4>EPA
United States
Environmental Protection
Agency
Report on the Regional Environmental Monitoring
and Assessment Program Study of Wadeable
Streams in the Driftless Area Ecoregion in Western
Wisconsin
RESEARCH AND DEVELOPMENT
-------
EPA600/R-06/165
November 2006
www.epa.gov
Report on the Regional Environmental Monitoring and
Assessment Program Study of Wadeable Streams
in the Driftless Area Ecoregion in Western Wisconsin
Michael A. Miller1, Alison C.C. Colby1, and Paul Kanehl2
1Wisconsin Department of Natural Resources
Bureau of Fisheries Management and Habitat Protection
101 South Webster St.
Madison, Wisconsin 53707
2Wsconsin Department of Natural Resources
Bureau of Integrated Science Services
2801 Progress Rd.
Madison, Wsconsin 53716
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.
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
-------
Notice
The United States Environmental Protection Agency (EPA) through its Office of
Research and Development funded and managed the research described here via a
grant (#R-829537-01-0). It has been reviewed by the EPA and approved for
publication.
Mention of trade names or commercial products does not constitute endorsement
or recommendation by EPA for use.
-------
Foreword
The mission of the Ecological Exposure Research Division (EERD) within the National Exposure
Research Laboratory (NERL) of the U.S. Environmental Protection Agency (EPA) is to improve
the scientific basis for understanding, measuring, and protecting biological integrity so that EPA
and other resource agencies can make sound, defensible environmental decisions. Our
research is primarily focused on the development, evaluation, and implementation of new
methods to assess ecosystem condition, to evaluate biotic responses to environmental
stressors, and to predict future vulnerability of natural populations, communities, and
ecosystems. The scale of our research ranges from molecular to ecosystem levels of biological
organization and addresses immediate as well as emerging environmental threats.
EPA's Environmental Monitoring and Assessment Program (EMAP) is a research program to
develop the tools necessary to monitor and assess the status and trends of national ecological
resources. EMAP's goal is to develop the scientific understanding for translating environmental
monitoring data from multiple spatial and temporal scales into assessments of current ecological
condition and forecasts of future risks to our natural resources. EMAP focuses on surveys
based on probability (i.e., random) sampling designs to estimate condition with a known level of
uncertainty. Regional EMAP (REMAP) was initiated to test the applicability of the EMAP
approach to answer questions about ecological conditions at regional and local scales. REMAP
proposals are submitted through EPA's Regional Offices to the Office of Research and
Development (ORD). ORD carries out scientific review of proposals, and qualified proposals
are funded as cooperative agreements.
This report describes the results of a REMAP agreement awarded to the Wisconsin Department
of Natural Resources (WDNR) to compare random and modified-random sampling designs.
The EPA Project Officer was Bhagya Subramanian, and Karen Blocksom, a statistician in
EERD, provided assistance with data analysis. The report was prepared by WDNR and has
undergone EPA review. EERD is publishing this report to make these findings more widely
available, given their potential significance for state or tribal agencies that, like WDNR, are
considering wider adoption of random-sampling approaches in their water quality monitoring
programs.
Florence Fulk
Acting Director
Ecological Exposure Research Division
-------
Table of Contents
Notice 2
Foreword 3
Tables 6
Figures 8
1. Project Summary 10
1.1 Background and Rationale 10
1.2 Project Objectives and Focus of This Report 10
1.3 Methods 11
1.4 Results 11
1.5 Importance to the Science of Environmental Monitoring 12
1.6 Other Analyses Planned for Wisconsin REMAP Data 13
2. Study Area 14
2.1 Watershed Delineation, Land Use Quantification, and Water Quality
Stressor Gradient Development 14
2.2 Stream Resources in Study Area 14
3. Stream Population Sampling Design 17
3.1 Random, Modified-Random, Reference, and Targeted Stream Sites
Selection 17
3.2 Identification of Sampling Sites 18
3.3 Temporal Sampling Frame 18
4. Stream Site Sampling Protocols 19
4.1 Physical Habitat 19
4.2 Water Quality and Water Chemistry 19
4.3 Periphyton 20
4.4 Benthic Macroinvertebrates 20
4.5 Fish 21
5. Data Analytical Methods 22
5.1 Random, Bridge, and Reference Site Comparisons 22
5.2 Probability Estimates and Evaluation of Population Distributions 22
5.3 Land Use Evaluation 23
6. Results 24
6.1 Sample Size Summary 24
6.2 Watershed Land Use Identification and Quantification - Relationships
Among Agricultural Land Use and Physical, Chemical, and Biological
Measures 25
6.3 Random, Modified-Random, and Reference Site Comparisons 30
-------
6.4 Comparison of Reference Sites with Random and Modified-Random
Assessment Sites 38
6.5 Probability Estimates and Evaluation of Population Distributions 46
6.6 Targeted vs. Random Stream Assessment 58
6.7 Effects of Geographical Distances Between Random and Modified-Random
Sites 59
7. Discussion 61
8. Acknowledgements 66
9. References 67
Appendix A. Comparison of Spring and Fall Water Chemistry Parameters 70
Appendix B. Relationships of Stream Order with Physical Habitat, Water Quality and
Water Chemistry, and Biological Measures 72
Appendix C. Fish Assemblage Data Analyses 76
-------
Tables
Table 1. Pearson correlations between watershed land use and in-stream physical
habitat measures (N = 78) 25
Table 2. Pearson correlations between watershed land use and water quality and
water chemistry measures 26
Table 3. Pearson correlations between in-stream physical habitat, watershed land
use, and biological measures 28
Table 4. Pearson correlations between water quality and water chemistry
measures, and biological measures 29
Table 5. Weighted paired t-test comparisons of stream physical habitat measures
from X and B stream assessment sites 31
Table 6. Weighted paired t-test comparisons of water quality and water chemistry
measures from the X and associated B assessment sites 33
Table 7. Weighted paired t-test comparisons of macroinvertebrate metrics and
indices collected from X and associated B assessment sites 35
Table 8. Weighted paired t-test comparisons offish metrics collected from X and B
assessment sites 37
Table 9. Two sample t-test comparisons for physical habitat measures collected at
the X and reference sites 39
Table 10. Two-sample t-test comparisons of water quality and water chemistry
values between the X and reference stream sites 41
Table 11. Two sample t-test comparisons of the X and reference site
macroinvertebrate metrics 43
Table 12. Two-sample t-test comparisons of the X and reference site fish metrics 45
Table 13. Physical habitat reference condition thresholds 46
Table 14. Water chemistry and water quality reference condition thresholds 49
Table 15. Macroinvertebrate metric reference condition thresholds 53
Table 16. Fish metric reference condition thresholds 55
Table A 1. Paired t-test comparisons between spring and fall water quality and
water chemistry measures collected at the X stream sites 70
-------
Table B 1. ANOVA results of stream order and physical habitat measures
collected at the X sites 72
Table B 2. Tukey HSD pairwise comparison probabilities for physical habitat
measures that produced significant ANOVA results 72
Table B 3. ANOVA of stream order and water quality and water chemistry
measures from the X stream sites 73
Table B 4. Tukey HSD pairwise comparison for the percent oxygen saturation 73
Table B 5. ANOVA of stream order and macroinvertebrate measures from the X
stream sites 74
Table B 6. ANOVA of stream order and fish assemblage measures from the X
stream sites (N=42) 75
Table B 7. Tukey HSD pairwise comparison probabilities for fish metrics that
produced significant ANOVA results 75
Table C 1. Fish species captured at X stream sites 77
Table C 2. Fish species captured at reference stream sites 78
-------
Figures
Figure 1. Random and least disturbed reference stream sites in the REMAP
study area 16
Figure 2. Bivariate scatterplots comparing physical habitat measurements
collected at the X and associated B sites 30
Figure 3. Bivariate scatterplots comparing water quality and water chemistry
measurements collected from the X and associated B sites 32
Figure 4. Bivariate scatterplots comparing macroinvertebrate metrics collected at
the X and associated B sites 34
Figure 5. Bivariate scatterplots comparing fish metrics collected at the X and
associated B sites 36
Figure 6. Bar charts of weighted means and standard deviations (SD) for X and B
sites and simple random sample mean and SD for reference sites for physical
habitat measures 38
Figure 7. Bar charts of means and standard deviations for water quality and water
chemistry measures collected at the X, B, and reference stream sites 40
Figure 8. Bar charts of mean and standard deviation of macroinvertebrate metrics
collected at the X, B, and reference stream sites 42
Figure 9. Bar charts of mean and standard deviation of fish metrics collected at
the X, B, and reference stream sites 44
Figure 10. Cumulative distribution function curves of physical habitat measures
collected at the X and B sites 47
Figure 11. The percentages of stream miles not meeting the reference condition
threshold values for physical habitat measures 48
Figure 12. Cumulative distribution function curves of water quality measures
collected in situ at the X and B sites 50
Figure 13. Cumulative distribution function curves of laboratory analyzed water
chemistry measures collected at the X and B sites 51
Figure 14. The percentages of stream miles not meeting the reference condition
threshold values for water quality and water chemistry measures 52
Figure 15. Cumulative distribution function curves of macroinvertebrate metrics
collected at the X and B sites 54
8
-------
Figure 16. The percentages of stream miles not meeting the reference condition
threshold values for macroinvertebrate metrics 55
Figure 17. Cumulative distribution function curves for fish metrics collected at the
Xand B sites 56
Figure 18. The percentages of stream miles not meeting the reference condition
threshold values for fish metrics 57
Figure 19. Cumulative distribution function curves of fish IBI scores from
biologists' targeted sites and REMAP X sites, compared to the reference
condition threshold value. Box plots and distribution of the biologist's targeted
and REMAP X sites data is shown in figure on right 58
Figure 20. Scatterplots showing relationships between the distance between the
X and B sites, and the absolute value of the differences in physical, chemical, and
biological measures collected at these sites 59
Figure A 1. Box and whisker plots of spring and fall water quality and water
chemistry measures collected at the X-sites 71
Figure C 1. Percent frequency of occurrence of fish species at the X, B, and
reference sites 79
-------
1. Project Summary
1.1 Background and Rationale
Hundreds of millions of dollars have been spent in Wisconsin on watershed management and
restoration activities, yet certain farming, and suburban and urban development practices
continue to significantly impact a large portion of the state's waters, and in some areas of the
state severe stream degradation is readily apparent. Improvements in the assessment of
Wisconsin stream resources are needed to provide a comprehensive evaluation of the state's
waters. Greater understanding of land use factors affecting water resources will help improve
land and water resource management, and broader dissemination of this information is needed
to improve the political decision-making processes for these management activities.
To date, stream monitoring conducted by the WDNR has primarily been targeted sampling to
provide information for stream-specific management issues. Physical, chemical, and biological
data are often collected from either highly degraded streams affected by polluted run-off, or from
high quality streams where game fish management or stream habitat enhancement efforts are
being evaluated. This resulting data set can be strongly biased if used for making inferences
about broad-scale conditions of stream resources. Spatial clustering of the WDNR's current
sampling effort on a relatively small proportion of the State's streams and a focus on larger
streams that support adult game fish also limit the ability to make meaningful statements about
Wisconsin's entire stream population.
Beginning in 2003, the Wisconsin Department of Natural Resources (WDNR), with support from
the EPA's Regional Environmental Monitoring and Assessment Program (REMAP), conducted
an assessment of the physical, chemical, and biological conditions of wadeable streams in the
Driftless Area ecoregion in western Wisconsin using a probabilistic sampling design. The
Driftless Area ecoregion encompasses 20 percent of Wisconsin's total land area and contains
21 percent of the State's perennial stream miles.
1.2 Project Objectives and Focus of This Report
The Wisconsin REMAP study was conducted to address the following objectives:
Objective 1:
Use a statistically valid probabilistic sampling design to assess the condition of the entire
wadeable stream population in the Driftless Area ecoregion, and evaluate whether a modified-
random sample survey design (sampling randomly selected stream reaches near road-
accessible access points) characterizes individual or populations of streams similarly to a truly
randomized survey sampling design.
Objective 2:
Evaluate whether targeted stream sampling routinely conducted by the WDNR characterizes
stream fish populations similarly to a probabilistic sampling design.
Objective 3:
Investigate whether the effects of agricultural land use can be detected in the quality of in-
stream physical habitat or water chemistry measures, and whether changes in stream physical
or chemical characteristics influence the biological assemblages in streams.
10
-------
Objective 4:
Evaluate whether one biotic assemblage is more discriminating of specific stressors such as
percent agricultural land within a watershed, in-stream or riparian physical habitat degradation,
or chemical pollutants, than others, and whether there are significant differences in the
estimated miles of degraded streams depending upon the biotic assemblage used in the
estimation.
Objective 5:
Compare macroinvertebrate taxonomic data collected using two different field sampling
protocols, differing laboratory sub-sample sizes, and differing levels of laboratory taxonomic
resolution, to evaluate whether certain field or lab protocols are more effective at detecting and
measuring various types of environmental stressors. Additionally, develop a multi-metric
macroinvertebrate index for the Driftless Area ecoregion and subsequently the entire state,
using the measured physical and chemical explanatory data, and the more rigorous protocols
identified.
This report presents the results of data analyses addressing Objectives 1-4 for
macroinvertebrate and fish assemblages. Algae were collected and processing of these
samples is underway. Analyses to address Objective 5 are in progress.
1.3 Methods
Watershed land use, riparian and in-stream habitat, field and laboratory-analyzed water
chemistry, periphyton, macroinvertebrate, and fish data were collected from randomly selected
stream sites (n = 60), and an associated "modified-random" sampling site on each of these
streams, that was accessed via a road crossing nearest the randomly selected sampling site.
These data allowed us to evaluate whether sampling randomly selected stream sites at the
nearest road crossings would significantly bias the assessment results compared to sampling
truly random sites. Least disturbed reference sites were sampled to develop stream-quality
expectations based on objective reference condition criteria (n = 22). Also, WDNR biologists
provided fish assemblage data from 60 targeted stream sites they had sampled, and which were
thought to be representative of the range and modal condition of stream resources in the study
area. We compared fish data from these targeted sites to that from the REMAP random sites to
evaluate whether targeted sampling characterized the Driftless Area ecoregion stream
population similarly to the stratified random survey sampling design.
1.4 Results
The mean distance between the random (probabilistic) and the associated modified-random
assessment site on each stream was 701 meters. Study results show no significant differences
between the random and modified-random assessment sites for all 9 physical habitat measures,
11 water chemistry measures, 7 macroinvertebrate metrics, or 8 fish metrics.
We provide evidence that targeted sampling routinely done by WDNR characterized the
biological condition of stream resources in the Driftless Area ecoregion differently than
probabilistic sampling. Cumulative distribution plots of fish Index of Biotic Integrity (IBI) scores
from targeted stream sites sampled by WDNR regional biologists indicate that 65% of the
stream sites in the study area were not meeting the reference condition threshold value of 60,
versus the estimate of 80% of the random sample sites not meeting this threshold based on the
probabilistic sampling design results.
11
-------
Row cropping is the dominant land use in the study area followed by forest cover. Study results
indicate there were no significant relationships between the percentage of agricultural land in
each watershed and the degradation of various in-stream physical habitat features, but water
column nitrate-nitrite concentrations and conductivity were positively correlated with percent
watershed agricultural land. Fish assemblages appeared to be insensitive indicators of water
quality, but sensitive macroinvertebrate taxa declined with increased Kjeldahl nitrogen,
ammonia, total phosphorus and total dissolved phosphorus concentrations, and were positively
correlated with dissolved oxygen concentration, percent dissolved oxygen saturation and water
transparency.
Physical, chemical, and biological measures from least disturbed reference stream sites were
used to develop reference conditions. There were significant differences between the random
sample population (and by inference the entire stream population in the Driftless Area ecoregion
of Wisconsin) and the reference condition for a number of stream physical, chemical, and
biological measures. As examples, we documented that 75% of the random sample population
streams were degraded based on measures of stream bank erosion; 70% of the random sample
population had higher total dissolved phosphorus concentrations than the reference condition;
approximately 75% of the streams were degraded based on the percentage of sensitive
macroinvertebrate taxa present; and 74% of the streams showed impairment based on fish IBI
scores.
1.5 Importance to the Science of Environmental Monitoring
Randomization is an important aspect of sample survey design as it allows objectivity in the
sample selection process, and in the evaluation of sources of error or sampling variability
(Larsen 1995, Peterson et al. 1999). Probabilistic sampling designs can reduce sampling bias
and increase the representativeness of sample data used to make inferences about the target
population. Previous experience from employing probabilistic sampling designs in Wisconsin
have shown disadvantages when using this type of sampling design including: 1) Reduced
accessibility and increased travel times to more remote sites can significantly increase sampling
effort and cost; 2) Reduced site accessibility can result in having to use more transportable and
often less effective sampling gear; and 3) Streams highly degraded by poor land management
are often associated with landowners unwilling to allow access to streams on their property due
to distrust of government agencies or fear of legal action. With the REMAP project, we
evaluated whether random, modified-random, and non-random (targeted) sampling designs
provided similar results. These findings are being used to evaluate fundamental aspects of
stream sampling designs, which should help improve the monitoring and assessment of stream
resources in Wisconsin and elsewhere.
Knowledge of what the physical, chemical, or biological conditions of a stream should be in the
absence of human disturbances, and understanding of how streams respond to anthropogenic
stressors, are important to guide assessment and management efforts (Karr and Chu 1999,
Davies et al. 2000). The Wisconsin REMAP project initiated the collection of physical, chemical,
and biological data from least disturbed stream sites throughout western Wisconsin to help
define ecoregional expectations (reference conditions). This methodology will subsequently be
applied to the entire state. Land use types within the watersheds upstream from each study site
were quantified to evaluate how land use influence stream habitat and water quality
characteristics, which in turn can influence the biotic assemblages.
12
-------
1.6 Other Analyses Planned for Wisconsin REMAP Data
Numeric biological criteria can provide objective quantifiable measures of ecosystem health.
Biological indices have been developed for Wisconsin streams using macroinvertebrate and fish
assemblage data (Hilsenhoff 1987, Lyons 1992a, Lyons et al. 1996). Hilsenhoffs Biotic Index
(HBI) uses macroinvertebrate taxa data from samples collected from riffles to assess organic
pollution that manifests itself as reduced dissolved oxygen concentrations (Hilsenhoff 1987).
Alternative or refined macroinvertebrate field or lab protocols may result in the collection of data
that are more discriminating of specific types of stream impairment other than organic pollution.
Refined macroinvertebrate indices would also add additional measures of stream condition to
corroborate physical, chemical, and biotic measures such as fish or diatom indices, and can be
applied to streams where these other biological indices are insensitive. Two commonly used
macroinvertebrate field sample collection protocols (single (riffle) habitat and proportional multi-
habitat sampling), differing laboratory sub-sampling levels (100, 300, 500 fixed-count), and
differing levels of taxonomic identification (family and genus levels, versus lowest practical level)
are also being evaluated as part of the Wisconsin REMAP project. These results will be used to
characterize the method performance (rigor) of various aspects of the WDNR's
macroinvertebrate field and lab protocols. These data will also be used to develop a multi-
metric macroinvertebrate index for western Wsconsin streams, and this process will
subsequently be applied to develop multi-metric macroinvertebrate indices for the entire state.
13
-------
2. Study Area
The geographic extent of the REMAP study area is 7,418,000 acres (20.6% of Wisconsin's total
land area), and encompasses the Western Coulee and Ridges and the Southwest Savanna
ecoregions in western Wisconsin (Figure 1), which are based on the United States Department of
Agriculture - Forest Service's National Hierarchical Framework of Ecological Units (Keys et al.
1995). These ecological units closely approximate Omernik's Level III Driftless Area Ecoregion
(Omernik 1987) and are collectively referred to as the Driftless Area in this report. The Driftless
Area extends into the states of Minnesota, Iowa, and Illinois, and remained unglaciated during the
most recent glacial period, whereas the rest of Wisconsin and most of the upper Midwest was
covered with ice. As a result, the Driftless Area has a "mature" drainage system characterized by
deeply incised valleys and few natural lakes or wetlands relative to the rest of Wisconsin. The
relatively steep topography creates a significant water flow gradient and valley bottoms are close
to the water table, resulting in streams with high baseflow that are dominated by coldwater fish
assemblages. The soils of the Driftless Area are well-drained silty loess over dolomite, limestone,
or sandstone. Land use patterns closely follow spatial differences in slope, with field corn,
soybean, alfalfa, and pastureland situated on the ridge tops and in the valley bottoms, and
hardwood forests dominate the steeply sloped valley sides. The Driftless Area was selected for
the REMAP study area because the amount of topographic relief in this area provides clearly
delineated watersheds, and land use (and presumably stream quality) varies significantly among
watersheds, relative to other areas of the state (Keys et al. 1995).
2.1 Watershed Delineation, Land Use Quantification, and Water Quality Stressor
Gradient Development
We used Geographic Information System (GIS) technology and a digital elevation model to
delineate the watershed area upstream of each random and least disturbed reference stream site.
Wisconsin Initiative for Statewide Cooperation on Landscape Analysis and Data (WISCLAND)
GIS land cover data were used to quantify land use within the study area watersheds. The
WISCLAND data were derived from LANDSAT Thematic Mapper satellite imagery acquired from
fly-overs beginning in August 1991 and ending in May 1993. The on-the-ground resolution of
each WISCLAND digital pixel is 30 square meters. We quantified land use data to develop a land
use disturbance gradient based on the total proportion of agricultural (row crop) acreage found
within each watershed upstream of each stream assessment site. Higher proportions of row crop
acreage within each watershed were equated with greater potential in-stream impacts.
Agricultural land use is the major land "disturbance" and cause of degradation to Wisconsin
streams (Robertson et al. 2006). In particular, row crops such as field corn and soybeans are the
major contributors of sediment, and chemical fertilizers and livestock manure are the major
sources of nitrogen and phosphorus for the study area streams. It is recognized that riparian land
use, buffer width, plant species composition and linear extent, cropland slope, crop rotations, and
numerous geological and meteorological factors strongly influence pollutant delivery to streams,
but the level of effort needed to adequately characterize and quantify these factors were beyond
the scope of this study.
2.2 Stream Resources in Study Area
The REMAP study area has an estimated 3,560 individual (19.5 % of statewide total number) and
8,840 miles (21.1% of Wisconsin's total stream mileage) of perennial streams. Perennial streams
14
-------
in the study area were identified and numbers and miles quantified from a 1:24,000-scale GIS
statewide hydrography layer developed by the WDNR from digitized 7.5 minute U. S. Geological
Survey (USGS) quadrangle maps. The numbers of perennial stream miles by Strahler (1957)
stream order for the REMAP stream population are as follows: order 1 (1,031 mi.), order 2 (2,448
mi.), order 3 (2,282 mi.), and order 4 (1,548 mi.). The REMAP target population of first through
fourth order streams comprise 3,540 individual (19.3% of Wisconsin's total number) and 7,310
miles (17.4% of Wisconsin's total mileage of perennial streams) of streams. In western
Wsconsin, fifth order streams typically transition from wadeable to non-wadeable; therefore, only
first through fourth order streams were included in the target and sample populations.
15
-------
Minimally-Impaired Reference Sites
Random X Sites
SW Savanna Ecoregion
Western Coulee & Ridges Ecoregion
40 0 40
80 Miles
Figure 1. Random and least disturbed reference stream sites in the REMAP study area.
16
-------
3. Stream Population Sampling Design
3.1 Random, Modified-Random, Reference, and Targeted Stream Sites Selection
The EPA's Environmental Monitoring and Assessment Program (EMAP) probabilistic sampling
design was used to randomly select stream sites from the target population (Stevens and Olsen
1999, 2004). The sampling design was weighted so that equal numbers of first through fourth
order streams were included in the sample population. The EPA Office of Research and
Development (ORD) staff in Corvallis, Oregon used an unequal probability random tessellation
stratified (RTS) sampling design described by Stevens and Olsen (1999) and Stevens (1997), and
the WDNR 1:24000 - scale hydrography layer to identify 100 potential sample sites. From these
sample sites, 15 perennial stream sites in each of stream orders one through four (n = 60) were
randomly selected throughout the Driftless Area ecoregion. EPA - ORD also identified an
additional 100 over-sample sites to use as replacements if original sites were rejected. The
sample population for each stream order was weighted to account for the number of stream miles
these populations represented in the target population. These initial weights were calculated
assuming that only the first 100 sites would be evaluated. Because many of the original sites
were rejected and replaced with over-sample sites, weights were adjusted by multiplying the total
stream length calculated for the study area by the original weight, and dividing this number by the
sum of the weights of the final sample population. The WDNR hydrography layer was also used to
generate summary statistics on numbers and miles of streams statewide and in the study area.
Reference Site Selection
Selection of reference stream sites was based on evaluation of watershed land ownership (i.e.,
county, state, or federal lands tended to have less agricultural or urban land) and land use, using
the WISCLAND database. This a priori method used to select candidate watersheds and
reference stream sites was based on guidance developed by EPA and others (Hughes et al.
1986, Gibson et al. 1996). Reconnaissance of each candidate reference site was conducted to
verify that there were no apparent watershed or riparian land use factors significantly degrading
the site, and to do a cursory evaluation of in-stream physical habitat conditions prior to
designating it as a reference site. To develop reference conditions we collected physical,
chemical, and biological data from 22 least disturbed reference stream sites located throughout
the study area (Figure. 1). We sampled a total of four first-order, four second-order, ten third-
order, and four fourth-order least disturbed reference sites during the 2003 field season for the
same parameters as the random and modified-random sites.
Regional Targeted Sampling Sites
Regional WDNR biologists that assess and manage stream resources in the REMAP study area
were asked to select a total of 60 stream sites that were recently sampled and thought to be
representative of the range and modal condition of stream resources in the study area. Fish data
were collected by the biologists from these non-random sites using the same sampling protocols
used in the REMAP study, and these data were then compared to the fish data collected at the
REMAP random sites.
17
-------
3.2 Identification of Sampling Sites
EPA-ORD provided the latitude and longitude coordinates to identify the mid-point of each
randomly selected stream sampling site. Each corresponding modified-random sampling site was
located either upstream or downstream of the randomly selected site, at the nearest "easy"
access point (typically accessed from a roadway or driveway bridge that crossed each stream).
We dropped candidate streams from the sample population if there were intervening tributaries
between the random and modified-random sites that resulted in a Strahler stream order difference
of more than one order between the random and modified-random sites. In this report, random
sites are subsequently referred to as "X" sites and the modified-random sites are referred to as
"B" (Bridge) sites. The B sites were located sufficient distances (typically 10 x the mean stream
wetted width (MSW)) away from road crossings or driveways to avoid hydraulic influences of
bridge abutments, culverts, or other manmade structures on stream physical habitat
characteristics, or that created artificial fish habitat.
3.3 Temporal Sampling Frame
We followed the sampling index periods of the WDNR's standard operating procedures (SOPs)
when collecting stream physical habitat and fish assemblage data, and macroinvertebrate and
water chemistry samples. We collected stream physical habitat data and water chemistry
samples during stream baseflow conditions following spring snowmelt and rain event high-flow
conditions. Fish sampling was done from mid-June through mid-September. The fish sampling
index period avoids spring migratory movements of spawning catostomids, and high numbers of
young-of-the-year salmonids and centrarchids that exceed typical stream carrying capacities, as
well as fall upstream spawning migrations of salmonids and downstream overwintering migrations
of ictalurids and centrarchids. We collected macroinvertebrate samples in the fall of 2003. The
fall index period for benthic macroinvertebrate sampling allows for a greater portion of aquatic
invertebrates to be identified since most taxa are of sufficient size and development for
identification. In-situ water chemistry (instantaneous readings from an electronic water quality
meter) and laboratory-processed grab samples were collected in the spring (June) and again in
the fall (August - September) during baseflow conditions at each X, B, and reference stream site
during the 2003 field season (Appendix A).
18
-------
4. Stream Site Sampling Protocols
4.1 Physical Habitat
In-stream and riparian physical habitat characteristics were measured or visually estimated at 12
transects within each X, B, and reference stream assessment reach. We used WDNR SOPs for
assessing stream physical habitat at all of the study sites. Stream assessment reach lengths for
physical habitat and fish assemblage sampling were based on 35 x MSW. An assessment length
of 35 X MSW generally encompasses three run-riffle-pool sequences, and typically most fish
species found within stream reaches in Wisconsin will be encountered within this length of stream
surveyed (Lyons 1992b). For streams with a MSW less than 2.9m, the minimum assessment
reach was 100m long. At some sites, the reach lengths were slightly lengthened if it resulted in
starting and/or ending the assessment reach at a riffle, which enabled more efficient fish collection
since blocknets were not used. At each habitat assessment transect, measures of water and
sediment depth, bankfull water depth, overhead canopy; and visual estimates of percent substrate
composition were collected at four points equally spaced along the transect line. Bank erosion
was measured and riparian land use and land cover characteristics were visually estimated along
each of the transect lines extending laterally 10 meters into the upland riparian zones. The
distance between bends was measured and divided by the mean stream width. Throughout the
report this ratio is referred to as 'distance between bends' to abbreviate.
4.2 Water Quality and Water Chemistry
We collected 6 in situ (metered) water quality measures at all X, B, and reference stream sites in
the spring and fall during baseflow conditions. A Yellow Springs Instruments Company Model 85
electronic water quality meter was used to collect instantaneous measures of water temperature,
dissolved oxygen concentration, dissolved oxygen percent saturation, and conductivity. An Orion
Quikchek pen was used to measure pH. Both meters were calibrated following the
manufacturer's instructions prior to the start of each field day using air calibration for dissolved
oxygen, and chemical standards for conductivity and pH. To measure water column transparency
we used a transparency tube. This device consists of a 125 cm long x 4.5 cm diameter Plexiglas
cylinder sealed at the bottom with a 4.5 cm diameter Secchi disk. The transparency tube was
filled with stream water while at each assessment site, and the water was drained from the tube
(reducing the water column height) until the Secchi disk became visible when looking down into
the tube through the water column. The height of the water column was measured by reading a
height scale on the side of the tube. For some assessment sites (particularly reference sites), the
transparency of the water was often high enough that the Secchi disk could be seen when the
tube was completely filled with water (122 cm), so that an exact transparency value could not be
measured, since the true value was "off-scale". In these instances, a value of 122 cm was
assigned.
Five water chemistry parameters were measured. We collected one laboratory-analyzed water
chemistry grab sample during baseflow in the spring and fall at each X, B, and reference site.
The parameters analyzed included: total phosphorus, total dissolved phosphorus, total Kjeldahl
nitrogen, ammonia, and nitrate - nitrite nitrogen. Samples were preserved (acidified) and placed
on ice in the field, refrigerated upon return from the field, and submitted to the Wsconsin State
Laboratory of Hygiene within specified holding times for analysis, following the lab SOPs.
19
-------
4.3 Periphyton
To assess diatom assemblages we collected periphyton samples at all X, B, and reference sites.
Samples were collected in August and September of 2003 from cobble or smaller rocks, or in the
absence of mineral substrates, soft sediment samples were collected. At sites with rock
substrate, nine rocks (5cm - 25cm in diameter) were taken from riffle areas. An area 4 cm square
was measured on the upper surface of each of the rocks with a ruler and delineated with a
scalpel. The scalpel was then used to scrape the delineated area on each rock, and the material
adhering to the scalpel and loosened on the rock was washed with de-ionized water into a clean
pan. The same delineated areas were then scrubbed with a clean nylon brush and both the brush
head and brushed area of the rocks were again rinsed into the sample collection pan. The
composite sample was poured from the pan into a 1L bottle and diluted to the smallest volume
possible, either 250ml_, 500ml_, 750ml_, or 1000ml_, and the dilution volume was recorded. The
capped 1L bottle was vigorously shaken to suspend and homogenize the sample. A clean turkey
baster was then used to remove a 40ml_ aliquot from the 1L bottle and transfer it into a 60ml_
bottle. The 40ml_ aliquot was preserved with 1.6ml_ of a 50% glutaraldehyde solution. The
preserved samples were placed on ice in the field and refrigerated upon arrival to the lab. At sites
lacking coarse substrate, a glass Petri dish (49mm inside diameter) was inverted and placed on
streambed fine sediment (sand, clay, silt), and a stainless steel spatula without openings in the
blade was used to trap the fine sediment within the Petri dish. The fine sediment was then
deposited into a clean sample pan. A total of nine fine-sediment samples were collected and
composited in the sample pan. The fine sediment periphyton samples were homogenized, sub-
sampled, and preserved using the sample methods applied to the coarse substrate periphyton
samples.
Although the collection and analysis of periphyton samples from the X, B, and reference sites
were not part of the original study proposal, these samples were collected and are currently being
processed. Upon completion of the analytical results, a manuscript will be submitted to a
scientific journal.
4.4 Benthic Macroinvertebrates
We collected macroinvertebrate samples at all X, B, and reference stream sites in the fall of 2003
using a D-framed net with 500 micron mesh, following WDNR SOPs. One kick sample was
collected from a single riffle (course gravel or cobble substrate) located within the habitat and fish
sampling reach where water velocities created erosional habitat. The net frame was placed at
arm's-length downstream of the sample collector, who using the toe or heel of their boot disturbed
the substrate to a depth of approximately 5 cm. Sampling continued for approximately 3 minutes,
typically at which time there was a fist-sized wad of debris in the net and it was evident that over
100 organisms had been collected. In the absence of coarse substrate, overhanging riparian
vegetation, or grass and leaf "snags" were sampled. The D-frame net was used to sweep and jab
the overhanging vegetation for approximately 5 minutes, or the net was placed downstream of
snags, and the tree branches holding the snags were disturbed until the organic debris was
dislodged and washed into the net. In addition, a 20-jab proportional-habitat sample was
collected along a 100m-long reach within the habitat and fish assessment reach at each X, B,
and reference stream site following protocols described by Barbour et al. (1999). A rangefinder
was used to estimate each 100m-long sampling reach. For both the single riffle and proportional
habitat samples, rocks, twigs, and other coarse debris were removed from the net while making
certain to rinse or remove by hand organisms clinging to the debris and placing them back into the
20
-------
sample. Samples were placed in containers identified with internal and external labels, preserved
with an 80% ethanol solution, and transported to the lab for processing.
We used a 125+ organism fixed-count random sub-sampling method in the lab to process all of
the macroinvertebrate samples, following WDNR SOPs. In the lab, each field sample was evenly
distributed in a gridded pan, and a random number was used to select a grid square. All
organisms within the chosen grid-square were removed, enumerated, and stored for later
taxonomic identification. If fewer than 125 organisms were removed from the initial grid square,
subsequent randomly selected grid squares were picked in their entirety until the target number of
125+ organisms was reached. Guidelines provided by Hilsenhoff (1987) were followed for
determining which taxonomic groups were identified and the level of taxonomic resolution that
was applied to the taxa in the sub-samples. Lifestages and groups of organisms not included in
the enumeration and identification included: adult insects, empty or sealed Trichoptera cases,
Hemiptera, Coleoptera (non-dryopids), Collembola, Mollusca, Annelida, Decapoda, Nematoda,
Nematomorpha, Hydracarina, and Turbellaria.
The taxonomic analyses are still being conducted on the 20-jab samples and only the results of
the riffle samples are included in this report. The single-riffle and 20-jab sample data will be part
of a detailed Performance-Based System study (Miller et al.: in prep.) that follows guidelines
developed by Diamond et al. (1996). The study will evaluate the rigor of macroinvertebrate field
and lab methods to help refine WDNR biological assessment protocols and macroinvertebrate
indices for streams.
4.5 Fish
We used WDNR SOPs to assess fish assemblages by electrofishing each of the X, B, and
reference sites. The majority of the stream sites were sampled by a three-person crew that used
two handheld electrodes powered by a DC generator mounted in a small tow barge, pulled
upstream by one crew member. Each crewmember used a hand-held net to capture fish. In
streams too narrow or shallow to be negotiated by the tow barge, a two-person crew, each with a
hand-held net, used one battery-powered backpack electrofishing unit with one anode to stun fish.
All stream habitats were thoroughly sampled, and an effort was made to capture all fish greater
than 25mm total length. The fish were placed in the tow barge live well or in buckets, and
subsequently identified to species, measured, enumerated, and released. If field identification of
fish specimens were uncertain, specimens were preserved for laboratory identification.
21
-------
5. Data Analytical Methods
5.1 Random, Bridge, and Reference Site Comparisons
A total of 9 physical habitat, 11 water quality and water chemistry, 7 macroinvertebrate, and 8 fish
metrics were analyzed and are presented in this report. Prior to analyses, variables were
"weighted" using an assigned weight that accounted for the total number of stream miles
represented by that site in the target population. Weighted means and Horvitz-Thompson
estimates of standard deviation (Diaz-Ramos et al. 1996) were calculated for X and B sites using
the PSURVEY.ANALYSIS package (v. 2.7) developed by EPA's EMAP program (available at
http://www.epa.gov/nheerl/arm/analysispages/software.htm) for the free software R (R: A
Language and Environment for Statistical Computing, Vienna, Austria; http://www.r-project.org/).
An alpha level of 0.05 was applied to determine the significance of all tests reported.
Bivariate scatterplots were created for each metric to visually compare the X and B site data, and
Pearson correlation coefficients were calculated to identify any significant correlations. Paired t-
tests incorporating the weighted means and standard deviations were used with a Bonferroni
adjustment to determine if significant differences existed for any of the variables collected,
between the X and B sites.
The weighted means and standard deviations were plotted alongside the simple random sample
mean and variance estimates calculated from reference sites for each variable. Two-sample t-
tests assuming unequal variances were performed using the weighted mean and variance values
for the X sites and the simple random sample mean and variance for reference sites. A
Bonferroni adjustment was applied to each group of variables (i.e., habitat, chemistry,
macroinvertebrates, fish) to determine significance of differences at an overall Type I error rate of
0.05.
5.2 Probability Estimates and Evaluation of Population Distributions
Empirical cumulative distribution functions (CDFs) of X and B site data were generated to
estimate the percentage of the X and B stream populations that met reference condition
thresholds for various physical habitat, chemical, and biological metrics. The 25th percentile (for
those metrics where lower values indicated lower environmental quality) or the 75th percentile (for
those metrics where higher values indicated lower environmental quality) of physical habitat,
chemical, and biological measures calculated from the reference sites were used to develop the
reference condition thresholds. The CDFs and percentage estimates were calculated using the
PSURVEY.ANALYSIS package in R. A cumulative distribution function summarizes the overall
distribution of some variable (x-axis) measured at many random sites. The probability that any
variable within the population will be less than or greater than some specified value can be
estimated from the CDF curve (Sokal and Rohlf, 1981). If X and B CDFs provide equivalent data,
then their population estimates should be similar and their confidence intervals should overlap.
Kolmogorov-Smirnov two-sample tests were used to determine whether the X and B probability
distributions differed statistically (Sokal and Rohlf, 1981). CDF plots were also used to compare
the WDNR regional biologist's targeted fish sample data with the REMAP X sites fish sample
data.
22
-------
5.3 Land Use Evaluation
A digital elevation model was used to delineate watershed areas upstream of each X, B, and
reference site. WISCLAND land cover data were used to estimate percent land use types within
each of these watersheds. Pearson correlation coefficients were calculated to evaluate
relationships between stream reach physical habitat, water chemistry, and biological assemblage
attributes, and the percent row crop agriculture, grassland, and forest cover within each
watershed. Stepdown Bonferroni adjustments were applied to determine p-value significance
(Legendre and Legendre, 1998).
23
-------
6. Results
6.1 Sample Size Summary
Of the 100 random sites and 100 additional over-sample sites provided by EPA-ORD, we rejected
a total of 71 candidate sites during their initial reconnaissance. Reasons for dropping sites
included: stream channel was dry at assessment site (38 sites), with the majority of these in
Strahler first order (68%) and second order (24%) streams; Strahler stream order changes
between the random and modified-random assessment sites were greater than 1 stream order
(10 sites); assessment site was in a wetland without a well-defined stream channel (6 sites);
greater than 50% of the assessment site fell within an artificial impoundment or area dammed by
beaver (Castor canadensis) making these sites too deep to wade (5 sites); assessment site was
actually part of Mississippi River backwater channels (5 sites); access to private property was
denied (4 sites); no stream channel was located at the random site coordinates (2 sites); a 3
meter-tall barrier fence prevented access to the assessment site (1 site). Two of the 60 streams
accepted after reconnaissance were subsequently dropped because they dried-up prior to
collecting all of the fish, habitat, or macroinvertebrate data. The final study population consisted
of 58 stream sites. In all analyses, the term "percentage of stream miles" refers specifically to
stream miles that were accessible, had water present, and were wadeable.
Physical habitat data were collected at 57 of the 58 X-B paired sites. One of the 58 sites could
not be sampled at the X site due to the presence of dangerous livestock. Four streams had their
X and B sites combined for habitat sampling, because upon delineating the assessment site it
was found that the X and B sampling sites overlapped.
Water chemistry data were collected at the X and B sites from 56 of the 58 streams. Two of the
58 streams had their X and B sites combined when collecting water chemistry data. Some
sampling sites did not have water quality or chemistry data collected from either the spring or fall
sampling period. Reasons included: suspect meter readings (significantly outside the range of
values routinely encountered, or a meter was not meeting the calibration standard while in the
field); grab samples were not collected because recent rainfall resulted in non-baseflow conditions
at the site; or as with conductivity, no spring data were collected due to a meter malfunction. Only
spring water quality and chemistry data were used for analyses, with the exception of conductivity
measures, since the spring sample size was greater.
Macroinvertebrates were sampled from 57 of the 58 random sites, and fish were surveyed at 55 of
the 58 sites. Fish were not sampled at two sites with dangerous livestock, and inclement weather
prevented sampling at another site. Four streams had their X and B sites combined for fish
assessments, because upon delineating the assessment sites it was found that the X and B
assessment reaches overlapped. Only those stream sites where a minimum of 25 individual fish
were captured were used in computing fish IBI scores.
24
-------
6.2 Watershed Land Use Identification and Quantification - Relationships
Among Agricultural Land Use and Physical, Chemical, and Biological
Measures
The primary land use in the study area, determined from the 1991 - 1993 WISCLAND
(LANDSAT) data, was agriculture (40%) followed by forest cover (37%), grassland (13%), wetland
(5%), open water (2%), urban land (1%), and miscellaneous other (2%). Watershed area
upstream from the X sites ranged from 102 acres to 81,881 acres, with an average size of 7,268
acres. For the reference sites, watershed sizes ranged from 123 acres to 34,450 acres, with an
average watershed size of 5,532 acres.
The total percentage of land area identified as row cropland within the X site watersheds ranged
from zero to 82%, with an average of 19%. The total land area identified as row cropland in the
reference site watersheds ranged from zero to 41%, with an average of 14.5%. Forest cover in
the X site watersheds ranged from zero to 86% with an average of 43%, and in the reference site
watersheds forest cover ranged from 15% to 99%, with an average of 41%. Pearson correlations
were calculated to evaluate whether percent agricultural land, percent grassland, or percent forest
cover within watersheds had detectable relationships with various riparian and in-stream physical
habitat measures. Data collected at both the random X sites and the minimally disturbed
reference sites were used in these analyses. No significant correlations were observed among
the in-stream physical habitat measures and the surrounding watershed land use (Table 1).
Stepdown Bonferroni adjusted p-values equal 1.00 for all correlations in Table 1.
Table 1. Pearson correlations between watershed land use and in-stream physical
habitat measures (N = 78).
Habitat Measures
% Sand, silt, and
clay sediments
Bank
erosion
Depth of
fine substrate
Sinuosity
Distance
between bends
Riffle
length
Pool
length
Riparian
buffer
Width: Depth
ratio
rs =
p-value =
rs =
p-value =
rs =
p-value =
rs =
p-value =
rs =
p-value =
rs =
p-value =
rs =
p-value =
rs =
p-value =
rs =
p-value =
% Agriculture
-0.163
0.153
-0.069
0.550
-0.001
0.993
0.049
0.667
0.132
0.250
-0.042
0.716
0.020
0.866
-0.185
0.106
-0.139
0.224
% Grassland
0.138
0.230
0.114
0.318
0.087
0.446
0.093
0.420
0.063
0.581
0.058
0.615
-0.022
0.851
0.062
0.590
-0.080
0.486
% Forest
0.125
0.276
0.041
0.721
-0.068
0.552
-0.052
0.649
-0.178
0.120
0.049
0.670
-0.001
0.993
0.178
0.120
0.173
0.129
25
-------
Pearson correlations and associated p-values were calculated to evaluate relationships between
watershed land use and stream water quality and water chemistry measures, and these original p-
values are presented in Table 2. In addition, Stepdown Bonferroni adjustments were applied and
the resulting adjusted p-values are provided in parentheses. Nitrate-nitrite and conductivity were
found to be significantly correlated with percent agriculture and forest cover.
Table 2. Pearson correlations between watershed land use and water quality and water
chemistry measures. P values in bold are significant when a Stepdown Bonferroni
adjustment is applied.
Water Quality and Chemistry
N-Kjeldahl
NH3
NO3NO2
Total
Dissolved
Phosphorus
Total
Phosphorus
Dissolved
02
%O2
Saturation
Temperature
Transparency
PH
Conductivity
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
% Agriculture
0.109
0.332 (1.00)
80
0.118
0.298 (1.00)
80
0.606
<.001 (0.003)
80
0.113
0.317 (1.00)
80
-0.017
0.878 (1.00)
80
-0.064
0.572 (1.00)
80
0.099
0.380 (1.00)
80
0.235
0.036 (0.999)
80
-0.069
0.544 (1.00)
80
0.272
0.018 (0.513)
76
0.544
<.001 (0.003)
75
% Grassland
-0.058
0.610 (1.00)
80
-0.015
0.892 (1.00)
80
-0.149
0.188 (1.00)
80
-0.143
0.207 (1.00)
80
-0.0417
0.714 (1.00)
80
0.228
0.042 (1.000)
80
0.113
0.319 (1.00)
80
-0.170
0.132 (1.00)
80
0.194
0.085 (1.00)
80
-0.051
0.661 (1.00)
76
-0.093
0.429 (1.00)
75
% Forest
-0.152
0.177 (1.00)
80
-0.171
0.130 (1.00)
80
-0.586
<.001 (0.003)
80
-0.169
0.135 (1.00)
80
-0.037
0.742 (1.00)
80
0.047
0.678 (1.00)
80
-0.101
0.373 (1.00)
80
-0.207
0.066 (1.00)
80
0.031
0.788 (1.00)
80
-0.217
0.060 (1.00)
76
-0.543
<.001 (0.003)
75
Stepdown Bonferroni adjusted p-values are in parentheses.
26
-------
Pearson correlations were also calculated to evaluate the sensitivity of the macroinvertebrate and
fish assemblage metrics to specific environmental stress factors such as high percentages of
cropland in the watersheds, in-stream or riparian physical habitat degradation, or chemical
pollutants. The percentage of EPT macroinvertebrate genera present, HBI score, percentage of
'tolerant' fish individuals present, and fish IBI score were compared with 9 in-stream habitat
measures, 3 watershed-scale land use measures, and 11 water quality and water chemistry
measures (Tables 3, 4). Macroinvertebrate measures showed stronger correlations with physical
habitat data than the fish measures, based on the Pearson correlations presented in Table 3, and
statistically significant correlations were observed between macroinvertebrate measures and 7 of
the 11 water quality and water chemistry measures when Stepdown Bonferroni adjustments were
applied (Table 4).
27
-------
Table 3. Pearson correlations between in-stream physical habitat, watershed land use,
and biological measures.
Habitat Parameters
% Sand, silt,
and clay
sediments
Bank
erosion
Depth of
fine
substrate
Sinuosity
Distance
between
bends
Riffle
length
Pool
length
Riparian
buffer
Width: Depth
ratio
%
Agriculture
%
Grassland
% Forest
rs =
p-value* =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
% EPT
Genera
-0.247
0.030 (1.00)
77
-0.360
0.001 (0.062)
77
-0.126
0.275 (1.00)
77
-0.227
0.048 (1.00)
77
-0.284
0.012 (0.533)
77
0.232
0.043 (1.00)
76
0.052
0.656 (1.00)
77
0.271
0.017 (0.688)
77
0.191
0.096 (1.00)
77
-0.259
0.020 (0.796)
80
0.209
0.063 (1.00)
80
0.232
0.038 (1.00)
80
HBI Score
0.177
0.127 (1.00)
76
0.261
0.023 (0.866)
76
0.281
0.014 (0.592)
76
0.112
0.336 (1.00)
76
0.144
0.214 (1.00)
76
-0.332
0.003 (0.150)
76
0.012
0.915 (1.00)
76
-0.168
0.146 (1.00)
76
-0.249
0.030 (1.00)
76
0.283
0.012 (0.533)
79
-0.274
0.015 (0.599)
79
-0.254
0.024 (0.881)
79
% Tolerant Fish
Individuals
0.055
0.676 (1.00)
61
0.372
0.003 (0.150)
61
0.036
0.768 (1.00)
61
0.281
0.028 (1.00)
61
0.130
0.319 (1.00)
61
-0.181
0.163 (1.00)
61
0.007
0.958 (1.00)
61
-0.273
0.033 (1.00)
61
-0.241
0.061 (1.00)
61
0.071
0.589 (1.00)
61
0.012
0.928 (1.00)
61
-0.062
0.635 (1.00)
61
IBI Score
-0.062
0.640 (1.00)
60
-0.198
0.130 (1.00)
60
-0.104
0.430 (1.00)
60
-0.256
0.049 (1.00)
60
-0.119
0.365 (1.00)
60
0.204
0.117 (1.00)
60
0.016
0.902 (1.00)
60
0.328
0.011 (0.477)
60
0.120
0.362 (1.00)
60
-0.265
0.041 (1.00)
60
0.198
0.129 (1.00)
60
0.205
0.116 (1.00)
60
*Stepdown Bonferroni adjusted p-values are in parentheses.
28
-------
Table 4. Pearson correlations between water quality and water chemistry measures, and
biological measures. P-values in bold are significant when a Stepdown Bonferroni
adjustment is applied.
Water Quality and
Chemistry
N-Kjeldahl
NH3
NO3NO2
Total
Dissolved P
Total P
Dissolved
02
%O2
Saturation
Temperature
Transparency
PH
Conductivity
rs =
p-value* =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
rs =
p-value =
N =
% EPT
Genera
-0.491
<.001 (0.004)
79
-0.429
<.001 (0.004)
79
-0.011
0.921 (1.00)
79
-0.346
0.002 (0.058)
79
-0.361
<.001 (0.004)
78
0.312
0.005 (0.143)
79
0.150
0.187 (1.00)
79
-0.350
0.002 (0.007)
79
0.435
<.001 (0.004)
79
-0.218
0.060 (1.00)
75
-0.138
0.243 (1.00)
74
HBI Score
0.564
<.001 (0.004)
78
0.537
<.001 (0.004)
78
0.101
0.377 (1.00)
78
0.418
<.001 (0.004)
78
0.388
0.005 (0.143)
78
-0.487
<.001 (0.004)
78
-0.397
0.003 (0.099)
78
0.215
0.058 (1.00)
78
-0.413
0.002 (0.007)
78
-0.036
0.763 (1.00)
74
-0.018
0.882 (1.00)
73
% Tolerant
Fish
Individuals
0.263
0.041 (0.851)
61
0.232
0.073 (1.00)
61
-0.259
0.044 (0.874)
61
-0.059
0.651 (1.00)
61
0.055
0.674 (1.00)
61
-0.357
0.005 (0.143)
61
-0.216
0.094 (1.00)
61
0.280
0.030 (0.669)
61
-0.267
0.038 (0.834)
61
0.091
0.491 (1.00)
60
0.157
0.235 (1.00)
59
IBI Score
-0.374
0.003 (0.099)
60
-0.348
0.007 (0.176)
60
-0.001
0.996 (1.00)
60
-0.241
0.064 (1.00)
60
-0.318
0.013 (0.335)
60
0.332
0.010 (0.250)
60
0.172
0.190 (1.00)
60
-0.316
0.014 (0.335)
60
0.373
0.003 (0.099)
60
-1.071
0.594 (1.00)
59
-0.159
0.232 (1.00)
58
*Stepdown Bonferroni adjusted p-values are in parentheses.
29
-------
6.3 Random, Modified-Random, and Reference Site Comparisons
Stream Physical Habitat Measures
Each of the 9 physical habitat measures collected at the X sites was compared with its associated
B site value using Pearson correlations and bivariate scatterplots (Figure 2). Large correlation
coefficients (r) and significant p-values were observed for 8 of the 9 habitat variables. The only
physical habitat variable that did not show a significant correlation between the X and B sites was
mean total sum of pool habitat length.
2.5
r = 0.869
p< 0.001
0 2 4 6 8 10
X-site riparian buffer width (m)
0 20 40 60 80 100
X-site pool length (m)
0 0.5 1 1.5 2
X-site sinuosity
2.5
40
30
20
|S
s
.2 10-
V
m
0 0.2 0.4 0.6 0.8 1
X-site bank erosion (m)
r = 0.587
p< 0.001
0 10 20 30 40
X-site width : depth ratio
0 10 20 30 40
X-site riffle length (m)
150
0 25 50 75 100 125 150
X-site distance between bends (m)
0 0.3 0.6 0.9
X-site fine sediment depth (m)
0 20 40 60 80 100
X-site % sand/silt/clay
Figure 2. Bivariate scatterplots comparing physical habitat measurements collected at
the X and associated B sites.
30
-------
Weighted paired t-tests were used to determine if significant differences existed in physical habitat
data between the X and B sites. The weighted paired t-tests did not produce statistically
significant p-values for any of the 9 physical habitat measures, indicating that there were no
significant differences between the X and B sites for these variables (Table 5).
Table 5. Weighted paired t-test comparisons of stream physical habitat measures from X
and B stream assessment sites.
X vs B Site Physical Habitat Measures
(n = 57)
Mean Riparian Buffer Width (m)
Mean Length of Pools (m)
Sinuosity
Mean Width of Bank Erosion (m)
Width : Depth Ratio
Mean Length of Riffles (m)
Mean Distance Between Bends (m)
Mean Depth of Fines (m)
Percent Sand, Silt, and/or Clay Sediments
t-value*
1.68
-1.61
1.84
0.36
-1.64
1.74
1.85
-0.52
-0.05
p-value
0.0981
0.1125
0.0712
0.7218
0.1060
0.0878
0.0697
0.6081
0.9609
*Degrees of freedom = 56.
31
-------
Water Chemistry and Water Quality Measures
Each of the 11 water quality and water chemistry measures collected at the X sites was compared
with its associated B site value using Pearson correlations and bivariate scatterplots (Figure 3).
1.6
D)
1.2 -
0.8
r = 0.735
p < 0.001
0.8
D)
cT ฐ-6-l
"ro
S. 0.4
0
0.2
r = 0.860
p < 0.001
9
8.5
8
7.5
7
6.5
0 0.4 0.8 1.2 1.6
X-site N-Kjeldahl (mg/L)
0.2 0.4 0.6 0.8 1
X-site total P (mg/L)
6 6.5 7 7.5 8 8.5
X-site pH
14
12
10
8
6
4
2
0
r = 0.804
p< 0.001
125
-Si 100
>.
S 75
50
^ 25
r = 0.881
p< 0.001
0.1 0.2 0.3
X-site NH3 (mg/L)
0 2 4 6 8 10 12 14
X-site dissolved O2 (mg/L)
0 25 50 75 100 125
X-site transparency (cm)
25
30
y.
25
20
15
10
& 5]
V
co
r = 0.901
p< 0.001
5 10 15 20 25
X-site N03-N02 (mg/L)
26 52 78 104
X-site % O2 saturation
130
0 5 10 15 20 25 30
X-site water temperature (C)
2000
1500
| 1000
S 500 -
r = 0.950
p< 0.001
0.2 0.3 0.4
X-site total dissolved P (mg/L)
0 500 1000 1500 2000
X-site conductivity (uS)
Figure 3. Bivariate scatterplots comparing water quality and water chemistry
measurements collected from the X and associated B sites.
32
-------
Large correlation coefficients (r) with significant p-values were observed for all 11 water quality
and chemistry variables (Figure 3), showing a strong linear relationship between X site and B site
data. The results of weighted paired t-tests evaluating water quality and water chemistry
measures show no significant differences between the X and B site data for all 11 variables
(Table 6).
Table 6. Weighted paired t-test comparisons of water quality and water chemistry
measures from the X and associated B assessment sites.
X vs B Site Water Chemistry Measures
(n = 57)
Total Phosphorus (mg/L)
Total Dissolved Phosphorus (mg/L)
NH3(mg/L)
NO3-NO2 (mg/L)
N-Kjeldahl (mg/L)
Dissolved Oxygen (mg/L)
Percent Dissolved Oxygen Saturation
pH (df=55)
Conductivity (uS/cm) (df=52)
Water Temperature (C)
Transparency (cm)
t-value*
1.69
2.88
-0.08
-2.28
1.09
1.04
1.06
-1.89
-1.26
0.21
-1.53
p-value
0.0965
0.0056 (0.061 7)t
0.9344
0.0264 (0.2904)
0.2809
0.3010
0.2946
0.0644
0.2140
0.8377
0.1319
*Degrees of freedom = 56, unless otherwise noted.
fValue in parentheses is the Bonferroni-adjusted p-value.
33
-------
Macro!nvertebrate Metrics
Figure 4 shows the relationships between the 7 macroinvertebrate metrics collected at each X
and associated B site. Pearson correlation coefficients and bivariate scatterplots show significant
p-values and correlations (r) between X and B site data for all 7 macroinvertebrate metrics,
indicating strong linear relationships between the X and B sites.
40 -
o 30 -
20 -
ซ 10
CD
r = 0.733
p< 0.001
10 20 30 40
X-site species richness
r = 0.782
p < 0.001
2 4 6 8 10
X-site % EPT genera
01 2345
X-site Shannons Diversity Index
10
6 -
2 -
r = 0.805
p< 0.001
246
X-site HBI score
10
10 20 30 40
X-site % 'shredders'
10 20 30 40 50 60 70
X-site % 'scrapers'
0 20 40 60 80 100
X-site % EPT individuals
Figure 4. Bivariate scatterplots comparing macroinvertebrate metrics collected at the X
and associated B sites.
34
-------
Weighted paired t-tests were used to compare macroinvertebrate samples collected at the X and
associated B sites. The results show no significant differences between these paired sites for any
of the 7 metrics (Table 7).
Table 7. Weighted paired t-test comparisons of macroinvertebrate metrics and indices
collected from X and associated B assessment sites.
X vs B Site Macroinvertebrate Metrics
(n = 56)
Species richness
Percent EPT genera present
Shannon's Diversity Index score
Hilsenhoff s Biotic Index score
Percent Shredders
Percent Scrapers
Percent EPT individuals present
t-value*
0.42
0.02
-0.46
-0.69
-1.23
0.33
0.12
p-value
0.6739
0.9865
0.6473
0.4962
0.2244
0.7406
0.9056
*Degrees of freedom = 55.
35
-------
Fish Assemblage Metrics
Bivariate scatterplots and Pearson correlation coefficients show that significant correlations exist
between X and B site data for all 8 fish metrics analyzed (Figure 5).
20 i
15
c
tft 5 -
m
cS
3
ฅ
CO
5 10 15
X-site species richness
20
012345
X-site # 'intolerant' species
20 40 60 80
X-site % brook trout
2000
1500
1000
*
a)
500
100
ฅ
m
100
40
20
eg
S
ฅ
m
500 1000 1500 2000
X-site # fish captured
20 40 60 80
X-site % top carnivores
100
40
20 i
20 40 60 80
X-site IBI score
0 20 40 60 80 100
X-site % tolerant' individuals
20 40 60 80
X-site % stenotherms
100
Figure 5. Bivariate scatterplots comparing fish metrics collected at the X and associated
B sites.
36
-------
Weighted paired t-tests were used to determine if significant differences existed between the X
and B sites. Of the 58 streams sampled, 42 streams had sufficient numbers of fish at both the X
and associated B sites with which to make paired-site t-test comparisons. The results of the
weighted paired t-tests show no significant differences between X and B sites (Table 8).
Table 8. Weighted paired t-test comparisons of fish metrics collected from X and B
assessment sites.
X vs B Site Fish Community Metrics
(n = 42)
Species Richness
Number of Intolerant Species
Percent of Salmonid Individuals that are Brook Trout
Percent of Individual that are Top Carnivore Species
Coldwater IBI Score
Percent of Individuals that are Tolerant' Species
Percent of Individuals that are Stenothermal Cool / Coldwater Species
t-value*
-2.17
0.06
1.08
-1.58
-0.77
-0.34
0.20
p-value
0.0358
(0.2509)f
0.9553
0.2883
0.1219
0.4471
0.7368
0.8406
*Degrees of freedom = 41.
|Value in parentheses is the Bonferroni-adjusted p-value.
37
-------
6.4 Comparison of Reference Sites with Random and Modified-Random
Assessment Sites
Stream Physical Habitat Measures
Bar charts were used to visually compare data for 9 physical habitat measures collected at the X,
B, and reference sites (Figure 6).
0.25
0.20
0.15
8-0.10
0.05
0.00
0)
m
0)
100
80
60
40
20
15
10
Ftef Bridge X-Site
Ref Bridge X-Site
Ftef Bridge X-Site
100
>, 80
_CD
O
i 60
OT
CD
OT
20
40
30
i
c
0)
20
ฃ
ro 10
0)
Ref Bridge X-Site
Ref Bridge X-Site
1.0
3. 0.8
c
g
'ง 0.6
0)
ro 0.4
_a
I 0.2
0.0
Ref Bridge X-Site
2.5
2.0
.? 1.5
5 1.0
0.5
0.0
30
a) 20
c
0)
o
o
Q.
CD
0)
10
i I T
Ref Bridge X-Site
Ref Bridge X-Site
100
CD
ce
ฃ 60
Q.
0)
? 40
20
0
Ref Bridge X-Site
Figure 6. Bar charts of weighted means and standard deviations (SD) for X and B sites
and simple random sample mean and SD for reference sites for physical habitat
measures.
38
-------
Two-sample t-tests were used to compare the 9 physical habitat measures collected at the X sites
with data collected at the least disturbed reference sites. The mean width of stream bank erosion
and the mean width of riparian buffer were the only measures that showed a significant difference
between the X sites and the reference sites (Table 9).
Table 9. Two sample t-test comparisons for physical habitat measures collected at the X
and reference sites. Tests are based on the difference (X site mean - reference mean)
and use weighted values for X sites.
X vs Reference Site Habitat Measures
Mean Depth of Fines (m)
Mean Distance Between Bends (m)
Mean Riparian Buffer Width (m)
Percent Sand, Silt, and/or Clay Sediments
Mean Length of Riffles (m) (df=75)
Mean Width of Bank Erosion (m)
Sinuosity
Mean Length of Pools (m) (df=75)
Width: Depth Ratio
t-value
0.631
0.287
-4.869
1.855
-0.504
3.570
0.742
-1.913
-1.575
p-value*
0.5300
0.7746
<0.0001
0.0675
0.6159
0.0006
0.4605
0.0596
0.1195
*P-values in bold are significant (p < 0.006) when a Bonferroni correction is applied for an overall Type I error rate of
0.05. Degrees of freedom = 76, unless otherwise noted.
39
-------
Water Quality and Water Chemistry Measures
Bar charts were used to visually compare data for 11 water chemistry and water quality measures
among the X, B, and reference sites (Figure 7).
1.0
0.8
ง 0.6
ซ 0.4
0.2
0.0
0.15
0.12
o.o9
0.06
0.03
0.00
10
Ftef Bridge X-Site
Ftef Bridge X-Site
2
0
0.30
^IT
|> 0.24
CL
o 0.18
| 0.12
Q
S 0.06
H
0.00
Ftef Bridge X-Site
i i
0.5
0.4
2~
t 0.3
CL
3 ฐ-2
o
H
0.1
0.0
12
o 4
100
c 80
g
I 60
CO
CO
S *
s?
20
25
O
Ftef Bridge X-Site
20
S
D
2 15
CL
| 10
| 5
0
Ftef Bridge X-Site
Ftef Bridge X-Site
Ftef Bridge X-Site
150
100
50
Ftef Bridge X-Site
10
4
2
0
1000
1" 800
o
5)
5- 600
>,
I 400
O 200
Ftef Bridge X-Site
Ftef Bridge X-Site
Ftef Bridge X-Site
Figure 7. Bar charts of means and standard deviations for water quality and water
chemistry measures collected at the X, B, and reference stream sites. Values for X and B
sites are weighted.
40
-------
Two-sample t-tests were used to compare the 6 in-situ water quality and 5 lab-analyzed grab-
sample water chemistry measures collected at the X and reference sites. Eight of these 11
measures were found to be significantly different at the X sites versus the reference sites (Table
10).
Table 10. Two-sample t-test comparisons of water quality and water chemistry values
between the X and reference stream sites. Tests are based on the difference (X site
mean - reference mean) and use weighted values for X sites.
X vs Reference Site Water Chemistry Measures
Kjeldahl-N (mg/L)
Total Phosphorus (mg/L)
Transparency (cm)
NH3 (mg/L)
Dissolved Oxygen (mg/L)
pH (df = 74)
NO3-NO2 (mg/L)
Percent Dissolved Oxygen Saturation
Conductivity (uS/cm) (df = 73)
Total Dissolved Phosphorus (mg/L)
Water Temperature (C)
t-value
5.410
4.376
-5.243
5.137
-4.850
0.953
1.100
-2.910
0.122
4.470
3.745
p-value*
<0.0001
<0.0001
<0.0001
<0.0001
<0.0001
0.3436
0.2748
0.0047
0.9035
<0.0001
0.0003
*P-values in bold are significant (p < 0.005) when a Bonferroni correction is applied for an overall Type I error rate of
0.05. Degrees of freedom = 78, unless otherwise noted.
41
-------
Macro!nvertebrate Metrics
Bar charts were used to visually compare data among X, B, and reference sites for
macroinvertebrate metrics (Figure 8).
40
30
20
10
70
60
i 50
I 40
ฃ 30
HI
s? 20
10
0
30
20
10
2 3
Ref Bridge X-Site
Ref Bridge X-Site
20
15
10
Ref Bridge X-Site
Ref Bridge X-Site
0
100
80
60
40
20
0
Ref Bridge X-Site
Ref Bridge X-Site
Ref Bridge X-Site
Figure 8. Bar charts of mean and standard deviation of macroinvertebrate metrics
collected at the X, B, and reference stream sites. Values for X and B sites are weighted.
Two sample t-tests comparing 7 macroinvertebrate metrics for data collected at the X and
reference sites show significant differences in HBI scores, and the percentage of genera and
individuals in the sample that are Ephemeroptera, Plecoptera and Trichoptera (EPT) taxa. The
lower HBI scores and the higher percentages of EPT genera or EPT individuals found at the
reference sites relative to the X sites indicate that the reference streams have greater biological
quality and a greater percentage of environmentally sensitive taxa than the X stream sites. No
significant differences were detected between the X and reference stream populations for species
richness, Shannon's Diversity Index, the percent 'shredders', and the percent 'scrapers' present
(Table 11).
42
-------
Table 11. Two sample t-test comparisons of the X and reference site macroinvertebrate
metrics. Tests are based on the difference (X site mean - reference mean) and use
weighted values for X sites.
X vs Reference Site Macroinvertebrate Metrics
Species Richness
Percent Scrapers
Shannon's Diversity Index Score
Hilsenhoffs Biotic Index Score (HBI)
Percent Shredders
Percent EPT Individuals
Percent EPT Genera
t-value
1.900
0.091
0.691
5.307
-1.689
-5.439
-14.772
p-value*
0.0611
0.9279
0.4915
<0.0001
0.0953
<0.0001
<0.0001
*P-values in bold are significant (p < 0.007) when a Bonferroni correction is applied for an overall Type I error rate of
0.05. Degrees of freedom = 77.
43
-------
Fish Metrics
Bar charts were used to visually compare distributions of fish metrics among X, B, and reference
sites (Figure 9).
'o
0)
Q.
OT 3
2
o> 2
o
0
100
g so
'>
"g 60
2 40
"o
I
^o ^U
0
100
M 80
o
| 60
CD
O
& 40
H
S? 20
0
Ref Bridge X-Site
Ref Bridge X-Site
150
E 100
OT
50
Ref Bridge X-Site
125
100
75
m 50
25
100
2 80
o
o
OT
5 60
ro 40
2
O 20
15
10
Ref Bridge X-Site
Ref Bridge X-Site
Ref Bridge X-Site
Ref Bridge X-Site
Figure 9. Bar charts of mean and standard deviation of fish metrics collected at the X, B,
and reference stream sites. Values for X and B sites are weighted.
44
-------
Fish IBI scores were significantly different between the X and reference sites, when compared
using two-sample t-tests. The mean IBI score for the X sites population was 34 (narrative rating
of "Fair"), versus 70 (narrative rating of "Good") for the reference condition. In addition, there
were significant differences found in species richness, percentage of individuals captured that are
considered "tolerant", percentage of individuals captured that are top carnivores, and the
percentage of individuals that are cool/coldwater stenotherms (as defined by Lyons 1996). There
were no significant differences observed between the X stream population and the reference
condition for the number of intolerant species present or the percentage of salmonids captured
that were brook trout. (Table 12)
Table 12. Two-sample t-test comparisons of the X and reference site fish metrics. Test is
based on the difference (X site mean - reference mean).
X vs Reference Site Fish Metrics
Number of Intolerant Species (df=59)
Percent of Individuals that are Stenothermal Cool or Coldwater
Species
Coldwater IBI score
Percent of Individuals that are Tolerant Species
Percent of Salmonid Individuals that are Brook Trout (df=59)
Species Richness (df=59)
Percent of Individuals that are Top Carnivore Species
t-value
-2.319
-7.427
-7.021
6.165
-2.087
3.183
-4.632
p-value*
0.0239
<0.0001
<0.0001
<0.0001
0.0412
0.0023
<0.0001
*P-values in bold are significant (p < 0.007) when a Bonferroni correction is applied for an overall Type I error rate of
0.05. Degrees of freedom = 58, unless otherwise noted.
45
-------
6.5 Probability Estimates and Evaluation of Population Distributions
Physical Habitat Cumulative Distributions
Empirical CDFs were plotted to further compare differences between the X and B site data, and
to estimate the percentage of the X and B sample populations that met reference condition
threshold values. For each physical habitat measure, either the 25th or 75th percentile calculated
from the least disturbed reference sites population was used as a reference condition threshold
(Table 13). The upper 75th percentiles calculated from least disturbed reference sites were
used as the thresholds for the percentage of sand, silt, and clay substrates, the mean depth of
fine sediments, the width:depth ratio, the mean distance between bends, and the mean erosion
width. The 25th percentile was used as the threshold for mean riparian buffer width.
The X and B site cumulative distribution functions and their confidence intervals overlap for all of
the physical habitat measures analyzed (Figure 10), indicating the population estimates of these
measures are not significantly different between the random and modified-random sites. The
results of Kolmogorov-Smirnov tests (Dmax values) also found no significant differences (all Dmax
p-values > 0.05).
Table 13. Physical habitat reference condition thresholds.
Habitat Measures
% sand, silt, and clay sediments
mean depth of fine sediment (m)
width : depth ratio
mean distance between bends (m)
mean riparian buffer width (m)
mean bank erosion width ( m)
Reference
Condition
Threshold
66.8%
0.122
18.4
42
9.5
0.104
Impairment Criteria
> 66.8% sand, silt and clay
> 0.122m mean depth of fines
> 18.4 width : depth ratio
> 42m distance between bends
< 9.5m mean buffer width
> 0.104m mean bank erosion
46
-------
100
100
100
100
20 40 60 80
% Sand, Silt, and Clay Sediment
0.2 0.4 0.6 0.8
Mean Depth of Fines (m)
1.0
100
5 10 15 20 25
Width:Depth Ratio
0 20 40 60 80 100 120 140
Mean Distance Between Bends (m)
100
01 2345678
Mean Riparian Buffer Width(m)
9 10
0.0 0.2 0.4 0.6 0.8
Mean Bank Erosion Width (m)
1.0
X site CDF ... X site 95% C.I. _B site CDF . B site 95% C.I.
Figure 10. Cumulative distribution function curves of physical habitat measures
collected at the X and B sites. The stippled lines represent 95% confidence intervals
around the distribution plots. The vertical lines represent the reference condition
threshold values.
47
-------
Reference condition threshold values were applied to the CDF data to approximate the percent of
the stream population (miles) that did not meet these criteria and therefore received a 'poor' rating
(Figure 11). Overlapping error bars show no significant differences between the X site population
and the B site population for all physical habitat measures presented in Figure 11.
o
o
.
100
90
80
70
60
50
40
30
20
10
D B sites
nX sites
\
% sand,
silt, clay
depth of
fines (m)
distance
between
bends (m)
erosion
width (m)
riparian
buffer (m)
width :
depth ratio
Figure 11. The percentages of stream miles not meeting the reference condition
threshold values for physical habitat measures. The error bars equal 1 standard error.
48
-------
Water Quality and Water Chemistry Cumulative Distributions
Cumulative distribution functions were plotted for 10 water quality and water chemistry measures.
Either the 25th or 75th percentile of each water quality and water chemistry measure calculated
from the least disturbed reference sites were used to set reference condition thresholds (Table
14). The 75th percentile calculated from the reference sites was used as the threshold for
Kjeldahl-N, NH3, NO3-NO2, total dissolved P, and total P. The 25th percentile was used as the
threshold for the percentage dissolved oxygen saturation, dissolved oxygen concentration, and
water clarity.
Table 14. Water chemistry and water quality reference condition thresholds.
Water Chemistry & Quality
Measures
% Oxygen Saturation
Transparency (cm)
Dissolved Oxygen (mg/L)
Kjeldahl-N (mg/L)
NH3(mg/L)
NO3-NO2 (mg/L)
Total Dissolved P (mg/L)
Total P (mg/L)
Reference Condition
Threshold
73.3%
122
7.6
0.15
0.028
2.63
0.04
0.07
Impairment Criteria
< 73.3% dissolved oxygen
saturation
< 122cm water transparency
< 7.6 mg/L dissolved oxygen
cone.
> 0.15 mg/L concentration
> 0.028 mg/L concentration
> 2.63 mg/L concentration
> 0.04 mg/L concentration
> 0.07 mg/L concentration
49
-------
100
100
246 8 10 12
Dissolved O2 (mg/L)
0 20 40 60 80 100 120
Transparency (cm)
_X site CDF ...X site 95% C.I.
ro
"w
LU
Q.
O
CL
20
50 70 90 110
% Oxygen Saturation
130
B site CDF
B site 95% C.I.
Figure 12. Cumulative distribution function curves of water quality measures collected in
situ at the X and B sites. The stippled lines represent 95% confidence intervals around
the distribution functions. The vertical lines represent the reference condition threshold
values.
50
-------
100
100
0.0 0.2 0.4 0.6 0.8
Total P (mg/L)
1.0
0 2 4 6 8 10 12 14 16 18 20
N03-N02 (mg/L)
0.00 0.05 0.10 0.15 0.20 0.25 0.30
NH3 (mg/L)
_X site CDF ... Xsite 95% C.I.
0 0.1 0.2 0.3
Total Dissolved P (mg/L)
100 -,
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Kjeldahl-N (mg/L)
B site CDF .. B site 95% C.I.
Figure 13. Cumulative distribution function curves of laboratory analyzed water
chemistry measures collected at the X and B sites. The stippled lines represent 95%
confidence intervals around the distribution plots. The vertical lines represent the
reference condition threshold values.
51
-------
All 10 water quality and water chemistry measurements have overlapping X and B site cumulative
distribution function curves (Figures 12, 13), showing no significant differences between the X and
B sampling sites. The results of Kolmogorov-Smirnov tests (Dmax values) also found no significant
differences (all Dmax p-values > 0.05). Reference condition threshold values were applied to the
CDF data and estimates of the percent of stream miles not meeting these criteria were
determined (Figure 14). The overlapping error bars show similarities between the X site
population and the B site population for the 8 water quality and water chemistry measures
analyzed in Figure 14.
o
o
CL
T3
CD
"Jo
E
"w
LU
100
90 -
80 -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
*
I
1
^
D B site
DXsite
T
1
% O2
Saturation
Clarity
(cm)
Diss. O2
(mg/L)
Kjeldahl-N
(mg/L)
NH3
(mg/L)
N03-N02
(mg/L)
Total Diss.
P (mg/L)
Total P
(mg/L)
Figure 14. The percentages of stream miles not meeting the reference condition
threshold values for water quality and water chemistry measures. The error bars equal 1
standard error.
52
-------
Macro!nvertebrate CDFs
Cumulative distribution function curves were created for 4 macroinvertebrate measures. The
75th percentile of reference site measurements was used as the threshold for the HBI scores.
The 25th percentile was used as the threshold for species richness, the percentage of EPT
genera and EPT individuals present. (Table 15)
Table 15. Macroinvertebrate metric reference condition thresholds.
Macroinvertebrate
Metrics
HBI
Species Richness
% EPT Genera
% EPT Individuals
Reference Condition
Threshold
3.92
16
35%
31%
Impairment Criteria
HBI score > 3.92
< 16 species
< 35% of genera are EPT taxa
< 31% of individuals are EPT taxa
53
-------
5 10 15 20 25 30 35 40
Species Richness
40 60 80 100
%EPT Individuals
123456789 10
HBI Score
X site CDF
X site 95% C.I.
.Bsite CDF
B site 95% C.I.
Figure 15. Cumulative distribution function curves of macroinvertebrate metrics
collected at the X and B sites. The stippled lines represent the 95% confidence intervals
around the cumulative distribution estimates. The vertical lines represent the reference
condition threshold values.
The X site and B site data produced overlapping CDFs and confidence intervals for species
richness, HBI score, the percentage of EPT genera and the percentage of EPT individuals
(Figure 15). The results of Kolmogorov-Smirnov tests (Dmax values) also found no significant
differences (all Dmax p-values > 0.05). Reference condition threshold values were applied to the
CDF data and estimates of the percent of stream miles not meeting these criteria were
determined (Figure 16). The overlapping error bars show no statistical differences between the
X site population and the B site population for all macroinvertebrate metrics in Figure 16. The
reference condition threshold value for the percentage of EPT genera present in a sample was
35 percent, and no sites in the sample populations met this criteria.
54
-------
100 n
90
80
o 70
o
Q_
ซ 60
JD
1 50
E
jg 40
* 30
20
10
0
I
1
HBI
I
1
I
1
^
Species
n B site
nXsite
I
1
I
1
1 i
% EPT % EPT
Richness Genera Individuals
Figure 16. The percentages of stream miles not meeting the reference condition
threshold values for macroinvertebrate metrics. The error bars equal 1 standard error.
Fish Cumulative Distributions
Cumulative distribution functions were plotted for the 4 fish metrics (Figure. 17). The 75th
percentile calculated from the reference site measurements was used as the threshold for the
percentage of 'tolerant' individuals. The 25th percentile was used as the threshold for the
percentage of stenothermal individuals present, the percentage of top carnivore individuals
present, and the fish IBI score. (Table 16)
Table 16. Fish metric reference condition thresholds.
Fish Metrics
% Top Carnivore Individuals
% Stenothermal Individuals
% Tolerant' Individuals
IBI Score
Reference Condition
Threshold
24.4%
83.5%
3.40%
60
Impairment Criteria
Any site where < 24.4% of
individuals were not top
predators
Any site where < 83.5% of
individuals were not
stenotherms
Any site where > 3.40% of
individuals are tolerant species
Any site where fish IBI score is
<60
55
-------
The CDF results show significant overlap between the X site and B site data, indicating that X
and B site data provide equivalent population estimates for fish metrics. The results of
Kolmogorov-Smirnov tests (Dmax values) also found no significant differences (all Dmax p-values >
0.05).
0 20 40 60 80 100
% Stenothermal Individuals
,100
0 20 40 60 80 100
% Top Carnivore Individuals
X site CDF
X site 95% C.I.
0 20 40 60 80 100
% Tolerant Individuals
,100
20 40 60 80 100
IB I Score
.B site CDF
B site 95% C.I.
Figure 17. Cumulative distribution function curves for fish metrics collected at the X and
B sites. The stippled lines represent 95% confidence intervals around the distribution
plots. The vertical lines represent the reference condition values.
56
-------
Fish metric reference condition threshold values were applied to the CDF data and estimates of
the percent of stream miles not meeting these criteria were determined (Figure 18). The
overlapping error bars show no statistical differences between the X site population and the B site
population for all fish metrics.
100
90
80
8 70
Q_
8 6ฐ
^ 50
E
| 40
-i '
CO
ฃ 30
20
10
n
D B site
I
1
I
1
T
1
T
1
I
1
I
1
nXsite
T T
L
1
% Top % % Tolerant IBI score
Carnivore Stenothermal
Figure 18. The percentages of stream miles not meeting the reference condition
threshold values for fish metrics. The error bars equal 1 standard error.
57
-------
6.6 Targeted vs. Random Stream Assessment
Fish Index of Biotic Integrity Cumulative Distribution Function Plots of Random and
Regional Biologist's Targeted Sampling Sites
Cumulative distribution function curves and box plots show differing distributions offish IBI scores
between the REMAP X and WDNR regional biologist's targeted sampling sites (Figure. 19). The
cumulative distribution functions show that 80 percent of the REMAP random sample population
(using weighted data) and 65 percent of the WDNR targeted stream population received an IBI
score of 60 or less, thereby not meeting the reference condition threshold for fish IBI score.
100%
20 40 60 80
Coldwater Fish IBI Score
100
100
CD 80
8
60
m
I 40
20
0
Random
Sample
10 ฐ
10
Targeted
Sample
Figure 19. Cumulative distribution function curves of fish IBI scores from biologists'
targeted sites (solid bold line) and REMAP X sites (stippled bold line), compared to the
reference condition threshold value (stippled vertical line). Box plots and distribution of
the biologist's targeted and REMAP X sites data is shown in figure on right.
58
-------
6.7 Effects of Geographical Distances Between Random and Modified-Random
Sites
Excluding 3 sites where the X and B reaches overlapped, the average distance from the X sites to
the B sites was 701 meters, with a minimum distance of 106 meters and a maximum distance of
2,283 meters. We investigated the relationships between the distance from X to B sampling sites
for the following measures: width:depth ratio, percentage of fine substrate, percentage of EPT
genera present, HBI score, number of fish captured, IBI score, fish species richness, dissolved
total phosphorus, and NH3.
20
^
0)
Q
15
.c
^
^
.^ 10
0)
o
c
1 5
Q
0
0) ^
5
1 0.5
w
ฃ 0.4
E
^ 0.3
w
8 0.2
c
CD
fe 0.1
Q
O.C
0)
8 3-
ro
m
^
~ 2-
0)
o
0)
1 . .
b n
ฐo
1 1 1 1
rs = -0.138
-
.*
~ .
** * *
_ ^
* *
x '. /
3 500 1000 1500 2000 25
Distance (m)
% rs = 0.112
** .
9 9
j * *. *
\ t '
*f i i ซ i i
0 500 1000 1500 2000 2
Distance (m)
rs = 0.209
_
_
i
ซ
' " *
" .*"
~J /. /, , '. , ,
500 1000 1500 2000 250
Distance (m)
6
&
c 5
0)
O
t 4
LU
^
c ^
OJ
o
c 2
OJ Z
S
Q 1
0
00 0
5C
ง 4C
0
CO
^ 3C
OJ
o
S 2C
b
1C
500
w 6
1
1 5
o:
CD 4
Q.
ro 3
W
L_
c 2
OJ
o
c ^
OJ 1
n n
I I U
0 0
1 1 1 1
rs = 0.033
-
-.M
- M
500 1000 1500 2000 25
Distance (m)
i i i i
rs = 0.077
-
- MB
0 500 1000 1500 2000 2
Distance (m)
rs = -0.208
- M -
'_ ..I . 1 ... I
500 1000 1500 2000 25
Distance (m)
600
ฃ
i> 500
O
-g 400
IT
* n
c 300
o 200
1
Q 100
X,
2- 0.4
"a
ง 0.3
1
8=
Q 0.2
CL
-a
a)
w 0 1
.2!
Q
"ro
"o
500 H ฐ'ฐc
0.3
"a
80.2
tj
(D
e
b
>* -'
3 500 1000 1500 2000 25
Distance (m)
rs = 0.250
-
-
-
1^ ^, m . \g^_ | ^ ** | ป
500 1000 1500 2000 25
Distance (m)
.rs = 0.195
~
WS v ^ซ i * i *" i "
500 1000 1500 2000 25
Distance (m)
00
30
30
Figure 20. Scatterplots showing relationships between the distance between the X and B
sites, and the absolute value of the differences in physical, chemical, and biological
measures collected at these sites.
59
-------
Spearman correlation coefficients (rs) indicate no significant relationships between the distance
between the X and associated B assessment sites and any of the physical, chemical, or biological
measures reported (Figure 20). From these coefficients, we determined that the distance
between a random sampling site and the closest bridge sampling site did not appear to influence
the strength of the correlations between the X and B sites data.
60
-------
7. Discussion
Characterization of Stream Resources in the Driftless Area
A primary objective of this study was to characterize the physical, chemical, and biological
conditions of stream resources in the Driftless Area ecoregion using the EMAP probabilistic
sampling design. Previous sampling efforts in the Driftless Area and entire state have primarily
been targeted sampling to gather data to address stream specific data needs. These targeted
sampling efforts may have induced intentional or unintentional biases, and have produced data
for which confidence values cannot be estimated. This REMAP study is the first broad-scale
assessment of stream resources by the WDNR that has produced data of known quality and
applied objective numeric criteria to judge whether individual or populations of streams are
physically, chemically, or biologically degraded. The 25th or 75th percentile values for various
physical, chemical, and biological measures from the reference stream sites were used to set
reference condition values (management expectations). Study results indicate that the
percentages of the sample population (and by inference the entire Driftless Area stream
population) not meeting their potential ranged from 77 to 100 percent depending upon the
physical, chemical, or biological criteria used. While setting management goals at the 25th or
75th percentiles is a scientifically defensible approach, setting resource management goals
should perhaps be viewed as a societal decision. Stream assessment findings and the
methods used to set reference conditions will stimulate further discussion within WDNR on
which individual or combinations of measures should be used to assess stream quality, and
what numeric criteria thresholds for these various parameters should be applied to judge stream
quality.
Differences Between Random and Modified-Random Sample Sites
Another primary objective of this study was to evaluate whether sampling randomly selected
stream segments at sites accessed from road crossings would provide results comparable to a
truly randomized sampling design. None of the 35 physical, chemical, or biological, parameters
evaluated were significantly different between the random (X) and modified-random (B) sampling
sites.
Given the relative homogeneity of water chemistry parameters due to mixing in lotic systems,
intuitively, few differences would be expected to exist for these measures between the X and B
sites. However, it is possible that tributaries intervening between the X and B sampling sites can
change the concentration of chemical parameters, water temperature, or dissolved oxygen
concentrations, if the tributaries have differing water chemistry concentrations, temperature, or
dissolved oxygen concentrations than the receiving stream. Intervening groundwater or point-
source inputs could also have similar affects.
None of the seven macroinvertebrate metrics analyzed were significantly different between and X
and B sites. Most macroinvertebrate taxa are relatively sessile organisms and have been shown
to be more strongly influenced by local habitat or reach-scale environmental factors than fish,
which show a stronger response to watershed-scale influences (Barbour et al. 1999, Lammert and
Allan 1999). Given that macroinvertebrates are thought to respond more strongly to site-specific
or reach-scale influences, the lack of X and B site differences in the macroinvertebrate measures
61
-------
may be a more sensitive test of bias being induced by the modified-random sampling design than
the fish assemblage data.
Probabilistic Sampling Design Issues
Fundamental principles of probabilistic sampling are that every population element in the target
population has a known (and non-zero) probability of being sampled, and it is critically important
to rigorously define both the target population and the elements it's comprised of (Cochran 1977).
A major objective of the Wisconsin REMAP study was to estimate the number of stream miles in
the study area that were meeting physical, chemical, and biological, reference condition criteria.
Therefore, a continuous sampling design was applied, where assessment measurements were
taken at or in the vicinity of randomly selected points (Larsen 1997). For population elements that
are spatially-static such as stream physical habitat (and perhaps less influenced by dynamic
elements such as flowing water), sampling at the modified-random sites violates key principles of
probabilistic sampling (in essence, the target population becomes the lengths of all streams that
are within some distance of road crossings that field crews are willing to travel to reach an
assessment site). For spatially-dynamic population elements (water chemistry parameters, fish,
or macroinvertebrates that are strongly influenced by upstream land use and the ambient
conditions of flowing water) is it less clear (at least to the authors) what the spatial boundaries of
these population elements are. It is hoped that this REMAP report will generate further research
and discussion on the validity of using road accessible stream sampling sites to characterize
stream target populations.
The mean distance between the random and modified-random assessment sites in this study was
approximately 701 meters (0.44 mi.). For other ecoregions in Wisconsin or areas outside the
state with lower road density, greater land cover or land use heterogeneity, higher potential for
intervening point sources of pollution, or greater topographic relief than the REMAP study area,
there may be a greater potential for differences between random and modified-random
assessment reaches. For example, in a significant proportion of northern Wisconsin road density
is about 30 percent less than that of the REMAP study area. This increases the potential distance
between random and road-accessible sites, and may result in greater observed differences
between these sites. However, this portion of northern Wisconsin is also characterized by more
homogeneous land cover and land use, and human population densities are less than in the
REMAP study area, which may result in fewer observed differences between sites.
It will be interesting to evaluate these types of interactions in Wisconsin's other ecoregions.
Sample Population Site Selection
The finding that 71 of the randomly selected stream sites were rejected during the field
reconnaissance effort to reach a sample population of 60 streams (nearly 120% of the original
sample population) is of significance for stream assessment studies and programs that use
probabilistic sampling designs. Sample designs must include sufficient over-sample populations
to maintain target sample sizes, given that significant numbers of random sites are likely to be
rejected. Also, project planning must incorporate a sufficient amount of time for map work to
identify site locations, and for field time for reconnaissance of assessment sites and subsequent
replacement of rejected sites. The fact that over half of sites were dropped due to dry stream
channels indicates a significant amount of error in the WDNR's perennial stream hydrography
database for the Driftless Area Ecoregion. Only 5% of the random sites were dropped due to
landowner access denial, which is significantly less than what the investigators had expected prior
to the start of the study.
62
-------
Selection of Reference Sites and Application of Reference Condition Data
While macroinvertebrate and fish assemblage indices are increasingly used to objectively assess
stream resources in Wisconsin, development of reference conditions for stream physical habitat
and water chemistry measures provides additional objective numeric criteria with which to
evaluate the condition of individual and populations of streams. In addition, the macroinvertebrate
and fish assemblage data collected at reference sites allowed calibration of statewide biological
indices specifically for the study area streams, thereby increasing the accuracy of the stream
assessments.
The selection of least disturbed reference sites for this study was the first time WDNR has applied
an a priori site selection of reference sites based on GIS data of percentages of watershed land
use types and stream physical habitat characteristics (Hughes et al. 1986). Streams in
watersheds with high proportions of county, state, or federal land ownership, and low proportions
of agricultural land were evaluated as potential least disturbed reference sites. Study results
show that on average the size of the least disturbed watersheds were smaller than the mean size
for the random sample population watersheds. As watershed area increases there is a greater
likelihood that agricultural and other land uses that negatively affect stream quality will occur
within a watershed, increasing the potential cumulative negative impacts in higher order streams.
Since stream size (Strahler order, or flow volume) is strongly influenced by watershed size, there
presumably are fewer least disturbed large streams relative to small streams in Wisconsin. The
influence of stream order on physical, chemical, and biological parameters is further evaluated in
Appendix B. Stream order may have implications for setting management goals for larger
streams. If reference conditions are developed based on data from lower order (smaller) streams,
applying these criteria to higher order streams may result in unrealistic management goals for
larger wadeable streams.
A total of 60% (n = 21) of the 35 parameter comparisons between the random and reference
stream populations were significantly different. The remaining 14 parameters that were not
significantly different may vary little among streams in the study population, and may be
indiscriminate measures of stream condition in the study area. Thus, the lack of differences
between the random and associated modified-random sample pairs for these same parameters
may be more a function of insensitive ecological measures than a lack of differences between
random and modified-random sampling sites.
Response of Stream Physical Habitat and Water Chemistry to Watershed Land Use
No significant relationships were detected between watershed percent cropland, percent
grassland, or percent forest cover, and riparian or in-stream physical habitat characteristics (e.g.,
sediment depth). With increasing agricultural land, particularly row cropping, the potential for
sediment delivery to streams increases. While the proportion of cropland in individual watersheds
can influence sediment delivery, proximity of cropland to streams, cropland slope, riparian land
use, width and linear extent, and plant species composition of riparian buffers, and a number of
other factors can influence nutrient and sediment delivery to streams. An earlier study conducted
by the USGS in Wisconsin showed that streams in the Driftless Region had twice the sediment
and nutrient loading rate compared to the statewide average (Corsi et al. 1997). Based on
Natural Resource Conservation Service cropland erosion rate data and the USGS study findings,
it is evident that significant amounts of sediment are being delivered to the study area streams,
but given the relatively steep topography and resulting stream gradient, the majority of the
sediment flows through these stream systems and is not deposited on the streambeds. In
63
-------
addition, the WISCLAND land use data used to quantify percent watershed agricultural land was
based on 10-12 year old LANDSAT data. So while the general land use patterns have changed
little in these watersheds over recent time, the dated land use data could have weakened the
relationships between land use and measurable in-stream physical habitat degradation. The fact
that some of the reference stream watersheds had significant amounts of agricultural land
suggests that with sufficient riparian protection and proper agricultural land management,
agricultural productivity and high quality streams can co-exist.
Response of Biotic Indices to Percent Watershed Agriculture, Stream Sediment and
Water Chemistry Variables
The results of the Pearson correlation analyses indicate stronger responses of
macroinvertebrates than fish to increasing watershed disturbance. Based on adjusted p-values,
neither the macroinvertebrate nor the fish metrics were strongly related to habitat or land use
variables. However, the two macroinvertebrate metrics examined were significantly correlated to
several water chemistry and water quality measures. The fish IBI score was only weakly
correlated with Kjeldahl nitrogen and water clarity. These results are surprising, as various
studies have provided evidence that the structure and function of stream fish communities are
altered with increasing proportions of agricultural land within watersheds (Fausch et al., 1990,
Meador and Goldstein, 2003). The EPT metric indicated reductions in the quality of the benthic
community in response to increasing total dissolved phosphorus concentration and, more weakly,
streambed sedimentation. Various studies have shown that as nutrient concentrations increase in
streams there is an increase in filamentous algae growth which displaces the periphytic (diatom)
community that is an important food source for many macroinvertebrate taxa (Lillie et al. 2003).
Similarly, with increasing sedimentation there is a corresponding loss of available coarse
substrate (rubble, cobble, and gravel) and the associated interstitial spaces that are critical habitat
for many macroinvertebrate taxa (Lillie et al. 2003). The lack of strong relationships between
metrics and measures of habitat and land use may be a function of the metrics selected, as some
metrics may show responses only to specific types of disturbance. Other factors related to the
land cover data also could affect the relationships observed, including the age of the LANDSAT
data, the distribution of agricultural land within individual watersheds, and the quality of the
riparian corridor along individual streams.
Comparison of Random Sample Population with WDNR Biologist's Targeted Sample
Population
The WDNR has primarily relied on targeted sampling to provide data on stream conditions for
site-specific, regional and statewide resource assessments. While it is critically important to
continue to collect data for stream-specific assessments and management decisions, the REMAP
study results indicate that using targeted data to make broad spatial inferences about the quality
of stream populations can lead to erroneous estimations of stream quality. In general, the
emphasis of stream monitoring in western Wsconsin has been to target high quality streams to
assess the condition of trout fisheries or to determine the potential for these streams to support
brown trout or the more environmentally-sensitive native brook trout. A key finding of this study
was that the bias of targeted sampling resulted in an overestimation of stream quality when
extrapolated to the population-level, findings similar to that of Hughes et al. (2000). Bias induced
from targeted sampling can of course also underestimate the quality of regional stream
populations if the focus of the targeted sampling is directed toward streams impaired by industrial
or municipal waste discharges, or polluted urban or agricultural run-off, as is typically the focus of
Total Maximum Daily Load watershed assessments.
64
-------
Implications for WDNR's Wadeable Stream Monitoring Program
The REMAP study results provide valuable insights for improving the rigor of the WDNR's
wadeable stream monitoring program. This study is the first time the WDNR has conducted a
probabilistic survey on a broad geographic scale. Results also indicate macroinvertebrate and
fish assemblages differ in their power to detect physical habitat degradation and chemical
pollution. In addition, this study provides evidence that the targeted sampling design routinely
used overestimates the quality of stream resources in western Wisconsin. The preliminary
findings that a modified-random sampling design appears to induce little bias in the assessment of
stream quality may allow a more efficient and cost-effective stream sampling effort in Wsconsin,
but additional study is needed to more rigorously evaluate the utility of applying a road-accessible
sampling design. Finally, the process of using numeric criteria to objectively determine whether
individual or populations of streams are meeting their potential will be applied statewide which will
help reduce WDNR resource managers' reliance on subjective, qualitative, resource evaluations.
65
-------
8. Acknowledgements
A number of WDNR staff contributed to this study. Among those, we thank Sue Acre, Bob
Busch, Bill Ceelen, Walt Jaeger, and Joanne Tooley. In addition, a number of EPA staff were
key contributors to this project and the authors are very thankful for their support. Dr. Florence
Fulk- NERL, and Sarah Lehmann and Edward Hammer of Region 5 helped secure funding and
initiate the project. Dr. Tony Olsen, NHEERL-WED provided the sample population - draw and
statistical guidance; Drs. Robert Hughes, Tony Olsen, Steve Paulsen, and Paul Ringold of
NHEERL provided review comments; and Karen Blocksom of NERL provided statistical
guidance, and technical reviews of earlier report drafts.
66
-------
9. References
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.
Cochran, W.G. 1977. Sampling Techniques, 3rd edition. J. Wley and Sons, New York, New
York.
Corsi, S.D., J. Graczyk, D.W. Owens, and R.T. Bannerman. 1997. Unit-area Loads of
Suspended Sediment, Suspended Solids, and Total Phosphorus from Small Watersheds
in Wisconsin. U.S. Geological Fact Sheet No. USGS-FS-195-97.
Davies, N.M., R.H. Norris, and C.T. Martin. 2000. Prediction and assessment of local stream
habitat features using large-scale catchment characteristics. Freshwater Biology
45:343-369.
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.
Diaz-Ramos, S., D.L. Stevens, Jr., and A.R. Olsen. 1996. EMAP Statistical Methods Manual.
EPA/620/R-96/002. U.S. Environmental Protection Agency, Office of Research and
Development, National Health Effects and Environmental Research Laboratory, Western
Ecology Division, Corvallis, Oregon.
Fausch, K.D., J. Lyons, J.R. Karr, and P.L. Angermeier. 1990. Fish communities as indicators
of environmental degradation. American Fisheries Society Symposium 8:123-144.
Gibson, G.R., Jr., M.T. Barbour, J.B. Stribling, J. Gerritsen, and J.R. Karr. 1996. Biological
Criteria: Technical Guidance for Streams and Small Rivers. EPA/822-B-96-001. U.S.
Environmental Protection Agency, Office of Science and Technology, Washington, DC.
Hilsenhoff, W.L. 1987. An improved biotic index for organic stream pollution. Great Lakes
Entomologist 20:31-39.
Hughes, R.M., D.P. Larsen, and J.M. Omernik. 1986. Regional reference sites: a method for
assessing stream potentials. Environmental Management 10:629-635.
Hughes, R.M., S.G. Paulsen, and J.L. Stoddard. 2000. EMAP-Surface Waters: a
multiassemblage probability survey of ecological integrity in the U.S.A. Hydrobiologia
422:429-443.
Karr, J.R. and E.W. Chu. 1999. Restoring Life in Running Waters: Better Biological Monitoring.
Island Press, Washington, D.C.
67
-------
Keys, J.E., Jr., C.A. Carpenter, S.L Hooks, F.G. Koeneg, W.H. McNab, W.E. Russell, and M.L
Smith. 1995. Ecological Units of the Eastern United States-First Approximation.
Technical Publication R8-TP 21. Map (scale 1:3,500,000). U.S. Department of
Agriculture, Forest Service, Atlanta, Georgia.
Lammert, M., and J.D. Allan. 1999. Assessing Biotic Integrity of streams: effects of scale in
measuring the influence of land use/cover and habitat structure on fish and
macroinvertebrates. Environmental Management 23:257-270.
Larsen, D.P. 1995. The role of ecological sample surveys in the implementation of biocriteria.
In: Davis, W.S. and T.P. Simon (Editors), Biological Assessment and Criteria: Tools for
Water Resource Planning. CRC Press Inc. Boca Raton, Florida.
Larsen, D.P. 1997. Sample survey design issues for bioassessment of inland aquatic
ecosystems. Human and Ecological Risk Assessment 3:979-991.
Legendre, P. and L. Legendre. 1998. Numerical Ecology. Elsevier Science, Amsterdam, The
Netherlands.
Lillie, R.A., S.W. Szczytko, and M.A. Miller. 2003. Macroinvertebrate Data Interpretation
Guidance Manual. Wisconsin Dept. of Natural Resources, Madison, Wisconsin. Pub.
SS-965-2003.
Lyons, J. 1992a. Using the Index of Biotic Integrity (IBI) to Measure Environmental Quality in
Warmwater Streams of Wisconsin. General Technical Report NC-49. North Central
Forest Experiment Station, U.S. Department of Agriculture, St. Paul, Minnesota.
Lyons, J. 1992b. 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.
Lyons, J., L. Wang, and T.D. Simonson. 1996. Development and validation of an index of biotic
integrity for coldwater streams in Wisconsin. North American Journal of Fisheries
Management 16:241-256.
Meador, M.R., and R.M. Goldstein. 2003. Assessing water quality at large geographic scales:
relationships among land use, water, physiochemistry, riparian conditions, and fish
community structure. Environmental Management 31:505-517.
Miller, M.A., J.D. Diamond, J.B. Stribling, and K.L. Schmude. (In prep.) Development and
application of a performance-based comparability framework using a comparison of
macroinvertebrate field and laboratory protocols.
Omernik, J.M. 1987. Ecoregions of the conterminous United States. Annals of the Association
of American Geographers 77:118-125.
Petersen, S.A., N.S. Urquhart, and E.B. Welch. 1999. Sample representativeness: a must for
reliable regional lake condition estimates. Environmental Science and Technology
33:1559-1565.
68
-------
Robertson, D.M., D.J. Graczyck, P.J. Garrison, L. Wang, G. LaLiberte, and R. Bannerman.
2006. Nutrient concentrations and their relationships to the biological integrity of
streams in Wisconsin. U.S. Geological Survey Prof. Paper 1722.
Sokal, R.R., and F.J. Rohlf. 1981. Biometry 2ndEdition. W.H. Freeman and Co. New York,
NY.
Stevens, D.L., Jr. 1997. Variable density grid-based sampling designs for spatial populations.
Environmetrics 8:167-95.
Stevens, D.L., Jr., and A.R. Olsen. 1999. Spatially restricted surveys over time for aquatic
resources. Journal of Agricultural, Biological, and Environmental Statistics 4: 415-428.
Stevens, D.L., Jr., and A.R. Olsen. 2004. Spatially-balanced sampling of natural resources.
Journal of the American Statistical Association 99:262-278.
Strahler, A.N. 1957. Quantitative analysis of watershed geomorphology. Transactions of the
American Geophysical Union 38:913-920.
69
-------
Appendix A. Comparison of Spring and Fall Water Chemistry
Parameters
Evaluation of Seasonal Effects
Water quality and water chemistry data collected in the spring and fall of 2003 were compared
using Bonferroni-adjusted paired t-tests to determine if seasonal differences existed. In the spring
and fall of 2003 we measured instantaneous values of dissolved oxygen concentration, dissolved
oxygen percent saturation, pH, and water temperature in situ with electronic meters, and water
transparency with a transparency tube at the X, B and reference stream sites. Total phosphorus,
total dissolved phosphorus, NH3 (ammonia), NO3-NO2 (nitrate-nitrite), and total Kjeldahl nitrogen
concentrations were also measured using laboratory-analyzed grab samples. Paired t-tests (with
Bonferroni adjustments) were used to compare within-site spring and fall differences for these 10
parameters. Water transparency, dissolved oxygen concentration and dissolved oxygen percent
saturation showed statistically significant differences between the spring and fall sampling periods
(Table A1, Figure A1). Dissolved oxygen concentration, the percent dissolved oxygen saturation,
and water column transparency had higher values during the fall sampling period than the spring.
Total dissolved phosphorus, NH3, and Kjeldahl nitrogen also displayed statistically significant
concentration differences between the spring and fall samples. Total dissolved phosphorus
values were significantly higher in the fall. In contrast, both NH3 and Kjeldahl nitrogen were
significantly higher in the spring samples. Because of these seasonal differences in various water
chemistry measures, and because the spring dataset had a greater number of sites sampled, only
water chemistry samples collected in spring were used for analyses in this study.
Table A 1. Paired t-test comparisons between spring and fall water quality and water
chemistry measures collected at the X stream sites.
Spring Vs Fall Water Chemistry Measures
Total Phosphorus (mg/L)*
Total Dissolved Phosphorus (mg/L)*
NH3 (mg/L)*
NO3-NO2 (mg/L)*
N-Kjeldahl (mg/L)*
Dissolved Oxygen (mg/L)
Transparency (cm)
% Oxygen Saturation
PH
Water Temperature (C)
No. of
samples
47
49
49
49
49
49
43
43
38
49
Degrees of
Freedom
46
48
48
48
48
48
42
42
37
48
p-value
0.240
0.001
0.001
0.670
0.043
0.000
0.034
0.000
0.798
0.294
*Data were Log-io transformed prior to conducting the paired t-tests. P-values in bold are significant.
70
-------
3
D)
ง2-
ro
p=0.034
D"
3" LJ ]~
3
spring fall
-
-
Figure A 1. Box and whisker plots of spring and fall water quality and water chemistry
measures collected at the X-sites. P-values represent paired t-test results.
71
-------
Appendix B. Relationships of Stream Order with Physical Habitat,
Water Quality and Water Chemistry, and Biological Measures
Habitat Measures and Stream Order
To evaluate relationships between stream order and physical habitat measures, an ANOVA was
applied to each habitat measure collected at the X sites. Significant relationships were detected
between stream order and sinuosity, mean distance between bends, and mean length of bank
erosion (Tables B1, B2). Sinuosity differed between first and second order sites compared to
third and fourth order streams. Width of bank erosion and distance between bends both differed
significantly between second and fourth order streams only.
Table B 1. ANOVA results of stream order and physical habitat measures collected at the
X sites.
X Site Habitat Measures Vs Stream Order
Sinuosity*
Width : Depth Ratio
Mean Depth of Fines*
Percent Sand, Silt, and/or Clay Sediments
Mean Distance Between Bends (m)*
Mean Length of Riffles (m)*
Mean Length of Pools (m)*
Mean Width of Bank Erosion (m)*
# of Samples
53
56
53
53
53
53
53
53
F-ratio
12.551
0.869
1.493
2.144
3.367
0.366
1.552
6.795
p-value
<0.001
0.463
0.228
0.107
0.026
0.778
0.213
0.001
* Data were Logio(x+1) transformed prior to conducting ANOVA. The percent sand, silt, and clay values were arcsine
square root transformed prior to conducting ANOVA. P values in bold are significant.
Table B 2. Tukey HSD pairwise comparison probabilities for physical habitat measures
that produced significant ANOVA results. P-values in bold indicate significantly different
parameter values between the corresponding stream orders.
Parameter
Sinuosity
Mean Distance
Between Bends
Mean Width of
Bank Erosion
Stream Order
2
3
4
2
3
4
2
3
4
1
0.992
0.006
0.000
0.987
0.895
0.069
0.100
0.936
0.317
2
0.002
0.000
0.696
0.027
0.211
0.000
3
0.589
0.167
0.061
72
-------
Water Quality and Water Chemistry Measures and Stream Order
Similar to the ANOVA conducted with the habitat data, we evaluated water quality and water
chemistry data collected from the X sites to determine if these measures varied by stream order.
Only dissolved oxygen percent saturation showed statistical differences among stream orders
(Tables B3, B4).
Table B 3. ANOVA of stream order and water quality and water chemistry measures from
the X stream sites.
X Site Water Chemistry Vs Stream Order
Total Phosphorus (mg/L)*
Total Dissolved Phosphorus (mg/L)*
NH3 (mg/L)*
N03-N02 (mg/L)*
N-Kjeldahl (mg/L)*
Dissolved Oxygen (mg/L)
Percent Oxygen Saturation
Transparency (cm)
PH
Conductivity (uS)*
Water Temperature (C)
# of Samples
56
56
56
56
56
57
57
57
57
47
57
F-ratio
0.506
0.528
0.647
2.709
0.767
2.522
3.662
0.536
2.769
0.439
0.398
p-value
0.680
0.665
0.588
0.054
0.518
0.068
0.018
0.660
0.051
0.726
0.755
*Data were Log-io transformed prior to conducting the ANOVA analysis. P values in bold are significant.
Table B 4. Tukey HSD pairwise comparison for the percent oxygen saturation. The p-
value in bold indicates significantly oxygen saturation values between second and third
order streams.
Stream Order
2
3
4
1
1.000
0.060
0.247
2
0.049
0.224
3
0.900
73
-------
Macroinvertebrate Measures and Stream Order
ANOVA results comparing macroinvertebrate data and stream order indicated no significant
relationships between stream order and any of the 7 macroinvertebrate metrics or indices. (Table
B5)
Table B 5. ANOVA of stream order and macroinvertebrate measures from the X stream
sites.
X Site Macroinvertebrate Measures* Vs Stream Order
Species Richness
HBI Score
Percent EPT Genera
Percent EPT Individuals
Shannon's Diversity Index Score
Percent Shredders
Percent Scrapers
F-ratio
0.765
1.288
2.075
0.867
0.713
0.224
1.303
p-value
0.520
0.290
0.117
0.465
0.549
0.879
0.285
*Percent values were arc-sine square root transformed prior to conducting ANOVAS. n = 49
74
-------
Fish Assemblage Measures and Stream Order
In contrast to the macro!nvertebrate data presented above, ANOVA results showed some
significant differences among stream orders and fish assemblage measures. Specifically, species
richness and the number of "intolerant" species present are significantly related to stream order.
(Tables B6, B7)
Table B 6. ANOVA of stream order and fish assemblage measures from the X stream
sites (N=42).
X Site Fish Assemblage Measures Vs Stream Order
Species Richness
Number of Intolerant Species*
Percent of Tolerant Individuals
Percent Top Carnivores
Percent Stenothermal Cool/Coldwater
Percent of Salmonids that are Brook Trout
IBI Score*
F-ratio
4.672
4.601
0.558
1.887
0.879
1.228
0.993
p-value
0.007
0.008
0.646
0.148
0.461
0.313
0.406
* Data were Logio(X+1) transformed prior to conducting ANOVAS. **Data were Log-io(X) transformed prior to
conducting ANOVAs. Percent values were arc-sine square root transformed prior to conducting ANOVA. P-values in
bold indicate significant tests.
Table B 7. Tukey HSD pairwise comparison probabilities for fish metrics that produced
significant ANOVA results. P-values in bold indicate significantly different parameter
values between the corresponding stream orders.
Parameter
Species
Richness
Number of
Intolerant
Species
Stream Order
2
3
4
2
3
4
1
0.770
0.712
0.018
0.200
0.143
0.005
2
1.000
0.056
0.999
0.250
3
0.043
0.247
75
-------
Appendix C. Fish Assemblage Data Analyses
We captured a total of 54 species and 19,999 individual fish at the X, B, and reference stream
sites. Forty different species of fish were captured at the X sites, and eighteen different species of
fish were captured at the reference sites (Tables C1, C2). No fish were captured at 11 of the
individual X or B stream sites and at 2 of the reference sites. Numbers of individual fish caught
per X, B, or reference site ranged from 0 to 1,755. On average, 179 fish were captured at the X
and B sites and 101 fish at the reference sites. The number of fish species caught at individual X
and B sites ranged from 1 to 18, with an average of 6 species captured per site. The number of
fish species captured at each reference site ranged from 1 to 8, with an average of 3 species
captured per site.
Brook Trout (Salvelinus fontinalis) was the most common species (650 individuals) captured at
the reference stream sites, and Central Stoneroller (Campostoma anomalum) was the most
common species captured at the X sites (1,190 individuals). Two-thirds of the fish species
captured at the X sites were not found at the reference sites; conversely, Burbot (Lota lota),
Rainbow trout (Oncorhynchus mykiss), and Slimy Sculpin (Coitus cognatus) were captured at the
reference sites but at none of the X sites. Species with the greatest dissimilarity in frequency of
occurrence between the X and reference sites were Creek Chub (Semotilus atromaculatus)
occurring in 57% of the X sites and only 5% of the reference sites, White Sucker (Catostomus
commersoni) occurring in 55% of the X sites and only 14% of the reference sites, and Johnny
Darter (Etheostoma nigruni) occurring in 45% of the X sites and only 10% of the reference sites.
Brook Trout occurred in 57% of the reference sites and only 30% of the X sites, Mottled Sculpin
(Cottus bairdi) occurred at 38% of the reference sites and only 24% of the X sites, and Brown
trout (Salmo trutta) occurred at 48% of the reference sites and only 24% of the X sites (Fig. C1).
76
-------
Table C 1. Fish species captured at X stream sites.
Species
American Brook Lamprey
Bigmouth Shiner
Blacknose Dace
Blacknose Shiner
Blackside Darter
Bluegill
Bluntnose Minnow
Brassy Minnow
Brook Stickleback
Brook Trout
Brown Trout
Central Mudminnow
Central Stoneroller
Channel Catfish
Common Carp
Common Shiner
Creek Chub
Fantail Darter
Fathead Minnow
Green Sunfish
Hornyhead Chub
Johnny Darter
Longnose Dace
Mottled Sculpin
Northern Brook Lamprey
Northern Hog Sucker
Pearl Dace
Pumpkinseed
Quillback
Rosyface Shiner
Sand Shiner
Shorthead Redhorse
Smallmouth Bass
Southern Redbelly Dace
Spotfin Shiner
Stonecat
Suckermouth Minnow
Tiger Trout
Walleye
White Sucker
Total Number of Fish
Scientific Name
Lampetra appendix
Notropis dorsalis
Rhicnichthys atratulus
Notropis heterlepis
Percina maculata
Lepomis macrochirus
Pimephales notatus
Hybognathus hankinsoni
Culaea inconstans
Salvelinus fontinalis
Salmo trutta
Umbra limi
Campostoma anomalum
Ictalurus punctatus
Cyprinus carpio
Notropis cornutus
Semotilus atromaculatus
Etheostoma flabellare
Pimephales promelas
Lepomis cyanellus
Nocomis biguttatus
Etheostoma nigrum
Rhinichthys cataractae
Cottus bairdi
Ichthyomyzon fossor
Hypentelium nigricans
Semotilus margarita
Lepomis gibbosus
Carpiodes cyprinus
Notropis rubellus
Notropis stramineus
Moxostoma
macrolepidotum
Micropterus dolomieu
Phoxinus erythrogaster
Notropis spilopterus
Noturus flavus
Phenacobius mirabilis
Salvelinus fontinalis X
Salmo trutta
Sander vitreus
Catostomus commersoni
Total No. of
Fish Caught
17
49
284
26
4
16
542
2
180
511
208
4
1,190
1
4
1,154
945
156
106
22
745
345
95
260
3
2
238
1
4
34
3
3
25
227
29
42
32
1
2
762
8,275
No. ofX-Sites
with Species
Present
5
4
17
4
3
3
9
1
20
15
12
3
7
1
1
9
29
14
6
4
7
23
8
12
2
1
3
1
1
5
1
3
3
8
3
3
2
1
1
28
Frequency of
species at X-
sites
9.8%
7.8%
33.3%
7.8%
5.9%
5.9%
17.7%
2.0%
39.2%
29.4%
23.5%
5.9%
13.7%
2.0%
2.0%
17.6%
56.9%
27.5%
1 1 .8%
7.8%
13.7%
45.1%
15.7%
23.5%
4.0%
2.0%
5.9%
2.0%
2.0%
9.8%
2.0%
5.9%
5.9%
15.7%
5.9%
5.9%
3.9%
2.0%
2.0%
54.9%
77
-------
Table C 2. Fish species captured at reference stream sites.
Species
American Brook Lamprey
Blacknose Dace
Blacknose Shiner
Bluntnose Minnow
Brook Stickleback
Brook Trout
Brown Trout
Burbot
Central Mudminnow
Central Stoneroller
Creek Chub
Fantail Darter
Johnny Darter
Longnose Dace
Mottled Sculpin
Rainbow Trout
Slimy Sculpin
White Sucker
Total Number of Fish
Scientific Name
Lampetra appendix
Rhicnichthys atratulus
Notropis heterolepis
Pimephales notatus
Culaea inconstans
Salvelinus fontinalis
Salmo trutta
Lota lota
Umbra limi
Camostoma anomalum
Semotilus
atromaculatus
Etheostoma flabellare
Etheostoma nigrum
Rhinichthys cataractae
Cottus bairdi
Oncorhynchus mykiss
Cottus cognatus
Catostomus
commersoni
Total No. No. of Reference
of Fish Sites with
Caught Species Present
55
35
2
6
63
650
625
43
1
5
8
35
35
76
547
3
118
90
2,397
5
4
1
1
6
12
10
2
1
1
1
2
2
3
8
3
1
3
Frequency of
species at
Reference Sites
23.8%
19.1%
4.8%
4.8%
28.6%
57.1%
47.6%
9.5%
4.8%
4.8%
4.8%
9.5%
9.5%
14.3%
38.1%
14.3%
4.8%
14.3%
78
-------
Percent Frequency of Occurrence
10
20
30
40
50
60
Big mouth Buffalo
Burbot
Finescale Dace
Golden Shiner
Golden Redhorse
Largemouth Bass
Log perch
~ฉ
ฉ
-ฉ
0
-ฉ
Northern Redbelly Dace < h ฉ
Pirate Perch < h ฉ
Rainbow Trout <
Redside Dace
Rock Bass
Slimy Sculpin
Yellow Bullhead
Brassy Minnow 4- -ฉ
Channel Catfish ซh ฎ
Common Carp ซ I- 0
Northern Hogsucker ซ h 0
Pumpkinseed
Quillback
Sand Shiner
Tiger Trout ซ H 0
O
ฉ-ป
CD
Walleye
Northern Brook Lamprey
ff\ Suckermouth Minnow
T3 Blackside Darter ^ a
Bluegill
Central Mudminnow
JJJ Pearl Dace
Shorthead Redhorse
Smallrnouth bass
Spotfin Shiner
Stonecat ^ 9
Bigmouth Shiner
Blacknose Shiner -( ฉ -4
Green Sunfish
American Brook Lamprey
Rosyface Shiner
Fathead Minnow
Central Stoneroller
Hornyhead Chub
X Site
ฎ B Site
Ref Site
- ~ฉ
a
e 0
O
--ฉ
--ฉ--
-ซe-
ฉ
Long nose Dace ฉ- <
Souttiern Redbelly Dace I- ฉ
Btuntnose Minnow - -
Common Shiner
Brown Trout -
Mottled Sculpin - -
Fantail Darter - -
Brook Trout
Blacknose Dace 4 ฉ
Brook Stickleback - -
Johnny Darter
White Sucker
Creek Chub
-ฉ
5>
--ฎ
Figure C 1. Percent frequency of occurrence of fish species at the X, B, and reference
sites.
79
-------
vvEPA
United States
Environmental Protection
Agency
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
D; 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
Office of Research
and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
EPA600/R-06/165
November 2006
www.epa.gov
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free
------- |