Full text of DOI:10.1002/etc.5620210607

Environmental Toxicology and Chemistry, Vol. 21, No. 6, pp. 1156–1167, 2002
2002 SETAC
Printed in the USA
0730-7268/02 $9.00 .00
᭧
ϩ
USE OF PHYSICAL, CHEMICAL, AND BIOLOGICAL INDICES TO ASSESS IMPACTS
OF CONTAMINANTS AND PHYSICAL HABITAT ALTERATION IN URBAN STREAMS
C
ATRIONA E. ROGERS,*† DANIEL J. BRABANDER,‡ MICHAEL T. BARBOUR,§ and HAROLD F. H EMOND†
†Civil and Environmental Engineering, Massachusetts Institute of Technology, 48-311, Cambridge, Massachusetts 02139, USA
‡Environmental, Coastal and Ocean Sciences, University of Massachusetts Boston, Boston, Massachusetts 02125-3393, USA
§Tetra Tech, 10045 Red Run Boulevard, Suite 110, Owings Mills, Maryland 21117, USA
(
Received 31 January 2001; Accepted 6 November 2001)
Abstract—Human activities in urban areas can lead to both chemical pollution and physical alteration of stream habitats. The
evaluation of ecological impacts on urban streams can be problematic where both types of degradation occur. Effects of contaminants,
for example, may be masked if stream channelization, loss of riparian vegetation, or other physical stressors exert comparable or
larger influences. In the Aberjona watershed (near Boston, MA, USA), we used physical, chemical, and biological indices to discern
the relative impacts of physical and chemical stressors. We used standard protocols for assessing the biological condition of low-
gradient streams, sampling macroinvertebrate communities from several different habitat types (e.g., overhanging bank vegetation,
undercut bank roots, and vegetation on rocks). We strengthened the linkage between chemical exposure and macroinvertebrate
response by measuring metal concentrations not only in sediments from the stream bottom but also in the vegetative habitats where
the macroinvertebrates were sampled. Linear regression analysis indicated that biological condition was significantly dependent
(95% confidence level) on contaminants in vegetative habitats, but not on contaminants in sediments from the stream bottom.
Biological condition was also significantly dependent on physical habitat quality; regression analysis on both contaminants and
physical quality yielded the best regression model (r2
ϭ 0.49). Similar biological impairment was observed at sites with severe
contamination or physical impairment or with moderate chemical and physical impairment. These results have implications for the
management of urban streams.
Keywords—Multiple stressors
Metals
Habitat quality
Macroinvertebrates
Urban streams
INTRODUCTION
This is not accomplished by current protocols. For example,
the U.S. Environmental Protection Agency (U.S. EPA) and
several state agencies recommend sampling a combination of
macroinvertebrate habitats to assess a stream’s biological con-
dition [4,9–11], but to our knowledge, no recommended meth-
od exists for measuring metal concentrations in the same ma-
croinvertebrate habitats.
Metal concentrations in vegetative habitats could be quite
different from metal concentrations in sediments at the stream
bottom, so we developed a method for measuring exposures
in vegetative habitats. Sansone et al. [12] demonstrated that
the retention of suspended particles transported by river flow
onto surfaces of freshwater plants is a potentially important
process in the contamination of aquatic biota. Thus, our meth-
od analyzes whole samples of macroinvertebrate habitats; sed-
iments associated with the vegetative material were not washed
off before chemical analysis.
An often-neglected factor in ecological risk assessments of
contaminated streams is physical habitat quality. Physical hab-
itat, as structured by in-stream and surrounding topographical
features, is a major determinant of aquatic community potential
[1–7]. In urban watersheds, chemical contamination commonly
accompanies physical alteration of stream habitats. Our ob-
jective was to determine whether simple indices of chemical,
physical, and biological conditions could be used to separately
estimate the influence of chemical and physical degradation
on macroinvertebrate communities. In particular, we were in-
terested in whether consideration of physical habitat quality
could be used to improve chemical risk assessments. Thus, we
included evaluation of a stream’s physical habitat quality (e.g.,
sedimentation, channelization, and loss of vegetation) in our
assessment of the impacts from metal contaminants on ma-
croinvertebrates in an urban watershed in near Boston, Mas-
sachusetts, USA.
We further refined our assessment by improving estimates
of metals exposure. Applying the principle that contaminants
must contact receptors to cause effects [8], we argue that un-
derstanding the effects of metals on macroinvertebrate com-
munities requires measurement of metal concentrations in the
same habitats from which the macroinvertebrates are collected.
MATERIALS AND METHODS
This study was carried out in the Aberjona watershed (Fig.
1), located 12 miles north of Boston (MA, USA) in the Boston
Basin ecoregion [13]. Two Superfund sites (Industri-Plex 128
and Wells G&H, Woburn, MA, USA) are located on the Aber-
jona River, and elevated concentrations of metals in the river’s
sediments are believed to be the result of historical industrial
activities, notably leather processing and chemical manufac-
turing [14–16]. The Aberjona River and its tributary streams
have also been physically altered as the watershed has been
developed for residential, commercial, and industrial uses [17].
* To whom correspondence may be addressed
(rogers.catriona@epa.gov). The current address of C. Rogers is U.S.
Environmental Protection Agency, 1200 Pennsylvania Avenue NW,
Mail Code 8601-D, Washington, DC 20460.
Presented at the 20th Annual Meeting, Society of Environmental
Toxicology and Chemistry, Philadelphia, Pennsylvania, November
14–18, 1999.
1156

Contaminants and physical alteration in urban streams
Environ. Toxicol. Chem. 21, 2002
1157
Fig. 1. Maps of the sampling sites. a. Study sites in the Aberjona watershed (Boston, MA, USA). b. Reference sites shown with the Aberjona
sites.
Identifying metals that could be present in toxic amounts
mg/kg dry wt), Cu (190 mg/kg dry wt), Fe (28% dry wt), Mn
(1,700 mg/kg dry wt), Ni (45 mg/kg dry wt), Pb (99 mg/kg
dry wt), and Zn (550 mg/kg dry wt). Iron concentrations never
exceeded ERL or ERM values in the sediment samples from
any of the sites; thus, Fe was considered to be an unlikely
candidate for posing a toxic risk. Ingersoll et al. did not report
an ERL or ERM value for Hg, but concentrations of Hg were
lower than or equal to the detection limit of our analytical
method (5 mg/kg dry wt). The Al, Mn, and Ni concentrations
in sediment samples rarely exceeded ERM values, and similar
concentrations for each of these metals were found in reference
sites and study sites. (Ranges for reference and study sites
overlapped for each metal. The number of ERM exceedances
was four or fewer, and concentrations more than 15% in excess
of the ERM value were never observed.) The history of the
Aberjona watershed does not suggest that Al, Mn, or Ni is an
important industrial contaminant of this watershed.
In contrast, concentrations of As, Cd, Cr, Cu, Pb, and Zn
in sediments frequently exceeded ERM values, with some val-
ues being two- to sixfold as high as the ERM values. Each of
these metals is a known contaminant in the Aberjona water-
shed, and their presence is consistent with the watershed’s
industrial history [14]. In addition, concentrations of these
metals in sediment sites from the Aberjona River were con-
sistently higher than those found at the reference sites. Thus,
We used three criteria to focus this study on those metals
that were most likely to be present in toxic amounts: (1) Is
the metal present in concentrations that have been observed
to have toxic effects on macroinvertebrates elsewhere? (2) Is
the metal present in higher concentrations at the study sites
than at the reference sites? (3) Does the history of the Aberjona
watershed suggest that toxic amounts of this metal could have
been released into the watershed’s streams?
For the first criterion, we employed an approach developed
by the National Status and Trends Program to evaluate sedi-
ment contamination concentrations [18]. This approach uses
toxicity levels that are based on laboratory spiked-sediment
bioassays, equilibrium partitioning calculations, and field stud-
ies. Three concentration ranges are defined in this approach:
No-effects, possible effects, and probable effects. The effects
range low (ERL) is the lower 10th percentile concentration in
sediments that is associated with toxic effects, and the effects
range median (ERM) is the median metal concentration in
sediment associated with toxic effects [18,19]. No toxicity
effects are expected for concentrations below the ERL, where-
as effects are probable for concentrations in excess of the ERM.
Ingersoll et al. [20] reported ERM values for freshwater-sed-
iment samples for the following metals: Al (58,000 mg/kg dry
wt), As (50 mg/kg dry wt), Cd (3.9 mg/kg dry wt), Cr (270

1158
Environ. Toxicol. Chem. 21, 2002
C.E. Rogers et al.
this study focused on As, Cd, Cr, Cu, Pb, and Zn. (Other
stressors, including organic contaminants, may also be im-
portant; these are considered in the Discussion section.)
10 jabs (or kicks) were taken from all major habitat types in
the reach, resulting in sampling of approximately 2.5 m2 of
habitat. In this study, three of the sampling sites were chosen
at random, and duplicate, but not overlapping, samples were
collected from the major habitat types.
Sampling locations
Following procedures outlined by the MA DEP [10], sam-
ples (all material collected in the 10 jabs or kicks, including
benthic macroinvertebrates, bits of vegetation, small pebbles,
All sampling sites were 100-m segments of low-gradient
streams located within the Boston Basin ecoregion. Six stream
segments of the Aberjona River were selected for study and
were designated as Ab1 to Ab6, with Ab6 being the location
farthest downstream (Fig. 1a). Ten stream segments were se-
lected in minimally contaminated streams within the Aberjona
watershed and were designated as H2 to H8 and as Tr1 to Tr3
(Fig.1a). (Site H1 was rejected as a study area because of very-
low- to no-flow conditions.) These sites represented a range
of physical habitat conditions, from highly altered to relatively
natural.
and sand) were rinsed in a 500-␮m sieve; large organic material
was rinsed, visually inspected, and discarded. Samples were
spread evenly across trays marked with grids and subsampled
randomly. Organisms were sorted under a dissecting micro-
scope. Subsamples of 100
Ϯ 20 organisms preserved in 95%
(v/v) alcohol were identified to the lowest practical taxon,
generally genus or species. A representative of every species
is maintained in a reference collection.
The criteria for selecting reference sites were relatively
undeveloped headwaters, no or minimal evidence of human
alteration of the physical habitat, presence of wide (Ͼ18 m)
Characterization of the biological condition of
macroinvertebrate assemblages
vegetated riparian zones, and no evidence of pollution [21].
None of the Aberjona watershed sites met all these criteria,
although site H7 had wide vegetated riparian zones and little
evidence of pollution or human alteration of the physical hab-
itat. Due to the intensity of human development in the Aber-
jona watershed, four reference sites outside the watershed but
within the Boston Basin ecoregion were selected to represent
minimally impaired conditions (Fig. 1b). These four stream
segments are located in sparsely developed residential areas:
On Sawmill Brook 250 m upstream of Monument Street in
Concord (MA, USA) (site C); on the Fish River 1.5 km up-
stream of Chandler Road in Andover (MA, USA) (site F); on
Mill River at Miller Street in Norfolk (MA, USA) (site M);
and on Trout Brook at Haven Street in Dover (MA, USA) (site
T). More detailed maps of these locations have been published
elsewhere [17]. These locations were selected after consulta-
tion with resource managers and scientists from the U.S. Army
Corps of Engineers, the U.S. Geological Survey, the Massa-
chusetts Department of Environmental Protection (MA DEP),
and the U.S. EPA. Geological and hydrological properties (i.e.,
stream order, drainage area, and gradient) were similar for both
Aberjona watershed sites and reference sites (U.S. Geological
Survey maps: Reading, MA, USA [1987]; and Boston, MA-
North, USA [1985]).
An aggregate index of the biological condition of macroin-
vertebrate assemblages was constructed using a method similar
to that described by Barbour et al. [23] and Gibson et al. [24].
The complete method is described by Rogers [17]. Briefly,
biological condition was characterized with an aggregate index
composed of benthic metrics selected from among four cate-
gories: Taxa richness, composition metrics, tolerance/intoler-
ance measures, and feeding measures. At least one metric from
each category was included in the index. Twelve candidate
metrics were chosen for testing based on widespread use [22]
and their applicability to the Aberjona watershed (Appendix).
The candidate metrics were tested for their ability to dis-
criminate impaired from unimpaired conditions and to provide
nonredundant information using the method described by Bar-
bour et al. [23]. The metrics that resulted in providing the most
useful information for inclusion into the biological index were
the total number of macroinvertebrate taxa; percentage of or-
ganisms that were Ephemeroptera, Plecoptera, or Trichoptera;
percentage of organisms considered to be tolerant; percentage
of Trichoptera within the relatively tolerant subgroup Hy-
dropsychidae; and number of scraper and piercer taxa. These
five metrics were normalized into dimensionless scores based
on the scoring method described by Karr et al. [5] and Karr
[6]. Table 1 displays statistics for the observed values and
shows how each metric was scored. Figure 3 shows the score
that each site received for each of the five metrics; the aggre-
gate index of biological condition is the sum of the scores for
the five metrics.
Macroinvertebrate sampling, subsampling, and species
identification
Macroinvertebrate sampling was performed in August 1996
according to the protocols used by the MA DEP [10]. The
protocols are similar to Method 7.2 Multihabitat Approach:
D-Frame Dip Net described by Barbour et al. [4,22]. Briefly,
the multihabitat approach consisted of sampling macroinver-
tebrate habitats, such as plant roots in undercut stream banks,
streambank vegetation where it hangs over into the stream,
rocks or cobble in the stream, snags, floating or submerged
vegetation, and sand, in proportion to their visually estimated
representation in the stream. Several of these macroinverte-
brate habitat types are illustrated in Figure 2. The stream was
visually assessed, and approximate percentages were assigned
to in-stream habitat belonging to each category. The habitat
type that was assigned the highest percentage was considered
to be the dominant habitat type. Samples were collected by
kicking the substrate or jabbing with a rectangular dip net
Sampling for metal analysis: Sediments from the stream
bottom and vegetative macroinvertebrate habitats
At each site, the finest-grained sediments from the bottom
of the stream were identified by visual inspection, and the top
2 to 3 cm were sampled using a Russian corer. Sediment was
transferred to glass jars with a clean, plastic spatula and trans-
ported to the laboratory on ice.
We sampled the dominant macroinvertebrate habitat at each
site for metal analysis. Table 2 lists the habitat types that were
sampled for macroinvertebrates using standard protocols
[4,10] and describes the technique that we developed to sample
the habitats for metal analysis. Several of these macroinver-
tebrate habitat types are illustrated in Figure 2. Using polyvinyl
chloride gloves, samples of the plant material (along with sed-
iments suspended within the plant material) that constituted
(width, 0.5 m; height, 0.3 m; mesh size, 500 ␮m). A total of

Contaminants and physical alteration in urban streams
Environ. Toxicol. Chem. 21, 2002
1159
Fig. 2. Physical habitat parameters in (a) high-quality habitat and (b) low-quality habitat. a. In a high-quality habitat, rocks, fallen branches, and
undercut banks provide fish many niches for feeding, laying eggs, and refugia (in-stream cover). Snags, submerged logs, and other hard substrates
provide habitat for macroinvertebrates (epifaunal substrate). The variety of velocity–depth combinations (i.e., mixture of slow-moving pools with
faster and shallower areas) provides a stable and diverse aquatic environment and moderates surges in flow associated with storms. Diverse and
abundant bank vegetation provides habitat (including plant roots in undercut stream banks and vegetation hanging over into the water) and
organic inputs (leaves provide food for shredders and other macroinvertebrates), and vegetation roots hold soil in place and prevent erosion
(bank vegetative protection, bank stability, and riparian vegetative zone width). b. In a low-quality habitat, the stream lacks riparian vegetation
(bank vegetative protection, bank stability, and riparian vegetative zone width). Buildings and roads contribute to sediment deposition, which
creates an unstable environment for many organisms. Embeddedness, caused by the deposition of fine sediments, reduces or degrades available
habitats. Straightened stream channels (channel alteration) have fewer velocity–depth combinations. The lack of submerged branches, rocks, and
undercut banks results in little habitat for fish or macroinvertebrates (in-stream cover, epifaunal substrate). The water level is low (poor channel
flow status). Channel flow status is important, because it determines whether habitats such as cobbles, overhanging bank vegetation, and undercut
bank roots are submerged (available for aquatic fauna) or exposed to the air (unavailable for aquatic fauna).
each vegetative habitat were grasped and pulled, or gently
scraped, from their attachment points, placed in glass jars, and
transported on ice (Table 2). Sediments associated with the
vegetative material in macroinvertebrate habitats were not
washed off.
Sample preparation and analysis of metal concentrations
Samples were held under refrigeration for no more than 2
d before being dried to constant weight at 85ЊC. Five grams
of each dried sample were homogenized with a Spex mixer
Table 1. Descriptive statistics and scores for the core metrics
Statistics
50%
Score
Metric
Min
5%
95%
Max
6
4
2
0
No. of total taxa
15
0
2
0
0
17
0
4
2
0
23
4
20
100
3
34
21
54
100
6
37
54
69
100
6
Ͼ
25
16
24
24
17–25
10–16
24–47
24–47
3–4
9–16
5–9
48–72
47–73
2
Ͻ
Ͻ
9
5
% EPTa
Ͼ
Ͻ
Ͻ
% Tolerant taxa
% Hydropsychidae/Trichoptera
No. of scraper & piercer taxa
Ͼ
Ͼ
72
73
Ͼ
4
0–1
a EPT
ϭ Ephemeroptera, Plecoptera, Trichoptera.

1160
Environ. Toxicol. Chem. 21, 2002
C.E. Rogers et al.
Accuracy and precision of the metal concentration
analyses
Average measured values of metals in the NIST samples
compared favorably ( 5% difference between measured and
Ͻ
reported values for all metals except Cu) with values reported
by the NIST (Table 3). Copper concentrations were measured
precisely, but our measurements averaged approximately 20%
higher than the reported NIST values. This did not have a
significant impact on toxicity estimates, because less than 20%
of the total toxic units for each sample were attributable to
[Cu]; thus, even errors of 20% in [Cu] would result in errors
of less than 4% in total toxic units. Figure 4 shows the con-
centrations of each element in sediment and vegetative habitat
samples at each sampling site.
Fig. 3. Biological condition: Index scores for each study site. The
total index score is the sum of the scores for the five core metrics.
Table 1 shows the derivation of these scores and provides descriptive
Metal concentrations: Use of benchmark concentrations to
develop normalized metal concentrations for statistical
analysis
statistics for each metric. EPT
Trichoptera.
ϭ Ephemeroptera, Plecoptera, and
A measure of overall metal concentrations in stream-bottom
sediments and in vegetative habitats was developed for use in
regression analyses. For each individual metal, if [(observed
mill (CertiPrep, Metuchen, NJ, USA) using a tungsten carbide
ball mill for 5 min. One-half gram of Copolywax binder
௢
concentration)/ERM]
Ͼ 1, then the probability of sediment
(Cargille TAB-PRO, Cedar Knolls, NJ, USA) was added to
the sample, which was mixed again for 1 min. This mixture
was then poured into an aluminum sample cup (diameter, 31
mm) and pressed for 1 min using 12 t of force while pulling
a vacuum created by a rotary pump.
toxicity is high. Therefore, a toxic unit index was constructed
by adding these ratios for multiple contaminants:
Toxic unit index
ϭ
[M1]/[ERM1]
[M3]/[ERM3]
where [Mi] is the observed concentration of the
ϩ
[M2]/[ERM2]
ϩ
ϩ
. . .
(1)
Elemental concentrations in each pellet were determined
by wavelength dispersive x-ray fluorescence (XRF; used for
As, Cd, and Zn) using a Philips PW1480 (Philips Analytical,
Almelo, The Netherlands) wavelength dispersive instrument
and Uniquant (Omega Data Systems BV, Veldehoven, The
Netherlands) data-processing software, or by energy dispersive
XRF (used for Cr, Cu, and Pb) using a Spectro X-Lab 2000
instrument (St. Lawrence, PQ, Canada) with Turboquant
(Spectro, Kleve, Germany) data-processing routines. Standard
reference material 2709 (San Joaquin soil), supplied by the
National Institute of Standards and Technology (NIST; Gai-
thersburg, MD, USA), was used to monitor instrumental re-
sponse during the data collection period and to verify cali-
bration accuracy. The average and standard deviation of 15
measurements (covering the period when samples were ana-
lyzed) of the concentrations of As, Cd, Cr, Cu, Pb, and Zn
were compared to the concentrations and 95% prediction in-
terval reported by the NIST for this reference material (Table
3). The optimal detection limits for As, Cd, Cr, Cu, Pb, and
Zn ranged from 0.2 to 1.0 mg/kg dry weight.
i
-th metal and
[ERMi] is the ERM value for that metal. Adding these ratios
together to create a toxic unit index is an approximation, be-
cause toxic effects are not necessarily additive: They could be
synergistic or antagonistic. Ingersoll et al. [20] found that for
freshwater sediments collected in the field, sediments that had
multiple (two or more) exceedances were more frequently tox-
ic (to Hyalella azteca in laboratory tests) than sediments with
only one exceedance (i.e., only one metal was present in a
concentration
Ͼ ERM). Higher metal concentrations were as-
sociated with greater frequency of effects; thus, although tox-
icity may not increase linearly with metal concentrations, a
positive relationship exists. Therefore, it is reasonable to use
a toxic unit index as a measure of metal contamination for
metals that exceed toxic thresholds.
Assessment of physical habitat condition
Physical habitat assessment was performed at each site ac-
cording to the protocols used by the MA DEP [10] but modified
for slow-flow, low-gradient streams. These protocols use a
Table 2. Types of vegetative faunal habitat identified for chemical analysis
Habitat type
Sites sampled
Description of sampling technique
Overhanging bank vegetation
Submerged vegetation
Undercut bank roots
Ab4
Vegetation was gently pulled free of the bank, retaining as much
of the sediment suspended in the vegetation as possible.
Submerged vegetation was grasped and pulled from its
attachment point.
Handfuls of muddy undercut bank roots were grabbed and
pulled from the banks.
Ab3, H7, M, T
Ab2, Ab5, Ab6, H3, H8, Tr3
H2, H3, H6, C, F
Vegetation growing on rocks in
shallow sections of streams
Floating vegetation
Vegetation was pulled and gently scraped from rocks.
Ab1, H4, H5, Tr1, Tr2
Handfuls of floating vegetation (e.g., watercress/Cruciferae
nasturtium) were pulled from the water.

Contaminants and physical alteration in urban streams
Environ. Toxicol. Chem. 21, 2002
1161
Table 3. Accuracy and precision of the metal concentration analyses
Ratio of
measured
to
Average measured
NIST
SDa
(mg/kg dry wt)
Reported
NIST SD
Ϯ
Ϯ
Analysis
methodb
Detection
limitc
Metal
(mg/kg dry wt) reported
As
Cd
Cr
Cu
Pb
Zn
18
Ϯ
2,
n
ϭ
4
18
0.4
130
35
19
Ϯ
1
1.00
—
1.01
1.20
1.00
0.96
EDXRF
EDXRF
WDXRF
WDXRF
WDXRF
EDXRF
3
0.5
23
1
2
1
Ϯ
Ϯ
Ϯ
Ϯ
Ϯ
Ϯ
2
0.2
6
0.4
0.1
0.1
d
Below detection limit,
131
42
19
101
n
ϭ
4
Ϯ
Ϯ
Ϯ
Ϯ
10,
n
ϭ
15
Ϯ
Ϯ
Ϯ
Ϯ
4
1
1
3
3,
4,
4,
n
n
n
ϭ
ϭ
ϭ
15
15
4
106
a NIST
ϭ
National Institute of Standards and Technology; SD
ϭ energy dispersive x-ray fluorescence; WDXRF ϭ wavelength dispersive x-ray fluorescence.
ϭ
standard deviation.
b EDXRF
c Ideal detection limits were lower (0.2–1.0 mg/kg dry wt). The values in this column are actual detection
limits observed in typical samples.
d The detection limit for Cd was higher than the National Institute of Standards and Technology (NIST)
standard, so this ratio could not be determined.
scoring sheet that describes the conditions associated with each
score. For example, the embeddedness score requires an eval-
uation of the percentage of snags and submerged logs that are
surrounded by fine sediment. If less than 25% of these habitats
are surrounded, then a score of 16 to 20 is assigned. Con-
versely, if more than 75% of these surfaces are surrounded by
fine sediment, a score of 0 to 5 is assigned. Scores are deter-
mined by visually matching the observed conditions to those
described by the scoring sheet. Physical habitat condition is
judged to be poor (score of 0–5), marginal (score of 6–10),
suboptimal (score of 11–15), or optimal (score of 16–20).
Physical habitat features are illustrated in Figure 2. Ten habitat-
quality parameters (in-stream cover, epifaunal substrate, em-
beddedness, channel alteration, sediment deposition, variety
of velocity–depth combinations, channel flow status, bank veg-
etative protection, bank stability, and riparian vegetative zone
width) were scored from 0 to 20. The relationship of Figure
2 to these 10 habitat-quality parameters is explained in the
caption. The physical habitat-quality index is the sum of the
scores for the 10 habitat parameters (thus ranging from 0–
200). Detailed results of the scores for individual parameters
are displayed in Figure 5. Water-quality parameters were also
measured (i.e., temperature, dissolved oxygen concentrations,
pH, and conductivity), but these results were unremarkable
[17] and, thus, are not presented here.
The biological condition index was also regressed on the
total habitat-quality score and on both total habitat quality and
the toxic unit indices to determine whether habitat quality was
a statistically significant explanatory variable and whether its
inclusion yielded a better regression model. The forms for
these regressions were
Biological condition index
ϭ ␤0 ϩ ␤1 · TotalHabScore
ϩ
error
(4)
and
Biological condition index
ϭ ␤0 ϩ ␤1 · TOX ϩ ␤2 · TotalHabScore
ϩ
error
(5)
where ␤2 is also a regression coefficient, TotalHabScore is the
total habitat-quality score, and TOX is TOXSed or TOXVeg. We
only report in detail the results for TOXVeg for Equation 5,
because neither Equation 2 nor Equation 5 showed TOXSed to
be a significant explanatory variable.
RESULTS
Figures 3 to 5 present the scores and subscores for biolog-
ical condition (Fig. 3); the concentrations of As, Cd, Cr, Cu,
Pb, and Zn in sediment and vegetative samples (Fig. 4); and
the scores and subscores of physical habitat quality (Fig. 5).
These results are summarized in Figure 6. The chemical (Fig.
6a), biological (Fig. 6b), and physical (Fig. 6c) indices provide
a basis for assessing the influence of chemical and physical
degradation on macroinvertebrates.
As expected, sites located on the Aberjona River had ele-
vated concentrations of metals and, thus, high numbers of toxic
units (Figs. 4 and 6a). For the ERM values of As, Cd, Cr, Cu,
Pb, and Zn at least one exceedance occurred for each metal.
Exceedances were most frequent and of greatest magnitude
for sites Ab1 to Ab6.
Regression
The aggregate index of biological condition was regressed
using Microsoft௡ Excel (Redmond, WA, USA) on the indi-
vidual toxic unit indices for stream-bottom sediments and for
vegetative habitat samples (e.g., undercut bank roots, over-
hanging bank vegetation) to determine if either toxic exposure
estimate was a statistically significant predictor. The forms for
these regressions were
Biological condition index
If metal concentrations in bottom sediments and/or vege-
tative habitats were the dominant factor controlling biological
condition, then biological condition would be expected to be
impaired at contaminated sites and most impaired at the most
contaminated sites. However, this was not the case (Fig. 6a
and b). Biological condition at sites Ab1 to Ab6 was poorer
than biological condition at the reference sites (C, F, M, and
T), but biological condition at some uncontaminated sites
(Tr1–Tr3) was equally poor. Moreover, biological condition
ϭ ␤0 ϩ ␤1 · TOXSed
ϩ
error
(2)
and
Biological condition index
ϭ ␤0 ϩ ␤1 · TOXVeg
0 is the intercept,
ϩ
error
(3)
where
␤
␤
1 is a regression coefficient, TOXSed
is the toxic unit index for sediments, and TOXVeg is the toxic
unit index for vegetative habitats.

1162
Environ. Toxicol. Chem. 21, 2002
C.E. Rogers et al.
Fig. 4. Metal concentrations (mg/kg dry wt) at study sites, with con-
centrations in vegetative samples illustrated as solid diamonds and
concentrations in sediment samples illustrated as stars, for (a) As, (b)
Cd, (c) Cr, (d) Cu, (e) Pb, and (f) Zn. To facilitate comparisons with
the effects range median (ERM) values (toxic thresholds), the ERM
Fig. 4. Continued.
the graphs (Figs. 3 to 6) displays physical habitat condition
decreasing from left to right for all sites except Ab1 to Ab6,
which are shown at the far right. A general, but scattered,
relationship between biological condition and habitat quality
can be seen in Figure 6b and c. That some of the contaminated
sites had relatively good habitat quality and poor biological
condition is consistent with the existence of a stressor other
than habitat degradation (e.g., contamination).
for each metal is used as the interval for gridlines on the
y-axis (e.g.,
50 for As, 4 for Cd). Sed sediment; Veg vegetation.
ϭ
ϭ
was not the most impaired at the most contaminated sites (Ab3
and Ab4).
The observed trends (Fig. 6) do, however, suggest that metal
contamination had an adverse effect on macroinvertebrates,
but that this effect was confounded by variability in physical
Similar levels of biological impairment were observed at
sites with severe contamination or physical habitat impairment
habitat condition. The ordering of the sites along the x-axis in

Contaminants and physical alteration in urban streams
Environ. Toxicol. Chem. 21, 2002
1163
Fig. 5. Physical habitat condition at each study site. The total index
score is the sum of the scores for the 10 habitat metrics. Prot
protection; Veg vegetation.
ϭ
ϭ
or with moderate contamination and habitat impairment. The
reference sites (C, F, M, and T) were selected to represent
minimally impaired conditions for the Boston Basin ecoregion
[21]. As expected, these sites had low concentrations of con-
taminants, high scores for physical habitat quality, and high
scores for biological condition. Sites with high concentrations
of contaminants and moderately high scores for physical hab-
itat quality have lower scores for biological condition (i.e.,
Ab3 and Ab4). Relatively uncontaminated sites with poor hab-
itat quality (e.g., H8, Tr1, H2, and Tr2) tend to have poor
biological condition. Sites with intermediate levels of both
contamination and habitat-quality impairment (i.e., Ab6, Ab1,
Ab2, and Ab5) also have low scores for biological condition.
Regression of biological condition on physical and
chemical stressors
To determine whether biological condition was significantly
related to metal concentrations in sediments from the stream
bottom or in vegetative habitats, the aggregate macroinver-
tebrate index was regressed using Equations 2 and 3 (see Ma-
terials and Methods section). The results in Table 4 show that
the index of biological condition was statistically dependent
(at the 95% confidence level) on the toxic unit index for ma-
croinvertebrate habitat samples, but not on the toxic unit index
for bottom-sediment samples. In addition, the variance (r2
explained by the regression rises from 8 to 22% when TOXVeg
is used in place of TOXSed
)
.
To determine whether physical habitat quality was an im-
portant explanatory variable for biological condition, biolog-
ical condition was first regressed on physical habitat quality
alone (i.e., the total habitat score) and then on both the total
habitat score and TOXVeg using Equations 4 and 5 (see Ma-
terials and Methods section). The results in Table 4 indicate
that total habitat score alone is a significant explanatory var-
iable (at both the 95% and 99% confidence levels) that ac-
counts for approximately 30% of the variance. When both
TOXVeg and total habitat score are included in the model, both
retain their statistical significance (at the 95% confidence
level), and the variance explained jumps to 49% (Table 4).
These results leave a significant fraction of the variance un-
explained, but the combined model represents a great im-
provement over previous models (e.g., TOXSed, TOXVeg, or
habitat quality alone).
Fig. 6. The influence of toxic metals on biological condition becomes
apparent if physical habitat quality is considered. a. Toxic unit indices
exceed six only for sites Ab1 to Ab6. (Significant contamination is
indicated at these sites, because if each of the six metals were present
in a concentration exactly equal to its effects range median [ERM]
value, then the index for that site would be equal to six.) b. Based
on biological and chemical evidence alone, little support is found for
the hypothesis that toxic metals have adversely impacted biological
condition. c. Physical habitat data help to explain poor biological
condition at uncontaminated sites. Regression of biological condition
on chemical and physical condition provides a more refined analysis
(see Table 4). TOX
࿞Sed ϭ toxic unit index for sediments; TOX࿞Veg
ϭ
toxic unit index for vegetative habitats.

1164
Environ. Toxicol. Chem. 21, 2002
C.E. Rogers et al.
Table 4. Regression of the biological condition index (BCI) on toxic unit index for sediments (TOXSed),
toxic unit index for vegetative habitats (TOXVeg) and total habitat scorea
Single-variable linear regressionb
Single independent
variable
Parameter
estimate
Standard
error
Intercept
p
r2
TOXSed
TOXVeg
Total habitat score
17.88
18.78
3.46
Ϫ
Ϫ
0.41
0.42
0.11
0.33
0.19
0.04
0.22
0.04
0.01
0.08
0.22
0.30
Multivariable linear regressionc
Variables in the
multivariable regression
Parameter
estimate
Standard
error
p
r2
Intercept
TOXVeg
Total habitat score
6.96
0.39
0.10
4.29
0.16
0.03
0.12
0.02
0.008
0.49 for model with both
variables
Ϫ
a BCI was significantly dependent on TOXVeg and total habitat score, but not on TOXSed. Explanatory
power of the model increased to 49% when TOXVeg and total habitat score were both included in the
model.
b Regressed BCI on each parameter singly. Information for each of the three regression equations is
shown as a row in the table. Refer to Equations 2 to 4 in the text.
c Regressed BCI on both TOXVeg and total habitat score. Table shows the results for the regression with
the intercept and parameter estimates shown as rows;
5 in the text.
r
2 is given for the full equation. Refer to Equation
DISCUSSION
nonindigenous species, and human alterations in streamflow.
Because they may be correlated, the effects of organic con-
taminants and toxic metals might be difficult to distinguish.
The two sites with the highest metal concentrations (Ab3 and
Ab4) are downstream of both the Industri-Plex 128 and Wells
G&H Superfund sites (Woburn, MA, USA), where other stud-
ies have found high concentrations of both organic and metal
contaminants [25–27].
Influence of metal contamination and physical habitat
quality on biological condition
We sought to improve estimates of the exposure of ma-
croinvertebrates to metals in their environment by measuring
metal concentrations in the same habitats from which ma-
croinvertebrates were collected. We hypothesized that biolog-
ical condition would be more strongly correlated with metal
concentrations in vegetative habitats than with metal concen-
trations in bottom sediments. We found that biological con-
dition was significantly dependent (95% confidence level, r2
Another limitation is that total metal concentrations were
used as measures of exposure, because these measures do not
account for factors such as complexation with iron oxyhy-
droxides, organic matter, or sulfide. Processes controlling tox-
icity have been extensively studied for fine-grained sediments
such as those found at the bottom of a stream (e.g., [28–34]).
These studies can be very useful for characterizing exposure
of deposit-feeding animals living in sediments, but they are
less applicable to characterizing exposures in vegetative hab-
itats. A lack of bioavailable metals in bottom sediments does
not necessarily imply a lack of bioavailable metals in vege-
tative habitats. Indeed, undercut bank roots, watercress, and
overhanging bank vegetation exist near the water surface of
streams. These habitats differ from surficial bottom sediments
both chemically and physically. For example, they are likely
to have greater exposure to oxygen and, thus, lower concen-
trations of sulfide than those in bottom sediments. Macroin-
vertebrates in vegetative habitats may be exposed to metals in
the water associated with the habitat, to metals attached to
sediments associated with the vegetation, to metals adsorbed
to plants, or to metals in plants. The simple empirical rela-
tionship between biological condition and metal concentra-
tions in vegetative habitats, thus, does not address the com-
plexity of bioavailability or multiple exposure pathways. De-
spite limitations, bulk chemistry measurements are useful as
triggers for further analysis (e.g., [19]), and we propose that
measurement of metal concentrations in vegetative habitats
could usefully supplement analyses of bottom sediments.
It must be also acknowledged that the index of biological
condition and the total habitat score are both surrogate mea-
ϭ
0.22) on contaminants in vegetative habitats, but not on
contaminants in bottom sediments. This supports our hypoth-
esis and suggests that, in at least some cases, analysis of metal
concentrations in vegetative habitats could improve the as-
sessment of the ecological risks of contaminants in streams.
We hypothesized that physical habitat impairment is also
an important determinant of biological condition and, thus,
could mask contaminant effects in some cases. Thus, we in-
cluded physical habitat quality in the regression model. We
found that biological condition was significantly dependent on
physical habitat quality (95% confidence level, r2
ϭ 0.30).
Inclusion of both the toxic unit index for contaminants in
vegetative habitats and the total habitat-quality score increased
the r2 for the regression model from 0.22 (regression on the
toxic unit index alone) to 0.49 (regression on both variables).
Inclusion of physical habitat quality is not a panacea; it does
not explain all the variance in biological condition scores.
However, it does significantly improve our understanding of
the factors influencing biological condition, and it does help
to reveal the effects of contaminants.
Method strengths and limitations
The measures of physical habitat quality and metal con-
tamination that were used in our study explain approximately
ϭ 0.49) in biological condition. The un-
explained variance could be partly due to factors not included
in our study, such as organic contaminants, nutrients, invasive
half the variance (r2

Contaminants and physical alteration in urban streams
Environ. Toxicol. Chem. 21, 2002
1165
sures that aggregate a variety of indicators. The index of bi-
ological condition used in this study encompasses several as-
pects of the structure and function of the macroinvertebrate
community, making it a valuable tool for assessing a stream’s
ability to support aquatic life [23]. As an aggregate index, it
provides only an overall picture of stream condition, not a
detailed portrait of specific responses to physical and chemical
stressors, although individual metrics are often response sig-
natures [35]. Similarly, the habitat condition index is the sum
of a set of scores, each of which is an imperfect indicator for
the quality of stream habitat for macroinvertebrates. When risk
managers require more detailed information regarding specific
causes and effects, additional analysis will be needed.
Analysis of individual components of the biological con-
dition, total habitat, and toxic unit indices may provide ad-
ditional insight. As an example, the number of scraper and
piercer taxa was regressed on the total habitat score and the
toxic unit index by Rogers [17], and a statistically significant
relationship was found for both. Scraper and piercer taxa feed
on living plant material, so they are expected to be important
components of macroinvertebrate communities in streams
where plant habitat is more abundant than riffle habitat (e.g.,
the Aberjona watershed and other low-gradient streams). The
observed dependence of the number of scraper and piercer
taxa on physical habitat quality may be explained by the fact
that a loss of riparian vegetation (which would be reflected in
an impaired physical habitat-quality score) involves a loss of
habitat for scrapers and piercers. Similarly, the observed de-
pendence on the toxic unit index may be due to the fact that
scrapers and piercers living in vegetative macroinvertebrate
habitats such as undercut bank roots and overhanging stream-
bank vegetation would be exposed to contaminants in these
habitats.
but also to put the risks of contamination in context with other
risks. We found similar levels of biological impairment for
sites with severe contamination or physical habitat impairment
or with moderate contamination and habitat impairment. Two
Superfund sites drain into the Aberjona River, including one
of national fame. (Contamination associated with the Wells
G&H site was the subject of a best-selling book [37] and a
major motion picture, A Civil Action.) It may seem surprising
that physical habitat degradation in the Aberjona watershed,
which has experienced a level of urbanization comparable to
that in the suburbs of many eastern U.S. cities, appears to be
associated with a similar level of biological impairment as has
been caused by contamination in this watershed. Little regu-
latory attention has been given to protecting habitat from deg-
radation by channelization, loss of riparian vegetation, sedi-
mentation, and other alterations [6,21,38]. Yet, the vital im-
portance of physical habitat quality has not gone unrecognized.
The loss of habitat quality has resulted in extinctions, local
extirpations [39], and population reductions [40,41] of fish
species and other aquatic fauna [42] in the United States [38].
Consequently, although chemical pollution continues to be a
problem in many freshwater steam ecosystems, many inves-
tigators believe that habitat degradation is presently respon-
sible for more ecological impairment than is caused by chem-
ical pollution (e.g., [38,43]).
Comparative risk information can inform remediation and
restoration efforts
Information regarding comparative risk is available to man-
agers only if assessments address more than one stressor. Man-
agers can use these results to make informed decisions related
to the protection and restoration of stream ecosystems. If a
contaminated site is also physically impaired, it might be un-
realistic to expect chemically oriented remediation to restore
biological condition. In addition, the process of chemically
remediating long stretches of contaminated streams might in-
duce further physical degradation of stream habitats. Physical
remediation also cannot substitute for chemical remediation
in addressing human health risks. However, once health risks
have been addressed, managers seeking to restore the biolog-
ical resources of streams can balance the restoration of physical
habitat, reduction of contaminant concentrations, and spatial
distribution of each effort to maximize the benefits obtained.
Resources required to apply these methods
In deciding whether to use the methods recommended in
this paper, risk assessors and risk managers will consider the
value of the information gained relative to the cost of obtaining
that information. We believe that a major strength of the meth-
ods proposed here is their relatively low cost. Conducted in
conjunction with chemical and/or biological sampling, phys-
ical habitat-quality assessment by a trained investigator adds
only approximately 15 min to each site visit. The investigator
is required to visually assess each of 10 parameters, to assign
a score for each parameter, and to sum the scores to calculate
the total habitat score. An initial investment in training is also
necessary [36]. Chemical analysis of vegetative habitats costs
no more than chemical analysis of sediments. The dominant
macroinvertebrate habitat is identified during macroinverte-
brate sampling. Collecting a sample of the dominant vegetative
habitat along with the bottom-sediment samples would require
less than 10 min in most cases. Sample processing and analysis
costs are equal for sediments and vegetative habitats (using
the method we developed for this study). In addition, currently
available, field-portable XRF units are capable of analyzing
many metals regulated under the Comprehensive Environ-
mental Response, Compensation, and Liability Act on-site at
ERL/ERM concentration levels, potentially enabling the rapid
characterization of metal concentrations required for toxic unit
analysis in the field.
Acknowledgement—This research was supported in part by the Na-
tional Institute of Environmental Health Sciences Superfund Basic
Research Program (P42-ES04675) and the Alliance for Global Sus-
tainability. C.E. Rogers thanks the National Science Foundation for
supporting this research with her graduate fellowship and the U.S.
EPA for support during the final preparation of this manuscript. The
authors also thank Margaret Rogers, Terry Keating, Susan Norton,
and two anonymous reviewers for comments on the manuscript.
REFERENCES
1. Ohio Environmental Protection Agency. 1989. The qualitative
habitat evaluation index (QHEI): Rationale, methods, and appli-
cation. Columbus, OH, USA.
2. Ohio Environmental Protection Agency. 1987. Biological criteria
for the protection of aquatic life: Volumes I–III. Columbus, OH,
USA.
3. Barbour MT, Stribling JB. 1991. Use of habitat assessment in
evaluating the biological integrity of stream communities. EPA
440-5-91-005. U.S. Environmental Protection Agency, Washing-
ton, DC.
4. Barbour MT, Gerritsen J, Snyder BD, Stribling JB. 1999. Rapid
bioassessment protocols for use in streams and wadable rivers:
Periphyton, benthic macroinvertebrates, and fish, 2nd ed. EPA
Physical habitat quality strongly influences biological
condition
The inclusion of physical habitat quality in the assessment
not only helped to reveal the effects of metal contaminants

1166
Environ. Toxicol. Chem. 21, 2002
C.E. Rogers et al.
841/B/98/010. U.S. Environmental Protection Agency, Washing-
ton, DC.
5. Karr JR, Fausch KD, Angermeier PL, Yant PR, Schlosser IJ. 1986.
Assessing biological integrity in running waters: A method and
its rationale. Special Publication 5. Illinois Natural History Sur-
vey, Champaign, IL, USA.
6. Karr JR. 1991. Biological integrity: A long-neglected aspect of
water resource management. Ecol Appl 1:66–84.
7. Plafkin JL, Barbour MT, Porter KD, Gross SK, Hughes RH. 1989.
Rapid bioassessment protocols for use in streams and rivers, ben-
thic macroinvertebrates and fish. EPA/440/4-89-001. U.S. Envi-
ronmental Protection Agency, Washington, DC.
8. U.S. Environmental Protection Agency. 1998. Guidelines for eco-
logical risk assessment. EPA/630/R-95/002F. Washington, DC.
9. Florida Department of Environmental Protection. 1996. Standard
operating procedures for biological assessment. Biology Section.
Tallahassee, FL, USA.
24. Gibson GR, Barbour MT, Stribling JB, Gerritsen J, Karr JR. 1996.
Biological criteria: Technical guidance for streams and rivers.
EPA/822-B-94-001. U.S. Environmental Protection Agency, Of-
fice of Science and Technology, Washington, DC.
25. Kim HK, Hemond HF, Krumholz LR, Cohen BA. 1995. In-situ
biodegradation of toluene in a contaminated stream. 1. Field stud-
ies. Environ Sci Toxicol 29:108–116.
26. Tay STL, Hemond HF, Polz MF, Cavanaugh CM, Dejesus I,
Krumholz LR. 1998. Two new mycobacterium strains and their
role in toluene degradation in a contaminated stream. 1998. Appl
Environ Microbiol 64:1715–1720.
27. Tay STL, Hemond HF, Polz MF, Cavanaugh CM, Krumholz LR.
1999. Importance of Xanthobacter autotrophicus in toluene bio-
degradation within a contaminated stream. Syst Appl Microbiol
22:113–118.
28. Metal Bioavailability in Sediments (special section). 1996. En-
viron Toxicol Chem 15:2053–2233.
10. Massachusetts Department of Environmental Protection. 1995.
Massachusetts DEP preliminary biological monitoring and as-
sessment protocols for wadable rivers and streams. Grafton, MA,
USA.
11. Mid-Atlantic Coastal Streams Workgroup. 1996. Standard oper-
ating procedures and technical basis: Macroinvertebratecollection
and habitat assessment for low-gradient non-tidal streams. Del-
aware Department of Natural Resources and Environmental Con-
servation, Dover, DE, USA.
12. Sansone U, Belli M, Riccardi M, Alonzi A, Jeran Z, Radojko J,
Smodis B, Montanari M, Cavolo F. 1998. Adhesion of water-
borne particulates on freshwater biota. Sci Total Environ 219:
21–28.
13. Griffith GE, Omernik JM, Pierson SM, Kiilsgaard CW. 1994.
Massachusetts ecological regions project. Publication 17587-74-
70-6/94. Massachusetts Department of Environmental Protection,
Division of Water Pollution Control, Boston, MA, USA.
14. Durant JL, Zemach JJ, Hemond HF. 1990. The history of leather
industry waste contamination in the Aberjona watershed: A mass
balance approach. Journal of the Boston Society of Civil Engi-
neers 5:41–65.
15. Aurilio AC, Durant JL, Hemond HF, Knox ML. 1995. Sources
and distribution of arsenic in the Aberjona watershed, eastern
Massachusetts. Water Air Soil Pollut 81:265–282.
16. Knox ML. 1991. The distributional and depositional history of
metals in surface sediments of the Aberjona river watershed. MS
thesis. Massachusetts Institute of Technology, Cambridge, MA,
USA.
17. Rogers CE. 1998. A method to assess the ecological integrity of
urban watersheds that integrates chemical, physical and biological
data. PhD thesis. Massachusetts Institute of Technology, Cam-
bridge, MA, USA.
18. Long ER, Morgan LG. 1990. The potential for biological effects
of sediment-sorbed contaminants tested in the National Status and
Trends Program. NOS OMA 52. National Oceanic and Atmo-
spheric Administration, Seattle, WA, USA.
29. Ankley GT. 1996. Evaluation of metal/acid-volatile sulfide rela-
tionships in the prediction of metal bioaccumulation by benthic
macroinvertebrates. Environ Toxicol Chem 15:2138–2146.
30. Chapman PM, Wang FY, Janssen C, Persoone G, Allen HE. 1998.
Ecotoxicology of metals in aquatic sediments: Binding and re-
lease, bioavailability, risk assessment, and remediation. Can J
Fish Aquat Sci 55:2221–2243.
31. Di Toro DM, Zarba CS, Hansen DJ, Berry WJ, Swartz RC, Cowan
CE, Pavlou SP, Allen HE, Thomas NA, Paquin PR. 1991. Tech-
nical basis for establishing sediment quality criteria for nonionic
organic chemicals using equilibrium partitioning. Environ Toxi-
col Chem 10:1541–1583.
32. Lee BG, Griscom SB, Lee LS, Choi HJ, Koh CH, Luoma SN,
Fisher NS. 2000. Influences of dietary uptake and reactive sulfides
on metal bioavailability from aquatic sediments. Science 287:
282–284.
33. U.S. Environmental Protection Agency. 1999. Integrated ap-
proach to assessing the bioavailability and toxicity of metals in
surface waters and sediments. EPA 822-E-99-001. Washington,
DC.
34. U.S. Environmental Protection Agency. 2000. An SAB report:
Review of an integrated approach to metals assessment in surface
waters and sediments. EPA-SAB-EPEC-00-05. Washington, DC.
35. Yoder CO, Rankin ET. 1995. Biological response signatures and
the area of degradation value: New tools for interpreting multi-
metric data. In Davis, WS and Simon T, eds, Biological Assess-
ment and Criteria: Tools for Water Resource Planning and De-
cision Making. Lewis, Boca Raton, FL, USA, pp 263–286.
36. Hannaford MJ, Resh VH. 1995. Variability in macroinvertebrate
rapid-bioassessment surveys and habitat assessments in a northern
California stream. J North Am Benthol Soc 14:430–439.
37. Harr J. 1996. A Civil Action. Vintage Books, New York, NY,
USA.
38. Rankin ET. 1995. Habitat indices in water resource quality as-
sessments. In Davis WS, Simon T, eds, Biological Assessment
and Criteria: Tools for Water Resource Planning and Decision
Making. Lewis, Boca Raton, FL, USA, pp 181–208.
39. Karr JR, Toth LA, Dudley DR. 1985. Fish communities of mid-
western rivers: A history of degradation. BioScience 35:90–95.
40. Rankin ET, Yoder CO, Mishne DA. 1992. Ohio water resource
inventory, status and trends. Ohio Environmental Protection
Agency, Division of Water Quality Planning and Assessment,
Columbus, OH, USA.
19. Long ER, MacDonald DD, Smith SL, Calder FD. 1995. Incidence
of adverse biological effects within ranges of chemical concen-
trations in marine and estuarine sediments. Environ Manag 19:
81–97.
20. Ingersoll CG, Haverland PS, Brunson EL, Canfield TS, Dwyer
FJ, Henke CE, Kemble NE. 1996. Calculation and evaluation of
sediment effect concentrations for the amphipod Hyalella azteca
41. Trautman MB. 1981. The Fishes of Ohio. The Ohio State Uni-
and the midge Chironomus riparius. J Gt Lakes Res 22:602–
623.
versity Press, Columbus, OH, USA.
42. Williams JD, Warren ML Jr, Cummings KS, Harris JL, Neves JL.
1993. Conservation status of freshwater mussels of the United
States and Canada. Fisheries 18:6–22.
43. Stevens WK. 1993. River life through U.S. broadly degraded.
New York Times (Tuesday January 26) B5–B8.
44. Clements WH. 1991. Community responses of stream organisms
to heavy metals: A review of observational and experimental
approaches. In Newman MC, McIntosh AW, eds, Ecotoxicology
of Metals: Current Concepts and Applications. CRC, Boca Ra-
ton, FL, USA, pp 363–391.
45. Fore LS, Karr JR, Wisseman RW. 1996. Assessing invertebrate
responses to human activities: Evaluating alternative approaches.
J North Am Benthol Soc 15:212–231.
21. Hughes RM, Whittier TR, Rohm CM, Larsen DP. 1990. A regional
framework for establishing recovery criteria. Environ Manag 14:
673–683.
22. Barbour MT, Gerritsen J, Snyder BD, Stribling JB. 1997. Revi-
sions to rapid bioassessment protocols for use in streams and
rivers: Periphyton, benthic macroinvertebrates, and fish. EPA/
841-D-97-002. U.S. Environmental Protection Agency, Washing-
ton, DC.
23. Barbour MT, Gerritsen J, Griffith GE, Frydenborg R, McCarron
E, White JS, Bastian ML. 1996. A framework for biological cri-
teria for Florida streams using benthic macroinvertebrates.
J
North Am Benthol Soc 15:185–211.

Contaminants and physical alteration in urban streams
Environ. Toxicol. Chem. 21, 2002
1167
APPENDIX
Twelve candidate metrics chosen for testing based on their poten-
tial importance in the Aberjona watershed
Benthic metric
Reason for inclusion in analysis
Total no. of taxa
No. of EPTa taxa
% EPT
Widespread use
Widespread use
Widespread use
% Dominant taxon
Hilsenhoff Biotic Index
% Filterers
Widespread use
Widespread use
Widespread use
% Orthocladiinae to
chironomids
% Orthoclads correlated positively
with contamination by As, Cr, Cu,
Pb, and Zn in several descriptive
studies of water pollution and to Cu
and Zn in experimental studies [44].
To account for the dominance of the
relatively pollution-tolerant
caddisflies in the Boston Basin
ecoregion. Hydropsychidae are
highly tolerant of heavy metals (Cr,
Hg, Pb, and Ni) [44].
% Hydropsychidae to
Trichoptera
No. of scraper and piercer Taxa that feed on living plant material
taxa
are especially important and
sensitive in streams where plant
habitat is more abundant than riffle
habitat.
% Clingers
To represent behavior (habit) measures
of the structure and function of the
benthic assemblage [45].
No. of intolerant taxa
A richness measure of the most
sensitive organisms that decreases
when perturbation increases.
A composition measure that increases
when the more sensitive organisms
are lost.
% Tolerant organisms (of
total fauna)
a EPT
ϭ Ephemeroptera, Plecoptera, Trichoptera.