4>EPA
United States
Environmental Protection
Agency
EPA/600/R-10/023F | March 2011 | www.epa.gov/ncea
A Field-Based Aquatic Life
Benchmark for Conductivity
in Central Appalachian Streams
Office of Research and Development
National Center for Environmental Assessment
-------
EPA/600/R-10/023F
March 2011
A Field-Based Aquatic Life Benchmark for
Conductivity in Central Appalachian Streams
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20460
-------
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
Cover Photo:
Used by permission, from Randall Sanger Photography. Photo of Ramsey Branch, West Virginia.
Back Cover Photos:
Used by permission, from Guenter Schuster. Photo of Barbicambarus simmonsi.
Used by permission, from North American Benthological Society. Photos of various invertebrates. © 2009 North
American Benthological Society. http://www.benthos.org/Education-and-Outreach/Media-
Galleries/Invertebrates.aspx?Page= 1.
Preferred Citation:
U.S. EPA (Environmental Protection Agency). 2011. 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-10/023F.
11
-------
CONTENTS
LIST OF TABLES v
LIST OF FIGURES vi
LIST OF ABBREVIATIONS AND ACRONYMS vii
PREFACE viii
AUTHORS, CONTRIBUTORS, AND REVIEWERS ix
ACKNOWLEDGMENTS xiii
EXECUTIVE SUMMARY xiv
1. INTRODUCTION 1
1.1. CONDUCTIVITY 1
1.2. APPROACH 3
2. DATA SETS 7
2.1. DATA SET SELECTION 7
2.2. DATA SOURCES 7
2.3. DATA SET CHARACTERISTICS 10
3. METHODS 15
3.1. EXTIRPATION CONCENTRATION DERIVATION 15
3.2. TREATMENT OF POTENTIAL CONFOUNDERS 19
3.3. DEVELOPING THE SPECIES SENSITIVITY DISTRIBUTION 20
3.4. CONFIDENCE BOUNDS 21
3.5. EVALUATING ADEQUACY OF NUMBER OF SAMPLES 23
3.6. ESTIMATING BACKGROUND 24
4. RESULTS 25
4.1. EXTIRPATION CONCENTRATIONS 25
4.2. SPECIES SENSITIVITY DISTRIBUTIONS 25
4.3. HAZARDOUS CONCENTRATION VALUES AT THE 5th CENTILE 25
4.4. UNCERTAINTY ANALYSIS 25
5. CONSIDERATIONS 27
5.1. CHOOSING TO USE FIELD VERSUS LABORATORY DATA 27
5.2. SELECTION OF THE EFFECTS ENDPOINT 28
5.3. TREATMENT OF MIXTURES 28
5.4. DEFINING THE REGION OF APPLICABILITY 29
5.5. BACKGROUND 35
5.6. SELECTION OF INVERTEBRATE GENERA 36
5.7. INCLUSION OF OTHER TAXA 36
5.8. TREATMENT OF LISTED SPECIES 36
5.9. INCLUSION OF REFERENCE SITES 37
in
-------
CONTENTS (contuined)
5.10. SEASONALITY, LIFE HISTORY, AND SAMPLING METHODS 37
5.11. FORMS OF EXPOSURE-RESPONSE RELATIONSHIPS 39
5.12. USE OF MODELED OR EMPIRICAL DISTRIBUTIONS 40
5.13. DUPLICATE SAMPLES 41
5.14. TREATMENT OF CAUSATION 41
5.15. TREATMENT OF POTENTIAL CONFOUNDERS 42
6. AQUATIC LIFE BENCHMARK 43
REFERENCES 45
APPENDIX A. CAUSAL ASSESSMENT A-l
APPENDIXB. ANALYSIS OF POTENTIAL CONFOUNDERS B-l
APPENDIX C. DATA SOURCES AND METHODS OF LAND USE/LAND COVER
ANALYSIS USED TO DEVELOP EVIDENCE OF SOURCES OF
HIGH CONDUCTIVITY WATER C-l
APPENDIX D. EXTIRPATION CONCENTRATION VALUES FOR GENERA IN
THE WEST VIRGINA DATA SET D-l
APPENDIX E. GRAPHS OF OBSERVATION PROBABILITIES FOR GENERA IN
THE WEST VIRGINIA DATA SET E-l
APPENDIX F. GRAPHS OF CUMULATIVE FREQUENCY DISTRIBUTIONS FOR
GENERA IN THE WEST VIRGINIA DATA SET F-l
APPENDIX G. VALIDATION OF METHOD USING FIELD DATA TO DERIVE
AMBIENT WATER QUALITY BENCHMARK FOR
CONDUCTIVITY USING A KENTUCKY DATA SET G-l
APPENDIX H. EXTIRPATION CONCENTRATION VALUES FOR GENERA IN A
KENTUCKY DATA SET H-l
APPENDIX I. GRAPHS OF OBSERVATION PROBABILITIES FOR GENERA IN
KENTUCKY DATA SET 1-1
APPENDIX J. GRAPHS OF CUMULATIVE FREQUENCY DISTRIBUTIONS FOR
GENERA IN KENTUCKY DATA SET J-l
IV
-------
LIST OF TABLES
1. Differences between the method used to derive the conductivity benchmark and
the method in Stephan et al. (1985) and the section of the report in which each is
discussed 3
2. Summary statistics of the measured water-quality parameters 11
3. Number of samples with reported genera and conductivity meeting our acceptance
criteria for calculating the benchmark value 12
4. Samples excluded from the original data sets of 2,668 samples used to develop
benchmark value 12
5. Genera excluded from 95th centile extirpation concentration calculation because
they never occurred at reference sites 13
6. Genera of threatened species included in the SSD 37
-------
LIST OF FIGURES
1. Points are sampling locations used to develop the benchmark from Level III
Ecoregions 69 and 70 in West Virginia
2. Box plot showing seasonal variation of conductivity (uS/cm) in the reference
streams of Ecoregions 69 and 70 in West Virginia from 1999 to 2006
3. Box plot showing seasonal variation of conductivity (uS/cm) from a probability-
based set of sample streams of Ecoregions 69 and 70 in West Virginia from 1997
to 2007 8
4. Box plot showing seasonal variation of conductivity (uS/cm) from the data set
used to develop the benchmark 9
5. Histograms of the frequencies of observed conductivity values in samples from
Ecoregions 69 and 70 from West Virginia sampled between 1999 and 2006 16
6. Examples of weighted CDFs and the associated 95th centile extirpation
concentration values 18
7. Three typical distributions of observation probabilities 18
8. The species sensitivity distribution 20
9. Species sensitivity distribution (expanded) 21
10. Diagram depicting the process for estimating the uncertainty of the HC0s 22
11. The cumulative distribution of XCgs values for the 36 most sensitive genera and
the bootstrap-derived means and two-tailed 95% confidence intervals 22
12. Adequacy of the number of samples used to model the HCos 23
13a. Anions 30
13b. Cations 31
13c. Dissolved metals 32
13d. Total metals 33
13e. Other water quality parameters 34
14. Comparison of full data set and subsets of spring and summer collected samples 38
15. Relationship of conductivity values sampled from the same site in spring and
summer 39
vi
-------
LIST OF ABBREVIATIONS AND ACRONYMS
CDF cumulative distribution function
CVs chronic values
DCX depletion concentration
GAM Generalized Additive Model
HCX hazardous concentration
LCX lethal concentration
LOWESS locally weighted scatterplot smoothing
SSD species sensitivity distribution
TMDL total maximum daily load
U.S. EPA United States Environmental Protection Agency
WABbase Water Analysis Database
WVDEP West Virginia Department of Environmental Protection
WVSCI West Virginia Stream Condition Index
XCX extirpation concentration
vn
-------
PREFACE
At the request of U.S. Environmental Protection Agency's (EPA) Office of Water and
Regions, the EPA Office of Research and Development has developed an aquatic life benchmark
for conductivity for the Appalachian Region. The benchmark is applicable to mixtures of ions
dominated by salts of Ca2+! Mg2+, SO42 and HCOs at a circum-neutral to alkaline pH. The
impetus for the benchmark is the observation that high conductivities in streams below surface
coal mining operations, especially mountaintop mining and valley fills, are associated with
impairment of aquatic life. However, application of the benchmark is not limited to that source.
The benchmark was derived by a method modeled on the EPA's 1985 methodology for
deriving ambient water-quality criteria for the protection of aquatic life. The methodology was
adapted for use of field data, by substituting the extirpation of stream macroinvertebrates for
laboratory toxicity data.
The methodology and derivation of the benchmark were reviewed by internal reviewers,
external reviewers, and a review panel of the EPA's Science Advisory Board (SAB). The SAB
panel's review was in turn reviewed by the Chartered SAB. The SAB review is available at
http://yosemite.epa.gov/sab/sabproduct.nsf/02ad90bl36fc21ef85256eba00436459/984d6747508
d92ad852576b700630f32!OpenDocument.
The SAB concluded that the benchmark is applicable to the regions in which it was
derived and the benchmark and the methodology may be applicable to other states and regions
with appropriate validation. In addition, hundreds of public commenters provided their views.
Comments from all of these sources were considered and used to improve the clarity and
scientific rigor of the document.
Vlll
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
AUTHORS
Susan M. Cormier, PhD
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Cincinnati, OH 45268
Glenn W. Suter II, PhD
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Cincinnati, OH 45268
Lester L. Yuan, PhD
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Washington, DC 20460
Lei Zheng, PhD
Tetra Tech, Inc.
Owings Mills, MD21117
CONTRIBUTORS
R. Hunter Anderson, PhD
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Cincinnati, OH 45268
Jennifer Flippin, MS
Tetra Tech, Inc.
Owings Mill s,MD 21117
Jeroen Gerritsen, PhD
Tetra Tech, Inc.
Owings Mill s,MD 21117
Michael Griffith, PhD
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Cincinnati, OH 45268
IX
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
CONTRIBUTORS (continued)
Michael McManus, PhD
U.S. Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
Cincinnati, OH 45268
John Paul, PhD
U.S. Environmental Protection Agency
National Health and Environmental Effects Research Laboratory
Research Triangle Park, NC 27711
Gregory J. Pond, MS
U.S. Environmental Protection Agency
Region III
Wheeling, WV 26003
Samuel P. Wilkes, MS
Tetra Tech, Inc.
Charleston, WV 25301
REVIEWERS
Charles Delos, MS
U.S. Environmental Protection Agency
Office of Water, Health and Ecological Criteria Division
Washington, DC 20460
Charles P. Hawkins, PhD
Western Center for Monitoring and Assessment of Freshwater Ecosystems
Department of Watershed Sciences
Utah State University
Logan, UT 84322
Christopher C. Ingersoll, PhD
U.S. Geological Survey
Columbia Environmental Research Center
4200 New Haven Road
Columbia, MO 65201
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
REVIEWERS (continued)
Charles A. Menzie, PhD
Exponent
2 West Lane
Severna Park, MD 21146
Margaret Passmore, MS
U.S. Environmental Protection Agency
Office of Monitoring and Assessment, Freshwater Biology Team
Wheeling, WV 26003
John Van Sickle, PhD
U.S. Environmental Protection Agency
National Health and Environmental Effects Research Laboratory, Western Ecology Division
Corvallis, OR 97333
Science Advisory Board Panel on Ecological Impacts of Mountaintop Mining and Valley
Fills
Duncan Patten, Chairman, PhD
Montana State University, Bozeman, MT
Elizabeth Boyer, PhD
Pennsylvania State University, University Park, PA
William Clements, PhD
Colorado State University, Fort Collins, CO
James Dinger, PhD
University of Kentucky, Lexington, KY
Gwendelyn Geidel, PhD
University of South Carolina, Columbia, SC
Kyle Hartman, PhD
West Virginia University, Morgantown, WV
Robert Hilderbrand, PhD
University of Maryland Center for Environmental Science, Frostburg, MD
XI
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
REVIEWERS (SAB Panel continued)
Alexander Huryn, PhD
University of Alabama, Tuscaloosa, AL
Lucinda Johnson, PhD
University of Minnesota Duluth, Duluth, MN
Thomas W. La Point, PhD
University of North Texas, Denton, TX
Samuel N. Luoma, PhD
University of California - Davis, Sonoma, CA
Douglas McLaughlin, PhD
National Council for Air and Stream Improvement, Kalamazoo, MI
Michael C. Newman, PhD
College of William & Mary, Gloucester Point, VA
Todd Petty, PhD
West Virginia University, Morgantown, WV
Edward Rankin, MS
Ohio University, Athens, OH
David Soucek, PhD
University of Illinois at Urbana-Champaign, Champaign, IL
Bernard Sweeney, PhD
Stroud Water Research Center, Avondale, PA
Philip Townsend, PhD
University of Wisconsin-Madison, Madison, WI
Richard Warner, PhD
University of Kentucky, Lexington, KY
xn
-------
ACKNOWLEDGMENTS
Susan B. Norton, Teresa Norberg-King, Peter Husby, Peg Pelletier, Treda Grayson, Amy
Bergdale, Candace Bauer, Brooke Todd, David Farar, Lana Wood, Heidi Glick, Cristopher
Broyles, Linda Tackett, Stacey Herron, Bette Zwayer, Maureen Johnson, Debbie Kleiser, Crystal
Lewis, Marie Nichols-Johnson, Sharon Boyde, Katherine Loizos, 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.
Xlll
-------
2-
EXECUTIVE SUMMARY
This report uses field data to derive an aquatic life benchmark for conductivity that can
be applied to waters in the Appalachian Region that are dominated by salts of Ca2+' Mg2+, SO.
and HCO3 at a 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 EPA's standard methodology
for deriving water-quality criteria (i.e., Stephan 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. 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 for deriving
water-quality criteria, as explained in Stephan et al. (1985), in that it used the 5th centile of a
species sensitivity distribution (SSD) as the benchmark value. SSDs represent the response of
aquatic life as a distribution with respect to exposure. Data analysis followed the standard
methodology in aggregating species to genera and using interpolation to estimate the centile. It
differs primarily in that the points in the SSDs are extirpation concentrations (XCs) rather than
median lethal concentrations (LCsos) 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 centile of the distribution of the probability of occurrence of a genus with respect to
conductivity was used as a 95th centile 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 centile 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 sensitive species or higher quality, exceptional waters designated by state and
federal agencies.
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.
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
xiv
-------
impairments must be assessed. Also, any variables that are correlated with conductivity and 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 can cause impairments and the relationship between conductivity
and biological responses apparently is not appreciably confounded.
The chronic aquatic life benchmark value for conductivity derived from all-year data
from West Virginia is 300 uS/cm. It is applicable to parts of West Virginia and Kentucky within
Ecoregions 68, 69, and 70 (Omernick, 1987). It is expected to be applicable to the same
ecoregions extending into Ohio, Pennsylvania, Tennessee, Virginia, Alabama, and Maryland, but
data from those states have not been analyzed. This is because the salt matrix and background is
expected to be similar throughout the ecoregions. The benchmark may also be appropriate for
other nearby ecoregions, such as Ecoregion 67, but it has only been validated for use in
Ecoregions 68, 69, and 70 at this time. This benchmark level might not apply when the relative
concentrations of dissolved ions are not dominated by salts of Ca2+' Mg2+, SC>42 and HCOs or
the natural background exceeds the benchmark. However, the salt mixture dominated by salts of
SO42 and HCO3 is believed to be an insurmountable physiological challenge for some species.
xv
-------
-------
1. INTRODUCTION
At the request of U.S. Environmental Protection Agency's (U.S. EPA) Office of Water
and Regions, 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 Ca2+! Mg2+, SO42 and HCOs at a 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 EPA's standard methodology for deriving water-quality criteria
(i.e., Stephan 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: Ca2+, Mg2+, Na+, K+
Anions: HCO3 , CO32 , SO42 , Cl
The salinity of water may be expressed in various ways, but the most common is specific
conductivity. Specific conductivity (henceforth simply referred to as "conductivity") is the
ability of a material to conduct an electric current measured in microSiemens per centimeter
(uS/cm) standardized to 25°C. (In this report, "conductivity" refers to the measurement, and
resulting data and "salinity" refers 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.
The ionic composition of mixtures of salts affects their toxicity (Mount et al., 1997). Therefore,
a measure such as conductivity is necessary because the effects of the salts are a result of the
magnitude of the exposure and the relative proportion of 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 benchmark based on an individual ion could be under-protective. Therefore, this aquatic
-------
life benchmark for conductivity is only appropriate for a mixture of salts dominated by the Ca2+,
Mg2+, SC>42 , and HCOs ions at a circum-neutral to mildly alkaline pH (6.0-10.0) in the
Appalachian Region.
Salinity has numerous sources (Ziegler et al., 2007). Freshwater can become increasingly
salty due to evaporation, which concentrates salts such as those in 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; U.S. EPA, 2011), and runoff from treating pavements for icy
conditions (Environment Canada and Health Canada, 2001; Evans and Frick, 2000; Kelly et al.,
2008).
Exposure of aquatic organisms to salinity is direct. Fish, amphibians, mussels, and
aquatic macroinvertebrates are especially exposed on their gills or other respiratory surfaces that
are in direct contact with dissolved ions in water. All animals have specific structures to
transport nutrient ions and control their ionic and osmotic balance (Bradley, 2009; Evans,
2008a, b, 2009; Wood and Shuttleworth, 2008; Thorp and Covich, 2001; Komnick, 1977; Smith,
2001; Sutcliff, 1962; Hille, 2001). However, these cell membrane and tissue structures function
only 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 in freshwater, their epithelium is selectively permeable to the uptake of certain
ions and less permeable to larger ions and water. Many freshwater organisms depend heavily on
specialized external mitochondria-rich chloride cells on the epithelium of their gills for the
uptake of salts and export of metabolic waste (Komnick, 1977). 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 offish
(Tarin et al., 2000).
Retention of ions is insufficient to maintain homeostasis and the actual uptake and export
of ions occurs at semipermeable membranes. Anion, cation, and proton transport occurs by
passive, active, uniport, and cotransport processes often in a coordinated fashion (Nelson and
Cox, 2005; Hille, 2001). These numerous specific mechanisms are involved in the toxicity of
solutions with relative ion concentrations different from what an organism typically encounters
(see Appendix Section A.2.3 for more details on physiological mechanisms).
-------
1.2. APPROACH
The approach used to derive the benchmark is based on the standard method for the
EPA's published Section 304(a) Ambient Water-Quality Criteria. Those criteria are the
5th centiles of species sensitivity distributions (SSDs) based upon laboratory toxicity tests, such
that the goal is to protect at least 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 respond at exposure level X\, the second most
sensitive species respond 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 (as seen in Posthuma et al., 2002). Centiles of the distribution can be derived
using interpolation, parametric regression, or nonparametric regression. It should be noted that
because SSDs are models of the distribution of sensitivity—and not just descriptions of the
relative sensitivity of a particular set of species—they can be broadly applicable. In particular,
SSDs derived using species from different continents are consistent for some chemicals (Hose
and Van den Brink, 2004; Maltby et al., 2005).
For the conductivity benchmark, the SSDs are derived from field data. 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, and field data have some
advantages for benchmark development (see Section 5.1). The differences between the method
used here and the traditional method for deriving water-quality criteria are presented in Table 1,
and the advantages are listed in Section 5.1.
Table 1. Differences between the method used to derive the conductivity
benchmark and the method in Stephan et al. (1985) and the section of the
report in which each is discussed
Difference
Used field rather than laboratory data
Used extirpation as the response rather than a LC50 or CV
Used an integrative measure of a mixture rather than a single chemical
Used data from a particular region
Used the macroinvertebrate taxa from biological surveys rather than
test species
Section
5.1
5.2
5.3
5.4
5.6, 5.7, and 5. 8
-------
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 more subject to a broader array of 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 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 rocks. 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 large-scale surface mining, but they may also come from
slurry impoundments, coal refuse fills, or deep mines. Other sources include effluent from waste
water treatment facilities and brines from natural gas drilling and coalbed methane production.
This benchmark for conductivity applies to waters influenced by current inputs from these
sources in Ecoregions 69 and 70 with salts dominated by SC>42 and HCOs anions at a
circum-neutral to mildly alkaline pH.
-------
N
A
. WV Sampling Points
Level 3 Ecoregions
70
Other Ecoregions
WV State Boundary
30
60
90
120 150 Kilometers
Figure 1. Points are sampling locations used to develop the benchmark from
Level III Ecoregions 69 (light grey) and 70 in West Virginia.
-------
-------
2. DATA SETS
Data are required to develop and substantiate the benchmark. 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; Pond, 2010; Gerritsen et al., 2010).
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. Conductivity values from WVDEP's reference sites were low and similar in different
months collected over several years (see Figure 2), providing evidence that the sites were
reasonable reference sites. The 75th centiles were below 200 uS/cm in most months. The
25th centiles from samples from a probability-based sample and from the full data set were below
200 uS/cm in most months (see Figures 3 and 4). Also, a wide range of conductivity levels were
sampled, which is useful for modeling the response of organisms to different levels of salinity.
2.2. DATA SOURCES
All data used for benchmark derivation were taken from the WVDEP's in-house Water
Analysis Database (WABbase) 1999-2007. The WABbase contains data from Level III
Ecoregions 66, 67, 69, and 70 in West Virginia (U.S. EPA, 2000; Omernik, 1987; Woods et al.,
1996). In this assessment, only data from Ecoregions 69 and 70 were used (see Figure 1).
Chemical, physical, and/or biological samples were collected from 2,542 distinct locations
(2,668 samples) 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-based 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
-------
500 -
E
o
200 --
I
o
O
100 -
50 -
20 -
I I I I I I I I I
Jan Feb Apr May Jun Jul Aug Sep Dec
Month
Figure 2. Box plot showing seasonal variation of conductivity (uS/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 centiles were below 200 uS/cm in all months except in June.
10000 -
— 1000 -
w
£• .—
Conductiv
->• c
->• o c
I I I
o
o
o o o
o ° o . o 4_
§ 1 J_ -I -3- !
1 1 1 1 ^^^~ I
1
1
1
1
1
1 — ' —
— ' — 1 1
1 1 1 1 1 1 1
Apr May Jun Jul Aug Sep Oct
Month
Figure 3. Box plot showing seasonal variation of conductivity (uS/cm) from a
probability-based set of sample streams of Ecoregions 69 and 70 in West
Virginia from 1997 to 2007. A total of 1,271 samples were used for this
-th
analysis. The 25 centiles were below 200 uS/cm (horizontal dashed line) except
in the September and October samples.
-------
10000 -
5000 -
—2000 -
o
^ innn —
^ 500 -
~
-a 200
o
0 100 -
50
20 -
o
o
0 ° §
0 a i
» o a
o R a
8 1 1 1 7
1 1 7 '
1 § -*- !
4 rh ~
Q ~^~ ' i — ' — i
— -i-
--
~^ _^ -*- """ -L~
1 1 1 1 1 1 1 1
Jan Feb Mar Apr May Jun Jul Au
§
e
o
o
0 °
- 1- 7
1 1
1
— 1=1
^^ o
— 1 —
1
1 1 1 1
g Sep Oct Nov Dec
Month
Figure 4. Box plot showing seasonal variation of conductivity (uS/cm) from
the data set used to develop the benchmark. A total of 2,210 samples from
2000 to 2007 from Ecoregions 69 and 70 in West Virginia are represented. The
25th centiles were below 200 uS/cm except in the August and November (n = 2)
samples. The wide range of conductivities allows the XCgs to be well
characterized.
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), and supporting
information are used by the State to develop 305(b) and 303(d) reports to the EPA (WVDEP,
2008b). All sites were in perennial reaches of streams.
Quality assurance and standard procedures are described by WVDEP (2006, 2008a).
WVDEP collects macroinvertebrates from a 1-m2 area of a 100-m reach at each site. When
using 0.5-m-wide rectangular kicknet (595-|i mesh), four, 0.25-m2 riffle areas are sampled. In
narrow or shallow water, nine areas are sampled with a 0.33-m-wide D-frame dipnet of the same
mesh size. Composited samples are preserved in 95% denatured ethanol. A random subsample
of 200 individuals +20% are identified in the laboratory. All contracted analyses for chemistry
and macroinvertebrate identification follow West Virginia's internal quality-control and
quality-assurance protocols. This is a well-documented, regulatory database. EPA judged the
quality assurance to be excellent based on the database itself, supporting documentation, and the
experience of EPA Region 3 personnel.
-------
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.
2) Information on the effects of dissolved salts on freshwater invertebrates was taken from
standard texts and other physiological reviews.
3) An EPA Region 3 data set was obtained from Gregory Pond which includes the original
data for Table 3 in Pond et al. (2008) and data collected for the Programmatic
Environmental Impact Assessment (Bryant et al., 2002). It was used to evaluate the
relative contribution of different ions in drainage from valley fills of large scale surface
mining and for other analyses related to causation (see Appendix A) and confounding
(see Appendix B). Some of these data were added since the 2010 public review draft.
4) The constituent ions for Marcellus Shale brine were provided by EPA Region 3 based on
analyses by drilling operators (see Appendix A).
5) Data for Kentucky are from the Kentucky Department of Water database and are
described in Appendix G, and results are presented in Appendices A, G, H, I, and J.
6) Geographic and related information are from WVDEP and various public sources and are
described in Appendix C and also used in Appendix A.
2.3. DATA SET CHARACTERISTICS
Biological sampling usually occurred once per year with minimal repeat biological
samples from the same location (5%). Multiple samples from the same location were not
excluded from the data set (see Section 5.13). Summary statistics for ion concentrations and
other parameters for the data set are provided in Table 2. The benchmark applies to waters with
a similar composition to those in Table 2.
A total of 2,210 samples from Ecoregions 69 and 70 were used in the determination of
the benchmark (see Figure 1 and Table 3). Data from a sampling event at a site were excluded
from calculations if they lacked a conductivity measurement (see Table 4). Samples were
excluded if the samples were identified as being from a large river (>155 km2) because the
sampling methods differed (Flotermersch et al., 2001). They were excluded if the salt mixture
was dominated by Cl rather than SC>42 (conductivity > 1,000 uS/cm, SC>4 < 125 mg/L, and
Cl > 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 Marcellus Shale brines.
10
-------
Table 2. Summary statistics of the measured water-quality parameters
Parameter
Conductivity
Hardness
Alkalinity
S042~
cr
Ca, total
Mg, total
TSS
Fe, total
NO2-NO3
Al, total
Al, dissolved
Fe, dissolved
Mn, total
Mn, dissolved
Total
phosphate
Se, dissolved
Se, total
Fecal coliform
DO
pH
Catchment
area
Temperature
Habitat
Units
uS/cm
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
counts/
100 mL
mg/L
standard
units
km2
°C
RBP score
Min
15.4
0.5
0.2
1
1
0.002
0.05
1
0.005
0.01
0.01
0.01
0.001
0.003
0.01
0.01
0.001
0
0
1.02
6.02
0.173
-0.28
49
25th
centile
146
50.2
30.5
17
o
J
13.6
3.7
3
0.123
0.1
0.09
0.02
0.02
0.02
0.03
0.02
0.001
0.001
36
8.2
7.27
2.311
15.1
115
Median
261
91.1
66.7
37
5.2
25.1
6.3
4
0.26
0.2
0.11
0.05
0.042
0.04
0.07
0.02
0.001
0.001
170
9.2
7.62
6.965
18.4
130
75th
centile
563
188
117
159
11.95
49.2
14
6
0.5
0.37
0.23
0.06
0.06
0.1
0.22
0.03
0.001
0.005
600
10.3
7.96
25.836
21.3
145
Max
11,646
1,492
560
6,000
1,153
430
204
190
110
30
12
0.93
11.8
7.25
1.06
2.36
1.26
1.26
250,000
18.35
10.48
153.014
31.9
192
Mean
281.5
97.1
55
51.6
6.5
25.5
7.3
4.3
0.26
0.20
0.15
0.04
0.05
0.05
0.07
0.03
0.001
0.002
151
9.3
7.59
7.644
17
127.8
Valid
N
2,210
1,148
1,425
1,428
1,118
1,154
1,150
1,442
1,433
1,178
1,436
1,287
1,259
1,430
20
1,181
313
496
2,035
2,182
2,210
717
2,210
2,186
Note: K+ and Na+ not measured; all means are geometric means except pH, DO, Temperature, and Habitat Score.
DO = dissolved oxygen; RBP = rapid bioassessment protocol; TSS = total suspended solids.
11
-------
Table 3. Number of samples with reported genera and conductivity meeting
our acceptance criteria (see Table 4) for calculating the benchmark value.
Number of samples is presented for each month and ecoregion
Region
69
70
Total
Month
Jan
8
4
12
Feb
4
33
37
Mar
1
4
5
Apr
63
187
250
May
187
232
419
Jun
103
179
282
Jul
79
194
273
Aug
269
237
506
Sep
232
120
352
Oct
54
8
62
Nov
0
2
2
Dec
6
4
10
Total
1,006
1,204
2,210
Table 4. Samples excluded from the original data sets of 2,668 samples used
to develop benchmark value
Parameter
Catchment
Conductivity
PH
High Cl"
Taxonomic
identification
Exclusion level
>155km2
No measurement
<6
>1,000 uS/cm, SO4 <125 mg/L, and Cl" >250 mg/L
Ambiguous taxa, family, or higher level
N of samples
excluded
295
10
147
4
2
The effects of low pH were eliminated 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). The freshwater chronic water-quality criterion requires waters to be maintained
between a pH of 6.5 and 9 (U.S. EPA, 1986). The conductivity benchmark was derived from
waters having a pH between 6.0 and 10. Thus the circum-neutral application brackets pH of
7 primarily in the range where pH is usually not toxic to freshwater organisms.
An organism was excluded from calculations if it was not identified to the genus level,
and a genus was excluded if it was never observed at reference sites or it was observed in
<25 samples. Invertebrate genera that did not occur at WVDEP Tier 1 reference sites
represented 11.4% of all genera (see Table 5). They were excluded so that the data would be
relevant to potentially unimpaired conditions—and so as to not include opportunistic salt-tolerant
organisms. Genera were excluded that were observed at fewer than 25 sampling locations in the
composited ecoregion thus ensuring reasonable confidence in the evaluation of the relationship
12
-------
-th
Table 5. Genera excluded from 95 centile extirpation concentration
calculation because they never occurred at reference sites
Argia
Baetisca
Calopteryx
Corbicula
Dineutus
Ferrissia
Fossaria
Leucotrichia
Nanocladius
Oecetis
Palpomyia
Paracladopelma
Paratendipes
Prostoma
Saetheria
Sphaerium
Stenochironomus
Stictochironomus
Tokunagaia
Tribelos
Tricorythodes
between conductivity and the presence and absence of a genus. This decision was made because
an analysis showed that the benchmark varied within <5% when SSD models were constructed
from >20 occurrences of each genus; whereas the benchmark steadily became lower when
XC95 values were derived from <15 occurrences.
In the WABbase, 497 benthic invertebrate genera were identified in Ecoregions 69
and 70. Those ecoregions had 308 genera in common. Of these, 220 genera occurred at least
once at 1 of the 70 reference sites in the two ecoregions. Genera that did not occur at reference
sites were excluded from the SSD (see Table 5). Greater than 95% 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. Of the 220 genera, 163 occurred at
>25 sampling locations in Ecoregions 69 and 70. Of the genera occurring at >25 sampling sites,
162 genera occurred in Ecoregion 69 and 163 in Ecoregion 70.
13
-------
-------
3. METHODS
The derivation of the benchmark for conductivity includes three steps: first, the
extirpation values (XCs) for each invertebrate genus was derived. Second, the XCgs values for
all genera were used to generate an SSD and the 5th centile of the distribution, the 5th centile
hazardous concentration (HCos). (The HCy terminology for concentrations derived from SSDs
is not in the EPA method [Stephan et al., 1985], but its usage 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.
We used the statistical package R, version Version: 2.12.1 (December 2010), for all
statistical analyses (http://www.r-project.org/).
3.1. EXTIRPATION CONCENTRATION DERIVATION
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 XCgs conductivity value.
This is a chronic-duration endpoint because the field data set reflects exposure over the entire life
cycle of the resident biota. The 95th centile was selected because it is more reliable than the
maximum value, yet it still represents the extreme of a genus's tolerance of conductivity. The
maximum value is sensitive to occurrences due to drifting organisms, misidentifications, or other
misleading occurrences.
The XCgs is estimated from the cumulative distribution of probabilities of observing a
genus at a site with respect to the concurrently measured conductivity at that site. Observed
conductivity values were nonuniformly distributed across a range of possible values (see
Figure 5), 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, a weighted cumulative distribution function was used to estimate the XCgs 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
15
-------
o
o
CM
O
O
O
CO
o
CD
o _
O
CM
O -
n~i m n n
T
T
T
32
100
316
1000 3162 10000
Conductivity (uS/cm)
Figure 5. Histograms of the frequencies of observed conductivity values in
samples from Ecoregions 69 and 70 from West Virginia sampled between
1999 and 2006. Bins are each 0.017 logic conductivity units wide.
0.017 log™ conductivity units wide, that spanned the range of observed conductivity values, a
total of 60 bins. We then calculated the number of samples that occurred within each bin (see
Figure 5). Each sample was then assigned a weight wt = lint, where nt 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:
16
-------
M,
where % is the conductivity value in the/h sample of bin /',
Nb is the total number of bins,
Mi is the number of samples in the ith bin,
Gy is true if the genus of interest was observed mfh sample of bin /',
/is an indicator function that equals 1 if the indicated conditions are true, and 0
otherwise.
The XCgs value is defined as the conductivity value, x where F(x) = 0.95. Equation 1 is
an empirical cumulative distribution function, and the output is the proportion of observations of
the genus that occur at or below a given conductivity level. 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 6
for the mayfly, Drunella. The horizontal dashed line indicates the point of extirpation where
F(x) = 0.95 intersects the CDF. The vertical dashed line indicates the XCgs conductivity value
on the x-axis.
This method for calculating the XCgs will generate a value even if the genus is not
extirpated. For example, the occurrence of Nigronia changes little with increasing conductivity
(see Figure 6). In order to examine the trend of taxa occurrence along the conductivity gradient,
we used a nonparametric function (Generalized Additive Model [GAM] with 3 degrees of
freedom) to model the likelihood of a taxon being observed with increasing conductivity (see
Figure 7). The solid line is the mean smoothing spline fit. The dots are the mean observed
probabilities of occurrence, estimated as the proportion of samples within each conductivity bin.
The conductivity at the red, vertical, dashed line is the estimated XCgs from the weighted
cumulative distribution (see Appendix E).
Because of the data distributions, not all 95th centiles correspond to extirpation, and some
imprecisely estimate the extirpation threshold. The following rules were applied to the XC95
values. If the GAM mean curve at maximum conductivity is approximately equal to 0 (defined
as less than 1% of the maximum modeled probability), then the XCgs is listed without
qualification. If the GAM mean curve at maximum conductivity is >0, but the lower confidence
17
-------
Drunella
Nigronia
32 100 316 1000
Conductivity (u,S/cm)
1 III\
32 100 316 1000 3162 10000
Conductivity (u,S/cm)
-th
Figure 6. Examples of weighted CDFs and the associated 95 centile
extirpation concentration values. The step function shows weighted proportion
of samples with Drunella or Nigronia present at or below the indicated
conductivity value (uS/cm). The XCgs is the conductivity at the 95th centile of the
cumulative distribution function (CDF) (vertical dashed line). In a CDF, genera
that are affected by increasing conductivity (e.g., Drunella) show a steep slope
and asymptote well below the maximum 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 clear asymptote.
Lepidostoma
Diploperla
Cheum atopsyche
I
•§.
tt.
8
S.
32 100 316 1000 3162 10000
Conductivity (nS/cm)
32 100 316 1000 3162 10000
Conductivity (nS/cm)
32 100 316 1000 3162 10000
Conductivity (nS/cm)
Figure 7. Three typical distributions of observation probabilities 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 lines fitted to the probabilities are for
visualization. The vertical red line indicates the XCgs- Note that different genera
respond differently to increasing salinity. Lepidostoma declines, Diploperla has
an optimum, and Cheumatopsyche increases. The XCgs for genera like
Cheumatopsyche are reported as "greater than" because extirpation did not occur
in the measured range.
18
-------
limit is approximating to 0 (<1% of the maximum mean modeled probability) the value is listed
as approximate (~). If the GAM lower confidence limit is >0, then the XCgs is listed as greater
than (>) the 95th centile. All model fits and the scatter of points were also visually inspected for
anomalies and if the model poorly fit the data, the uncertainty level was increased to either (~) or
(>). This procedure was applied to the distributions in Appendices E and I, and the results
appear in Appendices D and H. Also, these models were used to evaluate when genera began to
decline as evidence of alteration and sufficiency in Appendix A.
For example, the XCgs for Cheumatopsyche (an extremely salt tolerant genus) is
>9,180 uS/cm (see Appendices D and E). Whereas, although Pteronarcys is declining, the upper
confidence bound is >0; therefore, its XCgs is -634 uS/cm. The assignation of (>) and (~) does
not affect the HCos, but are provided to alert users of the uncertainty of the XCgs values for other
uses such as comparison with toxicity test results or with results from other geographic regions.
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. As mentioned
previously in Section 2.3, because low pH waters violate existing water-quality criteria and
because the data set was large, sites were excluded with pH <6 before identifying the XC95
values.
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. Because high and low exposures occurred in all seasons, we chose to
include the occurrence of a genus whenever it was observed. Therefore, although we explored
season, we could not justify excluding an observation of a genus just because it was seen outside
an imposed time frame.
Other potential confounders were evaluated by weighing the available evidence. Because
confounders are by definition correlated with the cause of concern and the effect, we determined
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 sensitive
genera. Other evidence of confounding was included when appropriate data were available.
19
-------
3.3. DEVELOPING THE SPECIES SENSITIVITY DISTRIBUTION
The SSDs are cumulative distribution plots of XCgs values for each genus relative to
conductivity (see Figures 8 and 9). The cumulative proportion for each genus P is calculated as
P = RI (N + 1) where R is the rank of the genus and TVis 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
TV when calculating the proportions responding because they fall in the upper portion of the SSD.
The HCos was derived by using a 2-point interpolation to estimate the centile between the XCgs
values bracketing P = 0.05 (i.e., the 5th centile of modeled genera). The benchmark is obtained
by rounding the HCos to two significant figures as directed by Stephan et al. (1985).
0)
c
0)
O
M—
o
c
o
••e
o
Q.
O
CO
CD
CD
CD
(N
CD
O
CD
O
O—
295 |jS/cm
200 500 1000 2000
Conductivity (|jS/cm)
l i
5000 10000
Figure 8. The species sensitivity distribution. Each point is an XCgs value for
a genus. There are 163 genera. The HCos (295 (is/cm) is the conductivity at the
intercept of the SSD with the horizontal line at the 5th centile.
20
-------
CD
M—
O
o
Q.
o
100 200 500
Conductivity (pS/cm)
1000
2000
Figure 9. Species sensitivity distribution (expanded). The dotted horizontal
line is the 5l centile. The vertical arrow indicates the HCos of 295 uS/cm. Only
the lower 50 genera are shown to better discriminate the points in the left side of
the full distribution.
3.4. CONFIDENCE BOUNDS
The purpose of this analysis is to characterize the uncertainty in the benchmark value by
calculating confidence bounds on the HCos values. Because the XC9s values were estimated
from field data and then the HC0s 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 (see Figure 10). Bootstrapping is commonly
used in environmental studies to estimate confidence limits of a parameter, and the method has
been used in the estimation of HCos values (Newman et al., 2000, 2002).
Bootstrap estimates of the XC95 were derived for each genus used in the derivation of the
benchmark by resampling 2,210 times (the number of observations in the data set) with
replacement (see Figure 10) (Efiron and Tibshirani, 1993). From each bootstrap sample, the
XC95 was calculated for each genus 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 XC9s values for
each genus. These distributions were used to calculate a two-tailed 95% confidence interval on
21
-------
the XCgs for each genus. The XCgsS from the original data set, the mean XCgsS of the bootstrap
distributions, and the confidence intervals are shown for the 36 most sensitive genera (see
Figure 11).
Original
2,210
observations
Repeat 1000 times
I
r
Randomly pick
2,2 10 samples
from observations
with replacement
-
Select reference
taxa that occurred
at 25 or more
samples
\^ J
-
Compute
XC95 and
HCos
values
^ J
-
f
Store
results
V
Calculate 95%
confidence
bounds
Figure 10. Diagram depicting the process for estimating the uncertainty of
the HC05.
o
I
•3.
a.
s 4
0
100
200 500
Conductivity (jjS/cm)
1000
Figure 11. The cumulative distribution of XCgs values for the 36 most
sensitive genera (red circles) and the bootstrap-derived means (blue
x symbol) and two-tailed 95% confidence intervals (whiskers). The 5l centile
is shown by the dashed line.
Uncertainty in the HCos value was evaluated by generating an HCos from each of the
1,000 sets of bootstrapped XC95 estimates. The distribution of 1,000 HC0s values was used to
generate two-tailed 95% confidence bounds on these bootstrap-derived values.
22
-------
3.5. EVALUATING ADEQUACY OF NUMBER OF SAMPLES
Bootstrapping was performed to evaluate the effect of sample size on the HCos and their
confidence bounds. This process is similar to the method used to calculate confidence bounds on
the HCos values (see Figure 10). A data set with a selected sample size was randomly picked
with replacement from the original 2,210 samples. From the bootstrap data set, the XCgs was
calculated for each genus by the same method applied to the original data and the HCos was also
calculated. The uncertainty in the HCos value was evaluated by repeating the sampling and HCos
calculation 1,000 for each sample size. The distribution of 1,000 HC0s values was used to
generate two tailed 95% confidence bounds on these bootstrap-derived values. The whole
process was repeated for a selected sample size range from 100 to 2,210 samples. The mean
HCos values, the numbers of genera used for HC0s calculation, and their 95% confidence bounds,
were plotted to show the effects of sample sizes. The HCos values stabilize at approximately
800 samples in this data set, which suggests that 800 is a minimum sample size for this method
(see Figure 12). Note that, the mean HCos value is lower than the actual HCos value at a similar
sample size, because the Monte Carlo results are asymmetrical (i.e., there are more ways that the
sample variance can result in lower values than higher values).
900
800
,o
4£ 700
~ 600
o
T3
o 500-
^ 400
300
L
t
I
k
\\ '
t
4
k
)
1
,
A
k
» t «
A
L...
^ J.
...A
T
d>
166
145
123 E
o>
0)
inoo
102M_
o
k_
or, -°
80 £
z
59
37
500
\ \ \
1000 1500 2000
Sample Size
Figure 12. Adequacy of the number of samples used to model the HCos. As
sample size increases the number of genera included in the SSD increases
(triangles). As sample size increases, the confidence bounds on the HC0s
decreases and the mean HCos is asymptotic at <300 (is/cm (circles).
23
-------
3.6. ESTIMATING BACKGROUND
In general, a benchmark should be greater than natural background. The background
conductivities of streams were estimated using that portion of the WABbase that consists of
probability-based samples. Those are samples from locations that were selected to represent
streams within a stream order with equal probability. The 25th centile of the probability-based
samples was selected as the estimate of the upper limit of background because disturbed and
even impaired sites are included in the sample (U.S. EPA, 2000). A total of
1,271 probability-based samples were collected from Ecoregions 69 and 70. The background
values on the 25th centile were 72 uS/cm for Ecoregion 69, 153 uS/cm for Ecoregion 70, and
116 uS/cm when samples from Ecoregions 69 and 70 are combined (see Figure 3). We also
estimated the background conductivity using reference sites in WABbase (see Figure 2). The
75th centiles from 43 sites in Ecoregion 69 and 27 sites in Ecoregion 70 are 66 uS/cm for
Ecoregion 69, 214 uS/cm for Ecoregion 70. When samples from Ecoregions 69 and 70 are
combined, the 75th centile is 150 uS/cm. Sampling locations were among the least disturbed
based on WVDEP's best professional judgment (WVDEP, 2008a, b); therefore, the 75th centile
was selected (U.S. EPA, 2000). The bases for selecting centiles are explained in Section 5.5.
24
-------
4. RESULTS
4.1. EXTIRPATION CONCENTRATIONS
The XCgs values are presented in Appendix D. Values are calculated for all
macroinvertebrate genera that were observed at a reference site and at a minimum of
25 sampling sites in the two ecoregions. Distributions of occurrence with respect to conductivity
are presented for each genus of macroinvertebrate in Appendix E and the CDFs used to derive
the XC95 values are presented in Appendix F.
4.2. SPECIES SENSITIVITY DISTRIBUTIONS
A SSD for invertebrates is derived from XC95 values of 163 genera (see Figure 8). 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 10 for better viewing of the points near the
5th centile of genera.
4.3. HAZARDOUS CONCENTRATION VALUES AT THE 5™ CENTILE
The hazardous concentration value at the 5th centile of the SSDs is 295 uS/cm. Rounding
the HCos to two significant figures yields a benchmark value of 300 uS/cm.
4.4. UNCERTAINTY ANALYSIS
The bootstrap statistics yield 95% confidence bounds of 228 and 303 uS/cm (see
Figure 11). The asymmetry of the confidence bounds with respect to the point estimate of
295 uS/cm is not unusual. In bootstrap-generated estimates, such as those 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 HCos values using the SSD approach,
given the data and the model. Conceptually, these confidence bounds may be thought of
representing the potential range of HCos values that one might obtain by returning to West
Virginia 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
25
-------
sampling using different protocols. The contributions of those sources of uncertainty—in
addition to the sampling uncertainty—can best be evaluated by comparing the results of
independent studies. One estimate of that 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 laboratory processing protocols,
the values (West Virginia: 295 uS/cm, Kentucky: 282 uS/cm) differ by <5% (see Appendix G
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 West Virginia results.
26
-------
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. CHOOSING TO USE FIELD VERSUS LABORATORY DATA
The standard methodology for deriving water-quality criteria uses results from laboratory
toxicity studies (Stephan et al., 1985); however, we have adapted the method to use field data
because suitable laboratory data are not available. Furthermore, SSDs based on laboratory
studies cannot replicate the range of conditions, effects, or interactions that occur in the field
(Suter et al., 2002). Although field data require additional assurance of attributable causation
due to potential confounders (Section 5.15, Appendices A and B), field data have many
advantages over laboratory data.
1) Field exposures include realistic levels, proportions, and variability of pollutant mixtures.
2) Field exposures occur in inherently realistic physical and chemical conditions.
3) Field exposures include regionally appropriate taxa and relative abundances of taxa.
4) Field studies can include more taxa than are available in laboratory data sets.
5) Field data include appropriately sensitive taxa and life stages.
6) Field data include pollutant interactions with migration, predation, competition, and other
behaviors.
7) Organisms in the field have realistic nutrition and levels of stress.
8) Organisms in the field realistically integrate effects of pollutants and other conditions into
a population response.
9) The field chronic endpoint (extirpation of a population) is inherently relevant, but the
chronic laboratory test endpoints correspond to no particular effect (chronic values—
CVs).
This study can benefit from these inherent advantages of field data because of the availability of
large, high quality data sets with clear effects of the pollutant and little evidence of confounding.
27
-------
5.2. 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 (DC*) 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. DCX 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 sampled species within that genus, that 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 could have evolved to partition niches afforded by
naturally occurring causal agents such as conductivity (Remane, 1971). 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 Stephan et al. (1985) by using
genera until such time that the advantages and disadvantages of using species can be more fully
studied.
Because this endpoint is based on full life-cycle exposures and responses of populations
to multigenerational exposures, it is considered a chronic-duration endpoint.
5.3. TREATMENT OF MIXTURES
In natural waters, salinity is a result of mixtures of ions. A metric is required to express
the strength of that mixture. We use conductivity because it is a measure of the ionic strength of
the solution, because it is related to biological effects, and because it is readily measured
accurately. However, conductivity per se is not the cause of toxic effects, and 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 Ca+,
Mg+, SC>42 , and HCOs ions at circum-neutral to mildly alkaline pH (pH 6-10). 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 are described by Pond et al. (2008). Whereas Ca2+,
Mg2+, SO42 , and HCO3 are the four most common ions to drain from surface coal mines
(Bryant et al., 2002), ions of Na+ and Cl are the two most common in seawater and brines from
28
-------
Marcellus Shale drilling operations (see Appendix A, Table A-16). Because the few sites with
very elevated Cl~ were found to be outliers in the distributions of occurrence, they were deleted
from the data set used to derive the XCgs 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 13a~e).
5.4. DEFINING THE REGION OF APPLICABILITY
If the method for developing a benchmark as 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, >95% of
genera in common, and a common dominant source of the stressor of concern.
Although Ecoregions 69 (Central Appalachia) and 70 (Western Allegheny Plateau) are
very similar, including similar bedrock types, the relative abundances differ. The coal-bearing
subregions of the Central Appalachians are 69a (Forested Hills and Mountains), and 69d
(Cumberland Mountains). According to Woods et al. (1996), "Ecoregion 69 ... is a high,
dissected, and rugged plateau made up of sandstone, shale, conglomerate, and coal of
Pennsylvanian and Mississippian age. The plateau is locally punctuated by a limestone valley
(the Greenbrier Karst; subregion 69c) and a few anticlinal ridges" (p.30). Ecoregion 70 has more
heterogeneous bedrock formations than subregions 69a and 69d. It is underlain by shale,
siltstone, limestone, sandstone, and coal, including the interbedded limestone, shale, sandstone,
and coal of the Monongahela Group and the Pennsylvanian sandstone, shale, and coal of the
Conemaugh and Allegheny Groups (Woods et al., 1996).
Individual analyses of Ecoregions 69 and 70 result in a somewhat lower HC0s value for
Ecoregion 69 and a somewhat higher value for 70 (254 |iS/cm in Ecoregion 69 and 345 |iS/cm in
Ecoregion 70). This difference might be attributed to the background water chemistry (see the
following section). However, if the genera were adapted to high conductivity in Ecoregion 70
and low conductivity in 69, or if they were represented by more resistant species in 70 and more
sensitive species in 69, it would be expected that the XC95 values would consistently go up in
Ecoregion 70 and down in Ecoregion 69 relative to the values in the combined data set.
However, XCgs values go up and down in both ecoregions when they are analyzed individually.
29
-------
Conduct
-0.5 0.5 1.5 2.5
I I I I I I I
0.78
i i i i i r
1.5 2.0 2.5 3.0 3.5 4.0
Alkalinity
0.89
0.0 1.0 2.0 3.0
I I I I I I
0.6
Sulfate
0.64
0.56
0.41
Chloride
Figure 13a. Anions. Matrix of scatter plots and absolute Spearman correlation
coefficients between conductivity (uS/cm), alkalinity (mg/L), sulfate (mg/L), and
chloride (mg/L) concentrations 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).
30
-------
0.0
1.0 2.0
J I
3.0
-2-101 2
I I I I I
Conduct
0.95
0.93
0.92
i i i i i r
1.5 2.0 2.5 3.0 3.5 4.0
Hardn
0.96
Mg
0.99
0.91
Ca
Figure 13b. Cations. Matrix of scatter plots and absolute Spearman correlation
coefficients between conductivity (uS/cm), hardness (mg/L), Mg (mg/L), and Ca
(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).
31
-------
-2.0 -1.0 0.0
I I I I
-2.0 -1.5 -1.0 -0.5 0.0
I I I I
Conduct
0.64
0.14
0.12
0.09
Dis Mn
NA
0.01
0.28
vtf
Dis Se
0.21
0.06
Dis Al
0.07
Dis Fe
1.5 2.5 3.5
\ \ \ \ \ \
-3.0 -2.0 -1.0 0.0
\ \ \ \ \
-3 -2-10 1
Figure 13c. Dissolved metals. Matrix of scatter plots and absolute Spearman
correlation coefficients among conductivity (uS/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).
32
-------
-2.5 -1.5 -0.5 0.5
I I I I I I
-2-101 2
I I I I
Conduct
0.35
0.09
0.03
0.13
Mn
0.13
0.57
0.27
Se
0.08
0.13
a
Fe
0.59
Al
1.5 2.5 3.5
-3 -2
\ I I I I \
-2.0 -1.0 0.0 1.0
Figure 13d. Total metals. Matrix of scatter plots and absolute Spearman
correlation coefficients between conductivity (uS/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).
33
-------
024
60 140
5 15
-2.0 0.0
Conduct
0.52
PH
0.4
0.31
0.26
0.2
0.29
0.25
0.24
0.35
0.25
0.07
0.22
0.18
0.01
0.08
0.11
0.14
0.49
0.04
0.06
0.06
0.07
0.02
0.08
Fecal
«°*
m
Watshed
0.12
0.25
0.1
Hab Sc
0.13
0.1
0.64
Errfced
1.5 3.0
-1.0 1.0
TT I TT
0 10 20
0.11
0.14
0.13
0.08
DO
0.11
0.05
0.03
0.09
0.07
TP
Tr
-2.0 -0.5
0.03
0.02
0.23
0.11
0.05
0.14
NO23
Figure 13e. 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).
Conductivity is logarithm transformed specific conductance (uS/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); NO23 is logarithm-transformed
nitrate and nitrite (mg/L).
34
-------
The differences in HCos values appear to be due primarily to random differences in which rarer
genera do not meet the minimum sample size of 25 occurrences in a region. When the data set is
split by ecoregion, the SSD model is reduced by 3 Igenera for Ecoregion 69 and 35 genera for
Ecoregion 70. Furthermore, the two Ecoregions had similar genera, and, although Ecoregion 70
had a slightly higher estimated background, there were sites that had conductivity below 100
suggesting that the truly undisturbed background would be low. Overall we could not we could
not justify the increase in uncertainty associated with the reduced sample size and number of
genera. Therefore, EPA did not derive benchmarks for individual ecoregions.
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. Background levels may be estimated from
reference sites, which are sites that are judged to be among the best within a category. However,
because disturbance is pervasive, reference sites are not necessarily pristine or representative of
natural background. Many reference sites have unrecognized disturbances in their watersheds or
have recognized disturbances that are less than most others in their category. Some may 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 centile is based on
precedent and on the collective experience of EPA field ecologists (U.S. EPA, 2000). Estimated
background conductivities for Ecoregions 69, 70, and both combined are 66, 214, and
150 uS/cm, respectively, using 75th centiles of reference sites in West Virginia.
Alternatively, background values may be estimated using samples from a
probability-based design. Such samples include all waters within the sampling frame, including
impaired sites, with defined probability. In some regions, there are no undisturbed streams. To
characterize the best streams, the 25th centile is commonly used by EPA field ecologists
(U.S. EPA, 2000). Based on the 25th centiles, estimated background conductivities for
Ecoregions 69, 70, and both combined are 72, 153, and 116 uS/cm for probability-based samples
in West Virginia.
Background between Ecoregions 69 and 70 appear to be different; however, none of
these values exceed the benchmark value of 300 [j,S/cm. The higher estimates of background
conductivity in Ecoregion 70 relative to Ecoregion 69 may be attributed to the variable
occurrence of limestone and limestone-derived soils. The higher level of development and
population density in Ecoregion 70 may also contribute, but it was not evaluated.
35
-------
5.6. SELECTION OF INVERTEBRATE GENERA
Selection of genera to model can affect the results. Using the data set of all taxa includes
species 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 the HCos by less than 2%.
Genera are also selected for statistical reasons. We restricted genera used in the analyses
to those recorded at a minimum of 25 sampling sites to reduce the chance that an apparent
extirpation is due to sampling variance and to increase the likelihood that the models and
quantitative analyses for potential confounding are reasonably strong.
5.7. INCLUSION OF OTHER TAXA
Inclusion of other taxa are recommended under the EPA's 1985 criteria derivation
methodology (Stephan et al., 1985) solely to ensure that other taxonomic groups are not more
sensitive than those already evaluated. Fish were not included because their occurrence is
strongly affected by stream size making it difficult to determine XCgs values. Indeed, 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 25 samples probably owing to the WVDEP sampling
methods. Additional analyses may be necessary to ensure protection of federally or state listed
rare, threatened, or endangered species offish, amphibians, and mussels.
5.8. TREATMENT OF LISTED 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 in
Table 6, that were included in the SSD, be identified to species in subsequent monitoring to
evaluate the risk to these threatened taxa. Also, some genera of listed species were not included
36
-------
Table 6. Genera of threatened species included in the SSD (WVDNR, 2007)
Genus
Allocapnia
Alloperla
Caecidotea
Calopteryx
Cambarus
Cordulegaster
Crangonyx
Common Family Name
stonefly
stonefly
isopod
j ewelwing
crayfish
spiketail
amphipod
Genus
Diploperla
Ephemera
Orconectes
Pteronarcys
Stenonema
Sweltsa
Utaperla
Common Family Name
stonefly
mayfly
stonefly
stonefly
mayfly
stonefly
stonefly
in the SSD because the genus was not collected in sufficient numbers, such as from the genera
Gomphus, Hansonoperla, Macromia, and Ostrocerca. 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 centile when necessary to protect
threatened, endangered, or otherwise important species. Rare species may be ecologically
important.
5.9. 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 could not be derived by this
method in that region. In this case, WVDEP's reference sites were included as well as many
probability-based sites with >90% forest cover, which are believed to be representative of
good-to high-quality systems.
5.10. 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.
The effects of seasonality and life history were evaluated by comparing HC05 values
partitioned for season. The data set was partitioned into spring and summer based on seasonal
37
-------
patterns of conductivity in the full data set (see Figure 4) and the HCos was calculated. The
spring season is March through June. The summer season is July through October. The HCos
values in the truncated data sets are 317 uS/cm for spring that included 132 genera and
415 uS/cm for summer that included 120 genera. The greater summer HCos is due to the loss of
sensitive taxa from the SSD. The lower end of the SSD for the full data set and spring samples
are fairly similar (see Figure 14). Lower effects levels in the spring were not due to an
insufficient test range of conductivities because exposures as high as 5,200 uS/cm occurred in
the spring samples. Because the spring data set included both sensitive genera and a full range of
exposures, it was judged more reliable than the summer model.
CD
•6
200
500 1000 2000
Conductivity ftjS/cm)
5000
10000
Figure 14. Comparison of full data set (circles) and subsets of spring
(inverted triangles) and summer (triangles) collected samples. Spring consists
of 132 genera, summer of 120 genera. The SSD for the full data set and summer
are similar until XC95s of 1,000 |iS/cm. The summer SSD lacks sensitive genera.
Because we cannot be sure whether the greatest exposures in summer are tolerated by the
spring-emergent genera, we estimated the likelihood that conductivity would increase in the
summer. Sampling locations with at least one spring and summer conductivity measurement
were identified. The spring season is March through June. The summer season is July through
October. High and low conductivity streams are represented in both spring and summer samples.
The conductivity in certain streams was three times greater in the summer than the spring.
38
-------
However, streams with conductivity <300 uS/cm in summer are below the benchmark in
spring 98% of the time (see Figure 15). So, if a stream meets the benchmark in summer, it is
likely to meet it year-round. Therefore, seasonal variation should be considered when planning
monitoring of conductivity and should include ample samples in the spring to ensure inclusion of
sensitive genera.
o
8
o
o
> o
= o -
o •<-
T3
O
O
CD
I O
E o
10
100
1000
r
10000
Spring Conductivity
Figure 15. Relationship of conductivity values sampled from the same site in
spring and summer. When conductivity is <300 uS/cm (broken lines) in March
thru June, the conductivity is <300 uS/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.
5.11. FORMS OF EXPOSURE-RESPONSE RELATIONSHIPS
The diversity of the forms of the exposure-response relationships (i.e., decreasing,
unimodal, decreasing, and no relationship) (see Figure 7 and Appendix E) has required some
39
-------
methodological decisions. The forms are expected given the nature of the salts and the variance
in sensitivity. The salt mixture includes nutrient elements, and, like other pollutants that are
nutrients at low exposure levels (e.g., copper and selenium), the response to this mixture is
expected to have a unimodal distribution (see Figure 7, Diploperld). In the ascending (left) limb,
nutrient needs are increasingly being met. In the descending (right) limb, toxicity is increasing.
However, many of the empirical exposure-response relationships do not display both limbs.
They may show: (a) the descending portion of the curve because none of the observed
conductivity levels are sufficiently low to show deficiency for the taxon (see Figure 7,
Lepidostomci); (b) the ascending portion because none of the observed conductivity levels are
sufficiently high to show toxicity for the taxon (see Figure 7, Cheumatopsyche); (c) the entire
unimodal curve because their optimum is near the center of observed conductivities and the
range from deficiency to toxicity is relatively narrow (see Figure 7, Diploperla); or (d) no trend
because the optimum is more of a plateau than a peak so it extends across the range of observed
conductivities (see Appendix E, Nigronid).
In order to estimate effects to sensitive taxa, it may be necessary to exclude genera
favored by the pollutant if the region is highly modified. This was not done with the
Appalachian data set. All genera regardless of the exposure-response form were included in the
SSD. However, the XC values for those such as Cheumatopsyche that do not descend to zero in
the observed range are treated as "greater than values." Because the 5th centile of the SSD is
derived by interpolation, it is not necessary to provide point estimates of the XC values for
resistant taxa. The setting of the benchmark in a conductivity range in which the occurrence of
some genera is increasing suggests that the benchmark could result in the extirpation of some
genera. However, that is not the case. All but one of the 163 genera occur in sites with low
conductivity (<100 |iS/cm). Even if that were not the case, the concern for resistant taxa is
unwarranted. The EPA sets water-quality criteria to protect the taxa that occur prior to
pollution—not taxa that require pollution.
5.12. USE OF MODELED OR EMPIRICAL DISTRIBUTIONS
When deriving XC and HC values, one might use a centile 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 centiles 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 centiles of interest. In particular, standard models are symmetrical but many SSDs
are not, so the data are poorly fit at the extremes. The use of a nonparametric regression method
to alleviate the problem of assuming a particular functional form can result in biologically
40
-------
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 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 the
interest of conceptual and operational simplicity, we identify the XCgs as the conductivity value
at which the empirical cumulative probability is 0.95. The HCos is determined by2-point
interpolation of points on the empirical distributions of XC95 values as described in Stephan et al.
(1985).
5.13. DUPLICATE SAMPLES
Although most sites in the WABbase were sampled only once, 4% were sampled more
than once and 5% of samples were from sites with duplicates. This situation may be confused
with pseudoreplication, but that statistical error is not an issue in this analysis because we are
estimating a value rather than testing a hypothesis. Duplicates provide more information
especially when samples originate from different seasons when different genera may be present,
but they could be problematical if they introduce a bias (e.g., if low conductivity sites were more
likely to be sampled repeatedly). However, the duplicated sites do not appear to be biased in this
case. In fact, if a simple inverse weighting scheme is applied (e.g., if a site is sampled twice;
each observation is weighted 0.5) it does not materially change the result (HCos = 293).
Therefore, for the sake of simplicity and to avoid the possibility of inadvertently introducing bias
by inappropriately weighting, we have not deleted or differentially weighted the duplicated
samples. However, if there is a potential for bias due to duplication of some samples in future
applications of this method, an appropriate weighting scheme could be applied. It was not
necessary in this case.
5.14. TREATMENT OF CAUSATION
Causation should not be an issue in laboratory toxicity tests, but, even with rigorous
treatment of confounders, scientists 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).
41
-------
5.15. TREATMENT OF POTENTIAL CONFOUNDERS
The use of field data to understand and manipulate causal relationships is limited by the
possibility that the apparent relationship used to estimate the benchmark is confounded.
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 can bias a causal model
resulting in uncertainty concerning the actual magnitude of the effects. Therefore, a variety of
types of evidence are used to determine whether confounders significantly affect the results (see
Appendix B). This is done because statistics alone cannot determine the causal nature of
relationships (Pearl, 2009; Stewart-Oaten, 1996).
Potential confounders include the following: habitat, organic enrichment, nutrients,
deposited sediments, pH, selenium, temperature, lack of headwaters, catchment area, settling
ponds, dissolved oxygen, and metals. One potential confounder, low pH, was known to cause
effects and was controlled by removing sites with pH <6 (see also Section 2.3). The influence of
selenium is unclear 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. These variables do affect species in the region, but their
effects do not alter the signal from conductivity or the aquatic life benchmark.
42
-------
6. AQUATIC LIFE BENCHMARK
The aquatic life benchmark of 300 uS/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 HC0s point estimate is 228 uS/cm and
the upper bound is 303 uS/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,
282 uS/cm with a lower confidence bound of 169 uS/cm and an upper bound of 380 uS/cm (see
Appendix G).
The method used to develop the benchmark is an adaptation of the standard method for
deriving water-quality criteria for aquatic life (i.e., Stephan et al., 1985), so it is supported by
precedent. Because the organisms are exposed throughout their life cycle, this is a chronic value.
Acute exposures were not evaluated.
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 Ca+, Mg+, SC>42 , and HCOs at circum-neutral pH. The aquatic life benchmark
for conductivity is applicable to the parts of West Virginia, that provided the data for its
derivation, and to Kentucky, which gave essentially the same result. It may be applicable to
Ohio, Tennessee, Pennsylvania, Virginia, Alabama, and Maryland in Ecoregions 68, 69, and 70.
This is because the salt matrix and background is expected to be similar throughout the
ecoregions. (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 G]). The aquatic life benchmark may also be appropriate for other nearby
regions. However, this benchmark level may not apply when the relative concentrations of
dissolved ions are different (see Table 2 for the ranges of concentrations in the data set used to
derive the benchmark value).
43
-------
-------
REFERENCES
Bailey, J. (2009) Watershed assessment section's 2009 standard operating procedures. West Virginia Department of
Environmental Protection, Division of Water and Waste Management, Watershed Branch, Charleston, VA.
Available online at http://www.dep.wv.gOv/WWE/getinvolved/sos/Documents/SOPs/WABSOP.pdf
Bear, JA; Cheng, HD; Sorek, S; et al. (1999) Seawater intrusion in coastal aquifers: concepts, methods and practices.
Norwell, MA: Kluwer Academic Publishers.
Bradley, TJ. (2009) Animal osmoregulation. New YorkOxford University Press.
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-chemistiy/MTMVFChemistryPartl.pdf.
Clark, ML; Miller, KA; Brooks, MH. (2001) US Geological Survey monitoring of Powder River Basin stream-water
quantity and quality. U.S. Geological Survey, Cheyenne, WY. Available online at
http://pubs.usgs.gov/wri/wri014279/pdf/wri014279.pdf.
Cormier SM; Paul, JF; Spehar, RL; et al. (2008) Using field data and weight of evidence to develop water quality
criteria. Integr Environ Assess Manag 4(4):490-504.
Cormier, SM; Suter, GW, II; Norton, SB. (2010) Causal characteristics for ecoepidemiology. Hum Ecol Risk
Assess 16(l):53-73.
Efron, B; Tibshirani, R. (1993) An introduction to the bootstrap. Monographs on statistics and applied
probability, 57. Boca Raton, FL: Chapman & Hall/CRC.
Environment Canada and Health Canada. (2001) Priority substances list assessment report: road salts. Available
online at http://www.ec.gc.ca/Substances/ese/eng/psap/final/roadsalts.cfm.
Evans, DH. (2008a) Teleost Fish Osmoregulation: What have we learned since August Krogh, Homer Smith, and
Ancel Keys? Am J Physiol Regul Integr Comp Physiol 295: R704-R713.
Evans, DH. (2008b) Osmotic and ionic regulation: cells and animals. Boca Raton: CRC Press.
Evans, M; Frick, C. (2002) The effects of road salts on aquatic ecosystems. National Water Research Institute
(NWRI), Saskatoon, Saskatchewan NWRI Contribution No 02-308.
Flotemersch, JE; Cormier, SM; Autrey, BC. (2001) Comparisons of boating and wading methods used to assess the
status of flowing waters. U.S. Environmental Protection Agency, Office of Research and Development, Cincinnati,
OH. EPA/600/R-00/108. Available online at http://www.epa.gov/nerleerd/MCD_nocover.pdf.
Gerritsen, J; Zheng, L; Burton, J; et al. (2010) Inferring causes of biological impairment in the Clear Fork
Watershed, West Virginia. U.S. Environmental Protection Agency, Office of Research and Development, National
Center for Environmental Assessment, Cincinnati, OH. EPA/600/R-08/146. Available online at
http://oaspub.epa.gov/eims/eimsco mm. getfile?p_download_id=496962
GNU (2010) "R" version 2.12.1. GNU operationg system. Available online at http://www.r-project.org
Hille, B. (2001) Ion channels of excitable membranes, 3rd edition. Sunderland, MA: Sinauer Associates, Inc.
Hose, GC; Van den Brink, P. J. (2004) Confirming the species-sensitivity distribution concept for endosulfan using
laboratory, mesocosm, and field data. Arch Environ Contain Toxicol 47:511-520.
45
-------
Kelly, VR; Lovett, GM; Weathers, KC; et al. (2008) Long-term sodium chloride retention in a rural watershed:
legacy effects of road salt on streamwater concentration. Environ Sci Technol 42(2):410-415.
Komnick, H. (1977) Chloride cells and chloride epithelia of aquatic insects. Int Rev Cytol 49:285-328.
Kriebel, D. (2009) How much evidence is enough? Conventions on causal inference. Law Contemp Probl
72:121-136.
Maltby, L; Blake, N; Brock, TCM; et al. (2005) Insecticide species sensitivity distributions: Importance of test
species selection and relevance to aquatic ecosystems. Environ Toxicol Chem 24(2):379-388.
Mount, DR; Gulley, DD; Hockett, R; et al. (1997) Statistical models to predict the toxicity of major ions to
Ceriodaphnia dubia, Daphnia magna, and Pimephalespromelas (fathead minnows). Environ Toxicol Chem
16(10):2009-2019.
Nelson, D; Cox, M. (2005) Lehninger principles of biochemistry. 4th edition. New York: W. H. Freeman & Co.;
pp. 395-397.
Newman, MC; Ownby, DR; Mezin, LCA; et al. (2000) Applying species sensitivity distributions in ecological risk
34 assessment: Assumptions of distribution type and sufficient number of species. Environ Toxicol Chem
19(2):508-515.
Newman, MC; Ownby, DR; Mezin, LCA; et al. (2002) Species sensitivity distributions in ecological risk
assessment: distributional assumptions, alternate bootstrap techniques, and estimation of adequate number of
species. In: Posthuma, L; Suter, GW, II; Traas, TP; eds. Species sensitivity distributions in ecotoxicology. Boca
Raton, FL: Lewis Publishers; pp 119-132.
Omernik, JM. (1987) Ecoregions of the conterminous United States. Ann Assoc Am Geograph 77:118-125.
Pearl, J. (2009) Causality: models, reasoning, and inference, 2nd edition. Cambridge, UK: Cambridge University
Press.
Pond, GJ. (2004) Effects of surface mining and residential land use on headwater stream biotic integrity in the
eastern Kentucky coalfield region. Kentucky Department of Environmental Protection, Division of Water,
Frankfort, KY.
Pond, GJ. (2010) Patterns of Ephemeroptera taxa loss in Appalachian headwater streams (Kentucky, USA).
Hydrobiologia. 641(1): 185-201.
Pond, GJ; Passmore, ME; Borsuk, FA; et al. (2008) Downstream effects of mountaintop coal mining: comparing
biological conditions using family- and genus-level macroinvertebrate bioassessment tools. J N Am Benthol Soc
27(3):717-737.
Posthuma, L; Suter, GW, II; Traas, TP; eds. (2002) Species sensitivity distributions in ecotoxicology. Boca Raton,
FL: Lewis Publishers.
Remane, A. (1971) Ecology of brackish water. In: Remane A; Schlieper, C, eds. Biology of brackish water. New
York, NY: John Wiley and Sons.
Rengasamy, P. (2002) Transient salinity and subsoil constraints to dryland farming in Australian sodic soil: an
overview. Aust JExp Agri42(3):351-361.
Smith, DG. (2001) Pennak's freshwater Invertebrates of the United States: Porifera to Crustacea, 4th edition. New
York, NY: John Wiley & Sons Inc.
Smithson, J. (2007) West Virginia Stream/River Survey Design 2007—2111. WV Department of Environmental
Protection, Division of Water and Waste Management, Charleston, WV 25304.
46
-------
Stephan, CE; Mount, DI; Hanson, DJ; et al. (1985) Guidelines for deriving numeric National Water Quality Criteria
for the protection of aquatic organisms and their uses. U.S. Environmental Protection Agency, Office of Research
and Development, Washington, D.C. PB85-227049. Available online at
http://www.epa.gov/waterscience/criteria/library/85guidelines.pdf.
Stewart-Oaten, A. (1996) Problems in the analysis of environmental monitoring data. In: Detecting environmental
impacts, Schmitt, PJ; Osenberg, CW; ed. New York, NY: Academic Press, p. 109-131.
Sutcliffe, DW. (1962) The Composition of Haemolymph in Aquatic Insects. Exp. Biol 39:325-343.
Suter, GW, II. (2007) Ecological risk assessment, 2nd edition. Boca Raton, FL: CRC Press.
Suter, GW, II; Traas, T; Posthuma, L. (2002) Issues and practices in the derivation and use of species sensitivity
distributions. In: Posthuma, L; Suter, II, GW; Traas, T; eds. Species sensitivity distributions in ecotoxicology. Boca
Raton: Lewis Publishers; pp 437-474.
Tarin, JJ; Cano, A. (2000) Fertilization in protozoa and metazoan animals: Cellular and molecular aspects. Berlin:
Springer-Verlag.
Thorp, JH; Covich, AP. (2001) Ecology and classification of North American freshwater invertebrates. New York,
NY: Academic Press.
U.S. EPA (Environmental Protection Agency). (1986) Quality criteria for water 1986. Office of Water, Washington,
DC; EPA-440/5-86/001. Available online at
http://water.epa.gov/scitech/swguidance/waterquality/standards/criteria/aqlife/upload/2009_01_13_criteria_goldboo
k.pdf
U.S. EPA (Environmental Protection Agency). (2000) Nutrient criteria technical guidance manual: rivers and
streams. Office of Water, Washington, DC; EPA/822/B-00/002. Available online at
http://www.epa.gov/waterscience/criteria/nutrient/guidance/rivers/rivers-streams-full.pdf.
U.S. EPA (Environmental Protection Agency). (2003) Generic ecological assessment endpoints (GEAEs) for
ecological risk assessment. Risk Assessment Forum Washington, DC; EPA/630/P-02/004B. Available online at
http://oaspub.epa.gov/eims/eimsco mm. getfile?p_download_id=429201.
U.S. EPA (Environmental Protection Agency). (2006) Framework for developing suspended and bedded sediments
(SABS) water quality criteria. Office of Water, Washington, DC; EPA/822/R-06/001. Available online at
http://www.epa.gov/med/Prods_Pubs/framework_sabs_20060601 .pdf.
U.S. EPA (Environmental Protection Agency). (2007) Level III ecoregions of the Continental United States
(revision of Omernik 1987). National Health and Environmental Effects Research Laboratory, Corvallis, OR.
Available online at http://kgs.uky.edu/kgsweb/download/geology/useco.pdf.
U.S. EPA (Environmental Protection Agency). 2011. The Effects of Mountaintop Mines and Valley Fills on
Aquatic Ecosystems of the Central Appalachian Coalfields. Office of Research and Development, National Center
for Environmental Assessment, Washington, DC. EPA/600/R-09/138A.
Werner, AD. (2009) A review of seawater intrusion and its management in Australia. Hydoegeol J 18:281-285.
Available online at http://www.springerlink.com/content/q75147m6j6766470/fulltext.pdf (accessed 10/14/2009).
WVDEP (West Virginia Department of Environmental Protection). (2006) Quality assurance project plan for
watershed branch monitoring activities. Division of Water and Waste Management., WVDEP, Charleston, WV;
pp. 279.
47
-------
WVDEP (West Virginia Department of Environmental Protection). (2008a) West Virginia integrated water quality
monitoring and assessment report. West Virginia Department of Environmental Protection, Charleston, WV.
Available online at
http://www.dep.wv.gov/WWE/watershed/IR/Documents/IR_2008_Documents/WV_IR_2008_Supplements_Comple
te_Version_EPA_Approved.pdf.
WVDEP (West Virginia Department of Environmental Protection). (2008b) 2008 Standard operating procedures.
Vol. 1. Watershed Assessment Branch, WVDEP, Charleston, WV.
WVDNR (West Virginia Department of Natural Resources). (2007) Rare, threatened and endangered animals. West
Virginia Natural Heritage Program. February 2007. Available online at
http://www.wvdnr.gov/Wildlife/documents/Animals2007.pdf (accessed 10/12/2009).
Wood, CM; Shuttleworth, TJ. (2008) Cellular and molecular approaches to fish ionic regulation. Vol 14: Fish
Physiology. San Diego, CA: Academic Press, Inc.
Woods, AJ; Omernik, JM; Brown, DD; et al. (1996) Level III and IV ecoregions of Pennsylvania and the Blue
Ridge Mountains, the Ridge and Valley, and the Central Appalachians of Virginia, West Virginia, and Maryland.
U.S. Environmental Protection Agency, Office of Research and Development, Corvallis, OR; EPA/600/R-96/077.
Ziegler, CR; Suter, GW, II; Kefford, BJ. (2007) Candidate cause: ionic strength. CADDIS (The Causal
analysis/diagnosis decision information system. Available online at www.epalgov/caddis (accessed 10/13/09).
48
-------
APPENDIX A
CAUSAL ASSESSMENT
ABSTRACT
Because associations in the field are not necessarily causal, this appendix reviews the
evidence that salts are a cause of extirpation of aquatic macroinvertebrates in streams in
Ecoregions 69 and 70 of West Virginia. The goal is to establish that salts composed primarily of
Ca+, Mg+, HCO3 , and SO4 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. Appendix B addresses the
potential for other variables to confound the model of the effects of salts used to select the
benchmark.
A.l. INTRODUCTION
To assure 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)
and defined in Table A-l. They are 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 which are potentially available to
demonstrate characteristics of causation. Hill's considerations are a mixture of types of
evidence, sources of information, and quality of information.
A-l
-------
For the general causal question, "Can salts cause biological impairments in the region?"
the best support is evidence that salts have already caused biological impairment in the region.
We have relied on this type of evidence whenever possible.
Table A-l. Definitions of causal characteristics
Characteristic
Co-occurrence
Preceding causation
Interaction
Alteration
Sufficiency
Time order
Description
The cause co-occurs with the unaffected entity in space and time
Each causal relationship is a result of a larger web of cause-and-effect
relationships
The cause physically interacts with the entity in a way that induces the
effect
The entity is changed by the interaction with the cause
The intensity, frequency, and duration of the cause are adequate, and the
entity is susceptible to produce the type and magnitude of the effect
The cause precedes the effect
Source: Cormier etal. (2010).
A.1.1. Assessment Endpoints
This causal assessment evaluates whether aqueous salinity, as measured by conductivity,
is capable of causing local extirpation of stream biota in an area of Central 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, Virginia, Alabama, and Tennessee. The entities of concern are benthic invertebrates,
possibly including rare and threatened species. The effect is local extirpation of genera from
streams in their natural range. Because the endpoint for the benchmark is the extirpation of
multiple genera, a single measurement endpoint is sometimes needed to represent those multiple
individual responses. 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 many of the sensitive genera are Ephemeroptera and the number of
genera is a summary of the consequences of extirpation (see Figure A-l). However, the
assessment is of general causation in the regions of concern, not for any specific taxon or
location.
A-2
-------
o>
o>
O
o
Q.
E
Q.
oq
CD
CD
CD
CN
CD
O
CD
295 |jS/cm
.*
200 500 1000 2000
Conductivity (uS/cm)
I I
5000 10000
Figure A-l. The genera in the Order Ephemeroptera, as a group, are
extirpated at lower conductivity levels than many other taxonomic groups.
The plot is a species sensitivity distribution (SSD). Open circles represent the
95th centile extirpation concentration (XCgs) for a genus. The closed circles are
genera of the Order Ephemeroptera. The genus at 230 |iS/cm is Cinygmula and at
3,923 |iS/cm is Caenis.
A.1.2. Data Sets
The same data set used in the derivation of the aquatic life benchmark,the West Virginia
Department of Environmental Protection's (WVDEP's) Water Analysis Data Base (WABbase),
was used in the causal assessment (see Sections 2.2 and 2.3). In addition, other sources were
used. (1) Toxicity test results were obtained from peer-reviewed literature. (2) Information on
the effects of dissolved salts on freshwater invertebrates was taken from standard texts and other
physiological reviews. (3) An EPA Region 3 data set was obtained from Gregory Pond, which
includes the original data for Table 3 in Pond et al. (2008a) and data collected for the
Programmatic Environmental Impact Assessment (Bryant et al., 2002). (4) The constituent ions
for Marcellus Shale brine were provided by EPA Region 3, based on analyses by drilling
operators. (5) Data for Kentucky are from the Kentucky Department of Water database and are
described in Appendix G. (6) Geographic and related information is from various public sources
and WVDEP and is described in Appendix C.
A-3
-------
A.1.3. Analyzing and Weighing Evidence
Causal evidence is data that have been analyzed or organized in some way to show a
characteristic of causation or a lack of one. In this assessment, most of the evidence was
developed from analyses of the West Virginia field data. Other evidence was drawn from the
literature involving manipulations in the laboratory, field observations in the region and
elsewhere, and from general theories of physiology and ecology. Because the types of evidence
are diverse, each is described as it is presented.
After the evidence is developed, we used a form of criteria-guided judgment to weight
evidence, to weigh consolidated evidence for each causal characteristic, and to weigh the body of
evidence of the causal relationship. The overall process for synthesizing the evidence is depicted
in Figure A-2. (1) First, the evidence is sorted by type and by causal characteristic. (2) The
types of evidence are then evaluated for relevance to the assessment, consistency with scientific
theory, and quality of the study. Evidence that did not provide relevant or credible evidence was
not used in the assessment. The remaining types of evidence are weighted by scoring them
based on logical implications and the strength of the signal, and corroboration. (3) The overall
qualities of the collected evidence for each characteristic are weighed and then scored.
(4) Lastly, the body of evidence for the causal relationship is evaluated based on the evidence
that the hypothesized relationship possesses the characteristics of causation. The methods for
weighting and weighing steps are provided in Tables A-2 through A-8. The types of evidence
are scored for the relative strength of quantitative evidence (see Tables A-4, A-5, and A-6).
(1)
Sort evidence by
type and causal
characteristic
^
^
(2)
• Set aside
irrelevant or
poor quality
evidence
• Score remaining
evidence
^A
^
(3)
Weigh and score
collected types of
evidence of each
causal
characteristic
^A
^
(4)
Determine
causation by
weighing the body
of evidence
Table A-3
Table A-7
Table A-8
Figure A-2. A criteria-guided process to weight (score) and weigh the
evidence for or against causation. Tables called out below each box contain the
criteria for that step.
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
A-4
-------
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.
+ + + or Strongly supports or discounts
+ + or — Clearly supports or discounts
+ or - Somewhat supports or discounts
0 No effect
Note that these scores may be for particular types of evidence or a body of evidence for a
causal characteristic, but not for causation as a whole. For example, several studies may
convincingly demonstrate that a source exists that is associated with elevated conductivity in the
region, so that causal characteristic is scored + + +, but alone, it is not convincing evidence that
conductivity causes extirpation of biota.
A. 1.3.1. Sorting Evidence
Evidence is sorted into types by the kind of association or information, the source of the
information (from observation, manipulation, or general knowledge), and the source of the
association (from the case, from elsewhere, or from theory). For example in Table A-15, the first
type of evidence includes three pieces of evidence in the form of contingency tables (the kind of
association) of cause and effect from field surveys (the source of information: observational data
from the region). Then, the types of evidence are grouped by causal characteristics (see
Figure A-2, Step 1). The contingency table example is evidence of co-occurrence.
A. 1.3.2. Scoring Types of Evidence
Each type of evidence is dichotomously evaluated as credible or not based on
(1) relevance to the assessment, (2) coherence with scientific theory, and (3) quality of the study
(see Table A-2). Evidence that was not credible according to any of these criteria was not used
in the assessment. For example, in evaluating sufficiency, we did not include toxicity test studies
of taxa or ionic mixtures that were substantially different from those used to construct the causal
model (see Table A-21). No studies were found to be inconsistent with scientific theory. Low
relevance studies were rejected based on content. Data from non-peer-reviewed studies were not
used, but an exception was made for a data set of chemical analysis of brine drilling waste.
A-5
-------
The remaining evidence was weighted by scoring the types of evidence using a system of
plus (+) for supporting conductivity as a cause, minus (-) for weakening, and (0) for ambiguous
qualities (see Figure A-2, Step 2). Three qualities of the evidence may contribute to the score.
(1) A single score is applied to register the logical implication of the evidence: to decrease (-) or
increase (+) support for the causal relationship or to have neither tendency (0). (2) Especially
strong evidence receives an additional score, based on logical properties (e.g., the effect occurred
before the cause) or the quantitative strength of the evidence (e.g., high correlation coefficients
or large quantitative differences (see Tables A-4, A-5, and A-6). (3) A type of evidence may
receive an additional score if there is consistency among multiple studies for that type of
evidence. For example, for Co-occurrence of Cause and Ephemeroptera, the evidence in
Tables A-9, A-10, and A-l 1 all show that, where conductivity is high, individuals of the family
Ephemeroptera are less likely to occur. This supports the causal hypothesis, and a + is assigned
for logical implication. A change of 50% or more is large (see Table A-4), so another + is
assigned for strength. The evidence was consistently corroborated in three independent data sets
and, therefore, receives another + for a total of + + + (see Table 15).
Table A-2. Abbreviations used for scoring the different types of evidence
Score
NE
na
0
+ or-
Meaning
No evidence
Quality is not applicable
Evidence is ambiguous or
neutral
Logical implication
Table A-3. Standardized scoring for assigning weights to types of evidence
Rationale
Logical implication
Especially strong or
logically
compelling
Corroborated
Description
Registers that the evidence is relevant and
either supports or discounts the causal
relationship.
The association was quantitatively strong (see
Tables A-4, A-5, and A-6), or predicted from
first, principles of chemistry or physics, or
logically excludes or confirms the relationship.
An independent data set corroborated the
evidence.
Score assignment
+ or-
Increase score
Increase score
A-6
-------
When scoring evidence based on correlations, contingency tables, or quantitative
comparisons, we used standard criteria for logical implication and strength described in
Tables A-4, A-5, and A-6. Other qualities, which are not simple and quantitative, must be scored
based on judgment and explained in each case.
Table A-4. Scoring the logical implication and strength of evidence for
co-occurrence from contingency tables
Assessment
Effect endpoints differ and are explained in the
accompanying text for each association. For example,
Table A-9 supports the causal hypothesis because high
levels of conductivity increase the probability that a
site lacks Ephemeroptera, and low levels of
conductivity increase the probability that
Ephemeroptera are present.
Strength
Increased effect >25%
Increased effect >50%
Increased effect <25%
Increased effect <5%
Decreased effect
Score
+
+ +
0
-
—
An additional score may be added for corrob oration, for a total not to exceed three pluses or
minuses.
Table A-5. Scoring the logical implication and strength of evidence for
co-occurrence from correlations (for consistency all correlations are
Spearman's)
Assessment
The sign of the correlation coefficient depends on the
relationship. For toxic relationships, such as the
correlation between conductivity and Ephemeroptera,
the sign should be negative. Weak or positive
correlations discount the causal relationship. For
example, see Table A-22.
Strength
0.75
>r>
r >
0.1
0.75
-------
Table A-6. Scoring the logical implication and strength of evidence for
magnitude of effects
Assessment
Differences among sites or in levels of exposures
or effects were scored based on their magnitudes.
Small differences are ambiguous, and differences
counter to logical expectation are negative
evidence for causation. For example, see scoring
of Tables A- 12 and A- 13 summarized in
Table A- 18.
Different by a factor of
>2
>10
<2
wrong sign
wrong sign >2
Score
+
+ +
0
-
—
An additional score may be added for corrob oration, for a total not to exceed three
pluses or minuses.
A. 1.3.3. Weighing and Scoring the Collected Evidence for Each Causal Characteristic
We continued the process to assess the causal relationship by weighing the strength,
diversity, and consistency of the evidence for each causal characteristic and noting any
discrepancies and any aspects of the body of evidence that could be improved (see Figure A-2,
Step 3). The evidence is weighed using a system with the same symbols as for weighting the
types of evidence.
The summary score for each causal characteristic was assigned the median score for the
body of evidence. A score was reduced if the evidence for that characteristic was inconsistent.
The score was increased if the evidence included at least three types of consistent evidence not to
exceed 3 +'s or -'s. (see Table A-7).
Table A-7. Standardized scoring for assigning weights to collected types of
evidence for each causal characteristic
Rationale
Median score of evidence with the logical implication
indicated by the sign
Inconsistent evidence
Consistency among three or more types of evidence
Not to exceed
Score
three pluses or
minuses
+, 0, -, ne
Reduce by one or more + or -
Increase by one
+ or-
NE = No evidence.
A-8
-------
A.l.3.4. Weighing the Body of Evidence
The scores for the evidence of the causal characteristics were used to evaluate the body of
evidence for the causal relationship (see Figure A-2, Step 4). The system for evaluating the
evidence is outlined in Table A-8. A causal relationship was judged to be reliable if there was no
evidence that weakened the relationship and if there was supporting evidence for all six
characteristics of causation. Evidence for some causal characteristics is difficult to obtain, thus,
the cause was judged very likely if there was evidence of five characteristics and some of these
were strong. In this assessment, several types of evidence were weighted and then weighed for
five causal characteristics.
A summary of the evidence for each of the causal characteristics is described in
Section A.2.7 Evaluation of the Body of Evidence.
Table A-8. Rules for determining causation by weighing the body of
evidence for the causal relationship
Body of evidence
Evidence refuting21 1 or more characteristics
Evidence discounting13 4, 5, or 6 characteristics
Evidence discounting 1, 2, or 3 characteristics, others
supporting
Evidence strongly documenting 6 characteristics
Evidence documenting 5 or 6 characteristics and none
discounting
Evidence strongly documenting 3 or 4 characteristics and none
discounting
Evidence strongly documenting 2 characteristics and none
discounting
Evidence documenting 1 characteristic
Causal relationship
Refuted causation
Unlikely causation
Unlikely causation but low
confidence
Confirmed causation
Very probable causation
Probable causation
Probable causation but low
confidence
Insufficient evidence to make a
determination
aRefuting is the logical process of demonstrating the impossibility of a candidate cause, thus allowing it to be
eliminated from further consideration.
bDiscounting is the weighting of evidence that weakens the case for a candidate cause but is insufficient to refute.
A-9
-------
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 (Hill, 1965; U.S. EPA, 2000).
The summary of evidence is presented at the end of Section A.2.1 in Table A-15.
A.2.1.1. Co-occurrence of Cause and Ephemeroptera
The genera in the family Ephemeroptera, as a group, are extirpated at lower conductivity
levels than many other taxonomic groups (see Figure A-l). We constructed a contingency table
of the presence of Ephemeroptera at sites near background conductivity (<200 uS/cm) and high
conductivities (>1,500 (j,S/cm) and recorded the number and relative percentage of the presence
or absence of Ephemeroptera (see Table A-9). It shows that Ephemeroptera co-occur with low
conductivity but that all ephemeropteran species are absent from more than 55% of sites where
conductivity is high. This analysis emphasizes the difference between high and low conductivity
sites with respect to a clear endpoint, the absence of all Ephemeroptera.
We repeated the analysis with the EPA Region 3 data set and the Kentucky data set, with
similar results. To ensure a sufficient number of samples, low conductivity was <300 uS/cm,
and high conductivity was evaluated at >1,500 uS/cm. In the EPA Region 3 data set, 81% of
high conductivity sites lacked Ephemeroptera (see Table A-10), and in the Kentucky data set
30.8% of high conductivity sites lacked Ephemeroptera (see Table A-l 1).
We also compared the number of ephemeropteran genera at sites with lower
conductivities and higher conductivities with and without the co-occurrence of other parameters
that are somewhat correlated with conductivity or are known biological stressors (see
Appendix B). Whatever the level of the other parameter, when conductivity was low,
Ephemeroptera occurred, and they occurred much less often at high conductivity. Hence, those
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.
Scoring—This evidence supports the causal relationship between conductivity and extirpation of
genera (+). Where conductivity is high, individuals of the family Ephemeroptera are less likely
to occur. A change of 50% or more is large (+). The evidence is corroborated in three
independent data sets (+). The total score assigned is + + +.
A-10
-------
Table A-9. Presence of Ephemeroptera contingent on stream conductivity
Near background conductivity
(<200 uS/cm)
High conductivity
(>1,500 uS/cm)
Total
Ephemeroptera present
852
(99.2%)
50
(45%)
902
Ephemeroptera absent
7
(0.8%)
61
(55%)
68
Total
859
111
970
Source: data from WABbase.
Table A-10. Presence of Ephemeroptera contingent on stream conductivity
(EPA Region 3 data set)
Conductivity <300
Conducti vity> 1,5 00
Total
Ephemeroptera present
7
(100%)
4
(19%)
11
Ephemeroptera absent
0
(0%)
17
(81%)
17
Total
7
21
28
Source: data from EPA Region 3 data set.
Table A-ll. Presence of Ephemeroptera contingent on stream conductivity
(Kentucky data set)
Conductivity <300
Conducti vity > 1,5 00
Total
Ephemeroptera present
150
(97.4%)
9
(69.2%)
159
Ephemeroptera absent
4
(2.6%)
4
(30.8%)
8
Total
154
13
167
Source: data from Kentucky data set.
A.2.1.2. Co-occurrence in Nearby Catchments
Two valley-filled tributaries and one unmined tributary were identified in the Twenty
Mile Creek Watershed from the WABbase. The conductivity is lower in the unmined sites
A-ll
-------
compared to the valley-filled streams, and all of the biological metrics are greater than in the
mined sites (see Table A-12). In another study, sites in three reclaimed mined watersheds were
compared with three nearby unmined watersheds by Pond et al. (2008a) (see Table A-13). The
conductivity is lower in the unmined sites compared to the reclaimed mined sites, and all of the
biological metrics are greater in the unmined sites, even though habitat scores are similar. The
number of ephemeropteran genera is 2-3-fold greater in the unmined sites.
Scoring—This evidence supports the causal relationship (+); the biological effect is 2 to 3 times
less than at the low conductivity sites (no additional score). The results are consistent and
corroborated (+). Total score assigned is + +.
Table A-12. Temporal increase of conductivity after permitting of mining
operations
uS/cm
%E
#E
#P
#EPT
TT
Never mined
Ash Fork
1998
44a
2003
39b
27.23
6
5
20
41
2006
51a
29.21
4
6
14
24
2007
37a
31
9
8
22
27
Permit 1994, 1996
Boardtree Branch
1998
l,396a
2003
3,015b
1.23
2
0
5
20
2007
3,390a
Permit 1996
Stillhouse Branch
1998
511a
2003
3,199b
0
0
0
O
8
2007
3,970a
"Single measurement.
bMean value.
E = Ephemeroptera; P = Plecoptera; T = Trichoptera; TT = total taxa.
A-12
-------
Table A-13. Multimetric indices, selected metric values, specific
conductance, and total Rapid Bioassessment Protocol (RBP) habitat scores
for reclaimed mined and unmined sites
Stream
Specific conductance
(uS/cm)
GLIMPSS
WVSCI
Total genus richness
EPT genus richness
Ephemeropteran genus
richness
Total RBP habitat score
Unmined
Rushpatch
1999
60
75
68
42
17
9
147
2006
70
75
90
40
19
7
144
Spring
1999
51
74
90
33
17
8
163a
2006
66
79
95
37
21
8
149
White Oak
1999
64
75
91
32
17
9
161
2006
88
85
88
30
20
8
163
Reclaimed Mined
Ballard
1999
1,201
51
55
33
12
3
148
2006
1,195
38
52
20
9
3
149
Stanley Fork
1999
1,387
21
25
14
2
0
145
2006
2,010
34
38
28
6
0
155
Sugartree
2000
1,854
32
52
22
4
0
141
2007
191
29
36
20
4
0
154
aRBP from spring 2000.
GLIMPSS = genus-level index of most probable stream status; WVSCI = West Virginia Stream Condition Index;
EPT = Ephemeroptera, Plecoptera, Trichoptera
Source: Pond et al. (2008a).
A.2.1.3. Co-occurrence between Conductivity and Extirpation of Genera
All 163 benthic invertebrate genera appearing in the West Virginia species sensitivity
distribution (SSD) list are observed at some sites below 100 uS/cm except Hydroporus (lowest
occurrence at 168 uS/cm); therefore, low conductivity is not a limiting factor. However, 24.5%
of genera (40/163) are never observed above 1,500 uS/cm (see Table A-14).
Table A-14. Presence of genera contingent on stream conductivity
Near background conductivity (<150 uS/cm)
High conductivity (> 1,500 uS/cm)
Genera present
West Virginia
163
(99.9%)
123
(75.5%)
Kentucky
104
(100%)
58
(55.8%)
Genera absent
West Virginia
0
(0.01%)
40
(24.5%)
Kentucky
0
(0.0%)
46
(44.2%)
Source: data from WABbase and Kentucky Division of Water database.
A-13
-------
Scoring—This evidence supports the causal relationship (+); extirpation of 40 genera in West
Virginia and 46 in Kentucky in streams with conductivity >1,500 uS/cm is a strong effect (+).
The two analyses corroborated one another (+). The total score assigned is + + +.
Table A-15. Weighing and scoring evidence for co-occurrence
Type of
evidence
Co-occurrence
of cause and
Ephemeroptera
Co-occurrence
in nearby
watersheds
Co-occurrence
between
conductivity
and extirpation
of genera
Description of evidence
Contingency Tables A-9, A- 10, and A-l 1
provide quantitative evidence that high
conductivity is strongly associated with severe
effects. Ephemeroptera are present at >99% of
low conductivity and absent at 55-73% of high
conductivity sites in three data sets.
In two studies (see Tables A-12 and A-13),
there is a 2-3-fold difference between high and
low conductivity sites for several effect
endpoints despite similar habitat quality among
sites.
Table A-14 show that 37% of genera are never
seen >1,500 (is/cm, while all genera in the
study set were observed at sites <150 (is/cm
except for one. These findings were confirmed
with independent data sets from West Virginia
and Kentucky.
Logical
implication
+
+
+
Strength
+
+
Corroborated
+
+
+
Summary of co-occurrence — In summary, the causal relationship exhibits the causal characteristic of
co-occurrence of loss of susceptible taxa with conductivity greater than natural background (+). Many
genera are never seen at high conductivity in two independent data sets. Also, Ephemeroptera are present
where conductivity is low even when other stressors are present. Ephemeroptera are frequently absent where
conductivity is high, even when other stressors are absent. Loss of many genera is a strong effect (+). In
paired watersheds, various biological metrics are diminished in co-occurrence with elevated conductivity.
Each type of evidence was independently corroborated (+). A summary score of + + + was assigned.
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
EPA's SI and CADDIS process, causal pathway. The summary of evidence is presented at the
end of Section A.2.2 in Table A-l8.
A-14
-------
A.2.2.1. Complete Source to Cause Pathway from the Literature
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 of aqueous salts is assessed for this type of evidence. Potential sources
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 operations, treatment of wastewater,
human and animal waste, scrubbers at coal fired electric plants, and demineralization of crushed
rock (Ziegler et al., 2007, U.S. EPA, 2011). In particular, high conductivity leachate has been
shown to flow from valley fills created during coal mining operations (Bryant et al., 2002;
Merricks et al., 2007). General ecological studies have shown that conductivity increases only
slightly following clear-cutting and burning. Dissolved mineral loading may be increased
slightly by harvesting but also declines quickly as vegetation re-establishes (Swank and
Douglass, 1977). Golladay et al. (1992) 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.
Scoring—This evidence from the literature indicates that there are sources of aqueous salts in the
region (+). Multiple studies are consistent in the description of the ion types associated with
different sources (+). Strength is not scored. Total score is + +.
A.2.2.2. Co-occurrence of Sources and Conductivity from the Region
Conductivity is shown to increase after the construction of valley fill coal mining
operations in two catchments (see Table A-12). Conductivity is elevated where surface mining
operations occur in a watershed and not in an adjacent unmined watershed (see Tables A-12 and
A-13) and salts are higher overall in mined watersheds with valley fill than in unmined
watersheds (see Table A-16). Similar results are reported in mined and unmined sites in
Kentucky (Pond, 2010). Principal component analysis sorted mined and residential sites from
reference sites primarily on the basis of specific conductance and pH (Pond et al., 2008a).
Scoring—This evidence supports the causal relationship (+). The magnitude of the difference in
conductivity at mined sites is 10 to 50 times greater than at unmined sites (+). The source of
increased conductivity is corroborated and consistent (+). Total score is + + +.
A-15
-------
!•* ^ r^
^* ' ^ n-5
•^ O U
5 « C
* § S
s -^ s
•
o
K
O)
•s o
•a
a>
a
(L)
S °
o>
w
I
0/j
o ^
'^ 'a
S3 1>
±1 e
"^ m
— c
s3 o
' '5b
rl
t-;
2
ts
w>
t-;
ts°
OS
oo
ts
••*
ts'
oo
o
^
0
Ti
n
r*~i
o
oo
0
OS
o
^-
f-
0
c*^
t^
o
"'"J-
O
00
O
o
C3
o
^
'—
o
O O
•a o
^
-§
a
a
CS
I
A-16
-------
A.2.2.3. Characteristic Composition of Identified Sources
Correlation and regression analyses suggest that, in Ecoregions 69 and 70, conductivities
above 500 uS/cm contain high levels of the ions of Ca2+, Mg2+, HCO3 , and SO42 (see
Figure 13a~b), which is consistent with surface coal mining and valley fill sources (Pond et al.,
2008a; Pond, 2010). In the WABbase data set, 98% of the sample sites were characterized by
anions with (HCOs + SC>42 ) / Cl > 1. In mined and unmined sites, the dominant cations are
Ca2+ and Mg2+, and anions are HCOs and SC>42 . This pattern results from calcareous geology
and the fact that, in these regions, surface mining is the activity that greatly increases the
leaching of salts from those rocks. Other saline effluents including human and livestock wastes
and road salts are dominated by NaCl. Particularly high concentrations of NaCl occur in
Marcellus shale brines (see Table A-16). The median difference is very large; 99% of anions are
HCOs + SC>42 in both mined and unmined sites, and >99% of the anions are Cl in brines (see
Table A-16). Therefore, the causal assessment relates primarily to mixtures of salts typical of
alkaline coal mine drainage and associated valley fill discharges.
Scoring—This evidence supports the causal relationship (+) by showing that there are sources of
high conductivity with a consistent matrix of ions. Both mined and unmined sites have similar
proportions of Ca2+, Mg2+, HCOs , and SC>42 but very different concentrations. The difference
between the ionic concentrations is very large, with a >99% difference from other sources of
salts such as brines (+). The evidence from the WABbase data set and two other Appalachian
studies consistently supported the ionic makeup associated with land disturbance, especially
surface mining (+). The mined and unmined data are from a peer-reviewed publication (Pond et
al., 2008a), and the brine values are from reports from extraction permittees. Although the brine
analyses are not peer reviewed, the findings are qualitatively similar to other non-peer-reviewed
reports of the makeup of these brines. Total score is + + +.
A.2.2.4. Correlation of Conductivity with Sources
Scatter plots of conductivity levels were generated for nine land cover classifications to
determine if conductivity increased with any particular sources. The methods and results are
presented in greater detail in Appendix C. Briefly, 190 records of <20-km2 watersheds in the
WVDEP WABbase in Ecoregion 69D were found that had macroinvertebrate samples identified
to the genus level, at least one chemistry sample, and total maximum daily load land cover
information. Small (<20-km2) subwatersheds were selected to reduce confounding from multiple
sources. These subwatersheds drained to the Coal, Upper Kanawha, Gauley, and New Rivers.
Scatter plots and Spearman rank correlations of nine land use categories and geometric mean
conductivity are shown in Figure A-3: total percentage area in mining (% Total Mining);
A-17
-------
percentage in mountaintop mining valley fill (% MTM-valley fill); percentage of abandoned
mine lands (% Abandoned Mine); percentage of mining (% Mining) minus % MTM-Valley Fill
and % Abandoned Mine; percentage barren land use (% Barren); percentage of residences,
buildings, and roads (% Urban/residential); percentage in agriculture and pasture
(% Agricultural); percentage in forest (% Forest), and percentage in open water (% Water).
The two land use types that are most strongly and positively correlated with conductivity
are % MTM-Valley Fill and % Total Mining (see Table A-17). In contrast, % Forest is
negatively correlated with ion concentrtions. % Urban/residential is not well correlated and in
this region is confounded somewhat by mining land uses. The ions that are more strongly
correlated with land use are total calcium and magnesium (also captured together as hardness),
bicarbonate measured as alkalinity, and sulfate. Noticeably, chloride is not strongly correlated,
owing to fewer measurements of chloride, but also due to the low concentrations except at one
site. Chloride was 629 mg/L at the site with the greatest residential and mining land uses.
At relatively low % Urban/residential, conductivity is highly variable (see Figure A-3).
In contrast, there is a clear pattern of increasing conductivity as % MTM-Valley Fill increases
and of decreasing conductivity with increasing % Forest. When area in valley fill is subtracted
from the total nonacid mining area, the correlation decreases by 25% (see Figure A-3d). The
scatter plots illustrate that there are clear sources of increased conductivity, but that
% MTM-Valley Fill has the strongest correlation with conductivity (r = 0.65) and the percentage
of mining without a valley fill has a moderate correlation (r = 0.39).
Of the land uses in the small watersheds analyzed, only mining especially associated with
valley fills is a substantial source of the salts that are measured as conductivity. Disturbances
associated with agriculture and human habitation may also contribute, but 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. Furthermore, despite the natural bedrock of
shale, limestone, dolomite, and calcareous cemented sandstone, natural background is
exceedingly low.
Although conductivity typically increases with increasing land use (Herlihy et al., 1998),
at relatively low urban land use, 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 percentage of area in valley fill increases and decreasing
conductivity with increasing forest cover.
Scoring—This evidence supports the causal relationship (+). The correlations for mountaintop
mining with valley fill (r = 0.65), mining minus valley fill and abandoned mine lands (r = 0.39),
A-18
-------
and forestry (r = -0.55) are moderately strong. This study has not been independently
corroborated, although it is consistent with the findings of Pond et al. (2008a). The association
seems to be specific for extensive geologic disturbances, which in these regions, are from mining
and valley fills. The total score is +.
Table A-17. Correlation coefficients between pairs of land use and water
quality parameters in the land use data set
Water quality
parameter
Conductivity
Alkalinity
Hardness
Sulfate
Calcium total
Magnesium total
% MTM- Valley
Fill
0.65
0.51
0.69
0.64
0.67
0.66
% Total Mining
0.52
0.49
0.63
0.52
0.61
0.65
% Mining
0.39
0.37
0.55
0.39
0.52
0.58
% Forest
-0.54
-0.51
-0.63
-0.53
-0.64
-0.59
A-19
-------
a.
o
O
d.
.£•
I
o
O
8.
8-
8.
8-
r = 0.53
0 2.2 9 30.6 99
%Total Mining
r=0.4
0 2.2 9 30.6 99
% Mining
t^
b. £1
o
O
O
O
r = 0.65
0 2.2 9 30.6
%MTM-ValleyFill
r = 0
2.2 9 30.6
% Barren
C.
o
O
f.
O
O
8_
8-
8_
8-
r=0.02
0 2.2 9 30.6
% Abandoned Mine
009 O
r=0.13
0 0.6 1.5 3 5.3 9 14.8
% Urban/residential
o
-i
= -lD.03
0 0.6 1.5 3 5.3 9 14.8
% Agricultural
•& ^°
I S
! §
o
o %>
r = -0.55
0 20 40 60 80 100
% Forest
I 8_
•< O ^
CO
•& *° —
i °^
o
r=0.27
0 0.3 0.6 1 1.5 2.2
% Water
Figure A-3. Geometric mean conductivity associated with different land uses
in 190 watersheds in Ecoregion 69D and Spearman's correlation coefficient.
Conductivity increases with increasing % MTM-Valley Fill and % Total Mining,
and decreases with increasing % Forest, but there is less clear or no pattern with
other land use. From left to right, they are (a) % Total Mining (percentage
of deep, surface, quarry mining, MTM-Valley Fill, and abandoned mine land),
(b) % MTM-Valley Fill (from mountaintop mining overburden),
(c) % Abandoned Mine, (d) % Mining (inclusive of all types of mining except
MTM-Valley Fill and Abandoned Mine), (e) % Barren, (f) % Urban/residential,
(g) % Agricultural, (h) % Forest, and (i) % Water. Fitted LOWES S line with span
set at 0.75.
A-20
-------
Table A-18. Weighing and scoring evidence for preceding causation
Type of evidence
Complete
source-to-cause
pathway from
literature
Co-occurrence of
sources and
conductivity in the
region
Characteristic
composition of
identified sources
Correlation of
conductivity with
sources
Description of evidence
Multiple publications link conductivity
to sources in the region and eliminate
some other land uses as sources. Sources
are present, and no intermediate steps in
the pathway are required.
When valley fills are present,
conductivity is 14- to 90-fold greater
than at unmined sites (see Tables A- 12
andA-13). This is very strong
quantitative evidence from the case.
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. The salt mixture
consistently contains ions of HCO3 +
SO4 2 / Cl that are >1 (see Table 1 and
Table A- 16).
Correlation of % MTM Valley Fill is
r = 0.65; see Figure A-3. This is
moderately strong quantitative evidence
from the case.
Logical
implication
+
+
+
+
Strength
+
+
Corroboration
+
+
+
Summary of Preceding Causation. In summary, large-scale surface mining and associated valley fills
constitute a common source of high conductivity water in this region (+). Some of the evidence is very
strong and specific to sources associated with coal mining (+). Four types of evidence are provided from
different investigators (+). A summary score of + + + was assigned. Hence, the evidence of preceding
causation leading to high conductivity is conclusive.
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 C ADDIS's mechanistically plausible cause. The summary of
evidence is presented at the end of Section A.2.3 in Table A-19.
A-21
-------
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 and other permeable surfaces (Sutcliff, 1962;
Bradley, 2009; Evans, 2008a, b, 2009; Wood and Shuttleworth, 2008; Thorp and Covich, 2001).
Ionic concentration is greater than natural background levels (see Section 3.6 and Figures 2, 3,
and 4). Many benthic invertebrates inhabit low conductivity streams (see Appendix D).
Therefore, the pollutant is present and the animals are exposed.
Scoring—Evidence is from knowledge that the ions are present in Appalachian waters (see
Table 1 and Table A-16) and from general knowledge of animal physiology and the anatomy of
Ephemeroptera and other aquatic invertebrates (+). The exposure is 15 to 100 times greater than
background (+) (see Tables 1, A-12, A-13, and A-16). Many studies support this inference (+).
The total score is + + +.
A.2.3.2. Biochemical Mechanism of Effect
Living cells, and the organisms they comprise, must maintain a relatively narrowly
defined internal composition of ions that varies with function and that is different from their
environment. Maintaining homeostasis involves osmotic and ionic regulation by cells and
tissues. Homeostasis is achieved by surrounding cellular compartments with selectively
permeable and energy-converting membranes equipped with ion-transport proteins.
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 water content, charge balance, and specific ionic concentrations. 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, 2009; Wood and Shuttleworth, 2008; Thorp and Covich, 2001; Komnick, 1977; Smith,
2001; Sutcliff, 1962; O'Donnell, 2011). Many freshwater invertebrates have mitochondrion-rich
chloride cells on gills and other surfaces that take up chloride and other ions (Komnick, 1977;
Bradley, 2009, Evans 2009). Exclusion of ions is insufficient to maintain homeostasis, and the
actual uptake and export of ions occurs at semipermeable membranes. Anion, cation, and proton
transport occurs by passive, active, uniport, and cotransport processes often in a coordinated
fashion (Nelson and Cox, 2005).
Numerous specific mechanisms are involved in the toxicity of high-conductivity
solutions. One that is used by invertebrates and vertebrates is discussed here to illustrate how
ions are moved against a concentration gradient through a selectively permeable membrane. The
example ion-regulation system involves antiport anion exchange proteins that cotransport Cl
A-22
-------
against the concentration gradient into the cell simultaneously with HCOs movement down the
concentration gradient and out of the cell (Larsen et al., 1996; Nelson and Cox, 2005; Bradley,
2009, Evans 2009) (see Figure A-4). Normally, HCOs concentrations are relatively low in the
water and HCOs can be made from the waste products of respiration so that HCOs
concentration becomes greater inside the cell than in the surrounding water. Under these
conditions the HCOs gradient is strong enough and the antiport protein swaps out HCOs for Cl
despite the higher amounts of Cl inside the cell compared to in the water. However, when
external HCO3 is high, the gradient is not favorable for HCO3 export and Cl~ uptake (Avenet
and Lingnon, 1985). As a result, internal regulation of the Clconcentration must depend on the
active transport of CF against a concentration gradient, which is energetically costly or
impossible to maintain. In addition, the normal export of HCOs must occur against a gradient to
rid cells of metabolic waste CC>2 and to balance internal pH. Furthremore, there is also some
evidence that SO42 can pass through some HCO3 channels and high external concentration of
SO42 could also affect the concentration gradient and outward flow of HCO3 (Pritchard and
Renfiro, 1983). Furthermore, the internal concentration of CF affects the balance of other ions
such as Na+, K+, FT, and NH4+. This example illustrates how membrane-transport pathways are
inhibited by too much ambient salinity in the form of bicarbonate salts, which interfere with the
uptake and balance of necessary chloride and sodium ions. The gills of Ephemeroptera have an
abundance of mitochondrion-rich chloride cells that use the cellular physiological mechanisms
illustrated in Figure A-4.
The previous example illustrates only two types of passive co-transport proteins and four
ions. It does not show the roles of other ions on the stream side of the cell and does not depict
any of the mechanisms on the basal side (organism-side) of the cell. The full complement and
relative abundance of ions are necessary for homeostasis. Because all dissolved ions interact,
there are many types of ionic transport proteins that work together to regulate pH and ionic
concentrations and cell volume. Some of the types of transport proteins are depicted in
Figure A-5. These proteins are folded into the plasma membrane and are specific for certain
ions. Some are passive channels (depicted as tubes). Others require the expenditure of energy
(indicated by the ATP as part of the protein symbol). For these, the conversion of ATP to ADP
momentarily changes the shape of the protein to regulate transport or to move an ion against a
concentration gradient. Some transporters move one ion (single arrow and circle). Others
co-transport more than one type of ion thereby leveraging the electromotive force of the
concentration gradient of another ion to reduce the concentration of some ions and increase the
concentration of others (two or three arrows and circle[s]).
A-23
-------
a)
HCO3~
ci-
Na+
HCO3-
Anhydrase QO
Stream
b)
HCO3-
ci-
Na+
HCO,
J
Carbonic H2O +
r
V,
Gill epithelium
i Organism's
i body
Carbonic H2O +
Anhydrase QO
Figure A-4. Schematic of a mechanism altered by elevated bicarbonate salts.
(a) Dilute water with low HCO3~ and Cl~. (b) High conductivity water with high
HCOs" and low Cl~. Filled arrows indicate transport readily occurs in (a) but
unfilled arrows in (b) indicate transport is inhibited.
A-24
-------
Na+
K+
ci-
SO,2-
HCO,-
Ca2+
M92
n
n
n
ci-
ci-
Na+
Na+ H+ K+ Ca2'
ci-
H2O
Figure A-5. Depiction of a variety of types of transport proteins. Passive
transport by individual ions and water (tube and arrow), passive co-transport of
ions (two arrows and circle), and energy dependent transport (circle with ATP).
Transport proteins are depicted to show many types rather than a functional
example as in Figure A-4.
The type, distribution and abundance of transport proteins are different on each cell
membrane and on different sides of a cell, thus creating arrangements that concentrate the
different ions at different levels in organelles, cell types, and bodily fluids. The different
concentrations of ions in body compartments create a complex ionic circuit that stores specific
ions as potential energy that enables all cell functions and creates conditions for the proper
chemical reactions that cells and organisms use to grow, reproduce, and continue living. Some
transport proteins are altered by pressure and affect the regulation of water volume or signal
touch in a sensory cell. Some voltage-gated channels are involved in embryonic activation,
secretion, and nerve and muscle activity. The variety of organized combinations is as various as
life itself. Selectively permeable membranes are a universal attribute of living things. Every
physiological process of animals, plants, and microbes uses ionic gradients made possible by
these membranes and their ionic transport proteins. In all living things, when the ionic balance is
disrupted, organs fail and death ensues.
A-25
-------
Scoring—This evidence supports the causal relationship by providing evidence that the typical
ion matrix in the region can create ionic gradients that can interfere with proper homeostasis (+).
However, direct evidence of the ionic regulatory mechanism or membrane potential
measurement from affected species and tolerant species in Appalachia is not available. Evidence
from the literature about mitochondrion-rich chloride cells in aquatic animals including insects,
amphibians, and fish, logically leads to disruption of ionic regulation in organisms highly
dependent on passive ionic regulation by a HCOs / Cl~ antiport anion exchange, such as is
present on ephemeropteran gill epithelium. Other ion transport systems are also affected by
increases in the concentration of the ion mixture, which is measured as increased conductivity in
the region of concern. A large body of peer-reviewed physiological studies supports this
inference (+). The total score is + +.
A.2.3.3. Physiological Mechanism of Effect
In aquatic systems, organisms are capable of coping with different environmental
challenges presented by different concentrations of dissolved ions. However, the extent and rate
of adaptation to changes of salinity varies depending on the physiological potential of a
particular species (Bradley, 2009; Evans, 2009). As noted previously, osmotic and ionic cellular
mechanisms involve selectively permeable membranes. However, it is the disruption of the ionic
balance throughout a physiological system of specialized tissues and organs with specialized
functions that reduces fitness and survival. Some examples include slight or large differences in
ionic composition between cell compartments, cells, or external media that are used to release
energy from food; transcribe and translate RNA into proteins; regulate pH and water volume;
excrete metabolic waste (ammonia and CC^); enable secretion of enzymes, hormones, and
neurotransmitters; guide embryonic development (Evans, 2009; Bradley, 2009); and propagate
action potentials in nerves and muscles, thus enabling complex behaviors and activation of
fertilized eggs (Evans, 2009; Hagiwara and Jaffe, 1979; Tarin et al., 2000). These physiological
functions enable organisms to develop, grow, move, and sense their environment. When the pH
or ionic balance is disrupted, death is usually near at hand.
For the sake of illustration, the role of chloride ions within inhibitory neural circuits is
described. Chemical transmission of nerve impulses can excite or inhibit nerve conduction, thus
modulating signaling. Gama-aminobutyric acid is an inhibitory neurotransmitter that binds and
opens chloride channels on the postsynaptic membrane (Bloomquist, 1993, 1996). Cl ions flow
into the postsynaptic neuron, hyperpolarizing the cell (i.e., making the cell more negative than a
normal resting neuron and interfering with the propagation of action potentials). Too much or
too little Cr disrupts normal neural activity. Too much Cl~ excessively inhibits nerve activity,
whereas, insufficient Cl results in hyperexcitability. Interruption of the function of chloride
A-26
-------
channels has been exploited to develop insecticides, such as dieldrin, endrin, lindane, and
endosulfan that block Cl~ permeability, resulting in ataxia and insecticides such as avermectins
that activate Cl channels, resulting in paralysis. Exposure to these Cl channel blockers and
enhancers have similar effects in insects, fish, and mammals.
In dilute water, mitochondrion-rich epithelial cells and tissues of many aquatic organisms
help maintain the balance of Cl, which enables modulation of neural activity as well as
regulating pH and other ions. This example provides evidence that disruption of ionic imbalance
in insects is a known mechanism that can cause dysfunction of the nervous system, leading to
death. In this causal assessment, the ionic imbalance is not caused by chemicals binding to ionic
channels as with insecticides, but by altering the amount of ions dissolved in the water (see
Section A.2.3.2). Classic neurophysiological studies have demonstrated that changing the ionic
constituents outside and inside cells can block the propagation of neural signaling in all animals
with a nervous system (Hille, 2001).
Scoring—This evidence supports the causal relationship (+) by demonstrating that the loss of
ionic regulation can affect an animal's physiology leading to severe effects. Studies of the
physiology of affected species and tolerant species in Appalachia are not available. The effects
of ionic disruption are supported by a large body of peer-reviewed physiological studies, some of
which are presented in an example (+). The total score is + +.
A-27
-------
Table A-19. Weighing and scoring evidence for interaction
Type of
evidence
Mechanism of
exposure
Biochemical
mechansim of
effect
Physiological
mechanism of
effect
Description of evidence
Salts readily dissolve in water and interact
directly with aquatic organisms.
Organisms living in dilute streams
exchange intracellular bicarbonate for Cl
and H+ and NH4+ for Na+ and K+. This
transport is blocked when the
concentration gradient does not favor
movement of HCO3 out of the cell. No
studies of ionic compensation were found
for invertebrates in the region, but the
basic mechanism is well established for
the example and other ion channels.
Many mechanistic studies show that
disruption of ion and water regulation
leads to organ failure by interfering with
cell functions such as enzyme and
hormone secretion, nerve conduction,
muscle contraction, waste removal, and
other physiological functions. No studies
are available for invertebrates in the
region.
Logical
implication
+
+
+
Strength
+
Corroboration
+
+
+
Summary of interaction — In summary, 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 pH and 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 (+). A summary score of + + is
assigned.
A.2.4. Alteration
A cause alters or changes a susceptible entity. In this case, the alteration is failure to
maintain viable populations of sensitive species. Documentation that a change occurs is
evidence of causation, but that evidence is much stronger if a specific effect of a cause is
characterized. 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. The summary of evidence is presented at
the end of Section A.2.4 in Table A-20.
A-28
-------
A.2.4.1. Change of Occurrence of Genera
Ephemeroptera and Plecoptera do not occur in mesohaline waters, whereas other insect
families do occasionally occur in brackish water (Remane, 1971) (see also Figure A-l). In a
paper focusing on Ephemeroptera (Pond et al., 2008a), a nonmetric multidimensional scaling
model strongly associated Cinygmula, Drunella, Ephemerella, Epeoms, and Ameletus with the
low conductivity reference sites and Stenonema, Isonychia, Baetis, and Caenis with the high
conductivity sites. The first group has 95th centile extirpation concentration (XCgs) values of
230, 297, 299, 307, and 591 uS/cm, and the second group has XC95 values of 745, 1,180, 1,395,
and 3,923 uS/cm (see Appendix D). Another study using data from Kentucky showed similar
results (Pond, 2010); however, habitat alteration may have confounded the relationship with
conductivity in that data set. Nevertheless, the relative frequency of the sensitive genera
identified in the West Virginia study (Pond et al., 2008a) decreased by more than half at mined
sites in Kentucky and, except for Baetis, which was relatively unchanged, the relative frequency
of the insensitive genera increased at mined sites with high conductivity. This evidence indicates
that specific genera tend to be more or less tolerant of ionic stress found in the region. Johansen
(1918, as cited in Remane [1971]) also mentioned isolated records of Baetis and Caenis at
1.6 ppt; however, these salt matrices are marine in nature.
Both the XCgs values and species sensitivity distributions in this document demonstrate
that a characteristic set of genera, including many Ephemeroptera, were extirpated at relatively
low conductivities and others were resistant. The relative sensitivities are consistent with the
findings of Pond et al. (2008a), Pond (2010), and with our analyses of data from Kentucky (see
U.S. EPA [2010], 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.
Taxa that are sensitive to high conductivity are similar in Kentucky and West Virginia.
Extirpation levels can be found in Appendix D for West Virginia and Appendix H for Kentucky.
Genera that began to decrease in occurrence at levels 500 (is/cm were identified from the fitted
lines on generalized additive model plots in Appendix E for West Virginia and Appendix I for
Kentucky.
In the WABbase data set, 14 genera with XCgs values below 500 (is/cm also occur in the
Kentucky data set. Among these 14 genera, 9 (64.3%) have XCgs values below
500 (is/cm in the Kentucky data set.
A-29
-------
A total of 88 genera (85%) of the 104 in Kentucky used to develop the SSD were also
used in the West Virginia SSD. Of these 104 genera, 54 showed declines below
500 |iS/cm in at least one data set (44 declined in both data sets, 4 only in Kentucky, and
6 only in West Virginia). Therefore, the West Virginia and Kentucky data sets had 44 of
54 genera (81.5%) in common that showed declines below <500 |iS/cm.
Scoring—This evidence supports the causal relationship (+) by demonstrating that conductivity
greater than background levels causes a consistent set of sensitive animals to be extirpated.
Genera affected by increasing conductivity are consistent. The number of genera with similar
XCgs values (less than 10% difference) in Kentucky and West Virginia with XCgs < 500 (is/cm is
71.4% and for those with a similar pattern of decline is 81.5% (+). Multiple studies and data sets
confirmed the evidence (+). The total score is + + +.
A.2.4.2. Models of Change of Genera
Empirical models based on macroinvertebrate assemblage composition were used to
identify probable causes of biological impairments in a case study in Clear Fork Watershed in
West Virginia (Gerritsen et al., 2010). Eight weighted averaging regression models were
developed and tested using four groups of candidate stressors based on genus-level abundance.
The strongest predictive models were for acidic metals (dissolved aluminum) and conductivity,
r2 = 0.76 and r2 = 0.54, respectively.
In another approach, nonmetric multidimensional scaling and multiple responses were
used to examine the separation of "dirty" reference groups from "clean" reference groups based
on the biological communities observed in the two groups. Four "dirty" reference groups were
identified consisting of sites primarily affected by one of the following stressor categories:
dissolved metals (Al and Fe), excessive sedimentation, high nutrients and organic enrichment
(using fecal coliform as a surrogate measure of wastewater and livestock runoff), and increased
ionic strength (using sulfate concentration as a surrogate measure). Of the "dirty" reference
groups, the dissolved metals group was significantly different from the other three "dirty"
reference groups (p < 0.001). The other three "dirty" reference groups, though overlapping in
ordination space to some extent, were also significantly different from one another (p < 0.05).
Overall, each of the five reference models (the fifth model was "clean" reference sites) was
significantly different from the others (p < 0.001), indicating that differences among stressors,
including ionic strength, apparently led to unique macroinvertebrate assemblages.
In another study with a different data set collected in West Virginia, nonmetric
multidimensional scaling was applied to invertebrate genera, and sites were sorted into distinct
ordination space characterized by low, medium, and high conductivities associated with surface
A-30
-------
mines with valley fills (Pond et al., 2008a). A study in Kentucky found similar results (Pond,
2010).
Scoring—This evidence supports the causal relationship (+) by demonstrating that conductivity
greater than background levels causes a consistent set of sensitive animals to be extirpated. The
prediction was statistically strong (+). The effect is specific enough to be used to clearly
separate groups by nonparametric statistical methods in two different data sets. Independent data
sets and investigators confirmed that different assemblages of invertebrates occur with different
stressors, including neutral-to-alkaline waters with increased salinity (+). The total score is
Table A-20. Weighing and scoring evidence for alteration
Type of
evidence
Change in
occurrence of
genera
Models of
Change of
Genera
Description of evidence
Many genera exhibit sensitivity to
increasing conductivity. These same genera
are consistently sensitive to conductivity in
another data set from Kentucky. This
quantitative evidence is independently
confirmed. Although the effect is consistent
and strong, other causes may extirpate the
same genera.
Empirical models based on specific biology
discriminated effects of conductivity
associated with mining.
Logical
implication
+
+
Strength
+
+
Corroboration
+
+
Summary of alteration. In summary, exposure to saline waters in Appalachia is associated with the
declines of specific genera (+). The specific genera are not diagnostic because they may be affected by
other causes; however, statistical tests could reliably sort and predict stressors based on biological
assemblages (+) in different data sets from two states (+). The total score is + + +.
A.2.5. Sufficiency
For an effect to occur, susceptible entities must experience a sufficient magnitude of
exposure, and the magnitude of the alteration should be commensurate. 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.
The summary of evidence is presented at the end of Section A.2.5 in Table A-22.
A-31
-------
In this section, we describe evidence that can be credibly used to evaluate whether the
level of ionic stress is sufficient or not to cause extirpation. The evidence is primarily from field
observations. Several laboratory studies (see Table A-21) were not used to evaluate sufficiency
for the following reasons: (1) the ionic constituents were not similar to those in high salinity
waters in the region of concern; (2) the study organisms infrequently or never occurred in
streams in the region and are not closely related to the affected species; (3) the test species are
physiologically tolerant of higher salinity; or (4) only acute lethality effects were reported. Such
toxicity tests serve to show that the salt mixture is highly toxic at some levels to some test
species, but they do not provide evidence to support or discount that the levels observed are
sufficient to cause the extirpation of genera found by the analyses in this report. The fact that
these test were not useful for this purpose does not imply that they are not useful for other
purposes such as WET testing or criterion development.
Table A-21. Laboratory toxicity tests of saline mixtures and reasons that
they were not useful for determining the sufficiency of the field salts to cause
the field effects
Reference
Mount etal. (1997)
Lasier and Hardin
(2010)
Merricks et al.
(2007)
Echols etal. (2010)
Kefford et al.
(2003, 2004, 2005,
2006, 2007),
Hassell et al.
(2006)
Mixture
Binary salt
mixtures
Salts of HCO3 ,
SO42 , and Cl and
effluents
dominated by Na
salts
Waters from
below valley fills
Coal-processing
effluent
Tests of
NaCl-dominated
waters in Australia
Test species
Ceriodaphnia dubia,
Daphnia magna,
Pimephales
promelas
Ceriodaphnia dubia
Ceriodaphnia dubia
Isonychia bicolor
Various Australian
macroinvertebrates
Summary
Acute lethality tests
indicated that high levels of
mixtures of common salts
can be toxic to common
laboratory organisms
Reproductive tests showed
that bicarbonate is the most
toxic of the anions
Waters with high levels of
conductivity had a higher
prevalance of toxicity in 48-
hr tests than waters with
lower levels of conductivity
7-d lethality tests of an
NaCl-dominated effluent
Various test protocols and
endpoints
Reason to
exclude
1,2,3,4
1,2,3
2,3,4
1,4
1,2
A-32
-------
A.2.5.1. Laboratory Tests of Reconstituted Mine Discharges
Kennedy et al. (2003, 2004, 2005) tested simulated coal mine discharge waters in Ohio
with the cladoceran crustacean Ceriodaphnia dubia and an ephemeropteran (Isonychia bicolor).
In 7-day lethality tests, the ephemeropteran was about three times more sensitive than the
crustacean. Lowest observed effect concentrations (LOECs) for survival of Isonychia (mid-to-
late-instars) at 20°C occurred at 1,562, 966, and 987 uS/cm in three tests. These values bracket
the Isonychia XCgs of 1,180 uS/cm. However, when the assay was conducted at 12°C, the
LOEC was 4,973 uS/cm, suggesting that longer exposures are needed before effects occur at
cold temperatures. Ceriodaphnia tests with simulated effluent containing only major ions
indicated that the toxicity of this effluent was not due to heavy metals or selenium (Kennedy et
al., 2005).
Scoring—The laboratory tests by Kennedy et al. (2003, 2004, 2005) establish that the effect for
one insensitive ephemeropteran species, Isonychia bicolor, in the laboratory, occurred at a
similar conductivity level to that in the field. A total score of + was assigned.
A.2.5.2. Field Exposure-Response Relationships of Composite Metrics
As Hill (1965) suggested, a biological gradient in the field suggests that the exposures
reach levels that are sufficient to cause effects. Evidence from several studies was evaluated.
Our analyses, 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 (r = -0.61) (see Figure A-6). This analysis shows not only
the co-occurrence of elevated conductivity and the loss of stream biota but also that there is a
regular exposure-response relationship that extends to the lowest-observed concentrations
(evidence of sufficiency).
This relationship holds even when elevated levels of potential alternative causes
(confounders) are removed (see Figure A-7). The same data set was modeled after 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 moderately strong (r = -0.53) (see Figure A-7). The correlation of the
number of genera and conductivity increased slightly, from -0.41 to -0.49. See Appendix B for
additional evaluation of potential cofounders.
A-33
-------
50 -
2 40 -
0
c
0
O
1 30-
h-
"B
£ 20 -
z
10 -
o -
I
e
F
50 -
ro 40 -
0
c
0
O
1 30-
Number of "
5 8
i i
r = -0.41
*
ft ft *
ft * * *
*%» • •• ft* w ft *
* ** «•«*> * * * • * * *
ft*ft ftqit* • «*• •*• • m •* ft
ft ft ftftft>qw ft* ft* ***•>•*** ft * * ft
ft ft ft *T*» 4Mft **•**•*>** ftftft *
ftftftft ****•*•!• ***!••**• ft> * ft* ft> ft
.. .a^piegarffl!ar,,.7,T.'»y.;
.. * • — rrr.in.Min^ii.i. n.i»»»» jaw - • *
• •'•„-"•'£ -. ?~J!Ba=v*|5!g5;r. "• ' .'
.:.. .* : '--—ESSf-Sl^fe "
« ft ft* ft* **«*>ftft *« «•»••* • ft** ft""-"!
* ft* ft ft ft ft> ft ft ft *M» •
ft ft ft ftt ftft ft
• ft* •• •» ft
• . . t. • . "..
* ft* ft ft
ft ft * *
ft> ft *
14 -
ro
0 12 -
0
O
2 ID-
'S.
o
0 8
E
0
E 6~
LU
M—
O X
!_ 4
0
z 2
' ' r = -0.61
«« »
* • w * *
* w* *imi* imimmHi »«•
X^^
\
\
N^
^v
10 100 1000 10000 10 100 1000 10000
Conductivity ( j^S/cm) Conductivity ( j^S/cm)
Igure A-6. As conductivity increases, the number of total genera and
phemeropteran genera decreases. The fitted lines are locally weighted scatter
lot smoothing (LOWESS) lines (span = 0.75). Data source: WABase.
r = -0.49
ft
* ft *
• *
ft
ft* * «* * *
* ftft *N» * ft ft
* * *V » * ft *
*ft ftftft>* ft ft ft
"""•s^ft ftftftft) ftft* ft ft ft ft ft* ft
• *S*^^ * **» • •* * * *** ft
»»» »J^* ft*ftft*Vftft«* *
ft ftft^ •;* * ftft «Mft *ft ftftft * ft ft
ft ft ft **ft*^g*H*! ft *>**• ft * •>
ft • ft **VNv***ft*ft ft ft ft ft* ft
• ft * ft Mt <*> Jii*i%»xwii *n * *
* * ft* **•*> • • *» T*v»» * ft ft
ft ft ft ft*»ft • •»•» • ft ft ft ***ft)~^<* ft ft
* * * ft*ft ft* * MH * •> * iJ^v^*
• «M4 ft 4 ft ft** ft 4 4 ^^v*
• * * ftlft • ft *• ft ft
*M< » **. ft •
ft ft ft ft ft ft ftft} ft * ft
ft ft ft • •
ft ft ft ft
ft •
ft ft
*
ft ft ft ft
* *
ft
4 4
10 100 1000
ro 12 ~
0
c
0 10 -
2
B
& 8-
0
Number of Emphe
O K> -N O3
I I I I
r = -0.53
4 «
ft * «• * **>ft>
* ** * « ft***ft •*• • « **
ft 4Nftft *•> ft* * *»» ft * ftft*1*>* ft
•> ft ft ftft ft ft ****** ft* ft** **Mt< ftM*»^* ftft** ft*
4* ftft*ftftft**ft *mm ft ft^w 4
10 100 1000
Conductivity (j^S/cm)
Conductivity (j^S/cm)
Figure A-7. As conductivity increases, the number of total and
ephemeropteran genera decreases, even when potentially confounding
parameters are removed. (Excluded: streams with higher temperatures [>22°C],
low pH [<6], poor habitat [<135], and high fecal coliform
[>400 colonies/100 mL]). The fitted lines are LOWESS lines (span = 0.75).
A-34
-------
In a study of the effects of valley fills in West Virginia by Pond et al. (2008a, b),
ephemeropteran genera and conductivity were highly negatively correlated (r = -0.90) with
conductivity and less so with habitat (r = -0.64). Pond (2010) and Pond et al. (2008a, b) also
reported that the number of Ephemeroptera and the number of taxa decreases as conductivity
increases. In a recalculation of the Pond et al. (2008a) data with additional data to create the
EPA Region 3 data set, the ephemeropteran genera and total genera were both moderately
negatively correlated with conductivity (r = -0.72 and -0.35, respectively) (see Figure A-8).
ro
CO
c
0)
O
1
"o
1
E
3
z
35 -
30 -
25 -
20 -
15 -
10 -
5 -
n -
° r = -0.35
CD
o
o^^
\ o o
^\ o
O \ O «D
\ O O
by O
O ONV CD O
-p* — P (-P°
o
° °o
o
0 0
CD
O
O
0)
c
0)
O
f
Q
0)
1
Q.
LU
7 -
6 ~
5 ~
4 -
3 -
2 -
1 -
0 -
r = -0.72
O O O\O OOO
OXXD
50 100 200 500 1000 2000
Conductivity (j^S/cm)
50 100 200
Conductivity (
I i
500 1000 2000
Figure A-8. As conductivity increases, the number of total genera and
Ephemeroptera genera decreases. The fitted lines are LOWESS lines
(span = 0.75). Data from EPA Region 3.
Scoring—The field observations show that as conductivity increases, the number of
Ephemeroptera and total number of genera decrease and, thus, the level of salt in streams is
sufficient to cause effects (+). The correlation is strong to moderately strong depending on the
data set. The effect was specific for the types of salts and species native to the region. The
correlations were corroborated with independent data sets and different investigators (+). A total
score of + + was assigned.
A.2.5.3. Field Exposure-Response Relationships of Composite Indices
The relationship between conductivity and the West Virginia Stream Condition Index
(WVSCI) score, which is a composite of six family level metrics, was also modeled from the
A-35
-------
WABbase data set. A low WVSCI score indicates poorer stream condition. Mean WVSCI
scores from 60 bins were regressed with conductivity (see Figure A-9). A stream location with a
WVSCI score of <68 attained on multiple visits is assessed by WVDEP as impaired (Gerritsen et
al. 2000, WVDEP 2010). Based on the modeled relationship, a WVSCI score of 68 corresponds
to 180 uS/cm. At the benchmark of 300 |iS/cm, the corresponding WVSCI score is 64, which is
impaired based on West Virginia's biocriteria. Using logistic regression, the probability of
impairment at 500 uS/cm is 0.72 and at 300 uS/cm is 0.59.
1000
Conductivity (|aS/cm)
r
10000
Figure A-9. As conductivity increases, the West Virginia Stream Condition
Index (WVSCI) score decreases. Points represent mean WVSCI score for
conductivity bins. Bars are 90% confidence intervals. The dotted line is the
95% confidence bound for the modeled line. A WVSCI impairment score of 68
intercepts the regression line at 180 jiS/cm (dashed arrow). The model estimates
a WVSCI value of 64 at 300 jiS/cm (solid arrow).
In Pond et al. (2008a), the genus-level index of most probable stream status (GLIMPSS)
and WVSCI scores were strongly correlated with conductivity (r = -0.90 and -0.80,
respectively). In an earlier study completed in 2006 and published in 2010, Gerritsen et al.
identified 180 jiS/cm as a plausible stressor response threshold and 300 jiS/cm as a substantial
A-36
-------
effects threshold for the association of conductivity and the WVSCI biological index using a data
set from the WABbase.
Scoring—This set of evidence indicates that, in multiple data sets and by a variety of biological
responses and analytical methods, as conductivity levels observed in the region increase, stream
condition becomes impaired, and the assemblage of macroinvertebrates is different from best
available reference sites in the region. This is supporting evidence of sufficient salt in the
streams to cause widespread effects (+). The correlations are strong (+). The correlations were
corroborated with different methods in three studies (+). A total score of + + + was assigned.
A.2.5.4. Field Exposure-Response Relationships: Susceptible Genera
As conductivity increases, the occurrence and capture probability decreases for many
genera in West Virginia (see Appendices C, D, and E) and Kentucky (see Appendices H, I, and
J) at the conductivity levels predicted to cause effects. The loss of these genera is a severe and
clear effect.
In the West Virginia data set at 500 jiS/cm, 17% of genera (14/163) are extirpated and an
additional 50% of genera are declining. In the Kentucky data set, 11.5% of genera (12/104) are
extirpated at 500 jiS/cm, and a total of 76% of genera are in decline. This evidence shows that
exposures are sufficient to extirpate susceptible genera in two geographic areas. The
associations show that relatively low exposures are sufficient to adversely affect susceptible
genera.
Scoring—The observed effects logically support the causal relationship between increased
conductivity and declining survival of susceptible genera and indicate that effects occur at
relatively low conductivity levels (+). The effect is strong, with complete extirpation of many
genera (+). The results were corroborated with a separate data set from Kentucky (+). The total
score is + + +.
A-37
-------
Table A-22. Weighing and scoring evidence for sufficiency
Type of
evidence
Laboratory
tests of
ambient
waters
Field
exposure-
response
relationships
of composite
metrics
Field
exposure-
response
relationships
of composite
indices
Field
exposure-
response
relationships:
susceptible
genera
Description of evidence
These tests showed acute lethality to an apparently
resistant species, Isonychia bicolor, at
conductivity levels similar to its XC95.
Ephemeroptera were negatively correlated with
conductivity in two data sets r = -0.61 and -0.72
(see Figures A-6 and A-8) and r = -0.90 in Pond
et al. (2008a). This evidence is highly relevant
and was obtained independently in two separate
data sets, with moderate-to-strong correlations.
Exposures were in the field with native species.
Removal of sites with poor habitat had little effect
on the correlation (see Figure A-7), the SSD or
benchmark (see Appendix B).
The field observations show that as conductivity
increases, indices of stream condition (WVSCI
and GLIMPSS) decrease (see Figure A-9).
Correlations were strong (r = -0.80; r = -0.90 in
Pond et al. [2008 a, b]). Results were further
corroborated by Gerritsen et al. (2000).
Exposures were in the field with native species.
At 500 (iS/cm, the capture probabilities of more
than 65% of genera have begun to decline.
Similar results were obtained with West Virginia
and Kentucky data sets.
Logical
implication
+
+
+
+
Strength
+
+
Corrob-
oration
+
+
+
Summary of sufficiency. In summary, exposure to saline waters in Appalachia is sufficient to cause the
declines of genera (+) with the salts found in the region's streams. The increases in effects of conductivity
are strong even when other stressors are present (+). Different analytical approaches demonstrate the level
of salinity associated with different effect endpoints in different data sets in two states (+). The evidence is
consistent. The total score is + + +.
GLIMPSS = genus-level index of most probable stream status.
A.2.6. Time Order
Logically, a causal event occurs before an effect is observed. Evidence of time order
would be 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, in the SI types of evidence, and to temporal sequence in CADDIS.
A-38
-------
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).
Scoring—NE
A.2.7. Evaluation of the Body of Evidence
In this assessment, the body of evidence is assessed based on completeness of evidence
for most characteristics of causation, and the logical implications, strength, consistency, and
diversity of the overall body of 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 resulting 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 (+ + +).
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. Based on first
principals of physics, ionic gradients in high conductivity streams would not favor the
exchange of ions across gill epithelia. Physiological studies over the last 100 years have
documented the many ways that physiological functions of organisms are affected by
excess salt (i.e., combinations of ions that they do not have mechanisms or the capacity to
regulate) (+ +).
4. Alteration—Some genera, composite metrics, and assemblages are affected at sites with
higher conductivity, while others are not. These differences are characteristic of high
conductivity (+ + +).
5. Sufficiency—Laboratory analyses report results of effects for tolerant taxa, but taxa,
ionic compositions and durations are not representative of exposure in streams.
However, increased exposure in both concentration and duration to salt affects
invertebrates based on field observations (+ + +).
6. Time order—Conductivity increases and local extirpation occurs after mining permits are
issued, but conductivity and biological data before and after mining are not available
(NE).
A-39
-------
A.3. CONCLUSION
This causal assessment presents clear evidence that the deleterious effects to benthic
invertebrates are caused by, not just associated with, the ionic strength of the water. Because this
is an assessment of general causation, the causal relationship describes how Ephemeroptera and
other salinity intolerant invertebrates, in general, respond to ionic stress and does not require that
the species or genera be the same in all applications or at all locations. Therefore, we expect that
ionic stress sufficient to cause extirpations would occur with a similar ionic matrix in other
regions with naturally low conductivity.
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. When these causes are absent
or removed, a relationship between conductivity and ephemeropteran richness is still evident (see
Appendix B).
This causal assessment does not attempt to identify constituents of the mixture that
account for the effects. Rather, it shows that the mixture of ions in streams with elevated
conductivity and neutral or somewhat alkaline waters in the region of concern is causing the
extirpation of sensitive genera of macroinvertebrates. The dominant ions, that is, those in the
greatest relative amounts, are HCO3 , SO42 , Ca2+, 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.
Bloomquist, JR. (1993) Toxicology, mode of action, and target site-mediated resistance to insecticides acting on
chloride channels. Mini Review. Comp Biochem Physiol 106(2):301-314.
Bloomquist, JR. (1996) Ion channels as targets for insecticides. Ann. Rev. Entomol. 41:163-90.
Bradley, TJ. (2009) Animal osmoregulation. New York, NY: Oxford University Press; pp. 149-151.
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/MTMWChemistryPartl.pdf.
Bunge, M. (1979) Causality and modern science. Third revised edition. New York, NY: Dover Publications, Inc.
Cormier, SM; Suter, GW, II; Norton, SB. (2010) Causal characteristics for ecoepidemiology. Hum Ecol Risk
Assess 16(l):53-73.
A-40
-------
Echols, B; Currie, R; Cherry, DS. (2010) Preliminary results of laboratory toxicity tests with the mayfly, Isonychia
bicolor (Ephemeroptera: Isonychiidae) for development as a standard test organism for evaluating streams in the
Appalachian coalfields of Virginia and West Virginia. Environ Monitor Assess. 169(l-4):487-500.
Evans, DH. (2008a) Teleost fish osmoregulation: what have we learned since August Krogh, Homer Smith, and
Ancel Keys? Am J Physiol Regul Integr Comp Physiol 295(2):R704-R713.
Evans, DH. (2009) Osmotic and Ionic Regulation: Cells and Animals. CRC Press, Taylor and Francis Group, Boca
Raton, FL.
Fox, GA. (1991) Practical causal inference for ecoepidemiologists. J Toxicol Environ Health 33(4):359-374.
Gerritsen, J; Burton, J; Barbour, MT. (2000). A stream condition index for West Virginia wadeable streams.
Prepared for U.S. EPA by Tetra Tech, Inc., Owings Mills, MD 21117. Available online at
http://www.littlekanawha.com/536_WV-Index.pdf. Accessed on 1/15/11.
Gerritsen, J; Zheng, L; Burton, J; et al. (2010) Inferring causes of biological impairment in the Clear Fork
Watershed, West Virginia. U.S. Environmental Protection Agency, Office of Research and Development, National
Center for Environmental Assessment, Cincinnati, OH. EPA/600/R-08/146. Available online at
http://oaspub.epa.gov/eims/eimsco mm. getfile?p_download_id=496962.
Golladay, SW; Webster, JR; Benfield, EF; et al. (1992) Changes in stream stability following forest clearing as
indicated by storm nutrient budgets. Arch Hydrobiol Suppl 90 (Monographische Beitrage) 1:1-33. Available online
at http://coweeta.uga.edu/publications/954.pdf.
Hagiwara, S; Jaffe, LA. (1979) Electrical properties of egg cell membranes. Annu Rev Biophys Bioeng 8:385-416.
Hassell, KL; Kefford, BJ; Nugegoda, D. (2006) Sub-lethal and chronic salinity tolerances of three freshwater
insects: Cloeon sp. and Centroptilum sp. (Ephemeroptera: Baetidae) and Chironomus sp. (Diptera: Chironomidae).
JExpBiol 209:4024-4032.
Herlihy, AT; Stoddard, JL; Johnson, CB. (1998) The relationship between stream chemistry and watershed land
cover data in the mid-Atlantic region, U.S. Water Air Soil Pollut 105(l-2):377-386.
Hill, AB. (1965) The environment and disease: Association or causation. Proceed Royal Soc Med 58(5):295-300.
Hille, B. (2001) Ion channels of excitable membranes, 3rd edition. Sunderland, MA: Sinauer Associates, Inc.
Johansen, AC. (1918) Randersfjords naturhistorie [Natural history of Randers]. Kopenhagen:CA Reitzel; 1-520 (As
cited in Ramane, 1971).
Kefford, BJ; Papas, PJ; Nugegoda, D. (2003) Relative salinity tolerance of macroinvertebrates from the Barwon
River, Victoria, Australia. Mar Freshwater Res 54:755-765.
Kefford, BJ; Dalton, A; Palmer, CG; et al. (2004) The salinity tolerance of eggs and hatchlings of selected aquatic
macroinvertebrates in south-east Australia and South Africa. Hydrobiol 517(1-3): 179-192.
Kefford, BJ; Nugegoda, D. (2005) No evidence for a critical salinity threshold for growth and reproduction of the
freshwater snail Physa acuta. Environ Pollut 134(3):377-383.
Kefford, BJ; Zalizniak, L; Nugegoda, D. (2006) Growth of the damselfly Ischnura hetemsticta is better in saline
water than freshwater. Environ Pollut 141(3):409^19.
Kefford, BJ; Nugegoda, D; Zalizniak, L; et al. (2007) The salinity tolerance of freshwater macroinvertebrate eggs
and hatchlings in comparison to their older life-stages. Aquat Ecol 41(2):335-348.
Kennedy, AJ; Cherry, DS; Currie, PJ. (2003) Field and laboratory assessment of a coal processing effluent in the
Leading Creek Watershed, Meigs County, Ohio. Arch Environ Contam Toxicol 44(3):324-331.
A-41
-------
Kennedy, AJ; Cherry, DS; Currie, RJ. (2004) Evaluation of ecologically relevant bioassays for a lotic system
impacted by a coal-mine effluent, usingIsonychia. Environ Monit Assess 95(1—3):37—55.
Kennedy, AJ; Cherry, DS; Zipper, CE. (2005) Evaluation of ionic contribution to the toxicity of a coal-mine effluent
using Ceriodaphnia dubia. Arch Environ Contam Toxicol 49(2): 155-162.
Komnick, H. (1977) Chloride cells and chloride epithelia of aquatic insects. Int Rev Cytol 49:285-328.
Larsen, EH; Christoffersen, BC; Jensen, LJ; et al. (1996) Role of mitochondria rich cells in epithelial chloride
uptake. ExpPhysiol81(3):525-534.
Lasier, PJ; Hardin, I. (2010) Observed and predicted reproduction of Ceriodaphnia dubia exposed to chloride,
sulfate and bicarbonate. Environ Toxicol Chem 29(2):347-358.
Likens, GE; Bormann, FH; Johnson, NM; et al. (1970) Effects of forest cutting and herbicide treatment on nutrient
budgets in the Hubbard Brook watershed-ecosystem. Ecol Monogr 40:23-47.
Merricks, TC; Cherry, DS; Zipper, CE; et al. (2007) Coal-mine hollow fill and settling pond influences on
headwater streams in southern West Virginia, USA. Environ Monit Assess 129(l-3):359-378.
Mount, DR; Gulley, DD; Hockett, R; et al. (1997) Statistical models to predict the toxicity of major ions to
Ceriodaphnia dubia, Daphnia magna, and Pimephalespromelas (fathead minnows). Environ Toxicol Chem
16(10):2009-2019.
Nelson, D; Cox, M. (2005) Lehninger principles of biochemistry. 4th edition. New York: WH Freeman & Co.;
pp. 395-397.
O'Donnel, MJ. (2011) Mechanisms of excretion and ion transport in invertebrates. Supplement 30: Handbook of
Physiology, Comparative Physiology. Published on line Jan 2011.
201 lhttp://www.comprehensivephysiology.com/WileyCDA/CompPhysArticle/refId-cpl30217.html Accessed
3/3/2011.
Pond, GJ. (2010) Patterns of Ephemeroptera taxa loss in Appalachian headwater streams (Kentucky, USA).
Hydrobiologia. 641(1): 185-201.
Pond, GJ; Passmore, ME; Borsuk, FA; et al. (2008a) Downstream effects of mountaintop coal mining: comparing
biological conditions using family- and genus-level macroinvertebrate bioassessment tools. J N Am Benthol Soc
27(3):717-737.
Pond, GJ; Bailey, JE; Lowman, B. (2008b) West Virginia GLIMPSS (genus-level index of most probable stream
status): a benthic macroinvertebrate index of biotic integrity for West Virginia's wadeable streams. West Virginia
Department of Environmental Protection, Division of Water and Waste Management, Watershed Assessment
Branch. Charleston, WV
Pritchard, JB; Renfro, JL. (1983) Renal sulfate transport at the basolateral membrane is mediated by anion
exchange. Proc. Natl. Acad. Sci. 80(9):2603-2607.
Remane, A. (1971) Ecology of brackish water. In: Remane A; Schlieper, C; eds. Biology of brackish water, 2nd
edition. New York, NY: John Wiley and Sons.
Smith, DG. (2001) Pennak's Freshwater Invertebrates of the United States: Porifera to Crustacea, 4th edition. New
York, NY: John Wiley & Sons Inc.
Suter, GW, II; Traas, T; Posthuma, L. (2002) Issues and practices in the derivation and use of species sensitivity
distributions. In: Posthuma, L; Suter, GW, II; Traas, T; eds. Species sensitivity distributions in ecotoxicology. Boca
Raton, FL: Lewis Publishers, pp 437-474.
A-42
-------
Swank, WT; Douglass, JE. (1977) Nutrient budgets for undisturbed and manipulated hardwood ecosystems in the
mountains of North Carolina. In: Cornell, DL; ed. Watershed research in eastern North America: A workshop to
compare results. Edgewater, MD: Smithsonian Institution Press; pp 343-363.
Tarin, JJ; Cano, A. (2000) Fertilization in protozoa and metazoan animals: cellular and molecular aspects. Berlin:
Springer-Verlag.
Thorp, JH; Covich, AP; eds. (2001) Ecology and classification of North American freshwater invertebrates, 2nd
edition. New York, NY: Academic Press.
U.S. DHEW (Department of Health Education and Welfare). (1964) Smoking and health: report of the advisory
committee to the Surgeon General., Washington, D.C. Public Health Service Publication 1103. Available online at
http://profiles.nlm.nih.gov/NN/B/B/M/Q/_/nnbbmq.pdf.
U.S. EPA (Environmental Protection Agency). (2000) Stressor identification guidance document. Office of Water,
Washington, DC; EPA/822/B-00/025. Available online at
http://water.epa.gov/scitech/swguidance/waterquality/sandards/loader.cfm?csModule=security/getfile&PageID=303
59.
U.S. EPA (Environmental Protection Agency). (2010) Causal analysis/diagnosis decision information system
(CADDIS). Available online at http://www.epa.gov/caddis.
U.S. EPA (Environmental Protection Agency). 2011. The Effects of Mountaintop Mines and Valley Fills on
Aquatic Ecosystems of the Central Appalachian Coalfields. Office of Research and Development, National Center
for Environmental Assessment, Washington, DC. EPA/600/R-09/138A.
WVDEP (West Virginia Department of Environmental Protection). (2010) 2010 Integrated water quality
monitoring and assessment report, pp 14-15. Available online at
http://www.dep.wv.gov/WWE/watershed/IR/Pages/303d_305b.aspx. Accessed 3/05/2011.
Wood, CM; Shuttleworth, TJ. (2008) Cellular and molecular approaches to fish ionic regulation. Vol 14: Fish
Physiology. San Diego, CA: Academic Press, Inc.
Woods, AJ; Omernik, JM; Brown, DD; et al. (1996) Level III and IV ecoregions of Pennsylvania and the Blue
Ridge Mountains, the Ridge and Valley, and the Central Appalachians of Virginia, West Virginia, and Maryland.
U.S. Environmental Protection Agency, Office of Research and Development, Corvallis, OR; EPA/600/R-96/077.
Ziegler, CR; Suter, GW, II; Kefford, BJ. (2007) Candidate cause: ionic strength. Available online at
http://www.epa.gov/caddis/ssr_ion_int.html. Accessed 10/20/2010.
A-43
-------
-------
APPENDIX B
ANALYSIS OF POTENTIAL CONFOUNDERS
ABSTRACT
The purpose of Appendix B is to evaluate the ability of factors that may co-occur with
conductivity (i.e., potential confounders) to weaken our ability to model the relationship between
conductivity and occurrence of genera. The analyses in this appendix do not determine whether
those factors cause effects in the region. Rather, they evaluate how the potential confounders
may affect our ability to model the relationship between conductivity and the loss of
macroinvertebrate genera.
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 substantive influence on the causal relationship
between salts and macroinvertebrate assemblages. Twelve potential confounders were
evaluated: habitat, organic enrichment, nutrients, deposited sediments, pH, selenium,
temperature, lack of headwaters, catchment area, settling ponds, dissolved oxygen, and metals.
The inference was performed by identifying potential confounders and then determining the
occurrence and strength of 10 types of evidence of confounding for each of them. The term
"confounding" refers to a bias in the analysis of causal relationships due to the influence of
extraneous factors (confounders), in this case, the stressors listed above.
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 signal from conductivity was strong, so that
potential confounders that were not strongly influential could be ignored with reasonable or
greater confidence. No analysis can demonstrate that these variables have no influence at any
place or time, but, this analysis does demonstrate that their influence on the relationship of
conductivity and extirpation of genera is minimal given the streams that would be affected by the
aquatic life benchmark.
B.I. INTRODUCTION
Having established that salt mixtures dominated by bicarbonate and sulfate cause
biological impairments in the region (see Appendix A), this appendix addresses other potential
causes of impairment in the region that might confound that relationship. 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
B-l
-------
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 current or historic disturbance. The overall goal of the Report is to
estimate conductivity levels that would protect against the unacceptable effects of salts in those
streams (i.e., typical streams receiving salty effluents in the region of concern). The goal of the
assessment in Appendix B is to determine if the model developed for that purpose is a reliable
predictor of harmful effects and protective levels. We do this by trying to discover if there are
factors that bias that model.
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 have two consequences. First, it can 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, 2009;
Stewart-Oaten, 1996). Second, confounding can bias a causal model resulting in uncertainty
concerning the actual magnitude of the effects. That can be addressed by considering the
magnitudes of correlations with and without the potential confounder or by considering the
change in the results when the potential confounder is removed.
A variety of types of evidence may be used to determine whether confounders
significantly affect the results; we have identified 10 types of evidence. 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.
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: 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.
B-2
-------
4. Co-occurrence of confounder and effect: If the frequency of the effect does not
diminish when the potential confounder is never present or is present in all cases, 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 (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 accepts the result of the
causal assessment and attempts to determine whether any of the known potential confounders
interfere with estimating the effects of conductivity to a significant degree. If there is significant
interference, the confidence in the model predictions would be weakened unless the model is
modified. 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 (mayfly) genera is used as a standard
metric for the effects of conductivity, which may or may not be confounded. Because the
endpoint effect is extirpation of 5% of genera and the sensitive genera are primarily
B-3
-------
Ephemeroptera, this is an appropriate metric. However, because of a resistant mayfly genus (see
Figure A-l), it is not expected that all Ephemeroptera will be missing at high conductivities.
Some commenters recommended using multivariate statistics in place of weight-of-
evidence analysis as the sole means to address potential confounders. However, because of the
goals of the analysis and the nature of the data, it is not appropriate to use multivariate statistics
alone to try to model the relationship between conductivity and extirpation or to eliminate the
effects of confounders or estimate the magnitude of their effects. First, no statistical test can
demonstrate that an association is causal. Second, violation of assumptions prevents reliable
estimation of the influence of one potentially causal variable on another. Multiple regression
depends on assumptions of independence, additivity, and normality that are not met. In sum,
multivariate statistical associations are just associations, and association is not causation.
However, they can be used as evidence in the weight-of-evidence analysis along with other
incomplete or imperfect pieces of evidence to help reach the best-supported conclusion.
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.
+ + + or Convincingly supports or weakens
+ + or — Strongly supports or weakens
+ or - Somewhat supports or weakens
0 No effect
Any relevant evidence receives a single plus, minus, or zero to register the evidence as
relevant 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 best professional judgments. After strength, the other possible unit of
weight is assigned depending on the type of evidence.
B-4
-------
Table B-l. Relationships between qualities of evidence and scores for
weighing evidence
Qualities of the evidence
Logical implications and relevance
Strength
Other qualities
Score, not to
exceed three minus or three plus
+ o,-
Increase score
Increase score
For co-occurrence (Evidence Types 1-4), strength or consistency of the association is the
primary consideration. The primary measure of association is Spearman's correlation
coefficients. For comparison to the potential confounders, the correlation coefficient for
conductivity and number of ephemeropteran genera are -0.61 for the West Virginia (WV) data
set and -0.72 for the EPA Region 3 data set, values in the upper end of the moderate range.
Correlations, as measures of co-occurrence, can be scored as in Table B-2.
The scores in this appendix are based on conventional expectations for a confounder that
is itself a cause. That is, a potential confounder—such as a metal by itself—might cause
extirpation of invertebrate genera (independent combined action) or might act in combination
with conductivity to extirpate invertebrate genera (additive or more than additive combined
action). However, sometimes correlations are anomalous. For example, a potential confounder
may actually decrease effects as when calcium reduces effects of metals. Such anomalous
results require case-specific interpretation based on knowledge of mechanisms and
characteristics of the ecosystems being analyzed.
Table B-2. Weighting co-occurrence using correlations for Evidence
Types 1-2
Assessment
Absent
Weak
Moderate
High
Strength
r<|0.1
|0.1|r>|0.25|
r>|0.75|
Score
—
-
+
+ +
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
B-5
-------
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 potential confounder is present at high levels,
this is supporting evidence for confounding. Note that the goal here is not to determine the
effects of 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
by, for example, a step function or other discontinuity in the relationship. Therefore, clearly high
and low levels of conductivity and the potential confounder are used in contingency tables.
When scoring evidence from 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.
Table B-3. Weighting co-occurrence for Evidence Type 3 using contingency
tables
Assessment
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
Strength
Increased effect >25%
Increased effect >75%
Increased effect <25%
Increased effect <10% or
decreased effect
Score
+ for co-occurrence
+ + for co-occurrence and strength
- for co-occurrence
— for co-occurrence and strength
B-6
-------
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 from the correlation of conductivity and Ephemeroptera in the full data set (see
Table B-4). However, no more than one negative score was given if less than 10% of the data
were removed.
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
Removal of elevated
levels of a confounder
should change the
correlation coefficient
Strength
Coefficients decrease by <10%
(0.55 20%
(0.49 >r for WV data)
Coefficients increase
Score
— for a lack of change in effect
with removal of confounder
- for a small change in effect with
removal of confounder
+ for a strong change in effect
with removal of confounder
— because removal of a true
confounder should decrease the
effect of conductivity
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.
The primary data source for evidence of confounding is the Watershed Analysis Data
Base (WABbase), which was used to derive the benchmark. Except where indicated, reported
results are derived from those data, which are referred to as the West Virginia data. However,
where possible and appropriate, the EPA Region 3 data set from West Virginia samples (referred
to as the EPA data set) is used for independent corroboration. The EPA data set is much smaller
and often does not have enough extreme values of the potential confounder to calculate reliable
contingency tables or regressions of censored data.
B-7
-------
Table B-5. Considerations used to weight the evidence concerning the
influence of potentially confounding variables
Quality of evidence
Logical implication
Directness of cause
Specificity
Relevance to effect
Nature of the association
Strength of association
Consistency of information
Quantity of information
Quality of information
Descriptor
Negative or positive
Proximate cause, sources, or intermediate causal connections
Effect attributable to only one cause or to multiple causes
From the case or from other similar situations
Quantitative or qualitative
Strong relationships and large range or weak relationships
small range
and
All consistent or some inconsistencies
Many data or few data
Good study or poor study
Source: Cormier etal. (2010).
B.3. WEIGHING
After the individual pieces of evidence have been weighted, the body of evidence for a
potential confounder is weighed based primarily on the consistency of the evidence and
secondarily on the strength of the pieces of evidence (see Table B-6). The body of evidence—
rather than any one piece of evidence—determines how strongly these potential confounders
might affect the model.
B.4. POTENTIAL CONFOUNDERS
Potential confounders were chosen because they were believed to be associated with
mountaintop mining, valley fills, or other sources of salts or because of suggestions from
reviewer or public comments. Each of the discussions in this section begins with a statement of
the reason that the potential confounder was chosen for evaluation.
B-8
-------
Table B-6. Weighing confidence in the body of evidence for a potential
confounder
Assessment
Very confident
Moderately confident
Reasonably confident
Undetermined
Potential confounding
Score
—
-
0
+
Body of evidence
All minus, some strongly negative
evidence
All minus, no strongly negative
evidence
Majority minus
Approximately equal positive and
negative, ambiguous evidence, or
low quality evidence
Majority plus
Action
No treatment for confounding
No treatment for confounding
No treatment for confounding
Additional study advised
Correction for confounding may
be advised
B.4.1. Habitat Quality
Stream habitat may be modified by physical disturbance, changes in flow or increased
sediment loads in reaches that receive high conductivity effluents. Habitat quality was
represented by a qualitative index, the Rapid Bioassessment Protocol Habitat Evaluation (RBP)
derived by the WVDEP, which increases as habitat quality increases. Component metrics were
not used because they were less correlated with Ephemeroptera than the index.
Habitat quality was analyzed as part of groups of variables that were judged a priori to be
more likely than others to have combined effects. Therefore, sites at which RBP and pH were
low and fecal coliform count was high were removed to determine whether the 5th centile
hazardous concentration (HCos) was affected (see Figure B-l). Similarly, RBP was used with
fecal coliform count and temperature in a multiple linear regression with conductivity (see
Table B-7).
The body of evidence was mixed. Habitat scores were moderately correlated with both
conductivity and biological response, which indicates a potential for confounding. However,
removal of poor habitat had little effect on the correlation of conductivity with Ephemeroptera or
on the derivation of the HCos for conductivity (see Table B-7 and Figure B-l). Habitat score had
a very slight effect on the intercept and the slope for conductivity in a multiple regression (see
Table B-7). In addition, Ephemeroptera occur even when habitat is poor (see Table B-8). The
weight of the scored body of evidence indicated habitat was not a confounder (see Table B-9).
B-9
-------
0)
c
0)
CD
"5
c
o
••e
a
2
0.
00
Ci
CD
Ci
CN
Ci
q
ci
\
I
100 200
500 1000 2000
Conductivity (|jS/cm)
I \
5000 10000
in
o
O>
o>
O
c
g
t
8.
E
Q.
CO
C3
CSj
C3
n
100
I I I
200 500 1000
Conductivity (uS/cm)
2000
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 score
>135 and fecal coliform <400 colonies/100 mL (closed circles), (a) Uncensored
with 163 genera and censored dataset with 117 genera, (b) Only the lower half of
the SSD is shown to better discriminate the points in the left side of the full
distribution. Habitat disturbance and organic enrichment have little influence; the
HC05 for the constrained data set is 326 uS/cm based on 117 genera. The upper
and lower confidence bounds on that value are 229 uS/cm and 343 uS/cm,
respectively.
B-10
-------
Table B-7. An output table for two linear regression models. The first is the
simple model predicting ephemeropteran genera from conductivity. The second
is a multivariate model with the additional covariates RBP score, temperature, and
fecal coliform count. These variables were chosen a priori as likely confounders
that could co-occur and have combined effects.
Parameter
Estimate
Standard error
Univariate model
Intercept
Conductivity slope
3.65
-0.93
0.055
0.024
Multivariate model
Intercept
Conductivity slope
RBP slope
Temperature slope
Fecal coliform slope
3.39
-0.92
0.0014
0.0068
0.037
0.11
0.029
0.0005
0.0026
0.012
Table B-8. Number and percent of sites with high and low quality habitat
and high and low conductivity with Ephemeroptera in streams (pH >6)
Habitat score <115
Habitat score > 140
Conductivity <200 uS/cm
140/142
(99%)
373/375
(99%)
Conductivity > 1,500 uS/cm
12/31
(39%)
13/22
(59%)
B-ll
-------
Table B-9. Evidence and weight for confounding by habitat quality
Type
Score
Evidence
1. Correlation of
cause and
confounder
RBP score was barely moderately correlated with conductivity,
(r =-0.25, n = 2,192).
2. Correlation of
effect and
confounder
RBP score was barely moderately correlated with the number of
ephemeropteran genera (r = 0.26, n = 2,192).
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 uS/cm) even
when habitat is poor (<115). However, with high conductivity,
Ephemeroptera are present at only about half of sites regardless
of habitat.
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 = -0.55, n = 747).
The SSD and HCos are very similar when the XCgs values were
calculated with 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).
7. Multivariate
statistics
Habitat quality, temperature and fecal coliform together had
essentially no effect on the slope in multiple regression and the
slope for RBP score is particularly small (see Table B-7).
Weight of evidence
Reasonably confident. The correlations are marginal; RBP
explains only 6.7% of the variance in ephemeropteran
occurrence, based on r2. However, the contingency table gives
relatively strong negative evidence (Ephemeroptera occur even
when habitat is poor), and elimination of poor habitat (along with
high coliform counts) has almost no effect on the SSD or HCos
(see Figure B-l). Habitat has very little effect in the multiple
regression. Therefore, we did not correct for habitat, but more
detailed habitat studies could be worthwhile.
SSD = species sensitivity distribution.
B-12
-------
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 evidence is mixed, but, overall, the evidence against significant confounding
associated with fecal coliform counts was much stronger than the supporting evidence (see
Tables B-7, B-10, andB-11).
Table B-10. Number of sites with high and low conductivity with high and low
levels of fecal coliform with Ephemeroptera present in streams
Coliform <400 colonies/100 mL
Coliform >400 colonies/100 mL
Conductivity <200 uS/cm
610/613
(99%)
184/187
(98%)
Conductivity > 1,500 uS/cm
30/69
(43%)
14/34
(41%)
Table B-ll. Evidence and weights for confounding by organic enrichment
Type
1 . Correlation of cause
and confounder
2. Correlation of effect
and confounder
3. Contingency of high
level of cause and
confounder
6. Removal of
confounder
7. Multivariate statistics
Weight of evidence
Score
+
+
Evidence
Fecal coliform counts were barely moderately correlated with
conductivity (r = 0.26, n = 2,040).
Coliform counts were barely moderately correlated with the
number of ephemeropteran genera (r = -0.25, n = 2.040).
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.
When samples >400 colonies/100 mL were removed from the
analysis, the correlation of conductivity with Ephemeroptera
barely changed (r = -0.61, n = 1,364).
The species sensitivity distribution (SSD) and HC05 are very similar
to those used in the benchmark, when calculated from subset of
the data with sites removed with pH of <6, RBP score <135, and
fecal coliform >400 colonies/100 mL (see Figure B-l).
Habitat quality, temperature and fecal coliform together had
essentially no effect on the slope for conductivity in multiple
regression (see Table B-7).
Reasonably confident: the correlations producing the two positive
scores were exactly on the margin, and negative evidence was
strong. No treatment for confounding.
B-13
-------
B.4.3. Nutrients
Nitrogen and phosphorus may 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-12).
Table B-12. Evidence and weights for confounding by nutrients
Type
Score
Evidence
1. Correlation of
cause and
confounder
Conductivity was uncorrelated with nitrate and nitrite in the
WV data set (r = 0.07, n = 1,182) and moderately correlated
in the EPA data set (r = 0.33, n = 39).
Conductivity was uncorrelated with total phosphorus in the
WV data set (r = 0.04, n = 1,185) and the EPA data set
(/• = 0.03, » = 45).
2. Correlation of
effect and
confounder
Ephemeroptera was uncorrelated with nitrate and nitrite in
the WV data set (r = -0.04, n = 1,182) and barely
moderately correlated in the EPA data set (r = -0.26,
w = 39).
Ephemeroptera was uncorrelated with total phosphorous
(r = 0.001, n = 1,185) and the EPA data set (r = 0.06,
w = 45).
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 little changed
(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 little changed
(r = -0.56, w = 998).
Weight of evidence
Very confident: all negative, some strongly negative. No
treatment for confounding.
NA = not applicable.
B-14
-------
B.4.4. Deposited Sediment
Mining and other activities that result in crushing and exposing rocks are sources of salts
and potentially of silt that may affect stream organisms. A qualitative measure of embeddedness
(WABase embeddedness score) was evaluated by contingency table and by correlation (see
Table B-13 and B-14). No evidence supported embeddedness as a confounder (see Table B-14).
Table B-13. Number of sites with high and low embeddedness scores and
high and low conductivity with Ephemeroptera present in streams (pH >6)
Embeddedness score <7
Embeddedness score >15
Conductivity <200 uS/cm
42/44
(95%)
210/211
(99%)
Conductivity >1,500 uS/cm
7/16
(44%)
6/15
(40%)
Table B-14. Evidence and weights for confounding by deposited sediment
Type
1 . Correlation of
cause and
confounder
2. Correlation of
effect and
confounder
3. Contingency of
high level of cause
and confounder
6. Removal of
confounder
Weight of evidence
Score
Evidence
The WABbase embeddedness score is weakly correlated
with conductivity (r = -0.18, n = 2,197).
The WABbase embeddedness score is weakly correlated
with Ephemeroptera (r = 0.22, n = 2,197).
In a contingency table (see Table B-13), high
embeddedness (score >15) has little effect at either high or
low conductivity.
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 = -0.62, n = 1,088).
Very confident: all negative, some strongly. No treatment
for confounding.
B-15
-------
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 HCOs that raises the pH is also a major anion moiety that
contributes to conductivity. Hence, pH directly reflects a major constituent of conductivity
(HCOs ), so it could not be a conventional confounder. In addition, salts influence hydrogen ion
activity — which is measured as pH. In any case, available evidence indicates that the variance in
pH has little effect on the derivation of the HC05 for conductivity in waters above pH 7 (see
Tables B- 1 5 and B- 16).
B.4.6. LowpH
Because low pH from acid mine drainage is known to be an important cause of
impairment where it occurs and was judged a priori to be a potentially important environmental
variable. That preconception was supported by the evidence summarized here (see Table B-17).
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-17). In sum, although the benchmark
applies to waters with neutral or basic pH, high conductivity appears to also cause effects at
low pH.
Table B-15. Number of sites with high and low conductivity with high and
low levels of pH with Ephemeroptera present
pH <4.5
pH>8.5
Conductivity <200 uS/cm
16/19
(84%)
3/3
(100%)
Conductivity > 1,500 uS/cm
0/14
(0%)
4/8
(50%)
B-16
-------
Table B-16. Evidence and weights for confounding by high pH
Type
1 . Correlation of cause
and confounder
2. Correlation of effect
and confounder
3. Contingency of high
level of cause and
confounder
5 . Level s of confounder
known to cause effects
6. Removal of
confounder shows it is
important
8. Potential confounding
evaluated by
frequency
Weight of evidence
Score
0
—
—
Evidence
Conductivity was moderately correlated with pH between 7
and 9 in the WV data set (r = 0.45, n = 1,900) and weakly
correlated in the EPA data set (r = 0. 14, n = 45).
High pH was weakly correlated with Ephemeroptera in the
WV data set (r = -Q.19,n= 1,906) and in the EPA data set
(r = -0.10, w = 45).
In a contingency table (see Table B-15), high pH at high
conductivities has the same frequency of Ephemeroptera as
high conductivity without elevated levels of another variable
in other contingency tables (approximately 50%).
EPA (1976) Water Quality Standards indicate that water
with pH 6.5-9 is protective of freshwater fish and nearly all
data were within that range.
Tests of the mayfly Isonychia bicolor found sublethal effects
at pH 10 and lethality at pH 1 1 (Peters et al., 1985).
When samples with pH >8.5 are removed from the analysis,
the correlation of conductivity with the number of
Ephemeroptera was unchanged (r = -0.62, n = 2,151).
However, this evidence is weak because relatively few sites
were removed.
The number of sites with a pH >8.5 is a very small
proportion of the sample (<2.5%), so high pH is unlikely to
influence the conductivity relationship.
Reasonably confident: majority negative. No treatment for
confounding.
B-17
-------
Table B-17. Evidence and weights for confounding by low pH
Type
1 . Correlation of cause
and confounder
2. Correlation of effect
and confounder
3. Contingency of high
level of cause and
confounder
5 . Level s of confounder
known to cause
effects
Weight of evidence
Score
+
+
+
+
Evidence
Conductivity was moderately correlated with pH <6
(/• = -0.48, H = 145).
Low pH was moderately correlated with Ephemeroptera
(r = 0.46, n= 145).
Even at low pH some low conductivity streams support
some Ephemeroptera but not at high conductivities (see
Table B-15).
Hatching success of the mayfly Habrophlebia vibrans was
reduced a pH of 5.0 and lower (Rowe et al., 1988).
WVSCI was not reduced at pH 4-6 unless aluminum was
elevated in the Clear Fork, WV, study (Gerritson et al.,
2010).
Potential confounding: majority positive. Correction for
confounding was preformed.
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, but the evidence does not support confounding
(see Table B-18). No correlations were found between selenium and Ephemeroptera or between
selenium and conductivity in the West Virginia data set or in the EPA Region 3 data set. This
result is unreliable because most of the selenium values were detection limits, and many of the
detection limits were relatively high, even equaling the water quality criterion of 5.0 ug/L. In
addition, there were too few high selenium concentrations in the West Virginia data to perform a
contingency table analysis. For these reasons, correlational evidence of confounding was
ambiguous.
Evidence of the sufficiency of observed selenium levels to cause extirpation of stream
macroinvertebrates is weakly negative. The National Ambient Water Quality Criterion (5 |ig/L)
is irrelevant because it is based on more sensitive vertebrates (U.S. EPA, 2004). Field and
laboratory studies have found invertebrates to be relatively insensitive and unaffected at levels
observed in WV streams (Lemly, 1993; Chapman et al., 2010). In outdoor artificial streams
dosed with selenium, insects were less sensitive than fish, crustaceans, and oligochaetes; baetid
mayfly nymphs (Baetis, Callibaetis\ damselfly nymphs (Enallagma), and chironomid larvae
were not statistically significantly reduced—even at 30 |ig/L (Swift, 2002). Relatively few
invertebrate species have been tested and highly sensitive species may be identified in the future
B-18
-------
(DeBruyn and Chapman, 2007), but the available toxicological evidence does not indicate that
selenium confounds the relationship between conductivity and invertebrate extirpation.
The effects of removing high selenium on the conductivity relationship (Evidence
Type 6) were addressed using the West Virginia data set. When data from streams with
selenium concentrations above the water quality criterion (5 ug/L) were removed, the linear
correlation coefficient for number of ephemeropteran genera and log conductivity is barely
changed (r = -0.56, n = 339) relative to the full data set. When the same analysis was performed
with the EPA data set, the correlation was actually greater than that for the full data set
(r = -0.84, n = 32) (see Figure B-2), which is contrary to expectations for a confounder. This
result indicates that the conductivity relationship is not confounded by toxic effects of selenium.
CD
CD 8 ~
c
CD
O
1 • \J
Q.
0
CD
E
xi 4 -
Q.
LJJ
M —
O
CD o _
.Q ^
E
D
Z
o -
oo o
r=-0.84
o o o o
000 00
o o o o
o
o o
o
O CD
O (BDDCKD
i i i i '
50 100 200 500 1000 2000
Conductivity (|j,S/cm)
Figure B-2. Spearman's correlation coefficient and scatterplot between the
number of ephemeropteran genera and conductivity for 32 sites with low
selenium concentrations (<5 ug/L). Data from the EPA Region 3 data set.
B-19
-------
Table B-18. Evidence and weights for confounding by selenium
Type
1 . Correlation of cause
and confounder
2. Correlation of effect
and confounder
5 . Level s of confounder
known to cause
effects
6. Removal of
confounder shows it
is important
9. Specific effects of the
confounder
Weight of evidence
Score
0
0
—
—
—
Evidence
Conductivity was not correlated with total selenium in the
WV data set (r = 0.09, « = 501) and in the EPA data set
(r = -0.07, n = 46), but the evidence is ambiguous due to
poor selenium data.
Ephemeroptera were not correlated with total selenium in the
WV data set (r = -0.04, n = 501) and in the EPA data set
(r = -0.07, n = 46), but the evidence is ambiguous due to
poor selenium data.
In the most relevant toxicity test, effects on insects in an
artificial stream over an exposure of >2 years, occurred at
>0.030 mg/L (Swift, 2002). The 90th centiles for dissolved
and total selenium were (0.003 and 0.005 mg/L).
After removing high selenium sites (>5 |ig/L), the
correlation of Ephemeroptera with conductivity is barely
changed (r = -0.56, n = 339) relative to the full data set, but
the evidence is not strong because few sites have high
selenium. The same analysis performed with the EPA data
set also found no reduction in correlation (Figure B-2).
Selenium affects fish more than invertebrates and, in Swift
(2002), crustaceans and oligochaetes more than insects,
which is not the pattern seen in the streams.
Selenium causes characteristic deformities in fish, which
have not been seen in the streams.
Selenium effects occur primarily in top predators, not
herbivores and detritivores such as the Ephemeroptera.
Selenium at ambient concentrations causes effects in lentic
systems but not lotic systems such as the streams sampled in
WV. Deformities typical of selenium have been found in a
reservoir in the region but not in streams (WVDEP, 2010).
Reasonably confident: majority negative. No treatment for
confounding.
Consideration of the specific effects of selenium (Evidence Type 9) suggests that it is not
an important contributor to the impairment. First, the most sensitive organisms to aqueous
selenium are fish and other oviparous vertebrates (Chapman et al., 2010) but, in this case,
relatively selenium-insensitive insects are most affected. Second, selenium causes characteristic
deformities in fish, which have not been reported in WV streams. Third, the effects of selenium
B-20
-------
at low concentrations are seen in lentic ecosystems (lakes, reservoirs, ponds, wetlands)—not in
streams like those from which the conductivity relationship and benchmark were derived
(Chapman et al., 2010). Finally, because selenium is biomagnified, it primarily affects top
predators not the herbivores and detritivores that are affected in this case. This specificity is
supported by the fact that, in the region, the only reported effects of selenium are greatly elevated
body burdens and associated deformities in a top predator fish (largemouth bass) in a lentic
system (Upper Mud River Reservoir) (WVDEP, 2009, 2010).
The weight of evidence does not support confounding by selenium, so no action was
taken to adjust the dataset or analysis. However, because existing selenium data are poor, the
occurrence of selenium 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 reduced stream shading or if saline effluents are warmed. In an evaluation using
contingency tables, Ephemeroptera were present at 99-100% of sites at low conductivity at both
high and low temperature (see Table B-19). However, the differences between low and high
temperature are not large and that in itself suggests that temperature would not be a confounder
(a variant of Evidence Type 5). Correlations of temperature with conductivity are inconsistent
(see Table B-20). More importantly, elevated temperature does not appear to be associated with
the loss of Ephemeroptera and the relationship of conductivity to Ephemeroptera is not
influenced by elevated temperatures.
Table B-19. Number of sites with high and low temperatures and high and
low conductivity with Ephemeroptera present in streams (pH >6)
Temperature <17°C
Temperature >22°C
Conductivity <200 uS/cm
468/474
(99%)
78/78
(100%)
Conductivity > 1,500 uS/cm
9/27
(33%)
24/43
(56%)
B-21
-------
Table B-20. Evidence and weights for confounding by temperature
Type
Score
Evidence
1. Correlation of cause
and confounder
0
Temperature was moderately correlated with conductivity
year-round in the WV data set (r = 0.39, n = 2,216) but
weakly correlated in the EPA data set (r = 0.17, n = 46).
2. Correlation of effect
and confounder
Temperature was weakly correlated with Ephemeroptera
year round in the WV data set (r = -0.22, n = 2,216) and
uncorrelated in the EPA data set (r = -0.06, n = 46)
3. Contingency of high
level of cause and
confounder
Ephemeroptera were present at 99-100% 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-19),
which is contrary to expectations, if temperature were
contributing to the impairment.
5. Level s of confounder
known to cause
effects
Temperature limits are highly taxon specific but
temperatures rarely exceeded the WV limits for reference
sites (<30.6°C May-November and <22.8°C
December-April) and, therefore, are not likely to cause
extirpation.
6. Removal of
confounder shows it
is important
When high temperatures (>22°C) were deleted, the
correlation of conductivity and Ephemeroptera was
unchanged (r = -0.61, n = 1,787).
7. Multivariate statistics
Habitat quality, temperature and fecal coliform together
had essentially no effect on the slope in multiple
regression (see Table B-7).
Weight of evidence
Moderately confident: none positive, some strongly
negative. No treatment for confounding.
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
individual watersheds shows that many if not most of the sampled sites have at least one
upstream intact headwater. Two examples are presented here.
B-22
-------
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-3, B-4,
and B-5). Conductivity in Spring Branch measured <44-66 uS/cm. Conductivity in Ballard
Fork was 464-2,300 uS/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 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. Also, habitat quality (total RBP
habitat score), embeddedness, and pH are not related to biological quality, so they are not
confounders in these streams (see Table B-21).
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.
Figure B-3. Topographical map of Spring Branch (blue triangle) and
Ballard Fork (red triangle) sampling stations.
B-23
-------
Figure B-4. 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 by white arrows as examples.
Figure B-5. Aerial imagery (April 10,1996) with superimposed sampling
locations of spring branch (turquoise square) and Ballard Fork (yellow
square). Same area as Figure 3. The many upstream valley fills in Ballard Fork
are easily seen.
B-24
-------
Table B-21. Comparison of low conductivity Spring Branch with high
conductivity Ballard Fork
Stream name
Spring Branch
Spring Branch
Spring Branch
Ballard Fork
Ballard Fork
Ballard Fork
Ballard Fork
Ballard Fork
Date
5/9/2006
4/18/2000
4/20/1999
5/9/2006
4/18/2000
1/25/2000
7/26/1999
4/20/1999
Embed.
16
16
14
12
Total RBP
score
149
163
149
148
pH
7.7
7.5
7.7
8.1
7.1
7.9
8.2
8.1
uS/cm
66
44
51
1,195
464
1,050
2,300
1,201
#E
8
6
8
3
1
0
0
O
%E
29.27
44.76
34.72
2.96
2.08
0
0
4.12
Total
count
205
143
337
203
48
52
88
291
Embed. = embeddedness score from RBP; RBP = Rapid Bioassessment Protocol Habitat Evaluation; # E = Number
of ephemeropteran genera; % E = percent of ephemeropteran individuals in the sample; Total count = count of all
individuals of all taxa.
Source: data from U.S. EPA mountaintop mining studies (Green et al, 2000; Pond et al, 2008).
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-22). WVSCI scores, number of
ephemeropteran families and number of ephemeropteran, plecopteran, and trichopteran (EPT)
families were low when conductivity was high regardless of the condition of catchments that
provided sources of benthic macroinvertebrates including salt-sensitive genera.
In these two examples, the evidence indicates that 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-25
-------
Table B-22. Twentymile Creek sampling locations, conductivity, habitat
score, number of EPT taxa, and WVSCI scores
Year
2003
2004
1998
2003
2003
2004
2003
2004
2003
2004
1998
River
kilometer
44.6
44.6
25.1
25.1
11.9
11.9
1.8
1.8
0.5
0.5
0
Tributary
catchment
land usea
Forested
Forested
Mined
Mined
Mixed Forest
and Mine
Mixed Forest
and Mine
Mixed Forest,
Mine, & Urban
Mixed Forest,
Mine, & Urban
Mixed Forest,
Mine, & Urban
Mixed Forest,
Mine, & Urban
Mixed Forest,
Mine, & Urban
Max reported
conductivity
(uS/cm)
44
37
805
2,087
1,702
1,282
987
1,138
845
836
590
RBP
habitat
score
148
155
153
157
—
—
—
146
—
131
#E
4
3
1
2
—
—
—
2
—
O
#EPT
15
7
5
7
—
—
—
6
—
8
WVSCI
90.72
-
67.62
58.45
64.74
—
—
—
66.73
—
65.94
"Land use refers to catchment land use of tributaries upstream from the sampled sites in Twentymile Creek.
# E families = Number of ephemeroptera families; #EPT = ephemeropteran, plecopteran, and trichopteran families;
WVSCI = West Virginia Stream Condition Index.
Source: data from WABbase.
B.4.10. Catchment Area
Larger streams tend to have more moderate chemical properties than small streams
because they receive waters from more sources, both natural and anthropogenic. Consequently,
extreme values, in this case both low and high conductivity, tend 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
B-26
-------
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 of smaller streams? We examined these issues by
analyzing the influence of stream size (as catchment area) on the effects of conductivity and on
the occurrence of Ephemeroptera.
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 (15.5 km2 to 155 km2), and large
catchments greater than 60 mi2 (155 km2). In all three stream size categories, if conductivity was
<200 uS/cm, 99% or more of all streams had Ephemeroptera, but if conductivity was above
1,500 uS/cm, fewer streams had Ephemeroptera (see Table B-23). The number of
Ephemeroptera taxa declines with increasing conductivity in all streams with measured
catchment areas, independent of classification of catchment area (r = -0.59). Correlation of log
conductivity with log catchment area is weak (see Table B-24).
The weight of evidence for confounding by catchment area (see Table B-24) is uniformly
negative, so we conclude that catchment area has little or no effect on invertebrate response to
conductivity.
Table B-23. Number and percent of streams with Ephemeroptera present:
small, medium and large streams and low and high conductivity (pH >6)
Small streams
(<15.5 km2)
Medium streams
(>15.5km2and
<155 km2)
Large streams
(>155 km2)
Conductivity <200 uS/cm
302/303
(100%)
118/119
(99%)
37/37
(100%)
Conductivity >1,500
uS/cm
6/15
40%)
10/14
(71%)
1/2
(50%)
B-27
-------
Table B-24. Evidence and weights for confounding by catchment area
Type
1 . Correlation of cause
and confounder
2. Correlation of effect
and confounder
3. Contingency of high
level of cause and
confounder
6. Removal of
confounder
Weight of evidence
Score
—
—
Evidence
Log catchment area was very weakly correlated with log
conductivity (r = 0.18, n = 926).
Log catchment area was not correlated with the number of
ephemeropteran genera (r = -0.009, n = 926).
In a contingency table (see Table B-23), catchment area did
not affect the probability of finding Ephemeroptera at low
conductivity. Medium size somewhat increased the
probability of occurrence at high conductivity.
When large streams were removed, the correlation of
conductivity and number of ephemeropteran genera was
barely changed (r = -0.60, n = 837).
Very confident: all negative, some strongly negative. No
treatment for confounding.
B.4.11. Ponds
The effluents from most valley fills flow into settling ponds, and it has been suggested
that those ponds are the actual cause of downstream community impairments. This issue was
addressed using the EPA Region 3 data set because it identifies the presence of ponds. When
data from only streams with ponds are used (i.e., the occurrence of ponds is removed as a
variable—Evidence Type 4), the correlation coefficient for number of ephemeropteran genera
and log conductivity is r = -0.84 (see Figure B-6). This result is somewhat higher than those for
the uncensored EPA Region 3 data set (r = -0.73), which is contrary to the expectation if ponds
were the cause. This result clearly shows that the conductivity relationship is not a result of
co-occurrence with ponds. In addition, when ponds are removed and the streams are reclaimed,
conductivity remains high and the effects continue. For example, Venter's Branch and Jones
Branch in Martin County, KY, were mined in the mid 1990s, and the ponds were removed.
When the streams were sampled in 2009, conductivity was >2,000 uS/cm and no Ephemeroptera
were found in either stream (Greg Pond, U.S. EPA, personal communication).
The weight of evidence for confounding from ponds is uniformly negative, so we
conclude that the presences of ponds have little or no effect on invertebrate response to
conductivity.
B-28
-------
0
c
0
O
0
-i—'
Q.
2
0
E
0
.c
Q.
LU
0
.Q
E
8 -
6 -
4 -
2 ~
0 -
r=-0.84
o o ooo
O O
-------
Table B-25. Number of sites with high and low dissolved oxygen and high
and low conductivity with Ephemeroptera present in streams (pH >6)
DO >10.3 mg/L
DO <8.2 mg/L
Conductivity <200 uS/cm
244/246
(99%)
172/174
(99%)
Conductivity > 1,500 uS/cm
11/28
(39%)
12/30
(40%)
Table B-26. Evidence and weight for confounding by dissolved oxygen (DO)
Type
Score
Evidence
1. Correlation of
cause and
confounder
DO was weakly correlated with conductivity (r = -0.11,
n = 2,188).
2. Correlation of
effect and
confounder
DO was uncorrelated with the number of ephemeropteran
genera (r = 0.09, n = 2,188).
3. Contingency of
high level of cause
and confounder
In a contingency table (see Table B-25), Ephemeroptera are
present at 99% of sites with low conductivity (<200 uS/cm)
even when DO is low (<8 mg/L) and at high conductivity the
presence of Ephemeroptera is unaffected by DO.
5. Level of
confounder known
to cause effects
The 30 day mean water quality criteria for DO of 6.5 mg/L
for coldwater and 5.5 mg/L for warm water (U.S. EPA,
1986) are below the lower 10
(7.3 mg/L).
,th
centile of WV sites
Reduced body condition and survivorship in
ephemeropterans occur below 7 mg/L DO in laboratory
studies (Love et al., 2005; Pucket and Cook, 2004).
In the Clear Fork watershed, WV, a plausible threshold for
DO of 5 mg/L and a substantial threshold of 4 mg/L were
derived (Gerritsen et al., 2010).
6.
Removal of
confounder
When sites with moderate to low DO (<8.2 mg/L) were
removed from the analysis, the correlation of conductivity
with the number of Ephemeroptera is slightly increased
(r = -0.63, w= 1,642).
Weight of evidence
Very confident: all negative, some strongly negative. No
treatment for confounding.
B-30
-------
B.4.13. Metals
Iron (Fe), aluminum (Al), and manganese (Mn) are the metals most associated with acid
mine drainage and commenters have suggested that they may cause the impairment associated
with conductivity. However, for the following reasons, the circum-neutral to moderately
alkaline streams are unlikely to experience toxicity from these metals (Luoma and Rainbow,
2008). The most toxic form of iron (free Fe2+) does not occur in oxygenated waters above pH 4.
Under those conditions, iron occurs as hydroxide particles or, if significant dissolved organic
matter is present, as iron colloids. In these forms, iron is thought to serve primarily to reduce the
toxicity of co-occurring metals by adsorption and co-precipitation. Toxic divalent aluminum
precipitates similarly above pH 5 as hydroxide floes or polymeric aluminum. Divalent
manganese is converted to insoluble Mn4+ in mildly alkaline waters. The precipitates of these
metals may adversely modify habitats and directly affect organisms. However, the valley fill
effluents that are primarily responsible for the relationship between conductivity and extirpation
of invertebrates are not equivalent to the acid drainage into neutralizing streams that results in
heavy accumulations of precipitates. Finally, the toxicity of these divalent anions is mitigated by
divalent calcium, which is the dominant cation in the saline mixtures. Hence, it is expected that,
as conductivity increases, the toxicity of these metals will decrease per unit concentration.
Because of concern for combined effects of metals, multiple linear regression of
conductivity, iron, aluminum and manganese was performed. The metals reduced the coefficient
for conductivity by only 8.6% (see Table B-27).
Iron and aluminum are clearly not confounders, based on contingency table analyses (see
Tables B-28 and B-29), weak correlations (see Tables B-3 land B-32) and other evidence (see
Tables B-31 and B-32).
However, manganese is more ambiguous since it is moderately correlated with both
conductivity and ephemeropteran genera (see Tables B-30 and B-33). Manganese has been
relatively poorly studied because it has seldom been found at toxic levels. Like other divalent
cationic metals, Mn2+ is less toxic in hard (i.e., high Ca) waters and the high conductivity waters
in this region are inherently hard. Based on a linear relationship of hardness to conductivity in
the WV data, 300 |iS/cm conductivity is equivalent to a hardness of approximately 200 mg/L
CaCOs. The equivalent hardness-adjusted British Columbia Chronic Water Quality Guideline
for manganese is 1.5 mg/L (BC, 2001). Dittman and Buchwalter (2010) provide the laboratory
study with the most directly relevant taxa: aquatic insects from Appalachia. They quantified
bioaccumulation and performed biomarker studies that found reduced levels of cysteine and
glutathione at 0.10 and 0.50 mg/L but they saw no overt toxic effects. The most relevant
conventional toxicity tests of aquatic invertebrates were 21 day reproduction tests ofDaphnia
magna which yielded IC25 values of 5.4 and 9.4 mg/L for hardness levels of 100 and 250 mg/L,
B-31
-------
respectively (Reimer, 1999). A recent assessment of the Clear Fork watershed, WV, concluded
that total manganese at 0.002-0.50 mg/L was a minor contributor to biotic impairment, because
manganese was weakly correlated (r = -0.16) with the WVSCI index when corrected for
stronger causes (Gerritsen et al., 2010).
In sum, iron and aluminum are clearly not confounders. Equivocal evidence suggests
that manganese is potentially a weak confounder.
Table B-27. An output table for two linear regression models. The first is the
simple model predicting ephemeropteran genera from conductivity. The second
is a multivariate model with the additional covariates iron, aluminum, and
manganese.
Parameter
Estimate
Standard error
Univariate model
Intercept
Conductivity slope
3.65
-0.93
0.056
0.024
Multivariate model
Intercept
Conductivity slope
Iron
Aluminum
Manganese
3.05
-0.85
-0.028
-0.066
-0.30
0.092
0.031
0.042
0.044
0.033
Table B-28. Number of sites with high and low total iron and high and low
conductivity with Ephemeroptera present in streams (pH >6)
Iron >0. 5 mg/L
Iron <0. 12 mg/L
Conductivity <200 uS/cm
122/124
(98%)
139/140
(99%)
Conductivity > 1,500 uS/cm
8/28
(29%)
13/24
(54%)
B-32
-------
Table B-29. Number of sites with high and low total aluminum and high and
low conductivity with Ephemeroptera present in streams (pH >6)
Aluminum >0.23 mg/L
Aluminum <0.09 mg/L
Conductivity <200 uS/cm
177/178
(99%)
103/103
(100%)
Conductivity > 1,500 uS/cm
5/22
(23%)
14/31
(45%)
Table B-30. Number of sites with high and low total manganese and high
and low conductivity with Ephemeroptera present in streams (pH >6)
Mn>0.1 mg/L
Mn <0.02 mg/L
Conductivity <200 uS/cm
69/72
(96%)
158/158
(100%)
Conductivity > 1,500 uS/cm
13/50
(26%)
3/5
(60%)
B-33
-------
Table B-31. Evidence and weight for confounding by iron
Type
Score
Evidence
1. Correlation of cause
and confounder
Dissolved iron was uncorrelated with conductivity in the WV data
(r = -0.08, n = 1,265) and weakly correlated in the EPA data (r = -0.17,
n = 12). Both signs are incorrect for confounding.
Total iron was uncorrelated with conductivity in the WV data (r = 0.03,
n = 1,439) and weakly correlated with the wrong sign in the EPA data
(r = -0.14, w = 46).
2. Correlation of effect
and confounder
Dissolved iron was uncorrelated with the number of ephemeropteran
genera in the WV data (r = -0.08, n = 1,265) and in the EPA data
(r = -0.04, n = 12).
Total iron was weakly correlated with the number of ephemeropteran
genera in the WV data (r = -0.14, n = 1,436) and in the EPA data with
the wrong sign (r = 0.12, n = 46).
3. Contingency of high
level of cause and
confounder
In a contingency table (see Table B-28), Ephemeroptera are present at
>98% of sites with low conductivity (<200 uS/cm) even when total iron
is high (>0.1 mg/L). There are too few observations at extreme
conductivities to derive a contingency table for dissolved iron.
5. Level of confounder
known to cause effects
The most relevant criteria are the British Columbia Chronic Water
Quality Guidelines of 1 mg/L for total iron and 0.35 mg/L for dissolved
iron (BC, 2008), which are above the 90th centiles in WV (0.93 and
0.14 mg/L, respectively).
The most relevant conventional toxicity tests were 120 h tests of the
mayfly Leptophlebia marginata with an LC50 of 106.3 mg/L and reduced
predator avoidance at 70 mg/L at pH 7 and low conductivity (7.0 uS/cm)
(Gerhardt, 1994) which are well above the maximum dissolved iron in
WV.
Two highly relevant field studies use data from the same source. Total
iron caused no or minimal change at 0.21 mg/L and slight to moderate
changes at 1.74 mg/L using benthic macroinvertebrate abundances in the
WVDEP data set (Linton et al., 2007). Acid drainage sites were not
excluded. Gerritson et al. (2010) found no effects of iron in the WVDEP
data set.
6. Removal of confounder
When sites with moderate to high dissolved iron (>0.06 mg/L) were
removed from the analysis, conductivity is more negatively correlated
with the number of Ephemeroptera (r = -0.72, n = 949), which is
contrary to expectations for a confounder. This result is corroborated by
the EPA data set (r = -0.77, n = 9).
When sites with moderate to high total iron (>0.5 mg/L) were removed
from the analysis, conductivity is slightly more negatively correlated
with the number of Ephemeroptera (r = -0.66, n = 1,076), which is
contrary to expectations for a confounder. This result is corroborated by
the EPA data set (r = -0.64, n = 34).
7. Multivariate statistics
In the multiple linear regression, the slope for iron is less than a tenth
that of conductivity (see Table B-27).
Weight of evidence
Very confident: all negative, some strongly negative. No treatment for
confounding.
B-34
-------
Table B-32. Evidence and weight for confounding by aluminum
Type
1 . Correlation of cause
and confounder
2. Correlation of effect
and confounder
3. Contingency of high
level of cause and
confounder
5. Level of confounder
known to cause effects
6. Removal of confounder
7. Multivariate statistics
Weight of evidence
Score
-
0
0
—
—
Evidence
Dissolved aluminum was weakly correlated with conductivity in the WV
data (r = 0.12, n = 1,293) and in the EPA data (r = 0.18, n = 12).
Total aluminum was weakly correlated with conductivity and in the
wrong direction in the WV data (r = -0. 12, n = 1,442) and uncorrelated
in the EPA data (r = 0.03, n = 46).
Dissolved aluminum was weakly correlated with the number of
ephemeropteran genera in the WV data (r = -0. 16, n = 1,293) and
uncorrelated in the EPA data (r = -0.02, n = 12).
Total aluminum was uncorrelated with the number of ephemeropteran
genera in the WV data (r = 0.03, n = 1,442) and weakly correlated in the
EPA data (r = 0. 15, n = 46); both have the wrong sign
In a contingency table (see Table B-29), Ephemeroptera are present at
>99% of sites with low conductivity (<200 uS/cm) even when total
aluminum is high (>0. 1 mg/L). However, there are fewer Ephemeroptera
at high conductivity with high total aluminum so some confounding is
possible but only at levels far above the benchmark. There are too few
observations at extreme conductivities to derive a contingency table for
dissolved aluminum.
The most relevant criteria are the British Columbia Acute and Chronic
Water Quality Criteria of 0.1 and 0.05 mg/L, respectively, for dissolved
aluminum above pH 6.5 (BC, 2001). The chronic value equals the
median value in WV; the acute value equals the 90thcentile. However,
the criteria are based on effects on sensitive fish.
The most relevant conventional toxicity tests were 48 h tests of
Ceriodaphnia dubia in neutralized acid mine drainage which gave a
mean LC50 for total aluminum of 2.9 mg/L (Soucek et al., 2001). This
value is well above the 90th centile (0.1 mg/L) but its relevance to stream
insects is unclear.
In the most relevant field study, the plausible and substantial effects
thresholds were >0.2 mg/L and >0.4 mg/L dissolved aluminum in WV
Ecoregion 69 (Gerritson et al., 2010). These are above the 90th centile in
WV (0.1 mg/L).
When sites with moderate to high dissolved aluminum (>0.06 mg/L)
were removed from the analysis, conductivity is slightly more negatively
correlated with the number of Ephemeroptera (r = -0.68, n = 973) which
is contrary to expectations for a confounder.
When sites with moderate to high total aluminum (>0.23 mg/L) were
removed from the analysis, conductivity is slightly more negatively
correlated with the number of Ephemeroptera (r = -0.66, n = 1,063)
which is contrary to expectations for a confounder. This result is
corroborated by the EPA data (r = -0.79, n = 15).
In the multiple regression, the slope for aluminum is less than a tenth that
of conductivity (see Table B-27).
Moderately confident: none positive, some strongly negative. No
treatment for confounding.
B-35
-------
Table B-33. Evidence and weight for confounding by manganese
Type
1 . Correlation of cause
and confounder
2. Correlation of effect
and confounder
3. Contingency of high
level of cause and
confounder
5. Level of confounder is
known to cause effects
6. Removal of confounder
7. Multivariate statistics
Weight of evidence
Score
0
+
+
+
0
—
Evidence
Dissolved Mn was moderately correlated with conductivity in the WV
data (r = 0.64, n = 20) but weakly correlated in the EPA data (r = 0.22,
n = 12).
Total Mn was moderately correlated with conductivity in the WV data
(r = 0.35, n = 1,436) and in the EPA data (r = 0.55, n = 46).
Dissolved Mn was moderately correlated with the number of
ephemeropteran genera in the WV data (r = -0.73, n = 20) and in the
EPA data (r = -0.37, n = 12).
Total Mn was moderately correlated with the number of ephemeropteran
genera in the WV data (r = -0.41, n = 1,436) and in the EPA data
(r = -0.49, n = 46).
In a contingency table (see Table B-30), Ephemeroptera are present at
>96% of sites with low conductivity (<200 uS/cm) even when total Mn
is high (>0. 1 mg/L). However, there are fewer Ephemeroptera at high
conductivity with high total Mn suggesting that some confounding is
possible at levels far above the benchmark. There are too few dissolved
Mn observations at extreme conductivities to derive a contingency table.
The most relevant criterion is the British Columbia Chronic Water
Quality Guideline for Mn of 1.5 mg/L (BC, 2001). This is above the
maximum dissolved Mn.
The most relevant conventional toxicity tests were 21 day reproduction
tests ofDaphnia magna which yielded IC2s values of 5.4 and 9.4 mg/L
for hardness levels of 100 and 250 mg/L, respectively (Reimer, 1999).
This is far above the 90th centile and maximum dissolved Mn (0.29 and
1.06 mg/L), but its relevance to stream insects is uncertain.
In the most relevant field study, total Mn in the Clear Fork watershed,
WV, was weakly correlated (r = -0. 16) with the WVSCI index when
corrected for stronger causes and there were no substantial effects
(Gerristonetal., 2010).
When sites with moderate to high dissolved Mn (>0.05 mg/L) were
removed from the analysis, the correlation of conductivity with the
number of Ephemeroptera is little changed (r = -0.58, n = 16). There
were too few high dissolved Mn sites in the EPA data to corroborate.
When sites with moderate to high total Mn (>0. 1 mg/L) were removed
from the analysis, the correlation of conductivity with the number of
Ephemeroptera is slightly increased which is contrary to expectations for
a confounder (r = -0.63, n = 1,067). This result is corroborated by the
EPA data (r = -0.74, n = 34, compared to r = -0.72).
In the multiple regression, the slope for Mn is only 35% that of
conductivity and the conductivity slope is reduced by only 8.6% relative
to the univariate slope (see Table B-27).
Reasonably confident. Majority negative. No treatment for
confounding.
B-36
-------
B.5. SUMMARY OF ACTIONS TAKEN TO ADDRESS POTENTIAL CONFOUNDING
Low pH is an apparent confounder, but sites with pH <6 were removed from the data set
when calculating the benchmark value. Other potential confounders were eliminated from
consideration with some confidence. We do not argue that these variables do not cause
impairment at some locations in the region. Neither do we argue that they have no influence at
all on salt-impaired sites. Rather, given the inevitable variability in sites to which the benchmark
would be applied and the relatively strong relationship of conductivity and loss of sensitive
genera, the evaluated confounders do not substantially affect the model that is used to develop
and apply the conductivity benchmark.
REFERENCES
BC (British Columbia) (2001) Ambient water quality guidelines for manganese. Ministry of Environment, Province
of British Columbia, Canada. Available online at
http://www.env.gov.bc.ca/wat/wq/BCguidelines/manganese/manganese.html.
BC (British Columbia) (2008) Ambient water quality guidelines for iron. Ministry of Environment, Province of
British Columbia, Canada. Available online at
http://www.env.gov .be.ca/wat/wq/BCguidelines/iron/iron_overview.pdf
Chapman, PM; Adams, WJ; Brooks, ML; et al. (2010) Ecological assessment of selenium in the aquatic
environment. Boca Raton: SETAC/CRC Press.
DeBruyn, AMH; Chapman, PM. (2007) Selenium toxicity to invertebrates: will proposed thresholds for toxicity to
fish and birds also protect their prey? Environ Sci Technol 41(5):1766—1770.
Dittman, E; Buchwalter, DB. (2010) Manganese bioaccumulation in aquatic insects: Mn oxide coatings, molting
loss, and Mn(II) thiol scavenging. Environ Sci Technol 44(23):9182-9188.
Gerhardt, A. (1994) Short term toxicity of iron (Fe) and lead (Pb) to the mayfly Leptophlebia marginata (L.)
(Insecta) in relation to freshwater acidification. Hydrobiologia 284:157-168.
Gerritsen, J; Zheng, L; Burton, J; et al. (2010) Inferring causes of biological impairment in the Clear Fork
Watershed, West Virginia. U.S. Environmental Protection Agency, Office of Research and Development, National
Center for Environmental Assessment, Cincinnati, OH. EPA/600/R-08/146. Available online at
http://cfpub.epa. gov/si/si_public_record_report.cfm?dirEntryId=201963
Green, J; Passmore, M; Childers, H. (2000) A survey of the condition of streams in the primary region of
mountaintop mining/valley fill coal mining. Mountaintop mining/valley fill programmatic environmental impact
statement. Region 3, U.S. Environmental Protection Agency, Philadelphia, Pennsylvania. Available online at
http://www.epa.gOv/region03/mtntop/pdf/appendices/d/streams-invertebrate-study/FINAL.pdf
Hill, AB. (1965) The environment and disease: Association or causation. Proc Royal Soc Med 58(5):295-300.
Lemly AD. (1993) Guidelines for evaluating selenium data from aquatic monitoring and assessment. Environ
Monitor Assess 28:83-100.
Linton, TK; Pacheco, MAW; Mcintyre,DO; et al. (2007) Development of bioassessment-based benchmarks for iron.
Environ Toxicol Chem 26(6):1291-1298.
Love, JW; Taylor, CM; Warren, ML. (2005) Predator density and dissolved oxygen affect body condition of
Stenonema tripunctatum (Ephemeroptera, Heptageniidae) from intermittent streams. Hydrobiologia 543:113-118.
B-37
-------
Luoma, SN; Rainbow, PS (2008) Metal contamination in aquatic environments: science and lateral management.
Cambridge, UK: Cambridge University Press.
Pearl, J. (2009) Causality: models, reasoning, and inference, 2nd edition. Cambridge, UK: Cambridge University
Press.
Peters, GT; Cherry, DS; Cairns, J Jr. (1985) Response oflsonychia bicolor to alkaline pH: an evaluation of survival,
oxygen consumption, and chloride cell ultrastructure. Can. J. Fish. Aquat. Sci. 42:1088-1095.
Pond, GJ; Passmore, ME; Borsuk, FA; et al. (2008) Downstream effects of mountaintop coal mining: comparing
biological conditions using family- and genus-level macroinvertebrate bioassessment tools. J N Am Benthol Soc
27(3):717-737.
Puckett, RT; Cook, JL. (2004) Physiological tolerance ranges of larval Caenis latipenniss (Ephemeroptera:
Caenidae) in response to fluctuations in dissolved oxygen concentration, pH and temperature. Texas J Sci
56(2):123-130.
Reimer, PS. (1999) Environmental effects of manganese and proposed freshwater guidelines to protect aquatic life
in British Columbia. U. British Columbia, Vancouver, Canada. Available online at
http://www.env.gov .be.ca/wat/wq/wq_guidelines.html.
Rowe, L; Hudson, J; Berrill, M. (1988) Hatching success of mayfly eggs at low pH. Can J Fish Aquat Sci 45:
1649-1652.
Soucek, DJ; Cherry, DS; Zipper, CE. (2001) Aluminum-dominated acute toxicity to the cladoceran Ceriodaphnia
dubia in neutral waters downstream of an acid mine drainage discharge. Can J Fish Aquat Sci 58:2396-2404.
Stewart-Oaten, A. (1996) Problems in the analysis of environmental monitoring data. In: Detecting environmental
impacts, Schmitt, PJ; Osenberg, CW; ed. New York, NY: Academic Press, p. 109-131.
Swift, MC. (2002) Stream ecosystem response to, and recovery from, experimental exposure to selenium. J Aqua
Ecosyst Stress Rec 9(3): 159-184.
U.S. EPA (Environmental Protection Agency). (1976) Quality criteria for water. Office of Water, Washington, DC;
EPA/440/9-76/023. Available online at
http://water.epa.gov/scitech/swguidance/waterquality/standards/current/upload/2009_0 l_13_criteria_redbook.pdf.
U.S. EPA (Environmental Protection Agency). (1986) Quality criteria for water 1986. Office of Water, Washington,
DC; EPA 440/5-86/001. Available online at
http://water.epa.gov/scitech/swguidance/waterquality/standards/criteria/aqlife/upload/2009_01_13_criteria_goldboo
k.pdf.
U.S. EPA (Environmental Protection Agency). (2004) Draft aquatic life water quality criteria for selenium.
Washington, DC; EPA-822-D-04-001. Available online at
http://www.epa.gov/waterscience/criteria/selenium/pdfs/complete.pdf.
WVDEP (West Virginia Department of Environmental Protection). (2009) Selenium bioaccumulation among select
stream and lake fisheries. West Virginia Department of Environmental Protection. Available online at
http://www.dep.wv.gOv/W WE/watershed/wqmonitoring/Documents/Selenium/Se_Fish_Tissue_Summary_Paper_fin
al_Feb09.pdf.
WVDEP (West Virginia Department of Environmental Protection). (2010) Selenium-induced developmental effects
among fishes in select West Virginia waters. West Virginia Department of Environmental Protection. Available
online at
http://www.dep.wv.gov/WWE/watershed/wqmonitoring/Documents/Selenium/Se%20Larvae%202010%20final.pdf.
B-38
-------
APPENDIX C
DATA SOURCES AND METHODS OF LAND USE/LAND COVER ANALYSIS USED
TO DEVELOP EVIDENCE OF SOURCES OF HIGH CONDUCTIVITY WATER
ABSTRACT
Potential sources of elevated conductivity were characterized for subwatersheds within
the Coal, Upper Kanawha, Gauley, and New Rivers. From a large monitoring data set developed
by the West Virginia Department of Environmental Protection (WVDEP), 190 <20-km2
watersheds were found for which there was total maximum daily load (TMDL) and land cover
information in southwestern West Virginia and macroinvertebrate samples identified to the
genus level with at least one chemistry sample. Small <20-km2 subwatersheds were selected to
reduce confounding from multiple sources. Scatter plots of conductivity, SC>42 , and Cl, and
alkalinity levels were generated for nine land cover classifications: open water, agriculture,
forest, residential, barren, total mining, valley fill, abandoned mine lands, and mining excluding
valley fill and abandoned mine lands. Conductivity was negatively correlated with the
percentage of forest area and most strongly negatively associated with catchments with the
greatest percentages of valley fills, and the HCOs and SC>42 concentrations were greater than
Cl concentration. Areas with more residences and farm buildings also had elevated conductivity
but rarely exceeded 1,000 jiS/cm, and Cl~ often exceeded SO42~and HCO3 concentrations.
These findings confirm sources of high conductivity waters that are used as evidence in the
causal assessment that salts are a cause of impairment of aquatic macroinvertebrates in streams
in West Virginia (see Section A.2.2.4).
C.I. INTRODUCTION
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 for
different sources. This information was used as evidence of preceding causation in the causal
assessment described in Appendix A of this report.
C.2. METHODS
C.2.1. General Approach
Small catchments were delineated, and the proportions of land covers were regressed
against water quality parameters. Watershed size was limited to <20 km2 to minimize the variety
of land use and cover types within a single watershed, thereby providing a clearer signal for each
potential source of salinity. However, because the region has a long history of mining and land
C-l
-------
cover information may not include legacy mining, persistent effects of mining are potentially
present even when there is no current record of past or present mining activity in the publically
available land cover databases. Also, residences are present in areas where mining occurs.
Therefore, there are potential influences from multiple sources in most of the 190 watersheds,
but these are minimized by using small catchments.
The final data set consisted of 190 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 available. The
190 sites are located in the Coal, Upper Kanawha, Gauley, and New Rivers of Ecoregion 69D.
Water quality parameters are from the WVDEP's Watershed Assessment Branch Data Base
(WABbase). For each watershed, scatter plots for several parameters were generated for
nine land cover classifications: open water, agriculture, forest, urban/residential, barren, total
mining, valley fill, and mining excluding valley fill and abandoned mine lands.
C.2.2. 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 data, the WVGISTC (2011), maintains publicly available
shapefiles. WVDEP (201 la) also maintains a publicly available repository of statewide GIS data
sets (http://gis.dep.wv.gov/). All relevant GIS metadata are available for the data housed at each
repository site. All GIS coverages used in this EPA study are in universal transverse mercator
(UTM) 1983 Zone 17, and the units are in meters. Table C-l describes some of the publicly
available GIS shapefiles that were originally used to develop base files for WVDEP's TMDL
program. These base files were the beginning point for determining the 190 stations selected for
the analyses described in Section C.2.3 and were used to estimate land uses (see Table C-2). The
area in valley fill was from a 2003 coverage developed by WVDEP.
C-2
-------
Table C-l. 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
WVDEP GIS data sets
General West Virginia Universities
GIS data repository location
General WVDEP' s GIS data
repository location
http : //wvgis . wvu . edu/data/data.php
(WVGISTC, 2011)
http://gis.dep.wv.gov/ (WVDEP, 201 la)
Base Land use/land cover
GAP
NLCD2001
GAP land use
NLCD land use
http://wvgis.wvu.edu/data/dataset.php7ID
=62 (WVGISTC, 2002)
http://wvgis.wvu.edu/data/dataset.php7ID
=269 (WVGISTC, 2001)
Other files
Watershed Boundary
Data sets
NHD Streams
Abandoned Mine Lines
(AML-Highwalls) and
Polygons (AML Areas)
DMR Mining NPDES
Permits and Outlets
Mining Related Fills,
Southern West Virginia
Mining Permit
Boundaries
Roads Paved
Roads Unpaved
USGS 8 -digit Hydrologic Unit
Code boundaries
National Hydrography Data set
Streams
West Virginia abandoned mine
lands coverages. Highwall mine
coverage and AML area
WVDEP Office of Mining and
Reclamation NPDES permit and
outlet coverages
WVDEP valley fills coverage from
2003
WVDEP Mining permit boundaries
2000 TIGER/Line GIS and
WV Roads shapefiles
2000 TIGER/Line GIS shapefile
and digitized from aerial
photographs and topographic maps
http://wvgis.wvu.edu/data/dataset.php7ID
=123 (WVGISTC, 2004)
http://wvgis.wvu.edu/data/dataset.php7ID
=23 5 (WVGISTC, 2010)
http://wvgis.wvu.edu/data/dataset.php7ID
=150 (WVGISTC, 1996)
http ://gis .dep .wv.gov/data/omr.html
(WVDEP, 20 lib)
http ://gis .dep .wv.gov/data/omr.html
(WVDEP, 20 lie)
http ://gis .dep .wv.gov/data/omr.html
(WVDEP, 20 lid)
http://wvgis.wvu.edu/data/dataset.php7ID
=149 (WVGISTC, 2011)
http://www.census.gov/geo/www/tiger/tig
er2k/tgr2000.html (U.S. Census Bureau,
2000a)
http : //wvgis .wvu . edu/data/data.php
(WVGISTC, 2011)
http://www.census.gov/geo/www/tiger/tig
er2k/tgr2000.html (U.S. Census Bureau,
2000b)
http : //wvgis .wvu . edu/data/data.php
(WVGISTC, 2011)
GAP = Gap Analysis Program; GIS = geographic information system; NHD = National Hydrography Data Set;
NLCD = National Land Cover Database; NPDES = National Pollutant Discharge Elimination System; DMR =
Division of Mining Reclamation; USGS = U.S. Geological Survey; WVDEP = West Virginia Department of
Environmental Protection.
C-3
-------
Table C-2. Detailed WV TMDL land use category derivation and land use
derivation used in Appendix A. Base land use categories are highlighted in
grey.
Detailed WV
TMDL land use
category
Water
Wetland
Forest
Grassland
Cropland
Urban pervious
Urban impervious
Barren
Pasture
Paved roads
Unpaved roads
Revoked mining
permits
Abandoned mine
land
Quarry
Highwall
Oil and gas
Data source
Water — base LU coverage
Wetland — base LU coverage
Forest — consolidated all
forested types from base LU
coverage
Grassland — base LU
coverage
Cropland — consolidated all
cropland types from base LU
coverage
Urban — consolidated
urbanized types from base
LU coverage
Urban — consolidated
urbanized types from base
LU coverage
Barren — base LU coverage
Source tracking
Roads shapefiles
Roads shapefiles
AML information
AML shapefile
Mining shapefile
AML shapefile
Oil and Gas shapefile
Base land use from
which new source area
was subtracted
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
New area subtracted from
Grassland
New area subtracted from
Urban Impervious
New area subtracted from
Urban Pervious
New area subtracted from
Barren
New area subtracted from
Barren
New area subtracted from
Barren
New area subtracted from
Barren
New area subtracted from
Barren
Land use categories
used in scatter plots
in Appendix A
Water
Water
Forest
Agriculture
Agriculture
Urban/residential
Urban/residential
Barren
Agriculture
Urban/residential
Urban/residential
AML
AML
Mining
Mining
Mining
C-4
-------
Table C-2. Detailed WV TMDL land use category derivation and land use
derivation used in Appendix A (continued)
Detailed WV
TMDL land use
category
Surface Mine Water
Quality permits
Surface Mine
Technology permits
Comingled mine
deep ground gravity
discharge
Comingled mine
deep ground pump
discharge
Undeveloped surface
mine WQ permits
Undeveloped surface
mine technology
permits
Undeveloped
comingled mine
gravity discharge
Undeveloped
comingled mine
pump discharge
Burned Forest
Harvested Forest
Skid Roads
TMDL land use
considers Valley
Fill3 area as part of
the Surface Mine
Water Quality and
Technology Permit
information
Data source
Mining shapefile
Mining shapefile
Mining shapefile
Mining shapefile
Mining shapefile
Mining shapefile
Mining shapefile
Mining shapefile
Forestry Dept. information
Forestry Dept. information
Forestry Dept. information
WVDEP valley fills
coverage from 2003
Base land use from
which new source area
was subtracted
New area subtracted from
Barren
New area subtracted from
Barren
New area subtracted from
Barren
New area subtracted from
Barren
New area subtracted from
Forest
New area subtracted from
Forest
New area subtracted from
Forest
New area subtracted from
Forest
New area subtracted from
Forest
New area subtracted from
Forest
New area subtracted from
Forest
New area subtracted from
Mining, Barren, and
Forest, as appropriate
Land use categories
used in scatter plots
in Appendix A
Mining
Mining
Mining
Mining
Mining
Mining
Mining
Mining
Barren
Barren
Barren
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 for this EPA report. See Table 1 for the source file.
AML = Abandoned Mine Line, LU = Land use, TMDL = total maximum daily load, WQ = water quality,
WV= West Virginia.
C-5
-------
C.2.3. Selection of Catchments
Catchments with available data that met the needs of the analysis involved a six-step
selection process that resulted in 190 catchments. The steps were preformed in the following
sequence:
• Select all WVDEP WAB stations located within Ecoregion 69D. This generated
2,151 stations.
• Select 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 was required to be associated with the
macroinvertebrate sample from the same station location. This narrowed the available
stations to 825.
• Select stations with detailed TMDL-associated land use located within the Coal, Upper
Kanawha, Gauley, and New River watersheds. This narrowed the selection to
382 stations.
• Eliminate stations if the detailed land use was not created during the TMDL process.
This eliminated 38 stations for a total of 344 stations.
• Eliminate stations located on undelineated tributary streams contained within a larger
mainstem subwatershed. This eliminated an additional 33 stations for a total of
311 stations.
• Select stations with a total watershed drainage area <20 km2 (4,942.08 acres). The total
number of remaining stations in TMDL watersheds within Ecoregion 69D after this last
reduction was 190 (see Figure C-l), and the data from these stations were assembled
from 1997 to 2007, with the majority of samples collected from 2001 to 2006.
C.2.4. Land Use Analysis
To create the land use for the 190 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 National Land Cover Data [NLCD] 2001) 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 C-2 contains the WVDEP TMDL
C-6
-------
N
A
WV State Boundary
190 stations
TMDL Watersheds
60
120
180 Kilometers
Figure C-l. Sampling locations used to develop evidence of sources of high
conductivity inputs. The 190 stations (black dots) at the terminus of each
>20-km2 catchment are shown within the larger 8-digit HUCs in southwestern
West Virginia.
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. In brief, nine land use categories were generated: total percentage area in mining
(% Total Mining) which is the sum of % Abandoned Mine, % MTM-Valley Fill and % Mining;
percentage in mountaintop mining valley fill (% MTM-Valley Fill); percentage of abandoned
mine lands (% Abandoned Mine); percentage of mining (% Mining) excluding % MTM-Valley
Fill and % Abandoned Mine; percentage barren land use (% Barren); percentage of residences,
buildings, and roads (% Urban/residential); percentage in agriculture and pasture
(% Agricultural); percentage in forest (% Forest), and percentage in open water (% Water).
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
C-7
-------
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 190 sampling stations. The
land use representation used 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 190 WAB sampling stations used seven categories consolidated from the TMDL land use
(see Table C-2) and then included the addition of the valley fill area. The % Total Mining
category is simply the sum of the % Mining, % MTM-Valley Fill, and % Abandoned Mine land
categories. The % Mining land use represents all other types of mining activities except for
abandoned mines and valley fill areas.
C.3. RESULTS
C.3.1. Characterization of Catchments and Ionic Matrix
The 190 small catchments used in the analysis are located near the borders of the 8-digit
hydrologic unit codes (HUCS) where elevations are greater and headwaters of these small
perennial streams are located (see Figure C-l). The ionic composition of these waters is not
uniform, but bicarbonate and sulfate are usually greater than chloride (see Table C-3) (see also
Table 1 and Table A-16, Pond et al., 2008). Because we were interested in all ions as well as the
mixture, we did not exclude high Cl~ sites. Only one site, New West Hollow, had a conductivity
measurement >300 (j,S/cm and higher chloride (629 mg/L) than sulfate (89 mg/L). That
watershed had the greatest area in residences, 16.4% urban, and a conductivity of 2,767 jiS/cm.
The potential presence of methane coal brine production was not ruled out.
C.3.2. Correlations with In-stream Biological and Water Quality Parameters
Pairs of land use and water quality parameters are listed in Table C-4 with at least one
Pearson's correlation coefficient with an r > |0.50|, except for a few with spurious points or
composed of only two points.
The two land use types that are most strongly and positively correlated with conductivity
are percentage of mining and percentage of valley fill. Percentage of forest is negatively
correlated with ion concentrations. Percentage of residential land use is not well correlated, and
in this region, is somewhat confounded by mining land uses. Among the ions that are more
strongly correlated, are total calcium and magnesium, also captured together as hardness,
-------
Table C-3 Summary statistics of water quality parameters in the
190 catchments
Parameter
Conductivity
Fecal
Alkalinity
Hardness
Sulfate
Chloride
TSS
Al, total
Al, dissolved
Ca, total
Cu, total
Cu, dissolved
Fe, total
Fe, dissolved
Mg, total
Mn, total
Se, total
Se, dissolved
Zn, total
Zn, dissolved
Flow
Temperature
pH
DO
Units
uS/cm
counts/mL
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
mg/L
ft3/s
°C
standard
units
mg/L
Min
6
1
0.02
11.26
5
1
0.3
0.02
0.02
1.93
0.001
0.001
0.02
0.02
1.28
0.003
0.001
0.001
0.005
0.005
0.004
0.05
3.03
1.22
25th
centile
254
4
14.45
37.95
84.55
2.0
3
0.06
0.02
7.63
0.003
0.003
0.09
0.02
4.3
0.025
0.005
0.001
0.009
0.005
0.41
8.705
7.105
9.26
Median
474
42
46
84.24
192
3.8
5
0.15
0.05
22.5
0.004
0.003
0.21
0.03
8.0
0.10
0.005
0.003
0.01
0.005
1.45
12.65
7.6
10.43
75th
centile
851
330
99.15
235.57
358
12.3
8
0.59
0.07
49.43
0.005
0.004
0.51
0.07
26.3
0.40
0.005
0.006
0.021
0.01
4.545
17.77
7.97
11.75
Max
3,964
60,000
710
862.6
2915
629
1217
23.6
23.5
184
0.014
1.91
32.8
13.1
97.9
27.3
1.26
1.26
0.18
0.726
63.01
30.72
12.99
11.81
Mean
445
47.5
36.73
85.11
168.32
5.5
5.97
0.21
0.063
17.716
0.004
0.004
0.24
0.045
9.97
0.116
0.005
0.003
0.014
0.009
1.25
13.139
7.355
11.23
Valid
N
1,671
1,181
1,348
48
1,350
45
1,348
1,342
1,335
50
24
40
1,341
1,329
49
1,340
436
23
25
40
839
1,672
1,671
1,666
TSS = Total suspended solids, Mean is geometric mean except for temperature, pH, and DO =dissolved oxygen.
C-9
-------
Table C-4. Correlation coefficients between pairs of land use and water
quality parameters in the land use data set
Water quality parameter
Conductivity
Alkalinity
Hardness
Sulfate
Calcium Total
Magnesium Total
% Valley fill
0.65
0.51
0.69
0.64
0.67
0.66
% Total mining
0.52
0.49
0.63
0.52
0.61
0.65
% Mining
0.39
0.37
0.55
0.39
0.52
0.58
% Forest
-0.54
-0.51
-0.63
-0.53
-0.64
-0.59
Parameters yielding only r < |0.50 are not shown.
bicarbonate measured as alkalinity, and sulfate. Noticeably chloride is not strongly correlated,
owing to fewer measurements of chloride, but also due to the low concentrations except at one
site. Chloride was 629 mg/L chloride at the site with the greatest residential and mining land
uses.
Individual scatter plots and associated correlation coefficients for conductivity can be
found in Appendix A, Section 2.2.2.4 but are reproduced here for the convenience of the reader
(see Figure C-2). At relatively low urban land use, the range of conductivity is highly variable.
In contrast, there is a clear pattern of increasing conductivity as percentage of area in valley fill
increases and of decreasing conductivity with increasing forest cover. When area in valley fill is
subtracted from the total nonacid mining area, the correlation decreases by 25% (see
Figure C-2d). The scatter plots illustrate that there are clear sources of increased conductivity,
but that percentage area in valley fill has the strongest correlation with conductivity (r = 0.65),
and percentage mining without a valley fill has a moderate correlation (r = 0.39).
Assuming that the lower conductivity values represent current best practices, we modeled
the lower 25th quantile of the percent valley fill scatter plot (see Figure C-3).
From the 10th quantile regression, the intercept for 300 |iS/cm is 4% valley fill and the
intercept for 500 |iS/cm is 8% valley fill. Using logistic regression at 300 and 500 |iS/cm, the
probability of impairment, based on a WVSCI score <68, is around 0.59 and 0.72, respectively.
At 300 jiS/cm, 5% of genera are extirpated, and at 500 jiS/cm, 17% of genera are extirpated (see
Figure 9). Because these estimates do not take into account the volume of the fill, construction
practices, distance from the fill, or dilution from tributaries, the estimate of conductivity
associated with percent valley fill is useful as a general characterization but will vary for specific
cases.
C-10
-------
a.
co
d.
o
O
8.
^ CD
> S
't3
o
8.
r = 0.53
0 2.2 9 30.6 99
%Total Mining
r=0.4
0 2.2 9 30.6
% Mining
99
b.
o
55
.£•
'>
'•S
D
T3
C
O
O
e.
o
O
8.
r = 0.65
0 2.2 9 30.6
%MTM-Valley Fill
8.
r = 0
0 2.2 9
% Barren
30.6
C- 5
I §-l
co
O
f.
O
O
8_
r = 0.02
0 2.2 9 30.6
% Abandoned Mine
om o
r = 0.13
0 0.6 1.5 3 5.3 9 14.8
% Urban/residential
E 8
O
8-
e a
8°
= -1D.03
0 0.6 1.5 3 5.3 9 14.8
% Agricultural
h.
.£•
>
'•S
o
O
8-
r = -0.55
•">.•
o %>
0 20 40 60 80 100
% Forest
E o
I 8H
co ^
^ CD
'> m '
'•s
o
O
r = 0.27
0 0.3 0.6 1 1.5 2.2
% Water
Figure C-2. Geometric mean conductivity associated with different land uses
in 190 watersheds in Ecoregion 69D and Spearman's correlation coefficient.
Conductivity increases with increasing % MTM-Valley Fill and % Total Mining,
and decreases with increasing % Forest, but there is less clear or no pattern with
other land use. From left to right, they are (a) % Total Mining (percentage
of deep, surface, quarry mining, MTM-Valley Fill, and abandoned mine land),
(b) % MTM-Valley Fill (from mountaintop mining overburden),
(c) % Abandoned Mine, (d) % Mining (inclusive of all types of mining except
MTM-Valley Fill and Abandoned Mine), (e) % Barren, (f) % Urban/residential,
(g) % Agricultural, (h) % Forest, and (i) % Water. Fitted LOWESS line with span
set at 2/3.
C-ll
-------
,O
OD
T3
C
O
O
CN
CD
CO
O
O
O
O
O
CN
CO
O
0
nr
0
Alldata - Mean Model
NonZero - Mean Model
10th Quantile Model
25th Quantile Model
\
Q
\
31
% Valley Fill
Figure C-3. Quantile regression of percentage of area in valley fill and
conductivity in 190 small watersheds in Ecoregion 69D. Assuming the lowest
conductivity points represent some of the best fill construction practices, the 10th
and 25th quantile regression lines are shown. The intercepts for 500 |iS/cm
(horizontal dashed line) are approximately 4% and 8% valley fill and for 300
|iS/cm are 1.5% and 3.9% valley fill for the 10th and 25th quantiles, respectively.
The mean model based on samples minus those with zero percent valley fill
shows that the relationship is unaffected by the removal of sites without valley
fills.
C-12
-------
C.4. CONCLUSIONS
Of the land uses in the small watersheds analyzed, only mining associated with valley
fills are significant sources of the salts that are measured as conductivity. Disturbances
associated with agriculture and human habitation may also contribute, but 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. Furthermore, natural background is
exceedingly low. For Ecoregion 69, the 25th centile from a probability-based sample from the
WABbase data set was 72 jiS/cm, N= 617 (see Section 5.5).
Although conductivity typically increases with increasing land use (Herlihy et al., 1998),
conductivity is highly variable at relatively low urban land use. This may be caused by unknown
mine drainage, deep mine break-outs, road applications, poor infrastructure condition (e.g.,
leaking sewers or combined sewers), gas drilling, or other practices. In contrast, there is a clear
pattern of increasing conductivity as the percentage of valley fill area increases and decreasing
conductivity with increasing percentage of forest cover area. This is evidence of at least one
strong source of high conductivity in the region (see Appendix A for causal assessment).
REFERENCES
Herlihy, AT; Stoddard, JL; Johnson, CB. (1998) The relationship between stream chemistry and watershed land
cover data in the mid-Atlantic region, U.S. Water, Air Soil Pollut 105(l-2):377-386.
Pond, GJ; Passmore, ME; Borsuk, FA; et al. (2008) Downstream effects of mountaintop coal mining: comparing
biological conditions using family- and genus-level macroinvertebrate bioassessment tools. J N Am Benthol Soc
27(3): 717-737.
Shank, M. (2004). Development of a mining fill inventory from multi-date elevation data. Presented at the
Advanced Integration of Geospatial Technologies in Mining and Reclamation Conference, December 7-9, 2004,
Atlanta, GA.
U.S. Census Bureau. (2000a) 2000 TIGER/Line CIS and WV roads shapefiles. Geography division. Available
online at http://www.census.gov/geo/www/tiger/tiger2k/tgr2000.html.
U.S. Census Bureau. (2000b). 2000 TIGER/Line GIS shapefiles aerial photographs and topographic maps.
Geography division. Available online at http://www.census.gov/geo/www/tiger/tiger2k/tgr2000.html.
WVDEP (West Virginia Department of Environmental Protection). (2011a) General WVDEP's data repository
locations. WVDEP GIS data sets. State of West Virginia, Charleston, WV. Available online at
http://gis.dep.wv.gov/.
WVDEP (West Virginia Department of Environmental Protection). (201 Ib). Mining permits, point locations.
Division of Mining and Reclamation Data. State of West Virginia, Charleston, WV. Available online at
http://gis.dep.wv.gov/data/omr.html.
WVDEP (West Virginia Department of Environmental Protection). (2011c) Mining related fills, Southern West
Virginia as of September 2003. Division of Mining and Reclamation Data. State of West Virginia, Charleston, WV.
Available online at http://gis.dep.wv.gov/data/omr.html.
C-13
-------
WVDEP (West Virginia Department of Environmental Protection). (20 lid) Mining permit boundaries. Division of
Mining and Reclamation Data. State of West Virginia, Charleston, WV. Available online at
http://gis.dep.wv.gov/data/omr.html.
WVGISTC (West Virginia GIS Technical Center). (1996) Mining - abandoned mine lands. University of West
Virginia, Morgantown, WV. Available online at http://wvgis.wvu.edu/data/dataset.php?ID=150.
WVGISTC (West Virginia GIS Technical Center). (2001) Land cover (NLCD 2001). WV NLCD (National land
cover database). University of West Virginia, Morgantown, WV. Available online at
http://wvgis.wvu.edu/data/dataset.php?ID=269.
WVGISTC (West Virginia GIS Technical Center). (2002) Land cover (GAP). WV GAP analysis program.
University of West Virginia, Morgantown, WV. Available online at http://wvgis.wvu.edu/data/dataset.php?ID=62.
WVGISTC (West Virginia GIS Technical Center). (2004) Watershed boundary data sets (8, 10, and 12 digits).
University of West Virginia, Morgantown, WV. Available online at http://wvgis.wvu.edu/data/dataset.php?ID= 123.
WVGISTC (West Virginia GIS Technical Center). (2010) Streams - national hydrography data sets (24k).
University of West Virginia, Morgantown, WV. Available online at http://wvgis.wvu.edu/data/dataset.php?ID=235
WVGISTC (West Virginia GIS Technical Center). (2011) General West Virginia Universities GIS data repository
locations. University of West Virginia, Morgantown, WV. Available online at http://wvgis.wvu.edu/data/data.php.
C-14
-------
APPENDIX D
EXTIRPATION CONCENTRATION VALUES FOR
GENERA IN THE WEST VIRGINIA DATA SET
ABSTRACT
The purpose of Appendix D is to provide the reader with a list of the extirpation
concentration (XCgs) values used to develop the species sensitivity distribution and the
hazardous concentration (HCos). Genera are ordered alphabetically (see Table D-l). The
numbers of occurrences in the data set and at West Virginia Department of Environmental
Protection (WVDEP) reference sites are noted in the right-hand columns.
Not all 95th centiles correspond to extirpation, and some imprecisely estimate the
extirpation threshold. The following rules were applied to the XCgs values using the fitted curve
and the confidence bounds from the plots in Appendix E. If the generalized additive model
(GAM) mean curve at maximum conductivity is approximately equal to 0 (defined as less than
1% of the maximum modeled probability), then the XCgs value is listed without qualification. If
the GAM mean curve at maximum conductivity is >0 but the lower confidence limit is
approximating to 0 (<1% of the maximum mean modeled probability), then the XCgs value is
listed as approximate (~). If the GAM lower confidence limit is >0, then the XCgs value is listed
as greater than (>) the 95th centile. All model fits and scatter of points were also visually
inspected for anomalies, and if the model poorly fit the data, the uncertainty level was increased
to either (~) or (>).
The assignation of (~) and (>) does not affect the HCos. They are provided to alert users
to the uncertainty of some XCgs values for other uses such as comparison with toxicity test
results or with results from other geographic regions.
D-l
-------
Table D-l. Extirpation concentration and sample size from West Virginia
data set. XCgs values reported without a preceding symbol indicate evidence of
extirpation within the tested range. XCgs values preceded by a (~) or (>) indicate
extirpation with greater uncertainty or extirpation at a level above the reported
value.
Order
Diptera
Ephemeroptera
Ephemeroptera
Plecoptera
Trichoptera
Plecoptera
Plecoptera
Ephemeroptera
Plecoptera
Diptera
Diptera
Isopoda
Diptera
Diptera
Ephemeroptera
Ephemeroptera
Diptera
Odonata
Diptera
Diptera
Isopoda
Ephemeroptera
Decapoda
Diptera
Family
Chironomidae
Baetidae
Baetidae
Perlidae
Glossosomatidae
Capniidae
Chloroperlidae
Ameletidae
Nemouridae
Culicidae
Tipulidae
Asellidae
Athericidae
Ceratopogonidae
Ephemerellidae
Baetidae
Ceratopogonidae
Aeshnidae
Tipulidae
Chironomidae
Asellidae
Caenidae
Cambaridae
Chironomidae
Genus
Ablabesmyia
Acentrella
Acerpenna
Acroneuria
Agapetus
Allocapnia
Alloperla
Ameletus
Amphinemura
Anopheles
Antocha
Asellus
Atherix
Atrichopogon
Attenella
Baetis
Bezzia
Boyeria
Brachypremna
Brillia
Caecidotea
Caenis
Cambarus
Cardiocladius
XC95
>1 1,646
1,337
-649
>2,630
365
542
246
591
812
>2,768
>6,468
960
>1 1,646
>2,257
-698
>1,395
380
>7,340
408
>2,005
>4,713
>3,923
>1,274
>2,257
N
162
752
27
512
27
33
101
219
589
26
565
33
157
43
34
1,527
62
175
27
95
141
552
472
191
TV from
reference
locations
5
31
3
60
6
15
15
30
42
2
18
2
3
3
1
71
2
5
2
6
1
8
44
2
D-2
-------
Table D-l. Extirpation concentration and sample size from West Virginia
data set (continued)
Order
Ephemeroptera
Trichoptera
Diptera
Diptera
Trichoptera
Trichoptera
Diptera
Diptera
Ephemeroptera
Diptera
Diptera
Diptera
Odonata
Megaloptera
Diptera
Amphipoda
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Ephemeroptera
Trichoptera
Plecoptera
Diptera
Family
Baetidae
Hydropsychidae
Chironomidae
Empididae
Hydropsychidae
Philopotamidae
Chironomidae
Tabanidae
Heptageniidae
Chironomidae
Empididae
Chironomidae
C ordul egastri dae
Corydalidae
Chironomidae
Crangonyctidae
Chironomidae
Chironomidae
Ceratopogonidae
Chironomidae
Chironomidae
Tipulidae
Chironomidae
Baetidae
Hydropsychidae
Perlodidae
Dixidae
Genus
Centroptilum
Ceratopsyche
Chaetocladius
Chelifera
Cheumatopsyche
Chimarra
Chironomus
Chrysops
Cinygmula
Cladotany tarsus
Clinocera
Conchapelopia
Cordulegaster
Corydalus
Corynoneura
Crangonyx
Cricotopus
Cryptochironomus
Dasyhelea
Demicryptochironomus
Diamesa
Dicranota
Dicrotendipes
Diphetor
Diplectrona
Diploperla
Dixa
XC95
1,092
>6,468
>5,057
>3,341
>9,180
>3,972
>1 1,646
>1 1,646
230
>1 1,646
>4,713
546
>1,436
>1 1,227
>2,006
>2,169
>1 1,227
>3,489
>3,341
322
>4,713
>7,010
>1 1,646
632
>2,527
315
>704
N
90
909
184
152
1,665
516
105
76
90
104
61
135
43
317
149
105
617
287
66
81
486
355
197
148
618
106
70
TV from
reference
locations
6
27
4
9
57
11
1
1
15
5
6
7
3
1
4
7
21
3
3
6
14
43
1
17
59
2
16
D-3
-------
Table D-l. Extirpation concentration and sample size from West Virginia
data set (continued)
Order
Trichoptera
Ephemeroptera
Coleoptera
Plecoptera
Coleoptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Diptera
Ephemeroptera
Amphipoda
Trichoptera
Trichoptera
Plecoptera
Diptera
Coleoptera
Diptera
Ephemeroptera
Diptera
Coleoptera
Trichoptera
Trichoptera
Ephemeroptera
Plecoptera
Diptera
Diptera
Odonata
Family
Philopotamidae
Ephemerellidae
Elmidae
Perlidae
Psephenidae
Heptageniidae
Ephemeridae
Ephemerellidae
Chironomidae
Ephemerellidae
Gammaridae
Glossosomatidae
Goeridae
Chloroperlidae
Chironomidae
Dryopidae
Empididae
Heptageniidae
Tipulidae
Dytiscidae
Hydropsychidae
Hydroptilidae
Isonychiidae
Perlodidae
Chironomidae
Chironomidae
Gomphidae
Genus
Dolophilodes
Drunella
Dubiraphia
Eccoptura
Ectopria
Epeorus
Ephemera
Ephemerella
Eukiefferiella
Eurylophella
Gammarus
Glossosoma
Goera
Haploperla
Heleniella
Helichus
Hemerodromia
Heptagenia
Hexatoma
Hydroporus
Hydropsyche
Hydroptila
Isonychia
Isoperla
Krenopelopia
Krenosmittia
Lanthus
XC95
>863
297
>7,370
497
>1,380
307
696
299
>1,876
490
>4,713
>1,652
-738
418
>1,697
-11,646
>9,790
326
>9,790
822
>7,010
>1 1,227
1,180
460
>2,320
-1,115
>2,087
N
356
176
144
65
324
414
148
405
519
189
216
157
25
253
62
333
615
68
846
32
999
281
740
520
62
27
66
TV from
reference
locations
46
18
3
6
32
53
20
38
28
19
10
7
4
27
7
18
8
3
65
1
21
4
16
39
2
3
7
D-4
-------
Table D-l. Extirpation concentration and sample size from West Virginia
data set (continued)
Order
Diptera
Trichoptera
Ephemeroptera
Ephemeroptera
Plecoptera
Diptera
Diptera
Diptera
Isopoda
Ephemeroptera
Coleoptera
Plecoptera
Coleoptera
Diptera
Diptera
Hemiptera
Diptera
Diptera
Trichoptera
Megaloptera
Diptera
Ephemeroptera
Trichoptera
Coleoptera
Decapoda
Diptera
Coleoptera
Family
Chironomidae
Lepidostomatidae
Leptophlebiidae
Heptageniidae
Leuctridae
Tipulidae
Chironomidae
Tipulidae
Asellidae
Heptageniidae
Elmidae
Perlodidae
Elmidae
Chironomidae
Chironomidae
Veliidae
Tipulidae
Chironomidae
Uenoidae
Corydalidae
Chironomidae
Heptageniidae
Hydroptilidae
Elmidae
Cambaridae
Chironomidae
Elmidae
Genus
Larsia
Lepidostoma
Leptophlebia
Leucrocuta
Leuctra
Limnophila
Limnophyes
Limonia
Lirceus
Maccaffertium
Macronychus
Malirekus
Microcylloepus
Micropsectra
Microtendipes
Microvelia
Molophilus
Natarsia
Neophylax
Nigronia
Nilotanypus
Nixe
Ochrotrichia
Optioservus
Orconectes
Orthocladius
Oulimnius
XC95
-2,630
-121
251
424
>2,087
-1,503
>5,120
>5,057
-1,323
-1,035
>1,890
>904
>3,341
>6,468
>3,489
>2,523
-2,169
>1,842
316
>9,790
>2,266
319
>2,791
>9,790
>3,162
>3,427
>2,791
N
96
91
87
225
1,199
54
88
62
72
214
44
27
94
227
532
46
28
54
166
746
112
77
32
1,471
205
277
227
TV from
reference
locations
3
12
8
29
84
10
1
1
6
13
4
6
2
24
33
3
2
1
35
36
O
O
1
63
2
9
27
D-5
-------
Table D-l. Extirpation concentration and sample size from West Virginia
data set (continued)
Order
Diptera
Plecoptera
Diptera
Plecoptera
Diptera
Ephemeroptera
Diptera
Diptera
Diptera
Plecoptera
Plecoptera
Diptera
Basommatophora
Veneroida
Ephemeroptera
Trichoptera
Diptera
Diptera
Diptera
Ephemeroptera
Coleoptera
Diptera
Coleoptera
Diptera
Diptera
Trichoptera
Plecoptera
Family
Chironomidae
Capniidae
Chironomidae
Perlidae
Chironomidae
Leptophlebiidae
Chironomidae
Chironomidae
Chironomidae
Peltoperlidae
Perlidae
Chironomidae
Physidae
Pisidiidae
Baetidae
Polycentropodidae
Chironomidae
Chironomidae
Chironomidae
Baetidae
Elmidae
Simuliidae
Psephenidae
Chironomidae
Tipulidae
Psychomyiidae
Pteronarcyidae
Genus
Pagastia
Paracapnia
Parachaetocladius
Paragnetina
Parakiefferiella
Paraleptophlebia
Parametriocnemus
Paraphaenocladius
Paratany tarsus
Peltoperla
Perlesta
Phaenopsectra
Physella
Pisidium
Plauditus
Polycentropus
Polypedilum
Potthastia
Procladius
Procloeon
Promoresia
Prosimulium
Psephenus
Pseudochironomus
Pseudolimnophila
Psychomyia
Pteronarcys
XC95
>1,800
334
>1,147
2,087
>1,757
463
>4,713
>6,468
>3,489
>694
3,314
-2,332
>9,790
>1,795
996
>4,713
>4,884
>1,886
>1 1,227
702
-672
-531
>9,119
>1 1,646
>1,357
>1,131
-634
N
46
37
169
40
75
449
1,501
71
110
126
315
89
145
34
289
380
1,648
62
28
78
79
106
886
31
135
39
113
TV from
reference
locations
2
13
27
3
2
46
72
2
2
12
8
1
1
2
12
41
70
1
1
3
5
20
35
2
11
3
25
D-6
-------
Table D-l. Extirpation concentration and sample size from West Virginia
data set (continued)
Order
Trichoptera
Plecoptera
Hemiptera
Diptera
Diptera
Diptera
Trichoptera
Ephemeroptera
Megaloptera
Diptera
Diptera
Diptera
Ephemeroptera
Coleoptera
Ephemeroptera
Odonata
Diptera
Plecoptera
Diptera
Plecoptera
Plecoptera
Diptera
Diptera
Diptera
Diptera
Diptera
Plecoptera
Family
Limnephilidae
Perlodidae
Veliidae
Chironomidae
Chironomidae
Chironomidae
Rhyacophilidae
Ephemerellidae
Sialidae
Simuliidae
Chironomidae
Chironomidae
Heptageniidae
Elmidae
Heptageniidae
Gomphidae
Chironomidae
Chloroperlidae
Tabanidae
Taeni opterygi dae
Peltoperlidae
Chironomidae
Chironomidae
Chironomidae
Tipulidae
Chironomidae
Chloroperlidae
Genus
Pycnopsyche
Remenus
Rhagovelia
Rheocricotopus
Rheopelopia
Rheotanytarsus
Rhyacophila
Serratella
Sialis
Simulium
Stempellina
Stempellinella
Stenacron
Stenelmis
Stenonema
Stylogomphus
Subletted
Sweltsa
Tabanus
Taeniopteryx
Tallaperla
Tanytarsus
Thienemanniella
Thienemannimyia
Tipula
Tvetenia
Utaperla
XC95
295
121
>2,030
>3,489
-1,457
>3,489
>1,890
535
>1 1,227
>6,468
644
>927
-782
>9,790
745
>6,468
>2,421
-750
>9,790
260
478
>9,180
>9,790
>6,468
>1,979
>2,613
255
N
44
35
52
559
126
949
415
49
264
1,095
35
309
258
1,232
922
118
182
315
61
30
89
1,232
395
1,345
621
760
47
TV from
reference
locations
10
3
3
11
4
28
57
2
3
26
8
26
15
26
57
1
2
42
1
12
16
64
9
56
36
40
2
D-7
-------
Table D-l. Extirpation concentration and sample size from West Virginia
data set (continued)
Order
Trichoptera
Plecoptera
Diptera
Diptera
Family
Philopotamidae
Perlodidae
Chironomidae
Chironomidae
Genus
Wormaldia
Yugus
Zavrelia
Zavrelimyia
XC95
>1,553
655
413
>2,768
N
79
75
81
244
Nfrom
reference
locations
8
12
6
11
D-8
-------
APPENDIX E
GRAPHS OF OBSERVATION PROBABILITIES
FOR GENERA IN THE WEST VIRGINIA DATA SET
ABSTRACT
The purpose of Appendix E is to help the reader visualize the changes in the occurrence
of each genus in the West Virginia data set as conductivity increases. Each figure depicts a
general additive model (GAM) of the relationship between capture probabilities of a genus and
conductivity. Genera are ordered from the lowest to the highest extirpation concentration (XCgs)
value. Open circles are the probabilities of observing the genus within a range of conductivities.
Circles at zero probability indicate no individuals were found in any sample with those
conductivities. The GAM line (solid line) fitted to the probabilities is for visualization and
dashed lines are 90% confidence bounds. The vertical dotted line marks the XCgs as listed in
Appendix D. Note that, because of differences in sensitivity, different genera respond differently
within the observed range of salinity. For example, Lepidostoma declines, Diploperla has an
optimum, and Cheumatopsyche increases. The fitted lines and confidence bounds were used to
assign qualifiers to the XCgs values in Appendix D.
E-l
-------
-2
"3
5
I
O
9'0
CD
I- S
o
,- "O
oo C
O
O
i i i i r
vo e'o z'o i/o o'o
.CO
&
CD
\- s
Sco
,- "O
oo C
O
O
_ m
oi/o go'o oo'o
c
CD
S
CD
o:
CO
I- S
]
]
]
]
9'0 VO Z'O Z'O
I
\.'0
.
.>
"o
O
O
O'O
-C
t
s-
CO
S
.
.>
"o
O
O
O'l 8'0 9'0 VO Z'O O'O
CD
CO
-J
CO
^1-
co
SO
Sco
.
.>
"o
oo C
O
O
CO
^1-
co
SO
.>
"o
oo C
O
O
9'0
O'O
E-2
9'0
e'o z'o \.'o o'o
-------
-2
"CD
c
w
\- s
9'0
.
o
O
O'O
.CO
.0.
Q
CD
I- S
oo C
O
O
_ m
I I
oi'o go'o oo'o
CD
-C
£
V)
Q
S
£
(
<4
0 4
° <-/>i
0 , * / ,' °
w
- 1
r*E
- $\ -°-
CM (jj
3=
CO >>
- CM ~
- ^ .>
"o
~ oo C
O
O
- £
O'l. 8'0 9'0
O'O
I I I I
•Q 9'0 VQ Z'O O'O
.
_>
"o
O
O
•2
CL
o
5
CD
U 5-9
CO
^t-
co
o
oo C
O
O
oro go'o oo'o
CD
-C
\
O'l
CO
- co
u 5-y
.
.>
"o
oo C
O
O
'0 9'0 VO Z'O O'O
E-3
-------
CD
CO
1
o
c
8
I
I
CD
Q
oz'o
o
oo C
O
O
_ m
oro go'o oo'o
CO
.5
CD
&
D)
CO
\- s
CO !"7*
•- 3 >
oo C
O
O
n i \ \ \
gz'o oz'o gi/o oi/o go'o oo'o
CD
I- S
.
.>
"o
O
O
O3'o gro oro go'o oo'o
CD
5
CL
8
CO
I- S
.
.>
"o
O
O
oz'o gi/o oi/o go'o oo'o
CD
o
CD
CO
- CO
u 5-y
.
.>
"o
oo C
O
O
VO Z'O Z'O TO O'O
c
CD
D)
CD
n i r
gz'o oz'o gr
i i r
oi/o go'o oo'i
CO
CO
CO ^P
o
oo C
o
o
E-4
-------
CO
I
a
CD
I- S
o
c/5
^.
_>%
'>
'•5
c
o
O
9Z'o oz'o 91/0 oi/o 90'o oo'o
_ m
'0 9'0 VO Z'O O'O
CO
c
s
I
2
QQ
w
\- $
m
"o
c
o
O
93'0 03'0 91/0 OI-'O 90'0 OO'O
CO
c:
CD
\
O'l
CO
•*
- CO
\ \
3'0 9'0
.>
"o
oo C
O
O
\ \
ro z'o o'o
E-5
-------
o
-c
a
CD
•*
CO
Soo
.
.>
"o
oo C
O
O
17'0 E'O Z'O I'O O'O
-
e
CD
GO
CO
•*
- CO
.>
"o
oo C
O
O
I I
01/0 90'0 OO'O
O
O
3
S
55
p
E
,0
00
CO
CM
•*
c
o
O
9'0
VQ E'O Z'O I/O O'O
17'0
E'O
3'0
I/O
O'O
-C
t
&
fi
CO
\- s
O
GO
^.
_>%
'>
'•s
c
o
O
o
a
CO
\- s
So
CM ^r
'•s
oo C
O
O
_ m
9'0
fO
Z'O
O'O
93'0 OZ'O 91/0 OI-'O 90'0 OO'O
E-6
-------
o
S
CD
I- S
CO >>
h 3 >
'•5
oo C
O
O
_ m
9'0 fO E'O Z'O I/O O'O
CO
.C
"o
I
-2
GO
CO
\- s
CO ?
- CM ±r
^ >
oo C
o
o
n \ \ \ \ r
9Z'0 03'0 91.'0 Ol'O 90'0 DOT
CO
I
"o
c
o
O
CD
I- S
_ CM U
Soo
h§:t
"o
oo C
O
O
9Z'0 03'0 91.'0 OI-'O 90'0 OO'O
I
CO
c
CO
s
.
.>
"o
O
O
i \ \ \ \ \ r
L'Q 9'0 9'0 fO E'O Z'O I'O O'O
.CO
5
o
o
CO
- CO
CM ^
SGO
CO _£P
"* '^
"o
oo C
O
O
ZVQ Ol'O 80'0 90'0 WO 30'0 OO'O
o
s—*
CD
-C
.0.
Q
CO
- CO
u 5-y
.
.>
"o
oo C
O
O
9Z'0
\ \ \ r
91/0 01/0 90'0 OO'O
E-7
-------
co
8
o
5
I
CD
I- S
o
oo C
O
O
i i i i r
g'o vo e'o z'o i/o o'o
-2
"55
c
s
CD
\- s
oro
go'o
Sco
oo C
O
O
_ m
oo'o
CO
3
I
CD
I- S
.
.>
"o
O
O
CD
5
CD
-C
CO
k S
CM ^J
Sco
§f
"o
oo C
o
O
VO Z'O
Z'O
I'O O'O
9'o fo e'o z'o i/o o'o
CD
c
c
CD
e-
CD
O
CO
- CO
.
.>
"o
oo C
O
O
CD
d
o
o
CO
•*
CO
So
Sco
.
.>
"o
oo C
O
O
31.'0
80'0
WO
OO'O
01 8'0 9'0 fO 3'0 O'O
E-8
-------
CD
O
CD
CO
I- S
I I I
go'o wo eo'o
Sco
oo C
O
O
i r
I-O'O OO'O
c
§
CO
c
•2
CO
9'0
CO
s
Sco
oo C
O
O
_ m
O'O
Q
\
\
\
\
\
\
§f
"o
oo C
O
O
9'0 fO E'O Z'O I/O O'O
CO
42
I
CO
CO
s
E
CO
^.
>%
-^
;>
"o
c
o
O
O'l. 8'0 9'0 VQ Z'O O'O
c
o
CD
.O
O
O
CO
•*
CO
So
.>
"o
oo C
O
O
CO
5
CD
C
O
c
£
CO
CO
- CO
SO
.>
"o
oo C
O
O
oro go'o
OO'O
O'l.
8'0 9'0 VQ Z'O O'O
E-9
-------
CO
t
o
Q
CD
s
Sco
oo C
O
O
v)
"53
co
?
% •>
'13
oo C
O
O
_ m
O'l Q'O 9'Q VQ Z'O O'O
90'0
WO
30'0
OO'O
i
O
f
CO
I- S
.
.>
"o
O
O
\ \ I I I I
o'o go'o wo eo'o zo'o wo oo'o
3
.c
"53
I
CD
S—*
CO
\
O'l
CO
S
8'0
9'0
fO
3'0
E
CO
^.
>%
-^
'>
"o
c
o
O
O'O
S
CD
.C
-C
I
CO
^t-
co
.
.>
"o
oo C
O
O
I
|
"CD
.>
"o
oo C
o
O
O'l
8'0 9'0
VQ Z'O
O'O
oro
go'o
OO'O
E-10
-------
I
C
CD
O
ZVQ
O I
° o'
CO
I- S
i i i r
80'0 WO
O
oo C
O
O
OO'O
CO
O
£
CD
CD
-C
O
2
fi
CO
I- S
o
.
O
O
_ m
i i i i i i r
9'0 9'0 VQ E'O Z'O I/O O'O
I
CD
8
CD
CO
s
.
.>
"o
O
O
9Z'0 03'0 91.'0 OI-'O 90'0 OO'O
CD
t
O
CO
I- S
h§:f
"o
oo C
O
O
n i i i i r
9Z'0 03'0 91/0 01/0 90'0 OO'O
CO
CD
5:
CO
•*
- CO
CM ^-'
Sc75
CO ^P
"* '^
"o
oo C
O
O
CD
CO
o
c
CO
^t-
co
.
.>
"o
oo C
O
O
E'O Z'O
I/O O'O
VO Z'O Z'O I/O
O'O
E-ll
-------
CO
3
CD
p
o
oo C
O
O
90'0 WO ZO'O OO'O
-
a
I \ \
9'0 VO E'O
CD
^t-
CO
E
CO
^_
>%
-^
l>
"o
c
o
O
Z'O I-'O O'O
i
CD
-Q
S
CD
O
SO
Sco
co-^
^ >
'•s
oo C
O
O
c
S
i
3
CD
O
O
Q'O 9'0 VO E'O 3'0 l-'O O'O
9'0 17'0 E'O Z'O I-'O O'O
CD
1
O
CO
CO
S
SO
SCO
CO ^*
^ >
'•s
oo C
O
O
CD
I
C
_ m
co
s
oo C
O
O
9'0 VQ E'O 3'0 I-'O O'O
O'l. 8'0 9'0 VO Z'O O'O
E-12
-------
CO
I
CD
CD
-C
o:
CD
^t-
CO
CO
CM
•*
_O
co
^_
>%
-^
\>
"o
c
o
O
9'0 VO
\ \ \ r
e'o z'o i-'o O'O
CO
5
o
CO
o
CO
CO
o
CD
9'0
VQ E'O
Z'O
CO
^t-
co
E
CO
^_
>%
-^
\>
"o
c
o
O
I'O O'O
co
CO
D)
o
O
o
o
E
,0
>>
'>
'•s
c
o
O
CO
;5
CO
O
O
80'0 90'0 WO 30'0
OO'O
9'0
E'O Z'O I'O O'O
CD
CO
QQ
CO
\- s
oo C
O
O
C
.s
~J
o , -
CO
S
o
co
^.
_>%
'>
'•5
c
o
O
_ m
8'0 9'0
O'O
9'0 fO E'O 3'0 l-'O O'O
E-13
-------
I
ZI/0
\ \ \
QO'O WO
CD
s
O
oo C
O
O
OO'O
amjdeo
CD
-2
•3
CD
s
9'0
fO
Z'O
o
c/5
^.
_>%
'>
'•o1
c
o
O
_ m
O'O
.o
1
CD
«
\- S
E
c/5
>
"o
c
o
O
CD
•B CD
CO
s
CM ^J
3c75
§f
"o
oo C
O
O
gi/o
oro
go'o
OO'O
oro
go'o
OO'O
CD
5
CD
1
9'0 VO
E'O
Z'Q
i
ro
CO
^t-
co
CM ^
^(75
CO _£P
"* '^
"o
oo C
O
O
O'O
CO
CD
D)
CO
CO
E
(75
^_
&
_>
"o
c
o
O
80'0 90'0 WO 30'0
OO'O
E-14
-------
CD
-
o
i
O'l 8'0 9'0 VO
Z'O
o
c/5
^.
_>%
'>
'•5
c
o
O
O'O
fTT
5
CD
c
o
c
b
o
0
o
__ ________ —O. ______
0
o
J
o'
/{
0 C
° :
1
1
0
1 1 1 1 1
9'0 VQ E'O Z'O I/O
1 'f
1 1
/ 1
/ 1
/ 1
/;.
<$]' "*
i?('%
•>°o
i /' o
J»
7 '
' °
i ,
1 .
•
i
i j
i i
0
i
>
i
I
i
0
w
- S
^~
_ S
C\l
CM
%
'>
'•g
•Q
C
o
O
amjdeo
CO
CO
E
(75
^.
>%
-^
'_>
"o
c
o
O
Zl'O Ol'O 80'0 90'0 WO ZO'O OO'O
.CO
t
1
1
1
t
0 t
••• °r
0 01
0°°
o 0
o
o
0 J
o o <
o/°
1
0 ° I
' S
1
1
1
1
t- -8
'o °
1 0
o
0
p
° 0
oa
b
1 O
1 0
'• 8
i 8
.
.>
"o
O
O
\ \ \ \ r
93'0 03'0 91/0 01'0 90'0 OO'O
.CO
Sw
CO
CO
80'0 90'0
WO ZO'O
CO
^l-
co
CO
CM
•*
_O
c/5
^.
^
.>
"o
c
o
O
OO'O
CO
CO
^t-
co
.>
"o
C
O
O
O'O
E-15
-------
CD
-J
CO
I- S
O
oo C
O
O
CO
CD
\- S
««
CO >>
3 >
'•o1
oo C
O
O
_ m
8'0
9'0
fO
Z'O
O'O
oz'o gi/o oi/o go'o oo'o
c
co
\
\
\
CO
S
oro go'o oo'o
.
.>
"o
O
O
I
o
D)
C
c§
CO
\- s
.>
"o
oo C
O
O
oz'o
\ \
oi/o go'o oo'o
.co
1
o
D)
CO
§
CO
- CO
.>
"o
oo C
O
O
O3'o gro oro go'o oo'o
CO
.C
-G
CD
C
D)
CO
- CO
u 5-y
.>
"o
oo C
O
O
oe'o
O3'o
oi/o go'o oo'o
E-16
-------
CD
§
I
CO
oo C
O
O
\ \ \ \ \
9Z'o oz'o 91/0 oi'o 90'o oo'o
CO
CO
9'0 VO E'O Z'O
CO
s
o
CO
^.
_>%
'>
'•o1
C
o
O
_ m
I'O O'O
CO
^
.CO
o
.0
CO
O
N
\
\
N
\
O ^
0 °\°
n
/
f
l_
,'°°
1 ° 0
\ '. o
CO
\- S
o
oo C
O
O
£
O
CD
CO
C
CD
CO
CO
s
o
O
O
03'0 91/0 OI-'O
OO'O
oro go'o oo'o
C
o
D)
O
Q-
•s
.g
T*J
_
0 ^
.
\t
-A
O
0 *
\
J
1
°L
t
0 '
0 o ,
_ /
o
)3'0 91/0 01/0 90'0
1
1
I
1
'
T ^"
• 8
1 0
1 1
' o
10
? 8
10
T 8
i R
' 8
1 §
:e
001
.>
"o
oo C
O
O
CO
I
CD
I
OZ'O
91.'0
CO
•*
CO
CO
CM
•*
_o
c/5
^.
^
.>
"o
C
o
O
OI-'O
90'0
OO'O
E-17
-------
CD
S*J
I
1 ' 8
1 ' 8
1 ' D
•0 g
' §
_ 0 4.J, °.
° 0 ' 1 ' °
0° ' (,'
at' °
° 8$
4l0 0
•t • °
® ' T o
0° ° ' '
° "
"o
O
O
O'O
CD
CD
-J
\
\
CO
s
Sco
.
.>
"o
O
O
03'0 91.'0 01/0 90'0 OO'O
CD
1
.§
CO
•*
CO
CO
CM
•*
_O
CO
^.
&
_>
"o
C
o
O
.CD
S
CD
C
§
CO
^t-
co
CO
CM
•*
_O
CO
^.
&
_>
"o
C
o
O
01/0 80'0 90'0 WO 30'0 OO'O
9'0
O'O
E-18
-------
CD
CO
'5
o
oo C
O
O
L'Q 9'0 9'0 fO E'O Z'O I/O O'C
i
"o
g
CD
s
oo C
O
O
_ m
n i i i i
9'0 fO E'O Z'O I/O O'O
.CO
-c
.0
C:
s
•8
O
.
I
I
i
.o____ _____ __________ +
° 4
&
J
j
c
o ^
° ,
1
1
1
1
1
p
j'
e
8
o
8
s
1
:
|
CO
- S
^
^?
_ >
^ .>
"o
-^
~ m d
o
O
- ¥
9'0 fO E'O 3'0 I-'O O'C
4S
CO
CD
I- S
CM ^J
So5
§f
"o
oo C
O
O
9'0 9'0 fO E'O 3'0 I/O O'O
CO
|
"^
CO
N
CO
- CO
_ CM
CO ^P
"1
"o
oo C
O
O
i i i i i
9Z'0 OZ'O 91.'0 OI-'O 90'0 OO'O
co
^>
o
CD
C
o
fc
O
9'0
°cP
CD
^1-
CO
.
.>
"o
oo C
O
O
I I I I
VO E'O Z'O I'O O'O
E-19
-------
CD
CO
CD
O
O
O
oo C
O
O
9'0 VQ £'0 Z'O I'O O'O
I
I
1
I
1
n J
0 1
0 1
1
%
O 1
O ,
O
1
0
o
0
°(
o
0 0
8
8
0
o
)
•o
0°
o
P
•
P0
' R
1 °
1 ft
18
CO
s
o
C
O
O
_ m
\
9'0 9'0 VQ £'0 Z'O I'O O'O
CO
CD
o
I
CO <7*
i- ^;|
"o
oo C
o
o
I I I I I I
oe'o oz'o oi/o oo'o
CO
CD
c
CD
Hw
§
9'0
fO
Z'O
CO
s
E
c/5
^.
>%
-^
;>
"o
C
o
O
O'O
CD
"o
f
CD
Q
OS'O OZ'O
o
T3
c
o
o
Ol'O OO'O
CO
S
5
o
c
.8
-§
O
CO
^t-
co
SP
.>
"o
oo C
O
O
I \ \ \ \
9'0 VO Z'O Z'O I/O O'O
E-20
-------
c
CD
CD
O
CO
s
O
CO >>
3 >
'•5
oo C
O
O
O'l Q'O 9'Q VQ Z'Q O'O
2
CD
O
O
.C
3
i
i
.__ i
o '
1
1
I
o1
o
0
c
I
'
° $
4
0 1
1
1
1 1 1 1 1
O'l 8'0 9'Q VQ Z'O (
_
H
fi
Q
fl"
10
18
1
'8
P
^
o
_j
to
8
!„
€
8
«
D'O
CO
- S
^~
^~
^"^^s
^ E
— JN ^^
CM C/5
n.
**~*
CO ^*
^ !>
'-l_l
^J
-Q
- 5 c
o
O
_ m
CO
o
CD
CO
s
CM ^
3c75
§f
"o
oo C
O
O
9'0 9'0 VQ Z'O Z'O I/O O'O
CD
O
CD
CD
O
CO
s
CO <7*
"1
"o
oo C
O
O
\ \ \ \ \
Q'l 8'0 9'0 VQ Z'O O'O
CO
O
o
'5
o
CD
CO
^t-
co
.>
"o
oo C
O
O
I I I I
VO Z'O Z'O I'O O'O
CD
.S
-C
O
\
O'l
CO
^t-
co
SP
.
.>
"o
oo C
O
O
'0 9'0 VO Z'O O'O
E-21
-------
CO
3
5
CD
8
.O
"CD
I
I
CD
So
Sco
8 =
o
oo C
O
O
\ \ \ \ \ r
n Q'O 9'0 VQ Z'O O'O
CO
4? O
CD
CD
-C
O
CO
\- s
oo C
O
O
_ m
SZ'O OZ'O Sl'Q Ol'O 90'0 OO'O
CO
s
CD
CD
CD
CO
s
Sco
.
.>
"o
O
O
\ \ \ \ \ r
O'l Q'O 9'0 VQ Z'O O'O
CD
§
£
CO
s
Sco
.
.>
"o
O
O
O'l. 8'0 9'0 VO Z'O O'O
CD
CO
CD
S
CD
Q
CO
^t-
co
E
CO
^.
^
.>
"o
C
o
O
CO
s
I
C
1
£
CO
^t-
co
SO
SCO
.>
"o
oo C
O
O
O'l. 8'0 9'0 VO Z'O O'O
O'l 8'0 9'0 fO Z'O O'O
E-22
-------
CO
•8
o
O'l 8'0 9'0 VO Z'O
CD
s
o
c/5
^.
_>%
'>
'•o1
c
o
O
O'O
CO
"o
o
c
CD
CO
-C
\
\
\
\
\
co
s
C
O
O
_ m
O'l. 8'0 9'0 VO Z'O O'O
CO
I
§-
c
.s
O3'o
CO
s
E
c/5
^.
>%
-^
'_>
"o
c
o
O
oro go'o oo'o
£
o
CD
CO
\
O'l
'0 9'0 VO Z'O
CO
s
.
.>
"o
O
O
O'O
c:
o
5
CO
^t-
co
.
.>
"o
oo C
O
O
CD
CD
O
CO
^t-
co
So
.>
"o
oo C
O
O
oe'o
oz'o
oi'o
OO'O
'0 9'0
Z'O O'O
E-23
-------
CO
I
5
c
co
S
CD
C
CD
\
\
\
\
8'0 9'0 VO Z'O
w
\- s
u ^-°.
Sc75
CO >>
h 3 >
'•s
oo C
O
O
_ m
O'O
CO
I
QQ
CO
\- s
oo C
O
O
oe'o oz'o oi/o oo'o
CO
-c
I
o
D)
O
ct
1J •' B
CD
s
.
.>
"o
O
O
CD
i
f
CO
s
E
c/5
^.
>%
-^
;>
"o
C
o
O
'0 9'0 VO Z'O
O'O
O'l
8'0 9'0 VO Z'O
O'O
I
"3
.g
\
O'l
'0 9'0 VO Z'O
CO
•*
- CO
.
.>
"o
oo C
O
O
O'O
o
c
2
o
Q
CO
^t-
co
.>
"o
oo C
O
O
O'l 8'0 9'0
O'O
E-24
-------
CD
I
t
CD
I
CD
g
i-;-
O'l 8'0 9'0 VO
Z'O
CO
I- S
O
oo C
O
O
O'O
CD
5
5
CD
$
:c
o t
\
o 3,
O |
0 1
1
1
) t
'
'•
b
D
8
n i i i i
O'l Q'O 9'Q VQ Z'O
CD
s
O
(7)
^.
_>%
'>
'•o1
c
o
O
_ m
O'O
CO
s
C
CD
-C
CL
CD
CD
s
E
c/5
^.
>%
-^
|>
"o
c
o
o
O'l 8'0 9'0
VO
Z'O
O'O
CD
8
CD
5
\
O'l
8'0
9'0
VO
CO
s
E
c/5
^.
>%
-^
;>
"o
c
o
o
i
Z'O O'O
.CD
-C
I
•Q
S
Q
CD
^1-
CO
.>
"o
oo C
O
O
CO
0 '
•o 1 —
O 1
1
° 0 '
0 i
0 S
I,1.
o
0 (
0 „
0
i
1
0 ° ''
0 1
8 ° ;
O '
v
§
1 0
*
f Q
1
• °0
1
Is
'o
t
10 0
— _
1 ®
10
CD
- CD
"*"
CM
CM
- CN
•^~
oo
- £
.>
"o
O
O
O'l.
'0 9'0 P'O Z'O O'O
8'0 9'0
Z'O
O'O
E-25
-------
-2
"o
\
01
'0 9'Q VQ ZO
CO
s
Sco
CO >>
3 >
'•o1
oo C
O
O
O'O
o
5
c
co
5
CD
C
CD
CO
s
>
'•o1
C
O
O
_ m
'0 9'0 VO Z'O
O'O
CO
I
CD
CO
.O
n •
« _
0
§
o
0 °
0
0
°oo
o
1
1
e
o
<
8
0
\ \
01 8'0
/ /
/ / *
/ '° 8
//
/ /0
/o8
I
o
t
T \
s •
' A V °
•V'*.-
1 1 1 1
9'0 P'O Z'O O'O
CO
- i
,*—+.
^ E
^.
s.^^
CO <^
- 3;!
"o
~ 5 c
o
O
- £
CO
c
CO
-Q
SO
CO <7*
h "1
"o
oo C
O
O
01 8'0 9'0 VO Z'O O'O
.CO
S
,
* , -
1
0 !
t
#8;
°\
0«
Q 1
%°l
£
° 1°
1
1 1 1 1
01 8'0 9'0 VO
L ' 0
• T f)'
' a
§
1 0
1 0
'
1
lo
*
D
' 0
1
1
1
9
' ° 0
i §
1 l
ZO O'O
CO
- CO
^~
T~
- s-^
^_
^ !>
"o
- 5 c
o
O
_ m
CO
I
CD
C
-2
CO
CO
^t-
co
SO
Sco
.>
"o
oo C
O
O
01
'0 9'0 fO 3'0 O'O
E-26
-------
CO
I
O'l
CD
s
o
CO >>
3 >
'•5
oo C
O
O
'0 9'Q VO Z'O O'O
CO
t
CO
CD
•Q
CO
-Q
\
\
\
\
O'l. 8'0 9'0 VO
Z'O
CD
s
oo C
O
O
_ m
O'O
•8-
8
o
o
f
8
O
O'l
'0 9'0
CO
s
E
(75
^.
>%
-^
;>
"o
c
o
O
Z'O O'O
.CO
"co
CO
s
CM ^-'
3c75
§f
"o
oo C
O
O
Z'O 9'0 9'0 VO Z'O Z'O I'O O'O
o
O
«-
o
o aU
CO
^t-
co
.
.>
"o
oo C
O
O
CO
o
!
CO
CO
H3|
"o
oo C
o
o
O'l 8'0 9'0 VO Z'O O'O
g'o vo e'o z'o vo 00
E-27
-------
CO
o
g
/ 1
c^P
•
oo C
O
O
O'l Q'O 9'Q VQ Z'Q
O'O
CO
3
•§
1
\ \ \ \ \
O'l 8'0 9'0 VQ Z'Q Q'Q
amjdeo
CO
s
S
o
c
8
g
CD
s
.
.>
"o
O
O
Q'l 8'0 9'0 VQ Z'Q Q'Q
CO
CD
.0.
•D
C
•2
§
Q
\
\
O'l 8'0
9'0
CD
s
So
.
.>
"o
O
O
I
Z'Q Q'Q
CD
-C
CO
^t-
co
So
.
.>
"o
oo C
O
O
O'l 8'0 9'0 VQ Z'Q Q'Q
CO
I
CD
I
CD
O
\
O'l
CO
CO
o
.>
"o
C
o
O
'0 9'0 VO Z'O
O'O
E-28
-------
CO
5
0
c
.8
_2
0
.0
Q)
CO
0.
"° ~ ~ * .. ^\ ""»
^Ov
o \ \ *
°° %\
V
<
o
0
1 1 1 1 1
O'L Q'O 9'0 VQ Z'Q C
8
o
0
>
>
?
i
>
)
I
|
!
i
•Q
CD
- CD
\±J
~ CM
CM
- OJ
T;:'"
~ 00
- !?
73
C
o
O
E-29
-------
-------
APPENDIX F
GRAPHS OF CUMULATIVE FREQUENCY DISTRIBUTIONS
FOR GENERA IN THE WEST VIRGINIA DATA SET
ABSTRACT
The purpose of Appendix F is to help the reader visualize the changes in the occurrence
of each genus in the West Virginia data set as conductivity increases and understand how the
extirpation concentration (XCgs) values are derived. Each plot contains the weighted cumulative
distribution function (CDF) for the occurrence of a genus with respect to conductivity. For each
genus, the points in the CDF represent the weighted proportions of occurrences of the genus in
samples less than the indicated conductivity value (uS/cm), calculated using Equation 1. In a
CDF, genera that are affected by increasing conductivity (e.g., Drunelld) show a steep slope and
asymptote well below the maximum conductivity, whereas genera unaffected by increasing
conductivity (e.g., Nigronid) have a steady increase over the entire range of measured exposure
and do not reach a perceptible asymptote. The 95* centile is found at the intersection of the
dashed horizontal line with the CDF. The conductivity at the 95th centile is the XCgs value and is
found at the intersection of the vertical line and the x-axis.
F-l
-------
CO
- CO
CM
CM
CM
CO
CM
00
\ \ \
O'l 8'0 9'0
\ r
Z'O O'O
,0
0)
Q.
re
o
o
CO
^r
- CO
CM
CM
CM
\ \ \
O'l 8'0 9'0
\
Z'O
00
,0
CO >,
3 1
"o
^
~o
O
O
O'O
x =>
x =>
(A
3
C
0)
E
0)
CO
^r
co
CM
CM
CM
CO
CM
00
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 fr'O Z'O O'O
>,
\>
"o
O
O
\ \ \ \ \ r
O'l 8'0 9'0 fr'O Z'O O'O
x =>
x => uo|}jodojd
\ \ \
O'l. 8'0 9'0
co
^i-
co
CM
CM
CM
CO
CM
00
i r
Z'O O'O
O
O
CO
^1-
- CO
CM
CM
CM
CO >^
S '5
00
\ \ i i i r
O'l. 8'0 9'0 VQ Z'O O'O
O
O
x => uo|}jodojd
F-2
x => uo|}jodojd
-------
c
3
CO
^r
- co
CM
CM
CM
CO
CM
^r
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
,0
w
o
^
T3
c
o
o
re
CO
^r
co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 fr'O Z'O O'O
x=>
,0
>?
'>
"o
^
T3
O
O
co
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x=>
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 fr'O Z'O O'O
"o
^
73
C
O
O
x => uo|}jodojd
Q.
O
'c
0)
co
^r
co
CM
CM
CM
00
O'l. 8'0 9'0 VQ Z'O O'O
x => uo|}jodojd
E _
O (w
t f
co >.
T3
O
O
Q.
UJ
co
^1-
- CO
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
x => uo|}jodojd
T3
O
O
F-3
-------
CO
- CO
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
CO
^r
- co
CM
CM
CM
\ \ \
O'l 8'0 9'0
x=>
\
Z'O
00
O'O
,0
CO >*
3 1
"o
^
T3
O
O
0)
x
CO
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x=>
,0
O
O
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 fr'O Z'O O'O
O
O
x => uo|}jodojd
co
^i-
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 fr'O Z'O O'O
x => uo|}jodojd
>,
i>
"o
o
O
ro
'c
0)
D)
4S
Q.
0)
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
x => uo|}jodojd
&
.>
"o
O
O
F-4
-------
re
I
re
N
CO
- CO
CM
CM
CM
CO
CM
00
\ \ \
O'l 8'0 9'0
\ r
Z'O O'O
,0
re
o
o
CO
^r
- CO
CM
CM
CM
\ \ \
O'l 8'0 9'0
\
Z'O
00
,0
CO >,
3 1
"o
^
~o
O
O
O'O
x =>
x =>
CQ
CO
^r
co
CM
CM
CM
CO
CM
00
o
o
CO
^r
- CO
CM
CM
CM
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 fr'O Z'O O'O
>,
\>
"o
O
O
\ \ \ \ \ r
O'l 8'0 9'0 fr'O 3'0 O'O
x =>
x => uo|}jodojd
CO
^r
co
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 V'O Z'O O'O
,0
w
^
.>
"o
O
O
0)
Q.
£
Q.
re
CO
^1-
- CO
CM
CM
CM
CO >^
S '5
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
O
O
x => uo|}jodojd
F-5
x => uo|}jodojd
-------
CO
- CO
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
,0
"o
^
73
C
O
O
§
0)
(O
\ \ \
O'l 8'0 9'0
x=>
\
Z'O
CO
^r
- co
CM
CM
CM
CO
CM
00
O'O
>?
'>
"o
O
O
0)
Q.
O'l. 8'0 9'0 V'O Z'O O'O
x=>
00
,0
O
O
x => uo|}jodojd
co
^1-
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 fr'O 3'0 O'O
x => uo|}jodojd
O
O
uo|}jodojd
&
.>
"o
O
O
F-6
-------
(A
4->
0)
0)
CO
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
,0
ro
o.
E
0)
O
O
CO
- CO
CM
CM
CM
\ \ \
O'l 8'0 9'0
x=>
\
Z'O
00
O'O
,0
CO >*
9 '5
O
O
CO
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x=>
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 fr'O Z'O O'O
O
O
x => uo|}jodojd
re
'c
Q.
ro
o
o
CO
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 fr'O Z'O O'O
x => uo|}jodojd
^ 3
2? "5
i f
=! Q
c
o
o
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
x => uo|}jodojd
&
.>
"o
O
O
F-7
-------
CO
- CO
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
CO
^r
- CO
CM
CM
CM
\ \ \
O'l 8'0 9'0
x=>
\
Z'O
00
O'O
,0
CO >*
9 '5
O
O
CO
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x=>
>,
\>
"o
^
T3
O
O
ro
0)
E
0)
^
Q.
LU
CO
^1-
- CO
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 VQ Z'O O'O
"o
^
T3
C
O
O
x => uo|}jodojd
re
c
c
0)
Q.
0)
O
CO
^r
co
CM
CM
CM
--' 0)
§ I &
00
O'l. 8'0 9'0 VQ Z'O O'O
x => uo|}jodojd
o
^
T3
o
O
0)
Q.
CO
^1-
- CO
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 VQ Z'O O'O
x => uo|}jodojd
&
.>
"o
T3
O
O
F-8
-------
2
0)
o
CO
- CO
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
CO
^r
- co
CM
CM
CM
\ \ \
O'l 8'0 9'0
x=>
\
Z'O
00
O'O
,0
CO >*
9 '5
O
O
re
x
CO
^1-
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x=>
re
±!
0)
(O
O
O
CO
^1-
- CO
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 fr'O Z'O O'O
,0
O
O
x => uo|}jodojd
c
o
0)
_0
u
p
CO
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 VQ Z'O O'O
x => uo|}jodojd
E
,0
O
O
re
E
0)
c
o
c
^
(O
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
x => uo|}jodojd
&
.>
"o
O
O
F-9
-------
(A
0)
Q.
^
O
Q
co
^r
co
CM
CM
CM
CO
CM
00
,0
o
o
(A
(A
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ \ \ \ r
O'l 8'0 9'0 fr'O Z'O O'O
,0
>?
'>
"o
^
T3
O
O
\ \ \ \ \ r
O'l 8'0 9'0 fr'O Z'O O'O
x =>
x =>
\ \ \
O'l. 8'0 9'0
co
^r
co
CM
CM
CM
CO
CM
00
E
o
&
.>
"o
T3
O
O
re
0)
Q.
E
0)
4->
(O
co
^r
- co
CM
CM
CM
CO
CM
00
i r
Z'O O'O
,0
>,
^>
"o
O
O
\ \ i i i r
O'l. 8'0 9'0 fr'O 3'0 O'O
x =>
x => uo|}jodojd
re
3
E
0)
Q.
E
CO
^r
co
CM
CM
CM
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 V'O Z'O O'O
o
W
.>
"o
o
O
X.
0)
75
CO
^r
- co
CM
CM
CM
CO >*
S '5
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
o
O
x => uo|}jodojd
F-10
x => uo|}jodojd
-------
.
2
"c
0)
O
CO
- CO
CM
CM
CM
CO
- CM
00
\
O'l
\
8'0
\
9'0
\
fr'O
\
Z'O
r
O'O
.>
"o
T3
O
O
(A
O
O
"5
re
.c
o
2
re
Q.
CO
^r
- co
CM
CM
CM
00
\
O'l
\
8'0
\
9'0
\
3'0
,0
co >,
3 1
"o
^
T3
o
o
O'O
x =>
x =>
E
3
re
o
o
re
CO
^r
co
CM
CM
CM
CO
CM
00
o
o
CO
^r
- CO
CM
CM
CM
CO
CM
00
,0
>,
^>
"o
o
o
\ \ \
O'l. 8'0 9'0
i r
Z'O O'O
\ \ \ \ \ \
O'l 8'0 9'0 fr'O Z'O O'O
x =>
x => uo|}jodojd
(A
3
re
CO
^1-
co
CM
CM
CM
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 V'O Z'O O'O
O
O
CO
^r
- co
CM
CM
CM
CO >^
S '5
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
O
O
x => uo|}jodojd
x => uo|}jodojd
F-ll
-------
(A
3
0)
u
CO
- CO
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
,0
o
^
73
C
O
O
re
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ \
O'l 8'0 9'0
x=>
\
Z'O
O'O
>?
'>
"o
O
O
co
^i-
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x=>
.> ^
O
O
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 fr'O Z'O O'O
O
O
x => uo|}jodojd
re
CO
^r
co
CM
CM
00
O'l. 8'0 9'0 fr'O Z'O O'O
x => uo|}jodojd
o
w .55
CO >? i
3 '5 C
O
O
CO
^1-
- CO
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
x => uo|}jodojd
E
,0
^
.>
"o
o
O
F-12
-------
CO
- CO
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
,0
O
O
CO
^r
- co
CM
CM
CM
\ \ \
O'l 8'0 9'0
x=>
\
Z'O
00
O'O
,0
CO >*
9 '5
O
O
0)
4->
(A
ro
D)
O
O
co
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x=>
"o
^
73
C
O
O
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 fr'O Z'O O'O
"o
^
T3
C
O
O
x => uo|}jodojd
0)
re
CD
O'l. 8'0 9'0 fr'O 3'0 O'O
x => uo|}jodojd
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
x => uo|}jodojd
00
&
.>
"o
T3
O
O
F-13
-------
E
3
jo
'w
Q.
CO
- CO
CM
CO
CM
00
O'l. 8'0 9'0 VQ Z'O O'O
x => uo|}jodojd
* £
i i
~U ^3
o LU
O
\ \ I
O'l. 8'0 9'0
x=>
\
Z'O
CO
^r
- CO
CM
CM
CM
00
O'O
,0
co >,
3 1
"o
^
T3
o
O
re
0)
!^
re
re
Q.
CO
- CO
_ CM
^ CM
CM
CO >.
h CM -^
00
O'l. 8'0 9'0 fr'O 3'0 O'O
x=>
O
O
re
re
\ \ I
O'l. 8'0 9'0
x=>
CO
^r
- co
CM
CM
CM
CO
CM
00 Q
i r
Z'O O'O
re
c
0)
0)
X
CO
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 fr'O 3'0 O'O
x => uo|}jodojd
O
O
re
*j
(A
re
D)
re
Q.
co
^1-
- CO
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 VQ Z'O O'O
x => uo|}jodojd
&
.>
"o
O
O
F-14
-------
re
CO
- CO
CM
CM
CM
CO
CM
00
\ \ \
O'l 8'0 9'0
\ r
Z'O O'O
,0
o
o
re
3
0)
c
o
I
o
o
CO
- CO
CM
CM
CM
\ \ \
O'l 8'0 9'0
\
Z'O
00
,0
CO >,
3 1
"o
^
~o
o
o
O'O
x =>
x =>
(A
3
O
>
I
O
re
\ \ \
O'l. 8'0 9'0
CO
^r
co
CM
CM
CM
CO
CM
00
,0
"o
^
T3
c
o
o
re
CO
^r
- CO
CM
CM
CM
CO
CM
00
i r
Z'O O'O
,0
>,
\>
"o
^
~o
o
o
\ \ \ \ \ r
O'l 8'0 9'0 fr'O 3'0 O'O
x =>
x => uo|}jodojd
re
"Jo
re
S
o
Q.
CO
^1-
co
CM
CM
CM
CO
CM
00
CO
^1-
- CO
CM
CM
CM
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 V'O Z'O O'O
T3
O
O
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
x => uo|}jodojd
F-15
x => uo|}jodojd
-------
re
0)
CO
- CO
CM
CM
CM
CO
CM
00
\ \ \
O'l 8'0 9'0
\ r
Z'O O'O
,0
"o
^
73
c
o
o
(A
3
Q.
_0
o
CO
^r
- CO
CM
CM
CM
n i i
O'l. 8'0 9'0
i
3'0
00
,0
CO >,
3 1
"o
^
~o
O
O
O'O
x =>
x =>
(A
3
^
4->
(B
\ \ \
O'l. 8'0 9'0
CO
^1-
co
CM
CM
CM
CO
CM
00
CO
^i-
- co
CM
CM
CM
CO
CM
00
i r
Z'O O'O
>,
\>
"o
^
~o
o
o
\ \ i i i r
O'l. 8'0 9'0 fr'O 3'0 O'O
x =>
x => uo|}jodojd
co
^r
- co
CM
CM
CM
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 VQ Z'O O'O
>,
^>
"o
^
-a
o
O
\ \ \ \ \ \
O'l. 8'0 9'0 ¥0 Z'O O'O
x => uo|}jodojd
x => uo|}jodojd
F-16
-------
CO
- CO
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
,0
ro
0)
= .Q
^ =
•c (O
o
O
CO
^r
- CO
CM
CM
CM
\ \ \
O'l 8'0 9'0
x=>
\
Z'O
00
O'O
,0
co >*
S '5
O
O
CO
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x=>
O
O
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 fr'O Z'O O'O
"o
^
T3
C
O
O
x => uo|}jodojd
co
^i-
co
CM
co
CM
00
O'l. 8'0 9'0 fr'O Z'O O'O
x => uo|}jodojd
T3
o
o
Q.
£
0)
Q.
O
0)
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
x => uo|}jodojd
&
.>
"o
T3
O
O
F-17
-------
re
'c
0)
"5
CO
- CO
CM
CM
CM
CO
CM
00
\ \ \
O'l 8'0 9'0
\ r
Z'O O'O
,0
o
o
CO
^r
CO
CM
CM
CM
CO
CM
00
\ \ \ \ \ r
O'l 8'0 9'0 fr'O 3'0 O'O
,0
>?
'>
"o
^
T3
O
O
x =>
x =>
\ \ \
O'l. 8'0 9'0
CO
^r
co
CM
CM
CM
CO
CM
00
i r
3'0 O'O
ro
'S2
re
O
O
CO
^1-
- CO
CM
CM
CM
00
\ \ \ \ \ r
O'l 8'0 9'0 fr'O 3'0 O'O
,0
CO >*
S '5
O
O
x =>
x => uo|}jodojd
re
"55
CO
^1-
co
CM
CM
CM
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 V'O 3'0 O'O
O
O
re
'i_
3
0)
2
o
CO
^1-
- co
CM
CM
CM
CO >*
S '5
00
\ \ i i i r
O'l. 8'0 9'0 V'O 3'0 O'O
O
O
x => uo|}jodojd
x => uo|}jodojd
F-18
-------
(A
_3
'c
O
\ \ \
O'l 8'0 9'0
CO
- CO
CM
CM
CM
CO
CM
00
\ r
Z'O O'O
,0
"o
^
T3
c
o
o
re
I
0)
.c
o
\ \ \
O'l 8'0 9'0
\
Z'O
CO
^r
- CO
CM
CM
CM
00
,0
CO >,
3 1
"o
^
~o
O
O
O'O
x =>
x =>
re
o
o
O
\ \ \
O'l. 8'0 9'0
CO
^r
co
CM
CM
CM
CO
CM
00
i r
Z'O O'O
o
o
re
4->
(A
0)
Q.
CO
^1-
- CO
CM
CM
CM
CO
CM
00
\ \ \ \ \ r
O'l 8'0 9'0 fr'O Z'O O'O
>,
\>
"o
^
~o
o
o
x =>
x => uo|}jodojd
re
re
N
CO
^r
co
CM
CM
CM
CO
CM
00
T3
O
O
(A
0)
ts
0)
O
o
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 V'O Z'O O'O
T3
O
O
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
x => uo|}jodojd
F-19
x => uo|}jodojd
-------
(A
3
b
_re
o
p
CO
- CO
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
,0
&
.>
"o
T3
O
O
(A
3
S2
re
">
re
4->
re
re
Q.
\ \ \
O'l 8'0 9'0
x=>
\
Z'O
CO
^r
- CO
CM
CM
CM
00
O'O
,0
CO >,
9 '5
O
O
(A
3
Q.
0)
O
O
CO
^1-
co
CM
CM
CM
co
O'l. 8'0 9'0 V'O Z'O O'O
x=>
E «
_O fl)
w Q.
-3 ^
^ C
° i
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 fr'O Z'O O'O
O
O
x => uo|}jodojd
re
_a>
0)
(A
re
Q
\ \ \
O'l 8'0 9'0
co
^r
co
CM
CM
CM
CO
CM
00
O
O
(A
3
E
o
c
p
o
p
O
Z'O O'O
\
O'l
\ I
8'0 9'0
\
co
^r
- co
CM
CM
CM
CO
CM
00
\
E
,0
&
.>
"o
O
O
3'0 O'O
x => uo|}jodojd
F-20
x => uo|}jodojd
-------
(A
'c
0)
re
o
CO
- CO
CM
CM
CM
CO
CM
00
,0
c/3 re
a o>
I 1
"o .—
•D O
O
O
\ \ \ \ \ r
O'l 8'0 9'0 fr'O Z'O O'O
CO
^r
co
CM
CM
CM
CO
CM
00
,0
>?
'>
"o
^
T3
O
O
\ \ \ \ \ r
O'l 8'0 9'0 fr'O Z'O O'O
x =>
x =>
\ \ \
O'l. 8'0 9'0
co
^r
co
CM
CM
CM
CO
CM
00
,0
&
.>
"o
o
o
re
0)
o
0)
re
o
CO
^r
- CO
CM
CM
CM
CO
CM
00
i r
Z'O O'O
>,
^>
"o
O
O
n i i i i r
O'l. 8'0 9'0 fr'O 3'0 O'O
x =>
x => uo|}jodojd
(A
3
Q.
O
4->
O
o
'l_
o
o
0)
co
^1-
co
CM
CM
CM
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 V'O Z'O O'O
E
,0
o
o
§
re
E
!c
o
co
^r
- co
CM
CM
CM
CO >*
S '5
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
o
O
x => uo|}jodojd
F-21
x => uo|}jodojd
-------
(A
3
E
0)
c
o
g
'i_
"5
E
re
re
Q.
CO
- CO
CM
CM
CM
CO
CM
00
\ \ \ \ \ r
O'l 8'0 9'0 fr'O Z'O O'O
,0
(A
3
o
"o ^j
-n «
O
O
O
CO
^r
- co
CM
CM
CM
\ \ \
O'l 8'0 9'0
\
Z'O
00
,0
CO >,
3 1
"o
^
T3
o
O
O'O
x =>
x =>
(A
3
re
E
E
re
\ \ \
O'l. 8'0 9'0
CO
^r
co
CM
CM
CM
CO
CM
00
i r
Z'O O'O
,0
E
3
~ fl)
^ Q.
o _>
"D O
§ Q-
o
CO
^1-
- CO
CM
CM
CM
co >*
S '5
00
\ \ \ \ \ r
O'l 8'0 9'0 fr'O Z'O O'O
O
O
x =>
x => uo|}jodojd
re
(A
0)
E
re
b
CO
^r
co
CM
CM
CM
CO
CM
00
\ \ \ i i r
O'l. 8'0 9'0 V'O Z'O O'O
E (A
.o 3
W Q.
a 2
f =
•^ 0)
o O
-s >
c O
(3 £
CO
^r
- co
CM
CM
CM
co >*
S '5
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
o
O
x => uo|}jodojd
x => uo|}jodojd
F-22
-------
re
.c
u
o
CO
- CO
CM
CM
CM
CO
CM
00
\ \ \
O'l 8'0 9'0
\
\
,0
o
O
(A
3
T3
_re
u
o
c
0)
re
o.
re
re
Q.
\
\
\
\
\
\
CO
^r
CO
CM
CM
CM
CO
CM
00
,0
O
O
3'0 O'O
O'l. 8'0 9'0 fr'O 3'0 O'O
x =>
x =>
(A
0)
>
Q.
O
C
E
\ \ \
O'l. 8'0 9'0
CO
^r
co
CM
CM
CM
CO
CM
00
,0
o
o
2
•5
0)
(A
Q.
o
CO
^1-
- CO
CM
CM
CM
CO
CM
00
i r
3'0 O'O
,0
>,
\>
"o
O
O
\ \ \ \ \ r
O'l 8'0 9'0 fr'O 3'0 O'O
x =>
x => uo|}jodojd
re
'c
o
E
CO
^r
co
CM
CM
CM
00
\ \ \ i i r
O'l. 8'0 9'0 VQ 3'0 O'O
O
O
0)
§ f I
^ = 5
^ re
O
CO
^1-
- CO
CM
CM
CM
CO >^
S '5
00
\ \ \ \ \ \
O'l. 8'0 9'0 VQ 3'0 O'O
O
O
x => uo|}jodojd
F-23
x => uo|}jodojd
-------
CO
^r
co
CM
CM
CM
CO
CM
00
,0
&
>
"o
T3
O
O
re
'l_
0)
£
CQ
CO
- CO
CM
CM
CM
CO
CM
00
,0
>?
'>
"o
^
T3
O
O
\ \ \
O'l 8'0 9'0
\ r
3'0 O'O
\ \ \ \ \ \
O'l 8'0 9'0 fr'O 3'0 O'O
x =>
x =>
CO
^r
co
CM
CM
CM
00
\ \ \
O'l. 8'0 9'0
i r
3'0 O'O
0)
o
co >. «g
CM --^ Q.
^ -a o
O i-
O
O
CO
^1-
- CO
CM
CM
CM
CO
CM
00
n i i i i r
O'l. 8'0 9'0 fr'O 3'0 O'O
,0
>,
\>
"o
^
T3
O
O
x =>
x => uo|}jodojd
"5
E
(0
CO
^1-
co
CM
CM
CM
CO
CM
00
en (0
1 2
o O
I o
o
o
CO
^1-
- CO
CM
CM
CM
CO
CM
00
T3
o
o
\ \ \ i i r
O'l. 8'0 9'0 VQ 3'0 O'O
\ \ \ \ \ r
O'l. 8'0 9'0 VQ 3'0 O'O
x => uo|}jodojd
x => uo|}jodojd
F-24
-------
0)
(A
Q.
O
4->
(Q
E
3
0)
-C
O
CO
- CO
CM
CM
CM
CO
CM
00
\ \ \ \ \ r
O'l 8'0 9'0 fr'O 3'0 O'O
O
1
o
O
55
E
3
re
CO
^r
- co
CM
CM
CM
00
\ \ \
O'l 8'0 9'0
\
3'0
,0
CO >,
3 1
"o
^
~o
o
o
O'O
x =>
x =>
(A
3
C
0)
.C
Q.
0)
(A
Q.
co
^1-
co
CM
CM
CM
CO
CM
00
a 2
1 1
o fl)
§
o
CO
^1-
- CO
CM
CM
CM
CO
CM
00
,0
>,
\>
"o
o
O
\ \ \
O'l. 8'0 9'0
i r
3'0 O'O
n i i i i r
O'l. 8'0 9'0 fr'O 3'0 O'O
x =>
x => uo|}jodojd
ro
!c
Q.
(Q
3
Q
co
^i-
co
CM
CM
CM
00
\ \ \ i i r
O'l. 8'0 9'0 VQ 3'0 O'O
(A
§ 11
y c
O
o
CO
^r
- CO
CM
CM
CM
CO >^
S '5
00
\ \ \ \ \ r
O'l. 8'0 9'0 VQ 3'0 O'O
O
O
x => uo|}jodojd
x => uo|}jodojd
F-25
-------
(A
CO
- CO
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
,0
ro
O
O
CO
^r
CO
CM
CM
CM
CO
CM
00
\ \ \ \ \ r
O'l 8'0 9'0 fr'O Z'O O'O
x=>
,0
>?
'>
"o
O
O
(A
I
0)
(A
g
"•!-•
Q.
o
CO
^1-
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x=>
a 3
>, C
1 5
o
O
O
(Q
CO
^1-
- CO
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 fr'O Z'O O'O
,0
O
O
x => uo|}jodojd
re
I
D)
CO
^1-
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 fr'O Z'O O'O
x => uo|}jodojd
O
O
(A
E
0)
C
0)
w
CO
^r
- co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 V'O Z'O O'O
x => uo|}jodojd
^
.>
"o
O
O
F-26
-------
Q.
2
n i i
O'l 8'0 9'0
CO
^r
co
CM
CM
CM
00
i r
Z'O O'O
I .55
-3 E
co >, (A
CM -^ a)
^ 2 .Q
o ,
3 1
"o
^
T3
o
o
O'O
x =>
x =>
(A
3
Q.
O
4->
O
o
'l_
o
CO
^1-
co
CM
CM
CM
CO
CM
00
,0
"o
^
T3
c
o
O
re
n i i i i r
O'l. 8'0 9'0 fr'O 3'0 O'O
CO
^i-
co
CM
CM
CM
CO
CM
00
>,
\>
"o
^
T3
O
O
\ \ \ \ \ r
O'l 8'0 9'0 fr'O 3'0 O'O
x =>
x => uo|}jodojd
(A
_3
(Q
O
O
CO
^r
co
CM
CM
CM
CO
CM
00
&
.>
"o
T3
O
O
(A
3
_
o
p
n i i i i r
O'l. 8'0 9'0 V'O 3'0 O'O
CO
^r
co
CM
CM
CM
CO
CM
00
T3
O
O
\ \ i i i r
O'l. 8'0 9'0 V'O 3'0 O'O
x => uo|}jodojd
x => uo|}jodojd
-------
(A
Q.
O
(A
O
CO
^r
co
CM
CM
CM
CO
CM
00
O'l. 8'0 9'0 V'O Z'O O'O
x =>
,0
"o
^
73
c
o
O
(A
3
.C
O
"55
\ \ \
O'l 8'0 9'0
x=>
\
3'0
CO
^r
co
CM
CM
CM
00
O'O
,0
co >*
9 '5
O
O
(A
3
E
o
c
p
O
CO
^r
co
CM
CM
CM
O'l. 8'0 9'0 V'O Z'O O'O
x=>
(3
(A
<])
Q.
2
CO
^i-
co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 VQ Z'O O'O
,0
O
O
x => uo|}jodojd
X
'l_
0)
CO
^1-
co
CM
CM
CM
00
O'l. 8'0 9'0 VQ Z'O O'O
x => uo|}jodojd
3
S2
(Q
n 2? f'
CM --^ C
^ -^ ra
y o
§ «
O o
CO
^1-
co
CM
CM
CM
CO
CM
00
\ \ i i i r
O'l. 8'0 9'0 VQ Z'O O'O
x => uo|}jodojd
&
.>
"o
O
O
F-28
-------
(A
3
O
c
2
IE
o
o
TJ
3
0)
(A
0.
CD
s
CM
CM
CM
CO
CM
t
00
_>,
'>
"o
73
O
O
\ \ \ \ \ I
O'l Q'O Q'O P'O Z'O O'O
x =>
F-29
-------
-------
APPENDIX G
VALIDATION OF METHOD
USING FIELD DATA TO DERIVE AMBIENT WATER QUALITY BENCHMARK FOR
CONDUCTIVITY USING A KENTUCKY DATA SET
ABSTRACT
The method for developing the aquatic life benchmark for conductivity was validated by
developing extirpation concentration (XCgs) and hazardous concentration (HCos) values using a
data set independently collected by the Kentucky Division of Water (KDOW) and comparing
results with those found using the larger West Virginia 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 HCos value was 282 uS/cm for the full
Kentucky data set, which is very close to the West Virginia result.
G.I. 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 Figures G-l and G-2). Although the Kentucky data set is smaller than the West Virginia
data set, it was judged to be large enough for validation of the method (see Section 3.5). 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, which were judged to be similar within the state of Kentucky in
terms of water quality, resident biota, and sources of conductivity (Pond 2004, 2010).
Background conductivity was not estimated due to the lack of designation of reference sites in
the data set or a probabilistic sample of sufficient size. However, the 25th centile of the entire
data set, which includes impaired sites, is 118 uS/cm (see Figure G-3). Although not a
background estimate, it does indicate conductivity levels are generally low in these ecoregions
and is within the range of background values for West Virginia.
G-l
-------
0 37.5 75
150
225
I Kilometers
. Kentucky Sampling Points
Ecoregions
68
69
• 70
— Other
Figure G-l. Location of Southern Appalachia (68), Central Appalachia (69),
and Allegheny Plateau (70) and sampling points.
G-2
-------
37.5 75
150
225
I Kilometers
Kentucky Sampling Points
I HUC8
Figure G-2. Location of sampling points used to develop the Kentucky HCos,
shown with 8-digit HUC catchments.
G-3
-------
2000 -
1000 -
'"•"•*•'
§ 500 -
o
(/)
^3;
:^± 200
"o
"v 100 -
o
0
20 -
_ ,_
_i_
i
i
i
-
o
@
0
1
1
1
1
1
1
1
1
1
§
~l~
1
1
— *-
o
1
1
1
1
1
1
1
1
1
o
j
.
1
o
o
[___[
1
1
1
~*~
o
i i i i i i i i r
Feb Mar Apr May Jun Jul Aug Sep Oct
Month
Figure G-3. Box plot showing seasonal variation of conductivity (uS/cm)
from the data set used to develop the Kentucky HCos. A total of 291 samples
from 1998-2004 from Ecoregions 68, 69, and 70 in Kentucky are
represented. Small sample sizes and targeted sampling could obscure
seasonal patterns, if any.
G.2. DATA SOURCES
All data used in this study were taken from the Kentucky Division of Water, Water
Quality Branch database, Ecological Data Application System (KY EDAS). Chemical, physical,
or biological samples were collected from 274 distinct locations during February-October from
1998-2004 (see Table G-l). Like the West Virginia Department of Environmental Protection
(WVDEP), the KDOW obtains biological data from both probability biosurvey and targeted
ambient biological monitoring programs. The probability biosurvey program provides a
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
G-4
-------
Table G-l. Number of samples with reported genera and conductivity.
Number of samples is presented for each month and ecoregion
Ecoregion
68
69
70
Month
Jan
Feb
0
7
0
Mar
0
14
9
Apr
10
44
21
May
0
16
2
Jun
6
16
17
Jul
18
42
21
Aug
2
18
0
Sep
3
0
0
Oct
0
25
0
Nov
Dec
Total
39
182
70
291
for streams of interest as reference or impaired sites or for other reasons. Most sites have been
sampled once during February to October. Quality assurance and standard procedures are
described by KDOW (2008). Briefly, KDOW follows similar field and laboratory quality
assurance methods as WVDEP. Macroinvertebrates are collected from mid-riffle/runs.
Although KDOW also collects a separate sample from multihabitat survey, for consistency, only
the riffle samples are quantified for this analysis. Four, 0.25-m2 samples were collected
mid-riffle in a 100-m sampling reach using a 1-m-wide, 600-|im mesh net. The four samples
were composited, and all macroinvertebrates were removed and stored in 95%-ethanol. Samples
were composited, and macroinvertebrates were identified to the lowest possible taxonomic level.
A notable difference in the WVDEP and KDOW methods is that KDOW picks the entire sample
in the laboratory, as opposed to WVDEP's fixed count of 200 organisms.
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.
G.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 West Virginia data used to derive the aquatic life benchmark for conductivity. A sample
was excluded from calculations if (1) it lacked a conductivity measurement, or (2) the pH was
low. XCgs values were calculated for genera that occurred at >25 sampling locations.
Organisms were not included unless identified to the genus level. Reference sites were not
identified in the data set, so no genera were excluded in the species sensitivity distribution
(SSD). Future analyses should identify invasive and opportunistic genera for a benchmark for
G-5
-------
Kentucky. 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 measures of individual
ions were available, so no sites with high chloride and low sulfate were identified or removed
from the Kentucky data set. Eighty-five percent of the 104 genera used to develop the SSD for
Kentucky also occurred in the West Virginia SSD. Genera from both states were judged to be
similarly susceptible to the effects of conductivity after exploratory analysis (see also Sections
A.2.4.1 and A.2.1.2). Conductivity ranged from 16-2,390 uS/cm for the Kentucky data set (see
Table G-2, Figure G-4) and 15-11,646 uS/cm for the West Virginia data set.
Table G-2. Summary statistics of the measured water quality parameters for
the Kentucky data set
Specific conductance
pH
Total HAB score
Embeddedness
N
291
291
291
291
Min
16
6.03
56
0
25th
118.5
7.1
115
8
50th
272.1
7.5
138
13
75th
674.4
7.92
161
16
Max
2,390
9.26
191
19
Mean3
265.4
7.49
136.3
12
"Conductivity reported as geometric mean.
In the Kentucky database, 359 benthic invertebrate genera were identified. Of the
359 genera collected, 104 occurred in at least 25 sampling locations in Ecoregions 68, 69, and 70
(see Appendix H). All genera used to construct the SSD occurred in all three ecoregions.
Because of the data distributions, not all 95th centiles correspond to extirpation, and some
imprecisely estimate the extirpation threshold. The following rules were applied to the XCgs
values. If the generalized additive model (GAM) mean curve at maximum conductivity is
approximately = 0 (<1% of the maximum modeled probability), then the XC95 is listed without
qualification. If the GAM mean curve at maximum conductivity is >0 but the GAM lower
confidence limit is approximating to 0, the value is listed as approximate (~). If the GAM lower
confidence limit is >0, then the XC95 is listed as greater than (>) the 95th centile. All models fits
and the scatter of points were also visually inspected for anomalies, and if the model poorly fit
the data, the uncertainty level was increased to either (~) or (>). This procedure was applied to
G-6
-------
CM _
O _
0 CO -
c
CD
CJ CD —
2
j_
^r -
CM -
o -
fill-
3
-
,-
-
-
r-i
™
^
-
r-i
PI
i>
2 100 316 1000
Conductivity (|jS/cm)
Figure G-4. Histograms of the frequencies of observed conductivity values in
samples from Ecoregions 68, 69, and 70 in Kentucky sampled from
1998-2004.
plots in Appendix I, and the XC95 values appear in Appendix H. Also these models were used to
evaluate when genera began to decline as evidence of alteration and sufficiency in Appendix A.
Many genera are marked as approximate because the Kentucky data set is small, the XCgs
models are based on a smaller number of occurrences, and the maximum conductivity measured
is lower than in the West Virginia data set. The assignation of (>) and (~) does not affect the
HCos but alerts users of the uncertainty of the XCgs values for other uses such as comparison
with toxicity test results or with results from other geographic regions.
G.4. RESULTS
Appendix H lists the genera used to construct the SSD from the Kentucky sample and
their corresponding XCgs values. The cumulative distribution functions (CDFs) used to develop
them can be found in Appendix J. The full SSD is shown in Figure G-5, and an enlargement of
the lower half of the model is shown in Figure G-6. Despite the differences in sampling method
and geographic location, the HCos values were similar: 282 uS/cm for Kentucky compared to
G-7
-------
SE
CD
CD
CD
o
•-E
o
Q.
O
CO
O
CD
O
eg
o
p
o
282 |jS/cm
200
500
1000
2000
Conductivity (uS/cm)
Figure G-5. Species sensitivity distribution (SSD) for Kentucky. A total of
104 genera are included in the SSD. The HC0s is the conductivity at the
th
intercept of the CDF while the horizontal line at the 5 centile is 282 uS/cm
G-8
-------
10
O
SE
CD
c
CD
CD
c
g
t
o
Q_
O
CO
O
CN
o
p
o
282 |jS/cm
200
i i i r
400 600 800
Conductivity (uS/cm)
Figure G-6. Species sensitivity distribution (SSD) for Kentucky. Only the
lower half of the total genera are shown to better discriminate the points in
the left side of the SSD.
295 uS/cm for West Virginia (see Figures G-2 and G-3, Table G-3). The 95% confidence
bounds for the Kentucky HCos are 169 and 380 uS/cm, which overlap with the West Virginia
data set's 95% confidence bounds of 228 and 303 uS/cm. Genera that exhibited a decreasing
occurrence with increasing conductivity were among those with the lowest XC95 values in both
states. Table G-4 shows the 10 lowest XCgs values for both West Virginia and Kentucky
samples. The 5th centile occurs near the eighth genus for West Virginia samples and fifth genus
for Kentucky samples.
G-9
-------
Table G-3. HCos values for Kentucky and West Virginia data sets
HCos
95% CI
Months
represented
N Sample
TV Genera in SSD
Kentucky
282 uS/cm
169-380
February-October
291
104
West Virginia
295 uS/cm
228-303
January-December
2,210
163
Table G-4. Comparison of the sensitive genera and XC95 values
1
2
O
4
5
6
7
8
9
10
West Virginia
Genus
Lepidostoma
Remenus
Cinygmula
Alloperla
Leptophlebia
Utaperla
Taeniopteryx
Pycnopsyche
Drunella
Ephemerella
West Virginia XC95
-121
121
230
246
251
255
260
295
297
299
Kentucky Genus
Lepidostoma
Cinygmula
Diphetor
Wormaldia
Dolophilodes
Oulimnius
Drunella
Epeorus
Neophylax
Yugus
Kentucky XC95
149
165
190
235
270
320
321
321
353
354
G.5. CONCLUSIONS
Based on the similar results, EPA judged the field-based method to be robust. The same
aquatic life benchmark appears to be applicable to West Virginia and Kentucky streams in
Ecoregions 68, 69, and 70. However, analysis of a larger statewide data set, removal of
nonreference taxa, and verification of the basic water chemistry for the region are recommended.
G-10
-------
REFERENCES
KDOW (Kentucky Division of Water). (2008) Standard methods for assessing biological integrity of surface waters
in Kentucky. Commonwealth of Kentucky Environmental and Public Protection Cabinet Department for
Environmental Protection, Division of Water February 2008, Revision 3. 120 pp. Available online at
http://www.water.ky.gov/sw/swmonitor/sop/and at http://www.water.ky.gov/NR/rdonlyres/714984BB-54F8-46B9-
AF27-290EF7A6D5CE/0/BiologicalSOPMainDocument03_08.pdf (accessed 12/19/2009).
Omernik, JM. (1987) Ecoregions of the conterminous United States. Ann Assoc Am Geograph 77:118-125.
Pond, GJ. (2004) Effects of surface mining and residential land use on headwater stream biotic integrity in the
eastern Kentucky coalfield region. Kentucky Department of Environmental Protection, Division of Water,
Frankfort, KY.
Pond, GJ. (2010) Patterns of Ephemeroptera taxa loss in Appalachian headwater streams (Kentucky, USA).
Hydrobiologia. 641(1): 185-201.
Stevens, DL, Jr.; Olsen, AR. (2004) Spatially balanced sampling of natural resources. J Am Stat Assoc
99(465):262-278.
U.S. EPA (U.S. Environmental Protection Agency). (2000) Nutrient criteria technical guidance manual: rivers and
streams. Office of Water, Office of Science and Technology, Washington, DC. EPA/822/B-00/002. Available
online at http://www.epa.gov/waterscience/criteria/nutrient/guidance/rivers/rivers-streams-full.pdf.
Woods, AJ; Omernik, JM; Brown, DD; et al. (1996) Level III and IV ecoregions of Pennsylvania and the Blue
Ridge Mountains, the Ridge and Valley, and the Central Appalachians of Virginia, West Virginia, and Maryland.
U.S. Environmental Protection Agency, National Health and Environmental Effects Research Laboratory, Corvallis,
OR. EPA/600R-96/077. 50pp.
G-ll
-------
-------
APPENDIX H
EXTIRPATION CONCENTRATION VALUES FOR
GENERA IN A KENTUCKY DATA SET
ABSTRACT
The purpose of Appendix H is to provide the reader with a list of the extirpation
concentration (XCgs) values used to develop the species sensitivity distribution and the
hazardous concentration (HCos) for Kentucky. Genera are ordered alphabetically (see
Table H-l). The numbers of occurrences in the data set are noted in the right-hand column.
Genera highlighted in gray do not occur at West Virginia reference locations, but were included
in the Kentucky species sensitivity distribution (SSD). If they were removed the hazardous
concentration, HCos would be slightly lower.
Not all 95th centiles correspond to extirpation, and some imprecisely estimate the
extirpation threshold. The following rules were applied to the XCgs values using the fitted curve
and the confidence bounds from the plots in Appendix I. If the generalized additive model
(GAM) mean curve at maximum conductivity is approximately equal to 0 (defined as less than
1% of the maximum modeled probability), then the XCgs value is listed without qualification. If
the GAM mean curve at maximum conductivity is >0 but the lower confidence limit is
approximating to 0 (<1% of the maximum mean modeled probability), then the XCgs value is
listed as approximate (~). If the GAM lower confidence limit is >0, then the XCgs value is listed
as greater than (>) the 95th centile. All model fits and scatter of points were also visually
inspected for anomalies, and if the model poorly fit the data, the uncertainty level was increased
to either (~) or (>).
The assignation of (~) and (>) does not affect the HCos. They are provided to alert users
to the uncertainty of some XCgs values for other uses such as comparison with toxicity test
results or with results from other geographic regions.
H-l
-------
Table H-l. Extirpation concentration and sample size from Kentucky data
set. Highlighted genera are not found at WV reference sites but were included in
the SSD for Kentucky. XCgs values reported without a preceding symbol indicate
evidence of extirpation within the tested range. XCgs values preceded by a (~) or
(>) indicate extirpation with greater uncertainty or extirpation at a level above the
reported value. Genera highlighted in gray do not occur at West Virginia
reference locations.
Order
Diptera
Ephemeroptera
Plecoptera
Ephemeroptera
Plecoptera
Coleoptera
Diptera
Odonata
Diptera
Ephemeroptera
Odonata
Ephemeroptera
Odonata
Decapoda
Trichoptera
Trichoptera
Trichoptera
Diptera
Ephemeroptera
Veneroida
Megaloptera
Diptera
Diptera
Diptera
Family
Chironomidae
Baetidae
Perlidae
Ameletidae
Nemouridae
Elmidae
Tipulidae
Coenagrionidae
Athericidae
Baetidae
Aeshnidae
Caenidae
Calopterygidae
Cambaridae
Hydropsychidae
Hydropsychidae
Philopotamidae
Chironomidae
Heptageniidae
Corbiculidae
Corydalidae
Chironomidae
Chironomidae
Chironomidae
Genus
Ablabesmyia
Acentrella
Acroneuria
Ameletus
Amphinemura
Ancyronyx
Antocha
Argia
Atherix
Baetis
Boyeria
Caenis
Calopteryx
Cambarus
Ceratopsyche
Cheumatopsyche
Chimarra
Chironomus
Cinygmula
Corbicula
Corydalus
Cricotopus
Cryptochironomus
Diamesa
XC95
>1,410
>619
>697
>579
>1,269
798
>958
>1,410
>1,650
>1,410
>1,318
>1,410
>2,082
>1,090
>1,577
>1,630
>2,000
>1,670
165
>1,829
>1,650
>2,037
>1,037
>2,074
N
43
98
105
69
107
30
49
51
61
170
92
85
35
157
102
230
90
31
39
84
121
98
27
54
H-2
-------
Table H-l. Extirpation concentration and sample size
from Kentucky data set (continued)
Order
Diptera
Diptera
Coleoptera
Ephemeroptera
Trichoptera
Plecoptera
Trichoptera
Ephemeroptera
Coleoptera
Plecoptera
Lumbriculida
Coleoptera
Neotaenioglossa
Odonata
Ephemeroptera
Ephemeroptera
Ephemeroptera
Diptera
Ephemeroptera
Basommatophora
Odonata
Plecoptera
Coleoptera
Diptera
Diptera
Trichoptera
Trichoptera
Ephemeroptera
Family
Tipulidae
Chironomidae
Gyrinidae
Baetidae
Hydropsychidae
Perlodidae
Philopotamidae
Ephemerellidae
Elmidae
Perlidae
Lumbriculidae
Psephenidae
Pleuroceridae
Coenagrionidae
Heptageniidae
Ephemeridae
Ephemerellidae
Chironomidae
Ephemerellidae
Ancylidae
Gomphidae
Chloroperlidae
Dryopidae
Empididae
Tipulidae
Hydropsychidae
Hydroptilidae
Isonychiidae
Genus
Dicranota
Dicrotendipes
Dineutus
Diphetor
Diplectrona
Diploperla
Dolophilodes
Drunella
Dubiraphia
Eccoptura
Eclipidrilus
Ectopria
Elimia
Enallagma
Epeorus
Ephemera
Ephemerella
Eukiefferiella
Eurylophella
Ferrissia
Gomphus
Haploperla
Helichus
Hemerodromia
Hexatoma
Hydropsyche
Hydroptila
Isonychia
XC95
>484
>1,437
>874
190
>958
>997
270
321
>1,650
>1,649
>1,294
>582
-1,131
>959
321
-559
-467
>1,842
>499
>872
>1,063
485
>1,050
>2,000
>1,134
>1,650
>1,680
>1,524
N
25
29
45
25
102
35
31
37
86
31
92
66
33
31
65
42
70
54
84
29
36
38
148
123
106
161
58
132
H-3
-------
Table H-l. Extirpation concentration and sample size
from Kentucky data set (continued)
Order
Plecoptera
Odonata
Trichoptera
Ephemeroptera
Plecoptera
Isopoda
Odonata
Coleoptera
Diptera
Diptera
Diptera
Trichoptera
Megaloptera
Trichoptera
Coleoptera
Decapoda
Diptera
Coleoptera
Ephemeroptera
Diptera
Plecoptera
Plecoptera
Basommatophora
Ephemeroptera
Trichoptera
Diptera
Ephemeroptera
Diptera
Family
Perlodidae
Gomphidae
Lepidostomatidae
Heptageniidae
Leuctridae
Asellidae
Corduliidae
Elmidae
Chironomidae
Chironomidae
Chironomidae
Uenoidae
Corydalidae
Leptoceridae
Elmidae
Cambaridae
Chironomidae
Elmidae
Leptophlebiidae
Chironomidae
Peltoperlidae
Perlidae
Physidae
Baetidae
Polycentropodidae
Chironomidae
Baetidae
Simuliidae
Genus
Isoperla
Lanthus
Lepidostoma
Leucrocuta
Leuctra
Lirceus
Macromia
Macronychus
Micropsectra
Microtendipes
Natarsia
Neophylax
Nigronia
Oecetis
Optioservus
Orconectes
Orthocladius
Oulimnius
Paraleptophlebia
Parametriocnemus
Peltoperla
Perlesta
Physella
Plauditus
Polycentropus
Polypedilum
Procloeon
Prosimulium
XC95
>1,176
>1,564
149
>686
>1,029
-958
-772
>1,722
-462
>681
>1,630
353
>1,197
>1,337
>1,563
>1,291
>1,480
320
-420
>1,583
>1,520
>1,399
>1,856
-703
>570
>1,251
>800
>866
N
81
34
30
45
131
35
27
54
25
58
45
73
153
31
178
115
50
31
76
185
37
51
52
55
82
158
42
54
H-4
-------
Table H-l. Extirpation concentration and sample size
from Kentucky data set (continued)
Order
Coleoptera
Ephemeroptera
Diptera
Trichoptera
Hemiptera
Diptera
Diptera
Trichoptera
Megaloptera
Diptera
Ephemeroptera
Coleoptera
Diptera
Ephemeroptera
Odonata
Plecoptera
Diptera
Diptera
Diptera
Trichoptera
Ephemeroptera
Diptera
Trichoptera
Plecoptera
Family
Psephenidae
Baetidae
Tipulidae
Limnephilidae
Veliidae
Chironomidae
Chironomidae
Rhyacophilidae
Sialidae
Simuliidae
Heptageniidae
Elmidae
Chironomidae
Heptageniidae
Gomphidae
Chloroperlidae
Chironomidae
Chironomidae
Tipulidae
Leptoceridae
Leptohyphidae
Chironomidae
Philopotamidae
Perlodidae
Genus
Psephenus
Pseudocloeon
Pseudolimnophila
Pycnopsyche
Rhagovelia
Rheocricotopus
Rheotanytarsus
Rhyacophila
Stalls
Simulium
Stenacron
Stenelmis
Stenochironomus
Stenonema
Stylogomphus
Sweltsa
Tanytarsus
Thienemannimyia
Tipula
Triaenodes
Tricorythodes
Tvetenia
Wormaldia
Yugus
XC95
>750
-861
>1,051
>775
>600
>1,117
>1,601
>574
>1,843
>1,580
>862
>1,520
>824
>993
>1,720
558
>1,316
>1,697
>1,814
>938
>2,000
>1,254
235
354
N
111
36
40
64
27
51
115
94
64
179
90
168
35
178
90
55
118
155
150
31
48
46
38
25
H-5
-------
-------
APPENDIX I
GRAPHS OF OBSERVATION PROBABILITIES
FOR GENERA IN A KENTUCKY DATA SET
ABSTRACT
The purpose of Appendix I is to help the reader visualize the changes in the occurrence of
each genus in the Kentucky data set as conductivity increases. Each figure depicts a general
additive model (GAM) of the relationship between capture probabilities of a genus and
conductivity. Genera are ordered from the lowest to the highest extirpation concentration (XCgs)
value. Open circles are the probabilities of observing the genus within a range of conductivities.
Circles at zero probability indicate no individuals were found in any samples with those
conductivities. The GAM line (solid line) fitted to the probabilities is for visualization and
dashed lines are 90% confidence bounds. The vertical dashed line indicates the XCgs taken
from Appendix H. The fitted lines and confidence bounds were used to assign uncertainty levels
of the XCgs values in Appendix H.
1-1
-------
CD
-C
.0.
Q
o
en
CO
CM
CO fO
? >
-I—"
O
CQ-0
in CZ
O
O
CO
.3
£
.5
~3
O
o
en
CO
CM
CO fO
? >
-I—"
O
CZ
O
O
O'l 8'0 9'0
Z'Q O'O
O'l. 8'0 9'0 VO Z'Q O'O
I
O
o
_ en
^ co
CM
CO
- o>
E
c/5
ii
_>%
>
^4—"
O
CO
^
cz
o
O
_ CO
I I I I
O'l 8'0 9'0 VO Z'O O'O
CO
CD
•
O
Q
i
9'0
o
_ en
^ co
CM
CO
- O)
E
c/5
ii
_>%
>
^4—"
O
CO
^
CZ
o
O
_ CO
O'O
CD
s
5
CO
-J
o
en
co
CM
CO fO
I- ? >
-I—'
O
Z3
in CZ
O
O
_ CO
i i i i
O'l 8'0 9'0 17'0 Z'O O'O
CD
;S
"CD
i
o
_ en
^ co
CM
I I I I
O'l 8'0 9'0 VO Z'O
CO
en
O
Z5
-D
CZ
O
O
_ co
O'O
1-2
-------
X
CO
CD
o
o>
CO
CM
E
oo ~^
w
-I—"
O
CO (">
10 C
O
O
£
O
CD
CO
o
o>
CO
CM
3
CO
_>%
>
^4—"
O
CO
IO
o
o
O'l 8'0 9'0 P'O Z'O O'O
O'l 8'0 9'0
O'O
CO
S
o
CD
o
o>
CO
CM
E
oo ~^
w
-I—"
O
CD
^
cz
o
O
_ CO
I I I I
O'l 8'0 9'0 VO Z'O O'O
.CD
-Q
O'l 8'0 9'0
o
o>
CO
CM
E
oo ^^
w
-I—"
O
CO
IO
o
o
_ CO
I
3'0 O'O
o
c
o
O)
CO
CM
CO fO
? >
-I—"
O
o
O
_ CO
O'l 8
I I I I
0 9'0 VO Z'O O'O
CO
3
I
o
O)
CO
CM
CO
O)
>
^4—"
O
o
O
_ CO
O'l. 8'0
I I I
9'0 VO Z'O O'O
1-3
-------
-2
CD
Q.
o
&
3:
\
O'l
'0 9'0 P'O Z'O
o
o>
CO
CM
CO
10
I
to
^.
_>%
>
^—'
o
d
o
O
O'O
CD
6
CD
-C
\
O'l
o
o>
CO
CM
I
to
^.
_>%
>
^4—"
O
_ CO
o
o
'0 9'0 VO Z'Q O'O
4S
o
c
Q
\
O'l
'0 9'0
\ \
VO Z'O
o
o>
CO
CM
I
to
^.
_>%
>
^4—"
o
CO
10
o
o
_ CO
r
O'O
CD
42
1
CO
o
O)
CO
CM
E
oo ^^
w to
co <^*
? >
-i—"
O
CO
10
O
O
_ co
n i i i i r
O'l 8'0 9'0 VO Z'O O'O
CD
S
CD
-C
M-B.
o
o>
CO
CM
^ E
oo ^^
w to
CO <^*
? >
-I—'
O
CO
10
o
o
_ CO
n i i i i r
O'l 8'0 9'0 fO 3'0 O'O
CD
-c
o
O)
CO
CM
-y
CO
CO fO
? >
-i—'
O
o
O
_ CO
\ i i i i r
O'l 8'0 9'0 fO Z'O O'O
1-4
-------
CO
CD
6
o
o>
CO
CM
CO
o>
_O
tt)
^_
&
>
"o
CO
10
o
o
CO
1
c
CD
O
o
o>
CO
CM
£.-
E
oo -b£
WW
%~
o
CO
10
O
O
O'l 8'0 9'0 VO Z'Q O'O
O'l 8'0
9'0
O'O
o
I
o:
O'l 8'0 9'0 VO Z'Q
o
o>
CO
CM
E
oo -b£
WW
%~
CO
10
O
O
O'O
CO
1
o
^= CO
o ^:
\
o
o>
CO
CM
CO
o>
_o
c/j
^.
^
>
"o
T
T
O'l. 8'0 9'0 VO Z'O
CO
10
O
O
O'O
CO
3
Q.
fi
C
CD
S.
O'l 8
o
O)
CO
CM
O
"o
o
O
_ co
I I I I
0 9'0 VO Z'O O'O
CO
I
o
o>
co
CM
O
I I
O'l Q'O 9'Q
oo
"CO
"o
O
O
_ CO
\
Z'O O'O
1-5
-------
CD
CD
C
O
o
o>
CO
CM
S^
w fQ
«._
CO <-*
10 C
O
O
O'l 8'0 9'0 P'O Z'O O'O
o
o
o>
CO
CM
3
CO
CO <^*
? >
-I—"
O
CO (">
10 C
O
O
_ CO
n i i i i i
9'0 9'0 VO Z'O Z'O I/O O'O
3
3
O
§
3
CD
-J
\
O'l
'0 9'0
\
Z'O
o
O)
CO
CM
I
c/5
^.
_>%
>
^4—"
o
CO
10
O
O
_ CO
r
O'O
CO
3
C
CD
-C
Q.
CD
o
en
CO
CM
-I—"
O
_ CO
O
O
_ CO
T i i i r
''0 9'0 VO Z'O O'O
CO
CD
.0.
•5
c
!
o
o>
CO
CM
^ E
oo ^^
w CO
CO <^*
? >
-I—"
O
CO
IO
O
O
_ CO
n i i i i r
O'l 8'0 9'0 VO Z'O O'O
CO
CD
CL
o
_ en
^ CO
CM
\ I I I I
O'l 8'0 9'0 fO Z'O
CO fO
^ « >
-i—"
O
O
O
_ CO
O'O
1-6
-------
c
o
CO
CM
E
oo ~^
w 00
w <^*
? >
-I—"
O
CO
IO
o
o
c
§
CO
c
5
00
o
o>
CO
CM
^
oo
CO fO
? >
-I—"
O
o
o
9'0 fO Z'O O'O
O'l. 8'0 9'0 VO Z'Q O'O
I
I
C
o
CD
CO
CM
E
oo ~^
w 00
CO <^*
? >
-I—"
O
CO
IO
o
o
_ CO
I I I I
9'0 VQ E'O Z'O I/O O'O
c
o
CD
O
"o
CD
£
9'0
o
o>
CO
CM
^
00
CO fO
? >
-I—"
O
o
O
_ CO
O'O
CD
O'l 8
o
o>
CO
CM
^ E
oo ^^
w 00
CO <^*
? >
-I—"
O
CO
IO
o
o
_ co
I I I I
0 9'0 VO Z'O O'O
CO
1
p
c
s
-s
p
c
CO
o
o>
CO
CM
-y
00
CO fO
? >
-i—"
O
o
O
_ CO
I I I I I
9'0 9'0 VQ E'O 3'0 l-'O O'O
1-7
-------
CO
CD
.C
Q
o
o>
CO
CM
cS
CO
CO <^*
? >
-I—"
o
CO
10
O
O
CO
c
o
•92
.0.
Q
o
o>
CO
CM
3
CO
CO <^*
? >
-I—"
O
CO
10
o
o
Q'O
P'O
Z'O
O'O
O'l Q'O Q'O P'O Z'O O'O
CO
CO
.to
4
o
CD
CO
CM
CO fO
? >
-i—"
O
o
O
_ CO
\ \ \ i i r
9'0 17'0 E'O 3'0 I/O O'O
CO
•8
\
O'l
\
Q'O
\
Q'O
O
O>
CO
CM
E
c/5
^.
_>%
>
^4—"
O
CO
10
o
o
_ CO
P'O Z'O O'O
3
.6
CO
p
\
O'l
\
Q'O
\
Q'O
o
o>
CO
CM
E
%
>
^4—"
O
CO
10
o
o
_ CO
VQ Z'O O'O
CO
c
CD
CO
\
\
\
\
o
o>
CO
CM
CO fO
? >
-I—'
O
O
O
_ CO
9'0 9'0 fO E'O 3'0 I/O O'O
1-8
-------
CO
s
CD
C
O
c
^
CO
o
o>
CO
CM
E
oo ~^
w CO
CO <^*
? >
-I—"
O
CO (">
10 C
O
O
CO
3
6
o
-
5
I
o
_ CD
^ CO
CM
^
CO
CO fO
I" ? >
-I—"
O
o
O
O'l VQ Q'O P'O Z'Q O'O
amjdeo
OE'O OZ'O Ol'O OO'O
CO
co
o
O)
CO
CM
E
oo ~^
w CO
CO <^*
? >
-I—"
O
CO
IO
o
o
_ CD
I I I I I
9'0 9'0 VQ E'O Z'O I'O O'O
CD
-J
o
o>
CO
CM
E
CO
^.
_>%
>
^4—"
o
CO
IO
o
o
_ CO
I I I I
O'l 8'0 9'0 VO Z'O O'O
CO
3
CD
O
o
o>
CO
CM
-y
CO
CO fO
03 -^
^4—"
O
o
O
_ CO
I I I I
9'0 VO Z'O Z'O I/O O'O
CD
&
.0.
Q
o
o>
CO
CM
CO
CO fO
? >
-i—'
O
o
O
_ CO
I I I I
O'l 8'0 9'0 VO Z'O O'O
1-9
-------
-c
I
o
CD
o
o>
CO
CM
E
oo -b£
w CO
%~
o
CO
10
O
O
CO
I
3
o
o>
CO
CM
E
oo -b£
w CO
%~
o
CO
10
O
O
9'0 VQ E'O Z'O I-'O O'O
9'0
E'O Z'O I'O O'O
o
•D
3
CD
9'0
o
o>
CO
CM
CO
o>
_O
CO
^.
^
>
"o
CO
10
O
O
O'O
CO
3
Q.
5
O
O
'8
o
CD
o
o>
CO
CM
S^
"CO
"o
o
O
I I I I T
9'0 fO E'O 3'0 I/O O'O
•s
I
o
o>
CO
CM
•* o
oo -X
<° CO
c^^
o
CO
10
O
O
- CO
i i i i i r
O'l 8'0 9'0 fO 3'0 O'O
CO
s
CO
•Q
o
o>
CO
CM
_o
CO
"o
O
O
_ CO
i i i i r
O'l 8'0 9'0 fO 3'0 O'O
1-10
-------
CO
5
i
o
o>
CO
CM
cS
CO
_>%
>
^4—"
O
CO
10
O
O
6
CD
.C
-C
I
o
o>
CO
CM
3
CO
_>%
>
^4—"
O
CO
10
O
O
O'l
8'0 9'0 P'O Z'O O'O
amjdeo
O'l 8'0 9'0 P'O Z'O
amjdeo
O'O
Q.
O
CO
o
o>
CO
CM
3
CO
CO <^*
? >
-I—'
O
CO
10
O
O
_ CO
I I I I
O'l 8'0 9'0 VO Z'O O'O
I
o
o>
CO
CM
o
o
_ CD
I I I I I I
9'0 9'0 fO E'O Z'O I'O O'O
CO
6
8 '
o
o>
CO
CM
cS
CD
CO <^*
? >
-I—'
O
CD
IO
O
O
_ co
i i i i
O'l 8'0 9'0 fO 3'0 O'O
s
.5
^B
CD
§
£
o
o>
CO
CM
CO f^T*
? >
-I—'
O
o
O
_ CD
I I I I
O'l 8'0 9'0 fO Z'O O'O
1-11
-------
CO
I
I
o
o>
CO
CM
CO fO
^
O
co-0
o
o
4S
CO
_ CO
o
o>
CO
CM
CO fO
? >
-I—"
O
.
O
O
_ CO
O'l 8'0 9'0
O'O
8'0 9'0 VO Z'Q O'O
CO
3
1
.5.
o
en
CO
CM
0)
CO !*O
„
-i—"
O
co-0
O
O
_ CO
I I I I I
'0 9'0 VQ Z'O O'O
CD
O
CD
O
O
O>
CO
CM
CO i"T*
? >
-I—"
O
O
O
_ co
I I I \ \
9'0 9'0 17'0 E'O Z'O I'O O'O
CO
^
o
CD
C
O
g
O
o
o>
CO
CM
E
%
>
^4—"
O
CO
IO
O
O
_ CO
\
\
\
O'l 8'0 9'0 VO Z'O O'O
CD
I
QQ
o
o>
CO
CM
CO !^V*
„
-I—"
O
O
O
_ CO
O'l 8
I I I
0 9'0 VO Z'O O'O
1-12
-------
.to
-G
CD
CO
QQ
O'l 8'0 9'0
o
o>
CO
CM
3
CD
_>%
>
^4—"
O
CO
10
o
o
Z'O O'O
CO
3
•5
.CO
O
o
o
o
o>
CO
CM
3
CO
CO <^*
? >
-I—"
O
CO
10
o
o
I I I I I
9'0 VO E'O Z'O I/O O'O
CO
O'l
o '
\ \
8'0 9'0
o
o>
CO
CM
_O
c/5
_>%
>
^4—"
O
CO
10
o
o
_ CD
VO Z'O O'O
CO
CD
.0.
& c
1
Q
Q_
O
o
0
0 °
V
\
o
0
»
f
/
\
O'l 8'
o
o>
CO
CM
3
CO
CO <^*
? >
-I—'
O
CO
10
o
o
_ CO
I I I I
'0 9'0 fO Z'O O'O
CO
t
CO
CD
-Q
CO
-Q
o
o>
CO
CM
CO !<^
? >
-I—'
O
o
O
_ CO
I I I I I
O'l 8'0 9'0 fO 3'0 O'O
CO
5
CD
CO
o
o
o>
CO
CM
CO !*O
? >
-i—'
O
o
O
_ CO
I I I I
O'l 8'0 9'0 fO 3'0 O'O
1-13
-------
CO
o
CO
o
o>
CO
CM
^
co
CO fO
? >
-I—"
O
O
O
CD
I
CD
O
o
o>
CO
CM
^
CO
CO fO
? >
-I—"
O
O
O
8'0 9'0 P'O Z'O O'O
8'0 9'0 VO Z'Q O'O
CO
CD
C
CO
O'l 8'0
I
9'0
I
vo
\
Z'O
o
CD
CO
CM
E
co -~^.
WV)
CO <^*
? >
-I—"
O
_ co
o
o
8
O'O
_ CO
CO
C
CO
-J
O'l. 8'0
I
9'0
I
vo
\
Z'O
o
o>
CO
CM
I
CO
^.
_>%
>
^4—"
O
CO
IO
O
O
_ CO
O'O
CO
c:
CD
t
I
o
o>
CO
CM
E
CO
^.
_>%
>
^—'
o
co
10
O
O
_ CO
CU 8'0 9'0
i i
VO Z'O O'O
CO
I
CD
CO
.O
x \°
0 X V
° ° °A
* \ 1
0 ° > \
0
1
• *
o ° !°
8 °'
o ,
8
o (
8 •
° 0
0
0
f
1
1 0
\ _
| O Q
°\ 0
> 0
1 0
10
to
' 8
^ °
» 0
» 8
0\
O)
CO
CM
^^.
E
m ^^
w to
13.
s.^^
CO <^
- « >
-I—"
O
O
O
_ co
n i r
O'l. 8'0 9'0 VO Z'O
O'O
1-14
-------
CO
I
CD
O
CD
§
o
o>
CO
CM
CD fO
03 -^
"o
O
O
_ CO
O'l 8'0 9'0 VO Z'O O'O
2
a
1
a
0 •
\
\
° d
1
8'
i ,
ID
/ (
_ ° ii
° '
°'V;
v*/.1
' / '
/ 1
8 ' / '
* /
O'l 8'0 9'0 17'0 Z'O
'0
'1
1 H
. B
' fi
, S
00000 CO 00 00 C
°
8
o
8
Q
H
"
8
O'O
o
a>
CO
CM
3
CO
_>%
>
^4—"
O
CO
10
O
O
_ CO
CO
3
6
CD
g
.0
O'l
o'
I
T I I I
rO 9'0 VO Z'O
o
_ CD
^ CO
CM
CO f^>
h 03 .^
"o
in C
O
O
_ CO
O'O
CD
CD
./...§.
o
O)
CO
CM
E
oo ^^
w CO
CO <^*
? >
-I—"
O
CO
10
O
O
_ CD
9'0 VO Z'O Z'O I'O O'O
£
.:$
~5
£
I I I I
O'l 8'0 9'0 VO Z'O
o
o>
CO
CM
E
CO
CO fO
"1
"o
O
O
_ CD
O'O
CD
!
t
CD
6
3
CD
g
O'l
I I
8'0 9'0
I I
VO Z'O
o
_ CD
^ CO
CM
\- S-y
w CO
CO <^*
- o> rtr
O
3
co-D
10 C
O
O
_ CO
O'O
1-15
-------
-2
-c
.1
•§
Q
o
o>
CO
CM
CO fO
? >
-I—"
O
CQ-0
O
O
-2
-G
f
o
o>
CO
CM
CO fO
? >
-I—"
O
O
O
8'0 9'0 VO Z'Q O'O
O'l 8'0 9'0
O'O
o
O
o
O)
CO
CM
E
oo ~^
^tt)
CO <^*
? >
-I—"
O
CO
IO
O
O
_ CO
I I I I
O'L 8'0 9'0 VO Z'O O'O
CO
3
6
o
c
8
6
0
o
/ /
--;-/;'--
VQ
i
E'O
o
o>
CO
CM
E
oo ^^
10 O)
CO <^*
? >
-I—"
o
CO
IO
O
O
_ CO
3'0 I-'O O'O
CD
-C
o
O)
CO
CM
CO fO
? >
-I—"
O
O
O
_ CO
I I I I I I
9'0 9'0 VO E'O 3'0 l-'O O'O
CD
I
f
o
O)
CO
CM
CO fO
? >
-i—"
O
O
O
_ co
I I I I
O'l. 8'0 9'0 VO Z'O O'O
1-16
-------
o
CO
o
o>
CO
CM
3
CO
CO <^*
? >
-I—"
O
CO
10
O
O
-
CD
UJ
_ CO
o
o>
CO
CM
E
GO
^.
_>%
>
^4—"
O
CO
10
O
O
_ CO
9'0
fO
Z'O
O'O
O'l
'0 9'0 fO Z'O
O'O
amjdeo
CO
3
-C
I
a
o
^
CO
\
\
\
O'l 8'0 9'0 fO
o
O)
CO
CM
E
GO
^.
_>%
>
^4—"
O
CO
10
O
O
_ CO
Z'O O'O
O
-e
o
O
9'0
I
VQ
\
Z'O
o
o>
CO
CM
E
oo ~^
w GO
CO <^*
„
-I—'
O
CO
10
O
O
_ CO
O'O
CO
5,
.5
c
c
CO
5
CD
C
CD
S
°»o \
• V\.T
o • ¥
° o '• \
8
° °
0 '
°°
o A
8
/
o /
/
' o
0
8
2
o
^°:°°
h ••
1 0
1 0
• '•. '
1 a
< 8
a>
CO
CM
t
- ro -~i
w GO
^^ CO
-8| 1
o
O
_ CO
O ' V
v \ 6"
V*°\
0 ° \ \
o i
0 1
01
0 1
°°l
8 ° // '
8 ' / to
1 // ;»
/ / 10
1 0
0 0
o
o
• 0
O 1
01 0
d
' 0
' 8
o
o
o
0
a>
CO
CM
t
- ro -~i
w GO
CO <^*
- « >
-1—'
O
^
w-o
& c
o
O
_ CO
O'l.
\ \ r
8'0 9'0 fO Z'O
O'O
i r
O'l 8'0 9'0 fO Z'O
O'O
1-17
-------
g
CO
.5
6
o
o>
CO
CM
CD fO
? >
-I—"
O
O
O
CO
O
O
2
o
o>
CO
CM
CO fO
? >
-I—"
O
O
O
O'l 8'0 9'0
Z'O O'O
O'l. 8'0 9'0 VO Z'Q O'O
o
O'l 8
01,
o
o>
CO
CM
E
oo ~^
<° (7)
CO <^*
? >
-I—"
O
CO
IO
o
o
_ CD
I I I I I
'0 9'0 VQ ZO O'O
CO
CD
•Q
O
I
O
.0
9'0
I
vo
I
ZO
o
o>
CO
CM
c?
CO
co <^*
? >
-i—"
O
co
10
O
O
_ co
O'O
.CO
"co
Q_
O - i
o o°;
1
1
0 ° '
1
1
10°
1
1
** 0
1
\ w
o
O)
co
CM
^ E
oo ^^
w (7)
CO <^*
? >
-I—"
O
CO
IO
o
o
_ co
I I I I I
O'l 8'0 9'0 VQ ZO O'O
CO
8
CD
6
o
o>
CO
CM
^
(7)
CO !^V*
? >
-I—"
O
o
O
_ CD
I I I I
O'l 8'0 9'0 VO ZO O'O
1-18
-------
2,
CD
O
CO
CM
'•^
o
in C
O
O
\ \ \ \ \ \
9'0 VQ C'O Z'Q VQ O'O
CD
CO
CD
5
.TO
Q
\
\
\
O'L 8'0 9'0
o
O I,
I I I
O „ ill
\
\
CO
CM
in
o
C
O
O
Z'O O'O
1-19
-------
-------
APPENDIX J
GRAPHS OF CUMULATIVE FREQUENCY DISTRIBUTIONS
FOR GENERA IN A KENTUCKY DATA SET
ABSTRACT
The purpose of Appendix J is to help the reader visualize the changes in the occurrence of
each genus in the Kentucky data set as conductivity increases and understand how the extirpation
concentration (XCgs) values are derived. Each plot contains the weighted cumulative
distribution function (CDF) for the occurrence of a genus with respect to conductivity. For each
genus, the points in the CDF represent the weighted proportions of occurrences of the genus in
samples less than the indicated conductivity value (uS/cm), calculated using Equation 1. The
95* centile is found at the intersection of the dashed horizontal line with the CDF. The
conductivity for the 95th centile is the XCgs value and is found at the intersection of the vertical
line and the x-axis.
J-l
-------
o
O)
CO
CM
CO
O)
_ CO
o
&
tj
o
O
3
O
O'l 8'0 9'0
VQ
Z'Q
O'O
n
O'l.
O
O)
CO
CM
- oS
CD
O)
CD
1X5
O
>,
'>
•i=
o
O
O
'0 9'0
V'Q
Z'Q
O'O
cc
"3
E
I
O
o
O)
CO
CM
oo -p-
w I co
w -S
ZL <3
CO
O)
CO
1X5
_ CO
o
O
o
Q
o
O)
CO
CM
oo
CO
CO
O)
CO
1X5
_ CO
_0
o
o
O'l 8'0 9'0
VQ
Z'Q
O'O
O'l
'0 9'0
VQ
Z'Q
O'O
cc
E
5
co
-J
o
_ CD
^ CO
CM
CO
a>
CO
1X5
_ CO
o
CC
•t i
o
O
O'l
'O 9'Q
VQ
Z'Q
O'O
n
O'l.
O
O)
CO
CM
- oS
CO
a>
co
1X5
_ CO
'O 9'Q VQ Z'Q
r
O'O
o
o
o
J-2
-------
X
-2
I
o
o>
CO
CM
O
CO
O)
_ co
1
o
co
§•
O
O
o
O)
CO
CM
- oS
CD
O)
CO
1X5
O
>,
'>
•i=
o
73
O
O
O'l
8'0 9'0
V'Q
Z'Q
O'O
\
O'l
'0 9'0
V'Q
Z'Q
O'O
co
5
o
o
O)
CO
CM
co -p- CC
w I 3
CO
O)
CO
1X5
_ CO
73
O
O
s
ff
o
O)
CO
CM
oo
CO
CO
O)
CO
1X5
_ CO
_0
73
O
O
O'l
8'0 9'0
VQ
Z'Q
O'O
O'l
'0 9'0
VQ
Z'Q
O'O
-2
"33
§
o
_ en
^ co
CM
\
O'l
CO
a>
o
^ co
l|
o >•
73
C
CO (§
1X5 u
_ CO
o
a>
co
CM
- oS
CO
a>
co
1X5
_ CO
o
73
O
O
8'0 9'0
VQ
Z'Q
O'O
O'l
'O 9'Q VQ Z'Q
O'O
J-3
-------
-
I
o
_ en
^ co
CM
CO
O)
CO
1X5
_ CO
o
o
D
73
C
O
O
E
,
'>
•*=
o
73
O
O
O'l 8'0 9'0
V'Q
Z'Q
O'O
\
O'l
'0 9'0
V'Q
Z'Q
O'O
O
c
§
Q
o
O)
CO
CM
CD
O)
- CO
73
O
O
s
1
00
o
o>
CO
CM
oo
CO
CO
O)
CO
1X5
_ CO
_0
73
O
O
O'l 8'0 9'0
VQ
Z'Q
O'O
O'l
'0 9'0
VQ
Z'Q
O'O
-
•&
o
_ en
^ co
CM
\
O'l
o
-
CO
O)
•=L CD
11
'•s ^
73
c
CO R
1X5 u
_ CO
O
O)
CO
CM
- oS
CO
O)
CO
1X5
_ CO
o
73
O
O
'O 9'Q
VQ
Z'Q
O'O
O'l
'O 9'Q VQ Z'Q
O'O
J-4
-------
3
E
o
O)
CO
CM
- oS
CO
O)
_ CO
_ CO
O
o
O
C
O
o
O'l
8'0 9'0
V'Q
Z'O
O'O
\
O'l
o
o>
CO
CM
- oS
CO
O)
_ co
_ CO
O
>,
'>
•*=
o
o
o
'0 9'0
V'Q
Z'O
O'O
-
o
I
CC
o
O)
CO
CM
oo -p-
W I
CO
O)
73
C
o
CO (•)
1X5 U
a
cc
_ CO
o
o>
CO
CM
CO >^
05 -
"o
D
73
C
o
CO (•)
10 U
_ CO
O'l.
8'0 9'0
VQ
Z'Q
O'O
O'l
'0 9'0
Z'O
O'O
CO
I
C
«£
o
O)
CO
CM
- oS
CO
O)
CO
1X5
_ CO
o
a -2
^ S-
•5 O
73
O
O
O
O)
CO
CM
- oS
CO
O)
CO
1X5
_ CO
o
73
O
O
O'l
Q'O 9'Q
VO
O'O
n
O'l
'O 9'Q VO Z'O
O'O
J-5
-------
cc
i
o
c
s
o
o
O)
CO
CM
o
CD
O)
CD
1X5
E
8
o
O
O
o
O)
CO
CM
- oS
CO
O)
_ CO
O
>,
'>
•i=
o
O
O
O'l 8'0 9'0
V'Q
Z'O
O'O
O'l.
'0 9'0
VQ
Z'Q
O'O
g
o
o>
CO
CM
oo -p-
<° I
g .f ^
"-..gO.
'o CU
CO
1X5
_ CO
o
O
o
o>
CO
CM
oo
CO
CO
O)
CO
1X5
_ CO
_0
o
o
O'l 8'0 9'0
VQ
Z'O
O'O
O'l
'0 9'0
VQ
Z'O
O'O
co
O
.0.
•Q
C
.8
o
_ en
^ co
CM
\
O'l
CO
O)
CO
1X5
_ CO
o
>
•*=
o
D
T3
O
O
CC
a:
o
O)
CO
CM
oS
CO
O)
CO
1X5
_ CO
o
o
o
'O 9'Q
VO
Z'O
O'O
O'l
'O 9'Q YQ Z'O
r
O'O
1-6
-------
c
o
I
o
o>
CO
CM
CO
o>
_ CO
_ CO
o
c
s
o
cc
C
-ffi
00
o
o
o
o>
CO
CM
CO
O)
_ CO
_ CO
O
>,
'>
•i=
o
o
o
O'l
8'0 9'0
V'Q
Z'Q
O'O
O'l
'0 9'0
V'Q
Z'Q
O'O
_ CO
o
O)
CO
CM
CO >^
05 -
CO
10
_ CO
O'l
8'0 9'0
VQ
Z'Q
O'O
O'l
'0 9'0
VQ
Z'Q
O'O
o
_ en
^ co
CM
CO
a>
CO
1X5
_ CO
o
o
O
E
o
c
.8
5
o
o
c
-ffi
00
o
a>
CO
CM
- oS
o
CO
a>
CO
1X5
_ CO
o
o
O'l
Q'O 9'Q
VQ
Z'Q
O'O
O'l
'O 9'Q VQ Z'Q
r
O'O
J-7
-------
co
C
Q
o
o>
CO
CM
CO
O)
CO
1X5
" CO
C/D p*
a §
t ^
o
O
o
O)
CO
CM
- oS
CD
O)
CO
1X5
o
>,
'>
•i=
o
73
O
O
O'l Q'O
9'0
V'Q
Z'Q
O'O
O'l
Q'O 9'0
V'Q
I
Z'Q
O'O
o
O)
CO
CM
CO
O)
CC
•8
73
C
o
CO (•)
1X5 U
_ CO
o
o>
co
CM
oo
CO
CO
O)
CO
1X5
_ CO
_0
o
o
O'l 8'0
9'0
VQ
Z'Q
O'O
O'l
8'0 9'0
VQ
Z'Q
O'O
3
.5
co
I
O
_ CD
^ CO
CM
_ CO
o
CO
a>
C
CO (§
in *—'
co
•8
o
C
o
.55
o
a>
co
CM
- oS
CO
a>
co
1X5
_ CO
o
o
O
O'l
'Q 9'Q
VQ
Z'Q
O'O
O'l
8'0 9'0
\
VO
\
Z'Q
Q'Q
J-8
-------
CO
E
o>
c
o
c
o>
S-j
CO
o
o>
CO
CM
CO
o>
_ CO
_ CO
co
=!- C
t 1
« o
o
O
O
o
O)
CO
CM
CO
O)
O
&
tj
fS o
_ CO
O'l
8'0 9'0 V'Q Z'Q
O'O
n
O'l.
8'0
9'0
V'Q
Z'Q
O'O
cc
I
-2
"CD
O
O)
CO
CM
oo -p-
<° I
CO
O)
CO
1X5
_ CO
73
O
O
o
o>
CO
CM
CO
O)
CO
1X5
_ CO
73
O
O
O'l.
8'0 9'0 VQ Z'Q
O'O
O'l
8'0
9'0
VQ
Z'Q
O'O
co
3
s
o
o>
CO
CM
- oS
CO
O)
CO
1X5
_ CO
\
O'l
\ \
8'0 9'0
o
o
O
o
&
.0.
Q
o
O)
CO
CM
- oS
CO
O)
CO
1X5
_ CO
o
o
O
V'Q
Z'Q
O'O
O'l
I I
8'0 9'0
VQ Z'Q
O'O
J-9
-------
3
I
O
CD
o
O)
CO
CM
oS
CD
CD
O)
CO
1X5
O
&
tj
O
O
cc
o
O)
CO
CM
- oS
CO
O)
_ CO
o
>,
'>
•i=
o
o
O
O'l 8'0 9'0
Z'O
O'O
\
O'l
'0 9'0
VQ
Z'Q
O'O
•£.
Q.
o
c
.5
1
O
O)
CO
CM
OO
CO
CO
O)
CO
1X5
£ ^
^ I
f o
^ o
> "C
1 8
o
o
-C
cc
o
O)
CO
CM
oo
CO
CO
O)
CO
1X5
_ CO
_0
o
o
O'l 8'0 9'0
V'Q
Z'O
O'O
O'l
'0 9'0
V'Q
Z'O
O'O
-s
I
o
_ c>
^
c^t
\
O'l.
o
CO
O)
CO
1X5
_ CO
o
O
1
cc
-Q
E
CC
O
o
O)
CO
CM
- oS
o
CO
O)
CO
1X5
_ CO
o
o
''0 9'0
V'O
Z'O
O'O
O'l
'0 9'0 V'O Z'O
r
O'O
J-10
-------
C
o
_ en
^ co
CM
CO
o>
co
1X5
_ CO
| 5
(/} ^
a E
•t .c
I
O
O
o
o>
CO
CM
- S
CD
O)
CO
1X5
o
o
D
73
C
O
O
O'l
8'0 9'0 VQ Z'Q
O'O
\
O'l
8'0
9'0
V'Q
Z'O
O'O
-
co
o
O)
CO
CM
CO
O)
- CO
73
O
O
C
-ffi
I
o
O)
CO
CM
CO
O)
CO
1X5
_ CO
73
O
O
O'l
8'0 9'0
3'0
O'O
O'l
8'0
9'0
VQ
Z'O
O'O
cc
E
o
_ en
^ co
CM
O'l
\ I
Q'O Q'O
CO
a>
CO
1X5
_ CO
o
73
O
O
E
.
o
a>
co
CM
- cS
CO
a>
CO
1X5
_ CO
o
73
O
O
VQ
Z'O
O'O
\
O'l
\ \
Q'O Q'O
VO
O'O
J-ll
-------
co
I
o
O)
CO
CM
oS
CD
CD
O)
CD
1X5
o
o
D
73
C
O
O
.2
co
O'l
8'0 9'0
V'Q
Z'Q
O'O
\
O'l
O
O)
CO
CM
- oS
CO
O)
_ CO
o
>,
'>
•i=
o
O
O
'0 9'0
V'Q
Z'Q
O'O
co
..
s
o
O)
CO
CM
oo -p-
w I
CO
O)
73
C
o
CO (•)
1X5 U
O
CO
CM
oo
CO
CO
O)
CO
1X5
_ CO
_0
o
O
O'l
8'0 9'0
VQ
Z'Q
O'O
O'l
'0 9'0
VQ
Z'Q
O'O
co
-------
cc
QQ
o
_ en
^ co
CM
CO
O)
CO
1X5
_ CO
,0 CO
E ^
>, -<5
•t O
O
O
O
o
O)
CO
CM
- cS
CD
O)
CO
1X5
O
o
D
73
C
O
O
O'l
8'0 9'0 V'Q Z'Q
O'O
\
O'l
8'0
9'0
V'Q
Z'Q
O'O
cc
o
O)
CO
CM
co -p-
<° I
CO
O)
_ co
73
O
O
CO
O
.g.
s
o
Q
o
O)
CO
CM
CO
O)
CO
1X5
_ CO
73
O
O
O'l.
8'0 9'0 VQ Z'Q
O'O
O'l
8'0
9'0
VQ
Z'Q
O'O
o
_ en
^ co
CM
O'l
\
8'0
o
CO
a>
o
73
C
CO (§
1X5 u
.co
C
O
3
_ CO
o
a>
co
CM
- cS
CO
a>
CO
1X5
_ CO
o
73
O
O
9'0
V'Q
Z'Q
O'O
\
O'l
\ i
8'0 9'0
VQ Z'Q
O'O
J-13
-------
I
O
co
o
_ en
^ co
CM
CO
O)
CO
1X5
_ CO
o
O
O
I
O
o
o>
CO
CM
CO
O)
_ CO
_ CO
O
>,
'>
•i=
o
o
o
O'l
8'0 9'0
VO
Z'O
O'O
O'l
'0 9'0
VO
Z'O
O'O
.05
3
c
CO
CM
CO
O)
_ CO
_ CO
73
O
O
C
cc
-J
o
o>
co
CM
CO >^
05 -
"o
D
73
C
o
CO (•)
10 U
_ CO
O'l
8'0 9'0
VQ
Z'Q
O'O
O'l
'0 9'0
3'0
O'O
-2
05
£
S
o
_ en
^ co
CM
\
O'l
CO
O)
_ CO
o
f 8
= .o
o *C
73
C
CO (§
1X5 u
O
O)
CO
CM
- oS
CO
O)
CO
1X5
_ CO
o
73
O
O
8'0 9'0
VO
Z'O
O'O
\
O'l
'O 9'Q
VO
O'O
J-14
-------
.
o
O)
CO
CM
CO
O)
O
&
tj
S O
1
-
UJ
O'l
8'0 9'0
V'Q
Z'Q
O'O
\
O'l
o
O)
CO
CM
- oS
CD
O)
CO
1X5
O
>,
'>
•i=
o
73
O
O
'0 9'0
V'Q
Z'Q
O'O
co
3
E
o
s
.o
ff
o
O)
CO
CM
| |
tj
73
C
o
CO (•)
1X5 U
CC
2
.2
cc
_ CO
o
O)
CO
CM
CO >^
05 -
"o
D
73
C
o
CO (•)
10 U
_ CO
O'l
8'0
9'0 VQ
Z'Q
O'O
O'l
'0 9'0
VQ
Z'Q
O'O
oo
\
O'l
o
_ en
^ co
CM
_ CO
O
CO
O)
o
73
c
CO (§
1X5 u
I
t
CC
•C
O
o
O)
CO
CM
- oS
CO
O)
CO
1X5
_ CO
o
73
O
O
Q'O 9'Q
VQ
Z'Q
O'O
O'l
'O 9'Q VQ Z'Q
O'O
J-15
-------
•2
I
I
Q
o
O)
CO
CM
- oS
CD
O)
CD
1X5
o
.35
It
= ^
73 -C
O
O
O
O)
CO
CM
- oS
CO
O)
_ CO
o
>,
'>
•i=
O
73
O
O
O'l 8'0 9'0
V'Q
Z'Q
O'O
O'l
'0 9'0
Z'Q
O'O
co
3
o
O)
CO
CM
oo -p-
W I
CO
O)
CO
1X5
_ CO
CO
3
E
o
c
o -^
73
O
O
o
O)
CO
CM
CO >^
05 -
CO
10
_ CO
O'l 8'0 9'0
VQ
Z'Q
O'O
O'l.
'0 9'0
VQ
Z'Q
O'O
.><
1
^
o
_ en
^ co
CM
\
O'l
_ CO
o
CO
a>
c
CO (§
in *—'
I
o
a>
co
CM
- oS
o
CO
a>
CO
1X5
_ CO
o
o
'O 9'Q
VQ
Z'Q
O'O
O'l
'O 9'Q VQ Z'Q
r
O'O
J-16
-------
co
3
•c
o
o
cc
o
a>
co
CM
oS
CD
CD
O)
CO
1X5
O
O
O
-2
[gS
I
.i>
-2
•3
O'l
8'0 9'0
V'Q
Z'Q
O'O
\
O'l
o
O)
CO
CM
CO
O)
_ co
o
>,
'>
•*=
o
o
o
'0 9'0
V'Q
Z'Q
O'O
I
a
o
o
o>
CO
CM
oo -p-
<° I
CO
O)
CO
1X5
o
o
cc
3
o
-8
o
o
o
O)
CO
CM
oo
CO
CO
O)
CO
1X5
_ CO
_0
o
o
O'l
8'0 9'0
VQ
Z'Q
O'O
O'l
'0 9'0
VQ
Z'Q
O'O
c
c
cc
\
O'l
o
a>
co
CM
CO
a>
CO
1X5
_ CO
o
CC
3
.0.
o
O
Q'O Q'O
V'Q
Z'Q
O'O
n
O'l.
o
a>
co
CM
- oS
CO
a>
co
1X5
_ CO
'O 9'Q VQ Z'Q
r
O'O
o
o
O
J-17
-------
CD
.E
-c
O
CO
CM
00
CD
CD
O)
CD
in
_ CD
O
C/3
T3
O
O
§
t
o
o
5
CO
CM
00 C>
(o E
o
CD
CD
in
_ CD
i
i
8'0 9'0
VQ
i
Z'O
Q'Q
O'l
Q'Q
\
Q'Q
\
P'Q
\
Z'O
r
O'O
73
O
O
D
CO
CM
00
CD
CD
CD
in
_ CD
o
C/3
_>,
\>
"o
73
O
O
00
o
I
O
O
CO
CM
00 C>
(o E
o
w
CD
CD
in
_ CD
i
i
8'0 9'0
VQ
i
Z'Q
Q'Q
\^
O'l
Q'Q
\
Q'Q
\
P'Q
\
Z'O
r
O'O
T3
O
O
CD
00
CO
CM
00
CD
CD
CD
in
_ CD
73
O
O
CD
8
CD
E
CD
\
\
8'0 9'0
VQ
\
Z'Q
O'O
J-18
o
O)
CO
CM
00 /^'
(o E
o
CD
CD
in
_ CD
73
O
O
O'l
\
8'0
\
9'0
\
P'Q
\
Z'O
Q'Q
-------
CO
CM
3
CD
CD _>>
,— ^>
"o
73
CD
IT)
CD
O
O
\
\
O'l
8'0 9'0
T
O'O
CD
00
CD
E
CD
Q
CO
CM
CD
CD _>>
.5— ^>
"o
73
CD
IT)
CD
O
O
i
i
8'0 9'0
T
O'O
J-19
-------
500 1000 2000
Conductivity (uS/cm)
&EPA
United States
Environmental Protection
Agency
Office of Research and Development
National Center for Environmental Assessment
Cincinnati, OH 45268
Official Business
Penalty for Private Use
$300
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT NO. G-35
Printed on 100% recycled/recyclable paper
with a minimum 50% post-consumer
fiber using vegetable-based ink.
------- |