Full text of DOI:10.1002/etc.5620210606

Environmental Toxicology and Chemistry, Vol. 21, No. 6, pp. 1147–1155, 2002
Printed in the USA
0
730-7268/02 $9.00 ϩ .00
ASSESSING RELATIONSHIPS BETWEEN HUMAN LAND USES AND THE DECLINE OF
NATIVE MUSSELS, FISH, AND MACROINVERTEBRATES IN THE CLINCH AND
POWELL RIVER WATERSHED, USA
J
EROME M. DIAMOND,*† DAVID W. BRESSLER,† and VICTOR B. SERVEISS
Tetra Tech, 10045 Red Run Boulevard, Suite 110, Owings Mills, Maryland 21117, USA
U.S. Environmental Protection Agency (8623-D), 1200 Pennsylvania Avenue, NW, Washington, DC 20460
‡
†
‡
(
Received 31 January 2001; Accepted 25 November 2001)
Abstract—The free-flowing Clinch and Powell watershed in Virginia, USA, harbors a high number of endemic mussel and fish
species but they are declining or going extinct at an alarming rate. To prioritize resource management strategies with respect to
these fauna, a geographical information system was developed and various statistical approaches were used to relate human land
uses with available fish, macroinvertebrate, and native mussel assemblage data. Both the Ephemeroptera, Plecoptera, Trichoptera
(
EPT) family-level index, and the fish index of biotic integrity (IBI) were lowest in a subwatershed with the greatest coal mining
activity (analysis of variance [ANOVA], 0.05). Limited analyses in two other subwatersheds suggested that urban and agricultural
land uses within a specified riparian corridor were more related to mussel species richness and fish IBI than land uses in entire
catchments. Based on land uses within a riparian corridor of 200 m 2 km for each biological site in the watershed, fish IBI was
inversely related to percent cropland and urban area and positively related to pasture area (stepwise multiple regression,
0.05). Sites less than 2 km downstream of urban areas, major highways, or coal mine activities had a significantly lower mean
IBI value than those more than 2 km away (ANOVA, .05). Land use effects included poorer instream cover and higher substrate
embeddedness ( test, 0.05). Weaker land use relationships were observed for EPT and mussel species richness. Episodic spills
p
Ͻ
ϫ
2
R
ϭ 0.55,
p
Ͻ
p
Ͻ
t
p Ͻ
of toxic materials, originating from transportation corridors, mines, and industrial facilities, also have resulted in local extirpations
of native species, particularly mussels. The number of co-occurring human activities was directly related to stream elevation in the
Clinch River, with more human land uses in headwater areas. Approximately 60% of known U.S. Fish and Wildlife mussel
concentration sites in the watershed are located within 2 km of at least two land use sources identified as potentially stressful in
our analyses. Our results indicate that a number of land uses and stressors are probably responsible for the decline in native species.
However, protection of naturally vegetated riparian corridors may help mitigate some of these effects.
Keywords—Watershed
Mining
Land use
Geographical information system
Endangered species
INTRODUCTION
federally threatened and endangered aquatic species (mostly
mussels and fish) than almost any other watershed in the United
States; the fauna are susceptible to multiple physical, chemical,
and biological stressors that could be present in this watershed;
and local resource managers and researchers were actively
interested in further analyzing the extensive body of infor-
mation that had been collected cooperatively in this watershed
over many years.
The upper Clinch and Powell rivers in Virginia represent
some of the last free-flowing sections of the expansive Ten-
nessee River system. This watershed supports more native
mussel and fish species than any other basin in Virginia and
most streams in North America [4,5]. However, it is well es-
tablished that most of the historic (pre-1910) mussel beds in
the watershed have declined dramatically or been eliminated
Protection of valued ecological resources requires knowl-
edge of the autecological factors affecting their abundance and
distribution as well as their sensitivity and resilience to po-
tential stressors. Interest has been increasing on the part of
resource managers and the public to evaluate protection of
resources in watersheds, or even larger spatial scales, in part
because effects of habitat fragmentation and other stressors
can be underestimated at smaller spatial scales [1,2]. Further-
more, use of a watershed approach often is necessary to address
effects of nonpoint sources because traditional regulatory ap-
proaches do not adequately address this issue [3].
Assessing relationships between human land use activities
and ecological resources is especially complex in a watershed
or larger spatial scales in which multiple land uses, and in-
teractions among those uses, are likely to be present. The
Clinch and Powell rivers watershed in southwestern Virginia,
USA, was selected by the U.S. Environmental Protection
Agency as one of five watersheds to demonstrate the use of
the ecological risk assessment paradigm in identifying major
causes of declines in valued ecological resources. This wa-
tershed was selected because it harbors a higher number of
(
Fig. 1) [6]. Similar extirpations have been documented for
several endemic fish species as well [7,8]. Analyses of historic
data for certain sites in the watershed indicate that declines in
the number of mussel species have been somewhat continuous
over the past 70 to 80 years, despite implementation of re-
covery plans and reforestation over the last century [8,9]. Re-
cent mussel sampling in Copper Creek, a subwatershed of the
Clinch–Powell river basin also has indicated continued, un-
explained declines in mussel fauna (S. Ahlstedt, U.S. Geolog-
ical Survey, unpublished report). The loss of native mussel
and fish species in the Clinch–Powell watershed and in North
America as a whole is extensive [5], indicating that the prob-
lem is widespread.
*
To whom correspondence may be addressed
jerry.diamond@tetratech.com).
Presented at the 20th Annual Meeting, Society of Environmental
Toxicology and Chemistry, Philadelphia, Pennsylvania, November
4–18, 1999.
(
1
1
147

1
148
Environ. Toxicol. Chem. 21, 2002
J.M. Diamond et al.
marily by coal mining and agriculture. More than 40% of
Virginia’s coal production lies within the five counties in the
basin, where the Cumberland Plateau is composed of Penn-
sylvanian sandstone and shale. Most of the watershed was
intensively logged for agricultural production in the late 18th
and early 19th centuries and the forest industry has largely
declined through the 19th and 20th centuries. Beef cattle and
burley tobacco are the primary agricultural products in the
watershed [11]. Topographic constraints limit most of these
agricultural activities to the floodplain, where livestock and
row crop production is most feasible and productive.
Data sources
Several biological and habitat measures were available with
which to examine relationships with human land uses. The
majority of these data were collected by the Tennessee Valley
Authority (TVA) as part of regular monitoring programs. The
Clinch–Powell River Action Team Survey (CPRATS) included
habitat quality measures (12 metrics, each scored categorically
from excellent to poor [1–4], as well as a composite habitat
score [13]), a fish index of biotic integrity (IBI as formulated
by Karr et al. [14]), and macroinvertebrate Ephemeroptera,
Plecoptera, Trichoptera (EPT) family-level index data col-
lected between 1995 and 1998. Component metric values com-
prising the fish IBI scores were unavailable for analysis. The
CPRATS is a fixed-station monitoring network with 155 sam-
pling sites located throughout the watershed on both low- and
high-order streams. Although land use and other landscape
information was available for all CPRATS sites, biological
and habitat data were not collected at all sites. A second source
of biological data was the Cumberlandian Mollusc Conser-
vation Program, which includes native mussel abundance and
species richness data collected at 60 sites between 1981 and
Fig. 1. Map of the Clinch–Powell watershed (VA, USA) showing the
comparison between historic (pre-1910) and present locations of na-
tive mussel concentrations. Bold river lines indicate observed native
mussel concentrations.
Although several hypotheses have been advanced to explain
the decline in native mussel and fish species in this and other
watersheds, definitive answers have been lacking. Resource
managers recognized that a comprehensive examination of the
available data for the watershed as a whole was needed to
evaluate the relative effects of different human activities on
native mussels and fish. This study attempted to define current
human land use factors that may be associated with native
species decline and to identify beneficial resource management
strategies that might further protect remaining productive ar-
eas.
1
986. The Cumberlandian Mollusc Conservation Program has
sites located mostly in larger mainstem areas.
Mussel reproduction depends on successful parasitization
of fish hosts by mussel larvae or glochidia [5] and fish species
distribution is related, at least in part, to aquatic insect abun-
dance and distribution. Thus, a relationship between fish, ma-
croinvertebrates, and mussel distribution is expected [5,8,10].
Although none of the IBI or EPT sites coincided with mussel
sites in our data sets, some fish and mussel sites were near to
one another, allowing for qualitative assessment of fish–mussel
relationships. This was advantageous because mussel data
were not as widely distributed as either fish IBI or EPT data.
Because of the temporal variation between IBI, EPT, and mus-
sel data, definitive relationships between these biological as-
semblages could not be determined. Therefore, data for these
fauna are presented separately, although it is likely that inter-
actions among them also are related to native species decline.
Other data used in our analyses included land cover (derived
from Landsat Thematic Mapper imagery, Earth Resources Ob-
servation Systems, Sioux Falls, SD, USA), elevation and slope
data obtained from a mosaic of 30-m resolution U.S. Geolog-
ical Survey digital elevation models, stream drainages (U.S.
Geological Survey Stream Reach File 3), road density, and
locations of point source dischargers and mines (U.S. Envi-
ronmental Protection Agency’s Permit Compliance System).
In addition, peer-reviewed biological studies obtained from
other institutions, as well as historic data (pre-1920), were used
to help interpret results.
MATERIALS AND METHODS
Study area
The Clinch and Powell rivers flow into the upper reaches
of the Tennessee River at Norris Lake in northeastern Ten-
nessee, USA (Fig. 1). This region was unglaciated in recent
geologic time and has been isolated from other nearby river
systems by the mountainous terrain, thus favoring the evolu-
tion of a diverse, endemic assemblage of fish and freshwater
mussel species in this watershed [9,10]. The upper regions of
the Clinch and Powell rivers drain approximately 7,542 and
2
2
,429 km , respectively [11]. A significant reduction in aquatic
native species diversity is associated with impoundments in
this watershed [6,8,12]. Therefore, this assessment treated Nor-
ris Lake in northeastern Tennessee (the furthest upstream im-
poundment on the Tennessee River system) as the terminal
boundary of the study.
The economy in the Clinch–Powell region is driven pri-

Relationships between human land uses and stream biota
Environ. Toxicol. Chem. 21, 2002
1149
Table 1. Comparison of land cover for four subwatersheds examined
in the Clinch–Powell watershed (VA, USA) risk assessment
Upper
Upper
Clinch Powell
Guest
River
Copper
Creek
River
River
Forest (%)
Cropland (%)
Pasture (%)
53.7
0.6
44.5
1.1
8
89.6
3.1
2.4
4.2
21
84.1
Ͻ0.1
10.4
2.6
26
57.7
1.3
40.9
Urban (%)
Ͻ
0.1
0
Number of coal minesa
a
Mines and coal preparation plants.
Analyses
Land use, habitat, and biological data were entered into a
geographical information system (Arc/INFO, Ver 7.04, En-
vironmental Systems Research Institute, Redlands, CA, USA)
and partitioned in various ways by ACCESS (Microsoft, Red-
mond, WA, USA) to obtain databases that were amenable to
statistical analyses (Statistica, Ver 5.0, Statsoft, Tulsa, OK,
USA). Both univariate and multivariate analyses were used to
identify relationships between land uses, habitat quality, and
biological measures.
We performed our initial analyses with fish IBI and ma-
croinvertebrate EPT data from four subwatersheds: Copper
Creek, the Guest River, upper Clinch River, and upper Powell
River (Fig. 1). The four subwatersheds comprise a range of
different human land uses, particularly with respect to coal
mining, pasture and crop area, and urban or developed area
Fig. 2. Macroinvertebrate Ephemeroptera, Plecoptera, and Trichoptera
EPT) and fish index of biotic integrity (IBI) scores by subwatershed
in the Clinch–Powell basin (VA, USA) based on Tennessee Valley
Authority’s Clinch–Powell River Action Team Survey data set. Boxes
are standard errors, whisker bars are standard deviations, and the
square is the mean value.
(
(
Table 1). More rigorous analyses were not feasible because
of unbalanced representation of biological data across sub-
watersheds and stream types.
Analysis (Environmental Systems Research Institute), a 200-
Before conducting land use analyses for the watershed as
a whole, we performed pilot analyses to identify the appro-
priate spatial scale with which to evaluate relationships be-
tween human land use and biological measures. Copper Creek
subwatershed was chosen for this analysis because it was the
most data-rich subwatershed and because agricultural uses
were the major source of anthropogenic activity (Table 1),
making the data easier to interpret. ArcView (Ver 3.0, Envi-
ronmental Systems Research Institute) was used to calculate
land uses in areas having several different stream riparian
widths (50-, 100-, 200-, and 400-m transects across the stream)
and different distances upstream (100, 200, 500, 1,000, and
m
ϫ 2-km area was created upstream of each CPRATS sam-
pling site and land use coverages were identified at each site.
Land use information included percentages (cropland, pasture,
forest, and urban), and proximity to mining, urban centers, or
major highways, defined categorically as either less than 1 km,
1
to 2 km, or more than 2 km upstream of the biological
sampling point. Land use and biological data for each site
were then aggregated into a single table and imported into
Statistica for analysis.
Land use percentage data were related to fish IBI, number
of mussel species, or EPT with both Pearson product moment
correlation (for individual land use relationships) and forward
2
,000 m) of each of nine CPRATS sampling points. Percent
stepwise multiple regression analyses (p Ͻ 0.05; for multiple
agricultural land cover (pasture plus cropland) was then cal-
culated for each different combination of riparian width and
distance upstream for each site. Relationships between percent
agricultural land and fish IBI, macroinvertebrate EPT, and hab-
itat quality indices for the different combinations were deter-
land use relationships). Fish IBI, EPT, and habitat scores were
also converted to categorical variables (categorized as either
impaired or unimpaired based on TVA’s scoring system) and
then analyzed with respect to land use percentages with a one-
tailed
t
test or one-way analysis of variance (ANOVA;
p
Ͻ
mined with spearman rank correlation (p ϭ 0.05). The IBI,
0
.05). Land use proximity data (e.g., distance to nearest urban
EPT, and habitat quality index were also analyzed in relation
to riparian percent agricultural land cover for the entire sub-
watershed upstream of each sampling point. After the full wa-
tershed analyses (see below), a similar riparian corridor anal-
ysis was conducted, with mussel data reported by Jones et al.
center) were related to biological data with ANOVA (
.05).
p
Ͻ
0
RESULTS
Subwatershed characterization
Both EPT and IBI were lowest in the Guest River subwa-
tershed (ANOVA, 0.05; Fig. 2), which has had intense coal
[
15] (15 sites) from the upper Clinch River, to identify potential
errors due to extrapolating fish IBI results to other fauna and
to other areas of the watershed.
p
Ͻ
Results of Copper Creek pilot analyses were used to define
the approach for investigating effects of human land uses for
each biological sampling location in the Clinch–Powell wa-
tershed in subsequent analyses. By using ArcView Spatial
mining activity (Table 1), acid mine drainage for many years
[16], and few other potential land use stressors such as urban
or pasture influences. Ephemeroptera, Plecoptera, Trichoptera
was also lower in the upper Powell than either Copper Creek

1
150
Environ. Toxicol. Chem. 21, 2002
J.M. Diamond et al.
Table 3. Correlation (Pearson product moment R values) between
mussel data and land use percentages in whole drainage and different
sized riparian corridors. Mussel data were obtained from Jones et al.
Table 2. Summary of Spearman rank correlation coefficients between
percent agricultural area and fish index of biotic integrity (IBI) as a
function of different combinations of riparian corridor width and
distance upstream. The IBI values were obtained from Tennessee
Valley Authority’s Clinch Powell River Action Team Survey.
[15]. Italicized values are significant at
p
Ͻ
0.10
Coefficients marked with italics were significant at
p
Ͻ
0.05
Mussel
Mussel
Riparian corridor
Land use
Forest
Urban
Agriculture
Forest
Urban
Agriculture
Forest
Urban
Agriculture
Forest
Urban
Agriculture
Forest
Urban
Agriculture
Forest
Urban
Agriculture
Forest
Urban
Agriculture
Forest
richness
abundance
0.11
Upstream riparian length (m)
Riparian
width (m) 100
100 m
100 m
100 m
ϫ
ϫ
ϫ
ϫ
ϫ
ϫ
ϫ
ϫ
1 km
2 km
5 km
10 km
1 km
2 km
5 km
10 km
0.62
Ϫ
200
500
1,000
2,000
0.42
0.33
0.68
0.49
0.39
0.77
0.40
0.59
0.65
0.08
0.15
0.62
0.43
0.39
0.69
0.46
0.43
0.62
0.28
0.47
0.62
0.01
0.19
0.12
0.35
0.20
Ϫ
Ϫ
Ϫ
Ϫ
0.05
0.07
0.19
0.06
0.25
0.07
0.13
0.18
0.07
0.14
0.22
0.11
0.04
0.08
0.16
0.01
0.21
0.16
0.05
0.19
0.05
0.08
0.21
0.20
0.037
0.20
Ϫ
5
0
0.09
0.19
0.14
0.09
0.14
0.24
0.18
0.14
0.10
0.26
0.30
0.18
0.06
0.09
0.34
Ϫ90.04
1
2
4
00
00
00
Ϫ
0.11
0.11
0.19
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
0.20
Ϫ
Ϫ
Ϫ
Ϫ
1
00 m
or the upper Clinch River subwatershed (ANOVA,
p
Ͻ 0.05;
Ϫ
Ϫ
Ϫ
Fig. 2). Index of biotic integrity was not significantly different
among these three subwatersheds ( 0.10). The upper Powell
subwatershed also has substantial coal mining activity (Table
), although it has slightly more urban area and cropland than
p
Ͼ
200 m
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
1
2
00 m
in the Guest River subwatershed (Table 1). These data indicate
that coal mining is the human land use most associated with
insect and fish distribution in the watershed as a whole.
Ϫ
Ϫ
200 m
Ϫ
Ϫ
Ϫ
Ϫ
Subwatershed pilot analyses
2
00 m
Fish IBI showed a weak negative relationship with total
Urban
Agriculture
Forest
Urban
Agriculture
Ϫ
Ϫ
Ϫ
Ϫ
catchment agricultural area (
EPT seemed to be unrelated to upstream agricultural area (
0.23, 0.20). The TVA’s habitat quality index also seemed
to be unrelated to catchment agricultural land use upstream (
ϭ Ϫ0.14, 0.21). Thus, given the constraints of this analysis
9), total catchment agricultural area seemed to have little
relationship with biological or habitat measures examined.
More significant trends between percent agricultural land and
fish IBI were evident in intermediate-sized riparian areas (200-
m buffer width and 500–1,000 m upstream) as opposed to
either larger (i.e., greater riparian width, greater distance up-
stream, or both) or smaller (i.e., narrower riparian widths,
shorter distances upstream, or both) riparian areas (Table 2).
Habitat quality and EPT did not exhibit significant relation-
ships with percent agricultural land in any of the riparian cor-
r
ϭ Ϫ0.30,
p
ϭ
0.08), whereas
Whole drainage
r
ϭ
p Ͼ
Ϫ
r
p
ϭ
(n ϭ
Mollusc Conservation Program mussel data collected six years
later in Copper Creek, we observed a similar spatial pattern
between fish and mussel species richness with river mile (Fig.
3
; Wilcoxon matched pairs test, p Ͻ 0.05). EPT exhibited
relatively little variability with stream location in Copper
Creek (total range of EPT values was 13–19) and appeared to
be unrelated to either fish IBI or mussel species richness (
r Ͻ
0
.12; p Ͼ 0.10).
Watershed analyses
A 200-m 20-km riparian corridor was applied to all bi-
ological sites in the watershed based on results of Copper
ridor dimensions examined (r Ͻ 0.10, p Ͼ 0.20).
ϫ
Similar analyses performed with mussel data in a small
portion of the upper Clinch River [15] indicated that the cor-
relation between riparian urban land use and mussel species
richness was highest in 2-km-long corridors as opposed to
longer or shorter corridors, similar to the Copper Creek results
with fish IBI (Table 3). Agricultural land exhibited significant
correlation with mussel species richness in longer (5-km) ri-
parian corridors (Table 3). Mussel richness was positively as-
sociated with percent forest in all corridor lengths examined
(
1–10 km) but not in the drainage as a whole (Table 3). Mussel
abundance was uncorrelated with land uses in any of the ri-
parian dimensions examined. These additional pilot analyses
suggested that mussel species richness is more sensitive to
human land uses than is abundance. Furthermore, 2-km-long
riparian corridors, as suggested by analysis of the Copper
Creek data, may underestimate agricultural effects on mussels
at some sites in the watershed.
Too few sites were available in which mussels, fish, and
EPT were simultaneously sampled in Copper Creek, and in
the watershed as a whole, for us to compare statistically re-
sponses of these fauna to the same land use patterns. By using
fish species data collected by Masnik [12] and Cumberlandian
Fig. 3. Fish and mussel species richness by river mile in Copper Creek
VA, USA). Fish data are based on U.S. Soil Conservation Service
data [11]; mussel data are based on Tennessee Valley Authority’s 1981
Cumberlandian Mussel Conservation Program data. ⅜ ϭ fish; □ ϭ
mussels.
(

Relationships between human land uses and stream biota
Environ. Toxicol. Chem. 21, 2002
1151
Table 4. Summary of stepwise multiple regression and analysis of variance (ANOVA) analyses of fish index of biotic integrity (IBI) values in
relation to human land uses in the Clinch–Powell watershed (VA, USA). Data were obtained from the Tennessee Valley Authority’s Clinch–
Powell River Action Team Survey data set
Partial
R
Mean IBI
2 km
Mean IBI
2 km
R2
Land use factors
Association
2
p
value
Ͻ
Ͼ
n
3
8
0.55
Percent pasture area
Percent cropland
Percent urban
Highway proximity
Mining proximity
ϩ
Ϫ
Ϫ
Ϫ
Ϫ
0.15
0.06
0.04
0.019
0.15
0.23
0.05
0.01
ANOVA
ANOVA
F
F
ϭ
3.01
34.2
31.7
39.5
41.6
ϭ
17.27
Creek analysis. Agricultural land was split into two categories
for these analyses (cropland and pasture land), because both
of these uses occur in the watershed as a whole. We observed
a significant relationship between elevation and either fish IBI
or EPT for the watershed as a whole (Pearson correlation co-
urban area showed the clearest association with EPT with cat-
egorical EPT data (mean riparian urban area 25.2 and 7.4%
for impaired and unimpaired EPT condition respectively,
tests, 0.05). The ANOVA indicated that proximity to
mining and major roads were also factors associated with EPT
(Table 5).
ϭ
t
p
Ͻ
efficient [
r
]
ϭ
0.54,
p Ͻ 0.01, and r ϭ Ϫ0.19, p ϭ 0.05,
respectively,
n
ϭ 137). Because many of the CPRATS sites
that had concurrent habitat and biological information were
located between 350 and 450 m elevation ( 52 out of 80
sites), and no correlation was found between elevation and
either IBI or EPT within this range ( 0.2, 0.10), sub-
sequent watershed analyses concentrated on sites within 350
and 450 m elevation. A broader elevation range resulted in
Mussels
n
ϭ
Stepwise multiple regression analysis indicated a significant
negative association between mussel species richness and ur-
ban and cropland land use (Table 6). However, like EPT, less
than a third of the variance in mussel species richness was
explained by riparian human land uses in our analysis. Given
the results of pilot analyses in the upper Clinch River noted
earlier, the relationship with agricultural uses possibly was
underestimated in this analysis.
r
Ͻ
p Ͼ
significant elevation effects in multiple regression analyses (
0.05). Approximately 5 and 30% of the sites were less than
50 m or more than 450 m in elevation, respectively.
p
Ͻ
3
Variation in the data set was insufficient to evaluate rela-
tionships between mussel assemblage characteristics and prox-
imity to mining, urban, or major highways. However, episodic
spills of a variety of contaminants from coal mines, industrial
facilities, and transportation accidents clearly have contributed
to declines in native mussels (and other fauna) in many lo-
cations in this watershed. For example, mussel species richness
data collected over a 93-year period at two different locations,
one downstream of coal mining in the Powell River [10] and
one downstream of a power generation facility in the upper
Clinch River [17], indicated sharp declines in native mussels
after spills of toxic materials (Fig. 4).
Index of biotic integrity
Fish IBI was inversely related to percent cropland and urban
land use and positively related to percent pasture land (forward
stepwise multiple regression,
km downstream of mining or major transportation routes
had a significantly higher mean IBI value than sites less than
km downstream (ANOVA; Table 4). Mean riparian urban
n ϭ 52; Table 4). Sites more than
2
2
area was significantly higher at sites in which fish IBI was
categorized as impaired (based on TVA’s IBI scoring; urban
area
respectively;
ϭ
5.6 and 12.2% for unimpaired and impaired IBI sites,
test, 52, 0.05).
t
n
ϭ
p Ͻ
Effects of urban and cropland uses on IBI were due, at least
in part, to poorer instream cover and higher substrate embed-
dedness, as evidenced by analyses of categorical IBI data
Cumulative effects of land uses
(
mean instream cover values
embeddedness values 2.7 and 3.2 for impaired and unim-
paired IBI condition, respectively, tests, 0.05, 52).
ϭ
2.9 and 3.9 and mean stream
In an effort to address cumulative effects of human activ-
ities, riparian land use sources determined to be significant in
previous analyses and in other research [18] (mining, urban
areas, major transportation corridors, and percentage of crop-
land area) were calculated and summed for each biological
sampling site. Land use sources were expressed as a binary
function: 0 if the particular source was not present or nearby
(within 2 km upstream) and a 1 if the source was present. A
cumulative stressor index was then computed for each site
ϭ
t
p
Ͻ
n ϭ
Ephemeroptera, Plecoptera, and Trichoptera
Forward stepwise multiple regression analyses indicated
that EPT was negatively related to percent urban area and
positively related to percent pasture, but the overall
much lower than that for IBI (
2
R
was
2
R ϭ 0.29; Table 5). Percent
Table 5. Summary of stepwise multiple regression and analysis of variance (ANOVA) analyses of macroinvertebrate Ephemeroptera, Plecoptera,
Trichoptera (EPT) values in relation to human land uses in the Clinch–Powell watershed (VA, USA). The EPT values were obtained from the
Tennessee Valley Authority’s Clinch–Powell River Action Team Survey
Partial
R
p
Mean EPT
Mean EPT
2 km
R2
Land use factors
Association
2
value
Ͻ
2 km
Ͼ
n
3
4
0.29
Percent urban area
Percent pasture
Mining proximity
Highway proximity
Ϫ
ϩ
Ϫ
Ϫ
0.14
0.08
0.033
0.127
0.02
ANOVA
ANOVA
F
F
ϭ
5.01
3.70
9.8
10.4
12.5
12.3
ϭ
0.03

1
152
Environ. Toxicol. Chem. 21, 2002
J.M. Diamond et al.
Table 6. Summary of stepwise multiple regression analysis of native
mussel species richness as a function of riparian human land use
factors. Data were obtained from the Tennessee (USA) Valley
Authority’s Cumberlandian Mussel Conservation Program database
Table 7. Fish index of biotic integrity (IBI) scores in relation to the
number of co-occurring land use sources present within 2 km upstream
of the sampling point. Means with the same letter in common are not
significantly different (Duncan multiple range test,
I
Ͻ
0.05)
Partial
R
Number of
co-occurring
land uses
n
R2
Land use factors
Association
2
p
value
Mean IBI
Minimum
Maximum
SDa
3
3
0.26
Percent urban area
Percent cropland
Ϫ
Ϫ
0.09 0.007
0.02 0.30
0
1
2
3
4
48.8 A
42
26
20
20
18
58
58
56
48
28
5.8
10.3
10.1
9.1
41.5 A, B
35.6 B, C
31.3 C
ranging between 0 and 4 (representing the four human land
uses above).
21.0 D
4.8
a
SD
ϭ standard deviation.
Fish IBI decreased as more riparian land use sources co-
occurred (ANOVA,
p Ͻ 0.01; Table 7). Fish IBI was consis-
tently low (rated as poor based on TVA ecoregional reference
site data) at sites having all four stressors present. Approxi-
mately 66% of the sites having two of the four stressors present
had IBI scores less than 35, indicating impaired fish com-
munity integrity at those sites according to TVA. In nearly all
of these cases (88%), the sources present were urban areas and
mining. Cropland and transportation corridors were potential
sources at 12% of the sites having impaired fish community
integrity.
We could not statistically analyze relationships between
either mean number of mussel species or mean mussel abun-
dance and the number of human land uses present because of
low sample size in certain land use categories. However, we
observed lower mussel abundance and fewer mussel species
at sites having two or more of the four human land uses present
as compared to sites with one or no human land use within
the riparian corridor (Fig. 5). The maximum number of native
mussel species observed in this data set (18) is comparable to
some of the better mussel sites in the entire watershed, al-
though still far less than the number of species historically
reported (Ͼ35 species at many of these sites [9]).
In an attempt to examine vulnerability of native mussels
to multiple land use sources of potential stress, we used the
geographical information system (ArcInfo) to map significant
mussel concentration sites, identified by the U.S. Fish and
Wildlife Service, in relation to mines, major roadways, urban
centers, and riparian agricultural areas. Mussel concentration
sites typically harbor federal- as well as state-listed species of
concern [10,19] and they are located in several locations within
the Clinch River and some of its tributaries. No mussel con-
centration sites currently are known in the upper Powell and
Guest rivers, consistent with the more intensive coal mining
in that area. Of the 18 mussel concentration sites examined,
none are reasonably isolated from major roads, mines, urban
areas, or agricultural uses (Table 8). Ten of the 18 sites (55%)
are within 2 km of a major transportation corridor, which, in
some cases, have been sources of episodic toxic spills, as
mentioned previously. Eight of the sites (44%) are vulnerable
to two or more human land uses shown to be negatively as-
sociated with either fish IBI, EPT, mussel species richness, or
a combination of these (Table 8).
More human sources of potential stress tended to co-occur
Fig. 4. Number of native mussels and number of mussel species col-
lected as a function of the number of human land uses co-occurring.
Data are taken from Tennessee Valley Authority’s Cumberlandian
Fig. 5. Number of mussel species recorded over time at two sites in
the Clinch–Powell watershed (VA, USA), each affected by large toxic
spill events.
Mollusc Conservation Program, 1986 (n ϭ 48).

Relationships between human land uses and stream biota
Environ. Toxicol. Chem. 21, 2002
1153
Table 8. Riparian human land uses within 2 km of known mussel
concentration sties identified by the U.S. Fish and Wildlife Service
(
(
corresponding to elevation in our data set) or catchment slope
which may or may not be related to elevation), for example,
in the Clinch–Powell watershed (VA, USA). The plus symbol (
denotes that source is present
ϩ)
are expected to influence biota distribution and abundance
19], independent of land uses. In the present study, we ob-
[
Major
served more significant relationships between land uses and
fish IBI when using sites from a limited elevation range. Per-
haps if other natural watershed factors can be controlled in
analyses, relationships with land uses may be more readily
observed. However, a difficulty with limiting our analyses to
sites within a prescribed elevation range is that sample size
decreased accordingly. As a result, the variability in land use
percentages or combinations of land uses may not have been
broad enough to allow large-scale relationships to be distin-
guished accurately. For example, we observed that percentage
of pastureland was negatively associated with fish IBI in the
Copper Creek subwatershed analysis (in which pasture was
the major land use other than forest). However, in the water-
shed analysis, pasture was positively associated with fish IBI
Site
Mines
Urban
Agriculture transportation
1
2
3
4
5
6
7
8
9
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
1
1
1
1
1
1
1
1
1
0
1
2
3
4
5
6
7
8
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
ϩ
(
and EPT as well). These disparate results may have been due
to truly positive effects of pasture on fish and macroinverte-
brates (e.g., nutrient enrichment) as compared to effects of
other land uses such as mining. Alternatively, these results
may have been due to too few sites with certain land use
combinations [20].
Data quantity and variability also may have been factors
affecting our ability to distinguish land use relationship with
mussels. Sample size was limited more for mussels as com-
pared to either fish IBI or EPT, and most sampling locations
were on the mainstem rivers where local riparian land uses
may have less influence on stream fauna as compared to small-
er streams [21]. However, other researchers in this watershed
also have reported limited associations between mussels and
land uses [10,22]. Possible reasons include episodic toxic spills
that did not originate from sources in the immediate vicinity
in the upper Clinch River as compared to the upper Powell
River (means test, and 0.05, 33, respectively).
This was due to the greater frequency of urban and industrial
activity and major highways in the upper Clinch River. Fur-
thermore, the cumulative source index increased as one pro-
r ϭ 0.74, p Ͻ 0.05, n
14) because of the addition of more mining and urban in-
fluences in the upstream part of the river. A direct relationship
between number of land use sources and river mile was not
ϭ
t
p
Ͻ
n ϭ
gressed upstream in the Clinch River (
ϭ
observed in the upper Powell River (
0
n ϭ 19, r ϭ 0.03, p Ͼ
.10), probably because of the concentration of mining activity
of the site (i.e.,
Ͼ2 km upstream) and are too infrequent to
in much of that subwatershed. Given that most mussel con-
centration sites presently are located in the Clinch River drain-
age, the above information suggests that native mussel pop-
ulations are relatively vulnerable in this watershed and that
further extinctions or extirpations are likely to occur unless
resource protection measures are taken.
yield consistent land use associations; fish host abundance in
the area varies from year to year; or site-specific habitat char-
acteristics such as orientation of bedrock strata or stream slope
(
shown to be a possible factor affecting mussel communities
in the upper Clinch River) are more important in defining
abundance and distribution of native mussel species [23].
Available information in this watershed, and results of this
study, suggest that sedimentation and other forms of habitat
degradation from urban uses, mining, and agricultural areas
are likely to be limiting aquatic fauna in this watershed. For
example, sedimentation due to accumulation of coal fines (fine
particulate coal and refuse rock material) has been reported in
many areas downstream of active coal mines and coal slurry
ponds in the upper Powell River drainage [22,24]. Sedimen-
tation and substrate embeddedness are major factors affecting
mussel and fish distribution and abundance in lotic systems
[24]. In most of these cases, little evidence has been found of
mussel recolonization or recruitment to the affected areas de-
spite improved water quality and documented recovery of fish
populations [22,25].
Point and nonpoint source contaminants from coal mine
activities, agricultural uses, and urban areas are also likely to
be limiting aquatic fauna distribution. Insufficient water qual-
ity data were available for analyses in this study. However,
previous site-specific studies in this watershed have docu-
mented water quality problems in certain locations in the wa-
tershed [26]. Deep coal mines, the dominant form of mining
in the watershed, also have discharged toxic concentrations of
DISCUSSION
Results of our analyses suggest that riparian human land
use activities are important factors affecting native mussel,
aquatic insect, and fish distribution in the Clinch–Powell wa-
tershed. Similar findings have been reported in many other
watersheds [18,19]. However, our results indicate that mod-
erate differences in the aerial extent of the riparian corridor
can have large effects on the strength of observed land use–
biota associations. The two subwatershed pilot analyses per-
formed in this study indicated that agricultural land use as-
sociations with mussel species richness may have been un-
derestimated when using a 2-km-long riparian corridor for all
sites. Depending on the range of stream sizes, types of land
uses, and perhaps other catchment characteristics observed
across biological sampling sites, it may be necessary to cali-
brate appropriate riparian corridor dimensions to determine
whether particular land use combinations are significantly af-
fecting biota or their habitat.
2
The fairly low
R
values observed in multiple regression
analyses in this study demonstrate the difficulties in relating
land use activities to biological or habitat measures on large
spatial scales. Natural ecological features such as drainage area

1
154
Environ. Toxicol. Chem. 21, 2002
J.M. Diamond et al.
group, who provided extensive information and reviews of earlier
reports. Dennis Yankee and Jeff White provided graphical information
system support and database management.
hydraulic emulsion oils originating from underground coal
mining equipment [27]. Non–point-source inputs of agricul-
tural pesticides, particularly in the more fertile bottomlands
and valleys, also are a potential source of toxic stress on native
fish and mussels in this watershed.
Although others have observed that riparian vegetation can
reduce deleterious land use effects on water quality if large
enough [28,29], it is not clear that improvement of the riparian
corridor alone in this watershed will result in recovery of native
mussel and fish populations. Several researchers have reported
significant effects of upland land uses on surface water quality
depending on catchment topography and the spatial pattern of
land uses in the watershed [18,30]. For example, a 1999 mussel
survey of Copper Creek indicated a loss of mussel species as
compared to a similar survey performed in the 1980s, appar-
ently because of more pervasive sedimentation in the stream,
even though riparian land uses have not changed appreciably
REFERENCES
1
. Robinson G, Holt R, Gaines M, Hamburg S, Johnson M, Fitch
H, Martinko E. 1992. Diverse and contrasting effects of habitat
fragmentation. Science 257:524–526.
2
3
. Suter GW II. 1993. Ecological Risk Assessment. Lewis, Boca
Raton, FL, USA.
. U.S. Environmental Protection Agency. 1996. Watershed ap-
proach framework. EPA- 840-S-96-001. Office of Water, Wash-
ington, DC.
4
5
. Jenkins RE, Burkhead NM. 1994. The Freshwater Fishes of Vir-
ginia. American Fisheries Society, Bethesda, MD.
. Neves RJ. 1991. Mollusks. In Terwilliger K, ed, Virginia’s En-
dangered Species. McDonald and Woodward Publishing, Blacks-
burg, VA, USA, pp 251–320.
6
. Angermeier PL, Smogor RA. 1995. Estimating number of species
and relative abundances in stream-fish communities: Effects of
sampling effort and discontinuous spatial distribution. Can J Fish
Aquat Sci 52:936–949.
(
S. Ahlstedt, U.S. Geological Survey, unpublished data). If
agricultural or livestock use upstream is great enough, sedi-
mentation effects and subsequent loss of habitat may result
for some distance downstream even though forested riparian
areas may be present directly upstream of a given site.
In the Clinch–Powell watershed, several reports have been
made of little or no recovery of threatened or endangered
mussel or fish species [10,31]. Mussel introduction efforts have
had very limited success in most cases [32; R. Neves, U.S.
Fish and Wildlife Service, personal communication] and would
not be expected to improve mussel recruitment unless no hab-
itat limitations exist and sufficient host fish are present in the
area [33]. For some threatened and endangered mussel species,
glochidia may be able to parasitize only a few fish host species
7. Feeman JC Jr. 1987. Results of fish surveys in the Tennessee
River drainage, 1979–1981. Brimleyana 13:99–121.
8
9
. Ahlstedt SA. 1984. Twentieth century changes in the freshwater
mussel fauna of the Clinch River (Tennessee and Virginia). Walk-
erana 5:73–122.
. Ortmann AE. 1918. The nayades (freshwater mussels) of the up-
per Tennessee drainage with notes on synonomy and distribution.
Proc Am Philos Soc 57:521–626.
1
0. Ahlstedt SA. 1991. Cumberlandian Mollusk Conservation Pro-
gram: Mussel surveys in six Tennessee Valley streams. Walkerana
5
:123–160.
11. U.S. Soil Conservation Service. 1992. Clinch River basin land
and water resources study for hydrological units. Abington, VA.
1
2. Masnik M. 1974. Composition, longitudinal distribution and zoo-
geography of the fish fauna of the upper Clinch system in Ten-
nessee and Virginia. PhD thesis. Virginia Polytechnic Institute
State University, Blacksburg, VA, USA.
[
5,33]. If fish host species are reduced in number or avoid an
area because of certain stressors, mussels dependent on those
species will also be affected.
13. Barbour MT, Gerritsen J, Snyder BD, Stribling JB. 1999. Rapid
bioassessment protocols for use in streams and wadeable rivers:
Periphyton, benthic macroinvertebrates and fish, 2nd ed. EPA/
We observed a greater number of co-occurring potential
land use stressors in the upper parts of this watershed, which
is particularly significant in terms of resource conservation
efforts. This pattern, combined with naturally fewer mussel
and fish species at higher elevations and decreased drainage
area, suggests that mussels and other aquatic life are especially
vulnerable in upstream reaches of the Powell and Clinch rivers
and their tributaries. The vulnerability of fauna in headwater
areas of this watershed has been evidenced by the fact that
many current threatened and endangered mussel species were
historically present in fair numbers in small tributaries and
headwater areas of the Clinch–Powell watershed (Fig. 1) [8,9].
With any endemic population, a high risk exists of extir-
pation due to habitat fragmentation resulting in populations
that are too inbred, small in size, and more susceptible to
stressors [1]. Native mussels and fish in the Clinch–Powell
watershed are no exception. Populations are now more widely
separated than they were historically [8], which could lead to
reduced recruitment success and declining populations. For
this reason, it may be a higher priority to further protect those
populations that appear most vulnerable due to proximity to
mining, urban and agricultural areas, or transportation corri-
dors. Protection or enhancement of the riparian corridor at
these sites, as well as protection from toxic spills and sedi-
mentation, may help to sustain endemic species in the water-
shed.
8
41-B-99-002. U.S. Environmental Protection Agency, Office of
Water, Washington, DC.
14. Karr JR, Fausch KD, Angermeier PL, Yant PR, Schlosser IJ. 1986.
Assessing biological integrity in running waters: A method and
its rationale. Ill Nat Hist Surv Spec Publ 5.
5. Jones J, Patterson M, Good C, DiVittorio A, Neves R. 2000.
Survey to evaluate the status of freshwater mussel populations
in the upper Clinch River, VA. Final Report 05-2000. U.S. Fish
and Wildlife Service, Abington, VA.
6. Cherry DS, Rutherford LG, Dobbs MG, Zipper C, Cairns J. 1995.
Acidic pH and heavy metal impacts into stream watershed and
river ecosystems by abandoned mined lands, Powell River, Vir-
ginia. Bulletin 118. Virginia Water Resources Research Center,
Virginia Polytechnic and State University, Blacksburg, VA, USA.
17. Crossman JS, Cairns J Jr, Kaesler RL. 1973. Aquatic invertebrate
recovery in the Clinch River following hazardous spills and
floods. Bulletin 63. Virginia Water Resources Research Center,
Virginia Polytechnic Institute and State University, Blacksburg,
VA, USA.
1
1
18. Cooper JR, Gilliam JW, Daniels JB, Robarge WP. 1987. Riparian
areas as filters for agricultural sediment. Soil Sci Soc Am J 51:
4
16–420.
1
9. Osborne LL, Kovacic DA. 1993. Riparian vegetated buffer strips
in water-quality restoration and stream management. Freshwater
Biol 29:243–258.
20. Boward D, Kazyak P, Stranko S, Hurd M, Prochaska A. 1999.
From the mountains to the sea: The state of Maryland’s freshwater
streams. EPA/903/R-99/023. U.S. Environmental Protection
Agency, Office of Research and Development, Washington, DC.
2
1. Vannote RL, Minshall GW, Cummins KW, Sedell JR, Cushing
CE. 1980. The river continuum concept. Can J Fish Aquat Sci
3
7:130–137.
2
2. Wolcott L, Neves R. 1994. Survey of the freshwater mussel fauna
Acknowledgement—We wish to thank Don Gowan, Roberta Hylton,
and other members of the Clinch Watershed Ecological Risk Work-
in the Powell River, Virginia. Banisteria 3:3–14.
23. Church GW. 1991. Survey of the family Unionidae in the upper

Relationships between human land uses and stream biota
Environ. Toxicol. Chem. 21, 2002
1155
Clinch River and Little River, Virginia. PhD thesis. Biology De-
partment, Virginia Polytechnic Institute and State University,
Blacksburg, VA, USA.
28. Lowrance R, Todd R, Fail J Jr, Hendrickson O Jr, Leonard R,
Asmussen L. 1984. Riparian forests as nutrient filters in agri-
cultural watersheds. BioScience 34:374–377.
29. Richards C, Host G. 1994. Examining land use influences on
stream habitats and macroinvertebrates: A GIS approach. Water
Resour Bull 30:729–738.
2
2
4. Dennis SD. 1985. Distributional analysis of the freshwater mussel
fauna of the Tennessee River system with special reference to
possible limiting effects of siltation. Report 85–2. Tennessee
Wildlife Resources Agency, Nashville, TN, USA.
3
3
3
0. Omernik JM, Abernathy AR, Male LM. 1981. Stream nutrient
levels and proximity of agricultural and forest land to streams:
Some relationships. J Soil Water Conserv 36:227–231.
1. Neves RJ, Pardue GB, Benfield EF, Dennis SD. 1980. An eval-
uation of endangered mollusks in Virginia. Project E-F-1. Virginia
Commission of Game and Inland Fisheries, Roanoake, VA, USA.
2. Sheehan RJ, Neves RJ, Kitchel HE. 1989. Fate of freshwater
mussels transplanted to formerly polluted reaches of the Clinch
and North Fork Holston rivers, Virginia. J Freshwater Ecol 5:
5. Zipper CE, Holtzman GI, Rheem S, Evanylo GK. 1991. Under-
standing surface water quality trends in southwestern Virginia.
Report 91-1. Virginia Polytechnic Institute and State University,
Water Resources Research Center, Blacksburg, VA, USA.
6. Goudreau SE, Neves RJ, Sheehan RJ. 1993. Effects of wastewater
treatment plant effluents on freshwater molluscs in the upper
Clinch River, Virginia, USA. Hydrobiologia 252:211–230.
7. Biological Monitoring. 1990. Toxicity effects of a deep mine
water sump discharge containing a hydraulic oil at Westmoreland
Coal Company, Big Stone Gap, VA. BMI-90-16. Blacksburg, VA,
USA.
2
2
1
39–149.
33. Neves R, Weaver L, Zale A. 1985. An evaluation of host fish
suitability for glochidia of Villosa vanuxemi and V. nebulosa
(Peleypoda: Unionidae). Am Midl Nat 113:13–19.