Full text of DOI:10.1002/etc.5620210603

Environmental Toxicology and Chemistry, Vol. 21, No. 6, pp. 1112–1124, 2002
Printed in the USA
0730-7268/02 $9.00
ϩ .00
DETERMINING PROBABLE CAUSES OF ECOLOGICAL IMPAIRMENT IN THE LITTLE
SCIOTO RIVER, OHIO, USA: PART 1. LISTING CANDIDATE CAUSES AND
ANALYZING EVIDENCE
S
USAN B. NORTON,*† SUSAN M. CORMIER,‡ GLENN W. SUTER II,§ BHAGYA
AVID LTFATER OUNTS
†U.S. Environmental Protection Agency, National Center for Environmental Assessment (8623-D), 1200 Pennsylvania Avenue,
Washington, DC 20460
SUBRAMANIAN,‡ EDITH LIN,‡
D
A
,
࿣
and BERNIE
C
࿣
‡U.S. Environmental Protection Agency, National Exposure Research Laboratory, Cincinnati, Ohio 45268
§U.S. Environmental Protection Agency, National Center for Environmental Assessment, Cincinnati, Ohio 45268
࿣
Ohio Environmental Protection Agency, Ecological Assessment Section, Groveport, Ohio 43125, USA
(
Received 31 January 2001; Accepted 25 November 2001)
Abstract—The Little Scioto River in north-central Ohio, USA, is considered to be biologically impaired based on the results of
fish and invertebrate surveys. The causes for these impairments were evaluated by means of a formal method. Two of the impairments
identified on the stream reach were characterized in detail to support the causal assessment. A list of six candidate causes was
developed that included habitat alteration, polycyclic aromatic hydrocarbon contamination, metals contamination, low dissolved
oxygen, ammonia toxicity, and nutrient enrichment. Evidence for the causal evaluation was developed with data from the site that
associated each candidate cause with the biological responses. Evidence was also developed that drew on data from other locations
and laboratory studies, including comparisons of site exposures with screening values and criteria. The formal method increased
the transparency of the assessment; candidate causes were clearly listed and the pathways by which they may have produced effects
were shown. Analysis of the evidence maximized the utility of available data, which were collected as part of monitoring and
research programs rather than to specifically support a causal assessment. This case study illustrates how the stressor identification
method can be used to draw conclusions from available data about the most likely causes of impairment and to show what additional
studies would be useful.
Keywords—Biological assessment
Ecoepidemiology
Ohio, USA
Macroinvertebrates
Fish
INTRODUCTION
nating impossible causes, and comparing the strength of evi-
dence supporting the remaining causes [1].
As the use of biological surveys and monitoring increases,
interest has increased in improving methods for determining
the causes of observed biological impairments. Although caus-
es of impairments are often obvious to trained investigators,
a formal method can improve the ability to communicate the
evidence and logic that formed the basis of conclusions about
cause. A formal method provides additional benefits, including
minimizing opportunities for lapses in logic, improving the
likelihood that obscure causes are identified, strengthening the
defense of the results, and increasing confidence when un-
dertaking costly remediation.
This case study is not meant to illustrate an ideal causal
assessment. The data used were those that were available from
various studies, none of which were designed to support the
stressor identification process and none of which were as ex-
tensive or intensive as we would have liked. However, the data
are typical of the early phases of many investigations. This
case study illustrates how the stressor identification method
can be used to draw conclusions from available data about the
most likely causes of biological impairment and to show what
additional studies would be useful.
The case study described in this paper and its companion
[1] followed a process described in the U.S. Environmental
Protection Agency’s stressor identification guidance [2] and
Suter et al. [3]. This paper describes the beginning phases of
a causal evaluation prompted by a biological impairment de-
tected through monitoring of fish and macroinvertebrate as-
semblages. It presents the background of the investigation, the
development of the list of candidate causes, and the organi-
zation and analysis of available evidence. The companion pa-
per then uses the evidence to characterize causes by elimi-
MATERIALS AND METHODS
Study area
The Little Scioto case study involves a 15-km reach of a
river in north-central Ohio, USA, near Marion. The Little Sci-
oto River drains primarily farmland in the northeastern quad-
rant of the Eastern Corn Belt Plains ecoregion (Fig. 1) [4].
The soils in this area are glacial till overlying limestone, do-
lomite, and shale bedrock. The water table has been lowered
in much of the watershed by extensive use of tile drainage in
crop fields.
Many point and nonpoint sources of pollutants are asso-
ciated with the Little Scioto River (Fig. 1). Point sources in-
clude a wastewater treatment plant and combined sewer over-
flows that enter between 9.5 km and 10.5 km, respectively,
upstream of the confluence with the Scioto River (Rkm 9.5
* To whom correspondence may be addressed
(norton.susan@epa.gov).
Presented at the 20th Annual Meeting, Society of Environmental
Toxicology and Chemistry, Philadelphia, Pennsylvania, November
14–18, 1999.
1112

Listing and analyzing causes for Little Scioto River, OH
Environ. Toxicol. Chem. 21, 2002
1113
and 10.5). Nonpoint sources include runoff from agricultural
land uses and from the city of Marion. Releases also may
originate from several industrial areas, including an abandoned
wood treatment plant, a landfill, an appliance plant, and a rail
facility [5]. Finally, the stream was channelized in the early
1900s starting from Rkm 15 and continuing downstream to
the confluence with the Scioto River.
Data from the case
Information useful for causal evaluations can come from
the case at hand, from other similar cases, and from biological
knowledge [3]. Data from the Little Scioto included chemical
analyses (sediment, water, and fish tissue), qualitative rankings
of physical habitat quality, and biological surveys. The data
were collected as part of Ohio Environmental Protection Agen-
cy (Ohio EPA) monitoring programs, and U.S. Environmental
Protection Agency methods development research (Table 1).
Data from 1992 provided the principal basis of the evaluation
because they were collected during the same year as the bi-
ological data that were used to establish the impairment. In
addition, the chemistry and habitat data were collected at lo-
cations that were considered spatially equivalent to the loca-
tions where the biological data were collected. Results from
all 1992 sampling efforts are shown in Ohio EPA [5]. Relevant
data also were available from sampling conducted during 1987,
1991, and 1998 [6–8]. These additional years of data were
used to verify that the values observed in 1992 were not atyp-
ical, to evaluate the extent of year-to-year variability that might
be expected, and to provide insights into uncertainties and gaps
in the 1992 data. All years of data are shown in the U.S.
Environmental Protection Agency stressor identification guid-
ance document [2].
Fig. 1. Map of the Little Scioto River, Ohio, USA, showing sites
where fish were sampled. Approximate locations of significant phys-
ical features, tributaries, and point-source inputs are noted. The lo-
cations where impairments A and B were observed are also shown.
The small inset shows the location of the study area in the state of
Ohio. CSO
ϭ combined sewer overflow.
Biological surveys. The fish and macroinvertebrate assem-
blages both were surveyed at seven locations, by means of
methods described by the Ohio EPA [9–11]. Macroinvertebrate
communities were sampled with Hester–Dendy artificial sub-
Table 1. Summary of sampling locations for 1992 data collected in the Little Scioto River, Ohio, USA [5]
Site designation and sample location
(km upstream from confluence with Scioto River)
Upstream
site
Number of samples
Sample medium
Parametersa
Site A Site B
Downstream sites
per location
Fish assemblage
Macroinvertebrate
assemblage
IBI and MIWB metrics
ICI metrics
14.9
14.9
12.7
12.7
10.5
10.5
9.2
9.2
7.1
7.1
4.4
3.4
0.48
0.65
2–3
1
Sediment
Volatile chemicals semivol-
atile chemicals, metals
and cyanide, PCBs and
pesticides
15.3
12.7
10.5
9.4
7.1
4.4
0.65
1–3
Water
Metals, phenolic com-
pounds, ammonia, ni-
trates and nitrites, total
phosphorus, hardness,
and biochemical oxygen
demand
14.9
12.7
10.5
9.4
7.1
4.4
0.65
1
Water
Dissolved oxygen
14.8
12.7
12.7
9.4
9.2
7.1
7.1
4.4
0.58
0.48
Measured continu-
ously over 3 d
1
2–5 composite sam-
ples
Physical habitat
Fish tissue
QHEI metrics
14.9
14.9
10.5
10.5
4.4
4.4
Pesticides, PCBs, metals,
semivolatile compounds,
and percent lipid
White sucker bile
PAH metabolites
12.7
10.5
9.2
7.1
4.4
0.48
5–7
a IBI
biphenyl; QHEI
ϭ
index of biotic integrity; MIWB
ϭ
modified index of well being; ICI
ϭ invertebrate community index; PCB ϭ polychlorinated
ϭ
qualitative habitat evaluation index; PAH
ϭ
polycyclic aromatic hydrocarbon.

1114
Environ. Toxicol. Chem. 21, 2002
S.B. Norton et al.
less steel scoops. Samples from 45 to 60 cm were taken with
Teflon௡ core samplers. Sediment samples were analyzed for
metals and cyanide, volatile organic chemicals, semivolatile
organic chemicals (including polycyclic aromatic hydrocar-
bons [PAHs]), pesticides, and polychlorinated biphenyls. The
full list of analytes is provided in Ohio EPA [5]. Volatile
organic chemicals, pesticides, and polychlorinated biphenyls
were generally not detected in the upper 15 cm. The exceptions
were at Rkm 9.4 where 4,4Ј-dichlorodiphenyldichloroethane
and 4,4 -dichlorodiphenyldichloroethylene were detected at
Ј
concentrations considered to be elevated over background con-
centrations. Results for selected PAHs and metals are shown
in Table 2. In addition, field investigators noted that a black
material with a creosote odor occurred just below the surface
layer of sediments from site B to Rkm 7.1 [5].
Water chemistry. Methods for the collection and analysis of
surface water samples are described in Ohio EPA [5,12]. One
water sample was taken at each location. Water samples were
analyzed for metals, phenolic compounds, ammonia, nitrates and
nitrites, total phosphorus, hardness, and biochemical oxygen de-
mand (BOD). The full list of analytes is provided in Ohio EPA
[5]. Dissolved oxygen measurements were taken at six locations
over a 3-d period with Datasonde continuous monitors. Water
hardness ranged from 278 to 329 mg/L CaCO3. Metals were
detected in the water column at only a few locations. Copper
Fig. 2. Values of the fish index of biotic integrity (IBI) (A) and benthic
macroinvertebrate invertebrate community index (ICI) (B) in the Little
Scioto River, OHIO, USA, in 1992 [5]. WWH
MWH modified warm-water habitat.
ϭ warm-water habitat;
was detected at site B at a concentration of 15
detected at 3 g/L at site B and one of the downstream sites.
Zinc was detected only at sites downstream of site B at concen-
trations between 12 and 18 g/L. Concentrations of other selected
␮g/L. Lead was
ϭ
␮
␮
strate samplers, deployed for a six-week period. Fish were
sampled by electrofishing. The upstream site at Rkm 2 was
sampled in two passes over 250 m by a wading method. All
other sites were sampled by electrofishing from a boat in three
passes over 500 m of stream [11].
The Ohio EPA used the biological survey data to calculate
the index of biotic integrity (IBI) and invertebrate community
index (ICI) [9–11]. Average IBI and ICI values are shown in
Figures 2A and B, respectively. A subset of metrics, described
below, was used to characterize the impairments.
parameters are shown in Table 3.
Physical habitat quality. The quality of the habitat was
characterized with the qualitative habitat evaluation index [13].
The qualitative habitat evaluation index incorporates ordinal
scores of habitat attributes including substrate type and quality;
in-stream cover type and amount; channel morphology; ripar-
ian width and quality and bank erosion; pool and riffle char-
acteristics including depth, current, pool morphology, substrate
stability, and riffle embeddedness; and stream gradient. The
scores for channel morphology, degree of siltation, and degree
of substrate embeddedness are particularly relevant for this
case and are shown in Table 4. Higher scores indicate higher
quality sites.
Sediment chemistry. Methods for the collection and analysis
of sediment samples are described in Ohio EPA [5,12]. Fine-
grained sediment samples were collected in the upper 15 cm
of bottom material at each location with decontaminated stain-
Table 2. Average sediment concentrations (mg/kg dry wt) of selected polycyclic aromatic hydrocarbons and metals [5]. Samples were taken
from the upper 15 cma
Downstream site (river km)
Upstream
Chemical
site
Site A
Site B
9.4
7.1
4.4
0.65
Anthracene
NDb
ND
ND
ND
ND
ND
ND
7.34
7.44
ND
ND
ND
ND
ND
ND
ND
13.6
17.2
19.1
79
ND
8.2c
49.5
14.8c
16.5
8.2c
ND
208
79
172
173
0.33
27.1
16.5
11.2
15.8
20.8
37.6
7.0
60.9
56
84.6
7.9
6.9
4.9
7.2
9.9
13.5
4.0
302
76.8
93.4
226
0.79
ND
2c
3.3
15.8
6.9
11.5
ND
22.4
ND
48.6
24.5
38
Benzo[
Benzo[ghi]perylene
Benzo[ ]pyrene
a]anthracene
ND
ND
1.6c
ND
ND
71.2
42.4
108
408
0.12
a
Chrysene
Fluoranthene
Fluorene
Cr
Cu
Pb
Zn
Hg
12.1
3.06
ND
141
0.24
96.8
ND
ND
a Polycyclic aromatic hydrocarbons shown were either significantly correlated with biological responses, or exceeded sediment quality guidelines.
b ND
Not detected.
c Estimated values (below quantitation limit).
ϭ

Listing and analyzing causes for Little Scioto River, OH
Table 3. Concentrations (mg/L) of chemical in water [5]. All data are from 1992 except as noted
Downstream sites (river km)
Environ. Toxicol. Chem. 21, 2002
1115
Upstream
site
Chemical
Site A
Site B
9.4
7.1
4.4
0.65
BODa
Nitrate and nitrite
Ammonia
total phosphorus
Dissolved oxygen (minimum)
1
1
2.3
4.7
8.1
4.2
6.6
3.5
4.5
2.1
1.8
3.0
2.0d
2.2
1.22
NDb
0.06
8.8
1.44
ND
0.07
5.7
0.81
0.12
0.09
NAc
1.9d
4.47
0.58
1.34
4.4
1.16
2.17
4.2
1.44
1.96
4.3
NAd
2.8d
4.2d
3.2d
2.5d
a BOD
ϭ biochemical oxygen demands.
not detected.
no data were collected.
b ND
c NA
ϭ
ϭ
d Data collected in 1987 by using continuous monitors over 3 d [6].
Fish tissue chemistry. Fish were collected at three locations
for whole-body analyses of pesticides, polychlorinated biphe-
nyls, metals, semivolatile compounds, and percent lipid (Table
1) [5,14,15]. Zinc was detected in common carp (Cyprinus
carpio) from the upstream site (79.6 mg/kg), site B (68.3 mg/
kg), and at Rkm 4.4 (15.8 mg/kg). Lead was detected in white
suckers (Catostomus commersoni) at site B (81.4 mg/kg) and
Rkm 4.4 (0.34 mg/kg). Bile samples from white suckers were
analyzed for benzo[a]pyrene and naphthalene-type metabo-
lites; results are shown in Figure 3.
Experiments. Ireland et al. [16] performed studies in which
Ceriodaphnia dubia was experimentally exposed in situ cham-
bers to epibenthic water in the Little Scioto at the upstream
site and at Rkm 7.1.
Defining the impairment
The term impairment is used in two senses in this paper.
First is the official definition used by the state of Ohio to
determine that a stream is not achieving its designated use.
Impairment under this official definition provides the impetus
for performing a causal analysis; it means that a community
is in an unacceptable condition. Second, within officially im-
paired waters, it is necessary to identify impairments in terms
of distinct and reasonably consistent sets of biological re-
sponses that are likely to have an identifiable cause. The causal
analysis is facilitated by characterizing these distinct impair-
ments in detail.
In Ohio, biological impairment is officially defined by stan-
dard multimetric indices including the IBI, the ICI, and the
modified index of well being [11]. These indices were used
to identify the reaches of the Little Scioto River that were
considered to have impaired aquatic assemblages in compar-
ison to the criteria for warm-water habitat (WWH) or modified
warm-water habitat (MWH) designated uses.
The Little Scioto River is considered WWH above site A
(Rkm 12.7), a designation shared by the majority of Ohio’s
rivers and streams [17]. This designation is narratively defined
as supporting a balanced, reproducing aquatic community.
Quantitatively, the minimum criteria required to be in attain-
ment of WWH standards are defined as the 25th percentile
values of reference condition scores for a given index, site
type, and ecoregion.
The Little Scioto River is classified as MWH at and below
site A (see Fig. 1). The MWH criteria are based on comparisons
to a different reference condition than are used for the WWH
criteria [17]. The MWH designation is a nonfishable aquatic
life use, and is designed to protect streams that have been too
physically impacted, or modified, to meet WWH standards.
Modified warm-water habitat streams are unlikely to recover
Data from other similar cases and biological knowledge
Inferences concerning causes may be supported by data
from laboratory studies or studies at other sites of the effects
of the agents that comprise the candidate causes. For the chem-
ical contaminants, we relied on numeric water quality criteria,
benchmark values, summaries of aquatic toxicity data, and
studies of similarly polluted rivers in Ohio. For the habitat
factors, we relied on published reviews. These sources con-
stitute evidence if they provide exposure–response relation-
ships that can be related to exposures at the site or to the
responses that constitute the distinct impairments.
Data analyses
The data from the Little Scioto and other cases were an-
alyzed to define the impairment for the causal evaluation, to
develop a list of candidate causes, and to examine associations
between candidate causes and biological responses. This body
of evidence was then used to characterize causes, as described
in the companion paper [1].
Table 4. Selected physical habitat quality scores [5]
Downstream sites (river km)
Upstream
Parameter
site
Site A
Site B
9.4
7.1
4.4
0.65
Channel
17
2
2
10
1
1
10
1
1
10
1
1
10
1
1
10
1
1
7
1
1
Substrate embeddednessa
Siltb
a Levels of embeddedness were scored from most to least: 1
b Levels of silt were scored from most to least: 1
heavy; 2
ϭ
extreme; 2
ϭ
moderate; 3
ϭ
ϭ
low; 4
free.
ϭ none.
ϭ
ϭ moderate; 3 ϭ normal; 4

1116
Environ. Toxicol. Chem. 21, 2002
S.B. Norton et al.
Fig. 4. Changes in fish and macroinvertebrate metrics in the lower 15
km of the Little Scioto River, Ohio, USA, 1992. Normalized values
were calculated by dividing the value at the each site by the highest
Fig. 3. Bile metabolites (␮g/mg protein) measured in white suckers
from the Little Scioto River, Ohio, USA, in 1992. Median levels of
polycyclic aromatic hydrocarbon (PAH) metabolites below river ki-
lometer 10.5 were as much as four times greater than the exposure
criteria (dashed horizontal line), which are upper limits of concentra-
tions considered background for the state of Ohio. The number of
fish sampled is shown above each bar. Vertical lines are standard
value for all sites. IBI
community index; DELT
mors; U upstream site; A
ϭ
index of biotic integrity; ICI
deformities, eroded fins, lesions, or tu-
site a; B site B.
ϭ invertebrate
ϭ
ϭ
ϭ
ϭ
errors. A
naphthalene.
ϭ site A; B ϭ site B; BaP ϭ benzo[a]pyrene; NAPH ϭ
Development of the list of candidate causes
The list of candidate causes and associated conceptual mod-
els were developed by examining the available data on stress-
ors and sources, by discussing the case with Ohio EPA biol-
ogists and other experts, and by applying our professional
experience of stressors and their effects.
sufficiently to meet WWH designation. Consequently, MWH
criteria are typically lower than WWH criteria. Modified
warm-water habitat streams are able to support permanent as-
semblages of species tolerant of low DO, elevated ammonia
and nutrient levels, increased siltation, and poor quality hab-
itat.
Analyzing evidence
Suter et al. [3] discussed four types of associations that
provide evidence useful for causal evaluations: associations
between measurements of causes and effects, combining site
exposure data with effects data from elsewhere, associations
of measurements with mechanisms, and associations of the
mitigation of effects with mitigation of exposure. The follow-
ing paragraphs discuss these types of evidence.
Spatial associations between measurement of the candidate
causes and effects were analyzed with data from the 15-km
stream reach. Although we sought to use biological and chem-
ical data from the same locations, in some cases, chemical
measurements were recorded at a location that did not exactly
coincide with the location of biological assessment (e.g., Rkm
15.3 and Rkm 14.9, for sediment chemistry and fish assem-
blage, respectively). Table 1 shows sampling locations that
were considered equivalent by Ohio EPA biologists. Although
sediment samples were taken at depth, we used only sediment
chemistry values collected from the top 15 cm.
We used data from the site to analyze the relationship be-
tween candidate causes and response. The first analysis was
limited to the measurements taken at the upstream site, site
A, and site B. We evaluated whether any measurement asso-
ciated with a candidate cause was detected at sites A or B and
whether measurements indicated lower environmental quality
as compared with the closest upstream location. Because each
comparison used only two values, we used simple differences
to conclude whether environmental quality improved or de-
clined.
Data from all seven sites were used to correlate measure-
ments associated with candidate causes and the biological re-
sponses. Spearman’s rank correlations were used because many
variables were not normally distributed. Chemistry concen-
trations that were below detection limits were given the lowest
To characterize distinct impairments, the responses of the
biological assemblages were analyzed further by evaluating
the changes in individual metrics at different locations on the
Little Scioto. If the responses of an assemblage at different
locations are distinct, then different causes are likely to be
operating. Metrics selected for this analysis exhibited sensi-
tivity to changing conditions along the stream. Because a short-
er length of stream was sampled at the upstream location, we
selected fish metrics that were expressed as a proportion of
the fish sampled, or could be normalized to distance sampled.
For fish (Fig. 4), we used the weight of fish normalized to a
1 km distance (relative fish weight) and the percent of fish
having deformities, eroded fins, lesions, or tumors (DELT
anomalies). For invertebrates, we used the percentage of in-
dividuals that were mayflies (percent mayflies), and the per-
centage of individuals that were from taxa considered to be
tolerant of stress (percent tolerant invertebrates). Metrics not
included were based on only a few organism counts (e.g.,
number of sunfish species), were unaltered over the stream
reach (e.g., percentage of intolerant fish species), or were
known to be correlated with confounding variables (e.g., the
percentage of pioneering fish species is highly correlated with
stream size). To simplify the analyses, we did not include
metrics that were highly correlated with those already selected.
Distinct impairments were identified by comparing the val-
ue of individual metrics at adjacent locations, and by evalu-
ating the general trend of metrics over the stream reach. Im-
pairments were considered to be different if the metric values
changed substantially, or if an overall trend changed direction.
The determination was based on inspection of plots and pro-
fessional judgment.

Listing and analyzing causes for Little Scioto River, OH
Environ. Toxicol. Chem. 21, 2002
1117
Table 5. Summary of the distinct impairments considered in the Little
Scioto River, Ohio, USA
rank. Because our objective was to conservatively include
causes, we used a
p value of Ͻ0.10 to conclude that a cor-
relation was significantly different from that expected from
random processes.
Impairment A:
Change in
assemblage at site B in assemblage
Impairment B:
Additional changes
We compared site exposure data to available criteria and
screening values. Sediment screening values were available
for PAHs and all of the metals except for mercury, although
not for the community parameters or species of greatest interest
for this study. Sediment effect concentrations are expressed as
threshold effect level (TEL) and probable effect level (PEL)
[18]. The interpretation is that toxicity to test organisms (Hy-
alella azteca and Chironomus riparius) would rarely occur at
sediment concentrations below the TEL but would frequently
occur above the PEL. Both TELs and PELs were developed
from a large database consisting of chemistry data from field-
collected sediments and associated toxicity test results. Be-
cause most of the sediments in the database contained multiple
chemicals, these values should be interpreted with caution.
Sediment effect concentrations developed for H. azteca and
C. riparius were considered, but only the H. azteca values
were used because C. riparius was always less sensitive. We
used a modified toxic unit approach to account for exposure
to multiple chemicals; the ratios of sediment concentration to
PELs were added for the PAHs and for the metals.
relative to
upstream site
at site B relative
to site A
Response
Fish
Relative fish weight
DELTa anomalies
Invertebrates
% Mayflies
Increased
Increased
Decreased
Increased
Decreased
Increased
Decreased
Increased
% Tolerant taxa
a DELT
ϭ deformities, eroded fins, lesions, or tumors.
taminated or if the fish themselves contained the chemicals or
metabolites. As discussed above, data were available on fish
tissue concentrations of chemicals and metabolites. For PAHs,
we concluded that the pathway was complete if the PAH me-
tabolite data indicated that exposure had occurred. For metals,
fish tissue concentrations were available only at site B. Because
data were not available for food items, we conservatively con-
cluded that the pathway was complete if metals were found
in the sediments.
The analysis of the exposure pathway for nutrients was
straightforward; we assumed exposure occurred where nutri-
ents were detected in the water column. The DO pathway was
evaluated by examining colocation with precursors, such as
BOD.
Symptoms that occur in response to exposure to a particular
candidate cause can provide strong evidence of the mechanism.
Deformities, fin erosion, tumors, and lesions on fish are po-
tentially useful for this purpose. For example, some patholo-
gies have been linked with toxic substances (e.g., liver tumors
with PAHs) and others (e.g., skin lesions) with certain bacteria
[23,24]. Unfortunately, insufficient detail was available during
the case study to use anomalies to distinguish among these
stressors.
We compared aqueous ammonia and metal concentrations
with U.S. Environmental Protection Agency ambient water
quality criteria, which are based on laboratory tests measuring
effects on survival, growth, and reproduction [19,20]. The total
ϩ
ammonia criterion (which includes NH3 and NH4 ) varies with
pH. High concentrations of hydroxide ions (i.e., high pH) in-
ϩ
crease the dehydration of the ammonium ion (NH4 ) to the
more toxic unionized form (NH3). In 1998, pH values in the
Little Scioto ranged between 7.4 and 8.4 and seemed to be
independent of location [8].
We compared DO concentrations with Ohio EPA criteria.
Ohio’s criteria for minimum DO concentrations are 4.0 mg/L
and 3.0 mg/L for WWH and MWH streams, respectively [21].
Nutrient concentrations were compared with Ohio’s proposed
statewide criterion [22]. The criterion for nitrate and nitrite is
1.6 mg/L for wadeable, WWH streams in the Eastern Corn
Belt Plains having a drainage greater than 50 km2 and less
than 500 km2. For total phosphorus, the proposed state-wide
criterion for MWH is 0.28 mg/L.
The third type of evidence is based on analyses of causal
mechanisms and pathways. In the best case, available data
would document a stressor’s course from its sources and pre-
cursors, to the environmental medium, to contact or co-oc-
currence with organisms, into the organisms (when appropri-
ate), to distinct physiological responses observed in exposed
organisms. If a necessary step in the pathway can be shown
to be missing, the candidate cause can be eliminated from
consideration. In the Little Scioto, exposure pathways were
evaluated primarily by examining the colocation of the avail-
able data representing different steps of the pathways. If no
data were available for an intermediate step, the step may have
occurred.
RESULTS
Defining the impairments
Of the seven sites sampled in 1992, the highest IBI score
was 33 (out of a possible score of 60); it occurred at the
upstream site (Fig. 2A). This score translates to a fair ranking
according to WWH standards [17]. The remaining sites were
described as severely impaired, with IBI scores between 25
and 12 (the lowest possible IBI score) [5,17]. Similarly, the
ICI met WWH aquatic life use standards at the upstream site
with a score of 38 (Fig. 2B). The ICI score from site A down-
stream was below MWH aquatic life use standards.
Examination of the spatial distribution of the IBI, ICI, and
component metrics indicates that at least two distinct impair-
ments occurred (Table 5). Impairment A was measured at site
A (Rkm 12.7) where a marked drop in both the IBI and ICI
occurred relative to the upstream site. Relative fish weight and
DELT anomalies increased. For the invertebrates, the per-
centage of mayflies decreased, and the percentage of tolerant
invertebrates increased. Impairment B was observed at site B
(Rkm 10.5) and corresponded with an additional decrease in
both the IBI and the ICI. Compared with the upstream site,
the relative weight of fish and DELT anomalies increased. For
invertebrates, the percentage of mayflies decreased, and the
For the toxic chemical stressors, sediment concentrations
were available. The exposure pathway was assumed to be com-
plete for benthic invertebrates, which are in direct contact with
the contaminated sediments, but may also be exposed via water
or food to chemicals that have partitioned from the sediment
into these other media. For fish, the pathway could be shown
to be complete if water or food items were shown to be con-

1118
Environ. Toxicol. Chem. 21, 2002
S.B. Norton et al.
Fig. 5. A conceptual model of the six candidate causes for the Little Scioto River, Ohio, USA. Potential sources are shown in the topmost
rectangles. Potential stressors and interactions are located in ovals. Candidate causes are numbered 1 through 6. Note that some causes are
associated with more than one stressor or more than one step. The impairments are shown in the lowest rectangle. DO
BOD biochemical oxygen demand; PAH polycyclic aromatic hydrocarbon; UV ultraviolet; DELT deformities, eroded fins, lesions, or
tumors.
ϭ dissolved oxygen;
ϭ
ϭ
ϭ
ϭ
percentage of tolerant invertebrates increased. Compared with
the changes in metric values between the upstream site and
site A, the changes measured between sites A and B are dis-
tinctive enough to suggest that either additional or completely
different causes may be influencing the assemblage at site B.
In particular, fish weight increased between the upstream site
and site A, but then decreased again at site B. The other metrics
indicated that environmental quality declined further. Relative
to site A, at site B the DELT anomalies increased, and per-
centage of tolerant invertebrates increased; the percentage of
mayflies decreased. To analyze the different causes influencing
the assemblage at site B, we defined impairment B in terms
of the additional changes relative to site A.
The biological assessment data for the remaining locations
indicate that distinct effects also occur at the downstream lo-
cations. As discussed in the stressor identification guidance
document [2], midges in the tribe Tanytarsini disappear com-
pletely at Rkm 9.2, with a concurrent increase in the percentage
of Cricotopus spp. Further downstream, the pattern of im-
pairment remained generally similar, with the possibility of
intensification at Rkm 7.1 and some improvement in metric
scores occurring at Rkm 4.4 and Rkm 0.65. The causal analysis
of the impairment observed at Rkm 9.2 is discussed in the
stressor identification guidance document [2], but is not in-
cluded here for the sake of brevity.
would be expected to occur together. We separated toxic sub-
stances from nutrients or physical stressors, and further sep-
arated the toxic substances into chemical classes. Based on the
knowledge of the sources and effects, six candidate causes
were identified. A conceptual model of these candidates is
provided in Figure 5.
Habitat alteration. Habitat alteration, resulting from chan-
nelization, includes several interacting stressors. Channeliza-
tion can alter biological communities by changing the physical
structure of the stream and the flow characteristics of the water,
ultimately lowering DO, increasing siltation, and reducing sub-
strate complexity. This complex suite of stressors also includes
decreased woody debris, which reduces available substrate and
changes the energy source; altered flow characteristics; in-
creased erosion of banks; embedded substrates; increased
channel depth; loss of pools; and loss of riffles that oxygenate
water and transport sediment [25–28].
PAHs and Metals. Biological impairment also could have
been caused by toxic stress. Historically, the river has provided
a means of waste disposal for various industries, whose efflu-
ents have contained metals, PAHs, and creosote. Waste ma-
terials also may have been buried in the landfill downstream
of site B [5]. All are potentially toxic to aquatic life, and some
have the ability to bioaccumulate through the food web
[29,30]. Thus, two candidate causes emerge: PAH exposure
and metal exposure. Exposure of fish and macroinvetebrates
to these contaminants could occur through food, water, and
direct contact with sediment. Exposure to ultraviolet light en-
hances the toxicity of PAHs in sediments and water to both
fish and invertebrates [16,31–33].
List of candidate causes
As discussed in Suter et al. [3], the causal evaluation fo-
cuses on the stressors that may actually contact organisms
living in the stream, rather than the sources of those stressors.
These stressors include toxic chemicals and excessive nutrients
in the water column and sediment as well as physical stressors
such as smothering of sediments. To simplify the analysis, we
grouped stressors that emanate from the same source and
Ammonia toxicity. Ammonia is directly discharged into
streams by point sources [34,35]. Ammonia also can be formed
as the result of nutrient enrichment. When DO levels are low,
nitrates are reduced to ammonium ion. Some of the ammonium

Listing and analyzing causes for Little Scioto River, OH
Environ. Toxicol. Chem. 21, 2002
1119
Table 6. Available measurements relevant to each candidate cause
ion is converted to unionized ammonia, which is toxic to aquat-
ic organisms [34]. The proportion of total ammonia present as
unionized ammonia is a function of pH, and the proportion
increases as pH increases. Moreover, pH may rise during pe-
riods of high photosynthetic rates from bicarbonate depletion.
High concentrations of nutrients often lead to increased algal
growth rates, and the conversion of ammonium to unionized
ammonia is expedited [36].
Candidate cause
Habitat alteration
Relevant measurements
Channel score; embeddedness
subscore; siltation subscore;
dissolved oxygen
PAHa toxicity
PAH concentrations in sedi-
ments; bile metabolites in
fish
High BOD and low DO. Depletion of dissolved oxgyen
commonly occurs from organic enrichment [37]. Organic en-
richment is the most common cause of increased BOD [28].
Potential sources of excess organic matter within the study
area include a wastewater treatment plant and several com-
bined sewer outfalls, as well as upstream, nonpoint sources.
Organic matter also is produced by excess algal growth from
nutrient enrichment [36], and nocturnal respiration of algae
contributes further to DO depletion. Because no chlorophyll
Metal toxicity
Metal concentrations in sedi-
ments, water, and fish tissue
Ammonia concentrations in wa-
ter
Dissolved oxygen, biological
oxygen demand of water
Nitrate and nitrite concentra-
tions in water; total phospho-
rus concentrations in water
Ammonia toxicity
Low dissolved oxygen and high
biological oxygen demand
Nutrient enrichment
a PAH
ϭ polycyclic aromatic hydrocarbon.
a
or algal biomass data were collected in this study, the path-
way of DO depletion can be traced back only to BOD.
Nutrient enrichment. The sixth and final candidate cause is
a less extreme form of nutrient enrichment. Primary production
and organic matter loading to the sediments are increased, but
not enough to reduce DO. This can cause changes in fish and
benthic macroinvertebrate assemblages, including changes in
dominant species, and greatly increased abundance and bio-
mass [22,36–38]. This form of nutrient enrichment is also
associated with DELT anomalies [22].
ammonia, DO, BOD, and nutrients. Finally, the percentage of
tolerant invertebrates increased with increasing concentrations
of PAHs and metals.
Combining exposure data from the site with effects data
from elsewhere
The analyses presented in this section combined exposure
data from the Little Scioto River with effects data from other
studies. The objective was to evaluate whether stressors in the
Little Scioto River were present in sufficient amounts that
effects would be expected.
Habitat alteration. Although some studies have been per-
formed on the depth of channel required for the survival of
larger species, and the amount of siltation that reduces survival
of invertebrates, none of the actual response relationships were
readily available for direct comparison with existing condi-
tions. Furthermore, data from the Little Scioto were not avail-
able for the actual depth of the stream or quantitative measure
of stream embeddedness with which to make comparisons.
However, reviews indicate that many community metrics are
drastically altered by channelization but anomalies are not
[39].
Evidence for causal characterization
Although high-quality data are essential, they are not, in
themselves, causal evidence. Causal evidence consists of re-
lationships between the observed effects (impairments) and
the candidate causes. This section shows how data are asso-
ciated with the candidate causes and then with data concerning
effects.
Associating candidate causes with effects by using data
from the site
The relationships between candidate causes and effects were
analyzed by using data from the site. Two approaches were used.
First, we evaluated whether the measurements relevant to each
candidate cause were spatially colocated with biological respons-
es. Second, we examined whether biological responses were cor-
related with an increase or decrease in the candidate cause. All
of these analyses evaluated the candidate causes by using the
associated measurements shown in Table 6.
Five of the candidate causes, habitat alteration, metals, low
DO, and nutrient enrichment were detected at site A and so
were colocated with impairment A (Table 7). However, only
four had decreased quality as compared with the upstream site:
habitat alteration, metals, high BOD and low DO, and nutrient
enrichment. All candidate causes were observed at site B. All
of the candidate causes except for habitat alteration exhibited
decreased quality at site B as compared with site A; habitat
alteration was unchanged.
The second analysis evaluated whether changes in candi-
date causes were correlated with the biological responses along
the entire study reach of the Little Scioto (Table 8). Relative
fish weight decreased with increased concentrations of am-
monia, nutrients, and BOD. Relative fish weight increased with
increasing DO concentrations. The percentage of fish having
DELT anomalies increased and the percentage of mayflies de-
creased with increasing concentrations of two of the metals,
Table 7. Spatial colocation of candidate causes with impairments A
and B
Decreased
quality at
site A
compared
with
Decreased
quality at
site B
compared
with
Detected at upstream Detected at
Candidate cause
site A?
site?
site B?
site A?
Habitat alteration
PAHsb
Metals
Yes
No
Yes
Yes
Yes
Yes
No
Yes
No
Yes
Yes
Yes
Yes
Yesd
Noa
Yes
Yes
Yes
Yesd
Ammonia
High BOD and low
dissolved oxygen
Nutrient enrichment
Yes
Yes
Yes
Yes
Yes
a Although dissolved oxygen concentrations declined between sites A
and B in 1987, no change occurred in the habitat alteration metrics.
b PAHs
c BOD
ϭ
ϭ
polycyclic aromatic hydrocarbons.
biochemical oxygen demand.
d Dissolved oxygen concentrations declined between site A and B
(based on 1987 data), and BOD increased (based on 1992 data).

1120
Environ. Toxicol. Chem. 21, 2002
S.B. Norton et al.
Table 8. Correlation coefficients and associated
between selected measurements and biological responses. Correlations
(underlined) are significant at 0.10. Stressor measurements with
no significant correlations (e.g., channel, silt, and embeddedness
p values (in italics)
although the PAH toxic units were substantially higher. Table
11 compares aqueous concentrations of copper, lead, and zinc
with the ambient water quality criteria for these chemicals. No
criteria were exceeded.
p
ϭ
scores, fluoranthene, and mercury) are not shown.
n
ϭ
7 except where
noteda
In other studies, PAHs and metals have been associated with
anomalies and tumors in fish. Polycyclic aromatic hydrocarbons
characteristically cause liver tumors in fish at relatively low con-
centrations, but that effect is unlikely to be detected by exami-
nations for DELT anomalies. However, PAHs also have been
associated with more obvious anomalies at high concentrations
in sediment (21,200 mg total PAH/kg) [40]. Lead causes spinal
deformities [29], which would be included in the DELT index if
sufficiently severe. Chromium, copper, and zinc are not reported
to cause DELT anomalies [29]. Fin erosion and deformities were
found in fish collected from the Ottawa River in Lima, Ohio,
USA, in the presence of elevated levels of a mixture of metals
(primarily chromium, lead, and zinc) and PAHs [41]. Subsequent
studies confirmed that the fish had taken up the PAHs [15,42].
These studies did not exclude other pollutants, such as nitrogen
compounds, as the cause of anomalies. However, they did indicate
that these anomalies occur in white suckers from a metal- and
PAH-contaminated river in the absence of physical habitat mod-
ification. Hence, in the absence of information concerning the
actual anomalies at site B, the elevated DELT anomalies provide
weak evidence for PAHs or lead as causes of impairment B.
Studies in the western United States and New Zealand in-
dicate that species richness and abundance of mayflies are
sensitive to metal contamination, although some of the streams
are also impacted by low pH [43–45]. Hence, the decline in
mayflies is consistent with effects of mixed metals.
Percent
tolerant
Relative
DELT
Percent
fish weight anomalies mayflies invertebrates
Anthraceneb
]anthraceneb
Benzo[ghi]peryleneb
]pyreneb
Ϫ
Ϫ
0.45
0.25
0.29
0.45
0.074
0.89
0.037
0.89
0.33
0.39
0.26
0.50
0.45
0.26
0.5
0.77
0.066
0.59
0.16
0.33
0.44
0.44
0.30
0.63
0.13
0.63
0.13
0.76
0.070
0.75
0.073
0.64
0.13
0.54
0.20
0.75
0.073
0.92
0.027
0.86
0.040
0.88
0.034
0.96
0.020
0.77
0.074
Ϫ0.45
0.25
Ϫ0.41
0.29
Ϫ0.11
0.75
Ϫ0.19
0.62
Ϫ0.48
0.22
Ϫ0.33
0.39
Ϫ0.45
0.26
Ϫ0.71
0.073
Ϫ0.54
0.18
Ϫ0.64
0.11
Ϫ0.89
0.026
Ϫ0.81
0.042
Ϫ0.75
0.060
Ϫ0.99
0.014
Ϫ0.86
0.032
0.94
0.041
0.45
0.29
0.88
0.034
0.96
0.021
0.93
0.026
0.74
0.076
0.74
0.076
0.81
0.053
0.61
0.15
0.82
0.049
0.68
0.11
0.43
0.31
0.59
0.16
0.14
0.76
0.31
0.48
0.54
0.20
0.6
Benzo[
a
Benzo[
a
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
Chryseneb
Fluorantheneb
Fluoreneb
Crb
0.20
0.29
0.46
0.39
0.31
0.71
0.073
0.72
0.070
0.75
0.06
0.72
0.019
0.75
0.06
0.89
0.055
Cub
Pbb
Znb
BODc
Nitrate and nitritec
Total ammoniac
Total phosphorusc
Minimum DOc, d
Ammonia, DO, and nutrient enrichment. Table 11 compares
measurements of ammonia, DO, nitrogen, and phosphorus at
sites A and B with available criteria. No criteria were violated
at any location in 1992. However, because DO was not mea-
sured at site B in 1992, we evaluated data collected in 1987
with the same methods. In 1987, DO levels were lower than
criteria at both sites A and B. In Ohio, nutrient enrichment
has been associated with DELT anomalies, particularly at sites
influenced by point sources [46].
Ϫ
Ϫ
0.16
a DELT
ϭ deformaties, eroded fins, lesions, or tumors; BOD ϭ bio-
chemical oxygen demand; DO
ϭ dissolved oxygen.
b Concentrations measured in sediment.
c Concentrations measured in water.
d n
ϭ 6.
Measurements associated with causal mechanisms and
pathways
PAHs and metals. Tables 9 and 10 compare the TEL and
For several candidate causes, evidence was available for
intermediate steps in causal pathways. The pathways for hab-
itat alteration and nutrient enrichment did not include any in-
termediate steps, and so these candidate causes are not dis-
cussed in this section.
PEL values with the sediment concentrations of PAHs and
metals measured at the upstream site and sites A and B. No
sediment quality values were exceeded at the upstream site or
site A. At site B, five PAHs exceeded the PEL values and the
metals exceeded either the TEL or the PEL. The cumulative
toxic units for PAHs and metals both exceeded 1.0 at site B,
PAH. For fish, the exposure pathway for PAHs was con-
Table 9. Polycyclic aromatic hydrocarbon (PAH) concentrations (mg/kg) at the upstream site, site A, and site B (Hyalella azteca sediment effects
concentration, probable effect level [PEL], and threshold effect level [TEL], normalized to sediment dry wt)
PAH sediment concentration
Chemical
TEL
PEL
Upstream site
Site A
Site B
Benzo[
Fluoranthene
Benzo[ ]anthracene
Chrysene
Benzo[ghi] perylene
a
]pyrene
0.03
0.04
0.03
0.02
0.01
0.32
0.32
0.28
0.41
0.25
NDa
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
14.8b
8.2b
b
8.2b
16.5b
49.5b
339.4
Cumulative toxic units based on PEL
a ND
b Exceeds PEL and TEL.
ϭ below detection.

Listing and analyzing causes for Little Scioto River, OH
Environ. Toxicol. Chem. 21, 2002
1121
Table 10. Metals concentrations (mg/kg) at the upstream site, site A, and site B (threshold effect level [TEL] and probable effect level [PEL]
values are for Hyalella azteca and are normalized to sediment dry wt)
Chemical
TEL
PEL
Upstream site
Site A
Site B
Cr
Cu
Pb
Zn
32.3
28
37.2
98.1
119.4
101.2
81.7
7.3
7.4
12.1
30.6
0.28
13.6
17.2
19.1
79
208a
79b
172a
173b
4.94
544
Cumulative toxic units based on PEL
0.66
a Exceeds PEL and TEL.
b Exceeds TEL.
cluded to be incomplete at site A, and complete at site B.
Biomarkers of naphthalene and benzo[ ]pyrene are elevated
the exposure pathway is considered incomplete. Site B is more
difficult to evaluate because data are scanty and are used from
different years. In 1992, BOD was slightly elevated. Although
DO was not measured in 1992, in 1987, DO was lower at site
B than at the upstream site. Thus, the pathway was considered
to be complete.
a
from site B to the mouth of the river, providing evidence that
the exposure pathway is complete at these locations. Concen-
trations of metabolites were above background for the state of
Ohio from site B downstream to the confluence with the Scioto
River [15].
Metals. Data gaps for site A made the evaluation of exposure
pathways more difficult for the metals. Data from fish tissue
confirm the uptake of lead and zinc at site B.
Associations of effects with mitigation or manipulation of
causes
The experimental field exposures showed that epibenthic
water downstream of site B reduced survival of C. dubia when
exposed to ultraviolet radiation but survival was not reduced
when the experimental chambers were shaded [16]. Because
PAH toxicity is increased by ultraviolet radiation, this study
established that PAHs cause toxicity in the Little Scioto, albeit
at a site with higher concentrations than measured at site B.
If the DELT anomalies are liver tumors, then the reduction
in such tumors in fish from the Black River after reductions
in PAH levels provides supporting evidence that PAHs cause
these anomalies in Ohio rivers [47].
Ammonia. As shown in Figure 5, several interweaving path-
ways exist by which unionized ammonia can be produced in
the river and cause effects. Because unionized ammonia is
formed at high pH, and the pH of samples collected from the
Little Scioto in 1998 ranged from 7.5 to 8, the presence of
total ammonia also provides evidence of the toxic form. Total
ammonia was detected at site B, but not at site A.
High BOD and low DO. Dissolved oxygen can be depleted
by high BOD because of the bacterial respiration associated
with allochthonous organic matter or decaying algal mats.
Nocturnal algal respiration can further decrease DO concen-
trations. We have measurements of several relevant parame-
ters: nitrates and nitrites, total phosphorus, BOD, and DO con-
centrations. Algal concentration was not measured. This ex-
posure pathway was considered complete under two scenarios:
BOD is elevated and DO is reduced compared with the
upstream site; or, if BOD data are unavailable, nitrates and
nitrites and phosphorus are elevated and assumed to cause algal
growth, and DO is reduced as compared with the upstream
site. At site A, DO is reduced, but BOD is unchanged, so that
DISCUSSION
The application of the first steps of the causal inference
methodology method to the Little Scioto case study illustrates
how information can be organized to support causal evalua-
tions in complex cases. This paper illustrates the complexities
of distinguishing distinct impairments, identifying candidate
causes, hypothesizing the relationships between candidate
causes and effects, obtaining and summarizing data, and con-
verting the data into evidence concerning causation.
Table 11. Comparison of the reported concentration of water quality parameters with available criteria.
All data are from 1992 unless noted otherwise
Criterion (units)
Upstream site
Site A
Site B
Ammonia (mg/L)a; 0.57 at pH 8.5, 1.27 at pH 8.0
Dissolved oxygen (mg/L)c: 3.0 for MWHd
NDb
8.8
ND
5.7
0.12
NA
1.9f,g
0.8
0.09
15
NAe
1.2
2.8f,g
1.4
Nitrate–nitrite (mg/L)h: 1.6
Total phosphorus (mg/L)i: 0.28
0.06
ND
ND
ND
0.07
ND
ND
ND
Copper (
Lead (
g/L)f: 7.7
Zinc (
g/L)j: 190
␮
g/L)j: 21
␮
3
ND
␮
a U.S. Environmental Protection Agency (U.S. EPA) ammonia criterion [19].
b ND
not detected.
c Ohio EPA dissolved oxygen criterion [21].
d MWH
modified warm-water habitat.
e NA
no data.
ϭ
ϭ
ϭ
f U.S. EPA ambient water quality criteria; 4-d averages at hardness of 200 mg/L [20].
g Violates criteria.
h Proposed nitrate–nitrite criterion [22].
i Proposed total phosphorus criterion [22].
j Minimum dissolved oxygen concentrations from continuous monitoring over 3 d in 1987 [6].

1122
Environ. Toxicol. Chem. 21, 2002
S.B. Norton et al.
Where multiple sources and stressors are present, different
stressors may dominate at different locations and result in
multiple distinct impairments. In the Little Scioto, the patterns
of the biological community at sites A and B were different
enough to define two distinct impairments. However, the
stressors overlapped in space and time so that the impairments
were not independent. Our solution to the complexity was to
define impairment B in terms of the incremental responses
between sites A and B. In this way, the candidate causes that
were responsible for the distinct responses could be deter-
mined. However, at the conclusion of the investigation, all
cumulative stressors responsible for the impairment noted at
site B would need to be considered.
in communicating among the work group members. As noted
in other papers, conceptual models can serve to illuminate
causal relationships [49,50]. Although the information in a
conceptual model may seem self-evident after the fact, the
authors have found that scientific specialists are likely to miss
causal pathways unless they participate in the development of
a conceptual model by a multidisciplinary team.
Data quality issues included the potential for sampling,
analytical, and transcriptional errors. Detection limits, partic-
ularly in media that are difficult to analyze such as sediment
and fish, may be above levels of concern. In this study, the
quantification limits of most of the PAHs were above the sed-
iment quality concentrations. For this reason, the toxic poten-
tial of PAHs was likely underestimated.
Disaggregation of the multimetric indices aided the ability
to distinguish different impairments. Specific definition of the
impairments increases the likelihood that stream reaches with
potentially different causes can be selected for analysis. In this
case study, if IBI and ICI were used solely rather than the
component metrics, the differences between impairments A
and B would have been ambiguous or even unnoticed.
We hypothesized seven candidate causes for this section of
the Little Scioto River. The development of the list is clearly
a critical step in causal assessments. If the true cause is not
included, then the impairment may be attributed to the wrong
candidate. In the worst case, expensive remedial actions might
target the wrong stressor, and result in no improvement. On
the other hand, if the list of candidates is too large, then a
large amount of effort may be spent on very unlikely causes.
In the worst case, the assessment may become so complicated
that the true cause is missed in the confusion. In the Little
Scioto, the list of candidate causes may have been expanded
to include overfishing, habitat alteration from boating, in-
creased hydrologic variance from impervious surfaces in the
watershed, and toxic chemicals from spills. Overfishing was
not included because fishermen avoid this river because the
fishing is poor, a fish consumption advisory is in effect, and
fish are tainted with an oily odor. For similar reasons, boating
recreation also was considered to be unlikely. Chemical ex-
posure from spills could have been considered, but the bio-
logical impairments had been observed in both 1987 and 1992,
and so were considered to be long-term, rather than acute,
changes. Spills of materials into the Little Scioto and its trib-
utaries have included red food dye, oil, paint waste, and sewage
sludge, but none were reported between July 1990 and the
1992 sampling season [5].
Limitations of water sampling and analysis also raise con-
cerns that some candidate causes may have been missed. Al-
though sediments were analyzed for persistent pesticides, other
pesticides frequently detected in agricultural streams by the
U.S. Geological Survey (e.g., atrazine and chlorpyrifos) were
not included [48]. In this study, water quality data for wet
weather flows were not available. Storm events could con-
tribute toxic chemicals, nutrients, and flow extremes, espe-
cially through the combined sewer overflow at Rkm 9.7. Some
pathways may have been incompletely specified. For example,
organic chemical and ammonia concentrations may have been
sufficiently high that the chemical oxygen demand may have
contributed to low DO, but no data were available to provide
evidence for this pathway. Finally, some potential causes may
be missed because they are obscure or not influential relative
to the identified candidate causes. Remediation of one cause
may reveal others that were masked.
Only data from 1992 were presented in this paper to sim-
plify the presentation. In reality, all available years of data
should be used. Data were available for the Little Scioto for
1987, 1991, and 1998. In most cases, these data supported the
conclusions reached with 1992 data, increasing our confidence
in our analyses. The exception was DO data, which violated
criteria in 1987, but not in 1992. In addition, PAH concentra-
tions tended to increase with depth at site B and the down-
stream locations, indicating that their toxic potential would be
greater if these sediments became accessible.
The analysis was limited by the fact that data were not
collected for the purpose of the investigation. Establishing
causation is more demanding than determining that a water
body is chemically or biologically impaired. For causal anal-
ysis, sampling for characterization of habitat, stressors, and
responses must be effectively collocated in space and time.
An additional issue in this investigation is that fish were sam-
pled at the upstream site by a wading method, whereas all
other sites were sampled by the boating method. Because the
boating method uses more electrical power, it may sample more
larger fish in deeper pools, although, in shallow areas such as
the upstream site, more power would make a minimal differ-
ence in the size or number of fish sampled. Still, the difference
in relative fish weight between the upstream site and site A
may be partially due to the difference in sampling method.
Finally, sampling with sufficient intensity to characterize var-
iance also is important. In this study, we had low confidence
in the aqueous chemical concentrations, which likely are more
variable than sediment concentrations. In addition, although
we believe that the patterns of biological responses in im-
pairments A and B are real, the fact that each impairment is
characterized at only one point is a significant weakness.
Although more samples at more times and locations would
be highly useful, designing studies to support causal analysis
is not simply a matter of increasing expense and effort. Often,
if impairments and candidate causes are well defined, it will
be possible to focus studies on a few critical mechanistic stud-
ies that clearly eliminate or implicate a cause. Toxicity iden-
tification evaluation studies are one example [51]. Liver pa-
thology studies of the fish are another.
This analysis also illustrates the importance of pausing be-
tween the collection of data and the determination of the most
likely cause to organize the data into causal evidence. Partic-
ularly when existing data are used, important relationships
among types of data are likely to be overlooked if this step is
skipped. Further, organizing the data as causal evidence sim-
plifies the determination of the causes. Finally, proceeding
through the assessment in a stepwise and deliberate manner
can help communicate the reasoning and evidence that underlie
The development of the conceptual model was a useful step

Listing and analyzing causes for Little Scioto River, OH
Environ. Toxicol. Chem. 21, 2002
1123
eds, Biological Assessment and Criteria: Tools for Water Re-
source Planning and Decision Making. CRC, Boca Raton, FL,
USA, pp 236–286.
the intuitive leaps that often characterize causal assessments
[1].
18. U.S. Environmental Protection Agency. 1996. Calculation and
evaluation of sediment effects concentrations for the amphipod
Hyallela azteca and the midge Chironomus riparius. Assessment
and Remediation of Contaminated Sediments (ARCS) Program.
EPA 905 R96-008. Great Lakes National Program Office, Chi-
cago, IL.
19. U.S. Environmental Protection Agency. 1998. 1998 Update of
ambient water quality criteria for ammonia. EPA 822-R-98-008.
Office of Water, Washington, DC.
20. U.S. Environmental Protection Agency. 1987. Quality criteria for
water 1986. EPA 440/5-86-001. Office of Water Regulations and
Standards, Washington, DC.
21. Ohio Environmental Protection Agency. 1996. Justification and
rationale for revisions to the dissolved oxygen criteria in the Ohio
water quality standards. Technical Bulletin MAS/1995-12-5. Di-
vision of Surface Water, Monitoring and Assessment Section,
Columbus, OH, USA.
22. Rankin ET, Miltner R, Yoder CO, Mishne D. 1999. Associations
between nutrients, habitat, and the aquatic biota in Ohio rivers
and streams. Technical Bulletin MAS/1999-1. State of Ohio En-
vironmental Protection Agency, Columbus, OH, USA.
23. Hinton DE. 1993. Toxicologic histopathology of fishes: A sys-
temic approach and overview. In Couch JA, Fournie JW, eds,
Pathobiology of Marine and Estuarine Organisms. CRC, Boca
Raton, FL, USA, pp 177–215.
Acknowledgement—We gratefully acknowledge the contributions of
Donna Reed-Judkins and Jeroen Gerritsen, the comments of three
anonymous reviewers, and the biologists of the State of Ohio Envi-
ronmental Protection Agency. The views expressed in this paper are
those of the authors and do not necessarily reflect the views and
policies of the U.S. Environmental Protection Agency. Mention of
trade names or commercial products does not constitute endorsement
or recommendation for use.
REFERENCES
1. Cormier SM, Norton SB, Suter GW II, Altfater D, Counts B.
2002. Determining the causes of impairments in the Little Scioto
River, Ohio, USA: Part 2. Characterization of causes. Environ
Toxicol Chem 21:1125–1137.
2. U.S. Environmental Protection Agency. 2000. Stressor identifi-
cation guidance document. EPA/822/B-00/025. Office of Water,
Washington, DC.
3. Suter GW II, Norton SB, Cormier SM. 2002. A methodology for
inferring the causes of observed impairments in aquatic ecosys-
tems. Environ Toxicol Chem 21:1101–1111.
4. Omernik JM. 1987. Ecoregions of the conterminous United States
(map supplement). Ann Assoc Am Geogr 77:118–125.
5. Ohio Environmental Protection Agency. 1994. Biological, sedi-
ment and water quality study of the Little Scioto River, Marion,
Ohio. OEPA Technical Report EAS/1994-1-1. Division of Surface
Water, Ecological Assessment Section, Columbus, OH, USA.
6. Ohio Environmental Protection Agency. 1988. Biological and wa-
ter quality study of the Little Scioto River Watershed. Division
of Surface Water, Columbus, OH, USA.
7. Ohio Environmental Protection Agency. 1992. Bottom sediment
evaluation, Little Scioto River, Marion, Ohio. Division of Water
Quality Planning and Assessment, Ecological Assessment Sec-
tion, Columbus, OH, USA.
8. Ohio Environmental Protection Agency. 1998. Biological and wa-
ter quality study of Marion area streams, Marion County, Ohio.
OEPA Technical Report MAS/1999-12-6. Division of Surface
Water, Ecological Assessment Section, Columbus, Ohio, USA.
9. Ohio Environmental Protection Agency. 1987. Biological Cri-
teria for the Protection of Aquatic Life, Vol II–Users Manual for
Biological Assessment of Ohio Surface Waters. Procedure
WQMA-SWS-6. Division of Water Quality Planning and As-
sessment, Ecological Assessment Section, Columbus, OH, USA.
10. Ohio Environmental Protection Agency. 1989. Addendum to: Bi-
ological Criteria for the Protection of Aquatic Life, Vol II–Users
Manual for Biological Assessment of Ohio Surface Waters. Di-
vision of Water Quality Planning and Assessment, Ecological
Assessment Section, Columbus, OH, USA.
11. Ohio Environmental Protection Agency. 1989. Biological Cri-
teria for the Protection of Aquatic Life, Vol III–Standardized
Biological and Field Sampling and Laboratory Methods for As-
sessing Fish and Macroinvertebrate Communities. Procedure
WQPA-SWS-3. Division of Water Quality Planning and Assess-
ment, Ecological Assessment Section, Columbus, OH, USA.
12. Ohio Environmental Protection Agency. 1989. Manual of Ohio
EPA Surveillance Methods and Quality Assurance. Division of
Environmental Services, Columbus, OH, USA.
24. Austin B, Austin DA. 1993. Bacterial Fish Pathogens. Ellis Hor-
wood Limited, Chichester, UK.
25. Tarplee WH Jr, Louder DE, Weber AJ. 1971. Evaluation of the
effects of channelization on fish populations in North Carolina’s
Coastal Plain streams. North Carolina Wildlife Resources Com-
mission, Raleigh, NC, USA.
26. Karr JR, Fausch KD, Angermeier PL, Yant PR, Schlosser IJ. 1986.
Assessing biological integrity in running waters: A method and
its rationale. Ill Nat Hist Surv Spec Publ 5.
27. Yount JD, Niemi GJ. 1990. Recovery of lotic communities and
ecosystems from disturbance: A narrative review of case studies.
Environ Manage 14:547–569.
28. Allan JD. 1995. Stream Ecology: Structure and Function of Run-
ning Waters. Chapman & Hall, London, UK.
29. Eisler R. 2000. Handbook of Chemical Risk Assessment. Lewis,
Boca Raton, FL, USA.
30. Eisler R. 2000. Polycyclic Aromatic Hydrocarbons. Handbook
of Chemical Risk Assessment, Vol 2. Lewis, Boca Raton, FL,
USA.
31. Oris JT, Giesy JP. 1985. The photoenhanced toxicity of anthracene
to juvenile sunfish (Lepomis spp.). Aquat Toxicol 6:133–146.
32. Oris JT, Giesy JP. 1987. The photo-induced toxicity of polycyclic
aromatic hydrocarbons to larvae of the fathead minnow (Pime-
phales promelas). Chemosphere 16:1395–1404.
33. Sasson-Brickson G, Burton GA. 1991. In situ and laboratory
sediment toxicity testing with Ceriodaphnia dubia
icol Chem 10:201–208.
. Environ Tox-
34. Russo RC. 1985. Ammonia, nitrite, and nitrate. In Rand GM,
Petrocelli SR, eds, Fundamentals of Aquatic Toxicology. Hemi-
sphere, New York, NY, USA, pp 455–471.
35. Miltner R, Rankin ET. 1998. Primary nutrients and the biotic
integrity of rivers and streams. Freshwater Biol 40:145–158.
36. Dodds WK, Welch EB. 2000. Establishing nutrient criteria in
streams. J North Am Benthol Soc 19:186–196.
37. Smith VH, Tilman GD, Nekola JC. 1999. Eutrophication: Impacts
of excess nutrient inputs on freshwater, marine, and terrestrial
ecosystems. Environ Pollut 100:179–196.
38. Carpenter SR, Caraco NF, Correll DL, Howarth RW, Sharpley
AN, Smith VH. 1988. Nonpoint pollution of surface waters with
phosphorus and nitrogen. Ecol Appl 8:559–568.
13. Ohio Environmental Protection Agency. 1989. The qualitative
habitat evaluation index (QHEI): Rationale, methods, and appli-
cation. Division of Water Quality Planning and Assessment, Eco-
logical Assessment Section, Columbus, OH, USA.
14. Lin ELC, Cormier SM, Torsella JA. 1996. Fish biliary polycyclic
aromatic hydrocarbon metabolites estimated by fixed-wavelength
fluorescence: Comparison with HPLC-fluorescence detection.
Ecotoxicol Environ Saf 35:16–23.
39. Yoder CO, Rankin ET. 1995. Biological response signatures and
the area of degradation value: New tools for interpreting multi-
metric data. In Davis WS, Simon TP, eds, Biological Assessment
and Criteria: Tools for Water Resource Planning and Decision
Making. CRC, Boca Raton, FL, USA, pp 263–286.
40. Albers PH. 1995. Petroleum and individual polycyclic aromatic
hydrocarbons. In Hoffman DJ, Rattner BA, Burton GA, Cairns
J, eds, Handbook of Ecotoxicology. Lewis, Boca Raton, FL, USA,
pp 330–335.
15. Cormier SM, Lin ELC, Fulk F, Subramanian B. 2000. Estimation
of exposure criteria values for biliary polycyclic aromatic hydro-
carbon metabolite concentrations in white suckers (Catostomus
commersoni). Environ Toxicol Chem 19:1120–1126.
16. Ireland DS, Burton GA, Hess GG. 1996. In situ toxicity evalu-
ations of turbidity and photoinduction of polycyclic aromatic hy-
drocarbons. Environ Toxicol Chem 15:574–581.
17. Yoder CO, Rankin ET. 1995. Biological criteria program devel-
opment and implementation in Ohio. In Davis WS, Simon TP,

1124
Environ. Toxicol. Chem. 21, 2002
S.B. Norton et al.
41. Ohio Environmental Protection Agency. 1992. Biological and wa-
ter quality study of the Ottawa River, Hog Creek, Little Hog Creek
and Pike Run. Hardin, Allen, and Putnam counties, OH. Technical
Report EAS/1992-9-7. Division of Water Quality Planning and
Assessment, Ecological Assessment Section, Columbus, OH,
USA.
42. Cormier SM, Lin ELC, Millward MR, Schubauer-Berigan MK,
Williams DE, Subramanian B, Sanders R, Counts B, Altfater D.
2000. Using regional exposure criteria and upstream reference
data to characterize spatial and temporal exposures to chemical
contaminants. Environ Toxicol Chem 19:1127–1135.
43. Clements WH, Carlisle DM, Lazorchak JM, Johnson PC. 2000.
Heavy metals structure benthic communities in Colorado moun-
tain streams. Ecol Appl 10:626–638.
44. Kiffney PM, Clements WH. 1994. Effects of heavy metals on a
macroinvertebrate assemblage from a Rocky Mountain stream in
experimental micocosms. J North Am Benthol Soc 13:511–523.
45. Hickey CW, Clements WH. 1998. Effects of heavy metals on
benthic macroinvertebrate communities in New Zealand streams.
Environ Toxicol Chem 17:2338–2346.
46. Ohio Environmental Protection Agency. 1999. Association be-
tween nutrients, habitat and aquatic biota in Ohio Rivers and
streams. Technical Bulletin MAS/1999-1-1. Division of Surface
Water, Monitoring and Assessment Section, Columbus, OH, USA.
47. Baumann PC, Harshbarger JC. 1995. Decline in liver neoplasms
in wild brown bullhead catfish after coking plant closes and en-
vironmental PAHs plummet. Environ Health Perspect 103:168–
170.
48. U.S. Geological Survey. 1999. The Quality of Our Nation’s Wa-
ters. U.S. Department of the Interior, Reston, VA.
49. Suter GW II. 1999. Developing conceptual models for complex
ecological risk assessments. Hum Ecol Risk Assess 5:375–396.
50. Gentile J, Harwell MA, Cropper W, Harwell CC, DeAngelis D,
Davis S, Ogden JC, Lirman D. 2001. Ecological conceptual mod-
els: A framework and case study on ecosystem management for
south Florida sustainability. Sci Total Environ 274:231–253.
51. U.S. Environmental Protection Agency. 1993. Methods for aquat-
ic toxicity identification evalutations, phase II toxicity identifi-
cation procedures for samples exhibiting acute and chronic tox-
icity. EPA 600/R-92/080. Office of Research and Development,
Environmental Research Laboratory, Duluth, MN.