductivity appears to also cause effects at low pH.
B.4.7. Selenium
Selenium (Se) is a potential confounder because it is commonly associated with coal, and
elevated levels have been reported in the region. In an analysis of a small data set, Pond et al.
(2008a) found that the number of ephemeropteran genera was highly correlated with Se
concentration (r = -0.88, n = 20). In contrast, weak correlations were found in our analysis of
the West Virginia data. This result is unreliable, because most of the Se values were detection
limits, and many of the detection limits were relatively high, equaling or exceeding the water
quality criterion of 5.0 |ig/L, In addition, there were too few high Se concentrations in the West
Virginia data to perform a contingency table analysis. For these reasons, we did not include a
quantitative analysis of potential confounding by Se. The effects of Se in central Appalachian
streams should be investigated further.
B.4.8. Temperature
Elevated temperature may occur with elevated conductivity if the sources of salts are
associated with lack of stream shading or if saline effluents are heated. Although temperature is
moderately correlated with conductivity on an annual basis, the correlation is greatly reduced by
seasonal partitioning (see Tables B-17 and B-18). More importantly, elevated temperature does
not appear to be associated with the loss of Ephemeroptera.
B.4.9. Lack of Headwaters
The loss of headwaters due to mining and valley fill eliminates a source of recolonization
for downstream reaches. Hypothetically, this could result in extirpation of invertebrates, if the
sampled sites are sink habitats that must be recolonized by headwater source habitats. This is
plausible in stream reaches immediately below valley fills. However, where there are other
headwaters on tributaries above the sampling site, they serve as alternative sources for
recolonization. No regional data are available to address this issue. However, examination of
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 73 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
individual watersheds shows that many if not most of the sampled sites have at least one
upstream intact headwater. Two examples are presented here.
Ballard Fork, a tributary to the Mud River in West Virginia, is downstream of several
valley fills but has unmined tributaries upstream such as Spring Branch (see Figures B-2, B-3,
B-4). Conductivity in Spring Branch measured <44-66 [j,S/cm. Conductivity in Ballard Fork
was 464-2,300 [j,S/cm. In Spring Branch, the benthic invertebrate assemblages in the springs of
1999, 2000, and 2006 had 6-8 genera of Ephemeroptera representing 29-45% of the sample. In
contrast, on the same dates Ballard Fork had 1-3 genera of Ephemeroptera representing only
2-4% of the sample and those may be indicative of immigrant specimens. Hence, even when a
source of recolonization was available from Spring Branch, ephemeropteran genera were
extirpated in Ballard Fork where conductivity was elevated. Other potential confounders are
apparently not responsible for differences between the creeks, because biological quality is not
related to habitat quality (embeddedness, total RBP habitat score, and pH). Data are from U.S.
EPA Mountain Top Mining studies (Green et al., 2000; Pond et al., 2008a) (see Table B-19).
In the Twentymile Creek watershed, the most upstream catchment above river kilometer
(RKm) 44 is a small headwater that is 99% forested. Between RKm 44 and 13, the tributary
catchments are heavily mined with valley fills. Below RKm 25 to the mouth, benthic
invertebrate assemblages are depauperate. Two catchments that enter Twentymile Creek near
Rkm 17 and 14 are 100% forested with diverse benthic invertebrate assemblages. Nevertheless,
at RKm 12, the benthic assemblage in Twentymile Creek remains depressed. Downstream from
RKm 12, there are mixed mining and forest land uses. Near RKm 2 there are legacy mining and
urban land uses (see Table B-20). WVSCI scores and numbers of EPT taxa were low when
conductivity was high regardless of the condition of catchments that provided sources of benthic
macroinvertebrates including salt-sensitive genera. Data are from WABbase.
In these two examples, the reduction in ephemeropteran genera or EPT is not caused by a
lack of sources of recolonization from headwaters. This is not to say that recolonization is never
an issue. The sources of salts in this region are primarily chronic and localized, so lack of
recolonization is unlikely to confound their effects. However, if an episodic agent caused the
loss of aquatic organisms (e.g., drought or forest treatment with insecticides), sources of
recolonization could be important.
B.4.10. Catchment Area
Larger streams tend to have more moderate chemical properties than small streams
because they receive waters from more sources than small streams, both natural and
anthropogenic. Consequently, extreme values, in this case both low and high conductivity, tend
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 74 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
to occur less frequently in large streams. One of the initial data filters for this analysis was to
exclude streams larger than 155 km2 (or 60 mi2). Small streams are numerically more abundant
than large streams and the inclusion of large streams might introduce extraneous variance. This
raises the issue whether stream size is a potential confounder and whether the results from small
streams might be extrapolated to larger streams. That is, do the same effects of conductivity
occur in larger streams as were found in the detailed analysis? We examined these issues by
analyzing the influence of stream size (as catchment area) on the effects of conductivity.
Correlation of log conductivity with log catchment area is extremely low (r = 0.12).
Owing to the large number of sites (N= 1,750), the regression is statistically significant, but it is
almost negligible and accounts for less than 2% of the variability in conductivity. Nearly all
reference sites, even those identified as Level II and Level III, had conductivities less than
300 [j.S/cm.
We categorized streams by catchment area into three groups: small catchments less than
6 mi2 (15.5 km2), medium catchments of 6 to 60 mi2 (155 km2), and large catchments greater
than 60 mi2. The number of Ephemeroptera (mayfly) taxa declines with increasing conductivity
in all streams, independent of classification of catchment area (r = -0.62).
We likewise categorized conductivity into three groups by defining low conductivity as
<200 [j,S/cm, and high conductivity as >1,500 [j,S/cm (see Table B-21). In all three stream size
categories, if conductivity was <200 [j,S/cm, 99% or more of all streams had mayfly populations,
but if conductivity was above 1,500 [j,S/cm, only 50% or fewer streams had mayflies (see
Table B-21). Evidence for confounding by catchment area is summarized in Table B-22; the
evidence is uniformly negative and we conclude that catchment area has little or no effect on
invertebrate response to conductivity.
B.5. SUMMARY OF ACTIONS TAKEN TO ADDRESS POTENTIAL
CONFOUNDING
Low pH is a potential confounder, but sites with pH <6 were removed from the data set
when calculating the benchmark value. The influence of Se could not be evaluated due to poor
data and should be investigated. However, toxic levels of Se appear to be relatively uncommon.
Other potential confounders were eliminated from consideration with some confidence. We do
not argue that these variables have no influence, but their effects appear to be minimal given the
inevitable variability in sites to which the benchmark would be applied.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 75 DRAFT—DO NOT CITE OR QUOTE
-------
1 REFERENCES
2 Green, J, M Passmore, H Childers. 2000 GREEN, J., M.. A survey of the condition of streams in the primary region
3 of mountaintop mining/valley fill coal mining. Appendix in Mountaintop mining/valley fills in Appalachia. Final
4 programmatic environmental impact statement. Region 3, US Environmental Protection Agency, Philadelphia,
5 Pennsylvania. (Available from: http://www.epa.gov/ region3/mtntop/pdf/Appendices/Appendix%20D%
6 20Aquatic/Streams%20Invertebrate%20Study-%20EPA/ FINAL.pdf)
7
8 Hill, AB. (1965) The environment and disease: Association or causation. Proceed Royal Soc Med 58:295-300.
9 Pearl, J. (2000) Causality: Models, Reasoning, and Inference. Cambridge U. Press, Cambridge, UK.
10 Pond, GJ; Passmore, ME; Borsuk, FA; L. Reynolds; CJ Rose. (2008) Downstream effects of mountaintop coal
11 mining: comparing biological conditions using family- and genus-level macroinvertebrate bioassessment tools. J N
12 Am Benthol Soc 27:717-737.
13 Stewart-Oaten, A. (1996) Problems in the analysis of environmental monitoring data. In: Detecting Environmental
14 Impacts, Schmitt, RJ; Osenberg, CW; ed. New York, NY: Academic Press, p. 109-131.
15 U.S. EPA (Environmental Protection Agency). (1976) Quality criteria for water. Washington, DC.
16
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 76 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-l. Relationships between qualities of evidence and scores for
weighing evidence
Qualities of the evidence
Score, not to exceed three minus or three plus
Logical implications and relevance
+ o
Strength
Increase score
Other qualities
Increase score
Table B-2. Weighting co-occurrence using correlations for Approaches 1-2
Assessment
Strength
Score
Absent
r < 0.1
—
Weak
0.1 r> 0.25
+
High
r > 0.75
+ +
Table B-3. Weighting co-occurrence for Evidence Type 3 using contingency
tables
Assessment
Strength
Score
High levels of a confounder
should increase the
probability that a site lacks
Ephemeroptera at low
conductivity, and low levels
of the confounder should
decrease the effect at high
conductivities
Increased effect >25%
+ for co-occurrence
Increased effect >75%
+ + for co-occurrence and strength
Increased effect <25%
- for co-occurrence
Increased effect <10% or
decreased effect
— for co-occurrence and strength
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 77 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-4. Weighting sufficiency for Evidence Type 6: alteration of the
correlation of conductivity with the number of ephemeropteran genera after
removal of elevated levels of a confounder
Assessment
Strength
Score
Removal of elevated
levels of a
confounder should
change the
correlation
coefficient
Coefficients deviating by <10%
0.56 < r < 0.69
— for a lack of change in effect with
removal of confounder
Coefficients deviating by <20%
0.50 20%
0.50 >r> 0.75
+ for a strong increase or decrease in
effect with removal of confounder
Table B-5. Considerations used to weight the evidence concerning the
influence of potentially confounding variables
Quality of evidence
Descriptor
Logical implication
Negative or positive
Directness of cause
Proximate cause, sources, or intermediate causal connections
Specificity
Effect attributable to only one cause or to multiple causes
Relevance to effect
From the case or from other similar situations
Nature of the association
Quantitative or qualitative
Strength of association
Strong relationships and large range or weak relationships and
small range
Consistency of information
All consistent or some inconsistencies
Quantity of information
Many data or few data
Quality of information
Good study or poor study
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 78 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-6. Weighing confidence in the body of evidence for a potential
confounder
Assessment
Score
Body of evidence
Action
Very confident
All minus, some strongly
negative evidence
No treatment for confounding
Moderately confident
—
All minus, no strongly negative
evidence
No treatment for confounding
Reasonably confident
-
Majority minus
No treatment for confounding
Undetermined
0
Approximately equal positive
and negative, ambiguous
evidence, or low quality
evidence
Additional study advised
Potential confounding
+
Majority plus
Correction for confounding
may be advised
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 79 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-7. Evidence and weight for confounding by habitat quality
Approach
Score
Evidence
1. Correlation of cause
and confounder
+
RBP score was moderately correlated with conductivity,
(r = -0.29, n = 2,344).
2. Correlation of effect
and confounder
+
RBP score was (barely) moderately correlated with the
number of ephemeropteran genera (r = 0.26, n = 2,193).
3. Contingency of high
level of cause and
confounder
In a contingency table (see Table B-8), Ephemeroptera are
present at >99% of sites with low conductivity (<200 (j,S/cm)
even when habitat is poor (<115). However, with high
conductivity, Ephemeroptera are present at only 40% of sites
with poor habitat and 60% of sites with good habitat.
6. Removal of
confounder
When sites with moderate to poor habitat (an RBP score
<140) were removed from the analysis, conductivity is a little
less negatively correlated with the number of Ephemeroptera
(r = -0.50, n = 768) (see Table B-8).
The SSD and HCos are very similar when the XC95 values
were calculated with a year-long data set and a subset of the
data set with sites removed with pH of <6, RBP score <135,
and fecal coliform >400 colonies/100 mL (see Figure B-l).
W eight of evidence
Somewhat confident, evidence is mixed, but the contingency
table gives relatively strong negative evidence
(Ephemeroptera occur even when habitat is poor), while RBP
explains only 6.7% of the variance in ephemeropteran
occurrence.
SSD = species sensitivity distribution.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 80 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-8. Number of sites with high and low quality habitat and high and
low conductivity with Ephemeroptera present in streams (pH > 6)
Conductivity <200 jiS/cm
Conductivity >1,500 jiS/cm
Habitat score <115
155/157
12/31
(98.7%)
(39.7%)
Habitat score >140
388/390
13/22
(99.5%)
(59.1%)
Table B-9. Evidence and weights for confounding by organic enrichment
Approach
Score
Evidence
1. Correlation of cause
and confounder
—
Fecal coliform counts were weakly correlated with
conductivity (r = 0.25, n = 2,044).
2. Correlation of effect
and confounder
—
Coliform count was not correlated with the number of
ephemeropteran genera (r = -0.14, n = 1,349).
3. Contingency of high
level of cause and
confounder
In a contingency table (see Table B-10), the presence of
high coliform counts did not change the probability of
finding Ephemeroptera at either high or low conductivity
(see Table B-10).
6. Removal of
confounder
(a) When samples >400 colonies/100 mL were removed
from the analysis, the correlation of conductivity with
Ephemeroptera was unchanged (r = -0.63, n = 1,671).
(b) The SSD and HCos are very similar when the XC95 is
calculated with a year-long data set and a subset of the data
set with sites removed with pH of <6, RBP score <135, and
fecal coliform >400 colonies/100 mL (see Figure B-l).
W eight of evidence
Very confident: All negative, some strongly negative. No
treatment for confounding.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 81 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-10. Number of sites with high and low conductivity with high and
low levels of fecal coliform with Ephemeroptera present in streams
Conductivity <200 jiS/cm
Conductivity >1,500 jiS/cm
Coliform <400 colonies/100 mL
658/662
36/77
(99.4%)
(46.7%)
Coliform >400 colonies/100 mL
233/237
20/42
(98.3%)
(47.6%)
Table B-ll. Evidence and weights for confounding by nutrients
Approach
Score
Evidence
1. Correlation of
cause and
confounder
Conductivity was not correlated with nitrate and nitrite
(r = 0.08, n = 1,265) or total phosphorus (r = 0.05,
n= 1,190).
2. Correlation of
effect and
confounder
Ephemeroptera was not correlated with nitrate and nitrite
(r = 0.037, n = 1,184) or total phosphorous (r = 0.001,
n= 1,186).
3. Contingency of
high level of cause
and confounder
NA
Contingency table analyses were not used because extreme
nutrient levels were rare at high conductivities.
6. Removal of
confounder
When samples with nitrate plus nitrite >0.6 mg/L were
removed from the analysis, the correlation of conductivity
with the number of Ephemeroptera was similar (r = -0.54,
n = 999).
When samples with total phosphorus >0.04 mg/L were
removed from the analysis, the correlation of conductivity
with the number of Ephemeroptera was similar (r = -0.56,
n = 999).
W eight of evidence
Moderately confident: all negative, none strongly negative.
No treatment for confounding.
NA = not applicable.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 82 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-12. Evidence and weights for confounding by deposited sediment
Approach
Score
Evidence
1. Correlation of
cause and
confounder
The WABbase embeddedness score is weakly correlated
with conductivity (r = -0.18, n = 2,202).
2. Correlation of
effect and
confounder
The WABbase embeddedness score is weakly correlated
with Ephemeroptera (r = -0.22, n = 2,198).
3. Contingency of
high level of cause
and confounder
In a contingency table (see Table B-13), high
embeddedness (score >15) has little effect at either high or
low conductivity (see Table B-13).
6. Removal of
confounder
When samples with an embeddedness score <13 are
removed from the analysis, the correlation of conductivity
with the number of Ephemeroptera was virtually
unchanged (r = -0.61, n = 1,089).
W eigh! of evidence
—
Very confident: all negative, some strongly. No treatment
for confounding.
Table B-13. Number of sites with high and low embeddedness scores and
high and low conductivity with Ephemeroptera present in streams (pH >6)
Conductivity <200 jiS/cm
Conductivity >1,500 jiS/cm
Embeddedness score <7
56/58
7/16
(96.6%)
(43.8%)
Embeddedness score >15
208/225
6/15
(92.4%)
(40%)
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 83 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-14. Evidence and weights for confounding by high pH
Approach
Score
Evidence
1. Correlation of cause
and confounder
+
Conductivity was moderately correlated with pH between 7
and 9.0, (r = 0.45, n= 1,911).
2. Correlation of effect
and confounder
—
High pH was weakly correlated with Ephemeroptera
(r = 0.19, n = 1,907).
3. Contingency of high
level of cause and
confounder
0
In a contingency table (see Table B-15), there were too few
streams with high pH to provide evidence for or against
confounding.
5. Levels of confounder
is known to cause
effects
U.S. EPA (1976) Water Quality Standards indicate that
water quality 6.5-9 is protective of freshwater fish.
6. Removal of
confounder shows it
is important
When samples with pH > 8.5 are removed from the analysis,
the correlation of conductivity with the number of
Ephemeroptera was unchanged (r = -0.63, n = 1,089) (see
Table B-15).
8. Potential confounding
evaluated by
frequency
The number of sites with a pH >8.5 is a very small
proportion of the sample (<2.5%).
W eight of evidence
—
Reasonably confident: majority negative. No treatment for
confounding.
Table B-15. Number of sites with high and low conductivity with high and
low levels of pH with Ephemeroptera present
Conductivity <200 jiS/cm
Conductivity >1,500 jiS/cm
pH <4.5
16/19
0/14
(84.2%)
(0%)
pH >8.5
3/3
4/8
(100%)
(50%)
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 84 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-16. Evidence and weights for confounding by low pH
Approach
Score
Evidence
1. Correlation of
cause and
confounder
+
Conductivity was moderately correlated with pH <6
(r = 0.48, n = 145).
2. Correlation of
effect and
confounder
+
Low pH was moderately correlated with Ephemeroptera
(r = 0.46, n = 145).
3. Contingency of
high level of cause
and confounder
Even at low pH some low conductivity streams support
some Ephemeroptera but not at high conductivities (see
Table B-15).
W eight of evidence
+
Potential confounding: majority positive. Correction for
confounding may be advisable.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 85 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-17. Evidence and weights for confounding by temperature
Approach
Score
Evidence
1. Correlation of cause
and confounder
0
Conductivity was moderately correlated with year-round
temperature, r = 0.39, n = 2,221. When the correlation was
recalculated for the spring and summer index periods,
conductivity was less correlated, but still moderately, with
the summer (r = 0.29, n = 961). Spring temperatures,
however, were weakly correlated with conductivity
(r = 0.16, n = 1,199).
2. Correlation of effect
and confounder
Temperature was weakly correlated with Ephemeroptera
year round (r = ~0.196, n = 2,363) and in summer
(r = -0.12, n = 961) and not correlated in spring (r = -0.04,
n= 1,195).
3. Contingency of high
level of cause and
confounder
Ephemeroptera were present at >98-100% of sites at low
conductivity at both high and low temperature. In the high
conductivity categories, Ephemeroptera occurred in more
sites with elevated temperatures (see Table B-18), which is
contrary to expectations, if temperature were contributing to
the impairment.
5. Levels of confounder
is known to cause
effects
Temperatures rarely exceeded 20°C and, therefore, are not
likely to cause extirpation of genera.
6. Removal of
confounder shows it
is important
When high temperatures (>22°C) were deleted, the
correlation of conductivity and Ephemeroptera was barely
changed (r = -0.61, n = 1,788).
W eight of evidence
Moderately confident: all negative, some strongly negative.
No treatment for confounding.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 86 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-18. Number of sites with high and low temperatures and high and
low conductivity with Ephemeroptera present in streams (pH >6)
Conductivity <200 jiS/cm
Conductivity >1,500 jiS/cm
Temperature <17°C
468/474
9/27
(98.7%)
(33.3%)
Temperature >22°C
78/78
24/43
(100%)
(55.8%)
Table B-19. Comparison of low conductivity Spring Branch with high
conductivity Ballard Fork
Stream Name
Date
Embd.
Total RBP
Score
pH
fiS/cm
# E
% E
Total Count
Spring Branch
5/9/2006
16
149
7.7
66
8
29.27
205
Spring Branch
4/18/2000
16
163
7.5
44
6
44.76
143
Spring Branch
4/20/1999
7.7
51
8
34.72
337
Ballard Fork
5/9/2006
14
149
8.1
1,195
3
2.96
203
Ballard Fork
4/18/2000
12
148
7.1
464
1
2.08
48
Ballard Fork
1/25/2000
7.9
1,050
0
0
52
Ballard Fork
7/26/1999
8.2
2,300
0
0
88
Ballard Fork
4/20/1999
8.1
1,201
3
4.12
291
Embd. = embeddedness score from RBP; RBP = Rapid Bioassessment Protocol Habitat Evaluation; # E = Number
of ephemeroptera genera; % E = percent of ephemeroptera individuals in the sample; Total count = count of all
individuals of all taxa.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 87 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-20. Twentymile Creek sampling locations, conductivity, habitat
score, number of EPT taxa, and WVSCI scores
Year
River
Kilometer
Tributary
Catchment
Land Use"
Max Reported
Conductivity
(fiS/cm)
RBP Habitat
Score
# EPT Taxa
WVSCI
2003
44.6
Forested
44
148
6
90.72
2004
44.6
Forested
37
-
1998
25.1
Mined
805
155
3
67.62
2003
25.1
Mined
2,087
153
1
58.45
2003
11.9
Mixed Forest
and Mine
1,702
157
2
64.74
2004
11.9
Mixed Forest
and Mine
1,282
_
_
_
2003
1.8
Mixed Forest,
Mine, & Urban
987
_
_
_
2004
1.8
Mixed Forest,
Mine, & Urban
1,138
_
_
_
2003
0.5
Mixed Forest,
Mine, & Urban
845
146
3
66.73
2004
0.5
Mixed Forest,
Mine, & Urban
836
_
_
_
1998
0
Mixed Forest,
Mine, & Urban
590
131
3
65.94
"Land use refers to catchment land use of tributaries upstream from the sampled sites in Twentymile Creek.
# EPT taxa = Number of Ephemeroptera, Plecoptera, and Trichoptera taxa; WVSCI = West Virginia Stream
Condition Index.
Table B-21. Number (and percent) of streams with Ephemeroptera present:
small, medium and large streams and low, medium and high conductivity
(pH > 6).
Conductivity < 200 jiS/cm
Conductivity > 1,500 jiS/cm
Small streams
426/430 (99%)
16/39 (41%)
Medium streams
205/207 (99%)
19/38 (50%)
Large streams
70/70 (100%)
1/2 (50%)
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 88 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-22. Evidence and weights for confounding by catchment area
Approach
Score
Evidence
1. Correlation of cause
and confounder
—
Log catchment area was very weakly correlated with log
conductivity (r = 0.12, n = 1,750).
2. Correlation of effect
and confounder
Log catchment area was not correlated with the number of
ephemeropteran genera (r = 0.05, n = 1,750).
3. Contingency of high
level of cause and
confounder
In a contingency table (see Table B-21), large catchment
area did not change the probability of finding
Ephemeroptera at either high or low conductivity.
6. Removal of
confounder
Overall correlation of conductivity with Ephemeroptera in
all sites was r = - 0.62, n = 1,750). When correlation was
repeated for each of the size classes, the correlations were
-0.48 (large streams; n = 165), and - 0.64 (small and
medium streams; n = 942 and 653).
W eight of evidence
Very confident: All negative, some strongly negative. No
treatment for confounding.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 89 DRAFT—DO NOT CITE OR QUOTE
-------
LO
o
•-
^r
o
CO
o
CM
o
o
o
o
100 200 500 1000 2000
Conductivity (pS/cm)
Figure B-l. Species sensitivity distribution for all year, pH >6 and all sites
(open circles) and for sites with pH >6, Rapid Bioassessment Protocol >135
and fecal coliform <400 colonies/100 mL (closed circles). Habitat disturbance
and organic enrichment have little influence; the HCos for the constrained data set
is 300 [j,S/cm based on 111 genera. The upper and lower confidence bounds on
that value are 225 [j,S/cm and 350 [j,S/cm, respectively.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 90 DRAFT—DO NOT CITE OR QUOTE
-------
Figure B-2. Topographical map of Spring Branch (blue triangle) and
Ballard Fork (red triangle) sampling stations.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 91 DRAFT—DO NOT CITE OR QUOTE
-------
¦ valleytill
Spring Branch
iage ® 2009 DiglUIGIobe
©2009
Figure B-3. Aerial imagery (June 13, 2007) with superimposed sampling
locations of Spring Branch (turquoise square) and Ballard Fork (yellow
square). Mined land drains into Ballard Fork (upper section of image) and
forested land drains into Spring Branch (lower right quadrant). Two valley fills
indicated as examples.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 92 DRAFT—DO NOT CITE OR QUOTE
-------
Figure B-4. Same area as Figure 2. Aerial imagery (April 10, 1996) with
superimposed sampling locations of spring branch (turquoise square) and Ballard
Fork (yellow square). The many upstream valley fills in Ballard Fork are easily
seen.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 93 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX C
EXTIRPATION CONCENTRATION VALUES FOR INVERTEBRATES
-------
Genus
Both
Spring
Summer
XC95
N
Ref.
XC95
N
XC95
N
1
Ablabesmyia
>11,646
162
5
3,162
56
11,646
106
2
Acentrella
1,289
748
31
1,289
422
769
326
3
Acroneuria
2,630
480
60
1,649
138
2,320
342
4
Alloperla
228
96
15
319
82
5
Ameletus
599
192
30
388
189
6
Amphinemura
805
561
42
1,468
556
7
Antocha
>6,468
538
18
3,725
162
6,468
376
8
Argia
9,790
75
NA
9,790
71
9
Asellus
925
33
2
10
Atherix
>11,646
156
3
11,646
149
11
Atrichopogon
2,257
42
3
2,257
40
12
Attenella
574
34
1
13
Baetis
1,383
1,509
72
1,383
642
1,494
867
14
Baetisca
918
47
NA
646
32
15
Bezzia
381
62
2
563
39
11,227
127
16
Bezzia/Palpomyia
4,713
306
26
3,725
179
17
Boyeria
>7,340
173
5
1,468
52
7,340
121
18
Brillia
1,746
91
6
1,083
51
2,768
40
19
Caecidotea
>4,713
137
1
1,083
51
4,713
62
20
Caenis
3,884
541
8
1,175
168
3,884
373
21
Calopteryx
3,489
52
NA
3,489
45
22
Cambarus
1,228
464
44
1,649
289
1,340
175
23
Cardiocladius
>2,257
185
2
1,530
86
2,257
99
24
Centroptilum
1,075
90
6
1,175
59
1,075
31
25
Ceratopsyche
>6,468
885
29
3,314
232
6,468
653
26
Chaetocladius
>5,057
182
4
1,650
76
5,057
106
27
Chelifera
>3,341
152
9
1,650
64
3,341
88
28
Cheumatopsyche
>9,180
1,612
57
2,493
562
9,180
1,050
29
Chimarra
>3,972
490
11
1,175
100
3,972
390
30
Chironomus
>11,646
105
1
5,120
48
11,646
57
31
Chrysops
>11,646
76
1
11,646
51
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 95 DRAFT—DO NOT CITE OR QUOTE
-------
Genus
Both
Spring
Summer
XC95
N
Ref.
XC95
N
XC95
N
32
Cinygmula
224
81
17
347
80
33
Cladotanytarsus
>11,646
103
5
3,162
57
11,646
46
34
Clinocera
>4,713
60
6
1,573
33
35
Conchapelopia
518
135
7
1,175
120
36
Corbicula
9,790
184
NA
1,175
51
9,790
133
37
Cordulegaster
1,468
42
3
38
Corydalus
>11,227
311
1
1,117
49
11,227
262
39
Corynoneura
2,006
149
5
1,650
82
2,768
67
40
Crangonyx
2,169
105
7
796
65
2,169
40
41
Cricotopus
>11,227
605
24
3,725
274
11,227
331
42
Cricotopus/Orthocladius
>6,468
1,054
13
3,725
493
7,340
561
43
Cryptochironomus
>3,489
287
3
3,162
129
3,489
158
44
Dasyhelea
>3,341
66
3
3,341
51
45
Demi cryptochironomus
322
81
6
322
67
46
Diamesa
>4,713
457
14
3,725
294
5,057
163
47
Dicranota
>7,010
327
43
1,649
160
7,010
167
48
Dicrotendipes
>11,646
192
1
3,314
79
11,646
113
49
Dineutus
9,790
46
NA
50
Diphetor
648
134
17
653
88
701
46
51
Diplectrona
2,523
594
60
3,725
233
2,523
361
52
Diploperla
318
99
2
357
96
53
Dixa
722
68
16
1,650
34
794
34
54
Dolophilodes
864
339
46
1,323
145
618
194
55
Drunella
294
172
18
660
153
56
Dubiraphia
>7,370
141
3
3,162
33
7,370
108
57
Eccoptura
462
65
6
518
43
58
Ectopria
1,386
311
32
1,175
121
1,570
190
59
Epeorus
316
359
57
500
280
247
79
60
Ephemera
736
138
20
1,175
45
627
93
61
Ephemerella
302
362
38
434
347
62
Eukiefferiella
1,930
501
28
1,649
271
1,979
230
63
Eurylophella
476
173
19
280
98
554
75
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 96 DRAFT—DO NOT CITE OR QUOTE
-------
Genus
Both
Spring
Summer
XC95
N
Ref.
XC95
N
XC95
N
64
Ferrissia
4,884
91
NA
4,884
80
65
Fossaria
5,057
30
NA
66
Gammarus
>4,713
215
10
1,800
67
4,713
148
67
Glossosoma
1,650
154
7
1,650
38
925
116
68
Haploperla
423
235
27
497
182
603
53
69
Heleniella
1,700
62
7
2,768
35
70
Helichus
>11,646
328
18
1,650
164
11,646
164
71
Hemerodromia
>9,790
607
8
3,725
109
9,790
498
72
Heptagenia
313
68
3
269
55
73
Hexatoma
>9,790
818
65
1,059
393
9,790
425
74
Hydropoms
822
32
1
810
30
75
Hydropsyche
>7,010
981
21
3,725
234
7,010
747
76
Hydroptila
>11,646
278
4
3,162
51
11,227
227
77
Isonychia
1,177
712
16
1,175
234
1,068
478
78
Isoperla
459
485
39
694
428
704
57
79
Krenopelopia
2,320
61
2
2,320
36
80
Lanthus
2,087
66
7
1,175
34
1,702
32
81
Larsia
2,630
96
3
1,289
76
82
Lepidostoma
109
88
12
796
74
83
Leptophlebia
224
85
8
805
70
84
Leucrocuta
425
219
29
1,175
158
418
61
85
Leuctra
2,087
1,170
84
1,175
665
2,257
505
86
Limnophila
2,768
49
10
322
33
87
Limnophyes
>5,120
88
1
5,120
31
1,979
57
88
Limonia
>5,057
62
1
5,057
57
89
Lirceus
1,303
70
6
517
51
90
Maccaffertium
1,111
174
13
680
82
1,177
92
91
Macronychus
1,890
39
4
92
Microcylloepus
3,341
94
2
3,341
82
93
Micropsectra
>6,468
220
25
5,120
107
6,468
113
94
Microtendipes
>3,489
507
34
1,383
198
3,489
309
95
Microvelia
2,523
46
3
2,523
32
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 97 DRAFT—DO NOT CITE OR QUOTE
-------
Genus
Both
Spring
Summer
XC95
N
Ref.
XC95
N
XC95
N
96
Nanocladius
1,485
50
NA
97
Natarsia
1,842
54
1
1,842
36
98
Neophylax
323
122
36
578
116
99
Nigronia
>9,790
726
37
3,162
204
7,340
522
100
Nilotanypus
2,630
112
3
731
49
2,630
63
101
Nixe
316
77
3
357
73
102
Ochrotrichia
2,791
32
1
103
Optioservus
9,790
1,429
65
1,890
500
7,370
929
104
Orconectes
3,162
205
2
3,162
56
1,978
149
105
Orthocladius
3,341
272
10
805
117
3,341
155
106
Oulimnius
2,791
219
27
1,650
73
2,440
146
107
Pagastia
1,800
46
2
108
Palpomyia
1,870
40
NA
109
Parachaetocladius
1,239
151
27
509
33
1,205
118
110
Paragnetina
2,087
39
3
111
Parakiefferiella
1,700
75
2
1,006
39
1,896
36
112
Paraleptophlebia
439
432
46
496
295
488
137
113
Parametriocnemus
>4,713
1,450
72
2,493
687
4,713
763
114
Paraphaenocladius
>6,468
71
2
6,468
46
115
Paratanytarsus
>3,489
108
2
3,314
48
3,489
60
116
Paratendipes
11,227
78
NA
1,800
34
11,227
44
117
Peltoperla
659
124
12
1,650
73
745
51
118
Perlesta
3,314
314
8
3,162
289
119
Phaenopsectra
2,332
89
2
2,332
64
120
Physella
>9,790
143
1
1,276
61
9,790
82
121
Pisidium
1,795
34
2
122
Plauditus
927
286
12
847
209
2,257
77
123
Polycentropus
4,713
357
41
1,443
154
2,768
203
124
Polypedilum
>4,884
1,604
75
1,890
686
5,057
918
125
Potthastia
1,896
60
1
1,534
33
126
Procloeon
701
78
3
347
31
524
47
127
Promoresia
589
78
5
467
53
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 98 DRAFT—DO NOT CITE OR QUOTE
-------
Genus
Both
Spring
Summer
XC95
N
Ref.
XC95
N
XC95
N
128
Prosimulium
565
89
21
808
82
129
Prostoma
2,553
41
NA
130
Psephenus
>9,790
853
36
1,479
329
2,553
39
131
Pseudochironomus
>11,646
31
2
7,370
524
132
Pseudolimnophila
1,418
130
11
731
78
1,740
52
133
Psychomyia
1,106
38
3
1,106
33
134
Pteronarcys
660
105
25
499
41
907
64
135
Pycnopsyche
299
40
10
502
31
136
Remenus
101
35
3
183
35
137
Rhagovelia
2,030
51
3
138
Rheocricotopus
3,489
556
11
1,346
225
3,489
331
139
Rheopelopia
1,247
125
4
1,534
37
1,014
88
140
Rheotanytarsus
>3,489
938
28
1,346
252
3,489
686
141
Rhyacophila
1,890
379
58
1,650
225
5,057
154
142
Serratella
500
46
2
143
Sialis
>11,227
261
3
3,725
52
11,227
209
144
Simulium
>6,468
1,084
26
1,800
408
6,468
676
145
Sphaerium
>9,790
39
NA
146
Stempellina
617
33
8
147
Stempellinella
892
304
26
562
120
1,075
184
148
Stenacron
769
249
15
316
105
850
144
149
Stenelmis
>9,790
1,217
27
3,162
539
9,790
678
150
Stenochironomus
1,613
40
NA
151
Stenonema
729
905
61
875
331
687
574
152
Stictochironomus
3,162
39
NA
153
Stylogomphus
>6,468
117
1
6,468
99
154
Sublettea
2,440
179
2
929
43
2,087
136
155
Sweltsa
761
294
43
1,650
126
111
168
156
Tabanus
>9,790
60
1
3,162
35
157
Tallaperla
452
87
16
302
63
158
Tanytarsus
9,180
1,194
64
1,650
481
9,180
713
159
Thienemanniella
>9,790
389
9
2,493
203
9,790
186
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 99 DRAFT—DO NOT CITE OR QUOTE
-------
Genus
Both
Spring
Summer
XC95
N
Ref.
XC95
N
XC95
N
160
Thienemannimyia
>6,468
1,326
56
1,649
536
6,468
790
161
Tipula
1,979
590
36
1,649
320
2,169
270
162
Tokunagaia
1,070
43
NA
1,070
35
163
Tribelos
2,257
45
NA
164
Tricorythodes
2,006
44
NA
2,006
43
165
Tvetenia
>2,768
727
40
1,649
370
5,057
357
166
Utaperla
240
47
2
198
32
167
Wormaldia
1,533
73
8
796
31
1,746
42
168
Yugus
603
72
12
796
48
169
Zavrelia
413
81
6
347
60
170
Zavrelimyia
>2,768
240
11
834
112
4,884
128
Empty cells indicates fewer than 30 occurrences during that season.
XC95 = 95th percentile extirpation concentration reported as (iS/cm; NA = not applicable because it never occurs at
WVDEP reference locations; Both = March through October; Spring = Sampled March through June;
Summer = July through October; Ref. = number of times the genus was observed at a reference site.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 100 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX D
GRAPHS OF OBSERVATION PROBABILITIES AND CUMULATIVE DISTRIBUTION
FUNCTIONS FOR EACH GENUS
-------
The purpose of Appendix D is to help the reader visualize the changes in the occurrence
of each genus as conductivity increases. Figure D-l contains general additive models of the
relationship between capture probability of the genus and conductivity, ordered from the lowest
to the highest XC95 value. Open circles are the probabilities of observing the genus within a
range of conductivities. Circles at zero probability indicate no individuals at any sites were
found at these conductivities. The line fitted to the probabilities is for visualization. The vertical
red line indicates the XC95. Note that different genera respond differently to increasing salinity.
For example, Lepidostoma declines, Diploperla has an optimum, and Cheumatopsyche increases.
The XC95 for genera like Cheumatopsyche are reported as "greater than" because extirpation did
not occur in the measured range.
Figure D-2 contains the weighted cumulative distribution function (CDF) and the
associated 95th percentile extirpation concentration value arranged in alphabetical order by
genus. Each point shows the weighted proportion of samples with each genus present at (F(x))
the conductivity less than the indicated conductivity value ([j,S/cm). The conductivity at the 95th
percentile is the XC95 (arrow). The CDF was calculated from observations from March through
October (all year; black connected points) from March through June (spring; green connected
points), and from July through October (summer; red connected points). As there were fewer
than 30 observations such as for Drunella between July and October, no CDF was developed for
the summer index period. In a CDF, genera that are affected by increasing conductivity
(e.g., Drunella) show a steep slope and asymptote well below the measured range of exposures,
whereas genera unaffected by increasing conductivity (e.g., Nigronia) have a steady increase
over the entire range of measured exposure and do not reach a perceptible asymptote.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 102 DRAFT—DO NOT CITE OR QUOTE
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
O
I
Cl
si
5
3
>5*
ft
Sk
:
0
1
i,
f
Co
a
o
*-+.
o
o
a
Co
5?
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
o
-fs.
|
ci
a
5
3
>5*
ft
Sk
:
0
<5-
1
i,
f
Co
a
o
o
o
a
Co
5?
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
O
L/i
>3*
|
ci
a
5
3
>5*
ft
Sk
:
0
<5-
1
i,
f
Co
a
o
*-+.
o
o
a
Co
5?
Q_ P
0) °
L_
3 "*
+-> ©
Q_ o
05
O .
¦+-»
CO
© -
-Q
©
cn
Q
o 8 _
CL
o
(I)
u.
3
3 -
Q.
CO
O
O o
i_
a.
2> , o
— to
¦8 d
_Q o
O
-------
25*
O
On
|
Cl
a
5
3
>5*
ft
Sk
:
0
<5-
1
i,
f
Co
a
o
*-+.
o
O
a
Co
5?
, t~;
'— o
JD
CD
o
o
0)
Q_ to
CD o
o °
Conductivity (jjS/cm)
Isoperla
I'S
JD
CD
_Q co
O 6
k_
CL
£
i_
3
•+—'
CL
CD
o
-------
s?
I
SS
Capture Probability and Relative Abundance Along Conductivity Gradient
Serratella
o
3
>5*
ft
Sk
s
<3*
*
si
§
in
0
<5-
1
i,
f
Co
a
o
*-+.
o
o
a
Co
5?
O °
,
d
4%
to
s
to
I
>>
00 ™
\
d
\ \
§ 1q « _
A\
^Ss\ °
\
CO o
s V*
o CO
X \
\ *
_Q
_Q
\ \\
- s e
ro
0.2
I
*~*o " %q\o
0.2
\ \ \ o
o n
u- cm
0-
'AO
\ \ *
CD d
'at
d 0
,Yp
L_
_ q 3
*\\
13
A1
o \\\°°
° Q_
iYo
CL -
Pf,
o
CO *-
r- CO O
'¦V,_
o
_ & o d
o
^ o
- d O
V
cP o^Sfc*
^_o
O {-)
aar9 o ; . . .
aSfw
ifS
n
. n
o
oo *^13 x -L
- S § -
o o
d d —
,
d
15 81 423 2221
Conductivity ((JS/cm)
11646
15 81 423 2221
Conductivity (pS/cm)
11646
15 81 423 2221
Conductivity (pS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
Conchapelopia
-------
O
00
I
Cl
a
5
3
>5*
ft
Sk
:
0
<5-
1
i,
f
Co
a
o
*-+.
o
O
a
Co
5?
.
CD °
-2
-*—> w
CL
CO
O 8
15 81 423 2221
Conductivity (pS/cm)
11646
Diphetor
15 81 423 2221
Conductivity (pS/cm)
—r
11646
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
o
'O
|
ci
a
5
3
>5*
ft
Sk
:
0
<5-
1
i,
f
Co
a
o
o
o
a
Co
5?
Cl
CO
O
¦=-*4- o
"i r
81 423 2221
Conductivity (|jS/cm)
11646
Stenonema
1 s
o
CL
o
CD
3 «*
Q. O
CO
O ..
n r
81 423 2221
Conductivity (pS/cm)
Amphinemura
11646
>,
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
>3*
|
a
5
3
>5*
ft
Sk
:
0
1
i,
f
• ^>3
a
o
*-+.
o
o
a
Zn
5?
CL
CO
O
"i 1 r
15 81 423 2221
Conductivity (|jS/cm)
11646
Stempellinella
n
CO fo
S*
Q_
o> {s,
3 <=>
+->
CL
CO
o -
v
\ Y o
vV" °
£3
So - „
81 423 2221
Conductivity ((jS/cm)
~T
11646
Dolophilodes
_Q O
CO
_Q
O
a. *
a> °
i—
3
-w
CL
CO
&_
"m 8
o<=
_ 8
~i 1 r
15 81 423 2221
Conductivity (pS/cm)
11646
Figure D-t. Observation probabilities for each genus (continued).
-------
Capture Probability arid Relative Abundance Along Conductivity Gradient
>3*
|
a
5
3
>5*
ft
Sk
:
0
1
i,
f
• ^>3
a
o
*-+.
o
o
a
Zn
5?
;
^ <
-»—«
Q.
CO
O
Q
\ \ O O 00
N. \ O
~G - bo®9w
o
o o
15
81
423
2221
r
11646
Conductivity ((jS/cm)
Cambarus
Is
O
L—
Q_ «
CD °
i—
3
Q_ ^
5^" °
CO
O
\Vc
^ TOO
°
0 o
15
81
423
2221
11646
Conductivity (pS/cm)
Maccaffertium
:>. °
-Q
CO
JD
O
i_
Q_
CD
L—
D
-*—•
Q_ o
CO
O
B -
& -
n 1 1 —i r
15 81 423 2221 11646
Conductivity (pS/cm)
Parachaetocladius
JD
CD
_Q
2
Q.
CD
¦*->
Q.
ra
O
81 423 2221 11646
Conductivity (|jS/cm)
Isonychia
_Q to
CO o
_Q
O
i_
CL
CM
CD d
ZJ
¦+-»
Q_
CO
O O
n —r
81 423 2221
Conductivity (fjS/cm)
Rheopelopia
11646
_Q
CO
_Q
2 !
Q_ ,
CD
L_
ZJ
-f—'
Q_ -
CO :
O 1
~i 1 1 —i r
15 81 423 2221 11646
Conductivity (pS/cm)
Figure D-l. Observation probabilities for each genus (continued).
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
Acentrella
Lirceus
_Q
CO
"8 <
S <
a.
a>
L_ I
U I
¦M
Q_
OJ
O
81 423 2221
Conductivity (pS/cm)
11646
Baetis
~i 1 1 1 r
15 81 423 2221 11646
Conductivity (jjS/cm)
±1 to
CO
_Q
O
£ s
Cl>
CL
CO Csl
o o
— - -^^.0
o.a
"y' O CP % D
-« -
rtr <—i
JD
CO !
_Q '¦
O
a;
CD
Q.
CO
o
Ectopria
°cP ~-NPp
_ S
T
15 81 423 2221
Conductivity ((JS/cm)
11646
JD
CO
.O
2
Q.
CD
I
o
Pseudolimnophila
~i r
15 81 423 2221
Conductivity (jjS/cm)
11646
"fr
O
><. <=>
_Q
CO
_Q
O
u_
Q_
a>
Q.
CO
O
s _
15
Cordulegaster
\
0
o
i
i
i
o
\o O
o
o
O " "sO^E
O °
O \ "V->D
81
423
2221 11646
Conductivity (pS/cm)
Figure D-l. Observation probabilities for each genus (continued).
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
Wormaldia
Glossosoma
Heleniella
_Q
CD
.O
O
a
2
D '
¦M
Q_
CD
o
\ \ i3
o
QO
%vsn o
o
o
o cd _
o o!
i 1 1 i r
15 81 423 2221 11646
Conductivity (pS/cm)
sAooo
~i r
15 81 423 2221
Conductivity (pS/cm)
11646
_Q
CD
_Q
O
qI
cd
L_
D
¦+-' ¦
Q_
CD
o
~T~
81 423 2221
Conductivity (pS/cm)
11646
Parakiefferiella
Brillia
Pisidium
n p
CD °
_Q
o
i—
Q- §
0)
t_
Q. P
(0 °
O
\ Q
°\ o
o
o o
15
81
423
2221
r
11646
O iCt
CD 3
_Q °
O
i—
CL
£ S
3 o
-W
CL
CD
o
: j
Conductivity ((jS/cm)
~i 1 1 n r
15 81 423 2221 11646
Conductivity (jjS/cm)
<*¦*1 - >
-Q o
CD o
JD
O X
o
Q_ o
CD co
3 S
§"8
O d
- 8
~T
81 423 2221
Conductivity ((jS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
Pagastia
Natarsia
Macronychus
-S s
^ o
_Q
O
CL co
0) §
3
Q_
TO
O d
~i 1 i r
15 81 423 2221 11646
Conductivity (pS/crn)
_Q °
CO
JD
O %
Q_ o
0
a ?
CL o
CO
° N
- oo o
~r
81 423 2221 11646
Conductivity ((jS/cm)
o
CO
'
CO
O -
d
- p
o
CO
'
o
*
o
jd
CO
JD
1
90 0
o
' /
90 0
l
to
o
o
k_
' /
d
CL
0>
3 _
_ 3
3
i_
D
o
o /
o
o
CL
CO
O /'/
0 ~ ''
'''
CM
O
O
;
D
-*-•
CL
CO
o :
1 1 i r
15 81 423 2221 11646
Conductivity (pS/cm)
8
d
^8 -
_c
C'j
n
o
o
Q_
0 fO
!— O
D o
-w
Q_
to fM
O p
n 1 r
15 81 423 2221
Conductivity {(jS/cm)
11646
JD
CO
-Q
O
D_
CD
3
Q_
CO
O
~i r
81 423 2221
Conductivity ((JS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
Tipula
Corynoneura
Rhagovelia
-Q
CO
JD
O
CD
¦
3
' .
CL
CD
O
~i 1 1 1 r
15 81 423 2221 11646
Conductivity (|jS/cm)
_Q
CO
JD
o
CL
Cl>
I—
3
"5.
CO
O
G_
81 423 2221 11646
Conductivity (jjS/cm)
~i 1 r
15 81 423 2221 11646
Conductivity (pS/cm)
Lanthus
Leuctra
Paragnetina
fsH
ro 8
-Q o
0
01 s
a) O
i—
3
-t—
CL
CO
O
3 _
%v\\ °
oo
cCSN
0
" O
DQOQO OO * "
" " " " ¦
_ 8
o
_ 3
o
s
15
81
423
2221
11646
^3 "
_Q
CO
_G ID
O o
I—
CL
CD
33
CL
CD
o
Conductivity ((jS/cm)
15 81 423 2221 11646
Conductivity (jjS/cm)
in
CO -
_Q o
O
3 °
Q_
CO tn
o s
°
o £®s#£p
o
O OTW1.
81 423 2221 11646
Conductivity (pS/cm)
Figure D-l. Observation probabilities for each genus (continued).
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
Crangonyx
Atrichopogon
Cardiocladius
~i 1 r
15 81 423 2221 11646
Conductivity ((jS/cm)
H
o- O
CD S
3 °
¦+->'
ro £
O p
\ °
*». o CD o °0
\ OO
o
O o Q) o
¦* «.
i 1 1 1 r
15 81 423 2221 11646
Conductivity (pS/cm)
_Q
CO
Ss
CL °
CD
i_
D
Q.
^ to
O q
~T
81 423 2221 11646
Conductivity (pS/cm)
G\
_Q
CD
-Q £0
O P
cl °
CD
S
CL °
CO
o
Phaeriopsectra
Sublettea
_ 3
Figure D-t. Observation probabilities for each genus (continued).
15 81 423 2221 11646
Conductivity (pS/cm)
>
°
-5
CO
SI
CL
-------
Capture Probability arid Relative Abundance Along Conductivity Gradient
^1
>3*
|
ci
a
5
3
>5*
ft
Sk
:
0
1
i,
f
Co
a
o
*-+.
o
o
a
Co
5?
i—
3
-w
Q_
CO
O
S -
_ 8
_ 8
_ 8
n 1 r
15 81 423 2221 11646
Conductivity (pS/cm)
Acroneuria
jQ
CO
_Q
O
¦-
CL
CD
i_
D
¦+-«
CL O
CO
O
fN
~i 1 1 r r
15 81 423 2221 11646
Conductivity (|jS/cm)
Limnophiia
JD <=>
CO
o £
£ d
2 o
u r
Q. °
CO
O g
c>
o
o
"i r
81 423 2221
Conductivity (pS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
-------
00
>3*
I
a
5
3
>5*
ft
Sk
:
0
<5-
1
i,
f
Co
a
o
*-+.
o
O
a
Co
5?
D
Q.
CD
o
8
©
3 _
Zoo \
©
>T \°
fN
oo^J r-^K X
O
;]/ \\
O
*9* "" \
O
:J:° ° \ \
\
— o
©
/// \
_ s
* // i.
o
* /'O o
\ \
- ?
/ /'o
o
Q'//
Cs)
/'A'o
-
o
©
- o
l 1 1 r
15 81 423 2221
Conductivity ((jS/cm)
11646
Ochrotrichia
COO o o
- s
~i 1 1 1 r
15 81 423 2221 11646
Conductivity ([jS/cm)
Perlesta
-Q id
CD *"
_q °
o
L—
o- o
E? 5
3
¦+-'
Q_
CD to
o §
~i r
15 81 423 2221
Conductivity (pS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
SB
25*
ci
si
5
3
>5*
ft
Sk
:
0
1
i,
f
Co
a
o
o
o
a
Co
5?
V_
3 O
-t—'
TO 2
o o
~r
81 423 2221
Conductivity (pS/cm)
11646
Dasyhelea
_Q
Cl;
JD
O
i_
Q_
0)
CL
TO
O 8
n—'—¦—i r
15 81 423 2221 11646
Conductivity (pS/cm)
Cryptochironomus
JD
TO
JD
2 <
CL ,
2>
3
-w
CL
TO ,
o
~r
81 423 2221
Conductivity (pS/cm)
11646
Microcyiioepus
81 423 2221
Conductivity (|jS/cm)
Microtendipes
11646
81 423 2221
Conductivity (pS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
Paratanytarsus
Rheocricotopus
Rheotanytarsus
n 1 1 1— r
15 81 423 2221 11646
Conductivity ((jS/cm)
8
o
>«.
•— in
— CM
_Q d
CO
-O o
O tN
t °
d) in
i— *-
D o
-t—
Q_
cu 2
O b
/ \ o
'/y° ^ \ *
//,' o A \
Ht ° \
\
/1*'0 \ \
o \
/ o/ '
\ \
15
81
423
2221
T
11646
_Q
CD
_Q
O
L_
Q_
Q)
i_
3
Q.
CD
o
Conductivity ((jS/cm)
T
81 423 2221
Conductivity (pS/cm)
11646
Caenis
Chimarra
_o
CO
o
£•
0) :
v—
3
Gi-
ro
O
n r
15 81 423 2221 11646
Conductivity (pS/cm)
15
fl m
CD o
3
Q_ B
CD b
o
Caecidotea
~i r
15 81 423 2221
Conductivity ((jS/cm)
11646
-Q
CO
_Q
O
¦—
Ql
a?
3
CL
CO
o
/ °
/
o /
° °
.-''Vv'o
0 O Q
15 81 423 2221
Conductivity (pS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
° ^
S?
|
s
5
>5*
ft
Sk
'<&
¦
m s
H S
B
O
2
Sn
H
!?¦
a
o
z
o
H
°r5r
HH (/Q
H
W|
o ®?
o
H
a
Clinocera
s -j
>,s
£ °
-Q ur>
03 2
-O s
O °
t_
A- o
0) o
3 °'
s~ ^
*? 8
O
b 3
3
o
o
8
T
81 423 2221
Conductivity (|jS/cm)
Parametriocnemus
11646
-&¦ °
s s
CO o
JD
Q ®
D_ o
0
¦5 8
q_ °
O ">
o
N. o
'9%
\ \ 0
» - D \
0^S\.
o°'-.
00
\
81 423 2221
Conductivity (pS/cm)
11646
Diamesa
_Q
CO
_q
o
0. <
a?
3 ,
Q- i
CD
O
_Q
CO
_Q
O
i_
CL
£
3
¦*—*
CL
CO
O
1 1 1 1
15 81 423 2221
Conductivity ((jS/cm)
I
11646
Polycentropus
S
O
\
s
0
<£>
O
V.
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
S?
I
ci
S
5
«s
>5*
ft
Sk
5r
n
0
1
i,
f
*>:
©
^•k
rs
©
H
!?¦
a
o
z
o
H
°r5r
HH (/Q
H
MJ
O ®?
o
H
a
Chaetocladius
CM _
O
_Q
CD
_Q
O
L_
CL
a>
i_
3 .
-~—»
Q_
CD
o
o
/
* Q"
° /' /
o qc^j^oo'"
,o- '/V ° o
- - ' o
°o°
15
81
423
2221
11646
Conductivity (jjS/cm)
Antocha
o°°
CL
CD
o
o C33G
"i r
15 81 423 2221
Conductivity (pS/cnn)
11646
Micropsectra
£ °
5
CD
_Q fo
O o
CL
CD
=
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
Cl
si
5
>5*
ft
Sk
5r
:
©
^•k
rs
©
H
!?¦
a
o
z
o
H
°r5r
HH (/Q
H
W|
o ®?
o
H
a
Paraphaenocladius
~i 1 1 1 —r
15 81 423 2221 11646
Conductivity (|jS/cm)
Thienemannimyia
r>
ro
£>
O
L_
CL
a>
3
Q_
S- S
CD q
o
o cm
81 423 2221
Conductivity (|jS/cm)
Hydropsyche
11646
£1
CD to
o o
o
L_
CL
CD
32
Q_
CO
O
Conductivity (pS/cm)
Figure D-l. Observation probabilities for each genus (continued).
15 81 423 2221 11646
—I T
2221 11646
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
Boyeria
Dubiraphia
Cheumatopsyche
T
81 423 2221
Conductivity (pS/cm)
11646
~r
81 423 2221
Conductivity (pS/cm)
11646
-Q
CD
_q
o
Q_
CD
!—
3
CL
CD
o
n r
15 81 423 2221
Conductivity (nS/cm)
11646
-Q
CD
_Q
O m
v— in
Q_ o
0>
la
O d
15
Tanytarsus
Hemerodromia
Hexatoma
0
o
o
0
o
o v-
o
o
,'''"o v -,
o * s
0
o
81
423
2221
r
11646
CD
_Q
O
Q_
Q)
i_
3
¦+-»
CL
CD
o
Conductivity (pS/cm)
~i 1 r
15 81 423 2221 11646
Conductivity (|jS/cm)
-G {£>
CD o
JD
O
D_
CD
u_
D ,
¦+—-
CL
CD
o
~i r
15 81 423 2221
Conductivity (pS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
Cl
si
5
>5*
ft
Sk
5r
"i 1 r
15 81 423 2221
Conductivity (pS/cm)
11646
Physella
81 423 2221
Conductivity (|jS/cm)
Tabanus
11646
.Q a
CO O
.Q
O
IX <0
2 §
3
CL
o a
~r
81 423 2221
Conductivity (pS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
-------
I
ci
S
5
«s
>5*
ft
Sk
:
3
©
^•k
rs
©
-r
^k
H
!?¦
a
o
z
o
H
°r5r
HH (/Q
H % o
_Q
CD
_Q
O
CL
Cl>
CL
CD
o
"i 1 1 r
15 81 423 2221 11646
Conductivity ((JS/crn)
Conductivity (pS/cm)
Atherix
~i r
15 81 423 2221
Conductivity (pS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
Capture Probability and Relative Abundance Along Conductivity Gradient
Thienemanniella Corydalus Cricotopus
-n r
2221 11646
11646
_Q
CD
_Q
O
— :
a. 1
Cl>
L_
3
-*—>
CL .
5
_q
CD
_Q
O
Q_
2 :
3
-~—»
CL
CD
o
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
Cl
si
5
>5*
ft
Sk
5r
:
©
^•k
rs
©
H
!?¦
a
o
z
o
H
°r5r
HH (/Q
H
W|
o ®?
o
H
a
Chironomius
Chrysops
Cladotanytarsus
15 81 423 2221 11646
Conductivity (pS/cm)
¦S <=
CD
o s
a. °
CD «o
l_ ¦*-;
D o
O o
_Q
CD
_Q
O
t_
0-
CD
i_
D
-t—'
CL
CD
o
81 423 2221
Conductivity (pS/cm)
Helichus
11646
JD
CD
_Q
O
u_
0L
o
81 423 2221
Conductivity ((jS/cm)
_ s
d
11646
11646
-------
Capture Probability and Relative Abundance Along Conductivity Gradient
s;
3
3
>;*
5r
¦
*
s;
s
to S
00
0
3
1
§•
>;
s
o
^•k
o
o
s
>1
3"
O
o
o
H
HH OtO
H
W|
o ^
o
H
W
Pseudochironomus
CO
.Q
o
S
d
a.
O
$
o
t 1 r
15 81 423 2221
Conductivity (|jS/cm)
11646
Figure D-l. Observation probabilities for each genus (continued).
-------
loq conductivity
loq conductivity
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus. Sites with pH <6 excluded.
Acroneuria
1.5 2.0 2.5 3.0 3.5
log conductivity
Amphinemura
Ablabesmyia
—\ 1 : 1 r~
1.5 2.0 2.5 3.0 3.5
log conductivity
Alloperla
1 5 2.0 2.5 3.0
log conductivity
Ameletus
Acentrella
-------
Antocha
Argia
Asellus
~\ 1 r
2,0 2.5 3.0
log conductivity
Atherix
3.5 4.0
log conductivity
Atrichopogon
2.0 2.5
log conductivity
Attenella
log conductivity
log conductivity
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
Baetis
Baetisca
Bezzia
1 1 1 1 r~
1.0 1.5 2.0 2.5 3.0 3.5
log conductivity
Boyeria
2.0
log conductivity
Caecidotea
—\ 1 1 1 r~
1.5 2.0 2.5 3.0 3.5
log conductivity
—\ 1 1 1 r~
1.5 2.0 2.5 3.0 3.5
log conductivity
1.5 2.0 2.5 3.0 3.5
log conductivity
~I 1 1-
2.0 2.5 3.0
log conductivity
Brillia
-------
Caenis
Calopteryx
Cambarus
log conductivity
Cardiocladius
log conductivity
Centroptilum
1.5 2.0 2.5 3.0
log conductivity
Ceratopsyche
1.5 2.0 2.5 3.0
loq conductivity
1.5 2.0 2.5 3.0
loq conductivity
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
Chaetocladius
Chelifera
Cheumatopsyche
i 1 1 1 r~
15 2.0 2.5 3.0 3.5
log conductivity
Chimarra
—\ 1 1 1 r~
1.5 2.0 2.5 3.0 3.5
loq conductivity
n 1 1 1 r~
1.5 2.0 2.5 3.0 3.5
log conductivity
h 1 1 1 r~
1.5 2.0 2.5 3.0 3.5
loo conductivity
2.0 2.5 3.0 3.5
log conductivity
Chironomus
2.0 2.5 3.0 3.5 4.0
log conductivity
Chrysops
-------
Cinygmula
Cladotanytarsus
Clinocera
1.5 2.0 2.5
log conductivity
Conchapelopia
~~i—
1.5
~T
~r
~T
2.0 2.5 3 0
ioq conductivity
2.0
log conductivity
Corbicula
2.5 3.0 3.5
Ioq conductivity
2.5 3.0
log conductivity
Cordulegaster
2.0 2.5
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Corydalus
Corynoneura
2.0 2.5
log conductivity
Cricotopus
loq conductivity
Crangonyx
log conductivity
Cryptochironomus
loq conductivity
2.0 2.5
log conductivity
Dasyhelea
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Demicryptochironomus
Diamesa
Dicranota
2.0 2.5
log conductivity
Dicrotendipes
log conductivity
2.0 2.5 3.0
log conductivity
Dineutus
log conductivity
2.0 2.5 3.0
log conductivity
Diphetor
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Diplectrona
Diploperla
Dixa
2.0 2.5 3.0 3.5
log conductivity
Dolophilodes
log conductivity
Drunella
1.5 2.0 2.5 3.0 3.5
loq conductivity
2.0 2.5 3.0
log conductivity
Dubiraphia
2.0 2.5 3.0
log conductivity
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Eccoptura
Ectopria
log conductivity
Ephemera
Epeorus
2.5 3.0 3.5
log conductivity
Ephemerella
2.0 2.5 3.0
loq conductivity
log conductivity
2.0 2.5 3.0
log conductivity
Eukiefferiella
1.5 2.0 2.5 3.0
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Eurylophella
Ferrissia
Fossaria
2.0 2.5
log conductivity
Gammarus
2.0 2.5 3.0 3.5
log conductivity
Glossosoma
3.0
log conductivity
Haploperla
log conductivity
2.0 2.5 3.0
log conductivity
2.5 3.0
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Heleniella
Helichus
Hemerodromia
2.0 2.5 3.0
log conductivity
Heptagenia
1.5 2.0 2.5
loq conductivity
log conductivity
Hexatoma
2.0 2.5 3.0
loq conductivity
log conductivity
Hydroporus
2.4 2.6
loo conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
Hydropsyche
1.5 2.0 2.5 3.0
log conductivity
Isoperla
1.5 2.0 2.5 3.0
loq conductivity
1.5 2.0 2.5 3.0 3.5
ioq conductivity
2.5
loo conductivity
Hydroptila
i 1 1 1 r
1.5 2.0 2.5 3.0 3.5
log conductivity
Lanthus
2.0 2.5 3.0 3.5
log conductivity
Krenopelopia
Isonychia
-------
Larsia
Lepidostoma
log conductivity
Leucrocuta
Leptophlebia
2.0 2,5 3.0
log conductivity
Leuctra
2.0 2.5
log conductivity
Limnophila
log conductivity
1.5 2.0 2.5 3.0
log conductivity
1.5 2.0 2.5 3.0
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Limnophyes
Limonia
Lirceus
1.5 2.0 2.5
log conductivity
Maceaffertium
log conductivity
1.5 2.0 2.5 3.0
log conductivity
Macronychus
log conductivity
2.0 2.5
log conductivity
Microcylloepus
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Micropsectra
Microtendipes
Microvelia
1 1 1 1 1 r
1.5 2.0 2.5 3.0 3.5 4.0
log conductivity
Nanocladius
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-1 1 r~
2.0 2.5 3.0
loq conductivity
-1 1 1 1 r~
1.5 2.0 2.5 3.0 3.5
log conductivity
Natarsia
1.5 2.0 2.5 3.0
ioq conductivity
1.5 2.0 2.5 3.0
log conductivity
2.0 2.5 3.0
log conductivity
Neophylax
-------
Nigronia
Nilotanypus
Nixe
2.0 2.5
log conductivity
Ochrotrichia
log conductivity
2.5 3.0
log conductivity
Optioservus
2.0 2.5 3.0 3.5
loq conductivity
1.5 2.0
log conductivity
Orconectes
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Pagastia
h 1 1 1 r~
1.5 2.0 2.5 3.0 3.5
log conductivity
Palpomyia
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
Orthocladius
Oulimnius
1.5 2.0 2.5 3.0
log conductivity
Paragnetina
q
~l 1 1 1 l-
1.5 2.0 2.5 3.0 3.5
log conductivity
Parachaetocladius
-1 1 r~
2.0 2.5 3.0
log conductivity
1 : 1 r~
1.5 2.0 2.5 3.0 3.5
log conductivity
2.0 2.5
log conductivity
-------
Parakiefferiella
Paraleptophlebia
Parametriocnemus
1.5 2.0 2.5 3.0
log conductivity
Paraphaenocladius
1.5 2.0 2.5 3.0 3.5
loq conductivity
1.5 2.0 2.5
log conductivity
Paratanytarsus
2.5 3.0 3.5
loq conductivity
2.0 2.5 3.0
log conductivity
Paratendipes
1.5 2.0 2.5 3.0
loq conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Peltoperla
Perlesta
2.0
log conductivity
Physella
Phaenopsectra
1.5 2.0 2.5
log conductivity
Pisidium
log conductivity
Plauditus
log conductivity
2.0 2.5
log conductivity
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
Poiypedilum
2.0 2.5 3.0
log conductivity
Promoresia
1.5 2.0 2.5 3.0
loq conductivity
Polycentropus
1 1 1 1 r~
.0 1.5 2.0 2.5 3.0 3.5
log conductivity
Procloeon
Potthastia
1.5 2.0 2.5 3.0
log conductivity
1.5 2.0 2.5 3.0
log conductivity
2.0 2.5
log conductivity
Prosimulium
-------
Prostoma
Psephenus
Pseudochironomus
2.0 2.5 3.0
log conductivity
Pseudolimnophila
2.0 2.5 3.0 3.5
log conductivity
Psychomyia
1.5 2.0 2.5 3.0
log conductivity
2.0 2.5 3.0
log conductivity
Pteronarcys
2.5 3.0
log conductivity
1.5 2.0 2.5 3.0 3.5
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Pycnopsyche
Remenus
2.0 2.5
log conductivity
Rheocricotopus
loq conductivity
CO
©
¦t
©
-------
Rhyacophila
Serratella
Sialis
log conductivity
Simulium
1.5 2.0 2.5 3.0
log conductivity
Sphaerium
1.5 2.0 2.5 3.0
loq conductivity
2.0 2.5 3.0
log conductivity
Stempellina
2.5
loq conductivity
2.0 2.5
loq conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Stempellinella
Stenacron
Stenelmis
2.5 3.0
log conductivity
Stenochironomus
1.5 2.0 2.5 3.0
log conductivity
Stenonema
2.0 2.5 3.0
log conductivity
2.0 2.5 3.0
log conductivity
Stictochironomus
1.5 2.0 2.5 3.0
log conductivity
2.5 3.0
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Stylogomphus
Sublettea
Sweltsa
log conductivity
Tabanus
2.0 2.5 3.0
log conductivity
Tallaperla
2.5 3.0
log conductivity
Tanytarsus
log conductivity
1.5 2.0 2.5 3.0
log conductivity
1.5 2.0 2.5
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Thienemanniella
Thienemannimyia
1.5
2.0 2.5 3.0
log conductivity
2.0 2.5 3.0
loq conductivity
~r
Tipula
2.5 3.0
log conductivity
Tribelos
2.0 2.5 3.0
loq conductivity
log conductivity
Tricorythodes
2.5
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
Tvetenia
lltaperla
Wormaldia
1.5 2.0 2.5 3.0
log conductivity
Yugus
~r
~T~
~r
1.8 2.0 2.2 2.4
tog conductivity
Zavrelia
1.5 2.0 2.5 3.0
loq conductivity
2.0 2.5
log conductivity
2.0 2.5 3.0
log conductivity
Zavrelimyia
log conductivity
Figure D-2. Cumulative distribution functions of observation probabilities weighted by sampling frequency for
each genus (continued).
-------
APPENDIX E
VALIDATION OF METHOD USING FIELD DATA TO DERIVE AMBIENT WATER
QUALITY BENCHMARK FOR CONDUCTIVITY USING KENTUCKY DATA SET
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
The method for developing the aquatic life benchmark for conductivity was validated by
developing XC95 and HC05 values using a data set independently collected by the Kentucky
Division of Water (KDOW) and comparing results with those found using the larger WV
database. Because samples were also drawn from the Central Appalachians (Ecoregion 69) and
Western Allegheny Plateau (Ecoregion 70) the two data sets were expected to give similar
results. Some differences were expected due to the different collection and taxa identification
protocols, shorter sampling window, inclusion of the Southwestern Appalachians
(Ecoregion 68), and the fewer number of samples in the Kentucky data set. Nevertheless, the
HC05 value was 319 [j,S/cm for the full Kentucky data set, which is very close to the West
Virginia result.
E.l. DATA SET SELECTION
The Southwestern Appalachians (68), Central Appalachia (69), and Western Allegheny
Plateau (70) ecoregions were selected for validation, because they are physiographically similar
to Ecoregions 69 and 70 in West Virginia (U.S. EPA, 2000; Omernik, 1987; Woods et al., 1996)
(see Figure E-l). Although the data set is smaller than the West Virginia data set, it was judged
to be large enough for validation of the method. These regions have heavily forested areas as
well as extensive areas developed for coal mining, and, as in West Virginia, conductivity has
been implicated as a cause of biological impairment in the three Kentucky ecoregions. The three
ecoregions were judged to be similar within the state of Kentucky in terms of water quality,
resident biota, and sources of conductivity. Confidence in the quality of reference sites was
relatively high owing to the extensively forested areas of the region. Background conductivity
was estimated from a probability sample from the U.S. EPA Wadeable Stream Assessment
(U.S. EPA, 2006) at the 25th percentile using the Spatial Survey Design Package (sp. survey R
package) (Stevens and Olsen, 2004). Background conductivity at the 25th percentile was
63 [j,S/cm for the Southern Appalachians, which includes Ecoregions 68, 69, and 70. When this
value is compared to the 25th percentile from a probabilistic subset of the WV data set, it was
similar to the 72 [j,S/cm value for Ecoregion 69, but much lower than the 153 [j,S/cm value for
Ecoregion 70.
E.2. DATA SOURCES
All data used in this study were taken from the Kentucky Division of Water, Water
Quality Branch database (KY EDAS). Chemical, physical, or biological samples were collected
from 274 distinct locations during February to October from 1998-2004 (see Table E-l). Like
WVDEP, the KDOW obtains biological data from both probability biosurvey and targeted
ambient biological monitoring programs. The probability biosurvey program provides a
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 158 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
condition assessment of the overall biological and water quality conditions for both basin and
state levels. Targeted ambient biological monitoring involves intensive data collection efforts
for streams of interest as reference or impaired sites or for other reasons. Most sites have been
sampled once during February to September. Quality assurance and standard procedures are
described by KDOW (2008). All contracted chemical analyses and macroinvertebrate
identifications followed internal quality control and quality assurance protocols. This is a
well-documented, regulatory database. The quality assurance was judged to be excellent based
on the database itself, supporting documentation, and experience of EPA Region 4 personnel.
E.3. DATA SET CHARACTERISTICS
Biological sampling usually occurred once during (February-October) with the KDOW
(1998-2004) wadeable sampling protocol. The Kentucky data set was treated in the same way
as the WV data used that was used to derive the aquatic life benchmark for conductivity. A
sample was excluded from calculations if (1) it lacked a conductivity measurement, (2) the
organisms were not identified to the genus level, or (3) the pH was low. Repeat biological
samples from the same location at the same time (or within a month) were excluded, but samples
collected in different months/years were not excluded from the data set. These repeat biological
samples from different years were retained and represented about 8% of the samples. All
samples were from wadeable streams. No sites with high chloride and low sulfate were
identified or removed from the Kentucky data set. We evaluated the effects of spring benthic
invertebrate emergence, seasonal differences in temperature and conductivities by partitioning
the data set into spring (February-June) and summer (July-October) subsets. Eighty-one
percent of the 95 genera used to develop the SSD for Kentucky also occurred in the WV SSD.
This indicates that sensitive genera still exist in both states. Genera from both states were judged
to be similarly susceptible to the effects of conductivity after exploratory analysis. Conductivity
ranged from 16 to 2,390 [j,S/cm for the Kentucky data set and 15 to 11,646 [j,S/cm for the WV
data set.
In the Kentucky database, 365 benthic invertebrate genera were identified. XC95 values
were not calculated for genera that occurred at <30 sampling sites and therefore, these genera
were not used to generate the SSD. Genera that did not occur at reference sites in West Virginia
were excluded from the SSD. Of the 365 genera collected, 95 occurred in at least 30 sampling
locations in Ecoregions 68, 69, and 70 (see Table E-4). Of the genera occurring in 30 or more
samples, all genera occurred in all three ecoregions.
KDOW samples benthic macroinvertebrates using methods similar to WVDEP (KDOW,
2008). KDOW collects 4-0.5 m2 kick samples in riffle/run habitat and composites them to yield
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 159 DRAFT—DO NOT CITE OR QUOTE
-------
1 aim2 sample. KDOW also supplements collections with multi-habitat qualitative sampling.
2 However, for consistency, these qualitative sampling data were not used in model construction,
3 only the riffle/run samples. Another notable difference in the WVDEP and KDOW methods is
4 that KDOW picks the entire sample in the laboratory, as opposed to WVDEP's fixed-count of
5 200 organisms. KDOW follows similar field and laboratory quality assurance methods as
6 WVDEP.
7
8 E.4. CONCLUSIONS
9 Despite the differences in method and in location, the HC0s was similar: 319 [j,S/cm for
10 Kentucky compared to 297 [j,S/cm for West Virginia (see Figures E-2 and E-3, Table E-2). The
11 95% confidence bounds for the Kentucky data set value are 180 [j,S/cm and 439 [j,S/cm which
12 overlap with the West Virginia data set confidence bounds of 225 [j,S/cm and 305 jaS/c. Genera
13 that exhibited a decreasing occurrence with increasing conductivity were among those with the
14 lowest XC95 values in both States. Table E-3 shows the 10 lowest XC95 values for both West
15 Virginia and Kentucky samples. The 5th percentile occurs near genus 7 for West Virginia
16 samples and genus 5 for Kentucky samples. Table E-4 lists the genera used to construct the SSD
17 from the Kentucky sample and their XC95 values.
18 Based on the similar results, we judged the method to be robust. The same aquatic life
19 benchmark appears to be applicable to West Virginia and Kentucky streams in Ecoregions 68,
20 69, and 70.
21
22 REFERENCES
23 KDOW (Kentucky Division of Water). (2008) Standard methods for assessing biological integrity of surface waters
24 in Kentucky. Commonwealth of Kentucky Environmental and Public Protection Cabinet Department for
25 Environmental Protection, Division of Water February 2008, Revision 3. 120 pp. Available online at
26 http://www.water.ky.gov/sw/swmonitor/sop/ and at http://www.water.ky .gov/NR/rdonlyres/714984BB-54F8-46B9-
27 AF27-290EF7A6D5CE/0/BiologicalSOPMainDocument03_08.pdf (accessed 12/19/2009).
28 Omernik, JM. (1987) Ecoregions of the conterminous United States. Ann Assoc Am Geograph 77:118-125.
29 Stevens, DL, Jr.; Olsen, AR. (2004) Spatially balanced sampling of natural resources. J Am Stat Assoc
30 99(465):262-278.
31 U.S. EPA (U.S. Environmental Protection Agency). (2000) Nutrient criteria technical guidance manual: rivers and
32 streams. Office of Water, Office of Science and Technology, Washington, DC. EPA/822/B-00/002. Available
33 online at http://www.epa.gov/waterscience/criteria/nutrient/guidance/rivers/rivers-streams-full.pdf.
34 U.S. EPA (U.S. Environmental Protection Agency). (2006) Wadeable streams assessment: a collaborative survey of
35 the Nation's STREAMS. Office of Research and Development, Office of Water, Washington, DC.
36 EPA/84l/B-06/002. December. Available online at http://www.epa.gov/owow/streamsurvey/(accessed
37 12/20/2009).
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 160 DRAFT—DO NOT CITE OR QUOTE
-------
1 Woods, AJ; Omernik, JM; Brown, DD; et al. (1996) Level III and IV ecoregions of Pennsylvania and the Blue
2 Ridge Mountains, the Ridge and Valley, and the Central Appalachians of Virginia, West Virginia, and Maryland.
3 U.S. Environmental Protection Agency, National Health and Environmental Effects Research Laboratory, Corvallis,
4 OR. EPA/600R-96/077. 50 pp.
5
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 161 DRAFT—DO NOT CITE OR QUOTE
-------
Table E-l. Number of samples with reported genera and conductivity.
Number of samples is presented for each month, ecoregion, and database.
Region
Month
Total
1
2
3
4
5
6
7
8
9
10
11
12
68
10
6
18
2
3
39
69
7
14
44
16
16
42
18
182
70
9
21
2
17
21
10
70
291
Table E-2. HCos values for Kentucky and West Virginia
Kentucky
HCos
West Virginia
HCos
February-October
319 [j,S/cm
March-October
297 [j,S/cm
February-June
397 [j,S/cm
March-June
322 [j,S/cm
July-October
641 [j,S/cm
July-October
479 [j,S/cm
Table E-3. Comparison of the sensitive genera and XC95 values
WV
KY
Rank
Genus
XC95
Rank
Genus
XC95
1
Remenus
101
1
Lepidostoma
132
2
Lepidostoma
109
2
Cinygmula
161
3
Cinygmula
224
3
Wormaldia
161
4
Leptophlebia
224
4
Dolophilodes
317
5
Alloperla
228
5
Drunella
320
6
Utaperla
240
6
Epeorus
324
7
Drunella
294
7
Neophylax
324
8
Pycnopsyche
299
8
Oulimnius
378
9
Ephemerella
302
9
Paraleptophlebia
400
10
Heptagenia
313
10
Ephemerella
467
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 162 DRAFT—DO NOT CITE OR QUOTE
-------
Table E-4. Extirpation concentration and sample size from Kentucky data
set. Highlighted genera are not found at WV reference sites. For a genus with a
decreasing capture probability with increasing conductivity, the XC95 is reported
directly. If the genus did not occur at the higher conductivity levels, it is reported
as a greater than or equal to value (>). Genera with increasing capture probability
with increasing conductivity are reported as greater than (>).
Genus
Both
Spring
Summer
XC95
N
XC95
N
XC95
N
1
Ablabesmyia
>1,410
43
879
30
2
Acentrella
>618
98
762
67
626
31
3
Acroneuria
>703
105
926
56
703
49
4
Ameletus
>507
69
762
68
5
Amphinemura
>1,287
107
1,980
105
6
Ancyronyx
>841
30
7
Antocha
>958
49
8
Aruia
>1,410
51
950
34
9
Atherix
>1,650
61
2,000
48
10
Baetis
>1,410
170
1,410
79
1,176
91
11
Boyeria
>1,410
92
1,410
36
879
56
12
Caenis
>1,410
85
1,410
35
2,340
50
13
( alopim \
>1,980
35
14
Cambarus
>1,132
157
1,163
95
841
62
15
Ceratopsyche
>1,580
102
1,980
34
1,203
68
16
Cheumatopsyche
>1,630
230
1,980
106
17
Chimarra
>2,000
90
2,260
34
2,000
56
18
Chironomus
>2,340
31
19
Cinygmula
161
39
183
39
20
CoiMaila
>1,863
84
1,863
60
21
Corydalus
>1,650
121
1,480
33
2,000
88
22
Cricotopus
>2,000
98
1,980
44
2,340
54
23
Diamesa
>1,980
54
1,980
43
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 163 DRAFT—DO NOT CITE OR QUOTE
-------
Table E-4. Extirpation concentration and sample size from Kentucky data set
(continued)
Genus
Both
Spring
Summer
XC95
N
XC95
N
XC95
N
24
Dinailus
879
45
841
33
25
Diplectro
>958
102
1,228
76
26
Diploperla
>702
35
1,157
34
27
Dolophilodes
317
31
28
Drunella
320
37
762
35
29
Dubiraphia
>1,650
86
1,650
65
30
Eccoptura
>1,228
31
31
Eclipidrilus
>1,322
92
1,271
47
934
45
32
Ectopria
>561
66
505
40
33
Elimia
>879
33
34
Ellagma
894
31
35
Epeorus
324
65
324
59
36
Ephemera
561
42
37
Ephemerella
467
70
485
67
38
Eukiefferiella
>1,650
54
2,260
38
39
Euryl ophella
>505
84
526
67
40
Gomphus
>1,047
36
41
Haploperla
485
37
485
32
42
Helichus
>1,050
147
1,520
80
1,132
67
43
Hemerodromia
>2,000
123
2,260
50
2,000
73
44
Hexatoma
>1,069
105
762
62
2,000
43
45
Hydropsyche
>1,650
160
1,980
66
2,000
94
46
Hydroptila
>1,863
58
1,863
41
47
Isonychia
>1,580
132
894
38
1,863
94
48
Isoperla
>1,176
80
1,157
77
49
Lanthus
1,520
34
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 164 DRAFT—DO NOT CITE OR QUOTE
-------
Table E-4. Extirpation concentration and sample size from Kentucky data set
(continued)
Genus
Both
Spring
Summer
XC95
N
XC95
N
XC95
N
50
Lepidostoma
132
30
51
Leucrocuta
>703
45
52
Leuctra
>1,113
131
889
91
2,000
40
53
Lirceus
958
35
54
Macronychus
>1,650
54
1,863
41
55
Microtendipes
>675
58
805
32
56
Natarsia
>1,630
45
57
Neophylax
324
73
431
67
58
Nigronia
>1,203
153
1,113
72
1,203
81
59
Oecetis
>2,000
31
60
Optioservus
>1,560
178
1,410
70
61
Orconectes
>1,302
115
1,287
41
1,277
74
62
Orthocladius
>1,480
49
1,480
39
63
Oulimnius
378
31
64
Paraleptophlebia
400
76
762
54
65
Parametriocnemus
>1,630
184
1,980
111
2,000
73
66
Peltoperla
>1,520
37
67
Perlesta
>1,287
51
1,520
33
68
Physella
>2,260
52
69
Plauditus
>703
55
70
Polycentropus
570
82
635
43
570
39
71
Polypedilum
>1,247
157
1,271
69
950
88
72
Procloeon
>802
42
768
34
73
Prosimulium
>958
53
401
46
74
Psephenus
>738
111
762
59
879
52
75
Pseudocloeon
855
36
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 165 DRAFT—DO NOT CITE OR QUOTE
-------
Table E-4. Extirpation concentration and sample size from Kentucky data set
(continued)
Genus
Both
Spring
Summer
XC95
N
XC95
N
XC95
N
76
Pseudolimnophila
>1,050
39
77
Pycnopsyche
>802
64
889
45
78
Rheocricotopus
>1,163
51
79
Rheotanytarsus
>1,580
115
1,113
36
1,863
79
80
Rhyacophila
>565
94
820
74
81
Sialis
>1,287
63
1,980
32
891
31
82
Simulium
>1,580
179
1,410
92
1,863
87
83
Stenacron
879
90
762
37
879
53
84
Stenelmis
>1,520
168
1,480
81
1,863
87
85
Stenochironomus
>802
35
86
Stenonema
>993
178
658
68
87
Stylogomphus
>1,863
89
1,480
40
1,863
49
88
Sweltsa
507
55
435
39
89
Tanytarsus
>1,287
118
1,163
54
1,863
64
90
Thienemannimyia
>1,630
139
1,410
71
2,000
68
91
Tipula
>,630
150
1,980
106
2,340
44
92
Triaenodes
841
31
93
Tricorythodes
>,000
48
2,340
44
94
Tvetenia
>,203
46
95
Wormaldia
161
38
1,980
34
XC95 = 95th percentile extirpation concentration reported as |iS/cm: NA = not applicable because it never occurs
at WVDEP reference locations; Both = February through October; Spring = Sampled February through June;
Summer = July through October; Empty cells indicates fewer than 30 occurrences during that season.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 166 DRAFT—DO NOT CITE OR QUOTE
-------
Legend
I | Ecoregion 68, 69, and 70
| | Kentucky
b 70 I
69
— r
160 80
0
160 Kilometers
Figure E-l. Location of Southern Appalachia (68), Central Appalachia (69),
and Allegheny Plateau (70).
Data source: State outlines from the U.S. EPA Base Map Shapefile Omernik
Level III Ecoregions from National Atlas (NationalAtlas.gov), projection:
NAD 1983 UTM 17 N. Map made December 21, 2009, by M. McManus.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 167 DRAFT—DO NOT CITE OR QUOTE
-------
o
CD
c
3
O"
CD
CD
>
'¦4—'
E
o
CO
o
CD
o
C\l
o
p
o
° Spr& Summer
A Spring
+ Summer
<4-
641 |jS/cm
397 |jS/cm
319 |jS/cm
A—
A&
9
\/+
T
T
200
500 1000
Conductivity (|jS/cm)
2000
Figure E-2. The species sensitivity distributions for all year (February
through October [black circles], February through June [blue triangles], and
July through October [red crosses +]). Ninety-five genera are included in SSD
using the all year data set. The HC0s is the conductivity at the intercept of the
CDF with the horizontal line at the 5th percentile. For all year, it is 319 [j,S/cm
with 95% confidence bounds at 180 ^iS/cm and 439 ^iS/cm.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 168 DRAFT—DO NOT CITE OR QUOTE
-------
LO
d
d
° Spr& Summer
A Spring
+ Summer
~nr~
0-
g?A +
G
319 pS/cm
©A-
o-
G-
* +
co
o
CM
O
641 |jS/C0J_0
o
o
397 |jS/cm
200
500 1000
Conductivity (|jS/cm)
2000
Figure E-3. Species sensitivity distribution for all year (black circles), spring (blue
triangles ), and summer (red crosses +). Only the lower half genera are shown to better
discriminate the points in the left side of the distribution.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 169 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX F
DATA SOURCES AND METHODS OF LANDUSE/LAND COVER ANALYSIS USED
TO DEVELOP EVIDENCE OF SOURCES OF HIGH CONDUCTIVITY WATER
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
F.l. OVERVIEW
Analysis of land use and cover was used to determine if there was a source of high
conductivity, to assess if land use was associated with conductivity levels, and to confirm the
relative proportion of ions associated with land use and cover types reported in the literature.
This information was used as evidence of preceding causation in the causal assessment described
in Appendix A of this report. A strategy used in this analysis was to limit the watersheds to
<20 km2 to minimize the number of land use and cover types within a single watershed, thereby
providing a clearer signal. However, because the region has a long history of mining, persistent
effects of mining were potentially present even when there was no current record of past or
present mining activity.
The final data set consisted of 191 small watersheds for which macroinvertebrate samples
were identified to genus, water chemistry was available from at least one sampling effort,
subwatershed area was <20 km2, and detailed land cover information was also available. The
data set of 191 sites was drawn from 2,151 sites in Ecoregion 69D described in the West Virginia
Department of Environmental Protection's (WVDEP) Watershed Assessment Branch Data Base
(WABbase). These 191 tributary watersheds were from the Coal, Upper Kanawha, Gauley, and
New Rivers. From each watershed, scatter plots for several parameters were generated for
eight land cover classifications: open water, agriculture, urban/residential, barren, valley fill,
mining, abandoned mine lands, and forested lands.
Although conductivity typically increases with increasing land use (Herlihy et al., 1998),
the densities of agricultural and urban land cover are relatively low, and a clear pattern of
increasing conductivity and increasing land use is not evident. At relatively low urban land use,
the range of conductivity is highly variable. This may be caused by unknown mine drainage,
deep mine break-outs, road applications, poor infrastructure condition (e.g., leaking sewers or
combined sewers), or other practices. In contrast, there is a clear pattern of increasing
conductivity as percent area in valley fill increases and decreasing conductivity with increasing
forest cover. Pairs of land use and water quality parameters with moderately strong correlation
coefficients (r > |0.50|) are listed in Table F-l. All other pairs exhibit r < |0.50| except a few
with spurious points or composed of only 2 points from which no evaluations could be made.
Biological effects measured as the West Virginia Stream Condition Index (WVSCI) score or the
genus level index of most probable stream status (GLIMPSS) score were weakly correlated with
percent forest cover and percent valley fill with r > |0.30|.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 171 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
F.2. GENERAL GEOGRAPHICAL INFORMATION SYSTEMS (GIS) DATA
DESCRIPTIONS
Numerous geographic information system (GIS) data sets are available for the State of
West Virginia. One such repository for West Virginia data, the West Virginia GIS Technical
Center (http://wvgis.wvu.edu/data/data.php), maintains publicly available shapefiles. WVDEP
also maintains a publicly available repository of statewide GIS data sets (http://gis.wvdep.org/).
All relevant GIS metadata are available for the data housed at each repository site. All GIS
coverages used in this U.S. EPA study are in universal transverse mercator (UTM) 1983 Zone 17
and the units are in meters. Table F-2 describes some of the publicly available GIS shapefiles
that were used as the total daily maximum load (TMDL) land use base files and the beginning
point for determining the 191 stations selected for the analyses described in Section F.3, and as
the beginning point for the 191 stations land use analysis described in Section F.4.
F.3. METHODS
The analysis for Appendix A proceeded in two steps; (1) selection of the 191 stations and
(2) land use analysis of the 191 stations. Section F.3 describes the selection process for selecting
the 191 sample stations, while Section F.4 describes the detailed land use evaluation for each of
the 191 stations. Figure F-l depicts the Ecoregion 69D in relation to the West Virginia State
boundary and the 8-digit watershed boundaries, while Figure F-2 shows the locations of the
191 stations within Ecoregion 69D.
191 Stations Selection Process within Ecoregion 69D with TMDL Land Use
• All WVDEP WAB stations located within Ecoregion 69D were selected. This generated
2,151 stations.
• The next station reduction occurred by selecting only stations where a macroinvertebrate
sample was collected and identified to the genus level. During this selection process,
stations had to have both a WVSCI and a GLIMPSS score. At least one chemistry
sample must accompany the macroinvertebrate sample from the same station location.
This narrowed the available stations to 825.
• To obtain the TMDL associated land use, stations located within the Coal, Upper
Kanawha, Gauley, and New River TMDL watersheds were selected. This narrowed the
selection to 382 stations.
• Stations were eliminated if detailed land use was not created during the TMDL process.
This eliminated 38 stations for a total of 344 stations.
• Next, a station was eliminated if it was located on an undelineated tributary stream that
was contained within a larger main stem subwatershed. Failure to remove these would
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 172 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
generate an overestimation of land use from the entire upstream contributing land use and
not simply the land use from the tributary where the sample was actually located. This
eliminated an additional 33 stations for a total of 311 stations.
• Lastly, the EPA workgroup limited the upstream contributing land use to stations with a
total watershed drainage area <20 km2 (4,942.08 acres). The total remaining stations in
TMDL watersheds within Ecoregion 69D after this last reduction was 191 stations (see
Figure F-2), and the data from these stations were assembled from 1997 to 2007, with the
majority of samples having been collected from 2001 to 2006. EPA workgroup consisted
of scientists from EPA Office of Research and Development and Region 3, contracted
scientists from Tetra Tech Inc., and scientists from the WVDEP.
F.4. LANDUSE MANIPULATIONS
To create the land use for the 191 stations, the original TMDL land uses from the Coal,
Upper Kanawha, Gauley, and New Rivers were used as the starting point. These land uses were
originally created by consolidating the available base land use (Gap Analysis Program [GAP]
2000 or NLCD) into more general categories and then adding more detailed source land use
categories (e.g., mining, oil and gas, roads) from detailed source information. To add these new
land use categories, GIS shapefiles were used to locate sources and assign areas. These areas
were then subtracted from the category they most likely would be attributed to in the original
base land use. For example, a disturbed mine site would likely be classified as barren in GAP, so
any area assigned as mining would be subtracted from barren to keep the total land use area in
the watershed the same. Table F-3 contains the WVDEP TMDL land use categories, the data
source from which the extent of the area and its location were determined, and the base land use
from which any newly created land use categories were subtracted.
Because the WVDEP TMDL land use manipulation process has undergone revisions and
enhancements since the initiation of the TMDL program, WVDEP TMDL land use data sets for
the Upper Kanawha, Coal, Gauley, and New Rivers were manipulated to have equivalent land
use when necessary and resulted in the consolidated land use for the 191 sampling stations. The
land use representation used in TMDL development for more recently developed TMDLs is
more detailed than that for TMDLs completed in earlier efforts. Therefore consolidation of the
detailed TMDL land use to seven basic land use categories was necessary. The valley fill GIS
coverage was then incorporated into the TMDL land use by subtracting the valley fill acreage
from Shank (2004) from the mining land use category. If more area was present in the valley fill
coverage than was present in the TMDL mining area for each TMDL subwatershed, the
remainder was subtracted from barren and then forest, respectively. The eight land use
categories calculated for each of the 191 WAB sampling stations used seven categories
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 173 DRAFT—DO NOT CITE OR QUOTE
-------
1 consolidated from the TMDL land use (see Table F-3), then included the addition of the valley
2 fill area.
3 F.5. CORRELATIONS WITH IN STREAM BIOLOGICAL AND WATER QUALITY
4 PARAMETERS
5 Spearman rank correlations of eight land use categories with conductivity and ion
6 concentration were calculated (see Table F-l). Individual scatter plots and associated correlation
7 coefficients for conductivity can be found in Appendix A (see Figure A-3). Land use and land
8 cover were arc sine square root transformed to better depict the upper and lower portions of the
9 distribution.
10
11 REFERENCES
12 Herlihy, A.T.; Stoddard, J.L.; Johnson, CB. (1998) The relationship between stream chemistry and watershed land
13 cover data in the mid-Atlantic region, U.S. Water, Air, and Soil Pollution 105:377-386.
14
15 Shank, M. (2004) Advanced integration of geospatial technologies in mining and reclamation conference,
16 December 7-9, 2004, Atlanta, GA.
17
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 174 DRAFT—DO NOT CITE OR QUOTE
-------
Table F-l. Pairs of land use and water quality parameters with correlations
coefficients >0.5 in the land use data set
Land use
Water quality parameter
r
Percent Forest
Conductivity
-0.56
Alkalinity
-0.51
Hardness
-0.65
Sulfate
-0.54
Calcium-total
-0.64
Magnesium-total
-0.58
Percent Mined Area
Hardness
0.56
Calcium-total
0.51
Magnesium-total
0.58
Percent Valley Fill Area
Conductivity
0.65
Alkalinity
0.50
Sulfate
0.64
Calcium-total
0.66
Magnesium-total
0.66
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 175 DRAFT—DO NOT CITE OR QUOTE
-------
Table F-2. Publicly available GIS data used to generate land cover estimates
Data information
Data description
Source
General sources of land use/land cover information
West Virginia GIS
Technical Center
General West Virginia Universities
GIS data repository location
http: //wvgis. wvu. edu/data/ data.php
WVDEP GIS data sets
General WVDEP's GIS data
repository location
http: //gis. wvdep. org/
Base Land use/land cover
GAP
GAP land use
http: //wvgis .wvu. edu/data/ dataset .php ?
ID=62
NLCD 2001
NLCD land use
http: //wvgis .wvu. edu/data/dataset .php ?
ID=269
Other files
Watershed Boundary
Datasets
USGS 8-digit Hydrologic Unit
Code boundaries
http: //wvgis .wvu. edu/data/dataset .php ?
ID=123
NHD Streams
National Hydrography Dataset
Streams
http: //wvgis .wvu. edu/data/dataset .php ?
ID=235
Abandoned Mine Lines
(AML-Highwalls) and
Polygons (AML Areas)
West Virginia abandoned mine
lands coverages. Highwall mine
coverage and AML area
http://wvgis.wvu.edu/data/dataset.php?
ID=150
OMR Mining NPDES
Permits and Outlets
WVDEP Office of Mining and
Reclamation NPDES permit and
outlet coverages
http://gis.wvdep.org/data/omr.html
Mining related Fills,
Southern West Virginia
WVDEP valley fills coverage from
2003
http: //gis .wvdep. org/data/omr .html
Mining Permit
Boundaries
WVDEP Mining permit boundaries
http: //wvgis .wvu. edu/data/dataset .php ?
ID=149
RoadsPaved
2000 TIGER/Line GIS and
WV_Roads shapefiles
http: //wvgis .wvu. edu/data/ data.php
RoadsUnpaved
2000 TIGER/Line GIS shapefile
and digitized from aerial
photographs and topographic maps
http: //wvgis .wvu. edu/data/ data.php
GAP = Gap Analysis Program; GIS = geographic information system; NHD = National Hydrography Dataset;
NLCD = National Land Cover Database; NPDES = National Pollutant Discharge Elimination System;
OMR = Office of Mine Reclamation; USGS = U.S. Geological Survey; WVDEP = West Virginia Department of
Environmental Protection.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 176 DRAFT—DO NOT CITE OR QUOTE
-------
Table F-3. Detailed WV TMDL Land Use Category Derivation and Land
Use Derivation used in Appendix A. Base land use categories highlighted in
grey.
Detailed WV TMDL
Land Use Category
Data source
Base land use from
which New Source Area
was subtracted
Land use categories
used in scatter plots
in Appendix A
Water
Water—base LU coverage
N/A
Water
Wetland
Wetland—base LU coverage
N/A
Water
Forest
Forest—consolidated all
forested types from base LU
coverage
N/A
Forest
Grassland
Grassland—base LU
coverage
N/A
Agriculture
Cropland
Cropland—consolidated all
cropland types from base
LU coverage
N/A
Agriculture
Urban Pervious
Urban—consolidated
urbanized types from base
LU coverage
N/A
Urban/residential
Urban Impervious
U rban—consolidated
urbanized types from base
LU coverage
N/A
Urban/residential
Barren
Barren—base LU coverage
N/A
Barren
Pasture
Source tracking
New area subtracted
from Grassland
Agriculture
Paved roads
Roads shapefiles
New area subtracted
from Urban Impervious
Urban/residential
Unpaved roads
Roads shapefiles
New area subtracted
from Urban Pervious
Urban/residential
Revoked Mining
Permits
AML information
New area subtracted
from Barren
AML
Abandoned Mine Land
AML shapefile
New area subtracted
from Barren
AML
Quarry
Mining shapefile
New area subtracted
from Barren
Mining
Highwall
AML shapefile
New area subtracted
from Barren
Mining
Oil and Gas
Oil and Gas shapefile
New area subtracted
from Barren
Mining
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 111 DRAFT—DO NOT CITE OR QUOTE
-------
Table F-3. Detailed WV TMDL Land Use Category Derivation and Land
Use Derivation used in Appendix A (continued)
Detailed WV TMDL
Land Use Category
Data source
Base land use from
which New Source Area
was subtracted
Land use categories
used in scatter plots
in Appendix A
Surface Mine Water
Quality permits
Mining shapefile
New area subtracted
from Barren
Mining
Surface Mine
Technology permits
Mining shapefile
New area subtracted
from Barren
Mining
Comingled mine deep
ground gravity
discharge
Mining shapefile
New area subtracted
from Barren
Mining
Comingled mine deep
ground pump
discharge
Mining shapefile
New area subtracted
from Barren
Mining
Undeveloped surface
mine WQ permits
Mining shapefile
New area subtracted
from Forest
Mining
Undeveloped surface
mine technology
permits
Mining shapefile
New area subtracted
from Forest
Mining
Undeveloped
comingled mine
gravity discharge
Mining shapefile
New area subtracted
from Forest
Mining
Undeveloped
comingled mine pump
discharge
Mining shapefile
New area subtracted
from Forest
Mining
Burned Forest
Forestry Dept. information
New area subtracted
from Forest
Barren
Harvested Forest
Forestry Dept. information
New area subtracted
from Forest
Barren
Skid Roads
Forestry Dept. information
New area subtracted
from Forest
Barren
TMDL land use
considers Valley Fill3
area as part of the
Surface Mine Water
Quality and
Technology Permit
information
WVDEP valley fills
coverage from 2003
New area subtracted
from Mining, Barren and
Forest, as appropriate
Valley fill
aValley fill land use was not part of the base TMDL land use and was specifically incorporated into the detailed land
use analysis during the EPA ion study for the 191 stations. See Table F-2 for the source file.
LU = land use; WQ = water quality.
This document is a draft for review purposes only and does not constitute Agency policy.
3/1/10 178 DRAFT—DO NOT CITE OR QUOTE
-------
§•
s;
3
3
>;*
¦
*
^S
s;
s
, ^
"--J >;
VO
0
s
1
§•
>;
s
o
^•k
o
o
s
>1
^k
3"
O
o
2
o
H
HH OtO
H
-------
° y
JS-
C's'
§•
s;
3
s
>;*
^s
^s
¦
*
^S
s;
^ §
|—'
00 Co
° O
s
I
s^
s §-
^
Tj »
H ©
3 §
2 §
o ^
H <£
hh Oq
H <3
W|
o ^
o
H
W
Upper Kanawha
River
Coal River
| p7 [Gaule^RiveJ
# #
## # &
%
t
*#* *
#« . #.
Lower New
River
Greenbrier
River
Upper Guyandotte
River
Upper New
River
Eco Region 69d
# 191 stations
WV_8_d ig it_h uc_bou ndaries
Eco region 69d
90 Miles
w
Figure F-2. One hundred ninety-one station locations used in the detailed land use analysis.
-------
|