Full text of DOI:10.1002/etc.5620210604

Environmental Toxicology and Chemistry, Vol. 21, No. 6, pp. 1125–1137, 2002
Printed in the USA
0730-7268/02 $9.00
ϩ
.00
DETERMINING THE CAUSES OF IMPAIRMENTS IN THE LITTLE SCIOTO RIVER,
OHIO, USA: PART 2. CHARACTERIZATION OF CAUSES
S
USAN M. CORMIER,*† SUSAN B. NORTON,‡ GLENN W. SUTER II,§ DAVID ALTFATER,ͦͦ and BERNIE COUNTSͦͦ
†U.S. Environmental Protection Agency, National Exposure Research Laboratory, 26 Martin Luther King Drive, MS 642,
Cincinnati, Ohio 45268
‡U.S. Environmental Protection Agency, National Center for Environmental Assessment, Washington, DC 20460
§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—Two stream reaches in the Little Scioto River (OH, USA) were characterized for the causes of impairments measured
at two locations. By inductive inference, six candidate causes were winnowed down to three and five candidate causes for each of
the two stream reaches. Using a formal strength-of-evidence process, a single cause was determined. At the most upstream location,
habitat alterations, including fine-textured substrates and low DO, were characterized as the probable causes for an increased
percentage of anomalies of fish, a decreased percentage of mayflies, and an increased percentage of tolerant macroinvertebrates.
An increase in the relative weight of fish was attributed to an artificially narrow, deepened channel. Approximately 2 km downstream,
polycyclic aromatic hydrocarbon (PAH)-contaminated sediments were identified as the cause for both fish and macroinvertebrate
impairments. Causal characterization using first elimination and then a strength-of-evidence approach narrowed and defined the
causes of ecological impairment even in this situation, where many complex and interacting candidate causes existed. Applying a
formal method highlighted types of data and associations that can strengthen and present a more convincing determination of the
causes of impairment.
Keywords—Causation
Stressor
Fish
Invertebrate
INTRODUCTION
increase in the relative weight of fish compared to the upstream
site and a decrease compared with impairment A, an increase
in the percentage of DELT, a decrease in the percentage of
mayflies, and an increase in the percentage of tolerant ma-
croinvertebrates. See part I of the case study in Norton et al.
[3] for a detailed discussion of the impairments.
We hypothesized six candidate causes for these impair-
ments: habitat alteration, polycyclic aromatic hydrocarbon
(PAH) contamination, metal contamination, organic enrich-
ment resulting in low dissolved oxygen, ammonia toxicity, and
nutrient enrichment. We analyzed evidence for each of these
candidate causes using four broad categories: associations be-
tween measurements of candidate causes and effects from the
site, associations using effects data from elsewhere with ex-
posure data from the site, measurements associated with a
causal mechanism, and associations of effects with mitigation
or manipulation of causes. These associations are presented in
detail in Norton et al. [3].
In this sequel, we used the four types of associations as
evidence to eliminate candidate causes where possible and then
applied a strength-of-evidence approach to identify the most
likely cause from those remaining after elimination. Because
impairment B occurs downstream from an existing impair-
ment, we evaluated candidate causes for the incremental im-
pairment at site B from three perspectives: as an incremental
increase of the same cause identified for impairment A, as a
different and sufficient cause independent of impairment A,
and, if the different cause was not sufficient, then as a com-
bination of candidate causes.
A process for characterizing the causes of biological im-
pairments to ecosystems and presenting the evidence is de-
scribed in the U.S. Environmental Protection Agency’s (U.S.
EPA) Stressor Identification Guidance Document [1] and by
Suter et al. [2]. In this process, hypothesized candidate causes
are refuted in the elimination step. Remaining causes are then
characterized by diagnosis or strength of evidence. The
strength-of-evidence step sorts evidence into three categories:
data from the site, associations between data at the site and
data from elsewhere, and consistency across all lines of evi-
dence. It evaluates each candidate cause with respect to each
type of evidence and then determines the most likely cause.
In this case study, we present a causal characterization for
two sites that fail to meet the state of Ohio’s biological criteria
that are based on multimetric indices for fish and macroin-
vertebrates. To simplify discussion, we refer to two distinct
biological impairments as impairments A and B as described
in Norton et al. [3]. These impairments are defined relative to
the upstream site (site U) at river kilometer (Rkm) 14.9. Im-
pairment A occurs near Rkm 12.7 (site A) and includes an
increase in the relative weight of fish; an increase in the per-
centage of deformities, erosion, lesions, and tumors (DELT)
of fish; a decrease in the percentage of mayflies; and an in-
crease in the percentage of tolerant macroinvertebrates. Im-
pairment B occurs near Rkm 10.5 (site B) and includes an
* To whom correspondence may be addressed
(cormier.susan@epa.gov).
Presented at the 20th Annual Meeting, Society of Environmental
Toxicology and Chemistry, Philadelphia, Pennsylvania, November
14–18, 1999.
METHODS AND DATA SOURCES
The methods for elimination, diagnosis, and strength-of-
evidence characterization for the causes of biological impair-
1125

1126
Environ. Toxicol. Chem. 21, 2002
S.M. Cormier et al.
Table 1. Candidate causes remaining after elimination
ment are described in the U.S. EPA’s Stressor Identification
Guidance Document [1], and Suter et al. [2]. Candidate causes
were eliminated if the evidence indicated that they did not co-
occur with the biological impairment, if the biological im-
pairment decreased with increasing influence of the candidate
cause, or if the exposure pathway was incomplete. In the Little
Scioto River, gradients of biological responses are not suffi-
ciently well defined to allow for elimination; however, we
cautiously used the presence and absence of biological gra-
dients in the strength-of-evidence analysis. Different causes
may be responsible for impairments at sited A and B and at
other sites downstream to the mouth of the river. In fact, no
gradient of recovery was observed; both the degree of bio-
logical impairments and the concentrations of environmental
contaminants increased or remained relatively constant.
Diagnosis is the identification of causes based on charac-
teristic signs or symptoms [2]. No evidence strong enough to
support diagnosis was available for any of the candidate caus-
es.
Strength-of-evidence analysis uses all the evidence gen-
erated in the analysis phase to examine the credibility of each
remaining candidate cause [3]. The causal considerations for
the strength-of-evidence analyses were organized into three
types of considerations: case-specific evidence, evidence from
other situations or biological knowledge, and evidence based
on multiple lines of evidence. This last consideration evaluated
the consistency of the evidence and, when inconsistent, the
coherence of the evidence in support of the hypothesized caus-
es. We conducted the strength-of-evidence analysis by using
the causal considerations discussed in U.S. EPA [1] and Suter
et al. [2] to array the information relevant to each cause and
impairment. We used table 6 in Suter et al. [2] to score each
piece of evidence by assigning it pluses, minuses, zeros, no
evidence, or not applicable. For evidence of biological gra-
Impairment A
Impairment B
Habitat alteration
PAH contaminationa
Metal contamination
Ammonia
X
X
X
X
X
X
X
Low BOD/low DOb
Nutrient enrichment
X
a PAH
b DO
ϭ
ϭ
polycyclic aromatic hydrocarbon.
dissolved oxygen; BOD biochemical oxygen demand.
ϭ
about 30% lower than upstream, but biochemical oxygen de-
mand (BOD) concentrations were not different from the up-
stream site, and thus we eliminated high BOD/low DO as a
candidate cause.
We retained habitat alteration, metal contamination, and
nutrient enrichment as candidate causes because they co-oc-
curred with impairment A and increased compared to upstream
locations. Habitat alteration was evident by lower values for
channel and substrate scores and lower DO concentrations
compared to the upstream site. All metals measured in sedi-
ments were slightly greater at site A compared to the upstream
site; however, aqueous concentrations of metals were below
detection limits at site A and the upstream site. Nitrates and
Nitrites (NOx) increased from 1.2 to 1.4 mg/L, and total phos-
phorus (P) increased from 0.06 to 0.07 mg/L. Although these
nutrient shifts were small, they precluded logical elimination
of nutrient enrichment as a candidate cause.
Impairment B. At this site, we eliminated only one can-
didate cause, habitat alteration. At site B, the fish and inver-
tebrate metrics changed even though the channel, substrate
embeddedness, and silt scores remained the same. Because
channel modification was unchanged, it is unlikely that habitat
alteration is the cause of the decline in the concentration of
DO. Furthermore, DO levels increase downstream of site B
even though the stream is still channelized.
dient, we used a
and for a weak association 0.05
assigned as follows: strong (ϩϩϩ),
0.70 and 0.90; ambiguous (0),
lations with the wrong sign ( ). No score of triple minus was
p
value of
Ͻ
0.05 for a strong association
0.10. Scores were
0.90, weak ( ),
0.70; and weak corre-
Ͼ
p Ͻ
r
Ն
ϩ r Ն
Ͻ
r Ͻ
We were unable to eliminate the remaining candidate caus-
es. The PAH contaminants in sediments were present and were
elevated above concentrations at site A. Exposure to these
organic chemicals was demonstrated by internal concentrations
of PAH metabolites in fish and induction of ethoxyresorufin-
Ϫ
assigned since no correlations showed strong associations with
a counterintuitive sign. Finally, we evaluated the scores to-
gether with our confidence in the underlying data to reach a
conclusion. The scores were not summed; they formed the
basis of comparison and were used to identify the most com-
pelling pieces of evidence.
Data pertinent to the study were obtained from the Ohio
EPA and U.S. EPA as described by Norton et al. [3]. The
associations derived from these data were used in this sequel
to eliminate candidate causes or were considered in a strength-
of-evidence process to determine the most likely causes of the
impairments. It is important to recognize that the evidence was
derived from data that were not collected for the purpose of
this causal analysis.
O
-deethylase activity. The metals chromium, copper, lead, and
mercury were elevated in sediment compared to upstream con-
centrations at site A. Lead and copper were detected in water
samples. Fish tissues had detectable levels of lead and zinc.
The DO levels were lower at site B than at site A in 1987,
and BOD levels were greater than upstream locations in 1992.
Data within the same years were not available. Ammonia con-
centrations were also greater, and total P concentrations were
0.02 mg/L greater. Therefore, five candidate causes remained
after elimination: PAH contamination, metal contamination,
ammonia toxicity, high BOD/low DO, and nutrient enrichment.
Table 1 summarizes the candidate causes that remained after
the elimination process. Only those causes remaining needed
to be evaluated by diagnostic or strength-of-evidence char-
acterization.
RESULTS
Characterize cause: Elimination
The two impairments (A and B) are discussed here in re-
lation to the elimination of specific causes. Conclusions about
which candidate causes remained for each impairment are also
listed.
Impairment A. The PAHs and ammonia were not elevated
at site A relative to the upstream site; thus, we eliminated these
candidate causes. Dissolved oxygen (DO) concentrations were
Characterize cause: Diagnosis
Although no evidence was strong enough to support di-
agnosis, the pattern of community change was considered to
be suggestive and was used later in the strength-of-evidence
characterization.

Characterization of causes: Little Scioto River, OH
Environ. Toxicol. Chem. 21, 2002
1127
The deformities, fin erosion, physical lesions, and tumors
(DELT) on fish are pathologies that are also potentially subject
to diagnosis. Some DELT are strongly associated with known
toxic substances and others with bacterial infections [4,5].
However, DELT cannot be used to distinguish among toxic
substances unless specific anomalies are identified, and even
these may be too nonspecific to diagnose without additional
information (e.g., histopathology).
available because of the low solubility of metals in hard water
(327 mg/L at site A). Aqueous concentrations of metals in
water were below ambient water quality criteria.
Many habitat alterations could contribute to the decrease
in the percentage of mayflies; however, low substrate texture
and low DO seemed to be the most likely causes for this
distinct impairment. The biological gradient between low DO
and the percentage of mayflies was significant. The greatest
change in the percentage of mayflies and concentration of DO
occurred from the upstream site to site A. Although metal
concentrations were elevated, they were so low that even cu-
mulatively they did not reach a toxic level. Nutrient enrichment
is unlikely because the increase in the concentration of nutri-
ents is not likely to increase the plant biomass and change the
type of food source and assemblages of macroinvertebrates.
The shift in the percentage of tolerant macroinvertebrates
was small and less representative of impairment A than the
other three distinct impairments but was consistent with the
evidence presented for mayflies except that no significant cor-
relations occurred between the percentage of tolerant macroin-
vertebrates and any water quality parameter. This may suggest
that substrate quality more than water quality influenced the
proportion of tolerant macroinvertebrates. We concluded that
the small increase in the percentage of tolerant macroinver-
tebrates is due to habitat alterations, but a more specific cause
could not be determined.
Characterize causes: Strength of evidence
The results of the strength-of-evidence analysis are pre-
sented in Tables 2 and 3 to permit comparisons of the quality
of the evidence across candidate causes. The scores are com-
bined in two shorter tables (4 and 5) in order to more easily
see the scoring patterns. Following the strength-of-evidence
analysis, we characterized the candidates by describing the
most compelling pieces of evidence and identifying the prob-
able cause(s).
Characterize causes: Identify probable causes
Impairment A. We identified habitat alterations as the prob-
able causes of impairment A. Here we describe the evidence
for each of the four distinct impairments: an increase in the
relative weight of fish, an increase in the percentage of DELT,
a decrease in the percentage of mayflies, and an increase in
the percentage of tolerant macroinvertebrates. We eliminated
the candidate causes PAH, ammonia, and high BOD/low DO
for all four distinct impairments because they were not elevated
compared to the nearest upstream location. Diagnosis was not
applied because of a lack of specific symptomology. In a
strength-of-evidence characterization (Table 2), we evaluated
the candidate causes: habitat alteration, metal contamination,
and nutrient enrichment.
The increase in the relative weight of fish at the site is
probably caused by an artificially deepened channel that allows
larger fish to survive at site A. Unaltered streams are broad,
shallow, and meandering. At site A, the stream channel is cut
in a straight, deep, narrow watercourse in the relatively flat
landscape in order to lower the water table to drain the soils
for farming. Quantitative evidence was not found to support
this cause, but it still seemed more likely when compared to
the two other candidate causes, which had inconcordant evi-
dence. Metal contamination is an improbable cause because
controlled laboratory studies reported decreases in growth
rather than increases [6]. Nutrient enrichment is unlikely be-
cause the increase in nutrient concentrations was minute and
not likely to increase the plant biomass that could be converted
to fish biomass.
The cause of the increase in DELT remains uncertain but
seems to be related to general stress associated with altered
habitat, including low DO [5]. The minimum DO concentration
decreased by 3.1 mg/L from the upstream site to site A. Nu-
trient enrichment is an unlikely cause because only minute
increases in the concentrations of nutrients were observed
compared with the upstream site. Rankin et al. [7] did report
an association for DELT and total P in Ohio streams but at
higher levels of DELT than the 1.5% detected at site A. The
significant biological gradients between the percentage of
DELT and total P appeared to have a greater bearing on im-
pairments downstream, where concentrations of P and the per-
centage of DELT are much greater. Metal contamination is an
unlikely cause because the concentrations of metals in sedi-
ment at site A were quite low and may not be readily bio-
Fine substrate texture, low DO, and a deepened channel
associated with channel modification are consistent with im-
pairment A. The magnitude of the alteration and clear differ-
ence from the unimpaired upstream location strongly support
this cause, and inconcordant stressor–response evidence weak-
ens the other candidate causes.
Impairment B. We found PAH contamination to be suffi-
cient to cause all of the specific biological impairments at site
B. Here we describe the evidence for each of the four distinct
impairments: a decrease in the relative weight of fish, an ad-
ditional increase in the percentage of DELT, an additional de-
crease in the percentage of mayflies, and an additional increase
in the percentage of tolerant macroinvertebrates. Habitat al-
teration was eliminated as a candidate cause for the incre-
mental impairment and changes in fish and macroinvertebrates
at site B because the impairments increased and changed while
scores for substrate and channel remained unchanged relative
to site A. Dissolved oxygen decreased, but this could not be
attributed to habitat alteration because channel, embeddedness,
and silt scores were unchanged. The decrease in DO was eval-
uated as a candidate cause associated with increased organic
matter. Diagnosis was not applied because of a lack of specific
symptomology. In a strength-of-evidence characterization (Ta-
ble 3), we evaluated the candidate causes: PAH contamination,
metal contamination, ammonia toxicity, low DO resulting from
organic enrichment (high BOD/low DO), and nutrient enrich-
ment.
Several lines of evidence support PAHs as the cause of the
decline in relative weight and increase in DELT anomalies
associated with impairment B. The measurement of PAH me-
tabolites in fish tissue provides evidence of a complete ex-
posure pathway. The concentrations of the contaminants in
sediments were at levels known to cause DELT anomalies
[8,9]. Reduced growth and declines in weight are sublethal
effects that occur at concentrations lower than those associated
with anomalies. The percentage of DELT weakly correlated
with PAH. Correlations with the other biological metrics may

1128
Environ. Toxicol. Chem. 21, 2002
S.M. Cormier et al.

Characterization of causes: Little Scioto River, OH
Environ. Toxicol. Chem. 21, 2002
1129

1130
Environ. Toxicol. Chem. 21, 2002
Table 3. Strength-of-evidence analysis for the five candidate causes of impairment Ba
S.M. Cormier et al.
PAH contamination
Evidence
Metals contamination
Causal consideration
Score
Evidence
Score
Case-specific considerations
Co-occurrence
Compatible: sediment PAH concentra-
tions were several orders of magni-
tude greater at site B than site A
[3].
ϩ
Compatible: lead, chromium, copper,
zinc, and mercury concentrations in
sediment were 2 to 10 times greater
at site B than at site A [3]. In sur-
face water grab samples, copper and
lead were greater at site B than at
site A. Other metals were below de-
tection limit.
ϩ
Temporality
Consistency of association
Biological gradient
No evidence
No evidence: only one location
Decreased relative weight: ambiguous
NE
NE
0
No evidence
NE
NE
ϩ
No evidence: only one location
Decreased relative weight: weak cor-
relation with zinc
Increased % DELT: weak correlation
with anthracene and fluorene
ϩ
Increased % DELT: weak correlations
with chromium and zinc
ϩ
ϩ
ϩ
Decreased % mayflies: ambiguous
0
Decreased % mayflies: weak correla-
tions with chromium and zinc
Increased % tolerant organisms: corre-
lations with all PAH except anthra-
cene, strong correlations with ben-
zo[ghi]perylene and benzo[a]pyrene
[3].
Fish metrics: actual evidence for all
steps: both BAP and NAPH metab-
olites were found in fish [3].
Invertebrate metrics: no evidence
Fish metrics: no evidence
Invertebrate metrics: concordant: field
exposures in the Little Scioto
showed that epibenthic water down-
stream of site B reduced survival of
C. dubia when exposed to UV radi-
ation, but survival was not reduced
when experimental chambers were
shaded [23].
ϩϩϩ
Increased % tolerant organisms: weak
correlation with copper
Complete exposure pathway
Experiment
ϩϩ
Fish metrics: Actual evidence for all
steps: zinc and lead were detected
in fish tissues [3].
Invertebrate metrics: no evidence
Fish metrics: No evidence
Invertebrate metrics: no evidence
ϩϩ
NE
NE
NE
NE
NE
ϩϩϩ
Considerations based on other situations or biological knowledge
Plausibility: mechanism
Decreased relative weight: plausible:
PAHs are known to reduce growth
and shorten life span, resulting in
smaller fish [9].
ϩ
Decreased relative weight: plausible:
metals are known to reduce growth.
Toxic compounds can shorten life
span, resulting in smaller fish
[6,13–15].
ϩ
ϩ
ϩ
Ϫ
Ϫ
Increased DELT: plausible: PAHs are
known to cause eroded barbels, fin
erosion, lesions, and internal and
external tumors [9].
ϩ
Increased DELT: plausible: metals
cause fin erosion and lesions [4].
Pb, Zn, Cu cause deformities [6,13–
15].
Decreased % mayflies and increased
% tolerant organisms: plausible:
PAHs are known to cause reproduc-
tive impairments, which could de-
crease % mayflies and favor toler-
ant species [9].
ϩ
Decreased % mayflies and increased
% tolerant organisms: plausible:
metals are known to cause lethal
and sublethal effects to inverte-
brates that could decrease % may-
flies and favor tolerant [6].
Plausibility: stressor–
response
Decreased relative weight: no evi-
dence
NE
ϩϩϩ
Decreased relative weight: inconsis-
tent: copper at site B was 15 mg/L,
and lead was 3
␮g/L. Ambient wa-
ter quality criteria for lead, copper,
and zinc (7.7 g/L, 21 g/L, 190
␮
␮
␮
g/L, water hardness 200 mg/L, re-
spectively) are not exceeded.
Increased DELT: quantitatively consis-
tent: PAHs were at levels that cause
tumors and other DELT [8].
Increased DELT: inconsistent: lordos-
coliosis is reported at 850
␮g/L
lead at a hardness of 353 mg/L.
Hardness at site B was 389 mg/L.
Lead at site B was 3
␮g/L. Copper
at site B was 15 mg/L. Also, ambi-
ent water quality criteria for lead
and copper (7.7 ␮g/L and 21 ␮g/L,
water hardness 200 mg/L, respec-
tively) were not exceeded.

Characterization of causes: Little Scioto River, OH
Ammonia toxicity
Environ. Toxicol. Chem. 21, 2002
1131
Table 3. Extended
High BOD/low DO
Evidence
Nutrient enrichment
Evidence
Evidence
Score
Score
Score
Compatible: ammonia concentra-
tion was doubled relative to
impairment A [3].
ϩ
Compatible: in 1992, BOD was
double the upstream value, and
in 1987, the minimal DO lev-
els measured were 0.9 mg/L
less than upstream [3].
ϩ
Compatible: compared to site A,
total P was elevated by 0.02
mg/L. NOx was less [3].
ϩ
No evidence
NE
NE
ϩ
No evidence
NE
NE
ϩ
No evidence
NE
NE
Ϫ
No evidence: only one location
Decreased relative weight: weak
correlation
No evidence: only one location
Decreased relative weight: weak
correlations with BOD and DO
No evidence: only one location
Decreased relative weight: weak
association with NOx and total
P with wrong sign.
Increased % DELT: weak correla-
tion
ϩ
Increased % DELT: strong corre-
lation with BOD, weak with
low DO
Decreased % mayflies: weak cor-
relation with BOD, strong with
low DO
Increased % tolerant organisms:
ambiguous [3]
ϩϩϩ
ϩϩϩ
0
Increased % DELT: weak correla-
tion with NOx, strong with to-
tal P
Decreased % mayflies: weak cor-
relations with both NOx and to-
tal P
Increased % tolerant organisms:
ambiguous [3]
ϩϩϩ
Decreased % mayflies: strong
correlation
ϩϩϩ
ϩ
Increased % tolerant organisms:
ambiguous [3]
0
0
Not applicable: no known inter-
mediate steps
NA
NE
Not applicable: no known inter-
mediate steps
NA
NE
No evidence: concentrations of
NE
NE
algae and chlorophyll
not measured.
a were
No evidence
No evidence
No evidence
Decreased relative weight: plausi-
ble: ammonia toxicity is
known to reduce growth and
survival. Low survival could
alter the age structure, resulting
in smaller fish and more juve-
nile fish [29,30].
Increased DELT: plausible: re-
ports of histopathologies of
gills and internal organs, but
not of DELT. Stress could in-
crease susceptibility to oppor-
tunistic pathogens [5].
ϩ
ϩ
ϩ
Ϫ
Ϫ
Decreased relative weight: plausi-
ble: stress could reduce growth
and survival [5]. Low survival
could alter the age structure,
resulting in smaller fish and
more juvenile fish.
ϩ
ϩ
ϩ
ϩ
ϩ
Decreased relative weight: im-
plausible: increased nutrients
are usually associated with in-
creased algal growth that aug-
ment the energy available for
growth.
Ϫ
Increased DELT: plausible:
chronic low DO may cause
stress, which then increases
growth deformities or suscepti-
bility to opportunistic patho-
gens and fungi [5,7].
Increased DELT: plausible: nutri-
ents are believed to create con-
ditions that favor opportunistic
pathogens and fungi that cause
lesions and fin erosion and that
interfere with wound healing
[5,7].
Decreased % mayflies and in-
creased % tolerant organisms:
plausible: switching to an au-
tochthonous energy source
could alter species survival and
community composition for
fish and invertebrates [19].
Decreased relative weight: not
applicable: implausible mecha-
nism
ϩ
Decreased % mayflies and in-
creased % tolerant organisms:
plausible: ammonia is known
to be toxic to invertebrates
[29].
Decreased % mayflies and in-
creased % tolerant organisms:
plausible: low DO can kill fish
and invertebrates [19].
ϩ
Decreased relative weight: incon-
cordant: in several species, re-
duced growth was reported at
greater than 0.5 mg/L [29]. At
site B, ammonia levels were
near 0.2 mg/L.
Decreased relative weight: con-
cordant
NA
Ϫ
Increased DELT: inconcordant:
most studies reported histopa-
thology to gills or internal or-
gans at these or higher concen-
trations [29]. Reports of gross
anomalies similar to those
Increased DELT: concordant
Increased DELT: inconcordant:
concentrations of NOx and P
are at or below background
levels. Nitrogen does not limit
algal growth in most streams
[19]. Ohio’s proposed criteria
for nitrogen and phosphorus
criteria were not exceeded [7].
found at site B were not found.

1132
Environ. Toxicol. Chem. 21, 2002
S.M. Cormier et al.
Table 3. Continued
PAH contamination
Metals contamination
Causal consideration
Evidence
Score
Evidence
Score
Decreased % mayflies and increased
% tolerant organisms: quantitatively
consistent: the Hyalella azteca
PELs were exceeded for all PAHs.
The cumulative PAH toxic units
were 339 times the PEL value [20].
ϩϩϩ
Decreased % mayflies and increased
% tolerant organisms: quantitatively
consistent: Hickey and Clements
[21] reviewed invertebrate species
richness, particularly of mayflies,
which declined in association with
metals in water column; however,
water quality concentrations at site
A are uncertain, and they reported
declines in taxa rather than % may-
flies. Lead and chromium exceeded
H. azteca PEL values. The cumula-
tive toxic unit values for all metals
was 4.9 times the PEL value [20].
Decreased relative weight: no evi-
dence
ϩϩϩ
Consistency of association
Decreased relative weight: no evi-
dence
NE
NE
NE
Increased DELT: invariant: tumors and
other DELT are associated with fish
exposed to high concentrations of
PAH in fresh and marine waters
[24]. Increased DELT was associat-
ed with complex toxic exposures
[12].
ϩϩϩ
Increased DELT: no evidence
Decreased % mayflies and increased
% tolerant organisms: invariant: at
more than 25 locations associated
with PAH contamination that ex-
ceeded exposure criteria in Ohio,
ICI scores were below 30 [25]. ICI
scores of less than 30 occur only
when mayflies are decreased and
tolerant species are relatively abun-
dant. IBI and ICI are known to be
depressed even when habitat quality
was good [26,27].
ϩϩϩ
Decreased % mayflies and increased
% tolerant organisms: no evidence
NE
Specificity of cause
One of many
0
One of many
0
Analogy
Experiment
Not applicable
NA
Not applicable
NA
NE
Decreased relative weight: concordant:
following dredging in the Black
River, Ohio, the age and weight of
brown bullheads increased. The
Black River is larger and deeper
than the Little Scioto [17].
ϩϩϩ
Decreased relative weight: no evi-
dence
Increased DELT: concordant: in the
Black River, Ohio, removal of
PAHs by dredging resulted in lower
levels of liver tumors [8] and PAH
bile metabolites [28].
ϩϩϩ
Increased DELT: no evidence
NE
Decreased % mayflies and increased
% tolerant organisms: no evidence
No evidence
NE
NE
Decreased % mayflies and increased
% tolerant organisms: no evidence
No evidence
NE
NE
Predictive performance
Considerations from multiple lines of evidence
Consistency of evidence
Decreased relative weight: all consis-
ϩϩϩ
Decreased relative weight: most con-
sistent
Increased DELT: most consistent
Decreased % mayflies and increased
% tolerant organisms: all consistent
ϩ
tent
Increased DELT: all consistent
Decreased % mayflies and increased
% tolerant organisms: all consistent
ϩϩϩ
ϩϩϩ
ϩ
ϩϩϩ
Coherence of evidence
Not applicable
NA
Decreased relative weight, increased
DELT: no known explanation
0
a NE
oxygen demand; DELT
naphthalene; PEL
water habitat.
ϭ
no evidence; NA
ϭ
ϭ
not applicable/not available; DO
deformities, erosion, lesions, and tumors; PAH
ϭ
dissolved oxygen; NOx ϭ total nitrate-nitrite; P
ϭ
phosphorus; BOD
polycyclic aromatic hydrocarbons; BAP benzo[
index of biotic integrity; MWH
ϭ
biochemical
ϭ
ϭ
a
]pyrene; NAPH
ϭ modified warm-
ϭ
ϭ
probable effects level; ICI
ϭ
invertebrate community index; IBI
ϭ

Characterization of causes: Little Scioto River, OH
Ammonia toxicity
Environ. Toxicol. Chem. 21, 2002
1133
Table 3. Extended Continued
High BOD/low DO
Evidence
Nutrient enrichment
Evidence
Evidence
Score
Score
Score
Decreased % mayflies and in-
creased % tolerant organisms:
inconcordant: the ammonia
concentrations were not great
enough to cause the dramatic
effects seen with impairment
B. Ammonia criteria were not
exceeded [29].
Ϫ
Decreased % mayflies and in-
creased % tolerant organisms:
concordant: DO levels are be-
low Ohio criteria for MWH
[31].
ϩ
Decreased % mayflies and in-
creased % tolerant organisms:
inconcordant: the magnitude of
change in P from site A was
not great enough to cause dra-
matic effects seen with impair-
ment B. Proposed P criterion
was not exceeded [7].
Ϫ
No evidence
NE
No evidence
NE
No evidence
NE
Decreased relative weight: one of
many
0
Decreased relative weight: one of
many
0
Decreased relative weight: not
applicable
NA
Increased DELT: one of many
Decreased % mayflies and in-
creased % tolerant organisms:
one of many
0
0
Increased DELT: one of many
Decreased % mayflies and in-
creased % tolerant organisms:
one of many
0
0
Increased DELT: one of many
Decreased % mayflies and in-
creased % tolerant organisms:
one of many
0
0
Not applicable
No evidence
NA
NE
Not applicable
No evidence
NA
NE
Not applicable
No evidence
NA
NE
No evidence
NE
No evidence
NE
No evidence
NE
Decreased relative weight: most
consistent: ammonia concentra-
tions were lower than those re-
ported to cause effects.
Increased DELT: most consistent:
magnitude of change was in-
consistent with magnitude of
effect.
Decreased % mayflies and in-
creased % tolerant organisms:
most consistent: magnitude of
changes was inconsistent with
magnitude of effect.
Decreased % mayflies and in-
creased % tolerant organisms:
ammonia levels were out mea-
sured at sediment/water inter-
face and may be higher than in
the water column.
ϩ
Decreased relative weight: all
consistent
ϩϩϩ
Decreased relative weight: multi-
ple inconsistencies: implausible
mechanism.
ϪϪϪ
ϩ
ϩ
Increased DELT: all consistent
ϩϩϩ
ϩϩϩ
Increased DELT: most consistent:
magnitude of change was in-
consistent with magnitude of
effect.
Decreased % mayflies and in-
creased % tolerant organisms:
multiple inconsistencies: mag-
nitude of change was inconsis-
tent with magnitude of effect.
Decreased relative weight, in-
creased DELT, decreased %
mayflies and increased % toler-
ant organisms: no known ex-
planation.
ϩ
ϩ
Decreased % mayflies and in-
creased % tolerant organisms:
all consistent
ϩ
Not applicable
NA
0

1134
Environ. Toxicol. Chem. 21, 2002
S.M. Cormier et al.
Nutrient
Table 4. Summary of strength of evidence analysis of impairment A
Habitat
Causal consideration
Metals
Case-specific considerations
Co-occurrence
Temporality
ϩ
ϩ
NE
NE
ϩ
NE
NE
NEa
NE
Consistency of association
Biological gradient
Increased relative weight
Increased % DELTb
Decreased % mayflies
Increased % tolerant
Complete exposure pathway
Experiment
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
NE
NE
Ϫ
ϩϩϩ
ϩ
0
NE
NE
ϩϩϩ
0
NAc
NE
Considerations based on other situations or biological knowledge
Plausibility: mechanism
Increased relative weight
Increased % DELT
Decreased % mayflies and increased % tolerant
Plausibility: stressor–response
Increased relative weight
Increased % DELT
Decreased % mayflies and increased % tolerant
Consistency of association
Specificity of cause
ϩ
ϩ
ϩ
Ϫ
ϩ
ϩ
ϩ
ϩ
ϩ
NE
NE
NE
NE
NA
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
NE
NE
Increased relative weight
Increased % DELT
0
0
NA
0
0
0
Decreased % mayflies and increased % tolerant
Analogy
Experiment
0
0
0
NA
NE
NE
NA
NE
NE
NA
NE
NE
Predictive performance
Considerations from multiple lines of evidence
Consistency of evidence
Increased relative weight
Increased % DELT
Decreased % mayflies and increased % tolerant
Coherence of evidence
ϩϩϩ
ϩϩϩ
ϩϩϩ
NA
ϪϪϪ
ϪϪϪ
ϩ
ϩ
0
ϩ
ϩ
0
a NE
b DELT
c NA
ϭ
no evidence.
deformities, erosion, lesions, and tumors.
not applicable/not available.
ϭ
ϭ
be obscured by additional changes occurring further down-
stream.
relative weight were equally great, but DO concentrations in-
creased. Finally, the role of BOD as a precursor is questionable
because higher concentrations of DO occur at greater BOD
concentrations at downstream locations. It is possible that the
incremental decrease in the percentage of mayflies is related
to a further decline in the concentration of DO. However, the
highly toxic, PAH-laden sediments seem to be a more probable
cause.
Applying stressor–response information weakened the cas-
es for the other three candidate causes. None of the available
criteria for ammonia, metals, total P, or NOx were exceeded.
The cumulative toxic units calculated for metals in sediments
exceeded 1 but were only 1.4% of those for PAHs. Only zinc
and lead were detected in fish tissue analysis. Total P correlated
strongly with the percentage of DELT anomalies; however, the
incremental change in total P was small, and the correlation
was probably reflective of much higher concentrations of total
P and higher incidence of DELT anomalies at downstream
locations. Similarly, the percentage of mayflies was very
strongly correlated with ammonia concentrations for the entire
stream, but ammonia concentrations were 10 times greater
downstream, and the correlation appeared to be relevant to
downstream impairments rather than impairments at site B.
The percentage of tolerant macroinvertebrates showed no cor-
relation with any water quality parameter while showing con-
sistent correlations with PAH.
Similarly, several lines of evidence support PAH contam-
ination as the cause of decline in the percentage of mayflies
and the increase in the percentage of tolerant macroinverte-
brates. Although no direct measurement was available to dem-
onstrate exposure of invertebrates, we assumed that if fish were
exposed, benthic invertebrates were also exposed. The PAH
levels were at concentrations known to be toxic to benthic
macroinvertebrates; in fact, the cumulative toxic units of PAH
were more than 300 times the probable effects level. The per-
centage of tolerant invertebrates increased sharply from site
A to site B and was strongly correlated with all the PAHs
except anthracene. Overall, the consistency of the evidence is
very good.
The evidence for low DO as a cause was fairly consistent,
but the precursors of the low DO and the magnitude of the
effect were uncertain. Low DO may contribute to stress that
results in lesions or reduced relative weight. Among the seven
sites on the Little Scioto, both the fish metrics correlated weak-
ly with DO. In 1987, the criteria for DO were exceeded at site
B. However, the decrease in DO concentrations from site A
to site B (0.9 mg/L) was moderate, whereas the changes in
percentage of DELT anomalies, relative weight of fish, and
decrease in the percentage of mayflies were large. In addition,
further downstream, changes in the percentage of DELT and

Characterization of causes: Little Scioto River, OH
Environ. Toxicol. Chem. 21, 2002
1135
Table 5. Summary of strength-of-evidence analysis of impairment B
Causal consideration
PAHa
Metals
NH3
BOD/DOb
Nutrient
Case-specific considerations
Co-occurrence
Temporality
Consistency of association
Biological gradient
ϩ
NEc
NE
ϩ
NE
NE
ϩ
NE
NE
ϩ
NE
NE
ϩ
NE
NE
Decreased relative weight
0
ϩ
0
ϩϩϩ
ϩϩ
ϩϩϩ
ϩ
ϩ
ϩ
ϩ
ϩϩ
NE
ϩ
ϩ
ϩ
ϩϩϩ
ϩϩϩ
0
NA
NE
Ϫ
ϩϩϩ
ϩ
0
NE
NE
Increased % DELTd
Decreased % mayflies
Increased % tolerant macroinvertebrates
Complete exposure pathway
Experiment
ϩϩϩ
0
NAe
NE
Considerations based on other situations or biological knowledge
Plausibility: mechanism
Decreased relative weight
Increased % DELT
Decreased % mayflies and increased % tolerant
Plausibility: stressor–response
Decreased relative weight
Increased % DELT
Decreased % mayflies and increased % tolerant
Consistency of association
Decreased relative weight
Increased % DELT
Decreased % mayflies and increased % tolerant
Specificity of cause
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
Ϫ
ϩ
ϩ
NE
ϩϩϩ
ϩϩϩ
Ϫ
Ϫ
ϩϩϩ
Ϫ
Ϫ
Ϫ
ϩ
ϩ
ϩ
NA
Ϫ
Ϫ
NE
ϩϩϩ
ϩϩϩ
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
Decreased relative weight
Increased % DELT
0
0
0
0
0
0
0
0
NA
0
Decreased % mayflies and increased % tolerant
Analogy
0
NA
0
NA
0
NA
0
NA
0
NA
Experiment
Decreased relative weight
Increased % DELT
Decreased % mayflies and increased % tolerant
Predictive performance
ϩϩϩ
ϩϩϩ
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
NE
Considerations from multiple lines of evidence
Consistency of evidence
Decreased relative weight
Increased % DELT
Decreased % mayflies and increased % tolerant
Coherence of evidence
ϩϩϩ
ϩϩϩ
ϩϩϩ
NA
ϩ
ϩ
ϩϩϩ
0
ϩ
ϩ
ϩ
ϩ
ϩϩϩ
ϩϩϩ
ϩϩϩ
NA
ϪϪϪ
ϩ
ϩ
0
a PAH
b BOD/DO
c NE
no evidence.
d DELT
deformities, erosion, lesions, and tumors.
e NA
not applicable/not available.
ϭ
polycyclic aromatic hydrocarbons.
biochemical oxygen demand/dissolved oxygen.
ϭ
ϭ
ϭ
ϭ
DISCUSSION
data and analyzing associations between candidate causes and
impairments, and using logical arguments to identify causes.
In presenting the arguments for identifying the causes, refu-
tation is clearly separated from diagnosis and strength of ev-
idence. These are complex activities, and using a formal pro-
cess helps ensure more complete documentation of the inte-
grated thought processes used by ecologists, toxicologists, and
resource managers to determine causes of impairment.
It is worth noting that the more precise the identification
of the impairment, the more likely that the list of candidate
causes can be reduced in the elimination step and that a cause
can be ultimately identified in later steps. Specific definition
of the impairments increases the likelihood that stream reaches
with potentially different causes can be recognized and eval-
uated. In this case study, if Index of Integrity and Invertebrate
community index were used rather than the constituent or com-
ponent metrics, differences between impairments A and B
would have been ambiguous. Likely causes could not have
been determined or narrowed down to only a few likely causes.
Use of the causal analysis approach described in the Stress-
or Identification Guidance Document [1] and in Suter et al.
[2] permitted us to determine the causes of biological im-
pairments in the Little Scioto River with reasonable confi-
dence. Most important, it led to the identification of distinct
impairments within an officially impaired river segment and
a way to evaluate and discriminate among physical and chem-
ical causes.
This causal characterization demonstrates that some basic
applications of logic can help narrow down a rather complex
list of candidate causes and might in other cases be sufficient
to select restorative or protective actions. At minimum, it high-
lights those types of data that need to be collected and can
point to types of associations that may be needed to strengthen
and present a more convincing determination of the causes of
impairment. For instance, the format helps organize complex
ecological information into three main activities: describing
the impairment and listing candidate causes, assembling the

1136
Environ. Toxicol. Chem. 21, 2002
S.M. Cormier et al.
Also, different causes for different specific impairments at the
same site could not be characterized.
multiple lines of evidence and different types of information
using a flexible but formal process described by the Stressor
Identification Guidance Document [1]. The process that is used
here may be useful for determining the causes of impairments
not only to streams but for other resource types. Furthermore,
research designs may benefit from considering methods for
eliminating alternative hypothesis that depend on inductive
reasoning rather than experiments alone.
Further definition of the component metrics could have
been useful. Characterization of the specific DELT anomalies
is an obvious example. It would have also been useful to know
whether the increases in percentage tolerant macroinvertebra-
tes were due to increases in specific tolerant taxa or declines
in intolerant taxa.
Many readers probably realize that some of the causes re-
maining after the elimination process are very unlikely. How-
ever, to accurately reflect the logical process that we use and
because only mean data or few data are available, these causes
are retained. In so doing, the elimination step is stronger, and
the entire process possesses greater integrity. For instance,
nutrient enrichment is retained as a candidate cause for im-
pairment A even though the increase in total P and NOx is
minute. The reason it is retained is because it fails to meet
criteria for elimination, namely, a reduction or unchanging
concentration of the candidate causes. Some may argue that
the amount of change in the concentration of NOx and total P
would not be statistically different and therefore could be elim-
inated. However, unless a large sampling program is imple-
mented, the power to detect real differences would remain
small and unreliable. More important, the most compelling
reason for nutrient enrichment being an improbable cause
comes from ecological knowledge about the amounts of nu-
trients that would be needed for effects to occur. This knowl-
edge comes from other watersheds and is not grounds for
refutation at this site. The proper way to show this type of
evidence is by the strength-of-evidence procedure.
In this particular case study, the diagnostic analysis is not
effective for identifying any causes. However, in other studies,
it has been a decisive tool for determining the causes of fish
kills [10]. In fish kills, pathological evidence is usually par-
ticularly useful. For instance, surface lesions may be associated
with bacterial infections, whereas liver tumors may be asso-
ciated with PAH. For alterations to community structure, di-
agnostic evidence is less well documented and studied. This
is being rectified by a number of researchers who are trying
to improve the specificity and confidence in using patterns of
biological metrics as a diagnostic tool [11,12].
The causal characterization of the Little Scioto River could
be strengthened by additional evidence from the literature that
could be used to evaluate the plausibility of mechanisms and
stressor–response relationships, consistency of associations,
specificity of causes, and results of experiments. The use of
readily available literature in this case study makes it repre-
sentative of the use of the method by scientists in regulatory
agencies given current resources. However, it would be de-
sirable to assemble the voluminous but diffuse results of field
studies of impaired waters and combine them with information
from the open literature to make these types of evidence ac-
cessible for future characterizations.
To rigorously apply the method described by U.S. EPA [1]
and Suter et al. [2] requires discipline to guard against the
intuitive leaps that often characterize conclusions about cause.
Scientists with long experience in particular streams can quick-
ly identify causes on the basis of their knowledge of the system
and its history. However, a formal process can help document
the critical pieces of the evidence used to draw conclusions,
increase consistency among investigations, and increase con-
fidence that the true cause has been identified.
Acknowledgement—The authors thank everyone who participated in
the Stressor Identification Workgroup, especially Bill Swietlik and
cochair Donna Reed-Judkins; special thanks to Jeroen Gerritson, Edith
Lin, Bhagya Subramanian, and Brad Autrey, who obtained some of
the data used here and helped review the manuscript. The views ex-
pressed 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 the trade names or commercial products does not
constitute endorsement or recommendation for use.
REFERENCES
1. U.S. Environmental Protection Agency. 2000. Stressor identifi-
cation guidance document. EPA/822-B-00-025. Final Report.
Washington, DC.
2. Suter G 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.
3. Norton SB, Cormier SM, Suter G II, Subramanian B, Lin E,
Altfater D, Counts B. 2002. Determining probable causes of eco-
logical impairment in the Little Scioto River, Ohio, USA: Part I.
Listing candidate causes and analyzing evidence. Environ Toxicol
Chem 21:1112–1124.
4. 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.
5. Austin B, Austin DA. 1993. Bacterial Fish Pathogens: Diseases
in Farmed and Wild Fish, 2nd ed. Ellis Horwood Limited, Chich-
ester, UK, pp 11–14.
6. Eisler R. 2000. Handbook of Chemical Risk Assessment, Vol I.
Lewis, Boca Raton, FL, USA.
7. Rankin E, Miltner R, Yoder C, Mishne D. 1999. Association
between nutrients, habitat, and the aquatic biota in Ohio rivers
and streams. Technical Bulletin MAS/1999-1. Ohio EPA, Colum-
bus, OH, USA.
8. 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.
9. Eisler R. 2000. Polycyclic aromatic hydrocarbons. In Handbook
of Chemical Risk Assessment, Vol 2. CRC, Boca Raton, FL, USA,
pp 1343–1411.
10. Meyer FP, Barclay LA. 1990. Field manual for the investigation
of fish kills. Resource Publication 177. U.S. Fish and Wildlife
Service, Washington, DC.
11. Norton SB, Cormier SM, Smith M, Jones RC. 2000. Can bio-
logical assessments discriminate among types of stress? A case
study from the eastern corn belt plains ecoregions. Environ Tox-
icol Chem 19:1113–1119.
12. 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 Deci-
sionmaking. Lewis, Boca Raton, FL, USA, pp 236–286.
13. U.S. Environmental Protection Agency. 1985. Ambient water
quality criteria for copper—1984. EPA 440/5-84-031. Final Re-
port. Office of Water, Washington, DC.
14. U.S. Environmental Protection Agency. 1985. Ambient water
quality criteria for lead—1984. EPA 440/5-84-027. Final Report.
Office of Water, Washington, DC.
15. U.S. Environmental Protection Agency. 1987. Ambient water
quality criteria for zinc—1984. EPA 440/5-87-0003. Final Report.
Office of Water, Washington, DC.
16. Karr JR, Schlosser IJ. 1977. Impact of nearstream vegetation and
stream morphology on water quality and stream biota. EPA-600-
This case study demonstrates the usefulness of combining

Characterization of causes: Little Scioto River, OH
Environ. Toxicol. Chem. 21, 2002
1137
3-77-097. Final Report. U.S. Environmental Protection Agency,
Environmental Research Laboratory, Athens, GA.
17. Yount JD, Niemi GJ. 1990. Recovery of lotic communities and
ecosystems from disturbance: A narrative review of case studies.
Environ Manage 14:547–569.
18. Rankin ET. 1995. Habitat indices in water resource quality as-
sessments. In Davis WS, Simon TP, eds, Biological Assessment
and Criteria. Lewis, Boca Raton, FL, USA, 181–208.
19. Allan JD. 1995. Stream Ecology: Structure and Function of Run-
ning Waters. Chapman & Hall, London, UK.
20. U.S. Environmental Protection Agency. 1996. Calculation and
evaluation of sediment effect concentrations for the amphipod
Hyallela azteca and the midge Chironomus riparius. EPA 905-
R96-008. Great Lakes National Program Office, Chicago, IL.
21. Hickey CW, Clements WH. 1998. Effects of heavy metals on
benthic macroinvertebrate communities in New Zealand streams.
Environ Toxicol Chem 17:2338–2346.
25. Cormier SM, Lin ELC, Fulk FA, Subramanian B. 2000. Esti-
mation of exposure criteria values for biliary polycyclic aromatic
hydrocarbon metabolite concentration in white suckers (Catos-
tomus commersoni). Environ Toxicol Chem 19:1120–1126.
26. 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.
27. 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, Ohio.
Division of Water Quality Planning and Assessment, Ecological
Assessment Section, Columbus, OH, USA.
28. Lin ELC, Neiheisel TW, Subramanian B, Williams DE, Millward
MR, Cormier SM. 2001. Historical monitoring of biomarkers of
exposure of brown bullhead in the remediated Black River and
Cuyahoga River, Ohio. J Gt Lakes Res 27:191–198.
29. U.S. Environmental Protection Agency. 1999. 1999 update of
ambient water quality criteria for ammonia. EPA-822-R-99-014.
Final Report. Office of Water, Washington, DC.
22. Lemly AD. 1998. Bacterial grown on stream insects: Potential
for use in bioassessment. J North Am Benthol Soc 17:228–238.
23. Ireland DS, Burton GA, Hess GG. 1996. In situ: Toxicity eval-
uations of turbidity and photoinduction of polycyclic aromatic
hydrocarbons. Environ Toxicol Chem 15:574–581.
30. World Health Organization Work Group. 1986. Ammonia. En-
viron Health Criter 54:1–210.
24. Abers P. 1995. Petroleum and individual polycyclic aromatic hy-
drocarbons. In Hoffman DJ, Rattner BA, Burton GA, Cairns J,
eds, Handbook of Ecotoxicology. CRC, Boca Raton, FL, USA,
pp 330–335.
31. 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.