United States
Environmental Protection
Agency
Office of Research and
Development
Washington, DC 20460
June 2005
EPA 620/R-05/006
&EPA
Environmental Monitoring and
Assessment Program (EMAP)
Western Streams and Rivers
Statistical Summary
Environmental Monitoring and
Assessment Program
-------
Environmental Monitoring and Assessment
Program (EMAP)
Western Streams and Rivers
Statistical Summary
Stoddard, J.L1, D.V. Peck1, A.R. Olsen1, D.P. Larsen1, J. Van Sickle1, C.P. Hawkins2,
R.M. Hughes3, T.R. Whittier4, G. Lomnicky4, AT. Herlihy3, P.R. Kaufmann1, S.A.
Peterson1, P.L. Ringold1, S.G. Paulsen1, R. Blair1
1 U.S. Environmental Protection Agency
Western Ecology Division
National Health and Environmental Effects Laboratory
Office of Research and Development
200 SW 35th Street
Corvallis, OR 97333
2 Department of Aquatic, Watershed, & Earth Resources
Western Center for Monitoring and Assessment of Freshwater Ecosystems
5210 Old Main Hill
Utah State University
Logan, UT 84322-5210
3 Department of Fish and Wildlife
Oregon State University
c/o U.S. Environmental Protection Agency
200 SW 35th Street
Corvallis, OR 97333
4 Dynamac Corp.
c/o U.S. Environmental Protection Agency
200 SW 35th Street
Corvallis, OR 97333
-------
Acknowledgments
The quality of this report was greatly improved by comments from Frank McCormick
(USDA Forest Service), Gail Sloane (Florida DEP), Dan McKenzie (EPA ORD), Dixon
Landers (EPA ORD), Bob Ozretich (EPA ORD) and Karl Hermann (EPA Region 8).
The information in this document has been funded wholly or in part by the U.S.
Environmental Protection Agency under contract 68-D-01-005 to Dynamac Corporation,
cooperative agreement CR831682 to Oregon State University (Herlihy and Hughes),
and EPA STAR grant R-82863701 (Hawkins). It has been subjected to review by the
National Health and Environmental Effects Research Laboratory and approved for
publication. Approval does not signify that the contents reflect the views of the Agency,
nor does mention of trade names or commercial products constitute endorsement or
recommendation for use.
-------
Introduction
This statistical summary reports data from the Environmental Monitoring and
Assessment Program (EMAP) Western Pilot (EMAP-W). EMAP-Wwas a sample survey
(or probability survey, often simply called 'random') of streams and rivers in 12 states of
the western U.S. (Arizona, California, Colorado, Idaho, Montana, Nevada, North
Dakota, Oregon, South Dakota, Utah, Washington and Wyoming), comprising the
conterminous portions of EPA Regions 8, 9 and 10.
The eventual objective of EMAP-W is to assess the ecological condition of, and relative
importance of stressors in, streams and rivers of the West at multiple scales. This
Statistical Summary is the first step in making that assessment, in that it reports on the
validated and verified, but largely uninterpreted, data collected by EMAP-W.
Field sampling was conducted from 2000 through 2004, using a combination of State,
Regional and contract crews. All crews were trained in the EMAP-W sampling protocols
described in detail in Peck et al. (2005a) and Peck et al. (2005b). Identical sampling
methods were used in all wadeable streams, and complementary methods were used in
large rivers.
The purpose of this report is to provide the reader with sufficient information to
understand how EMAP-Wwas conducted, and how the information can be interpreted.
The statistical distribution(s) of measured variables and calculated metrics are included
as appendices to each report section. Details of design, sampling and data analysis are
given in each of the following sections of the report:
• Design - how were the sites chosen, and what do they represent
• Quality Assurance - how did we evaluate and document the quality of the data,
during data collection, database development, and data analysis
• Reference Condition - several indicators require some estimate of reference
condition, or expected condition; how were these estimates made?
• Extent of Resource - what have we learned about the total length of streams and
rivers (and their size categories) in the West?
• Ecological Condition - we use biological indicators to measure ecological
condition:
o Benthic Macroinvertebrates - how we constructed metrics, a Multi-Metric
Index, and a Predictive Model to interpret macroinvertebrate assemblage
data
o Aquatic Vertebrates - how we constructed metrics and a Multi-Metric
Index to interpret aquatic vertebrate (fish and amphibians) assemblage
data
• Environmental Stressors - we use chemical, physical and biological indicators to
measure the stress to which streams and rivers are exposed:
o Water Chemistry - which variables might be considered measures of
stress and why
o Physical Habitat - indicators of 8 dimensions of stream and river habitat,
and how they indicate levels of stress on aquatic organisms
-------
o Fish Tissue Contaminants - levels of toxic contaminants that accumulate
in fish tissue and are considered contributions to stress
o Invasive Riparian Plants - information on the presence/absence of
selected invasive alien plants that are commonly found in riparian areas of
streams and rivers, and can be considered indicators of stress to riparian
areas
o Other alien species - information on the presence/absence of selected
invasive fish, amphibian and macroinvertebrate species that are potential
stressors to biotic integrity.
Results are presented a three different levels of geographic resolution (illustrated in
Figure 1):
• West-wide (12 states)
• Three major climatic/topographic regions - Mountains, Plains and Xeric (see
Table 1)
• Ten ecological regions - aggregated from Omernik Level III (Omernik 1987)
ecoregions (see Table 1)
References
Omernik, J. M. 1987. Ecoregions of the conterminous United States. Annals of the
Association of American Geographers 77:118-125.
Peck, D. V., D. K. Averill, A. T. Herlihy, R. M. Hughes, P. R. Kaufmann, D. J. Klemm, J.
M. Lazorchak, F. H. McCormick, S. A. Peterson, M. R. Cappaert, T. Magee, and P.
A. Monaco. 2005a. Environmental Monitoring and Assessment Program - Surface
Waters Western Pilot Study: Field Operations Manual for Non-Wadeable Rivers and
Streams. EPA 600/R-05/xxx, U.S. Environmental Protection Agency, Washington,
DC.
Peck, D. V., A. T. Herlihy, B. H. Hill, R. M. Hughes, P. R. Kaufmann, D. J. Klemm, J. M.
Lazorchak, F. H. McCormick, S. A. Peterson, P. L. Ringold, T. Magee, and M. R.
Cappaert. 2005b. Environmental Monitoring and Assessment Program - Surface
Waters Western Pilot Study: Field Operations Manual for Wadeable Streams. EPA
600/R-05/XXX, U.S. Environmental Protection Agency, Office of Research and
Development, Washington, DC.
-------
Reporting Units
Table 1. Aggregation of Level III ecological regions for reporting of EMAP West
data. Abbreviations in parentheses are shortened forms of aggregate names,
used throughout this report. Numbers in parentheses are the number of
probability sites in each ecoregion.
Climatic/
Topographic
Regions
Aggregated
Ecological Regions
Omernik Level III Ecoregion Names
(number)
Mountains (MT)
Southwestern
Mountains
(MT-SWEST)
Arizona/New Mexico Mountains (23)
Southern California Mountains (8)
Northern Rockies
(MT-NROCK)
Blue Mountains (11)
Northern Rockies (15)
Idaho Batholith (16)
Middle Rockies (17)
Canadian Rockies (41)
Pacific Northwest
(MT-PNW)
Coast Range (1)
Puget Lowland (2)
Willamette Valley (3)
Cascades (4)
Sierra Nevada (5)
North Cascades (77)
Klamath Mountains (78)
Eastern Cascades Slopes and Foothills (8)
Southern Rockies
(MT-SROCK)
Wasatch and Uinta Mountains (19)
Southern Rockies (21)
Plains (PL)
Cultivated Northern
Plains
(PL-NCULT)
High Plains (25)
Northern Glaciated Plains (36)
Western Corn Belt Plains (47)
Lake Agassiz Plain (48)
Rangeland Plains
(PL-RANGE)
Southwestern Tablelands (26)
Northwestern Glaciated Plains (42)
Northwestern Great Plains (43)
Nebraska Sand Hills (44)
in
-------
Climatic/
Topographic
Regions
Aggregated
Ecological Regions
Omernik Level III Ecoregion Names
(number)
Xeric (XE)
Northern Xeric Basins
(XE-NORTH)
Columbia Plateau (10)
Snake River Plain (12)
Northern Basin and Range (80)
Xeric California
Lowlands
(XE-CALIF)
Southern and Central California Chaparral
and Oak Woodlands (6)
Central California Valley (7)
Eastern Xeric Plateaus
(XE-EPLAT)
Southern Xeric Basins
(XE-SOUTH)
Wyoming Basin (18)
Colorado Plateaus (20)
Arizona/New Mexico Plateau (22)
Central Basin and Range (13)
Mojave Basin and Range (14)
Chihuahuan Deserts (24)
Madrean Archipelago (79)
Sonoran Basin and Range (81)
IV
-------
Aggregate Ecoregion Assessment Regions'
Western Forested Mountains
I I D^uirweMir Mcurolrs
CH Notnem Rockies
l l s
Bcfllc NortnweEt
Great Plains
CH Cu-vale-d Nolhsrr =lalrs
I I ^angsland Plains.
Xeric West
CH Kornem Xfiric- 53£lrs
CH Esjtrern Xerc Basins
l l E3s:sn xsr! PIS-.SSJB
Figure 1 Map of three scales used in reporting EMAP West results: (1) All of
EMAP West study area (12 states); (2) 3 major climatic/topographic regions
(Mountains, Plains, Xeric); and (3) 10 aggregate ecological regions.
-------
How to Use this Report
The introductory sections of this report (Design, Quality Assurance, Reference
Condition) provide the background necessary for the reader to understand the design of
EMAP-W. Each provides a narrative description of how EMAP-Wwas constructed to
assure that the data could be used to estimate the ecological condition of streams and
rivers throughout the West.
Each subsequent section (Extent, Ecological Condition, Stressors) presents the results
of data collection. In addition to narrative descriptions of sampling methods, index and
metric development, and summary statistical information, most of these sections
present a series of graphs illustrating the range of values found for each variable (direct
results of field or lab measurements), metric (a calculated variable, based on the raw
data collected in the field or lab), or index (a composite of metrics) at three geographic
scales: (1) West-wide; (2) Three climatic/topographic regions; and (3) Ten ecological
regions. Each graph page consists of three elements: (1) an Empirical Cumulative
Distribution estimate, (2) summary statistic estimates of percentiles, mean, and
standard deviation, and (3) an empirical density estimate.
An example Empirical Cumulative Distribution (also known as cumulative frequency
distributions, by convention abbreviated as CDF) is shown below with the following
guide to interpretation.
Empirical Cumulative Distribution Estimate
100 -
80
CD
E 6°
CD
CO
"o
•S 40 H
I
20 -
0 -
CDF Estimate
95% Confidence Limits
20
40 60
Response Variable
80
100
VI
-------
The solid line in the CDF (above) shows the entire cumulative distribution of the plotted
variable in the population of streams being presented (e.g., West-wide, or in a single
region). The left-hand Y axis shows the proportion of stream length with a particular
characteristic, while the right-hand Y axis shows the actual stream length. The value of
the plotted variable at 0% (or 0 km of stream length) is the minimum value in the region.
The variable value at 100% is the maximum value in the region. At any point along the
CDF, the corresponding value on the Y axis is the proportion (or length) of stream with a
value of the plotted variable less than or equal to the corresponding value on the X axis.
The median value, or 50th percentile, for example, is found by locating the 50% value on
the left axis, moving horizontally across the graph to the CDF line, then reading down
perpendicularly to the corresponding value on the X axis (see illustration, below). The
median value of the variable shown in this example is ca. 70. An equally valid
interpretation of this same information is that 50% of the stream length in this example
region has a value of 70 or less for the plotted variable.
Empirical Cumulative Distribution Estimate
GO
c;
0)
I
55
2
53
"B
100 -
80 -
60 -
t= 40 -
20 -
- 300000
- 250000
- 200000
-C
'S
150000
- 100000
- 50000
03
20
40
60
80
100
Response Variable
Each Empirical Cumulative Distribution also includes 95% confidence limits for the CDF.
The confidence limits are only plotted for the percent of stream length between 5% and
95%. One of the strengths of the sample survey design implemented in EMAP-W is
that it allows the calculation of uncertainty for any estimate we make. Using the example
above, we can state with 95% confidence that the proportion of stream length with a
variable value of 70 or less is between ca. 46% and 54% (the lower and upper
confidence bounds around the CDF at X=70).
A similar CDF graphic is used for discrete data (e.g., the number of species in a
particular taxonomic group, like EPT Taxa, where data values are integers). Its
interpretation and use is the same as for continuous data. For each observed discrete
Vll
-------
(integer) value a horizontal line segment is drawn at the estimated percent value. The
confidence limits are plotted similarly.
Empirical Cumulative Distribution Estimate
TO
0>
55
"o
to
Q.
100 -
80 -
60 -
40
20
0 -
CDF Estimate
95%Confidence Limits
- 300000
- 250000
- 200000 ~
- 150000 £
100000
- 50000
10
20
Response Variable
30
40
An example Empirical Density estimate is shown below. An empirical density is similar
to a smoothed histogram. For example, if the data were from a normal distribution, then
the empirical density would appear "bell-shaped." The purpose for including it is to aid
the reader in determining the "shape" of the distribution. It serves no other purpose.
The left plot below is an example for a continuous variable. It illustrates a skewed
distribution similar to a log normal distribution. The plot on the right is an example for a
discrete distribution, with a shape that is more "bell-shaped". No vertical axis is given
as the plots are scaled to have the total area (continuous) or total height (discrete) equal
to 1.0, so that they reflect a probability distribution.
i
10
!
20
i
30
!
.10
i
50
i
60
i
10
i
20
Vlll
-------
Quality Assurance
A comprehensive quality assurance (QA) program was developed and implemented for
EMAP-West. The principal QA-related activities implemented for various data
acquisition components of the study are presented in Table QA-1. These activities are
part of a systematic approach to ensure that
1. Collection and measurement procedures were standardized among all
participants in EMAP-West
2. Statistical control of measurement systems were maintained and feedback was
provided so that corrective actions were taken when necessary
3. The performance of measurement systems was assessed periodically
4. Data were reviewed and validated to be sufficiently representative, accurate,
precise, and complete for their intended use of developing appropriate indicators,
defining reference condition, and integrating these with an appropriate
probabilistic monitoring design to estimate ecological condition of streams in the
western U.S.
Documentation included a "programmatic" quality assurance project plan (QAPP)
developed for use with all E MAP-Surf ace Water research activities. Laboratory QAPPs
were developed specifically for chemical analyses of stream water and fish tissue, and
for benthic macroinvertebrates in fulfillment of EPA contract requirements. Field
protocols and other activities were documented in two field operations manuals (one for
wadeable streams, and one for non-wadeable streams and rivers), and each support
laboratory developed standard operating procedures (SOPs) for laboratory methods.
QA activities associated with the survey design and population estimation analysis
focused on accounting for the sampling status of all sites selected in terms of whether
or not they were sampled, and if not, why not. This was necessary to provide accurate
estimates of resource extent. The design included small sets of random samples,
termed "partitions". All sites in a partition that was used for field sampling had to be
accounted for to allow for proper calculation of weighting factors used in population
estimation. A subset of sites was selected each year as "revisit sites". Each revisit site
was attempted to be visited a total of four times (two times each in two successive
years). Data from revisit sites were used as part of the metric evaluation process for
various indicators, and to quantify various components of variability that affect either
status or trend estimates.
All field crew personnel participated in a standardized field training session, held in
different locations each year within the study area. Field trainers were experienced
EMAP principal investigators from the Western Ecology Division. Each training session
was 3-4 days, and included lectures, field demonstrations, and at least one practice field
exercise. The field operations manual served as the basis for the field training program.
Each field crew was visited at least once during the project by an experienced EPA
Regional person who had completed the field training program to ensure the protocols
were being implemented correctly and address questions the crew had regarding the
QA-1
-------
protocols. Field crews were offered an opportunity at the end of each year to suggest
improvements to the field operations manual and other aspects of field operations.
All water and fish tissue contaminant samples were shipped from the field to a single
laboratory (EPA Western Ecology Division, Corvallis, OR) for analysis. The laboratory
participated in the inter-laboratory performance evaluation (PE) program developed by
Environment Canada National Water Research Institute (NWRI) throughout the duration
of EMAP-West. Two PE studies were conducted each year, each consisting of 20
samples representing a range of surface water types and analyte concentrations, plus
10 additional samples that were analyzed for total phosphorus. Further details of the
QA activities related to water chemistry, and a summary of laboratory performance in
the NWRI studies, can be found in the Water Chemistry section of this report. For
analyses of metals in fish tissue contaminant samples, a Standard Reference Material
(SRM; DORM-2 dogfish obtained from the National Research Council of Canada) was
analyzed with every batch of samples. Results from the SRM analyses are summarized
in the Fish Tissue Contaminant section of this report (Tables FT-1 and FT-2).
The principal QA activities for the physical habitat indicator included an extensive
presentation on the collection of physical habitat data, including photographs of different
conditions expected to be encountered. Physical habitat data from the field data forms
were then subjected to a systematic, automated review process to produce validated
data files of the correct structure for calculating metric variables. In addition, the
accuracy of field crew identifications of invasive plants was assessed in two states
(Oregon and Montana) by having a separate field crew comprised of experienced field
botanists visit sites at a different time.
QA-related activities associated with the aquatic vertebrate assemblage indicator
included sampling an extended sampling reach (equal to three times the normal length,
or 300 times the mean channel width) at 1-2 non-wadeable sites each year. This
"oversampling" effort provided additional information regarding the sufficiency of the
sampling reach length in obtaining a representative sample of the aquatic vertebrate
species present. To ensure the accuracy of field identifications, voucher specimens of
aquatic vertebrates were obtained where allowed by scientific collecting permits and
sent to the National Museum of Natural History (part of the Smithsonian Institution) for
confirmation of the field identifications and archival in their permanent collection.
QA activities implemented for the benthic invertebrate assemblage indicator were
focused on obtaining a sufficient sample in the field, on consistent processing of
samples at the laboratory, and on taxonomy-related issues both within and among
laboratories. Two invertebrate laboratories were involved in EMAP-West, and they
collaborated closely on issues of taxonomic nomenclature and level of taxonomic
resolution to minimize compatibility problems in the final benthic database. Taxonomic
names were based on (or cross-referenced to) existing names in the Integrated
Taxonomic Information System (ITIS) maintained by the U.S. Geological Survey. One
or more suitable facilities will be identified for taxonomic reference specimens to be sent
to for archival in permanent collections.
All of the various data components of EMAP-West were managed at the Western
Ecology Division-Corvallis. A scanner based system was used to develop standardized
QA-2
-------
field data forms. Completed forms were scanned and fields the software could not
recognize were presented onscreen to the operator, who either made corrections from
the original field form or assigned a flag to the field for later reconciliation by a principal
investigator. After initial review, the data were exported into various data files. These
files, and data files from the various laboratories, were imported into a centralized
Surface Water Information Management system (SWIM). Data files were verified and
validated by various principal investigators prior to calculating metrics and indicator
variables. The SWIM system tracked changes to files as they were updated, and
provided a means for principal investigators and other project participants to access and
download data files for validation or data analysis activities.
QA-3
-------
Survey Design
Description of Study Requirements
The primary objectives of this study are to estimate (1) the extent (length) of perennial
and non-perennial streams and rivers, and (2) the condition of perennial streams and
rivers in conterminous states of EPA Regions 8, 9, and 10. The twelve states included
are Arizona, California, Colorado, Idaho, Montana, Nevada, North Dakota, Oregon,
South Dakota, Utah, Washington, and Wyoming. The target population of perennial and
non-perennial streams and rivers is defined by those present on the digital 1:100,000
scale U.S. Geological Survey (USGS) hydrologic maps that were incorporated into
EPA's River Reach File (Version 3). All or the lower portions of the Columbia, Snake,
Missouri, and Colorado Rivers are excluded. All of the Columbia River is excluded as
is the Missouri River from its beginning at Three Forks. The Snake River is excluded
below the Palisades Dam in Idaho to its confluence with the Columbia. The Colorado
River is excluded from Eagle River until it leaves the United States.
To address the two objectives, two integrated surveys were conducted: a non-perennial
survey and a perennial survey. Both surveys are used to provide information for the
extent estimation objective. Only the perennial survey is used to provide information for
the condition estimation objective.
Specific extent objectives are to estimate:
• Total stream and river channel length in the study region, by EPA Region, and by
state.
• Total perennial stream and river channel length in the study area, by EPA
Region, and by state.
• Total non-perennial stream and river channel length in the study area, by EPA
Region, and by state.
Specific condition objectives are to estimate the condition of streams and rivers for the
following subpopulations of perennial streams and rivers:
• All perennial streams and rivers (except the Columbia, Snake, Missouri, and
Colorado) in the study area.
• All wadeable perennial streams within each of Regions 8, 9, andIO within the
study area.
• All non-wadeable perennial rivers within each of Regions 8, 9, and 10 within the
study area
• All wadeable perennial streams within each of the 12 states in the study area.
• All perennial streams within the Upper Missouri Basin within Region 8.
• All perennial streams within the northern California coastal drainage in Region 9.
• All perennial streams within the southern California coastal drainage in Region 9.
• All perennial streams and rivers within the Deschutes and John Day Hydrologic
Units in Oregon in Region 10.
• All perennial streams and rivers within the Wenatchee Hydrologic Unit in
Washington in Region 10.
DE-1
-------
• All perennial streams and rivers in the Northern Glaciated Plains (Omernik Level
III ecoregions, January 1999) within North Dakota and South Dakota.
• All perennial streams and rivers in the Colorado Plateaus Ecoregion (Omernik
Level III ecoregions, January 1999) within Region 8
Figure DE-1 illustrates the 12-state study region and the geographic coverage for the
seven intensive study regions. For administrative reasons only, some of the intensive
studies are designated as REMAP studies and others as Special studies. Operationally
they are treated the same. REMAP identifies studies that are funded by the Regional
EMAP project within EMAP.
In the rest of this section the term "streams" refers to both streams and rivers.
Description of the Sampling Frame
The sampling frame comes from U.S. EPA's River Reach File 3 - Alpha (RF3) and the
USGS PNW River Reach File, both of which are based on digitized blue lines from
1:100,000 scale maps (Horn and Grayman, 1993). Based on prior information, it is
known that RF3 incorrectly codes some stream segments. Incorrect code information
occurs for (1) designating Strahler stream order; (2) delineating perennial and
intermittent, (3) defining natural versus constructed channels, including newly modified
channels, and (4) distinguishing irrigation return flow from irrigation delivery channels.
In some cases, RF3 includes stream channels that are not actually present, due to (1)
no definable channel present, (2) location is wetland/marsh with no defined channel, or
(3) channel may be an impoundment. RF3 may also exclude some stream channels
due to (1) mapping inconsistencies in construction of 1:100,000 maps, (2) digitization of
map blue lines, or (3) inadequacy of photo information used to develop maps, e.g.
heavily forested areas with low order streams. This study assumes that RF3 includes
all stream channels specified by the definition of the target population. That is, if stream
channels exist that are not included in RF3, they are not addressed by this study.
The sampling frame includes all RF3 stream channel segments coded as R, S, T, N, W,
and U in RF3 for Regions 8 and 9, or stream channel segments coded as 412, 413,
414, 415, and 999 in Region 10. As stated above all or portions of the Columbia,
Missouri, Snake, and Colorado are excluded from the sampling frame.
The sampling frame is subdivided into two major parts: (1) all RF3 stream segments
coded as perennial (RF3 perennial) and (2) all RF3 stream segments coded as non-
perennial, i.e., all other stream segments (RF3 non-perennial). The purpose of
subdividing the sampling frame is to use one survey design for RF3 non-perennial
streams and another survey design for RF3 perennial streams.
An additional concern with the sampling frame arises with rivers. Past experience has
indicated that Strahler order calculated from RF3 can be incorrect for some higher order
stream segments. This appears to be mainly due to "breaks" in the network as
delineated by RF3, which results in middle portions of the rivers being incorrectly coded
as 1st order (or other low stream orders). To alleviate this coding issue, we explicitly
constructed a list of all rivers with drainage basins greater than 12,950 km2.(5,000 mi2).
All river segments in RF3 associated with these "large rivers" are designated in the
sampling frame as large rivers and are assumed to be perennial. Although it would be
DE-2
-------
desirable to validate all rivers, this was not possible within the time and resource
constraints of the study. Note that no changes in the RF3 stream segments or
associated codes are made. All other rivers with smaller drainage areas are included in
the survey but are selected by Strahler order as coded in RF3.
Survey Design Description
The survey designs are different for the RF3 non-perennial and the RF3 perennial
streams and rivers. That is, the overall survey design consists of two strata: RF3 non-
perennial and RF3 perennial streams and rivers. Both strata are used for estimation of
extent and only the RF3 perennial stratum is used for estimation of condition.
Omernik Level III ecoregions (revised January 1999) are aggregated into two
categories: Mountainous/Humid and Arid. The purpose of the aggregation is to adjust
for an expected difference in miscoding by RF3 of perennial streams and to assure that
sites are selected from mountainous/humid and arid regions of a state. Definitions of
the categories are given in Table DE-1 and shown in Figure DE-1.
RF3 Non-Perennial Stream Survey
The RF3 non-perennial survey is solely connected with the objectives of estimating the
extent of the stream resource in the study area. The survey design is stratified by the
twelve states. Within each state an unequal probability, spatially-balanced sample was
selected (Stevens and Olsen, 2004). Unequal selection occurs by Strahler order
categories (1st, 2nd, and 3rd and higher) and by mountainous/humid and arid Omernik
Level 3 ecoregion groups (note that these 2 aggregations of Omernik ecoregions,
illustrated in Figures DE-1 and DE-2, are used only in the design of EMAP-West, and
are not the same as the three climatic/topographic regions used for reporting results).
Strahler 2nd order streams are selected with 3 and 5 times the probability of 1st order
streams for arid and mountainous/humid ecoregions, respectively. Similarly, Strahler 3rd
and higher-order streams are selected with 6 and 20 times the probability of 1st order
streams for arid and mountainous/humid ecoregions, respectively. The unequal
selection ensures that the number of site-evaluations on 2nd and higher order streams
and mountainous/humid regions is sufficient to estimate proportions of perennial and
non-perennial streams that RF3 categorizes as non-perennial. Table DE-2 summarizes
RF3 non-perennial streams and rivers by state, Strahler order category, and arid/humid
ecoregion. Table DE-3 summarizes the number of sites by state, Strahler order
category, and arid/humid ecoregion. Figure DE-2 shows their spatial distribution.
RF3 Perennial Stream Survey
The RF3 Perennial survey explicitly stratifies by the 12 states and within each state
uses an unequal probability, spatially-balanced survey design (Stevens and Olsen,
2004). Unequal probability categories are defined by Strahler order categories (1st,
2nd, 3rd, >4th, and large river) and by Humid and Arid aggregated ecoregions (Table
DE-1). Allocation of sites by order category gives an expected sample sizes resulting in
an equal number of sites for categories 1st, 2nd, 3rd, >4th order, and 120 sites for large
rivers. The expected sample size for the basic survey design is 50 sites per state, for a
total of 600 unique sites to be sampled across the study region. In addition to the basic
DE-3
-------
survey design for each state, five intensive studies are incorporated in the survey
design. Each EPA Region identified one or more intensive or REMAP studies. The
studies are a northern California coastal special study, a southern California REMAP
study, a Deschutes/John Day special study in Oregon, a Wenatchee basin REMAP
study in Washington, an Upper Missouri River Basin special study, and a Colorado
plains intensive study.
Table DE-4 summarizes the stream length (km) from RF3 that is coded perennial. The
twelve state total stream length is 628,625 km. Prior experience in Oregon and limited
information from other experiences in the West suggest that the miscoding of perennial
stream channels varies by stream order and by Humid/Arid aggregated ecoregion.
Landowner access denial may average approximately 20%. However, in the Central
California Valley, Hall et al (1998) report approximately 18% access denial and 33%-
46% no response from landowners. We combined these two sources of information to
estimate the percentage of sites expected to be available for field sampling (Table DE-
5). The final probability of selection is adjusted to incorporate these expected non-
accessible rates (Table DE-6). For example, rather than selecting an equal number of
sites by Strahler order category additional sites are selected for lower order Strahler
order categories with the expectation that the final set of sampleable sites would be
approximately equal.
Note that the site selection process also includes an over sample of sites that are
available for use if the prior estimates do not result in the base sample meeting the
sample size requirements. The over sample size was the same as the expected
sample size for each state. If the over sample sites were insufficient, then additional
over sample sites were selected until sufficient sampleable sites were found. Only
Arizona required additional over sample sites. The over sample sites are given in a
specified order to ensure that when they are added that the spatial-balance of the
survey design is preserved. Stevens and Olsen (2004) describe the reverse
hierarchical ordering process that makes this possible.
The five intensive study regions are incorporated by increasing the probability of
selection of streams within the study region to achieve the expected sample size for that
region. The same unequal probability selection is applied as for the state-wide sample.
For the Upper Missouri River Basin study, the expected sample size was allocated to
each of the four states in proportion to the Upper Missouri River Basin RF3 Perennial
stream length that occurred in each state. This is necessary since each state is a
separate stratum. Table DE-7 summarizes the expected sample size and realized
sample size by state. Expected sample sizes reported are the 50 state-wide sites
allocated to each state and the additional sites allocated for the five intensive studies
that include parts of a state. A total of 1035 sites were planned to be sampled.
Realized sample sizes reported are for the state-wide sample (assuming the intensive
studies had not occurred) and the total number of sites sampled in the intensive studies
(which includes state-wide sites as well as the additional intensive sites).
Consequently, some sites are counted in both the state-wide and intensive realized
sample size columns. The total realized sample size is the total number of unique sites
sampled within a state.
DE-4
-------
In addition, the design incorporates a survey over time panel structure that allocates an
equal number of samples for visits by year (2000-2003). The survey over time is
intended to allocate approximately an equal number of sites to be visited each year
within each state. This ensures that all four years are equally represented in the final
sample. It also provided a mechanism to terminate the study early if budget reductions
occurred. Each annual panel is a probability sample with the same properties as the
original design except with one-fourth the sites. Consequently, any combination of the
four panels is a probability sample, e.g., the study could have stopped after the first
three panels were completed. Note that due to field implementation difficulties, some
sites were not necessarily visited within the year planned, i.e., all panel 1 sites were not
visited in year 1. All sites within each panel were visited during the study so that the
resulting sites are a probability sample. No consequences are expected since most
sites are visited within the year planned.
Figure DE-3 shows the spatial pattern of all sites (3228) evaluated in the study. All
evaluated sites are used to estimate the extent of perennial streams. Note that some
states evaluated many more sites than subsequently required for field sampling. Figure
DE-4 shows the subset of sites valuated that were further investigated for potential field
visits (2342) and are used in estimating condition. Figure DE-5 shows the final subset
of sites that were perennial and were successfully sampled (965). Table DE-8
summarizes by state Evaluated sites, which are used for extent estimation and Used
sites, which are used for condition estimation. Approximately 42% of the Used sites are
non-target. Most non-target sites are non-perennial streams. The remaining non-target
sites are canals, ditches, impoundments, wetlands, tidal streams or non-existent stream
channels. Approximately 12% of the Used sites are sites where landowners denied
access or could not be contacted to acquire access. Approximately 5% of the Used
sites used could not to be physically accessed, mainly for safety reasons.
Estimates in the Statistical Summary are made for two sets of aggregated Omernik
Level 3 ecoregions (Figure DE-6). Two general criteria are used to define the
aggregated ecoregions: number of sites sampled and ecological similarity. Table DE-9
summarizes the number of sites by these ecoregions and Figure DE-7 shows the spatial
distribution within aggregated ecoregions.
Variance Component Study
The survey design includes a plan to revisit a subset of sites. The objective is to
estimate four sources of variability (see Kincaid et al. (2004) for a lake example). The
sources of interest are (1) population variation: site-to-site, (2) year variation that is
coherent that affects all sites, (3) site-by-year interaction: year-to-year site variation not
accounted for by the common year variation across sites, and (4) residual variation:
remaining variation which includes measurement error, analytical error, field crew
variation, and temporal variation within the index period. The index period is the
interval within a year when sampling is to be completed. A site-visit consists of a single
visit to a stream site and completion of a suite of field evaluations to assess condition of
the stream channel. Approximately, 10% of the total available site-visits for the study
were allocated to variance component estimation. The sites selected for revisit are
distributed according to Table DE-10. The sampling scheme for revisit sites is to
DE-5
-------
sample the site twice within the index period in two consecutive years, i.e., the site is
visited four times. Approximately 12-13 sites were revisited each year. The four revisits
at a site are necessary to estimate all four components of variation. This report does
not include results for the variance component study.
Precision for Estimates
The RF3 Perennial Survey objective is the estimation of proportions of stream length
with a specified characteristic, e.g., proportion of streams with a benthic
macroinvertebrate IB I score less than 50. Another objective is the estimation of total
stream length based on both the RF3 Perennial Survey and the RF3 Non-Perennial
Survey. Although the total number of samples in the two studies is constrained by the
available budget, the number of special study regions and sub-objectives of the study
are not only constrained by budget decisions but also by consideration of the precision
that could be expected. Approximate precision estimates for proportions can be
obtained by assuming the survey designs are simple random samples. Under this
condition the estimated confidence interval half-width (precision) can be estimated using
procedures given by Cochran (1987) for proportions. Given the survey designs are
actually based on the spatially-restricted survey designs described by Stevens and
Olsen (2004), the actual precision estimates are expected to be better (smaller
confidence intervals) than those stated below.
The confidence interval half-width (precision), as a percent, is determined from
Half-width = Zt.a * 100 * Sqrt[ p(1-p)/n]
To calculate precision requires knowledge of p, the proportion to be estimated.
However, a conservative estimate of precision can be obtained by assuming p to be 0.5,
which gives the maximum variance. Z^ is related to the level of confidence required
for the estimate. Table DE-10 gives the expected half-width of confidence intervals for
selected sample sizes and two alternative true proportions.
Each state was allocated a minimum of 50 samples. If the true proportion is 20% and
precision required is 95%, then the expected precision (confidence interval half-width) is
±11 %. Estimates based on the base state-wide part of the RF3 Perennial survey are
based on 600 sample sites. For proportions that are 50%, this results in an estimated
precision of ±3-5%.
Information from a total of approximately 150 to 200 sites per state is available to
calculate extent, 50 tolOO from RF3 perennial survey and 100 from RF3 non-perennial
survey. At 95% confidence and assuming proportion of 0.5, this gives an estimated
precision of approximately 7-8% for extent estimates for each state.
Statistical Analysis of Survey Data
The study uses a stratified, spatially-balanced probability survey design with unequal
probability of selection within strata. The objective is to estimate the empirical
cumulative distribution (percent and stream length), percentiles, and means for stressor
and condition indicators. To calculate these estimates, the statistical analysis must
DE-6
-------
incorporate the information about the survey design as well as the indicator values from
the sites sampled. The purpose of this section is to describe the statistical analysis
process.
The following steps are essential to the statistical analysis: (1) compiling evaluation
status for each site in the study, (2) adjusting the survey design weights, (3) estimating
the extent of perennial and non-perennial stream length, and (4) estimating the mean,
cumulative distribution, and percentile values for all indicators. These steps are
described in what follows.
The statistical analyses utilize the R statistical software (R Development Core Team
2004) and an R contributed library, psurvey.analysis (http://www.epa.gov/nheerl/arm),
developed specifically for the statistical analysis of probability survey design data.
Compiling site evaluation status
Information from the site evaluation is to estimate the extent of perennial and non-
perennial stream length in the study region. It is also used to estimate the extent of
stream length associated with access denial by landowners and physically inaccessible
streams.
Adjusting survey design weights
The survey design assigns a weight to each stream site selected for potential sampling.
These weights must be used in the statistical analyses. The weights are in units of
kilometers of stream length, e.g., a weight of 2.28 means that the sampled site
represents 2.28 kilometers of stream length. The weights differ by State, Stralher order,
and ecoregion categories used in the survey design. The initial weight assignments
assume that the survey will be implemented as planned. Rarely is a design
implemented exactly as planned. For example, suppose that a design has a sample
size of 1,000 sites and that the decision is made to return from the field with 1,000 sites
actually sampled. It may be necessary to evaluate 3,200 sites to identify 1,000 stream
sites that result in a field sample. The remaining 2,200 are sites that are non-target or
where landowners denied access, were physically accessible, or could not be sampled
for other reasons. The initial weights are based on an assumed sample size of 1,000
rather than the actual sample size of 3,200. Consequently, the weights must be re-
calculated, i.e., adjusted to account for the evaluation of 3200 stream sites rather than
the initial plan to evaluate 1000 stream sites.
The study plan states that when an additional stream site is required in a state, the next
stream site in the over sample list of stream sites will be used from that state. Under
this provision, the weight adjustment is completed by state.
Estimating extent of stream length
Data for estimating the extent of perennial and non-perennial stream length is the
evaluation status recorded for all stream sites evaluated for potential field sampling.
Cochran (1987) gives the statistical procedure for estimating a total from an unequal
probability sample. The local neighborhood variance estimate for the total is given by
DE-7
-------
Stevens and Olsen (2003). Both of these procedures are available in the
psurvey.analysis library. In addition to study region estimates, estimates can be made
by aggregated Omernik Level III ecoregions (e.g., Figure DE-1). Although an estimate
can be made for any sub-region, unless the sample size is sufficiently large, the
confidence intervals for the estimates may be so large that the estimate has high
uncertainty.
Empirical cumulative distribution estimation
Many measured variables and calculated metrics are available for statistical analysis
from the study. For this report, the mean, percentiles, and cumulative distribution for
each variable and metric are estimated. Note that these estimates apply to the entire
"population" of streams within each of the geographic regions reported. Since the
survey design is a stratified, unequal probability design, the statistical estimation must
account for the stratification and unequal probability of selection. This is done by
utilizing the weights associated with each stream site. The weight represents the
amount of stream length (km) that each site represents. The sum of the weights within
each stratum equals the total stream length within each stratum, i.e., state. Cochran
(1987) gives the equations for estimating the mean and Diaz-Ramos et al. (1996) give
the equations for estimating the cumulative distribution (Estimation Methods 1 and 2).
The percentiles are interpolated from the estimated cumulative distribution. The local
neighborhood variance estimator described by Stevens and Olsen (2003, 2004) is used
to calculate variance estimates for the mean and cumulative distribution. Confidence
intervals are calculated assuming the estimates are from a normal distribution for the
mean and cumulative distribution. Percentile confidence limits are interpolated from the
cumulative distribution confidence limits. Plots of the empirical cumulative distribution
estimates plot the entire range for the estimate but only the confidence limits for the
percent of stream length between 5% and 95%. Note that the confidence limits are for
the estimated percent and not for the estimated stream length. The confidence limits for
stream length are wider since the total stream length must also be estimated.
Empirical density estimation
Another estimate for each measured variable or calculated metric is its empirical
density. The empirical density is similar to a "smoothed" histogram for a variable. It
would be "bell-shaped" if the data were similar to a sample from a normal distribution.
The empirical density is estimated using averaged shifted histograms as described by
Scott (1985). Scott's procedures are extended to use unequally weighted data, as
arises in unequal probability surveys. Conceptually, the estimate is constructed by (1)
creating a series of equal bin-size histograms that differ only in where the first bin starts.
For example, if five histograms are to be averaged with a bin-size of 1 and the first bin
being at 0, then the five histograms would start at 0, 0.2, 0.4, 0.6, and 0.8 keeping all
bin-sizes at 1. The five histograms are then averaged and the plot constructed by
connecting the average bin heights. The final density estimate is scaled so that the
area under the curve is 1.
DE-8
-------
References
Cochran, W. G. (1987). Sampling Techniques. New York, John Wiley & Sons.
Diaz-Ramos, S., D. L. Stevens, Jr, et al. (1996). EMAP Statistical Methods Manual.
Corvallis, Oregon, U.S. Environmental Protection Agency, Office of Research and
Development, NHEERL-Western Ecology Division, ISBN EPA/620/R-96/002.
Hall, R. K., P. Husby, et al. (1998). "Site access and sample frame issues for R-EMAP
Central Valley, California, stream assessment." Environmental Monitoring and
Assessment 15: 357-367.
Horn, C.R. and Grayman, W.M. (1993) Water-quality modeling with EPA reach file
system. Journal of Water Resources Planning and Management, 119, 262-74.
Kincaid, T. M., D. P. Larsen, et al. (2004). "The Structure of Variation and Its Influence
on the Estimation of Status: Indicators of Condition of Lakes in the Northeast, U.S.A."
Environmental Monitoring and Assessment 98(1-3): 1-21.
R Development Core Team (2004). R: A language and environment for statistical
computing. Vienna, Austria, R Foundation for Statistical Computing, ISBN 3-900051-07-
0, http://www.R-proiect.org.
Scott, D. W. (1985). "Averaged shifted histograms: effective nonparametric density
estimators in several dimensions." The Annals of Statistics 13(3): 1024-1040.
Stevens, D. L., Jr. and A. R. Olsen (2003). "Variance estimation for spatially balanced
samples of environmental resources." Environmetrics 14: 593-610.
Stevens, D. L., Jr. and A. R. Olsen (2004). "Spatially-balanced sampling of natural
resources." Journal of American Statistical Association 99(465): 262-278.
DE-9
-------
Tables
Table DE-1 Aggregated arid and mountainous/humid ecoregions defined by
Omernik Level III ecoregions (revised January 1999)
Mountainous/Humid Ecoregions
Arid Ecoregions
1. Coastal Range
2. Puget Lowland
3. Willamette Valley
4. Cascades
5. Sierra Nevada
9. Eastern Cascades
11. Blue Mountains
15. Northern Rockies
16. Montana Valley/foothill
17. Middle Rockies
19. Wasatch and Uinta Mountains
21. Southern Rockies
41. Canadian Rockies
77. North Cascades
78. Kalamath Mountains
6. Southern/Central Calif
7. Central Calif Valley
8. Southern Calif Mountains
10. Columbia Plateau
12. Snake River Basin
13. Northern Basin and Rang
14. Mojave Basin and Range
18. Wyoming Basin
20. Colorado Plateaus
22. Arizona/New Mexico Plateau
23. Arizona/New Mexico Mountains
24. Southern Deserts
25. Western High Plains
26. Southwestern Tablelands
42. Northwestern Glaciated Plains
43. Northwestern Great Plains
44. Nebraska Sand Hills
46. Northern Glaciated Plains
47. Western Corn Belt Plains
48. Lake Agassiz Plain
79. Madrean Archipelago
80. Snake River High Desert
81. Sonoran Basin and Range
DE-10
-------
Table DE-2 Stream and river length (km) in RF3 non-perennial sampling frame by
state, Strahler order category, and arid/humid Omernik Level 3 ecoregions
State
Arizona
California
Colorado
Idaho
Montana
North Dakota
Nevada
Oregon
South Dakota
Utah
Washington
Wyoming
Total
Arid Ecoregions
1st
153,949
151,865
59,554
49,018
87,964
58,682
157,077
39,820
91,255
69,209
28,042
81,485
1,027,919
ond
31,093
25,939
14,605
6,729
25,588
15,811
33,660
7,552
24,791
14,939
5,157
18,275
224,138
3rd+
20,046
15,608
10,103
3,815
18,270
9,429
18,196
5,053
19,347
8,879
1,769
12,334
142,849
Humid Ecoregions
1st
0
26,044
27,427
19,419
55,217
0
221
33,992
4,318
14,938
15,011
14,584
211,172
2nd
0
2,630
3,212
1,579
7,033
0
25
4,413
982
2,031
1,221
1,789
24,916
3rd+
0
978
724
527
2,096
0
22
975
353
579
291
659
7,206
Total
205,088
223,063
115,625
81,087
196,168
83,922
209,201
91,806
141,046
110,574
51,492
129,127
1,638,200
DE-11
-------
Table DE-3 Sample size for RF3 non-perennial survey by state, Strahler order
category, and arid/humid Omernik Level 3 ecoregions
State
Arizona
California
Colorado
Idaho
Montana
Nevada
North Dakota
Oregon
South Dakota
Utah
Washington
Wyoming
Total
Arid Ecoregions
1st
44
43
30
38
27
47
33
26
37
34
34
40
433
ond
18
23
19
23
18
35
35
18
22
20
17
22
270
3rd+
38
20
22
15
22
18
32
14
35
36
13
26
281
Humid Ecoregions
1st
0
7
11
10
11
0
0
19
1
7
17
5
88
2nd
0
2
8
6
12
0
0
13
2
6
11
5
65
3rd+
0
5
10
8
10
0
0
10
3
7
8
2
63
Total
100
100
100
100
100
100
100
100
100
100
100
100
1200
DE-12
-------
Table DE-4 Stream and river length (km) in RF3 Perennial sampling frame by
state, Strahler order category, and arid/humid Omernik Level 3 ecoregions
State
AZ
CA
CO
ID
MT
ND
NV
OR
SD
UT
WA
WY
Total
Arid Ecoregions
1st
10,284
23,300
3,612
11,460
2,594
759
9,744
3,604
1,178
7,257
4,368
5,812
83,971
2nd
4,581
6,626
2,265
4,581
2,163
1,183
4,856
1,878
1,376
2,293
1,643
3,508
36,951
3rd
3,813
5,171
2,411
3,776
3,351
2,552
3,180
1,781
2,792
2,230
1,334
4,052
36,444
4th +
4,615
5,944
4,405
4,944
7,753
7,301
4,057
3,006
7,664
3,616
2,032
8,354
63,692
Large
River
799
925
1,184
1,228
2,576
1,403
458
657
2,324
927
260
891
13,631
Mountainous/Humid Ecoregions
1st
0
28,892
18,149
36,398
32,495
0
321
40,898
213
5,973
35,355
14,993
213,687
2nd
0
11,622
8,496
13,710
15,827
0
73
14,985
350
2,965
13,037
6,784
87,848
3rd
0
7,211
4,638
6,778
9,847
0
20
9,506
383
1,788
7,195
3,955
51,322
4th+
0
5,525
3,740
4,242
7,629
0
26
9,063
290
1,128
4,410
2,415
38,468
Large
River
0
553
46
615
425
0
0
689
18
66
200
0
2,609
Total
24,093
95,769
48,945
87,732
84,660
13,197
22,735
86,067
16,587
28,242
69,834
50,764
628,625
Table DE-5 Percentage of RF3 perennial stream sites expected to actually be
perennial and accessible.
Strahler Order
Category
1st
2nd
3rd
>4th
Validated rivers
Humid
Ecoregion
65%
75%
80%
100%
100%
Arid
Ecoregion
30%
50%
50%
90%
100%
DE-13
-------
Table DE-6 Unequal probability multipliers to achieve expected sample sizes.
Strahler Order
Category
1st
2nd
3rd
>4th
Validated
rivers
Humid
Ecoregion
1.55
1.33
1.25
1.00
1.00
Arid
Ecoregion
3.33
2.00
2.00
1.10
1.00
Table DE-7 Expected sample sizes by state and intensive study for the RF3
perennial survey
State
Arizona
California
Colorado
Idaho
Montana
Nevada
North Dakota
Oregon
South Dakota
Utah
Washington
Wyoming
Total
Expected Sample Size
State-Wide
50
50
50
50
50
50
50
50
50
50
50
50
600
Intensive
100
25
87
12
100
24
50
37
435
Total
50
150
75
50
136
50
94
150
102
50
100
87
1035
Realized Sample Size
State-Wide
47
50
51
48
49
51
54
61
55
55
49
49
619
Intensive
132
22
52
30
95
74
55
42
502
Total
47
169
67
48
69
51
63
146
76
55
100
75
966
DE-14
-------
Table DE-8 Number of sites evaluated, used, and sampled by state
State
Arizona
California
Colorado
Idaho
Montana
Nevada
North
Dakota
Oregon
South
Dakota
Utah
Washington
Wyoming
Total
Evaluated
Sites
384
528
203
138
198
208
204
424
221
168
290
262
3228
Used Sites
Total
384
475
125
108
124
106
151
329
99
132
186
123
2342
Sampled
47
169
67
48
69
51
63
146
76
55
100
74
965
No access
by
landowner
24
62
22
8
19
5
4
88
8
1
18
13
272
Physically
Inaccessible
4
49
0
14
8
2
1
2
0
5
27
3
115
Non-
Target
309
195
36
38
28
48
83
93
15
71
41
33
990
DE-15
-------
Table DE-9 Number of sites sampled by aggregated Omernik Level 3 ecoregions
Aggregated
Ecoregion
Mountain
Plains
Xeric
Total
MT-NRock
MT-PNW
MT-SRock
MT-SWest
PL- NCultivated
PL-Range
XE-CALIF
XE-EPLAT
XE-NORTH
XE-SOUTH
Total
Evaluated
for Extent
1480
647
1101
3228
555
577
157
191
230
417
189
300
202
410
3228
Used for
Condition
1109
383
850
2342
394
437
96
182
123
260
168
225
138
319
2342
Classification of Used Sites
Sampled
574
190
201
966
210
227
60
77
66
124
34
71
49
47
965
No access
by
landowner
160
39
73
272
67
67
7
19
8
31
19
13
28
13
272
Physically
Inaccessible
97
5
13
115
18
57
4
18
1
4
5
2
3
3
115
Non-
Target
277
150
563
990
99
86
24
68
49
101
110
139
58
256
990
Table DE-10 Number of site revisits by Strahler order category
Strahler Order Category
1st
2nd
3rd
>4th
Large rivers
Total
Number of First-Visit Sites
Humid
4
4
4
4
4
20
Arid
4
4
4
4
4
20
Total
8
8
8
8
8
40
Number of
Site Revisits
24
24
24
24
24
120
Total
Site-visits
32
32
32
32
32
160
DE-16
-------
Table DE-11 Confidence interval half-widths for 90 and 95 percent confidence
intervals for sample sizes of 25, 50, 100, 400, and 1000 for two assumed true
proportions.
Assumed
True
Proportion
20%
50%
90% Confidence Level
n=25
±13
±17
n=50
±9
±12
n=100
±7
±8
n=400
±3
±4
n=1000
±2
±3
95% Confidence Level
n=25
±16
±20
n=50
±11
±13
n=100
±8
±10
n=400
±4
±5
n=1000
±3
±3
DE-17
-------
Figures
i i i Upper MssoLl Rksr Easn
I I I Cclcraio PI; rj
I I I Northern California Coastal Dranage
SOLPIE-TI C-aircrria Coas-.al Drainage
Wsratefee R.'ver Sssln
Day River Easns
Figure DE-1. EMAP-West study area, with special interest areas highlighted.
Areas shaded grey are mountainous/humid aggregated ecoregions; arid
ecoregion areas are unshaded.
DE-18
-------
Figure DE-2. EMAP-West study area, with location of all evaluated non-perennial
sites (n = 1200).
DE-19
-------
i i i Upper M«SOL.I RK'=r Eai T
i i i Colorado Plans
I I I Ncrthgm Caltamla Coasta C-ansgs
I I I ScLtnen carsrria Casual Drainage
I I I Weralchee River Bffiln
OKchJteE'Jcflr Day Kh'sr Eaj is
Figure DE-3. EMAP-West study area, with location of all evaluated perennial sites
(n = 3228).
DE-20
-------
i i i Upper MSSSOLI Rksr Eas i
i i i Csicrado Plans
I I I Nalhsni Calfamla Coastal Drainage
Scttiem carsrr'.a Coaxal Drainage
I I I wsralcr'se R'
-------
i i i Upper MSSSOLI Rksr Eas i
i i i Csicrado Plans
I I I Nalhsni Calflamla Coastal Drainage
I I I Scttiem Carsrr a Coaxal Drainage
I I I wsralcr'se R'
-------
Aggregate Ecoregion Assessment Regions'
Western Forested Mountains
I I D^uirweMir Mcurolrs
CH Notnem Rockies
l l s
Bcfllc NortnweEt
Great Plains
CH Cu-vale-d Nolhsrr =lalrs
I I ^angsland Plains.
Xeric West
CH Kornem Xfiric- 53£lrs
CH Esjtrern Xerc Basins
l l E3s:sn xsr! PIS-.SSJB
Figure DE-6 Map of three scales used in reporting EMAP West results: (1) All of
EMAP West study area (12 states); (2) 3 major climatic/topographic regions
(Mountains, Plains, Xeric); and (3) 10 aggregate ecological regions.
DE-23
-------
Aggregate Ecoregion Assessment Regions'
Western Forested Mountains
I I 33uir*K:arr wojnr.alrs
CH No-tnem Rocfclss
l l 5:uirerfT RccKles
Xeric West
I I isomeTi Xe^c 5aslrs
I I Scutr-eni Xsrc Bssins
l l Ess:=n Xer: PIS:;JJS
O >:e1: Calm:T> a Lo.vards
Great Plains
I I CuKivaled Northern =lalrs
l l ^an^lana Pains
.:•.:•. 15
Figure DE-7. EMAP-West study area, with all scales of reporting units (West-
wide, 3 climatic/topographic regions, 10 aggregate ecoregions) and locations
sampled perennial sites.
DE-24
-------
Reference Condition
While the primary purpose of this report is to show the statistical distribution of key
ecological variables (ecological condition indices and stressors) at multiple scales,
development of many of the metrics and indices we report on require reference
condition information in their development and interpretation. For example, the metrics
and multimetric indices we report for macroinvertebrate and fish assemblages use sets
of least-disturbed (reference) and most-disturbed sites to evaluate those metrics and
indices. The predictive (0/E) modeling we report for macro-invertebrates is based solely
on data from reference sites.
It is beyond the scope of this report to provide much background on the science and
concept of reference condition; comprehensive discussions of the concept can be found
in numerous published sources (Hughes 1995, Stoddard et al. In Press (2005)).
Reference Condition represents natural or pre-Columbian conditions in the USA;
reference sites represent the least-disturbed sites available, which in many cases are
markedly disturbed. For the purposes of this report, it is important for the reader to
know that we used least-disturbed sites chosen through the methods described below
to represent the best (= least disturbed) ecological condition in each of the 10 ecological
regions of the West (Figure RC-1). Inherent in this definition is the characteristic that
reference conditions for one part of the West (e.g., any of the mountainous ecological
regions) may be significantly less disturbed than those in another (e.g., either of the
plains ecological regions). For example, Whittier et al. (In Press) reported that the
most-disturbed sites in the xeric and mountain regions were disturbed to a similar
degree as the least-disturbed sites in the plains.
Candidate reference sites were selected from three different sources:
• Hand-picked sites from State or other monitoring programs, often chosen through
best-professional judgment (BPJ), and sampled by EMAP crews with EMAP
protocols.
• EMAP probability sites that passed numerous chemical and physical criteria
(below)
• Hand-picked sites identified through a CIS screening process (described in
(Lattin et al. 2005, Lattin In Preparation), verified through BPJ, and sampled by
EMAP crews with EMAP protocols.
In each case, candidate reference sites were carefully evaluated to assure that they
represented the least-disturbed set of sites in their ecological region. This evaluation
was carried out in three ways. Two sets of filtering criteria were developed according to
the description in (Waite et al. 2000); the criteria were developed independently (by two
individuals: Alan Herlihy and John Stoddard) and are listed in Tables RC-1 and RC-2.
Sites were required to pass a series of chemical and physical criteria, developed by
ecological region, to be considered least-disturbed.
A variant on the filtering approach, the goal of which is to identify least-disturbed sites
along key environmental gradients, was also used (Whittier et al. In Press). In the
Whittier approach, sites were evaluated relative to their position along natural
RC-1
-------
gradients—for example, those with the lowest total phosphorus concentrations relative
to their elevation and stream size are more likely to be identified as reference sites. The
physical and chemical variables used in each aggregate ecoregion, as well as the
environmental gradients for each region, are listed in Table RC-3.
The three approaches yielded slightly different lists of candidate reference sites. To
resolve these differences each candidate was evaluated according to how it was rated
by each method. If all three methods agreed that a given site was in least-disturbed
condition for its region, the site became a reference site for further analyses. If two of
three methods identified a site as least disturbed, the data were re-evaluated to
determine which criterion was violated. In general, sites were listed as least disturbed if
they violated only one criterion (for one method), particularly if the only violation was for
a catchment-scale variable (e.g., human landuse) versus a site-scale variable.
An analogous approach (implementing all three methods, and a resolution of
differences) was used to identify the most-disturbed sites in each region. Criteria used
to filter most-disturbed sites by the Herlihy and Stoddard approaches are listed in
Tables RC-4 and RC-5. The Whittier method is very similar to the approach described
earlier for least-disturbed sites, except that the most-disturbed sites (relative to their
position along environmental gradients) are identified. Candidate sites that were
identified as neither least-disturbed nor most-disturbed were put into an intermediate
disturbance category.
In the case of benthic macroinvertebrates, we also used a large reference site database
created by researchers at Utah State University (as part of the U.S. EPA STAR grant
program, and through funding from the U.S. Forest Service), referred to here as
STAR/R5BIO sites (Hawkins et al. 2003). Macroinvertebrate sampling methods for
these sites were identical to the targeted riffle sampling method described for EMAP
sites in Peck et al. (2005). The STAR/R5BIO sites were selected through a best
professional judgment process, and carefully evaluated in the field by trained crews.
These sites lacked sufficient data to implement any of the three EMAP approaches to
identify least-, intermediate-, and most-disturbed sites. Instead, we relied on the
rankings made by the STAR/R5BIO field crews—each site was assigned to one of four
disturbance categories ("Pristine", Minimally Disturbed, Least Disturbed, Disturbed;
Hawkins et al. 2003; Table RC-6). In this report, we used all four categories of sites in
the Plains ecoregions, but eliminated the Disturbed sites from the reference site list in
the Mountains and Xeric ecoregions.
The results of the various efforts to identify least-disturbed sites in the West are shown
in Figure RC-1. Approximately 230 reference sites were available for use in all indicator
analyses; for the macroinvertebrates, an additional ca. 500 sites were available from the
STAR/R5BIO programs, for a total of 730 reference sites.
RC-2
-------
References
Hughes, R. M. 1995. Defining acceptable biological status by comparing with reference
conditions. Pages Chapter 4, pg. 31-47 in W. Davis and T. Simon, editors.
Biological Assessment and Criteria: Tools for Water Resource Planning and
Decision Making for Rivers and Streams. Lewis, Boca Raton, FL.
Lattin, P. D. In Preparation. A process for characterizing watershed level disturbance
using orthophotos.
Lattin, P. D., L. McAllister, and P. Ringold. 2005. A multi-scale screening process for
identification of least-disturbed stream sites: Finding the best of what's left. EOS
Transactions Suppl. 86:NB13D-05.
Peck, D. V., A. T. Herlihy, B. H. Hill, R. M. Hughes, P. R. Kaufmann, D. J. Klemm, J. M.
Lazorchak, F. H. McCormick, S. A. Peterson, P. L. Ringold, T. Magee, and M. R.
Cappaert. 2005. Environmental Monitoring and Assessment Program - Surface
Waters Western Pilot Study: Field Operations Manual for Wadeable Streams.
EPA 600/R-OS/xxx, U.S. Environmental Protection Agency, Office of Research
and Development, Washington, DC.
Stoddard, J. L., D. P. Larsen, C. P. Hawkins, R. K. Johnson, and R. H. Morris. In Press
(2005). Setting expectations for the ecological condition of running waters: the
concept of reference condition. Ecological Applications.
Waite, I. R., A. Herlihy, D. P. Larsen, and D. J. Klemm. 2000. Comparing strengths of
geographic and nongeographic classifications of stream benthic
macroinvertebrates in the Mid-Atlantic Highlands, USA. Journal of the North
American Benthological Society 19:429-441.
Whittier, T. R., J. L. Stoddard, R. M. Hughes, and G. Lomnicky. In Press. Associations
among watershed- and site-scale disturbance indicators and biological
assemblages at least- and most-disturbed stream and river sites in the western
USA. in R. M. Hughes, L. Wang, and P. W. Seelbach, editors. Influence of
landscapes on stream habitats and biological assemblages. American Fisheries
Society, Bethesda, Maryland.
RC-3
-------
Tables
Table RC-1 Criteria used by Alan Herlihy to identify least-disturbed sites in each of 10 ecological regions of the
West
Herlihy Criteria:
MT-PNW
MT-NROCK
NSROCK
MT-SWEST
PL-RANGE
PL-NCULT
XE-CALIF
XE-NORTH
XE-SOUTH
XE-EPLAT
Total
Phosphorus
(M9/L)
<25
<25
<25
<50
<150
<150
<50
<50
<50
<50
Total
Nitrogen
(M9/L)
<750
<750
<750
<750
<4500
<4500
<1500
<1500
<1500
<1500
Chloride
(Meq/L)
<200
<200
<200
<300
<1000
<1000
<1000
<1000
<1000
<1000
Sulfate
(Meq/L)
<200
<200
<200
PH
<9
<9
<9
<9
<9
<9
<9
<9
<9
<9
Turbidity
(NTUs)
<50
<50
<25
<25
<25
<25
Riparian
Disturbance
(W1JHALL)
<0.5
<0.5
<1.0
<0.5
<2.0
<2.0
<1.5
<1.5
<1.5
<1.5
% Fines
<15%
<15%
<15%
<15%
<90%
<90%
<50%
<50%
<50%
<50%
Canopy
Density
(XCDENBK)
>50%
>50%
>50%
>50%
>25%
>25%
>50%
>50%
>50%
>50%
RC-4
-------
Table RC-2 Criteria used by John Stoddard to identify least-disturbed sites in each of 10 ecological regions of the
West
Stoddard Criteria:
Region:
MT-PNW
MT-NROCK
NSROCK
MT-SWEST
PL-RANGE
PL-NCULT
XE-CALIF
XE-NORTH
XE-SOUTH
XE-EPLAT
Total
Phosphorus
(M9/L)
<25
<25
<25
<50
<100
<200
<50
<50
<50
<50
Total
Nitrogen
(M9/L)
<750
<750
<750
<750
<1000
<2000
<1000
<1500
<1500
<1500
Chloride
(Meq/L)
<200
<200
<200
<300
<1000
<1000
<1000
<1000
<1000
<1000
Sulfate
(Meq/L)
<200
<200
<200
<2000
< 10000
< 10000
< 10000
PH
<9
<9
<9
<9
<9
<9
<9
<9
<9
<9
Turbidity
NTUs
<50
<50
<25
<25
<25
<25
Riparian
Disturbance
(W1JHALL)
<0.5
<0.5
<1.0
<0.5
<1.5
<1.5
<1.5
<1.5
Relative
Bed
Stability
(LRBS
BW5)
>-2.0
>-2.0
>-2.0
>-2.0
>-2.5
>-3.5
>-2.0
>-2.0
>-2.0
>-2.0
Canopy
Density
(XCDENBK)
or mean RBP
Habitat
(RH_XMET)
XCDENBK
>50%
XCDENBK
>50%
XCDENBK
>50%
XCDENBK
>50%
RH_XMET>12
RH_XMET>12
RC-5
-------
Table RC-3 Variables used in Whittier ranking approach to identifying least-
disturbed and most-disturbed sites in 3 aggregate ecological regions.
Natural Gradients'.
Chemical Variables:
Total Phosphorus
Total Nitrogen
Turbidity
Chloride
Sulfate
DOC
Habitat Variables:
% Fines
Riparian Disturbance
Natural Fish Cover
Riparian Vegetation
Catchment Variables:
Road Density
Population Density
% Urban
% Agriculture
Mountains
Elevation
Reach Slope
Stream Size
X
X
X
X
X
X
X
X
X
X
X
X
Plains
Longitude
Elevation
X
X
X
X
X
X
X
X
X
X
X
X
Xeric
Elevation
Reach Slope
Stream Size
X
X
X
X
X
X
X
X
X
X
X
X
X
RC-6
-------
Table RC-4 Criteria used by Alan Herlihy to identify most-disturbed sites in each of 10 ecological regions of the
West
Herlihy Criteria:
MT-PNW
MT-NROCK
NSROCK
MT-SWEST
PL-RANGE
PL-NCULT
XE-CALIF
XE-NORTH
XE-SOUTH
XE-EPLAT
Total
Phosphorus
>100
>100
>100
>100
>500
>500
>500
>150
>150
>150
Total
Nitrogen
>1500
>1500
>1500
>1500
> 10000
> 10000
> 10000
>5000
>5000
>5000
Chloride
>1000
>1000
>1000
>1000
>5000
>5000
>5000
>5000
>5000
>5000
Sulfate
>1000
>1000
>1000
>1000
> 10000
PH
<6
<6
<6
<6
<6
<6
<6
<6
<6
<6
Turbidity
>10
>10
>10
>10
>100
>100
>75
>75
>75
>75
Riparian
Disturbance
(W1JHALL)
>3.0
>3.0
>3.0
>3.0
>3.0
>3.0
>3.0
>3.0
>3.0
>3.0
% Fines
>50%
>50%
>50%
>50%
>99%
>99%
>95%
>90%
>90%
>90%
Canopy
Density
(XCDENBK)
<10%
<10%
<10%
<10%
<5%
<5%
<10%
<10%
<10%
<10%
RC-7
-------
Table RC-5 Criteria used by John Stoddard to identify most-disturbed sites in each of 10 ecological regions of the
West
Stoddard Criteria:
Region:
MT-PNW
MT-NROCK
NSROCK
MT-SWEST
PL-RANGE
PL-NCULT
XE-CALIF
XE-NORTH
XE-SOUTH
XE-EPLAT
Total
Phosphorus
>200
>200
>200
>200
>900
>900
>300
>300
>300
>300
Total
Nitrogen
>1000
>1000
>1000
>1000
>3000
>4000
>4000
>4000
>4000
>4000
Chloride
>1000
>1000
>1000
>1000
>3000
>2750
>2500
>2500
>2500
>2500
Sulfate
>1000
>1000
>1000
> 15000
> 15000
> 15000
> 15000
PH
>9
>9
>9
>9
>9
>9
>9
>9
>9
>9
Turbidity
>50
>50
>50
>50
>200
>100
>50
>50
>50
>50
Riparian
Disturbance
(W1_HALL)
>3.0
>3.0
>3.0
>3.0
>3.0
>3.0
>3.0
>3.0
>3.0
>3.0
Relative Bed
Stability
(LRBS_BW5)
<-3.0
<-2.0
<-3.0
<-2.0
<-4.0
<-2.8
<-2.8
<-2.8
<-2.8
Mean RBP
Habitat
(RH_XMET)
<6
<6
<6
<6
<6
<6
<6
<6
<6
<6
RC-8
-------
Table RC-6 Variables used in STAR/R5BIO approach to screen reference sites. Site ranks were reduced by one
class if Lattin's orthophoto/GIS screening indicated watershed-wide disturbance.
Variables
Riparian Vegetative Cover
Hillslope Erosion
Riparian Livestock Browsing
Stream Incision
Bank Erosion
"Pristine"
>95%
0
0-5%
0
0-5%
Minimally-
disturbed
85-95%
Some, local
5-25%
Old, not vertical
5-15%
Least-
disturbed
75-85%
Obvious, some
stream
deposition
25-50%
Deep, new
floodplain
15-35%
Disturbed, but
best in wide
area
<75%
Evident
deposition;
alters stream
flow
>50%
Deep, active
>35%
RC-9
-------
Figures
EMAP AND STAR GRANT REFERENCE SITES
• EMAP HAND-PICKED REFERENCE SITES
• EMAP PROBABILITY SAMPLE REFERENCE SITES
A STAR GRANT HAND-PICKED REFERENCE SITES
| | STATE BOUNDARIES
| MOUNTAIN ECOREGIONS
PLAINS ECOREGIONS
XERIC ECOREGIONS
| | LEVEL 3 ECOREGION BOUNDARIES
Figure RC-1 Location of EMAP and STAR/R5BIO reference sites resulting from EMAP efforts to identify least-
disturbed sites across the West
RC-10
-------
Extent of Resource
Methods
For both the non-perennial and perennial surveys, the process to evaluate candidate
sites selected by the survey design was the same. For each site, three basic questions
were addressed to determine if a site met the criteria defined for the target population:
• Is there a stream channel present at the site coordinates?
• If a channel is present, is the flow perennial (i.e., believed to contain water all year
in most years)?
The site evaluation was conducted independently of any information contained in RF3,
using standardized procedures and data recording forms. The procedures allowed for
various approaches to obtain the information, including maps, photos, publications, local
contacts, and locally developed CIS coverages. In some cases, a field visit to a site
was conducted to acquire or confirm the information. Different individuals or groups
conducted site evaluations for each State; each group determined the best approach to
use in acquiring the necessary information.
The results of the site evaluation were used to determine the status of the sites in the
non-perennial survey. For the perennial survey, the site evaluation exercise also served
to identify candidate field sampling sites. The status of candidate sites in the perennial
survey was refined through reconnaissance visits conducted before sampling, and/or
actual sampling visits.
For analysis, each site was classified into one of the "status" classes listed in Table EX-
1. Target sites were those that met the explicit criteria defined for the target population
(see "Description of Study Requirements" in Design section). Some target sites could
not be sampled because of safety concerns, physical barriers that prevent access, or
because permission to access the site was denied. Sites that are inaccessible or where
permission to sample was denied represent part of the target population that cannot be
assessed for ecological condition. Nontarget sites included sites on non-perennial
streams, map errors (no stream channel at coordinates) or sites with perennial flow but
which were not natural freshwater stream channels (e.g., impoundments, wetlands,
tidally influenced, artificial canals or pipelines). For the non-perennial survey, two target
classes and two non-target classes were used. Additional target classes were not used
since the primary purpose of this survey was simply to identify the potential length of
stream that was evaluated as having perennial flow, and also would have required in
many cases the additional time and expense of a field visit. For the perennial survey,
four target classes and one non-target class were used. Target sites were either
confirmed as perennial, or presumed to be perennial in the absence of evidence to the
contrary, and included sites that were inaccessible for field sampling and sites where
access was denied. Non-target sites included non-perennial sites, perennial sites that
were not streams, and map errors (no water at the site coordinates).
Initial weights computed for each site as part of the survey design process for selecting
sites need adjustment prior to analysis so that they sum to the appropriate sampling
EX-1
-------
frame length derived from RF3 (Tables DE-2, DE-4; also see "Adjusting survey design
weights" in Design section). Extent estimates were based on all sites that were
evaluated (including those not considered later for field sampling). Sites that were
included as both statewide sites and intensive studies (see Table DE-7) were included
as part of the intensive study for the purpose of weight adjustment. Adjusted weights
were calculated as:
if 3
V n-1
where:
wad] = adjusted site weight
wimt = initial site weight
s
LRP3 = Stream length in RF3-based sampling frame for study, summed over strata
(=states) when study area included more than one state
wimt = Sum of initial weights of all sites in study
N
z
n=l
Six different adjusted weights were calculated to meet requirements for different types
of estimates or for different indicators.
WGT_NONP
WGT_EXT
WGT_COND
WGTJNVP
WGT_FTIS
WGT FTIS2
Used to estimate extent in non-perennial survey. Includes all
sites.
Used to estimate extent in perennial survey; includes all sites that
were evaluated.
Used to estimate condition for benthos, fish, chemistry, and habitat
indicators. Includes all sites in partition groups required for field
sampling and all panel years (0-3).
Used to estimate condition for the invasive plant indicator.
Includes all sites in partition groups required for field sampling in
panel years 1-3.
Used to estimate condition for fish tissue mercury indicator.
Includes all sites in partition groups required for field sampling in
panel years 0-2.
Used to estimate condition for fish tissue metals indicator. Includes
all sites in partition groups required for field sampling in panel
years 0-1.
EX-2
-------
For the non-perennial survey, a single weight adjustment was needed. For the
perennial survey, six separate adjustment calculations were required for the different
study components (Statewide, Upper Missouri Basin, N. California, S. California,
Deschutes-John Day Basin, and Wenatchee Basin). Adjusted weights were computed
using the Statistical Analysis System (SAS). Length estimation for both the non-
perennial and perennial surveys was done using version 2.0.1 of the R statistical
software (R Development Core Team 2004) and the cat.analysis function from version
2.5.1 of an R contributed library, psurvey.analysis. The output from this function was
the estimated stream length represented in each status class. Precision was estimated
as the 95% confidence interval as calculated using the local variance estimation
procedure developed by Stevens and Olsen (2003).
Extent Estimates for Non-perennial and Perennial Surveys
Length estimates from both non-perennial and perennial surveys for the entire study
area are presented in Table EX-2. One of the assumptions of EMAP-West is that the
sampling frame derived from RF3 includes all stream channels specified by the
definition of the target population. Approximately 113,600 km (7%) from the non-
perennial survey sampling frame was evaluated as perennial and target. This length is
not represented in the target population, and results in an underestimate of the length of
the target population equal to about 18% of the total frame length for the perennial
survey (628,625 km).
Non-target sites (resulting from coding errors or recent changes in the landscape) in the
perennial survey sample frame represent an overestimate of the target population
length. Approximately 186,000 km (30%) of the perennial survey sampling frame length
was determined to be non-target. These sites also represent a "contamination" of the
sampling frame that require additional reconnaissance to determine their target status,
and potentially result in wasted time and costs associated with field visits that yield no
samples or data.
As described in the Design section, unequal probability sampling was based in part on
prior experience with the RF3 sampling frame and potential errors. To obtain some
indication of the location and types of streams that were misclassified in both the non-
perennial and perennial surveys, sites were plotted on a map that included state
boundaries and the "arid" and "humid" ecoregion groups, and some indication of stream
order. These plots are presented in Figure EX-1. Non-perennial sites classified as
potential target population sites tended to be lower order streams in humid ecoregions
or larger streams in arid ecoregions; no patterns by state boundaries were evident.
Perennial survey sites classified as non-perennial sites tended to be smaller order
streams in humid ecoregions and larger streams in arid ecoregions, which might be an
indication of loss of water resources since the information used in RF3 to classify the
site was produced. Arizona and North Dakota appear to have much higher numbers of
misclassified perennial survey sites than other states, suggesting that differences in
photointerpretation or quality of aerial photographs used to develop the maps on which
RF3 is based on might also be a factor.
The results of the two surveys suggest that improvements in the sampling frame would
be worthwhile. Sites coded as non-perennial in RF3 should be considered for inclusion,
EX-3
-------
given the estimated extent of potential target sites. Using the spatial distribution of
misclassified sites (Figure EX-1) and information acquired from the site evaluation
exercise and field sampling, it may be possible to identify areas and/or stream types
that seem to be prone to misclassification and possible causes. This information could
then be used to refine specific sampling frames, and possibly even NHD. Future frame
development and site selection efforts would become more efficient and cost-effective,
and length estimates derived from them would be more accurate, and yield more robust
assessments of condition for reports such as the state 305b reports required by the
Clean Water Act. Misclassified sites in both surveys can also be studied further to
determine those characteristics that might be used to predict target (or non-target)
status with some level of confidence to reduce the reliance on field evaluation activities.
Estimated Extent of Target Population from Perennial Survey
The total frame length for the perennial survey is 628,625 km (Table EX-2, also see
Table DE-4). To estimate condition of the target population based on various indicators,
length estimates are made using only the sites within partition groups that were required
to obtain the required number of target and sampled sites. These partitions will also
include non-target sites as well as target sites that could not be sampled because they
were physically inaccessible or because permission to access was not obtained (see
Table DE-8). Thus the length of the target population that can actually be assessed is
less than the total represented in the sampling frame. By estimating the length
associated with target sites that were not sampled, a more accurate assessment is
achieved, and the potential impact of the non-sampled length can be evaluated in terms
of the proportion of total resource length it represents and whether or not it represents a
potential bias to the assessment of ecological condition.
Figure EX-2 summarizes the proportion of the perennial survey frame length that was
sampled (and can be assessed for condition) versus other target and non-target
categories. Out of the total frame length, about 48% (305,000 km) represents the
potential target population that can be assessed for condition (Target-Sampled). About
12% of the total frame length can not be assessed due to lack of access permission,
and about 6 % can not be assessed due to being physically inaccessible. Compare
these results to those from Table DE-8, which provides this information in terms of
numbers of sites rather than length of resource.
Figure EX-3 summarizes the proportion of perennial survey frame length in various
status categories for the subpopulations being presented in this report. The length
representing the portion of the target population that can be used to estimate condition
was slightly less than 50% of the entire sampling frame, and was typically above 40%
for the various subpopulations. The highest proportion of frame length classified as non-
target is found in the xeric subpopulations, and ranged from approximately 20% in the
Pacific Northwest Mountains subpopulation (MT-PNW) to over 70% in the southern
xeric subpopulation (XE-SOUTH). In some areas, lack of permission to access sampling
sites was a concern. However, the estimated proportion of frame length affected by this
was fairly low (around 10%) and constant across all subpopulations. This indicates that
in areas where the number of sites where permission could not be obtained was high,
they tended to be sites having low adjusted weight values.
EX-4
-------
The estimated length of the target population varies slightly based on the particular
adjusted weight variable used, due to differences in sample sizes, and the adjusted
weight values of the individual sites included. These estimates are presented in Table
EX-3 for both the total study area and for each subpopulation. These are the maximum
values for length for any CDF developed for an indicator variable using a particular
adjusted weight variable. Actual lengths assessed for a particular indicator are further
reduced if there are missing values for the indicator variable due to any number of
reasons that prevented collecting data at an individual site.
EX-5
-------
Tables
Table EX-1. Status classes used for estimating extent
NON-PERENNIAL SURVEY
TARGET CLASSES
Perennial
Inaccessible
Evaluated as perennial,
representing a potential
sampling site
Evaluated as perennial,
but site could not be
visited and sampled due
to unsafe conditions or
physical barriers
NON-TARGET CLASSES
Non-
perennial
Perennial-
Non-target
Evaluated as non-
perennial
Evaluated as perennial,
but is impounded,
tidally-influenced, a
non-target canal or
pipeline, or a map error
PERENNIAL SURVEY
TARGET CLASSES
Target-
Sampled
Target-Not
Sampled
Inaccessible
Access
Denied
Evaluated as perennial,
and sampled as part of
EMAP-West
Evaluated as perennial,
but not required to be
sampled to achieve the
required sample size for
population estimation
Evaluated as perennial,
but site could not be
visited or sampled due
to physical barriers or
unsafe conditions;
represents part of target
population that cannot
be assessed for
condition
Evaluated as perennial,
but access was denied
by landowner;
represents part of target
population that cannot
be assessed for
condition
NON-TARGET CLASSES
Non-Target
Evaluated as non-
perennial, or evaluated
as perennial, but
impounded, a non-
target canal or pipeline,
or a map error
EX-6
-------
Table EX-2. Estimated stream lengths (km) of target population from non-
perennial and perennial surveys.
NON-
PERENNIAL
SURVEY
PERENNIAL
SURVEY
TOTAL
RF3
FRAME
SIZE (km)
1,638,200
628,625
EVALUATED
NON-
PERENNIAL
1,477,521
(±62,704)
Included in non-
target estimate
EVALUATED
TARGET
113,570
(±21,008)
442,859
(±14,187)
556,429
EVALUATED
NON-
TARGET
47,109
(±18,103)
185,766
(±12,846)
EX-7
-------
Table EX-3. Estimated lengths (km) of the target population lengths (km) that can
be used to estimate condition based on different adjusted weight variables.
Number of sites used to estimate length is in parentheses.
SUB-POPULATION
WEST- WIDE
MOUNTAINS
MT-NROCK
MT-PNW
MT-SROCK
MT-SWEST
PLAINS
PL-NCULT
PL-RANGE
XERIC
XE-CALIF
XE-EPLAT
XE-NORTH
XE-SOUTH
WGT_COND
(965)
304,544
(573)
220,047
(210)
100,904
(226)
84,184
(60)
32,106
(77)
2,853
(190)
35,142
(66)
8,004
(124)
27,138
(201)
48,812
(34)
6,856
(71)
20,981
(49)
1 1 ,600
(47)
9,376
WGTJNVP
(671)
305,559
(392)
214,392
(147)
103,583
(153)
79,852
(38)
27,432
(53)
2,938
(131)
36,581
(43)
7,240
(88)
29,341
(147)
53,876
(26)
8,133
(50)
22,131
(36)
13,870
(34)
9,526
WGT_FTIS
(876)
300,830
(517)
218,516
(190)
102,646
(209)
82,285
(52)
30,881
(66)
2,704
(173)
34,270
(60)
7,797
(113)
26,473
(185)
47,501
(34)
6,856
(62)
20,870
(45)
10,414
(44)
9,361
WGT_FTIS2
(583)
304,721
(355)
224,072
(130)
102,495
(138)
85,240
(37)
33,186
(50)
3,153
(113)
32,926
(43)
8,256
(70)
24,671
(114)
46,534
(22)
7,136
(44)
20,657
(23)
8,535
(25)
10,206
EX-8
-------
Figures
CANDIDATE PERENNIAL SITES
. 0-1
. 2
• 3
• 4-5
• 6-7
H Humid Ecoregions
_ Arid Ecoregions
NON-PERENNIAL SITES
. 0-1
o 2
o 3
O 4-5
O 6-8
OTHER NONTARGET SITES
0- 1
_ 2
.. 3
,-. 4-5
A 6-8
Humid Ecoregions
Arid Ecoregions
Figure EX-1. Location of misclassified sites in non-perennial survey (top) and
perennial survey (bottom). Size of symbol indicates Strahler order.
EX-9
-------
Status Category
Figure EX-2. Estimated length of perennial survey sampling frame in various
status categories. Target-Sampled represents the length of the target population
for which condition can be estimated. Error bars are 95% confidence intervals.
80
TARGET-SAMPLED
TARGET-INACCESSIBLE
TARGET-NO PERMISSION
NONTARGET
SUBPOPULATION
Figure EX-3. Estimated proportion of perennial survey frame length in each
subpopulation in various status categories. Target-Sampled represents the
proportion of the target population for which condition can be estimated. Error
bars are 95% confidence intervals.
EX-10
-------
Benthic Macroinvertebrates
We use benthic macroinvertebrates (larval insects and other stream invertebrates such
as snails and worms) to help understand how human activities/disturbances/stresses/
pressures affect the biotic or ecological condition of streams and rivers (e.g., to meet
Clean Water Act mandates). In order to do this, we need to understand the structure
and function of the benthic macroinvertebrate (BMI) assemblage in situations with no or
low human disturbance and compare the current condition with that condition. We use
human disturbance and related stressors to represent human presence on the
landscape and consequent alterations to the fundamental processes that organize the
BMI assemblages (i.e., delivery of water, nutrients, exotic chemicals, sediment, wood,
energy; riparian functions; floodplain connections; imposition of barriers and channel
modifications). We identified places (watersheds, stream reaches) that are the least
disturbed by human activities/stressors and represent the broad range of natural factors
that affect BMI assemblages. Then we characterized the BMI at these sites as a
benchmark to evaluate the extent of change across all the probability sites. The process
we used to select the least disturbed sites is described in detail in Waite et al. (2000)
and Whittier et al. (In Press).
The taxonomic composition and richness of the BMI found in streams provide valuable
information about the condition of streams and the potential stressors acting upon them.
The challenge we face is extracting that information from the complex relative
abundance data collected for BMI and presenting it in an informative way for managers
and the general public. Two approaches have been developed and tested in various
ecological settings across the U.S., in Canada, Europe, and Australia, and we employed
both for EMAP-West:
• Multi-metric Index (MMI): This is the traditional approach used in the U.S. to
analyze macroinvertebrate assemblage data (e.g., Barbour et al. 1995, Barbour
et al. 1999, Karr and Chu 1999) - various composition, tolerance and richness
characteristics of the assemblages are summarized as metrics, e.g., the number
of mayfly species present. Each of a series of candidate metrics is evaluated
against an array of criteria, and a subset of 5 to 10 of the best performing
metrics are then combined into a multi-metric index, often called an Index of
Biotic Integrity (IBI).
• 0/E Index: This second approach posits that taxonomic composition across a
set of reference sites can be modeled as a function of natural gradients (such as
elevation, stream size, stream gradient, latitude, longitude) to estimate the
expected taxonomic composition in the absence of human stressors (Hawkins et
al. 2000, Wright 2000). This calibrated model is then used to estimate expected
composition, usually expressed as richness, at "test" sites. The list of expected
taxa that are observed at test sites is compared with the expected list as an
Observed:Expected ratio (0/E index). Departures from a ratio of one indicate
that the composition at a test site differs from that expected under less disturbed
conditions.
BN-1
-------
Two programs contributed BMI samples that were used as a basis for both MMI and
0/E development. EMAP-W contributed BMI samples derived from the probability
survey, along with a small number of hand selected "reference site" samples. Utah
State University, partially with support from the USEPA STAR grant program and from
the U.S. Forest Service (R5BIO), collected samples at hundreds of reference sites
across the 12 state EMAP-W region, as well as at reference sites in New Mexico (all
referred to as STAR/R5BIO sites). The BMI data derived from the set of reference and
probability sites that passed a series of screening criteria were used in model
construction described later.
Macroinvertebrate field sampling and lab processing
EMAP:
Wherever possible, macroinvertebrate samples were collected at field sites with two
protocols. In wadeable streams, EMAP crews collected:
1. A reach-wide sample consisted of a composite of 11 D-frame kicknet samples,
one from each of the 11 standard transects used to characterize a reach—each
kicknet sample collected all organisms within a one square foot area; and
2. A targeted riffle sample consisted of a composite of 8 samples taken randomly
from four riffles, 2 samples per riffle—each D-frame kicknet sample collected all
organisms within a one square foot area.
In non-wadeable streams and rivers, the reachwide protocol was implemented along the
shoreline, but no targeted riffle sample was taken (nor were they taken in wadeable
streams that had no riffle habitat present). Details of all sampling protocols are given in
Peck et al. (2005b) and Peck et al. (2005a). Composite samples were preserved in the
field with ethanol and transported to one of two laboratories for processing (samples
collected in California were analyzed by the California Department of Fish and Game,
all other samples were analyzed by EcoAnalysts in Moscow, Idaho).
STAR/R5BIO:
Targeted riffle samples were collected in the same way as the EMAP targeted riffle
sample, preserved in the field and processed by the Utah State University laboratory.
Reach-wide samples were not collected at STAR/R5BIO sites.
Lab processing and data files:
All labs used a fixed count protocol consisting of enumerating and identifying 500
individuals (+/-10%) drawn from the composite sample, or a complete count if the
composite sample contained fewer than 500 individuals. Individuals were identified to
the lowest practical taxonomic level (in most cases, the genus level).
Raw data are contained in a BMI count file containing a sample-by-taxon summary of
numbers of individuals of each taxon found in each sample and that taxon's full
taxonomy (phylum, class, order, etc.). Other summary files are derived from the BMI
count files, sometimes in combination with other files.
BN-2
-------
Operational Taxonomic Units: Not all individuals can be classified to the same
taxonomic level across samples and sites. For some analyses, an evaluation of the
taxonomic consistency across samples and sites necessitates combining taxa to a
coarser level than that indicated in the BMI count file in order to conduct cross site
analyses at a common level of taxonomy. A single class of taxa combined in such a way
is referred to as an "operational taxonomic unit" or OTU. The combined EMAP-Wand
STAR/R5BIO BMI data were evaluated to come up with OTU assignments, yielding
approximately 550 OTUs. For various reasons, not all taxa can be assigned an OTU.
Standardizing counts: Although the lab processing protocol calls for a fixed count of 500
individuals, obtaining an exact fixed count is impractical. As a result, a subsampling
technique is used to extract a true fixed count from the taxa enumerated for a sample.
We used a fixed count of 300 individuals, drawn at random (without replacement) from
each site's BMI count file.
We created several operational BMI files that were used for the BMI MMI and 0/E
index. The key files are:
For the BMI MMI: A BMI metric file consisting of the array of candidate metrics was
evaluated and culled to produce the BMI MMI. The BMI metric file used a 300 count
subset from the BMI count file and retained the original level of taxonomy (i.e., did not
use the OTU assignments). The types of metrics calculated are summarized in the next
section.
For the 0/E index: The BMI count file was first reduced to an OTU file (by eliminating
ambiguous taxa), and then subsampled to a 300 count subset for each sample.
In some cases, we retained samples that did not contain at least 300 individuals. For
sites classified as reference, we retained samples with at least 200 individuals; for all
"test" sites, we retained the full count (low counts can indicate BMI responses to
stressors).
We used the reach-wide data from each sites if a reach-wide sample was available,
otherwise we used the targeted riffle data. Exploratory analyses that compared reach-
wide richness with targeted riffle richness indicated very high correlation. In addition;
multivariate plots of the assemblage composition within sites (in reach-wide and
targeted riffle samples) showed little compositional difference compared with differences
across sites.
Characterizing the assemblage
Multi-metric Index
Our goal was to produce three coordinated MMIs for EMAP-W: one for each of the three
aggregated ecoregions presented in Figures 1 and DE-6. Because metric values varied
widely across the geography of the West (as did expected metric values), we
recognized that metrics would need to be selected and scored separately in the regions
shown (Mountains, Plains and Xeric), but that the same process would need to be used
for metric evaluation and scoring, so that the three MMIs could be combined in a single
BN-3
-------
assessment without introducing regional bias. The process we used to assure
comparability between the regions is described in this section.
A total of 76 metrics were calculated from the BMI data collected by the EMAP-West
and STAR/R5BIO programs. Each metric was then assigned to one of 6 metric classes,
each of which is intended to capture a separate dimension of biotic integrity (Karr et al.
1986, Karr 1993, Barbouretal. 1999):
• Richness - the number of different kinds of taxa
• Diversity - evenness of the distribution of individuals across taxa
• Composition - the relative abundance of different kinds of taxa
• Functional feeding groups - primary method by which the BMI feed
• Habit - predominant BMI behavior, e.g., do taxa cling to substrates, or burrow
into substrates?
• Tolerance - often expressed as a general tolerance to stressors
Assignments of functional feeding group, habit, and tolerance come from a master
autecology file maintained (as of this writing) by TetraTech, Inc (Owings Mills, MD;
Michael Barbour, personal communication). In most cases (the diversity metrics are the
only exceptions), each autecological characteristic is represented by three metrics: (1)
the total number of taxa with that characteristic (e.g., a single feeding group); (2) the
proportion of all taxa with that characteristic; and (3) the proportion of all individuals in a
sample with that characteristic. The complete list of candidate metrics and their metric
classes are shown in Table MI-1.
We screened the pool of candidate metrics using a series of tests (below), with the goal
of finding the one metric in each metric class with the best behavior (in terms of the
tests described below). The tests were applied sequentially, and by ecoregion. For any
ecoregion, metrics that failed a test were not considered for further evaluation and were
not subjected to subsequent tests. Some metric scores are correlated with natural
gradients; ideally, these correlations should be factored out. However, we did not
screen the full set of candidate metrics. Instead, we examined the correlations of the
final set of metrics with several natural gradients, and concluded that none of the final
metrics were sufficiently correlated with natural gradients (i.e., stream size or slope) to
warrant calibration.
• Range: If the range (difference between the maximum and minimum values) of a
metric is small, or if most of the values are identical, then the metric is unlikely to
provide information that helps differentiate sites from one another. We eliminated
richness metrics if their range was less than 4, and eliminated any metric if more
than 75% of the values were the same. Three metrics were eliminated from
further consideration by the range test (Table MI-1).
• Signal to noise (S:N): Signal to noise is the ratio of variance between sites and
the variance of repeated visits to the same site, and is a measure of how
repeatable metric values are. A low value indicates that a metric has nearly as
much variability within a site (over time) as it does across different sites, and thus
BN-4
-------
indicates a metric that does not distinguish well between sites. We calculated
S:N ratios separately for each metric in each of 3 climatic/topographic regions
(Table MI-2). We failed metrics with S:N values less than 1 in the Mountains (11
metrics failed), less than 0.7 in the Plains (23 failed), and less than 1.5 in the
Xeric region (21 failed).
• Responsiveness: We examined whether metrics were responsive to key stressor
indicators in two ways: by examining scatter plots of each metric vs. of a subset
of chemical (e.g., nutrients, acidity, turbidity) and habitat (e.g., relative bed
stability, riparian disturbance) variables; and by conducting F tests of the ability of
metrics to distinguish between least-disturbed (a.k.a. reference) sites and most-
disturbed (a.k.a., trashed) sites. Both analyses were conducted separately for
the 3 climatic/topographic regions. Results of the F tests are shown in Table Ml-
2. If significant F test results were corroborated by scatter plots with individual
stressors, we considered the metrics suitable for inclusion in the MMI. The list of
metrics included in each climatic/topographic region's MMI was built by first
taking the metric with the highest F score, then taking the metric with the next
highest F score that represented a different metric class, and continuing until all
of the metric classes were represented, provided that all of the metrics were not
redundant.
• Redundancy: Only metrics that did not contain redundant information were
included in the final MM Is. We estimated redundancy by creating a correlation
matrix of metric values at reference sites (to avoid eliminating metrics that are
correlated only because of their relationship to stressors that co-vary). Inclusion
of redundant metrics adds little information to the MMI. We considered metrics
redundant if their Spearman correlation coefficients were > 0.71 (corresponding
to an r2 value of 0.5). Metrics selected for inclusion first (i.e., those with higher F
scores) were retained, and its redundant metric replaced with the next non-
redundant metric in the same metric class. Spearman correlation coefficients for
all of the metrics included in the final MMIs are shown in Table MI-3.
The results of the sequential inclusion of metrics in the final MMIs are shown in Table
MI-4. Within each climatic/topographic region, the order in which metrics were included
(i.e., highest F score first, next highest F score in a "new" metric class next, and so on)
is also shown (in parentheses). Each MMI (for each climatic/topographic region) has
one metric representing each metric class. In all cases, the first metrics chosen for
inclusion (highest F scores) were Tolerance metrics. The least responsive metrics were
always either Diversity metrics or Feeding Group metrics.
Before being combined into an MMI, each raw metric needs to translated to the same
scale—a process we call 'scoring'. We chose to score the metrics continuously on a
scale from 0 to 10. Metrics were scored separately by ecoregion, using a scheme
intended to maximize differences in final IBI scores (Blocksom 2003): ceiling and floor
values for each metric were defined to be the 5th and 95th percentile values observed in
all sites. For positive metrics (e.g., those that are highest in reference sites), values less
than the 5th percentile were given a score of 0, those with values greater than the 95th
percentile were given scores of 10, and all metric values in between were interpolated
linearly. Negative metrics were scored similarly, with the floor (95th percentile) and
BN-5
-------
ceiling (5th percentile) values reversed. Ceiling and floor values for each metric are
listed in Table MI-4. Scored metrics were summed (for a maximum combined score of
60) and the summed score was scaled to 100 by multiplying each sum by 1.666.
O/E Index
We followed the usual series of steps in the development of the O/E index (Moss et al.
1987, Clarke et al. 1996, Hawkins et al. 2000, Clarke et al. 2003). These include the
following: (1) selection and sampling of reference and test sites; (2) development of the
fixed count OTU BMI file (described above) for both reference and test sites; (3)
identification of candidate predictor variables (and creation of corresponding database);
(4) calibration of the predictive model; and (5) application of the predictive model to test
sites.
Predictor variables: We considered the following list of natural variables that have been
commonly used in predictive models in the past, or that are particularly relevant to
Western systems:
• time of year (Julian day) when sample was collected
• site latitude
• site longitude
• site elevation (above mean sea level)
• watershed area (as an indicator of stream size)
• stream gradient
• flow variability (ratio of mean annual low flow to mean annual high flow)
• geology (dummy variables to indicate whether the dominant geology at a site
was carbonate, gneiss, granitic, mafic, quaternary, sedimentary or volcanic)
• alkalinity
• mean annual site temperature (from PRISM, see below)
• mean annual precipitation (from PRISM, see below).
PRISM (Parameter-elevation Regressions on Independent Slopes Model) is an
analytical model that uses point climate data and a digital elevation model to generate
estimates of monthly and annual precipitation and air temperature (Daly et al. 1994).
Model calibration: The following steps are used in the development of a predictive
model:
1. Reference sites are clustered into groups based on the similarity in composition
and relative abundance of their BMI assemblages. The likelihood that a taxon
occurs in a cluster (occurrence probability or frequency of occurrence in a
cluster) is recorded.
2. The list of candidate predictor variables is screened using discriminant analysis
to identify which subset of predictor variables best distinguishes among the
clusters. The best set of predictor variables is combined in a discriminant function
(DF) to determine the likelihood that any site occurs in any of the biologically
determined groups, i.e., the DF estimates probability of group membership.
BN-6
-------
3. The probabilities that a particular taxon occurs in the different reference groups
(from the group occurrence frequencies) are combined with the probabilities that
a site belongs to each group (from the DF) to estimate the probability that the
particular taxon would occur at any particular site, assuming that site was in
reference condition. Summing the probabilities of expected taxon occurrences at
a site across taxa yields the expected richness (E).
4. Comparison of a site's observed richness (for the set of reference site taxa) to its
expected richness is the 0/E index. Following Hawkins et al. (2000), only those
taxa with predicted occurrence probabilities >0.5 at a site are included when
calculating 0/E for that site. If the composition of taxa at the test site is similar to
that of the reference sites, the 0/E ratio will be close to 1; departures from 1
indicate difference from the reference site condition.
For EMAP-W, we initially constructed an west-wide predictive model across all
reference sites in the 13 state region (12 EMAP-W states plus New Mexico, where
STAR/R5BIO reference site data were also available). The model consisted of 42
clusters with the following predictors as important discriminant variables: Elevation,
Longitude, Latitude, site annual precipitation, Log Watershed area, day of the year, and
sedimentary geology (0 or 1).
Model evaluation indicated that performance was reasonable for the Mountain region,
poor for the Plains region, and intermediate for the Xeric region, judged against null
models (Van Sickle et al. 2005) and past experience with model performance. We were
also concerned that the preponderance of reference sites in the mountains controlled
model performance in the other regions. As a result, we explored several ways of
subdividing the reference sites into broad groups that were both ecologically sensible
and that contained a sufficient number of reference sites for model building. This
process entailed both examining the biological clusters from the 42-cluster model, and
groups of sites derived by applying the same clustering approach to the natural
environmental factors. After a series of iterations, we settled on five reasonably well
defined geographic groupings, illustrated in Figure BN-1. The distribution of key
geographic attributes for the 5 clusters are shown in Figure BN-2.
These clusters consist of sites that occur primarily in: 1) North and South Dakota and
eastern Montana (basically Plains sites); 2) Oregon and Washington Cascades and
coastal region, Northern California and coastal mountains, with some sites in the
northern Rocky Mountains (Idaho and Montana); 3) Arid Southwest (California, Arizona,
and New Mexico); 4) Interior high, forested mountains; and 5) Interior xeric plateaus.
We developed satisfactory predictive models for the first two of these five clusters of
sites; for clusters 3-5 we created regional (cluster-specific) null models. In addition, ca.
20 EMAP sites could not be reliably assigned to any of the 5 clusters; for these sites,
0/E scores are based on the west-wide null model.
Validation of MM/ and O/E Indices
There is always some concern that using the same set of sites to build BMI indices and
then to assess BMI integrity will lead to difficulties with circular reasoning. To avoid this
BN-7
-------
difficulty as much as possible, we chose to set aside the data from a random set of sites
in each ecoregion before model development. These sites could then be used to
validate the models by comparing the behavior of the indices in sites present in the
calibration and validation datasets.
For the MM I model construction, 25 least-disturbed (except in the Plains, where 20 sites
were chosen), 25 moderately-disturbed and 25 most-disturbed sites (where disturbance
class was based on chemical and physical habitat [non-biological] data) were chosen at
random in each of the three climatic/topographic ecoregions and their data set aside for
validation purposes. The small number of least-disturbed sites set aside in the Plains
reflects the difficulty of finding many least-disturbed sites in this region. The MMI was
developed using data from all non-validation (calibration) sites; scores were then
calculated for both the calibration and validation datasets. Comparisons of the
distribution of MMI scores in calibration and validation datasets are shown in Figures
BN-3, BN-4 and BN5 (for the Mountains, Plains and Xeric regions, respectively). The
validity of the MMI model is indicated by the lack of significant differences between the
two datasets within each of the disturbance classes. In addition, all three regions show
a general decrease in scores as one moves from the least-disturbed to the most-
disturbed end of the gradient.
Because only least-disturbed ("reference") sites are used in the 0/E model
development, and because the model(s) were built independent of ecoregions, creation
of validation and calibration datasets proceeded somewhat differently than for the MMI.
A total of 136 least-disturbed sites were chosen at random and set aside before the
models were developed. Once the models were constructed, 0/E scores were
calculated for all sites. The best test of the validity of the 0/E model is the comparison
of scores in the least-disturbed sites in the calibration and validation datasets (Figure
BN-6). This comparison shows no significant difference, and very similar ranges of
scores. For completeness, we also include the comparison of scores in moderately-
disturbed and most-disturbed sites (as for the MMI); these also show very similar
distributions between the two datasets.
One further test of the validity of the MMI and 0/E models is their performance in the
types of statistical tests used to evaluate the metrics described earlier. Table MI-5
shows the signal:noise ratio, F test (for discriminating least-disturbed from most-
disturbed sites), and the standard deviation of scores at reference sites (a standard
measure of the precision of 0/E models) for both indices. All of these tests suggest that
both indices have excellent properties, and can be used to describe the ecological
condition of streams and rivers across the West.
BN-8
-------
References
Barbour, M. T., J. Gerritsen, B. D. Snyder, and J. B. Stribling. 1999. Rapid
Bioassessment Protocols for Use in Streams and Wadeable Rivers. EPA/841/B-
99/002, U.S. Environmental Protection Agency, Washington, DC.
Barbour, M. T., J. B. Stribling, and J. R. Karr. 1995. Multimetric approach for
establishing biocriteria and measuring biological condition. Pages Chapters, pg.
63-77 in W. S. Davis and T. P. Simon, editors. Biological assessment and
criteria: tools for water resource planning and decision making. Lewis, Boca
Raton, FL.
Blocksom, K. A. 2003. A performance comparison of metric scoring methods for a
Multimetric Index for Mid-Atlantic Highlands streams. Environmental
Management 31:670-682.
Clarke, R. T., M. T. Furse, J. F. Wright, and D. Moss. 1996. Derivation of a biological
quality index for river sites: comparison of the observed with the expected fauna.
Journal of Applied Statistics 23:311-332.
Clarke, R. T., J. F. Wright, and M. T. Furse. 2003. RIVPACS models for predicting the
expected macronivertebrate fauna and assessing the ecological quality of rivers.
Ecological Modeling 160:219-233.
Daly, C., R. P. Nielson, and D. L. Phillips. 1994. A statistical-topographic model for
mapping climatological precipitation over mountainous terrain. Journal of Applied
Meteorolology 33:140-158.
Hawkins, C. P., R. H. Morris, J. N. Hogue, and J. W. Feminella. 2000. Development and
evaluation of predictive models for measuring the biological integrity of streams.
Ecological Applications 10:1456-1477.
Karr, J. R. 1993. Defining and assessing ecological integrity: Beyond water quality.
Environmental Toxicology and Chemistry 12:1521-1531.
Karr, J. R., and E. W. Chu. 1999. Restoring life in running waters: better biological
monitoring. Island Press, Washington, D.C.
Karr, J. R., K. D. Fausch, P. L. Angermeier, P. R. Yant, and I. J. Scholosser. 1986.
Assessing Biological Integrity in Running Waters: A Method and its Rationale.
Illinois Natural History Survey, Champaign, IL.
Moss, D., M. T. Furse, J. F. Wright, and P. D. Armitage. 1987. The prediction of the
macroinvertebrate fauna of unpolluted running-water sites in Great Britain using
environmental data. Freshwater Biology 17:41-52.
Peck, D. V., D. K. Averill, A. T. Herlihy, R. M. Hughes, P. R. Kaufmann, D. J. Klemm, J.
M. Lazorchak, F. H. McCormick, S. A. Peterson, M. R. Cappaert, T. Magee, and
P. A. Monaco. 2005a. Environmental Monitoring and Assessment Program -
Surface Waters Western Pilot Study: Field Operations Manual for Non-Wadeable
Rivers and Streams. EPA 600/R-05/xxx, U.S. Environmental Protection Agency,
Washington, DC.
BN-9
-------
Peck, D. V., A. T. Herlihy, B. H. Hill, R. M. Hughes, P. R. Kaufmann, D. J. Klemm, J. M.
Lazorchak, F. H. McCormick, S. A. Peterson, P. L. Ringold, T. Magee, and M. R.
Cappaert. 2005b. Environmental Monitoring and Assessment Program - Surface
Waters Western Pilot Study: Field Operations Manual for Wadeable Streams.
EPA 600/R-OS/xxx, U.S. Environmental Protection Agency, Office of Research
and Development, Washington, DC.
Van Sickle, J., C. P. Hawkins, D. P. Larsen, and A. T. Herlihy. 2005. A null model for the
expected macroinvertebrate assembalge in streams. Journal of the North
American Benthological Society 24:178-191.
Waite, I. R., A. Herlihy, D. P. Larsen, and D. J. Klemm. 2000. Comparing strengths of
geographic and nongeographic classifications of stream benthic
macroinvertebrates in the Mid-Atlantic Highlands, USA. Journal of the North
American Benthological Society 19:429-441.
Whittier, T. R., J. L. Stoddard, R. M. Hughes, and G. Lomnicky. In Press. Associations
among watershed- and site-scale disturbance indicators and biological
assemblages at least- and most-disturbed stream and river sites in the western
USA. in R. M. Hughes, L. Wang, and P. W. Seelbach, editors. Influence of
landscapes on stream habitats and biological assemblages. American Fisheries
Society, Bethesda, Maryland.
Wright, J. F. 2000. An introduction to RIVPACS. Pages 1-24 in J. F. Wright, D. W.
Sutcliffe, and M. T. Furse, editors. Assessing the Biological Quality of Fresh
Waters. Freshwater Biological Association, Ambleside, UK.
BN-10
-------
Tables
Table MI-1. Candidate Macroinvertebrate Metrics
(and results of range test)
Metric ID
BURRPIND
BURRPTAX
BURRRICH
CHIRPING
CHIRPTAX
CHIRRICH
CLMBPIND
CLMBPTAX
CLMBRICH
CLNGPIND
CLNGPTAX
GINGRICH
COFIPIND
COFIPTAX
COFIRICH
COGAPIND
COGAPTAX
COGARICH
DOM1PIND
DOM3PIND
DOM5PIND
EPHEPIND
EPHEPTAX
EPHERICH
EPT_PIND
EPT_PTAX
EPT_RICH
Metric Class
HABIT
HABIT
HABIT
COMPOSITION
COMPOSITION
RICHNESS
HABIT
HABIT
HABIT
HABIT
HABIT
HABIT
FEEDING
FEEDING
FEEDING
FEEDING
FEEDING
FEEDING
DIVERSITY
DIVERSITY
DIVERSITY
COMPOSITION
COMPOSITION
RICHNESS
COMPOSITION
COMPOSITION
RICHNESS
Metric Description
Burrower % Individuals
Burrower % Distinct Taxa
Burrower Distinct Taxa Richness
Chironomid % Individuals
Chironomid % Distinct Taxa
Chironomid Distinct Taxa Richness
Climber % Individuals
Climber % Distinct Taxa
Climber Distinct Taxa Richness
Clinger % Individuals
Clinger % Distinct Taxa
Clinger Distinct Taxa Richness
Collector-Filterer % Individuals
Collector-Filterer % Distinct Taxa
Collector-Filterer Distinct Taxa Richness
Collector-Gatherer % Individuals
Collector-Gatherer % Distinct Taxa
Collector-Gatherer Distinct Taxa Richness
Percent of Individuals in Dominant Taxa
Percent of Individuals in Top 3 Taxa
Percent of Individuals in Top 5 Taxa
Ephemeroptera % Individuals
Ephemeroptera % Distinct Taxa
Ephemeroptera Distinct Taxa Richness
EPT % Individuals
EPT % Distinct Taxa
EPT Distinct Taxa Richness
Range Test
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
BN-11
-------
Metric ID
FACLPIND
FACLPTAX
FACLRICH
HBI
HPRIME
INTLPIND
INTLPTAX
INTLRICH
MEGLPIND
MEGLPTAX
MEGLRICH
NOINPIND
NOINPTAX
NOINRICH
NTOLPIND
NTOLPTAX
NTOLRICH
OLLEPIND
OLLEPTAX
OLLERICH
OMNIPIND
OMNIPTAX
OMNIRICH
PLECPIND
PLECPTAX
PLECRICH
PREDPIND
PREDPTAX
PREDRICH
Metric Class
TOLERANCE
TOLERANCE
TOLERANCE
TOLERANCE
DIVERSITY
TOLERANCE
TOLERANCE
TOLERANCE
COMPOSITION
COMPOSITION
RICHNESS
COMPOSITION
COMPOSITION
RICHNESS
TOLERANCE
TOLERANCE
TOLERANCE
COMPOSITION
COMPOSITION
RICHNESS
FEEDING
FEEDING
FEEDING
COMPOSITION
COMPOSITION
RICHNESS
FEEDING
FEEDING
FEEDING
Metric Description
Facultative % Individuals
Facultative % Distinct Taxa
Facultative Distinct Taxa Richness
Hilsenhoff Biotic Index
Shannon Diversity
Intolerant % Individuals
Intolerant % Distinct Taxa
Intolerant Distinct Taxa Richness
Megaloptera % Individuals
Megaloptera % Distinct Taxa
Megaloptera Distinct Taxa Richness
Non-Insect % Individuals
Non-Insect % Distinct Taxa
Non-Insect Distinct Taxa Richness
Non-Tolerant % Individuals
Non-Tolerant % Distinct Taxa
Non-Tolerant Distinct Taxa Richness
Oligochaete/Leech % Individuals
Oligochaete/Leech % Distinct Taxa
Oligochaete/Leech Distinct Taxa Richness
Omnivore % Individuals
Omnivore % Distinct Taxa
Omnivore Distinct Taxa Richness
Plecoptera % Individuals
Plecoptera % Distinct Taxa
Plecoptera Distinct Taxa Richness
Predator % Individuals
Predator % Distinct Taxa
Predator Distinct Taxa Richness
Range Test
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
FAIL
FAIL
FAIL
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
BN-12
-------
Metric ID
SCRPPIND
SCRPPTAX
SCRPRICH
SHRDPIND
SHRDPTAX
SHRDRICH
SIMPSON
SPRLPIND
SPRLPTAX
SPRLRICH
SWIMPIND
SWIMPTAX
SWIMRICH
TOLRPIND
TOLRPTAX
TOLRRICH
TOTLRICH
TRICPIND
TRICPTAX
TRICRICH
Metric Class
FEEDING
FEEDING
FEEDING
FEEDING
FEEDING
FEEDING
DIVERSITY
HABIT
HABIT
HABIT
HABIT
HABIT
HABIT
TOLERANCE
TOLERANCE
TOLERANCE
RICHNESS
COMPOSITION
COMPOSITION
RICHNESS
Metric Description
Scraper % Individuals
Scraper % Distinct Taxa
Scraper Distinct Taxa Richness
Shredder % Individuals
Shredder % Distinct Taxa
Shredder Distinct Taxa Richness
Simpson Index
Sprawler % Individuals
Sprawler % Distinct Taxa
Sprawler Distinct Taxa Richness
Swimmer % Individuals
Swimmer % Distinct Taxa
Swimmer Distinct Taxa Richness
Tolerant % Individuals
Tolerant % Distinct Taxa
Tolerant Distinct Taxa Richness
Total Distinct Taxa Richness
Trichoptera % Individuals
Trichoptera % Distinct Taxa
Trichoptera Distinct Taxa Richness
Range Test
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
PASS
BN-13
-------
Table MI-2. Signal:Noise Ratios and F-Test Results for Candidate
Macroinvertebrate Metrics
Metric ID
BURRPIND
BURRPTAX
BURRRICH
CHIRPING
CHIRPTAX
CHIRRICH
CLMBPIND
CLMBPTAX
CLMBRICH
CLNGPIND
CLNGPTAX
GINGRICH
COFIPIND
COFIPTAX
COFIRICH
COGAPIND
COGAPTAX
COGARICH
DOM1PIND
DOM3PIND
DOM5PIND
EPHEPIND
EPHEPTAX
EPHERICH
EPT PIND
EPT PTAX
EPT RICH
FACLPIND
FACLPTAX
FACLRICH
HBI
HPRIME
INTLPIND
INTLPTAX
INTLRICH
MEGLPIND
Signal:Noise Ratios
Mountains
1.97
1.14
0.83
1.54
1.58
0.88
1.94
1.35
1.24
1.89
2.08
3.54
1.68
2.63
2.75
1.82
1.33
1.25
0.88
1.20
1.41
2.89
2.22
2.97
2.46
3.65
4.68
2.31
3.38
1.80
2.60
1.63
3.21
5.56
5.59
0.07
Plains
1.59
0.98
0.42
1.48
0.41
0.41
0.69
0.93
0.66
0.09
0.63
1.60
0.35
0.51
1.10
0.98
0.84
0.43
0.69
0.79
0.67
1.32
2.00
2.89
0.76
2.02
3.75
0.95
1.09
1.51
1.30
0.91
0.63
1.62
2.66
0.45
Xeric
2.32
2.18
1.45
2.89
3.15
2.29
2.27
1.41
1.05
1.31
3.59
5.43
1.24
1.58
2.67
2.80
1.14
2.23
1.46
2.01
2.48
3.77
4.19
6.67
2.58
5.90
8.83
2.54
1.82
3.06
4.25
2.86
3.48
7.18
9.75
0.33
F Tests (Least vs. Most
Disturbed)
Mountains
190.67
115.05
103.99
30.05
57.91
41.67
24.43
61.76
50.12
71.80
163.82
89.03
10.48
1.62
5.55
38.12
25.60
15.56
3.47
5.53
6.38
52.42
57.20
29.50
140.14
172.83
66.57
0.07
30.82
23.66
84.64
3.64
90.41
173.05
85.25
0.00
Plains
10.20
2.74
0.07
0.78
2.36
2.92
0.07
2.12
0.77
7.62
13.73
18.89
5.51
2.00
7.19
1.72
0.03
10.65
10.55
7.21
7.93
5.39
9.17
23.05
12.98
14.78
26.42
3.95
2.13
12.14
5.31
12.03
13.49
24.66
22.07
0.14
Xeric
81.45
61.38
15.62
5.14
14.84
0.06
7.09
18.24
5.19
31.57
93.15
83.79
0.75
0.68
14.76
22.05
0.02
25.04
17.74
30.94
33.59
19.27
25.79
40.46
56.58
66.07
69.64
31.41
0.00
20.54
38.37
36.66
50.77
78.14
72.28
3.61
BN-14
-------
Metric ID
MEGLPTAX
MEGLRICH
NOINPIND
NOINPTAX
NOINRICH
NTOLPIND
NTOLPTAX
NTOLRICH
OLLEPIND
OLLEPTAX
OLLERICH
OMNIPIND
OMNIPTAX
OMNIRICH
PLECPIND
PLECPTAX
PLECRICH
PREDPIND
PREDPTAX
PREDRICH
SCRPPIND
SCRPPTAX
SCRPRICH
SHRDPIND
SHRDPTAX
SHRDRICH
SIMPSON
SPRLPIND
SPRLPTAX
SPRLRICH
SWIMPIND
SWIMPTAX
SWIMRICH
TOLRPIND
TOLRPTAX
TOLRRICH
TOTLRICH
TRICPIND
TRICPTAX
TRICRICH
Signal:Noise Ratios
Mountains
0.20
0.26
2.90
2.21
1.42
1.60
1.60
1.60
2.51
1.20
0.72
5.54
1.11
0.99
3.52
3.42
3.67
1.67
1.09
1.13
2.96
2.03
2.79
3.44
1.85
2.11
1.42
2.08
0.97
0.72
7.46
1.75
1.60
1.05
1.67
0.82
1.79
2.28
1.64
2.36
Plains
0.37
0.43
1.67
3.18
1.53
0.70
0.70
0.70
1.54
1.05
0.52
1.11
0.89
1.54
3.36
0.72
1.48
2.86
0.86
0.83
1.34
1.31
1.83
1.06
1.07
0.50
0.80
1.87
1.30
0.89
4.97
1.20
0.70
1.34
2.27
0.74
0.96
0.23
1.25
1.96
Xeric
0.55
0.47
3.63
3.98
2.17
1.84
1.84
1.84
2.23
1.38
0.60
4.01
1.49
1.13
2.09
2.46
3.78
5.63
0.98
1.39
3.78
2.26
3.79
2.77
1.28
2.94
2.03
3.05
1.40
1.36
5.52
2.35
1.84
2.28
3.11
1.48
4.10
1.20
2.61
3.81
F Tests (Least vs. Most
Disturbed)
Mountains
0.59
0.36
190.27
178.95
143.62
130.28
182.55
81.47
100.77
62.29
55.85
39.35
97.74
89.90
31.26
66.81
45.63
9.86
2.86
0.09
17.80
26.28
15.60
9.04
31.38
18.09
3.42
32.87
87.81
64.33
38.89
122.25
106.81
169.08
265.04
193.11
0.74
25.16
79.73
42.97
Plains
0.26
0.07
12.66
2.28
0.13
25.84
28.73
24.99
7.91
1.59
0.06
3.59
1.08
0.00
7.44
9.65
9.36
6.86
1.87
1.19
1.26
1.51
4.16
0.15
0.14
3.01
10.74
0.01
1.29
3.34
5.68
7.66
4.51
3.73
17.72
0.83
11.35
6.96
5.22
11.83
Xeric
4.55
4.24
95.03
101.95
21.12
86.60
84.63
77.78
39.30
44.62
19.97
23.86
30.02
14.64
21.24
36.30
43.36
0.05
1.92
7.20
11.29
3.27
11.75
4.47
13.58
31.41
23.07
16.53
21.91
1.86
10.51
14.97
6.22
85.05
111.54
38.65
29.73
26.81
33.72
44.48
BN-15
-------
Table MI-3. Spearman Correlation Coefficients for Final Metrics
Metric ID
BURRPIND
CLNGPTAX
CLNGRICH
COGARICH
DOM5PIND
EPHERICH
EPT_PTAX
EPT_RICH
HPRIME
NOINPIND
NOINPTAX
OMNIPTAX
SHRDRICH
TL05PTAX
TL05RICH
TOLRPTAX
li
m °-
1.00
-0.42
-0.21
0.34
-0.12
-0.09
-0.38
-0.14
0.14
0.43
0.49
0.28
-0.02
-0.37
-0.03
0.45
CD X
52
o o.
-0.42
1.00
0.78
-0.28
-0.12
0.42
0.76
0.52
0.11
-0.36
-0.43
-0.36
0.07
0.69
0.35
-0.61
Ox
^
-0.21
0.78
1.00
0.22
-0.54
0.68
0.68
0.85
0.58
-0.14
-0.27
-0.27
0.43
0.65
0.79
-0.45
< x
°1
0 ^
0.34
-0.28
0.22
1.00
-0.57
0.36
-0.19
0.33
0.63
0.28
0.19
0.07
0.32
-0.13
0.53
0.28
ii
Si
-0.12
-0.12
-0.54
-0.57
1.00
-0.50
-0.19
-0.60
-0.97
-0.18
0.01
0.06
-0.50
-0.20
-0.70
0.11
xx
Sj E
-0.09
0.42
0.68
0.36
-0.50
1.00
0.66
0.83
0.54
-0.10
-0.13
-0.19
0.32
0.48
0.70
-0.29
IX
l- <
Q- H
LU Q_
-0.38
0.76
0.68
-0.19
-0.19
0.66
1.00
0.76
0.19
-0.34
-0.36
-0.37
0.26
0.75
0.47
-0.58
t'3
LU o:
-0.14
0.52
0.85
0.33
-0.60
0.83
0.76
1.00
0.65
-0.07
-0.16
-0.25
0.57
0.62
0.88
-0.36
HPRIME
0.14
0.11
0.58
0.63
-0.97
0.54
0.19
0.65
1.00
0.20
0.01
-0.05
0.54
0.21
0.77
-0.08
Z Q
O?
•Z. Q-
0.43
-0.36
-0.14
0.28
-0.18
-0.10
-0.34
-0.07
0.20
1.00
0.61
0.23
0.06
-0.31
0.04
0.36
=? X
M
0.28
-0.36
-0.27
0.07
0.06
-0.19
-0.37
-0.25
-0.05
0.23
0.31
1.00
-0.14
-0.38
-0.20
0.33
Q x
£o
W*
-0.02
0.07
0.43
0.32
-0.50
0.32
0.26
0.57
0.54
0.06
-0.10
-0.14
1.00
0.30
0.65
-0.19
P<
-Z. Q.
-0.37
0.69
0.65
-0.13
-0.20
0.48
0.75
0.62
0.21
-0.31
-0.45
-0.38
0.30
1.00
0.58
-0.68
n
ZL <£
-0.03
0.35
0.79
0.53
-0.70
0.70
0.47
0.88
0.77
0.04
-0.12
-0.20
0.65
0.58
1.00
-0.29
a: x
— i <
OH
1- Q-
0.45
-0.61
-0.45
0.28
0.11
-0.29
-0.58
-0.36
-0.08
0.36
0.63
0.33
-0.19
-0.68
-0.29
1.00
BN-16
-------
Table MI-4. Final Metrics, Order of Inclusion in MM/, and Ceiling/Floor Values
Metric
Class
Composition
Diversity
Feeding
Habit
Richness
Tolerance
Mountains
Metric
NOINPIND(S)
DOM5PIND (6)
OMNIPTAX(5)
BURRPIND (2)
EPT_RICH (4)
TOLRPTAX(1)
Ceiling
0
32
0
0
28
0
Floor
65
90
6
20
5
30
Plains
Metric
EPT_PTAX (4)
HPRIME (5)
COGARICH (6)
GINGRICH (3)
EPHERICH (2)
NTOLRICH(1)
Ceiling
45
3
16
7
10
25
Floor
2
0.1
4
0
0
3
Xeric
Metric
NOINPTAX (2)
HPRIME (5)
SHRDRICH (6)
CLNGPTAX (3)
EPT_RICH (4)
NTOLPTAX(1)
Ceiling
3
3.1
6
35
18
75
Floor
36
0.5
1
0
1
15
Table MI-5. Performance of MM/ and O/E Models
Multi-Metric Index
West-wide
Mountains
Plains
Xeric
Observed/Expected Index
West-wide
Mountains
Plains
Xeric
Signal: Noise Ratio
5.55
3.05
2.95
9.39
2.41
2.22
1.44
2.59
Ftest
781.1
408.0
47.7
210.3
353.9
151.4
20.0
156.3
S.D. of reference sites*
0.42
0.13
0.32
0.2
0.22
0.21
0.27
0.22
* MMI values re-scaled by dividing all scores by mean of reference sites (to mimic scale of O/E model)
BN-17
-------
Figures
Class
Figure BN-1. Location of Sampling Sites in Five Clusters Used in O/E Modeling
BN-18
-------
-3
123456
NEW K5
0 •
i
0123456
NEW K5
4
3
D 2
123456
NEW K5
w
<
g
10
123456
NEW K5
a 5 •
LU
D.
O
0 •
0
ttt*
0123456
NEW K5
0
t
123456
NEW K5
123456
NEW K5
Figure BN-2. Distribution of Key Geographic Variables in Five
Clusters Used in O/E Modeling
BN-19
-------
[Boxes in box-and-whisker plots indicate interquartile range and median (center line);
whiskers show 10th and 90th percentiles; dots indicate 5th and 95th percentile values)
Macroinvertebrate Multi-Metric Index
Mountainous Ecoregions
100
80 -
60 -
8
CO
40 H
20 -
0
^y
Calibration Validation Calibration Validation Calibration Validation
Least Disturbed Moderately Disturbed Most Disturbed
Figure BN-3. MMI Results for Calibration and Validation Datasets -
Mountain Ecoregions
BN-20
-------
[Boxes in box-and-whisker plots indicate interquartile range and median (center line);
whiskers show 10th and 90th percentiles; dots indicate 5th and 95th percentile values)
Macroinvertebrate Multi-Metric Index
Plains Ecoregions
100
o
o
CO
80 -
60 -
— 40 -
20 -
0
Calibration Validation Calibration Validation Calibration Validation
Least Disturbed Moderately Disturbed Most Disturbed
Figure BN-4. MMI Results for Calibration and Validation Datasets
Plains Ecoregions
BN-21
-------
[Boxes in box-and-whisker plots indicate interquartile range and median (center line);
whiskers show 10th and 90th percentiles; dots indicate 5th and 95th percentile values)
Macroinvertebrate Multi-Metric Index
Xeric Ecoregions
100
80 -
60 -
8
CO
40 H
20 -
0
1
T
Calibration Validation Calibration Validation Calibration Validation
Least Disturbed Moderately Disturbed Most Disturbed
Figure BN-5. MM/ Results for Calibration and Validation Datasets -
Xeric Ecoregions
BN-22
-------
[Boxes in box-and-whisker plots indicate interquartile range and median (center line);
whiskers show 10th and 90th percentiles; dots indicate 5th and 95th percentile values)
1.2 -
1.0 -
(U
x
UJ
0.6 -
S °-
0.2 -
0.0
Observed/Expected Macroinvertebrate Index
Westwide
Calibration Validation Calibration Validation Calibration Validation
Least Disturbed
Moderately Disturbed
Most Disturbed
Figure BN-6. O/E Results for Calibration and Validation Datasets
BN-23
-------
Presentation of Results for Indices and Selected Metrics
The following pages present empirical cumulative distribution (CDF) plots for the Multi-
Metric Index and its component metrics, the 0/E Index, and a small number of special
interest metrics. Please refer to the following table to decipher the somewhat cryptic
metric names used in plots. The distributions for each variable are presented West-
wide, for each of the three climatic/topographic regions, and for 10 aggregate
ecoregions (see Figures 1 and DE-6 for the locations of ecological regions), along with
a summary of each distribution's statistical parameters. For an explanation of how to
interpret CDFs, please see the section "How to Use this Report" earlier.
Metric or Index
Description
BUG_MMI
OBS_EXP
BURRPIND
CLNGPTAX
GINGRICH
COGARICH
DOM5PIND
EPHERICH
EPT_PTAX
EPT_RICH
HPRIME
INTLPTAX
INTLRICH
NOINPIND
NOINPTAX
NTOLPTAX
NTOLRICH
OMNIPTAX
SHRDRICH
TOLRPTAX
TOLRRICH
Multi-metric Index for Macroinvertebrates
Observed/Expected Macroinvertebrate Richness
Burrower % Individuals
Clinger % Distinct Taxa
Clinger Distinct Taxa Richness
Collector-Gatherer Distinct Taxa Richness
Percent of Individuals in Top 5 Taxa
Ephemeroptera Distinct Taxa Richness
EPT % Distinct Taxa
EPT Distinct Taxa Richness
Shannon Diversity Index
Intolerant % Distinct Taxa
Intolerant Distinct Taxa Richness
Non-Insect % Individuals
Non-Insect % Distinct Taxa
Non-Tolerant (Pollution Tolerance<6) % Distinct Taxa
Non-Tolerant (Pollution Tolerance<6) Distinct Taxa Richness
Omnivore % Distinct Taxa
Shredder % Distinct Taxa
Tolerant % Distinct Taxa
Total Distinct Taxa Richness
BN-24
-------
Figure BN-1 Indicator: BUG_MMI Subpopulation: West-wide
Empirical Cumulative Distribution Estimate
-------
Figure BN-2 Indicator: BUG_MMI Subpopulation: MT
Empirical Cumulative Distribution Estimate
-------
Figure BN-3 Indicator: BUG_MMI Subpopulation: PL
Empirical Cumulative Distribution Estimate
-------
Figure BN-4 Indicator: BUG_MMI Subpopulation: XE
Empirical Cumulative Distribution Estimate
-------
Figure BN-5 Indicator: BUG_MMI Subpopulation: MT-NROCK
Empirical Cumulative Distribution Estimate
8 55
100
Summary Statistics
Empirical Density Estimate
Est
37.07
39.04
61.49
71.60
79.02
85.61
89.58
69.06
14.41
LCB
24.99
37.11
58.15
70.86
77.43
83.48
86.52
66.80
12.91
UCB
38.95
50.25
66.76
73.96
81.99
89.02
90.83
71.33
15.90
20 40 60
Benthos MMI Score
I
80
100
BN-29
-------
Figure BN-6 Indicator: BUG_MMI Subpopulation: MT-PNW
Empirical Cumulative Distribution Estimate
•in
8
100
Summary Statistics
Empirical Density Estimate
Est
33.83
40.81
61.45
74.01
85.69
89.65
92.02
70.87
17.19
LCB
26.23
35.28
51.21
72.74
83.45
88.53
90.91
68.19
15.28
UCB
38.56
48.93
67.10
77.86
87.91
91.96
93.57
73.54
19.11
20 40 60
Benthos MMI Score
I
80
100
BN-30
-------
Figure BN-7 Indicator: BUG_MMI Subpopulation: MT-SROCK
Empirical Cumulative Distribution Estimate
-------
Figure BN-8 Indicator: BUG_MMI Subpopulation: MT-SWEST
Empirical Cumulative Distribution Estimate
- (u
•^ w
O)
-ID
ID
100
Summary Statistics
Empirical Density Estimate
Est
11.20
26.12
40.98
58.63
67.44
73.30
77
53.68
18.23
LCB
0
8.72
26.77
54.29
64.81
72.55
73.58
49.21
15.39
UCB
26.14
34.54
52.20
63.36
72.54
76.77
78.59
58.15
21.08
20 40 60
Benthos MMI Score
I
80
100
BN-32
-------
Figure BN-9 Indicator: BUG_MMI Subpopulation: PL-NCULT
Empirical Cumulative Distribution Estimate
-------
Figure BN-10 Indicator: BUG_MMI Subpopulation: PL-RANGE
Empirical Cumulative Distribution Estimate
-------
Figure BN-11 Indicator: BUG_MMI Subpopulation: XE-CALIF
Empirical Cumulative Distribution Estimate
-------
Figure BN-12 Indicator: BUG_MMI Subpopulation: XE-EPLAT
Empirical Cumulative Distribution Estimate
-------
Figure BN-13 Indicator: BUG_MMI Subpopulation: XE-NORTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-14 Indicator: BUG_MMI Subpopulation: XE-SOUTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-15 Indicator: OBS_EXP Subpopulation: West-wide
Empirical Cumulative Distribution Estimate
-------
Figure BN-16 Indicator: OBS_EXP Subpopulation: MT
Empirical Cumulative Distribution Estimate
-------
Figure BN-17 Indicator: OBS_EXP Subpopulation: PL
Empirical Cumulative Distribution Estimate
-------
Figure BN-18 Indicator: OBS_EXP Subpopulation: XE
Empirical Cumulative Distribution Estimate
CDF estimate
95% Confidence Limits
I
0.5
CO "
in E
D)
C
-------
Figure BN-19 Indicator: OBS_EXP Subpopulation: MT-NROCK
Empirical Cumulative Distribution Estimate
-------
Figure BN-20 Indicator: OBS_EXP Subpopulation: MT-PNW
Empirical Cumulative Distribution Estimate
E
•00 ^
O *-'
10
in
O>
•in
1.0
1.5
Summary Statistics
Est
0.22
0.38
0.59
0.82
1.01
1.12
1.17
0.79
0.27
LCB
0.22
0.23
0.53
0.74
0.97
1.10
1.15
0.75
0.24
Benthos O/E Score
UCB
0.37
0.46
0.66
0.86
1.04
1.17
1.28
0.83
0.29
Empirical Density Estimate
0.0 0.2 0.4 0.6 0.8 1.0
Benthos O/E Score
1.2
BN-44
-------
Figure BN-21 Indicator: OBS_EXP Subpopulation: MT-SROCK
Empirical Cumulative Distribution Estimate
-------
Figure BN-22 Indicator: OBS_EXP Subpopulation: MT-SWEST
Empirical Cumulative Distribution Estimate
- (u
•^ w
O)
-ID
ID
1.0
1.5
Summary Statistics
Est
0.11
0.24
0.60
0.72
0.84
0.96
1.08
0.70
0.25
LCB
0
0.11
0.35
0.69
0.83
0.96
0.96
0.63
0.20
Benthos O/E Score
UCB
0.24
0.48
0.60
0.81
0.96
1.16
1.30
0.77
0.30
Empirical Density Estimate
\
o.o
0.2 0.4 0.6 0.8 1.0
Benthos O/E Score
I
1.2
1.4
BN-46
-------
Figure BN-23 Indicator: OBS_EXP Subpopulation: PL-NCULT
Empirical Cumulative Distribution Estimate
-------
Figure BN-24 Indicator: OBS_EXP Subpopulation: PL-RANGE
Empirical Cumulative Distribution Estimate
-------
Figure BN-25 Indicator: OBS_EXP Subpopulation: XE-CALIF
Empirical Cumulative Distribution Estimate
-------
Figure BN-26 Indicator: OBS_EXP Subpopulation: XE-EPLAT
Empirical Cumulative Distribution Estimate
E
• <9 ™
_CO
^O)
CO
1.0
1.5
Summary Statistics
Est
0.35
0.46
0.51
0.64
0.81
0.92
0.92
0.65
0.19
LCB
0.23
0.32
0.46
0.58
0.72
0.86
0.92
0.61
0.17
Benthos O/E Score
UCB
0.43
0.46
0.58
0.69
0.92
1
1.05
0.70
0.21
Empirical Density Estimate
\ \ \ \ \
0.0 0.2 0.4 0.6 0.8
Benthos O/E Score
1.0
BN-50
-------
Figure BN-27 Indicator: OBS_EXP Subpopulation: XE-NORTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-28 Indicator: OBS_EXP Subpopulation: XE-SOUTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-29 Indicator: EPT_RICH Subpopulation: West-wide
Empirical Cumulative Distribution Estimate
-------
Figure BN-30 Indicator: EPT_RICH Subpopulation: MT
Empirical Cumulative Distribution Estimate
-------
Figure BN-31 Indicator: EPT_RICH Subpopulation: PL
Empirical Cumulative Distribution Estimate
-------
Figure BN-32 Indicator: EPT_RICH Subpopulation: XE
Empirical Cumulative Distribution Estimate
-------
Figure BN-33 Indicator: EPT_RICH Subpopulation: MT-NROCK
Empirical Cumulative Distribution Estimate
-------
Figure BN-34 Indicator: EPT_RICH Subpopulation: MT-PNW
Empirical Cumulative Distribution Estimate
-------
Figure BN-35 Indicator: EPT_RICH Subpopulation: MT-SROCK
-------
Figure BN-36 Indicator: EPT_RICH Subpopulation: MT-SWEST
Empirical Cumulative Distribution Estimate
- (u
•^ w
O)
-ID
ID
Summary Statistics
Empirical Density Estimate
Statistic Est LCB UCB
5Pct 1.20 0 1.90
10Pct 2.09 0.81 4.03
25Pct 6.36 3.43 9.31
SOPct 10.42 9.65 11.11
75Pct 13.86 12 15.83
90Pct 16.74 15.61 19.65
95Pct 19.15 17.38 21.17
Mean 10.71 9.62 11.80
Std Dev 5.18 4.58 5.78
0 5
III,,
10 15 20 25
No. of EPT Taxa
BN-60
-------
Figure BN-37 Indicator: EPT_RICH Subpopulation: PL-NCULT
Empirical Cumulative Distribution Estimate
-------
Figure BN-38 Indicator: EPT_RICH Subpopulation: PL-RANGE
Empirical Cumulative Distribution Estimate
-------
Figure BN-39 Indicator: EPT_RICH Subpopulation: XE-CALIF
Empirical Cumulative Distribution Estimate
-------
Figure BN-40 Indicator: EPT_RICH Subpopulation: XE-EPLAT
Empirical Cumulative Distribution Estimate
-------
Figure BN-41 Indicator: EPT_RICH Subpopulation: XE-NORTH
Empirical Cumulative Distribution Estimate
o
o _
00
o
to
o
^r
o
OM
o -
__
• . . — •
—
...'" —
. —
i i i
0 10 20
No. of EPT Taxa
Summary Statistics
Statistic Est LCB UCB
5Pct 0 0 0.56
10Pct 0 0 1.11
25Pct 1.58 0 6.81
SOPct 9.88 3.55 12.35
75Pct 13.81 10.71 19.89
90Pct 19.74 14.75 23
95Pct 22.24 18.12 23
Mean 9.61 6.95 12.26
0
Std Dev 7.02 6.08 7.95
CDF estimate
95% Confidence Limits
i
30 40
o
00
-O)
0
_s
r^
00
oo •£•
oo S
in £,
<° £
'DJ
C
CN
-O
Empirical Density Estimate
1 i, .
5 10 15 20 25
No. of EPT Taxa
-------
Figure BN-42 Indicator: EPT_RICH Subpopulation: XE-SOUTH
-------
Figure BN-43 Indicator: EPT_PTAX Subpopulation: West-wide
Empirical Cumulative Distribution Estimate
-------
Figure BN-44 Indicator: EPT_PTAX Subpopulation: MT
Empirical Cumulative Distribution Estimate
-------
Figure BN-45 Indicator: EPT_PTAX Subpopulation: PL
Empirical Cumulative Distribution Estimate
-------
Figure BN-46 Indicator: EPT_PTAX Subpopulation: XE
Empirical Cumulative Distribution Estimate
-------
Figure BN-47 Indicator: EPT_PTAX Subpopulation: MT-NROCK
Empirical Cumulative Distribution Estimate
8 w
Summary Statistics
Empirical Density Estimate
Est
21.02
26.98
36.59
43.58
51.43
59.43
62.12
43.08
11.77
LCB
11.31
24.52
33.07
42.27
49.57
54.31
59.50
41.26
10.34
UCB
26.20
30.68
37.99
46.54
53.17
62.03
66.01
44.89
13.21
20 40 60
% Taxa That are EPT
I
80
100
BN-71
-------
Figure BN-48 Indicator: EPT_PTAX Subpopulation: MT-PNW
Empirical Cumulative Distribution Estimate
•in
8
Summary Statistics
Empirical Density Estimate
Est
14.38
19.86
32.62
40.69
50.93
56.46
62.97
40.63
13.15
LCB
12
15.03
27.99
38.79
49.25
55.35
58.11
38.68
11.86
UCB
19.52
22.64
35.35
44.65
52.81
63.09
67.80
42.58
14.44
20 40 60
% Taxa That are EPT
I
80
100
BN-72
-------
Figure BN-49 Indicator: EPT_PTAX Subpopulation: MT-SROCK
Empirical Cumulative Distribution Estimate
-------
Figure BN-50 Indicator: EPT_PTAX Subpopulation: MT-SWEST
Empirical Cumulative Distribution Estimate
- (u
•^ w
O)
-ID
ID
Summary Statistics
Empirical Density Estimate
Est
6.31
8.72
23.43
30.40
34.78
42.21
44.73
28.25
11.81
LCB
0
2.85
13.25
27.11
33.11
40.64
41.64
25.60
10.25
UCB
8.66
13.32
25.48
32.85
40.71
48.26
54.76
30.90
13.38
20 40 60
% Taxa That are EPT
I
80
100
BN-74
-------
Figure BN-51 Indicator: EPT_PTAX Subpopulation: PL-NCULT
Empirical Cumulative Distribution Estimate
-------
Figure BN-52 Indicator: EPT_PTAX Subpopulation: PL-RANGE
Empirical Cumulative Distribution Estimate
-------
Figure BN-53 Indicator: EPT_PTAX Subpopulation: XE-CALIF
Empirical Cumulative Distribution Estimate
-------
Figure BN-54 Indicator: EPT_PTAX Subpopulation: XE-EPLAT
Empirical Cumulative Distribution Estimate
-------
Figure BN-55 Indicator: EPT_PTAX Subpopulation: XE-NORTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-56 Indicator: EPT_PTAX Subpopulation: XE-SOUTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-57 Indicator: EPHERICH Subpopulation: West-wide
-------
Figure BN-58 Indicator: EPHERICH Subpopulation: MT
-------
Figure BN-59 Indicator: EPHERICH Subpopulation: PL
-------
Figure BN-60 Indicator: EPHERICH Subpopulation: XE
-------
Figure BN-61 Indicator: EPHERICH Subpopulation: MT-NROCK
-------
Figure BN-62 Indicator: EPHERICH Subpopulation: MT-PNW
-------
Figure BN-63 Indicator: EPHERICH Subpopulation: MT-SROCK
-------
Figure BN-64 Indicator: EPHERICH Subpopulation: MT-SWEST
-------
Figure BN-65 Indicator: EPHERICH Subpopulation: PL-NCULT
-------
Figure BN-66 Indicator: EPHERICH Subpopulation: PL-RANGE
-------
Figure BN-67 Indicator: EPHERICH Subpopulation: XE-CALIF
-------
Figure BN-68 Indicator: EPHERICH Subpopulation: XE-EPLAT
-------
Figure BN-69 Indicator: EPHERICH Subpopulation: XE-NORTH
4 6 8 10 12 14
Std Dev 3.06 2.62 3.50 No. of Ephemeroptera Taxa
BN-93
-------
Figure BN-70 Indicator: EPHERICH Subpopulation: XE-SOUTH
Empirical Cumulative Distribution Estimate
o
o —
o _
00
o
CD
O —
"~
o -
CDF estimate
95% Confidence Limits
1 1 1
0 5 10
No. of Ephemeroptera Taxa
0
CO
O)
CM
00 §
If)
^ .c
'DJ
-------
Figure BN-71 Indicator: NOINPIND Subpopulation: West-wide
Empirical Cumulative Distribution Estimate
-------
Figure BN-72 Indicator: NOINPIND Subpopulation: MT
Empirical Cumulative Distribution Estimate
-------
Figure BN-73 Indicator: NOINPIND Subpopulation: PL
Empirical Cumulative Distribution Estimate
-------
Figure BN-74 Indicator: NOINPIND Subpopulation: XE
Empirical Cumulative Distribution Estimate
-------
Figure BN-75 Indicator: NOINPIND Subpopulation: MT-NROCK
Empirical Cumulative Distribution Estimate
8 55
80
100
Summary Statistics
Empirical Density Estimate
Est
2.02
3
4.99
9.63
20.12
33.88
41.98
14.32
12.99
LCB
0
2.17
4.20
7.20
16.35
25.63
35.46
12.38
11.43
UCB
2.53
3.83
5.33
10.69
25.03
42.05
63.82
16.25
14.55
20 40 60 80
% Indiv. that are Not Insects
100
BN-99
-------
Figure BN-76 Indicator: NOINPIND Subpopulation: MT-PNW
Empirical Cumulative Distribution Estimate
•in
8
80
100
Summary Statistics
Empirical Density Estimate
Est
1.80
2.98
4.98
9.62
20.23
34.75
47.65
15.43
13.88
LCB
1.14
1.80
4.08
7.99
16.27
30.64
39.76
12.94
11.66
UCB
2.18
3.84
5.99
11.22
25.95
46.81
67.42
17.92
16.09
20 40 60 80
% Indiv. that are Not Insects
100
BN-100
-------
Figure BN-77 Indicator: NOINPIND Subpopulation: MT-SROCK
Empirical Cumulative Distribution Estimate
-------
Figure BN-78 Indicator: NOINPIND Subpopulation: MT-SWEST
Empirical Cumulative Distribution Estimate
-------
Figure BN-79 Indicator: NOINPIND Subpopulation: PL-NCULT
Empirical Cumulative Distribution Estimate
^in
80
100
Summary Statistics
Empirical Density Estimate
Est
1.31
4.99
9.63
18.90
45.34
76.10
79.31
29.90
23.93
LCB
0
1.08
6.37
16.17
31.28
57.29
76.07
24.34
20.12
UCB
4.98
6.40
14.86
26.23
59.41
82.37
96.33
35.47
27.73
20 40 60 80
% Indiv. that are Not Insects
100
BN-103
-------
Figure BN-80 Indicator: NOINPIND Subpopulation: PL-RANGE
Empirical Cumulative Distribution Estimate
-------
Figure BN-81 Indicator: NOINPIND Subpopulation: XE-CALIF
Empirical Cumulative Distribution Estimate
-------
Figure BN-82 Indicator: NOINPIND Subpopulation: XE-EPLAT
Empirical Cumulative Distribution Estimate
-------
Figure BN-83 Indicator: NOINPIND Subpopulation: XE-NORTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-84 Indicator: NOINPIND Subpopulation: XE-SOUTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-85 Indicator: NOINPTAX Subpopulation: West-wide
Empirical Cumulative Distribution Estimate
-------
Figure BN-86 Indicator: NOINPTAX Subpopulation: MT
Empirical Cumulative Distribution Estimate
-------
Figure BN-87 Indicator: NOINPTAX Subpopulation: PL
Empirical Cumulative Distribution Estimate
-------
Figure BN-88 Indicator: NOINPTAX Subpopulation: XE
Empirical Cumulative Distribution Estimate
-------
Figure BN-89 Indicator: NOINPTAX Subpopulation: MT-NROCK
Empirical Cumulative Distribution Estimate
8 w
Summary Statistics
Empirical Density Estimate
Est
5
6.82
10.53
13.51
17.21
21.19
23.92
14.23
6.21
LCB
0
5.17
9.55
12.35
16.71
19.50
21.56
13.34
5.29
UCB
6.07
8.43
10.86
15.20
18.69
23.90
39.16
15.13
7.13
20 40 60 80
% of Taxa that are Not Insects
100
BN-113
-------
Figure BN-90 Indicator: NOINPTAX Subpopulation: MT-PNW
Empirical Cumulative Distribution Estimate
•in
8
Summary Statistics
Empirical Density Estimate
Est
6.17
6.78
9.96
13.68
18.30
22.85
28.33
15.18
6.93
LCB
4.77
6.20
8.24
13.02
16.53
22.20
23.66
13.98
5.80
UCB
6.69
7.60
10.90
14.94
21.88
28.30
43.10
16.37
8.05
20 40 60 80
% of Taxa that are Not Insects
100
BN-114
-------
Figure BN-91 Indicator: NOINPTAX Subpopulation: MT-SROCK
Empirical Cumulative Distribution Estimate
-------
Figure BN-92 Indicator: NOINPTAX Subpopulation: MT-SWEST
Empirical Cumulative Distribution Estimate
- (u
•^ w
O)
-ID
ID
Summary Statistics
Empirical Density Estimate
Est
5.79
6.44
8.84
14.63
20.77
26.60
34.65
15.52
7.86
LCB
2.99
3.46
7.36
12.09
17.10
21.91
26.61
13.56
6.62
UCB
6.43
7.42
11.71
17.02
24.58
38
38.07
17.49
9.09
20 40 60 80
% of Taxa that are Not Insects
100
BN-116
-------
Figure BN-93 Indicator: NOINPTAX Subpopulation: PL-NCULT
Empirical Cumulative Distribution Estimate
-------
Figure BN-94 Indicator: NOINPTAX Subpopulation: PL-RANGE
Empirical Cumulative Distribution Estimate
-------
Figure BN-95 Indicator: NOINPTAX Subpopulation: XE-CALIF
Empirical Cumulative Distribution Estimate
-------
Figure BN-96 Indicator: NOINPTAX Subpopulation: XE-EPLAT
Empirical Cumulative Distribution Estimate
-------
Figure BN-97 Indicator: NOINPTAX Subpopulation: XE-NORTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-98 Indicator: NOINPTAX Subpopulation: XE-SOUTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-99 Indicator: INTLRICH Subpopulation: West-wide
Empirical Cumulative Distribution Estimate
-------
Figure BN-100 Indicator: INTLRICH Subpopulation: MT
Empirical Cumulative Distribution Estimate
-------
Figure BN-101 Indicator: INTLRICH Subpopulation: PL
Empirical Cumulative Distribution Estimate
-------
Figure BN-102 Indicator: INTLRICH Subpopulation: XE
Empirical Cumulative Distribution Estimate
-------
Figure BN-103 Indicator: INTLRICH Subpopulation: MT-NROCK
CD
CL
O _
O _
OM^
Empirical Cumulative Distribution Estimate
ID
O)
-00
00
O)
CDF estimate
95% Confidence Limits
10
I i
15 20
No. Intolerant Taxa
25
I
30
i
35
-8
-8
O)
O)
Summary Statistics
Empirical Density Estimate
Statistic Est LCB UCB
5Pct 5.29 1 6.67
10Pct 7.24 5.85 7.92
25Pct 11.60 9.05 12.54
SOPct 16.03 14.83 17.64
75Pct 20.20 19.21 21.02
90Pct 23.05 21.97 25.57
95Pct 25.74 24.11 32.11
Mean 16.17 15.25 17.10
Illl
l i
0 5
Std Dev 5.93 5.37 6.49
Illl
10 15 20 25 30 35
No. Intolerant Taxa
BN-127
-------
Figure BN-104 Indicator: INTLRICH Subpopulation: MT-PNW
-------
Figure BN-105 Indicator: INTLRICH Subpopulation: MT-SROCK
Empirical Cumulative Distribution Estimate
-------
Figure BN-106 Indicator: INTLRICH Subpopulation: MT-SWEST
Empirical Cumulative Distribution Estimate
-------
Figure BN-107 Indicator: INTLRICH Subpopulation: PL-NCULT
Empirical Cumulative Distribution Estimate
4 6 8 10
No. Intolerant Taxa
BN-131
-------
Figure BN-108 Indicator: INTLRICH Subpopulation: PL-RANGE
Empirical Cumulative Distribution Estimate
-------
Figure BN-109 Indicator: INTLRICH Subpopulation: XE-CALIF
-------
Figure BN-110 Indicator: INTLRICH Subpopulation: XE-EPLAT
Empirical Cumulative Distribution Estimate
-------
Figure BN-111 Indicator: INTLRICH Subpopulation: XE-NORTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-112 Indicator: INTLRICH Subpopulation: XE-SOUTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-113 Indicator: INTLPTAX Subpopulation: West-wide
Empirical Cumulative Distribution Estimate
-------
Figure BN-114 Indicator: INTLPTAX Subpopulation: MT
Empirical Cumulative Distribution Estimate
-------
Figure BN-115 Indicator: INTLPTAX Subpopulation: PL
Empirical Cumulative Distribution Estimate
-------
Figure BN-116 Indicator: INTLPTAX Subpopulation: XE
Empirical Cumulative Distribution Estimate
-------
Figure BN-117 Indicator: INTLPTAX Subpopulation: MT-NROCK
Empirical Cumulative Distribution Estimate
8 w
Summary Statistics
Empirical Density Estimate
Est
18.64
21.98
31.47
39.09
47.34
53.43
57.01
38.02
11.56
LCB
9.63
19.46
27.52
35.76
43.69
49.69
53.93
36.22
10.28
UCB
21.18
25.85
32.47
40.93
48.93
56.95
59.57
39.82
12.83
20 40 60 80
% Taxa that are Intolerant
100
BN-141
-------
Figure BN-118 Indicator: INTLPTAX Subpopulation: MT-PNW
Empirical Cumulative Distribution Estimate
•in
8
Summary Statistics
Empirical Density Estimate
Est
9.41
15.19
27.09
35.85
45.66
49.88
53.55
35.34
12.70
LCB
8.51
10.05
22.44
34.43
43.92
47.97
51.45
33.41
11.44
UCB
13.22
17.87
30.31
38.30
46.94
54.52
64.28
37.27
13.95
20 40 60 80
% Taxa that are Intolerant
100
BN-142
-------
Figure BN-119 Indicator: INTLPTAX Subpopulation: MT-SROCK
Empirical Cumulative Distribution Estimate
-------
Figure BN-120 Indicator: INTLPTAX Subpopulation: MT-SWEST
Empirical Cumulative Distribution Estimate
- (u
•^ w
O)
-ID
ID
Summary Statistics
Empirical Density Estimate
Est
3.20
4.47
6.55
12.90
24.10
32.65
36.18
16.06
9.37
LCB
0
3.14
5.63
10.96
17.44
27.72
32.94
14.37
7.66
UCB
4.29
5.74
9.58
15.70
26.03
39.22
51.05
17.76
11.09
20 40 60 80
% Taxa that are Intolerant
100
BN-144
-------
Figure BN-121 Indicator: INTLPTAX Subpopulation: PL-NCULT
Empirical Cumulative Distribution Estimate
-------
Figure BN-122 Indicator: INTLPTAX Subpopulation: PL-RANGE
Empirical Cumulative Distribution Estimate
-------
Figure BN-123 Indicator: INTLPTAX Subpopulation: XE-CALIF
Empirical Cumulative Distribution Estimate
-------
Figure BN-124 Indicator: INTLPTAX Subpopulation: XE-EPLAT
Empirical Cumulative Distribution Estimate
-------
Figure BN-125 Indicator: INTLPTAX Subpopulation: XE-NORTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-126 Indicator: INTLPTAX Subpopulation: XE-SOUTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-127 Indicator: TOLRRICH Subpopulation: West-wide
-------
Figure BN-128 Indicator: TOLRRICH Subpopulation: MT
-------
Figure BN-129 Indicator: TOLRRICH Subpopulation: PL
-------
Figure BN-130 Indicator: TOLRRICH Subpopulation: XE
-------
Figure BN-131 Indicator: TOLRRICH Subpopulation: MT-NROCK
Empirical Cumulative Distribution Estimate
8 w
Empirical Density Estimate
I I
5 10
No. Tolerant Taxa
15
BN-155
-------
Figure BN-132 Indicator: TOLRRICH Subpopulation: MT-PNW
•in
8
Summary Statistics
Empirical Density Estimate
Statistic Est LCB UCB
5Pct 1.06 0.14 1.70
10Pct 1.92 1.30 2.17
25Pct 2.80 2.53 3.05
SOPct 4.05 3.69 4.67
75Pct 5.95 5.51 7.14
90Pct 8.38 7.63 9.38
95Pct 9.41 8.63 10.55
Mean 5.11 4.75 5.47
Std Dev 2.44 2.20 2.67
I , , ,
0 5 10 15
No. Tolerant Taxa
BN-156
-------
Figure BN-133 Indicator: TOLRRICH Subpopulation: MT-SROCK
-------
Figure BN-134 Indicator: TOLRRICH Subpopulation: MT-SWEST
-------
Figure BN-135 Indicator: TOLRRICH Subpopulation: PL-NCULT
Empirical Cumulative Distribution Estimate
-------
Figure BN-136 Indicator: TOLRRICH Subpopulation: PL-RANGE
-------
Figure BN-137 Indicator: TOLRRICH Subpopulation: XE-CALIF
Empirical Cumulative Distribution Estimate
-------
Figure BN-138 Indicator: TOLRRICH Subpopulation: XE-EPLAT
-------
Figure BN-139 Indicator: TOLRRICH Subpopulation: XE-NORTH
Empirical Cumulative Distribution Estimate
-------
Figure BN-140 Indicator: TOLRRICH Subpopulation: XE-SOUTH
-------
Figure BN-141 Indicator: TOLRPTAX Subpopulation: West-wide
Empirical Cumulative Distribution Estimate
-------
Figure BN-142 Indicator: TOLRPTAX Subpopulation: MT
Empirical Cumulative Distribution Estimate
-------
Figure BN-143 Indicator: TOLRPTAX Subpopulation: PL
Empirical Cumulative Distribution Estimate
-------
Figure BN-144 Indicator: TOLRPTAX Subpopulation: XE
Empirical Cumulative Distribution Estimate
-------
Figure BN-145 Indicator: TOLRPTAX Subpopulation: MT-NROCK
Empirical Cumulative Distribution Estimate
8 w
Summary Statistics
Empirical Density Estimate
Est
4.56
6.03
7.37
9.93
13.97
18.16
22.18
11.49
5.17
LCB
2.98
4.64
6.97
9.05
12.75
16.61
19.48
10.68
4.49
UCB
5.73
6.36
7.97
11.36
15.71
22.01
29.69
12.31
5.85
20 40 60 80
% Taxa that are Tolerant
100
BN-169
-------
Figure BN-146 Indicator: TOLRPTAX Subpopulation: MT-PNW
Empirical Cumulative Distribution Estimate
•in
8
Summary Statistics
Empirical Density Estimate
Est
3.61
4.83
6.53
9.79
15.96
19.88
24.60
11.44
6.31
LCB
0.93
3.81
5.71
8.64
13.22
17.75
20.42
10.48
5.50
UCB
4.51
5.59
7.09
11.09
17.09
24.68
30.20
12.40
7.11
20 40 60 80
% Taxa that are Tolerant
100
BN-170
-------
Figure BN-147 Indicator: TOLRPTAX Subpopulation: MT-SROCK
Empirical Cumulative Distribution Estimate
-------
Figure BN-148 Indicator: TOLRPTAX Subpopulation: MT-SWEST
Empirical Cumulative Distribution Estimate
- (u
•^ w
O)
-ID
ID
Summary Statistics
Empirical Density Estimate
Est
6.66
8.72
12.21
17.32
20.76
25.67
26.06
17.13
6.61
LCB
3.52
6.43
10.27
15.50
19.41
22.67
25.60
15.75
5.37
UCB
8.69
10.30
15.27
19.14
23.35
31.17
41.67
18.51
7.86
20 40 60 80
% Taxa that are Tolerant
100
BN-172
-------
Figure BN-149 Indicator: TOLRPTAX Subpopulation: PL-NCULT
Empirical Cumulative Distribution Estimate
-------
Figure BN-150 Indicator: TOLRPTAX Subpopulation: PL-RANGE
Empirical Cumulative Distribution Estimate
-------
Figure BN-151 Indicator: TOLRPTAX Subpopulation: XE-CALIF
Empirical Cumulative Distribution Estimate
|