United States
Environmental Protection
hI	Agency
Association among
invertebrates and habitat
indicators for large rivers
in the Midwest
How sampling distance, point-sampling
of habitat, and subsample size effect
measures of large river
macroinvertebrate assemblages
RESEARCH AND DEVELOPMENT

-------
EPA 600/R-04/177
October 2004
www.epa.gov
Association among
invertebrates and habitat
indicators for large rivers
in the Midwest
How sampling distance, point-sampling
of habitat, and subsample size effect
measures of large river
macroinvertebrate assemblages
Joseph E. Flotemersch*, Karen Blocksom,
John J. Hutchens, Jr. 1, Bradley C. Autrey
National Exposure Research Laboratory, U.S. Environmental Protection
Agency, 26 W. Martin Luther King Dr., Cincinnati, OH 45268 USA, -iCurrent
address: Department of Biology, Coastal Carolina University,
Conway, SC 29528 USA
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
Notice: Although this work was reviewed by EPA and approved for publication, it may not necessarily reflect official
Agency policy. Mention of trade names and commercial products does not constitute endorsement or
recommendation for use.

-------
Executive Summary
Six reaches in each of two large rivers (one each in Kentucky and Ohio)
were sampled using a prototype benthic macroinvertebrate sampling technique.
The intent was to better understand the relationship between large river
macroinvertebrate assemblages and habitat features. This information was to
determine an acceptable sampling design to support development of a large river
bioassessment protocol (LR-BP). Specific objectives included determining the
appropriate number of habitat point-samples to be collected, examining how
varying reach length affects assemblage characteristics, and determining an
appropriate laboratory subsample size to accompany the resulting field sampling
method.
At each site, both banks of 12 transects separated by increasingly larger
distances were sampled. Analyses were conducted using Monte Carlo methods.
Interpretation of results relied on the metric values of total taxa richness, mayfly
taxa richness, caddisfly taxa richness, Diptera richness, % mayflies, %
caddisflies, % Tanytarsini, % non-Tanytarsini dipterans and non-insects, and %
tolerant individuals.
This research indicates that, using the sampling technique discussed
herein, a representative sample of the benthic macroinvertebrate fauna in the
study reaches was collected by sampling both banks of 6 transects spaced at
100 m intervals over a 500-m distance. It is hypothesized that these results were
achieved because the sampling method and design effectively sampled the
benthic macroinvertebrate fauna of the dominant habitat types within the reach.
ii

-------
It is recommended that the field method be coupled with a fixed laboratory
subsample size of 300 organisms for bioassessment purposes, with the
recognition that a subsample size of 500 organisms may be needed to meet the
objectives of some studies. This recommendation is based on the response of
the tested metrics, and the observation that the ability to separate sites of
different macroinvertebrate composition generally did not increase with larger
subsample sizes.
It is likely this approach will over-sample sites of uniform composition, but
the goal was to develop a standardized LR-BP that would perform well across
sites of differing habitat composition. It should be noted that the LR-BP for
macroinvertebrates has only been tested in main-channel habitats. It may work
equally well on off-channel habitats, but this remains to be tested.
While the method has been designed to perform well in a variety of
habitats, resulting data should be interpreted with appreciation for coarse habitat
characteristics. This information would ideally be derived from habitat data
collected concomitantly with the faunal data. Sites can then be categorized into
river types (e.g., impounded vs. free-flowing or lowland vs. upland rivers) or even
habitat types within a specific river (e.g., sandy- vs. cobble-bottom) in a more
controlled environment (i.e., in the laboratory), thus increasing the overall
integrity of any and all site assessments. Possible modifications to the method to
streamline its future application in the field are provided.
111

-------
Key Words: benthic macroinvertebrates, bioassessment, sampling
method, LR-BP, non-wadeable streams.
iv

-------
Introduction
Wadeable streams and smaller rivers are abundant and relatively easy to
sample compared to large rivers. As a result, efforts to develop appropriate
sampling protocols for the bioassessment of lotic ecosystems have been focused
largely on smaller systems (e.g., Barbour et al. 1999). As these methods
become increasingly refined and accepted, a growing number of government
agencies are starting to better understand (Humphries et al. 1998) and develop
sampling protocols for non-wadeable large rivers.
Large rivers differ from wadeable systems in some important ways. In
rivers, stressor sources are generally more numerous (Sweeting 1994) and
almost certainly more rapidly diffused as discharge rates are generally higher in
higher-order streams (Allen 2000). Consequently, individual stressor effects are
masked by the presence of other stressors and their impacts less conspicuous.
Additionally, biological communities change with stream size, as do habitat type
and quality (Vannote et al. 1980). Assemblages adapted to deeper, wider
streams with limited canopy cover are more likely to occur in downstream higher
order reaches. Thus, expectations for communities in large rivers may be very
different from those in smaller systems. In addition, the thalweg of a large river
often may not be accessible for sampling as it is in wadeable streams, precluding
the use of some wadeable stream sampling protocols. Hence, resource
managers need clear and consistent protocols available for measuring ecological
integrity that are designed specifically for large river systems (Loucks 2003).
1

-------
To effectively support a bioassessment program, protocols for sampling
fauna of large rivers should be clear, consistent, and reproducible. In order to be
applicable to a wide audience, they should perform well across numerous non-
wadeable habitats and river types, represent site conditions accurately, and
ideally, identify the presence of stressors. Protocols should also be cost
effective, logistically feasible, and meet or be adaptable to multi-purpose
sampling needs of researchers and managers (e.g., trend analysis, point-source
and non-point source programs) if they are to be accepted by regulatory
organizations.
Benthic macroinvertebrates are one of the most common faunal
assemblages used in the bioassessment of aquatic ecosystems (Rosenberg and
Resh 1993, Metcalfe-Smith 1994, Barbour et al. 1999, Karrand Chu 1999).
Many macroinvertebrate collection methods currently used in non-wadeable
systems are derived from wadeable methods (Ohio EPA 1989, Barbour et al.
1999, Klemm et al. 2000, Flotemersch et al. 2001, Moulton et al. 2002,). These
methods often involve wading in shallow areas (e.g., near the shore) of larger
rivers or sampling from a boat in deep areas without additional modification. The
exception is the use of artificial substrates, which were developed largely for non-
wadeable invertebrate sampling applications (Cairns 1982).
Blocksom and Flotemersch (in press) compared six sampling techniques
used by three government agencies to sample benthic macroinvertebrate
assemblages of large rivers. They found that these methods resulted in different
metric values. Additionally, metric response (i.e., positive vs. negative) to certain
2

-------
stressors varied among sampling methods, and metrics detecting a specific
stressor were not consistent across methods (Blocksom and Flotemersch, in
press). Differences among the methods and the relatively poor performance of
some methods were hypothesized to be due partly to the inadequacy of using a
single sampling technique (e.g., kick-net, dip-net, artificial substrates) when
sampling large rivers. For example, a method that produced a representative
sample in a large river with abundant epifaunal substrate and low embeddedness
might not reflect a highly embedded reach well. The research of Bartsch et al.
(1998) and Poulton et al. (2003) corroborate this hypothesis, with both
concluding that an approach employing multiple sampling techniques was
needed to effectively sample all components of a macroinvertebrate assemblage
in riverine ecosystems.
To support the development of a more consistent Large River
Bioassessment Protocol (LR-BP) for benthic macroinvertebrates, three
fundamental issues must be addressed. First, a collection technique is needed
that secures a representative sample of benthic macroinvertebrates across the
broad range of habitats that occur within and across rivers. Second, an
appropriate sampling design will be needed for application of the developed
sampling technique. And third, an appropriate laboratory method (e.g.,
laboratory subsample size) must be determined. Through this entire process, it
is important to keep in mind the logistical realities faced by regulatory agencies.
To address this issue of collection technique, data from the different
sampling techniques compared by Blocksom and Flotemersch (in press) were
3

-------
analyzed to identify those techniques that when integrated, produced a
potentially more representative sample across a broad range of large river
habitat types. Consideration was given to how the critical elements of each
method could be combined and applied in a standardized manner to support a
Large River Bioassessment Protocol (LR-BP). The result was a sampling
technique that hypothetical^ should overcome the limitations of previous
approaches and permit standardized sampling across all non-wadeable habitats
and river types of varying impoundment status. The approach consisted of
features of the Environmental Monitoring and Assessment Program - Surface
Waters kick net sampling method (Klemm et al. 2000), and the multiple habitat
dip net methods of the Ohio Environmental Protection Agency (Ohio EPA 1989)
and the United States Geological Survey (Moulton et al. 2002) that sample all
available habitats.
Critical elements of the development of a scientifically sound sampling
design include the spatial scale over which sub-samples should be collected (i.e.,
reach length), the number of sub-samples needed, and the manner in which sub-
samples should be distributed within the sample reach. The use of "reach" in this
study follows that of Frissell et al. (1986) who defined it as a length of stream
between breaks in channel slope, local side-slopes, valley floor width, riparian
vegetation, and bank material.
For bioassessment purposes, determination of appropriate sample reach
lengths are typically linked to measures of geomorphology (e.g., channel widths,
meander wavelengths, riffle pool sequences) (Barbour et al. 1999, Herlihy and
4

-------
Lazorchak 2000, and Moulton et al. 2002) or evaluation of species accumulation
curves. Several studies have focused on appropriate reach lengths for
macroinvertebrates in wadeable streams (e.g., Li et al. 2001), and for fish in both
wadeable and non-wadeable streams (e.g., Lyons 1992, Hughes et al. 2002).
However, an appropriate assessment reach for macroinvertebrates in non-
wadeable streams has not been estimated. One difficulty is that benthic
macroinvertebrates are usually sampled at specific points, whereas sampling for
fish (e.g., fish by means of electrofishing techniques) is continuous over the
whole reach. Hence, the approach of determining an appropriate sampling reach
length for macroinvertebrates using species accumulation curves as a direct
function of distance is logistically impractical (due to the large number of
contiguous samples that would be required).
Similar challenges are encountered using measures of geomorphology for
reach determination on large rivers. The majority of streams in the U.S. have
been anthropogenically altered (especially through dam construction) to the
extent that <2% are of a quality worthy of federal protection status (Benke 1990).
This reality limits the utility of geomorphology (e.g., riffle-pool sequences,
multiples of the natural channel width) as a determinant of reach length because
it would only apply to a small subset of sites. Even if either of these traditional
approaches did work, the question still remains as to the appropriate number and
distribution of sub-samples within the designated reach to effectively represent
the reach for bioassessment purposes.
5

-------
As for the development of an appropriate laboratory method, the
procedures for sample processing are typically assumed to be readily
transferable from wadeable streams to large rivers, but this has not been
evaluated. Hence, this study also investigated the efficacy of sample processing
in the laboratory. The methods used for laboratory processing of invertebrates
can greatly influence sample results and ultimately determine the value of a
method for bioassessment purposes. A full count of all invertebrates may
provide a more accurate assessment (Doberstein et al. 2000), but is usually not
feasible when large numbers of organisms are collected (Barbour and Gerritsen
1996). As a result, samples are often subsampled in the laboratory, typically
using either a fixed-organism count or a fixed sample proportion (Barbour et al.
1999, Carter and Resh 2001).
The primary objectives of the study were to: 1) determine the appropriate
number of sampling points needed using a new LR-BP for macroinvertebrates in
nonwadeable rivers, 2) determine an appropriate laboratory subsample size to
accompany this sampling method, and 3) examine how varying reach lengths
affect assemblage characteristics. Examination of these features relied on
evaluation of the quantitative metrics of the Ohio EPA Invertebrate Community
Index (ICI).
Methods
Study Sites
6

-------
We collected data during late July through August 2001 from the Kentucky
(n = 6 sites) and Great Miami (n = 6 sites) rivers, both of which are major
tributaries of the Ohio River in the east-central United States (Figure 1). The 12
7

-------
Figure 1. Sample sites on the Great Miami and Kentucky rivers.
8

-------
sites were selected from 30 sites over 2 midwestern large rivers that were
sampled in a previous study to compare existing large river sampling methods
(Blocksom and Flotemersch, in press)(Table 1). Sites were selected to partition
the sampling effort evenly between impounded and relatively free-flowing sites,
and across a gradient of habitat conditions within each river. Gradients were
based on existing instream and riparian physical habitat data collected using
EMAP protocols (Kaufmann 2000), land use data, and best professional
judgment.
The Great Miami River flows through several urban and industrial
corridors in Ohio (e.g., Dayton, Springfield, Hamilton, and Middletown) before
reaching the Ohio River. However, the dominant land use in the basin is
agriculture (80.3%) (www.miamiconservancy.org). The river has sections with
exposed riffles and rapids and sections with restricted flow associated with low-
head dams that store, rather than regulate, waters.
Table 1. Physical characteristics and mean percent (standard deviation) of land
use types in the study basins, with means and standard deviations based on
sites used in analyses.
River Basin
Parameter	Great Miami (n=6)	Kentucky (n=6)
Drainage Basin (km2)
13,947a
18,130b
River Length (km)
233.4a
410.4b
Average Gradient (m/km)
0.748
0.13b
% Urban land usec
5.90 (2.64)
7.81 (0.52)
% Agriculture land usec
82.55 (3.21)
80.18(1.34)
% Forested land use0
10.10 (1.44)
10.83(1.60)
a Ohio EPA, 1997
b Kentucky River Authority, 1999
0 Multi-Resolution Land Characteristics Consortium (Vogelmann eta!., 1998)
9

-------
The Kentucky River has a series of 14 lock-and-dam structures that span
the length of the mainstem, which historically supported commercial traffic. The
watershed has some large forested sections and some small areas with mining,
agricultural and urban influences (e.g., Lexington). As a result of impoundment,
all Kentucky River sites sampled in this study were much deeper than those of
the Great Miami River (Table 2).
Final site selection resulted in sites well-distributed longitudinally along the
mainstem of each river and included a mixture of habitat types. Study reaches
were positioned so that stream confluences, bridges and obvious stressor
Table 2. Ranges and medians of chemical and physical habitat variables at
study sites.
Great Miami River Kentucky River
(N=6)	(N=6)

Range
Median
Range
Median
Physical habitat




Mean thalwag depth (m)
1.2-2.3
2.03
5.2-9.7
7.0
Mean wetted width (m)
45.8-154.3
94.4
69.9-97.3
80.5
Mean bankfull height (m)
0.7-2.9
1.3
1.7-2.2
2.1
LWD quantity
11-70
30.0
20-43.0
30.5
% Canopy density at bank
0-92
37
71-92
78
% Substrate as large gravel and
0-89.5
25.5
0-0.92
0.78
larger at bank




% Urban in riparian
2.1-83.1
42.4
0.4-7.7
1.1
% Agriculture in riparian
9.0-77.1
27.7
9.7-33.6
14.0
% Forest in riparian
6.8-57.5
12.2
65.4-85.8
83.6
Water Chemistry




Mean conductivity (uS/cm)
521.2-857.2
664.6
270.6-435.2
334.2
S04 (mg/L)
33.0-64.4
45.3
33.9-104.6
79.8
N03 (mg/L)
1.37-5.69
1.95
0.37-0.80
0.56
Chloride (mg/L)
25.06-71.64
44.9
3.30-6.61
5.74
Ammonia (mg/L)
0.07-0.23
0.09
0.02-0.07
0.04
Total Khejldahl Nitrogen (TKN)
0.53-0.95
0.61
0.15-0.29
0.23
(mg/L)




Total Phosphorus (ug/L)
0.05-0.28
0.17
0.01-0.04
0.02
10

-------
sources, such as major outfalls, did not occur within the reach as this might
complicate and confuse the analysis of data within sites.
Sampling Design
An appropriate reach length for macroinvertebrates in non-wadeable
streams has not been estimated. However, benthic macroinvertebrate and fish
assemblage structure are often correlated (e.g., Kilgour and Barton 1999).
Therefore, the available literature on appropriate assessment units for fish in
large rivers was used for setting a maximum size for the study reach.
Measures of fish species richness is a function of the number of channel
units sampled (Gormann and Karr, 1978; Angermeier and Schlosser 1989, Lyons
1992), and the size and spacing of these units are functions of stream size
(Leopold et al. 1964). The assessment unit length required can also vary by
study objectives (Cao et al. 2001, Hughes et al. 2002). Lyons (1992) concluded
that for assessments of environmental quality or community-level ecological
analyses, a distance of 35 times the mean stream width, or a length equal to
three complete riffle-pool sequences, was sufficient. Pilot studies for the
Environmental Monitoring and Assessment Program suggested that in eastern
non-wadeable streams and rivers, a length of 40 channel widths was necessary
to characterize the fish community of a site (Herlihy and Lazorchak 2000). Using
this information, a reach length of 40 times the estimated mean wetted width of
the channel at each river site was selected. In hydrologically formed channels,
11

-------
this reach length would include approximately four meander wavelengths
(Leopold et al. 1964).
The downstream end of the study reach at each site was set at a
randomly determined point on one bank and marked with flagging. A systematic
sampling design was applied to establish 12 transects within the reach. This
design has many desirable features for field studies, and as long as the first point
is selected at random, remaining points based on that point can be considered
random as well (Cochran 1977). The simplicity of the design makes it easy to
execute without mistakes and results in significant time saving in the field. It also
results in the drawn sample being spread more evenly over the population
(Cochran 1977, Manly 2001).
Proceeding upstream from the initial point, 11 transects spanning the
width of the river were identified and flagged. The first 4 were spaced at a
distance equal to the mean wetted width of the channel, followed by 2 spaced at
2 times the wetted width, 2 at 4 times the wetted width, and 3 at 8 times the
wetted width (Figure 2). This identified 24 stations in the reach (e.g., 2 per
transect, 12 on each bank) where macroinvertebrate samples would be collected.
The size of the sampling zone at each sampling station was proportional
to the mean wetted width of the river. At each sampling station, a shoreline
sampling zone was defined as 0.1 times the estimated wetted width in shoreline
length and extended from the shore to the non-wadeable point of the river.
Therefore, if the river was 70 m wide, the sampling zone for each station would
be 7 m. This served to keep the sampling zone in proportion to the increasing
12

-------
Flow
B-
D-
n-
K"
1X
2X
4X
8X
Figure 2. Sampling scheme used to examine the effect of distance on metric
values.
13

-------
size of habitat features as the size of a river increased. Zone placement was
centered on the station transect.
Invertebrate Sampling
The technique used to collect benthic macroinvertebrate subsamples in
each station zone was a hybrid of existing techniques and consisted of two
distinct sample-collection procedures. The use of two distinct collection
approaches provides a representative benthic macroinvertebrate sample from
the different types of habitats encountered within and across rivers.
At each sampling station, two samples were collected with a modified kick-
net (50-cm wide x 30-cm tall x 60-cm bag-depth; 595-|jm mesh). The area in
front of the net equal to the width and length of the net frame (0.5 m; total area =
0.25m2) was then vigorously kicked for 20 seconds (see Klemm et al. 2000).
Next, a D-frame net (25.4-cm wide x 30.5-cm high x 25.4-cm bag-depth;
595 |jm mesh) was used to sample other available habitats in the sampling zone
(e.g., root wads, undercut banks, steep banks, vegetation). These habitats can
be quite difficult to sample with a kick-net procedure, and therefore, are often
underrepresented. For example, a river constrained by valley walls can be
nearly impossible to sample safely with a kick net. At each station the sampling
effort was standardized to 3 minutes per 5 m of sample zone width. While a
standardized sampling time may be sufficiently quantitative (Hynes 1970) and
used for quantifying effort, the main purpose of the timed effort was to control for
14

-------
the amount of time field personnel spent at any single station, thus assuring
ample time to cover the entire reach or multiple reaches in a day.
At some stations, only one collection procedure was feasible and suitable
for collection of a representative sample of the fauna of the prevailing habitat.
However, both were performed at every sampling zone if logistically safe and
practical.
Samples from the kick and D-frame nets were composited into a single
transect sample (n=12 per reach) for use in determining the appropriate number
of point-samples that needed to be collected and examining the effects of reach
length on sample results.
Each sample was processed in the field with a 595-pm sieve, preserved
with 100% ethanol, and diluted to a final concentration approximating 70%
ethanol (following Klemm et al. 2000). In the laboratory, individual samples were
completely sorted. All organisms were identified to the lowest practical
taxonomic level, usually species when specimen condition was adequate and
taxonomic keys were available.
Additional information was collected from each sample station (n=24) to
supplement the macroinvertebrate data, characterize each station and the study
reach, and document the gradient of conditions over which samples were
collected. Crews collected physical habitat data following EMAP protocols for
nonwadeable streams (Kaufmann 2000). A single depth-integrated water sample
was collected from each site and analyzed for sulfate, nitrate, total Kjeldahl
nitrogen, ammonia, total phosphorus, and chloride concentrations. Chemical
15

-------
analyses were conducted using EMAP-SW laboratory protocols (Klemm et al.
1990). Conductivity and water temperature were measured in situ using a YSI
Model 85 meter at the center of the sampling reach. Land cover data developed
by the Multi-Resolution Land Characteristics Consortium (Vogelmann et al. 1998)
were overlaid on a riparian corridor 500 m in width on each side of the river for a
distance of 4 km upstream of the center of the sampling reach. Proportions of
forest, agriculture, and urban (including residential) land uses were then
calculated within the riparian corridor.
Data Analyses
The differences in assemblage characteristics between the two banks at a
given transect were sometimes quite large. To encompass the spatial variability
present at each transect, samples were combined from the two banks at each
transect. Thus, all analyses described in this paper use samples from both
banks composited at each transect.
Subsample size.—Prior to analyses for estimating the minimum number of
transects required per site, an appropriate laboratory subsample size was
determined. The entire combined sample for each site (all transects combined)
was used to simulate fixed-count subsamples of 100 to 1000 organisms in steps
of 100. Simulations were run in C++ (Borland C++ Builder 4.0, Inprise
Corporation, Scotts Valley, California). It was assumed that organisms were
distributed randomly within each sample. Random sampling without replacement
16

-------
was used to simulate each subsample, and 100 subsamples were generated for
each site at each fixed count size to estimate laboratory sampling variability.
The effect of subsample size was measured on the quantitative metrics in
the Ohio EPA Invertebrate Community Index (ICI) because these metrics are
used to assess the macroinvertebrate assemblage in larger streams and rivers in
Ohio (Ohio EPA 1998). These metrics included total taxa richness, mayfly taxa
richness, caddisfly taxa richness, Diptera richness, % mayflies, % caddisflies, %
Tanytarsini, % non-Tanytarsini dipterans and non-insects, and % tolerant
individuals.
Since taxa richness metrics did not tend to level off with increasing
subsample size, the difference in a metric value between sites was used as a
way to measure the effect of sample size. The change in this absolute value of
the difference in the metric from one sample size (Xi) to the next higher sample
size (Xi+i) is defined as the "return", and the percent of the return relative to the
maximum value achieved for that metric (|Xj-Xj+i |/max(Xi+i)) as the "relative
return". In this calculation, the maximum value was set as the maximum for the
next higher sample size.
The subsample size at which the average relative return leveled off for
most metrics was selected for subsequent analyses on the effect of the number
of transects on metrics.
Number of transects-After the subsample size was determined, the number of
transects needed per site was evaluated. Following the concept of a species
17

-------
area curve, metric values were plotted as a function of the number of transects or
samples. This required randomizing the order of the transects to ensure that the
results were not affected by sequence of samples. However, the nature of the
sampling design meant that transects were not equidistant from one another. If
there was strong spatial autocorrelation among samples, randomizing the order
of transects would not be appropriate. Thus, spatial autocorrelation in
assemblage composition was tested by calculating the Coefficient of Community
(CC) similarity index (Sorensen 1948) for each pair of transects within each site.
The CC for each pair of transects was plotted against the distance between them
(Figure 3). There was no strong trend apparent between the CC and distance
and it was concluded that spatial autocorrelation was not prevalent.
Next, 100 randomizations in C++ were used to determine the number of
transects required before metric values leveled off. For each randomization,
transects were randomly ordered within each site. Next, transect data for
successively larger numbers of transects within each site was combined,
beginning with the first transect in the sequence. At each step in each
randomization, a simulated subsample was generated based on the Subsample
size results. For each metric and site, the average metric value across the 100
randomizations was plotted against the number of transects. The point at which
each metric leveled off was identified by visual inspection. Finally, a similar set
of simulations was run for smaller subsample size(s) to examine the influence of
subsample size on these plots.
18

-------
0.7
^ °-6 " ~
§	o -°" «
| 0.5	4 °°,rf < ^ <
0	0.4^%:4vlg ^A
1	0.3^	5>S^D-
tt ^	~ A
8	~ * Ai < a
° 0.2-	* "
v	V .
¦ 0 1000 2000 3000 4000 5000
Distance between transects (m)
Figure 3: Average Coefficient of Community Index value for each possible pair of
transects within each site as a function of the distance between transects. Each
type of symbol (shape and fill) represents a different site.
19

-------
Distance between transects.-After determining the number of transects required
per site, the affect of distance between transects on metric values was examined.
In each site, pairs of transects were grouped by the distance between them. This
created up to four groups with five pairs of transects each with inter-transect
distances of 1, 2, 4, and 8 times the mean wetted width. Not all sites had four
groups because of shortened reaches as described earlier. For each site, data
for all samples within a group were combined and 100 simulations of a 500-
organism subsample on each group were performed. The within-site differences
among groups were assessed qualitatively from plots of the mean metric value (±
1 SE).
Results
Although all 12 selected sites were sampled, the distance between dams
prevented sampling of all transects at three sites, severe weather conditions at
two sites, and loss of daylight at one site. The impact of these logistical
limitations on data analysis was negligible. Total reach length sampled at
individual sites ranged from 1200 to 4480 m in the Great Miami River and from
1680 to 4000 m in the Kentucky River. The range and median of water chemistry
and physical habitat variables at study sites are presented in Table 2. The
number of organisms per transect sample ranged from 63 to 2369 with a mean of
477.
20

-------
Subsample size.-The metric values for simulated samples quickly leveled off for
percentage metrics, but not for richness metrics (Figure 4). However, the
21

-------
900 1200
300 600 900 1200
Subsample size
900 1200
Figure 4: Results of subsample size simulations for each site and metric. Solid
lines represent sites in the Great Miami River and dashed lines represent sites in
the Kentucky River. Error bars represent 1 standard error of the mean.
22

-------
difference in richness metric values between any two sites did not change as
rapidly after approximately 500 organisms. In fact, the relative return dropped
below about 2% beyond 500 organisms (Figure 5), indicating that additional
sorting would not provide sufficient additional information in separating sites from
one another. There was also a significant drop between 200 and 300 organisms,
resulting in relative returns below 5% for 300 or more organisms. This
information is useful to note because most state programs subsample 300 or
fewer organisms for bioassessment samples in streams. Nonetheless, a
subsample of 500 organisms was used for further simulation analyses.
Number of transects.-There was a strong leveling off of richness metric values at
approximately six transects (Figure 6). For percentage metrics, the asymptote
typically was reached in fewer transects. When this analysis was rerun using a
subsample size of only 300 organisms, similar results were achieved (Figure 7).
Distance between transects.-Across all sites, there was no consistent pattern in
metrics based on five transects one wetted width apart to five transects eight
wetted widths apart (Figure 8). However, within individual sites, there were
sometimes very strong differences among the four groups, particularly between
the group of transects separated by a distance of eight times the wetted width
and the other three groups of transects separated by smaller distances (Figure
8). Retrospective analysis of the physical habitat data suggests that at some
23

-------
sites, as the distance between transects increased, the likelihood of encountering
large
24

-------
Subsample size
Figure 5: Relative return as a function of subsample size for each richness (top)
and percentage (bottom) metric and site. Error bars represent 1 standard error
of the mean value.
25

-------

8

7

6
"c

\n
5
is

c
4
CO

H
3


0s
2

1

0


CO
0
Q_
b
10
15
90
80
70
60
50
40
30
20
10
0
10
15
60
50
40
30
20
10
0
10
i_i
15
Number of transects
Figure 6: Metric values with increasing number of transects based on 500
organism simulated counts. Solid lines represent Great Miami River and dashed
represent Kentucky River sites. Error bars represent 1 standard error.
Vertical dashed line represents estimated point at which leveling-off occurs.
26

-------

co 90r
BBSS

10
Number of transects
Figure 7: Metric values with increasing number of transects based on 300
organism simulated counts. Solid lines represent Great Miami River and dashed
represent Kentucky River sites. Error bars represent 1 standard error. Vertical
dashed line represents estimated point at which leveling-off occurs.
27

-------
I I I I I I I I I
20
m 15
03
X
03
-i—»
^ 10
03
5
0
i i i i i i i i

8

7
03
X
6
03


5
>*

M=
(/)
4
~o

"O
03
3
o
2

1

0
[\
j	i	!	i
0123456789
Wetted widths apart
			i	i	i	i	i	i	i	i	i
0123456789
Figure 8: Metric values across groups (define) equidistant by varying numbers of
channel widths for each site. Solid lines represent Great Miami River sites and
dashed lines represent Kentucky River sites.
28

-------
variability in one or more gross physical habitat features (e.g., thalweg depth,
substrate composition) increased, but the habitat feature(s) causing the variability
was not the same across sites.
Given the study results and the reality that the vast majority of non-
wadeable streams and rivers have been modified in ways that undermine the
rationale of using geomorphology as a determinant of reach length, a fixed reach
length rather than one based on multiples of the wetted widths is appropriate.
From the study data, the range of distances covered by six transects was 270 m
to 960 m, with a median of 480 m and a mean of 542.5. Based on these results,
and taking into consideration the logistics of field sampling, the reach length was
set at 500 m. This results in six transects separated by a fixed distance of 100 m
each for the LR-BP.
Discussion
A new sampling protocol was developed for sampling large river
macroinvertebrates that is specifically designed to perform well across all shore-
line habitats and river types, integrate different habitats, and thus represent site
conditions accurately. The appropriate fixed count for laboratory subsampling
size to use with this method was also determined, based on the ability to
separate sites of differing macroinvertebrate composition.
This research indicates that, using the sampling technique discussed
herein, a representative sample of the benthic macroinvertebrate fauna of the
study reaches was collected by sampling both banks of six transects. These
29

-------
results were achieved because the sampling method and design effectively
sampled the benthic macroinvertebrate fauna of the dominant habitat types
within the reach. It is further recommended that transects be evenly-spaced over
a 500-m distance, a distance equivalent to that used for many years by several
state agencies for sampling riverine benthic macroinvertebrate and fish
assemblages (e.g., Ohio EPA, 1989, Royer et al. 2001).
The recommended design results in a composite sample consisting of 24
20-second kick-net samples and 12 timed samples collected with a D-frame net
in habitats complementing those sampled by the kick-net. Therefore, the final
composite sample from the 500-m reach consists of 36 subsamples collected by
two complementary sampling techniques. It is likely this approach will over-
sample sites of uniform composition, but the goal was to develop a standardized
LR-BP that would perform well across sites of differing habitat composition.
These conclusions agree with Bartsch et al. (1998), who stated that 18 to 40
subsamples may be required to adequately sample large flood-plain rivers, and
that no single sampling technique would efficiently and adequately sample all
components of a riverine macroinvertebate community. It should be noted that
this LR-BP for macroinvertebrates has only been tested in main-channel habitats.
It may work equally well on off-channel habitats, but this remains to be tested.
It is recommended that the field method be coupled with a fixed laboratory
subsample size of 300 or 500 organisms to maximize effectiveness of the LR-BP
for bioassessment purposes. The fixed laboratory subsample size of 500
organisms does offer lower variability for percentage metrics, but variability for
30

-------
richness metrics was higher. Nonetheless, 300 is in all probability sufficient for
most study needs. This recommendation is based on the response of the tested
metrics and the observation that the ability to separate sites of different
macroinvertebrate composition generally did not increase with larger subsample
sizes. Studies on other systems have recommended a broad range of
subsample sizes as sufficient (Barbour and Gerritsen, 1996, Vinson and
Hawkins, 1996, Growns et al. 1997, Somers et al. 1998). However, a one-size-
fits-all subsample size should not be expected, since the quality of information
needed by researchers and managers can vary depending on individual studies
(Doberstein et al. 2000). The best strategy for determining an appropriate
subsample size is to first determine the data quality requirements to meet study
objectives and then determine the appropriate subsample size from collected
data. This seems especially appropriate when developing new or modifying
existing field methods.
An advantage of the proposed protocol is that field crews targeting non-
wadeable streams and rivers can be sent into the field with a single method that
works well across a variety of site types. While the method has been designed to
perform well in a variety of habitats, resulting data should be interpreted with
appreciation for coarse habitat characteristics. This information would ideally be
derived from habitat data collected concomitantly with the faunal data. Sites can
then be categorized into river types (e.g., impounded vs. free-flowing or lowland
vs. upland rivers) or even habitat types within a specific river (e.g., sandy- vs.
cobble-bottom) in a more controlled environment (i.e., in the laboratory).
31

-------
The structured nature of the LR-BP provides a standardized sampling
protocol that produces a representative sample from the varying habitats and
changing impoundment conditions (through time and space) encountered within
and across large rivers. Sites in this study varied from free-flowing to those with
hydrologic modifications associated with lock-and-dam systems, habitat
modifications due to channelization, and the presence of low-head dams. In
habitats where both sampling methods could be performed, one method did not
supersede the other, and both were performed. At others, for example, the
banks of a sampling station may have been too steep, rendering the collection of
a sample via the kick-net method logistically impossible. However, using the D-
frame net from the boat, the benthos would be sampled. Hence, a habitat that
would have gone unrepresented using a sampling approach that relied purely on
kick-net sampling, was still represented in the composite sample of the site.
As a result of the additional equipment required to work in non-wadeable
streams and rivers (i.e., boats and associated equipment), the effort required to
secure a representative sample for bioassessment generally exceeds that
required in wadeable streams. Given these realizations, the proposed sampling
method is cost effective, logistically feasible, and collects a representative
sample for bioassessment purposes. Critical elements of the LR-BP include the
complementary sampling techniques, the distance of the sample reach, the
number of transects at which both banks are sampled, and the subsample size in
the laboratory.
32

-------
Future Research.- With development of this initial design, additional field
sampling has been conducted to allow performance-based testing (Diamond et
al. 1996) of the field and laboratory components of the LR-BP. Additional
research may also be needed to determine applicability of the LR-BP for use in
riverine ecosystems functioning differently than those described in this study
(e.g., floodplain-river ecosystems, riverine-influenced reservoirs, fast-flowing
rivers). Possible modifications to the method to streamline its application in the
field include using the D-frame net configuration for both the kick- and dip-net
sampling, setting a depth criterion for kick-net sampling (e.g., 1 m), and using a
fixed distance for sample zones (e.g., 10 m). Experimenting with an area
quantification of the dip-net sampling may also be considered for use in studies
requiring full quantification of sampling effort.
33

-------
Literature Cited
Allen, J. D. 2000. Stream Ecology: Structure and function of running waters.
Kluwer Academic Publishers. Dordrecht, Netherlands.
Angermeier, P. L., and I. J. Schlosser. 1989. Species-area relationships for
stream fishes. Ecology 70:1450-1462.
Barbour, M. T., and J. Gerritsen. 1996. Subsampling of benthic samples: a
defense of the fixed-count method. Journal of the North American
Benthological Society 15:386-391.
Barbour, M. T., J. Gerritsen, B. D. Snyder, and J. B. Stribling. 1999. Rapid
bioassessment protocols for use in streams and wadeable rivers:
periphyton, benthic macroinvertebrates, and fish, second edition. EPA
841-B-99-002. U.S. Environmental Protection Agency; Office of Water;
Washington, D.C.
Bartsch, L. A., W. B. Richardson, and T. J. Naimo. 1998. Sampling benthic
macroinvertebrates in a large floodplain river: considerations of study
design, sample size, and cost. Environmental Monitoring and Assessment
52:425-439.
Benke, A. C. 1990. A perspective on American's vanishing streams. Journal of
the North American Benthological Society 9(1):77-88.
Blocksom, K. A., and J. E. Flotemersch. 2004. Comparison of
Macroinvertebrate Sampling Methods for Nonwadeable Streams.
Environmental Monitoring and Assessment (In Press).
34

-------
Cairns, J., Jr. 1982. Artificial Substrates. Ann Arbor Science Publications, Inc.,
Ann Arbor, Michigan.
Cao, Y., D. P. Larsen, and R. M. Hughes. 2001. Evaluating sampling sufficiency
in fish assemblage surveys: a similarity-based approach. Canadian
Journal of Fisheries and Aquatic Sciences 58:1782-1793.
Carter, J. L., and V. H. Resh. 2001. After site selection and before data
analysis: sampling, sorting, and laboratory procedures used in stream
benthic macroinvertebrate monitoring programs by USA state agencies.
Journal of the North American Benthological Society 20:658-682.
Cochran, W. G., 1977. Sampling Techniques. John Wiley and Sons, New York.
Diamond, J. M., M. T. Barbour, and J. B. Stribling. 1996. Characterizing and
comparing bioassessment methods and their results: a perspective.
Journal of the North American Benthological Society 15:713-727.
Doberstein, C. P., J. R. Karr, and L. L. Conquest. 2000. The effect of fixed-
count subsampling on macroinvertebrate biomonitoring in small streams.
Freshwater Biology 44:355-371.
Flotemersch, J. E., B. C. Autrey, and S. M. Cormier. 2001. Comparisons of
boating and wading methods used to assess the status of flowing waters.
U.S. Environmental Protection Agency, Cincinnati, Ohio, EPA/600/R-
00/108.
Frissell, C. A., W. J. Liss, C. E. Warren, and M. D. Hurley. 1986. A hierarchical
framework for stream habitat classification: viewing streams in a
watershed context. Environmental Management 10(2):199-214.
35

-------
Gormann, O. T., and J. R. Karr. 1978. Habitat structure and stream fish
communities. Ecology 59:507-515.
Growns, J. E., B. C. Chessman, J. E. Jackson, and D. G. Ross. 1997. Rapid
assessment of Australian rivers using macroinvertebrates: cost and
efficiency of 6 methods of sample processing. Journal of the North
American Benthological Society 16:682-693.
Herlihy, A. T., and J. Lazorchak J. 2000. Initial Site Procedures. Pages 4.1-4.8 in
J. M. Lazorchak, B. H. Hill, D. K. Averill, D. B. Peck, and D. J. Klemm
(editors) Environmental monitoring and assessment program-surface
waters: field operations and methods for measuring the ecological
condition of non-wadeable rivers and streams. EPA 620/R-00/007. Office
of Research and Development, US Environmental Protection Agency,
Cincinnati, Ohio.
Humphries, P., J. E. Growns, L. G. Serafina, J. H. Hawking, A. J. Chick, and P.
S. Lake. 1998. Macroinvertebrate sampling methods for lowland
Australian rivers. Hydrobiologia 364(2):209-218.
Hughes, R. B., P. R. Kaufmann, A. T. Herlihy, S. S. Intelmann, S. C. Corbett, M.
C. Arbogast, and R. C. Hjort. 2002. Electrofishing distance needed to
estimate fish species richness in raftable Oregon rivers. North American
Journal of Fisheries Management 22:1229-1240.
Hynes, H. B. N. 1970. The ecology of running water. Liverpool University
Press, Liverpool.
36

-------
Karr, J. R., and E. W. Chu. 1999. Restoring Life in Running Waters: Better
Biological Monitoring. Island Press, Washington, D.C.
Kaufmann PR. 2000. Physical Habitat Characterization - non-wadeable rivers.
Pages 6.1-6.29 in J. M. Lazorchak, B. H. Hill, D. K. Averill, D. B. Peck, and
D. J. Klemm (editors) Environmental monitoring and assessment program-
surface waters: field operations and methods for measuring the ecological
condition of non-wadeable rivers and streams. EPA 620/R-00/007. Office
of Research and Development, US Environmental Protection Agency,
Cincinnati, Ohio.
Kentucky River Authority. 1999. Kentucky River Basin Status Report.
http://water.nr.state.ky.us/dow/pubs/kybasin97/status1.htm. Kentucky
River Authority, Frankfort, Kentucky.
Kilgour, B. W., and D. R. Barton. 1999. Associations between stream fish and
benthos across environmental gradients in southern Ontario, Canada.
Freshwater Biology 41:553-566.
Klemm, D. J., P. A. Lewis, F. A. Fulk, and J. M. Lazorchak. 1990.
Macroinvertebrate field and laboratory methods for evaluating the
biological integrity of surface waters, EPA 600/4-90/030, U.S.
Environmental Protection Agency, Cincinnati, Ohio.
Klemm, D. J., J. M. Lazorchak, and D. V. Peck. 2000. Benthic
macroinvertebrates. Pages 9.1-9.14 in J.M. Lazorchak, B. H. Hill, D. K.
Averill, D.B. Peck, and D. J. Klemm (editors) Environmental monitoring
and assessment program-surface waters: field operations and methods for
37

-------
measuring the ecological condition of non-wadeable rivers and streams.
EPA 620/R-00/007. Office of Research and Development, US
Environmental Protection Agency, Cincinnati, Ohio.
Leopold, L. B., M. G. Wolman, and J. P. Miller. 1964. Fluvial processes in
geomorphology. W.H. Freeman: San Francisco.
Li, J., A. Herlihy, W. Gerth, P. Kaufmann, S. Gregory, S. Urquhart, and D.
Larsen. 2001. Variability in stream macroinvertebrates at multiple spatial
scales. Freshwater Biology 46:87-97.
Loucks, D. P. 2003. Managing America's rivers: who is doing it? International
Journal of River Basin Management 1:23-31.
Lyons, J. 1992. The length of stream to sample with a towed electrofishing unit
when fish species richness is estimated. North American Journal of
Fisheries Management 12:198-203.
Manly, B. F. J. 2001. Statistics for environmental science and management.
Chapman and Hall/CRC. Boca Raton, FL.
Metcalfe-Smith, J. L. 1994. Biological water-quality assessment of rivers: use of
macroinvertebrate communities. Pages 144-170 in P. Colow and G. E.
Petts (editors). The rivers handbook: Hydrological and ecological
principles. Blackwell Scientific. Oxford, United Kingdom.
Moulton, S. R., II, J. G. Kennen, R. M. Goldstein, and J. A. Hambrook. 2002.
Revised protocols for sampling algal, invertebrate, and fish communities
as part of the National Water Quality Assessment Program. U.S. Geologic
Survey Open File Report 02-150.
38

-------
Ohio EPA. 1997. Biological and Water Quality Study of the Middle and Lower
Great Miami River and Selected Tributaries, 1995. Division of Surface
Water, Monitoring and Assessment Section, Ohio Environmental
Protection Agency, Columbus, Ohio.
Ohio EPA. 1988. Biological criteria for the protection of aquatic life: Volume II:
users manual for biological field assessment of Ohio surface waters. Ohio
EPA, Division of Water Quality Monitoring and Assessment, Surface
Water Section, Columbus, Ohio.
Ohio EPA. 1989. Biological criteria for the protection of aquatic life: Volume III:
Standardized biological field sampling and laboratory methods for
assessing fish and macroinvertebrate communities. Ohio EPA, Division of
Water Quality Monitoring and Assessment, Surface Water Section,
Columbus, Ohio.
Poulton, B. C., M. L. Wildharber, C. S. Charbonneau, J. F. Fairchild, B. G.
Mueller, and C. J. Schmitt. 2003. A longitudinal assessment of the
aquatic macroinvertebrate community in the channelized lower Missouri
River. Environmental Monitoring and Assessment 85:23-53.
Rosenberg, D. M., and V. H. Resh. 1993. Freshwater Biomonitoring and
Benthic Macroinvertebrates, Chapman and Hall, London.
Royer, T. V., C. T. Robinson, and G. W. Minshall. 2001. Development of
macroinvertebrate-based index for bioassessment of Idaho rivers.
Environmental Management 27:627-636.
39

-------
Somers, K. M., R. A. Reid, and S. M. David. 1998. Rapid biological
assessments: how many animals are enough? Journal of the North
American Benthological Society 17:348-358.
Sorensen, T. 1948. A method of establishing groups of equal amplitude in plant
sociology based on similarity of species content. Det Kongelige Danske
Videnskabernes Selskab, Biologiske Skrifter, (Copenhagen) 5:1-34.
Sweeting, R. A. 1994. River Pollution. Pages 23-32 in P. Colow and G.E. Petts
(editors) The rivers handbook: Hydrological and ecological principles.
Blackwell Scientific. Oxford, United Kingdom.
Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R. Sedell, and C. E. Cushing.
1980. The river continuum concept. Canadian Journal of Fisheries and
Aquatic Science 37:130-137.
Vinson, M. R., and C. P. Hawkins. 1996. Effects of sampling area and
subsampling procedure on comparisons of taxa richness among streams.
Journal of the North American Benthological Society 15:392-399.
Vogelmann, J. E., T. L. Sohl, P. V. Campbell, and D. M. Shaw. 1998 Regional
land cover characterization using landsat thematic mapper data and
ancillary data sources. Environmental Monitoring and Assessment 51:415-
428.
www.miamiconservancy.org. 4/22/04. Great Miami River Watershed.
40

-------
ŁEPA
United States
Environmental Protection
Agency
Office of Research
and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
EPA 600/R-04/177
October 2004
www.epa.gov
Please make all necessary changes on the below label,
detach or copy, and return to the address in the upper
left-hand corner.
If you do not wish to receive these reports CHECK HERE
~ ; detach, or copy this cover, and return to the address in
the upper left-hand corner.
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
®
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free

-------