DRAFT
DO NOT CITE OR QUOTE
EPA/600/R-11/036A
March 2011
Implications of Climate Change for Bioassessment Programs and
Approaches to Account for Effects
NOTICE
THIS DOCUMENT IS A PRELIMINARY DRAFT. THIS INFORMATION IS DISTRIBUTED
SOLELY FOR THE PURPOSE OF PRE-DISSEMINATION PEER REVIEW UNDER
APPLICABLE INFORMA TION QUALITY GUIDELINES IT HAS NOT BEEN FORMALLY
DISSEMINATED BY THE U.S. ENVIRONMENTAL PROTECTION AGENCY. IT DOES NOT
REPRESENT AND SHOULD NOT BE CONSTRUED TO REPRESENT ANY AGENCY
DETERMINATION OR POLICY.
Global Change Research Program
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20460

-------
DISCLAIMER
This document is an internal draft for review purposes only. It has not been subjected to
peer and administrative review and does not constitute U.S. Environmental Protection Agency
policy. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
March 28, 2011
ii
External Review Draft

-------
ABSTRACT
Climate change will affect stream ecosystems directly, indirectly, and through interactions with
other stressors. Biological responses to these changes include altered community composition,
interactions and functions. Effects will vary regionally and present heretofore unaccounted
influences on biomonitoring, which water-quality agencies use to assess the status and health of
ecosystems as required by the Clean Water Act. Biomonitoring, which uses biological indicators
and metrics to assess ecosystem condition, are anchored in comparisons to regionally established
reference benchmarks of ecological condition. Climate change will affect responses and
interpretation of these indicators and metrics at both reference and non-reference sites, and
therefore has the potential to confound the diagnosis of ecological condition. This report
analyzes four regionally-distributed state biomonitoring data sets to inform on how biological
indicators respond to the effects of climate change, what climate-specific indicators may be
available to detect effects, how well current sampling detects climate-driven changes, and how
program designs can continue to detect impairment. Results can be used to identify methods that
assist with detecting climate-related effects and highlight steps that can be taken to ensure that
programs continue to meet resource protection goals.
Preferred citation:
U.S. Environmental Protection Agency (U.S. EPA). (2011) Implications of climate change for bioassessment
programs and approaches to account for effects. Global Change Research Program, National Center for
Environmental Assessment, Washington, DC; EPA/600/R-11/036A. Available from the National Technical
Information Service, Springfield, VA, and online at http://www.epa. gov/ncea.
March 28, 2011
iii
External Review Draft

-------
Page
ABSTRACT	iii
LIST OF ABBREVIATIONS AND ACRONYMS	viii
PREFACE	ix
AUTHORS, CONTRIBUTORS, AND REVIEWERS	x
EXECUTIVE SUMMARY	xi
1.	CLIMATE CHANGE AND BIOASSESSMENTS	1
2.	ANALYSIS OF ECOLOGICAL TRAITS TO DETECT CLIMATE CHANGE EFFECTS ... 1
2.1.	BACKGROUND ANALYSES	1
2.2.	TRENDS IN ECOLOGICAL TRAIT GROUPS	3
2.2.1.	Ecological Trait Groups and Climate Patterns	8
2.2.2.	Ecological Trait Groups - Spatial Patterns, Elevation, and Size	14
2.2.3.	Potential biological indicators of climate-related hydrologic changes	23
2.3.	TEMPORAL TRENDS IN TAX A AND COMMONLY USED METRICS	28
2.3.1.	Trends and Patterns- Utah and Western States	29
2.3.2.	Trends and Patterns- Maine and New England States	35
2.3.3.	Trends and Patterns - North Carolina and Southeastern States	39
2.4.	CONFOUNDING SOURCES OF TEMPORAL VARIATION	41
2.4.1.	Seasonal variation	41
2.4.2.	Interannual and multi-decadal climatic variation	42
2.4.3.	Interpretation of directional climate change effects	48
2.5.	OTHER SOURCES OF POTENTIAL SPATIAL CONFOUNDING	48
2.6.	COMPARISON OF REGIONAL TRENDS, VULNERABILITIES, AND
INDICATORS	50
3.	IMPLICATIONS TO MULTIMETRIC INDICES, PREDICTIVE MODELS, AND
IMPAIRMENT/LISTING DECISIONS	 1
3.1.	maim: and the NORTHEAST	2
3.2.	NORTH CAROLINA AND THE SOUTHEAST	11
3.2.1.	Vulnerability of the North Carolina Bioclassification - Simulation of taxa
replacement	12
3.2.2.	EPT Taxa Richness Metric	15
3.2.3.	The North Carolina Biotic Index (NCBI)	17
3.3.	OHIO AM) I I Ii: MIDWEST	20
3.4.	UTAH AM) THE SOUTHWEST	25
3.4.1.	Approach	26
3.4.2.	RIVPACS Responses - Utah Decision Vulnerabilities	27
3.5	CONCLUSIONS ACROSS PILOT STUDY STATES	31
3.6	RECOMMENDATIONS FOR MODIFYING METRICS	32
4.	REFERENCE STATION VULNERABILITIES	 1
4.1.	Vulnerabilities in the reference station sampling design	1
4.2.	Vulnerabilities in assessing reference condition	2
4.2.1. Reference Stations Used in this Study	3
4.2.1. Climate Change Vulnerabilities of Reference Stations	5
4.3.	Synergistic effects between climate change and land use	7
4.4.	Future vulnerabilities of reference stations to land use	9
4.5.	SENTINEL MONITORING NETWORK	 16
March 28, 2011
iv
External Review Draft

-------
5. CHARACTERISTICS OF EXISTING BIO AS SES SMENT PROGRAMS RELEVANT TO
DISCERNING CLIMATE CHANGE TRENDS	 1
5.1 SUFFICIENCY AND LIMITATIONS OF DATA TO DEFINE AND PARTITION
LONG-TERM TRENDS	1
5.2. OTHER BIOMONITORING METHODS CONSIDERATIONS	7
6.	CLIMATE CHANGE IMPLICATIONS FOR ENVIRONMENTAL MANAGEMENT	1
6.1 IMPAIRMENT LISTINGS AM) TMDLS	 1
6.1.1.	Overview of impacts on impairment listings and TMDL development	1
6.1.2.	Approaches to evaluate impairment listings and TMDL development in the context
of climate change	2
6.2. WATER QUALITY STANDARDS AND BIOCRITERIA	3
6.2.1.	Overview of impacts on the development of water quality standards and biocriteria
3
6.2.2.	Approaches to modify the development of water quality standards and biocriteria
in the context of climate change	4
7.	CONCLUSIONS	1
8.	REFERENCES	1
APPENDIX A: BASIC EVIDENCE FOR CLIMATE CHANGE EFFECTS:
LONG-TERM TRENDS IN ANNUAL AIR TEMPERATURE,
PRECIPITATION AND WATER TEMPERATURES
APPENDIX B: DATA PREPARATION AND MANAGEMENT
APPENDIX C: SITE SELECTIONS AND SITE GROUPINGS
APPENDIX D: DATA ANALYSIS METHODS
APPENDIX E: DETAILED RESULTS FOR MAINE
APPENDIX F: DETAILED RESULTS FOR UTAH
APPENDIX G: DETAILED RESULTS FOR NORTH CAROLINA
APPENDIX H: TEMPORAL CHANGE IN REGIONAL REFERENCE CONDITION AS
A POTENTIAL INDICATOR OF GLOBAL CLIMATE CHANGE:
ANALYSIS OF THE OHIO REGIONAL REFERENCE CONDITION
DATABASE (1980-2006)
APPENDIX I: SELECTED SUBSETS OF RESULTS FROM CORRELATION
ANALYSES FOR MAINE, UTAH AND NORTH CAROLINA
APPENDIX J: CASE STUDIES
APPENDIX K: EXPLORATIONS OF RELATIONSHIPS BETWEEN HYDROLOGICAL
AND BIOLOGICAL DATA
March 28, 2011
V
External Review Draft

-------
LIST OF TABLES
Page
Table 2-1. Summary of analysis approach, by analysis type, main methods, and
overlying questions	2-3
Table 2-2. Thermal preference metric values in coldest, normal, and hottest year samples at
long-term biological monitoring sites in Utah, Maine, and North Carolina	2-8
Table 2-3. Selected results from the Pearson and Kendall Correlations with
Ordination Axes	2-19
Table 2-4. Results of Pearson product moment correlation analyses that examine associations
between year and a selected group of metrics at long-term biological monitoring sites in Utah,
Maine, and North Carolina	2-41
Table 2-5. Results of Pearson product moment correlation analyses that examine associations
between PRISM mean annual air temperature and a selected group of metrics at long-term
biological monitoring sites in Utah, Maine, and North Carolina	2-44
Table 2-6. Results of Pearson product moment correlation analyses that examine associations
between PRISM mean annual precipitation and a selected group of metrics at long-term
biological monitoring sites in Utah, Maine, and North Carolina	2-45
Table 2-7. Summary of differences in elevation, PRISM mean annual air temperature and
precipitation and mean number and percent of cold and warm-water-preference taxa across and
within major ecoregions in each state. Only full-scale samples were used to derive the numbers
for North Carolina. Samples were not limited to particular seasons in Utah and North Carolina.
Mean % individuals of cold and warm water in the North Carolina Coastal ecoregion were not
calculated (our analyses were concentrated in the Mountain and Piedmont ecoregions)	2-44
Table 3-1. Number of Maine cold-water taxa in each order with EPT taxa in italics	3-3
Table 3-2. Number of Maine cold-water taxa in each order with EPT taxa in italics	3-3
Table 3-3. Temperature trait information for Maine Class A indicator taxa	3-9
Table 3-4. Final bioclassification scores at 3 reference Mountain sites before and after all cold-
water-preference taxa are dropped from the sites	3-13
Table 3-5. Number of North Carolina cold-water-preference taxa in each order. EPT orders are
italicized	3-15
Table 3-6. Number of North Carolina warm-water-preference taxa in each order. EPT orders are
italicized	3-15
Table 3-7. EPT species richness values and scores at 3 reference Mountain
sites before and after all cold-water-preference taxa are dropped from the sites... 3-16
March 28, 2011
vi
External Review Draft

-------
Table 3-8. NCBI values and scores at 3 reference Mountain sites before and after all cold-water-
preference taxa are dropped from the sites	3-19
Table 3-9. Correlations of benthic taxa grouped by temperature traits with BI at
North Carolina Mountain and Piedmont reference stations	3-20
Table 4-1. Time periods for which biological data were available at the long-term
monitoring sites in Utah (UT), Maine (ME), and North Carolina (NC).
Data used in these analyses were limited to autumn (September-November)
kick-method samples in the Utah data set, summer (July-September) rock-
basket samples in the Maine data set, and summer (July-August) standard
qualitative samples in the North Carolina data set.	4-15
Table 4-2. Percent of existing Florida reference stations (N=58, classified as
"exceptional"), that have >20% developed land use (with 25 houses
per square mile or more, categories 5-12 in the ICLUS data set) within
a 1-km buffer surrounding the station, for current and decadal time
periods through 2100	4-15
Table 5-1. Average distribution of reference and total stations by state, categorized
by duration of sampling	5-4
Table 6-1. Variables addressed in criteria and pathways through which they may be
affected by climate change (from Hamilton et al. 2009)	6-5
March 28, 2011
vii
External Review Draft

-------
LIST OF FIGURES
Page
Figure 1-1. Climate change can affect many bioassessment program activities from
the initial assessment design, to collecting and analyzing data, and to
developing responses to assessment outcomes	1-4
Figure 2-1. Distributions of cold-water-preference taxa richness values in coldest-, normal-, and
hottest-year samples at 4 Utah sites	2-8
Figure 2-2. Distributions of warm-water-preference richness values in coldest-, normal-, and
hottest-year samples at 4 Utah sites	2-9
Figure 2-3. Distributions of thermal preference metric values in driest-, normal-, and wettest-
year samples at Maine site 56817	2-10
Figure 2-4. Distributions of thermal preference metric values in coldest-, normal-, and hottest-
year samples at Maine site 56817	2-11
Figure 2-5. Trends in the thermal preference metrics and PRISM climatic variables over time at
Maine site 56817	2-11
Figure 2-6. Distributions of thermal preference metric values in Utah reference samples in the
Wasatch and Uinta Mountains and Colorado Plateaus ecoregions	2-12
Figure 2-7. Distributions of thermal preference metric values in Utah reference samples in two
elevation groups	2-12
Figure 2-8. Distributions of thermal preference metric values in Utah reference samples grouped
by Strahler order	2-13
Figure 2-9. Distributions of thermal preference metric values in Maine Class A and AA samples
in the Laurentian Plains and Hills, Northeastern Coastal Zone and Northeastern
Highlands ecoregions	2-14
Figure 2-10. Distributions of thermal preference metric values in Maine Class A and AA
samples in the two elevation groups	2-15
Figure 2-11. Distributions of thermal preference metric values in Maine Class A and AA
samples grouped by Strahler order	2-15
Figure 2-12. Distributions of thermal preference metric values in North Carolina reference
samples in the Coastal, Mountain and Piedmont ecoregions	2-17
Figure 2-13. Distributions of thermal preference metric values in North Carolina reference
samples in two elevation groups	2-18
March 28, 2011	viii	External Review Draft

-------
Figure 2-14. Distributions of thermal preference metric values in North Carolina reference
samples grouped by watershed area	2-21
Figure 2-15. CCA plot of a selected subset of the Utah biological-hydrological data	2-22
Figure 2-16. NMDS plot of macroinvertebrate taxonomic composition and its relationship with
hydrologic parameters for a subset of North Carolina data	2-23
Figure 2-17. NMDS plot (Axis 1-2) for Utah Station 4927250 	2-23
Figure 2-18. NMDS plot (Axis 1-2) for Utah Station 4951200	2-25
Figure 2-19. NMDS plot (Axis 1-2) for Utah Station 4927250 that shows which taxa are most
highly correlated with each axis	2-25
Figure 2-20. NMDS plot (Axis 1-2) for Utah Station 4951200 (Virgin) that shows which taxa
are most highly correlated with each axis	2-26
Figure 2-21. Distributions of total taxa richness values in coldest-, normal-, and hottest-year
samples at Utah sites 4927250 (Weber) (A) and 4951200 (Virgin) (B)	2-26
Figure 2-22. Distributions of EPT richness values in coldest-, normal-, and hottest-year samples
at Utah sites 4927250 (Weber) (A) and 4951200 (Virgin) (B)	2-27
Figure 2-23. Relationship between EPT taxa richness and PRISM mean annual air temperature
(°C) at (A) Utah site 4927250 (Weber) (r=0.57, p=0.01) and (B) site 4951200....2-27
Figure 2-24. Maine site 56817 NMDS plot (Axis 3-2)	2-29
Figure 2-25. Distributions of Class A indicator taxa metric values in driest-, normal-, and
wettest-year samples at Maine site 56817	2-30
Figure 2-26. Distributions of EPT and Dipteran-related metric values in lowest-, normal-, and
highest-flow year samples at Maine site 56817 (Sheepscot). Plot (A) shows relative
Diptera richness, (B) Tanypodinae abundance, (C) EPT generic richness relative to
EPT plus Diptera, and (D) EPT generic richness/Diptera richness. Year groupings
are based on IHA median monthly flows averaged across July-September. Data used
in these analyses were limited to summer (July-September) rock-basket samples.. .2-
35
Figure 2-27. Relationships between North Atlantic Oscillation (NAO) indices and PRISM
climatic variables at Maine site 56817	2-36
Figure 2-28. Cluster analysis using Bray-Curtis (Sorensen) similarity index, based on benthic
invertebrate composition using genus-level OTUs, at Maine station 56817 	2-37
Figure 2-29. Correlation between Euclidean distance calculated as a difference between
successive sampling years, as a measure of similarity between benthic assemblages,
and the NAO annual index	2-39
March 28, 2011	ix	External Review Draft

-------
Figure 2-30. Trends in selected metrics, PRISM climatic variables and chloride concentrations
over time at Utah site 4927250	2-39
Figure 3-1. Distributions of Ephemeroptera abundance metric values across classifications and
trends in this metric over time at Maine site 56817	3-5
Figure 3-2. Distributions of relative Ephemeroptera abundance metric values across
classifications and trends in this metric over time at Maine site 56817	3-5
Figure 3-3. Distributions of EPT generic richness metric values across classifications and trends
in this metric over time at Maine site 56817	3-6
Figure 3-4. Relationship between Maine cold and warm-water-preference taxa and Maine
enrichment tolerance scores	3-8
Figure 3-5. Site-condition classification scores at three reference Mountain sites (Station
NC0109 (New), Station NC0207 (Nantahala) and Station NC0209 (Cataloochee))
and two reference Piedmont sites (Station NC0075 (Little River) and Station
NC0248 (Barnes Creek)) averaged across three 10-year periods. The black bars
represent average scores at Mountain sites when Mountain criteria are applied; the
white bars represent average scores at Piedmont sites when Piedmont criteria are
applied; the gray bars represent average scores at Piedmont sites when Mountain
criteria are applied	3-12
Figure 3-6. Relationship between North Carolina cold and warm-water-preference taxa and
North Carolina enrichment tolerance scores	3-18
Figure 3-7. Plots of macroinvertebrate taxa maximum temperature Weighted
Stressor Values (WSVs) vs mean maximum values for taxa for
headwater streams (upper left) and wadeable streams (lower left), and
box and whisker plots of maximum temperature by Ohio EPA
macroinvertebrate tolerance values (derived for the ICI) for
headwater streams (upper right) and wadeable streams (lower right)	3-24
Figure 3-8. Scatter plots of taxa/species Hydro-QHEI (Qualitative Habitat
Evaluation Index (QHEI) based only on hydrologic variables) WSVs
vs mean Hydro-QHEI values for macroinvertebrate taxa for
headwater streams (upper left) and wadeable streams (lower left),
and box and whiskers plots of macroinvertebrates (upper right)
and fish (lower right) WSVs for Hydro-QHEI for these waters	3-25
Figure 3-9. Distributions of observed/expected (O/E) values in coldest-, normal-, and hottest-
year samples at Utah sites 4936750 (Duchesne) (A) and 4951200 (Virgin) (B) ... 3-28
Figure 3-10. Exploration of how observed, expected and observed/expected values from the Utah
Fall RIVPACS model may change as climate-related predictor variables change 3-32
March 28, 2011
X
External Review Draft

-------
Figure 3-11. Method for tracking changes in cold and warm-water-preference taxa and
commonly used metrics (in this case, total number of taxa at Maine site 56817
(Sheepscot) over time	3-36
Figure 4-1. Conceptual model showing relationship between climate change trends
and reference and stressed sites with an overlay of temporal variation
on the trend (black line). "MDC" = minimally disturbed condition;
"LDC" = least disturbed condition	4-9
Figure 4-2. Reference station drift (degradation of assessed site condition) over time at Blue
Ridge Mountain ecoregion stations as cold-preference taxa are lost over time due to
climate change	4-10
Figure 4-3. Relationship between richness of EPT taxa and flashiness (Baker's index)
of the stream for stream types in the North Carolina Piedmont	4-12
Figure 4-4. Florida's biomonitoring sampling stations, including "exceptional"
reference locations (light green dots), show in relation to current land
use	4-13
Figure 4-5. Relationship between human population density (i.e., degree of urban
development) and Ephemeroptera (mayfly) taxon richness among six
New England states (from Snook et al., 2007)	4-14
Figure 4-6. Distribution of Florida reference stations (N=58, classified as
"exceptional"), plus categories of developed land use from ICLUS	4-16
Figure 4-7. Distribution of Florida reference stations (N=58, classified as
"exceptional"), plus categories of developed land use from ICLUS	4-17
Figure 5-1. Maine biomonitoring stations, coded by data duration and reference condition	5-2
Figure 5-2. North Carolina biomonitoring stations, coded by data duration and reference
condition	5-2
Figure 5-3. Utah biomonitoring stations, coded by data duration and reference condition	5-3
March 28, 2011
xi
External Review Draft

-------
LIST OF ABBREVIATIONS AND ACRONYMS
ANOVA
Analysis of variance
BCG
Biological Condition Gradient
BI
Biotic Index
CCA
Canonical Correlation Analysis
cfs
cubic feet per second
CWA
Clean Water Act
DEP
Department of Environmental Protection
DEQ
Department of Environmental Quality
EMAP
Environmental Monitoring and Assessment Program
ENSO
El Nino/Southern Oscillation
EPT
Ephemeroptera, Plecoptera, Trichoptera
GCRP
Global Change Research Program
HBI
Hilsenhoff Biotic Index
IBI
Index of Biotic Integrity
ICI
Invertebrate Community Index
IHA
Indicators of Hydrologic Alteration
MMI
Multi-metric index
NAO
North Atlantic Oscillation
NCAR
National Center for Atmospheric Research
NCBI
North Carolina Biotic Index
NCDENR
North Carolina Department of the Environment and Natural Resources
NMDS
Non-metric Multidimensional Scaling
O/E
Observed to Expected ratio
OCH
Odonata, Coleoptera, Hemiptera
OTU
Operational Taxonomic Unit
PDO
Pacific Decadal Oscillation
PRISM
Parameter-elevation Regressions on Independent Slopes Model
RBI
Richards-Baker flashiness Index
RIVPACS
River InVertebrate Prediction And Classification System
UAA
Use attainability analyses
USGS
United States Geological Service
WSV
Weighted Stressor Values
March 28, 2011
xii
External Review Draft

-------
PREFACE
This report was prepared by Tetra Tech, Inc. and the Global Change Research Program
(GRCP) in the National Center for Environmental Assessment of the Office of Research and
Development at the U.S. Environmental Protection Agency (U.S. EPA). It is intended for
managers and scientists working on biological indicators, bioassessment, and biocriteria,
particularly in the EPA's Office of Water and Regions, and also at state agencies. The results
presented in this report are based on data primarily from four states, Maine, North Carolina,
Ohio, and Utah. The main findings of interest to manager and policymakers, the supporting
evidence, and management responses are presented in a separate summary at the beginning of
this report. The remainder of the report provides more detail to substantiate each of the findings.
Descriptions of specific analysis methods, underlying data, and supporting analyses are in the
appendices to this report.
March 28, 2011
xiii
External Review Draft

-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
The Global Change Research Program, within the National Center for Environmental
Assessment, Office of Research and Development, is responsible for publishing this report. This
document was prepared by Tetra Tech, Inc. under Contract No.GS-10F-0268K, U.S. EPA Order
No. 1107. Dr. Britta Bierwagen served as the Technical Project Officer. Dr. Bierwagen
provided overall direction and technical assistance, and she contributed as an author.
AUTHORS
Center for Ecological Sciences. Tetra Tech. Inc.. Owings Mills. MP
Anna Hamilton, Jen Stamp, Mike Paul, Jeroen Gerritsen, Lei Zheng, Erik Leppo
U.S. EPA
Britta G. Bierwagen
REVIEWERS
U.S. EPA Reviewers
Wayne Davis (OEI), Lilian Herger (RIO), Rachael Novak (OW/OST), Lester Yuan
(ORD/NCEA)
Other Reviewers
ACKNOWLEDGMENTS
The authors thank the Global Change Research Program staff in NCEA, especially S.
Julius, for their input and advice throughout the development of this project. We also thank staff
in the Office of Science and Technology in the U.S. EPA's Office of Water for their input and
assistance; staff in state offices who contributed data, reviewed approaches, and assisted with the
development of the traits database; and representatives on the regional workgroups for their input
and review during critical phases of the project. Discussions with M. Slimak and A. Grambsch
greatly improved the structure of this report. The comments of EPA reviewers substantially
improved this report.
March 28, 2011
xiv
External Review Draft

-------
EXECUTIVE SUMMARY
Bioassessment is used for resource management to determine the ecological
consequences of environmental stressors. All states utilize some form of bioassessment as part of
their implementation of the Clean Water Act. This assessment identifies the components of state
and tribal bioassessment programs that may be affected by climate change. The study
investigates the potential to identify biological response signals to climate change within existing
bioassessment data sets; analyzes how biological responses can be categorized and interpreted;
and assesses how they may influence decision-making processes. This study focused on benthic
macroinvertebrates, important indicators used in bioassessments of wadeable rivers and streams.
The ultimate goals of the report are to provide a foundation for understanding the potential
climatic vulnerability of bioassessment indicators and advancing the development of specific
strategies to ensure the effectiveness of monitoring and management plans under changing
conditions.
We selected four regionally distributed state bioassessment data sets from Maine, North
Carolina, Ohio, and Utah for this analysis. Bioassessment data were analyzed to determine the
relative sensitivity of benthic community characteristics and traits to historical trends in
temperature, precipitation, and other environmental drivers. The analysis allowed community
characteristics and traits to be classified as either sensitive or insensitive to climate change
effects.
Bioassessment programs rely on reference sites, often the most natural or pristine sites
available, to help provide a basis for comparison with impaired sites. However, climate change
will impact all sites in a region. Consequently, it will be necessary to understand the potential
impacts of climate change for the use of reference sites in bioassessments. We examined the
vulnerability of reference conditions to changes in climate and interactions between climate
change and other landscape-level stressors, especially land use.
This study describes biological responses to changes in temperature, precipitation, and
flow that will, in the long term, affect the metrics and indices used to define ecological status.
Not all regions are equally threatened or responsive because of large-scale variability in climate
and other environmental factors. We found that climatically vulnerable components of
bioassessment programs include:
•	Assessment design (e.g., multi-metric indices (MMIs), selection of reference sites,
determination of reference condition)
•	Implementation (e.g., data collection and analysis)
•	Environmental management (e.g., determination of impairment and water quality
standards)
March 28, 2011
XV
External Review Draft

-------
Ecological traits are useful tools for these analyses since traits are not location specific,
unlike some species. This facilitates comparisons among the state data sets used. This study
mainly focuses on traits related to temperature and hydrologic sensitivities. Effective
bioassessment designs rely on MMIs and predictive models to detect impairment. The
effectiveness of widely-used MMIs and predictive models may be undermined by changing
climatic conditions through the ecological trait of temperature sensitivity. Taxa with cold water-
and warm water-preferences are used in many MMIs and predictive models. The climate
responsiveness of these traits groups varies between states and ecoregions; however, they are
generally found to be highly sensitive to changing temperature conditions. Consequently, MMIs
and predictive models, which rely heavily on these sensitive taxa are likely to be vulnerable to
climate change. In many cases, it may be feasible to develop new MMIs and modify variables in
predictive models to partition sensitive taxa and reduce the potential for changing conditions to
confound efforts to detect impairment.
Another widespread and related finding is the moderate but significant relationship
between temperature sensitivity and sensitivity to organic pollution. These findings show that
metrics selected because the composite taxa are considered to be generally sensitive or because
they respond to conventional pollutants also have demonstrable sensitivities to climate-related
changes in temperature and flow conditions. These results reinforce the need to partition taxa
with climate-sensitive traits from MMIs and to account for these responses in predictive models.
The implementation of bioassessment programs often involves flexible sampling systems,
such as rotating basin designs. These systems ensure statistically adequate sampling over five-
year periods, at the expense of continuous monitoring of specific locations. This type of
probabilistic sampling creates challenges for reference-based comparisons to assess condition,
detect impairment, and identify causes of impairment under changing climatic conditions. While
rotating, probabilistic systems sample numerous reference locations across a state, detection of
climate change requires evaluation of trends at least at a few specific locations over time.
Consequently, states may have many reference locations, but lack enough stable, long-term
stations needed to detect climate-driven changes in biotic condition.
Climate change can cause other problems for reference-based bioassessment systems. We
note that climate change can drive changes in community composition that vary by location,
potentially further compounded by non-climatic landscape stressors. The result is variation in
responses between locations that can confound efforts to establish statistically significant
relationships or detect impairment. For example, our results show that high-flow metrics (e.g.,
flashiness, high pulse-count duration, one-day maximum flow) tend to reflect urbanization,
swamping climate change effects; whereas low flow metrics (e.g., short-duration minimum
flows, low pulse count) respond to climate change effects more so than to land use.
Responses to low-flow parameters were also documented using long-term water
temperature trends at USGS gaging stations. Most of the long-term stations in our study showed
March 28, 2011
xvi
External Review Draft

-------
slightly to distinctly increasing trends in benthic inferred temperatures over time, though not all
trends were significant. Inferred-temperature responses are evidence of climate change-related
increases in temperature, in that they reflect a progressive shift over time in composition of
temperature preferences integrated across the entire benthic community. The response over time
of any one taxon with a particular temperature preference (e.g., a cold-preference taxon) may or
may not be significant despite the expectation, but it is significant if the community as a whole is
reflecting an overall progressive shift in temperature preferences. This response was slightly
greater at higher elevation locations. Results from these analyses corroborate the results from the
landscape analyses that low flow parameters have better performance than high flow/pulse event
parameters in detecting climate change trends.
A synthesis of all results leads to several recommendations for bioassessment programs
in terms of modifying assessment design, implementation, and environmental management. With
respect to metrics and indices, it will be useful to partition climatically vulnerable indicators into
new metrics that account for temperature preferences of the component taxa. Analyzing
bioassessment data according to temperature preferences will facilitate tracking climate change-
related taxa losses and replacements. This traits-based approach for detecting and tracking
climate change effects is promising, given that there were few specific species that showed
consistent climate-related trends across multiple sites and states analyzed.
Although data limitations prevent explicit differentiation between inter-annual, cyclical,
and long-term directional climate effects, the net response of benthic community metrics to
climate-sensitive variables (i.e., water temperature, hydrologic patterns) provides useful
information. The responses can be used to (1) define the direction and nature of effects expected
due to climate change; (2) identify the most sensitive indicators to climate change; and (3)
understand implications to MMIs or predictive models and their application by managers to
characterize condition of stream resources for decision making.
The limited long-term data also illustrate that annual monitoring at least at some fixed
reference locations is needed to account for climate change effects. The ability to detect a real
trend is affected by signal-to-noise ratio and by the amount of data available to account for this
variation. Evidence from this study of the high among-site variability within ecoregions suggests
a trade-off in sampling effort between sampling many stations using a probability-based design
to understand regional variations and sampling selected locations more frequently to document
long-term trends. A mixture of targeted reference sites that can be maintained over the long-term
along with probabilistic sampling may be more appropriate for monitoring the effects of climate
change. This more comprehensive monitoring design will increase the robustness of water
program assessments to the confounding effects of climate change.
Long-term monitoring also requires that these reference locations are as protected as
possible from other stressors and landscape influences. Our analyses show that reference
conditions may be more vulnerable than impaired sites to climate-change effects, a result that
March 28, 2011
XV/'/'
External Review Draft

-------
undermines the current methods of condition assessment. Two approaches that can assist with
condition assessments in the context of climate change are to: 1) implement the Biological
Condition Gradient (BCG) framework, within which changes in condition of both high quality
and impaired locations can be more rigorously defined and tracked; and 2) promote protection of
high quality stream reaches that define reference conditions. Protection should focus on
minimization, mitigation, and/or buffering from non-point source runoff, erosion, and hydrologic
changes.
Documenting existing land use conditions surrounding established reference locations is
also important to establish a baseline for tracking future changes. Urbanization surrounding
reference stations will interfere with the ability to detect climate change and separate climate
responses from conventional stressors; this can interfere with managing aquatic resources, setting
permit limits, and meeting Clean Water Act requirements. Our results show that hydrologic
monitoring, especially using low-flow parameters, can assist with distinguishing changes due to
urbanization versus climate.
Reference sites that remain unprotected from stressors or land use changes are vulnerable
to deterioration due to conventional stressors as well as climate change. The deterioration of
reference conditions and climate impacts on biological indicators, metrics, and indices together
will affect the determination of stream reach impairment. Unless metrics are modified so that
climate effects can be tracked, thresholds for defining impairment are re-evaluated, and actions
to document and protect reference station conditions are taken, it is likely that in vulnerable
watersheds there may be fewer listings of impaired stream reaches and progressive under-
protection of water resources.
Actions that are associated with the listing of a stream reach as impaired, including
stressor identification and development of TMDLs, are also affected by climate changes. Stressor
identification should include biological responses to climate change effects. Climate-related
changes to flow may also need to be integrated into loading calculations and limits for new or
revised TMDLs.
Water quality standards that are resilient to changes in climate-related variables will
remain protective and should be identified. Climate change can be expected to alter some
designated uses and their attainability, especially in vulnerable streams or regions. Refinement of
aquatic life uses can be applied to guard against lowering of water quality protective standards.
The results from the analyses conducted as part of this assessment show that climate
change will affect many of the activities in bioassessment programs. Our results also identify
methods that can assist with detecting these effects and controlling for them analytically.
Implementing these recommendations will allow programs to continue to meet their resource
protection and restoration goals in the context of climate change.
March 28, 2011
XV/'/'/'
External Review Draft

-------
SUMMARY FOR MANAGERS AND POLICYMAKERS
Climate change is an important consideration for bioassessment programs because it can
affect almost all activities associated with these programs (Fig. SMP-1). This report uses data
from four pilot study states, Maine, North Carolina, Ohio, and Utah, to examine the implications
of climate change on these activities. This summary is intended for managers and policymakers
working with bioassessment data and results, who are making decisions about resource
impairment, designated uses, water quality standards, use attainability, and total maximum daily
loads (TMDLs). This summary highlights the ten most important findings of the underlying
report:
•	Multi-metric indices are vulnerable to climate change
•	Predictive models used in bioassessment may be less vulnerable to climate change
•	Detection of climate change effects requires a specifically designed climate change
monitoring network
•	Reference stations are vulnerable from changes in community composition
•	Vulnerability varies by location
•	Reference sites need protection from other stressors
•	Collecting abiotic data is also necessary
•	Reference station degradation diminishes the ability to detect impairment
•	Climate change may make TMDL development more difficult
•	Climate change may alter designated uses and their attainability
Following each finding is a brief description of the evidence to support each finding and
discussion of potential responses that can assist managers and policymakers in adapting
bioassessment programs to climate-change effects. The findings in this summary and the body of
the report are organized according to the steps shown in Fig. SMP-1.
March 28, 2011
xix
External Review Draft

-------
Bioassessmerit Program Activities
©
O)
c
CTJ
JZ
O
CD
CS
O
Assessment Design
-	Select reference sites
-	Select sampling design
-	Determine reference
condition
-	Select communities
-	Select Indicators
-	Create indices,
predictive models
Implementation
- Collect biotic samples - Collect abiotic data
- Analyze data
Environmental Management
-	Develop biocriteria	- Determine impairment
-	Determine water	- Develop TMDLs
quality standards
Figure SMP-1. Climate change can affect many bioassessment program activities from the initial
assessment design, to collecting and analyzing data, and to developing responses to assessment
outcomes.
Findings influencing assessment design
1. Multi-metric indices are vulnerable to climate change
Finding: Climate change affects specific biological metrics used in multi-metric indices (MMIs).
This is important because MMIs are used by many states as a basis for comparing between high
quality and potentially impaired sites and to assign site ratings.
Evidence: In the four states analyzed, though not at all stations or in all regions within each
state, cold water-preference taxa decreased in richness and/or abundance with increasing
temperatures, and in some areas, warm water taxa increased. Some responses are fairly
widespread, including total taxa richness; Ephemeroptera, Plecoptera, and
Ephemeroptera/Plecoptera/ Trichoptera (EPT) richness; and richness of cold or warm sensitive
taxa. Changes in these metrics alter MMIs through shifts in the proportion of cold to warm
water-preference taxa. Further evidence for this finding is presented for each state analyzed.
March 28, 2011
XX
External Review Draft

-------
Maine: Maine's longest-term reference location is at a relatively low elevation and has a
higher proportion of warm-preference taxa, including warm-preference EPT taxa. Therefore, an
increase in EPT taxa with increasing temperatures could improve overall station rating, unlike in
Utah and North Carolina (described below). This is because one metric, Ephemeroptera
abundance, does not have a linear relationship with station class (see Section 3.1). In Maine there
is an additional consideration associated with the use of a group of "Class A indicator taxa" as
one of the ways of separating Class A from B condition ratings. Maine's Class A indicator taxa
are fairly evenly divided between cold and warm-water-preference taxa. Therefore, application
of this metric with increasing temperature could confound results, because some of the Class A
indicators could decline with increasing temperatures, while others could increase.
North Carolina: In North Carolina, EPT taxa richness is one of two indices used for
bioassessment, along with the Hilsenhoff Biotic Index (HBI). Both indices contain cold water-
preference taxa. Though the loss of all cold water-preference EPT taxa due to increasing
temperatures is highly unlikely, this scenario would lead to a reduction of reference station
condition equivalent to one full category (e.g., from excellent to good). The HBI in North
Carolina is vulnerable to the loss of cold-preference taxa and gain in warm water-preference
taxa. This is due to the relationship between temperature preference and sensitivity to organic
pollution (Section 3.2). Since a high proportion of cold water-preference taxa have low pollution
tolerance ratings, the loss of cold water-preference taxa at reference stations due to climate
change also increases the HBI index value for that station, making its assessed condition appear
degraded.
Ohio: The Ohio MMI and the determination of the final station rating are also vulnerable
to climate change because of the positive association between temperature sensitivity and
pollution tolerance. Percent of tolerant taxa is one of the metrics used in the Ohio MMI. There
are also several EPT metrics in the Ohio MMI, including EPT taxa richness, Ephemeroptera and
Trichoptera richness, and relative abundance of Ephemeroptera and Trichoptera taxa. These
metrics contribute to the vulnerability of the Ohio MMI through the relative contribution of cold-
preference taxa within these groups (Section 3.3).
Utah: Fairly predictable losses in EPT taxa richness (especially cold water-preference
taxa) with increasing temperatures are found in both higher and lower elevation ecoregions at
reference stations in Utah. Up to a 25% loss of EPT taxa could occur with current scenarios of
temperature increases by 2050 (Section 3.4). A lower EPT richness metric value will cause an
March 28, 2011
xxi
External Review Draft

-------
overall decrease in the MMI, reducing the condition status of that station. Because the relative
composition of cold water-preference taxa among EPT taxa, as well as in the total community, is
associated with elevation, higher elevation regions with a greater proportion of cold-preference
taxa may have a greater vulnerability to this effect.
Adaptation response: The liability of using existing metrics is due to the inability to separate
temperature responses from other conventional pollution responses. Vulnerable and influential
metrics, such as EPT richness, the HBI, other metrics related to EPT taxa, should be separated
into new metrics that account for temperature preferences (e.g., a 'cold-EPT richness' metric,
etc). This would allow climate-related taxa losses or taxa replacements to be tracked. For
example, a ratio of cold-EPT-richness to warm-EPT-richness would provide a benchmark for
changes related to climate variables compared to conventional stressors. The ability to compare
cold metrics to warm or total metrics between reference and non-reference locations will help
support detection of climate change responses.
Adding a climate-tolerant metric can also assist in separating climate change and
conventional stressor responses. In this study, responses of Odonata, Coleoptera, and Hemiptera
(OCH) taxa were examined based on their reported tolerance to summer conditions, higher
temperatures, and lower flows (Bonada et al. 2007). OCH taxa showed positive trends over time,
with increasing temperatures, and/or with lower precipitation in some locations (Section 3 and
Appendix I); this supports their use as a climate-tolerant metric.
Current limitations to developing climate-sensitive or climate-tolerant metrics are due to
a lack of information on temperature preferences for some taxa and hydrologic data necessary to
determine flow-related preferences. This study used biomonitoring data to develop temperature
preference and tolerance information for many taxa common to Maine, Utah and North Carolina
(see Stamp et al., 2010, USEPA, 2011). However, this type of analysis is needed nationwide. In
addition, much of the extant temperature preference information is at the generic operational
taxonomic unit (OTU) level, rather than the species-level. Hydrologic data was also examined,
but these variables were even more limited and could not be developed into flow-related
preference information for taxa.
2. Predictive models used in bioassessment may be less vulnerable to climate change
March 28, 2011
XX/'/'
External Review Draft

-------
Finding: Some predictive models (e.g., River InVertebrate Prediction and Classification System
(RIVPACS)) used by states may be more resilient to climate change than MMIs because they
incorporate long-term (e.g., 30-year) averages of environmental predictor variables, including
climate parameters, and also because this baseline is recalibrated to a more recent timeframe.
Most RIVPACS models are currently not designed to consider changes in climate, but they could
be. Predictive modeling could be used to associate ranges of biological responses with the
natural range of variation for various climate parameters, and perhaps then differentiate this from
long-term changes. A limitation of modeling is that it assumes freedom of movement, when in
reality, dispersal barriers exist.
Evidence: The greatest vulnerability in applying the RIVPACS model in Utah for decision
making lies in the measured "observed" communities, since they change as a result of shifts in
cold- (and warm-) preference taxa; this drives differences in the observed/expected (O/E)
quantity used in the model. The model predictor variables themselves appear relatively robust to
near-term climate changes in temperature, especially if long-term averaging periods for predictor
variables are used (see Section 3.4). Thus the predictive modeling approach can track changes
without detecting the trend in expected ("reference") communities. However, changes in climate-
related parameters used as predictor variables will alter model precision in assigning the
probability of occurrence of a taxon in a class. Without model recalibration, this could alter the
expectation for inclusion of taxa in a community, and may therefore create larger differences
between observed and expected communities.
Adaptation response: Periodic model recalibration may address this vulnerability. Recalibration
allows shifts in the expected community to be incorporated in O/E calculations. Wider use of the
predictive model approach may be a good adaptation for state biomonitoring programs to climate
change influences on data interpretation. However, as 'expected' assemblages become more
'tolerant,' assemblages may be less likely to show responses to other stressors (i.e. nutrients).
This may reduce differences between expected communities (i.e., the reference baseline) and
observed communities exposed to anthropogenic stressors.
3. Detection of climate change effects requires a specifically designed climate change
monitoring network
Finding: Detection of climate change requires evaluation of changes at some specific locations
or strata over time; despite the relatively large number of reference stations, there are very few
March 28, 2011
XX/'/'/'
External Review Draft

-------
with long-term data, thus limiting the power of current monitoring schemes to detect effects due
to climate change.
Evidence: Inherent biomonitoring program characteristics tend to limit regular, long-term
sampling at reference locations. These include random sampling within a stream reach or
watershed that tends to maximize spatial sources of variation; infrequent sampling (e.g., once
every 5 years) in a rotating basin design; lack of replication (one sample per location per year);
and lack of measurements of covariates (Section 4.1). An additional consideration is the high
vulnerability of existing reference locations to impairment from encroaching land uses (Finding
6). These considerations illustrate the value of designing a monitoring scheme to account for
climate change within the biomonitoring framework.
Adaptation response: Climatic changes, as well as aquatic ecosystem vulnerability to climate
change, vary regionally. Some of the variability is related to elevation, topography, and geology.
Such conditions often cross state and tribal boundaries. Establishment of climate-specific
networks, their monitoring, and subsequent data analysis may require collaboration among states
with regard to technical considerations (e.g., site selection, sampling methods) and funding.
Regional or national support may be important to facilitate this process.
An initial climate-specific monitoring network could focus on climatically vulnerable
locations. Sites should be sampled at least annually. Less frequent data collection would extend
the time needed to detect climate change responses, because of interannual and cyclic climate
variability. Monitoring should occur at some fixed locations, rather than only using a probability-
based sampling approach. It is valid to identify fixed but representative reference locations
within a target stratum (e.g., ecoregion/watershed /vulnerability zone) for detection of trends and
evaluation of biological responses to climate change.
To separate climate change effects from other stressors both should be measured over
time; thus, climate-specific monitoring should be established along part of the stressor gradient
and be anchored in reference conditions (Finding 4). This allows temporal trends at reference
sites to be compared to temporal trends at impaired sites as a mechanism for differentiating
between climate effects and conventional stressors. Sampling the stressor gradient would
potentially allow different levels of stressor effects to be compared and synergistic effects to be
considered. A less resource-intensive alternative would be to establish long-term sentinel sites
only at high-quality reference locations.
March 28, 2011
xx iv
External Review Draft

-------
Many different groups are considering, or have already started, monitoring for climate
change effects. Collaboration across groups avoids duplication, saves resources, and encourages
consistency in data collection and use of a common database. Consistency among monitoring
networks will enhance the ability to detect climate-related changes.
4. Reference stations are vulnerable from changes in community composition
Finding: Climate change increases the vulnerability of reference (high quality) stations through
shifts in biological communities that lead to degraded states.
Evidence: This study documents climate change effects that can degrade reference station status
to be more similar to non-reference stations, at least in some regions more vulnerable to climate
change effects (e.g., high elevation sites, head-water or low order streams) (Section 4.2). In
addition, at non-reference stations, effects of climate change may be additive with other
stressors, or interactions between climate change and other stressors may augment or ameliorate
responses (Figure SMP-2).
Comparison of long-term trends between reference and impaired sites can assist in
separating climate change effects from other stressor effects. This implies the need for long-term
monitoring at more than just high quality sites (see Finding 3). In the absence of climate change-
specific monitoring data, long-term trend analysis to characterize climate change effects should
be conducted on data from reference locations to minimize confounding effects (Sections 4 and
5). This is important, because impacts from land use, nutrient runoff, and other sources are often
not measured and cannot be easily controlled in analyses.
March 28, 2011
XXV
External Review Draft

-------
Ideal reference
sites (MDC)
Real reference
sites (LDC)
O
CO
u
CC attenuates stress
Effect of climate
change
'5b
_o
o
Stressed
sites
CC increases stress
o relative effect
Time
Figure SMP-2. Conceptual model showing relationship between climate change trends and
reference and stressed sites with an overlay of temporal variation on the trend (black line).
"MDC" = minimally disturbed condition; "LDC" = least disturbed condition.
Adaptation response: Reference station condition should be documented using a consistent
framework such as the Biological Condition Gradient (BCG), which captures a more subtle
range of biological conditions with regulatory significance, compared to an "impaired/not
impaired" decision approach. Changing conditions can then be judged against a common
framework. The BCG delineates a meaningful and scaled framework within which the degree of
degradation attributable to climate change can be characterized (Figure SMP-3). A predictive
modeling approach is another framework that can be used to judge a gradient of changes in
condition against a reference baseline in a manner that could support differentiating climate
change effects from other stressor effects. The modeling approach of classifying regions based
on major predictive variables, and using those predictor variables to define expectations for taxa
occurrences within a class (region) uses a wide spatial distribution of reference samples to define
the range of 'natural' variability in each predictor variable. This is essentially doing a space for
time substitution, to the extent that the spatial range of variation can be used to characterize the
expected range of temporal variation. Predictive modeling could be used to associate ranges of
biological responses with the natural range of variation for various climate parameters, and
perhaps then be used to differentiate this from long-term changes.
March 28, 2011
xxvi
External Review Draft

-------
5. Vulnerability varies by location
Finding: Elevation seems to determine the relative vulnerability of community metrics and
MMIs to climate change effects; the magnitude of this response varies regionally. Other
contributing factors appear to be stream order (size) and watershed size.
Evidence: Trends in biota are more distinct in Utah at more locations than Maine or North
Carolina. This is likely related to regional differences in climate change scenarios. For example,
projections for temperature increases are lower for the southeast, including North Carolina, than
they are for the other three states1. Temperatures in the southwest (Utah) are expected to increase
slightly more than for the northeast (Maine) and central states (Ohio), although the differences in
projected temperature among those three areas are very small. Other factors may contribute to
observed regional differences in biological trends and apparent vulnerability, such as differences
in groundwater contribution to flow, stream order, or watershed size. There also are some
artifacts of the available biomonitoring data sets, such as the lack of long-term reference
locations in the northeast highlands ecoregion in Maine, where higher elevations and greater
proportion of cold-preference taxa in the community might have shown stronger trends and
provided a better comparison to the Utah results.
1 See NCAR website: http://rcpm.ucar.edu
March 28, 2011
xxv a
External Review Draft

-------
North Carolina Blue Ridge Mountain ecoregion stations
45.0%
40.0%
35.0%
30.0%
| 25.0%
'•P
$ 20.0%
^ 15.0%
10.0%
5.0%
0.0%
Excellent
¦ current benthic community composition
~lose 50% of cold-preference EPT taxa
alose 100% ofcoId-preference EPT taxa
Good
Good-Fair
Classification


Fair
Poor
Figure SMP-3. Reference station drift (degradation of assessed site condition) over time at
Blue Ridge Mountain ecoregion stations as cold-preference taxa are lost over time due to
climate change.
Adaptation response: If detection of climate change effects becomes a goal, then sampling at
sites more sensitive to these effects becomes important. Elevation may be the first criteria to
identify these sites. The importance of other criteria, such as groundwater contributions and
riparian conditions, still needs to be assessed.
Sites that are more sensitive to climate change can also be used to identify sensitive
indicators and test hypotheses on the relationships among habitat, abiotic variables and species
traits. These indicators, along with cold water-preference indices (Finding 1), can be used in a
monitoring network (Finding 3) to document changes in reference condition (Finding 4).
Findings influencing assessment design and implementation
6. Reference sites need protection from other stressors
Finding: Encroachments of landscape-scale anthropogenic influences, in particular increasing
urban/suburban development, on reference sites over time threaten reference conditions. This
March 28, 2011
xxviii
External Review Draft

-------
threatens both the documentation of climate change-associated biological trends and spatial
comparisons to reference conditions for impairment detection.
Evidence: Our analyses of reference sites (Section 5) show that existing vulnerabilities to land-
use effects are much greater and more widely distributed than previously conceived or
quantified. Urbanization may affect 20-25% of reference stations currently. By 2100
urbanization could cause measurable degradation at almost 50% of reference stations. This level
of land use encroachment would imperil the foundation of the reference condition approach.
Adaptation response: The high vulnerability of reference locations to land-use effects, along
with the importance of identifying and separating climate-change effects, emphasizes the need to
characterize reference conditions and document current status. Two aspects of reference stations
must be considered. One is the selection and siting of reference stations and the other is their
protection. Candidate reference stations should be screened using land-use data. Land-use
distribution by major category should be documented for all stations. Criteria related to the
maximum extent of developed and agricultural land uses should be created and applied to define
reference conditions; however, the criteria may need to be state or region specific and
accommodate existing realities of extent of development. Criteria should recognize that
"unaffected" reference locations may not exist. If current land-use data show low urban and
agricultural uses, it is a reasonable assumption that associated impacts, including urban-
associated hydrologic impacts and agriculture-associated nutrient loadings, are minimal. In
addition, temporal changes in land-use characteristics surrounding a reference site become
important information forjudging degradation of condition that is separate from climate change.
Concepts for protection of reference stations are primarily related to land-use changes
and must involve social, political, and economic components in addition to technical
considerations. In general, an appropriate scale for protecting reference sites is within a
watershed management scheme. Implementing protection actions such as zoning regulations,
incentives for limiting development in riparian areas or near headwater streams, or other
strategies to directly protect high-quality stream reaches from land-use encroachment is
inherently difficult. Identification of the most vulnerable and important sets of reference sites
within a state could be an initial strategy to better target protective actions.
7. Collecting abiotic data is necessary
March 28, 2011
xx ix
External Review Draft

-------
Finding: Abiotic data, such as air and water temperatures, precipitation, flow, and water
chemistry, along with habitat characteristics, need to be collected frequently and over time at
biotic sampling locations.
Evidence: Establishing a relationship between climate change and biotic responses is contingent
upon understanding the relationship between climate variables and hydrologic variables like
water temperature and flow. Long-term, continuous data were not available for all reference
stations (Section 5 and Appendix A). This makes it more difficult, sometimes impossible, to
establish relationships between observed changes in climate and the benthic community.
Adaptation response: Data loggers should be used at all reference stations, especially at sites
within a climate-change monitoring network. Water temperature and flow are the minimum
variables that should be collected; additional data about water chemistry and habitat
characteristics would be useful, although these data may be collected with biological samples.
Findings influencing environmental management
8. Reference station degradation diminishes the ability to detect impairment
Finding: Changes in biological metrics are sufficient to downgrade reference station condition.
Degradation of reference station condition is essentially causing references stations to become
more similar to non-reference stations and diminishes the ability to detect impairment due to
conventional stressors.
Evidence: Climate change does not discriminate between reference and non-reference stations.
This diminishes the effectiveness of reference comparisons to determine impairment. This study
documents changes in biological indicators, which are reasonably attributable to climate change
effects. Sections 2 and 3 document changes in cold- and warm-preference taxa at reference
stations due to climate-change-related trends in temperature and precipitation; these trends result
in changes in MMIs (Finding 1). Unless metrics are modified so that climate effects can be
tracked, and thresholds for defining impairment re-evaluated, degraded reference conditions will
cause fewer stream reaches to be defined as impaired. This will lead to less corrective action and
greater long-term degradation of stream conditions.
Adaptation response: Maintaining the ability to detect impairment will require modifications of
biological metrics (Finding 1), re-evaluation of impairment thresholds, and reference station
classification and protection (Findings 4 and 6). These actions, along with a monitoring network
March 28, 2011	xxx	External Review Draft

-------
(Finding 3), will improve tracking effects of climate change and comparing effects between
reference and non-reference locations to differentiate climate change from other stressors and
detect conventional stressor impairment.
9.	Climate change may make TMDL development more difficult
Finding: Climate change scenarios show greater variability in runoff and flow, which may result
in greater uncertainty in loadings expected from non-point sources. Critical low flows also drive
TMDLs, and these may become more uncertain and more difficult to predict.
Evidence: Changes in biological metrics are sufficient to downgrade reference station condition
(Section 4). This degradation causes reference sites to become more similar to impaired sites,
thereby diminishing the ability to detect impairment. Therefore, unless climate change effects are
tracked using modified metrics, degradation of reference sites will cause fewer stream reaches to
be defined as impaired, at least in the most vulnerable watersheds (Section 6.1).
Adaptation response: In addition to modifying metrics, watershed-specific modeling to predict
how flow dynamics change with climate is needed to provide support for estimating future
changes in low flows, and to modify loading calculations and limitations accordingly.
10.	Climate change may alter designated uses and their attainability
Finding: Climate change can be expected to alter some uses and their attainability, especially in
vulnerable streams or regions. Biological responses to climate change will likely impact water
quality standards and biocriteria through shifts in baseline conditions.
Evidence: Climate change will affect biological communities at reference locations, thereby
altering the characterization of expected levels of ecological integrity. Some cold water streams
could take on cool water characteristics, with declining abundances or richness of sensitive cold
water taxa and possible increases in warm-water taxa. Regulated parameters such as temperature,
dissolved oxygen, and ammonia, may also be sensitive to climate change effects, and their values
may need to be adjusted relative to revised designated uses.
This study illustrates several avenues through which climate change is affecting stream
communities in ways that have implications for biocriteria programs. Section 2 discusses how
trait groups, taxonomic groups, and to some extent, individual taxa appear to be responding over
March 28, 2011
xxx i
External Review Draft

-------
time to climate drivers in ways that are predicable and consistent with expectations relative to
climate change. Section 3 discusses implications of these changes to various MMIs and
predictive models (Findings 1 and 2). The cascading effects of climate change-related trends in
temperature and precipitation on watershed conditions, water quality, and aquatic biological
communities, will lead to shifting, most often degrading, baseline conditions (Finding 4).
Decreases in mean abundances or species richness of cold-preference or other sensitive taxa and
trait groups, increases in warm-preference or other tolerant taxa and groups, and also increases in
variability of these indicators drive reference sites to greater similarity with non-reference areas,
as well as greater difficulty in establishing statistical differentiation (USEPA, 2008). As a result,
reference-based standards will be liable to progressive under-protection (Section 6.2).
Adaptation response: There are numerous criteria, both biological and chemical, that are
addressed in water quality standards and which may be affected by climate change (Table SMP-
1). Biocriteria are of particular interest, as they tie closely to the indices and thresholds used to
determine condition and impairment. The climate-related causes of drifting (degrading) baseline
conditions cannot be directly controlled, but can be assessed. The concepts that support this
include clear documentation of reference conditions, tracking of changes in reference conditions
over time (Finding 4), and protection of reference conditions from other stressors, particularly
land-use changes (Finding 6). This may include monitoring a network of sites designed to detect
climate-change effects (Finding 3).
For watersheds that are particularly vulnerable to climate-change effects, including those
characterized by particularly vulnerable trait groups, more refined aquatic life uses should be
considered. Refinement of aquatic life uses can be applied to guard against lowering of water
quality-protective standards. More refined aquatic uses could create more narrowly defined
categories, which could accommodate potentially "irreversible" changes, but with sufficient
scope to maintain protection and support anti-degradation from regulated causes.
Further efforts to address climate change impacts to standards require examination of
which water quality standards are resilient to climate change effects and therefore remain
protective, and identification of susceptible standards that may need adjustment. Climate change
effects that contribute to degradation of water quality and biological resource condition bring
into question how anti-degradation policies can be managed considering the additional
influences of climate change. High quality water bodies may be most vulnerable to climate
change degradation, making application of anti-degradation policies in vulnerable water bodies
March 28, 2011
xxxii
External Review Draft

-------
important. Management approaches and special considerations for implementation of anti-
degradation policies may need attention. In addition, the application of use attainability analyses
(UAA) on vulnerable water bodies may be pertinent for characterizing climate impacts.
Table SMP-1. Variables addressed in criteria and pathways through which
they may be affected by climate change (from Hamilton et al. 2010)
Criteria
Climate change impacts
Pathogens
Increased heavy precipitation and warming water temperatures may require
the evaluation of potential pathogen viability, growth, and migration.
Sediments
Changing runoff patterns and more intense precipitation events will alter
sediment transport by potentially increasing erosion and runoff.
Temperature
Warming water temperatures from warming air temperatures may directly
threaten the thermal tolerances of temperature-sensitive aquatic life and
result in the emergence of harmful algal blooms (HABs), invasion of exotic
species, and habitat alteration.
Nutrients
Warming temperatures may enhance the deleterious effects of nutrients by
decreasing oxygen levels (hypoxia) through eutrophication, intensified
stratification, and extended growing seasons.
Chemical
Some pollutants (e.g., ammonia) are made more toxic by higher
temperatures.
Biological
Climate changes such as temperature increases may impact species
distribution and population abundance, especially of sensitive and cold-
water species in favor of warm-tolerant species including invasive species.
This could have cascading effects throughout the ecosystem.
Flow
Changing flow patterns from altered precipitation regimes is projected to
increase erosion, sediment and nutrient loads, pathogen transport, and stress
infrastructure. Depending on region it is also projected to change flood
patterns and/or drought and associated habitat disturbance.
Salinity
Sea level rise will inundate natural and manmade systems resulting in
alteration and/or loss of coastal and estuarine wetland, decreased storm
buffering capacity, greater shoreline erosion, and loss of habitat of high
value aquatic resources such as coral reefs and barrier islands. Salt water
intrusion may also affect groundwater.
pH
Ocean pH levels have risen from increased atmospheric CO2, resulting in
deleterious effects on calcium formation of marine organisms and
dependent communities and may also reverse calcification of coral
skeletons.
March 28, 2011
xxxiii
External Review Draft

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
1. CLIMATE CHANGE AND BIOASSESSMENTS
In the US, the Clean Water Act (CWA) of 1972 identified the restoration and
maintenance of physical, chemical and biological integrity as a long-term goal (Barbour et al.,
2000). Biological assessment has been recognized around the world as a valuable and necessary
tool for resource managers to determine biological integrity (Norris and Barbour, 2009).
Research on biological assessment approaches is ongoing, including studies on the development
and evaluation of the most effective bioindicators, metrics, indices, and models (Bressler et al.,
2009; Johnson and Hering, 2009; Stevenson et al., 2008; Stepenuck et al., 2008; Stribling et al.,
2008; Hawkins, 2006; Clarke and Murphy, 2006; Fellows et al., 2006; Hering et al., 2004; Dale
and Beyeler, 2001; Johnson et al., 2006; Sandin and Johnson, 2000); consideration of the best
assemblage groups or system attributes to monitor (Resh, 2008; Bonada et al., 2006; Hering et
al., 2006); regional classification within which to structure monitoring (Archaimbault et al.,
2005; Moog et al., 2004; Hawkins and Norris, 2000); and establishment of reference conditions
(Verdonschot, 2006; Stoddard et al., 2006; Walin et al., 2003; Bailey et al., 1998; Reynoldson et
al., 1997). Overall, incorporation of biological assessment methods to preserve ecological
integrity of waterways has proved to be far more effective than sampling only chemical
parameters (Karr, 2006). Because the structure and function of aquatic assemblages reflect all
sources of environmental disturbance to which they are exposed over time, assessment of
biological communities can provide information that may not be revealed by measurement of
concentrations of chemical pollutants or toxicity tests (Barbour et al., 1999; Rosenberg and Resh,
1993; Resh and Rosenberg, 1984). Biological assessment, coupled with multi-metric or other
predictive modeling analyses, is the strongest approach for diagnosing diminished ecological
integrity, and minimizing or preventing degradation of river systems (Karr and Chu, 2000).
The effectiveness of biological assessment to more reliably detect impairment of
ecological integrity than chemical sampling has been demonstrated in state programs over the
years (Yoder and Rankin, 1998). When biological assessment was incorporated into Ohio's
CWA assessment scheme, almost half of stream segments statewide that met the standards of
their designated uses based on chemical sampling were found to be impaired based on
assessment of biological indicators (Yoder and Rankin, 1998). In contrast, only a few (<3% of)
stream segments that had acceptable biological condition had chemical exceedances (Yoder and
Rankin, 1998).
December 23, 2010
1-1
External Review Draft

-------
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Stream benthic invertebrates are the most common assemblage used for biomonitoring
(USEPA, 2002a), although fish and algal assemblages also are frequently applied in the United
States (Resh, 2008). In this study, benthic communities are the primary focus of analysis using
state biomonitoring data sets. Their integrative characteristics make benthic assemblages
effective monitoring tools if all major sources of stress are accounted for in order to reliably
attribute observed responses to particular sources.
The main goal of this study is to determine what components of bioassessment programs
are threatened by climate change, a stressor that is currently not considered. Related objectives
are to investigate whether biological response signals to climate change are discernable within
existing bioassessment data sets, how responses can be categorized and interpreted, and how
they influence the decision-making process. There is substantial evidence that climate change is
affecting the environment (IPCC 2007), including aquatic ecosystems, and therefore reasons to
account for climate change within the context of bioassessment programs.
A growing number of studies document climate change responses in freshwater
ecosystems. Increased prevalence and/or distribution of warm water (thermophilic) taxa, and
changes in species richness have been found in fish communities (Daufresne and Boet, 2007;
Buisson et al., 2008; Hiddink and Hofstede, 2008). Long-term responses of benthic invertebrate
communities have included changes in stability and persistence, changes in community
composition, increases or decreases in prevalence of taxa groups based on thermophily and
rheophily, species replacements and range shifts, and changes in resilience of community states
(Chessman, 2009; Collier, 2008; Burgmer et al., 2007; Beche and Resh, 2007; Woodward et al.,
2002; Daufresne et al., 2007; Durance and Ormerod, 2007; Mouthon and Daufresne, 2006;
Daufresne et al., 2003). Climate change effects on stream benthos can be seen as long-term,
progressive changes that overlay other natural sources of variability, including other climate
drivers. As an example, patterns of stream benthic community persistence in England were found
to be related to fluctuations in the North Atlantic Oscillation (NAO), as well as to directional
climate change (Bradley and Ormerod, 2001). The magnitude of changes associated with
directional climate change are often subtle compared to other large-scale spatial (e.g., land use)
and temporal (e.g., the NAO) influences (Chessman, 2009; Sandin, 2009; Collier, 2008; Bradley
and Ormerod, 2001).
Climate change effects may be small and long term from certain perspectives, but they
are pervasive. This study documents biological responses to changes in temperature,
precipitation, and flow that will, in the long term, affect the metrics and indices used to define
December 23, 2010
1-2
External Review Draft

-------
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
ecological status. Not all regions are equally threatened or responsive, because of regional
variability in climate combined with spatial variability in vulnerability2 and resilience of the
affected aquatic ecosystems. Many factors can influence susceptibility to changing water
temperature or hydrologic regime from climate change, such as elevation (Chessman, 2009; Diaz
et al., 2008; Cereghino et al., 2003), stream order (Cereghino et al., 2003; Minshall et al., 1985),
degree of ground water influence, or factors that affect water depth and flow rate, such as water
withdrawals (Chessman, 2009; Poff et al., 2006a; Poff 1997).
The components of bioassessment programs that may be affected by climate change
include assessment design, implementation, and environmental management (Figure 1-1).
Awareness that climate change can have widespread effects on biological communities
introduces additional uncertainty into a system that requires interpretable patterns of biological
indicator responses to "conventional" stressors. This has the potential to cast doubt on all claims
of stressor-response relationships that are being evaluated within a regulatory context. It also
highlights that the biomonitoring tools applied must be appropriately tailored to the types of
stressors expected. With increasing knowledge of the types of climate change effects that are
appearing to different degrees in regions around the country, and of the categories of organisms
that are showing the most predictable responses, it becomes important to adjust assessment tools
to changing biota to enable a clearer interpretation of stressor identification and causal analysis.
This study investigates the potential effects of climate change on indicator organisms and
consequences for benthic communities. The results will provide insights on how climate change
may hinder the ability of state and tribal bioassessment and biocriteria programs to meet their
goals of: 1) detecting impairment within reasonable temporal and spatial frames, 2) identifying
probable causes of impairment, and 3) meeting a variety of resource management objectives. The
ultimate goal of this study is to further the development of strategies for adapting monitoring and
management plans to accommodate these expected environmental changes.
2 Vulnerability is generally defined as a combination of exposure (e.g., the expected climate changes in temperature
and precipitation); sensitivity or the degree of responses to the exposures; and resilience or ability of the
communities (or habitats) to adapt and cope with the exposures and responses (see also Poff et al. 2010). We refer to
the vulnerability of the habitat (features of the natural landscape), as well as the vulnerability of the biotic
communities. Vulnerability can be thought about on at different scales, e.g. the biological assemblage as a whole,
individual species, particular sites, stream types, etc.
December 23, 2010
1-3
External Review Draft

-------
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
Bioassessment Program Activities

-------
108
109
110
111
112
113
114
115
116
117
118
Study findings are summarized in the beginning of this report in the Summary for
Managers and Policymakers (SMP). The body of the report expands on the analyses that support
these findings. Section 2 describes analyses using ecological traits and Section 3 applies these
results to indices and predictive models used in state and tribal water programs. Section 4
examines reference station vulnerabilities and discusses design considerations for a monitoring
network to detect climate-change effects. Section 5 describes additional characteristics of
biomonitoring programs that are relevant to discerning climate-change effects. Finally, Section 6
analyzes implications to environmental management, including development of total maximum
daily loads (TMDLs) and water quality standards. Detailed results of all analyses are compiled in
appendices.
December 23, 2010
1-5
External Review Draft

-------
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
2. ANALYSIS OF ECOLOGICAL TRAITS TO DETECT CLIMATE
CHANGE EFFECTS
The underlying objective of conducting detailed analyses of several state bioassessment
data sets is to understand how assessments of biological condition may be affected due to climate
change. Several specific research questions contribute to this objective:
•	Are there biological responses, illustrated in temporal patterns or relationships with
climate variables, already discernable in long-term biomonitoring data?
•	What biological indicators, e.g., trait or taxonomic groups, are sensitive or robust to
climate change effects?
•	Are there spatial patterns or associations that help elucidate climate vulnerabilities that
may be important to bioassessment programs?
•	What modifications to metric analyses might help separate and track climate change
effects?
As more research is conducted and more trait information becomes available, it is likely
that more traits-based metrics will become good candidates for detecting responses to climate
change. In the United States, the value of traits-based versus taxa-based approaches is becoming
more widely recognized (Olden et al., 2008; Beche and Resh, 2007; Poff et al., 2006b). In
Europe, traits-based approaches are currently being used in researching climate-related trends on
aquatic ecosystems (Bonada et al., 2007b). There are many values to traits-based approaches.
Categorization by traits rather than species (or other taxonomic level) reduces variation across
geographic areas, making traits better suited for regional analyses. Traits can be less susceptible
to taxonomic ambiguities or inconsistencies (Moulton et al. 2000) in long-term datasets. Traits
also can be used to detect changes in functional community characteristics (e.g., Bonada et al.
2006, 2007, Beche and Resh 2007) and provide a consistent framework for assessing community
responses to gradients across local and regional scales (Vieira et al. 2006). Finally, use of trait
categories allows for aggregation of data into fewer categories, which can simplify analyses.
2.1. BACKGROUND ANALYSES
In order to begin answering these questions, several foundational analyses were needed to
establish the magnitude and direction of climate change trends in specific locations. The analyses
of long-term air temperature, precipitation, water temperature, and flow records assist in
December 23, 2010
2-1
External Review Draft

-------
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
partitioning bioassessment data into relevant groupings. These analyses also set up an
expectation for the strength of the biological responses that may be discerned in the available
data. For example, long-term air temperature increases are evident from PRISM3 annual air
temperatures; these show gradual, but significant increases from 1974 to 2006 in three of the
states analyzed, Utah, Maine, and North Carolina (Appendix A). Air temperatures differed
between ecoregions in each state, but the rates of increase in air temperature over time were
similar across ecoregions. No significant long-term trends in annual precipitation could be
defined using PRISM data.
Long-term water temperature trends are also evident from USGS gaging station records.
The rate of water temperature increases averaged 0.76 °C per 10-year period, but varied around
the country, partly in relation to stream size. It should be noted that the North Carolina stream
analyzed had a higher water temperature increase than the Utah stream, even though climate
change-related temperature projections are slightly greater for Utah4 (Appendix A), suggesting
that differences in stream size have a greater effect in this case.
Extensive and iterative analyses were conducted using the large bioassessment data sets
from multiple states (see Appendix B for details on data preparation). These data also informed
the selection of long-term stations (Appendix C). Most of the long-term stations or station groups
within ecoregions of all states that were tested showed slightly to distinctly increasing trends in
benthic inferred temperatures over time, though not all the trends were statistically significant
(Appendix A). Inferred temperature responses are evidence of climate change increases in
temperature, with slightly greater responses at higher elevation locations. The benthic inferred
temperature trends in Utah were statistically significant, equivalent to a rate of increase of
approximately 3 °C in 25 years. Using benthic invertebrate occurrence and abundance coupled
with temperature preferences is a reliable means of estimating water temperature at the time of
collection, and conversely, provides evidence of benthic community changes over time related to
long-term changes in temperature.
3	PRISM Climate Group, Oregon State University, http://www.prismclimate.org.
Documentation: http://prism.oregonstate.edu/docs/index.phtml
4	See NCAR website: http://rcpm.ucar.edu
December 23, 2010
2-2
External Review Draft

-------
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
An initial set of analyses was done to assure temporal consistency of the bioassessment data
(Appendix B), as well as to evaluate reference station5 conditions for possible contributions of
other stressors (especially from land use; Appendix J). Table 2-1 summarizes the overall analysis
approach, organized by question being addressed and analysis type. These analyses support the
discussion of results presented in this chapter on the usefulness of ecological traits to track
biological responses to climate change. They are also the foundation for assessing implications
of climate change to multi-metric indices (MMIs) and predictive models used by state and tribal
bioassessment programs, summarized in Section 3. Details about the analysis methods are
presented in Appendix D.
2.2. TRENDS IN ECOLOGICAL TRAIT GROUPS
The reason for evaluating traits is that it comes closer to a mechanistic understanding of
interactions and provides a pathway toward describing the functional implications of climate-
change effects on aquatic communities. The ecological traits of temperature and hydrologic
preferences or sensitivities (e.g., Poff et al. 2006) provide the most direct link to climate impacts.
Other traits such as feeding types, habit, or morphology are also important, but defining
expectations for responses to the effects of climate change is more challenging. For example,
responses of some feeding types to climate change may be indirect through effects on food
resources (phytoplankton, periphyton, allochthonous organic matter) (e.g., Hargrave et al., 2009;
Monters-Hugo et al., 2009; Moline et al., 2004; Tuchman et al., 2002). This study evaluated
many traits and trait suites for relationships to climate change effects, though not all potentially
relevant and fruitful analyses were possible due to limitations of the available biomonitoring
data. In addition to trait groups, analyses also focused on various taxonomic metrics and indices
commonly incorporated in biomonitoring programs.
5 We selected reference sites based on guidance from the respective state agencies. Selection criteria vary across
states. In Utah, reference criteria are based on a combination of a reference scoring sheet (multiple lines of scoring
(i.e., habitat, land use, chemistry) and independent ranking of sites from field crew/scientists. In North Carolina,
land use land cover in the upstream catchment area is an important selection criterion. In Maine, for purposes of our
analyses, we categorized Class A sites (these classifications are based on biology) as reference. Class A sites are not
necessarily designated as reference sites by Maine DEP. Maine DEP is in the process of developing strict reference
criteria; considerations will include factors such as land use land cover and proximity to NPDES discharges.
December 23, 2010
2-3
External Review Draft

-------
202 Table 2-1. Summary of analysis approach, by analysis type, main methods, and overlying questions.
Analysis
Type
Relationship of
(or patterns in):
Relationship to:
Method Highlights
Questions Addressed
Correlations
Biological
indicators,
Taxa;
Taxa groups;
Trait groups;
Indices;
Index components;
Predictive model
parameters.
Time (year)
Climate variables (temperature,
precipitation)
Pearson product moment; calculated
using Statistica software (Version 8.0,
Copyright StatSoft, Inc., 1984-2007);
considered significant if p < 0.05.
Are there biological
responses, illustrated in
temporal patterns or
relationships with climate
variables, already
discernable in long-term
biomonitoring data?
ANOVA
Biological
indicators,
Taxa;
Taxa groups;
Trait groups
(including cold and
warm water
temperature
indicator taxa);
Indices;
Index components;
Predictive model
parameters.
Hot/cold/normal years; and wet/dry/normal
years: defined these using extremes in
climate variables among existing data as
proxies for future climate conditions.
Partitioned data at long-term reference
stations in each state into years characterized
by hotter (>75th percentile of the
temperature distribution during years of
biological collections), colder (<25th
percentile of temperature), and normal (25th
to 75th percentile) average annual air
temperatures. Using similar thresholds, years
were partitioned based on average annual
precipitation into wetter, drier, and normal
years.
One-way ANOVA; calculated using
Statistica software (Version 8.0,
Copyright StatSoft, Inc., 1984-2007);
differences considered significant if: F
statistics p < 0.05, and Tukey honest
significant difference (HSD) test for
unequal sample size (N)
(Spjotvoll/Stoline) p<0.05.
Same as above.
Are certain metrics more
likely to be affected by
climate change than
others?
ANOVA
Trait groups (esp.
cold and warm
water temperature
indicator taxa)
Elevation categories;
Ecoregions. Size (Strahler order or
watershed area).
Same as above
Are there spatial patterns
or associations that
determine climate
vulnerabilities important
to bioassessment
programs?
ANOVA
Maine station
classification
Station classes
Same as above
How do model input
metric values differ among
December 23, 2010
2-4
External Review Draft

-------

descriminant
model metrics


the different station
classifications?
How much do metric
values have to change for
a sample to change
classification?
203
December 23, 2010
2-5
External Review Draft

-------
204	Table 2-1. continued.
Analysis
Type
Relationship of (or
patterns in):
Relationship to:
Methods Highlights
Questions Addressed
NMDS
Taxa composition of
benthic communities
collected each year at
long-term reference
stations, by state.
Hot/cold/normal years;
Several temperature
and precipitation
variables (annual
mean, departure from
mean).
Non-parametric multidimensional scaling;
performed using PCOrd: McCune, B. and M. J.
Mefford. 1999. PC-ORD. Multivariate Analysis
of Ecological Data. Version 4.41 MjM Software,
Gleneden Beach, Oregon, U.S.A.); using a
Sorensen distance measure.
Do changes in community
composition over time reflect
patterns consistent with climate
change effects?
Are taxa associated with
observed changes
sensitive/robust, as expected, to
those climate change effects?
Weighted
average (WA)
modeling
All major taxa, by
state.
temperature (and/or
precipitation, flow
when available)
Weighted average modeling or related
approaches (e.g., maximum likelihood estimates,
general linear modeling) to estimate the optima
and range of temperatures of occurrence for each
taxon from each state that had a sufficient
distribution and number of observations to
support the analysis (Yuan 2006); performed in R
code.
What are the temperature
(and/or precipitation, flow when
available) preferences and
tolerances of taxa collected in
each state?
Benthic
inferred
temperature
modeling
Taxon temperature
preferences, occurrence
and abundances
Time (long-term
trends), temperature
Use WA model results of temperature optima for
each taxon, and taxon occurrence and abundance
by station, to do a weighted-average estimate of
temperature [optimum temperature for each taxon
at a station, times the abundance of that taxon,
summed over all taxa, and divided by the sum of
taxa abundances].
Do benthic communities reflect
water temperatures at the time of
collection?
Do long-term changes in
inferred temperatures provide
evidence of benthic community
changes over time related to
temperature?
Re-running of
Utah
RIVPACS
model
Model input
parameters;
Model outputs.
Climate changes in
temperature,
precipitation, other
climate variables.
Done in R, using Utah DEQ model code, and
revising modify climate-related input parameters.
Are O/E predictive model
predictor variables and 0, E, and
O/E predictions sensitive to
climate change alterations in
temperature, precipitation, and
other pertinent variables?
December 23, 2010
2-6
External Review Draft

-------
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
Evaluation of climate change effects is a fundamentally temporal question. Trend
analysis is used to investigate long-term patterns in temperature, precipitation, flow, other habitat
variables, and in biologic response variables. Results are used as evidence of whether global
changes are contributing to the trends, and for considering other possible contributions. This type
of post-facto analysis of historic data sets is widely used to determine whether climate change
effects are already discernable in ecosystem responses (e.g., Daufresne et al., 2003; Durance and
Ormerod, 2007; Burgmer et al., 2007; Murphy et al., 2007). Long-term stream benthic data from
state biomonitoring programs are used in this study to look for temporal trends in various benthic
community characteristics as evidence of existing biological responses to climate change.
Limitations in the extent of statistically significant trends within the historic biological data are,
in part, related to characteristics of the existing biomonitoring data sets, and should be
understood in the context of the requirements and limitations of typical biomonitoring programs.
This will inform on how biomonitoring and biocriteria programs are likely to be affected in the
future.
Grouping macroinvertebrates based on temperature preferences and tolerances is
expected to (1) have a greater chance of detecting temperature-related climate-change effects if
they exist, (2) be interpretable with regard to causal relationships, (3) offer predictive ability and
transferability to other regions, and (4) serve as a basis for developing adaptive responses
(Verbeck et al., 2008a, 2008b; Poff et al., 2006b; Lamouroux et al., 2004). We developed
temperature indicator metrics by designating cold and warm-water-preference taxa derived from
weighted average or maximum likelihood modeling (methods in Appendix D), case studies,
literature reviews and best professional judgment of regional workgroups6. Hydrologic indicator
metrics were based on literature (e.g., the North Carolina perennial taxa list, Bonada et al.,
2007a) and trait information (i.e., rheophily, drought resistance) related to flow permanence and
current preference. Among the hydrologic indicator metrics were various 'scenario' metrics
(drier-vulnerable, drier-robust, wetter-vulnerable, wetter-robust). These scenario metrics capture
combinations of traits expected to impart an adaptive advantage (or not) under projected climate
change conditions. After developing a list of traits believed to be favorable for each future
climate change scenario, taxa that possessed the most number of those traits states were
6 See Appendix Attachments E2, F2, and Gl, Temperature Indicator descriptions for each state, for more detailed
descriptions of the process followed to develop the temperature indicator taxa lists.
December 23, 2010
2-7
External Review Draft

-------
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
considered potentially 'robust' and those that had the fewest favorable trait states and the most
number of unfavorable trait states were considered potentially 'vulnerable.' In addition, several
scenario metrics were created that take both temperature and hydrologic regime into
consideration (warmer drier vulnerable, warmer drier robust, warmer wetter vulnerable, warmer
wetter robust).
2.2.1. Ecological Trait Groups and Climate Patterns
In Utah, results of ANOVA on ecological trait and scenario metrics varied by site. Two
long-term reference stations, 4927250 (Weber) in the Wasatch Uinta Mountains and 4951200
(Virgin) in the Colorado Plateau, showed relatively strong temperature patterns, while two other
long-term reference sites, 5940440 (Beaver) in the Wasatch Uinta Mountains, and 4936750
(Duchesne) in the Colorado Plateau, showed no patterns (Figure 2-1). At Stations 4927250
(Weber) and 4951200 (Virgin), hottest-year (see Table 2-1 and Appendix D for definition of hot
and cold year definitions and proxy analysis approach) samples had significantly fewer cold-
water-preference taxa than coldest-year samples7 (Table 2-2). The greatest differences generally
occurred between hottest- and coldest-year samples, while normal-year samples were variable.
Warm-water-preference taxa showed even fewer responses, increasing during hottest
years only at Colorado Plateau station 4951200 (Virgin) of the four reference stations tested
(Figure 2-2). Neither cold- nor warm-water-preference taxa responded differently among wettest,
driest and normal years (see Table 2-1 and Appendix D for definition of wet and dry year
definitions and proxy analysis approach, and Appendix F for additional details of Utah analysis
results).
7 In data preparation and analyses, the attempt was to identify and limit potential confounding factors as much as
possible. However, factors other than (or in addition to) climate-related factors (such as changes in water chemistry,
which were documented to show significant yearly trends at some of the sites) also potentially influenced
assemblage composition. It should be emphasized that results for such confounding factors were generally site-
specific, and it is uncertain whether similar patterns are occurring at other sites (see also Section 3.4).
December 23, 2010	2-8	External Review Draft

-------
258
259
260
261
262
263
264
265
266
267
268
Table 2-2. Mean (±1 SD) richness and % individuals with cold- or warm-thermal
preferences in coldest (group 1), normal (group 2), and hottest (group 3) years at
long-term biological monitoring sites in Utah (UT), Maine (ME), and North Carolina
(NC). Year groups were based on Parameter-elevation Regressions on Independent
Slopes Model (PRISM) mean annual air temperature values at each site. One-way
analysis of variance (ANOVA) was done to evaluate differences in mean thermal-
preference metric values. Groups with the same superscripts within a site are not
significantly different (p < 0.05). NA = not applicable (warm-water-preference taxa
absent).
Cold
Warm
Site
Group
Richness
% individuals
Richness
% individuals
UT 4927250
(Weber)
1
2
4.9 ± 1.1A
3.4 ± 1.1A
6.5 ± 5.3a
6.7 ± 7.3a
2.3 ± 0.8a
1.1 ± 0.7a
0.6 ± 0.5a
0.4 ± 0.3a

3
1.0 ± 0.7b
1.0± 1.1A
1.0 ± 1.2a
0.3 ± 0.4a
UT 4951200
(Virgin)
1
2
4.5 ±2.4a
5.3 ± 1.2a
15.7 ± 10.9ab
23.4 ± 15.6a
1.5 ± 0.6a
1.5 ± 0.8a
7.7 ± 6.7a
18.1 ± 15.3a
3
0.8 ± 0.1B
0.2 ± 0.2b
3.8 ± 1.3b
27.8 ± 19.4a
UT 4936750
(Duchesne)
1
2
6.3 ± 1.5a
6.3 ± 1.0A
24.3 ±4.1a
14.9 ± 6.8a
0.3 ± 0.6a
0.7 ± 0.8a
0.03 ±0.1A
0.1 ± 0.2a

3
5.7 ± 2.9a
17.7 ± 8.5a
0.7 ± 1.2a
0.1 ± 0.2a
UT 5940440
(Beaver)
1
2
4.0 ± 2.6a
3.3 ±0.6a
12.1 ±6.2a
10.0 ± 9.2a
NA
NA
NA
NA

3
3.3 ± 1.2a
8.4 ± 5.9a
NA
NA
ME 56817
(Sheepscot)
1
2
0.5 ± 0.5a
0.5 ±0.8a
0.6 ± 0.6a
0.7 ± 1.7a
6.4 ± 2.4a
8.0 ± 1.4a
15.6 ± 7.4a
21.2 ± 1 1.5a

3
1.1 ±0.5a
1.0 ± 0.8a
8.5 ± 2.7a
19.6 ± 10.7a
ME 57011
1
0.7 ± 0.3a
0.4 ± 0.2a
7.2 ± 1.5a
16.1 ±7.3a
(W.Br.
Sheepscot)
2
3
1.5 ±0.5a
1.0 ± 0.7a
6.3 ± 5.4a
1.7 ± 0.3a
7.3 ± 2.8a
7.8 ± 0.8a
48.4 ± 9.6b
39.5 ± 15.4ab
ME 57065
(Duck)
1
2
2.4 ± 1.2a
1.7 ± 0.3a
7.8 ± 6.4a
5.3 ± 5.9a
6.3 ± 0.6a
6.8 ± 1.5a
44.0 ± 22.5a
32.8 ± 10.8a

3
1.6 ± 0.7a
5.0 ± 3.3a
4.8 ± 1.3a
46.6 ± 17.6a
NC 0109
(New)
1
2
4.3	± 1.5a
5.4	± 1.7a
2.3 ± 0.7a
3.6 ± 2.9A
8.3	± 0.6a
7.4	± 1.7a
7.7 ± 2.5a
7.6 ± 2.5a

3
4.0 ± 1.7a
2.2 ± 1.0A
7.3 ± 2.3a
7.0 ± 1.3a
December 23, 2010
2-9
External Review Draft

-------
269
270
271
272
273
274
275
276
277
278
279
280
I
a
5>


~
CoWest
Hottest
Coldest
Normal
Holes!
a Median
0	25%-TM
1	NavOuBtr Range
, s Outers
~ Extremw
Year Groupings
Figure 2-1. Distributions of cold-water-preference taxa richness values in coldest-, normal-,
and hottest-year samples at Utah sites 4927250 (Weber) (A), 4951200 (Virgin) (B), 4936750
(Duchesne) (C), and 5940440 (Beaver) (D). Year groupings are based on PRISM mean
annual air temperatures from each site during time periods for which biological data were
available. Average temperatures in hottest-year samples were 1.1 to 2.7°C higher than
coldest year samples. Mean metric values for cold-water-preference taxa were significantly
higher in coldest-year samples than in hottest-year samples at sites 4927250 and 4951200.
Data used in these analyses were limited to autumn (September-November) kick-method
samples.
December 23, 2010
2-10
External Review Draft

-------
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
Coldest	Normal	Hottest	Coktesl	Normal	Hottest * Extremes
Year Groupings
Figure 2-2. Distributions of warm-water-preference richness values in coldest-, normal-,
and hottest-year samples at Utah sites 4927250 (Weber) (A), 4951200 (Virgin) (B), 4936750
(Duchesne) (C), and 5940440 (Beaver) (D). Year groupings are based on Parameter-
elevation Regressions on Independent Slopes Model (PRISM) mean annual air
temperatures from each site during time periods for which biological data were available.
Average temperatures in hottest-year samples were 1.1 to 2.7°C higher than coldest year
samples. Mean metric values for warm-water-preference taxa were significantly higher in
hottest year samples than in coldest year samples at site 4951200. No warm-water-
preference taxa were present at site 5940440. Data used in these analyses were limited to
autumn (September-November) kick-method samples
In contrast to Utah, in Maine there was greater response to wet/dry years than to
temperature differences. Cold-water-preference taxa, which were present in low numbers at the
sites evaluated, were slightly more abundant and diverse during wet years at the longest-term
reference station (Sheepscot) in the Laurentian Hills and Plains, though warm-water-preference
taxa showed no response to a range of annual precipitation (Figure 2-3); the response of cold-
water-preference taxa was not found at the few other reference stations that could be tested
(Appendix E). Warm-water-preference taxa were generally more abundant and diverse during
hottest and normal years at this Maine location (Sheepscot) (Figure 2-4). There also was a
significant increase over time in richness and abundance of warm-water taxa at Station 56817
(Sheepscot) (Figure 2-5). This appears consistent with climate change expectations, given the
predominance of warm-water-preference taxa at this station, plus increasing temperatures over
December 23, 2010
2-11
External Review Draft

-------
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
time. However, neither abundance nor richness of warm-water-preference taxa was directly
correlated with temperature at this station. In addition, the temporal trend was not spatially
consistent. For example, the warm-water-preference taxa metrics did not increase at another
Laurentian Hills and Plains reference location (site 57011 - W.Br. Sheepscot) (Appendix E).
I
35
Driest
Normal
Wettest
Wettest
Wettest
Year Groupings
Figure 2-3. Distributions of thermal preference metric values in driest-, normal-, and
wettest-year samples at Maine site 56817 (Sheepscot). Plot (A) shows % cold-water-
preference individuals, (B) number of cold-water-preference taxa, (C) % warm-water-
preference individuals, and (D) number of warm-water-preference taxa. Year groupings
are based on PRISM mean annual precipitation from each site during time periods for
which biological data were available. Data used in these analyses were limited to summer
(July-September) rock-basket samples.
December 23, 2010
2-12
External Review Draft

-------
319
320
321
322
323
324
325
326
327
328
. I* -*	Normal	Holes!	Coldesl	Manual	Hottest ~ L-lrrtii'
Year Groupings
Figure 2-4. Distributions of thermal preference metric values in coldest-, normal-, and
hottest-year samples at Maine site 56817 (Sheepscot). Plot (A) shows % cold-water-
preference individuals, (B) number of cold-water-preference taxa, (C) % warm-water-
preference individuals, and (D) number of warm-water-preference taxa. Year groupings
are based on PRISM mean annual air temperatures during time periods for which
biological data were available. Data used in these analyses were limited to summer (July-
September) rock-basket samples.
December 23, 2010
2-13
External Review Draft

-------
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
40
35
30
25
20
15
to
S
0
40
35
30
25
20
15
10
5
0
- #cold water preference
taxa


-# wamt water
preference taxa

40
35 *
30 '
25 •
20 -
15 •
10
5 •
0 *
60
50
40
30
20 •
10
0
• % cold water preference
individuals
% warm water
preference individuals
PRISM mean annual air
temperature (°C)
¦ Adjusted PRISM mean
annual precipitation
(inches)
1985 1988 1991 1994 1997 2000 2003 2006
1985 1968 1991 1994 1997 2000 2003 2006
Figure 2-5. Trends in the thermal preference metrics and PRISM climatic variables over
time at Maine site 56817 (Sheepscot). Plot (A) shows number of cold-water-preference taxa,
(B) % cold-water-preference individuals, (C) number of warm-water-preference taxa, and
(D) % warm-water-preference individuals. Data used in these analyses were limited to
summer (July-September) rock-basket samples. In the plots, PRISM mean annual
precipitation values were adjusted to fit the scale by subtracting 30 from the original
values.
2.2.2. Ecological Trait Groups - Spatial Patterns, Elevation, and Size
We found differences in the distributions of thermal preference taxa between ecoregions,
largely related to elevation differences, in all states tested. In Utah, distributions of the cold-
water-preference taxa were significantly higher in the Wasatch Uinta ecoregion and at higher
elevation sites (Figures 2-6 and 2-7). Sites in the Colorado Plateau ecoregion and at lower
elevations had significantly more warm-water-preference taxa, but numbers of warm-water-
preference taxa were low at the Utah reference sites8. The prevalence and distribution of cold-
8 The relatively low number of taxa on the Utah warm-water-preference list was partially a consequence of the need
to use a family-level OTU for Chironomidae because of inconsistencies in the long-term data set that arose from a
change in taxonomic laboratories.
December 23, 2010
2-14
External Review Draft

-------
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
and warm-water-preference taxa also varied predictably with stream order (Figure 2-8). First and
second order streams in Utah had slightly greater relative abundance and richness of cold-water-
preference taxa, and fewer warm-preference taxa, compared to 3rd or higher order streams. These
results suggest that effects are likely to vary spatially within states, potentially reflecting spatial
differences in vulnerabilities. Biotic assemblages in the Wasatch and Uinta Mountains and at
higher elevations may be more vulnerable to increasing temperatures that are predicted to occur.
On the other hand, many of the higher elevation stations evaluated in Utah were also mid-order
streams, and may not contain the greatest proportion of cold-preference taxa, but may represent
transitional areas to higher elevation headwater reaches that may be vulnerable if they harbor
taxa near thermal thresholds.
1
!
t
Colorado Ffalisus
Wasatch Uinta Mountains
-10
~	I
~	25%-7S%
x Mon-O«tii§f Rang®
- o Outliers
Cefaratki Plateaus
Wasstch m 'Jims Mountain* * EMmum
EPA Level 3 Ecoregion
Figure 2-6. Distributions of thermal preference metric values in Utah reference samples in
the Wasatch and Uinta Mountains and Colorado Plateaus ecoregions. Plot (A) shows
number of cold-water-preference taxa, (B) number of warm-water-preference taxa, (C) %
cold-water-preference individuals, and (D) % warm-water-preference individuals. Data
December 23, 2010
2-15
External Review Draft

-------
361
362
363
364
365
366
367
368
369
370
371
372
used in these analyses were limited to autumn (September-November) kick-method
samples. The sample size (n) of the Wasatch and Uinta Mountains data set was 74 and
n=44 for the Colorado Plateaus.
<2000 ro
» 2000 m
f 4
= 10
I
1 36
¥ ^
T
Elevation
« 2008 m
> 2000 m
a I
O »-TS%
X Hsn-Oaltur Rangs
- o Outliers
* ExVtmtt
Figure 2-7. Distributions of thermal preference metric values in Utah reference samples in
two elevation groups (< 2000 m and > 2000 m). Plot (A) shows number of cold-water-
preference taxa, (B) number of warm-water-preference taxa, (C) % cold-water-preference
individuals, and (D) % warm-water-preference individuals. Data used in these analyses
were limited to autumn (September-November) kick-method samples. The sample size (n)
of the <2000 m data set is 74 and n=55 for the >2000 m data set.
December 23, 2010
2-16
External Review Draft

-------
2^2	Strahter Ord»r
374	Figure 2-8. Distributions of thermal preference metric values in Utah reference samples
375	grouped by Strahler order. Plot (A) shows number of cold-water-preference taxa, (B)
376	number of warm-water-preference taxa, (C) % cold-water-preference individuals, and (D)
377	% warm-water-preference individuals. Data used in these analyses were limited to autumn
378	(September-November) kick-method samples. Sample size for 1st order samples is 11, 2nd
379	order =29, 3rd order = 22, 4th order = 41, > 5th order = 21.
380
381	In Maine, the Northeastern Highlands sites had the highest mean number of cold-water-
382	preference taxa, followed closely by the Northeastern Coastal Zone sites (Figure 2-9). Overall,
383	the number of cold-water taxa in all the Maine ecoregions evaluated was low (1 to 2 taxa). The
384	mean number of warm-water-preference taxa at sites in the Laurentian Plains and Hills was
385	significantly higher than at sites in other ecoregions, while the Northeastern Highlands sites had
386	the lowest mean number of warm-water-preference taxa. These observed ecoregional differences
387	appear to be driven by elevation: there are more cold-water-preference taxa at higher elevation (>
388	150 m) sites and more warm-water-preference taxa at lower elevation (< 150 m) sites (Figure 2-
December 23, 2010
2-17
External Review Draft

-------
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
10). Although it was originally assumed that the high (northern) latitude of Maine also would
influence composition by cold water taxa, apparently elevation is a more influential factor.
is
16
14
| 12
1 10
I 8
I 4
*
2
0

-2

tui

yo

80


ra
70


i.M
it

S

-

¦e
40


f
Jt
r

"3
53®
2.0

n

-10
Laurentian
EPft Level 3 Ecoregion
o Median
O 25%-7S%
T Non-Outliar Range
, o Ouilisis
ME Coasts! Zon*
NEHlsManUs * Extremes
Figure 2-9. Distributions of thermal-preference-metric values in Maine Class A and AA
samples in the Laurentian Plains and Hills, Northeastern Coastal Zone and Northeastern
Highlands ecoregions. Plot (A) shows number of cold-water-preference taxa, (B) number of
warm-water-preference taxa, (C) % cold-water-preference individuals, and (D) % warm-
water-preference individuals. Data used in these analyses were limited to June-October
rock-basket samples, and each replicate was treated as an individual sample. The sample
size (n) of the Laurentian Plains and Hills data set was 747, n=41 for the Northeastern
Coastal Zone and n=433 for the Northeastern Highlands.
As observed in Utah, first and second order streams in Maine had slightly greater relative
abundances and richness of cold-water-preference taxa, while 4th and higher order streams had
more warm-water-preference taxa (Figure 2-11). Third order streams appeared transitional in
temperature preference composition. Based on the distribution of cold-water-preference taxa, it
might be expected that biotic assemblages at Northeastern Highland and other higher elevation
December 23, 2010
2-18
External Review Draft

-------
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
locations, especially in lower order streams, will be more vulnerable to increasing temperatures.
Unfortunately, none of the reference sites located in the Northeastern Highlands have enough
long-term data to support trend analyses. The three reference sites that had enough data to
analyze were located in the Laurentian Plains and Hills and Northeast Coastal Zone ecoregions
and were dominated by warmer-water-preference taxa.
12 ,
60
f IP
%
5
? 6
£
> *
-
2
0
•2
100
90 I
^fg I
1 I
Ck	I
J 20
8 « 1
< 150 m
> 150 m
*	39 I
E I
I »
*	10
0
-10 '
Elevation
< 150 m
0	Median
02S*-7SX
1	fton-Outlier Range
- o Oulliers
* Estlrtrnes
Figure 2-10. Distributions of thermal preference metric values in Maine Class A and AA
samples in the two elevation groups (< 150 m and > 150 m). Plot (A) shows number of cold-
water-preference taxa, (B) number of warm-water-preference taxa, (C) % cold-water-
preference individuals, and (D) % warm-water-preference individuals. Data used in these
analyses were limited to June-October rock-basket samples, and each replicate was treated
as an individual sample. The sample size (n) of the <150 m data set is 817 and n=404 for the
>150 m data set.
December 23, 2010
2-19
External Review Draft

-------
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
12
10
| 8
I 8
4
I ,
1S
14
| 12
1 10
L
g
i
i
q&
I
e£»
90 f£
sa |
70 J
60 l
E
50 |
at i
30 |
20 I

~

~ 25*-T5%
"J Sem-Otiltier Rsngt
- o OHIllfIS
* Extremes
Strahlei Ortfai
Figure 2-11. Distributions of thermal preference metric values in Maine Class A and AA
samples grouped by Strahler order. Plot (A) shows number of cold-water-preference taxa,
(B) number of warm-water-preference taxa, (C) % cold-water-preference individuals, and
(D) % warm-water-preference individuals. Data used in these analyses were limited to
June-October rock-basket samples, and each replicate was treated as an individual sample.
Sample size of lsl order samples is 230, 2nd order =149, 3ra order = 273, 4m order = 284, 5
rd
a th
-til
order = 95, 6th order = 32.
In North Carolina, ecoregions also vary in the predominance of cold- and warm-water-
preference taxa. The richness of cold-water-preference taxa is, on average, higher in the
Mountain ecoregion than in the other two ecoregions (Figure 2-12). The distribution of warm-
water-preference taxa is significantly different between all three ecoregions, with the highest
abundance occurring in the Coastal ecoregion and the lowest number occurring in the Mountain
ecoregion. This distributional pattern is reinforced by the finding that significantly more cold-
water taxa occur at higher elevation sites than at lower elevations (Figure 2-13). Conversely,
median richness and abundance of warm-water taxa is greater at lower elevation sites.
Distribution of cold-and warm-water-preference taxa was also related to watershed size. The
December 23, 2010
2-20
External Review Draft

-------
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
smaller watersheds in North Carolina (<35 mi2) had a greater proportion of cold-water-
preference taxa (based on both abundance and richness), while larger watersheds (>100 mi2) had
a greater proportion of warm-water-preference taxa (Figure 2-14). Based on the results from the
cold- and warm-water taxa distribution analysis, it is likely that biotic assemblages at Mountain
and higher elevation sites, and in smaller watersheds, will be more vulnerable to increasing
temperatures than others because greater numbers of cold-water taxa inhabit these sites.
However, in North Carolina, few trends over time were found for cold- or warm-water-
preference taxa. This may be attributable to the more limited time series of data available from
North Carolina, as well as to the use of categorical rather than abundance data (though this
would not affect evaluation of richness trends).
North Carolina Eeorogion
Figure 2-12. Distributions of thermal preference metric values in North Carolina reference
samples in the Coastal, Mountain and Piedmont ecoregions. Plot (A) shows number of cold-
water-preference taxa, (B) number of warm-water-preference taxa, (C) % cold-water-
preference individuals, and (D) % warm-water-preference individuals. Data used in these
analyses were limited to June-September standard qualitative samples. The sample size (n)
of the Coastal data set is 20, n=61 for the Mountain ecoregion and n=21 for the Piedmont
data set.
December 23, 2010
2-21
External Review Draft

-------
459
460
461
462
463
464
465
466
> -ses m
< too m
< SOO m
> SOC m
SJI
02S%-?5*
T Hea-Ootiitr Rjnj«
- o OuMttrs
Je Exlrtmii
Figure 2-13. Distributions of thermal preference metric values in North Carolina reference
samples in two elevation groups (< 500 m and > 500 m). Plot (A) shows number of cold-
water-preference taxa, (B) number of warm-water-preference taxa, (C) % cold-water-
preference individuals, and (D) % warm-water-preference individuals. Data used in these
analyses were limited to June-September standard qualitative samples. The sample size (n)
of the <500 m data set is 49 and n=50 for the >500 m data set.
467
December 23, 2010
2-22
External Review Draft

-------
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
18
I
o Median
q 25%-75%
X Non-Oullw Range
o Outliers
w Extreme
t -rt-.i {squ»r*
Figure 2-14. Distributions of thermal preference metric values in North Carolina reference samples
grouped by watershed area. Plot (A) shows number of cold-water-preference taxa, (B) number of
warm-water-preference taxa, (C) % cold-water-preference individuals, and (D) % warm-water-
preference individuals. Data used in these analyses were limited to June-September standard
qualitative samples. The sample size for the < 35 sq mi group is 17, 35 to 100 sq mi =15 and > 35 sq
mi = 24.
2.2.3. Potential biological indicators of climate-related hydrologic changes
This section focuses on analyses of paired hydrologic-biological datasets in Utah, Maine
and North Carolina. In the Utah analyses, weighted averaging was used to calculate taxa optima
and tolerance values for selected Indicators of Hydrologic Alteration (IHA) (Richter et al., 1996)
parameters derived from USGS flow data (Appendix K) and year. Data from 43 biological
sampling sites (=159 fall samples) and their associated USGS gages were used in the analyses.
Results showed that several low flow parameters performed well compared to high flow/pulse
event parameters. Indicator values for the IHA 3-day minima flow values show potential for
detecting climate change effects. Results for taxa that had more than 20 occurrences in the
dataset (which is statistically an adequate sample size) show that Leuctridae (rolled wing
r
1
1 20
1
i S
0
"JL
rii
X
3T
(tss than 3S
35 to 11#
«
¦21		
.r> "i 35
35 l.o 1##	wetter tl»n 1#
December 23, 2010
2-23
External Review Draft

-------
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
stoneflies), Asellidae (isopod crustaceans referred to as sowbugs) and Zapada (a stonefly in the
family Nemouridae) had the lowest 3-day minima optimum values (0.056 cfs or less), while
Hyalella (an amphipod crustacean) and Helicopsyche had the highest (0.11 cfs). Leuctridae and
Zapada had relatively low tolerance ranges, while Hyalella and Helicopysche (a caddisfly) had
large tolerance ranges. These results suggest that the stoneflies Leuctridae and Zapada (in the
family Nemouridae) are better adapted, perhaps partly due to their smaller sizes, to lower flow
conditions than other organisms that appeared in the Utah data set.
When taxonomic trends were examined using NMDS and CCA, both analyses indicated
that year had the strongest influence on taxonomic composition. The IHA parameters that were
used in the analysis had a weaker effect, although the high pulse and 3-day minima parameters
also explained a fair amount of variation. Results of the CCA are shown in Figure 2-15. The plot
shows which taxa were most closely associated with year, high pulse and 3-day minima values.
Correlation analyses were also performed on the seven sites in Utah that had the most number of
years of biological and hydrological data, which were analyzed individually. Results showed that
there were no taxa or metrics that were consistently associated with the IHA parameters across
all sites (Leuctridae was not present at any of the sites, and at the one site where Zapada was
present, it was not significantly correlated with the IHA parameters). The same was true for the
flashiness index. Therefore, no taxa or metrics emerged as good candidates for 'disturbance
indicator' metrics related to hydrology in Utah. Disturbance is of particular interest because
hydrologic regimes are expected to change (i.e. duration and/or frequency of high and/or low
flow events) as a result of climate change, so taxa that are better able to adapt to the changing
conditions are generally believed to have a greater chance for success.
NMDS ordinations were also performed on the North Carolina data (440 samples,
selected based on matches between biological sampling locations and USGS gaging stations, for
all available sampling dates). Results from one of the analyses are shown in Figure 2-16. In the
plot, samples are grouped by level 3 ecoregion. NMDS results show that baseflow index, which
is a parameter that represents low flow influence, had the strongest correlation with
macroinvertebrate species composition, though this strong relationship may be mostly due to
ecoregional distribution of taxa. A number of covariates, such as elevation, temperature, and
other factors may co-affect the observed pattern. The second factor that related to taxonomic
composition was number of reversals, which is a measurement of flashiness. The flashiness
December 23, 2010
2-24
External Review Draft

-------
517	index (RBI) had weaker correlation with the species axes. Other factors that showed correlations
518	were low pulse and high pulse parameters (Table 2-3).
519
CCA-Utah
SPECIES
A
ENV VARIABLES
La&vapex
Ciriygma ^ ^
N em our a
Hesperap
hydrid
Planar*!
Amphinem . . , Dyttiad
I aemone
Dolopkil
Tri
IfcaMIa
Gammams
A
^Jfaf&dae ^Heiicaps
A O
4p
Ca#timkMrr,pte A
Umnopho Caems AsdUda ,
^ Acraneur Nt
Salis
Pjophia
epteim
Pis Mum Ferrissi
Calopary fyptantict A Aexlwia
CuUcoid AValvJ£yc?™
Qchthebi
iztjsmoe
e&chus
520
521
522
523
524
Figure 2-15. CCA plot of a selected subset of the Utah biological-hydrological data.
December 23, 2010
2-25
External Review Draft

-------
525
526
527
528
Baseflow Index
CM
V)
No. of Reversal
T
8*
A A Jk ^#1,
A Aa
A
A
j'
A A M	_ ~
Level III Ecoregion
a Piedmont
Coastal Plains
~	SE Plains
~	Blue Ridge
Axis 1
Figure 2-16. NMDS plot of macroinvertebrate taxonomic composition and its relationship
with hydrologic parameters for a subset of North Carolina data. Baseflow index and
number of reversals were associated with Axis 2.
529
December 23, 2010
2-26
External Review Draft

-------
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
Table 2-3. Selected results from the Pearson and Kendall Correlations with
Ordination Axes (N= 440). Only those variables that had strong correlations are
included in this table. Correlations with r or tau values greater than 0.3 are in bold
print. Note: significance values are not available for these analyses because the
correlations associated with the ordinations are not meant for testing hypotheses.
Axis:	1	2	3
r	r-sq	Uni	r	r-sq	Uni	r	r-sq	Urn
January




0.364
0.312
February




0.361
0.33
March




0.368
0.327
April




0.322

XI dry min
-0.373





X3 day min
-0.375





X7davmin
-0.327





XI da y ma x




0.363
0.338
X3daymax




0.375
0.368
X7davmax




0.4
0.375
X3()davmax




0.395
0.37
X90daymax




0.387
0.359
Base flow
-0.51
-0.382
0.416
-0.317
0.365

LopulscL
0.364





H ipulsc
-0.328





HipulscL
0.442





Rise rale





0.303
Fa Urate





-0.332
Reversals
-0.512
-0.358




Highlpcak




0.443
0.358
Highldur
0.362





Highlfreq
-0.302





H igli 1 rise




0.383
0.323
High Hall




-0.338
-0.359
RBI


-0.391



Similar analyses were attempted in Maine, but there were not enough USGS gages
associated with biological sampling sites to estimate flow variable optima using weighted
averaging. Instead, correlation analyses were performed on data from Station 56817 (Sheepscot)
(22 samples). Some taxa and some metrics were significantly correlated with some IHA
parameters. For example, Hydropsyche (spotted sedge caddisfly), Promoresia (an elmid beetle)
and Rhyacophila (green sedge caddisfly) were significantly and positively correlated with 1- and
3-day minima flow values. However, it is difficult to draw conclusions about consistency of
patterns statewide based on data from one site, and also because unambiguous causal
December 23, 2010
2-27
External Review Draft

-------
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
relationships between the biological and hydrologic data cannot be established based on
correlation analyses.
2.3. TEMPORAL TRENDS IN TAX A AND COMMONLY USED METRICS
Although different states use different methods for assessing sites, there are certain
metrics, such as those related to EPT taxa, which are commonly used in state multi-metric
indices (MMIs). To provide regional relevance and a common basis for regional comparisons,
analyses were performed that examined associations between these commonly-used metrics and
climate-related variables. ANOVAs were used to determine differences in biological metrics
among hottest, coldest, wettest, driest, and normal years (Table 2-1) for reference sites that had
adequate long-term data. Correlation analyses were used to evaluate associations between
biological variables, annual air temperature and precipitation, variables related to inter-annual
climate variability, and these variables with lagged year effects.
NMDS were used to show in ordination space how samples collected over years at long-
term stations varied in species composition over time. Reference locations with sufficient long-
term data to perform ordinations included two Utah reference sites (Station 4927250 - Weber and
Station 4951200 - Virgin) and one Maine reference site (Station 56817 - Sheepscot). The
ordinations were used to evaluate differences in taxonomic composition among samples
collected during hottest, coldest, wettest, driest and normal years. Other environmental variables
were used to group the data while looking for trends, including temperature and precipitation
categories, PRISM9 mean annual air temperature and precipitation from the year the sample was
collected, PRISM mean annual air temperature and precipitation from the year prior to sample
collection (to look for lagged effects), and absolute difference between the PRISM mean annual
air temperature value and PRISM mean annual precipitation value from the year of the sample
collection and the year prior (to look for effects of climate variability). In addition, several IHA
parameters were included in the ordination of the Maine data: average of median monthly flows
from sample collection months (July-September), Richards-Baker Flashiness Index (which uses
flow data to quantify the frequency and rapidity of short-term changes in stream flow) (Baker et
al., 2004), and 1- and 3-day minimum and maximum flows. Presence/absence data from fall
9 PRISM data downloaded from http://www.prism.oregonstate.edu/.
December 23, 2010
2-28
External Review Draft

-------
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
samples in Utah and summer/fall samples in Maine were used to analyze changes in the
biological assemblage.
2.3.1. Trends and Patterns- Utah and Western States
In Utah, the NMDS ordinations tend to corroborate ANOVA findings regarding benthic
responses among years partitioned by climate parameters. At the two long-term Utah reference
stations tested, 'hottest year' samples formed distinct clusters from the 'coldest' and 'normal'
year samples (Figures 2-17 and 2-18).
Utah StationID 4927250
^1986
1987
A	1985
Cat_Temp
ii
ppt14_ab
,tmean14
\PrevYr_t
2001
2000
	~	A
Axis 1
Figure 2-17. NMDS plot (Axis 1-2) for Utah Station 4927250 (Weber). Cat Temp refers to
the temperature categories, which are: l=coldest years; 2=normal years; 3=hottest years.
Samples are labeled by collection year. tmeanl4=PRISM mean annual air temperature
from the year the sample was collected, PrevYr_t= PRISM mean annual air temperature
from the year prior to sample collection, and pptl4_ab= absolute difference between the
PRISM mean annual precipitation value from the year of the sample collection and the
year prior.
December 23, 2010
2-29
External Review Draft

-------
Utah StationID 4951200
	PrevYrJ
tmean14
592	Axis 1
593	Figure 2-18. NMDS plot (Axis 1-2) for Utah Station 4951200. Cat Temp refers to the
594	temperature categories, which are: l=cold years; 2=normal years; 3=hot years. Samples
595	are labeled by collection year. tmeanl4=PRISM mean annual air temperature from the
596	year the sample was collected, PrevYr_t= PRISM mean annual air temperature from the
597	year prior to sample collection, pptl4_ab= absolute difference between the PRISM mean
598	annual precipitation value from the year of the sample collection and the year prior, and
599	PrevYr_p= PRISM mean annual precipitation from the year prior to sample collection.
600
601	Figure 2-19 shows which taxa are the strongest drivers along Axes 1-2 at Station
602	4927250 (Weber). Pteronarcys, Chloroperlidae and Ephemerella have the strongest positive
603	correlations with Axis 2, and Optioservus, Lepidostoma and Hyallela have the strongest negative
604	correlations with Axis 2. The three taxa positively associated with Axis 2 tend toward cold-
605	water-preference - Chloroperlidae and Pteronarcys are absent from the 'hottest year' samples
606	and Ephemerella is present in all the 'coldest year' and 'normal year' samples and is only present
607	in one 'hottest year' sample. Some additional taxa that occurred during multiple years and were
608	not found in 'hottest year' samples include Rhithrogena, Nematoda, and Tubificidae. Warm-
609	water-preference taxa that are present in at least 4 of the 5 'hottest year' samples include
610	Optioservus, Lepidostoma and Hyallela, though they also are present in 'coldest year' and/or
611	'normal year'samples.
612
December 23, 2010
2-30
External Review Draft

-------
613
614
615
616
617
618
619
620
621
622
623
624
625
626
UtahStationID 4927250
A
Pteronar	Cttoro
Ephem
	A	A
Axis 1
Figure 2-19. NMDS plot (Axis 1-2) for Utah Station 4927250 (Weber) that shows which
taxa are most highly correlated with each axis.
Figure 2-20 shows which taxa are the strongest drivers along Axes 1-2 at Station
4951200 (Virgin). Ephemerella, Nematoda and Heptagenia have the strongest negative
correlations with Axis 1, and appear to tend toward a cold-water-preference. Nematoda is absent
from the 'hottest year' samples, and Ephemerella and Heptagenia are present in all 'coldest year'
samples, 6 of the 7 'normal year' samples and only 1 of the 'hottest year' samples.
ForcipomyialProbezzia, Microcylloepus, Caloparyphus and Chimarra have the strongest
positive correlations with Axis 1, and appear to be warn tolerant. These taxa are present in at
least 2 of the 4 'hottest year' samples and are absent from the 'coldest year' and/or 'normal year'
samples.
December 23, 2010
2-31
External Review Draft

-------
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
Utah StationID 4951200
CM

<
Zaitzevi
Neotrich
MayaNeo
Tubmci
v-horo
Leuco
AhsoLeu
Corydalu
Rhith
Brachy
etrophi
Planorb
Calopary
Nematoda
Taenio
Polycent
Chimarra
Heptagen
Apatania
Asellida
Clinocer
Athenx
Pteronar
Arcto
Micrasem
Tanyderi
Cinygmul
_ ForcProb
¦ Micro
Figure 2-20. NMDS plot (Axis 1-2) for Utah Station 4951200 (Virgin) that shows which
taxa are most highly correlated with each axis.
Only five of the metrics tested were significantly different between hottest, coldest,
wettest, driest, and normal year samples at more than one site. At Stations 4927250 (Weber) and
4951200 (Virgin) in Utah, the hottest year samples had significantly fewer total taxa, EPT taxa
and cold-water-preference taxa than the coldest year samples (Figures 2-21 and 2-22, and Figure
2-1). Figure 2-23 illustrates the relationship between EPT richness and PRISM mean annual air
temperature at Utah sites 4927250 (Weber) and 4951200 (Virgin). If a linear regression is used
to infer the relationship at site 4927250 (which is located in the Wasatch and Uinta Mountains),
approximately 3 EPT taxa are lost for every 1°C increase in (air) temperature. If the same is done
for site 4951200, which is a lower elevation site located in the Colorado Plateaus ecoregion, the
inferred loss rate is -1.5 EPT taxa for every 1°C increase in (air) temperature. If one were to take
this a step further, the median number of EPT taxa at site 4927250 (Weber) is -13 to 14 taxa.
Based on a projected temperature increase of 2°C over the next 40 y (i.e., by 205010), an average
of 6 taxa could be lost (>40% of total EPT richness). There is no particular reason to assume that
the actual rate of taxa losses will be linear over time, especially considering year-to-year and
10 Based on data from the National Center for Atmospheric Research website: http://rcpm.ucar.edu.
December 23, 2010	2-32	External Review Draft

-------
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
decadal-scale climate variations (see Section 2). However, if these types of trends were to occur,
they would likely affect MMIs.
~	Median
~	25%-75%
X Non-Outlier Range
o Outliers
Extremes
Year Groupings
Figure 2-21. Distributions of total taxa richness values in coldest-, normal-, and hottest-
year samples at Utah sites 4927250 (Weber) (A) and 4951200 (Virgin) (B). Year groupings
are based on PRISM mean annual air temperatures from each site during time periods for
which biological data were available. Average temperatures in hottest-year samples were
1.1 to 2.7°C higher than coldest year samples. At both sites, mean total taxa metric values
were significantly higher in coldest year samples than in hottest year samples. Data used in
these analyses were limited to autumn (September-November) kick-method samples.
December 23, 2010
2-33
External Review Draft

-------
661
662
663
664
665
666
667
668
669
670
671
22
20
18
16
14
12
nj 10
x
03
fc 8
£	6
0

-------
672
673
674
675
676
677
678
679
680
681
682
24
22
20
18
16
14
12
10
o
i2 = 0.3225
8
6
4
4.5
5.0
5.5
6.0
6.5
7.0
7.5
8.0
8.5
20
18
16
14
12
10
8
6
4
2
10.0
11
PRISM mean annual air temperature (°C)
13.0
14.0
10.5
12.0
12.5
13.5
Figure 2-23. Relationship between EPT taxa richness and PRISM mean annual air
temperature (°C) at (A) Utah site 4927250 (Weber) (r=0.57, p=0.01) and (B) site 4951200
(Virgin) (r=0.79, p<0.01). The dotted lines represent 95th confidence intervals.
2.3.2. Trends and Patterns- Maine and New England States
There were few consistent patterns in Maine that clearly relate trait or taxonomic metrics
to climate condition variables. Unlike Utah, the MMDS ordination at Maine's longest term
Station 56817 (Sheepscot) (in the Laurentian Hills and Plains) showed no distinct clusters
reflecting hottest, coldest, wettest, driest, and/or normal year groups (Figure 2-24).
December 23, 2010
2-35
External Review Draft

-------
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
Maine StationID 56817
A 1989

Cat_Temp


~ 1
1990

~ 2


3
1 937

1992 A

A

1991 ^2DD4

1986

^ 1998 2006

1993 1997

1994 A

~

1996

A. 2003

2001 J^.AbsD_P

~ 1985

~

1E,a9j 1988 2002

1995 A 2000

~

2005

~

Axis 2
Figure 2-24. NMDS plot (Axis 3-2). Cat Temp refers to the temperature categories, which
are: l=coldest years; 2=normal years; 3=hottest years. Samples are labeled by collection
year. Absolute difference between the PRISM mean annual precipitation from the
sampling year and the previous year (AbsD P) is the most strongly correlated
environmental variable with Axes 2 and 3.
Species composition did not change in a consistent way among hottest, coldest, wettest,
driest, and/or normal year groups. Several of the taxa that were drivers of interannual patterns in
community composition (Ablabesmyia (a midge), Tricorythodes (a mayfly), and Pseudocloeon
(blue-winged olive mayflies) occurred most often during "normal" precipitation years,
suggesting potential importance of hydrologic conditions in affecting community patterns in
Maine. Overall, taxa could not be consistently categorized as having singular temperature or
precipitation preferences. The lack of strong association between ecological trait groups and
community patterns of responses, in combination with the lack of regional consistency in
December 23, 2010
2-36
External Review Draft

-------
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
ecological trait group trends, makes it difficult to recommend a particular group of 'climate
change indicators' as being regionally important.
At Station 56817 (Sheepscot), precipitation and flow appear to have a greater influence
on the biotic assemblage than temperature. Five of the Maine bioclassification model input
metrics were significantly correlated with flow category (Appendix E). The mean richness and
abundance of cold-water-preference taxa were slightly higher during the wettest years (Figure 2-
3). On average, more Class A indicator taxa were present during wetter years (Figure 2-25), as
were EPT taxa relative to Diptera taxa (Figure 2-26). These patterns are consistent with
expectation if the wettest years provide a more thermally stable and hospitable environment for
the more sensitive cold-water-preference, Class A, and EPT taxa. The relative abundance of
collector-gatherers was higher during higher flow years (Appendix E), though this may reflect a
relationship between higher flows and the distribution of more fine and course particulate
organic matter for food. In contrast, Diptera richness and Tanypodinae abundance decreased
during higher flow years (Figure 2-26). These taxonomic groups include many environmentally
tolerant taxa, which may do well during more stressful low flow years (higher relative abundance
and richness), and decrease during wet years relative to the increase in sensitive taxa.
0.8 r
0.7 ¦
OJ
X
05
I-
0.6
o 0.5
o

-------
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
Figure 2-25. Distributions of Class A indicator taxa metric values in driest-, normal-, and
wettest-year samples at Maine site 56817 (Sheepscot). Year groupings are based on PRISM
mean annual precipitation from each site during time periods for which biological data
were available. Data used in these analyses were limited to summer (July-September) rock-
basket samples.
While Station 56817 in the Laurentian Hills and Plains showed faunal responses to
precipitation, responses related to temperature were more evident in ANOVA analyses
conducted on data from Station 57011, (W.Br. Sheepscot), also in the Laurentian Plains and Hills
(Appendix E Table E3-2). This reflects the importance of site-specific variability in Maine. At
Station 57011, a range of metrics, including percent abundance of collector-filterers, Hilsenhoff
Biotic Index (HBI) scores, Shannon-Wiener diversity index scores, and percent abundance of
Odonata (dragon/damsel flies), Coleoptera (beetles), and Hemiptera (true bugs) (OCH) taxa,
showed significant differences between samples grouped by hottest, coldest and normal years.
Some responses were logically consistent with climate change expectations (i.e. the mean
percent of warm-water-preference individuals was lower in the coldest year samples). Others
were not, perhaps because they were driven more by non-climatic factors, such as nutrient
enrichment, or may reflect indirect mechanisms (e.g., flow or temperature effects on food
resource availability, which then affects certain feeding groups), which cannot be directly
assessed using the biomonitoring data sets. Since we do not yet fully understand the mechanisms
behind these responses, interpreting these relationships and establishing expectations in the
context of climate change is complex.
December 23, 2010
2-38
External Review Draft

-------
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758

-------
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
111
778
779
780
781
782
783
784
785
786
787
788
789
predators (feeding type trait category) was positively correlated with mean annual air
temperature at one Blue Ridge site and negatively correlated with it at a Piedmont site.
Appendices G and I shows the range of significant trends found, most only occurring at single
reference sites.
More metrics were significantly correlated with annual precipitation variables than with
temperature variables; however, inter-site variability is still strong, even within an ecoregion.
Biological metrics that were related to precipitation at more than one site include the Hilsenhoff
Biotic Index (HBI; negatively correlated with mean annual precipitation at two Blue Ridge
sites); the percentage of climbers (negatively correlated with precipitation at one Blue Ridge and
one Piedmont site); the percentage of shredders (negatively correlated with the previous year's
precipitation at one Blue Ridge and one Piedmont site); and the percentage of burrowers
(negatively correlated with the precipitation difference (sampling year - previous year) at one
Blue Ridge site and positively correlated at one Piedmont site). Most of these relationships to
precipitation are consistent with functional expectations. For instance, lower HBI scores during
wetter years are consistent with the observed tendency for cold-water-preference taxa to have
lower HBI tolerance values and to be more abundant during wetter years. The climbing habit
may be disadvantaged during wetter years (e.g., more easily dislodged). On the other hand, the
relationship of invertebrates that burrow to precipitation cannot be interpreted based on the
limited information available and in any case exhibits contrasting responses among sites. Among
trait groups selected for their expected hydrologic relationships, the abundance of perennial taxa
was greater when precipitation was higher, and the richness of intermittent taxa was lower when
precipitation was greater, both as would be expected (Appendix G). Furthermore, both the
abundance and richness of cold-water-preference taxa increased as precipitation increased, which
is consistent with the generally inverse relationship between precipitation and temperature.
However, all these correlations were only significant at one Blue Ridge station. In addition,
abundance of drought-tolerant taxa increased with increasing precipitation at that same station.
This may not be entirely counter to expectation, in that taxa that can tolerate drought may still do
better in more favorable conditions. However, it calls into question the value and sensitivity of
the trait group designation for distinguishing precipitation trends related to climate change.
This notable spatial inconsistency in trends may result from high variability in site-
specific habitat or other environmental conditions (e.g., in factors not specifically accounted for
December 23, 2010
2-40
External Review Draft

-------
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
using biomonitoring data, such as groundwater contribution, riparian cover, substrate conditions,
etc.), and from the relatively limited (short duration) data that are currently available for North
Carolina. Variability between sites even within an ecoregion may also indicate variations in
factors that affect vulnerability to climate change. Elevation and its effect on the relative
contribution of cold-water-preference taxa to the community have been shown in other states to
help define vulnerability to climate-change effects. In the North Carolina Blue Ridge, the higher
elevations were associated with more cold-water taxa on average, but some Blue Ridge
locations, such as the long-term reference Station NC0109 (New River) tended to have more
warm-water-preference taxa. The reasons for this are not clear and warrant further consideration.
The strength of the relatively short (mostly one decade or less at any one station) duration of the
available data is too limited in the face of the magnitude of both spatial and temporal variation to
discern the more subtle long-term trends and relationships needed to define any existing climate-
change responses, and also to define the most effective climate-change indicators.
2.4. CONFOUNDING SOURCES OF TEMPORAL VARIATION
The ability to detect trends in a rigorous manner is affected by the amount and sources of
variation contained in the data, and the ability to control or account for the variation. Interannual
variation is expected to be larger in magnitude than incremental climate change responses (at
least with respect to near-term linear projections based on historic data), and seasonal variation is
often larger than that.
2.4.1. Seasonal variation
In a biomonitoring framework, seasonal variation is typically accounted for by limiting
sampling to a single season or index period. This is the case for the four states (Maine, Utah,
North Carolina, and Ohio) evaluated here, although not all focus on the same index period. In
addition, over the two or more decades of data examined, the range of months during which
sampling was actually conducted within any one state was found to vary over a wider range of
seasons than expected based on current definitions of index period for each state. For example,
Utah's defined index period is late summer to fall; however, sampling dates actually ranged from
March through November (see Appendix F). Incorporation of the full range of available data
over years would have introduced substantial variation in taxon occurrences and abundances due
December 23, 2010
2-41
External Review Draft

-------
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
primarily to seasonal variation. This predictable increase in variation would further obscure other
trends or patterns. The approach to minimizing seasonal variation was to subset data by season
and for most analyses focus only on the predominant index period sampled in each state. For
example for Utah data, sampling months were limited to August through November. Specifics on
sampling months included for each set of trend, correlation, community, and weighted average
modeling analyses conducted can be found in corresponding appendices (mainly see Appendices
E through I, and K). It should be noted that in many cases, this approach reduced the amount of
data available for long-term trend analyses. That is, elimination of "outlying" seasons frequently
eliminated one or more years of data at some locations. While deemed an important procedure to
account for predictable sources of variation, there is a practical impact of reducing the number of
data points for trend analysis, and thus reducing the power to detect trends.
The selection of an index period will also be affected by climate change. Projected
climate changes are likely to impact seasonal patterns through changes in flow conditions as well
as in temperature regimes. These will influence a variety of biological processes, including rates
of development, timing of emergence, and other components of reproduction (Seebens et al.,
2009; Harper and Pecarsky, 2006; Poff et al., 2002; Vannote and Sweeney, 1980). This may have
several ramifications to biomonitoring designs. If samples are collected at a fixed time during the
year, then in the future sampling may yield lower abundances of some species, different species
composition, or different relative abundances. This impacts temporal comparisons. Also, spatial
comparisons may now be based on communities of more limited seasonal diversity. More
extreme or extended summer low flows may, over the long term, become an impediment to
sampling for states that use summer or fall index periods. This may be a particular concern in
perennial streams vulnerable to a shift to intermittent conditions in the future. Biological
responses to reductions in flow can represent legitimate responses to climate change. However,
the eventual inability to sample during a late-season index period in some stream locations must
be considered and planned for. Though highly unlikely due to resource limitations, sampling
more than once per year, including once during the spring/high flow index period, could provide
valuable information on components of the benthic community that emerge early in summer.
2.4.2. Interannual and multi-decadal climatic variation
December 23, 2010
2-42
External Review Draft

-------
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
Biological data also reflect responses to interannual variations (e.g., year-to-year
variations in temperature, precipitation regime, etc); and to multi-year to multi-decadal "cyclic"
climate variations, such as the North Atlantic Oscillation (NAO), the Pacific Decadal Oscillation
(PDO), or the El Nino Southern Oscillation (ENSO) that drive differences in water temperature
and hydrologic regimes in a manner similar to the mechanisms linking to long-term climate
change responses. The NAO, for example, affects mainly winter weather conditions on decadal
time scales (Hurrell, 1995). A rigorous approach, were it supported by available data, could
examine what components of observable temporal variation in biological responses are
attributable to long-term directional climate change, and then apply general linear modeling or
another comparable approach to partition the variation within the observed biological responses
between interannual or cyclic and long-term directional climatic sources. However, most state
biomonitoring data sets, even the most critically developed (sensu Yoder and Barbour, 2009) and
long term, such as those analyzed in these pilot studies, are limited in duration and frequency of
sampling. These data are not able to support linear modeling, especially of several separate
variables, because the average scope of available data is typically 20 years or less, with 10 to
fewer than 20 annual data points over that time span. In addition, it often is the case that needed
covariates, including flow variables, continuous temperatures, and water chemistry parameters
such as nutrients, etc. are not available concurrently with the biological collections.
An alternative approach used here was to test for significant correlations between indices
of known cyclic climatic variation (e.g., the NAO, PDO, and ENSO) with biological metrics,
focusing on those that also showed long-term temporal responses, or correlations with
temperature or precipitation. Details of results can be found in Appendices E through G, and I. In
general, responses varied by state and region, as well as by taxon and trait group. Analyses were
limited to representative reference stations with long-term data sets. In North Carolina, there
were no strong correlations of major trait groups, especially cold- or warm-water-preference
taxa, with annual or winter NAO indices. This is very likely a reflection of the shorter data sets
available in North Carolina among reference stations in the Blue Ridge and Piedmont
ecoregions.
In Maine, only one reference station, 56817 (Sheepscot) in the Laurentian Hills and
Plains, was of sufficient length (23 years) to consider NAO effects. A few interesting
relationships appear, though none are significant. None of the major climate variables,
December 23, 2010
2-43
External Review Draft

-------
882
883
884
885
886
887
888
889
890
891
892
893
894
precipitation, air temperature, or water temperature are substantially related to either the mean
annual or the winter (DJFM) NAO index (Figure 2-27).
With regard to benthic community characteristics, clustering of years at Station 56817
(Sheepscot) based on the Bray-Curtis (Sorensen) similarity index show some evidence of a
temporal pattern (Figure 2-28), with the first four sampling years in cluster 1, and more of the
early sampling years (e.g., 1984-1992 inclusive) within clusters 1 and 2. Later years of sampling
occur more frequently in clusters 3 and 4. This suggests changes in community composition over
time that reflect progressive changes in similarity. However, there are several "misplaced" years,
e.g., 2004 and 2006 are in cluster 2, more similar to the mid- to late 1990's sampling years. One
of these, 2006, is a year with a low NAO index. But 2004 is an "average" NAO index year, and
overall, there is no pattern associating the distribution of years among clusters with the NAO
index (Figure 2-28).
December 23, 2010
2-44
External Review Draft

-------
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909

8,8

8.6
o
8.4

O)

3
8.2


0>
CL
8.0
E

0)
7,8
ro

m
7.6
3

a
c
7,4
flj

c
to
7.2


E


7.0
CO

K.
6.8
a.


6.6

64
A


0
0


>

o

0
o


Oo
0


0

0
o o


o

0 O
o <
o
0
o

o
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
NAO (DJFM PC)
B

o

° o


o
o


o

o

o

0 o
0
o
0
0
o
o


6>

o
° o

-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
NAO (Annual PC)
Figure 2-27. Relationships between North Atlantic Oscillation (NAO) indices and PRISM
climatic variables at Maine site 56817 (Sheepscot). Plot (A) shows NAO winter index
(December-January-February-March (DJFM)) vs. PRISM mean annual temperature (°C)
(r=-0.15, p=0.51) and (B) shows PRISM mean annual precipitation (inches) vs. NAO
annual index (r=0.03, p=0.90).
There is a modest relationship between benthic assemblages at Maine Station 56817
(Sheepscot) and NAO patterns when stability or persistence of the community is tested, based on
degree of change in community similarity among years. The community was more stable, that is
more similar between years, based on Euclidean distances during negative NAO phases, and
more variable during positive phases (Figure 2-29). This is consistent with findings in Wales,
where benthic community persistence was related to both long-term climate effects and cyclic
effects of the NAO (Bradley and Ormerod, 2001). Persistence reflected environmental
variability, with high persistence during negative NAO phases (cold, dry winters in northern
December 23, 2010
2-45
External Review Draft

-------
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
Europe) and low persistence (high variability) during positive NAO phases (milder, wetter
winters). In New England, positive winter indices are associated with more winter storms, while
negative winter indices are associated with fewer storms and drier winter conditions (New
England Regional Assessment Group, 2001). Apparently the greater environmental variability
introduced through more frequent winter storms to stream temperatures, flow conditions, and
water quality translate to more variable benthic assemblages. This suggests that the NAO may be
an important, climate-related influence on interannual patterns in benthic community responses,
even though in most of the correlations of NAO index with community, trait group, or taxonomic
group parameters were relatively weak.
[
[
22EC2
W
[
1984
1986
1987		
1999		
1988	—[_
1990		
1994		
1997
1991		
1992
2004		
2006 	
1983 	
2000		
2005		
2003 	
1995		
1999 	
1998		
2002 	
55E01
dLEter_53817
~stars (Ogective Fincticri)
1.1EKD
Irfcm^icn feraring (°/^
3D
1.6BC0
	I	
21EKD
25
High NAO index years
Figure 2-28. Cluster analysis using Bray-Curtis (Sorensen) similarity index, based on
benthic invertebrate composition using genus-level OTUs, at Maine station 56817
(Sheepscot).
The PDO, which influences western and southwestern regions, is generally considered to
be a much longer term, multi-decadal phenomenon (Brown and Comrie, 2004; Mantua et al.,
1997), while ENSO is found to vary in the range of multiple years to a decade or more. In Utah,
there were some intriguing relationships found at individual long-term reference stations
between trait groups (e.g., warm-water-preference taxa, perennial taxa, etc. - see earlier sections
of this chapter and Appendix I) and either the ENSO or PDO annual or monthly indices (see
Appendix I). However, none of these were consistent spatially; therefore, no particular trait or
December 23, 2010
2-46
External Review Draft

-------
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
taxonomic group is a strong indicator of PDO or ENSO responses. The complexity of the
patterns compared to the relatively short (20 years or fewer) data sets suggests the importance of
further investigation and long-term monitoring, including further study on the relative
contributions of each index.
0.25
J 0.15
y « 0 tt&fc * 0 0613
R'* 0 2088
0.1
0.06
y* -0 0061* + 0 0539
R>" 0 0127

-1,5
0	0.5	1
NAO, Annual
1.5
25
Figure 2-29. Correlation between Euclidean distance calculated as a difference between
successive sampling years, as a measure of similarity between benthic assemblages, and the
NAO annual index. In this case, Euclidean distance is plotted against the NAO index for
the first year (e.g., 1984-85 comparison against 1984 NAO index), creating a 1-year lag.
While it is important to consider NAO, PDO, and/or ENSO when evaluating
biomonitoring (or any other) data sets for climate change effects, there are still some practical
limitations, particularly in the biomonitoring application. Fundamentally, the analyses require
data spanning multiple (2-3) multi-decadal cycles to be able to model the cycle-associated
responses and extract the residual long-term trend on a rigorous basis. The Maine Station 56817
(Sheepscot) data series spanned 23 years, and this is long compared to most existing available
biomonitoring data. It also is likely that variations in the effects of the NAO interact with long-
term climate change effects, potentially damping increasing temperatures in negative years and
augmenting them in positive years (Durance and Ormerod, 2007). This is important, because the
composite of climate effects may underestimate long-term climate impacts during some periods,
or overestimate them during others. It would take proportionately more (longer-term) data to
separate these and confidently define the long-term climate change component.
December 23, 2010
2-47
External Review Draft

-------
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
2.4.3. Interpretation of directional climate change effects
Since the nature of most bioassessment data limit the ability to separate the magnitude of
observed trends among interannual, cyclical, and long-term directional climate effects, the results
obtained in this study cannot be interpreted as entirely attributable to directional climate change.
However, the net response of benthic or other aquatic community metrics to climate sensitive
variables, including water temperature and hydrologic patterns, can reasonably and effectively be
used to address the primary questions of this study. The direction and nature of the observed
climate responses can be applied to characterize the types of responses that can be expected due
to climate change, to identify the most sensitive indicators to climate change, and to understand
implications to multimetric indices or predictive models and their application by managers to
characterize condition of stream resources for decision making. These effects may be viewed in
some respects as maximum estimates of probable effects, since multiple components of climate
change could be included, though at times, the resulting estimates may also be undervalued.
2.5. OTHER SOURCES OF POTENTIAL SPATIAL CONFOUNDING
There are other potential sources of spatial confounding of temporal trends, which were
tested in this study. Land use and land cover within a 1 km buffer of the individual reference
sites indicated that anthropogenic influences were higher than desired (>5% urban or >10%
agricultural) at most sites. The urban land uses surrounding these sites generally consisted of
low-intensity and open-space development, and the agricultural land uses were mostly
pasture/hay, with occasional cultivated crops. We further explored these relationships by using
correlation analyses to determine whether any available chemistry and habitat variables were
significantly correlated with biological metrics. Data availability limited this pursuit. For
example, Utah only had chemistry data. At the two Utah long-term reference stations that
showed strong temperature-related trends (Stations 4927250 - Weber and 4951200 - Virgin),
some of the temperature preference metrics were significantly correlated with water chemistry
variables. Many of the correlations were driven by outliers, but a few of the water chemistry
variables, notably chloride, may have influenced trends in the biological assemblage (Figure 2-
30; Appendix I). Chloride could be an indirect indicator of human development, as increases are
sometimes associated with increasing road development and/or increasing application of road
December 23, 2010
2-48
External Review Draft

-------
986
987
988
salt over time (NRC, 1991). However, chloride concentrations may also vary naturally with
drought conditions.
989
990
991
992
993
35
30
25
20
15
10
5
0
30
25
20
15
10
0
30
25
20
15
10
¦ Number of total taxa
>C'
• Number of EPT taxa
-1	1	1	I	1	1—
¦ Number of cold water
preference taxa
't
»\
I \
I *
I t
\ ^ V"
V ^
x 	\..OC
1985 1988
1991
1994 2000
2004
PRISM mean annual air
temperature ("C)
¦ PRISM mean annual
precipitation (inches)
Chloride (mg/L)
Figure 2-30. Trends in selected metrics, PRISM climatic variables and chloride
concentrations over time at Utah site 4927250 (Weber). Plot (A) shows number of total
taxa, (B) number of EPT taxa, and (C) number of cold-water-preference taxa. Data used in
these analyses were limited to autumn (September-November) kick-method samples.
December 23, 2010
2-49
External Review Draft

-------
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
In Maine, limited chemistry and habitat information were available (mainly in situ water
quality measurements and visual substrate estimates). At site 56187 (Sheepscot), yearly trends in
the biological data were likely influenced by nonpoint-source pollution (pers.comm. Maine
DEP), but we lack the long-term chemistry data necessary to confirm this possibility. Some of
the habitat variables at site 56817 also showed trends over time. Percent boulders and % gravel
were significantly correlated with some of the biological variables. However, based on
conversations with Maine DEP, it appears that this 'trend' actually reflects observer bias, and it
is not considered a real change over time in substrate characteristics. A similar example occurred
in North Carolina, where visual substrate estimates for one site showed a fairly dramatic yearly
trend. Scientists at NCDENR believe this also to be observer bias. More problematically, there
were some fairly dramatic trends in canopy cover and water chemistry found at some North
Carolina sites, which turned out to be due to data entry errors. This seems a minor but important
cautionary note, as the "false" trend in canopy cover seemed feasible (increasing cover over time
would be possible if there were an earlier instance of logging), and a (non-significant) trend of
decreasing water temperature over time appeared to be logically consistent with increasing
canopy cover. In the end, this very "appealing" discovery was false.
2.6. COMPARISON OF REGIONAL TRENDS, VULNERABILITIES, AND
INDICATORS
Trends over time and in association with climate variables have been found within the
state bioassessment data sets and in particular with ecological or life history trait groups (Section
2.2) (Gallardo et al., 2009; Beche and Resh, 2007; Bonada et al., 2007b). Unfortunately, there
are numerous examples within this study in which observed trends were significant in some
places but not in others. Spatial consistency can be used as evidence that a particular trend or
relationship is important. But even with the extensive biomonitoring data sets analyzed in this
study, it was rare to have more than one or two reference sites within a region with sufficient
data to conduct satisfactory long-term trends analyses. The focus on reference stations is needed
to minimize contributions of effects from sources other than climate change (potential
confounding factors such as urban or agricultural land use, see Appendix C for land use and
other criteria used for reference station screening). We used an alternative approach of grouping
reference stations within an ecoregion (Appendix C) to increase spatial coverage for trend
December 23, 2010
2-50
External Review Draft

-------
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
analyses. However, this was marginally successful, and there was still high inter-site variability
in factors that affect community comparability (e.g., elevation, stream size, general
geography/topography). As a result, variation among sites was almost always greater than the
magnitude of any long-term temporal trends. Some component of spatial variability between
sites may reflect real spatial variation in degree of vulnerability to climate change effects.
Regional variation in trends for ecological trait groups defined by temperature
preferences also reflects the lack of spatial consistency in results. The number of warm-water-
preference taxa increased significantly over time at lower elevation locations in both Maine (site
56187 - Sheepscot, site 57011 - W. Br. Sheepscot) and Utah (site 4951200 - Virgin), but not in
North Carolina, and not at all stations (Table 2-4). The increasing temporal trend in warm-water-
preference taxa was corroborated by correlation with temperature in Utah, but not in Maine
(Table 2-5). At the longest-term station in Maine (56817) cold-water taxa also increased, but this
is generally counter to climate change expectations. Though significant (see Table 2-2), the
number of taxa was so low as to make the apparent trend largely meaningless. Cold-water-
preference taxa decreased over time at one of the higher elevation sites in Utah (site 4927250 -
Weber), and was also negatively correlated with temperature, as would be expected for a trait
group responding to climate change increases in temperature. Comparable associations with
temperature were not found in Maine or North Carolina. For these two states, cold-water-
preference taxa were instead more often related to precipitation (Table 2-6).
December 23, 2010
2-51
External Review Draft

-------
1046	Table 2-4. Results of Pearson product moment correlation analyses done to examine associations between year and a selected
1047	group of metrics at long-term biological monitoring sites in Utah (UT), Maine (ME), and North Carolina (NC). Significant
1048	relationships at/; < 0.05 are shown in bold with shading; at p<0.1 in bold. NA=not available (we did not calculate the Shannon-
1049	Wiener diversity index for North Carolina samples because the abundance data were categorical).

Utah
Maine
NC
Biological Metric
vs. YEAR
4927250
4951200
4936750
5940440
56817
57011
57065
NC0109
r
P
r
P
r
P
r
P
r
P
r
P
r
P
r
P
Cold water taxa
richness
-0.71
0.00
-0.62
0.02
-0.38
0.23
-0.64
0.07
0.49
0.02
0.04
0.90
0.54
0.13
0.55
0.08
Cold water taxa
relative abundance
-0.72
0.00
-0.63
0.02
-0.15
0.64
-0.12
0.76
0.47
0.03
-0.67
0.02
0.45
0.23
0.57
0.07
Warm water taxa
richness
-0.21
0.42
0.85
0.00
0.38
0.22
NA
NA
0.78
0.00
0.65
0.02
0.58
0.10
-0.58
0.06
Warm water taxa
relative abundance
-0.21
0.42
0.41
0.15
0.42
0.17
NA
NA
0.55
0.01
-0.59
0.04
-0.36
0.34
-0.04
0.90
Total taxa richness
-0.29
0.26
-0.28
0.34
0.08
0.81
-0.54
0.14
0.75
0.00
0.81
0.00
0.56
0.11
-0.67
0.02
EPT taxa richness
-0.59
0.01
-0.49
0.08
-0.21
0.52
-0.65
0.06
0.75
0.00
0.76
0.00
0.51
0.16
0.30
0.36
EPT relative
abundance
0.06
0.81
0.06
0.85
-0.26
0.42
0.44
0.23
0.06
0.80
-0.52
0.08
-0.36
0.34
0.74
0.01
Ephemeroptera
taxa richness
-0.57
0.02
-0.60
0.02
-0.04
0.89
-0.57
0.11
0.58
0.01
0.63
0.03
0.37
0.33
0.22
0.52
Plecoptera taxa
richness
-0.76
0.00
-0.53
0.05
-0.29
0.36
-0.71
0.03
-0.16
0.47
0.05
0.88
0.44
0.23
0.45
0.16
Shannon Wiener
diversity index
0.13
0.62
-0.43
0.12
-0.08
0.81
-0.25
0.52
0.64
0.00
0.12
0.72
0.43
0.25
NA
NA
OCH taxa richness
0.61
0.01
0.46
0.10
0.83
0.00
0.28
0.47
0.43
0.04
0.43
0.16
0.28
0.47
0.10
0.78
OCH taxa relative
abundance
0.66
0.00
0.40
0.15
0.32
0.31
-0.59
0.10
0.28
0.20
-0.52
0.09
0.37
0.33
0.14
0.68
HBI
-0.19
0.47
0.28
0.34
0.32
0.31
-0.46
0.21
-0.13
0.54
0.75
0.01
0.18
0.65
-0.63
0.04
1050
1051
December 23, 2010
2-52
External Review Draft

-------
1052
1053
1054
1055
1056
Table 2-5. Results of Pearson product moment correlation analyses done to examine associations between Parameter-
elevation Regressions on Independent Slopes Model (PRISM) mean annual air temperatures and a selected group of metrics at
long-term biological monitoring sites in Utah (UT), Maine (ME), and North Carolina (NC). Significant relationships at p <
0.05 are shown in bold with shading; at p<0.1 in bold. NA=not available (we did not calculate the Shannon-Wiener diversity
Biological Metric
vs. TEMP
4927250
4951200
4936750
5940440
56817
57011
57065
NC0109
r
P
r
P
r
P
r
P
r
P
r
P
r
P
r
P
Cold water taxa
richness
-0.63
0.01
-0.73
0.00
-0.08
0.82
-0.14
0.73
0.31
0.15
0.02
0.95
-0.58
0.10
-0.38
0.25
Cold water taxa
relative abundance
-0.30
0.24
-0.56
0.04
-0.20
0.53
-0.29
0.46
0.15
0.50
-0.16
0.62
-0.27
0.48
-0.32
0.34
Warm water taxa
richness
-0.44
0.08
0.76
0.00
-0.03
0.93
NA
NA
0.21
0.34
0.27
0.39
-0.73
0.02
-0.18
0.59
Warm water taxa
relative abundance
-0.35
0.17
0.62
0.02
0.01
0.98
NA
NA
0.13
0.55
0.37
0.23
0.05
0.90
0.00
1.00
Total taxa richness
-0.48
0.05
-0.68
0.01
-0.08
0.81
-0.20
0.60
0.29
0.18
0.10
0.76
-0.52
0.15
0.04
0.91
EPT taxa richness
-0.57
0.02
-0.79
0.00
-0.09
0.77
-0.43
0.25
0.17
0.44
0.25
0.43
-0.64
0.06
0.00
0.99
EPT relative
abundance
0.03
0.91
0.29
0.32
0.04
0.90
0.07
0.86
0.08
0.71
0.64
0.03
-0.07
0.87
-0.09
0.80
Ephemeroptera
taxa richness
-0.59
0.01
-0.81
0.00
-0.26
0.41
-0.22
0.56
0.19
0.39
0.51
0.09
-0.62
0.08
-0.18
0.60
Plecoptera taxa
richness
-0.45
0.07
-0.65
0.01
0.17
0.61
-0.72
0.03
0.09
0.70
-0.06
0.85
-0.56
0.12
0.12
0.73
Shannon Wiener
diversity index
-0.13
0.62
-0.67
0.01
-0.12
0.71
-0.29
0.46
0.45
0.38
0.45
0.14
-0.59
0.09
NA
NA
OCH taxa richness
0.11
0.68
0.12
0.68
0.27
0.40
0.59
0.09
0.13
0.54
0.35
0.26
-0.10
0.80
0.14
0.68
OCH taxa relative
abundance
0.44
0.07
0.27
0.36
-0.11
0.74
-0.01
0.98
0.01
0.98
-0.09
0.79
-0.33
0.38
0.30
0.37
HBI
-0.32
0.21
0.04
0.89
0.09
0.77
0.09
0.82
-0.07
0.76
-0.21
0.51
0.13
0.75
0.13
0.71
1057
December 23, 2010
2-53
External Review Draft

-------
1058
1059
1060
1061
1062
Table 2-6. Results of Pearson product moment correlation analyses done to examine associations between Parameter-elevation
Regressions on Independent Slopes Model (PRISM) mean annual precipitation and a selected group of metrics at long-term
biological monitoring sites in Utah (UT), Maine (ME), and North Carolina (NC). Significant relationships atp < 0.05 are
shown in bold with shading; at p<0.1 in bold. NA=not available (we did not calculate the Shannon-Wiener diversity index for

Utah
Maine
NC
Biological Metric
vs. PRECIP
4927250
4951200
4936750
5940440
56817
57011
57065
NC0109
r
P
r
P
r
P
r
P
r
P
r
P
r
P
r
P
Cold water taxa
richness
-0.11
0.68
0.44
0.12
0.42
0.17
0.01
0.98
0.44
0.04
0.18
0.59
-0.51
0.16
0.85
0.00
Cold water taxa
relative abundance
0.08
0.75
0.23
0.42
0.30
0.35
0.54
0.14
0.58
0.00
0.03
0.93
-0.02
0.97
0.63
0.04
Warm water taxa
richness
-0.05
0.84
-0.18
0.53
0.21
0.50
NA
NA
0.07
0.75
-0.04
0.91
-0.13
0.73
-0.65
0.03
Warm water taxa
relative abundance
-0.14
0.60
-0.34
0.23
0.33
0.29
NA
NA
0.04
0.85
-0.10
0.76
-0.44
0.23
-0.57
0.07
Total taxa richness
-0.15
0.56
0.57
0.04
0.43
0.16
-0.07
0.85
0.28
0.21
0.15
0.63
-0.28
0.47
-0.64
0.04
EPT taxa richness
-0.25
0.34
0.68
0.01
0.45
0.14
0.17
0.66
0.20
0.37
0.24
0.45
-0.12
0.76
0.36
0.28
EPT relative
abundance
-0.29
0.27
0.07
0.82
0.32
0.30
0.32
0.40
0.01
0.97
-0.05
0.88
0.17
0.66
0.82
0.00
Ephemeroptera
taxa richness
-0.20
0.45
0.58
0.03
0.34
0.28
0.02
0.97
0.35
0.11
0.45
0.14
-0.09
0.83
0.24
0.47
Plecoptera taxa
richness
-0.34
0.18
0.40
0.16
0.22
0.49
0.29
0.45
0.18
0.43
-0.19
0.56
-0.32
0.41
0.62
0.04
Shannon Wiener
diversity index
-0.49
0.05
0.10
0.74
0.49
0.10
-0.14
0.72
0.25
0.27
0.09
0.79
0.01
0.97
NA
NA
OCH taxa richness
0.46
0.06
0.27
0.35
0.15
0.64
-0.30
0.44
0.28
0.20
0.07
0.82
-0.44
0.24
-0.06
0.86
OCH taxa relative
abundance
-0.04
0.89
-0.29
0.32
0.25
0.43
-0.28
0.47
0.25
0.24
0.29
0.36
-0.33
0.39
-0.22
0.52
HBI
0.16
0.53
0.11
0.71
-0.55
0.06
-0.37
0.32
-0.22
0.31
0.25
0.43
-0.27
0.49
-0.86
0.00
1063
1064
December 23, 2010
2-54
External Review Draft

-------
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
As presented above, the distributions of cold-water-preference taxa (richness and relative
abundance) were significantly associated with elevation, stream size (order) and watershed size,
such that more cold-water-preference taxa were present at higher elevations and in smaller
streams and watersheds, and warm-water-preference taxa were more common at lower
elevations, and in larger streams and watersheds (Table 2-7, figures in Section 2.2.2). Though
not consistently demonstrated in all states, some higher elevation ecoregions with a greater
predominance of cold-water-preference taxa exhibited greater responsiveness (e.g., more
significant trends) to changes in climate variables (Appendix A). The sizes of streams sampled in
those ecoregions probably interacted with elevation differences. For example, in most of the
states, low-order streams tended to be under-sampled. Mid-order streams even at relatively
higher elevations might have fewer cold-water-preference taxa than lower order streams would.
Nevertheless, higher elevation regions, as well as areas subsetted by stream and watershed size,
should be evaluated for vulnerability to climate changes in temperature and hydrologic
conditions. This factor should be accounted for in assessing climate change monitoring priorities.
The hydrologic indicator metrics generally failed to show significant trends for a number
of reasons. The metrics as developed might not be effective at detecting shifts in hydrologic
regimes and may need to be further refined. Limited knowledge about life history, mobility,
morphology, and temperature preference and tolerance information is currently one of the major
limitations of traits-based metrics; more information could improve metric performance. It also
is possible that the metrics are effective and are simply documenting that there are no consistent
patterns yet. Precipitation tends to be highly variable and can be difficult to predict or model
(e.g., Brown and Comrie, 2004; 2002). Analyses conducted on Piedmont sites in North Carolina
as an adjunct to this study (Appendix J) indicate that natural stream communities appear to be
resilient within the range of natural hydrologic variability. Because of this resilience, effects
from hydrologic changes associated with climate change may not be seen unless these changes
are large. This may happen as the magnitude of effects increases. Analyses were conducted on
reference site data with natural flow regimes; it is possible that the metrics may be effective, but
that shifts in hydrology over the short periods of record have not followed consistent patterns.
Results of this study, as well as other research (Webb et al., 2009; Dewson et al., 2007;
Suren and Jowett, 2006; Lind et al., 2006; Poff, 2002; Extence et al., 1999; Stanley et al., 1994)
have demonstrated the importance of hydrologic changes on biological responses, and it will be
December 23, 2010
2-55
External Review Draft

-------
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
worthwhile to consider both "scenario-based" and hydrologic metrics further in the future. These
classes of indicators may be most valuable in regions, such as North Carolina, where associations
with precipitation are already strong, and where other evidence suggests the dominance of
hydrologic drivers (NCDENR, 2005). Hydrologic metrics are also likely to be valuable in
regions with strong future vulnerability to hydrologic impacts due to the combination of climate
change predictions for temperature and precipitation, such as in the arid west and southwest.
Analyses testing for relevant biological responses to climate patterns often lacked spatial
consistency both within and across states. Several biological metrics, evaluated for differences
between years partitioned based on temperature (hottest/coldest/normal years) or precipitation
(wettest/driest/normal years) regime showed patterns in one or another state (see the above
subsections of this chapter), but only a few showed statistically significant patterns at sites in
more than one state, and none showed common patterns among all states. Overall, more metrics
were significantly associated with temperature-related variables than with precipitation variables
(Appendix I). While long-term increasing trends in temperature already can be demonstrated for
many regions (see Section 2.1 and Appendix A), this is seldom the case for precipitation or flow-
related variables (Appendix I). Long-term data for flow (e.g., IHA) variables tend to be scarcer;
and climate change projections for precipitation are small in magnitude and variable for many
regions. Nevertheless, the importance of ongoing changes in precipitation effects on flow regime
should not be discounted.
Other biological metrics that were sometimes responsive to climate variables include
functional feeding groups (e.g., predators, collector-filterers) or life history habits (e.g.,
swimmers, climbers) (Appendix I). Feeding, life habit, and other functional trait groups are often
included as metrics in state MMIs. It is thus recommended that, on a case by case basis, the
vulnerability of this class of metrics be evaluated through trend and correlation analysis, as well
as through assessment of composition by temperature sensitive taxa.
December 23, 2010
2-56
External Review Draft

-------
Table 2-7. Summary of differences in elevation, PRISM mean annual air temperature and precipitation and mean number
and percent of cold and warm-water-preference taxa across and within major ecoregions in each state. Only full-scale samples
were used to derive the numbers for North Carolina. Samples were not limited to particular seasons in Utah and North
Carolina. Mean % individuals of cold and warm water in the North Carolina Coastal ecoregion were not calculated (our
analyses were concentrated in the Mountain and Piedmont ecoregions).




Air
Richness
Relative Abundance


#
Samples
Elevation
(m)




State
Ecoregion
temperature
(°C)
Cold water
Warm
water
Cold
water
Warm water

Northeastern Coastal Zone
576
29.3
8.3
1.7 ± 1.9
3.3 ± 2.8
5.4 ±9.9
17.0 ±20.6
Maine
Laurentian Plains &Hills
2830
65.2
6.5
1.1 ± 1.4
4.7 ±3.3
2.8 ±6.6
22.4 ±22.0

Northeastern Highlands
857
210.4
5.8
1.7 ±2.0
3.2 ± 2.7
7.1 ± 11.8
15.1 ± 17.5

Mojave Basin & Range
13
736.6
16.8
2.8 ±2.4
1.3 ±0.9
6.6 ±8.9
5.5 ±8.7

Central Basin & Range
111
1411.7
10.0
1.4 ± 2.0
2.4 ± 1.4
2.1 ±7.0
10.8 ± 16.5

Colorado Plateaus
205
1729.4
9.1
3.8 ±2.8
1.2 ± 1.2
9.8 ± 11.5
6.1 ± 11.6
Utah
Northern Basin & Range
6
1769.7
8.6
4.7 ± 1.0
1.2 ±0.8
3.2 ±2.9
12 ±20.1

Wyoming Basin
27
2002.0
5.7
6.1 ±4.0
1.3 ±0.9
13.2 ± 13.2
1.1 ±2.4

Wasatch & Uinta Mountains
644
2131.1
5.4
5.5 ±4.0
1.0 ± 1.3
13.1 ± 15.4
3.8 ± 11.0

Southern Rockies
7
2535.2
6.3
9.1 ±0.7
0±0
30.6 ± 14.6
0±0

Middle Atlantic Coastal Plain
173
4.7
16.7
0.1 ± 0.2
4.7 ±5.1
0.1 ±0.4
12.3 ±6.4
North
Southeastern Plains
317
34.1
16.3
0.1 ±0.4
8.8 ±3.4
0.1 ±0.4
12.1 ±5.1
Carolina
Piedmont
1106
183.5
15.0
1.5 ±2.0
5.2 ±3.1
1.8 ± 2.7
6.7 ±4.7

Blue Ridge
631
714.5
12.1
8.0 ±4.5
2.8 ±2.4
11.4 ± 7.9
3.1 ± 3.7
December 23, 2010
2-57
External Review Draft

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
The Shannon-Wiener diversity index is another metric that is often included in MMIs,
which showed an inverse relationship with temperature at one Utah station (site 4951200 -
Virgin) (Table 2-5). Both the value and the drawback of using overall community diversity as a
metric of condition is that it is a composite response of all community components. This study
shows that, in at least some regions, overall community diversity is reduced in hotter years due to
suppression of cold-water-preference taxa (Table 2-5). This response is mediated by the relative
composition of cold and warm taxa, which is also associated with elevation and stream size.
Potential modification of this metric to help track climate change effects will have to be site or
region specific, and should initially focus on the relative contribution of cold and warm-water-
preference taxa within the community.
The abundance or richness of OCH taxa are more rarely incorporated as an MMI metric.
It functions as a contrasting metric to EPT taxa, due to the generally high environmental
tolerances of these taxa and expectation that they do better in the summer and in drier, more
intermittent conditions (Bonada et al., 2007a). The potential robustness of OCH taxa to climate
change effects was considered important. In fact, the abundance of OCH taxa was higher during
hot years in some locations, though the trends were not statistically significant (Table 2-4,
Appendix I). Still, this may be a valuable indicator to consider in the future.
Climate change "scenarios" (e.g., warmer and drier conditions) were used to combine
temperature preference traits with other ecological (e.g., hydrologic preferences) and life history
traits in an attempt to improve both the detection of responses to climate variables, and impart
greater ability to explain the responses and use this information to develop more effective
indicators. Overall, temperature-preference metrics by themselves were more responsive to
climate change variables in more regions tested than were these composite scenario metrics.
Only one of the scenario metrics, percent drier-vulnerable taxa, showed significant patterns at
more than one site (Appendix I). Further consideration of this "climate scenario" trait suite
approach may still be fruitful in the future. The limitation is that selection for multiple traits
tends to reduce the number of member taxa, and therefore limits the amount of data for trend
analyses.
December 23, 2010
2-58
External Review Draft

-------
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
3. IMPLICATIONS TO MULTIMETRIC INDICES, PREDICTIVE
MODELS, AND IMPAIRMENT/LISTING DECISIONS
For states and tribes to assess stream condition, the extensive biological monitoring data
collected on macroinvertebrate, fish, and/or other stream and river communities must be distilled
to a format that accurately and reasonably reflects condition. That is, the result must be a good
"indicator" or "index" and must be readily compared between reference and affected conditions.
The main categories of such computational approaches are multi-metric indices (MMIs) and
predictive models.
MMIs are generally structured as a composite of biological metrics selected to capture
ecologically important community structural or functional characteristics and have been applied
to fish and benthic macroinvertebrate communities (Norris and Barbour, 2009; Bohmer et al.,
2004; Sandin and Johnson, 2000; Barbour et al, 1995; Yoder and Rankin, 1995; DeShonn, 1995;
Karr, 1991). Component metrics are selected based on their responsiveness to the environmental
impacts most often evaluated. Sites are assessed by comparing the test location MMI to that
calculated for applicable reference locations, grounded in the assumption that degradation in the
MMI reflects aquatic community responses to pertinent environmental stressors.
There is much variation among states and tribes in the particular components included in
MMIs or predictive models, because, as a rule, they are calibrated to the state, or more often, to
regions within a state to account for predictable (natural) variability (Barbour and Gerritsen,
2006). Added to this index variability is the regional variability in both climate change
projections and associated biological responses. These sources of variability make
generalizations about the implications of climate change for bioassessment indices challenging.
However, there are some commonalities among states, such as the categories of metrics used,
which we use to investigate vulnerabilities of these approaches to climate change.
Predictive models use regional reference conditions to develop relationships between
environmental predictor variables and macroinvertabrate taxon occurrence from which
predictions for an "expected" (E) community are based. A commonly applied model for
macroinvertebrate communities is the River InVertebrate Prediction And Classification System
(RIVPACS) (Wright, 2000). An important assumption is that the predictor variables are
minimally affected by human disturbance and are relatively invariant over ecologically-relevant
time (Utah State University, 2009; Tetra Tech, 2008; Hawkins et al., 2000; Wright, 2000; Wright
December 23, 2010
3-1
External Review Draft

-------
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
et al., 1984). The E community is then compared to various "observed" (O) communities at non-
reference locations. A basis for comparison is that any differences between O and E communities
reflect biological responses to the range of environmental pollutants or alterations that are
intended to be evaluated. This is similar to the MMI approach.
Among the four states evaluated in this study, three of them, Maine, North Carolina, and
Ohio, use some form of MMI. Utah uses a predictive model, RIVPACS, for assessing wadeable
streams. These states are representative of major regions of the US, encompassing large-scale
variations in climate, climate change projections, geography, topography, geology, and
hydrology. State-specific analysis results also inform a regional view of climate change
implications to commonly used MMIs and predictive models.
3.1. MAINE AND THE NORTHEAST
Maine uses 4 linear discriminant models that incorporate 30 input metrics to assign sites
to one of four classes (A, B, C, and NA, where A represents the best conditions, and NA is Non-
Attainment), applying the same criteria to all sites. Vulnerabilities of the component metrics to
climate change can be evaluated, but it is difficult to extend the results to impacts on station
classifications, because the discriminant model inherently looks at multiple variables
simultaneously. There are no firm thresholds or individual metric values at which a sample
changes classification levels. Analyses of the differences in each component metric among rating
classes (summarized in Appendix E) provide the basis for comparing climate-related sensitivities
of these metrics. Overall, stations with the following characteristics received better ratings:
•	High generic richness
•	High richness and abundance of EPT taxa
•	High Shannon-Wiener diversity index values
•	Low HBI scores
•	Low Chironomidae abundances
•	Low relative Diptera richness
•	Low relative Oligochaeta abundance
•	Greater presence of Class A indicator taxa
•	Greater scraper relative abundance
A variety of analyses were used to characterize possible vulnerabilities of Maine's
discriminant model approach (see Appendix E for details). For instance, ANOVA was used to
December 23, 2010
3-2
External Review Draft

-------
96	evaluate whether certain model components were more important than others in distinguishing
97	between different classes. Temperature preferences and tolerances of Class A indicator taxa were
98	examined, as were Biological Condition Gradient (BCG) assignments and tolerance values of
99	cold- and warm-water-preference taxa.
100	Climate change effects are likely to influence a number of Maine's discriminant model
101	input metrics. Eight of these are metrics related to EPT taxa, which are also used by other
102	northeastern states. In Maine, the vulnerabilities of EPT taxa are largely related to the ecological
103	trait of temperature preference. Twenty nine (29) of the Maine cold-water-preference taxa are
104	EPT taxa (Table 3-1). There are also 18 EPT taxa on the warm-water-preference list (Table 3-2).
105
106	Table 3-1. Number of Maine cold-water taxa in each order with EPT taxa in italics.
Order
Total
Plecoptera
16
Trichoptera
10
Diptera
7
Ephemeroptera
3
Coleoptera
2
Odonata
2
Megaloptera
1
107
108	Table 3-2. Number of Maine warm-water taxa in each order with EPT taxa in italics.
Order
Total
Diptera
10
Ephemeroptera
9
Trichoptera
6
Basommatophora
4
Plecoptera
3
Arhy nchob dellida
1
Coleoptera
1
Decapoda
1
Haplotaxida
1
Hoplonemertea
1
Hydroida
1
Mesogastropoda
1
Odonata
1
109
110	More of the ephmeropteran (mayfly) taxa are warm-water- than cold-water-preference
111	taxa. Two of the model input metrics used by Maine are specifically related to ephemeropterans
December 23, 2010
3-3
External Review Draft

-------
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
(metrics for both absolute and relative abundance). On average, higher values for the
Ephemeroptera abundance metric occur at Class B sites (Figure 3-1A), while the highest relative
abundances occur at Class A sites (Figure 3-2A). Thus, increases in warm-water
ephemeropterans as temperature increases with climate change can affect station classification.
For example,former Class C sites could become Class B sites due to the addition of these taxa,
while the same trend of increasing Ephemeroptera abundance might degrade former Class A
sites to Class B. If the relative abundance of ephemeropterans increases as their absolute
abundance increases (which would, of course, depend on the relative responses of other taxa as
well), then station classification of any condition class might increase. Increasing abundance of
warm water ephemeropterans at Maine's longest term reference station (56817 - Sheepscot in the
Laurentian Hills and Plains) over the 22-year sampling period has already resulted in a difference
in these ephemeropteran metrics comparable to the average difference between these metrics at
Class A and B sites (Figures 3-1 and 3-2). For example, ephemeropteran abundance (Figure 3-
1B) increased from just under 100 per sample in the first 5-years to close to 200 per sample (but
with high variability) in the last 5 years. This range approximates the mean difference between
Class A and B stations, or between Class B and C stations (Figure 3-1 A).
December 23, 2010
3-4
External Review Draft

-------
800
700
600
500
400
300
200
100
0
-100
700
B	C
Class
o Median
O 25%-?5%
T NorvOuttier Range
" Outliers
* Extremes
¦ PRISM mean annual air
temperature fC)
^ PRISM mean annual
precipitation finches)
- Ephemeroptera
flbundsnw
129
130
131
132
133
134
135
1985 1968 1991 1994 199? 2000 2KB 2008
Figure 3-1. Ephemeroptera abundance in Maine.. Plot (A) shows distributions of
Ephemeroptera abundance metric values across classifications (A, B, C, NA) and (B) shows
trends in Ephemeroptera abundance and PRISM climatic variables over time at Maine site
56817 (Sheepscot). Data used in these analyses were limited to summer (July-September)
rock-basket samples.
December 23, 2010
3-5
External Review Draft

-------
136
137
138
139
140
141
142
143
144
145
146
147
148
1.0
0,8
I
5 0,8
to
m
f
o
5
Q.
Ill
®
oc
0.4
0.2
0.0
-0,2
J=
NA
Class
~	Median
~	25%-75%
" X Non-Outlier Range
o Outliers
+ Evtiemes
...... prism mean annual air
temperature (*C)
	PRISM mean annual
precipitation (inches)
	Relative
Epherneroptera
abundance x100
1985 1988 1991 1994 1997 2000 2003 2006
Figure 3-2. Relative Epherneroptera abundance metric values in Maine.. Plot (A) shows
distributions of relative Epherneroptera abundance metric values across classifications (A,
B, C, NA) and (B) shows trends in relative Epherneroptera abundance and PRISM climatic
variables over time at Maine site 56817 (Sheepscot). Data used in these analyses were
limited to summer (July-September) rock-basket samples.
The Maine classification procedure also uses EPT taxa richness as a metric (as well as
EPT richness divided by Diptera richness), which is also used in the MMIs of other states. At site
56817 (Sheepscot), EPT richness has increased over time, as has the number of warm-water EPT
taxa (Figure 3-3a). When the first five years of data are compared to the last, results show that
the number of EPT taxa has increased by approximately 6 taxa over time. This difference is
much greater than the average difference in EPT taxa richness between Class A and B stations,
December 23, 2010
3-6
External Review Draft

-------
149
150
151
152
153
154
155
156
157
158
159
160
161
162
which is approximately 2-3 taxa (Figure 3-3b). Based on this, station quality and ranking could
improve due to warming temperatures that are projected to occur with climate change.

45

40

35
1
30
XT
o
25
E

o
'C
20
8

5
15
t

in
10

8

0

-5
NA
Class
a Median
~ 25%.75%
" X Non-Outlier Range
o Outliers
* Extremes
* * • PRISM mean annual air
temperature (*C)
-- PRISM mean annua!
precipitation (inches)
— EPT generic richness
70
60
50
40
30
20
10
0
1985 1988 1991 1994 1997 2000 2003 2006
Figure 3-3. EPT generic richness metric values in Maine. Plot (A) shows distributions of
EPT generic richness metric values across classifications (A, B, C, NA) and (B) shows
trends in EPT generic richness and PRISM climatic variables over time at Maine site 56817
(Sheepscot). Data used in these analyses were limited to summer (July-September) rock-
basket samples.
There are many more plecopteran (stonefly) taxa on the cold- than warm-water-
preference list. Three of Maine's discriminant model input metrics involve plecopterans:
Plecoptera abundance, Perlidae abundance and relative Plecoptera richness. For each metric,
highest abundances or richness values occur at Class A sites, and ratings decrease as plecopteran
abundance or richness values decrease. Although cold-water taxa like plecopterans should be
December 23, 2010
3-7
External Review Draft

-------
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
sensitive to climate change effects, the Maine data did not reveal any trends. One explanation is
that the longest-term reference station is a low elevation site where the benthic community is
predominantly warm water tolerant. Also, PRISM mean annual air temperature trends at this site
were small and variable over time, which suggests that water temperatures may not have
exceeded the thermal tolerance limits of the cold-water-preference taxa at this site. For reference
locations grouped in the Northeast Highlands, where cold-water taxa composed a much greater
proportion of the community, the duration of data records was too limited to define significant
trends.
Two other model input metrics are related to trichopterans: Hydropsyche abundance and
Cheumatopsyche abundance. These trichopteran metrics are not currently viewed as particularly
responsive to changes in temperature, because neither taxon is on the cold- or warm-water-
preference lists. These taxa are likely to be more resilient to climate change effects. However,
model metrics related to dipterans (true flies) may be vulnerable, as there are a large number of
dipterans on both the cold- and warm-water-preference lists. Seven of the cold-water taxa are
dipterans from the family Chironomidae (non-biting midges), and ten of the warm-water taxa are
dipterans (Tables 3-1 and 3-2). Although several dipteran genera are cold-preference taxa, in the
current Maine classification model a greater abundance or richness of dipterans tend to cause a
station to receive a lower rating. As the cold- and warm-water-preference components of the
dipterans are expected to respond differently to climate changes, the effects on outcomes of the
Maine model are likely to be variable and somewhat unpredictable. Depending on whether there
is any replacement of cold-water with warm-water dipteran taxa, increasing temperatures may
not change the associated metric values much.
The HBI is also a component of the Maine discriminant model, and is also used in other
northeastern states. Most of the Maine cold-water-preference taxa have low (< 3) HBI tolerance
values (to organic pollution) (Figure 3-4). Exceptions include two chironomids, Larsia and
Natarsia. There is a mix of tolerance values among the Maine warm-water taxa (Figure 3-4).
This results in a significant but weak correlation between temperature optima values and
pollution tolerance (r=0.29, p=001). The HBI metric is therefore also vulnerable to increases in
water temperature; any responses that involve decreases in cold-water taxa with low HBI
tolerance values or replacement by warm-water taxa with higher tolerance values could cause an
increase in the HBI metric. Since higher HBI values impart a more impaired station rating, there
December 23, 2010
3-8
External Review Draft

-------
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
would be a concomitant decrease in the station rating. Given the mixed relationship in Maine
between warm-water-preference taxa and HBI tolerances, there should be regional (spatial)
variability in HBI vulnerability, related to spatial differences in community composition of
warm-water-preference taxa.

18

16

14
1
12
i
10


£
8
E

3
6
z


4

2

0
a cold
¦ warm
Intolerant	Intermediate	"iuiwaiii
Enrichment Tolerance
Figure 3-4. Relationship between Maine cold and warm-water-preference taxa and Maine
enrichment tolerance scores. Taxa with enrichment tolerance scores of 0-3 were
categorized as Intolerant, those with scores of 4-6 were Intermediate and those with scores
of 7-10 were Tolerant.
Additional vulnerabilities of station quality classification are illustrated in aspects of the
BCG (Gerritsen and Craig, 2008) as applied in New England (USEPA, 2007), even though the
BCG is not a component of the Maine discriminant model classification scheme, or of the MMIs
applied in other northeastern states. The BCG provides a more refined and explicit approach for
defining and classifying condition, and includes five BCG attribute levels in New England:
2=highly sensitive taxa, 3=intermediate sensitive taxa, 4=taxa of intermediate tolerance,
5=tolerant taxa, 6=non-native or intentionally introduced taxa. Stations with communities
composed of more sensitive taxa (2 or 3) generally receive better BCG-level assignments, while
stations that have more tolerant taxa (5 or 6) are more likely to be classified in lower BCG levels.
Twenty of the Maine cold-water-preference taxa are considered to be sensitive taxa (2 or 3), and
two are considered to be tolerant (5) (Appendix E). Ten of the warm-water-preference taxa are
considered to be tolerant (5 or 6) and 7 are considered to be sensitive (2 or 3) (Appendix E). If
sensitive cold-water taxa are replaced by warm-water taxa that have higher BCG attribute
December 23, 2010
3-9
External Review Draft

-------
218	assignments, then samples may receive lower tier assignments as temperatures increase. This can
219	alter the assessment and rating of condition of a location based on biological composition over
220	time.
221	Maine defines seven "Class A indicator taxa" to separate Class A and B sites. This study
222	defines two of the seven taxa, Eurylophella and Glossosoma, as cold-water-preference taxa, and
223	three, Paragnetina, Serratella and Leucrocuta, as warm-water-preference taxa (Table 3-3).
224	Brachycentrus was initially classified as a warm-water-preference taxon. However, variation in
225	temperature preferences among species within this genus cause this designation to be dropped,
226	even though weighted average modeling (Stamp et al., 2010; USEPA, 2011) shows that
227	Brachycentrus tends to occur more at warmer sites. The fairly even split between temperature
228	preferences among Maine's Class A indicator taxa suggests that increasing temperatures may
229	have contradictory effects on components of this metric, and lead to variable results.
230
231	Table 3-3. Temperature trait information for Class A Indicator taxa. Temperature optima
232	(°C) and tolerance values are based on instantaneous water-temperature measurements
233	and occurrences of organisms. The values were derived from weighted average modeling,
234	using the guidelines of Yuan (2006). The rankings (Temp RankOpt = optima ranking;
235	Temp Rank_Tol=tolerance ranking) range from 1 to 7 and are based on percentiles within
236	each data set.
Class A
Indicator
Temp
Indicator
Temp
Optima
Temp
Tol
Temp
RankOpt
Temp
Rank Tol
Comments
Taxa






Eurylophella
cold
17.4
3.2
2
4

Glossosoma
cold
18.7
4.8
3
7







Occurred at one of






the warm water
Psilotreta

18.8
3.0
3
4
case study sites,
otherwise would
have been on the
cold water list.
Paragnetina
warm
20.7
3.6
5
6

Serratella
warm
20.8
3.8
5
6

Leucrocuta
warm
21.2
3.3
6
5







VT gave this a 'no'






for warm-water-
Brachycentrus

21.5
3.4
6
5
preference taxa due
to variation among
species within this
genus
December 23, 2010
3-10
External Review Draft

-------
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
At Station 56817 (Sheepscot), the Class A indicator taxa metric was significantly
correlated with several precipitation metrics and was higher in wet years compared to dry or
normal years (see Section 2 and Appendix E), showing that potential future changes in
precipitation may influence the Maine Class A Indicator metric at this and comparable locations.
As shown with respect to temperature-driven responses, the projection for slight increases in
precipitation in the northeast raises the possibility that Class A indicator taxa may fare better in
the future and contribute to better station ratings due to climate change, independent of any
actual change in environmental quality. Projected changes in precipitation are small, and
seasonal patterns of precipitation (e.g., projected increases during winter and spring with
possible decreases in summer) must be considered in concert with the season during which
biomonitoring occurs. Therefore, the magnitude of vulnerability of this metric to changes in
precipitation, and through this metric to station ratings, is probably small in the short term.
3.2. NORTH CAROLINA AND THE SOUTHEAST
North Carolina classifies sites as Excellent (5), Good (4), Good/Fair (3), Fair (2) or Poor
(1) using EPT richness and the North Carolina Biotic Index (NCBI). Different scoring criteria
are used for each major ecoregion (Mountain, Piedmont, Coastal). Details of each of these two
indices and how they are combined for final scoring can be found in Appendix G.
Several analytical approaches contribute information to the potential vulnerabilities of the
North Carolina MMI (the EPT richness metric and the NCBI) and bioclassification procedures.
In one scenario we removed all cold-water-preference taxa from three reference Mountain sites
(on average the Mountain sites have more cold-water taxa, see Appendix G). NCBI, EPT
richness and bioclassification scores were recalculated to evaluate effects on site scores. In a
second scenario we replaced taxa that typically inhabit Mountain sites with assemblages more
typical of the Piedmont ecoregion. This was accomplished by applying Mountain scoring criteria
to data from two Piedmont reference sites and evaluating by how much the scores changed (see
Appendix C3 for site descriptions).
We also explored relationships between temperature preference taxa, pollution tolerance
values, the biotic index, and climate-related variables. Applicable correlation analyses include:
1. temperature optima values vs. tolerance values;
December 23, 2010
3-11
External Review Draft

-------
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
2.	temperature indicator metrics at selected Mountain and Piedmont reference sites
(percent cold- and warm-water-preference individuals and number of cold- and
warm-water-preference taxa) vs. BI scores; and
3.	BI values vs. PRISM mean annual air temperature and precipitation.
The correlation analyses were performed on datasets that used genus-level tolerance
values. Tolerance values can vary within some genera, and therefore these BI scores may vary
somewhat from NCBI scores (though they are generally close).
We performed additional analyses relevant to other southeastern states that may not use
EPT taxa richness or the NCBI. Metrics commonly used in other southeastern states include total
taxa, EPT taxa, Ephemeroptera taxa, Plecoptera taxa, Trichoptera taxa, HBI, an assortment of
functional feeding group and habit metrics, and percent dominant taxon. Other metrics evaluated
include: metrics reflecting temperature preferences/ tolerances (developed using maximum
likelihood modeling on the NC dataset); trait metrics that reflect sensitivity to changes in
hydrologic regime; and metrics that incorporate combinations of traits that are most likely to be
favorable or unfavorable if changes to climate occur as projected by NCAR models (warmer
with a very slight increase in precipitation in North Carolina). Appendix G and Stamp et al.
(2010) contain the full list of metrics evaluated, of cold- and warm-water-preference taxa, and
calculated temperature-tolerance values.
3.2.1. Vulnerability of the North Carolina Bioclassification - Simulation of taxa
replacement
Many predictions and observations of biological trends in response to climate change
include shifts in ranges of sensitive taxa, often involving movements north and/or higher in
elevation, such that northerly or higher elevation communities tend to become more similar to
lower elevation or more southerly neighbors (e.g., Bonada et al., 2007b). Replacing higher
elevation, more cold-sensitive North Carolina Mountain communities with lower elevation, more
warm-tolerant Piedmont taxa, is a reasonable approximation of this type of biological response.
This sets a boundary on the range of vulnerability of North Carolina's bioclassification indices.
The first scenario tests whether the Mountain biotic index as currently formulated will still
accurately classify Mountain benthic communities that in the future may become increasingly
like Piedmont benthos in composition. The results, however, show that at the most extreme (i.e.
December 23, 2010
3-12
External Review Draft

-------
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
at the point of complete community replacement) classifications would decrease by one level
(Figure 3-5).
c
o
15
¦ Mountain/Mountain
~ Piedmont/Piedmont
E3 Piedmont/Mountain
o
1980-1989 1990-1999 2000 2009
Year
Figure 3-5. Site-condition classification scores at three reference Mountain sites (Station
NC0109 (New), Station NC0207 (Nantahala) and Station NC0209 (Cataloochee)) and two
reference Piedmont sites (Station NC0075 (Little River) and Station NC0248 (Barnes
Creek)) averaged across three 10-year periods. The black bars represent average scores at
Mountain sites when Mountain criteria are applied; the white bars represent average
scores at Piedmont sites when Piedmont criteria are applied; the gray bars represent
average scores at Piedmont sites when Mountain criteria are applied.
This second scenario is an upper bound of index vulnerability. It is unlikely that complete
community replacement will occur, and certainly not in the near term. Current biological trends
in the NC data are fairly weak and/or spatially inconsistent (see Section 2 and Appendix G).
This, in part, reflects the relative paucity of long-term data adequate to define climate-related
trends given high natural variability. Still, some sensitive taxa (e.g., EPT taxa) and trait groups
(e.g., cold-sensitive taxa) do exhibit trends in North Carolina, especially in response to
precipitation and in the Mountain region. Therefore, the nature of this vulnerability is valid,
although the actual magnitude of vulnerability is probably modest, especially in the near term.
Table 3-4. Final bioclassification scores at 3 reference Mountain sites (NC0109 - New
River, NC0209- Cataloochee and NC0207/2554 - Nantahala) before and after all cold-
water-preference taxa are dropped from the sites
Site
Year
Before
After
Difference

December 23, 2010	3-13	External Review Draft

-------


# cold-water-
Final
# cold-water-
Final
Final


preference Taxa
Score
preference Taxa
Score
Scores
NC0109
1983
6
5
0
4
1
NC0109
1984
5
5
0
4
1
NC0109
1985
4
4
0
4
0
NC0109
1986
3
4
0
4
0
NC0109
1987
4
4
0
4
0
NC0109
1988
3
4
0
4
0
NC0109
1989
6
4
0
4
0
NC0109
1990
4
4
0
4
0
NC0109
1993
4
5
0
4
1
NC0109
1998
5
4
0
4
0
NC0109
2003
7
5
0
5
0
Site
Year
Before
After
Difference


# cold-water-
Final
# cold-water-
Final
Final


preference Taxa
Score
preference Taxa
Score
Scores
NC0209
1984
18
5
0
4
1
NC0209
1986
19
5
0
4
1
NC0209
1989
20
5
0
5
0
NC0209
1990
19
5
0
4
1
NC0209
1991
21
5
0
4
1
NC0209
1992
18
5
0
4
1
NC0209
1997
23
5
0
5
0
Site
Year
Before
After
Difference


# cold-water-
Final
# cold-water-
Final
Final


preference Taxa
Score
preference Taxa
Score
Scores
NC0207
1984
14
5
0
4
1
NC0207
1986
15
5
0
4
1
NC0207
1988
17
5
0
5
0
NC0207
1990
17
5
0
5
0
NC0207
1991
20
5
0
5
0
NC0207
1994
19
5
0
4
1
NC0207
1999
17
5
0
4
1
NC2554
2004
19
5
0
4
1
325
326	A similar response of the bioclassification results is observed if the cold-water-preference
327	taxa are eliminated from the biotic assemblages at three references sites in the Mountain
328	ecoregion. The maximum drop in station classification score is one bioclassification level (from
329	Excellent to Good); this occurred for 3 of the 11 years at Site NC0109, 5 of the 7 years at Site
330	NC0209, and 5 of the 8 years at Site NC0207/NC2554 (Table 3-4).
December 23, 2010
3-14
External Review Draft

-------
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
3.2.2. EPT Taxa Richness Metric
EPT richness is one of the two components of the North Carolina bioclassification
scheme. EPT metrics also are used in other southeastern states. EPT metrics appear to be
particularly vulnerable because they include many cold-water taxa. In North Carolina, 20 of the
31 cold-water-preference taxa (genus-level OTUs) are EPT taxa (Table 3-5). There are
substantially fewer (5) EPT taxa on the warm-water-preference list (Table 3-6). Within the EPT
genera on the cold-water-preference list, there are 53 species that could potentially be counted
towards the EPT richness metric used in the bioclassification of sites in North Carolina, while
only 5 species could be potentially counted from the warm-water-preference list.
Losses of cold-water-preference taxa and/or replacement by warmer water taxa in
response to increasing temperatures may include loss of EPT taxa, potentially lowering
bioclassification scores. At high quality sites, a loss of 3 (Coastal sites) or 4 (Mountain or
Piedmont sites) EPT species would lower the EPT richness score by a full level, from a 5
(Excellent) to a 4 (Good) (see Appendix G). A greater loss, of 10 EPT taxa at Mountain sites, 8
taxa at Piedmont sites, or 7 at Coastal sites, would be needed to decrease scores by one level at
sites of lesser condition (currently rated Good (4) or lower).
Table 3-5. Number of North Carolina cold-water-preference taxa in each order. EPT
orders are italicized
Order	Total
Diptera	10
Plecoptera	8
Ephemeroptera	6
Trichoptera	6
Coleoptera	1
Odonata	1
The greatest effect of removing cold-water-preference EPT taxa from benthic
assemblages was observed for three Mountain ecoregion reference stations, because cold-water-
preference taxa comprise the greatest percentage of the benthic communities at higher elevations
In these cases, removal of cold-water taxa resulted in the loss of 9 to 14 EPT taxa, and decreases
in EPT richness scores ranging from 0.4 to 1.2 (see Table 3-7 and Appendix G). The third Blue
December 23, 2010
3-15
External Review Draft

-------
357	Ridge reference site (NC0109) has fewer cold-water taxa in the assemblage, and the removal of
358	cold-water taxa resulted in a loss of 4 species, and a decrease in EPT richness score of 0.6.
359
360	Table 3-6. Number of North Carolina warm-water-preference taxa in each order. EPT
361	orders are italicized
Order
Total
Odonata
7
Diptera
5
Trichoptera
4
Coleoptera
2
Rhynchobdellida
2
Arhynchobdellida
1
Basommatophora
1
Decapoda
1
Ephemeroptera
1
Hemiptera
1
Isopoda
1
Unionoida
1
362
363
364	Table 3-7. EPT species richness values (EPTS) and scores at 3 reference
365	Mountain sites (NC0109 - New River, NC0209- Cataloochee and
366	NC0207/2554 - Nantahala) before and after all cold-water-preference taxa
367	are dropped from the sites
Site
Year
Before
After
Difference


EPT
S EPT S
EPT S
EPT S
EPT S
EPT S



Score

Score

Score
NC0109
1983
50
5
47
5
3
0
NC0109
1984
45
5
42
4.6
3
-0.4
NC0109
1985
45
5
44
5
1
0
NC0109
1986
43
4.6
41
4.4
2
-0.2
NC0109
1987
41
4.4
38
4
3
-0.4
NC0109
1988
42
4.6
40
4.4
2
-0.2
NC0109
1989
43
4.6
39
4
4
-0.6
NC0109
1990
49
5
46
5
3
0
NC0109
1993
47
5
46
5
1
0
NC0109
1998
37
4
34
4
3
0
NC0109
2003
51
5
47
5
4
0
Site
Year
Before
After
Difference
December 23, 2010
3-16
External Review Draft

-------
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383


EPTS
EPTS
Score
EPTS
EPTS
Score
EPTS
EPTS
Score
NC0209
1984
42
4.6
32
3.6
10
1
NC0209
1986
47
5
35
4
12
1
NC0209
1989
53
5
42
4.6
11
0.4
NC0209
1990
51
5
39
4
12
1
NC0209
1991
48
5
34
4
14
1
NC0209
1992
42
4.6
31
3.4
11
1.2
NC0209
1997
50
5
37
4
13
1
Site
Year
Before
After
Difference


EPT
S EPT S
EPT S
EPT S
EPT S
EPT S



Score

Score

Score
NC0207
1984
45
5
36
4
9
1
NC0207
1986
48
5
40
4.4
8
0.6
NC0207
1988
49
5
38
4
11
1
NC0207
1990
53
5
43
4.6
10
0.4
NC0207
1991
54
5
41
4.4
13
0.6
NC0207
1994
48
5
36
4
12
1
NC0207
1999
49
5
39
4
10
1
NC2554
2004
49
5
37
4
12
1
3.2.3. The North Carolina Biotic Index (NCBI)
The second component of the North Carolina bioclassification scheme is the NCBI. This
is North Carolina's version of the HBI, which is commonly used in site assessments in other
states. The HBI documents the contribution of pollution tolerant taxa to the composition of the
community (Hillsenhoff, 1987). Taxa are assigned pollution tolerance values ranging from 1
(most sensitive) to 10 (most tolerant). The higher the HBI, the more strongly the community is
dominated by taxa tolerant of organic pollution, and the more impaired the site is considered.
Vulnerability of the NCBI (and HBI) is largely due to the high association of cold-water
taxa with low tolerance to organic pollution (Figure 3-6). Taxa that show preferences for lower
temperatures tend to have lower tolerance values and those that tend to occur more in warmer
water habitats tend to have higher tolerance values. Most (22 of the 30) of the North Carolina
cold-water-preference taxa for which the tolerance value is known have low tolerance values (<
3) (Figure 3-6). Only one of the cold-water-preference taxa (the chironomid Diamesa) has a
tolerance value > 7. In contrast, most of the warm-water taxa have higher tolerance values.
December 23, 2010
3-17
External Review Draft

-------
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
Twelve of the warm-water-preference taxa that have been assigned tolerance values have
tolerance values > 7. Only one of the warm-water taxa, Chimarra, has a tolerance value < 3.
Based on this information alone, it is likely that a loss of cold water taxa and an increase in
warmer water taxa would result in higher BI scores, which would contribute to lower
bioclassification scores. An increase in BI scores of 0.1 can lower the classification of an
Excellent site a full level from 5 to 4. At lower quality sites (those rated Good (4) or lower), it
would take a greater increase in BI scores (by at least 0.6) to lower bioclassification levels a full
level (i.e. go from a classification of 4 to 3, 3 to 2, or 2 to 1).

30

25
s

3
20
¦5

s
15
E

3
10
z

5

0
~ Cold
¦ Warm
Intolerant	intermediate
Enrichment tolerance
Tolerant
Figure 3-6. Relationship between North Carolina cold- and warm-water-preference taxa
and North Carolina enrichment tolerance scores. Taxa with enrichment tolerance scores of
0-3 were categorized as Intolerant, those with scores of 4-6 were Intermediate and those
with scores of 7-10 were Tolerant.
Again, three Mountain ecoregion reference sites with the greatest percentage composition
of cold-water taxa were most vulnerable in terms of NCBI scores. In these cases, removal of
cold-water taxa resulted in an increase in BI values ranging from 0.45 to 0.86 and decreases in
NCBI scores ranging from 0 to 1 (Table 3-8 and Appendix G). At the Blue Ridge reference site
(NC0109) with fewer cold-water-preference taxa, the loss of cold-preference taxa resulted in a
maximum increase in NCBI value of 0.24, maximum decrease in NCBI score of 0.2.
December 23, 2010
3-18
External Review Draft

-------
406	Table 3-8. NCBI values and scores at 3 reference Mountain
407
408
409
Site
Year
Before
After
Difference
BI
BI Score
BI
BI Score

BI
BI Score
NC0209
1984
3.32
5
3.90
5
+
0.58
0
NC0209
1986
3.46
5
4.29
4
+
0.83
-1
NC0209
1989
2.98
5
3.68
5
+
0.70
0
NC0209
1990
3.12
5
3.88
5
+
0.76
0
NC0209
1991
2.67
5
3.51
5
+
0.84
0
NC0209
1992
3.00
5
3.86
5
+
0.86
0
NC0209
1997
2.69
5
3.29
5
+
0.60
0
Site
Year
Before
After
Difference
BI
BI Score
BI
BI Score

BI
BI Score
NC0207
1984
3.77
5
4.43
4
+
0.66
-1
NC0207
1986
3.61
5
4.15
4
+
0.54
-1
NC0207
1988
3.41
5
3.89
5
+
0.48
0
NC0207
1990
3.00
5
3.47
5
+
0.47
0
NC0207
1991
2.39
5
3.04
5
+
0.65
0
NC0207
1994
2.60
5
3.13
5
+
0.53
0
NC0207
1999
3.38
5
3.83
5
+
0.45
0
NC2554
2004
3.19
5
3.79
5
+
0.60
0
410
411	Further evidence of the potential vulnerability of the NCBI to climate change effects is
412	that relative abundance of cold-water-preference taxa was significantly negatively correlated
413	with NCBI values at all 3 Mountain sites (NC0109 r2=0.46, p= 021; NC0207 r2=0.66, p= 007;
sites (NC0109 - New River, NC0209- Cataloochee and
NC0207/2554 - Nantahala) before and after all cold-water-
preference taxa are dropped from the sites	
Site
Year
Before
After
Difference
BI
BI Score
BI
BI Score

BI
BI Score
NC0109
1983
4.60
4
4.67
4
+
0.07
0
NC0109
1984
4.33
4
4.44
4
+
0.11
0
NC0109
1985
5.48
3
5.51
3
+
0.03
0
NC0109
1986
5.43
3
5.55
3
+
0.12
0
NC0109
1987
4.87
3.6
4.93
3.4
+
0.06
-0.2
NC0109
1988
5.37
3
5.46
3
+
0.09
0
NC0109
1989
4.21
4
4.28
4
+
0.07
0
NC0109
1990
4.87
3.6
4.91
3.4
+
0.04
-0.2
NC0109
1993
4.70
4
4.74
4
+
0.04
0
NC0109
1998
4.40
4
4.49
4
+
0.09
0
NC0109
2003
3.61
5
3.85
5
+
0.24
0
December 23, 2010
3-19
External Review Draft

-------
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
NC0209 r2=.85, p=.004) (Table 3-9 and Appendix G). The abundances of cold-water-preference
taxa were lower at sites that had higher NCBI scores. At one of the Mountain sites, the warm-
water metrics also were positively correlated with NCBI values. Replacement of colder water
preference taxa with warmer preference taxa would likely contribute to a site receiving a higher
NCBI score and therefore a poorer rating; this will most likely affect sites in the Mountain
ecoregion.
Table 3-9. Correlations of benthic taxa grouped by temperature traits with
BI at North Carolina Mountain and Piedmont reference stations. Significant
correlations in bold text.
Temperature Metric
Mountain
Piedmont
NC0109 NC0207 NC0209
NC007 NC024
5 8
Cold-water-preference taxa -
relative abundance
-0.68 -0.81 -0.92
N=ll N=9 N=7
p=. 021 p=.007 p=. 004
-0.25 -0.11
N=7 N=7
p= 587 p=.821
Warm-water-preference taxa -
relative abundance
0.66 0.12 0.63
N=ll N=9 N=7
p=.026 p=766 p=.127
-0.16 0.19
N=7 N=7
p=726 p=679
Cold-water-preference Taxa -
richness
-0.81 -0.46 -0.57
N=ll N=9 N=7
p=.003 p=.208 p=.182
-0.37 0.17
N=7 N=7
p=.416 p=.708
Warm-water-preference Taxa -
richness
0.77 0.17 0.65
N=ll N=9 N=7
p=.006 p=.664 p=. 116
-0.01 -0.51
N=7 N=7
p=.991 p=239
3.3. OHIO AND THE MIDWEST
Ohio also uses MMIs, the Index of Biotic Integrity (IBI) for fish communities, and the
Invertebrate Community Index (ICI) for macroinvertebrates. Evaluations are separated by stream
size categories and by level 3 ecoregions. Evaluations for Ohio were integrated with analyses of
reference location re-sampling conducted to determine whether biological reference condition
has changed since 1980 and as a foundation for the recalibration of Ohio biocriteria (Rankin
2008). The examination of trends in biological condition at reference sites and the exploration of
potential causes was an essential component of this effort. Although climate change effects may
be a contributing component to observed trends, there is evidence that other environmental
changes may be responsible.
December 23, 2010
3-20
External Review Draft

-------
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
Based on approximately 30 years of watershed assessments in Ohio, there have been a
variety of environmental changes identified that are associated with shifts in biological condition
at assemblage and taxon levels (Rankin, 2008; Yoder et al., 2005). Environmental factors that
have been identified as main contributors to these changes include reduction in point source
loadings; changes in land uses (e.g., increased urbanization); altered pollutant loadings from
agricultural lands (e.g., reductions in sediments and nutrients in response to increased
conservation tillage); loss of habitat quality due to agricultural drainage practices and
suburbanization; and localized improvement in habitat quality due to stream restoration. These
environmental changes make it difficult to detect responses to climate-related changes in
temperature and/or hydrology. The lack of readily available long-term data for temperature, flow
and biology needed to define and separate such effects compounds the problem.
To examine long-term trends, the Ohio IBI and ICI were recalculated based on data from
early and late sampling cycles, with an average of 14-16 years between data sets (Appendix H).
This analysis corrected for any changes in taxonomic resolution over time. Values of the ICIs
and IBIs for the most recent time period for each stream size and ecoregion category were almost
always higher than or similar to the original values (i.e., the direction of change was either
positive or neutral). This shows a strong pattern of environmental improvement reflected in the
condition of biological communities. Although the overall pattern is compelling, none of the
MMI differences are outside the range of natural variation for each index (Appendix H).
These effects, shown in changes over time in the Ohio MMIs, reflect, in some part,
reductions in pollutant loadings and habitat degradation (Rankin, 2008). They could incorporate
a climate change signal that is confounded or swamped by these apparent responses to improved
environmental management. Given the probable confounding factors, two possibilities exist. One
is that the actual improvement in environmental condition do to better management practices is
greater than that reflected in the magnitude of MMI improvement. The implied suppression of
the MMI response could result from climate change-related reductions in cold-water-preference
taxa that are also pollution sensitive, and/or from increases in warm-water taxa that are pollution
tolerant. This possibility would have the effect of reducing the MMI, leading to an underestimate
of the magnitude of improvement. Other scenarios also are possible. For example, climate-
related changes in precipitation and flow could have increased cold-sensitive taxa, as has been
observed in other states (Section 2). Cold-sensitive taxa are often pollution sensitive (earlier
December 23, 2010
3-21
External Review Draft

-------
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
sections of this Chapter); their increases would appear as an improvement in the MMI, leading to
an over-estimate of improvements attributable to management practices. Some preliminary
evidence of range extensions of flow-sensitive taxa into headwater streams in Ohio is presented
in Appendix H. Thus it is possible that climate change has augmented, or contributed to the
apparent improvements in environmental quality reflected in ICI and IBI improvements.
The direction of climate-related changes in BI scores could be positive or negative. The
expectation for the more likely direction is, in part, informed by apparent relationships between
temperature and/or hydrologic sensitivities of the Ohio fish and macroinvertebrate taxa and their
associated pollution tolerance values. Figures 3-7 and 3-8 show a general concordance between
these two taxon traits. With regard to macroinvertebrate temperature preferences, Figure 3-7
(upper and lower left graphs) shows that for both stream sizes plotted, many, though not all, of
the pollution tolerant taxa (shown with red dots) have higher temperature preferences. The
lowest temperature preferences are exhibited by taxa with "moderately intolerant" to "intolerant"
(i.e., sensitive) pollution designations. These figures also show substantial variation. A few
pollution-tolerant taxa exhibit relatively low temperature preferences, and there is a broad span
of temperature preferences exhibited by taxa with moderate pollution sensitivity. A slightly
clearer association is seen for hydrologic preferences and pollution tolerance, especially for fish
(Figure 3-8). In this case, fish with the greatest pollution tolerance could tolerate lower flows,
while the most pollution-sensitive fish had preference for higher flows.
In Ohio and other central states, climate change projections are for warmer temperatures
and slight increases in precipitation (Appendix H). There is an associated expectation for
increasing stream temperatures, although the expectation for changes in flow are more uncertain,
being affected by both increasing precipitation, which may increase flows, and increasing
temperatures, which can also increase evapotranspiration and contribute to decreasing flows at
least seasonally. Community composition also will contribute to determining how climate
change effects on component taxa will be reflected in the MMIs. For example, in the
macroinvertebrate community, the balance in composition between cold- and warm-water-
preference taxa will influence net response, as has been illustrated for other states. Results
suggest that because of the general concordance between temperature and/or hydrologic
sensitivity and pollution-tolerance sensitivity, it is plausible to expect the loss of sensitive taxa
due to climate change (Appendix H). This may occur through replacements by or increases in
December 23, 2010
3-22
External Review Draft

-------
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
occurrence and abundance of more tolerant taxa, with an associated apparent increase in the
calculated pollution tolerance of the community. Decreasing ICI or IBI scores would make
station conditions look more impaired, due only to climate change, or would mask detection of a
portion of environmental improvement.
Because most, if not all, of Ohio's reference stations are considered "best available"
(sensu Stoddard et al., 2006), conditions at reference locations in Ohio are changing, and mostly
improving, in response to management and pollution control efforts. The detection of
improvement is occurring in spite of potential climate change impacts. The ability to partition
these responses is hampered by the lack of reference locations unaffected by pollution or land
alterations. In addition, most stations are sampled on a regionally rotating basis, so that even over
two or more decades of sampling, many locations have relatively few data points to support
definition of trends. Without "natural" reference sites as an anchor, the definition of a gradient of
site conditions using the BCG (Davies and Jackson, 2006) would provide a basis for selecting
and sampling stations along a gradient of effects, which serve as an alternative approach for
separating climate change from more conventional pollution responses.
Stressor identification and related processes contribute to elucidating causes of
impairment through analysis of MMI and component responses. These same techniques are
useful to identify sources of improvement in environmental condition. This does not by itself
offer a mechanism for partitioning climate change responses from other causes, but would be
valuable in tandem with gradient sampling along a BCG to support such an effort.
December 23, 2010
3-23
External Review Draft

-------

25
- 20
WSVs for Maximum Temperature
Headwater Streams
Planorbella (Pfer'psoma)
pilsbryi \
Chironomus (C.)'sp 1
Baetis
tricaudatus
15
Acer pernio
macdunnough
H&yesomyia senate
	£ l
s( ~—f-— HydroptHa sp.
1—- Cricotopus (Isoc/a
i sytvestris grot
iius)
iP
N > 30 Stations
Unionids>.5 Stations
Very Tolerant
Tolerant
Moderately Tolerant
Intermediate
Moderately Intolerant
Intolerant
20
25
30
Mean Maximum Temperature (C)
30
2>
^ - oo
> U 28
Q ^
V S. 26
l a
=» t_
+- _Q
a u 24
Q- CO
£ >
5 E 22
h- p
I*
20
WSVs Temperature
Headwater Macroinvertebrate Data
.... 1 ... 1 1 .. 1 .
' T '
. . . , | . . . .
° i


; o
¦
: ^ ! i
O ;

Very Tolerant Moderately Inter- Moderately Intolerant
Tolerant	Tolerant mediate Intolerant
ICI Taxa Tolerance Categories
o)
WSVs for Maximum Temperature
Wadeable Streams

25
£
£
ZJ
E
x
o
$
- 20
* \Abtabesmyia annu/ata'
„ ! — Tanypus neopunctipennis
"" ^	Parachironomus "hirtalatus"
Polypedih
(Pen taped it urn)
5 Stations
A therix lantha
Phuditus punctivertris
Paragnetina media
Acerpenna macdunnough/) ^
Para iany tarsus r.sp 1
i—L-La—i—L_i—i—i
Very Tolerant
Tolerant
Moderately Tolerant
Intermediate
Moderately Intolerant
Intolerant
15
20
25
Mean Maximum Temperature (C)
30
•> V)
(r) ^ 28
u co
£ to
G ^ 26
u +-
s_ a
^ ^ 24
a co
»- +-
Q_ "
J .1
h o 20
xf u
| I 18
WSVs Temperature
Wadeable Macroinvertebrate bat a
¦


u
1

Very Tolerant Moderately Inter- Moderately Intolerant
Tolerant	Tolerant mediate Intolerant
ICI Taxa Tolerance Categories
518
519	Figure 3-7. Plots of macroinvertebrate taxa maximum temperature Weighted Stressor
520	Values (WSVs) vs mean maximum values for taxa for headwater streams (upper left) and
521	wadeable streams (lower left), and box and whisker plots of maximum temperature by
522	Ohio EPA macroinvertebrate tolerance values (derived for the ICI) for headwater streams
523	(upper right) and wadeable streams (lower right). Data for taxa represent data from
524	artificial substrates where at least 5 samples were represented for each stream size
525	category.
December 23, 2010	3-24	External Review Draft

-------
526
527
528
529
530
531
532
533
534
535
536
537
538
539
WSVs for Hydro-QHEI
Headwater Streams
LL)
I
a
~o
I
c
a
"O
O)
st
i
¦ Chi marra obscura
Psychomyia ftavidi
-¦	#•—		-p Pipuditus-ek/bfus--
• Hydropsyche frisen
> .	\ J
V
N > 30 Stations
Unionids> 5 Stations
Very Tolerant
Tolerant
Moderately Tolerant
Intermediate
Moderately Intolerant
Intolerant
5	10	15
Mean Hydro-QHEI
20
24
V)
ft)
'o 20
^ ft)
Q_
-
I
WSVs Hydro-QHEI
Headwater Macroinvertebrate Data
o
I ' 'D' ¦ I I 'O' ' i ' I ' I I ' ' ¦ ' I
Intolerant Moderately Inter- Moderately Tolerant Very
Intolerant mediate Tolerant	Tolerant
24
LLI
X
a
o
a
0
>
1
c
o
(1)
s
12
Weighted Hydro-QHEI
Wadeble Streams (>20-200 sq mi)
Tolerant
Moderately Tolerant
Intermediate
Sensitive
Intolerant
N > 5 Stations
Tonguetied Minnow
~
v	_ .	X Brown Trou'.
ilippecance Darte? ~ _ _
!	+ Streamtane Chub
~ Lorgnose Dace
^ American Brook Lamprey
-,~<—Varsgate Darter
Eastern Sand Darter
X Fathead Minnow
BlaekstrspejTopminnow
25
>	20
>
H u ig
LD
p ™ 10
-o
I
WSVs Hydro-QHEI (Overall Score)
Wadeable Fish Data


1 1 1 1 1
' « 1 1
i ... .
i i . i



T
		T



	X	
r

+
O





1



i . . , ,
¦ ¦ ¦ ¦
Intolerant Sensitive IntermediMaderately ToleranTolerant
Mean HYDRO-QHEI
Figure 3-8. Scatter plots of taxa/species Hydro-QHEI (Qualitative Habitat Evaluation
Index (QHEI) based only on hydrologic variables) WSVs vs mean Hydro-QHEI values for
macroinvertebrate taxa for headwater streams (upper left) and wadeable streams (lower
left), and box and whiskers plots of macroinvertebrates (upper right) and fish (lower right)
WSVs for Hydro-QHEI for these waters. Data from Ohio EPA.
3.4. UTAH AND THE SOUTHWEST
Utah rates its sites with a RIVPACS model, in which data from reference sites are used to
establish expected (E) macroinvertebrate assemblages and to which observed (O) assemblages at
sites are compared (Appendix F). The ratio of these values (O/E) can be interpreted as a measure
of taxonomic completeness. Values of O/E near 1 (one) suggest that the site is comparable to
reference, whereas values that vary substantially from 1 suggest that the site is degraded (Yuan,
2006a).
December 23, 2010
3-25
External Review Draft

-------
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
3.4.1. Approach
Utah DEQ developed two different RIVPACS models: one for fall samples and the other
for all seasons. They currently use the fall model for bioassessments, consistent with a focus on
fall as their primary sampling period. The model has 15 predictor variables, 7 of which are
related to climate (e.g., temperature, precipitation, freeze dates). Two fundamentally different
approaches were used to evaluate possible vulnerabilities of RIVPACS assessments to climate
change responses. One approach manipulates climate-related predictor variables within the
model, within ranges informed by the magnitude of climate change projections for the southwest
in temperature and precipitation. Half of the predictor variables included in the Utah fall
RIVPACS model are climate related, showing that some climate factors expected to change in
the future are important in controlling stream macroinvertebrate community composition across
regions in Utah. Alteration of these variables at existing reference locations is intended to
illustrate the range of model responses that might be expected over time due only to climate
change, and thus be a measure of vulnerability of model-based decisions. A related analysis
involved running the Utah fall RIVPACS model using only the climate-related predictor
variables, assuming this would maximize their influence on definition of the expected
community and thus, if possible, isolate components of the community most sensitive to climate
variables. Details of the model runs are summarized in Appendix F.
The other analysis used extremes in existing data as proxies for future climate conditions,
by partitioning data at long-term reference stations into years characterized by hottest ( >75th
percentile of the long-term temperature distribution), coldest (<25th percentile of temperature),
and normal (25th to 75th percentile) average annual temperatures. Using similar thresholds, years
were partitioned based on average annual precipitation into wettest, driest, and normal years
(Appendix D). Examination of RIVPACS model responses between year groups was used as an
indication of the direction and magnitude of responses in the RIVPACS O/E outcomes that might
result from climate change. An assumption is that these temperature and precipitation differences
drive responses in benthic communities that are reasonable proxies for the types of community
changes that can be expected over the long term with climate change.
December 23, 2010
3-26
External Review Draft

-------
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
3.4.2. RIVPACS Responses - Utah Decision Vulnerabilities
Comparison of RIVPACS model outputs among hotter, colder and normal years provides
evidence of potential vulnerabilities to climate change. Figure 3-9 shows results from two
reference sites in the Colorado Plateaus ecoregion (site 4951200 (Virgin) and site 4936750
(Duchesne)) where mean O/E values from hottest year samples were significantly higher than
mean O/E scores from coldest and normal year samples. These differences are in the range of
differences relevant to the Utah DEQ decision matrix for determining whether a test location
should be characterized as not supporting beneficial uses (i.e., classified as impaired). In this
matrix, an O/E score <0.74 represents the first threshold of impairment, with another at O/E
<0.54 (Appendix Attachment F6). The magnitude of average annual temperature differences
between the "hottest" and "coldest" year samples is about 2 °C, comparable to long-term climate
change projections for temperature increases in the Utah region by about 205011. Therefore, this
result is directly relevant to impairment decisions that Utah may make in the future, because it
will introduce a range of variation among reference locations similar to the impairment decision
threshold, and thus may be more difficult to determine impairment as temperatures increase.
One peculiarity is that the median O/E scores at sites 4951200 (Virgin) and 4936750
(Duchesne)) are significantly higher (closer to 1) in hottest year samples. This means that the
observed community in hottest years is closer to the expected community. A similar pattern
occurred at site 4929750 (Weber) but this result was not statistically significant. These patterns
might be partially explained by the fact that Utah DEQ calibrated their RIVPACS model based
on data collected from 1999-2005, which happens to be a period during which some of the
hottest and driest conditions occurred, in some cases in consecutive years (Appendix Section
F5). Although it is possible that other confounding factors might have also contributed to O/E
trends at these sites, results suggest that climatic variables likely influenced these changes in
community composition.
11 See NCAR website: http://rcpm.ucar.edu
December 23, 2010
3-27
External Review Draft

-------
596
597
598
599
600
601
602
603
604
605
606
607
no
~o
I
1.05
1,00
0.95
0,90
0.85
W 080
O
075
0.70
0.65
0.60
1 00
0 96
0.90
0.85
0.80
0.75
0,70
0.65
060
0,55
0,50
0.45
0 40
0.35
Coldest	Normal	Hottest
Year Groupings
~ Median
O 25-75%
X (ton-Outlier Range
- o Outliers
+ Extremes
Figure 3-9. Distributions of observed/expected (O/E) values in coldest-, normal-, and
hottest-year samples at Utah sites 4936750 (Duchesne) (A) and 4951200 (Virgin) (B). Year
groupings are based on Parameter-elevation Regressions on Independent Slopes Model
(PRISM) mean annual air temperatures from each site during time periods for which
biological data were available. Average temperatures in hottest-year samples were 1.1 to
2.7°C higher than coldest year samples. At both sites, O/E values were significantly higher
in hottest year samples than in coldest and normal year samples. Data used in these
analyses were limited to autumn (September-November) kick-method samples.
In addition to the O/E trend analysis, we also performed several exploratory analyses in
which we manipulated the climate-related predictor variables that are used in the Utah fall
December 23, 2010
3-28
External Review Draft

-------
608
609
610
611
612
613
614
615
616
617
618
RIVPACS models. When climate-related predictor variables were altered, there was very little
effect on O/E values (Figure 3-10). This occurred in both the original model (with probability of
capture (Pc) set at 0.5, as used by Utah DEQ) and the model re-run with Pc <0.1 which allowed
for inclusion of rare taxa. The greatest change in O/E occurred in the scenario in which all
climate-related predictor variables were altered by the greatest amounts ('scenario 5 in Figure 3-
11) (for more information on model manipulations, see Appendix F). This amounted to a change
in O/E of 0.03, which is within the range of natural variability (Appendix F). There also was
little effect on O/E values when 'unrealistic' changes were made to climate-related variables to
investigate possible thresholds (i.e. doubling temperature, halving precipitation variables), with
O/E values never varying by more than one standard deviation (Appendix E).
26
24
22
1 20
i
| 18
1 16
14
12
10
1.09
1.08
107
1.06
1,05
1.04
1.03
1.02
# Ol^fc!\t*U >Pi Mil]
o
A
~
A
O
A
¦ E«jv MiF.
O c	i P<
A E c ei-ted < F :
<>
&
r
o
A
« O/E {Pc > 0.1)
O O/E (Pc >0.5)
2	3	4
Scenario
December 23, 2010
3-29
External Review Draft

-------
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
Figure 3-10. Exploration of how observed (O), expected (E) and observed/expected (O/E)
values from the Utah Fall RIVPACS model may change as climate-related predictor
variables change. Plot (A) shows changes in O and E and (B) shows changes in O/E at two
different probabilities of capture (Pc) (0.1 and 0.5) under 5 different scenarios: l=baseline;
2=temperature predictor variable values + 2, precipitation predictor variable values - 0.05;
3=temperature predictor variable values + 4, precipitation predictor variable values - 0.1;
4= temperature predictor variable values + 1, precipitation predictor variable values - 1,
day of last freeze -1, day of first freeze + 1; 5= temperature predictor variable values + 2,
precipitation predictor variable values - 2, day of last freeze - 2, day of first freeze + 2.
There are a number of possible reasons why the alterations to the climate-related
predictor variables resulted in small changes to O/E values. One is the fact that the analyses were
based on reference site data. Reference sites are typically more stable than test sites. Another
potentially important factor was that we disregarded elevation in the model manipulations. It
would be worthwhile to explore how manipulations to the elevation-related predictor variables
affect O/E values, especially since elevation and temperature are linked. Other potential factors
relate to model development. The fall Utah RIVPACS model is comprised of 15 predictor
variables. Recent analyses suggest that models with fewer predictor variables may have better
performance (pers. Comm. Chuck Hawkins). Also, the Utah model is unique in that it uses a
Random Forest model (Breiman and Cutler 2009) instead of discriminant analysis to predict site
group membership. It is possible that using the random forest may make the RIVPACS model
more robust to CC effects.
These results illustrate several important points in considering how RIVPACS models
may be affected by future climate change effects. One is the importance of the calibration data
used in model development. Ideally models are calibrated using data that encompass a full range
of natural variability. Unfortunately, these types of long-term data sets are rare. However, it is
something to consider and strive for as biomonitoring programs gather more data and recalibrate
their models over time. Another important consideration has to do with the assumption that
climate-related predictor variables, which are typically based on long-term (30-year) averages,
are relatively invariant over ecologically-relevant time. If climate change is going to be an
important factor in years to come, it would be interesting to develop a second RIVPACS model
that includes predictor variables based on current climate (not just the historic benchmark
climate) and to compare O/E values across these models over time. In theory, this would allow
for partitioning of climate change effects over time.
December 23, 2010
3-30
External Review Draft

-------
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
3.5 CONCLUSIONS ACROSS PILOT STUDY STATES
There are a variety of regional differences in biological responses evident from this study.
More and stronger trends and responses were found in Utah, largely related to temperature
changes. Fewer significant trends were found in North Carolina and more were related to
precipitation (see also Section 2). There is much spatial variation in these patterns, in part due to
ecoregional, geographic, and climatological variations, and in part attributable to limitations of
the available data. The results point to several conclusions. One is the importance of categorizing
taxa based on ecological traits, especially temperature sensitivities, in order to evaluate responses
to climate change variables and to estimate future vulnerabilities to climate change. It is a
relatively consistent finding that biological metrics and indices used by states and tribes are
either composites of cold-water and warm-water-preference taxa, or are dominated by one or the
other. This composition defines the nature of responses, and therefore, the vulnerability of the
metric or index to climate change effects. The richness of cold-water-preference taxa is a metric
that was fairly consistently responsive, especially at higher elevations, because high-elevation
communities tend to have more cold-water-preference taxa. Metrics using cold-water-preference
taxa will help identify climate change 'sensitive' or vulnerable areas. Such information would
assist in detecting climate change effects and in identifying sites to monitor these changes.
Another widespread and related finding is the moderate but significant relationship
between temperature sensitivity and sensitivity to organic pollution. Metrics selected because the
composite taxa were considered to be generally sensitive, such as EPT taxa, or generally tolerant,
such as Diptera taxa, or to represent responses to conventional pollutants (e.g., organic pollution
as in the HBI), also have demonstrable sensitivities to climate-related changes in temperature and
flow conditions. We have shown these sensitivities to be related, at least in part, to the
predominance of cold- (and/or warm-) water preference taxa at a location. Assemblage
composition by cold and warm-water-preference taxa may be related to ecoregion, latitude,
watershed size, and/or stream order, and is also clearly affected by elevation. This association
between temperature and pollution sensitivities will affect how indices are interpreted with
regard to the conventional stressors for which the indices were originally developed.
From more limited evidence it also appears that the ability to categorize taxa according to
flow preferences and requirements could be useful. However, there are generally fewer data
December 23, 2010
3-31
External Review Draft

-------
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
available for this analysis. We augmented the approach of grouping taxa by traits responsive to
one climate variable (temperature) through consideration of a suite of traits. This was useful in
some cases, though it produced fewer significant results. This was probably due to the fact that
fewer taxa were included when categorized by a suite of several traits, resulting in more limited
and/or more variable data and smaller sample sizes with which to test responses. Still, this is
potentially a useful approach to apply as more data become available.
3.6 RECOMMENDATIONS FOR MODIFYING METRICS
In general, biological metrics (indicators) are selected for their diagnostic value
(Verdonschot and Moog 2006). However, the effects of global climate changes in temperature
and precipitation on biological metrics have, until now, been largely untested, because climate
change was not considered a "stressor of concern" until recently (Hamilton et al., 2010a). Given
our demonstrations of the vulnerabilities of traditional metrics to climate change, and associated
impacts to the classification of station conditions, it is important that state and tribal
biomonitoring programs consider adopting modified metrics with the purpose of tracking
climate-associated changes in MMI outputs (Hamilton et al. 2010a). This will support making
inferences about cause, helping differentiate climate change from other stressors as part of a
weight of evidence evaluation. It will allow resource managers to more effectively make
management and regulatory decisions on the basis of biomonitoring results in the face of climate
change impacts (Hamilton et al. 2010b).
Our initial focus here is on the relative contribution of cold- and warm-water-preference
ecological trait groups to the composition of traditional metrics. Our general recommendation is
that the cold- and warm-preference components of traditional metrics be documented and tracked
separately. A recommended approach for incorporating modified metrics into a biomonitoring
data analysis regime is to continue calculating the traditional metric (e.g., EPT richness, HBI),
while adding new cold- and warm-water-preference metrics. Proportional changes in cold- and
warm-water-preference taxa would provide a basis for estimating how much of the difference in
the total (traditional) metric can be accounted for by changes in temperature trait groups. This
then becomes evidence for comparing potential climate change effects to those of other stressors
in a weight of evidence assessment. Comparisons could be made over time, among locations,
and/or groups of sites (both reference and non-reference). An option for tracking climate-related
December 23, 2010
3-32
External Review Draft

-------
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
changes is to put traditional and modified metrics on the same plot and compare their trends over
time (i.e. Figure 3-13). Another option that requires further testing is to track the ratio of the
cold- or warm-modified metric to the original (total) metric. For example, separate tracking of
cold-to-total EPT and warm-to-total EPT richness metrics was able to account for trends in total
EPT richness over time in circumstances where changes in total EPT richness were caused by
losses of cold-water-preference taxa, and where changes include both losses of cold-water-
preference taxa plus gains of warm-water-preference taxa (i.e., taxon replacements) (Hamilton et
al. 2010a).
We examined evidence in this study for the value of adopting temperature-modified
metrics for diversity and total taxa richness metrics; for EPT-related metrics; and for pollution
tolerance metrics, such as the HBI or related indices. However, the principle of partitioning
metrics to separate component taxa based on cold or warm-water-preferences should be
considered for other biological metrics (Hamilton et al. 2010a). These could include trait metrics
related to functional feeding groups (e.g., predators, collector-filterers) or life history habits (e.g.,
swimmers, climbers). Such metric modification should be considered on a state or region-
specific basis, in particular for climate-vulnerable regions (e.g., high elevations, low order
streams, small watersheds). In addition, an OCH taxa metric may be valuable to track taxa that
are robust to warmer conditions and/or more intermittent flows. This may be especially valuable
in regions at lower elevations, where temperature increases may be large, and/or where summer
flow conditions are likely to be especially vulnerable to climate change effects.
We cannot yet make strong suggestions for metrics related to hydrologic sensitivity, in
part because the lack of flow data corresponding to biological collections has limited ability to
calculate flow metric preferences by taxon (see Appendix K). However, hydrology-related trait
characterizations can be based on known life history traits coupled with regional observations
and literature information, as with the intermittent taxa metric used in North Carolina. A metric
that accounts for tolerance to intermittent flows, requirement for perennial flows, or some similar
hydrologic-preference metric, may become valuable as changes in flow conditions are more
evident. Such a metric would have to be calibrated by region.
Calculation of modified metrics for incorporation into biomonitoring data evaluation will
require designation of cold-and warm-water-preference ecological trait groups. Cold-and warm-
water-preference taxa lists must be developed on a state- or region-specific basis, which is a
December 23, 2010
3-33
External Review Draft

-------
746
747
748
749
750
751
752
753
754
755
756
757
substantial undertaking. The efforts initiated in this study, including the process of applying
weighted average or maximum likelihood modeling in concert with literature information and
best professional judgment to estimate temperature preferences by taxon from biomonitoring
data, and the development of a traits database that documents the temperature preferences and
tolerance results calculated for the 3 states analyzed in this study (see Stamp et al., 2010;
USEPA, 2011) can be used as a starting point for future state efforts.
35
30
25
20
15
10
5
p.. jx	a—o—g—n-g-
0
—•——¦—¦—1——t-			 "X, Total Taxa
1398 2000 2002 2004 2006 20os ^ cold Water Taxa
Warm Water Taxa
-5 	'	1	'	-»'	¦	
1984 1986 1988 1990 1392 1994 1S96
Year
Figure 3-13. Method for tracking changes in cold- and warm-water-preference taxa and
commonly used metrics (in this case, total number of taxa at Maine site 56817 (Sheepscot)
over time.
December 23, 2010
3-34
External Review Draft

-------
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
111
778
779
780
781
782
783
784
785
786
787
788
4. REFERENCE STATION VULNERABILITIES
While partitioning indices and revising predictive models are important steps in assuring
that the bioassessment design will continue to meet program needs under changing climatic
conditions, several other program elements also need to be considered. The use of reference
stations and comparison to reference conditions are central to bioassessment. Therefore, this
study also examined potential vulnerabilities of the sampling design, the process to determine
reference condition, and the location of reference sites.
4.1. VULNERABILITIES IN THE REFERENCE STATION SAMPLING DESIGN
State and tribal bioassessment programs establish reference stations across their
jurisdictions for reference-based comparisons to assess condition, detect impairment, and
identify causes. The main objectives of these programs focus on spatial comparisons, and
program design elements reflect this. Assessment designs generally include random sampling
within a stream reach or watershed, or a combination of random plus some targeted sampling.
Random sampling tends to maximize spatial sources of variation. Rotating basin sampling
designs are often used, which typically include sampling once every 5 years. Collections are
usually of one sample per location per year, with measurements of few covariates.
In contrast to the original spatial objectives of biomonitoring designs, detection of
climate change requires evaluation of trends over time, whether at a specific location or for a
defined area or stratum. There are some commonly observed limitations of many existing
biomonitoring programs with regard to assessment of trends. Despite the relatively large number
of reference stations in the biomonitoring data sets analyzed, there are very few with long-term
data, and typically few if any replication of long-term reference sites within a particular region
(Table 4-1). In addition, samples are not collected from the same sites every year (Table 4-1), so
many data sets have discontinuities, which make analyzing and detecting trends difficult. This
limits the adequacy of many biomonitoring programs for detecting climate change effects.
Continued accounting for climate change effects is desirable within the framework of state and
tribal biomonitoring programs, to increase the robustness of their program assessments to the
confounding effects of climate change. Modifying existing sampling design, potentially
December 23, 2010
4-1
External Review Draft

-------
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
including establishing a sentinel monitoring network specifically to detect climate change effects
would contribute to this objective.
Table 4-1. Time periods for which biological data were available at the long-
term monitoring sites in Utah (UT), Maine (ME), and North Carolina (NC).
Data used in these analyses were limited to autumn (September-November)
kick-method samples in the Utah data set, summer (July-September) rock-
basket samples in the Maine data set, and summer (July-August) standard
qualitative samples in the North Carolina data set.	
Station ID
Water body
Number of
years of data
analyzed
Years




1985-1995,
1998, 2000, 2001,
UT 4927250
Weber
17
2003-2005




1985-1993,
1996, 2000-2002,
UT 4951200
Virgin
14
2004

UT 4936750
Duchesne
12
1985-1993,
1995, 2000, 2001
UT 5940440
Beaver
9
1996-1998,
2000-2005
ME 56817
Sheepscot
22
1985-2006

ME 57011
W. Br. Sheepscot
12
1995-2006

ME 57065
Duck
9
1997-2005

NC 0109
New
11
1983-1990,
1993, 1998, 2003
4.2. VULNERABILITIES IN ASSESSING REFERENCE CONDITION
Reference station comparisons are central to bioassessment. Both in the United States
(Clean Water Act) and in Europe (Water Framework Directive) the determination of ecological
status and integrity is based on a comparative approach ("reference based comparisons")
requiring reference locations that can be used to set expectations for "natural" conditions and
associated variability (Barbour and Gerritsen, 2006; Stoddard et al., 2006; Verdonschot, 2006;
Nijboer et al., 2004; Walin et al. 2003). Impairment in the regulatory context is representative of
an unacceptable level of departure from this "expected condition" defined based on selected
reference stations. The vulnerabilities of reference locations to climate change, as well as to
ongoing changes in land use, is a significant issue that will impact the continued viability of
bioassessment approaches as currently applied.
Under ideal circumstances, reference conditions are found in locations unaffected by
human influences and thus represent natural, undisturbed conditions. At such sites, only natural
December 23, 2010
4-2
External Review Draft

-------
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
determinants of environmental conditions should influence biological communities. In the
absence of other stressors, long-term patterns in climate-related variables and associated
biological responses could be attributed to climate change (with the caveat that multi-decadal
climatic cycles also influence these communities). But "pristine" reference sites are seldom
available. It is more often the case that reference stations represent "best available" or
"minimally disturbed" conditions (Stoddard et al., 2006; Baily et al., 2004). Human influences of
agriculture and development are both widespread and long term in their effects (Allan, 2004;
Paul and Meyers, 2001), and many states have determined that they have no pristine or
unaffected reference conditions existing (see, for instance, Snook et al., 2007; Appendix H).
There also is variation among states in how reference stations are defined and selected. Some
states apply land use criteria, or at least document land use conditions, for selection of reference
stations. However, in many cases the selection of reference conditions are determined post facto
using biological sampling results, or are based on best professional judgment. In these cases, the
distribution of urban and agricultural land uses, or other factors affecting condition, can be less
than ideal, and often are not documented.
4.2.1. Reference Stations Used in this Study
We focused analyses in this study on reference stations to minimize the potential
influence of confounding factors. Given the variability in approaches for reference station
selection, and in the information documented within each bioassessment data base, a set of
criteria was established for selecting appropriate reference locations for analyses in this study.
Land use distribution was the primary consideration, based on the assumption that there is a
reasonable correspondence between extent and intensity of urban and agricultural land uses
surrounding a station and the level of non-point as well as point source influences on the stream
(e.g., Paul and Meyer, 2001; Arnold and Gibbons, 1996). Other factors considered in selecting
reference stations for analysis were the length of the data set, the presence of dams upstream of
the station, the occurrence of sewage treatment plant discharges, and consistent application of
appropriate sampling methods (see Appendix C).
Land use composition among major categories such as urban, agricultural, forest,
wetland, and barren, were obtained for a defined buffer area (1-km radius) around each sampling
site using a Geographic Information System (GIS) (see Appendix C). Stations for all states were
December 23, 2010
4-3
External Review Draft

-------
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
initially screened at 1%, 2%, and 5% urban, and 5% and 10% agricultural land use levels. Final
levels applied were 5% urban/10% agricultural in Maine and North Carolina, and 2% urban/10%)
agricultural in Utah. These values were selected in part based on practical considerations,
specifically the need to not eliminate all stations with data that could be used for long-term
analyses. These criteria are more conservative than those used in several southeastern states.
Georgia, Alabama, and South Carolina apply land use criteria for selecting reference stations of
<15% urban/<20%> agricultural for high gradient streams, and <15% urban/<30% agricultural in
low gradient streams (Barbour and Gerritsen, 2006). It will take additional analysis to determine
on a more objective level whether these criteria are adequate to minimize confounding of climate
change effects.
It is reasonable and sometimes necessary to use less than "natural" conditions as a
baseline for spatial comparisons. For example, accessibility of a site for frequent (e.g., annual)
long-term sampling can be an important practical consideration. For example, the longest term
reference station in Maine, 56817 (Sheepscot) is generally (though not always) categorized as an
"A" station by MDEQ, but is surrounded by about 16% urban and 23% agricultural land uses
(see Appendix C for characteristics of reference stations used in analyses for each state). Though
higher than would be considered ideal for "unconfounded" analyses, the level of urban land use
was stable over time (at about 16%), although forested conditions decreased from 84% to 57%,
while the agricultural land use increased 0% to 23%. At Maine's Station 57065, there was an
increase from 0% to 16% urban land use, but a decrease from 4% to 0% agricultural land use. At
Maine's Station 57011, urban land use increased from about 4% to 9%, and agricultural use from
0%> to 18.5%) with the changes coming from both forested and wetland uses. It is possible that
such land use changes may have contributed to trends observed at these sites (Appendices C and
E). It is recommended for all sampling stations, but especially for reference stations, that
quantification of land-use categories be documented. This will support tracking changes in land
uses over time (although land-use data are often only available at infrequent intervals), which
will aid in separating this from degradation due to climate change effects (and other stressors).
To supplement the spatial coverage of trend analyses from the limited number of
individual long-term reference stations available in each state, we grouped other reference
stations to form long-term data sets that could support climate change analyses. The intent was to
subset regions based on major factors considered important in driving natural, and therefore
December 23, 2010
4-4
External Review Draft

-------
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
predictable, variation in aquatic habitat characteristics. Typical factors include climate zone,
elevation, geology, and topography and are often considered well represented if areas follow
level 3 ecoregions. We therefore tested groupings within level 3 ecoregions and at a more refined
stage of level 4 ecoregions. We also defined regions for station groupings by physiographic
province in some cases. We screened candidate stations within each defined area according to
land use, absence of dams, and as having at least two or more years of data. We combined data
from all stations within a group for trend analyses. Ordinations and correlation analyses for
station groups in each state showed that samples within each station (different annual
collections) tended to cluster more closely than stations within the group. Site differences were
often greater than climate-related trends, and drove observed temporal trends if some sites in a
group were sampled early in the period and some later.
Within-group variation is an important result to consider, because in the biomonitoring
context, reference conditions are often established not based on a single reference station, but on
a population of reference locations that together reflect the range of natural variability for a
region (Barbour and Gerritsen, 2006). Combining reference stations across major physiographic,
geomorphic or climatological regions inflates the range of measured variation in biological
parameters from predictable, natural sources (Barbour and Gerritsen, 2006). It is thus important
to account for predictable, natural sources of variation. This will affect how many reference
stations within a defined area must be sampled, how frequently they must be sampled, and the
sampling duration needed to have the power detect climate change response trends. In the current
study, groups of reference stations analyzed were typically not of sufficient duration to define
statistically significant trends within the context of natural spatial and interannual variation. On
the other hand, transfer of results on trends and other biological responses defined from
individual long-term reference sites to the corresponding regions or states may be problematic, in
that without sufficient spatial replication it is difficult to know whether the observed trends are
representative of the region as a whole.
4.2.1. Climate Change Vulnerabilities of Reference Stations
Climate change influences reference station vulnerability through changes in biological
communities at these stations. This study documents climate changes that have the potential of
degrading reference station biological status in a manner that will make existing reference
December 23, 2010
4-5
External Review Draft

-------
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
stations more similar to non-reference stations, at least in some vulnerable regions (e.g., high
elevation sites, head-water or low order streams). In addition, at non-reference stations, effects of
climate change may be additive with other stressors, or interactions between climate change and
other stressors may augment or ameliorate responses (Figure 4-1).
CO
u
-o
c
"ro
u
'5b
_o
o
2
Ideal reference
sites (MDC)
Real reference
sites (LDC)
Stressed
sites
Time
Effect of climate
change
CC attenuates stress
No relative effect
CC increases stress
Figure 4-1. Conceptual model showing relationship between climate change trends and
reference and stressed sites with an overlay of temporal variation on the trend (black line).
"MDC" = minimally disturbed condition; "LDC" = least disturbed condition.
With regard to documenting reference station condition and establishing a framework
within which changes in condition can be judged, BCG (Davies and Jackson, 2006) captures a
more subtle range of biological conditions that have regulatory significance and value compared
to an "impaired/not impaired" decision approach. The associated levels provide a uniform
framework within which the degree of degradation attributable to climate change can be
characterized. The BCG also provides a more meaningful basis for characterizing existing
reference conditions. The more numerous, subtle and well defined levels captured in the BCG
delineate a meaningful and scaled framework within which the degree of degradation attributable
to climate change can be characterized. Figure 4-2 shows that as cold-water-preference taxa are
lost from North Carolina biomonitoring stations, the percentage of stations that are characterized
December 23, 2010
4-6
External Review Draft

-------
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
as excellent or good decreases. This climate change degradation of reference conditions will
impact the stability of reference baselines and associated comparisons upon which management
and regulatory decisions are base. A BCG would allow reference stations to be more accurately
characterized, would support evaluation of reference station condition or drift over time, and
would similarly support characterization of non-reference station changes over time (Figure 4-2)
This affects the interpretation of the scope of response of reference communities to both climate
change and conventional stressors and the interpretation of vulnerability of existing reference
conditions to climate change.
North Carolina Blue Ridge Mountain ecoregion stations
45.0%
40.0%
35.0%
30.0%
25.0%
20.0%
15.0%
10.0%
5.0%
0.0%
Excellent
¦ current benthic community composition
~lose 50% of cold-preference EPT taxa
nlose 100%ofcold-preference EPTtaxa
Good
Good-Fair
Classification


-i-i-
H
:::::
i

H
Fair
Poor
Figure 4-2. Reference station drift (degradation of assessed site condition) over time at
Blue Ridge Mountain ecoregion stations as cold-preference taxa are lost over time due to
climate change.
4.3. SYNERGISTIC EFFECTS BETWEEN CLIMATE CHANGE AND LAND USE
Though slightly different in geographic scale, both climate and land-use change can be
considered large-scale impacts (Hamilton et al. 2010b). Global climate change drivers are well
described (IPCC, 2007). Land-use change is generally considered a landscape-scale stressor, but
December 23, 2010	4-7	External Review Draft

-------
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
is driven by global population growth (Nakicenovic et al., 2000). Land-use changes, such as
urban/suburban land development, have encroached on and impaired reference stations across
the US. However, documentation of such problems has been sparse and likely has been handled
on a local, case-by-case basis.
The successful use of biomonitoring data for evaluating pollution impairment in the
context of climate change is in part related to understanding synergistic effects between climate
change and conventional stressors, and how they can be separated. These synergistic effects can
impact approaches used for attributing causes through the stressor identification process (see
USEPA, 2000). Synergistic effects between climate change and other stressors are increasingly
documented (Clement et al., 2008; Collier, 2008; Kaushal et al., 2008).
We examined the relative responses to climate change compared to land-use change
(urbanization) through analyses of existing biomonitoring data (Appendix J). Hydrologic
response variables play important roles in defining habitat conditions and structuring aquatic
communities (e.g., Poff et al., 1997) and are responsive to both climate change and urbanization.
Results show differences in the types of hydrologic variables (IHA, sensu Richter et al., 1996)
that are likely to be most responsive to either climate change or urbanization effects. High flow
metrics, such as flashiness, high-pulse-count duration, one-day maximum flow, and others, tend
to strongly reflect urbanization, swamping inputs from climate change effects. In comparison,
several low-flow metrics, such as 1-, 3- and 7-day minimum flows and low-pulse count, show
responses to climate change effects more so than to land use (Appendix J). Where future climate
change effects are small compared to land use, expectations are for more frequent, shorter,
higher flows in urban-affected streams. Where future climate change effects are large compared
to land-use effects, expectations are for more frequent, longer, lower flows. Accordingly, low-
flow parameters should be selected as sensitive climate change indicators, and low-flow effects
on biota are correspondingly expected to be most influential.
We further evaluated the relative effects of climate change and urbanization on stream
condition through benthic invertebrate responses, using the sampling results from the Piedmont
regions of Maryland and North Carolina as a test case. EPT taxa were evaluated as the
responding biological metric (see Section 2). EPT taxa respond to both high flow metrics
(flashiness) and to low flow metrics. For example, extreme increases in frequency of low-flow
pulses (>20/y) are associated with EPT taxa loss, though low-pulse count did not differ much
December 23, 2010
4-8
External Review Draft

-------
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
between the natural and urban streams in this analysis (Appendix J). There was a
association of decreasing richness of EPT taxa with increasing flashiness (Figure
J), as well as confirmation of the greater flashiness of urban streams.
strong
4-3, Appendix
All Streams
50
45
40
35
30
25
20
m
"0
HI 5
S10
03

•






•
•




A
•*
*
•






» A
o
o




\ A
A*,*
L ~ ~
<0


: %
. A ^
4* A
SJA



MA
^ ^|
^ A <
o
¦


¦ -
o ¦¦



*
¦ ¦
¦
"a
¦ ¦
¦ ¦
¦ ¦




¦
¦
¦
¦
¦
0 0.2
Stable
0.4
0.6 0.8
Flashiness
1.0
1.2 1.4
Flashy
Natural
Agriculture
Urban
Other
Figure 4-3. Relationship between richness of EPT taxa and flashiness (Baker's index) of
the stream for stream types in the North Carolina Piedmont.
There is an apparent threshold response below minimum flows of about 15% in natural
streams, where richness of EPT taxa is lower and less variable compared to higher flows,
(Appendix J). In this component of the study, urban conditions were compared with natural
stream conditions, and flow minima were more extreme in the urban streams. These results
suggest that natural streams are more resilient to hydrologic changes within the range of recent
past climate. Large changes in minimum or low flows may take much longer to become
biologically meaningful, and in the shorter term, temperature effects may be more important.
4.4. FUTURE VULNERABILITIES OF REFERENCE STATIONS TO LAND USE
References stations are vulnerable to human-induced changes to the surrounding
landscape. We evaluated current and future vulnerabilities of existing reference stations to
urban/suburban development for three study states (Maine, Utah, and North Carolina), as well as
December 23, 2010
4-9
External Review Draft

-------
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
for Florida as a case study (Appendix J) representing a high level of population growth. Data on
current and future land uses comes from the Integrated Climate and Land Use Scenarios
(ICLUS) project (Bierwagen et al. 2010). Future land-use scenarios are consistent with IPCC
Special Report on Emissions Scenarios (SRES) social, economic, and demographic storylines
used in global climate models (USEPA, 2009a; Nakicenovic and Swart, 2000). The ICLUS
scenarios consider different levels of population growth, with different assumptions about
development patterns (USEPA, 2009a). The two most extreme scenarios are: A2, which has high
population growth rates and business-as-usual development patterns; and Bl, which has low-
population growth rates and compact development patterns. We used a total of 248 reference
sites compiled from Maine, Utah, and North Carolina to examine their vulnerability to current
and future land use. The number and distribution of reference stations for these states are
discussed in earlier sections of this report and in Appendices E, F, and G. Florida DEP has about
308 sampling locations, with 58 reference sites designated as "exceptional" (Figure 4-4).
rag 10 units/acre
j ¦ 5-9.9 units/acre
2-4.9 units/acre
0.5-1.6 acreAjnit
| || 1.7-4.9 acreAjnit
I j S9.9 acre/unit
i 110-19.9 acreJUnit
I 120-39.9 acreAjnit
I	140-79.9 acreAjnit
II	180-159.9 acreAjnit
|i | >160 acreAjnit
j' ^ Private undeveloped
December 23, 2010
4-10
External Review Draft

-------
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
Urbanization affects stream conditions through alterations in hydrology and
geomorphology, with typically increased loading of nutrients, metals, pesticides, and other
contaminants; these effects are associated with increases in impervious surface (Paul and Meyer,
2001). For the Florida case study, results of a broad spatial analysis in New England states of the
relationship between human population density and Ephemeroptera (mayfly) taxon richness were
used to estimate the degree of urbanization representing a threshold of impairment (Figure 4-5)
(Snook et al., 2007). At low population densities, up to approximately 50 persons (-25 houses)
per square mile, there are few detectable biological responses. From 50-500 people (25-250
houses) per square mile corresponds to a degradation gradient, and above 500 people (250
houses) per square mile, New England streams are degraded. Therefore, a threshold of housing
density >25 houses per square mile was selected to indicate potential degradation. Using the land
use composition within a 1-km (0.62-mi) radius buffer around each reference station,
vulnerability was defined as >20% of the buffer with a land use at or above the threshold of
housing density.
For the analysis conducted for Maine, Utah, and North Carolina, urban and suburban
(>0.6 units/acre, or about 384 per square mile) was used. However, a threshold of 10% of
development within a 1-km buffer was used to reflect expectations for impacts to the biological
communities from urbanization (Schueler 1994, Booth and Jackson 1997, Wang et al. 2001).
These differences in thresholds may account for some of the differences in results between the
evaluation of the 3 study state reference stations and the Florida case study. Given the low
threshold of development used and the high population growth rates for Florida, we take the
Florida results to represent a worst-case scenario.
This analysis was done for several ICLUS scenarios to bracket a range of future
projections, including the base case that approximates the current condition; the A2 scenario, that
essentially represents a high estimate of population growth and development expansion; and the
B1 scenario that represents a minimized estimate of population growth and compact
development (USEPA, 2009a).
December 23, 2010
4-11
External Review Draft

-------
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
New England
35
30
25
20
15
10
m
T3
0 0
3
CD 0.5
3
T3
r+
CD

~
	i	
1


Y< ax +


•
•

1
1



	•					
1
¦
• •
\ • «
1
1
1
~
~ ¦
¦ •
• ~
~ ~ ~
• • • •
• • ••
~ *V 1
A. - iv i
~
¦
•
•y V >
¦ V | 1
#¦ ~
. 1
Xi •

¦
#
•
¦ ~


i i -
5.0
50.0
500.0
Population Density, /miz
CT
Rl
VT
ME
NH
MA
Figure 4-5. Relationship between human population density (i.e., degree of urban
development) and Ephemeroptera (mayfly) taxon richness among six New England states
(from Snook et al., 2007).
Among the 58 "exceptional"-grade reference stations in Florida under year 2000
conditions, 19% of the stations can be classified as vulnerable to land-use impacts (Table 4-2).
That is, nearly 1/5 of Florida reference stations may already exhibit impacts from urbanization.
Within the next two decades, more than one third of existing reference stations will be
vulnerable, and by 2100, nearly half of current reference stations may be impacted by
urbanization under the base case and A2 scenario. This level of vulnerability is significant.
Figures 4-6 and 4-7 illustrate the distribution of this reference station land use vulnerability for
the current base case, and for future (2100) projections of the base case, A2 and B1 scenarios.
The spatial distribution of this vulnerability is broad. In Florida, most sampling stations are in the
northern half of the state. Future projections of urbanization generally follow current patterns of
development, with particularly dense future development projected for the northern half of the
Florida peninsula (Figures 4-6 and 4-7). Sampling stations in these areas become vulnerable to
future development, especially in the high population growth (A2) scenario (Figure 4-3, left
panel), compared to the cluster of reference stations in northwestern Florida. The only reference
December 23, 2010
4-12
External Review Draft

-------
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
locations that appear to be protected from future land development are those largely surrounded
by water, and/or those within government-owned or protected lands that cannot be developed. In
Florida this represents about 17% of existing reference locations.
Table 4-2. Percent of existing Florida reference stations (N=58, classified as "exceptional"),
that have >20% developed land use (with 25 houses per square mile (9.65 houses per
square kilometer) or more, categories 5-12 in the ICLUS data set) within a 1-km buffer

Scenario
Year
BC
A2
Bl
2000
19.0%
19.0%
19.0%
2010
36.2%
34.5%
36.2%
2020
36.2%
36.2%
36.2%
2030
37.9%
37.9%
36.2%
2040
41.4%
39.7%
36.2%
2050
44.8%
44.8%
36.2%
2060
44.8%
44.8%
36.2%
2070
44.8%
44.8%
36.2%
2080
44.8%
44.8%
36.2%
2090
44.8%
44.8%
36.2%
2100
44.8%
48.3%
36.2%
The results for Maine, North Carolina, and Utah show a somewhat lesser degree of
vulnerability. Under current (2000) conditions, 22% reference locations in these three states have
greater than 10% urban/suburban densities within a 1-km2 neighborhood (Table 4-3). Under the
worst case (A2) scenario, future housing development increased that to 34% by 2100. The
maximum amount of suburban and urban development within the 1-km2 neighborhood in 2000
was 58%; this increased to 99% by 2050. The average amount of development increased from
22% in 2000 to 28% in 2050 and 34% in 2100 using A2 scenario, while it leveled off at 26%
using a lower population growth and higher development density scenario (Bl) (Table 4-3). The
results for Utah are difficult to interpret, and the projections not very meaningful, as the number
of reference sites falling within the 10% development threshold as calculated for a 1-km2
neighborhood was very small.
Table 4-3. Percent urban and suburban development within a 1 km2 area surrounding
reference sites, for all sites and for sites at or above the impact threshold of 10%. Number
of sites is shown in parentheses. Scenario A2 has high population growth and business-as-
December 23, 2010
4-13
External Review Draft

-------
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
usual development pattern; scenario B1 has low population growth and compact
development pattern (USEPA 2009).				

Area
2000
A2
2050
A2
2100
B1
2050
B1
2100
Mean of
reference sites
(>10% threshold)
Combined
22%
(35)
28%
(37)
34%
(45)
26%
(37)
26%
(37)
Maine
23%
(26)
24%
(26)
30%
(32)
23%
(26)
23%
(26)

North
Carolina
20%
(9)
27%
(9)
40%
(10)
24%
(9)
24%
(9)

Utah
0% (0)
87%
(2)
64%
(3)
77%
(2)
77%
(2)
The specific patterns of reference station distribution and vulnerability to land
development will vary among states, although there are widely applicable lessons from these
results. The high level of current vulnerability to urbanization (about 20% in all states tested
except Utah) highlights the difficulties in siting reference locations in many areas and the
probability of encountering substantial existing urban influences, which impact baseline
(reference) conditions. This evidence suggests that protection of reference stations is of
substantial importance. Options for protection may differ regionally and include zoning changes,
limitations to development within buffer zones of selected stream reaches, incorporation into
land protection programs (USEPA, 2009b), or other sociological, economic, and/or political
solutions. If alternatives for protecting reference locations are limited or costly, it may be that
reference stations in already protected areas, such as national parks, other government lands, or
in otherwise inaccessible areas may represent the only "protected" references. This is likely to
leave many watersheds and regional ecotypes without good reference conditions for comparison.
In Florida, this would reduce the ratio of reference sites to total sampling sites from 19% to 3%.
If reference sites are too scarce, they will be unrepresentative.
December 23, 2010
4-14
External Review Draft

-------
1104
1105	Figure 4-6. Distribution of Florida reference stations (N=58, classified as "exceptional"),
1106	plus categories of developed land use from ICLUS. Red squares highlight current reference
1107	stations that in the future with scenario-associated land use change projected to 2100, will
1108	have >20% developed land use (with 25 houses per square inile or more, categories 5-12 in
1109	the ICLUS data set) within a 1-km buffer surrounding the station. Left panel is current
1110	(2000) land use distribution; right panel is the base case scenario in 2100.
1111
1112	The need to protect reference locations is an important issue for the future of
1113	bioassessment. If reference stations become urbanized, the ability to detect climate change, and
1114	separate climate responses from conventional stressors in order to continue to manage resources,
1115	set permit limits, and meet CWA requires, may be hampered. It may become important to
1116	consider and promote more broad-based alternatives than just local or state-specific protections,
1117	such as regional cooperation in the establishment and monitoring of long-term fixed "sentinel"
1118	locations. The shifting baseline of reference condition demonstrates that both communication
1119	and understanding are immensely improved by measuring biological condition in comparison to
1120	pristine, undisturbed condition instead of to present-day reference.
1121
1122
1123
L*9*nd
December 23, 2010
4-15
External Review Draft

-------
1124	Figure 4-7. Distribution of Florida reference stations (N=58, classified as "exceptional"),
1125	plus categories of developed land use from ICLUS. Red squares highlight current reference
1126	stations that in the future with scenario-associated land use change projected to 2100, will
1127	have >20% developed land use (with 25 houses per square mile or more, categories 5-12 in
1128	the ICLUS data set) within a 1-km buffer surrounding the station. Left panel is the A2
1129	scenario in 2100; right panel is the B1 scenario in 2100. See Figure 4-3 for the current
1130	(2000) condition for comparison.
1131
1132	4.5. SENTINEL MONITORING NETWORK
1133	Results of this study have demonstrated the importance of accounting for climate change
1134	effects in order to maintain sound bioassessment decision making. The next step is to consider
1135	possibilities for augmenting existing programs to address this need. Section 5 discusses many of
1136	the typical characteristics of biomonitoring program and their inherent limitations with regard to
1137	detecting trends that might be associated with climate change. Approaches to address some of
1138	those limitations are discussed here.
1139	A monitoring network designed to detect climate-related changes needs to account for
1140	regional variations in climate, geology (including soils), topography, elevation, latitude,
1141	vegetation, etc. Such conditions often cross state and tribal boundaries. Therefore, this kind of
1142	monitoring network may require collaboration among states and tribes with regard to technical
1143	considerations (e.g., site selection, sampling methods) and funding. Regional and national
1144	support may be important to facilitate this process.
1145	Thorough coverage across ecoregions and other environmental variants would require a
1146	large network of sites. A modest initial effort for sentinel site monitoring could focus on highly
1147	vulnerable areas and watershed types. Since not all watersheds or community types would be
1148	represented by such selective establishment of a sentinel site monitoring network, the
1149	classification of conditions and transferability of bioassessment results will be integral for
1150	extrapolation to other areas (e.g., Allan et al., 1997; Gerritsen et al., 2000; Wu and Li, 2006).
1151	In order to separate climate change effects from other stressors, both reference and some
1152	portion of impaired sites should be measured over time; thus, sentinel sites should be established
1153	along the BCG and be anchored in reference conditions. This would support an analysis
1154	approach in which temporal trends at reference sites could be compared to temporal trends at
1155	impaired sites, in order to differentiate between climate effects and conventional stressors.
December 23, 2010
4-16
External Review Draft

-------
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
Different levels of stressor effects could also be compared and synergistic effects considered. It
is possible that in a monitoring context, as opposed to a controlled study, synergisms between
climate change and conventional stressor responses could not be fully partitioned. Inference
using literature studies, especially through use of CADDIS and the stressor identification process
(Suter et al., 2002; USEPA, 2000) would contribute to data interpretation in a weight of evidence
approach. The efficacy of conducting long-term sampling along the BCG should be considered
through interactions with state and tribal biomonitoring managers, consideration of avenues of
funding support, and finally, through practical evaluation of existing opportunities for
establishing such a sentinel site monitoring network in representative and vulnerable regions.
If a sentinel site monitoring network along the BCG is infeasible, a less resource-
intensive alternative would be to establish long-term sentinel sites only at high-quality reference
locations. Lack of trend data from non-reference sentinel locations would present some
limitations to separating climate change from other stressors responses. Selection of such
locations would face some of the same difficulties as any reference selection effort conducted by
individual states. However, the larger spatial scale and regional perspective necessary for
implementation would offer opportunities to search for and select least-affected locations from a
larger area and share results across jurisdictional boundaries.
While typical bioassessment approaches include sampling watersheds on a rotating, often
5-year basis, biomonitoring at sentinel sites should be considered on a regular, repeating basis,
annually if possible. With less frequent data, temporal variations from interannual and cyclic
climatic sources would greatly extend the time frame needed to describe climate change
responses.
Another component of sentinel site monitoring for climate change is the recommendation
for continued monitoring at targeted locations, even if initial site selection is probability-based,
rather than application of a probability-based sampling approach in which all sites are reselected
each year. Probability sampling has important strengths in capturing the (often large) range of
variability within a defined stratum, such as low-order stream reaches (Barbour and Gerritsen,
2006; Hughes et al., 2000); it also provides valuable data about the status of our nation's waters
at any given time (Hughes et al., 2000; Paulsen et al., 1998). This is important for defining the
range of condition within the stratum at any one time, but it requires replication (multiple
December 23, 2010
4-17
External Review Draft

-------
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
reference sites) within the stratum. There is also a high likelihood of never sampling the same
location again. We found high among-site variability within ecoregions despite expectation that
partitioning by ecoregion should control major predictable sources of variation. This maximizes
the effects of "natural" site (spatial) variability on detection of temporal trends, and greatly
extends the time it will take to discern climate change effects. This suggests a trade-off between
gaining knowledge about regional status and knowledge about long-term trends. There is a valid
consideration of whether detection of climate change patterns at a fixed location has meaning if
it does not incorporate the real range of conditions that defines the stratum. However, replication
of targeted locations within a region or stratum accounts for natural spatial variability.
Combining some fixed with random sites in a pre-determined sampling pattern may be the most
likely design that accomplishes both trend detection and representation (Urquhart et al. 1998).
Many different groups are considering, or have already started, monitoring for climate
change effects. If possible, collaboration among at least some groups, particularly among
bordering states, would have many potential benefits. Some duplication of effort could be
avoided, results could be integrated in a more meaningful way, and resources could potentially
be saved. Collaboration would foster consistency across groups in types of data collected, as well
as potential use of a common database. Efforts to discuss and establish a sentinel monitoring
network might facilitate collaboration among existing efforts. A common vision of sampling and
agreement on types of data that could be incorporated into a common database related to a
potential climate change monitoring network could have a better chance of success.
December 23, 2010
4-18
External Review Draft

-------
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
5. CHARACTERISTICS OF EXISTING BIOASSESSMENT PROGRAMS
RELEVANT TO DISCERNING CLIMATE CHANGE TRENDS
There are some inherent qualities of biomonitoring data that limit the ability to define
long-term trends. Limitations on the long-term trend analysis approach should be understood in
the context of the nature of the data being analyzed, and the types of information needed about
climate change responses in order to assess how state and tribal biomonitoring and biocriteria
programs are likely to be affected in the future.
5.1 SUFFICIENCY AND LIMITATIONS OF DATA TO DEFINE AND PARTITION
LONG-TERM TRENDS
One significant limitation is the small number of long-term monitoring sites available to
support temporal analyses. As discussed in Section 2, the small number and limited distribution
of long-term reference stations reduces the ability (1) to confirm regional trends, (2) assert the
strength of any trends discerned, and (3) to compare biological responses between regions.
Essentially, the very low number of stations with sufficient long-term data limits replication for
testing climate change effects.
The small number of long-term stations is largely a product of the focus of most state and
tribal biological monitoring programs. Objectives of these programs typically include assessing
the status, health, and integrity of aquatic ecosystems in response to Clean Water Act (CWA)
requirements (Barbour et al., 2000). The basis for such assessments is the comparison of test
locations to reference locations to detect community differences concurrently. Temporal patterns
seldom figure into these spatial comparisons. Figures 5-1 to 5-3, show the spatial distribution of
biomonitoring locations in Maine, North Carolina, and Utah, and illustrate that spatial coverage
using all sampling sites can be relatively extensive. Total spatial coverage of stations represents
the composite of the stations periodically re-sampled across major watersheds to assess condition
of state-wide aquatic resources and list impaired stream reaches, plus occasional additional
spatial efforts that may arise for evaluation of a particular discharge or other local impact.
December 23, 2010
5-1
External Review Draft

-------
Legend
Stations (Ref Status, Num Years)
(N«?42]
#	R«f. 20*tN«l|
¦ R«f. 10-19 (N-1)
T Ret M(N-T»
~ R«t. 2-4 |N»21(
•	R«f 11N-36)
Otom. 20* (M-0)
~	10-19 (NM)
V Oth«.»-»
-------
NC Reference Stations NC Non-Reference Stations Ecoregion
Ref status, Yrs data
~ Ref 10-19 (N=3)
¦ Ref 5-9 (N=13)
~ Ref 2-4 (N=53)
• Ref 1 (N=36)
1239
1240
1241
Ref status, Yrs data
¦£? Other 20+(N=2)
~ Other 10-19 (N=28)
~ Other 5-9 (N=210)
Other 2-4 (N=841)
Other 1 (N=1600)
LEVEL3_NAME
| Blue Ridge
Middle Atlantic Coastal Plain
| Piedmont
Southeastern Plains
Figure 5-2. North Carolina biomonitoring stations, with data durations by reference and
non-reference locations.
December 23, 2010
5-3
External Review Draft

-------
^ o o

&
IS
Utah stations
Ref Stat, Num Yrs
1242
1243
1244
1245
1246
1247
~
Ref > 20 (N=3)
~
Ref 10-19 (N=1)
A
Ref 5-9 (N=2)
O
Ref 2-4 (N=6)
+
Ref 1 (N=54)
~
Other > 20 (N=27)
~
Other 10-19 (N=22)
A
Other 5-9 (N=77)

Other 2-4 (N=163)
«&
Other 1 (N=194)
Ecoregion
LEVEL3_NAM
Centra! Basin and Range
| Colorado Plateaus
| Mojave Basin and Range
Northern Basin and Range
| Southern Rockies
| Wasatch and Uinta Mountains
| Wyoming Basin
Figure 5-3. Utah biomonitoring stations, with reference condition and non-reference
locations.
Despite the large number of stations, very few are sampled in more than one or a few
years over the entire period of record. For example, Maine has at least 742 stations, but only 66
December 23, 2010
5-4
External Review Draft

-------
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
classified as reference locations (<10%) (Table 5-1). Only two of these have been sampled for
more than a decade (Table 5-1), and only one of these for more than two decades. North
Carolina's biomonitoring program has been operating for two decades; still, only one reference
station exists with more than 10 years of data (Table 5-1). The small number of reference
locations with long-term data is a surprising but important finding that likely applies to many
other biomonitoring data sets.
Table 5-1. Average distribution of reference and total stations by state,
categorized by duration of sampling.
Years
Sampled
Maine
North Carolina
1
Jtah
Avera
ee
Ref
Total
Ref
Total
Ref
Total
Ref
Total
% Ref
1 to 4
57
696
89
2530
61
482
207
3708
5.6
5 to 9
7
40
13
223
1
41
21
304
6.9
> 10
2
6
3
33
4
26
9
65
13.8
Total
66
742
105
2786
66
549
237
4077
5.8
Another limiting factor for long-term analyses is that data for trend analysis must be
selected from reference data sets to minimize contributions from conventional stressors (see
Section 4). Only qualified reference locations should be used. In addition, climate change
responses differ among regions (see Section 3). Partitioning by ecoregion often means there are
few (or no) individual long-term stations available for analyses.
A related factor is the actual length of the "long-term" data record. Reference locations in
this study yielded some valuable results, but also many non-significant patterns. This suggests
that the duration and data density from these stations borders on data sufficiency. As examples,
the longest-term reference station in North Carolina, NC0109, had 11 years of data over a 21-
year time span (1983-2003); the longest-term reference station in Maine had 23 years of data
over a 23-year time span (1984-2006); and three long-term reference stations in Utah had 19
years of data over a 21-year span (1985-2005, station 4927250 - Weber), 15 years of data over a
20-year span (1985-2004, station 4951200 - Virgin), and 14 years of data over an 18-year span
(1985-2002, station 4936750 - Duchesne). The sufficiency of data duration in combination with
number of stations sampled and frequency of sampling is being further explored in subsequent
work.
December 23, 2010
5-5
External Review Draft

-------
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
Data durations of about 15-20 years also appear in the literature as an apparent minimum.
For example, analyzing an 18-year data set from a large number of streams in the UK, Durance
and Ormerod (2008) found significantly increasing temperature trends and significant
correlations of some invertebrate variables with temperature, although they concluded that water
quality improvements confounded interpretation of results. Chessman (2009) found significant
climate change trends in benthic invertebrate taxomonic families and trait groups within a 13-
year data record in New South Wales, Australia. Daufresne et al. (2003) defined aquatic
community trends in the Rhone River based on data durations of 20 (macroinvertebrates) to 21
(fish) years. Although Daufresne et al. (2003) found several meaningful community patterns and
showed statistically significant trends in temperature, trends related to flow parameters were
generally not found to be significant based on the same duration of data. Two possibilities are 1)
in the Rhone River there were no temporal trends in flow and/or no relationships between flow
and invertebrate or fish communities; or 2) given the typically high variability of hydrologic
variables, the 20 to 21-year duration of data was not sufficient to detect any trends. Murphy et al.
(2007) examined relationships between climate variables and benthic invertebrate responses in
England based on about 20 years of data, indicating that while multi-decadal data sets required to
define climate-driven trends were rarely available for rivers, potential responses of biota to
climate forcing can be estimated based on relationships between climate variables and biological
indicators using past data. Even with a long-term data set, Durance and Ormerod (2008)
discounted stream benthic assemblage changes that were correlated with long-term (18 years)
temperature increases at sites in southern England because some of the faunal changes included
taxa with traits (e.g., preferences for high flows, high dissolved oxygen) that were contrary to
expected responses to climate-driven increases in stream temperatures. The existence of trends,
by themselves, is insufficient to assert climate change impacts, but must be interpreted based on
consistency with expectations for biological responses to climate change.
One observation that stands out regarding the Maine, North Carolina, and Utah reference
locations is that most of these have more frequent annual sampling than would be the case if they
were only sampled on a "rotating basin" basis. Utah adopted a rotating basin sampling scheme as
well as a probability-based station selection approach within the last decade (Utah DEQ, 2006).
However, they maintain regular annual sampling at a small number of fixed locations with long-
December 23, 2010
5-6
External Review Draft

-------
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
term historic records. Whether by formal decision or historic happenstance, some other states
also have regularly sampled stations outside of rotating and/or probabilistic designs.
5.2. OTHER BIOMONITORING METHODS CONSIDERATIONS
Each of the states analyzed in this study use different collection methods. Utah collects a
quantitative sample from riffle habitats during a September/October index period using the
Environmental Monitoring and Assessment Program (EMAP) kick method (note: prior to 2006,
samples were collected using the Hess method) (Utah DEQ, 2006). Maine uses artificial
substrates (rock bags or baskets) to collect quantitative samples during late summer, low flow
periods (July 1 to September 30) (Maine DEP, 2002). North Carolina uses several different
collection methods, but for this study we focused on the standard qualitative, or 'full-scale',
method. It is comprised of 2 kicks, 3 sweeps, 1 leaf pack sample, 2 fine mesh rock and/or log
wash samples, 1 sand sample and visual collections (NCDENR, 2006). Abundance data is
recorded as rare=l (1-2 specimens), common=3 (3-9 specimens) or abundant (>10 specimens).
Ohio uses a modified Hester-Dendy multiplate artificial substrate sampler that is placed in-
stream to colonize for six weeks between mid-June and late September (DeShonn, 1995).
Some methods are likely to be more effective than others for certain applications (e.g.,
Flotemersch et al., 2006). Artificial substrates specifically placed to remain wetted for the entire
colonization period may be less sensitive to shifts in hydrology. In Maine, rock baskets are
placed in run habitats that will have sufficient water for the entire deployment period. If there are
drought-like conditions that cause a loss of edge habitat, the rock baskets are less likely to reveal
the potential loss of edge taxa. Even protocols that sample only riffles may be less likely to
collect edge-specialized fauna. However, the multiple habitat protocol used in North Carolina is
more likely to detect such shifts.
It is difficult to define which sampling protocol is best suited for detecting climate change
effects. Use of artificial substrates were favored for pollution detection on the premise that
application of a uniform substrate eliminates the substrate variation among stations as a variable
that would confound detection of community responses to a pollution discharge or other
disturbance (e.g., Barbour et al., 1999; Cairns, 1982). At least in some regions, long-term
changes in climate variables are expected to contribute to responses that can include drought or
flood-related changes in flows and associated changes in nutrient loadings, sediment loadings,
December 23, 2010
5-7
External Review Draft

-------
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
habitat availability, and other inter-related factors. Given these considerations, the ability to
examine the full spectrum of naturally occurring biological community components may be
advantageous. In-stream, multi-habitat sampling may be more likely to provide realistic
estimates of abundance or richness of particular indicator taxa. On the other hand, there is a
significant disadvantage to changes in sampling methods, due to the disruption it causes in
temporal patterns that might otherwise be observed. Because of this, any consideration of
changing sampling methods should at least be accompanied by a period of time in which both
methods are applied simultaneously in order to develop translation models. It should be noted
that such translational models may not always be effective or overcome inherent sampling
biases. For example, if rock baskets almost never effectively collect edge taxa, then no factor can
be defined that would translate multiple years of near-zero results into meaningful estimates of
abundance.
Because of considerations such as these that bear on the consistency of results, states
have a vested interest in continued use of their own methods to assure that new data are
meaningful to their program. Additional sampling might be considered in representative and/or
especially vulnerable regions as an adjunct to standard biomonitoring methods. For instance, in
streams with a high likelihood of transitioning from perennial to intermittent status, collection of
samples from edge habitats could be considered.
Another potential hindrance to effective detection of climate change trends is relatively
low sampling effort and the lack of replication in station sampling. In most biomonitoring
programs the concept of collection of replicate samples is relinquished in favor of collecting
single composite samples. The composites can be either of multiple artificial substrates (e.g., in
Ohio, 5 Hester-Dendy samples per station are composited and processed as a single unit
(DeShon 1995)); or a single sample unit can be a composite of collections made in multiple
representative habitats (NCDENR, 2006). In general, increasing the number of samples collected
and composited for a site has been found to decrease variance among 'replicate' (similar) sites
and increase the precision of characterizing the assemblage at the site (Cao et al., 2003; Diamond
et al., 1996). Multi-habitat sampling, applied in many biomonitoring programs (e.g., Utah, North
Carolina) is considered to yield representative, and therefore precise, samples (Barbour et al.,
2006; Hering, 2004). Though replication is considered necessary to determine the precision of
the sampling method (Barbour et al., 2000), it is often only accomplished on about 10% of
December 23, 2010
5-8
External Review Draft

-------
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
collections (e.g., Stribling et al., 2008; Barbour et al., 2006; Flotemersch et al., 2006). However,
with regard to understanding the significance and implications of climate change temporal
trends, knowledge of spatial variation within a station (or stream reach), and between similar
sites within a watershed or ecoregion, may be valuable.
There are some environmental variables that are, or can be, measured along with
biological samples to aid in interpretation of results. For example, a detailed assessment of
substrate and related habitat condition, as is used in EMAP (Lazorchak et al., 1998), is valuable
in differentiating habitat disturbance from other stressors. If and when biomonitoring programs
consider climate change as an additional stressor, it becomes valuable to have good information
on water temperatures and flows from biological collection sites. Existing sampling protocols
usually include concurrent point measurements of temperature, and sometimes also of pH,
dissolved oxygen, and conductivity, as these values are relatively easy to obtain with portable
sondes. However, the analyses conducted in this study illustrate that point measurements of
temperature are not a good measure of the stream conditions to which an aquatic community is
exposed. They tend to include a large amount of variation from time of day as well as date
during the seasonal index period when that measurement happened to be taken.
In this study, the lack of long-term, site specific temperature and flow data impaired the
ability to conduct weighted average modeling (or use of related approaches) to determine
temperature or flow parameter preferences for many taxa. It also made it difficult to conduct
simple trend and correlation analyses (see Sections 2 and 3). It would be beneficial to consider
deploying in situ equipment to obtain continuous water temperature and flow measurements at as
many climate change monitoring sites as possible. Though such equipment is widely available
and much less expensive than it used to be, the sometimes severe resource limitations
experienced by states and tribes may limit the extent to which this recommendation can be
applied. Priorities could be set based on regional assessments of relative vulnerability to climate
change. For example, a limited number of deployments could be done at reference locations in
higher elevations, and/or in lower order streams. Such deployments also could be coordinated
with implementation of monitoring at sentinel sites (Section 4). There is also high value in
continued operation of USGS long-term flow and temperature gages.
December 23, 2010
5-9
External Review Draft

-------
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
6. CLIMATE CHANGE IMPLICATIONS FOR ENVIRONMENTAL
MANAGEMENT
6.1 IMPAIRMENT LISTINGS AND TMDLS
6.1.1. Overview of impacts on impairment listings and TMDL development
One of the central objectives of state programs for establishing a reference condition
baseline and conducting ongoing biomonitoring at reference and non-reference locations is to
detect locations, or stream reaches, that are sufficiently different from the established baseline to
be considered impaired. The approach and specific criteria used to make impairment decisions
are established by states and tribes, and vary among regions to reflect the appropriate range of
natural variability (Barbour and Gerritsen, 2006). But the assumptions inherent in the almost
universally applied reference comparison approach include that the stressors likely to impair
streams and rivers within a region are accounted for within the sampling and analysis scheme
applied, and that if a real impairment exists, it can be detected with a reasonable level of
confidence. The concept that all stressors must be accounted for presents an unusual problem
with regard to climate change effects, because climate change effects are "global", so reference
stations are equally at risk. This threatens the reference comparison paradigm.
Results of this study reveal changes in biological indicators and within specific ecological
traits groups that are reasonably attributable to climate change effects and are likely to interfere
with impairment determinations. Sections 2 and 3 document changes in cold- and warm-water-
preference taxa at reference stations due to climate-change-related trends in temperature and
precipitation. These trends result in corresponding changes in biological metrics used by states,
such as EPT taxa richness or abundance in the HBI index. The observed and projected changes
in biological metrics are sufficient to downgrade reference station condition (Section 4).
Degradation of reference station condition is essentially causing references stations to become
more similar to non-reference stations, and diminishes the ability to detect impairment (Section
4). Previous analyses presented preliminary evidence for this (USEPA, 2008). These findings
imply that unless metrics are modified so that climate effects can be tracked and thresholds for
defining impairment re-evaluated, degraded reference conditions will cause fewer stream reaches
to be defined as impaired, at least in the most climate-vulnerable watersheds. This will lead to
less corrective action and greater long-term degradation of stream conditions (see also Hamilton
etal. 2010b).
December 23, 2010
6-1
External Review Draft

-------
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
When a stream segment is found to be impaired, total maximum daily loads (TMDLs) of
pollutants are developed by states, and the cause(s) of the impairment are identified through the
stressor identification process (USEPA, 2000; Suter et al., 2002). In permitting (e.g., the National
Pollutant Discharge Elimination System (NPDES)), discharge limits must be set considering any
existing TMDLs. Beyond the possibility of under-protection with fewer impairment listings and
fewer requirements for TMDLs, there may be other direct climate change implications to TMDL
development. Climate change scenarios show greater variability in runoff and flow, which may
result in greater uncertainty in loadings expected from non-point sources. Critical low flows also
drive TMDLs, and these may become uncertain and more difficult to predict. The identification
of culpable stressors is also complicated by the effects of climate change on biological
indicators.
6.1.2. Approaches to evaluate impairment listings and TMDL development in the context
of climate change
The main approaches pertinent to preserving the ability to detect impairment concern
climate change-related modifications of biological metrics, associated re-evaluation of
impairment thresholds, and reference station classification and protection. These actions are
directed at improving the ability to track effects of climate variables, compare these between
reference and non-reference locations, and thus increase the information brought to bear on
differentiating climate change from other stressors and detecting conventional stressor
impairment. The stressor identification process, tailored to include detailed climate change
information, would facilitate partitioning biological responses between climate change and other
stressors. The paradigm for conventional stressor identification is based on spatial
(reference/non-reference) comparisons, combined with weight-of-evidence evaluation of
potential causes, augmented by research and other literature-based knowledge of major cause-
effect expectations (Suter et al., 2002; USEPA, 2000). The need to partition climate change
effects could add a relatively extensive time component to this framework if the process were to
rely primarily on site-specific, long-term field data. However, it is impractical and undesirable
from a decision-maker's point of view to obtain this degree of detailed, long-term sampling for
every case of impairment assessment. From a practical perspective, it also is likely to be outside
of the level of resources available to most states or tribes for routine bioassessment sampling. An
December 23, 2010
6-2
External Review Draft

-------
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
alternative approach includes monitoring a more limited network of sentinel sites (Section 4.5).
Documentation of trends for monitoring data, other aspects of weight-of-evidence evaluation of
potential causes, and an expanded knowledge data base on biological responses to climate
change could be included in an expanded stressor identification process.
With regard to other vulnerabilities in the TMDL development process, there is a need
for watershed-specific modeling to predict how flow dynamics change with climate, to provide
support for estimating future changes in low flows, and to modify loading calculations and
limitations accordingly.
6.2. WATER QUALITY STANDARDS AND BIOCRITERIA
6.2.1. Overview of impacts on the development of water quality standards and biocriteria
Biological responses to climate change will likely impact water quality standards and
biocriteria through shifts in baseline conditions. This study illustrates several avenues through
which climate change is affecting stream communities in ways that have implications for
biocriteria programs. Details are presented in Section 2, which discusses how trait groups,
taxonomic groups, and to some extent, individual taxa appear to be responding over time to
climate drivers, responding in ways that are predicable, and responding in ways that are
consistent with expectations relative to climate change. Section 3 discusses implications of these
changes to various MMIs and predictive models. The cascading effects of climate change-related
trends in temperature and precipitation on watershed conditions, water quality, and aquatic
biological communities, will lead to shifting, most often degrading, baseline conditions.
Decreases in mean abundances and/or species richness of cold-preference or other sensitive taxa
and trait groups, increases in warm-preference or other tolerant taxa and groups, and also
increases in variability of these indicators drive reference sites to greater similarity with non-
reference areas, as well as greater difficulty in establishing statistical differentiation (USEPA,
2008). As a result, reference-based standards will be liable to progressive under-protection.
By itself, climate change can be expected to alter some uses and their attainability,
especially in vulnerable streams or regions. For example, some cold-water streams could take on
cool-water characteristics, with declining abundances and/or richness of sensitive cold-water
taxa, possible increases in warm-water taxa, and other changes potentially related to altered
December 23, 2010
6-3
External Review Draft

-------
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
hydrologic patterns. Regulated parameters such as temperature, dissolved oxygen, and ammonia,
may also be sensitive to climate change effects, and their values may need to be adjusted relative
to revised designated uses.
6.2.2. Approaches to modify the development of water quality standards and biocriteria in
the context of climate change
There are numerous criteria, both biological and chemical, that are addressed in water
quality standards, and which may be affected by climate change (Table 6-1). Biocriteria are of
particular interest, as they tie closely to the indices and thresholds used to determine condition
and impairment. The climate-related causes of drifting (degrading) baseline conditions cannot be
directly controlled, but can be assessed, at least to the degree resources allow. The concepts that
support this include clear documentation of reference conditions, tracking of changes in
reference conditions over time, and to the extent possible, protection of reference conditions
from other encroaching impacts, particularly land-use changes (Section 4). This may be extended
to include repetitive regional monitoring of sentinel sites, carefully chosen to represent the best
conditions of the most vulnerable regional watersheds (Section 4). Further efforts to address
climate change impacts to standards would require examination of which water quality standards
are resilient to climate change impacts and will remain protective, and identification of
susceptible standards that may need adjustment.
For watersheds that are found to be particularly vulnerable to climate change effects,
including those that are characterized by particularly vulnerable trait groups, more refined
aquatic life uses should be considered for application. Refinement of aquatic life uses can be
applied to guard against lowering of water quality protective standards. Uses are designated for a
stream segment based on conditions at similar reference stream segments, using information on
habitat characteristic and associated biological communities, and potentially also consideration
of economics and human-related conditions. Criteria are set to protect designated uses, and often
differ between use levels. Application of refined aquatic uses could provide a greater number of
more narrowly defined categories, which could accommodate potentially "irreversible" changes
(e.g., increased temperatures driven by long-term climate change), but with sufficient scope to
maintain protection, and also support anti-degradation from regulated causes.
December 23, 2010
6-4
External Review Draft

-------
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
Table 6-1. Variables addressed in criteria and pathways through which they
may be affected by climate change (from Hamilton et al. 2009)	
Criteria
Climate change impacts
Pathogens
Increased heavy precipitation and warming water temperatures may require the
evaluation of potential pathogen viability, growth, and migration.
Sediments
Changing runoff patterns and more intense precipitation events will alter
sediment transport by potentially increasing erosion and runoff.
Temperature
Warming water temperatures from warming air temperatures may directly
threaten the thermal tolerances of temperature-sensitive aquatic life and result
in the emergence of harmful algal blooms (HABs), invasion of exotic species,
and habitat alteration.
Nutrients
Warming temperatures may enhance the deleterious effects of nutrients by
decreasing oxygen levels through eutrophication (hypoxia), intensified
stratification, and extended growing seasons.
Chemical
Some pollutants (e.g., ammonia) are made more toxic by higher temperatures,
and also by pH, which may change as a result of climate change.
Biological
Climate changes such as temperature increases may impact species distribution
and population abundance, especially of sensitive and cold-water species in
favor of warm-tolerant species including invasive species. This could have
cascading effects throughout the ecosystem.
Flow
Changing flow patterns from altered precipitation regimes is projected to
increase erosion, sediment and nutrient loads, pathogen transport, and stress
infrastructure. Depending on region it is also projected to change flood
patterns and/or drought and associated habitat disturbance.
Salinity
Sea level rise will inundate natural and manmade systems resulting in
alteration and/or loss of coastal and estuarine wetland, decreased storm
buffering capacity, greater shoreline erosion, and loss of habitat of high value
aquatic resources such as coral reefs and barrier islands. Salt water intrusion
may also affect groundwater.
pH
Ocean pH levels have risen from increased atmospheric CO2, resulting in
deleterious effects on calcium formation of marine organisms and dependent
communities and may also reverse calcification of coral skeletons.
Climate change effects that contribute to degradation of water quality and biological
resource condition bring into question how antidegradation policies can be managed considering
the additional influences of climate change. High quality water bodies may be most vulnerable to
climate change degradation, making application of anti degradation policies in vulnerable water
bodies important. Management approaches and special considerations for implementation of
anti degradation policies may need attention. In addition, the application of use attainability
analyses (UAA) on vulnerable water bodies may be pertinent for characterizing climate change
effects.
December 23, 2010
6-5
External Review Draft

-------
7. CONCLUSIONS
Climate change will affect many of the components of bioassessment programs,
including assessment design, implementation, and environmental management. Implementing
the recommendations derived from the results in this study can improve the resilience of
bioassessment programs and ensure that management goals can be met under changing climatic
conditions. These steps can help manage the risks associated with not meeting goals, even
though the magnitude and timing of climate change effects on aquatic resources is uncertain.
There are four main sets of recommendations from this study specific to adaptations of
biomonitoring programs:
1.	Multi-metric indices should be revised to reflect the sensitivity of taxa and trait groups to
climate change effects; predictive models should also reflect these changes in indicators
and periodically revise the expected community composition used in the analysis. At
present, the most accessible information relates to temperature sensitivities and
preferences; however, sensitivities to changing hydrologic conditions should be pursued
in the future.
2.	A monitoring network to detect climate change effects should be set up, at least for the
most climate-vulnerable regions. This network will need to be more comprehensive
spatially and sampled more frequently than current bioassessment sites. Detecting climate
change at these monitoring sites requires that they are protected from other stressors.
3.	Abiotic data needs to be collected more frequently and at more sites; a monitoring
network to detect climate change effects should incorporate abiotic data collection as
well, including water temperature and flow. The value of better water temperature and
flow data is great, and consideration should be given to deploying in situ temperature and
flow meters.
4.	TMDLs and water quality standards should be examined to ensure that these remain
protective of aquatic life uses under changing climatic conditions.
December 23, 2010
7-1
External Review Draft

-------
We have some additional recommendations for further study and collaboration that
would enhance our ability to track climate change effects and separate these from other stressor
responses in the context of biomonitoring:
1.	The use of thermal-preference metrics for detecting climate-related trends should be
further explored. Monitoring of thermal-preference metrics will increase the probability
of detecting community responses to warming trends and reduce the likelihood that they
will be obscured by taxonomic variability.
2.	The listing lists of cold- and warm-water-preference taxa developed in this study should
be refined and extended to more states and regions. Refinements can be made by using
continuous water-temperature data instead of instantaneous water-temperature data, by
calculating propensity scores to help improve the robustness of the analyses (Yuan 2010),
and by using species-level OTUs for genera in which differences in which species-level
thermal preferences are known to occur.
3.	Continue to further our knowledge of traits and how they relate to climate change. More
information is needed about which traits are most important in the context of climate
change, the influence of each trait on an organism's ability to adapt, and which
combinations of traits are most adaptive to particular environmental conditions (Stamp et
al. 2010). A key component of furthering the traits-based framework will be expansion
and unification of existing trait databases (Statzner and Beche, 2010).
December 23, 2010
7-2
External Review Draft

-------
8. REFERENCES
Allan, J.D. 2004. Landscapes and riverscapes: The Influence of Land Use on Stream
Ecosystems. Annu. Rev. Ecol. Evol. Syst. 35:257-84.
Allan, J.D., D.L. Erickson, and J. Fay. 1997. The influence of catchment land use on stream
integrity across multiple spatial scales. Freshwater Biology 37:149-61.
Archaimbault, V., P. Usseglio-Polatera, and J.-P. Vanden Bossche. 2005. Functional differences
among benthic macroinvertebrate communities in reference streams of same order in a given
biogeographic area. Hydrobiologia 551:171-182.
Arnold, C.L. and C.J. Gibbons. 1996. Impervious surface coverage: the emergence of a key
environmental indicator. Am. Planners Assoc. J. 62:243-258.
Bady, P. S. Dole'dec, C. Fesl, S. Gayraud, M. Bacchi, andF. Scholl. 2005. Use of invertebrate
traits for biomonitoring of European large rivers: the effects of sampling effort on genus
richness and functional diversity. Freshwater Biology 50:159-173.
Bailey, R. C., R. H. Norris, and T. B. Reynoldson. 2004. Bioassessment of Freshwater
Ecosystems: Using the Reference Condition Approach. Kluwer Academic Publishers, New
York, New York, USA.
Bailey, R.C., M.G. Kennedy, M.Z. Dervish, and R.M. Taylor. 1998. Biological assessment of
freshwater ecosystems using a reference condition approach: comparing predicted and
actual benthic invertebrate communities in Yukon streams. Freshwater Biology 39:765-774.
Baker, David B., R. Peter Richards, Timothy T. Loftus, and Jack W. Kramer. 2004. A New
Flashiness Index: Characteristics and Applications to Midwestern Rivers and Streams.
Journal of the American Water Resources Association (JAWRA) 40(2):503-522.
Barbour, M.T. and J. Gerritsen. 2006. Key features of bioassessment development in the United
States of America. Pages 351-366 In G. Ziglio, M. Siligardi, and G. Flaim (editors).
Biological Monitoring of Rivers: Applications and Perspectives. John Wiley and Sons,
Ltd., Chichester, England. 469 pgs.
Barbour, M.T., J.B. Stribling, and P.F.M. Verdonschot. 2006. The multihabitat approach of
USEPA's Rapid Bioassessment Protocols: Benthic macroinvertebrates Limnetica 25
(3):839-850.
Barbour, M.T., W.F. Swietlik, S.K. Jackson, D.L. Courtemanch, S.P. Davies, and C.O. Yoder.
2000. Measuring the attainment of biological integrity in the USA: A ciritial element of
ecological integrity. Hydrobiologia 422/423:453-464.
December 23, 2010
8-1
External Review Draft

-------
Barbour, M. T. and Yoder, C. O., 2000. The multimetric approach to bioassessment, as used in
the United States. In. Assessing the Biological Quality of Freshwaters: RIVPACS and Other
Techniques, Wright, J. F., Sutcltiffe, D. W. and Furse, M. T. (Eds). Freshwater Biological
Association: Ambleside, UK, pp. 281-292
Barbour, M.T., J. Gerritsen, B.D. Snyder, and J.B. Stribling. 1999. Rapid Bioassessment
Protocols for Use in Streams and Wadeable Rivers: Periphyton, Benthic Macroinvertebrates
and Fish, Second Edition. EPA 841-B-99-002. U.S. Environmental Protection Agency;
Office of Water; Washington, D.C.
Barbour, M.T., J.B. Stribling, and J.R. Karr. 1995. The multimetric approach for establishing
biocriteria and measuring biological condition. In W.S. Davis and T.P. Simon (eds).
Biological assessment and criteria: tools for water resource planning and decision making.
CRC Press, Boca Raton, FL. p. 63-76.
Beche, L.A. and V.H. Resh. 2007. Biological traits of benthic macroinvertebrates in California
mediterranean-climate streams: long-term annual variability and trait diversity patterns.
Fundamental and Applied Limnology Archiv fiir Hydrobiologie 169/1: 1-23.
Beche, L.A., E.P. McElracy and V.H. Resh. 2006. Long-term seasonal variation in the biological
traits of benthic-macroinvertebrates in two Mediterranean-climate streams in California,
U.S.A. Freshwater Biology 51: 56-75.
Beck, W.M. Jr. 1977. Environmental Requirements and Pollution Tolerance of Common
Freshwater Chironomidae. Report EPA-600/4-77-024. U.S. EPA, Washington, D.C. 260 p.
Bierwagen, B.G., D.M. Theobald, C.R. Pyke, A. Choate, P. Groth, J.V. Thomas, and P.E.
Morefield. 2010. National housing and impervious surface scenarios for integrated climate
impact assessment. Proceedings of the National Academy of Sciences.
Bohmer, J., C. Rawer-Jost, and A. Zenker. 2004. Multimetric assessment of data provided by
water managers from Germany: assessment of several different types of stressors with
macrozoobenthos communities. Hydrobiologia 516: 215-228.
Bonada, N., M. Rieradevall and N. Prat. 2007a. Macroinvertebrate community structure and
biological traits related to flow permanence in a Mediterranean river network.
Hydrobiologia 589: 91-106.
Bonada, N., S. Doledec and B. Statzner. 2007b. Taxonomic and biological trait differences of
stream macroinvertebrate communities between Mediterranean and temperate regions:
implications for future climatic scenarios. Global change biology 13: 1658-1671.
Bonada, N., N. Prat, V.H. Resh and B. Statzner. 2006. Developments in aquatic insect
biomonitoring: a comparative analysis of recent approaches. Annual Review of Entomology
51: 495-523.
December 23, 2010
8-2
External Review Draft

-------
Booth, D.B. and C.R. Jackson. 1997. Urbanization of aquatic systems: Degradation thresholds,
stormwater detection, and the limits of mitigation. Journal of the American Water Resources
Association 33: 1077-1090.
Borwick, J., J. Buttle and M.S. Ridgway. 2006. A topographic index approach for identifying
groundwater habitat of young-of-year brook trout (Salvelinus fontinalis) in the land-lake
ecotone. Can. J. Fish. Aquat. Sci. 63: 239-253.
Bradley, D.C. and S.J. Ormerod. 2001. Community persistence among stream invertebrates
tracks the North Atlantic Oscillation. Journal of Animal Ecology 70: 987-996.
Brandt, Darren. 2001. Temperature Preferences and Tolerances for 137 Common Idaho
Macroinvertebrate Taxa. Idaho Department of Environmental Quality. Coeur d'Alene, ID.
Breiman, L and A. Cutler. 2009. Internet site:
http://www.stat.berkelev.edu/ -breiman/RandomForests/cc home.htm#intro. Source areas:
random forests. Accessed April 23, 2009.
Bressler, D.W., M.J. Paul, A.H. Purcell, M.T. Barbour, E.T. Rankin, and V. H. Resh. 2009.
Assessment Tools for Urban Catchments: Developing Stressor Gradients. Journal of the
American Water Resources Association (JAWRA) 45(1): 1-15. DOI: 10.1111/j. 1752-
1688.2008.00278.x.
Brown, D.B. and Comrie, A.C., 2004: A winter precipitation 'dipole' in the Western United
States associated with multidecadal ENSO variability. Geophysical Research Letters 31,
doi: 10.1029/2003GL018726.
Brown, D.P. and Comrie, A.C., 2002: Sub-regional seasonal precipitation linkages to SOI and
PDO in the Southwest United States. Atmospheric Science Letters 3, 94-102.
Buisson, I	W. Thuiller, S. Lek, P. Lim, G. Grenouillet. 2008. Climate change hastens the
turnover of stream fish assemblages. Global Change Biology 14:2232-2248.
Bunn, S.E. and P.M. Davies. 2000. Biological processes in running waters and their implications
for the assessment of ecological integrity. Hydrobiologia 422/423:61-70.
Burgmer, T., H. Hillebrand, and M. Pfenninger. 2007. Effects of climate-driven temperature
changes on the diversity of freshwater macroinvertebrates. Oecologia 151:93-103.
Cajo J. F. ter Braak, C.J.F. and C.W.N. Looman. 1986. Weighted averaging, logistic regression
and the Gaussian response model. Vegetatio 65:3-11.
Cairns, J., Jr. 1982. Artificial substrates. Ann Arbor Science Publishers, Inc., Ann Arbor,
Michigan.
Caissie, D. 2006. The thermal regime of rivers: a review. Freshwat. Biol. 51:1389-1406.
December 23, 2010
8-3
External Review Draft

-------
Cao, Y., C.P. Hawkins, and M.R. Vinson. 2003. Measuring and controlling data quality in
biological assemblage surveys with special reference to stream benthic macroinvertebrates.
Freshwater Biology 48, 1898-1911.
Cereghino, R., Y.-S. Park, A. Compin, and S. Lek. 2003. Predicting the species richness of
aquatic insects in streams using a limited number of environmental variables. J. N. Am.
Benthol. Soc. 22(3):442-456.
Charvet, S., B. Statzner, P. Usseglio-Polatera and B. Dumont. 2000. Traits of benthic
macroinvertebrates in seminatural French streams: an initial application to biomonitoring in
Europe. Freshwater Biology 43:277-296.
Chessman, B. 2009. Climatic changes and 13-year trends in stream macroinvertebrate
assemblages in New South Wales, Australia. Global Change Biology (2009), doi:
10.1111/j. 1365-2486.2008.01840.X.
Chovanec, A., P. Jager, M. Jungwirth, V. Koller-Kreimel, O. Moog, S. Muhar and St. Schmutz.
2000. The Austrian way of assessing the ecological integrity of running waters: a
contribution to the EUWater Framework Directive. Hydrobiologia 422/423: 445-452.
Christensen, J.H., B. Hewitson, A. Busuioc, A. Chen, X. Gao, I. Held, R. Jones, R.K. Kolli, W-
T. Kwon, R. Laprise, V. Magana Rueda, L. Mearns, C.G. Menendez, J. Raisanen, A. Rinke,
A. Sarr and P. Whetton. 2007. Regional Climate Projections. In: Climate Change 2007: The
Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report
of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z.
Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge University
Press, Cambridge, United Kingdom and New York, NY, USA.
Christensen, N., and D. P. Lettenmaier. 2006. A multimodel ensemble approach to assessment of
climate change impacts on the hydrology and water resources of the Colorado River basin,
Hydrol. Earth Syst. Sci. Discuss. 3: 3737- 3770.
Christensen, N.S., A.W.Wood, N. Voisin, D.P. Lettenmaier, and R.N. Palmer. 2004. Effects of
climate change on the hydrology and water resources of the Colorado river basin. Climatic
Change 62:337-363.
Clarke, R.T. andJ.F. Murphy. 2006. Effects of locally rare taxa on the precision and sensitivity
of RIVPACS bioassessment of freshwaters. Freshwater Biology 51(10) pp.1924-1940.
Clarke, R.T., Wright, J.F, and Furse, M.T. (2002) RIVPACS models for predicting the expected
macroinvertebrate fauna and assessing the ecological quality of rivers. Ecological Modelling
160:219-233.
Clements, W.H., M.L. Brooks, D.R . Kashian, and R.E. Zuellig. 2008. Changes in dissolved
organic material determine exposure of stream benthic communities to UV-B radiation and
heavy metals: implications for climate change. Global Change Biology 14:2201-2214.
December 23, 2010
8-4
External Review Draft

-------
Collier, K.J. 2008. Temporal patterns in the stability, persistence and condition of stream
macroinvertebrate communities: relationships with catchment land-use and regional climate.
Freshwater Biology 53:603-616.
Cuffney, T.F., M.D. Bilger and A.M. Haigler. 2007. Ambiguous taxa: effects on the
characterization and interpretation of invertebrate assemblages. J. N. Am. Benthol. Soc.
26(2):286-307.
Dale, V.H. and S.C. Beyeler. 2001. Challenges in the development and use of ecological
indicators. Ecological Indicators 1:3-10.
Daufresne, M. and P. Boet. 2007. Climate change impacts on structure and diversity of fish
communities in rivers. Global Change Biology 13:2467-2478.
Daufresne, M., P. Bady, and J.-F. Fruget. 2007. Impacts of global changes and extreme
hydroclimatic events on macroinvertebrate community structures in the French Rhone
River. Oecologia 151:544-559.
Daufresne, M., M.C. Roger, H. Capra, and N. Lamouroux. 2004. Long term changes within the
invertebrate and fish communities of the upper Rhone River: effects of climate factors.
Global Change Biology 10:124-140.
Davies, P.E. 2000. Development of a national river bioassessment system (AUSRIVAS) in
Australia, in Pages 113-124. Wright, J.F., D.W. Sutcliffe, and M.T. Furse (editors).
Assessing the biological quality of freshwaters: RIVPACS and other techniques. Freshwater
Biological Association, Ambleside, Cumbria, United Kingdom.
Davies, S.P., and S.K. Jackson. 2006. The Biological Condition Gradient: A descriptive model
for interpreting change in aquatic ecosystems. Ecolog. Appl. 16: 1251-1266.
Davies, S.P. and L. Tsomides. 2002. Methods for Biological Sampling and Analysis of Maine's
Rivers and Streams. DEP LW0387-B2002. Prepared for the State of Maine Department of
Environmental Protection. Internet link (accessed July 2009):
http://www.maine.gov/dep/blwq/docmonitoring/biomonitoring/materials/finlmethl.pdf
Del Rosario, R.B. and V.H. Resh. 2000. Invertebrates in intermittent and perennnial streams: is
the hyporheic zone a refuge from drying? J. N. Am Benthol. Soc. 19(4): 680-696.
DeShon, J. 1995. Development and application of the Invertebrate Community Index (ICI). p.
217-243. In W.S. Davis and T.P. Simon (eds). Biological assessment and criteria: tools for
water resource planning and decision making. CRC Press, Boca Raton, FL.
Dewson, Z.S., A.B.W. James, and R. G. Death. 2007. Invertebrate community responses to
experimentally reduced discharge in small streams of different water quality. J. N. Am.
Benthol. Soc., 2007, 26(4):754-766.
December 23, 2010
8-5
External Review Draft

-------
Diamond, J. M., M. T. Barbour, and J. B. Stribling. 1996. Characterizing and comparing
bioassessment methods and their results: A perspective. Journal of the North American
Benthological Society 15:713-727.
Diaz, A.M., M.L. Suarez Alonso, and M. R. Vidal-Abarca Gutierrez. 2008. Biological traits of
stream macroinvertebrates from a semi-arid catchment: patterns along complex
environmental gradients. Freshwater Biology 53:1-21.
Doledec, S. B. Statzne r and M. Bournard. 1999. Species traits for future biomonitoring across
ecoregions: patterns along a human-impacted river. Freshwater Biology 42:737-758.
Doledec, S., B. Statzner and V. Frainay. 1998. Accurate description of functional community
structure: identifying stream invertebrates to species-level? Bulletin of the North American
Benthological Society 15: 154-155.
Doledec, S., J.M Olivier and B. Statzner. 2000. Accurate description of the abundance of taxa
andtheir biological traits in stream invertebrate communities—effects of taxonomic and
spatial resolution:Archiv fiir Hydrobiologie 148: 25-43.
Douglas, T. (RPBio) 2006. Review of Groundwater-Salmon Interactions in British Columbia.
Prepared for Watershed-Watch Salmon Society and Walter and Duncan Gordon Foundation.
Ducrot, V., P. Usseglio-Polatera, A.R. R. Pery, J. Mouthon, M. Lafont, M.-C. Roger, J. Garric
and J.-F. Ferard. 2005. Using aquatic macroinvertebrate species traits to build test batteries
for sediment toxicity assessments: accounting for the diversity of potential biological
responses to toxicants. Environmental Toxicology and Chemistry 24: 2306-2315.
Durance, I. and S.J. Ormerod. 2008. Trends in water quality and discharge confound long-term
warming effects on river macroinvertebrates. Freshwater Biology doi: 10.1111/j .1365-
2427.2008.02112.x.
Durance, I., and S. J. Ormerod. 2007. Climate change effects on upland stream
macroinvertebrates over a 25-year period. Global Change Biology 13:942-957.
Eaton, J.G. and R.M. Scheller. 1996. Effects of climate warming on fish habitat in streams of the
United States. Limnol. and Oceanogr. 41(5): 1109-1115.
Euro-limpacs Consortium: Freshwaterecology.info - The Taxa and Autecology Database for
Freshwater Organisms. Available from www.freshwaterecology.info (Version: 3.2 -
08/2008)
European Commission, 2000. Directive 2000/60/EC. Establishing a framework for community
action in the field of water policy. European Commission PE-CONS 3639/1/100 Rev 1,
Luxembourg.
December 23, 2010
8-6
External Review Draft

-------
Extence, C.A., D.M. Balbi, and R.P. Chadd. 1999. River flow indexing using British benthic
macroinvertebrates: a framework for setting hydroecological objectives. Regul. Rivers: Res.
Mgmt. 15: 543-574.
Fellows, C.S., J.E. Clapcott, J.W. Udy, S. E. Bunn, B.D. Harch, M.J. Smith, and P.M. Davies.
2006. Benthic metabolism as an indicator of stream ecosystem health. Hydrobiologia
572:71-87.
Flotemersch, J.B. Stribling, and M.J. Paul. 2006. Concepts and Approaches for the
Bioassessment of Non-wadeable Streams and Rivers. EPA/600/R-06/127. US
Environmental Protection Agency, Cincinnati, Ohio.
Furse, M.T., D. Moss, J.F. Wright, P.D. Armitage, and R.J.M. Gunn. 1984. The influence of
seasonal and taxonomic factors on the ordination and classification of running-water sites in
Great Britain and on the prediction of their macroinvertebrate communities. Freshw. Biol.
14:257-280.
Gallardo, B., S. Gascon, A. Cabezas, M. Gonzalez, M. Garcia, and F.A. Comin. 2009.
Fundamental and Applied Limnology Archiv fur Hydrobiologie 173/4:281-292.
Gayraud, S., B. Statzner, P. Bady, A. Haybach, F. Scholl, P. Usseglio-Polatera and M. Bacchi.
2003. Invertebrate traits for the biomonitoring of large European rivers: an initial
assessment of alternative metrics. Freshwater Biology 48:2045-2064.
Gerritsen, J. and M. Craig. 2008. The biological condition gradient: model development and
calibration. Prepared for U.S. EPA Office of Science and Technology.
Gerritsen, J., M.T. Barbour, and K. King. 2000. Apples, oranges, and ecoregions: on determining
pattern in aquatic assemblages. J. N. Am. Benthol. Soc. 19(3):487-496.
Griffith, G., J. Omernik and J. Comstock. 2002. Ecoregions of North Carolina. Corvallis, OR.
Hamilton, A.T., J.D. Stamp, andB.G. Bierwagen. 2010a. Vulnerability of biological metrics and
multimetric indices to effects of climate change. Journal of the North American
Benthological Society 29(4): 1379-1396.
Hamilton, A.T., M.T. Barbour, and B.G. Bierwagen. 2010b. Implications of global change for
the maintenance of water quality and ecological integrity in the context of current water
laws and environmental policies. In: R. J. Stevenson, S. Sabater (guest eds) Global Change
and River ecosystems - Implications for Structure, Function and Ecosystem Services.
Hydrobiologia 657:263-278.
Hamilton, A.T., A. Roseberry-Lincoln, M.T. Barbour, J. Stamp, E. Leppo, and D. Bressler. 2009.
Draft white paper on climate change and bioassessments. Prepared for Office of Science and
Technology/OW, MC 4304T, Washington DC.
December 23, 2010
8-7
External Review Draft

-------
Hargrave, C.W., K.P. Gary, and S.K. Rosado. 2009. Potential effects of elevated atmospheric
carbon dioxide on benthic autotrophs and consumers in stream ecosystems: a test using
experimental stream mesocosms. Global Change Biology doi: 10.1111/j. 1365-
2486.2009.01897.x.
Harper M.P. and B.L. Peckarsky. 2006. Emergence cues of a mayfly in a high-altitude stream
ecosystem: Potential response to climate change. Ecological Applications 16:612-621.
Harper, D.M., J.L. Kemp, B.Vogel, and M.D. Newson. 2000. Towards the assessment of
'ecological integrity' in running waters of the United Kingdom. Hydrobiologia 422/423:
133-142.
Harris, T.L., and T.M. Lawrence. 1978. Environmental Requirements and Pollution Tolerance of
Trichoptera. Report No. EPA-600/4-78-063. U.S. EPA, Washington, D.C. 316 p.
Hawkins, C.P., J.N. Hogue, L.M. Decker, and J.W. Feminella. 1997. Channel morphology, water
temperature, and assemblage structure of stream insects. Journal of the North American
Benthological Society 16(4):728-749.
Hawkins, C.P. 2006. Quantifying biological integrity by taxonomic completeness: its utility in
regional and global assessment. Ecological Applications, 16(4), 2006, pp. 1277-1294.
Hawkins, C.P. and R.H. Norris. 2000. Performance of different landscape classifications for
aquatic bioassessments: introduction to the series. J. N. Am. Benthol. Soc. 19(3):367-369.
Hawkins, C.P., R.H. Norris, 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.
Hayhoe, K., C.P. Wake, T.G. Huntington, L. Luo, M.D. Schwartz, J. Sheffield, E. Wood, B.
Anderson, J. Bradbury, A. DeGaetano, T.J. Troy, and D. Wolfe. 2007. Past and future
changes in climate and hydrological indicators in the US Northeast. Clim Dyn 28:381-407.
Herbst, D. and E.L. Silldorff. 2007. Development and Evaluation of Tolerance Values for
Lahontan Region Invertebrates- Preliminary Analysis Summary.
http://www.climatechange.umaine.edu/mainesclimatefuture/
http://www.epa.gov/emap/greatriver/rcw/Yuan.pdf
Herbst, D. and E.L. Silldorff. 2006. Comparison of the performance of different bioassessment
methods: similar evaluations of biotic integrity from separate programs and procedures. J.
N. Am. Benthol. Soc. 25(2):513-530.
Hering, D., R.K. Johnson, S. Kramm, S. Schmutz, K. Szoszkiewicz, and P. F.M. Verdonschot.
2006. Assessment of European streams with diatoms, macrophytes, macroinvertebrates and
fish: a comparative metric-based analysis of organism response to stress. Freshwater
Biology 51:1757-1785.
December 23, 2010
8-8
External Review Draft

-------
Hering, D., O. Moog, L. Sandin, and P.F.M. Verdonschot. 2004. Overview and application of the
AQEM assessment system. Hydrobiologia 516: 1-20.
Hiddink, J.G. and R. ter Hofstede. 2008. Climate induced increases in species richness of
marine fishes. Global Change Biology 14:453-460.
Hilsenhoff, W.L. 1987. An improved biotic index of organic stream pollution. Great Lakes
Entomologist 20: 31-39.
Hubbard, M.D., and W.L. Peters. 1978. Environmental Requirements and Pollution Tolerance of
Ephemeroptera. Report No. EPA-600/4-78-061. U.S. EPA, Washington, D.C. 468 p.
Huff, D.D., S.L. Hubler, Y. Pan and D.L. Drake. Detecting Shifts in Macroinvertebrate
Community Requirements: Implicating Causes of Impairment in Streams. 2008. DEQ06-
LAB-0068-TR. Oregon Department of Environmental Quality, Watershed Assessment, and
Portland State University, Portland, OR.
Hughes, R.N., S.G. Paulsen, and J.L. Stoddard. 2000. EMAP-Surface Waters: a
multiassemblage, probability survey of ecological integrity in the U.S.A. Hydrobiologia
422/423: 429-443.
Hughes, R.N., D.P. Larsen, and J.M. Omernik. 1986. Regional reference sites: a method for
assessing stream potentials. Environmental Management 10(5):629-635.
Hurrell, J.W. 1995. Decadal trends in the North Atlantic Oscillation: Regional temperatures
and precipitation. Science 269:676-679.
Idaho Department of Environmental Quality. 2002. Water Body Assessment Guidance Second
Edition - Final. Internet site:
http://www.deq.state.id.us/water/data reports/surface water/monitoring/wbag 02 entire.pdf
Indicators ofHydrologic Alteration. Version 7.0.4.0. Copyright 1996-2007, The Nature
Conservancy.
Intergovernmental Panel on Climate Change (IPCC). 2000. Emissions Scenarios. A Special
Report of Working Group III of the Intergovernmental Panel on Climate Change.
Cambridge University Press, UK. http://www.grida.no/climate/ipcc/emission/091 .htm
IPCC (Intergovernmental Panel on Climate Change). (2007) Climate change 2007: the physical
science basis. In: Solomon, S; Qin, D; Manning, M; et al.; eds. Contribution of Working
Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate
Change. Cambridge, UK: Cambridge University Press.
Jacobson, G.L., I.J. Fernandez, P.A. Mayewski, and C.V. Schmitt (editors). 2009. Maine's
Climate Future: An Initial Assessment. Orono, ME: University of Maine.
December 23, 2010
8-9
External Review Draft

-------
Johnson, R.K. and D. Hering. 2009. Response of taxonomic groups in streams to gradients in
resource and habitat characteristics. Journal of Applied Ecology 46:175-186.
Johnson, R.K., D. Hering, M.T. Furse, and P.F.M. Verdonschot. 2006. Indicators of ecological
change: comparison of the early response of four organism groups to stress gradients.
Hydrobiologia 566:139-152.
Karr, J.R. 2006. Seven Foundations of Biological Monitoring and Assessment. Biologia
Ambientale, 20 (2): 7-18.
Karr, J.R. and E.W. Chu. 2000. Sustaining living rivers. Hydrobiologia 422/423:1-14.
Karr, J.R. 1991. Biological integrity: A long-neglected aspect of water resource management.
Ecological Applications 1: 66-84.
Kaushal, S.S., P.M. Groffman, L.E. Band, C.A. Shields, R.P. Morgan, M.A. Palmer, K.T. Belt,
C.M. Swan, S.E.G. Findlay, and G.T. Fisher. 2008. Interaction between Urbanization and
Climate Variability Amplifies Watershed Nitrate Export in Maryland. Environmental
Science and Technology
Keddy P.A. 1992. A pragmatic approach to functional ecology. Functional Ecology 6: 621-626.
Lamouroux, N, S. Doledec, and S. Gayraud. 2004. Biological traits of stream macroinvertebrate
communities: effects of microhabitat, reach, and basin filters. J. N. Am. Benthol. Soc.
23 (3):449-466.
Lawton, J.H. 1999. Are there general laws in ecology? Oikos 84: 177-192.
Lazorchak, J.M. B.H. Hill, D.K. Averill, D.V. Peck, and D. J. Klemm. 2000. Environmental
Monitoring and Assessment Program -Surface Waters: Field Operations and Methods for
Measuring the Ecological Condition of Non-Wadeable Rivers and Streams U.S.
Environmental Protection Agency, Cincinnati OH.
Lazorchak, J. M., Klemm, D. J., and D. V. Peck (editors). 1998. Environmental Monitoring and
Assessment Program -Surface Waters: Field Operations and Methods for Measuring the
Ecological Condition of Wadeable Streams. EPA/620/R-94/004F. US Environmental
Protection Agency, Washington, D.C.
Lind, P.R., B.J . Robson, and B. D. Mitchell. 2006. The influence of reduced flow during a
drought on patterns of variation in macroinvertebrate assemblages across a spatial hierarchy
in two lowland rivers. Freshwater Biology 51, 2282-2295.
Ling, S.D., Johnson, C.R., Ridgway, K., Hobday, A. J., Haddon, M. 2009. Climate driven range
extension of a sea urchin: inferring future trends by analysis of recent population dynamics.
Global Change Biology 15(3):719-731.
December 23, 2010
8-10
External Review Draft

-------
Mantua, N.J. and S.R. Hare, Y. Zhang, J.M. Wallace, and R.C. Francis. 1997. A Pacific
interdecadal climate oscillation with impacts on salmon production. Bulletin of the
American Meteorological Society, 78, pp. 1069-1079. Available online:
http://www.atmos.washington.edu/~mantua/abst.PDO.html
Marchant, R., L. Barmuta, and B.C. Chessman. 1995. A preliminary study of the ordination and
classification of macroinvertebrate communities from running waters in Victoria, Australia.
Austral. J. Mar. Freshw. Res. 45:945-962.
Matthews, W.J. 1998. Patterns in freshwater fish ecology. Chapman & Hall, NY.
McGill B.J., B.J. Enquist, E. Weiher and M. Westoby. 2006. Rebuilding community ecology
from functional traits. Trends in Ecology and Evolution 21: 178-185.
Mearns, L.O., F. Giorgi, L. McDaniel, and C. Shields. 2003. Climate Scenarios for the
Southeastern U.S. Based on GCM and Regional Model Simulations. Climatic Change 60:7-
35.
Menendez, R. 2007. How are insects responding to global warming? Tijdschrift voor
Entomologie 150: 355-365.
Merritt R.W. and K.W. Cummins (eds.). 1978. An introduction to the aquatic insects of North
America. Kendall/ Hunt, Dubuque, Iowa.
Mieszkowska, N., M.A. Kendall, S.J. Hawkins, R. Leaper, P. Williamson, N.J. Hardman-
Mountford, and A.J. Southward. 2006. Changes in the range of some common rocky shore
species in Britain - a response to climate change? Hydrobiologia 555:241-251.
Merritt R.W. and K.W. Cummins. 1988. An introduction to the aquatic insects of North
America. 2nd ed. Kendall/Hunt, 758p
Merritt, R. W. and K. W. Cummins. 1996 (eds.). An introduction to the aquatic insects of North
America (3rd edition). Kendall/Hunt Publ. Co., Dubuque, Iowa. USA. 862 p
Merritt, R. W., K. W. Cummins, and M. B. Berg (eds). 2007. An Introduction to the Aquatic
Insects of North America (4th ed.). Kendall/Hunt Publ. Co., Dubuque, IA 1158 pp.
Milly, P.C.D., J. Betancourt, M. Falkenmark, R.M. Hirsch, Z.W. Kundzewicz, D.P. Lettenmaier,
R. J. Stouffer. 2008. Stationarity Is Dead: Whither Water Management? Science 319:573-
574.
Milly, P. C. D., K. A. Dunne and A. V. Vecchia. 2005. Global pattern of trends in streamflow
and water availability in a changing climate. Nature. 438: 347-350.
Minshall, G.W., R. C. Petersen, and C. F. Nimz. 1985. Species richness in streams of different
size from the same drainage basin. American Naturalist 125:16-38.
December 23, 2010
8-11
External Review Draft

-------
Mohseni, O. H.G. Stephan, and J.G. Eaton. 2003. Global warming and potential changes in fish
habitat in U.S. streams. Climate Change 59:389-409.
Moline, M.A., H. Claustre, T.K. Frazer, O. Schofield, and M. Veret. 2009. Alteration of the food
web along the Antarctic Peninsula in response to a regional warming trend. Global Change
Biology 10:1973-1980.
Montes-Hugo, M., S. C. Doney, H.W. Ducklow, W. Fraser, D. Martinson, S. E. Stammerjohn,
and O. Schofield. 2009. Recent Changes in Phytoplankton Communities Associated with
Rapid Regional Climate Change Along the Western Antarctic Peninsula. Science 323:1470-
1473.
Moog, O., A. Schmidt-Kloiber, T. Ofenbock, and J. Gerritsen. 2004. Does the ecoregion
approach support the typological demands of the EU 'Water Framework Directive'?
Hydrobiologia 516: 21-33.
Moss, D., M.T. Furse, J.F. Wright, and P.D. Armitage. 1987. The prediction of the macro-
invertebrate fauna of unpolluted running-water sites in Great Britain using environmental
data. Freshw, Biol. 17:41-52.
Mote, P.W. 2003. Trends in snow water equivalent in the Pacific Northwest and their climatic
causes. Geophysical Research Letters 30(12): 1601.
MouillotD., S. Spatharis, S. Reizopoulou, T. Laugier, L. Sabetta, A. Basset and T.D. Chi T.D.
2006. Alternatives to taxonomic-based approaches to assess changes in transitional water
communities. Aquatic Conservation: Marine and Freshwater Ecosystems 16: 469-482.
Mouthon J. and M. Daufresne. 2006. Effects of the 2003 heatwave and climatic warming on
mollusc communities of the Saone: a large lowland river and of its two main tributaries
(France). Global Change Biology, 12, 441-449.
Moulton, S.R., II, J.L. Carter, S.A. Grotheer, T.F. Cuffney and T.M. Short. 2000. Methods
foranalysis by the U.S. Geological Survey National Water Quality Laboratory—Processing,
taxonomy,and quality control of benthic macroinvertebrate samples: U.S. Geological Survey
Open-File ReportOO-212, 49 p.
Murphy, J.F., C.E. Davies, and T.A. Corbin. 2007. Developing indicators to climate-induced
change in UK river ecosystems, p. 8-14 In M. Kernan, R.W. Battarbee, and H.A. Binneu
(eds). Climate change and aquatic ecosystems in the UK: science, policy and management.
UCL Environmental Change Research Center.
Nakicenovic, N. and R. Swart, eds. 2000. Special report on emissions scenarios.
Intergovernmental Panel on Climate Change (IPCC). Cambridge UK: Cambridge University
Press. Available online at
http://www.grida.no/publications/other/ipcc_sr/?src=/climate/ipcc/emission/.
December 23, 2010
8-12
External Review Draft

-------
NC Division of Water Quality (NCDWQ). 2005. Identification Methods for the Origins of
Intermittent and Perennial Streams, Version 3.1. North Carolina Department of
Environment and Natural Resources, Division of Water Quality. Raleigh, NC.
NCDENR Memo by Trish MacPherson and Bryn Tracy. 2005. Post Hurricanes Frances, Ivan,
and Jeanne Biological Monitoring (French Broad and Watauga River Basins) Biological
Sampling, November 30-December 2, 2004.
NCDENR Memo by Kathy Herring. 2004 . Effects of Long Term Drought on Benthic
Macroinvertebrate Communities in NC Streams and Tracking Their Recovery, 2002-2004.
New England Regional Assessment Group. 2001. Preparing for a Changing Climate: The
Potential Consequences of Climate Variability and Change. New England Regional
Overview, U.S. Global Change Research Program, 96 pp., University of New Hampshire.
Nijboer, R.C., R. K. Johnson, P.F.M. Verdonschot, M. Sommerhauser, and A. Buffagni. 2004.
Establishing reference conditions for European streams. Hydrobiologia 516:91-105.
Norris, R.H. and M.T. Barbour. 2009. Bioassessment of Aquatic Ecosystems. In: Gene E.
Likens, (Editor) Encyclopedia of Inland Waters, volume 3, pp. 21-28. Oxford: Elsevier.
Norris, R.H. and M.C. Thorns. 2000. What is river health? Freshwater Biology 41:197-209.
North Carolina Department of Environment and Natural Resources (NCDENR), Division of
Water Quality Environmental Sciences Section. 2006. Standard Operation Procedures for
Benthic Macroinvertebrates Biological Assessment Unit.
North Carolina Department of Environment and Natural Resources, Division of Water Quality,
Environmental Sciences Section. 2006. Standard Operating Procedures for Benthic
Macroinvertebrates - Biological Asssessment Unit. Internet link (accessed July 2009):
http://h2o.enr.state.nc.us/esb/BAUvvvvvv/benthossop.pdf
North Carolina Department of Water Quality. 2005. Final Report - Development of Policy for
Protection of Intermittent Streams through the 401 Water Quality Certification Program CD
974043-00-0 Version 1.2. Raleigh, NC.
NRC. 1991. Special Report 235. Highway Deicing: Comparing Salt and Calcium Magnesium
Acetate. Chapter 2. Road salt use in the United States. Transportation Research Board,
National Research Council. 1991. ISBN 0-309-05123-1.
Olden. J.D., N.L. Poff, and K. R. Bestgen. 2008. Trait synergisms and the rarity, extirpation, and
extinction risk or desert fishes. Ecology 89(3):847-856.
Ostermiller, Jeff. Unpublished presentation titled 'Development of a Biological Assessment
Framework.
December 23, 2010
8-13
External Review Draft

-------
Palmer, M.A., D.P. Lettenmaier, N.L. Poff, S.L. Postel, B. Richter, and R. Warner. 2009.
Climate Change and River Ecosystems: Protection and Adaptation Options. Environmental
Management. DOI 10.1007/s00267-009-9329-l.
Palmer, M.A., D. Lettenmaier, NL. Poff, S. Postel, B. Richter, and R. Warner. 2008. Chapter 6
Wild and Scenic Rivers. In S.H. Julius and J. West (eds). Preliminary Review of Adaptation
Options for Climate-Sensitive Ecosystems and Resources: Final Report, Synthesis and
Assessment Product 4.4. U.S. Climate Change Science Program and the Subcommittee on
Global Change Research.
Parmesan, C. 2006. Ecological and evolutionary responses to recent climate change. Annual
Review of Ecology Evolution and Systematics 37:637-669.
Parmesan, C. and H. Galbraith. 2004. Observed impacts of global climate change in the U.S.
Prepared for the pew Center on Global Climate Change.
Paul, M.J. and J.L. Meyers. 2001. Streams in the urban landscape. Ann. Rev. Ecol. System.
32:333-365.
Paulsen, S.G., R.M. Hughes, and D.P Larsen. 1998. Critical elements in describing and
understanding our nation's aquatic resources. Journal of the American Water Resources
Association. 34(5):995-1005.
Persat, H., J.-M. Olivier and D. Pont. 1994. Theoretical habitat templets, species traits, and
species richness: fish in the Upper Rho" ne River and its floodplain. Freshwater Biology 31:
439-454.
Poff, N.L., M.I. Pyne, B.P. Bledsoe, C.C. Cuhaciyan, and D.M. Carlisle. 2010. Developing
linkages between species traits and multiscaled environmental variation to explore
vulnerability of stream benthic communities to climate change. Journal of the North
American Benthological Society 29(4): 1441-1458.
Poff, N.L., B.P. Bledsoe, and C.O. Cuhaciyan. 2006a. Hydrologic variation with land use across
the contiguous United States: Geomorphic and ecological consequences for stream
ecosystems. Geomorphology 79:264-285.
Poff, N.L., J.D. Olden, N.K.M. Vieira, D.S. Finn, M.P. Simmons, and B.C. Kondratieff 2006b.
Functional trait niches of North American lotic insects: traits-based ecological applications
in light of phylogenetic relationships. Journal of the North American Benthological Society
25(4):730-755
Poff, L.N. 2002. Ecological response to and management of increased flooding caused by
climate change. Phil. Trans. R. Soc. Lond. A 360, 1497-1510.
December 23, 2010
8-14
External Review Draft

-------
Poff, L.N., M.M. Brinson, and J.W. Day, Jr. 2002. Aquatic Ecosystems and global climate
change: potential impacts on inland freshwater and coastal wetland ecosystems in the
United States. Prepared for the Pew Center on Global Climate Change. 44pp.
Poff, N.L, M.M. Brinson and J.W. Day, Jr. 2002. Aquatic ecosystems & Global climate change
Potential Impacts on Inland Freshwater and Coastal Wetland Ecosystems in the United
States. Prepared for the Pew Center on Global Climate Change.
Poff, N. L. 1997. Landscape filters and species traits: towards mechanistic understanding and
prediction in stream ecology. Journal of the North American Benthological Society 16: 391—
409.
Poff, N. L., J.D. Allan, M.B. Bain, J.R. Karr, K.L. Prestegaard, B.D. Richter, R.E. Sparks, and
J.C. Stromberg. 1997. The natural flow regime: a paradigm for river conservation and
restoration. Bioscience 47(ll):769-784.
PRISM Group, Oregon State University, 2009. http://www.prismclimate.org Utah State
University. Source area: Utah Freeze Dates. Internet site:
http://climate.usurf.usu.edu/reports/freezeDates.php. Accessed April 23. 2009.
Rabeni, C.F. 2000. Evaluating physical habitat integrity in relation to the biological potential of
streams. Hydrobiologia 422/423:245-256.
Rankin, E.T. 2008. Recalibration of the Ohio Biocriteria: Initial Steps and Recommendations.
Prepared for Ohio EPA, Division of Surface Water by Institute for Local Government
Administration and Rural Development (ILGARD), Ohio University, The Ridges, Building
22, Athens, Ohio 45701.
Resh, V. H. 2008. Which group is best? Attributes of different biological assemblages used in
freshwater biomonitoring programs. Environmental Monitoring and Assessment 138:131—
138
Resh, V. H. 2007. Multinational, freshwater biomonitoring programs in the developing world:
Lessons learned from African and Southeast Asian river surveys. Environmental
Management, 39, 737-748.
Resh, V. and Rosenberg, D. 1984. The ecology of aquatic insects. New York, NY: Praeger.
Reynoldson, T. B., RH. Norris, V. H. Resh, K.E. Day, and D.M. Rosenberg. 1997. The
reference condition: a comparison of multimetric and multivariate approaches to assess
water-quality impairment using benthic macroinvertebrates. J. N. Am. Benthol. Soc.
16(4):833-852.
Ribera, I., S. Dole'dec, I. S. Downie and G. N. Foster. 2001. Effect of land disturbance and stress
on species traits of ground beetle assemblages. Ecology 82: 1112-1129.
December 23, 2010
8-15
External Review Draft

-------
Richards C., R.J. Haro, L.B. Johnson and G.E. Host. 1997. Catchment and reach-scale properties
as indicators of macroinvertebrate species traits. Freshwater Biology 37: 219-230.
Richter, BD; Baumgartner, JV; Powell, J; et al. 1996. A method for assessing hydrologic
alteration within ecosystems. Conservation Biology 10(4): 1163-1174.
Root, T. L., J. T. Price, K. R. Hall, S. H. Schneider, C. Rosenzweig, and J. A. Pounds. 2003.
Fingerprints of global warming on wild animals and plants. Nature 421:57-60.
Rosenberg, D. and V. Resh, 1993. Freshwater biomonitoring and benthic macroinvertebrates.
Chapman Hall Publishers.
Sandin, L. 2009. The relationship between land-use, hydromorphology and river biota at
different spatial and temporal scales: a synthesis of seven case studies. Fundamental and
Applied Limnology Archiv ftir Hydrobiologie 174(1): 1-5.
Sandin, L. and R.K. Johnson. 2000. The statistical power of selected indicator metrics using
macroinvertebrates for assessing acidification and eutrophication of running waters.
Hydrobiologia 422/423:233-243.
Schmidt-Kloiber, A., W. Graf, A. Lorenz, and O. Moog. 2006. The AQUEM-STAR taxa list - a
pan-European macro-invertebrate ecological database and taxa inventory. Hydrobiologia
566: 325-342.
Schueler, T. 1994. The importance of imperviousness. Watershed Protection Techniques 1: 100-
111.
Seager, R., M. Ting, I. Held, Y. Kushnir, J. Lu, G. Vecchi, H. Huang, N. Harnik, A. Leetmaa, N.
Lau, CI. Li, J. Velez, and N. Naik. 2007. Model Projections of an Imminent Transition to a
More Arid Climate in Southwestern North America. Science 316 (5828): 1181 - 1184.
Seebens, H., U. Cheinsle, and D. Straile. 2009. Copepod life cycle adaptations and success in
response to phytoplankton spring bloom phenology. Global Change Biology (2008), doi:
10.1111/j. 1365-2486.2008.01806.x.
Simpson, J.C., and R.H. Norris 2000. Biological assessment of river quality: development of
AUSRIVAS models and outputs, in Pages 125-142. Wright, J.F., D.W. Sutcliffe, and M.T.
Furse (editors). Assessing the biological quality of freshwaters: RIVPACS and other
techniques. Freshwater Biological Association, Ambleside, Cumbria, United Kingdom.
Snook, H., S.P. Davies, J. Gerritsen, B.K. Jessup, R. Langdon, D. Neils, E. Pizutto. 2007. The
New England Wadeable Stream Survey (NEWS): Development of Common Assessments in
the Framework of the Biological Condition Gradient. Prepared for USEPA Office of
Science and Technology, and USEPA Office of Watersheds Oceans and Wetlands
Washington, DC Washington, DC.
December 23, 2010
8-16
External Review Draft

-------
Southwood, T.R.E. 1977. Habitat, the templet for ecological strategies? Journal of Animal
Ecology 46: 337-365.
Southwood, T.R.E. 1988. Tactics, strategies and tern- plates. Oikos 52:3-18.
Stamp, J.D., A.T. Hamilton, L. Zheng, and B.G. Bierwagen. 2010. Use of thermal preference
metrics to examine state biomonitoring data for climate change effects. Journal of the North
American Benthological Society 29(4): 1410-1423.
Stanley, E.H., D.L. Buschman, A.J. Boulton, N.B. Grimm, and S.G. Fisher. 1994. Invertebrate
resistance and resilience to intermittency in a desert stream. Am.Midl. Nat. 131:288-3000.
State of Ohio Environmental Protection Agency, Environmental Assessment Section, Division of
Water Quality, Planning and Assessment. 1989 (last updated 2008). Biological Criteria for
the Protection of Aquatic Life: Volume III: Standardized Biological Field Sampling and
Laboratory Methods for Assessing Fish and Macroinvertebrate Communities. Internet link
(accessed July 2009): http://vvvvvv.epa.state.oh.us/dsvv/bioassess/BioCriteriaProtAqLife.html
Statzner, B., and L.A. Beche. 2010. Can biological invertebrate traits resolve effects of multiple
stressors on running water ecosystems? Freshwater Biology 55: 80-119.
Statzner B., A.G. Hildrew and V.H. Resh. 2001a. Species traits and environmental, constraints:
entomological research and the history of ecological theory. Annual Review of Entomology
46: 291-316.
Statzner B., B. Bis, S. Doledec, and P. Usseglio-Polatera. 2001b. Perspectives for biomonitoring
at large spatial scales: a unified measure for the functional composition of invertebrate
communities in European running waters. Basic and Applied Ecology 2:73-85.
Statzner B., S. Doledec and B. Hugueny. 2004. Biological trait composition of European stream
invertebrate communities: assessing the effects of various trait filter types. Ecography 27:
470-488.
Statzner B., V.H. Resh and S. Doledec. 1994. Ecology of the upper Rho"ne river: a test of habitat
templet theories. Freshwater Biology 31: 253-554.
Stepenuck, K.F., R.L. Crunkilton, M.A. Bozek, and L. Wang. 2008. Comparison of
macroinvertebrate-derived stream quality metrics between snag and riffle habitats. Journal
of the American Water Resources Association 44(3):670-678.
Stephan, H.G. and E.B. Preudhomme. 1993. Stream temperature estimation from air
temperature. Water Resour. Bull. 29:27-45.
Stevenson, R. J. Y. Pan, K.M. Manoylov, C.A. Parker, D.P. Larsen, and A.T. Herlihy. 2008.
Development of diatom indicators of ecological conditions for streams of the western US. J.
N. Am. Benthol. Soc., 2008, 27(4): 1000-1016.
December 23, 2010
8-17
External Review Draft

-------
Stewart, I.T., D.R. Cayan and M.D. Dettinger. 2004. Changes in snowmelt runoff timing in
western North America under a 'business as usual' climate change scenario. Climate
Change 62: 217-232.
Stoddard, J.L., D.P. Larsen, C.P. Hawkins, R.K. JOHNSON, and R.H. Norris. 2006. Setting
expectations for the ecological condition of streams: The concept of reference condition.
Ecological Applications 16(4): 1267-1276.
Stribling, J.B., B.K. Jessup, and D.L. Feldman. 2008. Precision of benthic macroinvertebrate
indicators of stream condition in Montana. J. N. Am. Benthol. Soc. 27(l):58-67.
Suren, A.M., and I.G. Jowett. 2006. Effects of floods versus low flows on invertebrates in a New
Zealand gravel-bed river. Freshwater Biology 51:2207-2227.
Suter, G.W. II, S.B. Norton, and S.M. Cormier. 2002. A methodology for inferring the causes of
observed impairments in aquatic ecosystems. Environmental Toxicology and Chemistry
21(6): 1101—1 111.
Surdick, R.F., and A.R. Gaufin. 1978. Environmental Requirements and Pollution Tolerance of
Plecoptera. Report No. EPA-600/4-78-062. U.S. EPA, Washington, D.C. 423 p.
Sweeney, B.W., J.K. Jackson, J.D. Newbold, and D.H. Funk. 1992. "Climate Change and the
Life Histories and Biogeography of Aquatic Insects in Eastern North America." In Global
Climate Change and Freshwater Ecosystems. P. Firth and S.G. Fisher, eds. Springer-Verlag,
New York, pp. 143-176.
Sweeney, B.W. and R.L. Vannote. 1978. Size variation and the distribution of hemimetabolous
insects: two thermal equilibrium hypotheses. Science 200(4340):444-446.
Tetra Tech, Inc. 2008. Draft Predictive Bioassessment Models for New Mexico Streams.
Prepared for New Mexico Environment Department, Santa Fe, NM by Tetra Tech, Inc.
Owings Mills, MD.
Thuiller, W. 2004a. Patterns and uncertainties of species' range shifts under climate change.
Global Change Biology 10, 2020-2027.
Thuiller, W., S. Lavorel, G. Midgley, S. Lavergne and T. Rebelo. 2004b. Relating plant traits and
species distributions along bioclimatic gradients for 88 Leucadendron taxa. Ecology 85:
1688-1699.
Thuiller W (2003) BIOMOD: optimising predictions of species distributions and projecting
potential future shifts under global change. Global Change Biology, 9, 1353-1362.
December 23, 2010
8-18
External Review Draft

-------
Townsend C.R., S. Dole'dec and M.R. Scarsbrook. 1997. Species traits in relation to temporal
and spatial heterogeneity in streams: a test of habitat template theory. Freshwater Biology
37: 367-387.
Townsend, C.R. and A.G. Hildrew. 1994. Species traits in relation to a habitat template for river
systems. Freshwater Biology 31:265-275.
Tuchman, N.C., R.G. Wetzel, S.T. Rier, K.A. Wahtera, and J.A. Teeri. 2002. Elevated
atmospheric C02 lowers leaf litter nutritional quality for stream ecosystem food webs.
Global Change Biology 8:163-170.
United States Environmental Protection Agency (USEPA). 2011. Freshwater Biological Traits
Database. (External Review Draft).
United States Environmental Protection Agency (USEPA). 2009a. Land-Use Scenarios:
National-Scale Housing-Density Scenarios Consistent with Climate Change Storylines
(Final Report). EPA/600/R-08/076F.
United States Environmental Protection Agency (USEPA). 2009b. State and County Land
Protection Programs: Review of Selection Processes and the Consideration of Climate
Change (External Review Draft).
United States Environmental Protection Agency (USEPA). 2008. Climate change effects on
stream and river biological indicators: A preliminary analysis. EPA/600/R-07/085. Global
Change Research Program, National Center for Environmental Assessment, Washington,
DC. Available from the National Technical Information Service, Springfield, VA, and
online at http://www.epa.gov/ncea.
United States Environmental Protection Agency (USEPA). 2007. The New England Wadeable
Stream Survey (NEWS): Development of Common Assessments in the Framework of the
Biological Condition Gradient. Prepared for USEPA Office of Science and Technology,
Washington DC.
United States Environmental Protection Agency (USEPA). 2006. Wadeable Streams
Assessment: A Collaborative Survey of the Nation's Streams. U.S. Environmental
Protection Agency, OWOW, Washington, D.C., EPA 841-B-06-002.
United States Environmental Protection Agency (USEPA). 2002a. Summary of Biological
Assessment Programs and Biocriteria Development for States, Tribes, Territories, and
Interstate Commissions: Streams and Wadeable Rivers. EPA-822-R-02-048. U.S.
Environmental Protection Agency, Office of Environmental Information and Office of
Water, Washington, D.C.
United States Environmental Protection Agency (USEPA). 2002b. Primary Distinguishing
Characteristics of Level III Ecoregions of the Continental United Stations. Accessed July
30, 2009 at the internet sites: http://www.epa.gov/wed/pages/ecoregions/level iii.htm.
December 23, 2010
8-19
External Review Draft

-------
United States Environmental Protection Agency (USEPA). 2000. Stressor identification
guidance document. Office of Water and Office of Research and Development, Washington
DC. EPA-822-B-00-025.
United States Environmental Protection Agency (USEPA). 1990. Freshwater Macroinvertebrate
Species List Including Tolerance Values and Functional Feeding Group Designations for
Use in Rapid Bioassessment Protocols. Prepared by: EA Mid Atlantic Regional Operations
Engineering, Science, and Technology, Inc. Prepared for: U.S. EPA Assessment and
Watershed Protection Division Washington, D.C. Report No. 11075.05
Urquhart, N. S., S. G. Paulsen and D. P. Larsen. 1998. Monitoring for policy-relevant
regional trends over time. Ecological Applications 8: 246-257.
Usseglio-Polatera, P ., M. Bournaud, P. Richoux and H. Tachet. 2000. Biomonitoring through
biological traits of benthic macroinvertebrates: how to use species trait databases?
Hydrobiologia 422/423:153-162.
Usseglio-Polatera, P. 1994. Theoretical habitat templets, species traits, and species richness:
aquatic insects in the Upper R" ho" ne and its floodplain. Freshwater Biology 31: 417-437.
Utah Division of Water Quality. 2006. Monitoring Manual. Internet link (accessed July 2009):
http://www.waterqualitv.utah.gov/Monitoring/06 DWO monitoring manual.pdf
Utah State University, The Western Center for Monitoring and Assessment of Freshwater
Ecosystems, Department of Aquatic, Watershed, and Earth Resources. Internet site accessed
April 23, 2009. Source areas: Predictive Models. Internet site:
http://129.123.10.240/WMCPortal/DesktopDefault.aspx?tabindex=2&tabid=27
Van Duinen. 2006. Biological traits successfully predict the effects of restoration management
on macroinvertebrates in shallow softwater lakes. Hydrobiologia 565: 201-216.
Vannote, R.L. and B.W. Sweeney. 1980. Geographic analysis of thermal equilibria: a
conceptual model for evaluating the effects of natural and modified thermal regimes on
aqualtic insect communities. American Naturalist 115(5):667-695.
Verdonschot, P.F.M. 2006. Evaluation of the use of Water Framework Directive typology
descriptors, reference sites and spatial scale in macroinvertebrate stream typology.
Hydrobiologia (2006) 566:39-58.
Verdonschot, P.F.M. and O. Moog. 2006. Tools for assessing European streams with
macroinvertebrates: major results and conclusions from the STAR project. Hydrobiologia
566:299-309.
Verberk, W.C.E.P., H. Siepel and H. Esselink. 2008a. Life-history strategies in freshwater
macroinvertebrates. Freshwater Biology 53: 1722-1738.
December 23, 2010
8-20
External Review Draft

-------
Verberk, W.C.E.P., H. Siepel and H. Esselink. 2008b. Applying life-history strategies for
freshwater macroinvertebrates to lentic water. Freshwater Biology 53: 1739-1753.
Vieira, N.K.M., N.L. Poff, D.M. Carlisle, S.R. Moulton II, M.K. Koski, and B.C.Kondratieff
2006. A database of lotic invertebrate traits for North America: U.S. Geological Survey
Data Series 187. Available at: http://pubs.water.usgs.gov/dsl87.
Wallin, M., T. Wiederholm, and R.K. Johnson . 2003. Guidance on establishing reference
conditions and ecological status class boundaries for inland surface waters (final). Produced
by CIS Working Group 2.3 -REFCOND. final version 7.0, 2003-03-05.
Walther, G.-R., E. Post, P. Convey, A. Menzel, C. Parmesan, T. J. C. Beebee, J.-M. Fromentin,
O. Hoegh-Guldberg, and F. Bairlein. 2002. Ecological responses to recent climate change.
Nature 416:389-395.
Walther, G. R., S. Berger, and M. T. Sykes. 2005. An ecological 'footprint' of climate change.
Proceedings of the Royal Society B-Biological Sciences 272:1427-1432.
Wang, L., J. Lyons, P. Kanehl and R. Bannerman. 2001. Impacts of urbanization on stream
habitat and fish across multiple spatial scales. Environmental Management 28: 255-266.
Webb, J.A., M. J . Stewardson, and W.M. Koster. 2009. Detecting ecological responses to flow
variation using Bayesian hierarchical models. Freshwater Biology, doi: 10.1111/j .1365-
2427.2009.02205.x.
Wiggins, G.B. and R.J. Mackay. 1978. Some relationships between systematics and trophic
ecology in nearctic aquatic insects, with special reference to Trichoptera. Ecology 59:1211—
1220.
Woodward, G, J. I. Jones, and A.G. Hildrew. 2002. Community persistence in Broadstone
Stream (U.K.) over three decades. Freshwater Biology 47:1419-1435.
Wright, J.F. 1995. Development and use of a system for predicting the macroinvertebrate fauna
in flowing waters. Australian J. Ecol. 20:181-197.
Wright, J.F. 2000. An introduction to RIVPACS. in Pages 1-24. Wright, J.F., D.W. Sutcliffe, and
M.T. Furse (editors). Assessing the biological quality of freshwaters: RIVPACS and other
techniques. Freshwater Biological Association, Ambleside, Cumbria, United Kingdom.
Wright, J.F., M.T. Furse, and P.D. Armitage. 1993. RIVPACS: a technique for evaluating the
biological water quality of rivers in the UK. Euro. Wat. Poll. Control 3:15-25.
Wright, J.F., D. Moss, P.D. Armitage, and M.T. Furse. 1984. A preliminary classification of
running-water sites in Great Britain based on macroinvertebrate species and the prediction
of community type using environmental data.. Freshwater. Biol. 14:221-256.
December 23, 2010
8-21
External Review Draft

-------
Wu, J. and H. Li. 2006. Perspectives and methods of scaling, pp. 17-44. In J. Wu, K.B. Jones,
H. Li, O.L. Loucks (Eds). Scaling and Uncertainty Analysis in Ecology: Methods and
Applications.
Wuebbles, D.J. and K. Hayhoe. 2004. Climate Change Projections for the United States
Midwest. Mitigation and Adaptation Strategies for Global Change 9(4): 335-363.
Yoder, C.O. and M.T. Barbour. 2009. Critical technical elements of state bioassessment
programs: a process to evaluate program rigor and comparability. Environ Monit Assess
(2009) 150:31-42.
Yoder, C.O. and 9 others. 2005. Changes in fish assemblage status in Ohio's nonwadeable
rivers and streams over two decades, pp. 399-429. in R. Hughes and J. Rinne (eds.).
Historical changes in fish assemblages of large rivers in the America's. American Fisheries
Society Symposium Series.
Yoder, C.O. and E.T. Rankin. 1998. The role of biological indicators in a state water quality
management process. Environmental Monitoring and Assessment 51:61-88.
Yuan, L. 2006b. Predictive models for bioassessment of large and great rivers. National Center
for Environmental Assessment, Office of Research and Development, US EPA
Yuan, L.L. 2004. Assigning macroinvertebrate tolerance classifications using generalized
additive models. Freshwater Biology 49:662-677.
Yuan, Lester L. 2006a. Theoretical predictions of observed to expected ratios in RIVPACS-type
predictive model assessments of stream biological condition. Journal of the North American
Benthological Society 25: 841-850.
Yuan, Lester. 2006b. Estimation and Application of Macroinvertebrate Tolerance Values. Report
No. EPA/600/P-04/116F. National Center for Environmental Assessment, Office of
Research and Development, U.S. Environmental Protection Agency, Washington, D.C.
Zhang, Y., J.M. Wallace, D.S. Battisti. 1997. ENSO-like interdecadal variability: 1900-93. J.
Climate, 10, 1004-1020.
Zuckerberg, B., A.M. Woods, and W,F. Porter. 2009. Poleward shifts in breeding bird
distributions in New York State. Global Change Biology (2009), doi: 10.1111/j. 1365-
2486.2009.01878.x.
December 23, 2010
8-22
External Review Draft

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
APPENDIX A
Basic Evidence for Climate Change Effects:
Long-term Trends in Annual Air Temperature,
Precipitation, and Water temperatures
There is reasonable evidence of climate change effects on both terrestrial and aquatic
biological assemblages at various levels, including changes in ecosystem process, community
composition, phenology of populations, number of reproductive cycles, evolutionary adaptations,
and genetic selection (e.g., Parmesan and Galbraith, 2004; Root et al., 2003; Poff et al., 2002;
Walther et al., 2002). More recently, there are also documented responses in freshwater
ecosystems (Chessman, 2009; Buisson et al., 2008, Hiddink and Hofstede, 2008; Collier, 2008;
Durance and Ormerod, 2007; Daufresne and Boet, 2007).
Stream water temperature regimes will be altered by air temperature increases and
modified by other influences (Cassie et al., 2006; Mohseni et al., 2003; Daufresne et al., 2003;
Hawkins et al., 1997). Temperature regimes determine the distribution and abundance of aquatic
species through temperature tolerances and evolutionary adaptations, along with competitive
interactions, effects on food supply, and other factors (e.g., Matthews, 1998; Hawkins et al.,
1997; Vannote and Sweeney, 1980; Sweeney and Vannote, 1978). Changes in prevailing
temperature regime, as well as climate change-associated increases in variability of temperature,
may have various biological effects.
Evidence for climate change effects can be pursued within both abiotic (e.g., temperature,
precipitation, flow) and biotic components of the environment (Figure A-l). Examples from each
category (climate change projections, stream changes, ecological responses) were examined for
existing evidence of climate change effects.

-------
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Responses
Habitat Responses
Climate
Change
Projections
Stream Changes
• Water chemistry,
Air temp
Precipitation
Hydrologic Regime
•	Magnitude
•	Frequency
•	Duration
•	Timing
Thermal Regime
• Substrate, other
phys/chem
nutrients, turbidity
characteristics
Ecological Responses
•	Structural
•	Functional
Figure A-l. Mechanisms linking climate changes with streams.
A.l Trends and Variability in Annual Air Temperature and Precipitation using PRISM
One way to detect whether climatic changes have occurred in a region that could have
implications for aquatic organisms is to examine air temperatures. There is a general
correspondence between air and stream temperatures, though the magnitude and seasonal
patterns of changes in stream water temperatures are likely to vary regionally, due to factors
including water source influences, watershed characteristics, and season (Caissie, 2006;
Daufresne et al., 2003). Stephan and Preudhomme (1993) estimated a linear relationship (factor
of 0.86 in °C) between weekly average water and air temperatures for eleven streams in the
Mississippi River Basin; and a similar linear relationship has been applied by others (e.g., Eaton
and Scheller, 1996). However, Mohseni et al. (2003) suggest the relationship between air and
water temperatures is better explained by an S-curve, such that at higher air temperatures, stream
temperature increases level off due to evaporative cooling.
Below are plots of Parameter-elevation Regressions on Independent Slopes Model
(PRISM) (PRISM Climate Group, Oregon State University, Corvallis, Oregon;
http://www.prismclimate.org, data) mean annual air temperature values1 at biological sampling
sites in each of the three states examined (Maine, Utah, and North Carolina). PRISM uses a
1 maximum and minimum air temperature values were averaged to derive what we refer to as
mean annual air temperature.

-------
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
digital elevation model and point measurements of climate data to generate estimates of annual,
monthly, and event-based climatic elements.
Annual air temperatures in these states have increased gradually from 1974 and 2006.
Stations in Utah showed the strongest trend and experienced the greatest increase in air
temperature (about 2 °C, r2=0.42) (Figure A-lb), with Maine and North Carolina showing
weaker trends (about 1 °C, r2=0.15 and r2=0.11, respectively) (Figure A-2a and A-2c). Absolute
air temperatures within each state differed by ecoregion, but the change in air temperatures (e.g.,
the increasing trends) were similar across ecoregions. Maine had the greatest difference, where
the Northeastern Highlands ecoregion had a stronger upward trend than the other two ecoregions
(r2=0.23 versus r2=0.12 and 0.13) (Figure A-2a). This ecoregion is at a higher elevation than the
other two as well.
Plots of annual precipitation patterns from PRISM data are also displayed below. Trends
in those patterns were highly variable and were not significantly correlated with year in any of
the states (r2 values ranging from 0.004 to .01; see figure A-3). The amount of annual
precipitation across ecoregions within a state often varied quite a bit. For instance, the mountain
ecoregions in both Utah and North Carolina had higher annual precipitation than the plateau or
coastal regions (true also for Maine, but to a lesser extent) (Figure A-4). However, mean annual
precipitation values in all ecoregions were highly variable over the 30 years analyzed (Figure A-
5).
From 1974 to 2006, fluctuations between years in temperature and precipitation have also
been highly variable. However, in Utah, the differences between consecutive years (i.e., current
year minus previous year) in both air temperature and precipitation have declined. Precipitation
showed a stronger trend in this than temperature (r2=0.12 temperature, r2=0.28 precipitation)
(Figures A-6b and A-7b). Unlike Utah, the trends in inter-annual climate variability in Maine and
North Carolina showed almost no trend for both annual air temperature and precipitation.

-------
Maine PRISM mean annual air temperature ("C)
73
74
75
76
77
YearAII Stations: r * 0.3845. p = 0.0272: r « 0.1478
2 7.2

5 6.0
1970 1975 1980 1985 1990 1995 2000 2005
Utah PRISM mean annual air temperature (*C)
0
«j
2
2
<5 7.5
£X
£
1	7.0
S 6.5
£
5
6.0
< 5.5
5.0
Year:AII_Stations: r = 0.6495. p ¦ 0.00004: r2 = 0.4219
C)
1970 1975 1980 1985 1990 1995 2000 2005 2010
Year
North Carolina PRISM mean annual air temperature (*C)

15.8

15.6
p


15.4
£

2
15.2
I)

§¦ 15.0
V

CO
14.8
w

9
c
146
CD

c
(0
14.4
E

5
142
CO

a.
a
14.0
4>

—
138
3j


13.6

13.4

13.2
| YearAII Stations: r = 0.3258. p = 0 0643: r = 0.1061
. o o
1970 1975 1980 1985 1990 1995 2000 2005 2010
Year
Figure A-2. Plots of PRISM mean annual air temperature (°C) values (averaged across all
stations) for Maine (a), Utah (b) and North Carolina (c).

-------
a)
Maine PRISM mean annual air temperature (°C)
YeanLaurentian: r = 0.3434. p = 0.0504; r = 0.1179
Year:NE Coastal: r = 0.3576, p = 0.0410; r2 = 0.1279
Year:NE Highlands: r = 0.4816, p = 0.0045; r2 = 0.2319
b)
1970 1975 1980 1985 1990 1995 2000 2005 2010
Year
Utah PRISM mean annual air temperature (°C)
Laurentian Plains & Hills
Northeast Coastal Zone
""V Northeastern Highlands
11
p 10
0>
3 9
CO
a>
Q.
I 8
'co
I 7
c
c
CO
£ 6
CO
a>
£
1 5
lY.
CL
Year:Colorado Plateau: r = 0.6636, p = 0.00003; r2 = 0.4404
Year:Wasatch Uinta: r = 0.6666, p = 0.00002; r2 = 0.4443
<



O
o

o
o
o
0

o


r—
o
o
o
o
0

o o 	



	
o
Q
o o

o o
o


o






~
~

~ D
~ n		—


~



~ _n^			

D ~





rfi	~

~
~
~
i
~
]°n



~

„ ~
~
~

~
~


3
1970 1975 1980 1985 1990 1995 2000 2005 2010 ^ Wasatch Uirta Mountains
Year
Colorado Plateau

-------
North Carolina PRISM mean annual air temperature (°C)
18
O
~ 17
a 16
15
14
% 13
fe 12
yi uiiiiti rmoivi mean annual an leinpciaiuie
>
= 0.0874
o ~
Coastal
>ov Mountain
1970 1975 1980 1985 1990 1995 2000 2005 2010 pjedmont
Year
Figure A-3. Plots of PRISM mean annual air temperature (°C) values (averaged across
each major ecoregion) for Maine (a), Utah (b) and North Carolina (c).

-------
a)
Maine PRISM mean anmial precipitation (inches)
65
2 50
45
E 40
5
<£>
(XL
a 35
Sv
to
¦5 30
| Year:AII_5tatiortsi r - 0.1021. p = 0.5718: r - 0.0104 |
25 	
1970
1975 1980 1965
1990
Year
1995 2000 2005 2010
Utah PRISM mean annual precipitation (inches)
| Year:All_Statkms: r = Q.0808. p = 0.6551; r; = 0.0Q65
1970 1975 1980 1985 1990 1995 2000 2005 2010
North Carolina PRISM mean annual precipitation linchest
Year:AI Stations: r = -0.0592. p = 0.7436: r * 0.0035
1970 1975 1980 1985 1990 1995 2000 2005 2010
Figure A-4. Plots of PRISM mean annual precipitation (inches) values (averaged across
stations) for Maine (a), Utah (b)_and North Carolina (c).

-------
Maine PRISM mean annual precipitation (inches)
70
65
60
55
50
45
40
35
30
25
Year: Laurentian: r
= 0.0672, p
=
0.7102: r2 =
0.0045
Year:NE Coastal: i
= 0.1308, p
=
0.4680: r2 =
0.0171
Year:NE Highlands:
r = 0.1724,
P
= 0.3374: r2
= 0.0297

~
~



~
~
e

o
D 8
8
o ft
~
c
~
9° :
0 o
o 		
"7—
n ((
~
~
~ O O
ft J—-=r
OD
o O
0°
6
		T3 n	
~ ° o ~
^8° o o
~ ° *
o o
° ~
o 5
o
o
Laurentian Plains & Hills
ME Coastal Zone
1970 1975 1980 1985 1990 1995 2000 2005 2010 Northeastern Highlands
Year
Utah PRISM mean annual precipitation (inches)
Year:Colorado Plateau: r = 0.1435, p = 0.4257; r2 = 0.0206
YearWasatch Uinta: r = 0.0661. p = 0.7146; r2 = 0.0044
~
~ ~
~ ~
~
36
34
32
30
28
26
24
22
20
18
16
14
12
10
1970 1975 1980 1985 1990 1995 2000 2005 2010
Year
~ ~
~ O o O
oo
Colorado Plateau
Wasatch Uinta Mountains

-------
0)
North Carolina PRISM mean annual precipitation (inches)
75
70
65
60
55
50
45
40
35
YearCoastal: r = 0.0824, p = 0.6484; r2 = 0.0068
Year:Mountain: r = -0.0896, p = 0.6201; r2 = 0.0080
Year: Piedmont: r = -0.0948, p = 0.5997; r2 = 0.0090
~ ~ o
o o
o ~
30
1970 1975 1980 1985 1990 1995 2000 2005 2010
Year
Coastal
^ Mountain
Piedmont
Figure A-5. Plots of PRISM mean annual precipitation (inches) values (averaged across
each major ecoregion) for Maine (a), Utah (b) and North Carolina (c).

-------
97
Maine Trends in Cimate VariabiSty - PRISM Annual Air Temperature
98
99
100
101
102
103
104
Yeaf .AII_Stanons: r = -0.1369. p = 0.4550: r = 0.01871
1970 1975 1980 1985 1990 1995 2000 2005 2010
Utah Trends in Climale Variability • PRISM Annual Air Temperature
YearVar Al; r - -0 3404. p » 0.0566; r - 0.1158
1970 1975 1980 1985 1990 1995 2000 2005 2010
North Carolina Trends in Climate VariabiSty - PRISM Annual AJr Temperature
Year Var All. r = -0.0058. p = 0.9747; r- = 0.0000
1970 1975 1980 1985 1990 1995 2000 2005 2010
Figure A-6. Trends in Climate Variability - PRISM Annual Air Temperature. Values
represent the absolute difference between average PRISM mean annual temperature
values (current year - previous year) for all stations in Maine (a), Utah (b) and North
Carolina (c).

-------
105
Maine Trends in Climate Variability - PRISM Annual Precipitation
106
107
Year:All_Stattons: r » -0.0198, p = 0.9T45; r1 = 0-00041
5 >. 14
55 *-
O a
1970 1975 1980 1985 1990 1995 2000 2005 2010
Utah Trends in Cimate Variability - PRISM Annual Precipitation
I Year:Var_AJt: r a -0.5278. p 0.0019; r- * 0.2786
ra
ai
North Carolina Trends in Cimate Vanabitty - PRISM Annual Precipitation
Year:var_AJi: r = -0,0497. p * 0.7870: r = 0.0025
1975 1980 1985 1990 1995 2000 2005 2010
108
109
110
111
112
Figure A-7. Trends in Climate Variability - PRISM Annual Precipitation. Values
represent the absolute difference between average PRISM mean annual precipitation
values (current year - previous year) for all stations in Maine (a), Utah (b) and North
Carolina (c).

-------
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
A.2. Long-term water temperature trends at USGS gage stations
Data from USGS gages with long-term water temperature records (30 years) were
compiled. Initially a screening process, outlined in Figure A-8, was applied to minimize the
likelihood of confounding effects (e.g., sewage treatment plant discharges, heavy urban/suburban
development, effects of dam releases), or temporal discontinuities from methods or data quality
issues. However, screening criteria had to be relaxed in certain regions because stations were not
meeting all the criteria. To expand stream site coverage nationwide, sites that did not meet all of
these criteria had to be added to the list (i.e. Colorado River sites were used even though they are
higher order and have dams, but this was the best data available for this region). Data were
downloaded from the USGS real-time water data website: http://waterdata.usgs.gov/nwis/rt.
About 25 stations were evaluated for trends. Plots of seasonal means, minimum, and maximum
temperatures were developed to partition seasonal variation when checking for long-term
patterns. Summer temperatures generally showed greater trends, and were used to evaluate rates
of temperature change per 10-year period at 23 of the stations (Table A-l).
USGS stations with long term (-30 years) temperature data (-138 stations)
Screen by data quality
data calculated as mean/max/min ratherthan instantaneous reading, other
temperature in fahrenheit (not mixed with Celsius)
no abrupt changes (e.g., flagging uncorrected gage relocations, etc)
Screen for nationwide geographic distribution
Screen by land use and
no treatment plants <5 mile
no upstream Dam
stream order(<=5)
low urban land uses
Final stream sites evaluated
Figure A-8. Flow chart showing the screening process that was followed when determining
which USGS stations to use in the water temperature trend analyses.

-------
Table A-l. Summary of results from water temperature trend analyses at 23 USGS stations that met the screening criteria.
Rates of temperature (°C) change per 10-year period were evaluated at 23 of the stations.
Site #
Stream Name
Stream
Order
NPDES
Land Use
State
TempA/
lOyear
R2
2423130
Cahaba River
3
no
FOR/AG (URB)
AL
0.73
0.024
10339400
Martis Creek
3
no
FOR
CA
0.28
0.02
7086000
Cache Creek
2
no
FOR
CO
1.48
0.151
9169500
Dolores River
5
no

CO
0.93
0.05
2266300
Reedy Creek
3
no
URB
FL
0.3
0.081
5474000
Skunk River
6
no
FOR
IA
0.25
0.006
13340600
Beaver Creek
4
no

ID
0.4
0.032
3354000
White River
5
no
AG
IN
0.32
0.017
1600000
North Branch Potomac River
5
no

MD
0.5
0.013
1021050
Saint Croix River
6
no
URB/FOR
ME
0.39
0.02
12363000
Flathead River
6
no
AG (URB)
MT
1.36
0.17
2077200
Hyco Creek
3
no
FOR
NC
0.7
0.192
6338490
Missouri River
1
no
GRASSLAND
ND
5.09
0.508
5056000
Sheyenne River
4
no
GRASSLAND
ND
0.41
0.013
5058700
Sheyenne River
1
no
GRASSLAND
ND
0.43
0.018
1466500
McDonalds Branch
1
no
FOR
NJ
0.33
0.03

-------
1428500	Delaware River	6
14138870	Fir Creek	2
14372300	Rogue River	6
2160700	Enoree River	5
8123800	Beals Creek	5
8181500	Medina River	5
408000000 Middle Branch Embarrass River	3
no	FOR	NY	0.42	0.019
no	OR	0.38	0.059
no	FOR	OR	0.16	0.011
no	FOR (urb)	SC	0.5	0.04
no	Shrub	TX	0.46	0.018
no	AG	TX	0.7	0.095
no	AG	WI	0.96	0.03

-------
1	Stations in Utah, Maine and North Carolina were of particular interest because biological
2	data from these states were analyzed for climate change effects. Summer temperatures have
3	increased gradually at stations in each of these three states (Figures A-9 to A-l 1). Hyco Creek in
4	North Carolina has shown the greatest increase in daily maximum temperature (from about 23 to
5	25°C over about a 40 year period, r =0.192). This may be influenced by stream size, as Hyco
6	Creek is a 3rd order stream, while the St. Croix in Maine and the Colorado in Utah are 6th order
7	or higher.
9180500 UT
w
3
CD
3
<
"O
c
co
>.
3
-D
C
O
O
IT
a
2
0
Gt
E
0
I-
E
3
E
"x
CD
CD
Q
COLORADO RIVER NEAR CISCO, UT
30
25
20
15
o ocSc
OCQCOO Q30 O
GO CD GOOD GO
OOGCOGCOCOD GOCOO
nrmrmorrrrrrm OO
O
OO OO
O O CGOCGD OGGOO O
OO	O
CDDGGO OCG0CDG3D GGGG3D O
„	GOD	O O
OXOGD OOOGCEO OGGdmBOCDC,
OGGOa
axco o
aamHBXEax) cd arraxmirr) gcccqo cco
O CO GCOEOXOO CO CO GOO O
axo aacoceo o ao annxao aco ao ao
Qdogo o ogo ceoceqd
°s
3GGQOO
CO

3 CD
O O

} O
O
o
O
O
o

O
o
O
O
o

O
o

O
ooo
o oo
go
o oo
O ODD
o o
CGGGO GGGGO GO
COD O
GDOOGGO
O
cooo
oo
o
o
R2=0.059
_L
_L
_L
_L
_L
_L
10
1940 1950 1960 1970 1980 1990 2000 2010
YEAR
8	Figure A-9. Summary of daily maximum temperature trends for July and August data
9	from USGS Gage 9180500 on the Colorado River near Cisco, Utah.
August 14, 2009
4-1
Internal Review Draft

-------
11
12
13
14
15
16
17
18
19
20
21
22
1021050 ME
St. Croix River at Milltown, ME
O OO
OOOO O
O OO O
OOOOO O
OOOOO O
OOOOOOO
OOOOOOO
OOOOOOO
OOOOOOO
OOOOOOO
qooooaa
FWS	OOOOOOO
>il	OOOOOOO
iSl	OOOOOOO
§1	OOOOOOO
III	OOOOOOO
isl oooooo
ill OOOOOOO
,gf§ OOOOO o
Q OOOO o
OO OOOO
i	OOOOO o
®	OO o o
8	ooo o
o	ooo o
1970 1980 1990 2000 2010
YEAR
Figure A-10. Summary of daily maximum temperature trend for summer data from USGS
Gage 1021050 on the St. Croix River at Milltown, Maine.
Around the country, temperature responses are quite variable, though long-term
increasing water temperature trends are observable in many rivers and streams (Table A-l).
Rates of temperature change per 10-year period for 23 stations in 18 different states range from
5.09 degrees in 10 years in a 1st order reach of the Missouri River, North Dakota (r2= 0.5), to
0.25 degrees in 10 years in a 6th order reach of the Skunk River, Iowa (r2= 0.006)' Results varied
across stations. The average rate of increase per 10-year period was 0.76 degrees. Similar
increases in stream and river water temperatures over recent decades have been documented
across the US (Kaushal et at. 2010) and in Europe (e.g., Webb and Nobilis (2007).
August 14, 2009
4-2
Internal Review Draft

-------
2077200 NC
00
HYCO CREEK NEAR LEASBURG, NC
CD
E
E
3
CO
c
O
o
¥
a
CO
0
I-
E
3
X
CO
35
15
R2=0.192

O O
o o
OCO 00
OOO o 00
0000000000000000
OOOOOOOOO 000000
00 00
OCOOOOoOOOOO
OOOO03qOO OOqq o o°° 0
caoaaoooo 000000 00 00
O OOcCoOO Oo o o
O O OOOO OOO o
o	00
o
_L
_L
_L
_L
^ 1960 1970
'co
G
1980 1990
YEAR
2000 2010
23	Figure A-11. Summary of daily maximum temperature trend for summer data from USGS
24	Gage 2077200 on Hyco Creek, North Carolina.
25
August 14, 2009
4-3
Internal Review Draft

-------
A.3 Benthic Macroinvertebrate Inferred Temperature
Annual water temperature values for selected sites were inferred based on relative
abundance and temperature optima data for macroinvertebrate taxa that occurred at each site.
The temperature optima values used in these calculations were derived from weighted averaging
or maximum likelihood modeling on appropriate subsets of the state biomonitoring data
(Appendix D). The "benthic inferred temperature" for a station is then calculated as another
weighted average, taking model results of temperature optima for each taxon occurring at a
station, multiplied by the abundance of that taxon, with those products summed over all taxa at
the station, and divided by the sum of taxa abundances. Questions that were addressed using this
approach include whether benthic communities reflect water temperatures at the time of
collection; and whether long-term changes in inferred temperatures provide evidence of benthic
community changes over time related to temperature.
Most of the long-term stations within ecoregions that were tested showed slight to
distinct increasing trends in benthic inferred temperatures over time, though not all the trends
were significant. In Maine, inferred temperatures for Station 56817, a long-term but low
elevation station in the Laurentian Hill and Plains ecoregion, showed a gradual upward trend
since 1984 (Figure A-12). A steeper upward trend was evident at the selected Maine East Coast
region reference sites, which included some higher elevation locations (Northeast Highlands
ecoregion) (Figure A-13). There is no real pattern for the group of relatively low elevation sites
in the Maine Central Interior biophysical region (Figure A-14). The greater inferred temperature
responses are evidence of climate change increases in temperature, with greater apparent
responsiveness in higher elevation locations. This is consistent with findings of greater climate
change effects at higher elevation areas based on other biologic metrics (Section 2).
The plot of inferred temperatures for multiple stations across all three ecoregions in
North Carolina (excluding the coastal plain) showed a gradual temperature increase since 1994
(Figure A-15), though the trend with year was not significant. The benthic inferred temperature
trend at three reference stations in Utah (sampled in October-November) showed a gradual, but
statistically significant, increase (Figure A-16). The rate of increase is equivalent to
approximately 3° C in 25 years. In the plots in which multiple sites were grouped together, site-
specific differences were often evident. In all of these cases, the close relationship between the
August 14, 2009
4-4
Internal Review Draft

-------
benthic inferred temperatures and the field-measured water temperatures shows that the approach
of using benthic invertebrate occurrence and abundance coupled with temperature preferences is
a reliable means of estimating water temperature at the time of collection. More importantly, it
provides evidence of benthic community changes over time related to long-term changes in
temperature.
Maine Site 56817
Figure A-12. Benthic macroinvertebrate inferred temperature trend for Maine Site 56817.
August 14, 2009
4-5
Internal Review Draft

-------
Maine East Coast region reference sites
25
Q_
STATION
20
® 20
-Q
8 10
20
25
30
•	57064
~ x 57065
57066
*	57104
_ t 57105
« 57198
STATION
•	57064
x 57065
+ 57066
*	57104
~ 57105
-< 57198
15 20 25
Field Temperature (°C)
Year
Figure A-13. Benthic niacroinvertebrate inferred temperature trend for selected reference
sites in the Maine East Coast region.
Maine Central Interior biophysical region
Q_
E
£ 23
I-
» 21
K 10
1980
1990
2000
2010
•	56817
x 56854
+ 57011
*	57040
STATION
• 56817
x 56854
57011
a 57040
STATION
• <
Year	Field Temperature (°C)
Figure A-14. Benthic niacroinvertebrate inferred temperature trend for selected sites in
the Maine Central Interior biophysical region. Note that Station 57040 has a statutory class
of AA but its use in this analysis is questionable because of its proximity to a Superfund
site.
August 14, 2009
4-6
Internal Review Draft

-------
North Carolina multiple full-scale samples
15 20 25 30
Field Temperature (C)
35
0
£
3
4—<
1
I 30
e
S 25
«—
i—
© 20
to
5
9
o
TO
15
10
1
•
I
•
•
•
•	t
•	t
1
« •
1
•
• I *
is* *!
*» * •
	=
	1-
•
~
. ;•
i *
•
•
*	* 1
*
• •
•
i
i."!
: •:
~
•
1
•
"
1
1990
1995
2000
Year
2005 2010
Figure A-15. Temperature inference model for multiple full-scale samples in North
Carolina.
Utah October and November reference samples
o
0
D
¦*—'
TO
i*.
0)
Q.
E
OJ
TD
0
20
15
j? 10
0
.Q
0
0 5
-t—*
0
>
C
o
o
ro
	1	1	1	
STATION
R-=0.147
•	4927250
•	4936750 P-0.017
•	4951200
..itf*•*
		* • .
.»¦	* . • • *
• I
~
_L

20
1980 1985 1990 1995 2000 2005
Year
0

I—
"O
a>
| 10
c
CD
4—'
2
-Q
aj
1
c
o
V—
o
to
1	T
STATION
•	4927250
*	4936750
« 4951200
• I
V *
5 10 15
Field Temperature 
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
APPENDIX B
Data preparation and management
The purpose of this appendix is to provide detailed information on how the state databases were
selected, what collection and assessment methods are used by each of the states, and how the
data for each of the states were prepared for analysis.
Bl. Selection of the 4 state databases: Maine, North Carolina, Ohio and Utah
B2. State collection methods
B3. Database Preparation
B4. Discussion
B-l

-------
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Bl. Selection of the Four State Databases: Maine, North Carolina, Ohio and Utah
Four state benthic macroinvertebrate and/or fish databases were selected for the regional
climate change pilot studies. Overall criteria for selection of existing state data sets were to
include representatives of a distribution of regions around the country that would reflect different
climatic, geographic, and ecological zones, as well as different ranges of future climate change
projections. We considered state programs that have been well-established for the longest times
and would have long-term data bases with consistent methods and strong quality control (QC).
Additional rationale for the final selections include:
Maine. Maine has a benthic macroinvertebrate dataset that is long-term with consistent
methods, and with repeat sampling at some locations (i.e. one site has over 20 years of data). It is
in an area with regional climate change modeling and is expected to show sensitive responses to
climate change given its northerly location.
North Carolina. North Carolina captures the unique expectations for climate responses
in the southeastern region. The North Carolina invertebrate data set is long-term, with consistent
methods and good quality control (QC).
Utah. Both the Utah and New Mexico datasets were strongly considered for analyses in
the western/southwestern region of the country. Utah was selected because it had more long-term
repeat sampling (up to 19 years of data over a 21 year time span) and had a better distribution of
sampling locations. A shortcoming of the Utah data was that (unlike the New Mexico data) most
of the historic data set (i.e., older than about the last 8 years) had only recently been entered into
an electronic format from hard-copy data sheets and had not been QC'd or previously analyzed
as a unified long-term data set.
Ohio. Both the Ohio and Wisconsin datasets were strongly considered for analyses in the
Midwestern region of the country because they both have long-term fish data in addition to
benthic invertebrate data. Wisconsin is expected to show shifts between cold- and cool- or warm-
water fauna. However, it was not clear that collection and reporting methods for the Wisconsin
fish data were standardized. Because Ohio has a long-term fisheries dataset with standard
methods, and had long-term benthic data as well that was already being analyzed for long-term
trends, it was selected.
B-2

-------
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
B2. State Collection Methods
Each of the four states uses different benthic macroinvertebrate collection methods. Utah
collects a quantitative sample from riffle habitats during a September/October index period using
the EMAP kick method (note: prior to 2006, samples were collected using the Hess method).
Maine uses artificial substrates (rock bags or baskets) to collect quantitative samples during late
summer, low flow periods (July 1 to September 30). North Carolina uses several different
collection methods and collects samples throughout the year, but for this study the focus was on
summer (June through September) samples collected using the standard qualitative ('full-scale')
method, which is comprised of 2 kicks, 3 sweeps, 1 leaf pack sample, 2 fine mesh rock and/or
log wash samples, 1 sand sample and visual collections. Ohio collects quantitative
macroinvertebrate samples using a modified multiple-plate Hester Dendy artificial substrate
sampler. A routine sample consists of a composite of five samplers that are colonized for a 6
week period that normally falls between June 15 and September 30. In addition to the artificial
substrate, a qualitative sample is taken from all available natural habitats within the reach. When
sampling for fish, Ohio uses pulsed direct current electrofishing techniques. Depending on
stream size, crews either use headwater, wading or boat site protocols.
B3. Database Preparation
Biomonitoring data from Maine, North Carolina, Utah and Ohio were compiled into
Ecological Data Application System (EDAS) databases, which are custom database applications
that are used with MS Access. The data from Maine were taken from Maine's existing
Oracle/Access database (EGAD). North Carolina data were provided in various formats (MS
Excel and MS Access). Data for Utah were obtained from STORET. For Ohio, data were
originally obtained from STORET; however, interactions with Ohio EPA revealed that data
generation, data base development and management, as well as ongoing analyses for Ohio are
conducted by Ed Rankin and Chris Yoder of Midwest Biodiversity Institute (MBI). Therefore,
the additional data manipulation and analyses needed for this study were conducted by MBI
under subcontract to Tetra Tech.
B2-1

-------
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
B3.1 Data screening
Data were screened to minimize chances of detecting false trends. Preliminary iterative
data summary and screening procedures included:
1.	Tabulating numbers of samples by station (e.g., station name, station ID number, and/or
sample ID number) and date. Results were examined for consistent number of samples by
station/date and for breaks in sample collection at stations across years. Problems
discovered through this approach included changes over time in collection and/or
reporting of replicates; and errors or changes in station naming that resulted in data for
the same location appearing under different station names. It also helped identify
locations with long-term data records.
2.	Tabulating total abundance and total number of taxa by station and collection date.
Results were examined for discontinuities in magnitude or trends in values between
stations and across dates. Problems discovered through this approach included changes in
reporting of abundances (e.g., from number per sample to number per square meter;
whether replicates were averaged, summed, or reported separately); and changes or errors
in whether sub-sampling was applied during sample analysis and how it was accounted
for in the data.
3.	Tabulating taxa (at the lowest levels reported) by collection date. For these, either taxa
abundance or occurrence was tabulated, and these were either averaged over all stations
within the state, or within each ecoregion and/or other appropriate subset (e.g., river basin
or watershed). For this purpose, the tabulations of taxa were placed in phylogenetic order,
and some higher-level phylogenetic structure (e.g., order and family names, or others as
needed) was included for reference. Results were examined for many types of patterns,
including:
a. changes in taxonomic naming over time (e.g., changes in genus or higher level
names, changes in placement within families). This not only revealed changes in
systematics over time, but also caught changes in taxonomists and/or labs used to
analyze samples.
B2-2

-------
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
b.	changes in level of attribution over time (e.g., increasing use of species names in
recent years where individuals were typically left at the genus or family level in
earlier samples);
c.	changes in other types of naming conventions (e.g., changes in level of placement
for taxa such as water mites).
Problems identified through these procedures included extensive changes in taxonomic
knowledge and systematics over the decades of sample analysis. For illustration, one
example is changes in the mayfly genus Ephemerella, including changes in the inclusion
of various species of Ephemerella between Ephemerella and Drunella. In addition, we
found many instances of changes in the higher-level groups under which various taxa
would be reported, so that in the data base the same genus (or species, or family) would
appear in more than one place. The effect of this was that these would act like separate
taxa when a taxa ID name or number was invoked for trend analysis. Many associated
corrections were applied to the phylogenetic structuring and naming conventions in the
data bases. In many cases, changes in taxonomic naming of genera and/or species, or
greater prevalence of species identifications in recent years, required standards to be set
for summing species back to the genus level (or similar procedures at other levels), or
for combining two or more genera that cannot always be reliably separated. This type of
correction falls into the category of developing 'Operational Taxonomic Units" (OTUs),
and is discussed in more detail below.
4.	Tabulations of station descriptive data, to identify reference locations and any data
documented in support of reference station status.
5.	Tabulations of 'ancillary' environmental data, such as temperature, water chemistry,
substrate characteristics, habitat characteristics, by station over time. These results were
compared for concordance with biological data.
6.	Data also were screened for changes in sampling methods over time and/or between
stations.
We used Non-metric Multidimensional Scaling (NMDS) to evaluated whether the
database 'fixes', and in particular the taxonomic corrections and application of OTU rules, were
B2-3

-------
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
effective in minimizing changes over time due to taxonomic identification procedures rather than
actual community changes. NMDS is an ordination that takes the taxa in the samples and shows
in ordination space how closely related the samples and stations are based on their species
composition. Grouping variables (i.e. year, month, collection method, taxonomy lab, ecoregion,
watershed, etc.) can be overlaid to look for trends. The NMDS ordinations were performed only
on reference sites in order to eliminate differences due to other disturbances. The NMDS
ordinations were run before and after generating genus level OTUs. Patterns were examined for
distinct shifts that might indicate changes in taxonomists or labs during the sampling period of
record, as well as ineffective OTU procedures (see results below).
B3.2 Development of operational taxonomic units (OTUs)
The intent of OTUs is to exclude ambiguous taxa from analyses (e.g., Cuffney et al.,
2007) and include only distinct/unique taxa. Since a complete and correct master taxa list is
required before OTUs can be established, the master taxa lists in each of the databases were first
verified through several iterative procedures (see above). Next, three levels of OTUs were
established: lowest taxonomic unit (generally species), genus and family. Rules were developed
based on a general procedure of Remove Parent / Merge Children (RPMC) (retain the Child taxa
(finer level of detail) and remove the Parent taxon or merge the Child taxa into the Parent taxon).
According to Cuffney et al. (2007), this appears to be the most robust method for retaining taxa
richness and abundance information for further analysis. All decisions were data set dependant.
Rules were created on the dataset as a whole and then applied to individual samples prior to
analysis. The last step in the process was to manually review the list of OTU designations and
make final corrections where necessary.
B3.2 Utah
Data for Utah were obtained from STORET and compiled. The process was less efficient
than originally hoped, in that data had to be gathered in sections by data type and pieced back
together. This was largely due to limitations placed on data downloads from the STORET
website. Jeff Ostermiller from the Utah Department of Environmental Quality (UT DEQ) was
the contact person for data interactions.
B2-4

-------
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
Through examination of results in the Utah data base, it was determined that genus-level
OTUs were most appropriate for the long-term analyses. One of the more noteworthy OTU
'fixes' that had to be made was that all midges had to be grouped to the family level
(Chironomidae), as subfamily and/or genus level identifications only occurred in later years in
the Utah data. As another example, changes also had to be made to OTU assignments for
Ephemerella and Drunella.
To check for trends in the Utah dataset, pre- and post-OTU NMDS analyses were
performed using the following grouping variables: taxonomy lab (pre- and post-1989), level 3
ecoregion, reference status, and HUC04 hydrologic basins. Trends related to latitude and
longitude were also evaluated in reference status plots. An obvious trend appeared in the pre-
OTU plot that used taxonomy lab as the grouping variable (Figure B3-la). This was due to the
change in taxonomy lab that occurred in 1989. The OTU sufficiently corrected for this change,
as can be seen in the post-OTU taxonomy lab plot (Figure B3-lb). Results from the other
NMDS ordinations can be found in Figures B3-2-B3-6.
Another issue that arose with the Utah data was that there was some uncertainty as to the
consistency of how abundance data was recorded over the years. These questions related to
whether the recorded abundances were corrected for subsampling in the laboratory, area
sampled, and/or replication. These questions could not be fully resolved based on institutional
knowledge of Utah DEQ scientists or from extant database metadata or other documentation.
Because of this uncertainty, relative abundances were used in analyses. We also found that
although Utah reports using a late-summer to fall index period for sample collection, the Utah
database includes samples collected throughout the year. For most analyses, only fall samples
were used to minimize variation associated with seasonal differences in taxonomic composition.
B3.4 Maine
Data for Maine were obtained from Maine's Oracle/Access database (EGAD) as output
in an Access database, and compiled. Susanne Meidel from the Maine Department of
Environmental Protection (ME DEP) was the contact person for data interactions.
As for Utah, it was determined based on evaluation of the Maine data that genus-level
OTUs were appropriate for the long-term analyses. To check for trends at the genus-level OTU
B2-5

-------
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
in the Maine dataset, NMDS analyses were performed using reference status, level 3 ecoregion,
year (in 5, 10 and 20-year increments) and taxonomy lab as grouping variables. Trends related to
latitude and longitude were also evaluated in the reference status plots. Unlike the Utah data,
there were no defined breakpoints (Figures B3-7-B3-14). Rather there were small breaks in the
data in 1990-91 and 1999, along with a subtle shift towards finer taxonomic resolution from the
early 1980's to the present (as one would assume due to improved taxonomic keys, etc.). The
improved resolution is evident in plots that show the average number of species and genus-level
identifications per year (Figure B3-15).
The break in the data that was detected in 1990-91 resulted from an increase in species-
level identifications that were recorded for a number of different Orders. This was particularly
evident for the order Trombidiformes (water mites). Water mites were identified to the suborder
level (Prostigmata) prior to 1991, but from 1991 onwards, there were 28 different identifications
associated with the water mites, with some to the species-level. This was accounted for by an
OTU correction in which all taxa from the Order Trombidiformes were grouped into the
suborder Prostigmata. An increase in taxonomic resolution for Chironomidae also tracks with the
1990-91 break in the data (Figure B3-16). We considered grouping all Chironomidae to the
family-level, but decided that this would result in the loss of too much information, and that the
trends were not consistent enough to warrant the change. The second more subtle break in the
data occurs in 1999. This is likely due to variability among the taxonomic labs, since four new
labs started doing taxonomic identifications for Maine in 1999.
The genus-level OTU procedures resolved most of these observed differences, as can be
seen in the post-OTU NMDS plots. Other possible refinements were problematic, because of the
multitude of taxonomy labs that were used over the years. In the 26 year period over which data
were collected, sixteen different taxonomy labs did identifications (Table B3-1) (NOTE - this
list is revised from that which appears in the Maine database, based on personal communication
with Leon Tsomides of Maine DEP). Seven of the labs did 10 or fewer samples, while 4 did 100
or more samples. Once sample size is factored in, there is not a clear difference in the
distribution of total taxa among labs (Figure B3-17). The NMDS plots that use taxonomy lab as
the grouping variable also failed to reveal any clear or consistent patterns.
B2-6

-------
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
Another issue that arose with the Maine data involved differences in collection methods.
Maine DEP typically collects samples using rock baskets and rock cones that are deployed for
about a month. During some years, Maine DEP experimented with different collection methods
(i.e. qualitative methods). To minimize variability due to collection method, we only used rock
basket and rock cone samples in their analyses. Another factor that was taken into account when
doing the analyses was temporal differences. The majority of samples were collected during the
summer and fall. However, some samples were collected in the winter and spring. Seasonal
variability was accounted for by limiting its analyses to samples that were collected from June
through November. Differences in subsampling were also investigated (mainly for effects on
richness metrics; abundances had already been adjusted for subsampling). In the Maine DEP
database, subsampling information is recorded in a field titled 'Dilution factor' (a value of 1
means that the entire sample was analyzed, a value of 2 means half the sample was analyzed, a
value of 4 means that a quarter of the sample was analyzed, etc.). This field had limited worth
because many entries were blank. However, for analyses in which long-term trends in generic
richness were investigated at specific sites, subsampling information was noted when available.
Due to the inconsistency in whether subsampling information was included for samples in the
data base, no corrections to taxa richness information were actually applied.
Abundance information appeared to be recorded in a consistent manner in the Maine
data. Maine DEP typically deploys three rock baskets or cones per site. Each rock basket is
considered to be a replicate. For purposes of the analyses, the replicates from each site were
grouped into a single 'BenSamp' and subsampling of the data was done to 200 organisms (±
20%) [160 - 240], Relative abundances of the taxa were calculated for all the BenSamps.
B3.5 North Carolina
Data for North Carolina were compiled from into a database from the raw data provided
by Trish MacPherson from North Carolina Department of Water Quality (NCDWQ). We found
that North Carolina records data by water body name, location description, latitude and
longitude, and date, but does not assign unique Station IDs to its sampling sites. We therefore
had difficulty determining whether some stations represented the same or different sites. Some
samples have similar water body names but with slightly different spellings (for example,
B2-7

-------
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
'Creek' might be spelled out in one sample record and abbreviated as 'Cr' in another). Samples
with similar water body names and location descriptions might have had slightly different
latitudes and longitudes. Some sites had the same water body name but slightly different location
descriptions. To address this issue, we created unique identifiers for sites (Station IDs) based on
matching a combination of water body name, location, and latitude-longitude, for the individual
stations that were analyzed for long-term trends.
As with the other states, evaluation of the North Carolina data confirmed that genus-level
OTUs were appropriate for the long-term analyses. To check for trends at the genus-level OTU
in the North Carolina dataset, NMDS analyses were performed using collection method,
reference status, level 3 ecoregion, and year (in 5-year increments) as grouping variables.
Because the same people in the North Carolina biomonitoring program have done all the
taxonomic identifications for the last 25-30 years, we felt it was unnecessary to include
taxonomy lab as a grouping variable. Any inconsistencies in taxonomic identifications over the
years are most likely due to changes in taxonomic keys.
An obvious trend occurred in the NMDS plot that used collection method as the grouping
variable (Figure B3-18). Samples that were collected using different collection methods
generally formed different groups. EPT samples in particular formed a very distinct group.
Another noticeable pattern in the data occurred in 1998, when there was a spike in the total
number of taxa identified, despite fact that the number of stations sampled in 1998 was only
slightly higher than in previous years (Figure B3-19a). Many of these taxa only occur in the
database during 1998. Upon further investigation, we found that a large number of estuarine sites
were sampled in 1998. Many of these sites were not sampled prior to 1998 and have not been
sampled since. By limiting the samples to full-scale method only, these trends were eliminated
(Figure B3-19b). Based on these results, only full-scale collection method samples were used in
analyses. This resulted in the loss of 4 years of data (1978-1981) and reduced the overall number
of taxa in the database, but was a necessary and effective step in minimizing the chances of
detecting false trends in the biological data (Figure B3-20).
The NMDS plots that used reference status as the grouping variable did not show a clear
or consistent pattern (Figure B3-21), but plots with samples grouped by level 3 ecoregion did
(Figure B3-22). Samples generally grouped together by ecoregion (both pre- and post-OTU).
B2-8

-------
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
We account for this by performing most analyses, as appropriate, on subsets of data specific to
particular ecoregions (an exception is the maximum likelihood temperature optima and tolerance
calculations, for which sample size was an issue and having a wide range of temperatures is
needed and appropriate). Temporal differences also had to be accounted. NCDWQ collects
samples throughout the year. Some taxa, such as the winter stoneflies, are strongly seasonal. To
minimize such predictable variation associated with seasonal differences in taxonomic
composition, the datasets used in most of the analyses were limited to samples collected from
June through November.
Abundance information appeared to be recorded in a consistent manner in the North
Carolina data. NCDWQ records its abundance data as categorical variables, l=rare (1-2
specimens), 3=common (3-9 specimens), and 10=abundant (10 or more specimens), which limits
the type of analyses that can be performed. Data were converted to presence-absence and/or
relative abundance (calculated using the categorical variables (1, 3 and 10)) when performing
analyses.
B3.6 Ohio
As mentioned above in Section B.3, data manipulation and analyses needed for this study
were conducted by MBI under subcontract to Tetra Tech. This included taxonomic comparisons,
OTU development, and analyses such as NMDS applied to assess the effectiveness of these data
management efforts (see Appendix H, especially H.3).
B4 Discussion
Preparing the data for the analyses was a very time consuming yet necessary step. It is
essential that proper quality assurance procedures are followed to ensure the validity of the
analyses. For this project in particular, the detection of false trends in the long-term data was a
major concern.
Factors that were shown to contribute to changes that had to be accounted for prior to
trend analysis include collection method (Maine and North Carolina), sample collection dates
(all three databases), and taxonomic labs (Utah). Although one cannot entirely eliminate these
B2-9

-------
311	issues, by selecting appropriate subsets of data and establishing appropriate OTUs, chances of
312	detecting false trends in the biological data can be minimized.
313
B2-10

-------
Utah (pre-OTU)
315
316
317
318
a pre-1989
post-1989


Axis 1
Figure B3-la. Pre-OTU (genus) NMDS plot when lab is used as the grouping variable.
Utah (post-OTU)
a pre-1989
post-1989

A l44
A AA A*
Axis 1
Figure B3-lb. Post-OTU (genus) NMDS plot when lab is used as the grouping variable.
B2-11

-------
Utah (pre-OTU)
Ecoregion (Level 3)
a 13 (Central Basin and Range)
a 14 (Mojave Basin and Range)
v 18 (Wyoming Basin)
~ 19 (Wasatch and Uinta Mountains)
o 20 (Colorado Plateaus)
21 (Southern Rockies)
o 80 (Northern Basin and Range)
320
Axis 1
321	Figure B3-2a. Pre-OTU (genus) NMDS plot when level 3 ecoregion is used as the grouping
322	variable.
Utah (post-OTU)

Ecoregion (Level 3)
a 13 (Central Basin and Range)
14 (Mojave Basin and Range)
18 (Wyoming Basin)
~ 19 (Wasatch and Uinta Mountains)
o 20 (Colorado Plateaus)
21 (Southern Rockies)
o 80 (Northern Basin and Range)
323
Axis 1
324	Figure B3-2b. Post-OTU (genus) NMDS plot when level 3 ecoregion is used as the grouping
325	variable.
B2-12

-------
Utah (pre-OTU)
326
Ref Status
a Ref
~ So-So
v Trash
NA
* t	» '
'	,T W
-^_Tf T	T *A T t
t ''' 't4
X-*p2S£
T t »A ' iX ^
327	Figure B3-3a. Pre-OTU (genus) NMDS plot when reference status is used as the grouping
328	variable.
Utah (post-OTU)
fi
V"A ^

Ref Status
ARef
i So-So
^ Trash
~ NA
329
Axis 1
330	Figure B3-3b. Post-OTU (genus) NMDS plot when reference status is used as the grouping
331	variable.
B2-13

-------
Utah (pre-OTU)
HUC04
a 1403
1404
1405
t 1406
o 1407
1408
° 1501
• 1601
~ 1602
¦ 1603
x 1704
Axis 1
332
333
334
335
336
Figure B3-4a. Pre-OTU (genus) NMDS plot when HUC04 is used as the grouping variable.
Utah (post-OTU)
'T S ° D ?y

Axis 1
HUC04
a 1403
a 1404
v 1405
~ 1406
o 1407
1408
o 1501
• 1601
~ 1602
¦ 1603
x 1704
Figure B3-4b. Post-OTU (genus) NMDS plot when HUC04 is used as the grouping
variable.
B2-14

-------
Utah (pre-OTU)
337
338
339
Ref Status
A Ref
So-So
V T rash
~ NA
-y*
.. ><;
¦ •'Ml?
Axis 1
Latitude
Axis 1
.139 tau =
Axis 2
-.010 lau = -.003
¦ V / ¦
" *£ ..•*****}>'«»¥.'. "*" * •
Figure B3-5a. Pre-OTU (genus) NMDS plot when reference status is used as the grouping
variable. Trends related to latitude are also evaluated.
Utah (post-OTU)
340
341
342
Ref Status
Ref
So-So
Trash
~ NA
t-im
•tj


Axis 1
Latitude




Axis 1
r = .119 tau = .078
Axis 2
r= .087 tau = .061
Figure B3-5b. Post-OTU (genus) NMDS plot when reference status is used as the grouping
variable. Trends related to latitude are also evaluated.
B2-15

-------
Utah (pre-OTU)
Ref Status
A Ref
~ So-So
V T rash
~ NA
343
344
345
346
347
348
-115 -113 -111 -109
Longitude
Axis 1
r= -.220 tau = -.188
Axis 2
r = .008 tau = .036
Axis 1

Figure B3-6a. Pre-OTU (genus) NMDS plot when reference status is used as the grouping
variable. Trends related to longitude are also evaluated.
Utah (post-OTU)
Ref Status
Ref
So-So
Trash
NA
«
-115 -113 -111 -109
Longitude
Axis 1
r = -.218 tau = -.208
Axis 2
r= .076 tau = .045
Axis 1
Figure B3-6b. Post-OTU (genus) NMDS plot when reference status is used as the grouping
variable. Trends related to longitude are also evaluated.
B2-16

-------
Maine (pre-OTU)
Year Group (5)
a 1970-4
* 1980-4
7 1985-9
» 1990-4
o 1995-9
2000-4
o 2005-6
350
Axis 1
351	Figure B3-7a. Pre-OTU (genus) NMDS plot using sample years (5-year increments) as the
352	grouping variable.
Maine (post-OTU)
353
354
355

'F
«/sV^°o<
;%C(Vt°	oA
Year Group (5)
a 1970-4
* 1980-4
v 1985-9
t 1990-4
o 1995-9
2000-4
o 2005-6
Figure B3-7b. Post-OTU (genus) NMDS plot using sample years (5-year increments) as the
grouping variable.
B2-17

-------
Maine (pre-OTU)
356
357
358

&
Axis 1
Year Group (10)
& 1970-9
1980-9
1990-9
~ 2000-6
Figure B3-8a. Pre-OTU (genus) NMDS plot using sample years (10-year increments) as the
grouping variable.
Maine (post-OTU)
*57 fc ~ ~ ljc'w.
15
Year Group (10)
a 1970-9
a 1980-9
v 1990-9
~ 2000-6
¦<
vW V if* i
v^7 v
V 7	7^
4
359
Axis 1
360
361
Figure B3-8b. Post-OTU (genus) NMDS plot using sample years (10-year increments) as
the grouping variable.
B2-18

-------
Maine (pre-OTU)
Year Group (20)
a NA
*1970-1989
v 1990-2006
363	Figure B3-9a. Pre-OTU (genus) NMDS plot using sample years (20-year increments) as the
364	grouping variable.
Maine (post-OTU)
Year Group (20)
366	Figure B3-9b. Post-OTU (genus) NMDS plot using sample years (20-year increments) as
367	the grouping variable.
B2-19

-------
Maine (pre-OTU)
A	# T
y? vy_ .aT Vv
VA ^£7	i wr
368
<\ >" *i T.^ ^
* V / V V
Axis 1
Ref Status
a AA
A
B
~ C
o NA
369	Figure B3-10a. Pre-OTU (genus) NMDS plot when reference status is used as the grouping
370	variable.
Maine (post-OTU)
371	Ms1
372	Figure B3-10b. Post-OTU (genus) NMDS plot when reference status is used as the
373	grouping variable.
B2-20

-------
Maine (pre-OTU)
Ecoregion (Level 3)
ANA
A 58 (NE Highlands)
v 59 (NE Coastal Zone)
~ 82 (Laurentine Plaines and Hills)
Axis 1
376	Figure B3-lla. Pre-OTU (genus) NMDS plot when level 3 ecoregion is used as the
377	grouping variable.
Maine (post-OTU)
Ecoregion (Level 3)
ANA
a 58 (NE Highlands)
^ 59 (NE Coastal Zone)
~ 82 (Laurentine Plains and Hills)
Axis 1
379	Figure B3-llb. Post-OTU (genus) NMDS plot when level 3 ecoregion is used as the
380	grouping variable.
B2-21

-------
Maine (pre-OTU)
20
40
Latitude

Axis 1

r = .014 tau =
.171
Axis 2

r = .009 tau =
.029
Axis 1
382

Ref Status
AAA

~ A

VB

TC

ONA



v_ "jv A .'">1 l.\







383	Figure B3-12a. Pre-OTU (genus) NMDS plot when reference status is used as the grouping
384	variable. Trends related to latitude are also evaluated.
385
386
387
Latitude
Axis 1
r = -.022 tau = -.178
Axis 2
r = -.009 tau = -.036
Axis 1
Figure B3-12b. Post-OTU (genus) NMDS plot when reference status is used as the
grouping variable. Trends related to latitude are also evaluated.
Maine (post-OTU)
	
Ref Status
AAA
AA
\ B
TC
ONA
B2-22

-------
Maine (pre-OTU)
Ref Status
AAA
AA
VB
~ C
ONA
Axis 1
388
389
390
Longitude
Axis 1
r = -.109 tail = -.095
Axis 2
r = .000 tau = .046
0
-20

Figure B3-13a. Pre-OTU (genus) NMDS plot when reference status is used as the grouping
variable. Trends related to longitude are also evaluated.
391
Longitude
Axis 1
= -.072 tau = -.084
Axis 2
= -.007 tau = -.056
Maine (post-OTU)
. »» , > V'
mzmk:

"» '» W a "T^	» V »
'J, "d
Axis 1

Ref Status
AAA
~ A
VB
~ C
ONA
392	Figure B3-13b. Post-OTU (genus) NMDS plot when reference status is used as the
393	grouping variable. Trends related to longitude are also evaluated.
B2-23

-------

CM
(/)
% y o "" Ow?	7
¦'
LabCode2
42
*4
v 6
~ 8
o 9
11
o 16
• 17
~ 18
¦ 20
x 99
^ A.
394
395
396
Axis 1
Figure B3-14a. Pre-OTU (genus) NMDS plot for Maine data when lab is used as the
grouping variable.
(N
(/)
	£
V VA„^* ^44* A " 7^7 A
A% A	, /A
A A "V 7
LabCode2
i 2
a 4
76
~ 8
o9
11
o 16
• 17
~ 18
¦ 20
x 99
397
398
399
Axis 1
Figure B3-14b. Post-OTU (genus) NMDS plot for Maine data when lab is used as the
grouping variable.
B2-24

-------
Q.
Q.
Q.
O)
1970
1975
1980
1985
1990
1995
2000
2005
2010
400	Year
401	Figure B3-15a. Average number of species-level identifications per replicate sample per
402	year in the Maine database (using original data (not adjusted for OTUs).
403
35
30
25
20
15
10
5
0
1970
1975
1980
1985
1990
1995
2000
2005
2010
Year
405	Figure B3-15b. Average number of genus-level identifications per replicate sample per
406	year in the Maine database (using original data (not adjusted for OTUs).
B2-25

-------
407
Chironomidae
Heptageniidae
Hydropsychidae
408	Figure B3-16a. Average number of species-level identifications per replicate sample per
409	year for selected families in the Maine database (using original data (not adjusted for
410	OTUs).
411
412
413
414
415
¦Bactidac
•Chironomidac
Heptageniidae
•Hydropsychidae
^j-roi_T)r^cn<—iroLor— cn<—iroLD
r^oooooooocricricncricnooo

-------
Maine, GTU
40 -
30 -
X
CO
t
TO
o
H 20 -
10 -
o -
418
419	Figure B3-17. Distribution of the total number of taxa (average per replicate) among
420	laboratories.
421	Table B3-1. Per communication with Leon Tsomides Maine DEP some
422	adjustments were made
Lab
YearMin
YearMax
#Samp
LabNum
BILLIE BESSIE
1996
1996
2
1
DAVID COURTEMANCH
1983
1983
5
2
B.A.R ENVIRONM
1994
1994
6
3
WOODWARD CLYDE
1981
1981
6
4
unknown
1995
1995
7
5
BBL SCIENCES
2004
2004
9
6
CF RABENI
1974
1974
10
7
QST ENVIRONMENTAL
1994
1996
20
8
(BOWATER)




CHRIS PINNUTO
2000
2000
22
9
NORMANDEAU
1989
1999
45
10
SUSAN DAVIES
1981
1989
74
11
NEW BRUNSWICK
1999
2001
84
12
IDAHO ECOANALYSTS
1999
2005
100
13
TERRY MINGO
1983
1987
254
14
LOTIC
1988
2006
743
15
MICHAEL WINNELL
1983
2006
2509
16
B2-27

-------
North Carolina (preliminary NMS)
423
424
425
A A
4 4 A.
A *A
A	jA A A
*
CM

X
A
A^T o
A	A
Collection Method
a Full Scale
aEPT
Boat
t 2Kicks
o EQUAL
Qual4
o Qual 5
• Qual 7
~ Swamp
Axis 1
Figure B3-18. Preliminary NMDS plot (genus-level OTU) using collection method as the
grouping variable.
B2-28

-------
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
NC - Taxonomy - GTU IDs
600
500
400
300
200
100
0

"*V^
V V_J_' V V
- -! nruuyvr,/uyi-fvvT-'
1978 1982 1986 1990 1994 1998 2002 2006
Num Taxa
¦ Taxa First
Taxa Last
Nurn Stations
Figure B3-19a. North Carolina (genus-level OTU or GTU) data using all collection
methods. "Num Taxa" refers to the total number of taxa recorded in a particular year;
"Taxa First" refers to the number of taxa that appear in the database for the first time in a
particular year; "Taxa Last" refers to the number of taxa that appear in the database for
the last time in a particular year; "Num Stations" refers to the number of stations sampled
in a particular year.
NC - Taxonomy - GTU IDs - (Full-scalemethodonly)
600
500
400
300
200
100

0 (wiK -•	-j-pJ'J\ryxjir..p
1978 1982 1986 1990 1994 1998 2002 2006
-Num Taxa
Taxa First
Taxa Last
Num Stations
Figure B3-19b. North Carolina (genus-level OTU or GTU) using data from only the Full-
scale collection method. "Num Taxa" refers to the total number of taxa recorded in a
particular year; "Taxa First" refers to the number of taxa that appear in the database for
the first time in a particular year; "Taxa Last" refers to the number of taxa that appear in
the database for the last time in a particular year; "Num Stations" refers to the number of
stations sampled in a particular year.
B2-29

-------
443
444
445
446
YrG rp05
— _ A A	A ~ ~
' oJ*	v _A A	T
*	AI.' ^
«
»
-------
A**
A A A A *	4	A A
a*	? A a
Reference Status
a Reference
Unknown
451
452
453
Axis 1
Figure B3-21a. Pre-OTU (genus) NMDS plot for North Carolina data using reference
status as the grouping variable and only full-scale collection method data is used.
Reference Status
a Reference
a Unknown
454	Axis 1
455	Figure B3-21b. Post-OTU (genus) NMDS plot for North Carolina data using reference
456	status as the grouping variable and only full-scale collection method data is used.
B2-31

-------
457
V
Ecoregion
V
a Unknown
~
± 45, Piedmont
~
~
v 63, Mid Atlantic CP
~ v A
~ 65, SE Plains
V? A ~ .
A A o
o 66, Blue Ridge
V ' T 'A T I*° k
„ '»* n' ° o <>
v a.* A._ <>
•AVwrnttibjiy* s

<
4s1 <
o

44 O

O

459	Figure B3-22a. Pre-OTU (genus) NMDS plot for North Carolina data using level 3
460	ecoregion as the grouping variable and only full-scale collection method data is used.
461
»	T * J. 0 o 0°
a*V*4 I "
^VA	*mAA	A	°
Ty*iS v t &
v	O	I Qo
Ecoregion (Level 3)
a Unknown
*45, Piedmont
v 63, Mid Atlantic CP
~ 65, SE Plains
o 66, Blue Ridge
452	Axis 1
463	Figure B3-22b. Post-OTU (genus) NMDS plot for North Carolina data using level 3
464	ecoregion as the grouping variable and only full-scale collection method data is used.
465
B2-32

-------
1
APPENDIX C
2
3
Site selections and site groupings
4
5	The purpose of this appendix is to provide comprehensive and detailed information on individual
6	biological sampling sites and groups of sites that were selected for long-term trend analyses in
7	Maine, Utah and North Carolina.
8
9	CI. Maine
li
10
C2. Utah
C3. North Carolina
c-i

-------
13
14
15
16
17
18
19
20
21
22
23
24
25
26
CI. MAINE
Cl.l Individual Station Selection
In this study we refer to Class A and AA stations (as determined by Maine DEP, based on
biological attainment) as reference stations1. Reference sites with the longest-term biological
data were identified for analysis of long-term trends. There were two reference stations in Maine
that had 10 or more years of data (Table Cl-1). These 2 stations plus the reference station with
the next longest data record (9 years) were included in the individual station analyses. Locations
of all the reference stations are shown in Figure Cl-1 and locations of the 3 individual stations
that were closely examined are shown in Figure Cl-2. Brief descriptions of the 3 stations are
given below and are summarized in Table Cl-2. Additional information (i.e. aerial photos) is
available upon request.
Table Cl-1. Summary of how many years of data were available for the
different classes of Maine stations	
# Years Reference Stations B & C Not Attaining
Sampled	(A & AA)	Stations	(NA)
10-19	2	4	0
5-9	10	30	0
2-4	94	183	0
1	116	302	1
1 Class A sites are not necessarily designated as reference sites by Maine DEP (Maine DEP is in
the process of developing strict reference criteria; considerations are not based on biology and
include land use land cover and proximity to NPDES discharges).
C-2

-------
Num_ Years
• 12
O 23
Oct07 DRAFT Ecoregions in Maine
LEVEL3_NAM
| Maine/New Brunswick Plains and Hills
~ Northeastern Coastal Zone
I Northeastern Highlands
27
28	Figure Cl-1. Distributions of reference (Class A & AA) sites in Maine among the different
29	level 3 ecoregions (the Maine/New Brunswick Plains and Hills was formerly called the
30	Laurentian Plains and Hills). The number of years of data available for each station is also
31	shown.
32
33	StationID 56817 (Latitude 44.22319, Longitude -69.59334). The station is located on the
34	Sheepscot River, Maine DEP Station 74, in the town of Whitefield. It is in the Laurentian Plains
35	and Hills (which has recently been updated to Maine/New Brunswick Plains and Hills) level 3
36	ecoregion and Central Interior Biophysical Region at an elevation of 104 feet. This station is
37	located on a 4" Strahler order reach and has a drainage area of 145 square miles. It is classified
38	as ' AA' but (per communication with Maine DEP) has been influenced by non-point source
39	pollution and has occasionally received 'B' ratings. The station has been monitored on an annual
40	basis since 1984. Long-term USGS gage flow data are available for this station.
41	StationID 57011 (Latitude 44.36791, Longitude -69.53129). The station is located on the
42	West Branch Sheepscot River, Maine DEP Station 268, in the town of China. It is in the
43	Laurentian Plains and Hills level 3 ecoregion and Central Interior Biophysical Region at an
44	elevation of 230 feet. This station is located on a 3rd Strahler order reach and appears (based on
C-3

-------
45
46
47
48
49
50
51
52
53
54
55
56
57
58
aerial photographs and land use land cover information) to be influenced by human activities.
Twelve years of continuous data (1995-2006) are available for this station.
StationID 57065 (Latitude 44.3934, Longitude -68.23461). This station is located on
Duck Brook, Maine DEP Stati on 322, in the town of Bar Harbor. It is in the Laurentian Plains
and Hills level 3 ecoregion and East Coastal Region Biophysical Region at an elevation of 179
feet. This station is located on a lsl Strahler order reach and has 9 years of continuous data (1997
to 2005). Forest is the dominant surrounding land use.
57065
57011
56817+
Siteswnhthe longestcontinuousdata Oct07 DRAFT Ecoregions in Maine
LEVEL3_NAM
! Maine/New Brunswck Plains and Hills
| Northeastern Coastal Zone
I Northeastern Highlands
Figure Cl-2. Locations of the 3 reference sites in
Maine that have the longest term biological data.
C-4

-------
59	Table Cl-2. Station information for the 3 Maine reference sites with the longest-term biological data. # years of
60	data refers to June-September samples only. Eco_L3 is level 3 ecoregion. Reference status (Class A & AA) was
61	designated by Maine DEP. % land use refers to the area within a 1 km buffer of the station (NLCD 2001).


# Yrs




Ref
Status




Station
WaterbodyName
of
Eco_L3
Biophysical
Order
Elevft
%URB
%AGR
%BAR
%FOR %WET


Data









SHEEPSCOT

LAURENTIAN
CENTRAL
INTERIOR







56817
RIVER-ME
23
PLAINS AND
4
103.8
AA
16.4
23
0
56.8 3.8

STATION 74

HILLS








WEST BRANCH

LAURENTIAN
PLAINS AND
HILLS








57011
SHEEPSCOT
RIVER - ME
12
CENTRAL
INTERIOR
3
229.9
AA
9.1
18.5
0
68.3 4

STATION 268










DUCK BROOK -
ME STATION 322

LAURENTIAN
EAST







57065
9
PLAINS AND
COASTAL
1
179.1
AA
15.9
0
0
75 8.9


HILLS
REGION








-------
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
C3.2 Site Group Selection
Due to the limited number of individual sites with long-term data, sites were grouped
together to obtain more long-term biological datasets to analyze. Our initial approach to identify
appropriate station groupings involved cluster analyses to determine which reference stations had
similar assemblages and could be grouped together. In one dendrogram, stations were coded by
biophysical region (Figure Cl-3) and in another, by level 3 ecoregion (Figure Cl-4). Results
show fairly strong site-specific differences among assemblages, which does not support grouping
them for trend analyses. The stations that showed the greatest similarities were more closely
examined, but did not have enough continuous data among them to make analyses worthwhile.
Maine Reference Stations
Distance (Objective Function)
6.6E+00
Information Remaining (%)
50
58758 -
57554	-
56854	-
57483	-
57479 -
57198 -
57555	-
56935 -
57243	-
57245 -
57036 -
57484	-
57029 -
57040 -
58780 -
56864 -
56855	-
57034 -
56817	-
56861 -
57237	-
58901 -
56980 -
57465 -
57238	-
56818	-
57244	-
57139 -
56875 -
57329 -
57011 -
57065	-
57242 -
57565 -
56881	-
58991	,
58992	l|
57076 J
57441 -
58988 -
56882	-
56934 -
57396 -
56905 -
57063	-
57295 -
57297 -
57064	-
57066	-
57105 -
57423 -
57515 -
57103	-
57553 -
57104	-
57562 -

Figure Cl-3. Dendrogram of Maine reference stations color-coded by biophysical region.
Biophysical region l=Aroostook Hills, 2=Mixed, 3=Central Interior, 4=Central Mountains,
5=East Coastal Region, 6=Eastern Interior, 7=Eastern Lowlands, 8=Southwest Interior,
9=Western Foothills, 10=Western Mountains.
C-6

-------
Maine Reference Stations
80
81
82
83
84
85
86
87
88
56758 -
57554 ¦
56854 ¦
57555 -
56935 ¦
57243 ¦
57036 ¦
57464	¦
57029 ¦
57040 ¦
56760 ¦
56862 ¦
56864 ¦
56855
57034
57545
56817
56861
56901
56980
57465
57238
56818
57139 -
56875 ¦
57329 ¦
57011 -
57242 ¦
57565 -
56881 ¦
56882 ¦
56934 ¦
57396 -
56905 ¦
57423 ¦
57103 ¦
57262
57267 -l~
Distance (Objective Function)
6.6E+00
Information Remaining
Figure Cl-4. Dendrogram of Maine reference stations color-coded by level 3 ecoregion.
Level 3 Code 58=Northeastern Highlands, 59=Northeast Coastal Zone and 82=Laurentian
Plains and Hills.
In addition to the cluster analysis, other site grouping options were explored. None of
these analyses proved any more successful in forming site groupings. Information on these other
options is available upon request.
C-7

-------
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
C2 UTAH
C2.1 Individual Station Selection
Reference stations (as designated by Utah DWQ2) with the longest-term biological data
were identified and analyzed for long-term trends. There were four stations in Utah that had 10
or more years of data (Table C2-1). Locations of these stations are shown in Figure C2-1. Brief
descriptions of the 4 sites are given below and are summarized in Table C2-2. Additional
information (i.e. aerial photos) is available upon request.
Table C2-1. Summary of how many years the reference and unclassified
stations in Utah had been sampled	
# Years Reference Unclassified
Sampled Stations Stations
10-19	4	3
5-9	4	29
2-4	7	178
1	54	300
StationID 4927250 (Latitude 40.7529444, Longitude -111.3735833). This station is
located on the Weber River about 0.5 miles above Rockport Reservoir in Summit County. It is in
the Wasatch Uinta Mountains/Mountain Valleys ecoregion at an elevation of 6059 feet. This
station has 19 years of data, ranging from 1985 to 2005. Samples were taken in the spring,
summer and fall. When limited to only fall samples, 17 years of data are available. Based on
aerial photographs, this station appears to be influenced by human activities.
StationID 5940440 (Latitude 38.28, Longitude -112.5671111). This station is located on
the Beaver River above a USGS gage in Beaver County. It is in the Wasatch Uinta
Mountains/Semiarid Foothills ecoregion at an elevation of 6249 feet. This station has 11 years of
data, ranging from 1994-2005. It has a mix of spring and fall samples. When limited to only fall
samples, 9 years of data are available. Examination of aerial photographs reveals that is located
near a road (Hwy 153), but there does not appear to be any other obvious confounding factors.
2 reference criteria are based on a combination of a reference scoring sheet (multiple lines of
scoring, i.e., habitat, land use, chemistry) and independent ranking of sites from field
crew/scientists.
C-8

-------
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
StationID 4951200 (Latitude 37.2848333, Longitude -112.9480833). This station is
located on the Virgin River below Zion Narrows in Washington County. It is in the Colorado
Plateaus/ Escarpments ecoregion at an elevation of 4492 feet. This station has 15 years of data,
ranging from 1985-2004. It is in close proximity to Zion National Park. The aerial photographs
that were examined did not provide much information because they were of poor quality.
StationID 4936750 (Latitude 40.4613889, Longitude -110.83). This station is located in
Duchesne County. It is in the Colorado Plateaus/Semiarid Benchlands and Canyonlands
ecoregion at an elevation of 6967 feet. This station has 14 years of data, ranging from 1985-2002.
When limited to only fall samples, 12 years of data are available. Examination of aerial
photographs shows that the station is surrounded by roads, is located in a valley, and that there is
agricultural land in the upstream catchment area.
4927250 -fa
•fe 4936750
5940440
it 4951200

Ecoregion
LEVEL3NAM
Colorado Plateaus
Individual Long-term Biological Sampling Sites


Mojave Basin and Range
Northern Basin and Range
Southern Rockies
Wasatch and Uinta Mountains

Figure C2-1. Locations of the 4 reference sites in
Utah that have the longest term biological data.
C-9

-------
129	Table C2-2. Station information for the 4 Utah reference sites with the longest-term biological data. # years of data refers to
130	fall samples only. Eco_L3 is level 3 ecoregion and Eco_L4 is level 4 ecoregion. Reference status was designated by Utah DWQ.
131	% land use refers to the area within a 1 km buffer of the station (NLCD 2001).
StationID
# Years
of Data
Eco_L3
Eco_L4
UT Watershed Group
Elevft
Ref
Status
%URB
%AGR
%BAR
%FOR
4951200
15
Colorado
Plateaus
Wasatch and
Escarpments
Scvicr/Virgin/Bcavcr
4492
REF
3.4
0.5
28.8
67.2
5940440
9
Uinta
Mountains
Wasatch and
Semiarid Foothills
Scvicr/Virgin/Bcavcr
6249.3
REF
3.9
0
0
96.1
4927250
17
Uinta
Mountains
Mountain Valleys
Bear/Weber
6058.5
REF
4.5
21.1
0
67.1


Colorado
Plateaus
Semiarid







4936750
12
Benchlands and
Uinta Basin
6967
REF
4.8
10.3
1.1
83.7


Canyonlands







132
133
C-10

-------
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
C2.2 Site Group Selection
Due to the limited number of individual sites with long-term data, sites were grouped
together to obtain more long-term biological datasets to analyze. Efforts were focused in the
Wasatch Uinta Mountains and Colorado Plateaus ecoregions, where most of the stations are
located. One limitation of the reference stations that were used in the individual station analyses
was that % urban and % agricultural land uses within a 1 km buffer of the stations was higher
than desired. Because of this, different reference criteria were used to screen for stations to
include in the site groups. Initially stations in the Wasatch and Uinta Mountains that had <1%
urban and <10% agricultural land uses were selected, but not enough stations met this criteria, so
the criteria was changed to <2% urban and <10% agricultural land use. This resulted in a site
group consisting of 150 sites with data from 1983-2005. Jeff Ostermiller of Utah DWQ
recommended that groups be further refined because there is a lot of variation among sites within
the level 3 ecoregions (particularly between mountain and valley sites). When the 150 sites were
divided into level 4 ecoregions, two of the resulting site groups had enough stations to work
with: Mid-elevation Uinta Mountains (39 sites) and Semiarid Foothills (62 sites). These datasets
were further refined so that they only contained stations with 2 or more years of data. Efforts
were also made to limit the number of basins or watershed groups within which stations were
located, because NMDS ordinations of the preliminary data showed that stations tended to
cluster together based on basin/watershed group. The Mid-elevation Uinta Mountains site group
was limited to sites within the Ashley-Brush and Duchesne basins and the Semiarid Foothills
group excluded sites in the Colorado Watershed Group.
A similar process was followed when selecting stations within the Colorado Plateaus
level 3 ecoregion. Stations that had <2% urban and <10% agricultural land use within the 1 km
buffer area were selected. The resulting group consisted of 60 stations. The best option was to
divide the groups into level 4 ecoregions, the Semiarid Benchlands and Canyonlands, with 21
stations. Because of the small sample size, all stations were included in the group regardless of
basin or number of years of data.
Locations of the stations in each of the 3 site groups are shown in Figure C2-2. Site
information for the stations in the 3 site groups is summarized in Table C2-3, and the lists of
C-ll

-------
163
164
165
166
167
168
169
170
171
172
173
174
stations comprising each site group and the years for which data were available for each station
are shown in Tables C2-4, -5 and -6.
A number of different analyses were performed on the site group datasets. Preliminary
ordinations showed that stations generally clustered together by watershed/basin groups.
Attempts were made to reduce the differences among stations but results still need to be
interpreted with caution because site specific differences were still evident in the ordinations that
were performed on the revised datasets, as well as in results from the correlation analyses. In
some instances, trends were detected but they were due to the presence/absence of certain taxa at
certain sites that were sampled during certain years, rather than due to differences associated
with changes in climatic variables. Selected results from the NMDS ordinations are shown in
Figures C2-3, -4 and -5. More results from these ordinations and also from the correlation
analyses are available upon request.
C-12

-------
Footiuii sue Group Ecoregion
UTWshedG	LEVEL3_NAM
•	BsarfWeOe*
O Jordan
•	SeeierJVirgiiVBeaver
Mid eK-v.flion Uinta Site Group LEVEL3 NA M
Central Basin and Range
Colorado Plateaus
Mojave Basin and Range
Uinta Basin
175
176
| Northern Basin and Range
| Southern Rockies
I Wasatch and Ursa Mountains
| Wyoming Baan
BasinName
O Ashley-Brush
• Duchesne
Ecoregion
| Central Basin and Range
| Colorado Plateaus
| Mojave Basin and Range
J Northern Basin and Range
| Soul hern Roctoes
| Wasatch and Uinta Mountains
| Wyoming Basn
Figure C2-2. Locations of stations in the 3 Utah site groups.
Colorado Plateau Site Group LEVEL3NAM
UT_Wshed_G
O Colorado
• SenerMrgirtfBeaver
O Unta Basin
Ecoregion
I Central Basin and Range
| Colorado Ptaleaus
I Mofave Basin and Range
| Northern Basin and Range
| Southern Rockies
| Wasatch and Uinta Maintains
| Wyoming Basil
C-13

-------
178	Table C2-3. Summary of site information for the 3 Utah site groups. WUSF refers to the Semiarid Foothills
179	site group, WU ME refers to the Mid-elevation Uinta Mountains site group and CP refers to the Colorado
180	Plateaus Semiarid Benchlands and Canyonlands site group. % land use refers to the area within a 1 km buffer
181	of the station (NLCD 2001).
SiteGroup
#
Sites
# Yrs of
Data
Samples
Used
Eco_L3
Eco_L4
Elevft
%URB
%AGR
%BAR
%FOR
%WET
WU_SF
8
20
June-
November
Wasatch
and Uinta
Mountains
Semiarid
Foothills
5164 to
8048
0 to 1.7
Oto 0.1
0 to 2.7
97 to
100
0 to 2



June-
November
Wasatch
Mid-elevation
7200 to
9776



86 to
100

WU ME
9
12
and Uinta
Uinta
0 to 1.9
0
0 to 5
Oto 3.9



Mountains
Mountains
Semiarid




CP
16
14
June-
Colorado
Benchlands
4126 to
Oto 1.1
0 to 8.9
0 to 33
62 to
0 to 5.6
November
Plateaus
and
Canyonlands
7479
99.9
182
C-14

-------
Table C2-4. Semi-arid Foothill stations and the years during which they were sampled
Year
StationID
4936660
4926370
5988610
4954450
4954230
4995710
4995820
4954440
1983




X



1984








1985








1986








1987



X
X


X
1988



X
X


X
1989


X
X
X
X

X
1990
X

X



X

1991
X







1992
X





X

1993
X




X


1994


X





1995
X







1996
X







1997
X





X

1998
X
X






1999

X






2000

X






2001

X






2002

X






2003

X






2004

X






2005

X



X



-------
Table C2-5. Mid-elevation stations and the years during which they were sampled
Year
StationID
Ashley-Brush
Duchesne
5987290
5987000
5987350
5987230
4936840
5987530
5987610
5987700
5987870
1983





X



1984









1985









1986









1987
X




X

X
X
1988
X





X
X
X
1989
X
X



X
X

X
1990
X





X
X

1991
X
X





X
X
1992







X
X
1993


X
X





1994









1995









1996
X

X
X





1997

X
X
X




X
1998









1999









2000









2001









2002




X




2003




X




2004









2005










-------
190 Table C2-6. Colorado Plateaus stations and the years during which they were sampled
Year
StationID
Colorado
SevierAirgm/Beaver
Uinta Basin
4930340
4954196
4954460
4955820
4956400
4958032
4958730
4958755
4954220
4954090
4954110
4954140
4954180
5987860
5987880
4936200
1977




X

X









1983
















1984











X
X



1985











X
X



1986











X
X



1987








X
X
X
X
X



1988


X





X

X
X
X



1989


X





X

X

X



1990















X
1992


X













1994















X
1996
















1997
X



X








X
X

1998
X



X











2002
















2003

X














2004



X












2005





X

X








C-17

-------
Basin
a Upper Weber
a Spanish Fork
a Straw Berry
a Fremont
Taxa
A
a A
Lat_Dec
Year
A URB
A	A
^ a
A
A
UAA
A AAA
A A A
\ A A a
aa	a a
A &	^ A
& A	A	1
A Lat Dec
Basin
a Upper Weber
a Spanish Fork
a Strawberry
a Fremont
Trait Metrics
Axis 1	Axis 1
Figure C2-3. Plots of Utah Semi-arid Foothills Site Group NMDS ordinations based on
taxonomic composition and selected trait metrics.
Basin
A Ashley-Brush
A Duchesne
a a
a
A A
Elevm a
A k Long_Dec
A A AA
A A M
Basin
a Ashley-Brush
a Duchesne
A

A *
A
A & A
A
aa a a a
A

A A t A
A
A
A \ £
a\ a

A \ A
A * A
A A A
Month
A
A
Julian36 A

* a
195
196
197
198
199
Taxa
Axis 1
Trait Metrics
Axis 1
Figure C2-4. Plots of Utah Mid-Elevation Site Group NMDS ordinations based on
taxonomic composition and selected trait metrics.
C-18

-------
201
202
203
204
205
ppt14
A Ai
~
~
A A
A
AA *
Taxa
Watershed
a SevierA/irgin/Beaver
a Uinta Basin
a Colorado
Watershed
k SevierA/irgin/Beaver
k Uinta Basin
^ Colorado
ppt14 //
A Year (
Trait Metrics
Axis 1	Axis 1
Figure C2-5. Plots of Colorado Plateaus Site Group NMDS ordinations based on
taxonomic composition and selected trait metrics.
C-19

-------
207	C3. NORTH CAROLINA
208
209	C3.1 Individual Station Selection
210	Reference stations3 (as designated by NCDENR) with the longest-term biological data
211	were identified and analyzed for long-term trends. There was one reference station (sampled
212	using the standard qualitative/full-scale collection method) in North Carolina that had 10 or more
213	years of data (Table C3-1). This station plus four other reference stations were included in the
214	individual station analyses. Locations of these reference stations are shown in Figure C3-1.
215	Brief descriptions of the five stations are given below and are summarized in Table C3-2.
216	Additional information (i.e. aerial photos) is available upon request.
217
218	Table C3-1. Summary of how many years of data were available for the
219	reference and unclassified biological sampling stations in North Carolina.
220	These numbers apply only to stations that were sampled using the standard
221	qualitative (full-scale) collection method.	
# Years
Reference
Unclassified
Sampled
Stations
Stations
10 +
1
8
5 to 9
2
146
3 to 4
4
182
2
8
237
1
12
933
222
223
224	StationID NC0109 (Latitude 36.5522, Longitude -81.1833). This station is located on
225	the New River at SR 1345 in Alleghany County. It is in the Blue Ridge EPA level 3 ecoregion
226	and Mountain NCDENR ecoregion. This station has the most number of years of biological data
227	(11 years: 1983-1990, 1993, 1998 & 2003). Land use/land cover within the 1 km buffer is 44%
228	forest, 44% agricultural (of this, 99.6% is pasture hay) and 3% urban.
229	StationID NC0207 (Latitude 35.126944, Longitude -83.61916). This station is located
230	on the Nantahala River at FSR 437 in Macon County. It is in the Blue Ridge EPA level 3
231	ecoregion and Mountain NCDENR ecoregion. This station has 9 years of biological data: 1984,
3 Land use/land cover in the upstream catchment area was a major consideration in reference site
selection. These sites were recommended by Trish MacPherson (formerly NCDENR).
C-20

-------
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
1986, 1988, 1990, 1991, 1993, 1994, 1999 & 2004. Land use/land cover within the 1 km buffer
is 96% forest and 2.6% urban. Much of the upstream watershed is located in the Nantahala
National Forest. A USGS gage is located at this site (USGS 03504000).
StationID NC0209 (Latitude 35.66722, Longitude -83.07277). This station is located on
Cataloochee Creek at SR 1395 in Haywood County. It is in the Blue Ridge EPA level 3
ecoregion and Mountain NCDENR ecoregion. This station has 8 years of biological data: 1984,
1986, 1989, 1990, 1991, 1992, 1997 & 2002. It is located in the Great Smokey Mountains
National Park. Land use/land cover within the 1 km buffer is 97% forest and 3% urban. Based on
aerial photography, the urban land use is comprised of a campground, a park road and some park
buildings. A USGS gage is located at this site (USGS 03460000).
StationID NC0075 (Latitude 35.38638, Longitude -79.8322). This station is located on
Little River at SR 1340 in Montgomery County. It is in the Piedmont ecoregion. This station has
8 years of data: 1983, 1985, 1988, 1989, 1995, 1996, 2001 & 2006. Land use/land cover within
the 1 km buffer is 1% urban, 19% shrub and grassland, and 80% forest. A USGS gage is located
at this site (USGS 02128000).
StationID NC0248 (Latitude 35.43861, Longitude -80.00055). This station is located on
Barnes Creek at SR 1303 in Montgomery County. It is in the Piedmont ecoregion. This station
has 7 years of data: 1984, 1985, 1987, 1989, 1996, 2001 & 2006. Land use/land cover within the
1 km buffer it is 0.6% urban,5 % agricultural and 88% forest. There is a nearby road. This site is
located in the Uwharrie National Forest, but there are more agricultural lands in this watershed
than at some of the other sites. Trish MacPherson identified it as an interesting site because there
are a few mountain taxa still hanging on in the Uwharrie Mountains, but these are old, eroded
mountains that don't look anything like the western mountains and are actually in the middle of
the Piedmont. She believes this could be an area where cold-water taxa such as Epeorus might
disappear first as temperatures rise. Trish considers this a "relict population" site that is fairly
undisturbed.
C-21

-------
NC0109
260
261
262
263
NC0207
--"iji 4 ?>
NC0248|mc0075
H IP®!

Q Piedmont Sites
9' Blue Ridge Sites
Ecoregions
LEVEL3_NAM
| 1 Blue Ridge
	 Middle Atlantic Coastal Plain
Piedmont
Southeastern Plains
Figure C3-1. Locations of the 5 reference sites in North Carolina that were examined for
long-term trends.
C-22

-------
264	Table C3-2. Station information for the 5 North Carolina reference sites with the longest-term biological data. #
265	years of data refers to standard qualitative/full-scale collection method samples only. Eco_L3 is level 3 EPA
266	ecoregion and Eco_L4 is level 4 EPA ecoregion. Reference status was designated by NCDENR. % land use
267		refers to the area within a 1 km buffer of the station (MLRC 2001).	
StationID
WaterbodyName
# Yrs of
Full Scale
Data
Eco_L3
Eco4_Name
Elevft
%URB
%AGR
%BAR
%FOR
%WET
Sample
Months
Used
NC0109
NEW R-SR 1345
11
Blue Ridge
New River
Plateau
2341.3
3.3
44*
0
44.1
0.2
July &
Aug
NC0207
NANTAHALA R
-FSRD437
9
Blue Ridge
Southern
Crystaline
Ridges and
Mountains
6162.4
2.6
0.4
0
96
0
July,
Aug and
Nov
NC0209
CATALOOCHEE
CR - SR 1395
7
Blue Ridge
Southern
Metasedimentary
Mountains
2483.3
3
0
0
97
0
July &
Aug
NC0248
NC0075
BARNES CR-
SR 1303
LITTLE R - SR
1340**
7
7
Piedmont
Piedmont
Carolina Slate
Belt
Carolina Slate
Belt
350.1
489.8
0.6
1.4
5.4
0.1
0.2
0
87.5
79.7
1
0.1
May,
July,
Aug,
Sept and
Oct
July,
Aug and
Nov
268	*99.6% pasture/hay
269	* 18.7% shrub and grasslands
C-23

-------
270
271
272
273
274
275
276
Til
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
C3.2 Site Group Selection
Due to the limited number of individual sites with long-term data, we also tried
performing an analysis in the Blue Ridge ecoregion in which sites were grouped together to
obtain more long-term biological datasets. The dataset that was used for this analysis was
derived from reference sites (as designated by NCDENR) that were: 1. sampled from June-
September; and 2. sampled using the standard qualitative/full-scale collection method. The 15
sites that comprised the dataset are listed in Table C3-3 Locations of these sites are shown in
Figure C3-2 The years during which each of the sites was sampled are listed in Table C3-4.
Eighteen years of (non-continuous) data are available from 1983 through 2006. A genus-level
OTU was used to derive the taxa list, and relative abundance was used. The raw data from which
the relative abundances were calculated is categorical (l=rare (1-2 specimens), 3=common (3-9
species) and 10=abundant (10 or more species). When multiple sites were sampled in a year, the
mean value was calculated so that there was only one value for each trait or taxa per year (i.e. in
1983, Sites NC0107 and NC0109 were sampled. The one value that was used for 1983 was the
average value from those two sites).
Two main types of analyses were performed. One involved looking for trends among
individual taxa and the other involved searching for trends among traits. First, correlation
analyses were performed to see whether any taxa or traits were significantly correlated with year.
Next, correlation analyses were performed to see whether any taxa or traits were significantly
correlated with PRISM air temperature or precipitation data. To briefly summarize the results, 41
taxa (19 EPT, 22 non-EPT) were significantly correlated with year, 27 (13 EPT, 14 non-EPT)
were significantly correlated with at least one of the PRISM air temperature variables (minimum,
maximum or mean) and 19 (4 EPT, 15 non-EPT) were significantly correlated with PRISM
annual precipitation. Thirteen of the % individual trait metrics were significantly correlated with
year and 6 were significantly correlated with a PRISM variable. Seventeen of the taxa richness (#
of taxa) variables were significantly correlated with year and 8 were significantly correlated with
a PRISM variable. Plots and summary tables of the significant correlations are available upon
request. Results from this analysis are not included or used anywhere else in this report.
C-24

-------
300
301
302
303
304
StationID
# Years
Sampled
WaterbodyName
Location
LatDec
Long^Dec
NC0107
7
N FK NEW R
NC 16
36.50389
-81.39028
NC0109
11
NEW R
SR 1345
36.55222
-81.18333
NC0200
7
MILLS R
SR 1337
35.39861
-82.59500
NC0366
3
N FK MILLS R
SR 1341
35.39667
-82.62472
NC0806
2
THREE TOP CR
SR 1100
36.42806
-81.62389
NC0812
3
LITTLE R
SR 1128
36.46778
-81.13333
NC1006
1
W FK FRENCH BROAD R
OFF NC 281
35.18583
-82.95889
NC1285
2
CROOKED CR
SR 1135
35.60556
-82.11694
NC1289
1
S HOMINY CR
NC 151
35.53444
-82.69222
NC1438
2
S FK MILLS R
SR 1340
35.37583
-82.61500
NC1540
1
MILL CR
SR 1400
35.63667
-82.21861
NC1573
3
BOONE FK
SR 1561
36.12306
-81.77000
NC1591
1
BEECH CR
US 321
36.26111
-81.89667
NC1827
1
MACKEY CR
BE US 70
35.66972
-82.11417
NC2757
1
E FK PIGEON R
US 276
35.41056
-82.81000
Table C3-4
Year
# of Sites
Sampled


Sites

1983
2
NC0107
NC0109


1984
2
NC0109
NC0200


1985
3
NC0107
NC0109
NC0366

1986
2
NC0109
NC0200


1987
2
NC0107
NC0109


1988
2
NC0109
NC0200


1989
2
NC0107
NC0109


1990
3
NC0109
NC0200
NC1006

1992
1
NC0200



1993
6
NC0107
NC0109
NC0366 NC0806
NC0812 NC1438
1994
3
NC1540
NC1573
NC1591

1997
3
NC0200
NC1285
NC1289

1998
4
NC0107
NC0109
NC0806 NC0812

1999
1
NC1573



2002
5
NC0200
NC0366
NC1285 NC1438
NC1827
2003
3
NC0107
NC0109
NC0812

2004
1
NC1573



2006
1
NC2757



C-25

-------
Blue Ridge Reference Sites - Years Sampled
_Years_Sa
306
307	Figure C3-2. Locations of Blue Ridge reference sites that were used in this analysis.
308
309
310
311
C-26

-------
, APPENDIX D
2
3	Data Analyses Methods
4
5	The intent of this appendix is to present more comprehensive descriptions of the analytical
6	approaches and methods applied to evaluate the selected state biomonitoring data sets. Each
7	major question or approach is presented separately, with common methods described first, and
8	then any state-specific variations.
9
10
D-l

-------
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Dl. GENERATION OF TEMPERATURE-PREFERENCE AND TOLERANCE DATA
Temperature is an environmental parameter of particular interest in this project. We
therefore attempted to gather as much existing relevant temperature-preference and tolerance
information as possible and to use analyses of the state data sets to generate temperature
preference and tolerance information for as many taxa, defined by generic-level operational
taxonomic units (OTUs), as possible. The specific sources and types of existing temperature-
preference and tolerance information gathered in this study and their application in categorizing
temperature traits of OTUs are described in U.S. EPA (2011).
We used weighted average modeling or related approaches (e.g., maximum likelihood
estimates, general linear modeling) to estimate the optima values and ranges of occurrence
(tolerances) for temperature, and in some cases flow parameters, for each OTU from each state
that had a sufficient distribution and number of observations to support the analysis. The
methods described in Yuan (2006) were applied to derive temperature and flow optima and
tolerance values.
Weighted averaging is a simple, robust approach for estimating the central tendencies of
different taxa, or in our case, temperature optima and tolerance values (ter Braak and Looman,
1986). The basic approach is a straightforward weighted average—the temperature at each site in
a state at which the species is observed, multiplied by the relative abundance of the species at
that site, with the sum over all sites of the weighted temperatures divided by the sum of the
abundances of that species from all sites. This mean temperature is taken as the preferred
temperature for the taxon, and the breadth of the distribution (size of the standard deviation or
other measure of spread) represents an estimate of the tolerance or sensitivity of the taxon. The
approached is illustrated in Table D-l and Figure D-l.
D-2

-------
38
39
40
41
42
43
44
45
46
47
48
49
Table D-l. Example to illustrate the derivation of a weighted average model
temperature optimum (weighted mean) estimate
Species A Temperature preference

StationID
Relative
Observed
RA *

Abundance
Temperature
Temp

(RA)
(Temp)

A
0.10
22
2.20
B
0.02
33
0.66
C
0.02
12
0.24
D
0.04
14
0.56
SUM
0.18

3.66




Weighted Average = 3.66 0.18 = 20.3333
35
30
25
1°
h
OjO
5
0
5
10
15
20
25
30
Temperature ( C)
Figure D-l. Illustration of weighted average temperature distribution, where the
weighted average mean (u) is taken as the temperature optimum (preference) for the taxon,
and the magnitude of standard deviation (sd) is taken as an estimate of the temperature
sensitivity or tolerance.
When using weighted averages, a wide distribution of samples across the environmental
gradient results in a more robust estimate of temperatures of occurrence, and therefore, of
D-3

-------
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
inferred preference. For a given state data set, weighted average tolerance values for each OTU
are computed using the same set of environmental data; therefore, any bias arising from an
uneven distribution of data will be the same for all OTU, and their relative placement along the
temperature gradient will generally be preserved.
The generalized linear model is also used to estimate taxon-environment relationships for
each combination of taxon and environmental variable. In addition to providing a means of
computing tolerance values, regression estimates of the taxon-environment relationship quantify
the strength of the association between a given environmental gradient and changes in the
occurrence probability or abundance of a taxon. In the case of presence/absence data, the
response variable is modeled as a binomial distribution; in the case of abundance data, a negative
binomial distribution is often assumed (maximum likelihood estimates).
In our analyses, weighted average calculations were used for the states that had absolute
(non-categorical) abundance data by taxon (Maine, Utah and Ohio). If only presence/absence
(categorical or qualitative abundance) data were available, a generalized linear model was used
(North Carolina). Calculations were made separately for each state. Since use of the widest range
of temperature variation available is desired in this type of analysis, all stations within each state
across all ecoregions were retained in each state analysis. However, data were subset to account
for seasonal variation (when needed), as well as for variation associated with different sampling
methods. For example, in Utah, only samples collected during the fall index period were used. In
North Carolina, only samples collected by the full collection method were analyzed.
Several statistical models were run (using R and C2 software), and model performance
was compared for possible improvement of the weighted average model, weighted average
partial least square regression (WA-PLSR), and maximum likelihood (ML). The WA-PLSR
model result was difficult to interpret and only slightly improved the WA model. The ML model
had similar performance to the WA model; therefore the WA model was used when sample sizes
were sufficiently large (>500 samples). Only taxa occurring at more than 9 sites were included in
the WA analyses. Low sample size affects the regression model and biases the optima and
breadth values for rare taxa, especially under extreme conditions.
Based on the derived optima and tolerance values for each OTU analyzed for each state,
we defined optima and tolerance rankings to support relative comparisons among taxa and
D-4

-------
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
regions. For example, relative rankings of taxa as cold or warm preference can be compared
between Utah, where most samples were collected in the fall, and Maine, where most were
collected in the summer, which might otherwise result in differences in absolute temperature
values. It also allowed comparison of the Utah results from this study to be compared to results
from other western datasets, which were generally based on summer samples and therefore had
noticeable differences in ranges of absolute temperatures used as thresholds for designating cold-
and warm-preference taxa (Herbst and Silldorff, 2007; Brandt, 2001). Comparison to these
results was used to support final designation of taxa membership in cold-preference and warm-
preference groups (U.S. EPA 2011).
Ranks were defined separately for temperature optima and tolerances using a scoring
system. Both the temperature optima for all taxa in a state and the standard deviations were
divided into the following percentiles: 0.1, 0.25, 0.4, 0.6, 0.75, 0.9, 1. Taxa associated with the
lowest temperature optima and those with the smallest standard deviations (narrowest tolerance
ranges), i.e., those in the lowest 10th percentile, received scores of 1. Those in the next percentile
category (>0.1 up to the 25th percentile) received a score of 2, and so on, up to the highest
temperature optima and widest tolerance ranges (the 90th percentile or greater) which received a
score of 7 (Figure D2-2). Lower ranks for temperature optima reflect preference for colder water,
and higher ranks reflect preference for warmer water. It was a relatively arbitrary judgment to
include taxa with optima rankings of 1, 2 or 3 as cold water taxa and those with rankings of 5, 6
or 7 as warm water taxa. Similarly, standard deviation ranks of 1, 2 or 3 were considered
sensitive (e.g., stenothermal), while ranks of 5, 6, or 7 were considered tolerant (eurythermal).
Percentile
Optimum
Breadth
0
4.57029
2.02959
0.1
6.847701
2.770389
0.25
7.722833
3.317888
0.4
8.411832
3.600784
0.6
9.188384
3.812142
0.75
9.689138
3.997378
0.9
10.5325
4.442977
1
15.7144
5.06721
Rank
2	\ Cold/Stenotherms
3	J
4
5	^
6	J Warm/Eurytherms
D-5

-------
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
Figure D2-2. Example taken from Utah analysis results to illustrate development of
ranking for temperature (or other environmental parameter) preference and tolerance
rankings from weighted average or GLM temperature distribution results.
D2. EVALUATION OF BIOLOGICAL RESPONSES TO CLIMATE VARIABLES
D2.1 Characterization of Years as Proxy for Future Climate Conditions
To evaluate responses of a variety of biological metrics, trait and taxonomic groups, as
well as indices and predictive-model results to differing climatic conditions that could be
expected, we used extremes in climate variables among existing data as proxies for future
climate conditions. We partitioned data at long-term reference stations in each state into years
characterized by hotter (>75th percentile of the temperature distribution during years of
biological collections), colder (<25th percentile of temperature), and normal (25th to 75th
percentile) average annual air temperatures. Using similar thresholds, years were partitioned
based on average annual precipitation into wetter, drier, and normal years. When flow data were
available, a similar partitioning of high and low flow years was applied. An assumption is that
these temperature, precipitation, and flow differences drive responses in benthic communities
that are reasonable proxies for the types of community changes that can be expected over the
long term with climate change. Another assumption is that PRISM air temperature is a
reasonable surrogate for water temperature, and PRISM precipitation for flow (see Appendix A,
Section A.l for substantiation and references). Table D2-1 summarizes how these categories
were grouped and designated for ANOVAs at long-term references stations among the three
states analyzed.
Table D2-1. Descriptions of the temperature, precipitation, and flow (IHA parameters)
categories that were used in ANOVA analyses for long-term reference stations in all states.
Note that flow was only available for the Maine long-term station 56817.	
Variable	Description
D-6

-------
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
Cat 1 Temp
CatlPrecip
Cat2_Temp
Cat2_Precip
Temperature categories: l=coldest years (defined as years when the
PRISM mean annual air temperature was < 25th percentile of the
overall temperature values); 2=normal years (25th-75th percentile),
3=hottest years (>75th percentile).
Precipitation categories: l=driest years (defined as years when the
PRISM mean annual precipitation was < 25th percentile of the overall
precipitation values) 2=normal precip year (25th-75th percentile),
3=wettest years (>75th percentile).
Temperature categories: l=coldest years & normal years (defined as
years when the PRISM mean annual air temperature was < 75th
percentile of the overall temperature values); 2=hottest years (>75th
percentile).
Precipitation categories: l=driest years (defined as years when the
PRISM mean annual precipitation was < 25th percentile of the overall
precipitation values); 2=normal years and wettest years (> 25th
percentile).
IHA median monthly flows averaged across July-September: l=years
Cat_Flow	with the lowest flow (<25th percentile; 2=years with normal flow (25th-
75th percentile), 3=years with the highest flow (>75th percentile).
D2.2 Reference Stations and Seasonal Data used in Analyses
For Maine, the various ANOVA and correlation analyses described in Section D2.3 were
conducted at stations 56817, 57011, and 57065 (see Appendix CI for details). These 3 sites, all
located in the Laurentian Plains and Hills ecoregion, are reference sites (rated as Class AA by the
Maine DEQ) that have the longest-term biological data. Only rock basket samples that were
collected from June-November were used in the analysis.
For Utah, the various ANOVA and correlation analyses were conducted at stations
4927250 and 5940440 in the Wasatch and Uinta Mountains and at stations 4951200 and 4936750
in the Colorado Plateau (see Appendix C2 for details). These represent reference locations with
the longest-term biological data records available. Many of the analyses also were performed on
reference stations grouped into three site groups: the Wasatch and Uinta Semi-arid Foothills, the
Wasatch and Uinta Mid-elevation Mountains, and the Colorado Plateau. Only samples collected
during the fall season were used in analyses.
For North Carolina, analyses were conducted at five reference sites—Stations NC0109,
NC0207, NC0209, NC0075 andNC0248; although only one, NC0109, had relatively long-term
D-7

-------
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
data (>10 years) (see Appendix C3 for details). Two site groups were used—the Blue Ridge and
Piedmont EPA level 3 ecoregions, which are very similar to the NCDENR Mountain and
Piedmont ecoregions. All samples in the NC database that were collected using the standard
qualitative method during the summer index period (June-September) were used in this analysis.
D2.3 ANOVA
One-way ANOVA tests were used to evaluate whether significant differences exist
among various mean metric values from samples collected at the selected long-term reference
sites for each state during hot, cold, wet, dry, and normal years. Numerous biological metrics
were tested for all states (Table D2-2). The O/E metric was also tested in Utah. The ecological
trait groups of cold-water and warm-water-preference taxa were tested for differences among
hot, cold, wet, dry and normal years. If the p-levels from the Tukey honest significant difference
(HSD) test for unequal sample size (N) (Spjotvoll/Stoline) were less than 0.05, the differences in
metric values among the different temperature and precipitation groups were considered to be
significant.
We examined the distributions of cold- and warm-water-temperature indicator taxa to try
to identify areas that are more likely to be 'vulnerable' to the effects of climate change, in
particular the increase in temperature. One-way ANOVAs were used to determine whether
significant differences exist between the number of cold- and warm-water taxa between
ecoregions and between elevations. All samples in each state data base were used in these
(spatial) analyses. Ecoregions included for each state were:
•	Maine—Laurentian Plains and Hills, Northeastern Highlands and Northeast Coastal
Zone.
•	Utah—Wasatch and Uinta Mountains and Colorado Plateaus level 3 ecoregions.
•	North Carolina— Mountain, Piedmont and Coastal ecoregions.
Specific elevation categories varied among states:
•	Maine—sites < 150 m and > 150 m.
•	Utah— sites < 2000 m and > 2000 m.
•	North Carolina—sites < 500 m and > 500 m.
D-8

-------
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
D2.3.1 Maine ANOVAs
The analyses using Maine's data used ANOVA to explore the relative importance of the
various input metrics used in the linear discriminant models for classification of station condition
and to relate these to any effects on the metrics due to climate change. These analyses also
examined (1) how model input values differ among the different station classifications; (2) how
much metric values have to change for a sample to change classification, e.g. from Class A to
Class B (or B to C, etc.); (3) whether certain metrics are more important than others in
contributing to classification changes; and (4) whether certain metrics are more likely to be
affected by climate change than others, and if so, how they are affected, and how this affects
overall classification. Understanding these aspects of the data is difficult, because Maine's
classification models look at multiple variables simultaneously, and because there are no firm
thresholds or metric values at which a sample goes from being a Class A to Class B, etc.
Instead, ANOVA was used on all the samples in the Maine database to see how mean
metric values differed among the different classes. At Station 56817, which has the longest-term
biological data and is considered by Maine DEP to be a reference site, 9 of the 22 annual
samples collected from 1985 to 2006 were classified as Class B, while all the others were Class
A. We used one-way ANOVA to determine which model input metrics had significantly
different mean values between the Class A and Class B samples.
To determine which of Maine's station classification discriminant model metrics are
affected by climate-related variables (temperature, precipitation and flow), one-way ANOVA
tests were used to evaluate whether significant differences exist between mean model input
metric values from samples collected during hot, cold, wet, dry, low flow, high flow, and normal
years.
Because Maine's linear discriminant models are not used in other northeastern states, we
performed ANOVAs on metrics that are commonly used to assess streams in northeastern states
and determine if significant differences occurred between hot, cold, wet, dry, and normal years.
Among the most commonly-used metrics are total taxa, EPT taxa, Ephemeroptera taxa,
Plecoptera taxa, Trichoptera taxa, Hilsenhoff Biotic Index (HBI), an assortment of functional
feeding group and habit metrics, percent dominant taxon and the Shannon Wiener diversity
index. In addition, a variety of other biological metrics were evaluated, as listed in Table D2-2.
D-9

-------
208
209
210
211
212
213
It should be noted that richness values are affected by the operational taxonomic unit (OTU) that
is used in the analysis. A mostly genus-level OTU was used in this analysis because this
taxonomic level was found to be most appropriate for the long-term Maine dataset.
Table D2-2. List of biological metrics that were evaluated in Maine, Utah, and North
Carolina.	
Metric	Descriptions
Total taxa
Ephemeroptera taxa
Trichoptera taxa
Plecoptera taxa
EPT Taxa
Percent Plecoptera
#	of Total taxa
#	of Ephemeroptera taxa
#	of Trichoptera taxa
#	of Plecoptera taxa
#	of Ephemeroptera, Plecoptera and Trichoptera taxa
Percent individuals in the Order Plecoptera
Percent EPT
Percent individuals - Ephemeroptera, Plecoptera and Trichoptera
HBI	Hilsenhoff Biotic Index (calculated using New Mexico tolerance values)
Clinger Taxa
Habit - number of clinger taxa
Swimmer Taxa
Habit - number of swimmer taxa
Burrower Taxa
Habit - number of burrower taxa
Climber Taxa
Habit - number of climber taxa
Sprawler Taxa
Habit - number of sprawler taxa
Percent Clinger
Habit - percent clinger individuals
Percent Swimmer
Habit - percent swimmer individuals
Percent Burrower
Habit - percent burrower individuals
Percent Climber
Habit - percent climber individuals
Percent Sprawler
Habit - percent sprawler individuals
Collector-gatherer Taxa
Functional Feeding group - number of collector-gatherer taxa
Collcctor-filtcrer Taxa
Functional Feeding group - number of collcctor-filtcrer taxa
Shredder Taxa
Functional Feeding group - number of shredder taxa
Herbivore/Scraper Taxa
Functional Feeding group - number of herbivore/scraper taxa
Predator Taxa
Functional Feeding group - number of predator taxa
Percent Collector-gatherer
Functional Feeding group - percent collector-gatherer individuals
Percent Collcctor-filtcrer
Functional Feeding group - percent collcctor-filtcrer individuals
Percent Shredder
Functional Feeding group - percent shredder individuals
Percent Herbivore/Scraper
Functional Feeding group - percent herbivore/scraper individuals
Percent Predator
Functional Feeding group - percent predator individuals
D-10

-------
Shannon Wiener DI
% DominantO 1 taxa
Shannon Wiener Diversity Index (log2)
Percent dominant taxon individuals
TempCoreColdPct
Thermal Preference and Tolerance -Percent cold water individuals
TempCoreWarmPct
Thermal Preference and Tolerance -Percent warm water individuals
TempCoreColdTax
Thermal Preference and Tolerance -Number of cold water taxa
T empCoreWarmT ax
PerennialPct
IntermitPct
DroughtPct
Drier WinPct
Thermal Preference and Tolerance -Number of warm water taxa
Percent perennial stream individuals (these taxa require water for their
entire life cycle).
Percent intermittent stream individuals (these taxa are found in perennial
streams but tend to be more dominant in numbers in intermittent
conditions).
Percent individuals that possess at least one of the following traits: ability
to survive desiccation, adult ability to exit, respiration plastron/spiracle
Percent individuals that possess the most number of traits states that are
predicted or have been shown to be most favorable in a drier climate
scenario
Drier LoserPct
Percent individuals that have the fewest favorable trait states and the
most number of unfavorable trit states in a drier climate scenario
WarmDrierLoserPct
OCH Pet
Percent individuals that have the fewest favorable trait states and the
most number of unfavorable trait states in a warmer drier climate
scenario
Percent individuals - Odonata, Coleoptera and Hemiptera
D-ll

-------
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
Table D2-2. Continued
Metric
Descriptions
PerennialTax
Number of perennial stream taxa (these taxa require water for their entire
life cycle)
IntermitTax
Number of intermittent stream taxa (these taxa are found in perennial
streams but tend to be more dominant in numbers in intermittent
conditions).
DroughtTax
Number of taxa that possess at least one of the following traits: ability to
survive desiccation, adult ability to exit, respiration plastron/spiracle
WarmDrier LoserTax
Number of taxa that have the fewest favorable trait states and the most
number of unfavorable trait states in a warmer drier climate scenario
Drier LoserTax
Number of taxa that have the fewest favorable trait states and the most
number of unfavorable trait states in a drier climate scenario
OCHTax
Number of Odonata, Coleoptera and Hemiptera taxa
D2.3.2 Utah ANOVAs
Biological metrics that are commonly used to assess streams in southwestern states were
selected for this analysis, for example from Idaho, New Mexico, Colorado, Nevada, Wyoming,
Montana and Arizona. The list of metrics that were evaluated is shown in Table D2-2 (O/E was
also analyzed). Richness values are affected by the OTU that is used in the analysis. A mostly
genus-level OTU was used in this analysis because this taxonomic level was found to be most
appropriate for the long-term Utah dataset. Some taxa, such as Chironomidae, were grouped to
family level and higher.
O/E scores were evaluated to provide information on the sensitivity of O/E scores to
changes in annual temperature and precipitation. Scores used in the ANOVAs were calculated
for Stations 4927250, 4951200, 4936750 and 5940440 using the fall Utah RIVPACS model.
These calculations involved changing O and keeping predictor variables, which are long-term
averages, constant. The OTUs used by the Utah DEQ in model construction were retained, so
that taxa lists would be consistent across years. Because the model was developed using a subset
of more recent data (about 5 years worth), it may not perform as well on older datasets.
D-12

-------
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
Therefore, these OTUs may not be as appropriate for the longer-term datasets that were analyzed
in this exercise.
D2.3.3 North Carolina ANOVAs
In Maine and Utah we were able to perform one-way ANOVA tests to evaluate whether
significant differences exist between mean metric values from samples collected at selected sites
during hot, cold, wet, dry, and normal years. For the North Carolina dataset, we only had
sufficient data to conduct this type of analysis on site NC0109 (11 yrs data). Correlation
analyses were also used to evaluate relationships between the selected metric values and mean
annual air temperature and precipitation variables at NC0109 as well as at the other sites (see
Section D2.4).
D2.4 Correlation Analyses
Correlation analyses were performed in all states to test the relationships between the
biological metrics listed in Table D2-2 and year to test for temporal trends; or annual average air
temperature or precipitation to examine basic relationships to climate variables. In Utah
correlation analyses were also performed to explore relationships between O/E values and
climatic variables for each site and sampling year.
In North Carolina, to further explore the relationship between temperature indicator taxa,
tolerance values, the NCBI and climate-related variables, three different correlation analyses
were performed: 1. correlation analysis of temperature optima values vs. tolerance values; 2.
correlation analysis of temperature-indicator metrics at selected Mountain and Piedmont
reference sites (percent cold- and warm-water-indicator individuals and number of cold- and
warm-water-indicator taxa) vs. NCBI scores; and 3. correlation analysis of BI values and PRISM
mean annual air temperature and PRISM mean annual precipitation. The correlation analyses
were performed on datasets that used genus-level tolerance values. Tolerance values can vary
within some genera, and therefore, these NCBI scores may vary somewhat from NCDENR BI
scores (but they are generally close).
Because not all southeastern states use EPT taxa richness and the NCBI to rate biological
sampling sites, we performed additional analyses on metrics listed in Table D2-2. Richness
D-13

-------
263
264
265
266
267
268
269
270
271
272
273
274
275
276
Til
278
279
values are affected by the OTU that is used in the analysis. A mostly genus-level OTU was used
in this analysis because this taxonomic level was found to be most appropriate for the long-term
NC dataset.
D2.5 NMDS Ordinations
Non-metric Multidimensional Scaling (NMDS) ordinations were performed on data from
selected reference stations with sufficient long-term data:
•	Maine—Station 56817;
•	Utah— Stations 4927250 and 4951200;
•	North Carolina—Insufficient data .
NMDS is an ordination that takes the taxa in the samples and shows in ordination space
how closely related the samples are based on their species composition. NMDS was performed
using PCOrd (McCune & Mefford, 1999), a Sorensen distance measure, and a maximum of 3
axes. Annual samples were categorized based on hot/cold/normal, and wet/dry/normal years to
assess patterns. We also used the environmental variables described in Table D2-3 to group the
data while looking for trends.
D-14

-------
281
282
283
284
285
286
287
288
289
290
291
Table D2-3. Summary of the environmental variables that were used to group data and
look for trends in the NMDS
Variable
Description
Temperature categories: l=coldest years (defined as years when the
PRISM mean annual air temperature was < 25th percentile of the
overall temperature values); 2=normal years (25th-75th percentile),
3=hottest years (>75th percentile).
CatTemp
CatPrecip
tmeanl4
pptl4
Pre vY rtmean 14
PrevYr_pptl4
tmeanl4_absdifc
ppt 14_absdifc
Precipitation categories: l=driest years (defined as years when the
PRISM mean annual precipitation was < 25th percentile of the
overall precipitation values) 2=normal precip year (25th-75th
percentile), 3=wettest years (>75th percentile).
PRISM mean annual air temperature
PRISM mean annual precipitation
PRISM mean annual air temperature from the previous year (lag
effects)
PRISM mean annual precipitation from the previous year (lag
effects_
Absolute difference between the PRISM mean annual air
temperature from the sampling year and the previous year
Absolute difference between the PRISM mean annual precipitation
from the sampling year and the previous year
In addition, for Maine:
MonthMed
Flash
ldmin
3d_min
ldmax
3d max
IHA - Average of median flow values from luly, August and
September
R-B Flashiness Index
IHA - 1 day minimum flow
IHA - d day minimum flow
IHA - 1 day maximum flow
IHA - 3 day maximum flow
D3 EFFECTS OF TEMPERATURE-SENSITIVITY TRAITS GROUP COMPOSITION
ON VARIOUS STATE METRICS AND INDICES
D3.1 How cold- and warm-water-indicator taxa may affect EPT metrics, HBIs and BCG
tier assignments and how these effects may vary across different ecoregions in Maine
Because annual air temperature is predicted to increase as a result of climate change,
temperature-preference and tolerance traits are of particular interest. We examined how changes
D-15

-------
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
in species composition resulting from replacement of cold-water-indicator taxa may affect state
assessment methods. Specifically, we examined potential effects on EPT metrics and the HBI
because these are commonly used in assessments in many states. In addition, in Maine, we
evaluated Biological Condition Gradient (BCG) attribute levels that were assigned to the
temperature indicator taxa during The New England Wadeable Stream Survey (NEWS) (US
EPA 2007); Class A indicator taxa temperature preferences and tolerances; and the distribution
of the temperature indicator across the different level 3 ecoregions to see whether some
ecoregions are more likely to be more vulnerable to climate change effects than others.
D3.2 How cold- and warm-water-indicator taxa may affect indices in North Carolina
We evaluated relationships between temperature indicator taxa and EPT taxa richness,
the NCBI and final bioclassification scores. This involved looking at the number of temperature-
indicator taxa that are EPT taxa and the tolerance values of the temperature-indicator taxa. In
addition, we evaluated two different scenarios: 1. a 'worst case' scenario in which all the cold-
water-temperature-indicator taxa at selected Mountain reference sites were dropped; 2. a scenario
in which Mountain criteria were applied to biotic assemblages at selected reference Piedmont
sites (this simulated a scenario in which taxa that are typically found in Mountain sites are
replaced by taxa that typically inhabit Piedmont sites). For both scenarios, we evaluated how this
affected the EPT richness, NCBI and final bioclassification scores.
D-16

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
APPENDIX E
Detailed Results for Maine
The intent of this appendix is to provide more comprehensive and detailed information on the
large number of analyses that were performed on the Maine data. Some of the analyses that are
covered in this appendix are also referenced (generally in less detail) in the main body of the
report. When this occurred, attempts were made to reduce any overlap or duplication in the
reporting of results.
El. Overview of Maine's Linear Discriminant Model
E2. Maine Ecoregion Descriptions
E3. Results
Attachment El - Results of the ANOVA analysis in which mean model
input metric values were compared across Class A, B, C and NA
samples
Attachment E2 - Temperature Indicator Taxa - Maine
Attachment E3 - Tolerance values and BCG attribute levels of Maine's
temperature indicator taxa
E-l

-------
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
El OVERVIEW OF MAINE'S LINEAR DISCRIMINANT MODEL
Information in this section was provided by Maine DEP (see Davies, SP; Tsomides, L.
2002. Methods for Biological Sampling and Analysis of Maine's Rivers and Streams. DEP
LW0387-B2002. Prepared for the State of Maine Department of Environmental Protection.
http://www.maine.gov/dep/blwq/docmonitoring/biomonitoring/materials/finlmethl.pdf).
Maine DEP rates sites using aquatic life decision models. These are four statistical
models that use 30 variables of the macroinvertebrate community to determine the strength of
association of a sample community to Maine's water quality classes. The first stage model acts
as a screen and gives the strength of association of the sample to each of the different water
quality classes. This model provides four initial probabilities that a given site attains one of three
classes (A, B, or C) or is in nonattainment (NA) of the minimum criteria for any class.
Association values are computed for each classification using one four-way model and three
two-way models. These probabilities have a possible range from 0.0 to 1.0 and are used, after
transformation, as variables in each of the three subsequent second stage or final decision
models. Each of the four linear discriminant models uses different variables, providing
independent estimates of class membership. The same criterion is applied to all sites. A flow
chart depicting decision criteria is shown in Figure El-1. The protocol is outlined in the Maine
DEP methods manual (Davies and Tsomides 2002).
E-2

-------
Process for Determining Attainment Class Using Association Vatass
Is tb* sample appropriate for IBM*
41
42
43
44
45
46
47
48
49
50
YSS
\

~



v.. - • y
1
r
Is the sample class C ar b«ttar?
pAt Z _ j 0

. - .pA5C - id
<_

f

1
| - - - J f A-1.V-: ")(j^unLuum' \ NA ")
"	4		
( NA
V. _J
Ii th« swplt class B nr btttw?

pAB>0.6

0.4 < pAB <0.6
pAB < 0.4
/ \
f At least 1 I-
r
		
y	-\ y
\ A; Nv B '|f iMa-im.u'S ^
	i	
c
Is the sample class A?

1.
0.4
JL ¦ ^
f , > f
i 	) t
T
1 r~» )
\	j
Figure El-1. Flow chart that outlines the process that Maine DEP uses for determining
attainment class using association values from its 4 linear discriminant models (chart by
Thomas J. Danielson, taken from ME DEP 2002 monitoring manual).
The variables used in the first stage model are variables important to the evaluation of all
classes. Of the nine variables used in the first modeling stage, 5 measure abundance, 2 measure
richness, and 2 variables are biotic indices involving tolerance to pollution and abundance. The
first stage model uses the following nine variables: total abundance, generic richness, Plecoptera
E-3

-------
51
52
53
54
55
56
57
58
59
60
61
62
and Ephemeroptera abundance, Shannon-Wiener Generic Diversity Index, Hilsenhoff Biotic
Index (HBI), Relative Abundance Chironomidae, Relative Richness Diptera and Hydropsyche
Abundance. A list of all the model input metrics can be seen in Table El-1.
The final decision models (the three, two-way models- C or Better Model, B or Better
Model, or A Model.) are designed to distinguish between a given class and any higher classes as
one group and any lower classes as another group (e.g. Classes A+B+C vs NA; Classes A+B vs
Class C+NA; Class A vs Classes B+C+NA). The equations for the final decision models use the
predictor variables relevant to the class being tested. The process of determining attainment
class using association values is outlined in Appendix F of the ME DEP methods manual (Davies
and Tsomides 2002). Application of the three second-stage models or two-group tests is
hierarchical.
E-4

-------
Table El-1. Metrics that are used in Maine's Linear Discriminant Models
#
Metric
Model
1
Total Abundance
First Stage Model
2
Generic Richness
First Stage Model
3
Plecoptera Abundance
First Stage Model
4
Ephemeroptera Abundance
First Stage Model
5
Shannon-Wiener Generic Diversity
First Stage Model
6
Hilsenhoff Biotic Index
First Stage Model
7
Relative Abundance Chironomidae
First Stage Model
8
Relative Richness Diptera
First Stage Model
9
Hydropsyche Abundance
First Stage Model
10
Probability (A+B+C) from First Stage Model

11
Cheumatopsyche Abundance
C or Better Model
12
EPT Generic Richness Divided by Diptera Generic Richness
C or Better Model
12
Relative Abundance Oligochaeta
C or Better Model
13
Perlidae Abundance
B or Better Model
14
Tanypodinae Abundance
B or Better Model
15
Chironomini Abundance
B or Better Model
16
Relative Abundance Ephemeroptera
B or Better Model
17
EPT Generic Richness
B or Better Model
18
Summed Abundance's of: Dicrotendipes (warm), Micropsectra,
Parachironomus and Helobdella
B or Better Model
19
Relative Generic Richness Plecoptera
A Model
20
Summed Abundances of: Cheumatopsyche, Cricotopus, Tanytarsus and
A Model
Ablabesmyia
21
Summed Abundances of: Acroneuria, Maccaffertium and Stenonema
A Model
22
EP Generic Richness/14
A Model
23
Class A Indicator Taxa/7
A Model

-------
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
E2 MAINE ECOREGION DESCRIPTIONS
Northeastern Highlands. This is a relatively sparsely populated region located in the
western part of Maine. It is characterized by hills and mountains, a mostly forested land cover,
nutrient-poor frigid and cryic soils (mostly Spodosols), and numerous high-gradient streams and
glacial lakes. Typical forest types include northern hardwoods (maple-beech-birch), northern
hardwoods/spruce, and northeastern spruce-fir forests. Recreation, tourism, and forestry are
primary land uses (Hellyer draft Ecoregion descriptions 2007). On average, biological sampling
sites in this ecoregion are located at higher elevations (average of 829 feet) and have lower urban
and agricultural land use within 1 km of the sites (averages of 12% and 7%, respectively).
Unfortunately, sites that met our selection criteria (<5% urban and <10% agricultural land use
within a 1 km buffer) lacked long-term data (note: the most number of years of data at
Northeastern Highland sites was 9 years, and these sites were classified as 'C'). At one of the
selected sites, there were 3 years of data. The remaining selected sites only had 1 or 2 years of
data, and this data had mostly been collected from 2000 onwards. When we attempted to search
for long-term trends in a limited dataset comprised of a group of Northeastern Highland sites,
there was not enough data to effectively work with and the site groups did not work well (see
Appendix C). If trends were observed, they appeared to be due to site-specific differences rather
than climate-related changes.
Northeastern Coastal Zone. This region is located in the southeastern corner of Maine.
This ecoregion contains much greater concentrations of human population than the Northeastern
Highlands (including the city of Portland). Current land use mainly consists of forests,
woodlands, and urban and suburban development, with only some minor areas of pasture and
cropland. Forests are mostly white, red, and jack pine and oak-hickory, and the soils are
generally Inceptisols and Entisols (Hellyer draft Ecoregion descriptions 2007). Sites in this
ecoregion are located at lower elevations (average of 97 ft) and have higher urban land use
within 1 km of the sites (average of 44%). Only one site in this ecoregion met our selection
criteria and it only had 2 years of data, so no attempts were made to perform trend analyses on
Northeastern Coastal Zone sites.
E-6

-------
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
Laurentian Plains and Hills (or Maine/New Brunswick Plains and Hills). This is a
mostly forested region located in the eastern part of Maine. It has dense concentrations of
continental glacial lakes, is less rugged than the Northeastern Highlands, and is considerably less
populated than the Northeastern Coastal Zone. Vegetation is mostly spruce-fir with some patches
of maple, beech, and birch. The majority of biological sampling sites are located in this
ecoregion. The average elevation of sites is 214 ft and average land use is 23% urban and 13%
agricultural.
E3 RESULTS
E3.1 ANOVA - Comparison of mean model input metric values among the different
classifications using all the samples in the Maine database
Detailed results of the ANOVA can be found in Attachment El. There were significant
differences between mean model input metric values among many of the classes. The amounts
that the mean metric values changed between the different classes varied and are therefore
difficult to summarize. Also, looking at each metric individually has limited value because the
linear discriminant models look at multiple variables simultaneously. Results from the Station
56817 analyses were used to identify which of the 24 metrics were most likely to be influenced
by climate-related changes. Those results are summarized below.
E3.2 ANOVA - Comparison of mean model input metric values between Class A and
Class B samples at Station 56817
Comparison of Class A versus Class B samples at Station 56817 showed that the only
model input metric that had a significantly different mean value between the Class A and Class B
samples was the Hilsenhoff Biotic Index (HBI). Mean HBI values were significantly higher in
the Class B samples (Figure E3-1). Because the HBI model input metric was not significantly
related to climatic variables (as shown in the correlation analyses with PRISM mean annual air
temperature and precipitation and by the ANOVA analyses comparing mean values across
hot/cold/wet/dry/normal years), it is likely that non-climatic factors, such as non-point source
pollution, contributed to the change in classification.
E-7

-------
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
5.0
n Median
~ 25%-75%
X N on-Outlier Range
o Outliers
s* Extremes
Figure E3-1. Box and whisker plots for the HBI model input metric at Maine Station
56817 for samples that received different classifications (Class A versus Class B).
E3.3 ANOVA - Station 56817- hot/cold/wet/dry/normal years
There were differences in some of the model input metric values from samples collected
at Station 56817 during hot, cold, wet, dry, low-flow, high-flow and normal years, but none of
them were significant when tested with the Tukey honest significant difference (HSD) test for
unequal sample size (N) (Spjotvoll/Stoline). There were, however, significant correlations
between six of the metrics and precipitation or flow variables (Table E3-1).
The Class A indicator taxa metric (which equals the number of Class A indicator taxa
divided by 7) was significantly correlated with both mean annual precipitation (pptl4) and the
categorical precipitation variable (l=dry years; 2=normal years; 3=wet years). Class A indicator
taxa include: Brachycentrus (Trichoptera: Brachycentridae), Serratella (Ephemeroptera:
Ephemerellidae), Leucrocuta (Ephemeroptera: Heptageniidae), Glossosoma (Trichoptera:
Glossosomatidae), Paragnetina (Plecoptera: Perlidae), Eurylophella (Ephemeroptera:
Ephemerellidae), and Psilotreta (Trichoptera: Odontoceridae). At Station 56817, on average,
more Class A indicator taxa were present during wetter years (Figure 2-25 in main report). The
relative abundance of collector-gatherers was higher during higher flow years (Figure E3-2)
E-8
4 8
4 6
4.4
4 2
4 0
o 38
z
u
3 8
34
3.2
3.0
2.8


~
Class

-------
145
146	Table E3-1. Summary of results of the correlation analysis using data from Station 56817.
147	Only the significant correlations are shown	
Climate-related Variables
Model Input Metric	Catl Precip ppt!4 Cat Flow Avg Median Flow
I30-PRESENCE OF A
INDICATOR TAXA
r=.4686
N=22
p=.028
r=.5407
N=22
p=.009


I08-RELATIVE DIPTERA
RICHNESS


r=-.4316
N=22
p=.045

I12-EPT GENERIC
RICHNESS DIVIDED BY
DIPTERA RICHNESS


r=.4350
N=22
p=.043
r=.6214
N=22
p=.002
I16-TANYPODINAE
ABUNDANCE


r=-.5049
N=22
p=.017

131-EPT GENERIC
RICHNESS RELATIVE TO
EPT PLUS DIP


r=.4505
N=22
p=.035
r=.5831
N=22
p=.004
I33-COLLECTOR-
GATHERERS RELATIVE
ABUNDANCE


r=.4346
N=22
p=.043

148
E-9

-------
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
0.40
UJ
o
<0 35
o
0 30
0 25
co
K
0.20
C£
O
o
o
0 15
0 10
U 05
Lowest
H qt t
~	Median
~	25%-75%
I Non-Outlier Range
o Outliers
* Extremes
Normal
Year Groupings
Figure E3-2. Box and whisker plot for the Collector-gatherers relative abundance metric.
Samples were grouped by the following flow categories (Station 56817): l=low flow years,
2= normal years, 3=high flow years.
E3.4 NMDS ordination - Station 56817 - hot/cold/wet/dry/normal years
Results from the NMDS ordination show that samples from Station 56817 do not form
distinct clusters when grouped by hot/cold/wet/dry/normal years, so species composition did not
change in a consistent way when the climate-related variables changed (Figure 2-24 in main
body of report and E3-3). The plots that are shown are for the 2nd and 3rd axes because these axes
explained the greatest amount of variance (Axis 3 in particular). The environmental variable that
is most highly correlated with Axis 3 is the absolute difference between the PRISM mean annual
precipitation from the sampling year and the previous year (r=-0.377). This variable is also the
most highly correlated variable with Axis 2. The 2 minimum flow IHA parameters (1-day and 3-
day minimum flow) have the next strongest correlations with Axis 3 (r=0.35 for both). There is
an outlying sample in the plots. In 2005, mean annual precipitation was much higher than normal
(and was much higher than the previous year) but minimum annual flows and median flows
during the sampling months were relatively low (a lot of the rain occurred in October).
E-10

-------
169
170
171
172
173
174
175
176
177
178
Maine StationID 56817
. 1989
1990
Cat_Prec
A1
~ 2
3
1992
~
19S7
~
1991
2004
CO

X
1994
~
1993
A
1986
A
^ 1997
1996
1998
2003
2001 Sk.AbsD P
2006
1985
1999
1988
2002
1995
A
A
± 2000
2005
A
Axis 2
Figure E3-3. NMDS plot (Axis 3-2). CatPrec refers to the precipitation categories, which
are: l=dry years; 2=normal years; 3=wet years. Samples are labeled by collection year.
Absolute difference between the PRISM mean annual precipitation from the sampling year
and the previous year (AbsD P) is the most strongly correlated environmental variable
with Axes 2 & 3.
Figure E3-4 shows which taxa are the strongest drivers along the 2"d and 3rd axes.
Tricorythodes, Oecetis and Ablabesmyia have the strongest negative correlations with Axis 3,
E-ll

-------
179	and Pseudocloeon has the strongest positive correlation with Axis 3. Closer examination of these
180	taxa plotted in ordination space shows that none of them occur exclusively in a particular
181	temperature or precipitation category, although Ablabesmyia, Tricorythodes and Pseudocloeon
182	did occur more often in samples collected during normal precipitation years (Figure E3-5).
183
Maine StationID 56S17
CO
CO
x
Arentrp^ Physella ^Heptagen
T ri aenodJ^WydTScfiri^
Stylaria ^
Ephemere
~
Amnicola
Ablabestr
Oecetis
Boyeria
Stenochi
Lepidcisf^phernere
Psephenu
i A
.EJineutus
"(§>ten
Helicops
Pentaneu
Tricoryt
184
185
186
Axis 2
Figure E3-4. NMDS plot (Axis 3-2) that shows which taxa are most highly correlated with
each axis.
E-12

-------
Maine Station 56817 Tricorythodes
Maine Station 56817 Ablabesmyia
Oil riH.:
187
188
Maine Station 56817 Oecetis
Pro.;
189
190
191
192
193
~ -ii r-.
Maine Station 56817 Pseudocloeon
r-H.
Figure E3-5. NMDS plots of the taxa that have the strongest correlations with Axis 3
(Tricorythodes, Oecetis, Ablabesmyia and Pseudocloeon). Cat Precip refers to the
precipitation categories, which are: l=dry years; 2=normal years; 3=wet years.
E-13

-------
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
E3.5 ANOVA - commonly used metrics in northeastern states - hot/cold/wet/dry/normal
years at 3 Maine reference stations with the most years of biological data (Station 56817,
57011 and 57065).
Metrics that had at least one significant difference when one-way analysis of variance
was done to evaluate differences in samples grouped by coldest, normal, and hottest or driest,
normal and wettest years are shown in Tables E3-2 and E3-3. These tables do not include results
for thermal preference metrics, which are shown in Table 2-2 of the main report. Additional
results are available upon request.
Table E3-2. These metrics had at least one significant difference when one-way analysis of
variance was done to evaluate differences in samples grouped by coldest, normal, and
hottest years. Year groups were based on Parameter-elevation Regressions on Independent
Slopes Model (PRISM) mean annual air temperature values at each site. Groups with the
Site
Metric
Coldest
Normal
Hottest
56817
(Sheepscot)
% Swimmer individuals
5.2 ± 3.6a
5.7 ± 2.4ab
10.6 ± 5.0b

% EPT individuals
23.3 ±9.4a
58 ± 6.8b
63.6 ± 18.7b

Hilsenshoff Biotic Index
5.4 ± ,01A
3.9 ± .07®
4.5 ± ,04a

% Collector-filterer individuals
68.2 ± 1 1.7a
16.6 ± 6.0b
28.2 ±27.0^
57011
Shannon-Wiener diversity index
2.6 ± 0.6a
3.7 ± 0.4b
3.5 ±0.4^
(W.Br.
% Most dominant individuals
59.7 ± 12.3a
23.9 ± 6.23b
32.1 ±3.8b
Sheepscot)
% Perennial individuals
19.9 ± 3.6a
57.1 ± 17.5b
64.8 ± 16.7b

% Intermittent individuals
67.4 ± 9.3a
22.0 ± 7.2b
24.4 ± 16.9b

% Drier vulnerable individuals
10.5 ± 2.6a
29.9 ± 10.3ab
39.2 ± 10.3b

% OCH individuals
4.6 ± 2.4a
17.2 ± 6.2b
8.4 ±2.6^
Table E3-3. These metrics had at least one significant difference when one-way analysis of
variance was done to evaluate differences in samples grouped by driest, normal, and
wettest years. Year groups were based on Parameter-elevation Regressions on Independent
Slopes Model (PRISM) mean annual precipitation values at each site. Groups with the
Site
Metric
Driest
Normal
Wettest
56817
(Sheepscot)
% Warmer-drier vulnerable individuals
0.43 ± 0.6^
0.04 ± 0.1A
1.3 ± 1.3b
# Warmer-drier vulnerable taxa
0.32 ±0.4^
0.05 ±0.1A
0.7 ± 0.4b
57065
(Duck)
% Climber individuals
8.2 ± 0.9ab
10.7 ± 3.4a
3.9 ± 2.2b
E-14

-------
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
E3.6 How cold- and warm-water-indicator taxa may affect EPT metrics and HBIs, how
they relate to BCG tier assignments and Class A indicator taxa and how changes in
temperature indicator taxa may vary across different ecoregions in Maine
Attachment E3 contains tables with lists of the temperature-indicator taxa, temperature
optima and tolerance values that were calculated from the weighted average modeling, the
tolerance values assigned by Maine DEP (which are used to calculate the HBI) and BCG
attribute levels assigned to each taxa during the New England Wadeable Streams (NEWS)
project (US EPA, 2007).
E3.7 Distribution of cold and warm-water temperature indicator taxa
Additional results are reported below (Tables E3-6, E3-7, E3-8, E3-9, E3-10 and E3-
11).
Tables E3-10 and E3-11 summarize distribution and abundance information for the
Maine temperature-indicator taxa at the 3 sites (Stations 56817, 57011 and 57065) and 2 site
groups that were analyzed for long-term trends. Boyeria and Eurylophella appear to be two of the
strongest cold-water indicators because they occurred at all or most of the sites and generally had
higher mean relative abundances than the other taxa. Nigronia, Pagastia, and Leuctra also
occurred at most of the sites. Overall, the cold-water taxa are not well-represented at the 3
individual stations, but have a greater presence among the site groups, especially the
Northeastern Highlands. The warm-water-indicator taxa show a different pattern. They are well-
represented at the individual sites and are poorly represented in the site groups, especially among
the Laurentian Plains and Hills site group. Stenonema and Stenelmis appear to be two of the
strongest warm-water indicators because they occur at all the sites and site groups and are
present in higher numbers than the other taxa. Acroneuria, Ceraclea, Hydra, Neureclipsis,
Nilotanypus and Oecetis are present at 4 of the 5 sites/site groups, and Leucrocuta also occurs in
relatively high abundances.
E-15

-------
243	Table E3-10. Summary of distribution and abundance information for the cold-
244	water temperature indicator taxa at the 3 sites (Stations 56817, 57011 and 57065) and 2 site
245	groups (Laur = Laurentian Plains and Hills, NEHigh = Northeastern Highlands). #Sites
246	refers to the number of sites or site groups at which the taxa occurs. A=absent. P=present
247	(highlighted in grey). Relative abundance codes: L=low (<0.01), M=medium (0.01-0.1),
248	H=high (>0.1) (M or H are in bold type). Guide to interpretation: P-1L = present, occurred
249	duringl year, low relative abundance (RA), P-11M = present, occurred during 11 years,
250	medium RA, etc.
FinallD
#Sites
ME56817
ME57011
ME57065
Laur
NEHigh
Ameletus
1
A
A
A
A
P-4L
Apatania
1
A
A
A
A
P-2L
Boyeria
5
P-11L
P-11M
P-9M
P-8M
P-3M
Capnia
0
A
A
A
A
A
Diplectrona
2
A
A
P-1L
A
P-1L
Epeorus
2
P-10L
A
A
A
P-4M
Eurylophella
4
A
P-1L
P-9M
P-7M
P-7M
Glossosoma
3
P-1L
A
A
P-4L
P-2L
Heterotrissocladius
2
A
A
P-1L
A
P-3L
Hydatophylax
3
A
P-1L
A
P-2L
P-1L
Lanthus
1
A
A
A
A
P-1L
Larsia
1
A
A
A
A
P-3M
Table E3-10. Continued





FinallD
#Sites
ME56817
ME57011
ME57065
Laur
NEHigh
Leuctra
4
A
P-4L
P-1L
P-5L
P-6M
Limnephilus
1
A
A
A
P-4M
A
Macropelopia
1
A
A
A
A
P-2L
Malirekus
0
A
A
A
A
A
Micrasema
2
P-9L
A
A
A
P-1L
Natarsia
1
A
A
A
A
P-1L
Nemoura
0
A
A
A
A
A
Nigronia
4
P-1L
P-6L
P-5L
P-1L
A
Oligostomis
2
A
A
A
P-6M
P-2L
Oulimnius
2
A
A
A
P-2L
P-4L
Pagastia
4
P-1L
P-1L
A
P-6M
P-5M
Palaeagapetus
0
A
A
A
A
A
Paracapnia
2
A
P-1L
A
P-5L
A
Paranemoura
0
A
A
A
A
A
Parapsyche
1
A
A
A
A
P-3L
Peltoperla
1
A
A
A
A
P-1L
E-16

-------
Perlodidae
3
P-2L
A
Prodiamesa
0
A
A
Prostoia
0
A
A
Pseudodiamesa
1
A
A
Psychoglypha
2
A
A
Pteronarcys
1
A
A
Rliitlirogena
1
A
A
Sweltsa
3
A
A
Taenionema
0
A
A
Tallaperla
1
A
A
Utacapnia
0
A
A
Utaperla
0
A
A
Zapada
0
A
A
A
P-4M
P-7M
A
A
A
A
A
A
A
A
P-1L
A
P-4L
P-1L
A
A
P-6M
A
A
P-2M
P-1L
P-4L
P-6M
A
A
A
A
A
P-4M
A
A
A
A
A
A
A
A
A
252
253
E-17

-------
255	Table E3-11. Summary of distribution and abundance information for the warm-water
256	temperature indicator taxa at the 3 sites (Stations 56817, 57011 and 57065) and 2 site
257	groups (Laur = Laurentian Plains and Hills, NEHigh = Northeastern Highlands). #Sites
258	refers to the number of sites or site groups at which the taxa occurs. A=absent. P=present
259	(highlighted in grey). Relative abundance codes: L=low (<0.01), M=medium (0.01-0.1),
260	H=high (>0.1) (M or H are in bold type). Guide to interpretation: P-1L = present, occurred
261	duringl year, low relative abundance (RA), P-11M = present, occurred during 11 years,
262	medium RA, etc.	
FinallD	#Sites ME56817 ME57011 ME57065 Laur NEHigh
Acroneuria
4
P-23M
P-12M
P-9M
A
P-3M
Amnicola
3
P-12L
P-2L
P-5L
A
A
Argia
3
P-7L
P-6L
P-1L
A
A
Attaneuria
1
P-3L
A
A
A
A
Caenis
1
A
P-3L
A
A
A
Cardiocladius
1
P-1L
A
A
A
A
Ceraclea
4
P-5L
P-3L
P-2L
A
P-1L
Chaetogaster
3
P-1L
A
P-2L
P-1L
A
Dicrotendipes
3
P-2L
P-2L
P-2L
A
A
Erpobdella
1
A
A
A
A
P-1L
Ferrissia
2
P-5L
P-2L
A
A
A
Helicopsyche
3
P-7L
P-8M
A
A
P-1L
Helisoma
1
A
A
P-2L
A
A
Hemerodromia
3
P-6L
P-11M
P-3L
A
A
Hydra
4
P-1L
P-1L
P-3L
P-2L
A
Hydroptila
3
P-14M
P-3L
A
P-1L
A
Isonychia
2
P-22M
P-3L
A
A
A
Labrundinia
2
P-2L
P-2L
A
A
A
Leucrocuta
3
P-19M
P-11M
P-6M
A
A
Macrostemum
2
P-16M
P-3L
A
A
A
Neureclipsis
4
P-22M
P-2L
P-1L
A
P-2L
Nilotanypus
4
P-5L
P-2L
P-1L
A
P-2L
Oecetis
4
P-8L
P-9M
P-8M
A
P-3L
Orconectes
1
P-1L
A
A
A
A
Parachironomus
0
A
A
A
A
A
Paragnetina
2
P-2L
P-1L
A
A
A
Pentaneura
2
P-13L
P-1L
A
A
A
Physa
2
A
A
P-4M
P-3L
A
Phy sella
2
P-8L
A
P-5M
A
A
Plauditus
3
P-6L
P-1L
A
A
P-1L
Prostoma
2
P-1L
P-1L
A
A
A
Psectrocladius
3
P-1L
A
P-8M
P-1L
A
Pseudocloeon
1
P-7L
A
A
A
A
Rheopelopia
3
P-8L
P-5L
P-1L
A
A
Serratella
3
P-15M
P-2L
A
A
P-1M
Stenacron
2
A
P-1L
P-9M
A
A
Stenelmis
5
P-19M
P-10M
P-1L
P-6M
P-4L
Stenonema
5
P-23M
P-12M
P-9M
P-7M
P-6M
Tribelos
1
A
A
A
P-1L
A
Tricorythodes
2
P-6L
P-11M
A
A
A
E-18

-------
264
265
266
267
268
269
270
271
272
273
274
275
276
Til
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
E3.8 Summary
•	In general, samples with the following characteristics received better ratings (=higher
classifications):
o High generic richness
o High richness and abundance of EPT taxa
o High Shannon-Wiener diversity index values
o Low HBI scores
o Low Chironomidae abundances
o Low relative Diptera richness
o Low relative Oligochaeta abundance
o Greater presence of Class A indicator taxa
o Greater scraper relative abundance
•	Results from the NMDS ordination show that samples from Station 56817 do not form
distinct clusters when grouped by hot/cold/wet/dry/normal years, so species composition
did not change in a consistent way when the climate-related variables changed. None of
the taxa that were the strongest drivers in the analysis occurred exclusively in a particular
temperature or precipitation category.
•	Although there were significant differences among certain metrics at certain sites, the
only 'consistent' pattern (=one that occurred at more than one site) was that the
percentage of swimmers was higher during the warmer years at 2 sites. The other
differences appeared to be site-specific.
•	At Station 56817, precipitation appeared to have a greater influence on the biotic
assemblage than temperature. At Station 57011, temperature had a greater influence on
metric values than precipitation.
•	At Station 56817, the mean richness and abundance of cold-water taxa was higher during
the wet years, as were the richness and abundance of taxa that are predicted to be more
vulnerable in a warmer drier climate scenario.
•	At Station 57011, samples collected during cold years had higher percent dominant taxon
individuals, higher percent collector-filterer individuals, higher HBI scores, lower
Shannon-Wiener diversity index scores, higher intermittent taxa individuals, lower EPT
percent individuals, lower percent perennial taxa individuals, lower percent
Odonata/Coleoptera/Hemiptera individuals, lower percent warm-water individuals and
lower percent 'drier vulnerable individuals.
•	Only one metric at Station 57065 had significant differences among the temperature and
precipitation categories. At that site, the number of collector-filterer taxa was higher in
samples collected during the normal/wet years, but this should be interpreted with caution
due to a low sample size.
E-19

-------
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
Many of the cold-water-indicator taxa in Maine are EPT taxa: 16 of the cold-water taxa
are Plecopterans, 10 are Trichopterans and 3 are Ephemeropterans. There are also a
relatively high number of EPT taxa on the warm-water indicator list: 9 of the warm-water
taxa are Ephemeropterans, 6 are Trichopterans and 3 are Plecopterans.
Eight of Maine's model input metrics are related in some way to EPT taxa. Two of the
model input metrics are specifically related to Ephemeropterans: Ephemeroptera
Abundance and Relative Abundance Ephemeroptera. Three model input metrics are
specifically related to Plecopterans: Plecoptera abundance, Perlidae abundance and
relative Plecoptera richness. Two of the model input metrics are specifically related to
Trichopterans: Hydropsyche abundance and Cheumatopsyche abundance.
Seven of the cold-water taxa are Dipterans from the family Chironomidae, and ten of the
warm-water taxa are Dipterans. Several model input metrics are specifically related to
Dipterans.
All but two of the tolerance values of the cold-water indicator taxa are low (< 3). Nine of
the warm-water taxa have tolerance values > 7, but it should be noted that 10 of the
warm-water taxa have low tolerance values (< 3), so there is a mix.
Tolerance values had a weak but significant correlation (r=0.29, p=001) with temperature
optima values.
When BCG attributes of temperature indicator taxa are examined, twenty of the cold-
water taxa are considered to be sensitive taxa (2 or 3) and two of the taxa are considered
to be tolerant (5).
Two of the Class A indicator taxa, Eurylophella and Glossosoma, are on the cold-water
list and three, Paragnetina, Serratella and Leucrocuta, are on the warm-water list.
Brachycentrus was initially on the warm-water list but was removed due to variation in
temperature preferences among species within this genus.
At Station 56817, on average, more Class A indicator taxa were present during wetter
years
The Northeastern Highlands sites have the highest mean number of cold-water indicator
taxa (followed closely by the Northeastern Coastal Zone sites). It should be noted that the
number of cold-water taxa in all the ecoregions is low (values generally range from 1 to 2
taxa). The mean number of warm-water indicator taxa at sites in the Laurentian Plains
and Hills is significantly higher than at sites in the other ecoregions, while the
Northeastern Highlands sites have the lowest mean number of warm-water indicator taxa.
On average, there are more cold-water indicator taxa at higher elevation (> 500 ft) sites
and more warm-water indicator taxa at lower elevation (< 500 ft) sites.
At Station 56817, 5 model metrics were significantly correlated with flow category. The
general pattern was that Dipteran Richness and Tanypodinae abundance decreased during
higher flow years, while the EPT:Diptera ratio metrics increased. The relative abundance
of collector-gatherers was higher during higher flow years.
Bottom lines: it is tough to predict effects on Maine's classification models because they
look at multiple variables simultaneously. There are no firm thresholds. We did the best
we could with what we had, but predictions at this point are only speculative.
Unfortunately we lack long-term data for Northeastern Highlands sites, which seems to
be the area where we would have been most likely to see a trend or detect a climate-
related shift in the biology.
E-20

-------
Attachment El
348
349	Results of the ANOVA analysis in which mean model
350	input metric values were compared across the different
351	classifications
352
353	This attachment contains a table and box plots that summarize mean model input metric values
354	for Class A, B, C and NA samples.
355
El-1

-------
356
357
358
359
360
Table El-1. Results of the one-way ANOVA that was performed on all the samples in the Maine database to see how mean
metric values varied among the different classes. Mean metric values, sample sizes (N) and standard deviations (Std. Dev.) are
shown. P-levels from the Tukey honest significant difference (HSD) test for unequal sample size (N) (Spjotvoll/Stoline) are
shown if they are significant (<0.5). Note: I35-PIERCERS RELATIVE ABUNDANCE and I38-PARASITES RELATIVE

CLASS
Significance level (Tukey p-level)

A
B
C
NA
Metric
N
Mean
Std.
Dev.
N
Mean
Std.
Dev.
N
Mean
Std.
Dev.
N
Mean
Std.
Dev.
A-
B
A-
C
A-
NA
B-C
B-NA
c-
NA
IOl-TOTAL
ABUNDANCE
577
460.5
548.8
448
909.8
948
353
1132.1
1800.2
227
598.8
1128.7
0
0

0.04
0.02
0
I02-GENERIC
RICHNESS
577
42.9
14.4
448
42.4
13.2
353
36.3
13.3
227
28.1
11.9

0
0
0
0
0
103 -PLECOPTERA
ABUNDANCE
577
21.1
38.7
445
11.5
15.6
170
6.3
11.4
19
5.6
15.7
0
0




104-
EPHEMEROPTERA
ABUNDANCE
576
99.5
132.9
448
144.8
163.9
353
90.7
155.7
142
10
31.7
0

0
0
0
0
I05-SHANNON-
WEINER GENERIC
DIVERSITY
577
3.8
0.7
448
3.4
0.7
353
3.2
0.8
227
2.7
0.9
0
0
0
0
0
0
I06-HILSENHOFF
BIOTIC INDEX
577
3.7
0.8
448
4.7
0.7
353
5.3
1
227
5.8
1.2
0
0
0
0
0
0
107-RELATIVE
CHIRONOMIDAE
ABUNDANCE
574
0.2
0.2
448
0.3
0.2
353
0.3
0.2
222
0.3
0.3

0
0
0
0.02

I08-RELATIVE
DIPTERA RICHNESS
576
0.4
0.1
448
0.4
0.1
353
0.4
0.1
224
0.4
0.2

0
0
0.01
0

I09-HYDROPS Y CHE
ABUNDANCE
472
103
154.4
395
216.4
301.1
263
343
600.2
138
134.8
618.6
0
0

0

0
111-
CHEUMATOPSYCHE
ABUNDANCE
362
22.4
48.5
383
98.8
283.8
264
175.5
433.4
124
118.2
371.7
0
0

0.01


I12-EPT GENERIC
RICHNESS DIVIDED
BY DIPTERA
RICHNESS
576
1.5
1
448
1.3
0.6
353
1
0.6
224
0.8
0.9
0
0
0
0
0
0.02
El-2

-------
361 Table El-1. Continued

CLASS
Significance level (Tukey p-level)

A
B
C
NA
Metric
N
Mean
Std.
Dev.
N
Mean
Std.
Dev.
N
Mean
Std.
Dev.
N
Mean
Std.
Dev.
A-B
A-
C
A-
NA
B-C
B-
NA
c-
NA
113-RELATIVE
OLIGOCHAETA
ABUNDANCE
277
0
0
221
0
0.1
218
0
0.1
168
0.1
0.2


0

0
0
I15-PERLIDAE
ABUNDANCE
401
10.2
8.9
376
9.7
12.3
104
6.2
11
10
5.5
9.7

0.04




I16-TANYPODINAE
ABUNDANCE
465
8.2
14.7
405
15.2
25
316
25.4
48.5
190
21.3
43
0.01
0
0
0


117-CHIRONOMINI
ABUNDANCE
511
27.7
70.7
432
52.3
122.9
331
109.5
323.9
187
68
192.2

0

0


I18-RELATIVE
EPHEMEROPTERA
ABUNDANCE
576
0.3
0.2
448
0.2
0.2
353
0.1
0.2
142
0
0.1
0
0
0
0
0
0
I19-EPT GENERIC
RICHNESS
577
19.3
5.7
448
17.3
4.6
353
11.5
4
216
5.9
3.3
0
0
0
0
0
0
121-SUMMED
ABUNDANCE OF
DICROTENDIPES,
MICROPSECTRA,
PARACHIRONO
302
9.5
22
231
17.2
88.8
231
36.6
132
149
55.3
193

0.05
0

0.02

123-RELATIVE
PLECOPTERA
RICHNESS
577
0.1
0.1
445
0.1
0
170
0
0
19
0.1
0
0
0




I25-SUMMED
ABUNDANCE OF
CHEUMATOPSYCHE,
CRICOTOPUS,
TANYTARSUS A
518
28.1
60.7
446
121.1
275.8
347
183.5
403.3
213
96.7
305.2
0
0
0.04
0.01

0.01
I26-SUMMED
ABUNDANCE OF
ACRONEURIA,
STENONEMA AND
MACCAFFERTIUM
463
32
37.1
417
51.7
59.2
262
37.6
64.7
60
6.4
19.9
0

0.03
0.01
0
0.01
El-3

-------
362
363	Table El-1. Continued

CLASS
Significance level (Tukey p-level)

A
B
C
NA
Metric
N
Mean
Std.
Dev.
N
Mean
Std.
Dev.
N
Mean
Std.
Dev.
N
Mean
Std.
Dev.
A-B
A-
C
A-
NA
B-C
B-
NA
c-
NA
I28-EP GENERIC
RICHNESS
DIVIDED BY 14
577
0.8
0.3
448
0.6
0.2
353
0.4
0.2
149
0.2
0.1
0
0
0
0
0
0
I30-PRESENCE OF
A INDICATOR
TAXA
522
0.4
0.2
348
0.2
0.1
152
0.2
0.1
43
0.2
0.1
0
0
0
0.04


131-EPT GENERIC
REICHNESS
RELATIVE TO
EPTPLUS
DIPTERA
576
0.6
0.1
448
0.5
0.1
353
0.5
0.1
214
0.4
0.2
0
0
0
0
0
0
I32-COLLECTOR-
FILTERERS
RELATIVE
ABUNDANCE
577
0.4
0.2
448
0.5
0.3
353
0.4
0.3
214
0.3
0.3
0
0.01
0.04
0
0
0
13 3-COLLECTOR-
GATHERERS
RELATIVE
ABUNDANCE
577
0.2
0.1
448
0.2
0.1
352
0.2
0.2
218
0.2
0.2
0
0
0



I34-PREDATORS
RELATIVE
ABUNDANCE
577
0.1
0.1
447
0.1
0.1
351
0.1
0.1
223
0.1
0.2



0
0.01

I36-SHREDDERS
RELATIVE
ABUNDANCE
567
0.1
0.1
433
0.1
0.1
336
0.1
0.1
200
0.2
0.2
0.03

0

0
0
I37-SCRAPERS
RELATIVE
ABUNDANCE
565
0.1
0.1
444
0.1
0.1
344
0.1
0.1
194
0.1
0.2






I39-ETO GENERIC
RICHNESS
577
17.5
6.1
448
17.1
5
353
12.6
4.5
222
7
3.5

0
0
0
0
0
El-4

-------
364
365
366
367
368
369
The next 32 graphs are categorical box and whisker plots for the Maine model input
metrics. They show the mean metric values for Class A, B, C and NA samples that are currently
in the Maine database. These plots were examined to gain more insight into the following
questions: how do model input metric values differ among the different classifications? Do
certain metrics appear to be more important than others in contributing to classification changes?
Maine Model Input Metrics
1400
1200
LU
o
<
Q
=>
CO
<
_l
£
o
1000
800
600
400
200
NA
~	Mean
~	Mean±SE
I Mean±1.96*SE
370
Class
El-5

-------
372
Maine Model Input Metrics
46
44
42
40
CO
CO
w 38
x
o
cc
0
01
LU
cy 36
34
LU
9 32
oj
o
30
28
26
24
373
B	C
Class
NA
~	Mean
~	MeantSE
X Mean±1.96*SE
Maine Model Input Metrics
374
375
26
24
22
20
$ 18
<
Q
16
14
12
z>
00
<
<
a:
w 10
CL
8 8
LU
NA
~	Mean
~	Mean±SE
I Mean±1.96*SE
Class
El-6

-------
180
Maine Model Input Metrics
B	C
Class
NA
a Mean
'~ MeaniSE
I Mean±1.96*SE
Maine Model Input Metrics
n Mean
~ Mean±SE
X Mean±1,96*SE

-------
Maine Model Input Metrics
a Mean
~ MeaniSE
I Mean±1.96*SE
Maine Model Input Metrics
B	C
Class
NA
~	Mean
~	MeaniSE
I Mean±1.96*SE

-------
0.43
Maine Model Input Metrics
0.42
co 0.41
co
LU
l 040
CsL
i 0.39
i-
Q.
Q
LU
0.38
0.37
- 0.36
0.35
0.34
382
B	C
Class
a Mean
~ Mean±SE
NA	I Mean±1.96*SE
Maine Model Input Metrics
450
383
Q 300
250
III
co 200
9 150
~	Mean
~	MeantSE
I Mean±1.96*SE
El-9

-------
Maine Model Input Metrics
384
I
T
~	Mean
~	Mean±SE
I Mean±1.96*SE
Maine Model Input Metrics
385
386
Q.
LU
1.8
<
CC
LU
H
CL
Q
>
CD
~
LU
Q
>
Q
CO
ro w
il
X O
O OC
QL
o
QL
LU
Z
LU
0
1.6
1.4
W 1.2
1.0
o.e
0.6
0.4
B	C
Class
NA
~	Mean
~	Mean±SE
I Mean±1.96*SE
El-10

-------
0.16
Maine Model Input Metrics
0.14
ill
o
q 0.12
3
m
<
<
0.10
LLi
<
O 0.08
o
o
LLI 0.06
>
H
<
w 0.04
QL
CO
0.02
0.00
387
B	C
Class
~	Mean
~	Mean±SE
NA	X Mean±1.96*SE
Maine Model Input Metrics
14
12
10
0
z:
<	r
Q 8
z
3
m
<	,
LU 0
<
Q
1	4
-2
388
B	C
Class
NA
~	Mean
~	Mean±SE
I Mean±1.96*SE
El-11

-------
Maine Model Input Metrics
32
30
28
26
24
22
20
3
m
<
LU
< 13
O 16
Q.
£
CD
389
B	C
Class
NA
~	Mean
~	Mean±SE
I Mean±1.96*SE
Maine Model Input Metrics
160
390
391

-------
Maine Model Input Metrics
392
lu 0.26
< 0.24
Q
ZD
00
<	0.20
<
en
lu 0.18
Q_
O 0.16
± 0-12
LU
UJ 0.10
Class
Q Mean
~ MeaniSE
I Mean±1.96*SE
Maine Model Input Metrics
393
Q_
tu 10
~	Mean
~	MeantSE
I Mean±1.96*SE
El-13

-------
UJ
ID
Un
I23-RELATIVE PLECOPTERA RICHNESS
o
o
-e*
h
CD
M
o
ffl
w
o
HD ~
3 S :
r1 CO
<£> m
ro

UJ
ID
4=>
121 - SUMMED ABUNDANCES OF DICROTENDIPE5,
MICROPSECTRA, PARACHIRONOMUS AND HELOBDELLA
2
0
1
I
S
a
=.
o
2
2.
2
o
I
HD d
sss
CD en m
§ § §
r4 W
§m
*
m
m

-------
240 r-
220 ¦
200 ¦
180 ¦
160 ¦
140 ¦
120 ¦
100 -
80 ¦
80 ¦
40 ¦
20 •
0 -
60
50 ¦
40 ¦
:
>
J 30 ¦
i
>
; 20 ¦
10 ¦
0 ¦
-10 ¦
Maine Model Input Metrics
NA
~	Mean
~	Mean±SE
I Mean±1.96*SE
Class
Maine Model Input Metrics
NA
Q Mean
~ MeantSE
I Mean±1.96*SE
Class

-------
0.9
Maine Model Input Metrics
398
0.8
0.7
>-
m
o
LU
d
>
Q 0.6
§
LU
x 0.5
o
K
O
E 0.4
LU
z
LU
o
q. 0.3
w
co
2
0.2
0.1
p I
g i
i T i
NA
Q Mean
~ Mean±SE
I Mean±1.96*SE
Class
Maine Model Input Metrics
P 0,32
y o.3o
LU
O 0.24
0	Mean
~ Mean±SE
1	Mean±1.96*SE
399
Class
El-16

-------
131-EPT GENERIC REICHNESS RELATIVE TO EPT PLUS
DIPTERA
o
K>
O O O O
^ 4^ U1
4* CD 00 O
O O
en " 00 o
o


H -




h d H

. (- ~ _l


HQ ~
g g:
CO
cd rn
03
CO
rn

-------
Maine Model Input Metrics
401
0.54
0.52
LU
z 0.50
g 0.48
m 0.46
W 0.44
F
<
LU
OH
0.42
0.40
K 0.38
LU
OH
LU
Ct.
P
0.36
0.34
0.32
O 030
LU
0.28
0.26
0.24
0.22
0.26
HI
o
<	0.24
Q
Z
3
CO
<
LU
>
0.22
0.18
LU _ __
ce 0.20
w
m
LU
Q-
LU
<
0
0£
O 0.16
I—
o
LU
o
o
ch
CO
0 14
0.12
NA
~ Mean
[j| Mean±SE
I Mean±1.96*SE
Class
Maine Model Input Metrics
NA
a Mean
~ Mean±SE
I Mean±1.96*SE
402
Class
El-18

-------
0.14
Maine Model Input Metrics
0.13
LU
O
2
2
0.12
3
00
<
LU
> 0,11
I—
<
LU
cc
w 0.10
or
O
i-
3
lu 0.09
DC
0.08
0.07
NA
a Mean
0 MeantSE
T Meant 1,96*SE
403
Class
Maine Model Input Metrics
0.024
0.022
w 0.020
Z
g 0.018
3
CD
<
LU
0.016
< 0.014
LU
DC
® 0.012
LU
O
g] 0.010
E
lO
£2 0.008
0.006
0.004
NA
a Mean
~ MeantSE
I Mean±1,96*SE
404
Class
El-19

-------
Maine Model Input Metrics
405
0.22
0.20
LU
O
z
<
D
0.18
g 0,16
<
in
p 0.14
Gj
I
QL
LU
Q
D 0.10
LU
a.
i
to

o.os
0.06
0.04
0.16
0.15
LU
o 0.14
<
0
1	0.13
CO
<
LU
> 0.12
LU
of 0.11
CO
IX
UJ
% 0.10
ce
o
»
ri 0.09
0.08
0.07
Class
Maine Model Input Metrics
NA
0	Mean
~ Mean±SE
1	Mean±1 96*SE
NA
~	Mean
~	MeantSE
I Mean±1.96*SE
406
Class
El-20

-------
Maine Model Input Metrics
407
408
0.00144
0.00142
0.00140
LU
1	0.00138
2
z 0.00136
3
<	0.00134
ill
p 0.00132 [
3
£ 0.00130
CO
W 0.00128
CO
<	0.00126
0.00124
0.00122
0.00120
0.00118
~	Mean
~	Mean+SE
I Mean±1,96*SE
Class
Maine Model Input Metrics
409
o: 14
co 10
a Mean
~ MeantSE
I Mean±1.96*SE
Class
El-21

-------
410
411
412
413
Attachment E2
Maine Temperature-Indicator Taxa
414
415	This attachment contains tables with lists of the Maine temperature-indicator taxa and describes
416	the process that we followed to develop these lists.
417
418
E2-1

-------
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
MAINE TEMPERATURE-INDICATOR TAX A
Sources. The Maine cold- and warm-water taxa lists were developed using several
different sources: 1. weighted average calculations based on a subset of the Maine
biomonitoring database (using site average temperature values (July, August, and September)
from 616 sites); 2. the thermal preference trait from the Poff et al. (2006) traits matrix; 3. the
thermal-preference trait from the USGS traits database (Vieira et al., 2006); 4. the thermal-
preference trait from the compilation of EPA Environmental Requirements and Pollution
Tolerance series from the late 1970's (Beck et al., 1977; Harris et al., 1978; Hubbard et al., 1978;
Surdick et al., 1978); and 5. best professional judgment of the New England Climate Change
traits feedback group1.
Designation as cold-water taxa. Taxa were placed on the Maine cold-water taxa list if
they met the following criteria: 1. They received a rank temperature optima value of 1 or 2 or 3
(the rank optima value is based on percentiles of the dataset; for these taxa, the weighted average
optima value was less than the 0.4 percentile value of the dataset it was derived from); or 2. the
thermal preference in the Poff et al. 2006 traits matrix was 'coldcool'; or 3. The thermal
preference in the USGS traits database (Vieira et al., 2006) was 'cold stenothermal' or 'cold-cool
eurythermal' (temperature preference of less than 15°C); or 4. The thermal preferences in the
EPA Environmental Requirements and Pollution Tolerance series were 'oligothermal' or
'stenothermal' or 'metathermal' (temperature preference of less than 15°C); or 5. If anyone in
the New England Climate Change feedback group felt a taxa should be added to this list.
Designation as warm-water taxa. Taxa were placed on the Maine warm-water taxa list
if they met the following criteria: 1. They received a rank temperature optima value of 5 or 6 or 7
(the rank optima value is based on percentiles of the dataset; for these taxa, the weighted average
optima value was greater than the 0.6 percentile value of the dataset it was derived from); or 2.
the thermal preference in the Poff et al. 2006 traits matrix was 'warm'; or 3. The thermal
preference in the USGS traits database (Vieira et al., 2006) was 'hot euthermal' or 'warm
eurythermal' (temperature preference of greater than 15°C); or 4. The thermal preferences in the
1 New England Climate Change group: Maine DEP (Leon Tsomides, TomDanielson, Dave Courtemanche, Susan
Davies), Vermont DEC (Doug Burnham, Steve Fiske, Jim Kellogg, Rich Langdon), New Hampshire (Dave Neils -
NH DES, Don Chandler- UNH), Mike Winnell (professional taxonomist who works on a lot of Maine samples).
E2-1

-------
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
EPA Environmental Requirements and Pollution Tolerance series were 'euthermal' or
'eurythermal' or 'mesodermal' (temperature preference of greater than 15°C); or 5. If anyone in
the New England Climate Change feedback group felt a taxa should be added to this list.
Limitations. These lists were developed using the best information available, but it
should be noted that the available information is limited. The weighted average calculations are
based on instantaneous water temperature measurements that were taken at the time of the
sampling event. Ideally, continuous water temperature data should be used, since this provides
more information about the thermal regime, especially during times of greatest thermal stress
(i.e. summer baseflow conditions). However, these data are generally unavailable. The weighted
average calculations also have limitations. One of the main concerns is that the analysis does not
take into account the confounding factors ('noise') that are not related to temperature. However,
the theory is that with a sufficient amount of data, the noise essentially cancels itself out. Another
limitation is that the operational taxonomic unit that was most appropriate for this analysis is at
the genus-level (in some instances, family-level was most appropriate). Within certain genera in
particular, the thermal preference among species varies, so the assigned thermal preference may
not be appropriate for all species within a genera. Attempts were made to note these genera (see
'species-variation' column in the worksheets).
We want to reiterate that when we developed these lists, we did the best we could with
the data that was available. These lists should be viewed as a first step, not a final product. It
would be very helpful if future research included a combination of short- and long-term field and
experimental studies designed to better evaluate climate change effects on freshwater
ecosystems.
Initial Results. Initially there were 106 taxa on the cold-water list and 82 taxa on the
warm-water list. These lists were based on weighted average calculations and literature. These
lists were further refined through the evaluation of additional evidence. This evidence included
analyses of other datasets, case studies, and best professional judgment. Taxa with the greatest
amount of evidence were designated as temperature-indicator taxa. More detailed information
about the steps that were used to develop the temperature indicator taxa lists is summarized
below:
Considerations
E2-2

-------
476	A. Results from weighted average or maximum likelihood thermal optima and tolerance
477	calculations were a major consideration. Results from the following eight analyses were used:
478	• California (taken from 'Herbst_CABW.2007_Sierra.climate.change.ppt')
479	• Idaho (taken from 'Temperature Preferences and Tolerances for 137 Common
480	Idaho Macroinvertebrate Taxa. Darren Brandt. Idaho DEQ. November 2001.')
481	• Maine (based on site average temperature values (July- September) from 616
482	sites in the Maine biomonitoring database)
483	• North Carolina (based on maximum likelihood calculations for the North
484	Carolina biomonitoring database, full-scale collection method only)
485	• Ohio (Ed Rankin, these are PRELIMINARY, and are based on average mean
486	temperature values)
487	• Oregon (Shannon Hubler (2007), based on the Oregon DEQ database)
488	• Utah (based on 572 fall samples from the Utah biomonitoring database)
489	• Yuan 2006 (Estimation and Application of Macroinvertebrate Tolerance
490	Values. Report No. EPA/600/P-04/116F, based on Western EMAP data).
491	A scoring system was developed to summarize results from the eight different analyses.
492	It takes into account thermal preference, thermal tolerance and sample size. Scores were
493	assigned (for each of the eight analyses) as follows:
494	COLD-WATER TAXA
495	• 2=cold stenotherm (rank optima of 1 or 2 or 3 and rank tolerance of 1 or 2 or
496	3), adequate sample size (20 or more counts)
497	• l=cold preference (rank optima of 1 or 2 or 3), adequate sample size (20 or
498	more counts)
499	• l=cold stenotherm (rank optima of 1 or 2 or 3 and rank tolerance of 1 or 2 or
500	3), low sample size (less than 20 counts)
501	• 0.5=cold preference (rank optima of 1 or 2 or 3), low sample size (less than 20
502	counts)
503	WARM-WATER TAXA
504	• 2=warm eurythermal (rank optima of 5 or 6 or 7 and rank tolerance of 5 or 6
505	or 7), adequate sample size (20 or more counts)
E2-3

-------
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
•	l=warm preference (rank optima of 5 or 6 or 7), adequate sample size (20 or
more counts)
•	1= warm eurythermal (rank optima of 5 or 6 or 7 and rank tolerance of 5 or 6
or 7), low sample size (less than 20 counts)
•	0.5=warm preference (rank optima of 5 or 6 or 7), low sample size (less than
20 counts)
In addition to the weighted average and maximum likelihood results, information on
thermal preferences was also derived from literature. The taxon received a score of 1 if it was
cited as a cold- or warm-water taxon in at least one of the following sources: Poff et al. 2006
traits matrix; or USGS traits database (Vieira et al., 2006); or EPA Environmental Requirements
and Pollution Tolerance series from the late 1970's (Beck et al., 1977; Harris et al., 1978;
Hubbard et al., 1978; Surdick et al., 1978). If the weighted-average results showed the taxon to
have a preference for cold- or warm-water but the literature showed conflicting results (i.e. based
on the weighted-average results, the taxon was a cold-water taxa, but the literature showed it to
be a warm-water taxa), then the taxon was not included on the temperature indicator list.
After scores were assigned as described above, they were summed so that each taxon
received a total score. The higher the total score, the more evidence there was in the eight
analyses and the literature that supported the designation of the taxon as a temperature indicator
taxon.
B. Several 'case studies' were performed to see whether the cold- or warm-water taxa
occurred at sites in Maine and Vermont that had the warmest or coldest summer water
temperatures. The following case studies were performed:
a.	Cold-Water Case Study #1. Vermont provided us with taxa lists from two sites that
they regard as cold-water habitat. They are located below a dam that does profundal
releases, and the water temperature remains around 8°C year round. The dam is a
confounding factor (although a study by the VT DEC indicates minor impacts on the
macroinvertebrate community from the Whitewater releases), but temperature is
regarded as a major factor influencing community composition at these sites.
b.	Cold-Water Case Study #2. Taxa lists from the following 3 sites in Maine: Station
57514 (Cold Brook (Dead River) - Maine DEP Station 772), Station 57513 (Cold
Brook (Dead River) - Maine DEP Station 771) and Station 57512 (Cold Brook (Dead
E2-4

-------
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
River) - Maine DEP Station 770). These sites were selected for the following
reasons: 1. Water temperature readings at these 3 sites were among the lowest in the
database, ranging from 7.8 to 13.9°C (these were July-Sept readings); based on the
surrounding land use land cover (1km buffer), these sites appear to have few
confounding factors (0-1% urban, 0% agricultural). Wetlands may influence the
biota at these sites, especially Station 57512 (23% wetland), but temperature is
believed to be a factor influencing community composition at these sites.
c.	Warm-Water Case Study #1. Taxa lists from two sites in Maine with the warmest
average water temperatures: Station 56834 (Mattanawcook Stream - Maine DEP
Station 91, below Lincoln Pulp and Paper (cooling water), which had an average
summer water temperature of 31°C; and Site 57055 (Birch Stream (Bangor) - Maine
DEP Station 312), which had an average summer water temperature of 30°C. These
are not reference sites. Within the 1 km buffer, Station 56834 is 24% urban and
Station 57055 is 60% urban.
d.	Warm-Water Case Study #2. Taxa lists from 5 sites in Maine with the warmest
average water temperatures that were <5% urban and <10% agricultural within a 1
km buffer. Average water temperatures ranged from 26-27°C. Sites included:
Station 57560 (West Seboeis Stream - Maine DEP Station 818), Station 56871
(Penobscot River - Maine DEP Station 128), Station 56953 (Dead River - Maine DEP
Station 210), Station 57228 (Pollard Brook - Maine DEP Station 485) and Station
56952 (Dead River - Maine DEP Station 209).
C. In addition to the case studies, best professional judgment from the New England
Climate Change group was taken into account.
Development of the Temperature Indicator Cold-Water Taxa List. Taxa were placed on the
cold-water list if the following criteria were met:
1.	The taxon was NOT present at the warm-water case study sites.
2.	The taxon was present at one or more of the cold-water case study sites and/or the New
England Climate Change feedback group believed that it should be on the list.
E2-5

-------
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
Development of the Temperature Indicator Warm-Water List. Taxa were placed on the
warm-water list if the following criteria were met:
1.	The taxon was NOT present at the cold-water case study sites.
2.	The taxon was present at one or more of the warm-water case study sites and/or the New
England Climate Change feedback group believed that it should be on the list.
Temperature Indicator Lists. The cold-water taxa list was comprised of 41 taxa and the warm-
water taxa list was comprised of 40 taxa. Temperature indicator taxa lists can be found in Tables
E2-1 and E2-2.
Important Notes - variation within genera. Some noteworthy genera were left off the
Maine warm-water taxa list. These included Brachycentrus, Hydropsyche, and Ceratopsyche.
Genera left off the Maine cold-water list included Eukiefferiella and Rhyacophila. The reason
they were not included is because there is variation in temperature preferences among species
within these genera, and this was noted by the New England Climate Change feedback group or
in the literature (Vermont DEC suggested a list of species to include on the lists - see Tables E2-
6 and E2-7).
It is also worth noting the absence of two other genera from the cold-water list - Antocha
and Dicranota. In the weighted average and maximum likelihood analyses, these two taxa were
often listed as cold-water taxa. However, in the case studies, it became apparent that these
genera were widespread and occurred at sites at which cold and warm temperatures had been
recorded.
Dispersal Ability. If temperature is a major factor influencing community composition,
then taxa that are able to adapt to warming temperatures or that are able to disperse to more
favorable habitats (generally believed to be upstream or to higher elevations) have a better
chance of surviving. Five mobility traits were examined for the taxa on the Maine temperature
indicator lists: dispersal (adult), adult flying strength, occurrence in drift, maximum crawling rate
and swimming ability. More information on these traits can be found in Table E2-3.
Dispersal (adult) and adult flying strength received the greatest amount of consideration.
Because movement is most likely to be upstream, taxa that are strong fliers are likely to have a
better chance of success. It will be difficult for taxa that disperse via occurrence in drift to
E2-6

-------
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
migrate upstream, and taxa that disperse via crawling or swimming are likely to have difficulty
moving the distances required to find more favorable habitats.
Two of the 41 taxa on the Maine temperature indicator cold-water taxa list (for which w
had trait information), Boyeria and Pteronarcys, are considered to have high dispersal ability and
strong adult flying strength. Another taxon, Lanthus, is categorized as having strong flying
ability but low adult dispersal ability. Eleven of the 40 taxa on the warm-water list are
categorized as having high adult dispersal ability. Four of these taxa are considered to be strong
fliers.
Abundance and Distribution. In addition to dispersal ability, abundance and
distribution are also important considerations. Those taxa that are widespread and common are
likely to have greater genetic diversity and greater chance of adapting than rare taxa that only
occur in isolated, localized populations (Sweeney et al. 1992). Moreover, the more abundant taxa
are more likely to affect the state biomonitoring assessments. Abundance and distribution
information for the temperature indicator taxa can be found in Tables E2-1 and E2-2.
The most abundant cold-water-temperature-indicator taxa are Leuctra (Plecopteran),
Epeorus (Ephemeropteran), Eurylophella (Ephemeropteran), Perlodidae (Plecopteran) and
Boyeria (Odonata). These taxa comprise only 0.3 to 0.4% of the total individuals in the Maine
database. Thirty-one of the cold-water taxa have overall abundances of less than 0.1%.
Stenonema and Neureclipsis are the most abundant warm-water taxa, with overall abundances of
5.2 and 2.6%, respectively. Nine of the warm-water taxa have overall abundances of less than
0.1%. Of the cold-water taxa, Boyeria occurs at the largest percentage of sites (38%), followed
by a Plecopteran, Perlodidae, which occurs at 25% of the sites. Thirty-one of the taxa occur at
less than 10% of the sites. Among the warm-water taxa, Stenonema occurs at the highest
percentage of sites (63%), followed by Acroneuria (39%) and Neureclipsis (38%). Eight of the
warm-water taxa occur at less than 10% of the sites.
Additional information - Cold-Water Taxa. Sixteen of the cold-water taxa are
Plecopterans, ten are Trichopterans, seven are Dipterans, and three are Ephemeropterans. The
rest are Coleopterans, Odonates and Megalopterans. The families with the most number of taxa
on the cold-water list are Chironomidae and Nemouridae (Table E2-4). It should be noted that
E2-7

-------
625	two of the taxa on the cold-water list, Malirekus and Taenionema, do not occur in the Maine
626	database. They were added per best professional judgment of the Vermont DEC.
627	Additional information - Warm-Water Taxa. Ten of the warm-water taxa are
628	Dipterans, nine are Ephemeropterans and six are Trichopterans. The families with the most
629	number of taxa on the warm-water list are Chironomidae and Perlidae (Table E2-5).
630
E2-8

-------
631	Table E2-1. List of Maine cold-water temperature indicator taxa. Distribution and abundance information is also included.
632	Sum_Individuals=the total number of individuals from that taxon in the Maine database; Pct_Abund=percent of total
633	individuals in the database comprised of that taxon; Num_Stations=number of stations in the database that the taxon occurred
634	at; Pet Stations=percent of stations in the database at which the taxon occurred	
Type
Order
Family
FinallD
Sum Individs
Pet Abund
Num Stations
Pet Stations
cold
Ephemeroptera
Ameletidae
Ameletus
63
0.01
26
3.06
cold
Trichoptera
Apataniidae
Apatania
48
0.01
23
2.71
cold
Odonata
Aeshnidae
Boyeria
1761
0.3
321
37.81
cold
Plecoptera
Capniidae
Capnia
71
0.01
5
0.59
cold
Trichoptera
Hydropsychidae
Diplectrona
1137
0.19
47
5.54
cold
Ephemeroptera
Heptageniidae
Epeorus
2132
0.36
172
20.26
cold
Ephemeroptera
Ephemerellidae
Eurylophella
1785
0.3
170
20.02
cold
Trichoptera
Glossosomatidae
Glossosoma
945
0.16
119
14.02
cold
Diptera
Chironomidae
Heterotrissocladius
447
0.08
73
8.6
cold
Trichoptera
Limnephilidae
Hydatophylax
114
0.02
49
5.77
cold
Odonata
Gomphidae
Lanthus
36
0.01
11
1.3
cold
Diptera
Chironomidae
Larsia
269
0.05
58
6.83
cold
Plecoptera
Leuctridae
Leuctra
2407
0.4
142
16.73
cold
Trichoptera
Limnephilidae
Limnephilus
889
0.15
62
7.3
cold
Diptera
Chironomidae
Macropelopia
322
0.05
43
5.06
cold
Plecoptera
Perlodidae
Malirekus
0
0
0
0
cold
Trichoptera
Brachycentridae
Micrasema
405
0.07
87
10.25
cold
Diptera
Chironomidae
Natarsia
430
0.07
65
7.66
cold
Plecoptera
Nemouridae
Nemoura
17
0
4
0.47
cold
Megaloptera
Corydalidae
Nigronia
713
0.12
170
20.02
cold
Trichoptera
Phryganeidae
Oligostomis
485
0.08
87
10.25
cold
Coleoptera
Elmidae
Oulimnius
237
0.04
37
4.36
cold
Diptera
Chironomidae
Pagastia
420
0.07
96
11.31
cold
Trichoptera
Hydroptilidae
Palaeagapetus
1
0
1
0.12
cold
Plecoptera
Capniidae
Paracapnia
52
0.01
17
2
E2-9

-------
636
Table E2-1. Continued
Type
Order
Family
FinallD
Sum Individs
Pet Abund
Num Stations
Pet Stations
cold
Plecoptera
Nemouridae
Paranemoura
3
0
3
0.35
cold
Trichoptera
Hydropsychidae
Parapsyche
398
0.07
27
3.18
cold
Plecoptera
Peltoperlidae
Peltoperla
9
0
4
0.47
cold
Plecoptera
Perlodidae
Perlodidae
1775
0.3
212
24.97
cold
Diptera
Chironomidae
Prodiamesa
392
0.07
28
3.3
cold
Plecoptera
Nemouridae
Prostoia
6
0
1
0.12
cold
Diptera
Chironomidae
Pseudodiamesa
139
0.02
12
1.41
cold
Trichoptera
Limnephilidae
Psychoglypha
329
0.06
37
4.36
cold
Plecoptera
Pteronarcyidae
Pteronarcys
248
0.04
80
9.42
cold
Ephemeroptera
Heptageniidae
Rhithrogena
193
0.03
23
2.71
cold
Plecoptera
Chloroperlidae
Sweltsa
640
0.11
66
7.77
cold
Plecoptera
T aeniopterygidae
Taenionema
0
0
0
0
cold
Plecoptera
Peltoperlidae
Tallaperla
126
0.02
12
1.41
cold
Plecoptera
Capniidae
Utacapnia
71
0.01
3
0.35
cold
Plecoptera
Chloroperlidae
Utaperla
2
0
2
0.24
cold
Plecoptera
Nemouridae
Zapada
2
0
1
0.12
637
638
E2-10

-------
640	Table E2-2. List of Maine warm-water temperature indicator taxa. Distribution and abundance information is also included.
641	Sum_Individuals=the total number of individuals from that taxon in the Maine database; Pct_Abund=percent of total
642	individuals in the database comprised of that taxon; Num_Stations=number of stations in the database that the taxon occurred
643	at; Pet Stations=percent of stations in the database at which the taxon occurred	
Type
Order
Family
FinallD
Sum Individs
Pet Abund
Num Stations
Pet Stations
warm
Plecoptera
Perlidae
Acroneuria
4857
0.82
331
38.99
warm
Mesogastropoda
Hydrobiidae
Amnicola
4589
0.77
160
18.85
warm
Odonata
Coenagrionidae
Argia
869
0.15
137
16.14
warm
Plecoptera
Perlidae
Attaneuria
172
0.03
36
4.24
warm
Ephemeroptera
Caenidae
Caenis
1783
0.3
169
19.91
warm
Diptera
Chironomidae
Cardiocladius
200
0.03
52
6.12
warm
Trichoptera
Leptoceridae
Ceraclea
876
0.15
152
17.9
warm
Haplotaxida
Naididae
Chaetogaster
342
0.06
70
8.24
warm
Diptera
Chironomidae
Dicrotendipes
1978
0.33
169
19.91
warm
Arhynchobdellida
Erpobdellidae
Erpobdella
265
0.04
65
7.66
warm
Basommatophora
Ancylidae
Ferrissia
594
0.1
102
12.01
warm
Trichoptera
Helicopsychidae
Helicopsyche
2563
0.43
104
12.25
warm
Basommatophora
Planorbidae
Helisoma
716
0.12
66
7.77
warm
Diptera
Empididae
Hemerodromia
1764
0.3
260
30.62
warm
Hydroida
Hydridae
Hydra
483
0.08
113
13.31
warm
Trichoptera
Hydroptilidae
Hydroptila
1799
0.3
189
22.26
warm
Ephemeroptera
Isonychiidae
Isonychia
5413
0.91
225
26.5
warm
Diptera
Chironomidae
Labrundinia
618
0.1
137
16.14
warm
Ephemeroptera
Heptageniidae
Leucrocuta
3320
0.56
208
24.5
warm
Trichoptera
Hydropsychidae
Macrostemum
4557
0.77
168
19.79
warm
Trichoptera
Polycentropodidae
Neureclipsis
15523
2.61
320
37.69
warm
Diptera
Chironomidae
Nilotanypus
413
0.07
133
15.67
warm
Trichoptera
Leptoceridae
Oecetis
3390
0.57
306
36.04
warm
Decapoda
Cambaridae
Orconectes
381
0.06
99
11.66
644
E2-11

-------
646 Table E2-2. Continued
Type
Order
Family
FinallD
Sum Individs
Pet Abund
Num Stations
Pet Stations
warm
Diptera
Chironomidae
Parachironomus
946
0.16
83
9.78
warm
Plecoptera
Perlidae
Paragnetina
625
0.11
103
12.13
warm
Diptera
Chironomidae
Pentaneura
881
0.15
139
16.37
warm
Basommatophora
Physidae
Physa
1373
0.23
115
13.55
warm
Basommatophora
Physidae
Physella
1681
0.28
155
18.26
warm
Ephemeroptera
Baetidae
Plauditus
1285
0.22
125
14.72
warm
Hoplonemertea
T etrastemmatidae
Prostoma
267
0.04
61
7.18
warm
Diptera
Chironomidae
Psectrocladius
1693
0.28
161
18.96
warm
Ephemeroptera
Baetidae
Pseudocloeon
1147
0.19
113
13.31
warm
Diptera
Chironomidae
Rheopelopia
729
0.12
144
16.96
warm
Ephemeroptera
Ephemerellidae
Serratella
2534
0.43
191
22.5
warm
Ephemeroptera
Heptageniidae
Stenacron
6503
1.09
196
23.09
warm
Coleoptera
Elmidae
Stenelmis
2638
0.44
280
32.98
warm
Ephemeroptera
Heptagenidae
Stenonema
30768
5.18
536
63.13
warm
Diptera
Chironomidae
Tribelos
1781
0.3
78
9.19
warm
Ephemeroptera
Leptohyphidae
Tricorythodes
2655
0.45
205
24.15
E2-12

-------
647	Table E2-3. Mobility traits that were evaluated. The source of most of this information was
648	the Poff et al. 2006 traits matrix. Some also came from the USGS traits database (Vieira et
649	al. 2006)
Mobility Trait	Trait States
low (<1 km flight before laying eggs), high (>1 km flight before
laying eggs)
weak (e.g. cannot fly into light breeze), strong
rare (catastrophic only), common (typically observed), abundant
(dominant in drift samples)
very low (<10 cm/h), low (<100 cm/h), high (>100 cm/h)
none, weak, strong
650
651	Table E2-4. Number of cold-water taxa in each family
Family	Total
Chironomidae	7
Nemouridae	4
Capniidae	3
Limnephilidae	3
Chloroperlidae	2
Elmidae	2
Hydropsychidae	2
Peltoperlidae	2
Perlodidae	2
Aeshnidae
Ameletidae
Apataniidae
Brachycentridae
Corydalidae
Ephemerellidae
Glossosomatidae
Gomphidae
Heptageniidae
Hydroptilidae
Leuctridae
Phryganeidae
Pteronarcyidae
Taeniopterygidae
652
Dispersal (adult)
Adult flying strength
Occurrence in drift
Maximum crawling rate
Swimming ability
E2-1

-------
654
Table E2-5. Number of warm-water taxa in each family
Family	Total
Chironomidae	9
Perlidae	3
Physidae	2
Leptoceridae	2
Heptageniidae	2
Baetidae	2
T etrastemmatidae
Polycentropodidae
Planorbidae
Naididae
Leptohyphidae
Isonychiidae
Hydroptilidae
Hydropsychidae
Hydrobiidae
Hydridae
Heptagenidae
Helicopsychidae
Erpobdellidae
Ephemerellidae
Empididae
Elmidae
Coenagrionidae
Cambaridae
Caenidae
Ancylidae	
655
E2-2

-------
657 Table E2-6. Potential cold-water species (per recommendation by Vermont DEC)
658
659
660
Order
Genus

Species
Diptera
Polypedilum

aviceps
Diptera
Neostempellina

reissi
Diptera
Tvetenia

bavarica
Ephemeroptera
Rhithrogena

sp
Ephemeroptera
Ameletus

sp
Trichoptera
Arctopsyche

sp
Trichoptera
Arctopsyche

ladogensis
Trichoptera
Rhyacophila

Carolina
Trichoptera
Rhyacophila

torva
Trichoptera
Rhyacophila

nigrita
Trichoptera
Rhyacophila

invaria
Trichoptera
Rhyacophila

acutiloba
Plecoptera
Peltoperla

sp
Plecoptera
Tallaperla

sp
Plecoptera
Taenionema

sp
Decapoda
Cambarus

Cambarus bartoni
Trichoptera
Palaeagapetus

sp
Diptera
Eukiefferella

brevicalar, brehmi, and tirolensis
Coleoptera
Oulimnius

latiusculus
Coleoptera
Promoresia

tardella
Table E2-7. Potential warm-water species (per recommendation by Vermont D
Order
Genus
Species

Diptera
Eukiefferella
claripennis
Diptera
Polypedilum
flavum

Diptera
Tvetenia
discoloripes
Trichoptera
Leucotrichia
sp

Trichoptera
Rhyacophila
mainensis
Trichoptera
Rhyacophila
manistee
Trichoptera
Rhyacophila
minora

Plecoptera
Neoperla
sp

Plecoptera
Taeniopteryx
sp

E2-3

-------
661
Attachment E3
662
663	Tolerance values and BCG attribute levels of
664	the cold and warm-water temperature indicator
665	taxa
666
667	This attachment contains tables with lists of the temperature indicator taxa, temperature optima
668	and tolerance values that were calculated from the weighted average modeling, the tolerance
669	values assigned by Maine DEP (which are used to calculate the HBI) and BCG attribute levels
670	assigned to each taxa during the New England Wadeable Streams (NEWS) project (US EPA
671	2007). These tables were used to examine whether temperature indicator taxa were considered to
672	be sensitive or tolerant taxa.
E3-1

-------
673
674
675
676
677
Table E4-1. Cold-water temperature indicator taxa. TempOpt is the temperature optima (°C) and TempTol is the
temperature tolerance calculated during the weighted average modeling. TolVal ME is the tolerance value that was assigned
by Maine DEP and that is used in the calculation of the HBI. BCGNEWS is the BCG attribute level assigned to each taxa
during the New England Wadeable Streams project and BCG Cat is the category associated with the BCG NEWS attribute
Order
Family
FinallD
TempOpt
TempTol
TolVal ME
BCG NEWS
Coleoptera
Elmidae
Oulimnius



3
Diptera
Chironomidae
Heterotrissocladius
16.3
2.8
0
3
Diptera
Chironomidae
Larsia
17.5
3.6
6
4
Diptera
Chironomidae
Macropelopia
15.5
1.9

5
Diptera
Chironomidae
Natarsia
16.6
2.5
8
5
Diptera
Chironomidae
Pagastia
17.1
3.7
1
4
Diptera
Chironomidae
Prodiamesa
15.6
2
3
2
Diptera
Chironomidae
Pseudodiamesa




Ephemeroptera
Ameletidae
Ameletus


0
2
Ephemeroptera
Heptageniidae
Epeorus
19.9
4.9
0
2
Ephemeroptera
Ephemerellidae
Eurylophella
17.4
3.2
3
3
Ephemeroptera
Heptageniidae
Rhithrogena


0
2
Megaloptera
Corydalidae
Nigronia
20.4
2.8
0
3
Odonata
Aeshnidae
Boyeria
20.4
2.9
2
4
Odonata
Gomphidae
Lanthus



3
Plecoptera
Capniidae
Capnia


1

Plecoptera
Leuctridae
Leuctra
16.3
3
0
2
Plecoptera
Nemouridae
Nemoura


1

Plecoptera
Capniidae
Paracapnia


1
3
Plecoptera
Nemouridae
Paranemoura




Plecoptera
Peltoperlidae
Peltoperla




Plecoptera
Perlodidae
Perlodidae
17.3
4.4

3
E3-2

-------
679 Table E4-1. Continued
Order
Family
FinallD
TempOpt
TempTol
TolVal ME
BCG NEWS
Plecoptera
Nemouridae
Prostoia




Plecoptera
Pteronarcyidae
Pteronarcys
19.1
3.9
0
2
Plecoptera
Chloroperlidae
Sweltsa
14.9
3.5

3
Plecoptera
Peltoperlidae
Tallaperla



2
Plecoptera
Capniidae
Utacapnia




Plecoptera
Chloroperlidae
Utaperla




Plecoptera
Nemouridae
Zapada




Trichoptera
Apataniidae
Apatania



3
Trichoptera
Hydropsychidae
Diplectrona
16.8
2.5
0
4
Trichoptera
Glossosomatidae
Glossosoma
18.7
4.8
0
3
Trichoptera
Limnephilidae
Hydatophylax
17.7
3.2
2
3
Trichoptera
Limnephilidae
Limnephilus
17.4
2.5
3

Trichoptera
Brachycentridae
Micrasema
18.6
5.3
2
3
Trichoptera
Phryganeidae
Oligostomis
16.6
2.8
2
2
Trichoptera
Hydroptilidae
Palaeagapetus




Trichoptera
Hydropsychidae
Parapsyche
12.9
2.2
0

Trichoptera
Limnephilidae
Psychoglypha
15.3
1.7
0

680
E3-3

-------
682
683
684
685
686
687
Table E4-2. Warm-water temperature indicator taxa. Temp Opt is the temperature optima (°C) and Temp Tol is the
temperature tolerance calculated during the weighted average modeling. TolVal ME is the tolerance value that was assigned
by Maine DEP and that is used in the calculation of the HBI. BCGNEWS is the BCG attribute level assigned to each taxa
during the New England Wadeable Streams project and BCG Cat is the category associated with the BCG NEWS attribute
levels (2=highly sensitive taxa, 3=intermediate sensitive taxa, 4=taxa of intermediate tolerance, 5=tolerant taxa, 6=non native
Order
Family
FinallD
Temp Opt
Temp Tol
TolVal ME
BCG NEWS
Arhynchobdellida
Erpobdellidae
Erpobdella
20.7
3.2

6
Basommatophora
Ancylidae
Ferrissia
21.8
2.9

4
Basommatophora
Planorbidae
Helisoma
21
2.7

5
Basommatophora
Physidae
Physa
21.6
3.3

4
Basommatophora
Physidae
Physella
20.5
3.3

4
Coleoptera
Elmidae
Stenelmis
21.6
2.5
5
4
Decapoda
Cambaridae
Orconectes
22.4
2.6

5
Diptera
Chironomidae
Cardiocladius
21.8
2.4
5
6
Diptera
Chironomidae
Dicrotendipes
21.2
3.3
8
6
Diptera
Empididae
Hemerodromia
20.8
3.1
3
5
Diptera
Chironomidae
Labrundinia
21.5
3.2
7
5
Diptera
Chironomidae
Nilotanypus
22.2
2.2
6
4
Diptera
Chironomidae
Parachironomus
21.7
2.2
10
5
Diptera
Chironomidae
Pentaneura
22.4
2.8
6
4
Diptera
Chironomidae
Psectrocladius
21.6
3.3
8
4
Diptera
Chironomidae
Rheopelopia
20.8
3.1


Diptera
Chironomidae
Tribelos
20.4
2.7
5
4
Ephemeroptera
Caenidae
Caenis
21.4
3.5
7
4
Ephemeroptera
Isonychiidae
Isonychia
22
2.8
2
3
Ephemeroptera
Heptageniidae
Leucrocuta
21.2
3.3
1
3
Ephemeroptera
Baetidae
Plauditus
21.2
2.8

3
Ephemeroptera
Baetidae
Pseudocloeon
21.4
3.2
4
4
E3-4

-------
688 Table E4-2. Continued
Order
Family
FinallD
Temp Opt
Temp Tol
TolVal ME
BCG NEWS
Ephemeroptera
Ephemerellidae
Serratella
20.8
3.8
2
3
Ephemeroptera
Heptageniidae
Stenacron
21.6
2.8
7
4
Ephemeroptera
Heptagenidae
Stenonema
21
3.1
4
4
Ephemeroptera
Leptohyphidae
Tricorythodes
22.1
2.1
4
4
Haplotaxida
Naididae
Chaetogaster
20.5
1.9

5
Hoplonemertea
T etrastemmatidae
Prostoma
23.1
2.4

4
Hydroida
Hydridae
Hydra
20.5
3.5


Mesogastropoda
Hydrobiidae
Amnicola
22.7
2.4

5
Odonata
Coenagrionidae
Argia
22.7
3
7
4
Plecoptera
Perlidae
Acroneuria
21.6
2.9
0
3
Plecoptera
Perlidae
Attaneuria


1

Plecoptera
Perlidae
Paragnetina
20.7
3.6
1
3
Trichoptera
Leptoceridae
Ceraclea
21.2
3
3
2
Trichoptera
Helicopsychidae
Helicopsyche
22
2.3
3
4
Trichoptera
Hydroptilidae
Hydroptila
20.4
4.2
6
4
Trichoptera
Hydropsychidae
Macrostemum
22.7
2
3
4
Trichoptera
Polycentropodidae
Neureclipsis
22.1
2.7
7
4
Trichoptera
Leptoceridae
Oecetis
21.5
2.7
8
4
689
E3-5

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
APPENDIX F
Detailed Results for Utah
The intent of this appendix is to provide more comprehensive and detailed information on the
large number of analyses that were performed on the Utah data. Some of the analyses that are
covered in this appendix are also referenced in the main body of the report. When this occurred,
attempts were made to reduce any overlap or duplication in the reporting of results.
Fl. Overview of RI VP ACS model
F2. Methods - RIVPACS model manipulation analyses
F3. Results - RIVPACS model manipulation analyses
F4. Methods - Trends associated with climate-related variables
F5. Results - Trends associated with climate-related variables
Attachment Fl. Extreme alterations of Utah fall RIVPACS model climate-
related predictor variable values
Attachment F2. Temperature-Indicator Taxa - Utah
Attachment F3. Utah Station 4927250
Attachment F4. Utah Station 4951200
Attachment F5. Utah Station 4936750
Attachment F6. Utah Station 5940440
F-l

-------
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Fl. OVERVIEW OF THE UTAH RIVPACS MODEL
A number of states use a predictive bioassessment approach called River InVertebrate
Prediction And Classification System (RIVPACS) to assess stream condition (Wright, 2000). In
the RIVPACS model, data from reference sites are used to establish expected (E)
macroinvertebrate assemblages, and the observed (O) assemblages at sites are compared to these
expected assemblages. The ratio of these values (O/E) can be interpreted as a measure of
taxonomic completeness. Values of O/E that are near 1 at a test site suggest that the site is
comparable to reference sites, whereas values that differ substantially from 1 suggest that the site
is degraded (Yuan, 2006a).
Utah recently started using a RIVPACS model (fall samples) to rate its sites (Ostermiller,
unpublished presentation titled 'Development of a biological assessment framework'). Sites are
scored based on the ratio of O to E assemblages (expected assemblages are established based on
reference site data). Differences in site characteristics are taken into account when sites are
scored. Flow charts depicting the criteria and decision-making process that go into rating sites
are shown in Figures Fl-1 and Fl-2.
F-2

-------
Review RIVPACS Model
Beneficial Use
Fully Supported
Were three (3) or
more samples
collected?
Y = 0.74?
OIE < 0.54
Beneficial Use
Not Supported
Y m
0.54< - O/E <0.74
Category 3A
(More data needed for
an accurate assessment)
Beneficial Use
Not Supported
Does the chemistry data
agree with the biology data?
Ym
Best Professional
Judgment
(see Figure XX
for criteria)
Chemistry
Validates
RIVPACS
assessment
42
43	Figure Fl-1. Summary of the decision-making process and criteria that go into rating sites
44	using the Utah fall RIVPACS model.
F-3

-------
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
Decision criteria for differing chemical
and biological assessments
He
Do dear boundaries
for cfvidmg AUs
x. exist*? .
Were samples coll ected in a
similar environ mental setting?
No
Place in category
3Aand collect data
to find
appropriate AU boundaries
Su bdivideAU
Is the model
applicable
to the sites'?
No
Place in category
3Aandfind additional
referen ce sites
Y«
NO
Place in categoiy
3Aand collect
additional data
Yea
Place in category
4C
(habitat related impairment)
/"fa there 8vldan<*\.
Ih at 1m palrmen t Is h afttat
Assess according to
biological assessment
resu Its
Mo
Figure F12-2. Summary of the decision criteria that is used for the Utah fall
RIVPACS model when results of chemical and biological assessments differ.
RIVPACS models are built using predictor variables that are minimally affected by
human disturbance and that are considered to be relatively invariant over ecologically relevant
time (Tetra Tech, 2008; Wright et al1984; Hawkins et al., 2000; Wright, 2000; Utah State
University, 2009). Variables that are typically used include those related to geographic position
(i.e. latitude, longitude, elevation), watershed area, climate, and surficial geology (Utah State
University, 2009). If alterable variables were used (i.e., nutrient concentrations, conductivity,
forest cover), it would be difficult to discriminate the natural gradient from that caused by human
activity, and confident prediction of an expected community in the absence of human disturbance
for a test site would be impossible (Tetra Tech, 2008).
F-4

-------
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
The development of RIVPACS models requires several steps: 1) group reference sites
into clusters with similar biological assemblages; 2) examine how natural factors vary within the
clusters of reference sites; 3) for each test site, use natural factors to predict the clusters in which
the site would most likely be grouped; 4) the expected biological composition of the test site is
predicted to be the same as that observed in the reference cluster (this is expressed as the capture
probability for each taxon); and 5) compare the observed taxa list to the expected taxa list, as
expressed by the O/E ratio (Yuan, 2006b). To elaborate further on the expected taxa list, it is
conceptually a weighted average of taxa frequencies found across all reference sites, where the
weights are the probability a site is in a particular group of reference sites; average taxa
frequencies from reference sites that are physically very similar to a test site are weighted most
(Tetra Tech, 2009). The expected taxa list can be set to different thresholds (e.g., to exclude rare
taxa, the threshold can be set to 50%).
Utah DEQ uses a RIVPACS model for assessing wadeable streams. During model
development, the random forests method was used to select predictor variables that best
discriminated among the site groups (Breiman and Cutler, 2009) (NOTE: this is oftentimes
accomplished using a discriminant model). A major benefit of using the random forests method
is that the calculations are done in a way that prevents the model from being overfit. Another
valuable feature is that it gives estimates of what variables are important in the classification,
both overall and within each site group (Breiman and Cutler, 2009).
For this assessment, we explored how climate-related shifts in macroinvertebrate
assemblages may affect Utah's predictive bioassessment approach. RIVPACS analyses were run
using a number of different scenarios in which climate-related predictor variables were altered.
Questions we examined include: How does site class (membership probability) shift with
changes to the climate influenced variables? Are the climate-related predictor variables being
changed "enough" to cause any shift in the site class (membership probability) and thus a change
in E? Do the climate predictive variables have enough predictive power to change the O/E score?
Which of the climate predictor variables are most important, both overall and among site groups?
F2. RIVPACS MODEL MANIPULATION ANALYSES
F-5

-------
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
The fall model (and not the 'all seasons' model) was evaluated in this analysis, because
the fall model is the one that Utah DEQ currently uses to assess wadeable stream sites. Jeff
Ostermiller of Utah DEQ provided the R Code and data input files for the model; it is available
upon request. The data input files contained information on the 88 reference sites that were used
in the construction of the fall model. The files contain data on taxa, predictor variables and site
groups (these are available upon request). Table F2-1 contains a list of the predictor variables
that are used in the model.
RUN 1. This approach examined model performance under different climate change
scenarios. Associated questions were: how much do O, E and O/E values change in each of the
different scenarios? Is the change in O/E greater than the natural variability among reference
scores (which equals 0.13=1 Standard Deviation)? Which of the climate-related predictor
variables were most important overall? Which predictor variables were most important in each of
the different site groups? We included several different approaches for this analysis. In the first
approach, we changed combinations of climate predictor variables while keeping the observed
(O) values constant (i.e., we kept the biology, which was based on about 5 years of reference
data, the same) and the probability of capture (Pc) limit at > 0.5 (this is the Pc value that Utah
uses when running the model).
We used the NCAR1 projections for the southwestern US for 2050 and 2090 as guidance
for how much to alter the climate- predictor variables. We also ran two scenarios in which the
freeze date and the temperature and precipitation variables (i.e., all the climate-related predictor
variables) were altered simultaneously. Since we did not have information on how much freeze
dates are likely to change, we used best professional judgment and long-term averages of freeze
dates (minimum, average, maximum) from reporting stations that were closest to the 4 sites to
estimate the numbers. There were 5 different 'alteration scenarios' that were used in the
analyses. These are summarized in Table F2-2. Compared to annual climatic variations, the
alteration increments may seem small, but they are realistic for purposes of this analysis because
1 Regional Climate-Change Projections from Multi-Model Ensembles (RCPM) data and analysis were provided by
the Institute for the Study of Society and Environment (ISSE) at the National Center for Atmospheric Research
(NCAR). based on model data from the World Climate Research Programme's Coupled Model Intercomparison
Project phase 3 (WCRP CMIP3) multi-model dataset. More information about the RCPM analysis can be found at
http://rcpm.ucar.edu. © 2006 University Corporation for Atmospheric Research. All Rights Reserved."
F-6

-------
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
the predictor variables that are used in the RIVPACS models are long-term averages (i.e. 1971-
2000).
RUN 2. One limitation of the approach described above is that by only including taxa
that have Pc >.5, we may be missing an important piece of the climate change picture, which is:
what is happening to the rare taxa that are at the edges of their ranges? Are their distributions
shifting, but the model is not detecting these changes because the Pc is set to > 0.5? To address
this question, we re-ran the approach described above with the Pc set to > 0.1 to evaluate how
things changed.
RUN 3. Another question of interest was how well the RIVPACS model would perform
with only climate-related predictor variables. How much variation would climate variables alone
explain? Which of the 7 climate-related predictor variables are the key drivers? To examine this,
we ran the random forests method (Breiman and Cutler, 2009) with only the climate-related
predictor variables and evaluated performance by calculating SD and RMSE of the reference site
O/E scores.
Table F2-1. Predictor variables that are used in the fall model. Climate-related variables
are in red italicized print	
Predictor Variables	Description
MINWD. WS	Watershed average of the annual minimum of the predicted mean monthly
number of days with measurable precipitation (days) derived from PRISM
data. Each watershed grid cell calculated as MIN[Xi], where Xi = the
predicted minimum mean number of days with me
BDH.AVE	Watershed mean high values of soil bulk density of soils types within the basin
(grams per cubic centimeter) from State Soil Geographic (STATSGO)
Database.
G.PH.STD	Predicted physical activity based on lithology from state geology maps and
estimated physical weathering rates based on known rock hardness. Ordinal
ranking from low activity (1, granitic, gneiss, limestone) to high activity (5,
siltstone, shale).
AWCHAVE	Watershed mean high values of available water capacity of soils (fraction)
from State Soil Geographic (STATSGO) Database.
GPT.VOLC	Dummy Variable indicating dominant geology (l=yes; 2=N0)
ELEV.MAX	Maximum watershed elevation (meters) from National Elevation Dataset
F-7

-------
135 Table F2-1. Continued
Predictor Variables
Description
FST32AVE
MEANP.PT
Watershed average of the mean day of year (1-365) of the first freeze derived
from the PRISM data.
Annual mean of the predicted mean monthly precipitation (mm) derived from
the PRISM datas for the sampling site. Calculated as sum Xi/12, where Xi =
the predicted mean precipitation for month i (1-12) derived from 29 years of
record(1961-1990).
SQ.KM	Watershed area in square kilometers.
TMEAN. WS	Watershed average of the annual mean of the predicted mean monthly air
temperature (tenths of degree Celsius) derived from PRISM data. Each
watershed grid cell calculated as sum Xi/12, where Xi = the predicted mean air
temperature for month i (1-12) deriv
MINP.PT	Annual minimum of predicted mean monthly precipitation (mm) derived from
the PRISM data for the sampling site. Calculated as MIN[Xi], where Xi = the
predicted minimum mean precipitatioin for month i (1-12) derived from 29
years of record (1961-1990).
ELEV.WS
Mean watershed elevation (meters) from National Elevation Dataset.
SLOPE.GIS
LST32AVE
TMEANNET
Average slope calculated from GIS
Watershed average of the mean day of year (1-365) of the last freeze derived
from the PRISM data.
Stream network average of the annual mean of the predicted mean monthly air
temperature (tenths of degree Celsius) derived from PRISM data. Each stream
network grid cell calculated as sum Xi/12, where Xi = the predicted mean air
temperature for month i
F-8

-------
136
137
Table F2-2. Descriptions of how the climate-related predictor variables were altered in each of the 5 runs of the RIVPACS
model
Scenario
Category
Altered Predictor variables

Rationale
1
Baseline
None - used original values

Obtain baseline values
2
Temperature &
TMEAN.WS + 2 & TMEAN.NET H
MEANP.PT - .05
b2&
NCAR annual temperature and
precipitation predictions (2050)
3
Precipitation
TMEAN.WS + 4 & TMEAN.NET H
MEANP.PT-.l
b4&
NCAR annual temperature and
precipitation predictions (2090)
LST32AVE-1, MINP.PT-1, MEANP.PT-1,
4	TMEAN.NET+1, TMEAN.WS+1, FST32AVE+1, Best professional judgment
MINWD.WS-1
All
LST32AVE-2, MINP.PT-2, MEANP.PT-2,
5	TMEAN.NET+2, TMEAN.WS+2, FST32AVE+2, Best professional judgment
MINWD.WS-1

-------
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
F3. RESULTS - RIVPACS MODEL MANIPULATION ANALYSES
RUN 1. Overall, both mean O and mean E decreased in each of the scenarios, with E
decreasing by a greater amount than O (Tables F3-1 and F3-2). The greatest changes in O and E
values (-0.70 and -1.12, respectively) occurred in Scenario 'All 2.' O/E scores increased by very
small amounts in each scenario. The maximum change in the O/E score occurred in Scenario
'All 2' but this was only an increase of 0.03 from the baseline O/E score. The output of results
from all sites is available upon request.
NOTE: We also ran some 'extreme' scenarios (i.e. doubling temperature, dividing
precipitation values by two, changing freeze dates by 30 days, etc.) to satisfy our curiosity about
how much alteration it would take in order to result in a substantial change to O/E scores. Even
with these extremes, the O/E scores never varied by more than one standard deviation (0.13) and
were therefore still within the realm of natural variability. Results are shown in Attachment Fl.
The overall importance of the 15 predictor variables used in the Utah fall RIVPACS
model was also evaluated. Table F3-3 and Figure F3-1 list the variables in order of highest
overall variable importance to lowest (this is measured by Mean Decrease Accuracy, see
Breiman and Cutler (2009) for more information). By importance, we are referring to how
important the variable is in predicting the class correctly. The 5 most important overall variables
are annual minimum of predicted mean monthly precipitation (MINP.PT), average slope
calculated from GIS (SLOPE. GIS), watershed average of the mean day of year of the last freeze
(LST32AVE), mean watershed elevation (ELEV.WS) and stream network average of the annual
mean of the predicted mean monthly air temperature (TMEANNET). Six of the top ten most
important variables are climate-related.
In addition to evaluating the overall dataset, results within the 8 different site groups were
examined. O and E values did vary among the site groups (Table F3-4). For example, Site
Group 4 had higher O and E values than the other groups and Site Group 6 had lower values.
However, O/E scores were very similar across all site groups and if O/E scores changed, they
only changed by small amounts. Differences from baseline O/E scores ranged from 0 to 0.10,
with the greatest changes generally occurring in the 'All 1' and 'All 2' Scenarios (Table F3-5).
Mean O/E scores in Site Group 4 changed the most, but were still within the range of natural
F-10

-------
168	variation (=1 StDev, 0.13). The 6 most important predictor variables in Site Group 4 are climate-
169	related, which may be part of the reason for the bigger change (Table F3-6). However, this is not
170	clear because the 3 most important predictor variables in Site Group 8 were climate-related, and
171	Site Group 8 O/E values were the same as the baseline values in each scenario. The small sample
172	size (5 sites) may be a contributing factor.
173
174	Table F3-1. Mean O, E and O/E values for each scenario
Scenario
O
E
O/E
Altered predictor variables
liusclinc
14.4U
13. ~<)
1.05
used original values
Temp/Precip 1
14.36
13.56
1.06
TMEAN.WS + 2 & TMEAN.NET + 2 &
MEANP.PT - .05
Temp/Precip 2
14.33
13.49
1.06
TMEAN.WS + 4 & TMEAN.NET + 4 &
MEANP.PT - .1
All 1
13.83
12.80
1.07
LST32AVE-1, MINP.PT-1, MEANP.PT-1,
TMEAN.NET+1, TMEAN.WS+1,
FST32AVE+1, MINWD.WS-1
All 2
13.69
12.59
1.08
LST32AVE-2, MINP.PT-2, MEANP.PT-2,
TMEAN.NET+2, TMEAN.WS+2,
FST32AVE+2, MINWD.WS-1
175
176
177	Table F3-2. Mean difference between baseline O, E and O/E values and mean O, E and
178	O/E values for each scenario

Mean Difference from
Baseline Values

Scenario
O
E
O/E
Altered predictor variables
Temp/Precip 1
-0.03
-0.15
0.01
TMEAN.WS + 2 & TMEAN.NET + 2 &
MEANP.PT - .05
Temp/Precip 2
-0.07
-0.21
0.01
TMEAN.WS + 4 & TMEAN.NET + 4 &
MEANP.PT - .1
All 1
-0.57
-0.91
0.03
LST32AVE-1, MINP.PT-1, MEANP.PT-1,
TMEAN.NET+1, TMEAN.WS+1, FST32AVE+1,
MINWD.WS-1
All 2
-0.70
-1.12
0.03
LST32AVE-2, MINP.PT-2, MEANP.PT-2,
TMEAN.NET+2, TMEAN.WS+2, FST32AVE+2,
MINWD.WS-1
179
F-ll

-------
fall.random.forest.G10L2
MINP.PT
SLOPE.GIS
LST32AVE
ELEV.WS
TMEANNET
MEANP.PT
TMEAN.WS
FST32AVE
MINWD.WS
SQ.KM
ELE V.MAX
G.PH.STD
BDH.AVE
GPT.VOLC
AWCH.AVE
n—i—i—r
4 5 6 7 8 9
MeanDecreaseAccuracy
SLOPE.GIS
MINP.PT
ELEV.WS
LST32AVE
TMEANNET
MEANP.PT
TMEAN.WS
SQ.KM
FST32AVE
ELE V.MAX
MINWD.WS
G.PH.STD
AWCH.AVE
BDH.AVE
GPT.VOLC
i—r
1 2 3 4 5
MeanDecreaseGini
180
181
182
183
184
Figure F3-1. Plot summarizing importance of the predictor variables (more important
variables have higher Mean Decrease Accuracy and Mean Decrease Gini scores). The
following R code command was used to obtain this table:
varImpPlot(fall.random.forest.G10L2).
F-12

-------
185	Table F3-3. Predictor variables are listed in order of highest overall variable importance to lowest (as measured by Mean
186	Decrease Accuracy). The importance of the predictor variables varies among site groups. The values under each site group
187	give the importance of the variable for predicting that class correctly. The Mean Decrease Gini calculation is performed by
188	adding up the Gini decreases for each individual variable over all trees in the forest. This gives a fast variable importance that
189	is often very consistent with the permutation importance measure. The following R code command was used to obtain this
190	table: importance(fall.random.forest.G10L2).	
Site Group
Predictor
Variable
1
2
3
4
5
6
7
8
Mean
Decrease
Accuracy
Mean
Decrease
Gini
MINP.PT
4.9
2.5
5.7
31.5
11.0
8.0
22.0
28.2
9.62
6.04
SLOPE.GIS
2.3
0.1
8.9
14.3
20.6
19.1
23.2
OO
00
9.39
6.55
LST32AVE
8.9
11.7
2.1
33.6
3.7
4.8
18.0
27.9
9.28
5.95
ELEV.WS
6.3
8.7
6.7
16.1
10.1
4.7
27.7
29.1
9.24
5.99
TMEANNET
5.1
5.8
2.2
25.1
6.4
5.5
22.7
33.3
9.18
5.86
MEANP.PT
1.1
9.9
2.9
21.3
14.3
21.8
14.9
26.5
9.08
5.82
TMEAN.WS
4.0
7.4
2.7
21.8
7.4
1.2
22.7
33.1
8.75
5.50
FST32AVE
7.9
9.9
1.7
13.6
0.6
3.0
28.6
33.0
8.39
4.90
MINWD.WS
3.6
0.2
-0.9
25.9
4.7
0.5
24.2
23.8
7.89
4.36
SQ.KM
5.6
-2.7
6.2
9.6
11.5
18.6
16.1
21.5
7.60
5.30
ELEV.MAX
4.9
7.8
-1.4
18.3
7.9
7.9
4.4
25.2
7.34
4.65
G.PH.STD
1.5
-0.5
9.0
11.4
10.8
5.8
3.2
-1.8
5.68
4.30
BDH.AVE
2.3
13.3
-0.4
7.7
-2.0
3.9
5.3
21.0
4.62
3.57
GPT.VOLC
10.8
5.0
-3.4
4.9
-3.1
7.5
5.7
4.8
3.67
1.01
AWCH AVE
-2.4
16.0
0.1
5.2
3.2
OO
00
-3.7
8.0
3.43
4.12
F-13

-------
191
192
193
Table F3-4. For each scenario and each site group, O, E and O/E values were calculated. N=number of sites in the dataset that
are assigned each site group. See Tables F3-1 or F3-2 for descriptions on how variables were altered in the Temp/Precip 1,

Baseline
Temp/ Precip 1
Temp/Precip 2
All 1
All 2
Site
Group
N
O
E
O/E
O
E
O/E
O
E
O/E
O
E
O/E
O
E
O/E
1
16
15.1
14.2
1.1
15.2
14.2
1.1
15.1
14.1
1.1
14.5
13.2
1.1
14.5
13.0
1.1
2
10
15.9
14.8
1.1
16.0
14.8
1.1
16.0
14.7
1.1
15.2
13.7
1.1
15.1
13.5
1.1
3
16
17.2
16.3
1.1
16.9
15.9
1.1
16.9
15.9
1.1
16.4
14.9
1.1
15.8
14.4
1.1
4
7
24.3
23.2
1.0
24.1
22.7
1.1
24.1
22.4
1.1
22.6
20.0
1.1
22.6
19.7
1.1
5
19
12.0
11.7
1.0
12.0
11.5
1.0
12.1
11.6
1.0
11.6
11.3
1.0
11.6
11.3
1.0
6
9
8.3
8.1
1.0
8.3
8.1
1.0
8.0
7.9
1.0
8.0
7.8
1.0
7.8
7.7
1.0
7
6
10.2
9.7
1.1
10.2
9.6
1.1
10.2
9.6
1.1
10.5
9.8
1.1
10.5
9.7
1.1
8
5
11.4
11.0
1.0
11.4
11.1
1.0
11.4
11.1
1.0
11.4
11.1
1.0
11.4
11.1
1.0
194
195
196
197
Table F3-5. For each scenario and each site group, differences between mean O/E values and mean baseline O/E values were
calculated. N=number of sites in the dataset that are assigned each site group. See Tables F3-1 or F3-2 for descriptions on how
variables were altered in the Temp/Precip 1, Temp/Precip 2, All 1 and All 2 scenarios.	
Mean Difference from Baseline O/E
Site
Group
N
Baseline
Temp/
Precip 1
Temp/
Precip 2
All 1
All 2
1
16
1.06
0.01
0.01
0.03
0.05
2
10
1.07
0.01
0.02
0.03
0.05
3
16
1.06
0.01
0.01
0.04
0.04
4
7
1.05
0.02
0.03
0.08
0.10
5
19
1.03
0.01
0.01
0.00
0.00
6
9
1.03
0.00
-0.03
0.00
-0.02
7
6
1.05
0.00
0.00
0.02
0.03
8
5
1.03
0.00
0.00
0.00
0.00
F-14

-------
198	Table F3-6. Predictor variables are listed in order of highest variable importance to lowest within each site group (this refers
199	to the importance of the variable for predicting class correctly). These results are derived from the Baseline scenario. Variable
200	importance was also evaluated for the other scenarios, but it was determined that results (at least for the 5 most important
201	variables in each site group) either did not change or varied only slightly and are therefore not reported.	
Site Group
1
2
3
4
5
6
7
8
uveraii
GPT.VOLC
AWCH.AVE
G.PH.STD
LST32AVE
SLOPE.GIS
MEANP.PT
FST32AVE
TMEANNET
MINP.PT
LST32AVE
BDH.AVE
SLOPE. GIS
MINP.PT
MEANP.PT
SLOPE.GIS
ELEV.WS
TMEAN.WS
SLOPE. GIS
FST32AVE
LST32AVE
ELEV.WS
MINWD.WS
SQ.KM
SQ.KM
MINWD.WS
FST32AVE
LST32AVE
ELEV.WS
MEANP.PT
SQ.KM
TMEANNET
MINP.PT
AWCH.AVE
SLOPE.GIS
ELEV.WS
ELEV.WS
SQ.KM
FST32AVE
MINP.PT
TMEAN.WS
G.PH.STD
MINP.PT
TMEAN.WS
MINP.PT
TMEANNET
TMEANNET
ELEV.WS
MEANP.PT
MEANP.PT
ELEV.WS
ELEV.MAX
TMEANNET
LST32AVE
MEANP.PT
ELEV.MAX
ELEV.MAX
TMEAN.WS
ELEV.MAX
ELEV.MAX
GPT.VOLC
MINP.PT
MEANP.PT
TMEAN.WS
MINP.PT
TMEAN.WS
TMEANNET
ELEV.WS
TMEAN.WS
G.PH.STD
LST32AVE
ELEV.MAX
FST32AVE
TMEAN.WS
TMEANNET
LST32AVE
SLOPE. GIS
TMEANNET
TMEANNET
SQ.KM
MINWD.WS
MINWD.WS
MINWD.WS
GPT.VOLC
FST32AVE
FST32AVE
MINWD.WS
LST32AVE
MEANP.PT
SQ.KM
SQ.KM
SLOPE.GIS
MINP.PT
AWCH.AVE
G.PH.STD
LST32AVE
ELEV.WS
GPT.VOLC
BDH.AVE
ELEV.MAX
BDH.AVE
MINWD.WS
BDH.AVE
SQ.KM
AWCH.AVE
BDH.AVE
BDH.AVE
SLOPE.GIS
G.PH.STD
G.PH.STD
SLOPE.GIS
MINWD.WS
BDH.AVE
FST32AVE
FST32AVE
ELEV.MAX
AWCH.AVE
BDH.AVE
MEANP.PT
G.PH.STD
ELEV.MAX
AWCH.AVE
BDH.AVE
TMEAN.WS
G.PH.STD
GPT.VOLC
GPT.VOLC
AWCH.AVE
SQ.KM
GPT.VOLC
GPT.VOLC
GPT.VOLC
MINWD.WS
AWCH.AVE
G.PH.STD
AWCH.AVE
202
203
F-15

-------
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
RUN 2. Normally the Utah fall RIVPACS model is ran with the probability of capture
(Pc) limit set at > 0.5. To evaluate model performance when rare taxa are included (i.e. taxa that
occur at the edges of their ranges and are likely to be more sensitive than others to climate
change), the model was run with the Pc set to > 0.1. This model did not perform as well. The
standard deviation of O/E scores for the Pc > 0.1 run was 0.18 versus 0.13 for the Pc > 0.5 run
(Table F3-7). The mean O/E score for the Pc > 0.1 run was slightly lower (a difference of 0.02).
When the two sets of O/E values are fitted with a linear regression line, the r2 value = 0.51
(Figure F3-2).
When mean O, E and O/E values are further compared among the Pc > 0.1 and Pc > 0.5
datasets, as expected, the mean O and mean E values in the Pc > 0.1 dataset are higher (by about
10 taxa) than those in the Pc > 0.5 but differences in mean O/E values are very small, ranging
from 0.02 to 0.03 (Tables F3-8 and F3-9). The alteration of the climate-related predictor
variables had little if any effect on mean O and mean E values in the Pc > 0.1 dataset, as well as
on O/E values. Only the 'All 1' and 'All 2' scenarios resulted in changes to the O/E values, and
they only increased by 0.02. A comparison of mean differences from baseline values shows that
the alteration of predictor variables generally caused a greater change in mean O, E and O/E
values in the Pc > 0.5 dataset. Mean E values changed the most in both datasets; the maximum
change in mean E values was -1.12 in the Pc > 0.5 dataset, whereas in the Pc > 0.1 dataset, the
maximum change was -0.55. This occurred in the 'All 2' Scenario.
Table F3-7. Comparison of the mean O/E scores and standard deviations that were
derived from the original Pc > 0.5 model versus the Pc > 0.1 model
Model
Mean O/E
St Dev
Pc > 0.1
1.03
0.18
Pc > 0.5
1.05
0.13
F-16

-------
1.4
0.7
0.6
228
229
230
0.5
0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
O.E_Pc0.1
Figure F3-2. Plot of O/E scores when the probability of capture (Pc) limit is set at > 0.5
versus O/E scores when the probability of capture (Pc) limit is set at > 0.1.
1.5
F-17

-------
231 Table F3-8. Mean O, E and O/E values for each scenario for both the Pc > 0.1 and Pc > 0.5 runs

Pc > 0.1
Pc > 0.5

Scenario
O
F
O/F
O
F
O/F
Altered predictor variables
liascline
23.4

1.03
14.4
13.-
1.U5
used original values
Temp/Precip 1
23.4
22.7
1.03
14.4
13.6
1.06
TMEAN.WS + 2 & TMEAN.NET + 2 & MEANP.PT - .05
Temp/Precip 2
23.4
22.6
1.03
14.3
13.5
1.06
TMEAN.WS + 4 & TMEAN.NET + 4 & MEANP.PT - .1
All 1
23.4
22.3
1.05
13.8
12.8
1.07
LST32AVE-1, MINP.PT-1, MEANP.PT-1,
TMEAN.NET+1, TMEAN.WS+1, FST32AVE+1,
MINWD.WS-1
All 2
23.4
22.2
1.05
13.7
12.6
1.08
LST32AVE-2, MINP.PT-2, MEANP.PT-2,
TMEAN.NET+2, TMEAN.WS+2, FST32AVE+2,
MINWD.WS-1
232
233
234	Table F3-9. Mean differences between baseline O, E and O/E values and mean O, E and O/E values for each scenario for both
235	the Pc > 0.1 and Pc > 0.5 runs

Mean Differences from Baseline Values


Pc > 0.1
Pc > 0.5

Scenario
O
E
O/E
O
E
O/E
Altered predictor variables
Temp/Precip 1
0.01
-0.03
0.00
-0.03
-0.15
0.01
TMEAN.WS + 2 & TMEAN.NET + 2 & MEANP.PT - .05
Temp/Precip 2
0.02
-0.09
0.00
-0.07
-0.21
0.01
TMEAN.WS + 4 & TMEAN.NET + 4 & MEANP.PT - .1
All 1
0.06
-0.45
0.02
-0.57
-0.91
0.03
LST32AVE-1, MINP.PT-1, MEANP.PT-1,
TMEAN.NET+1, TMEAN.WS+1, FST32AVE+1,
MINWD.WS-1
All 2
-0.03
-0.55
0.02
-0.70
-1.12
0.03
LST32AVE-2, MINP.PT-2, MEANP.PT-2,
TMEAN.NET+2, TMEAN.WS+2, FST32AVE+2,
MINWD.WS-1
F-18

-------
236
237
238
239
240
241
242
243
244
245
246
247
248
249
RUN 3. The performance of the RIVPACS model with only the climate-related predictor
variables was evaluated and compared to original model results. Model performance was
evaluated by looking at the standard deviations of the reference site O/E scores. When limited to
climate-related predictor variables only, the model performs well. The standard deviation of the
O/E scores using the original model is 0.13 versus 0.14 with the climate-related variables only
model (Table F3-10). Also, when the two sets of O/E values are plotted against one another,
there is a tight fit (r2 = 0.91) (Figure F3-3). The most important predictor variables are annual
minimum of predicted mean monthly precipitation (MINP.PT) and watershed average of the
mean day of year of the last freeze (LST32AVE) (Table F3-11).
Table F3-10. Comparison of the mean O/E scores and standard deviations that were
derived from the original model versus the model with only the climate-related variables
Model
Mean O/E St Dev
Original
Climate-related Only
1.05	0.13
1.05	0.14
F-19

-------
1.4
1.3
1.2
£ 1.1
O
¦a

| 0.9
o
LLI
o 0.3
0.7
0.6
O.E Original:O.E Climate-Related Only: r2 = 0.9133
250	O.E Original
251	Figure F3-3. Plot of O/E scores from the original model versus O/E scores from the
252	climate-related predictor variables only model.
253
F-20

-------
254	Table F3-11. Predictor variables are listed in order of highest overall variable importance to lowest (as measured by Mean
255	Decrease Accuracy). The importance of the predictor variables varies among site groups. The values under each site group
256	give the importance of the variable for predicting that class correctly. The Mean Decrease Gini calculation is performed by
257	adding up the Gini decreases for each individual variable over all trees in the forest. This gives a fast variable importance that
258	is often very consistent with the permutation importance measure. The following R code command was used to obtain this
259	table: importance(fall.random.forest.G10L2).	
Site Group
Predictor
Variable
1
2
3
4
5
6
7
8
Mean
Decrease
Accuracy
Mean
Decrease
Gini
MINP.PT
8.17
2.34
6.69
36.10
18.63
14.73
32.57
35.95
12.31
11.53
LST32AVE
12.88
15.40
0.94
50.80
6.89
3.03
18.58
32.54
12.07
11.89
TMEANNET
6.32
8.75
2.83
27.01
7.52
0.84
24.11
42.99
10.43
10.37
MEANP.PT
-2.79
9.85
2.45
25.92
18.09
26.67
14.86
33.50
10.30
11.54
TMEAN.WS
5.68
8.18
3.27
20.70
11.82
-2.56
24.17
41.54
10.24
9.96
FST32AVE
9.68
11.30
6.41
11.07
-2.89
-3.85
39.83
44.08
10.22
10.74
MINWD.WS
5.09
-2.36
0.78
27.29
4.33
10.02
31.69
28.09
9.22
7.89
260
F-21

-------
SUMMARY OF RIVPACS MODEL MANIPULATION RESULTS
•	Overall, altering the climate-related predictor variables had very little effect on O/E
values. The greatest change occurred in the 'All 2' scenario but this only amounted to a
change of 0.03, which is within the realm of natural variability (-0.13, 1 st dev) (see main
report section 3.4.2 for discussion on possible reasons for the small change).
•	There was also little effect on O/E values when 'unrealistic' changes were made to
climate-related variables (i.e. doubling temperature, halving precipitation variables). O/E
values never varied by more than one standard deviation (see main report section 3.4.2
for discussion on possible reasons for the small change).
•	In the Utah fall RIVPACS model, the (overall) most important climate-related variables
are annual minimum of predicted mean monthly precipitation (MINP.PT), watershed
average of the mean day of year of the last freeze (LST32AVE) and stream network
average of the annual mean of the predicted mean monthly air temperature
(TMEANNET). Six of the top ten most important variables are climate-related.
•	When O/E values were evaluated within the 8 different site groups, scores were very
similar across all site groups and if O/E scores changed, they only changed by small
amounts. Mean O/E scores in Site Group 4 changed the most, but were still within the
range of natural variation. The 6 most important predictor variables in Site Group 4 are
climate-related, which may be part of the reason for the bigger change within this site
group.
•	When the probability of capture (Pc) limit was changed from Pc > 0.5 to > 0.1, the Utah
fall RIVPACS model did not perform as well (st dev of 0.18 versus 0.13). The alteration
of the climate-related predictor variables had little if any effect on O/E values, (in fact,
alteration of climate-related variables generally had less of an effect in the Pc < 0.1 run).
Only the 'All 1' and 'All 2' Scenarios resulted in changes to the O/E values, and they only
increased by 0.02.
•	When run with climate-related predictor variables only, the Utah RIVPACS model
performed very well. The standard deviation of the O/E scores using the original model is
0.13 versus 0.14 with the climate-related variables only model. When the two sets of O/E
values are plotted against one another, there is a tight fit (r2 = 0.91).
•	See Section 3 of the report for discussion of possible explanations on why the RIVPACS
model appears to be insensitive to climate change effect, at least based on results from
our analyses. To briefly summarize, the long-term averages of the climate predictor
variables in the model capture major spatial differences between the regions (classes),
which at this point are probably bigger than long-term climate change difference are.
F-22

-------
F4. UTAH ECOREGION DESCRIPTIONS
Wasatch and Uinta Mountains. This ecoregion is composed of a core area of high,
precipitous mountains with narrow crests and valleys flanked in some areas by dissected plateaus
and open high mountains. The elevational banding pattern of vegetation is similar to that of the
Southern Rockies except that aspen, chaparral, and juniper-piny on and oak are more common at
middle elevations. This characteristic, along with a far lesser extent of lodgepole pine and greater
use of the region for grazing livestock in the summer months, distinguish the Wasatch and Uinta
Mountains ecoregion from the more northerly Middle Rockies (US EPA 2002).
Colorado Plateaus. Rugged tableland topography is typical of the Colorado Plateau
ecoregion. Precipitous side-walls mark abrupt changes in local relief, often from 300 to 600
meters. The region is more elevated than the Wyoming Basin to the north and therefore contains
a far greater extent of pinyon-juniper woodlands. However, the region also has large low lying
areas containing saltbrush-greasewood (typical of hotter drier areas), which are generally not
found in the higher Arizona/New Mexico Plateau to the south where grasslands are common (US
EPA 2002).
F-23

-------
1	Tables F4-1 and F4-2 summarize distribution and abundance information for the Utah
2	temperature indicator taxa at the 4 sites (Stations 4927250, 4951200, 4936750 and 5940440) and
3	3 site groups that were analyzed for long-term trends. Ephemerella seems to be the strongest
4	indicator because it occurred at all the sites and generally had higher mean relative abundances
5	than the other taxa. Chelifera, Chloroperlidae, Cinygmula, Lepidostoma and Rhitrogena also
6	occurred at all the sites, although generally in lower abundances. Overall, the cold-water taxa are
7	well-represented at most of the sites and site groups. Station 5940440 has the least number of
8	cold-water taxa, but nevertheless has moderate abundances of Chloroperlidae and Rhitrogena.
9	Leptohyphidae appears to be the strongest indicator among the warm-water taxa2 because it
10	occurred at 5 sites and generally had higher mean relative abundances than the other taxa. The
11	next strongest warm-water indicators appear to be Oecetis and Cheumatopsyche, which are
12	present at 6 sites but occurred in lower abundances.
13
14	Table F4-1. Summary of distribution and abundance information for the cold-water
15	temperature indicator taxa at the 4 sites (Stations 4927250, 4951200, 4936750 and 5940440)
16	and 3 site groups (WU_SF=Wasatch and Uinta Semi-arid Foothills, WU_ ME= Wasatch
17	and Uinta Mid-elevation Mountains, CP=Colorado Plateaus). #Sites refers to the number
18	of sites or site groups at which the taxa occurs. A=absent. P=present (highlighted in grey).
19	Relative abundance codes: L=low (<0.01), M=medium (0.01-0.1), H=high (>0.1) (M or H
20	are in bold type). Guide to interpretation: P-1L = present, occurred during 1 year, low
21	relative abundance (RA), P-11M = present, occurred during 11 years, medium RA, etc.
FinallD
#Sites
4927250
4936750
4951200
5940440
WU_SF
WU_ME
CP
Ameletus
5
A
IMI.
IMI.
A
l'-l.
I'-'M.
I'-M.
Anagapetus
0
A
\
\
A
A
A
A
Apatania
3
\
P-IIM
\>-:\.
A
\
ij-:i.
\
Bezzia
6

\>-:\.
IMI.
A
IM II.
\>-h\.
I'ol.
Bibiocephala
1
\
\
\
A
\
I'-2I.
\
Capniidae
6
IMI.
\
IMI.
\>-:\.
I'-'M.
P-IOM
I'ol.
Chelifera
7
i'-(>i.
\>-h\.
\>-:\.
IMI.
P-IIM
I'-XI.
I'-l.
Chloroperlidae
7
\>-h\.
IMUI.
IMI.
P-9M
P-I9M
P-I2M
I'-"I.
Cinygma
1
A
A
\>-:\.
A
A
A
A
2 There are noticeably fewer warm-water indicator taxa in Utah (when compared to cold-water taxa and
also when compared to warm-water indicator taxa lists in Maine and North Carolina). The most likely
explanation appears to be the fact that our OTU caused Chironomidae to be grouped to the family-level
(Number of warm-water taxa that are in the Chironomidae family in Maine=9 and in North Carolina=5).
There may be several other contributing factors as well.
F-24

-------
Cinygmula
7

IMI.
I'ol.
IMI.
P-I4M
P-IOM
P-SM
Cultus
6
I'ol.
IMI.
\
IMI.
I'-XI.
l'-(>l.
I'-M.
Dicranota
5
\
IMI.
\>-M.
\
IM II.
I'ol.
I'ol.
Ecclisomyia
0
\
\
\
\
\
\
\
Ephemerella
7
P-IJM
P-IOM
P-IIM
\>-:\.
P-16M
P-IOM
P-(.M
Glutops
1
A
IMI.
A
A
A
A
A
Heterlimnius
0
A
A
A
A
A
A
A
Ironodes
0
A
A
A
A
A
A
A
Kogotus
1
\
\
\
\
IMI.
\
\
Lepidostoma
7
I'-XI.
P-SM
\>-:\.
\>-h\.
IM II.
P-SM
P-4M
Leuctridae
3
A
A
A
A
I'-"I.
\>-h\.
\>-:\.
Megarcys
2
A
A
A
A
IMI.
\>-:\.
\
Nematoda
6

P-'JM
P-SM
A
P-IJM
\>-<>\.
l'-~L
Neothremma
4
\
1mm.
\
A
IMI.
P-SM
I'-M.
Oligophlebodes
3
\
P-IM
\
A
\
P-5M
\>-:\.
Oreogeton
2
A
A
A
A
\>-:\.
IMI.
\
F-25

-------
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Table F4-1. Continued
FinallD
#Sites
4927250
4936750
4951200
5940440
WIISF WU_ME
CP
Parapsyche
2
\
\
\
\
IMI.
i'-:i.
\
Pericoma
6
P-ll.
IMI.
I'-M.
\
P-I5M
i'-(>i.
I'-M.
Rhabdomastix
0
\
\
\
\
\
\
\
Rhithrogena
7
Pol.
P-6.M
i'-:i.
P-9M
P-I3M
P-SI.
\>-h\.
Taenionema
5
IMI.
\
P-5M
\
l'-l.
I'-M.
i'-:i.
Visoka
0
A
A
A
A
A
A
A
Wiedemannia
3
A
A
IMI.
IMI.
IMI.
A
A
Yoraperla
0
A
A
A
A
A
A
A
Table F4-2. Summary of distribution and abundance information for the warm-water
temperature indicator taxa at the 4 sites (Stations 4927250, 4951200, 4936750 and 5940440)
and 3 site groups (WU_SF=Wasatch and Uinta Semi-arid Foothills, WU_ ME= Wasatch
and Uinta Mid-elevation Mountains, CP=Colorado Plateaus). #Sites refers to the number
of sites or site groups at which the taxa occurs. A=absent. P=present (highlighted in grey).
Relative abundance codes: L=low (<0.01), M=medium (0.01-0.1), H=high (>0.1) (M or H
are in bold type). Guide to interpretation: P-1L = present, occurred duringl year, low
FinallD
#Sites
4927250
4936750
4951200
5940440
WU_SF
WU_ME
CP
Ambry sus
2
A
A
IMI.
A
A
A
IMI.
Asellidae
3
IMI.
A
IMI.
A
IMI.
A
A
Caenis
1
A
A
A
A
IMI.
A
A
Calineuria
1
A
A
A
A
\
IMI.
\
Caloparyphus
2
\
\
i'-:i.
A
I'-M.
\
\
Cheumatopsyche
(>
IMI.
I'-M.
I'-M.
A
\>-:\.
I'-ll.
I'-ll.
Coenagrionidae
3
\
\
I'-M.
A
IMI.
A
IMI.
Leptohyphidae
5
P-'M.
I'-M.
1*-1511
A
I'-XI.
A
P-6M
Maruina
2
A
A
A
A
P-ll.
A
I'-ll.
Microcylloepus
2
A
A
P-4M
A
A
A
P-2M
Nectopsyche
0
A
A
A
A
A
A
A
Ochrotrichia
3
IMI.
\
IMI.
A
IMI.
A
A
Oecetis
6
I'-"I.
IMI.
IMI.
A
I'-'M.
\>-2\.
P-ll.
Ordobrevia
0
A
A
A
A
A
A
A
Psephenus
0
A
A
A
A
A
A
A
Tinodes
1
Pol.
A
A
A
A
A
A
F-26

-------
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
F5. NMDS Ordination and ANOVA analyses
Results from the NMDS ordinations show that hottest year samples at Stations 4927250
(Weber) and 4951200 (Virgin) form distinct clusters when grouped by hot/cold/normal years,
and coldest and normal samples are generally mixed together (Figures 2-17 and 2-18,
respectively, in main body of report). The following environmental variables were most strongly
correlated with Axes 1 and 2, which are the axes that explained the most variance: PRISM mean
annual air temperature from the year the biological sample was collected; PRISM mean annual
air temperature from the previous year; absolute difference between collection year and previous
year PRISM mean annual precipitation. It should be noted that the hottest years occurred
sequentially (2000-2005). When grouped by precipitation category, samples at Stations 4927250
(Weber) and 4951200 (Virgin) do not form distinct clusters (Figures F5-1 and F5-2,
respectively).
F-27

-------
Utah StationID 4927250
1987
~
i 1986
1985
~
Cat_Prec
~ 1
4
1989
~
1995
A
1991
~
x
<
1992
~
1998
~
1994
A
1993
~
1990
~
2000
~
2001
A
2003	2004
	A	A
Axis 1
55	Figure F5-1. NMDS plot (Axis 1-2). Cat_Prec refers to the precipitation categories, which are:
56	l=dry years; 2=normal years; 3=wet years. Samples are labeled by collection year.
F-28

-------
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
Utah StationID 4951200
	PrevYrJ
tmean14
Axis 1
Figure F5-2. NMDS plot (Axis 1-2). Cat_Prec refers to the precipitation categories, which are:
l=dry years; 2=normal years; 3=wet years. Samples are labeled by collection year.
In addition, for the NMDS ordinations on coldest/normal/hottest year samples, we
evaluated which taxa were most strongly correlated with each axis. As shown in Figure 2-19 in
the main report, at Station 4927250 (Weber), Pteronarcys, Chloroperlidae and Ephemerella have
the strongest positive correlations with Axis 2, and Optioservus, Lepidostoma and Hyallela have
the strongest negative correlations with Axis 2. Closer examination of those taxa plotted in
ordination space shows that Chloroperlidae and Pteronarcys are absent from the hottest year
samples and Ephemerella is present in all the coldest and normal year samples and in only one
hottest year sample. Some additional taxa that occurred during multiple years that were not
found in hottest year samples include Rhithrogena, Nematoda, and Tubificidae. NMDS plots of
Optioservus, Lepidostoma and Hyallela show these taxa to be present in at least 4 of the 5 hottest
year samples. These taxa are also present in coldest and normal year samples. These plots and
associated information are available upon request.
F-29

-------
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
As shown in Figure 2-20 in the main report, at Station 4951200 (Virgin), Ephemerella,
Nematoda and Heptagenia have the strongest negative correlations with Axis 1, and
Forcipomyia/Probezzia, Microcylloepus, Caloparyphus and Chimarra have the strongest positive
correlations with Axis 1. Closer examination of those taxa plotted in ordination space shows that
Nematoda is absent from the hottest year samples. Ephemerella and Heptagenia are present in all
coldest year samples, 6 of the 7 normal year samples and only 1 of the hottest year samples.
Forcipomyia/Probezzia, Microcylloepus, Caloparyphus and Chimarra are present in at least 2 of
the 4 hottest year samples. These taxa are not present in coldest and/or normal year samples.
These plots and associated information are available upon request.
One-way ANOVA analyses were performed to evaluate differences in mean values of
commonly-used, ecological trait and scenario metrics when samples were grouped by coldest,
normal, and hottest or driest, normal and wettest years. Results varied by site. Two stations
(4927250 (Weber - Wasatch Uinta) and 4951200 (Virgin - Colorado Plateau)) showed relatively
strong temperature patterns, while one site (5940440 - Beaver) showed no patterns at all. The
greatest differences generally occurred between hottest and coldest year samples, while coldest
and normal year samples tended to be similar. Metrics that had at least one significant difference
between coldest, normal, and hottest or driest, normal and wettest years are shown in Tables F5-
1 and F5-2. These tables do not include results for thermal-preference metrics, which are shown
in Table 2-2 of the main report. Additional results are available upon request.
Table F5-1. These metrics had at least one significant difference when one-way analysis of
variance was done to evaluate differences in samples grouped by coldest, normal, and hottest
years. Year groups were based on Parameter-elevation Regressions on Independent Slopes
Model (PRISM) mean annual air temperature values at each site. Groups with the same
Station
Metric
Coldest
Normal
Hottest

# Ephemeroptera taxa
7.1 ± 1.2a
4.9 ±2.1ab
2.6 ±0.9b

# EPT taxa
17.4 ± 2.1a
13.6 ±4.9ab
8.8 ±2.2b
4927250
(Weber)
% Collector-filterer individuals
13.4 ± 7.6a
32.1 ± 15.4ab
40.1 ± 17.3b
% Collector-gatherer individuals
69.9 ± 16.9a
50.9 ±23.6^
33.5 ± 19.7b
# Collector-gatherer taxa
8.2 ± 0.8a
6.3 ±2.3ab
4.0 ± 1.4B

# Predator taxa
7.3 ± 1.6a
5.9 ±2^
3.8 ± 1.3B

# Clinger taxa
17.8 ± 1.3a
14 ±4.8ab
9 ± 1.2B
F-30

-------

# Plecoptera taxa
3.2 ± 0.8a
3.1 ± 1.5ab
0.8 ± 0.4b

# Total taxa
27.5 ± 3.5a
21.5 ±7.8^
17.2 ± 3.3b

# Warmer-drier vulnerable taxa
4.0 ± 1.2a
2.7 ± 1.1AB
1.0 ± 0.7b

# Drier vulnerable taxa
12.6 ± 0.9a
9.0 ± 2.9ab
5.4 ± 1.8b

# Ephemeroptera taxa
6.8 ± 2.2a
4.8 ± 0.8ab
2.5 ± 0.6b

# EPT taxa
12.3 ± 3.9a
9.5 ± 2.6ab
5.3 ± 1.5B

# Herbivore taxa
5.3 ± 1.5a
4.2 ± 1.6a
1.5 ± 0.6b
4951200
(Virgin)
# Burrower taxa
1 ±0AB
1.5 ± 0.5a
0.3 ± 0.5b
% Swimmer individuals
13.2 ± 3.7a
14.8 ± 8.4a
33 ± 10.9b
Shannon-Wiener diversity index
2.8 ± 0.4ab
2.9 ± 0.2a
2.2 ± 0.4b

# Total taxa
22.8 ± 6.6a
19.8 ± 3.2ab
14.5 ± 1.9b

% Drought-resistant individuals
45.8 ± 17.9a
40 ± 20.3^
13.1 ±4.4b

# Perennial taxa
10.3 ± 2.2a
7.7 ± 2.2ab
6 ± 0.8b
99
100
101	Table F5-2. These metrics had at least one significant difference when one-way analysis of
102	variance was done to evaluate differences in samples grouped by driest, normal, and
103	wettest years. Year groups were based on Parameter-elevation Regressions on Independent
104	Slopes Model (PRISM) mean annual precipitation values at each site. Groups with the
same superscripts are not significantly difl
'erent (p < 0.C
)5).
Station
Metric
Driest
Normal
Wettest
4927250
(Weber)
# Intermittent taxa
1.4 ± 0.5a
1.7 ± 0.5ab
2.6 ± 0.9b
4951200
(Virgin)
% Collector-gatherer individuals
47.9 ± 9.6a
73.0 ± 15.7b
75.9 ± 0.3b
106
107	Mean O/E values at the two Colorado Plateau stations (4951200-Virgin and 4936750-
108	Duchesne) were significantly different between hot year and cold/normal year samples (see
109	Section 3 of the report). O/E values at the 2 Wasatch Uinta reference sites were not significantly
110	different when grouped by temperature categories, but O/E values at Station 4927250 (Weber)
111	were higher during hot years. These results suggest that climate change effects on O/E values
112	will vary spatially and may result in classifications improving. There are no significant
113	differences in mean O/E values at any of the sites when the data are grouped into precipitation
114	categories.
F-31

-------
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
Results of the correlation analyses show that O/E scores are only significantly correlated
with one or two of the climatic variables at 2 of the sites. The significant correlations occur at
Stations 4951200 (Virgin) and 5940440 (Beaver) (Table F5-9). At site 4951200 (Virgin), O/E
values are positively correlated with two of the PRISM mean annual air temperature variables
(sample year and prior year). At Site 5940440 (Beaver), O/E values are negatively correlated
with the prior year mean annual precipitation variable.
It should be noted that there are other potential confounding factors that may be
influencing trends in O/E values at the sites. For example, at Station 4951200 (Virgin), pH is
significantly correlated with O/E values (r=0.77, p=.03) (pH ranges from 7.95 to 8.5 at this site).
Therefore results should be interpreted with caution. See Section 2 of the report for more
information on potential confounding factors.
NOTE: Additional NMDS and/or ANOVA results for each station (4927250, 4951200, 4936750
and 5940440) are available upon request.
Table F5-9. Results of the correlation analyses between O/E values and climatic variables.
R and p-level values that are significantly correlated are in red bold print.
Station
Climatic Variables
4927250
4936750
4951200
5940440
(N=17)
(N=12)
(N=14)
(N=9)
PRISM mean annual air temperature
-0.17
p=.517
0.35
p=.265
0.70
p=.006
0.27
p=.486
PRISM mean annual precipitation
0.18
p=.496
0.00
p=.998
-0.14
p=.628
-0.23
p=.551
Previous year PRISM mean annual air
0.10
0.49
0.72
-0.25
temperature
p=.714
p=. 103
p=.004
p=.518
Previous year PRISM mean annual
0.08
0.12
0.02
-0.79
precipitation
p=.768
p=.720
p=.940
p=.012
Absolute difference between the
PRISM mean annual air temperature
-0.02
-0.38
-0.26
0.28
from the sampling year and the
previous year
p=.936
p=.225
p=.361
p=.472
Absolute difference between the
0.16
-0.03
0.27
0.00
PRISM mean annual precipitation from




the sampling year and the previous year
p=.550
p=.918
p=.348
p=.997
F-32

-------
134
135
136
137
138
139
140
141
142
143
144
F6. TREND PLOTS
Trends in two commonly-used metrics (# of total taxa and # of EPT taxa) were plotted over time
at the 4 Utah stations. Figures F6-1 through F6-8 show how trends in these metrics related to
trends in PRISM mean annual air temperature and PRISM mean annual precipitation over time.
Utah Station 4927250

¦PRISM mean annual air
temperature (°C)
PRISM mean annual
precipitation (inches)
-TotalTax
Figure F6-1. Trends in number of total taxa, PRISM mean annual air temperature and
precipitation at Station 4927250 (Weber).
F-33

-------
Utah Station 4927250
25
20
15
10
r\

LnuDr^oocno^HrMro^-Lnoo
oooooooooocncncncncncncn
cncncncncncncncncncncncn
o	m ^	ld
o	o	o o	o
o	o	o	o	o
fM	(N	IN	(N	(N
¦PRISM mean annual air
temperature (°C)
¦ PRISM mean annual
precipitation (inches)
EPTTax
Figure F6-2. Trends in number of EPT taxa, PRISM mean annual air temperature and
precipitation at Station 4927250 (Weber).
35
Utah Station 4951200
1985 1986 1987 1988 1989 1990 1991 1992 1993 1996 2000 2001 2002 2004
PRISM mean annual air
temperature (°C)
¦ PRISM mean annual
precipitation (inches)
TotalTax
149	Figure F6-3. Trends in number of total taxa, PRISM mean annual air temperature and
150	precipitation at Station 4951200 (Virgin).
F-34

-------
Utah Station 4951200
151
152
153
154
155
156
157
¦PRISM mean annual air
temperature (°C)
¦ PRISM mean annual
precipitation (inches)
EPTTax
& ^^ ^ ^ 4-^
Figure F6-4. Trends in number of EPT taxa, PRISM mean annual air temperature and
precipitation at Station 4951200 (Virgin).
F-35

-------
Utah Station 4936750
¦PRISM mean annual air
temperature (°C)
¦ PRISM mean annual
precipitation (inches)
TotalTax

198519861987198819891990199119921993199520002001
159
160
161
162
Figure F6-5. Trends in number of total taxa, PRISM mean annual air temperature and
precipitation at Station 4936750 (Duchesne).
F-36

-------
Utah Station 4936750
25
20
15
10
	*"

1985 1986 1987 19881989 1990 19911992 1993 1995 2000 2001
¦PRISM mean annual air
temperature (°C)
¦ PRISM mean annual
precipitation (inches)
EPTTax
164	Figure F6-6. Trends in number of EPT taxa, PRISM mean annual air temperature and
165	precipitation at Station 4936750 (Duchesne).
166
Utah Station 5940440
35
30
PRISM mean annual air
temperature (°C)
PRISM mean annual
precipitation (inches)
TotalTax
25
20
1996 1997 1998 2000 2001 2002 2003 2004 2005
168	Figure F6-7. Trends in number of total taxa, PRISM mean annual air temperature and
169	precipitation at Station 5940440 (Beaver).
F-37

-------
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
Utah Station 5940440
¦PRISM mean annual air
temperature (°C)
¦ PRISM mean annual
precipitation (inches)
EPTTax
1996 1997 1998 2000 2001 2002 2003 2004 2005
Figure F6-8. Trends in number of EPT taxa, PRISM mean annual air temperature and
precipitation at Station 5940440 (Beaver).
SUMMARY OF RESULTS
•	Site-specific: two sites (4927250 and 4951200) showed stronger patterns than the others
when data were grouped into temperature categories. One site, 5940440, showed no
patterns at all.
•	Temperature appears to be a more important influence than precipitation: more
significant differences occurred when samples were grouped by temperature categories
vs. precipitation categories.
•	When patterns occurred, the greatest differences were between hot- vs. cold-year
samples: in the ANOVA analyses, the greatest number of significant differences occurred
between hot and cold year samples. In the NMDS ordination, hot-year samples formed
distinct clusters from the other samples when data from sites 4927250 and 4951200 were
grouped by temperature categories.
•	Not much difference between cold and normal year samples: In the ANOVA analyses, no
significant differences occurred between cold and normal year samples, and in the
ordinations, cold and normal year samples were generally mixed together.
F-38

-------
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
•	The metrics that had the most number of significant differences between hot-year and
cold- or normal-year samples are total taxa, Ephemeroptera taxa, Plecoptera taxa, EPT
taxa, cold-water taxa and herbivore/scraper taxa.
•	Not many metrics had significant differences when grouped by precipitation categories:
only 3 metrics were significantly different when grouped by precipitation categories, and
these occurred at single sites. The 3 metrics are: % collector-gatherer, %
herbivore/scraper and # of intermittent taxa.
•	The temperature metric that performed 'best' is # of cold-water taxa. The other
temperature metrics (except for % warm-water individuals) performed fairly well at the 2
sites where patterns occurred (meaning mean values were significantly different between
hot and cold and/or normal year samples at least at one site).
•	Are certain taxa consistently driving trends? The ordinations showed that Ephemerella,
which was considered to be a cold-water taxon in these analyses, was a key player at both
sites 4927250 and 4951200. Nematoda, also considered a cold-water taxon, was present
at these two sites. Other taxa (i.e. warm-water taxa) were important at one site but not the
other (i.e. Forcipomyia/Probezzia, Microcylloepus, Caloparyphus and Chimarra)
•	'Hydrologic' metrics (i.e. perennial, intermittent, drier scenario, etc.) did not show any
significant associations with the climate-related variables.
•	Results from the ANOVA analyses on the data from the 4 sites show that there are
significant differences in O/E values between hot year and cold/normal year samples at 2
of the 4 sites (4951200 and 4936750). At both sites, mean O/E scores from the hot year
samples are significantly higher than mean O/E scores from cold and normal year
samples.
•	Results from the correlation analyses on the data from the 4 sites show that O/E scores
are only significantly correlated with one or two of the climatic variables at 2 of the sites.
Patterns are not consistent between the 2 sites.
•	See Section 6 of the report for information on potential confounding factors that may
have influenced trends at these sites.
F-39

-------
Attachment F1
'Extreme' alterations of Utah fall RIVPACS
model climate-related predictor variable values
As referenced on page F-7 in this Appendix, we also ran some 'extreme' scenarios (i.e. doubling
temperature, dividing precipitation values by two, changing freeze dates by 30 days, etc.) to
further explore how much the climate-related predictor variables would have to change in order
to result in substantial changes to O/E scores. The tables in this attachment show which
scenarios were run and what the results were.
Fl-1

-------
Table Fl-1. Descriptions of how the climate-related predictor variables were altered in the 'extreme alteration' RIVPACS
analyses
Run#
Category
Altered Predictor variables
Rationale
1
Baseline
None - used original values
get baseline values and QC
2

TMEAN.WS + 2 & TMEAN.NET + 2
NCAR annual temperature predictions
(2050)
3

TMEAN.WS + 4 & TMEAN.NET + 4
NCAR annual temperature predictions
(2090)
4

TMEAN.WS + 10 & TMEAN.NET + 10
curiosity
5
Temperature
TMEAN.WS + 20 & TMEAN.NET + 20
curiosity
6

MEANP.PT - .05
NCAR annual precipitation predictions
(2050)
7
8

MEANP.PT -. 1
MEANP.PT - Minimum PRISM pptl4
NCAR annual precipitation predictions
(2090)
based on PRISM pptl4 minimum values
(1975-2006)
9

MEANP.PT/2
curiosity
10

MINP.PT/2
curiosity
11

MEANP.PT/2 & MINP.PT/2
curiosity
12
Precipitation

curiosity
13

TMEAN.WS + 2 & TMEAN.NET + 2 & MEANP.PT - .05
NCAR annual temperature and
precipitation predictions (2050)
14
Temperature &
Precipitation
TMEAN.WS + 4 & TMEAN.NET + 4 & MEANP.PT - .1
NCAR annual temperature and
precipitation predictions (2090)
15

LST32AVE - 2
best professional judgment
16

LST32AVE - 5
best professional judgment
17

FST32AVE + 5
best professional judgment
18
Freeze Date
LST32AVE - 5 & FST32AVE + 5
best professional judgment
19
LST32AVE - 10
curiosity
20MINWD.WS/2
FST32AVE + 10
curiosity
21

LST32AVE - 10 & FST32AVE + 10
curiosity
22

LST32AVE - 15
curiosity

-------
Table Fl-1.
Continued.


Run#
Category
Altered Predictor variables
Rationale
23
Freeze Date
LST32AVE - 15 & FST32AVE + 15
curiosity
24
25
Combine All
LST32AVE-1, MINP.PT-1, MEANP.PT-1, TMEAN.NET+1,
TMEAN.WS+1, FST32AVE+1, MINWD.WS-1
LST32AVE-2, MINP.PT-2, MEANP.PT-2, TMEAN.NET+2,
TMEAN.WS+2, FST32AVE+2, MINWD.WS-1
best professional judgment
best professional judgment

-------
17	Table Fl-2. Results for the scenarios in which temperature predictor variables were
18	altered

Baseline (original)
TMEAN.WS + 2 &
TMEAN.NET + 2

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
14
14.92
0.94
0.01
7
4951200
120184
10
9.58
1.04
10
9.56
1.05
0
1
4936750
118524
15
14.04
1.07
15
14
1.07
0
6
4927250
127718
8
8.74
0.92
8
8.74
0.92
0





Baseline (original)
TMEAN.WS + 4 &
TMEAN.NET + 4

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
14
14.8
0.95
0.02
7
4951200
120184
10
9.58
1.04
10
9.6
1.04
0
1
4936750
118524
15
14.04
1.07
15
14
1.07
0
6
4927250
127718
8
8.74
0.92
7
8.25
0.85
-0.07





Baseline (original)
TMEAN.WS +10 &
TMEAN.NET +10

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
14
14.65
0.96
0.03
7
4951200
120184
10
9.58
1.04
10
9.61
1.04
0
1
4936750
118524
15
14.04
1.07
15
13.89
1.08
0.01
6
4927250
127718
8
8.74
0.92
7
8.24
0.85
-0.07





Baseline (original)
TMEAN.WS + 20 &
TMEAN.NET + 20

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
13
14.08
0.92
0
7
4951200
120184
10
9.58
1.04
10
9.63
1.04
-0.01
1
4936750
118524
15
14.04
1.07
15
13.44
1.12
0.05
6
4927250
127718
8
8.74
0.92
7
8.24
0.85
-0.07
19
Fl-4

-------
21	Table Fl-3. Results for the scenarios in which precipitation predictor variables were
22	altered

Baseline (original)

MEANP.PT
-.05

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
14
15.1
0.93
0
7
4951200
120184
10
9.58
1.04
10
9.59
1.04
0
1
4936750
118524
15
14.04
1.07
15
14
1.07
0
6
4927250
127718
8
8.74
0.92
8
8.75
0.91
0





Baseline (original)

MEANP.PT
-.1

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
14
15.08
0.93
0
7
4951200
120184
10
9.58
1.04
10
9.58
1.04
0
1
4936750
118524
15
14.04
1.07
15
14.01
1.07
0
6
4927250
127718
8
8.74
0.92
8
8.74
0.92
0





Baseline (original)
MEANP.PT - Min
ppt!4 PRISM

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Difce
O.E
1
5940440
127636
14
15.09
0.93
14
14.78
0.95
0.02
7
4951200
120184
10
9.58
1.04
10
9.51
1.05
0.01
1
4936750
118524
15
14.04
1.07
15
13.79
1.09
0.02
6
4927250
127718
8
8.74
0.92
8
8.71
0.92
0





Baseline (original)
MLANP.PT/2

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Difce
O.E
1
5940440
127636
14
15.09
0.93
14
14.79
0.95
0.02
7
4951200
120184
10
9.58
1.04
10
9.43
1.06
0.02
1
4936750
118524
15
14.04
1.07
15
13.8
1.09
0.02
6
4927250
127718
8
8.74
0.92
8
8.68
0.92
0.01





Baseline (original)
MINP.PT/2

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Difce
O.E
1
5940440
127636
14
15.09
0.93
13
13.92
0.93
0.01
7
4951200
120184
10
9.58
1.04
10
9.46
1.06
0.01
1
4936750
118524
15
14.04
1.07
15
13.58
1.1
0.04
6
4927250
127718
8
8.74
0.92
8
8.69
0.92
0.01
Fl-5

-------
24 Table Fl-3. Continued

Baseline (original)
MEANP.PT/2 &
MINP.PT/2

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
13
13.69
0.95
0.02
7
4951200
120184
10
9.58
1.04
10
9.33
1.07
0.03
1
4936750
118524
15
14.04
1.07
15
13.38
1.12
0.05
6
4927250
12 IS
8
8.74
n Q2
8
8.16
0.98
0.07





Baseline (original)
MINWD.WS/2

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
13
13.81
0.94
0.01
7
4951200
120184
10
9.58
1.04
10
9.53
1.05
0.01
1
4936750
118524
15
14.04
1.07
15
13.47
1.11
0.05
6
4927250
127718
8
8.74
0.92
7
7.63
0.92
0
25
26
27
28
Table Fl-4. Results for the scenarios in which both temperature and precipitation






TMEAN.WS + 2 &




Baseline (original)
TMEAN.NET + 2 &








MEANP.PT -
.05

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Difce
O.E
1
5940440
127636
14
15.09
0.93
14
14.93
0.94
0.01
7
4951200
120184
10
9.58
1.04
10
9.56
1.05
0
1
4936750
118524
15
14.04
1.07
15
14.01
1.07
0
6
4927250
127718
8
8.74
0.92
7
8.24
0.85
-0.07










TMEAN.WS + 4 &




Baseline (original)
TMEAN.NET + 4 &








MEANP.PT
-.1

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Difce
O.E
1
5940440
127636
14
15.09
0.93
14
14.83
0.94
0.02
7
4951200
120184
10
9.58
1.04
10
9.58
1.04
0
1
4936750
118524
15
14.04
1.07
15
14.02
1.07
0
6
4927250
127718
8
8.74
0.92
7
8.26
0.85
-0.07
29
Fl-6

-------
Table Fl-5. Results for the scenarios in which freeze date predictor variab

Baseline (original)

LST32AVE
-2

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
14
15.05
0.93
0
7
4951200
120184
10
9.58
1.04
10
9.58
1.04
0
1
4936750
118524
15
14.04
1.07
15
14.01
1.07
0
<".
492"250
12 IS
8
8 "4
0 92
"
8 25
0 85
-o rr





Baseline (original)

LST32AVE
-5

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
14
14.733
0.95
0.02
7
4951200
120184
10
9.58
1.04
10
9.5648
1.05
0
1
4936750
118524
15
14.04
1.07
15
13.999
1.07
0
6
4927250
12^18
8
8^4
0.92
i
8 24^
0.85
-0.07





Baseline (original)
FST32AVE + 5

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Difce
O.E
1
5940440
127636
14
15.09
0.93
15
15.374
0.98
0.05
7
4951200
120184
10
9.58
1.04
10
9.5875
1.04
0
1
4936750
118524
15
14.04
1.07
15
14.028
1.07
0
6
4927250
127718
8
8.74
0.92
8
8.7184
0.92
0





Baseline (original)
LST32AVE - 5 &
FST32AVE + 5

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Difce
O.E
1
5940440
127636
14
15.09
0.93
13
14.128
0.92
-0.01
7
4951200
120184
10
9.58
1.04
10
9.5647
1.05
0
1
4936750
118524
15
14.04
1.07
15
13.992
1.07
0
6
4927250
127718
8
8.74
0.92
7
8.224
0.85
-0.06





Baseline (original)

LST32AVE
-10

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Difce
O.E
1
5940440
127636
14
15.09
0.93
13
14.02
0.93
0
7
4951200
120184
10
9.58
1.04
10
9.56
1.05
0
1
4936750
118524
15
14.04
1.07
15
13.7
1.09
0.03
6
4927250
127718
8
8.74
0.92
7
8.23
0.85
-0.07
es were altered
Fl-7

-------
34 Table Fl-5. Continued

Baseline (original)
FST32AVE +10

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
14
14.713
0.95
0.02
7
4951200
120184
10
9.58
1.04
10
9.6097
1.04
0
1
4936750
118524
15
14.04
1.07
15
13.797
1.09
0.02
(>
492"250
12—IS
S
X "4
0 92
-
X IS4'
0 X(i
-I) ()(>





Baseline (original)
LST32AVE - 10 &
FST32AVE +10

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
13
13.743
0.95
0.02
7
4951200
120184
10
9.58
1.04
10
9.6115
1.04
0
1
4936750
118524
15
14.04
1.07
15
13.532
1.11
0.04
6
4927250
127718
8
8.74
0.92
7
8.1706
0.86
-0.06





Baseline (original)

LST32AVE
- 15

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
13
13.945
0.93
0
7
4951200
120184
10
9.58
1.04
10
9.5818
1.04
0
1
4936750
118524
15
14.04
1.07
15
13.454
1.11
0.05
6
4927250
127718
8
8.74
0.92
7
8.2214
0.85
-0.06





Baseline (original)
LST32AVE - 15 &
FST32AVE +15

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
13
13.415
0.97
0.04
7
4951200
120184
10
9.58
1.04
10
9.6052
1.04
0
1
4936750
118524
15
14.04
1.07
14
12.787
1.09
0.03
6
4927250
127718
8
8.74
0.92
7
8.1713
0.86
-0.06
35
Fl-8

-------
37
38
Table Fl-6. Results for scenarios in which combinations of all climate-related predictor

Baseline (original)
Changed by 1

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
13
14.04
0.93
0
7
4951200
120184
10
9.58
1.04
10
9.51
1.05
0.01
1
4936750
118524
15
14.04
1.07
15
14.03
1.07
0
6
4927250
127718
8
8.74
0.92
8
8.71
0.92
0





Baseline (original)
Changed by 2

GROUP
SITE
SAMPLE
O
E
O.E
O
E
O.E
Dif'ce
O.E
1
5940440
127636
14
15.09
0.93
13
13.81
0.94
0.01
7
4951200
120184
10
9.58
1.04
10
9.49
1.05
0.01
1
4936750
118524
15
14.04
1.07
15
14.03
1.07
0
6
4927250
127718
8
8.74
0.92
7
8.23
0.85
-0.06
39
Fl-9

-------
, Attachment F2
2
3
4
s Utah Temperature-Indicator Taxa
6
7
8	This attachment contains tables with lists of the Utah temperature-indicator taxa and describes
9	the process that we followed to develop these lists.
F2-1

-------
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
F2. UTAH TEMPERATURE-INDICATOR TAXA
Sources. The Utah cold- and warm-water taxa lists were developed using several
different sources: 1. weighted-average calculations based on a subset of the Utah biomonitoring
database (using fall samples (sample size=572)); 2. the thermal-preference trait from the Poff et
al. 2006 traits matrix; 3. the thermal-preference trait from the USGS traits database (Vieira et al.,
2006); 4. the thermal-preference trait from the compilation of EPA Environmental Requirements
and Pollution Tolerance series from the late 1970's (Beck et al., 1977; Harris et al., 1978;
Hubbard et al., 1978; Surdick et al., 1978); and 5. best professional judgment of the Utah
Climate Change feedback group3.
Many of the same general criteria that were used to designate cold- and warm-water
indicator taxa in Maine were also used in Utah (see Attachment F2). Also, see Attachment F2
for general limitations of the weighted averaging, as well as for information on general
considerations that were taken into account.
Initial Results. Initially there were 76 taxa on the cold-water list and 53 taxa on the
warm-water list. These lists were based on weighted-average calculations and literature. These
lists were further refined through the evaluation of additional evidence. This evidence included
analyses of other datasets, case studies, and best professional judgment. Taxa with the greatest
amount of evidence were designated as temperature indicator taxa. More detailed information
about the steps that were used to develop the temperature indicator taxa lists is summarized
below:
Considerations (unique to Utah)
In addition to Considerations A and C in Attachment E2, a subset of the scores that
included only the western states (California, Oregon, Idaho, Utah, Yuan Western EMAP) was
also taken into account when developing the lists. The reasoning behind this is that the data from
these states is more similar and therefore more comparable to Utah than data from Ohio, North
Carolina and Maine. Therefore it was given more weight in the consideration process. Taxa that
3 Utah Climate Change group: Utah DWQ (Jeff Ostermiller), Utah State University Bug Lab (Mark Vinson), Eric
Dinger (formerly of the USU Bug Lab), David Herbst (California Sierra Nevada), Wyoming (Eric Hargett), Pyramid
Lake Paiute Tribe (Dan Mosely), and Shann Stringer (formerly New Mexico).
F2-2

-------
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
received higher scores had more evidence supporting their inclusion on the temperature indicator
taxa list. Cold- and warm-water taxa lists from these western states were also evaluated for
conflicting evidence. If a taxon showed a preference for cold- or warm-water in Utah but was
shown to have the opposite preference in the California, Oregon, Idaho, or Yuan Western EMAP
analyses (i.e. cold-water taxon in Utah was listed as a warm-water taxon in Oregon), it was not
included on the temperature indicator list.
Several 'case studies' were performed to see whether the cold- or warm-water taxa
occurred at sites in Utah that had the warmest- or coldest-water temperatures (June-September).
The following case studies were performed:
Cold-water Case Study #1. Taxa lists from 4 sites in the Wasatch and Uinta Mountains
level 3 ecoregion that had the coldest average water temperatures (using June-September
samples) and that had <2% urban and <10% agricultural land use/land cover within a 1 km
buffer were evaluated. Sites include: Station 4938910 (avg temp 5.75°C, 0% urban, 0%
agricultural), Station 4936700 (avg temp 9.1°C, 0.74% urban, 0% agricultural), Station 4935970
(avg temp 9.5°C, 0% urban, 0% agricultural), and Station 4995830 (avg temp 9.6°C, 0% urban,
0% agricultural).
Cold-water Case Study #2. Taxa lists from 4 sites in the Colorado Plateaus level 3
ecoregion that had the coldest average water temperatures (using June-September samples) and
that had <2% urban and <10% agricultural land use/land cover within a 1 km buffer were
evaluated. Sites include: Station 4937720 (avg temp 10.9°C, 0.17% urban, 1.4% agricultural),
Station 4936200 (avg temp 12.5°C, 0.11% urban, 0% agricultural), Station 4954140 (avg temp
14.1°C, 0% urban, 0% agricultural), and Station 4956480 (avg temp 14.2°C, 0% urban, 0%
agricultural).
Warm-water Case Study #1. Taxa lists from two sites in the Colorado Plateaus level 3
ecoregion that had the warmest average water temperatures (using June-September samples) and
that had <2% urban and <10% agricultural land use/land cover within a 1 km buffer were
evaluated. Sites include: Station 4933120 (avg temp 32°C, 1.6% urban, 3/4% agricultural) and
Station 4950790 (avg temp 26.2°C, 0% urban, 0% agricultural).
Development of the Temperature Indicator Cold-water Taxa List. Taxa were placed
on the cold-water list if the following criteria were met:
F2-3

-------
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
1.	The taxon was NOT present at the warm-water case study site.
2.	The taxon (and no species within the genera) was NOT on the warm-water
lists derived from the California, Oregon, Idaho, and Yuan Western EMAP datasets.
3.	The Utah Climate Change feedback group did not specify that they did not
think the taxon should be on the list (based on best professional judgment).
4.	The taxon had to be on the cold-water taxa list in at least two of the
western datasets, or if it was only listed in one dataset, it also had to be present at one or
more of the cold-water case study sites.
Development of the Temperature-Indicator Warm-Water List. Taxa were placed on
the warm-water list if the following criteria were met:
1.	The taxon was NOT present at the cold-water case study sites.
2.	The taxon (and no species within the genera) was NOT on the cold-water
lists derived from the California, Oregon, Idaho, and Yuan Western EMAP datasets.
3.	The Utah Climate Change feedback group did not specify that they did not
think the taxon should be on the list (based on best professional judgment).
4.	The taxon had to be on the warm-water taxa list in at least two of the
western datasets, or if it was only listed in one dataset, it also had to be present at the
warm-water case study sites.
Temperature-Indicator Lists. The cold-water taxa list was comprised of 33 taxa and
the warm-water taxa list was comprised of 16 taxa. Temperature indicator taxa lists can be found
in Tables F2-1 and F2-2.
Important Notes - variation within genera. Some noteworthy genera were left off the
Utah cold-water taxa list. These include Zapada, Epeorus, Drunella, Brachycentrus and
Rhyacophila. The reason they were not included is because there is variation in temperature
preferences among species within these genera. For example, Zapada cinctipes is on the warm-
water taxa lists in the Oregon and Idaho datasets, but the other species within this genus are
listed as cold-water taxa. Epeorus albertae is on the warm-water list in the Oregon dataset, but
other species within this genus are generally listed as cold-water taxa. Drunella grandis is listed
as a warm-water taxa (barely - it received a rank optima score of 5) in the Oregon and Idaho
datasets, but other species within this genus are generally listed as cold-water taxa. Within the
F2-4

-------
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
family Rhyacophilidae, there are a few taxa that are listed as warm-water taxa and several that
are listed as cold-water taxa. There is similar variation within the genus Brachycentrus.
Dispersal Ability. If temperature is a major factor influencing community composition,
then taxa that are able to adapt to warming temperatures and/or that are able to disperse to more
favorable habitats (generally believed to be upstream or to higher elevations) have a better
chance of surviving. Five mobility traits were examined for the taxa on the Utah temperature
indicator lists: dispersal (adult), adult flying strength, occurrence in drift, maximum crawling rate
and swimming ability. More information on these traits can be found in Table F2-3.
Dispersal (adult) and adult flying strength received the greatest amount of consideration.
Because movement is most likely to be upstream, taxa that are strong fliers are likely to have a
better chance of success. It will be difficult for taxa that disperse via occurrence in drift to
migrate upstream, and taxa that disperse via crawling or swimming are likely to have difficulty
moving the distances required to find more favorable habitats.
All of the taxa on the Utah temperature indicator cold-water taxa list (that we had trait
information for) are considered to have low dispersal ability and weak adult flying strength. Six
of the taxa on the temperature indicator warm-water taxa list (that we had trait information for)
are considered to have high dispersal ability (Cheumatopsyche, Microcylloepus, Ochrotrichia,
Oecetis, Calineuria and Nectopsyche). Two of these are categorized as strong fliers
(Cheumatopsyche and Calineuria).
Abundance and Distribution. In addition to dispersal ability, abundance and
distribution are also important considerations. Those taxa that are widespread and common are
likely to have greater genetic diversity and greater chance of adapting than rare taxa that only
occur in isolated, localized populations (Sweeney et al., 1992). Moreover, the more abundant
taxa are more likely to affect the state biomonitoring assessments.
Abundance and distribution information for the temperature-indicator taxa can be found
in Tables F2-1 and F2-2. The most abundant cold-water-temperature-indicator taxa are two
Ephemeropterans, Ephemerella and Cinygmula, which comprise 1.85 and 1.03 percent of the
total individuals, respectively. Twenty of the cold-water taxa have overall abundances of less
than 0.1%. Asellidae and Leptohyphidae are the most abundant warm-water taxa, with overall
abundances of 3.12 and 1.42%. Eleven of the warm-water taxa have overall abundances of less
F2-5

-------
127
128
129
130
131
132
133
134
135
136
137
138
139
140
than 0.1%. Of the cold-water taxa, Chloroperlidae occurs at the highest percentage of sites
(49%), followed by two Ephemeropterans (Ephemerella and Cinygmula), which occur at 44 and
46%) of the sites, respectively. Fifteen of the cold-water taxa occur at less than 10%> of the sites.
Among the warm-water taxa, Leptohyphidae occurs at the highest percentage of sites (31%>),
followed by Coenagrionidae (18%>) and Cheumatopsyche (17%>). Eleven of the warm-water taxa
occur at less than 10%> of the sites.
Additional information - Cold-water Taxa. Ten of the cold-water taxa are
Plecopterans, eight are Dipterans, seven are Trichopterans and six are Ephemeropterans (Table
F2-4a) The families with the most number of taxa on the cold-water list are Heptageniidae,
Empididae and Perlodidae (Table F2-4b).
Additional information - Warm-water Taxa. Five of the warm-water taxa are
Trichopterans, three are Coleopterans, and two are Dipterans and Ephemeropterans (Table F2-
5a). The families with the most number of taxa on the warm-water list are Elmidae and
Leptoceridae (Table F2-5b).
F2-6

-------
141	Table F2-1. List of Utah cold-water temperature indicator taxa. Distribution and abundance information is also included. Sum_Individuals=the
142	total number of individuals from that taxon in the Utah database; Pct_Abund=percent of total individuals in the database comprised of that
143	taxon; Num_Stations=number of stations in the database that the taxon occurred at; Pct_Stations=percent of stations in the database at which the
144	taxon occurred.
Type
Order
Family
FinallD
Sum Individs
Pet Abund
Num Stations
Pet Stations
cold
Ephemeroptera
Ameletidae
Ameletus
13157.6
0.03
137
21.57
cold
Trichoptera
Glossosomatidae
Anagapetus
42
0
2
0.31
cold
Trichoptera
Apataniidae
Apatania
20154.3
0.04
39
6.14
cold
Diptera
Ceratopogonidae
Bezzia
109267.1
0.23
232
36.54
cold
Diptera
Blephariceridae
Bibiocephala
2257
0
15
2.36
cold
Plecoptera
Capniidae
Capniidae
113578.8
0.24
228
35.91
cold
Diptera
Empididae
Chelifera
94014.1
0.2
261
41.1
cold
Plecoptera
Chloroperlidae
Chloroperlidae
203579.9
0.44
309
48.66
cold
Ephemeroptera
Heptageniidae
Cinygma
606.2
0
6
0.94
cold
Ephemeroptera
Heptageniidae
Cinygmula
479866.5
1.03
278
43.78
cold
Plecoptera
Perlodidae
Cultus
20419.7
0.04
97
15.28
cold
Diptera
Tipulidae
Dicranota
35439.2
0.08
220
34.65
cold
Trichoptera
Limnephilidae
Ecclisomyia
1262.8
0
14
2.2
cold
Ephemeroptera
Ephemerellidae
Ephemerella
859335.8
1.85
292
45.98
cold
Plecoptera
Pelecorhynchidae
Glutops
91
0
4
0.63
cold
Coleoptera
Elmidae
Heterlimnius
16463
0.04
50
7.87
cold
Ephemeroptera
Heptageniidae
Ironodes
551.6
0
6
0.94
cold
Plecoptera
Perlodidae
Kogotus
1288.7
0
14
2.2
cold
Trichoptera
Lepidostomatidae
Lepidostoma
353679.8
0.76
240
37.8
cold
Plecoptera
Leuctridae
Leuctridae
21176.5
0.05
106
16.69
cold
Plecoptera
Perlodidae
Megarcys
7129.9
0.02
65
10.24
cold
Dorylaimida
Dorylaimidae
Nematoda
141425.3
0.3
249
39.21
cold
Trichoptera
Uenoidae
Neothremma
129853.8
0.28
100
15.75
cold
Trichoptera
Uenoidae
Oligophlebodes
147256.9
0.32
101
15.91
cold
Diptera
Empididae
Oreogeton
228.5
0
13
2.05
F2-7

-------
146 Table F2-1. Continued
Type
Order
Family
FinallD
Sumlndivids
PctAbund
Num_Stations
PctStations
cold
Trichoptera
Hydropsychidae
Parapsyche
3552.5
0.01
40
6.3
cold
Diptera
Psychodidae
Pericoma
145582.7
0.31
210
33.07
cold
Diptera
Tipulidae
Rhabdomastix
8
0
1
0.16
cold
Ephemeroptera
Heptageniidae
Rhithrogena
198501.8
0.43
243
38.27
cold
Plecoptera
Taeniopterygidae
Taenionema
79949.8
0.17
87
13.7
cold
Plecoptera
Nemouridae
Visoka
50
0
1
0.16
cold
Diptera
Empididae
Wiedemannia
458
0
13
2.05
cold
Plecoptera
Peltoperlidae
Yoraperla
72.7
0
5
0.79
147
148	Table F2-2. List of Utah warm-water temperature indicator taxa. Distribution and abundance information is also included. Sum_Individuals=the
149	total number of individuals from that taxon in the Utah database; Pct_Abund=percent of total individuals in the database comprised of that taxon;
150	Num_Stations=number of stations in the database that the taxon occurred at; Pct_Stations=percent of stations in the database at which the taxon
occurred.
Type
Order
Family
FinallD
Sum Individs
Pet Abund
Num Stations
Pet Stations
warm
Hemiptera
Naucoridae
Ambrysus
25879.7
0.06
39
6.14
warm
Isopoda
Asellidae
Asellidae
1450840.4
3.12
81
12.76
warm
Ephemeroptera
Caenidae
Caenis
567
0
11
1.73
warm
Plecoptera
Perlidae
Calineuria
245
0
9
1.42
warm
Diptera
Stratiomyidae
Caloparyphus
9652
0.02
26
4.09
warm
Trichoptera
Hydropsychidae
Cheumatopsyche
172233.9
0.37
105
16.54
warm
Odonata
Coenagrionidae
Coenagrionidae
45144.1
0.1
117
18.43
warm
Ephemeroptera
Leptohyphidae
Leptohyphidae
659670.3
1.42
197
31.02
warm
Diptera
Psychodidae
Maruina
1140.2
0
16
2.52
warm
Coleoptera
Elmidae
Microcylloepus
114016
0.24
50
7.87
warm
Trichoptera
Leptoceridae
Nectopsyche
8434.7
0.02
35
5.51
warm
Trichoptera
Hydroptilidae
Ochrotrichia
6768.2
0.01
29
4.57
warm
Trichoptera
Leptoceridae
Oecetis
28993.3
0.06
90
14.17
warm
Coleoptera
Elmidae
Ordobrevia
360
0
5
0.79
warm
Coleoptera
Psephenidae
Psephenus
65.8
0
4
0.63
warm
Trichoptera
Psychomyiidae
Tinodes
12774.6
0.03
34
5.35
F2-8

-------
Table F2-3. Mobility traits that were evaluated. The source of most of this information
was the Poff et al. 2006 traits matrix. Some also came from the USGS traits database
(Vieira et al., 2006).
Mobility Trait	Trait States
low (<1 km flight before laying eggs), high (>1 km flight before
laying eggs)
weak (e.g. cannot fly into light breeze), strong
rare (catastrophic only), common (typically observed), abundant
(dominant in drift samples)
very low (<10 cm/h), low (<100 cm/h), high (>100 cm/h)
none, weak, strong
Table F2-4a. Number of cold-water taxa in each order
Order	Total
Plecoptera
10
Diptera
8
Trichoptera
7
Ephemeroptera
6
Coleoptera
1
Dorylaimida
1
Table F2-4b. Number of cold-water taxa in each family
Family	Total
Heptageniidae	4
Empididae	3
Perlodidae	3
Tipulidae	2
Uenoidae	2
Ameletidae	1
Apataniidae	1
Blephariceridae	1
Capniidae	1
Ceratopogonidae	1
Chloroperlidae	1
Dorylaimidae	1
Elmidae	1
Ephemerellidae	1
Glossosomatidae	1
Hydropsychidae	1
Lepidostomatidae	1
Leuctridae	1
Limnephilidae	1
Dispersal (adult)
Adult flying strength
Occurrence in drift
Maximum crawling rate
Swimming ability
F2-9

-------
Table F2-4b. Continued
Family	Total
Nemouridae	1
Pelecorhynchidae
Peltoperlidae	1
Psychodidae	1
Taeniopterygidae	1
1
Table F2-5a. Number of warm-water taxa in each order
Order	Total
Trichoptera
5
Coleoptera
3
Diptera
2
Ephemeroptera
2
Hemiptera
1
Isopoda
1
Odonata
1
Plecoptera
1
Table F2-5b. Number of
Family
Total
Elmidae
2
Leptoceridae
2
Asellidae
1
Caenidae
1
Coenagrionidae
1
Hydropsychidae
1
Hydroptilidae
1
Leptohyphidae
1
Naucoridae
1
Perlidae
1
Psephenidae
1
Psychodidae
1
Psychomyiidae
1
Stratiomyidae
1
i-water taxa in each family
F2-10

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
APPENDIX G
Detailed Results North Carolina
The intent of this appendix is to provide more comprehensive and detailed information on the
large number of analyses that were performed on the North Carolina data. Some of the analyses
that are covered in this appendix are also referenced in the main body of the APM report. When
this occurred, attempts were made to reduce any overlap or duplication in the reporting of
results.
G1. Overview
G2. North Carolina Ecoregion Descriptions
G3. Results
Attachment G1. Temperature Indicator Taxa - North Carolina
Attachment G2. Tolerance values of the cold and warm-water temperature
indicator taxa
G-l

-------
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Gl. Overview
North Carolina's Biological Assessment Unit rates sites as Excellent (5), Good (4),
Good/Fair (3), Fair (2) or Poor (1). Historically bioclassifications had been assigned based on
EPT richness alone or in combination with total taxa richness. However, interpretations were
sometimes troublesome because criteria often needed to be adjusted to account for differences in
factors like collection method, stream size, seasonal changes and ecoregion, so the North
Carolina Biotic Index (NCBI) was developed as an independent method of rating water quality.
The NCBI uses tolerance values that are derived from the NC database. The NCBI is a summary
measure of the tolerance values of organisms found in the sample relative to their abundance (see
NC Standard Operating Procedures (SOP) 2006 for more details). NCBI scores, which range
from 0 (best) to 10 (worst), are calculated for samples collected using the standard qualitative
(full-scale) or EPT collection methods.
For most sites, equal weight is given to both the NCBI value and EPT taxa richness when
assigning bioclassifications. Exceptions are outlined in the NC SOP (2006) and include such
things as pristine high altitude mountain streams, swamp streams, and Coastal B streams (see NC
SOP 2006 for more details). Under normal circumstances NCBI and EPT taxa richness measures
are averaged together. A rounding approach is used when the two scores differ by one
bioclassification and produce a final score midway between two ratings (1.5, 2.5, 3.5 or 4.5). In
this situation EPT abundance is taken into account when deciding whether to round up or round
down. Abundance of organisms is recorded as rare=l (1-2 specimens), common=3 (3-9
specimens) or abundant (>10 specimens). Scoring criteria are outlined in Figure G-l (see NC
SOP 2006 for more details). Figure G-l also shows that bioclassification criteria have been
developed for three major ecoregions (as defined by NCDENR): Mountain (MT), Piedmont (P)
and Coastal Plain (CA).
G-2

-------
Score
Mt
B1 Values
P
CA
EPT Values
MT P
CA
5
<4.00'
<5.14
<5.42
>43
>33
>29
4.6
4,00-4.04
5,14-5.18
5.42-5.46
42-43
•Ji- J O
28
4.4
4.05-4.09
5.19-5.23
5.47-5,51
40-41
30-31
27
4
4.10-4,83
5,24-5.73
5.52-6.00
34-39
26-29
22-26
3,®
4.84-4,88
5,74-5.78
6.01-6,05
32-33
24-25
21
3.4
4 89-4.93
5.79-5.83
6.06-6,10
30-31
22-23
20
3
4.94-5,89
5.84-8.43
6.11-6.67
24-29
18-21
15-19
2.6
5 70-5.74
6.44-6.48
6.68-6.72
OOOQ
jCnmL mL*
16-17
14
2.4
5 75-5.79
8.49-6.53
8.73 6 77
20-21
14-15
13
2
5.80-6,95
6.54-7,43
6.78-7,68
14-19
10-13
8-12
1.6
8.96-7.00
7.44*7.48
7.69-7.73
12-13
8-9
7
1.4
7.01-7.05
7.49-7.53
7.74-7.79
10-11
6-7
6
1
>7.05
>7,53
>7.79
0-9
0-5
0-5
Biotic
Index corrections for non-summer data:



Summer = Jun-Sep, Fall
= Oct-Nov, Winter
= Oec-Feb, Spring = Mar-May




Fall
Winter
goring


Mountain Correction
+0.4
+0,5
+0.5


Piedmont Correction
+0.1
+0.1
+0.2


Coastal A Correction
+0,2
+0,2
+0.3


Rounding Criteria: Round down if EFT N < criterion, otherwise round up.
Biootossifieatiort /Score)	MT	P			CA
Excellent (5) vs. Good (4)	191 135	108
Good(4) vs. Good-Fair {3)	125 103	91
Good-Fair (3) vs, Fair (2) 85 . 71	46
Fair (2) vs. Poor (1) 45 38	18
46
47	Figure G-l. These tables are used to determine the scores for EPT taxa richness values and
48	NCBI values for all standard qualitative samples after seasonal corrections are made. EPT
49	N refers to EPT abundance (NC SOP 2006).
50
51
52	G2 North Carolina Ecoregion Description
53
54	The major ecoregions defined by NCDENR differ slightly from EPA Level 3 ecoregions.
55	Sites in the NCDENR Mountain ecoregion generally fall within the Blue Ridge EPA Level 3
56	ecoregion, which runs along the western portion of the state. Sites in the NCDENR Piedmont
57	ecoregion are generally in the Piedmont EPA Level 3 ecoregion, which runs through the central
58	portion of North Carolina. The NCDENR Coastal ecoregion generally overlaps with the
59	Southeastern Plains and Middle Atlantic Coastal Plain EPA Level 3 ecoregions, which are
60	located in the eastern portion of the state.
G-3

-------
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
The major ecoregions are quite different. Terrain in the Mountain ecoregion ranges from
narrow ridges to hilly plateaus to more massive mountainous areas with high peaks. Elevations
generally range from 305-1524 meters, with Mount Mitchell, the highest point in North Carolina,
and highest in the U.S. east of the Mississippi River, reaching 2037 meters. There is a high
diversity of flora and fauna with high gradient, cool, clear streams with rocks and boulders.
Forest-related land uses occur along with some small areas of pasture, apple orchards, and
Christmas tree farms. Low-density recreational activities in forested settings have also become a
typical land-use. (Griffith et al., 2002)
The Piedmont ecoregion is a transitional area between the mostly mountainous ecological
regions of the Appalachians and the relatively flat coastal plain. Several major land cover
transformations have occurred in the Piedmont over the past 200 years, from forest to farm, back
to forest, and now in many areas, spreading urban- and suburbanization. Once largely cultivated
with crops such as cotton, corn, tobacco and wheat, most of the Piedmont soils were moderately
to severely eroded (Trimble, 1974). Much of this region is now in planted pine or has reverted to
successional pine and hardwood woodlands with some pasture in the land cover mosaic (Griffith,
et al. 2002).
The Coastal ecoregion consists of low elevation, flat plains, with many swamps, marshes,
and estuaries. Pine plantations for pulpwood and lumber are typical with some areas of cropland.
In some areas there is a mix of cropland, pasture, woodland, and forest. Over the past three
centuries, naval stores or pine tar production, logging, open range cattle and feral hog grazing,
agriculture, and fire suppression removed almost all of the longleaf pine forests. Streams in this
area are relatively low-gradient and sandy-bottomed (Griffith et al., 2002).
More biological sampling sites are located in the Mountain and Piedmont ecoregions
(1185 and 1007, respectively) than in the Coastal ecoregion (365). As expected, average
elevations of the Mountain sites are much higher than the other sites (637 meters in the
mountains vs. 155 meters in the Piedmont and 37 meters in the Coastal Plain).
G3 Results
G3.1 Distributions of temperature-indicator taxa
G-4

-------
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
Results are summarized in Section 2 of the report. We used PRISM mean annual
precipitation as a surrogate for flow at sites that did not have USGS gages. In the Figure G-2
example, PRISM mean annual precipitation tracked fairly well with mean monthly July-August
discharge (which corresponded to the collection period for benthic samples that we examined in
our analyses)
350
BOO
250
200
-PRISM mean annual
precipitation (inches)
Mean Monthly Discharge
(July & August) (cfs)
150
100
LntDrvooaiOH(Mm^miDNOO!JiOH(Mm^i/iu3
coajMcocoaicncnoioiaicriaioioiooooooo
cno^cncjioicnaiaicncncji^cnoicnooooooo
I *—I t—I tH *—I rH rH rH *—I rH rH *—I I t-H I CSl fNJ <~\| C\l C\1 C\l CSl
Year
Figure G-2. Relationship between PRISM mean annual precipitation and mean monthly
discharge over time at NC0207 (Nantahala River).
It was difficult to examine the relationship between flow and biology at most individual
sites due to lack of flow data or discontinuities in the biological data at sites with USGS flow
gages (Figure G-3). However, data from NC0109 (New River) showed significant relationships
between PRISM mean annual precipitation and several biological metrics (Figures G-4 and G-
5). This included thermal-preference richness metrics (# cold-water taxa r=0.85, p<0.01; #
warm-water taxa r= -0.65, p=0.Q3).
G-5

-------
109
110
111
112
350
250
m^rinvDr-oo<3"iO»-»rMifY>*fr Lr»iJ)r>oooiOr-ir>jmyr in ^
<7icno>cr>o>cr»cricncricr%cricricrvcnooc>ooc>o
WH^HHMHHnHH-IHWHXHrN'NNrgrgNIN
- Mean Monthly Discharge (July &
August) (cfs)
Figure G-3. Relationship between mean monthly discharge and % EPT individuals over
time at NC0207 (Nantahala River).
113
114
115
116
1983 1984 1985 1986 1987 1988 1989 1990 1993 1998 2003
Year
-¦-PRISM mean annual air temperature (X)
—^ PRISM mean annual precipitation (inches)
(adjusted to fit scale - subtracted 25)
—Number of cold-water-preference taxa
— Number of warm-water-preferencetaxa
Figure G-4. Relationship between PRISM mean annual air temperature, PRISM mean
annual precipitation and thermal preference richness metrics at site NC0109 (New River).
G-6

-------
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
80
70
/
-®- PRISM mean annual atr temperature fC)
——PRISM mean annual precipitation (Inches)
~ %EPT individuals
20
10
0
1983 1984 1985 1986 1987 1988 1989 1990 1993 1998 2003
Year
Figure G-5. Relationship between PRISM mean annual air temperature, PRISM mean
annual precipitation and % EPT individuals at site NC01Q9 (New River).
Tables G-l and G-2 summarize distribution and abundance information for the North
Carolina cold- and warm-water-preference taxa at the 5 sites (Stations NC0109 - New, NC0207 -
Nantahala, NC0209 - Cataloochee, NC0075 - Little and NC0248 - Barnes) that were analyzed
for long-term trends. At these stations, the most prevalent cold-water-preference taxa were
Antocha and Promoresia which occurred at all the sites and in low to moderate abundances.
Acentrella, Atherix, Dolophilodes, Epeorus, and Eukiefferiella occurred at 6 of the sites and
therefore also appear to be stronger indicators. Procladius, Placobdella and Stenochironomus
were the most prevalent warm-water-preference taxa. They occurred at 5 sites and generally had
higher mean relative abundances than the other taxa. Chimarra and Macromia also appear to
occur in higher abundances than most of the other warm-water-preference taxa.
G-7

-------
Table G-l. Summary of distribution and abundance information for the cold-water-
preference taxaat the 5 sites (Stations NC0109, NC0207, NC0209, NC0075 and NC0248).
#Sites refers to the number of sites at which the taxa occurs. A=absent. P=present
(highlighted in grey). Relative abundance codes: L=low (<0.01), M=medium (0.01-0.1),
H=high (>0.1) (M or H are in bold type). Guide to interpretation: P-1L = present, occurred
duringl year, low relative abundance (RA), P-11M = present, occurred during 11 years,
medium RA, etc.	
FinallD
#Sites
NO • HI')
NC0207
\ro2fto
\r0rr5
NOI24S
Acentrella
6
l'-l.
\
IMI.
p-ll.
P-5M
Agapetus
3
A
I'-M.
IMI.
A
A
Amphinemura
2
\
\>-:\.
\
\
\
Antocha
7
I'-M.
P-SM
l)-"7M
I'-M.
\>-V.
Apatania
3
A
IMI.
I'-ll.
A
A
Arctopsyche
3
\
IM.M
P-(»M
\
\
Atherix
6
\
l'-~l.
l)-"7M
P-5.M
IM.M
Cardiocladius
5
l'-5l.
\
I'-M.
\
I'-ll.
Cinygmula
0
\
\
\
\
\
Clioperla
3
\
\
\
I'-ll.
IMI.
Cultus
4
A
\>-:\.
I'-ll.
A
IMI.
Diamesa
3
A
imi.
\>-2\.
A
A
Dicranota
3
\
P-'JM
P-7M
A
A
Diploperla
3
\
IMI.
\
\
\
Dolophilodes
6
1mm.
P-'JM
l)-"7M
\
\>-:\.
Dranella
4
P-SI
P-SM
l>-7M
\
\
Epeoras
6
p-(.i
P-'JM
P-"7M
\
IM.M
Eukiefferiella
6
I'-M.
l'-(>l.
P-7M
i'-:i.
\
Glossosoma
4
I'-M.
P-SM
P-"7M
\
\
Heleniella
2
\
IMI.
\
\
\
Isoperla
5
\
IM.M
P-7M
\
i'-:i.
Lanthus
5
\
P-'JM
P-7M
\
I'-n.
Malirekus
3
\
IMI.
IM.M
\
\
Nixe
3
\
IMI.
I'ol.
\
\
Pagastia
4
IMI.
P-(»M
IMI.
\
\
Parapsyche
2
\
\
P-ll.
\
\
Potthastia
4
\

I'-ll.
\
I'-n.
Promoresia
7
IM<>\|

P-7M
IMI.
\>-v.
Rheopelopia
3
\

i'-:i.
\
\
Rhithrogena
5
I'-M.
IMI.
P-^.M
\
\
Tallaperla
5
I'ol.
P-'JM
P-7M
\
\
Zapada
0
A
A
A
A
A
140
141
G-8
133
134
135
136
137
138
139

-------
142
143
144
145
146
147
148
149
150
151
152
153
Table G-2. Summary of distribution and abundance information for the warm-water-
preference taxa at the 5 sites (Stations NC0109, NC0207, NC0209, NC0075 and NC0248).
#Sites refers to the number of sites at which the taxa occurs. A=absent. P=present
(highlighted in grey). Relative abundance codes: L=low (<0.01), M=medium (0.01-0.1),
H=high (>0.1) (M or H are in bold type). Guide to interpretation: P-1L = present, occurred
duringl year, low relative abundance (RA), P-11M = present, occurred during 11 years,
medium RA, etc.
FinallD
#Sites
NC0109
NC0207
NC0948
NC0075
NC0248
Belostoma
0
A
A
A
A
A
Berosus
1
IMI.
A
A
A
A
Caecidotea
3
\
A
IMI.
IMI.
\
Chimarra
4
P-IIM
A
A
P-(.\l
P-^M
Elliptio
2
\>-:\.
A
A
A
A
Epicordulia
0
\
A
A
A
A
ERPOBDELLA/






MOOREOBDELLA
5
IMI.
A
IMI.
IMI.
i'-:i.
Helobdella
4
\>-:\.
A
\
IMI.
i'-ii.
Helocordulia
3
\
A
A
P-l.
I'-n.
Hetaerina
1
I'-M.
A
A
A
A
Ischnura
0
A
A
A
A
A
Lioporeus
1
A
A
A
IMI.
\
Macromia
5
l'-~l.
A
A
P-^M
P-(»M
Macrostemum
2
\>-h\.
A
A
P-ll.
\
Neureclipsis
2
\>-~\.
A
A
\
\
Neurocordulia
4
IMI.
\
A
I'ol.
i'-:i.
Nilothauma
4
A
IMI.
A
A
i'-:i.
Palaemonetes
1
A
A
A
A
A
Parachironomus
0
A
A
A
A
A
Pentaneura
1
IMI.
\
\
\
\
Phylocentropus
5
\
\
\
IMI.
I'-ll.
Phy sella
4
P-XM
\
\
I'-M.
IMI.
Placobdella
6
\>-:\.
IMI.
IMI.
i'-:i.
I'-ll.
Procladius
7
P-IOM
\>-:\.
i'-:i.
i'-:i.
i'-:i.
Stenochironomus
6
P-SM
imi.
1mm.
IMI.
I'-M.
Tetragoneuria
1
\
\
\
I'-n.
\
Tricorythodes
2
P-IIM
A
A
A
A
G3,2 How cold- and warm-water indicator taxa may affect EPT taxa richness, the NCBI
and final bioclassification levels
G-9

-------
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
Attachment G2 contains tables with lists of the temperature indicator taxa, temperature
optima values that were calculated from the maximum likelihood modeling, and the tolerance
values assigned by NCDENR that are used to calculate the NCBI. It should be noted that the
tolerance values are assigned at the species level by NCDENR. Because the maximum likelihood
modeling was done at the genus-level, some of the tolerance values had to be averaged across
species to get one value for each genus. The number of species within each genus that have been
assigned tolerance values along with minimum and maximum tolerance values of species within
each genus are also included in these tables. There is a fair amount of variation within some
genera.
Results of the analyses that were performed to examine potential climate change effects
on the EPT taxa richness metric are discussed in Section 2 of the report. Potential effects on the
NCBI are discussed in Section 3 of the report (see also Attachment G2). One set of results that
was not included in the report, but that are shown here, is from the correlation analysis of BI
values and PRISM mean annual precipitation variables. Results show BI values1 and PRISM
mean annual precipitation variables to be significantly correlated at 3 of the sites (Figure G-6).
1 For this particular analysis, we used the original BI values that were provided to us by NCDENR- this is
important to note because, as mentioned earlier, NCDENR calculates the BI values based on species,
while Tetra Tech calculated it based on genus-level OTUs.
G-10

-------
a)
North Carolina Site NC0109
Bl = 7.1102-0.0526*x; 0.95 Conf.lnt.
5.6
5.4
ppt14:Bl: r: = 0.7882
5.2
5.0
CO
4.2
4.0
3.8
25
30
35
40
PRISM mean annual precipitation (inches)
45
50
55
60
65
North Carolina Site NC0207/2554
Bl = 4.8965-0.0252*x; 0.95 Conf.lnt.
b)
4.0
3.8
ppt14:BI: ('= 0.5067
3.4
3.2
3.0
2.4
2.2
50
55
60
65	70
PRISM mean annual precipitation (inches)
75
80
85
90
95
c)
North Carolina Site NC0248
Bl = 3.1412+0.0327*x; 0.95 Conf.lnt.
5.0
4.9
ppt14:BI: r2 = 0.8087
4.7
4.5
4.4
4.3
4.2
4.1
32
34
36
38
40
42
44
46
48
50
52
54
56
58
PRISM mean annual precipitation (inches)
170
171	Figure G-6. Bl values and PRISM mean annual precipitation variables were significantly
172	correlated at 3 of the reference sites that were used in the correlation analyses: a) NC0109 -
173	New, b) NC 0207 - Nantahala and c) NC0248 - Barnes. Bl values are original values provided
174	to us by NCDENR.
G-ll

-------
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
Two of the sites are Mountain sites and one is a Piedmont site. At the 2 Mountain sites,
NC0207 (Nantahala) and NC0109 (New), the variables are significantly negatively correlated
(r2=0.51 and r2=0.79, respectively), while at the Piedmont site, NC0248 (Barnes), the variables
are positively correlated (r2=0.81). This suggests that there are site-specific differences as well as
differences among major ecoregions.
Results of the analysis simulating the effects of the loss of cold-water taxa on
bioclassification scores for 3 Mountain reference sites are discussed in Section 3 of the report, as
are results of the analysis in which Mountain criteria were applied to biotic assemblages at
selected reference Piedmont sites.
G3,3 Correlation analyses - commonly used metrics and climate-related variables
Metrics that are significantly correlated with the PRISM air temperature variables are
summarized in Table G-3. There are not many strong or consistent relationships between the
commonly used metrics and the temperature variables. Results appear to be mostly site-specific.
Only two metrics were significantly correlated with a temperature variable at more than one site:
the % climbers metric was negatively correlated with previous year PRISM mean annual air
temperature at 2 Blue Ridge sites; and the % predators metric was positively correlated with
PRISM mean annual air temperature at a Blue Ridge site and negatively correlated with it at a
Piedmont site. The 2 Piedmont sites have the most number (5) of metric values significantly
correlated with mean annual average air temperature (from the sampling year). Station NC0207
(Nantahala) has the most number (5) of significantly correlated metrics with the temperature
difference (sampling year - previous year) variable.
Results of the correlation analyses using the PRISM mean annual precipitation variables
are summarized in Table G-4. More metrics were significantly correlated with precipitation
variables than with temperature variables. But as with the temperature variables, there are not
many strong or consistent relationships and results appear to be mostly site-specific. Four metrics
were significantly correlated with a precipitation variable at more than one site: the Hilsenhoff
Biotic Index (HBI) (which used NC tolerance values, averaged at the genus-level) was
negatively correlated with PRISM mean annual precipitation at 2 Blue Ridge sites; the % climber
metric was negatively correlated with mean annual precipitation at 1 Blue Ridge and 1 Piedmont
site; the % shredder metric was negatively correlated with previous year mean annual
G-12

-------
206
207
208
209
210
211
212
213
214
215
216
217
precipitation at 1 Blue Ridge and 1 Piedmont site; and the % burrower metric was negatively
correlated with the precipitation difference (sampling year - previous year) variable at 1 Blue
Ridge site and positively correlated at 1 Piedmont site. Station NC0109 has the most number
(16) of metric values significantly correlated with PRISM mean annual precipitation values.
Station NC0207 (Nantahala) has the most number (10) of significantly correlated metrics with
the precipitation difference (sampling year - previous year) variable. Closer examination of the
data shows that mean annual precipitation values increased by 30 inches from 1993-1994, which
likely affected the biota.
Table G-3. Metric values that are significantly correlated with the selected temperature
variables are shown. + means positively correlated; - means negatively correlated. Values
are in bold print if they are significant at more than one site.		
PRISM
mean
annual
average air
temperature
(tmeanl4)
Metric
Blue Ridge
Piedmont
NC0109
(New)
NC0207
(Nantahala)
NC0209
(Cataloochee)
NC0075
(Little)
NC0248
(Barnes)
% Burrowers



-

# Predator Taxa



-

% Predators

+

-

# Cold-water
Indicator Taxa




-
% Drier Losers




-
Previous
year
PRISM
mean
annual
average air
temperature
Metric
Blue Ridge
Piedmont
NC0109
NC0207
NC0209
NC0075
NC0248
# Trichoptera Taxa
+




# Climber Taxa


-


% Swimmers


+


% Climbers

-
-


% Sprawlers



-

% Collector-gatherers

-



G-13

-------
219 Table G-3. Continued
Absolute
difference
between
PRISM mean
annual average
air temperature
from the
sampling year
and the
previous year
Metric
Blue Ridge
Piedmont
NC0109
NC0207
NC0209
NC0075
NC0248
# Total Taxa

+



# Plecoptera Taxa

+



# Swimmer Taxa

+



# Climber Taxa

+



% dingers




-
# Predator Taxa

+



% Shredders


-


% Cold-water
Indicators




+
# OCH Taxa




+
220
221
G-14

-------
223
224
225
Table G-4. Metric values that are significantly correlated with the selected precipitation
variables are shown. + means positively correlated; - means negatively correlated. Values
PRISM
mean
annual
precipitatio
n (pptl4)
Metric
Blue Ridge
Piedmont
NC010
9 (New)
NC0207
(Nantahala
)
NC0209
(Cataloochee
)
NC0075
(Little)
NC0248
(Barnes)
# Total Taxa
-




# Ephemeroptera Taxa




-
# Plecoptera Taxa
+




% Plecoptera

+



% EPT
+




HBI
-
-



# Climber
-




% dingers
+




% Climbers
-


-

# Herbivore Taxa
-




# Predator Taxa
-




% Predators
-




% Cold-water Indicators
+




# Cold-water Indicator Taxa
+




# Warm-water Indicator Taxa
-




% Perennial
+




% Drought Resistant
+




# of Intermittent Taxa
-




Previous
year PRISM
mean
annual
precipitatio
n
Metric
Blue Ridge
Piedmont
NC010
9
NC0207
NC0209
NC0075
NC0248
# EPT Taxa

+



% Plecoptera

+



# Burrower Taxa


-


# Collector-gatherer taxa




+
% Shredders


-
-

% Drier Losers


+


G-15

-------

% Warm Drier Losers


+


# OCH Taxa
-




G-16

-------
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
Table G-4. Continued

Metric
Blue Ridge
Piedmont

NC0109
NC0207
NC0209
NC0075
NC0248

# Total Taxa

-




# Plecoptera Taxa




-

# EPT Taxa


+


Absolute
difference
# Swimmer Taxa

-



# Burrower Tax

-



between
PRISM
# Sprawler Taxa

-



mean
annual
precipitaton
% Burrowers
-



+
# Collector-gatherer taxa

-



from the
sampling
year and
# Predator Taxa

-



% Collector-filterer
+




the
previous
year
% Shredders

+



% Herbivores

-




% Predators




+

% Perennial
+





% Intermittent




+

# Perennial Taxa

-




# Intermittent Taxa

-



G3.4 Summary
The mean number of cold-water -preferencetaxa at sites in the Mountain ecoregion is
significantly higher than the mean number of cold-water-preference taxa at sites in the
other two ecoregions. The mean number of warm-water-preference taxa is significantly
different between all 3 ecoregions with the highest number occurring in the Coastal
ecoregion and the lowest number occurring in the Mountain ecoregion.
Significantly more cold-water-preference taxa are present at higher elevation sites than at
lower elevation sites. There is a significantly higher number of warm-water-preference
taxa at lower elevation sites.
Many of the cold-water-preference taxa in North Carolina are EPT taxa: 8 of the cold-
water-preference taxa are Plecopterans, 6 are Trichopterans and 6 are Ephemeropterans.
There are substantially fewer EPT taxa on the warm-water-preference list: 1 warm-water-
preference taxais an Ephemeroptera, 4 are Trichopterans and none are Plecopterans.
G-17

-------
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
•	Within the EPT genera on the cold-water-preference list, there are 53 species that could
potentially be counted towards the EPT richness metric that is used in the
bioclassification of sites in NC, while there are only 5 species that could be potentially
counted from the warm-water-preference list (therefore changes to the cold-water-
preference taxa are likely to have a greater effect on EPT richness scores).
•	EPT Richness: a loss of 3 (Coastal sites) or 4 (Mountain or Piedmont sites) EPT species
can lower the EPT richness score at a high quality site a full level, from a 5 (Excellent) to
a 4 (Good). To drop the EPT richness bioclassification by a full level at sites of lesser
quality, it would take a loss of 10 taxa at Mountain sites, 8 taxa at Piedmont sites and 7
taxa at Coastal sites.
•	NCBI: an increase in BI scores of 0.1 can lower the BI score at a high quality site a full
level, from a 5 (Excellent) to a 4 (Good). To drop the NCBI bioclassification by a full
level at sites of lesser quality, NCBI scores would have to increase by at least 0.6 (it
varies by ecoregion and bioclassification level).
•	When cold-water-preference taxa were eliminated from the biotic assemblages at 3
references sites in the Mountain ecoregion, the effects on EPT richness scores, NCBI
scores and overall bioclassification levels were relatively small and site-dependent:
o At Station NC0109, which had fewer cold-water-preference taxa than the other 2
sites, the loss of cold-water-preference taxa resulted in little if any change to EPT
richness ratings (maximum loss of 4 species, maximum decrease in EPTS score
of 0.6), little if any change to BI values and scores (maximum increase in BI
value of 0.24, maximum decrease in BI score of 0.2) and the maximum drop in
overall score was 1 bioclassification level (from Excellent to Good), and this
occurred 3 out of 11 years
o At Stations NC0209 and NC0207/2554, removal of cold-water-preference taxa
resulted in the loss of 9 to 14 EPT species decreases in EPT S scores ranging
from 0.4 to 1.2, an increase in BI values ranging from 0.45 to 0.86 and decreases
in BI scores ranging from 0 to 1, and the maximum drop in score was one
bioclassification level (from Excellent to Good), which occurred 5 out of 7 years
at Site NC0209 and 5 out of 8 years at Site NC0207/NC2554.
•	Effects at the other 2 sites were more noticeable because they have many more cold-
water-preference taxa. Removal of cold-water-preference taxa resulted in the loss of 9 to
14 EPT species decreases in EPT S scores ranging from 0.4 to 1.2.
•	22 of the 30 cold-water-preference taxa that have been assigned tolerance values have
low tolerance values (< 3). Tolerance values for most of the warm-water-preference taxa
are higher. Twelve of the warm-water-preference taxa that have been assigned tolerance
values have tolerance values > 7.
•	Temperature optima values are significantly and positively correlated with tolerance
values (r=0.53, p=00), indicating that taxa that show preferences for lower temperatures
tend to have lower tolerance values and those that tend to occur more in warmer water
habitats tend to have higher tolerance values.
•	Results from correlation analyses using thermal preference metrics and BI values suggest
that replacement of colder water taxa with warmer water taxa would likely contribute to a
site receiving a higher BI score and therefore a poorer rating, and that this is most likely
to affect sites in the Mountain ecoregion.
G-18

-------
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
•	BI values (which were provided to us by NCDENR) and PRISM mean annual
precipitation variables were significantly correlated at 3 of the sites. At the 2 Mountain
sites, the variables are negatively correlated (r2=0.51 and r2=0.79, respectively), while at
the Piedmont site, the variables are positively correlated.
•	When Mountain bioclassification criteria was applied to the biotic assemblages at the 2
Piedmont sites to simulate how ratings may change if taxa that typically inhabit Mountain
sites are replaced by assemblages that are more typical of the Piedmont ecoregion,
bioclassifications consistently dropped by 1 level.
•	There are not many strong or consistent relationships between the commonly used
metrics and the temperature variables. Results appear to be mostly site-specific.
•	Two metrics were significantly correlated with a temperature variable at more than one
site: the % climbers metric was negatively correlated with previous year PRISM mean
annual air temperature at 2 Blue Ridge sites; and the % predators metric was positively
correlated with PRISM mean annual air temperature at a Blue Ridge site and negatively
correlated with it at a Piedmont site.
•	More metrics were significantly correlated with precipitation variables than with
temperature variables. But as with the temperature variables, there are not many strong or
consistent relationships and results appear to be mostly site-specific.
•	Four metrics were significantly correlated with a precipitation variable at more than one
site: the HBI (which used NC tolerance values, averaged at the genus-level) was
negatively correlated with PRISM mean annual precipitation at 2 Blue Ridge sites; the %
climber metric was negatively correlated with mean annual precipitation at 1 Blue Ridge
and 1 Piedmont site; the % shredder metric was negatively correlated with previous year
mean annual precipitation at 1 Blue Ridge and 1 Piedmont site; and the % burrower
metric was negatively correlated with the precipitation difference (sampling year -
previous year) variable at 1 Blue Ridge site and positively correlated at 1 Piedmont site.
G-19

-------
>	Attachment G1
2
3	North Carolina Temperature Indicator Taxa
4
5	This attachment contains tables with lists of the North Carolina temperature-indicator taxa and
6	describes the process that we followed to develop these lists.
Gl-1

-------
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
ATTACHMENT Gl. NORTH CAROLINA TEMPERATURE-INDICATOR TAXA
Sources. The North Carolina cold- and warm-water taxa lists were developed using
several different sources: 1. maximum likelihood calculations based on a subset of the North
Carolina biomonitoring database (using full-scale collection method data); 2. the thermal-
preference trait from the Poff et al. (2006) traits matrix; 3. the thermal-preference trait from the
USGS traits database (Vieira et al., 2006); 4. the thermal preference trait from the compilation of
EPA Environmental Requirements and Pollution Tolerance series from the late 1970's (Beck et
al., 1977; Harris et al., 1978; Hubbard et al., 1978, Surdick et al., 1978); 5. best professional
judgment of the Southeast Climate Change traits feedback group2.
The same general criteria and guidelines that were used to designate cold- and warm-
water indicator taxa in Maine were also used in North Carolina (see Attachment D2). Also, see
Attachment D2 for general limitations of the analyses.
Initial Results. Initially there were 126 taxa on the cold-water list and 112 taxa on the
warm-water list. These lists were based on maximum likelihood calculations and literature.
These lists were further refined through the evaluation of additional evidence. This evidence
included analyses of other datasets, case studies, and best professional judgment. Taxa with the
greatest amount of evidence were designated as temperature indicator taxa. More detailed
information about the steps that were used to develop the temperature indicator taxa lists is
summarized below:
Considerations (unique to North Carolina)
Several 'case studies' were performed to see whether the cold- or warm-water taxa
occurred at sites in North Carolina that had the warmest or coldest summer water temperatures.
The following case studies were performed:
a. Cold-water Case Study #1. Taxa lists from two Blue Ridge reference sites(NC1560-
BEAR CR and NC1561-HAZEL CR) that have full-scale collection method data,
have <5% urban and <10% agricultural land use within a 1 km buffer, and have the
coldest recorded summer water temperatures (13-14° C in July). Note: there were a
number of sites with temperature readings of 0°C; these readings seemed
questionable so they were not used.
2 North Carolina DWQ (Trish MacPherson), South Carolina (Jim Glover) and Tennessee (Debbie Arnwine)
Gl-1

-------
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
b.	Cold-water Case Study #2. The taxa list from the Piedmont site (NC0634-TOWN
FORK CR) that has full-scale collection method data and has the coldest recorded
Piedmont summer water temperature (9° C in August). This site has 4.3% urban and
11% agricultural land uses within a 1 km buffer.
c.	Cold-water Case Study #3. Taxa lists from three Piedmont reference sites
(NC0248-BARNES CR, NC0713-CATAWBA R, NC1607-MARLOWE CR) that
have full-scale collection method data, have <5% urban and <10% agricultural land
uses within a 1 km buffer, and have the coldest recorded summer water temperatures
(16-17° C in August and September).
d.	Warm-water Case Study #1. Taxa lists from the two warmest reference sites in the
state (NC1466-CAPE FEAR R and NC1467-CAPE FEAR R) that have full-scale
collection method data, have <5% urban and <10% agricultural land uses within a 1
km buffer, and have the warmest recorded summer water temperatures (30-32° C in
July).
e.	Warm-water Case Study #2. Taxa lists from the two Piedmont reference sites
(NC0219-TAR R and NC0573-DEEP R) that have full-scale collection method data,
have <5% urban and <10% agricultural land uses within a 1 km buffer, and have the
warmest recorded summer water temperatures (28-29° C in July).
f.	Warm-water Case Study #3. Taxa list from the warmest Blue Ridge reference
site(NC1285-CROOKED CR) that has full-scale collection method data, has <5%
urban and <10% agricultural land uses within a 1 km buffer, and has the warmest
recorded summer water temperature (24° C in July).
Development of the Temperature-Indicator Cold-Water Taxa List. Taxa were placed
on the cold-water list if the following criteria were met:
1.	The taxon received a 'yes' per best professional judgment AND has been recorded at one
or more of the cold-water case study sites AND has NOT been recorded at either of the
two warm-water case study sites.
2.	The taxon received a 'yes' per best professional judgment AND received a Total Score of
5 or more.
3.	The taxon received a 'no comment' per best professional judgment AND has been
recorded at two or more of the cold-water case study sites AND no species variation was
noted.
Gl-2

-------
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
Development of the Temperature-Indicator Warm-Water List. Taxa were placed on
the warm-water list if the following criteria were met:
1.	The taxon received a 'yes' per best professional judgment AND has been recorded at one
or more of the warm-water case study sites AND has NOT been recorded at either of the
two cold-water case study sites.
2.	The taxon received a 'yes' per best professional judgment AND received a Total Score of
5 or more.
3.	The taxon received a 'no comment' per best professional judgment AND has been
recorded at two or more of the warm-water case study sites AND no species variation
was noted.
4.	The taxon received a 'no comment' per best professional judgment AND received a Total
Score of 5 or more AND has been recorded at one or more warm-water case study sites
AND NOT at any of the cold-water case study sites AND no species variation was noted.
Temperature Indicator Lists. The cold-water taxa list was comprised of 32 taxa and the
warm-water taxa list was comprised of 27 taxa. Tables Gl-1 and Gl-2 show the temperature
indicator taxa lists.
Important Notes - variation within genera. Some noteworthy genera were left off the
North Carolina cold-water taxa list. These included Ephemerella, Neophylax, Rhyacophila,
Goera, Eurylophella and Paragnetina. The reason they were not included is because there is
variation in temperature preferences among species within these genera, and this was noted by
the Southeast Climate Change feedback group. Genera that were left off the warm-water list due
to species variation included Hydropsyche, Oecetis and Polypedilum.
Dispersal Ability. If temperature is a major factor influencing community composition,
then taxa that are able to adapt to warming temperatures and/or that are able to disperse to more
favorable habitats (generally believed to be upstream or to higher elevations) have a better
chance of surviving. Five mobility traits were examined for the taxa on the North Carolina
temperature indicator lists: dispersal (adult), adult flying strength, occurrence in drift, maximum
crawling rate and swimming ability. Table Gl-3 lists more information on these traits.
Dispersal (adult) and adult flying strength received the greatest amount of consideration.
Because movement is most likely to be upstream, taxa that are strong fliers are likely to have a
better chance of success. It will be difficult for taxa that disperse via occurrence in drift to
Gl-3

-------
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
migrate upstream, and taxa that disperse via crawling or swimming are likely to have difficulty
moving the distances required to find more favorable habitats.
One of the 32 taxa on the North Carolina temperature indicator cold-water taxa list (that
we had trait information for), Clioperla, is categorized as having 'high' dispersal ability. Another
taxon, Lanthus, is categorized as having strong flying ability but low adult dispersal ability. Nine
of the 27 taxa on the warm-water list are categorized as having high adult dispersal ability. Six of
these taxa are considered to be strong fliers.
Abundance and Distribution. In addition to dispersal ability, abundance and
distribution are also important considerations. Those taxa that are widespread and common are
likely to have greater genetic diversity and greater chance of adapting than rare taxa that only
occur in isolated, localized populations (Sweeney et al., 1992). Moreover, the more abundant
taxa are more likely to affect the state biomonitoring assessments. Abundance and distribution
information for the temperature indicator taxa can be found in Tables Gl-1 and Gl-2. It should
be noted once again that the abundance data in the North Carolina dataset is categorical (l=rare
(1-2 specimens), 3=common (3-9 species) and 10=abundant (10 or more species).
The most abundant cold-water temperature indicator taxa are Epeorus (Ephemeropteran),
Antocha (Dipteran), Isoperla (Plecopteran) and Tallaperla (Plecopteran). These taxa comprise
only 0.4 to 0.6% of the total individuals in the North Carolina database. Seventeen of the cold-
water taxa have overall abundances of less than 0.1%. Physella (Basommatophora), Chimarra
(Trichopteran) and Macromia (Odonata) are the most abundant warm-water taxa, with overall
abundances ranging from 0.6 to 0.8%. Twelve of the warm-water taxa have overall abundances
of less than 0.1%. Of the cold-water taxa, Antocha occurs at the largest percentage of sites
(25%), followed by a Chironomidae, Eukiefferiella, and a Plecopteran, Isoperla, which occur at
18-19%) of the sites. Eighteen of the cold-water taxa occur at less than 10%> of the sites. Among
the warm-water taxa, Physella occurs at the highest percentage of sites (30%>), followed by
Macromia (29%) and Stenochironomus (21%). Nineteen of the warm-water taxa occur at less
than 10%) of the sites.
Gl-4

-------
129	Additional information - Cold-water Taxa.
130	Ten of the cold-water taxa are Dipterans, eight are Plecopterans, six are Ephemeropteran
131	and six are Trichopterans. The rest are Coleopterans and Odonates. The families with the most
132	number of taxa on the cold-water list are Chironomidae, Perlodidae and Heptageniidae (Table
133	Gl-4).
134	Additional information - Warm-water Taxa.
135	Seven of the warm-water taxa are Odonates, five are Dipterans and four are
136	Trichopterans. The families with the most number of taxa on the warm-water list are
137	Chironomidae and Corduliidae (Table Gl-5).
Gl-5

-------
138	Table Gl-1. List of North Carolina cold-water temperature indicator taxa. Distribution and abundance information is also included.
139	Sum_Individuals=the total number of individuals from that taxon in the North Carolina database; Pct_Abund=percent of total
140	individuals in the database comprised of that taxon; Num_Stations=number of stations in the database that the taxon occurred at;
141	Pet Stations=percent of stations in the database at which the taxon occurred.	
Type
Order
Family
FinallD
Sumlndivids
PctAbund
Num_Stations
PctStations
cold
Ephemeroptera
Baetidae
Acentrella
2745
0.33
427
15.19
cold
Trichoptera
Glossosomatidae
Agapetus
247
0.03
53
1.89
cold
Plecoptera
Nemouridae
Amphinemura
1210
0.14
281
10
cold
Diptera
Tipulidae
Antocha
5103
0.61
711
25.29
cold
Trichoptera
Apataniidae
Apatania
339
0.04
47
1.67
cold
Trichoptera
Hydropsychidae
Arctopsyche
222
0.03
40
1.42
cold
Diptera
Athericidae
Atherix
1236
0.15
240
8.54
cold
Diptera
Chironomidae
Cardiocladius
2300
0.27
376
13.38
cold
Ephemeroptera
Heptagenidae
Cinygmula
247
0.03
40
1.42
cold
Plecoptera
Perlodidae
Clioperla
574
0.07
155
5.51
cold
Plecoptera
PERLODIDAE
Cultus
296
0.04
70
2.49
cold
Diptera
Chironomidae
Diamesa
734
0.09
185
6.58
cold
Diptera
Tipulidae
Dicranota
1384
0.16
284
10.1
cold
Plecoptera
Perlodidae
Diploperla
393
0.05
122
4.34
cold
Trichoptera
Philopotamidae
Dolophilodes
2905
0.35
316
11.24
cold
Ephemeroptera
EPHEMERELLIDAE
Drunella
2846
0.34
218
7.76
cold
Ephemeroptera
Heptageniidae
Epeorus
5226
0.62
403
14.34
cold
Diptera
CHIRONOMIDAE
Eukiefferiella
2974
0.35
533
18.96
cold
Trichoptera
Glossosomatidae
Glossosoma
1755
0.21
309
10.99
cold
Diptera
Chironomidae
Heleniella
95
0.01
50
1.78
cold
Plecoptera
PERLODIDAE
Isoperla
4556
0.54
498
17.72
cold
Odonata
Gomphidae
Lanthus
1174
0.14
300
10.67
cold
Plecoptera
Perlodidae
Malirekus
753
0.09
132
4.7
cold
Ephemeroptera
HEPTAGENIIDAE
Nixe
64
0.01
16
0.57
cold
Diptera
Chironomidae
Pagastia
751
0.09
157
5.59
Gl-6

-------
142
143
144
145
146
147
148
149
Table Gl-1. Continued
Type
Order
Family
FinallD
Sumlndivids
PctAbund
Num_Stations
PctStations
cold
Trichoptera
Hydropsychidae
Parapsyche
280
0.03
52
1.85
cold
Diptera
CHIRON OMID AE
Potthastia
757
0.09
292
10.39
cold
Coleoptera
Elmidae
Promoresia
3020
0.36
332
11.81
cold
Diptera
Chironomidae
Rheopelopia
135
0.02
64
2.28
cold
Ephemeroptera
Heptageniidae
Rhithrogena
725
0.09
152
5.41
cold
Plecoptera
Peltoperlidae
Tallaperla
3337
0.4
377
13.41
cold
Plecoptera
NEMOURIDAE
Zapada
3
0
3
0.11
Table Gl-2. List of North Carolina warm-water temperature indicator taxa. Distribution and abundance information is also included.
Sum_Individuals=the total number of individuals from that taxon in the North Carolina database; Pct_Abund=percent of total
individuals in the database comprised of that taxon; Num_Stations=number of stations in the database that the taxon occurred at;
Pet Stations=percent of stations in the database at which the taxon occurred.	
Type
Order
Family
warm
Hemiptera
Belostomatidae
warm
Coleoptera
Hydrophilidae
warm
Isopoda
ASELLIDAE
warm
Trichoptera
Philopotamidae
warm
Unionoida
UNIONIDAE
warm
Odonata
Corduliidae
warm
Arhynchobdellida
ERPOBDELLIDAE
warm
Rhynchobdellida
Glossiphoniidae
warm
Odonata
Corduliidae
warm
Odonata
Calopterygidae
warm
Odonata
Coenagrionidae
warm
Coleoptera
Dytiscidae
FinallD
Sum Individs Pet Abund Num Stations Pet Stations
Belostoma
173
0.02
99
3.52
Berosus
1843
0.22
277
9.85
Caecidotea
3203
0.38
544
19.35
Chimarra
5178
0.62
554
19.71
Elliptio
1556
0.18
189
6.72
Epicordulia
178
0.02
78
2.77
ERPOBDELLA/
760
0.09
210
7.47
MOOREOBDELLA
Helobdella
835
0.1
225
8
Helocordulia
188
0.02
95
3.38
Hetaerina
854
0.1
153
5.44
Ischnura
318
0.04
101
3.59
Lioporeus
182
0.02
83
2.95
Gl-7

-------
150 Table Gl-2. Continued
Type
Order
Family
FinallD
Sumlndivids
PctAbund
Num_Stations
PctStations
warm
Odonata
Corduliidae
Macromia
5064
0.6
813
28.92
warm
Trichoptera
Hydropsychidae
Macrostemum
1753
0.21
134
4.77
warm
Trichoptera
Polycentropodidae
Neureclipsis
2092
0.25
241
8.57
warm
Odonata
Corduliidae
Neurocordulia

0.18
278
9.89
warm
Diptera
Chironomidae
Nilothauma
180
0.02
124
4.41
warm
Decapoda
Palaemonidae
Palaemonetes
2262
0.27
271
9.64
warm
Diptera
Chironomidae
Parachironomus
395
0.05
128
4.55
warm
Diptera
Chironomidae
Pentaneura 1511
771
0.09
154
5.48
warm
Trichoptera
Dipseudopsidae
Phylocentropus
576
0.07
201
7.15
warm
Basommatophora
Physidae
Physella
6677
0.79
853
30.35
warm
Rhynchobdellida
Glossiphoniidae
Placobdella
677
0.08
339
12.06
warm
Diptera
Chironomidae
Procladius
3460
0.41
706
25.12
warm
Diptera
Chironomidae
Stenochironomus
3419
0.41
750
26.68
warm
Odonata
CORDULIIDAE
Tetragoneuria
687
0.08
202
7.19
warm
Ephemeroptera
Leptohyphidae
Tricorythodes
4939
0.59
363
12.91
Gl-8

-------
151	Table Gl-3. Mobility traits that were evaluated. The source of most of this information was the
152	Poff et al. 2006 traits matrix. Some also came from the USGS traits database (Vieira et al., 2006).
Mobility Trait	Trait States
low (<1 km flight before laying eggs), high (>1 km flight before
laying eggs)
weak (e.g. cannot fly into light breeze), strong
rare (catastrophic only), common (typically observed), abundant
(dominant in drift samples)
very low (<10 cm/h), low (<100 cm/h), high (>100 cm/h)
none, weak, strong
153
154	Table Gl-4. Number of cold-water taxa in each family
Family	Total
Chironomidae
7
Perlodidae
5
Heptageniidae
4
Glossosomatidae
2
Hydropsychidae
2
Nemouridae
2
Tipulidae
2
Apataniidae
1
Athericidae
1
Baetidae
1
Elmidae
1
Ephemerellidae
1
Gomphidae
1
Peltoperlidae
1
Philopotamidae
1
155
Dispersal (adult)
Adult flying strength
Occurrence in drift
Maximum crawling rate
Swimming ability
G1--9

-------
157
158	Table Gl-5. Number of warm-water taxa in each family
Family	Total
Chironomidae	5
CORDULIIDAE	5
Glossiphoniidae	2
Asellidae	1
Belostomatidae	1
Calopterygidae	1
Coenagrionidae	1
Dipseudopsidae	1
Dytiscidae	1
ERPOBDELLIDAE	1
Hydrophilidae	1
Hydropsychidae	1
Leptohyphidae	1
Palaemonidae	1
Philopotamidae	1
Physidae	1
Polycentropodidae	1
Unionidae	1
159
160
161
Gl-10

-------
, Attachment G2
2
3
4
s	Tolerance values of the North Carolina
6	temperature-indicator taxa
7
8	This attachment contains tables with lists of the temperature-indicator taxa, temperature optima
9	and tolerance values that were calculated from the maximum likelihood modeling, and the
10	tolerance values assigned by North Carolina DWQ (which are used to calculate the NCBI).
11	These tables were used to examine whether temperature-indicator taxa were considered to be
12	sensitive or tolerant taxa.
G2-1

-------
13	ATTACHMENT G2 TOLERANCE VALUES OF THE NORTH CAROLINA TEMPERATURE INDICATOR TAXA
14
15	Table G2-1. Cold-water temperature indicator taxa. Temp Opt is the temperature optima (°C) calculated during the maximum likelihood
16	modeling (temperature tolerance could not be calculated for this dataset). Avg TolVal was calculated by taking the average of the
17	tolerance values assigned to species within the genera. The number of species within each genus that have been assigned tolerance values
18	(NumSpecies) along with minimum (Min TolVal) and maximum tolerance values (Max TolVal) of species within each genus are also
19	included.
Order
Family
Genus
Temp Opt
NumSpecies
Avg TolVal
Min TolVal
Max TolVal
Coleoptera
Elmidae
Promoresia
10.6
3
1.5
0
2.4
Diptera
Athericidae
Atherix
9
2
2.1
2.1
2.1
Diptera
Chironomidae
Cardiocladius
13.2
1
5.9
5.9
5.9
Diptera
Chironomidae
Diamesa
15.8
1
8.1
8.1
8.1
Diptera
CHIRONOMIDAE
Eukiefferiella
9
6
3.4
2.2
5.6
Diptera
Chironomidae
Heleniella
13
1
0
0
0
Diptera
Chironomidae
Pagastia
15.3
1
1.8
1.8
1.8
Diptera
CHIRONOMIDAE
Potthastia
15.2
3
5
2
6.5
Diptera
Chironomidae
Rheopelopia
9




Diptera
Tipulidae
Antocha
15.7
1
4.3
4.3
4.3
Diptera
Tipulidae
Dicranota
9
1
0
0
0
Ephemeroptera
Baetidae
Acentrella
16.9
4
4.3
3.6
5.5
Ephemeroptera
EPHEMERELLIDAE
Drunella
9
8
0.2
0
1
Ephemeroptera
Heptagenidae
Cinygmula

1
0
0
0
Ephemeroptera
Heptageniidae
Epeorus
9
4
1.3
1
1.8
Ephemeroptera
HEPTAGENIIDAE
Nixe
9
3
0.7
0
1
Ephemeroptera
Heptageniidae
Rhithrogena
14.6
5
0.3
0.3
0.3
Odonata
Gomphidae
Lanthus
9
3
1.8
1.8
1.8
Plecoptera
Nemouridae
Amphinemura
9
1
3.3
3.3
3.3
Plecoptera
NEMOURIDAE
Zapada





Plecoptera
Peltoperlidae
Tallaperla
9
1
1.2
1.2
1.2
Plecoptera
Perlodidae
Clioperla

1
4.7
4.7
4.7
G2-2

-------
21 Table G2-1. Continued
Order
Family
Genus
Temp Opt
NumSpecies
Avg TolVal
Min TolVal
Max TolVal
Plecoptera
PERLODIDAE
Cultus

1
1.6
1.6
1.6
Plecoptera
Perlodidae
Diploperla

2
2.1
1.4
2.7
Plecoptera
PERLODIDAE
Isoperla
9
12
1.7
0
5.4
Plecoptera
Perlodidae
Malirekus

1
1.2
1.2
1.2
Trichoptera
Apataniidae
Apatania

1
0.6
0.6
0.6
Trichoptera
Glossosomatidae
Agapetus

1
0
0
0
Trichoptera
Glossosomatidae
Glossosoma
9
1
1.6
1.6
1.6
Trichoptera
Hydropsychidae
Arctopsyche

1
0
0
0
Trichoptera
Hydropsychidae
Parapsyche
13.1
1
0
0
0
Trichoptera
Philopotamidae
Dolophilodes
9
1
0.8
0.8
0.8
22
23
24	Table G2-2. Warm-water temperature indicator taxa. Temp Opt is the temperature optima (°C) calculated during the maximum
25	likelihood modeling (temperature tolerance could not be calculated for this dataset). Avg TolVal was calculated by taking the average of
26	the tolerance values assigned to species within the genera. The number of species within each genus that have been assigned tolerance
27	values (NumSpecies) along with minimum (Min TolVal) and maximum tolerance values (Max TolVal) of species within each genus are
28	also included.
Order
Family
FinallD
Temp Opt
NumSpecies
Avg TolVal
Min TolVal
Max TolVal
Arhynchobdellida
ERPOBDELLIDAE
ERPOBDELLA/
MOOREOBDELLA
29.4
1
8.3
8.3
8.3
Basommatophora
Physidae
Physella
32
1
8.8
8.8
8.8
Coleoptera
Dytiscidae
Lioporeus
32
2
3
3
3
Coleoptera
Hydrophilidae
Berosus
32
1
8.4
8.4
8.4
Decapoda
Palaemonidae
Palaemonetes
31.5
2
7.1
7.1
7.1
Diptera
Chironomidae
Nilothauma
32
1
5
5
5
Diptera
Chironomidae
Parachironomus
32
4
8.5
6.5
9.6
Diptera
Chironomidae
Pentaneura
32
1
4.7
4.7
4.7
Diptera
Chironomidae
Procladius
32
1
9.1
9.1
9.1
G2-3

-------
29 Table G2-2. Continued
Order
Family
FinallD
Temp Opt
NumSpecies
Avg TolVal
Min TolVal
Max TolVal
Diptera
Chironomidae
Stenochironomus
32
1
6.5
6.5
6.5
Ephemeroptera
Leptohyphidae
Tricorythodes
32
1
5.1
5.1
5.1
Hemiptera
Belostomatidae
Belostoma
29.5
1
9.8
9.8
9.8
Isopoda
ASELLIDAE
Caecidotea
26.1
1
9.1
9.1
9.1
Odonata
Calopterygidae
Hetaerina
28.1
1
5.6
5.6
5.6
Odonata
Cocnagrionidac
Ischnura
32
1
9.5
9.5
9.5
Odonata
Corduliidae
Epicordulia
28.4
2
5.6
5.6
5.6
Odonata
Corduliidae
Helocordulia
27.8
2
4.9
4.8
4.9
Odonata
Corduliidae
Macro mia
32
2
6.2
6.2
6.2
Odonata
Corduliidae
Neurocordulia
32
4
3.5
1.8
5.2
Odonata
CORDULIIDAE
Tctragoncuria
32
2
8.6
8.5
8.6
Rhynchobdellida
Glossiphoniidae
Helobdella
28.9
3
9.1
8.6
9.5
Rhynchobdellida
Glossiphoniidae
Placobdella
27
3
8.9
8.7
9
Trichoptera
Dipseudopsidae
Phylocentropus
32
1
6.2
6.2
6.2
Trichoptera
Hydropsychidae
Macrostemum
32
1
3.5
3.5
3.5
Trichoptera
Philopotamidae
Chimarra
32
1
2.8
2.8
2.8
Trichoptera
Polycentropodidae
Neureclipsis
32
2
4.2
4.2
4.2
Unionoida
UNIONIDAE
Elliptio
32
3
4.2
2.4
5.1
30
G2-4

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
APPENDIX H
Temporal Change in Regional Reference
Condition as a Potential Indicator of Global
Climate Change: Analysis of the Ohio Regional
Reference Condition Database (1980-2006)
Analyses on Ohio data conducted by
Edward T. Rankin, Senior Research Associate, Voinovich Center for Leadership and Public Affairs, The
Ridges, Building 22, Ohio University, Athens, OH 45701
and
Chris O. Yoder, Research Director, Center for Applied Bioassessment and Biocriteria, Midwest
Biodiversity Institute, P.O. Box 21541, Columbus, OH 43221
The intent of this appendix is to provide more comprehensive and detailed information on the
analyses that were performed on the Ohio data. Some of the analyses that are covered in this
appendix are also referenced in the main body of the report. When this occurred, attempts were
made to reduce any overlap or duplication in the reporting of results.
HI. Overview of Ohio Indices and Associated Assessment Approach
H2. Ohio EPA Regional Reference Database - Background
H3. Data Analyses
H4. Results
H-l

-------
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
HI Overview of Ohio Indices and Associated Assessment Approach
The State of Ohio Environmental Protection Agency (Ohio EPA) calculates an
Invertebrate Community Index (ICI) to evaluate biological condition based on the benthic
macroinvertebrate assemblage and an Index of Biotic Integrity (IBI) used to evaluate fish
assemblages at wading sites, boat sites and headwaters stream sites. The metrics that go into the
ICI and IBI are shown in Figures Hl-1 and Hl-2 (State of Ohio Environmental Protection
Agency, Environmental Assessment Section, Division of Water Quality, Planning and
Assessment. 1989 (last updated 2008). Biological Criteria for the Protection of Aquatic Life:
Volume III: Standardized Biological Field Sampling and Laboratory Methods for Assessing Fish
and Macroinvertebrate Communities.
http://vvvvvv.epa.state.oh.us/dsvv/bioassess/BioCriteriaProtAqLife.html).
Sto-c-
Metric	0	2	4
1-	Tnfa 1 Nuff&i?!" of Tasta	Vj' 1 a:. with drainage area (Flo, 5 1
2-	Total Number of Mayfly Taxes	Varlej with  with drainage area 1M5.
4.	total Numoer of Qloteran Taxs	Va-lei with drainage area iFig. 5-4'
5.	Percent Mayfly Zorcsosl tlon	0	^0,2^
6.	Percent Caddlsfly Composition	Va-les with drainage area (Fig. !>-6*
7.	Percent 1 r 1 tie Tanytarslni
Compos 11 \ Oft	0	>0 ,<10 >10, <25 >25
8.	Percent other Ulpteran ana
Hon Insect Composition	Va-les with d-a1nage area iFig. S-fi'
§, Percent In It-rant nrganUntt	Varies with drainage area (Fig, t-Q
(fram Table 5-2}
10. Totd 1 Number of Qualitative
LPT T a >.i	V a • 1 c •» with drainage er ca >riy. S-1 0)
Figure Hl-1. Macroinvertebrate community metrics and criteria for calculating the
Invertebrate Community Index (ICI) and ICI scores for evaluating biological condition in
H-2

-------
40
41
42
43
44
45
46
47
48
49
50
51
52
Ohio. Taken from Table 5-1 in Ohio EPA's 'Standardized Biological Field Sampling and
Laboratory Methods for Assessing Fish and Macroinvertebrate Communities' (1989).
ill. Hie!
„ tfl 5
1"*1; * *'
He d icq
I
' t f ,
1. Tn!fil	u4 rrf, l u-:,4
2	. hull* tit Odr	6vilri
X	SriPr-,'1
3	Noo'icr w	ips.'. t:
NtitrOir o- Ik*s4w*-; «r-, pr'fi
A . ilimbc" et •jbi.l'er l;w> - r>-
'luiitu** i,f Winn-'* iptu'it'L
t. Wunti-f L>1 Ii'U.ie an, in"- =-
•innf «<• rf sprint* ivr cper
I , % b i L -1' j'
% ICkr.i'-T
j , % 3fc! i t ^ bf 4 S
f. X "r,- i'i I ivurriiL ^wp-lrics
31 " r.iff t > v rt u k yn t *
V *	.-r'-'H"'e
% i-ivM'siwP«ri*\
'ti hjr«ic „iv5 '.*« ' f!ti i' 1
1 , % 1Kb'' J J
X S'K.L ft' S.'lUOtT 1 S
Niion«-'- r! ' np I »• *i-.hajiM -> :	' •«
(' % .-i *< t'J I "111 N ' C Jtd 1}
X tin *• 3r- •"<:•/
1 jfn'.if". tr ilff*. W. tr,	uri-is	?^ar ?0 . it 5,
¦" the.*. sltei A'tf i£"'l' tf-'i fc-tl'niy iv i *: i?U	Itei t i ait i?rwt»-ri
With he,i1, #*»*(",ifi , 1 tii'tJc*1- M'-'tU £p*rJ f,; ^ i'.ilnnc"	'•i ^
*> im I tji». i l. • I 41 "-I m thf <3f.»r,i k-; n">U-1'ur , f: 
f r smi , ¦¦, ,	jd tr 5 vi\ i uf t f r i Cn-..-( >'i'"	H-cmpr! llrr,	end	' *—sreti l«". i-
Figure Hl-2. Index of Biotic Integrity metrics used to evaluate wading sites, boat sites and
headwaters stream sites in Ohio. Original metrics from Karr (1981) are given first with
substitute metrics following. Taken from Table 4-1 in Ohio EPA's 'Standardized Biological
Field Sampling and Laboratory Methods for Assessing Fish and Macroinvertebrate
Communities' (1989).
x
Is
X
*
X
*
*
*
X
X
I
H
*
X
X
X
X
*
A
X
X
I
*
I
X
X
X
X
X
K
X
*
H2 OHIO EPA REGIONAL REFERENCE DATABASE - BACKGROUND
Ohio was one of the early states to systematically use biological assemblage data to
determine aquatic life use designations and assess the condition of those uses dating back to the
H-3

-------
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
late 1970s. Ohio implemented standardized sampling methods for biological assessments early
on (late 1970s) hence their data represent a nearly thirty year span of standardized biological data
for two assemblage groups. From the late 1970s to 2006 the Ohio fish assemblage database
represents >10,000 unique sites and >24,000 unique sampling events; macroinvertebrate
assemblage data were also collected at most of these same sites. Qualitative Habitat Evaluation
Index (QHEI) data has also been included at the fish sites (Ohio EPA 2006; Rankin 1995, 1989).
While the QHEI is visually based, recent analyses have shown it to be as precise as a quantitative
habitat assessment tool to which it was compared (Miltner et al., 2009; Rankin, in preparation).
The purpose of analyses presented here is to analyze any changes in the reference dataset that
could represent signal or lack of signal related to the effects of global climate change.
In the 1980s and with assistance from the U.S. EPA, Office of Research and
Development, Ohio EPA began a focused sampling of least impacted reference sites in order to
determine the efficacy of level III ecoregions (Omernik, 1987) as a way to account for and
stratify natural variations in biological assemblages (Yoder ,1989; Ohio EPA, 1987a; Whittier et
al., 1987). Ohio EPA used this and other sampling data to establish a network of "least
impacted" regional reference sites that eventually supported the derivation of numerical
biocriteria for Ohio streams and rivers. This was also accomplished across all practically
sampleable stream and rivers from >1 mi2 up to the largest inland rivers ( 6000-8000 mi2). This
includes both wadeable and non-wadeable. Fish assemblage indices were stratified by three
stream- and river-size strata; headwater streams (<20 mi2), "wadeable" streams (20 --300 mi2),
and "boatable" (i.e., non-wadeable) rivers (>-150-200 mi2) (Yoder and Rankin, 1995).
Macroinvertebrate assemblage indices were calibrated continuously across the entire range of
stream and river sizes. The initial reference dataset was developed from a statewide network of
about 300 reference sites that was sampled over a ten year period (1980-89; Table H2-1). That
reference site network was maintained and expanded with the initial re-sampling during 1990-99
and a second re-sampling that will be completed at the end of 2009 (2000-09). Data on habitat
quality (QHEI), water quality, and other physical data such as temperature were also collected
and were based on multiple grab samples collected during "normal" seasonal flows within a
summer-fall seasonal index period (mid-June through mid-October).
H-4

-------
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
Table H2-1. Summary of Ohio EPA regional reference site network including original
sites (1980-89) and updates via first (1990-99) and second round re-sampling (2000-06)
that were used in data ana
yses.
Reference Network
Size Type
Fish: Latest (All
Data)
Macroinvertebrates

Headwaters
112/225

Original Reference Sites:
1980-89 (Sites/Samples)
Wadeable
166/399
242
Boatable
97/254


Headwaters
115/(149)/l 50 (296)

New Reference Sites:
1990-2006 (Sites/Samples)
Wadeable
184(231)/281(539)
309 (525)
Boatable
68(84)/127(278)

H3 DATA ANALYSES
Our primary goal is to examine the Ohio reference database for trends and the entire
dataset for candidate indicators of climate change. Important effects of climate change include
changes in not only temperature, but also rainfall patterns and resulting hydrological regimes in
Ohio streams and rivers thus we also explored the usefulness of using the QHEI as an indirect
measure of hydrological change.
H3.1 Trends in Ohio Reference Sites
We conducted an initial exploration of Ohio's reference database to determine whether
biological reference condition has changed overtime since 1980. This analysis directly
overlapped with an Ohio EPA-sponsored effort to conduct the initial data analysis steps for the
recalibration of the Ohio biocriteria (Rankin, 2008). Of particular importance to our analyses is
the examination of trends in biological condition at the reference sites and exploring the potential
causes that are associated with the observed changes. As such it is essential to understand the
environmental changes that have also occurred that could potentially confound any signals of
climate change-related effects. Based on nearly thirty years of intensive watershed assessments
Ohio EPA has identified a variety of environmental changes that are associated with shifts in
biological condition at the assemblage and species/taxa levels. Such environmental changes
H-5

-------
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
include, 1) a reduction in point source loadings (particularly important in non-wadeable rivers
where some reference sites are necessarily downstream of point sources), 2) changes in land uses
(e.g., increased urbanization), 3) changing loadings of pollutants from agricultural lands (e.g.,
declining sediments and nutrients in response to increased conservation tillage), 4) habitat
changes (e.g., loss of habitat quality from agricultural drainage practices [common],
suburbanization [common], improved habitat quality resulting stream restoration [rare and
localized]), and 5) potential climate change related influences from changes to the temperature
and/or hydrological regimes. These latter changes may be the most difficult to detect due to the
lack of readily available long-term data for temperature and flow and the indirect actions of any
adverse impacts. It is first important to identify any methodological differences in data collection
(environmental and biological) that could either confound or mask apparent trends. In the Ohio
dataset this is most likely represented by taxonomic refinements from an improving resolution in
the identification of macroinvertebrates over the past 30 years. Thus we included some initial
explorations and recommendations related to this factor for the Ohio data set. We focused
primarily on the mayflies because they are an important component of the Ohio ICI, taxonomic
refinements are known to have occurred, and taxonomic refinements would be expected to
influence multiple metrics (total taxa, mayfly taxa, qualitative EPT taxa, etc.).
H3.2 Taxonomic Analyses
We used the entire Ohio database to identify "earliest" and "latest" years for all taxa in
order to extract a list of possible taxa that could affect ICI scoring via taxonomic refinement
(splitting or lumping of taxa). We focused on the mayfly taxa at reference sites and identified
taxa and sites that occurred in the original reference sites, but not the new sites and vice versa.
Table H2-1 lists all mayfly taxa collected at the Ohio reference sites that appeared earlier and
then "disappeared" ("earlier") or those that "appeared" later, mostly at re-sampled reference sites
("later"). We then conferred with senior Ohio EPA taxonomists (Mike Bolton and Jack Freda,
Ohio EPA) and determined whether any of these taxa are purely a result of taxonomic changes
made in the intervening time. These taxa were identified (Table H3-1) and the ICI recalculated
with the same taxon designations as for the original references sites in order to attribute any
changes in the total taxa metric, the mayfly metric, and the qualitative EPT metric to observed
changes in the ICI. This effort primarily consisted of "lumping" individual taxa designations of
mayfly taxa back to "Baetis sp." or "Pseudocloeon sp." (Table H3-1).
H-6

-------
Table H3-1. Mayfly taxa from reference sites in Ohio that abruptly appeared (Later) or
disappeared (Earlier) in the Ohio dataset and explanation of change. Explanations were
provided by Mike Bolton and Jack Freda of Ohio EPA.	
Taxa
Code
Tax on Name
Appear
-ence
Explanation of Change
11010
Acentrella sp
Later
Improved taxonomy allow this taxa to be distinguished
Pseudocloeon sp.
11014
Acentrella turbida
Later
Improved taxonomy allow this taxa to be distinguished
from Pseudocloeon sp.
11015
Acerpenna sp
Later
Improved taxonomy allow this taxa to be distinguished
from Baetidae sp.
11018
Acerpenna
macdunnoughi
Later
Improved taxonomy allow this taxa to be distinguished
from Baetidae sp.
11020
Acerpenna pygmaea
Later
Improved taxonomy allow this taxa to be distinguished
from Baetidae sp.
11110
Acentrella parvula
Later
Improved taxonomy allow this taxa to be distinguished
from Pseudocloeon sp. or was renamed from
Pseudocloeon parvulum
11115
Baetis tricaudatus
Later
Improved taxonomy allow this taxa to be distinguished
from Baetidae sp.
11118
Plauditus dubius
Later
Improved taxonomy allow this taxa to be distinguished
Pseudocloeon sp.
11119
Plauditus dubius or
P. virilis
Later
Improved taxonomy allow this taxa to be distinguished
Pseudocloeon sp.
11120
Baetis flavistriga
Later
Improved taxonomy allow this taxa to be distinguished
from Baetidae sp.
11125
Pseudocloeon
frondale
Later
Improved taxonomy allow this taxa to be distinguished
from Baetidae sp.
11130
Baetis intercalaris
Later
Improved taxonomy allow this taxa to be distinguished
from Baetidae sp.
11150
Pseudocloeon
propinquum
Later
Improved taxonomy allow this taxa to be distinguished
from Baetidae sp.
11155
Plauditus
punctiventris
Later
Improved taxonomy allow this taxa to be distinguished
Pseudocloeon sp.
11175
Plauditus virilis
Later
Improved taxonomy allow this taxa to be distinguished
Pseudocloeon sp.
11250
Centroptilum sp
(w/o hindwing pads)
Later
Improved taxonomy allow this taxa to be distinguished
Cloeon sp.
11400
Centroptilum sp or
Procloeon sp
(formerly in Cloeon
Earlier
Improved taxonomy allow this taxa to be distinguished
Cloeon sp.
11430
Diphetor hageni
Later
Improved taxonomy allow this taxa to be distinguished
from Baetidae sp.
11503
Heterocloeon
curiosum
Later
Renamed Heterocloeon (H.) sp, Heterocloeon sp.
11600
Paracloeodes sp 1
Later
Improved taxonomy allow this taxa to be distinguished
from Paracloeodes sp

-------
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
Table H3-1. Continued
11625
Paracloeodes sp 3
Later
Improved taxonomy allow this taxa to be distinguished
from Paracloeodes sp
11645
Procloeon sp
Later
Was earlier classified as Centroptilum sp or Cloeon sp
11650
Procloeon sp (w/
hindwing pads)
Later
Was earlier classified as Cloeon sp
11651
Procloeon sp (w/o
hindwing pads)
Later
Was earlier classified as Centroptilum sp
11670
Procloeon irrubrum
Later
Improved taxonomy allow this taxa to be distinguished
from Cloeon sp
11700
Acentrella sp or
Plauditus sp
(formerly in
Pseudoc
Earlier
Renamed as Pseudocloeon sp
13010
Leucrocuta hebe
Earlier
Renamed as Heptagenia hebe
13030
Leucrocuta
maculipennis
Earlier
Renamed as Heptagenia maculipennis
14501
Leptophlebiidae
Earlier
Now coded as Leptophlebia sp
14900
Leptophlebia sp
Later
Leptophlebia sp
14950
Leptophlebia sp or
Paraleptophlebia sp
Later
Small specimens lumped
H3.3 Weighted Stressor Values (WSVs)
Candidate fish and macroinvertebrate taxa that could serve as indicators of climate
change (sensitive to temperature or other measures such as hydrological stressors) were
determined from weighted stressor values (WSVs) and "Taxa Indicator Values" (TIVs) for
temperature and habitat measures that would be correlated with hydrological alterations. The
WSVs were generated by relating historical taxa/species from sites in Ohio to chemical and
habitat stressors and calculating weighted average values for each taxa/stressor combination
where the weighting is the relative abundance of the taxa/species at a site. TIV values for taxa
were then ranked from most to least sensitive for each of the pertinent parameters and converted
to an ordinal scale of 1-10 where 1 is the most sensitive and 10 the most tolerant following the
methodology of Meador and Carlisle (2007). WSVs were then plotted vs. a simple means code
by Ohio taxa/species tolerance designations to identify the indicator taxa that occur at the
extremes of the distributions.
H3-4 QHEI Data
QHEI includes the habitat attributes of substrate, cover, channel, riparian, pools, riffle,
and stream gradient (Rankin 1995, 1989). Recent analyses of the QHEI shows it to be relatively
H-8

-------
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
precise (Miltner et al., 2009), and it has been collected by trained professionals since its
inception by Ohio EPA. We used a subset of the metric components to create a sub-index
(Hydro-QHEI) that extracts the habitat attributes that are responsive either directly (current speed
components) or indirectly (stream depth measures) to alterations of the flow regime. Scoring
calculations for the Hydro-QHEI are detailed in Table H3-2. Hydro-QHEI ranges from 0 to 25
and includes the two QHEI subcomponents most related to hydrology, current and depth. We
used the Hydro-QHEI and its two subcomponents to detect any trends in these components over
time as evidence for potential effects from hydrological alterations. We also calculated WSVs for
these components to identify taxa/species that could be sensitive to hydrological changes in
Ohio.
Table H3-2. Sub-components of the Ohio QHEI which were used to score a Hydro-
QHEI and current and depth sub-scores
Current Metric
Depth Metric
QHEI Current Attribute
Score
QHEI Depth Attribute
Score
Very Fast Current
+5
Deep Pools (Cover Metric)
+4
Fast Current
+3
Pool Depths > lm
+4
Moderate Current
+2
Pool Depths 0.7 - 1.0 m
+3
Slow Current
+ 1
Pool Depths 0.4 - 0.7 m
+2
Eddies
+2
Pool Depths 0.2 - 0.4 m
+1
Very Deep Riffles
+3
Pool Depths < 0.20
-1
Moderate Depth Riffles
+ 1
Deep Riffles
+3
Interstitial Flow
-1
Moderate Riffles
+2
Intermittent Flow
-3
Shallow Riffles
+1


Riffles Absent or Non-
functional
-1
H4 RESULTS
H3.1 Potential Trends in Ohio Reference Sites
Some of the following analyses were conducted for Ohio EPA in an initial assessment
towards re-calibrating Ohio EPA's biocriteria based on data after 1988 (Rankin, 2008). Ohio's
original reference site data was collected between 1978 and 1988. Table H4-1 summarizes the
ranges of years that represent the universe of original and re-sampled reference sites. For
analyzing trends in reference sites we used the latest data available for calculating updated
H-9

-------
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
biocriteria statistics. On average the latest data period was 13-16 years after the mean of the
original reference sample dates (Table H4-1).
Table H4-1. Average and range of years represented by original reference site data and
re-sampled (latest) data by index and stream size category pertaining to fish samples
Index/Stream Size
Mean Year Sampled (Range)
Original Reference
Sites
Re-Sampled Sites
ICI - All Sites
1984
(1980-1988)
2000
(1989-2007)
IBI - Headwaters
1984
(1978-1988)
2000
(1989-2006)
IBI - Wading
1984
(1979-1988)
2000
(1990-2006)
IBI - Boat
1984
(1979-1988)
1997
(1990-2005)
Table H4-1 reports the original biocriteria values and statistics, a re-calculation of those
statistics using refined variables, and "new" biocriteria values based on the latest re-sampled
reference sites. Because possible IBI or ICI scores based on single samples are always even
values, calculated percentile values were rounded upwards (e.g., 41 to a 42). Discrepancies
between the original calculations and our recalculations are highlighted in yellow. The original
biocriteria statistics were re-calculated in the database because there are a few minor
discrepancies related to uncertainties about the exact membership of the original reference sites
and gradual changes made to the database since 1990 due to changing taxonomy and a more
precise calculation of drainage area (Rankin, 2009).
The direction of change in the biocriteria between the original and latest reference site
data was either positive (an increase) or neutral (no change) with only three instances where the
new biocriteria were lower. These included: 1) the ICI biocriterion for the non-acidic mine
drainage modified use (-4 pts; possible small sample size); 2) the IBI for WWH headwater site
type in the EOLP ecoregion (-2 pts); and, 3) the IBI for WWH headwater site type in the WAP
ecoregion (-2 pts). None of these changes are considered to be greater than the non-significant
departure for each index.
H-10

-------
211	Table H4-2. Original Ohio biocriteria (O), recalculated biocriteria (R) using similar sites,
212	and new biocriteria (N) using the latest data from re-sampling of original reference sites.
213	Sites with discrepancies between original and recalculated criteria are highlighted in yellow
Ecoregion
Modified Warmwater Habitat (MWH)
WWH
EWH
Channelized
Non-Acidic
Mine
Drainage
Impounded
IBI - Headwater Site Type

O
R
N
O
R
N
O
R
N
O
R
N
O
R
N
HELP
20
20
26


28
-
-
50
50
52
IP
24
24
26
40
40
40
EOLP
40
38
36
WAP
24
24
a
44
44
42
ECBP

40
40
44
IBI - Wadeable Site Type
HELP
22
22
22


32
-
-
50
50
52
IP
24
24
30
40
40
44
EOLP
24
24
30
38
38
42
WAP
24
24
30
24
24
32
44
44
46
ECBP
24
24
30

40
40
40
IBI - Boatable Site Type
HELP
20
20
20



22
22
26
34
30
34
30
34
48
48
52
IP
24
24
24



30
28
34
38
38
47
EOLP
24
24
24



30
28
34
40
40
46
WAP
24
24
24
24
24
26
30
28
34
40
40
40
ECBP
24
24
24



30
28
34
42
42
42
Mlwb - Wadeable Site Type
HELP
5.6
5.9
6.4

7.3
-
-
9.4
9.4
9.5
IP
6.2
6.4

8.1
8.1
8.1
EOLP
6.2
6.4

7.9
7.9
8.2
WAP
6.2
6.4

5.5
4.7
6.1

8.4
8.3
8.8
ECBP
6.2
6.4


8.3
8.3
7.8
Mlwb - Boatable Site Type
HELP
5.7
5.7
7.5a

5.7
5.7
7.4
8.6
-
-
9.6
9.6
10.2
IP
5.8
5.7
6.1a
6.6
7.0
7.5
8.7
8.7
9.6
EOLP
5.8
5.7
6.1a
6.6
7.0
7.5
8.7
8.8
8.9
WAP
5.8
5.7
6.1a
5.4
5.4
6.4
6.6
7.0
7.5
8.6
8.6
9.2
ECBP
5.8
5.7
6.1a

6.6
7.0
7.5
8.5
8.5
9.7
H-ll

-------
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
Table H4-2. Continued
ICI - All Site Types Combined
HELP
22
22
24


34
34
42
46
46
50
IP
22
22
24
30
30
38
EOLP
22
22
24
34
34
44
WAP
22
22
24
O
CO
o
CO
36
36
40
ECBP
22
22
24

36
36
42
a - Non-acidic mining influenced modifiec
sites due to small sample size.
sites for headwaters com
jinec
with wading
The direction of climate-related changes in biological index scores could be in either
direction. However, the most plausible expectation would be for a decline due to the immediate
loss of highly intolerant species and taxa (i.e., temperature- and flow-sensitive taxa/species) and
a co-occurring increase in intermediate, moderately, and/or highly tolerant taxa/species. Such
expectations are supported by our analyses that identify a general concordance between
intolerant and sensitive species as categorized for the IBI and ICI and species sensitive to
temperature and habitat features indicative of altered flow conditions.
The largest positive changes in the biocriteria were in the WWH boatable fish sites (IBI
and Mlwb) and in the WWH ICI. The fish assemblage changes in large rivers are most
attributable to reduced pollution from point sources, mostly due to municipal wastewater
treatment plant upgrades after 1988 (Yoder et al., 2005). While it was necessary in the derivation
of the original Ohio IBI for boatable sites to include reference sites located in effluent dominated
rivers, the sites were positioned below known recovery points. Nevertheless, the lessening of
secondary impacts from nutrient enrichment by the aforementioned controls had positive effects
on the fish assemblages at these reference sites. Taxonomic changes in fish nomenclature did not
influence IBI scores between these time periods nor did the fish sampling technology as the
methodology and equipment was generally stable between these time periods.
H4.2 Influence of Taxonomic Changes on Trend Assessment in Ohio
The question concerning the relative contribution of taxonomic changes to the
macroinvertebrate assemblage trends in the Ohio biocriteria values at reference sites was also
examined during this phase of the data analysis. While fish data can be influenced by factors
such as sampling efficiency, their taxonomy has been comparatively stable during the period
over which the Ohio reference database was developed. As for sampling methodology, methods
used by Ohio EPA for both fish and macroinvertebrates have been stable over the period of the
Ohio reference database. However, there have been significant changes in macroinvertebrate
H-12

-------
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
taxonomy over this time period mostly in the form of improved discrimination within certain
genera (e.g., Baetid mayflies) that could result in changes to the ICI "number of' metrics for
mayflies and other taxonomic groups that are also identified to more refined taxonomic
resolution.
We developed a program to scan the Ohio EPA database and identify taxa that may have
been revealed by improved taxonomy which would result in two or more taxa in lieu of a single
taxon. This program resulted in a listing of all taxa and the first and last occurrence of each taxon
in the Ohio EPA database. We then focused on the taxonomic changes in mayflies to examine
the quantitative contribution of the refined taxonomy on ICI scoring for three metrics; total taxa,
mayfly taxa, and qualitative EPT taxa. We then recalculated the mean number of taxa for each
metric as it now occurs in the database ("refined" taxonomy) and then again with the taxonomy
"lumped" to match the level of taxonomy that was prevalent during the derivation of the original
biocriteria (Table H4-2). We also recalculated the biocriteria statistics (25th percentiles by
ecoregion for WWH; 75th percentiles statewide for EWH) based on the newly refined and
lumped taxonomy (Table H4-3).
The recalculation of ICIs from all sites indicated a 5.9 point increase in the mean ICI
score between the two time periods. When mayfly taxonomy was lumped between these time
periods the increase was 5.0 showing that taxonomic refinement in mayflies accounted for 14%
of the increase in the mean ICI between the two reference time periods (Table H4-3). Only two
cases showed a change in the biocriteria the HELP WWH biocriterion (38.5 compared to 42) and
the EOLP WWH biocriterion (42 compared to 44).
The changes in mayfly taxonomy reflect the greatest influence on ICI scoring in the Ohio
database; other taxa would likely have a lesser impact compared to the impact on mayfly metrics
(Jack Freda, personal communication). Future work should isolate all of the other taxonomic
refinements that could confound trends in metrics and index scores. Comparisons of similarity of
macroinvertebrate taxonomy in samples between European countries concluded that taxonomic
adjustments prior to analyses of the separate data sets reduced species richness from 45 to 81%
by country and 85% for all countries combined (Verdonschot and Nijboer, 2004). We are dealing
with much smaller changes in the Ohio database.
H-13

-------
276
Til
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
Table H4-3. Changes in ICI and mayfly influence ICI metrics related to increasing
taxonomic resolution over time in the Ohio EPA least impacted reference data set
Metric
Original Reference Sites
New Reference Sites
(Latest Data)

Standard
Taxonomy
Mean Taxa
(Mean Score)
Lumped
Taxonomy
Mean Taxa
(Mean Score)
Standard
Taxonomy
Mean Taxa
(Mean Score)
Lumped
Taxonomy
Mean Taxa
(Mean Score)
Total Taxa
35.97 (4.89)
35.93 (4.89)
38.36 (5.18)
37.65 (5.04)
Number of Mayfly Taxa
6.95 (4.20)
6.90 (4.17)
7.42 (4.59)
6.59(4.16)
QUAL EPT Taxa
11.29 (3.63)
11.24 (3.60)
15.16 (5.16)
14.23 (4.91)
ICI Score
39.59
39.53
45.35
44.56
Table H4-4. Table of original and recalibrated Ohio biocriteria with adjustments
made to equilibrate taxonomic advances made in the later time period. Highlighted
cells indicate where standardizing taxonomic resolution would have resulted in altered
criteria.

Warm water Habitat

Exceptional Warm water




Habitat





Latest


Latest



Reference


Referen



w/


ce w/

Original
Latest
Refined
Original
Latest
Refined

Referen
Referen
Taxonom
Referen
Refere
Taxono
Ecoregion
ce
ce
V
ce
nee
my
HELP
34
42
38.5



IP
30
38
38



EOLP
34
44
42
46
50
50
WAP
36
40
40



ECBP
36
42
42



H4.3 Weighted Stressor Values (WSVs)
We calculated WSVs for maximum temperature and Hydro-QHEI variables separately
for headwater streams (drainage area <20 mi.2) and wadeable streams (drainage area >20 to 300
mi.2). These data are ordered by WSV for each parameter to provide a sequential listing of
sensitive species/taxa that can be used to detect trends in relation to temperature or flow
alterations. It also provides a listing of tolerant species that might increase in predominance if
temperature were to increase or the hydrological regime became increasingly variable.
H-14

-------
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
H4.4. Temperature
We used the maximum temperature recorded from summer-fall grab samples collected
during the same period within which the biological data were collected to calculate WSVs for
headwater and wadeable streams. To visualize the distribution of these data with taxa
sensitivities we plotted the means of these values vs. the weighted means (WSVs) color coded by
the existing taxa tolerance rankings of Ohio EPA (Figure H4-1). Because the temperature
indicator was derived from a small number of grab samples, the precision of these data could be
rather low for a given site. However, when aggregated across the temporal and spatial extent of
Ohio EPA database we expect that relationships between taxa relative abundance and maximum
summer temperatures should be much more representative of taxa sensitivities. Figure H3
represents plots of WSVs based on maximum temperatures (°C) from grab samples at sites with
macroinvertebrate taxa collected from artificial substrates in headwater and wadeable streams.
The WSVs for maximum temperature generally track with the "general" tolerance categories
assigned by Ohio EPA for each taxon for both headwater (Figure H4-1, upper right) and
wadeable streams (Figure H4-1, lower right). A similar pattern was observed for fish species.
WSVs for temperature can be confounded with WSVs for other stressors, particularly habitat.
However, the extremes of these distributions can be useful for identifying possible indicator taxa
for future applications.
It is interesting to note that selected Chironomidae taxa occurred at both extremes of the
WSV for temperature. For example, Paratanytarsus n.sp 1 had the lowest WSV for temperature
at wadeable sites and Parachironomus "hirtalatus" and Tanypus neopunctipennis had among the
highest WSVs (Figure H4-1, lower left). Additional analysis using environmental traits could
help in determining the rare taxa that could exhibit some sensitive traits, but which may be too
rare by themselves to serve as useful indicators.
H4.5 Hydro-QHEI
We generated WSVs for Hydro-QHEI variables separately for headwater and wadeable
streams for both fish and macroinvertebrates. We plotted several examples of the WSVs for
these variables vs. the simple means for these same variables (Figure H4-2) in order to reveal the
distributions of tolerant and sensitive species along this gradient as we did for temperature. Fish
and macroinvertebrate WSVs for Hydro-QHEI and its subcomponents tracked relatively closely
to the Ohio EPA tolerance designations for macroinvertebrate taxa and fish species (Figure H4-
2). Outlier points and variability are often associated with small sample sizes for a given species
at a given stream size. Intolerant species are frequently rarer than "sensitive" species, especially
H-15

-------
328	so for fish, and as such may exhibit more variation than "sensitive" species where sample sizes
329	are typically larger. As expected, tolerant species generally have wider sensitivity ranges.
330
331
332
30
U
25
u
"O Q_
£ w
CT) I—
w E
20
15
H/S\/s for Maximum Temperature
Headwater Streams
Planorbella (Pier
pilsbryi
Chironomus (C.)
jsoma)
\ *•«

30 Stations Jnionids> 5 Stations - Baetis tricaudatus ¦ ¦¦ • Acerpenna macdunnoughi # Very T olerant ~ T olerant ~ Moderately T olerant * Intermediate ¦ Moderately Intolerant • Intolerant dius) V 15 20 25 Mean Maximum Temperature (C) 30 30 > w u 28 _ Q- G ^ V £ 26 a D L. +- _Q g w 24 W L_ Q- W e i I— 9 22 cj . 20 18 WSVs Tempera fure Headwa fer Macroin ver tebra te Da fa ¥ ¦ Very Tolerant Moderately Inter- Moderately Intolerant Tolerant Tolerant mediate Intolerant ICI Taxa Tolerance Categories 30 WSVs for Maximum Temperature Wadeable Streams Anthqpotamus sp- Laevapex fuscus ~ - \Ablabesmyia annutafa — Tanypus neopunctipennis ^ Parachironomus "hirtalatus" Polypedilum (Pentapediturn) tritum var. I M > 5 Stations - Atherix lontha - Plauditus punctiventris ' Paragnetina media . Acerpenna macdunnoughf. ! \ Paratanytarsus n. sp 1 Very T olerant Tolerant Moderately T olerant Intermediate Moderately Intolerant Intolerant 15 15 20 25 Mean Maximum Temperature (C) 30 30 > w 28 O ^ 26 s. 24 22 J .E I- p 20 18 16 WSVs Tempera fure Wadeable Macroinvertebrate Data ,, , , | ° i ¦ o i fl [. o rh * T ~! m ¦I T 8 ' ° i 0 ,, , , i a Very Tolerant Moderately Inter- Moderately Intolerant Tolerant Tolerant mediate Intolerant ICI Taxa Tolerance Categories Figure H4-1. Plots of macroinvertebrate taxa maximum temperature WSV values vs. mean maximum values for taxa for headwater streams (upper left) and wadeable streams (lower left) and box and whisker plots of WSVs for maximum temperatures by Ohio EPA macroinvertebrate tolerance values (derived for the ICI) for headwater streams (upper right) and wadeable streams (lower right). Data for taxa represents data collected from artificial substrates where at least five samples were represented for each stream size category. H-16


-------
LU
31
a
~a
>-
I
-a
u
w
25
20
15
10
WSVs for Hydro-QHEI
Headwater Streams
Chimarra obscura
Ptychomyia fiavidc
- - • - - - - - - - - P/pudttus -dubius
• Hydropsyche frisct
> ! I
N> 30 Stations
Unionids > 5 Stations
Very Tolerant
Tolerant
Moderately T olerant
Intermediate
Moderate ly Intolerant
Intolerant
5	10	15
Mean Hydro-QHEI
20
WSVs Hydro-QHEI
Headwater Macroinvertebrate Oaf a
V)
O
20
- ft)
^ Q_
I/) 20-200 sq mi)
LU
31
a
o
CL
d
>
31
Tolerant
Moderately Tolerant
Intermediate
Sensitive
Intolerant
N > 5 Stations
Tonguetied Minnow _
~
v	r, ,	Brown Trout
Tippecanoe Darter			
4 Streamline Chub
~ Longnose Dace
~ American Brook Lamprey
-Variegate Darter
Eastern Sand Darter
Goldfish
-athead Minnow
til a;kst ripejTopm innow
Mean HYDRO-QHEI
25
>
^ _
LLJ
20
I/)
I
0
-a
>
1
15
10
WSVs Hydro-QHEI (Overall Score)
Wadeable Fish Data

Intolerant Sensitive Inter mediM"@der ate ly To leranToler ant
Figure H4-2. Scatter plots of taxa/species Hydro-QHEI WSV values vs. mean
Hydro-QHEI values for macroinvertebrates taxa for headwater streams (upper
left) and for species in wadeable streams (lower left) and box and whisker plots
of macroinvertebrate (upper right) and fish (lower right) WSVs for Hydro-
QHEI for these waters. Data from Ohio EPA.
H-17

-------
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
Although QHEI is a visual habitat tool, recent analyses of variation from sites with multiple
QHEI values using signal/noise ratio analyses indicate the index is precise and the
subcomponents are moderately precise to precise (Miltner et al., 2009 Draft; Rankin et al., in
preparation). We chose Hydro-QHEI subcomponents that are expected to change in response to
flow alterations. For example the presence of fast current or the presence of eddies is a
characteristic of permanent summer base flows (QHEI assessments are generally conducted
during summer-fall low flow periods). Habitat attributes related to depth (i.e., deep pool and
deep runs) are also associated with permanent base flows. Thus the Hydro- QHEI is expected to
reflect a gradient of base flow stability, one of the attributes that would be expected to change
with changes in precipitation patterns as a result of climate change. Sensitive fish species and
macroinvertebrate taxa were positively correlated with the Hydro-QHEI, thus it promises to be a
useful tool for indicating hydrological changes that may be associated with climate change.
These data are commonly collected by states throughout the Midwestern U.S.
H4.6 Species Distribution by Stream Size
The identification of certain intolerant fish species in headwater streams at the "sensitive"
end of the Hydro-QHEI gradient suggests that the distribution of these species at the tails of their
preferred stream size range may reflect the degree of base flow. Fish species such as streamline
chub, variegate darter, river chub and stonecat madtom (all with high Hydro-QHEI WSVs) are
generally found in larger wadeable streams and their presence in headwater streams is associated
with high Hydro-QHEI scores that indicate more stable flow regimes. Year-to-year or long-term
trends of these species in headwater streams could represent a response to climate-induced
hydrologic changes. Thus we suggest that this could be an opportunity to explore whether the
stream size "tails" of sensitivity distributions shift with hydrological change.
The Ohio database does contain a stream-size bias because headwater streams were less
frequently sampled in the 1980s than in the 1990s and 2000s. With the knowledge of this bias as
a test of the ability to detect species distribution changes at the edge of their distribution we
divided the dataset into three time periods and examined whether a suite of sensitive species
distributions along stream size was apparent through time. We recognized that the distribution of
sites was different between these periods and we wanted to test whether it would be evident in
low percentiles (1st, 5th, and 25th) for species distributions across all sites in Ohio. The results of
this initial test showed that some bias between time periods exists for species distributions where
nearly all selected sensitive species had distributions that extended further into small streams
during the later (1998-2008) compared to the earliest (1978-1989) sampling periods (Table H8).
H-18

-------
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
In this table species with a + "increased" their distribution in small streams sampled in the most
recent years (Ohio EPA, 2002).
We then restricted this analysis to sites that had only been sampled in all three sampling
periods so that the resulting distributions were not an artifact of stream size bias. The distribution
of each species was then examined along a stream size gradient as measured by the same low
percentiles (1st, 5th and 25th) (Table H4-5). There is still a possible bias in this initial analysis
because some of these sites that were sampled across all three periods may have been sampled
more frequently during some periods which could increase the probability of capture. However,
as an initial exploratory analysis we were interested in whether any obvious trends were
apparent.
The results (Table H4-6) do not indicate evidence of the same patterns similar to what
was evident in Table H4-5 that were attributable to the sampling frequency among small streams.
This analysis assumes, however, that some strong long-term shifts would have occurred during
these time periods that would affect the tails of stream size distributions more than inter-annual
flow variation. A more sensitive analysis would control or consider year-to-year variability in
flow or temperature within each time periods that may confound the current analysis. We suggest
that these distributional shifts could be a fruitful path for analysis when annual variation and
regional variation in flows, which can be extracted from USGS flow data using IHA flow
indicators, are incorporated into the analyses. The initial analyses conducted herein establish a
basis for more detailed analyses.
H-19

-------
390	Table H4-5. Analysis of frequency of species collections by stream size as measure by
391	1st, 5th and 25th percentiles of drainage area at sites with these species collected
Sample Size	1st Percentile	5thPercentile 25th Percentile
Species
Code
Species Name
1978- 1938-
1989 2008
1978-
1989
1998-
2008 Trend
1978-
1989
1998-
2008 Trend
1978-
1989
1938-
2008
34-001
central mudminnow |T]
18
1
5.0
9.0 -
5.0
9.0 -
10.8
9.0
40-009
black redhorse [I]
211
153
10,1
28.2 4
35.6
69,8 4
105.3
140.8
40-010
golden redhorse [M]
507
353
7,8
8.0 4
27.6
24.5 4
111.3
122.0
40-015
northern hog sucker [M]
553
424
5.3
6.0 4
10.0
14.5 4
46.7
71.5
43-001
common carp [T]
562
287
11.5
8.4 4
28.4
19.3 4
207.0
130.0
43-004
homyhead chub [I]
37
19
1,5
1.5 4
9.4
4.0 4
32.0
32.8
43-005
river chub [I]
90
73
6,7
41.0 4
33.0
47.9 4
80.0
105.0
43-007
bigeye chub [1]
16
20
15,0
16.9 4
15.6
16.9 4
34.0
56.7
43-012
longnose dace [R]
0
0
0,0
0.0 4
0,0
0.0 4
0,0
0,0
43-014
tonguetied minnow [S]
12
5
34,0
7.5 -
34.0
7.5 4
34.0
27.4
43-017
redside dace [1]
22
17
5.0
5.0 4
5.0
5.8 4
7.5
7.5
43-021
silver shiner [I]
191
148
10,7
13.9 4
32.0
16.5 4
74.0
45.0
43-022
rosyface shiner [1]
133
114
6.8
5.0 4
32.0
9.1 4
79.5
99.0
43-034
sand shiner [M]
111
262
5.9
7.5 4
16.9
16.7 4
51.4
77.0
43-042
fathead minnow [T]
53
51
0.8
0.8 -
2.8
2.5 4
16.0
7.4
43-043
bluntnose minnow [T|
646
449
1.5
1.5 4
7.4
6.8 4
37.0
29.0
47-004
yellow bullhead [T]
296
176
2.1
1.8 4
8.0
9.0 4
32.5
23.0
47-008
stonecit madtom [1]
115
68
20.5
10.9 4
33.0
69.0 4
101.0
105.0
47-012
brindled madtom [I]
42
24
19.7
19.7 4
19.7
29.7 4
32.0
89.0
77-004
smalimouth bass [M]
520
428
11.4
8.7 4
24.5
23.9 4
104.0
125.5
77-008
green sunfish [T]
673
398
2,5
1,5 4
8.0
6.9 4
44.5
29.5
77-011
longear sunflsh |M]
510
339
8.6
6.3 4
28.2
15,2 4
132,0
122.0
80-004
dusky darter [M]
12
27
16.1
4.9 4
21.4
14.4 4
120,5
157.0
80-011
logperch |M]
138
248
1.5
14,3 4
28.5
32.0 4
74.0
151.0
80-011
logperch [M]
138
246
1.5
14.3 4
28.5
32.0 4
74.0
151.0
80-013
eastern sand darter [R]
2
3
288.0
174.0 4
286.0
174.0 4
286.0
202.0
80-015
greenside darter [M]
239
306
8.5
5.9 4
16.0
9.9 4
42.0
41.0
80-015
greenside darter [M]
299
306
8.5
5.9 4
16.0
9.9 4
42.0
41.0
80-022
rainbow darter [M]
267
253
2.5
2.6 4
6.8
8.2 4
28.2
26.5
392
393
H-20

-------
395
396	Table H4-6. Analysis of frequency of species collections by stream size as measured by
397	1st, 5th and 25th percentiles of drainage area at sites with these species collected
Sample Size	1st Percentile	SthPercentile	25th Percentile
Species
Code
Species Name
1978-
1989
1998-
2008
1978-
1989
1998-
2008
Trend
1978-
1989
1998-
2008
Trend
1978-
1989
1998-
2008
Trend
01-006
least brook lamprey [ ]
19
16
4.9
4.9

4.9
4.9
-
16.3
9.6
+
34-001
central mud mi mow [T]
18
1
5.0
9.0

5.0
9.0
-
10.8
9.0
+
4Q-009
black redhorse [I]
211
153
10.1
28.2
¦i
35.6
69.8
-
105.3
140.8
-
40-010
golden redhorse [M]
507
353
7.8
8.0
¦
27.6
24.5
+
111.3
122.0
-
40-015
northern hog sucker [M]
553
424
5.3
6.0
*
10.0
14.5
-
46.7
71.5
-
43-001
common carp [T]
562
287
11.5
8.4

28.4
19.3
+
207.0
130.0
#
43-004
hornyhead chub [I]
37
19
1.5
1.5
M
9.4
4.0
+
32.0
32.8
-
43-005
river chub [I]
90
73
6.7
41.0

33.0
47.9
-
80.0
105.0
-
43-007
bigeye chub [I]
16
20
15.0
16.9
¦i
15.6
16.9
-
34.0
56.7
-
43-012
iongnose dace fRJ
0
0
0.0
Q.O
m
0.0
0.0
-
0.0
0.0
-
43-014
tonguetied minnow [S]
12
5
34.0
7.5

34.0
7.5
~
34.0
27.4
+
43-017
redside dace [I]
22
17
5.0
5.0
m
5.0
5.8
-
7.5
7.5
-
43-021
silver shiner [I]
191
148
10.7
13.9

32.0
16.5
+
74.0
45.0
+
43-022
rosyface shiner [I]
133
114
6.8
5.0

32.0
9.1
+
79.5
99.0
-
43-034
sand shiner [M]
111
262
5.9
7.5

16.9
16.7
+
51.4
77.0
-
43-042
fathead minnow [T]
53
51
0.8
0.8

2.8
2,5
+
16.0
7.4
+
43-043
bluntnose minnow [T]
646
449
1.5
1,5

7.4
6.8
+
37.0
29.0
+
47-004
yellow bullhead [T]
296
176
2.1
1,8
*
8.0
9.0
-
32.5
23.0
+
47-008
stonecat madtom [I]
115
68
20.5
10.9
~
33.0
69.0
-
101.0
105.0
-
47-012
brindled madtom [I]
42
24
19.7
19.7
-
19.7
29.7
-
32.0
89.0
-
54-002
blackstripe topminnow [
74
38
2.6
1.5
+
10.3
2.6
*
32.0
25.0
+
77-004
smallmouth bass [M]
520
428
11.4
8.7
~
24,5
23.9
+
104.0
125.5
-
77-008
green sunfish [T]
673
396
2.5
1.5
+
8.0
6.9
+
44.5
29.5
+
77-011
longear sunfish [MJ
510
339
8.6
6,3
*
28.2
15.2
+
132.0
122.0
+
80-004
dusky darter [M]
12
27
16.1
4,9
*
21.4
14.4
+
120.5
157.0
-
80-011
fogperch [M]
138
248
1.5
14.3
-
28.5
32.0
-
74.0
151.0
-
80-013
eastern sand darter [R]
2
3
286.0
174.0
#
286.0
174.0
+
286.0
202.0
+
80-015
greenside darter [M]
299
306
8.5
5.9
#
16.0
9.9
+
42.0
41.0
#
80-022
rainbow darter [M]
267
253
2.5
2.6
-
6.8
8.2
-
28.2
26.5
+
90-002
mottled sculpin [ ]
107
137
1.5
1.5
-
2.5
4.9
-
9.6
16.7
-
95-001
brook stickleback [ ]
17
10
7.4
7.4
-
8,6
7.4
+
21.0
8.0
+
398
H-21

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Appendix I
Selected subsets of results from correlation
analyses for Maine, Utah and North Carolina
The purpose of this Appendix is to show selected subsets of results from correlation analyses
performed on data for Maine, Utah and North Carolina. Results are presented to allow for
comparisons of trends both within and across states. Numerous traits variables and taxa were
analyzed. The object was to identify biological variables that were significantly correlated with
year, PRISM annual air temperature or PRISM annual precipitation (PRISM variables were
typically the best site-specific climatic variables available). Metrics that are presented in this
Appendix relate to temperature preferences and tolerances, EPT taxa, HBI, OCH taxa, hydrology
and scenario metrics. Additional results are available upon request.
II. Overview
Attachment II.
Attachment 12.
Attachment 13.
Attachment 14.
Attachment 15.
Attachment 16.
Attachment 17.
Attachment 18.
NAO/ONI/PDO
Climate Variable - Year Trends
Year & Climate Variable -Temperature Metric Trends
Year & Climate Variable -EPT Metric Trends
Year & Climate Variable -HBI Trends
Year & Climate Variable -OCH Metric Trends
Year & Climate Variable -Hydrologic Metric Trends
Year & Climate Variable -Scenario Metric Trends
1-1

-------
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
II. OVERVIEW
Several approaches were used in analyzing data for long-term trends. To briefly summarize:
•	We did correlation analyses and ordinations (i.e.Non-metric Multidimensional Scaling
(NMDS) and Canonical Correlation Analyses (CCA)) on several of different subsets of
data from each of the states.
•	We evaluated subsets of data from individual long-term biological sampling sites and
from groups of sites.
•	We evaluated each site and site group for confounding factors (non-climate) that may be
influencing trends. Examples of factors that we evaluated (availability varied) included:
habitat (i.e. width, depth, visual substrate estimates), a variety of water chemistry
parameters, land use/land cover, and organic enrichment (using HBI calculations as
surrogates because long-term nutrient data were generally not available).
•	We used a 'two-pronged' approach and evaluated both taxonomic composition (mainly
using relative abundance) and traits metrics (percent individuals and number of taxa).
Table 11-1 contains metadata for the environmental and biological variables that were included
in the correlation analyses. The climate variables used in the analyses are PRISM mean annual
air temperature and PRISM mean annual precipitation. Variables associated with the North
Atlantic Oscillation (NAO), Oceanic Nino Index (ONI) and Pacific Decadal Oscillation (PDO)
were also analyzed (see Attachment II). The procedure was automated to run in R software (the
R code is available upon request - it produces a correlation matrix, a table with significant
correlations (with option to set the p-value) and plots of the significantly-correlated variables).
Because there are so many results, we selected the most relevant subset of summary tables and
plots to present in this Appendix (all the correlation matrices are available upon request). The
summary tables include side-by-side results from Maine, Utah and North Carolina sites and site
groups so that patterns can be compared across states and regions (see Table 1-2 for a list of the
sites and site groups). The Pearson product moment correlations were calculated using Statistica
software (Version 8.0, Copyright StatSoft, Inc., 1984-2007).
The following groups of results are presented in Attachments 12 through 18:
1-2

-------
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
12.	Climate Variable -Year Trends;
13.	Year & Climate Variable -Temperature Metric Trends;
14.	Year & Climate Variable -EPT Metric Trends;
15.	Year & Climate Variable -HBI Trends;
16.	Year & Climate Variable -OCH Metric Trends;
17.	Year & Climate Variable -Hydrologic Metric Trends;
18.	Year & Climate Variable -Scenario Metric Trends; and
There are a number of limitations that should be noted. Correlation analyses cannot establish
unambiguous causal relationships between the environmental and biological variables. We tried
to disentangle confounding factors from the climate change effects by using reference data, but
some reference stations still are influenced by anthropogenic factors. In addition, significant
correlations can sometimes be driven by outliers. We attempted to address this issue by
reviewing plots of significantly-correlated variables. Another issue is that the climate variables
used (PRISM mean annual air temperature and precipitation), while bearing relationships to in-
stream conditions, are not direct measures of actual stream thermal and hydrologic regimes at the
biological sampling sites. Ideally we would use continuous water temperature and flow data in
the analyses.
Some limitation with the traits analyses include:
•	Experimental evidence regarding which individual traits are most important in the
context of climate change is still lacking, so that application of trait analyses was related
to some published literature, but still requires some 'best professional judgment';
•	Redundancy of information among traits has been cited as an issue (Poff et al., 2006). We
also found a number of individual trait metrics to be correlated (r>0.8). Efforts to limit
impacts of redundancy (Poff et al., 2006) were dataset-dependent, making broad
generalizations about which trait metrics to exclude difficult.
•	We calculated trait metrics (% individual and number of taxa) for about 30 different traits
(which each had 2 to 5 trait states). There were a lot of significant correlations, but
interpretation was difficult, since few showed consistent patterns across sites and states.
1-3

-------
87
88
89
90
91
92
93
94
95
96
97
98
99
100
In selecting which trait metrics to summarize, we used the following guidelines to the
extent possible:
•	Focus on traits for which we have the most amount of information for the most
number of taxa (i.e. functional feeding group (FFG) and habit)
•	Focus on groups of traits rather than individual traits, to the extent support by
available literature information on trait characteristics in various trait categories
by taxon. We approached consideration of combinations of traits by developing
'scenario' traits metrics, where scenarios represent projected future climate
characteristics for a region (e.g., warmer and drier in Utah). Taxa were then
grouped based on suites of traits expected to confer an advantage in surviving
these projected future conditions. "Robust" were groups of taxa with the most
number of favorable traits states for that scenario, while "vulnerable" were groups
of taxa with the fewest favorable trait states and the most number of unfavorable
trait states.
1-4

-------
Table 11-1. Metadata for the environmental and biological variables that were used in the correlation analyses.
Category
Variable
Description

StationID
self-explanatory

Year
year sample was collected

lulianDate
collection (Julian) date of biological sample

Month
month sample was collected
PRISM
tmeanl4
PRISM mean annual air temperature (=avg of tmin and tmax) (°F)
PRISM
tmaxl4
PRISM mean annual maximum air temperature (=avg of tmin and tmax) (°F)
PRISM
tminl4
PRISM mean annual minimum air temperature (=avg of tmin and tmax) (°F)
PRISM
pptl4
PRISM mean annual precipitation (inches)
PRISM
tmeanl4_difc
difference between value from year of sampling event minus value from previous year. Calculation based
on PRISM tmeanl4 data.
PRISM
tmaxl4_difc
difference between value from year of sampling event minus value from previous year. Calculation based
on PRISM tmax 14 data.
PRISM
tminl4 difc
difference between value from year of sampling event minus value from previous year. Calculation based
on PRISM tmin 14 data.
PRISM
pptl4_difc
difference between value from year of sampling event minus value from previous year. Calculation based
on PRISM pptl4 data.
PRISM
tmeanl4_absdifc
absolute difference between value from year of sampling event minus value from previous year. Calculation
based on PRISM tmeanl4 data.
PRISM
tmaxl4_absdifc
absolute difference between value from year of sampling event minus value from previous year. Calculation
based on PRISM tmax 14 data.
PRISM
tminl4_absdifc
absolute difference between value from year of sampling event minus value from previous year. Calculation
based on PRISM tmin 14 data.
PRISM
pptl4_absdifc
absolute difference between value from year of sampling event minus value from previous year. Calculation
based on PRISM pptl4 data.
Taxa
Taxon
relative abundance of taxon
HBI
HBI NM
Hilsenhoff Biotic Index calculated using the tolerance values in the New Mexico database (this was used in
Utah only)
HBI
HBI
Hilsenhoff Biotic Index calculated using the tolerance values from the state being analyzed
Selected Metrics
PlecopPct
Percent individuals in the Order Plecoptera
1-5

-------
102 Table 1.continued...
Category
Variable
Description
Selected Metrics
ChiroPct
Percent individuals in the Family Chironomidae
Selected Metrics
EPTPct
Percent individuals - Ephemeroptera, Plecoptera and Trichoptera
Selected Metrics
PlecopTax
Number of taxa in the Order Plecoptera
Selected Metrics
ChiroTax
Number of taxa in the Family Chironomidae
Selected Metrics
EPTTax
Number of Ephemeroptera, Plecoptera and Trichoptera taxa
FFG
FFGCGPct
Functional Feeding group - percent collector-gatherer individuals
FFG
FFGCFPct
Functional Feeding group - percent collector-filterer individuals
FFG
FFGSHPct
Functional Feeding group - percent shredder individuals
FFG
FFGHBPct
Functional Feeding group - percent herbivore individuals
FFG
FFGPRPct
Functional Feeding group - percent predator individuals
FFG
FFGCFTax
Functional Feeding group - number of collector-filterer taxa
FFG
FFGCGTax
Functional Feeding group - number of collector-gatherer taxa
FFG
FFGHBTax
Functional Feeding group - number of herbivore taxa
FFG
FFGPRTax
Functional Feeding group - number of predator taxa
FFG
FFGSHTax
Functional Feeding group - number of shredder taxa
Habit
HabitCNPct
Habit - percent dinger individuals
Habit
HabitSWPct
Habit - percent swimmer individuals
Habit
HabitBUPct
Habit - percent burrower individuals
Habit
HabitSKPct
Habit - percent skater individuals
Habit
HabitCBPct
Habit - percent climber individuals
Habit
HabitSPPct
Habit - percent sprawler individuals
Habit
Habit CBTax
Habit - number of climber taxa
Habit
HabitCNTax
Habit - number of dinger taxa
Habit
HabitSPTax
Habit - number of sprawler taxa
Habit
Habit BUTax
Habit - number of burrower taxa
Habit
HabitSWTax
Habit - number of swimmer taxa
1-6

-------
104 Table 1.continued...
Category
Variable
Description
Hydro
PerennialPct
Percent perennial stream individuals (list of perennial taxa was based on NC intermittent stream report and
Del Rosario et al. (2000): includes taxa listed in Table 1 of NC report (except for Peltoperlidae, since those
occurred only at the intermittent site in Del Rosario), plus gilled snails (Subclass Prosobranchia) and
Simuliidae (Del Rosario et al. 2000). These taxa, which require water for their entire life cycle, should be
found a later instar larvae. Some are indicators of perennial stream features.
Hydro
IntermitPct
Percent intermittent stream individuals (list based on NC intermittent stream report: amphipods, isopods,
small elongate Dipteran larvae (Ceratopogonidae, Blephariceridae, Chironomidae, Deuterophlebiidae,
Psychodidae) winter stoneflies (Capniidae, Taeniopterygidae), Dytiscidae. Unique to intermittent=Helichus
larvae and Dasyhela (family Dolchopodidae). Rest are also found in perennial. They just tend to be more
dominant in numbers in intermittent conditions (probably due to loss of predators).
Hydro
DroughtPct
Percent individuals that possess at least one of the following traits: ability to survive desiccation, adult ability
to exit, respiration plastron/spiracle
Hydro
PerennialTax
Number of perennial stream taxa (list of perennial taxa was based on NC intermittent stream report & Del
Rosario et al. 2000 JNABS paper: includes taxa listed in Table 1 of NC report (except for Peltoperlidae, since
those occurred only at the intermittent site in Del Rosario), plus gilled snails (Subclass Prosobranchia) and
Simuliidae (Del Rosario et al. 2000). These taxa, which require water for their entire life cycle, should be
found a later instar larvae. Some are indicators of perennial stream features.
Hydro
IntermitTax
Number of intermittent stream taxa (list based on NC intermittent stream report: amphipods, isopods, small
elongate Dipteran larvae (Ceratopogonidae, Blephariceridae, Chironomidae, Deuterophlebiidae,
Psychodidae) winter stoneflies (Capniidae, Taeniopterygidae), Dytiscidae. Unique to intermittent=Helichus
larvae and Dasyhela (family Dolchopodidae). Rest are also found in perennial. They just tend to be more
dominant in numbers in intermittent conditions (probably due to loss of predators).
Hydro
DroughtTax
Number of taxa that possess at least one of the following traits: ability to survive desiccation, adult ability to
exit, respiration plastron/spiracle
Hydro
OCH Pet
Percent individuals - Odonata, Coleoptera and Hemiptera
Hydro
OCHDPct
Percent individuals - Odonata, Coleoptera, Hemiptera and Diptera
Hydro
OCHTax
Number of Odonata, Coleoptera and Hemiptera taxa
Hydro
OCHDTax
Number of Odonata, Coleoptera, Hemiptera and Diptera taxa
Temp
TempColdPct
Thermal Preference -Percent cold individuals (optima ranking of 1, 2 or 3)
Temp
TempColdStenoPct
Thermal Preference and Tolerance -Percent cold stenotherm individuals (optima ranking of 1, 2 or 3 and
tolerance ranking of 1, 2 or 3)
1-7

-------
106 Table 1.continued...
Category
Variable
Description
Temp
Temp InterPct
Thermal Preference -Percent intermediate individuals (optima ranking of 4)
Temp
TempWarmEuryPct
Thermal Preference and Tolerance -Percent warm eurythermal individuals (optima ranking of 5, 6 or 7 and
tolerance ranking of 5, 6 or 7)
Temp
TempWarmPct
Thermal Preference -Percent warm individuals (optima ranking of 5, 6 or 7)
Temp
TempCoreColdPct
Thermal Preference and Tolerance -Percent core cold individuals (see temperature indicator writeups)
Temp
TempCoreWarmPct
Thermal Preference and Tolerance -Percent core warm individuals (see temperature indicator writeups)
Temp
TempColdTax
Thermal Preference -Number of cold taxa (optima ranking of 1, 2 or 3)
Temp
T empColdStenoT ax
Thermal Preference and Tolerance -Number of cold stenotherm taxa (optima ranking of 1, 2 or 3 and
tolerance ranking of 1, 2 or 3)
Temp
TempWarmTax
Thermal Preference -Number of warm taxa (optima ranking of 5, 6 or 7)
Temp
TempWarmEury Tax
Thermal Preference and Tolerance -Number of warm eurythermal taxa (optima ranking of 5, 6 or 7 and
tolerance ranking of 5, 6 or 7)
Temp
Temp InterTax
Thermal Preference -Number of intermediate taxa (optima ranking of 4)
Temp
T empCoreColdT ax
Thermal Preference and Tolerance -Number of core cold taxa (see temperature indicator writeups)
Temp
T empCore W arm_T ax
Thermal Preference and Tolerance -Number of core warm taxa (see temperature indicator writeups)
Scenario
Drier WinPct
Percent individuals that possess the most number of traits states that are predicted or have been shown to be
most favorable in a drier climate scenario
Scenario
Drier VulnerablePct
Percent individuals that have the fewest favorable trait states and the most number of unfavorable trait
states in a drier climate scenario
Scenario
WarmDrier Vulnerable
Pet
Percent individuals that have the fewest favorable trait states and the most number of unfavorable trait
states in a warmer drier climate scenario
Scenario
WarmDrier WinPct
Percent individuals that possess the most number of traits states that are predicted or have been shown to be
most favorable in a warmer drier climate scenario
Scenario
Wet_WinPct
Percent individuals that possess the most number of traits states that are predicted or have been shown to be
most favorable in a wetter climate scenario
1-8

-------
108 Table 1.continued...
Category
Variable
Description
Scenario
WetLosPct
Percent individuals that have the fewest favorable trait states and the most number of unfavorable trait
states in a wetter climate scenario
Scenario
WarmWetLosPct
Percent individuals that have the fewest favorable trait states and the most number of unfavorable trait
states in a warmer wetter climate scenario
Scenario
WarmWet_WinPct
Percent individuals that possess the most number of traits states that are predicted or have been shown to
be most favorable in a warmer wetter climate scenario
Scenario
Drier WinTax
Number of taxa that possess the most number of traits states that are predicted or have been shown to be
most favorable in a drier climate scenario
Scenario
Drier VulnerableTax
Number of taxa that have the fewest favorable trait states and the most number of unfavorable trait states
in a drier climate scenario
Scenario
WarmDrierVulnerab
leTax
Number of taxa that have the fewest favorable trait states and the most number of unfavorable trait states
in a warmer drier climate scenario
Scenario
WarmDrier WinTax
Number of taxa that possess the most number of traits states that are predicted or have been shown to be
most favorable in a warmer drier climate scenario
Scenario
Wet_WinTax
Number of taxa that possess the most number of traits states that are predicted or have been shown to be
most favorable in a wetter climate scenario
Scenario
WetLosTax
Number of taxa that have the fewest favorable trait states and the most number of unfavorable trait states
in a wetter climate scenario
Scenario
WarmWetLosTax
Number of taxa that have the fewest favorable trait states and the most number of unfavorable trait states
in a warmer wetter climate scenario
Scenario
WarmWet_WinTax
Number of taxa that possess the most number of traits states that are predicted or have been shown to be
most favorable in a warmer wetter climate scenario
1-9

-------
Table 11-2. Results for these sites and site groups are presented in the summary tables in
Attachments 12-18.
State
Site/ Site Group

56817

57011
&
cd
57065

NE High (= Northeastern Highlands Site Group)

Laur (= Laurentian Plains and Hills Site Group)

4927250

4936750

4951200

5940440

WU SF (= Wasatch Uintas Semiarid Foothills Site Group)

WU ME (= Wasatch Uintas Mid-elevation Mountains Site Group)

CP (= Colorado Plateaus Site Group)
g
NC0109 (BR)
8
NC0207 (BR)
cd
O
NC0209 (BR)
£
o
NC0075 (Pied)
£
NC0248 (Pied)
1-10

-------
Attachment II
NAO/ONI/PDO
The purpose of this attachment is to provide background on the variables from the North Atlantic
Oscillation (NAO), Oceanic Nino Index (ONI) and Pacific Decadal Oscillation (PDO) datasets
that were used in the correlation analyses. Results from these analyses are available upon
request.
1-11

-------
ATTACHMENT II. NAO/ONI/PDO
Data sources:
North Atlantic Oscillation (NAO): http://www.cgd.ucar.edu/cas/ihurrell/indices.html
Oceanic Nino Index (ONI):
http://www.cpc.noaa.gov/products/analvsis monitoring/ensostuff/ensovears.shtml and
http://www.cdc.noaa.uov/data/climateindices/List/
Pacific Decadal Oscillation (PDO): http://iisao.washington.edu/pdo/PDO.latest
Correlation analyses were performed on NAO, ONI, and PDO datasets to evaluate whether these
cyclic climate indices have shown significant trends over the last 32 years, and also to examine
whether the indices are significantly associated with trends in biological data (i.e. do year-to-year
changes in species composition track the NAO, ONI and PDO?). We used NAO data in the
Maine and North Carolina analyses because the NAO affects the Eastern seaboard states
(personal communication with James Hurrell (Email: ihurrel 1 @ucar.edu)). The ONI and PDO
indices have greater relevance in western states and were therefore used in the Utah analyses.
There are many different variables associated with the NAO, ONI and PDO datasets (e.g.,
monthly values, various averages). It was difficult to know which ones had the greatest relevance
to our analyses, especially with the ONI and PDO datasets. This is because there is no unique or
universally accepted way to define the spatial structure of these phenomena. More information
seemed to be available on the NAO than the ONI and PDO. Bradley and Ormerod (2001) were
used as guidance in selecting the following 2 NAO variables: NAO PC-Based Seasonal
December-January-February-March (DJFM) index and the NAO PC-Based Annual index. We
used the December-January-February-March (DJFM) seasonal index because the main "season"
of the NAO is northern winter; this is when the atmosphere is most dynamically active (personal
communication with James Hurrell (Email: ihurrell@ucar.edu). We chose the Principal
Component (PC)-based time series data over the station-based indices because they were
available for the appropriate timeframe, and they provide better representations of the full NAO
spatial pattern than station-based indices. Station-based indices are limited because they are fixed
in space and can therefore only adequately capture NAO variability for parts of the year.
1-12

-------
Moreover, their pressures are significantly affected by small-scale and transient meteorological
phenomena not related to the NAO and, thus, contain more noise.
We were unable to find references to help guide our selection of ONI and PDO indices for the
Utah analyses, so we included all the variables. The ONI is based on sea surface temperature
departures from average in the Nino 3.4 region, and is a principal measure for monitoring,
assessing, and predicting the El Nino-Southern Oscillation (ENSO). ENSO is a combination of
atmospheric and oceanic phenomena in the tropical Pacific Ocean. It is manifested in the
atmosphere by changes in the pressure difference between Tahiti and Darwin, Australia and in
the ocean by warming of surface waters of the tropical Eastern Pacific Ocean. NOAA's
operational definitions of El Nino and La Nina are keyed to the ONI index. NOAA's Climate
Prediction Center (CPC) considers El Nino or La Nina conditions to occur when the monthly
Nino3.4 sea surface temperature departures meet or exceed +/-0.5°C along with consistent
atmospheric features. These anomalies must also be forecasted to persist for 3 consecutive
months1. The PDO is a long-lived El Nino-like pattern of Pacific climate variability. While the
two climate oscillations have similar spatial climate fingerprints, they have very different
behavior in time. Two main characteristics distinguish PDO from El Nino/Southern Oscillation
(ENSO): first, 20th century PDO "events" persisted for 20-to-30 years, while typical ENSO
events persisted for 6 to 18 months; second, the climatic fingerprints of the PDO are most visible
in the North Pacific/North American sector, while secondary signatures exist in the tropics - the
opposite is true for ENSO. More sources of information on the ONI and PDO are available upon
request.
1 For more information, see
http://www.cpc.noaa.gov/products/analvsis monitoring/lanina/enso evoliition-status-fcsts-\veb.pdf.
1-13

-------
Attachment 12
Climate Variable -Year Trends
In this Attachment, we summarize yearly trend results for PRISM mean, maximum and
minimum annual air temperature and PRISM mean annual precipitation at Maine, Utah and
North Carolina biological sampling sites and site groups.
1-14

-------
ATTACHMENT 12. CLIMATE VARIABLE-YEAR TRENDS
We examined yearly trends in PRISM mean, maximum and minimum annual air temperature and
PRISM mean annual precipitation at Maine, Utah and North Carolina biological sampling sites
and site groups. Results are summarized in Table 12-1. Utah had the most number of sites and
site groups that showed significant yearly temperature trends across the most number of
temperature variables; year is significantly correlated with mean and minimum annual air
temperature at 6 of the 7 sites/ site groups. Three of the 4 Maine sites/site groups were
significantly and positively correlated with mean and minimum annual air temperature, and 6 of
the 7 North Carolina sites/site groups were significantly correlated with minimum annual air
temperature. None of the sites/site groups had significant yearly trends in mean annual
precipitation (which tends to be highly variable).
1-15

-------
Table 12-1. Pearson product moment correlations of PRISM mean, maximum and minimum annual air temperature (tmeanl4, tmaxl4
and tminl4) and PRISM mean annual precipitation (pptl4) versus year for individual sites and site groups in Maine, Utah and North
Carolina. Correlations with NAO and ONI variables were also included (see Attachment II for more details on these variables).




PRISM/NAO/ONI
-YEAR





State
Site/ Site Group
N
tmeanl4
tmaxl4
tminl4
p
)tl4
NAO DJFM PC
r
P
r
P
r
P
r
P
r
N
P

56817
32
0.24
00
00
II*
p.
0.09
p=.620
0.33
p=.065
0.04
p=.820
-0.15
N=23
p=.506
£
57011
32
0.09
p=.635
-0.03
p=.888
0.17
p=.351
0.08
p=.674
-0.22
N=12
p=.502
"3
57065
32
0.42
p=.018
0.28
p=.123
0.46
p=.008
-0.16
p=.373
-0.23
N=9
p=.549
s
NE High
32
0.52
p=.002
0.41
p=.019
0.55
p=.001
0.22
p=.236
-0.35
N=8
p=.391

Laur
32
0.41
p=.019
0.28
p=. 119
0.46
p=.009
-0.13
p=.494
-0.15
N=8
p=.731
State
Site/ Site Group
N
tmeanl4
tmaxl4
tminl4
P
)tl4
ONI DJF
r
P
r
P
r
P
r
P
r
N
P

4927250
32
0.57
p=.001
0.60
p=.000
0.35
p=.051
-0.04
p=.824
0.18
N=17
p=.494

4936750
32
0.48
p=.005
0.14
p=.443
0.70
p=.000
0.11
p=.540
-0.14
N=12
p=.656
Utah
4951200
32
0.74
p=.000
0.71
p=.000
0.71
p=.000
-0.08
p=.674
-0.09
N=15
p=.736
5940440
32
0.71
p=.000
0.55
p=.001
0.74
p=.000
-0.13
p=.483
0.22
N=9
p=.567
WU SF
32
0.77
p=.000
0.52
p=.002
0.83
p=.000
-0.20
p=.262
-0.25
N=20
p=.284

WU ME
32
0.30
p=.093
0.18
p=.321
0.36
p=.042
0.04
p=.846
-0.28
N=12
p=.383

CP
32
0.65
p=.000
0.45
p=.009
0.75
p=.000
0.06
p=.724
0.27
N=14
p=.359
State
Site/ Site Group
N
tmeanl4
tmaxl4
tminl4
P
)tl4
NAO DJFM PC
r
p
r
P
r
P
r
P
r
N
P
"p
NC0109 (BR)
32
0.15
p=.412
0.19
p=.302
0.07
p=.688
0.04
p=.820
0.02
N=ll
p=.954
NC0207 (BR)
32
0.73
p=.000
0.58
p=.001
0.77
p=.000
-0.12
p=.497
-0.02
N=9
p=.969
o
NC0209 (BR)
32
0.68
p=.000
0.53
p=.002
0.69
p=.000
0.19
p=.292
0.09
N=32
p=.637
ti
NC0075 (Pied)
32
0.29
p=. 107
-0.09
p=.627
0.54
p=.001
-0.17
p=.348
-0.46
N=7
p=.294
£
NC0248 (Pied)
32
0.37
p=.039
0.01
p=.942
0.57
p=.001
-0.12
p=.527
-0.41
N=7
p=.365
1-16

-------
Attachment 13
Climate Variable -Temperature Metric Trends
In this Attachment, we show results for a selected subset of temperature-sensitive metrics, which
were examined for yearly trends and trends related to PRISM air temperature and precipitation
variables.
1-17

-------
ATTACHMENT I3.CLIMATE VARIABLE-TEMPERATURE METRIC TRENDS
Results for a selected subset of temperature-sensitive metrics are shown in Tables 13-1 through
13-6. Overall, there were few consistent trends among the metrics (or taxa) that occurred across
the site/site groups and across states. In addition, some of the significant correlations were found
to be driven by outliers. Some may also have been influenced by confounding factors (see
Section 2 of the report for more information).
There are some notable regional differences. In Utah, the temperature metric trends seemed more
evident (maybe because Utah has experienced a more noticeable temperature increase than the
other states). This may have been because of the availability of at least some long-term reference
data from sites/site groups in Utah at higher elevations. In Maine, the ecoregion with higher
elevations (Northeastern Highlands) had no sites with sufficient long term data detect such
trends.
For temperature preference trait groups in each state, 'long-list' metrics were based on the
original list of cold- or warm-water taxa (which were derived from weighted average and
maximum likelihood calculations and literature - see Appendix K for more information). The
'short-list' metrics were based on a subset of the 'long-list' taxa that are referred to (in other
sections of this report) as temperature indicator taxa. Data from other states, case studies (i.e.
evaluation of taxa lists at the coldest and warmest sites), and best professional judgment from the
regional climate change working groups were taken into account when developing the
temperature indicator lists (see Appendices E, F and G for more information on the working
groups; and Attachments to these appendices for information on temperature indicator taxa for
each state). The 'short-list' metrics were developed because it was believed that they would have
a greater chance of showing trends.
Overall, results from the 'short-list' and 'long-list' metrics were similar (Tables 13-1 to 6). We
examined what taxa might be driving the differences at certain sites. In Utah, Ephemerella
(designated as a cold-water taxon) was influential over trends at two of four long-term reference
sites. There are also a few taxa that were excluded from the 'short-list', but showed trends
1-18

-------
(although some trends were counter-intuitive trends and must be related to factors other than
climate or discounted). Nevertheless, in Utah, these taxa are worth further consideration:
Rhyacophilidae, Drunella, and Brachycentrus. In North Carolina, there were a few taxa on the
warm-water list that showed counter-intuitive trends (i.e. decreased as PRISM mean annual air
temperature increased). These were: Chimarra, Dromogomphus and Gomphus. This suggests
what has been noted in other sections of this report: that the cold- and warm-water taxa lists and
the derived biological temperature metrics are not final, that they do not necessarily capture all
relevant considerations. Although not presented here, another result worth noting is that the
'intermediate' taxa did not show strong or consistent trends.
1-19

-------
Table 13-1. Pearson product moment correlations of temperature-sensitive richness metrics versus year for individual sites and site
groups in Maine, Utah and North Carolina. Highlighted correlations were significant (p<0.05). The 'long-list' metrics were based on
the original list of cold and warm-water taxa (which were derived from weighted average and maximum likelihood calculations and
literature - see Appendix K for more information). The 'short-list' metrics were based on a subset of the 'long-list' taxa that are
referred to (in other sections of this report) as temperature indicator taxa. Data from other states, case studies (i.e. evaluation of taxa
lists at the coldest and warmest sites), and best professional judgment from the regional climate change working groups were taken
into account when developing the temperature indicator lists (see Attachments in Appendices E, F and G for more information on
temperature indicator taxa for each state). The 'short-list' metrics were developed because it was believed that they would have a
Temperature Richness Metrics - YEAR


# Cold-water Taxa
# Warm-water Taxa
# Cold-water Taxa
# Warm-water Taxa
State
Site/ Site Group

(Short-list)

(Short-list)

(Long-list)

(Long-list)


r
N
P
r
N
P
r
N
P
r
N
P

56817
0.49
N=23
p=.017
0.78
N=23
p=.000
0.41
N=23
p=.049
0.73
N=23
p=.000
£
57011
0.04
N=12
p=.896
0.65
N=12
p=.023
0.66
N=12
p=.019
0.77
N=12
p=.003
*c3
57065
0.54
N=9
p=.133
0.58
N=9
p=. 101
0.57
N=9
p=. Ill
0.56
N=9
p=. 115

Laur
0.82
N=8
p=.012
0.28
N=8
p=.494
0.43
N=8
p=.292
0.57
N=8
p=.141

NEHigh
-0.25
N=8
p=.554
-0.05
N=8
p=.900
0.26
N=8
p=.537
-0.13
N=8
p=.761

4927250
-0.71
N=17
p=.002
-0.21
N=17
p=.416
-0.59
N=17
p=.012
0.08
N=17
p=.762

4936750
-0.38
N=12
p=.227
0.38
N=12
p=.222
-0.32
N=12
p=.313
0.23
N=12
p=.468

4951200
-0.60
N=15
p=.017
0.81
N=15
p=.000
-0.64
N=15
p=.009
0.60
N=15
p=.019

5940440
-0.64
N=9
p=.065
0.00
N=9
p=1.00
-0.68
N=9
p=.043
-0.26
N=9
p=.502

WU SF
0.10
N=20
p=.677
0.67
N=20
p=.001
0.37
N=20
p=. 113
0.30
N=20
p=.206

WU ME
-0.27
N=12
p=.400
0.72
N=12
p=.009
-0.42
N=12
Jl
C\
vio
0.52
N=12
p=.080

CP
0.13
N=14
p=.662
0.50
N=14
p=.067
0.29
N=14
p=.321
0.61
N=14
p=.021
S3
NC0109 (BR)
0.55
N=ll
p=.080
-0.58
N=ll
p=.059
0.14
N=ll
p=.682
-0.56
N=ll
p=.074

NC0207 (BR)
0.38
N=9
p=.316
-0.04
N=9
p=.914
-0.21
N=9
p=.593
-0.52
N=9
p=. 151
o
NC0209 (BR)
0.82
N=7
p=.024
-0.53
N=7
p=.219
0.27
N=7
p=.562
-0.86
N=7
p=.013

NC0075 (P)
-0.39
N=7
p=.393
0.60
N=7
p=. 157
0.35
N=7
p=.441
0.45
N=7
p=.316
£
NC0248 (P)
0.28
N=7
p=.542
0.14
N=7
p=.772
0.57
N=7
Jl
<1
-0.50
N=7
p=.253
1-20

-------
Table 13-2. Pearson product moment correlations of temperature-sensitive % individual metrics versus year for individual sites and
site groups in Maine, Utah and North Carolina. Highlighted correlations were significant (p<0.05). The 'long-list' metrics were based
on the original list of cold and warm-water taxa (which were derived from weighted average and maximum likelihood calculations
and literature - see Appendix K for more information). The 'short-list' metrics were based on a subset of the 'long-list' taxa that are
referred to (in other sections of this report) as temperature indicator taxa. Data from other states, case studies (i.e. evaluation of taxa
lists at the coldest and warmest sites), and best professional judgment from the regional climate change working groups were taken
into account when developing the temperature indicator lists (see Attachments in Appendices E, F and G for more information on
temperature indicator taxa for each state). The 'short-list' metrics were developed because it was believed that they would have a




Temperature % Individual Metrics -
YEAR





State
Site/ Site Group
% Cold-water Individs
{Short-list)
% Warn Water Individs
{Short-list)
% Cold-water Individs
{Long-list)
% Warm-water Individs
{Long-list)


r
N
P
r
N
P
r
N
P
r
N
P

56817
0.47
N=23
p=.025
0.55
N=23
p=.006
0.11
N=23
p=.612
-0.06
N=23
p=.794
"3
57011
-0.67
N=12
p=.017
-0.59
N=12
p=.043
0.13
N=12
p=.687
0.02
N=12
p=.947
57065
0.45
N=9
p=.226
-0.36
N=9
p=.336
0.62
N=9
p=.076
-0.56
N=9
p=.121

Laur
-0.01
N=8
p=.990
-0.38
N=8
p=.347
-0.14
N=8
p=.745
0.23
N=8
p=.589

NEHigh
-0.45
N=8
p=.258
-0.02
N=8
p=.958
0.01
N=8
p=.986
0.05
N=8
p=.905

4927250
-0.72
N=17
p=.001
-0.21
N=17
p=.416
0.03
N=17
p=.918
-0.27
N=17
p=.291

4936750
-0.15
N=12
p=.635
0.42
N=12
p=.174
-0.42
N=12
cn
r-
II*
Ph
0.08
N=12
Jl
*00
o
Utah
4951200
-0.63
N=15
p=.013
0.40
N=15
p=.140
-0.63
N=15
p=.011
0.21
N=15
p=.460
5940440
-0.12
N=9
p=.764
0.00
N=9
p=1.00
0.27
N=9
p=.487
-0.33
N=9
p=.388
WU SF
-0.12
N=20
p=.603
0.60
N=20
p=.005
0.58
N=20
p=.008
-0.12
N=20
p=.610

WU ME
0.64
N=12
p=.026
0.63
N=12
p=.028
0.30
N=12
p=.346
-0.28
N=12
p=.375

CP
-0.02
N=14
p=.951
0.48
N=14
p=.084
0.20
N=14
p=.503
0.43
N=14
p=.127
cd
NC0109 (BR)
0.57
N=ll
p=.067
-0.04
N=ll
p=.904
0.36
N=ll
p=.273
-0.37
N=ll
p=.256

NC0207 (BR)
0.33
N=9
p=.391
0.09
N=9
p=.824
0.46
N=9
p=.212
-0.28
N=9
p=.462
8
o
NC0209 (BR)
0.29
N=7
p=.522
-0.33
N=7
p=.469
0.48
N=7
p=.277
-0.46
N=7
p=.300
NC0075 (P)
-0.02
N=7
p=.969
-0.73
N=7
p=.060
-0.26
N=7
p=.567
0.66
N=7
p=.107
o
£
NC0248 (P)
-0.14
N=7
p=.760
-0.55
N=7
p=.202
0.19
N=7
p=.685
-0.31
N=7
p=.494
1-21

-------
Table 13-3. Pearson product moment correlations of temperature-sensitive richness metrics versus PRISM mean annual air
temperature for individual sites and site groups in Maine, Utah and North Carolina. Highlighted correlations were significant
(p<0.05). The 'long-list' metrics were based on the original list of cold and warm-water taxa (which were derived from weighted
average and maximum likelihood calculations and literature - see Appendix K for more information). The 'short-list' metrics were
based on a subset of the 'long-list' taxa that are referred to (in other sections of this report) as temperature indicator taxa. Data from
other states, case studies (i.e. evaluation of taxa lists at the coldest and warmest sites), and best professional judgment from the
regional climate change working groups were taken into account when developing the temperature indicator lists (see Attachments in
Appendices E, F and G for more information on temperature indicator taxa for each state). The 'short-list' metrics were developed
Temperature Richness Metrics - PRISM mean annual air temperature


# Cold-water Taxa
# Warm-water Taxa
# Cold-water Taxa
# Warm-water Taxa
State
Site/ Site Group

{Short-list)

{Short-list)

{Long-list)

{Long-list)


r
N
P
r
N
P
r
N
P
r
N
P

56817
0.31
N=23
p=. 147
0.21
N=23
p=.341
0.08
N=23
p=.709
0.18
N=23
p=.423
£
57011
0.02
N=12
p=.947
0.27
N=12
p=.388
-0.09
N=12
p=.772
0.18
N=12
p=.586
'a
57065
-0.58
N=9
p=. 103
-0.73
N=9
p=.024
-0.62
N=9
p=.078
-0.53
N=9
p=. 143

Laur
-0.71
N=8
p=.049
-0.18
N=8
p=.675
-0.56
N=8
p=. 151
-0.56
N=8
p=. 150

NEHigh
-0.50
N=8
p=.203
0.54
N=8
p=.165
-0.54
N=8
Jl
C\
vio
0.48
N=8
p=.227

4927250
-0.63
N=17
p=.007
-0.44
N=17
p=.076
-0.61
N=17
p=.009
-0.23
N=17
p=.374

4936750
-0.08
N=12
p=.815
-0.03
N=12
p=.929
-0.04
N=12
p=.913
-0.13
N=12
p=.694

4951200
-0.65
N=15
p=.009
0.75
N=15
p=.001
-0.72
N=15
p=.003
0.35
N=15
p=.207
s
5940440
-0.14
N=9
p=.726
0.00
N=9
p=1.00
-0.32
N=9
p=.405
0.16
N=9
p=.682

WU SF
0.11
N=20
p=.639
0.53
N=20
p=.016
0.34
N=20
II*
Oh
0.01
N=20
p=.953

WU ME
-0.66
N=12
p=.018
0.65
N=12
p=.023
-0.74
N=12
p=.006
0.23
N=12
p=.469

CP
0.15
N=14
p=.619
0.51
N=14
p=.063
0.30
N=14
p=.302
0.37
N=14
p=. 196

NC0109 (BR)
-0.38
N=ll
p=.246
-0.18
N=ll
p=.592
0.17
N=ll
p=.614
-0.08
N=ll
p=.825
NC0207 (BR)
0.43
N=9
p=.245
0.27
N=9
p=.478
0.36
N=9
p=.341
-0.19
N=9
p=.626
2
o
NC0209 (BR)
0.00
N=7
p=.993
-0.48
N=7
p=.276
0.13
N=7
p=.784
-0.23
N=7
p=.617

NC0075 (P)
-0.09
N=7
p=.841
0.56
N=7
p=. 194
-0.15
N=7
p=.743
-0.08
N=7
p=.863
£
NC0248 (P)
-0.20
N=7
p=.667
-0.47
N=7
p=.291
0.27
N=7
p=.562
-0.92
N=7
p=.004
1-22

-------
Table 13-4. Pearson product moment correlations of temperature-sensitive % individual metrics versus PRISM mean annual air
temperature for individual sites and site groups in Maine, Utah and North Carolina. Highlighted correlations were significant
(p<0.05). The 'long-list' metrics were based on the original list of cold and warm-water taxa (which were derived from weighted
average and maximum likelihood calculations and literature - see Appendix K for more information). The 'short-list' metrics were
based on a subset of the 'long-list' taxa that are referred to (in other sections of this report) as temperature indicator taxa. Data from
other states, case studies (i.e. evaluation of taxa lists at the coldest and warmest sites), and best professional judgment from the
regional climate change working groups were taken into account when developing the temperature indicator lists (see Attachments in
Appendices E, F and G for more information on temperature indicator taxa for each state). The 'short-list' metrics were developed
Temperature Metrics - PRISM mean annual air temperature


% Cold-water Individs
% Warn Water Individs
% Cold-water Individs
% Warm-water Individs
State
Site/ Site Group

(iShort-list)

{Short-list)

{Long-list)

{Long-list)


r
N
P
r
N
P
r
N
P
r
N
P

56817
0.15
N=23
p=.506
0.13
N=23
p=.546
-0.25
N=23
p=.258
-0.09
N=23
p=.687
£
57011
-0.16
N=12
p=.617
0.37
N=12
p=.232
0.03
N=12
p=.921
-0.07
N=12
p=.834
'a
57065
-0.27
N=9
p=.480
0.05
N=9
p=.903
-0.23
N=9
p=.546
0.17
N=9
p=.666

Laur
0.42
N=8
p=.295
0.38
N=8
p=.358
0.66
N=8
p=.076
-0.21
N=8
p=.612

NEHigh
0.46
N=8
p=.250
0.20
N=8
p=.642
-0.56
N=8
p=.152
0.46
N=8
p=.252

4927250
-0.30
N=17
p=.236
-0.35
N=17
r-
II*
Oh
0.26
N=17
p=.313
-0.05
N=17
p=.848

4936750
-0.20
N=12
p=.534
0.01
N=12
p=.981
-0.10
N=12
p=.754
0.14
N=12
p=.664

4951200
-0.53
N=15
p=.044
0.62
N=15
p=.014
-0.54
N=15
p=.037
0.17
N=15
p=.547

5940440
-0.29
N=9
p=.455
0.00
N=9
p=1.00
-0.37
N=9
p=.326
-0.08
N=9
p=.846

WU SF
0.05
N=20
p=.826
0.44
N=20
p=.050
0.70
N=20
p=.001
-0.24
N=20
p=.307

WU ME
0.31
N=12
p=.324
0.75
N=12
p=.005
0.02
N=12
p=.945
-0.38
N=12
p=.225

CP
0.04
N=14
p=.898
0.65
N=14
p=.012
0.26
N=14
p=.362
0.36
N=14
p=.205

NC0109 (BR)
-0.32
N=ll
p=.344
0.00
N=ll
p=.998
-0.13
N=ll
p=.697
0.10
N=ll
p=.781
&
NC0207 (BR)
0.17
N=9
p=.665
0.26
N=9
p=.491
0.08
N=9
p=.848
-0.24
N=9
p=.537
C
o
NC0209 (BR)
0.07
N=7
p=.883
-0.45
N=7
p=.310
0.32
N=7
p=.491
-0.71
N=7
p=.073

NC0075 (P)
-0.10
N=7
p=.838
-0.16
N=7
p=.738
-0.05
N=7
p=.915
0.28
N=7
p=.545
£
NC0248 (P)
-0.10
N=7
p=.837
-0.54
N=7
p=.215
0.43
N=7
p=.331
-0.82
N=7
p=.024
1-23

-------
Table 13-5. Pearson product moment correlations of temperature-sensitive richness metrics versus PRISM mean annual
precipitation for individual sites and site groups in Maine, Utah and North Carolina. Highlighted correlations were significant
(p<0.05). The 'long-list' metrics were based on the original list of cold and warm-water taxa (which were derived from weighted
average and maximum likelihood calculations and literature - see Appendix K for more information). The 'short-list' metrics were
based on a subset of the 'long-list' taxa that are referred to (in other sections of this report) as temperature indicator taxa. Data from
other states, case studies (i.e. evaluation of taxa lists at the coldest and warmest sites), and best professional judgment from the
regional climate change working groups were taken into account when developing the temperature indicator lists (see Attachments in
Appendices E, F and G for more information on temperature indicator taxa for each state). The 'short-list' metrics were developed
Temperature Richness Metrics - PRISM mean annual precipitation


# Cold-water Taxa
# Wami-water Taxa
# Cold-water Taxa
# Warm-water Taxa
State
Site/ Site Group

(Short-list)

(Short-list)

(Long-list)

(Long-list)


r
N
P
r
N
P
r
N
P
r
N
P

56817
0.44
N=23
p=.035
0.07
N=23
p=.751
0.32
N=23
p=.130
0.05
N=23
p=.829
£
57011
0.18
N=12
p=.585
-0.04
N=12
p=.909
0.19
N=12
p=.561
0.01
N=12
p=.975
'a
57065
-0.51
N=9
p=.161
-0.13
N=9
p=.733
-0.12
N=9
p=.765
0.00
N=9
p=.993

Laur
-0.23
N=8
p=.581
-0.15
N=8
p=.725
0.03
N=8
p=.935
-0.16
N=8
p=.714

NEHigh
-0.19
N=8
p=.654
-0.03
N=8
p=.936
0.09
N=8
p=.832
0.14
N=8
p=.741

4927250
-0.11
N=17
p=.678
-0.05
N=17
p=.835
-0.07
N=17
p=.794
-0.11
N=17
p=.687

4936750
0.42
N=12
p=.169
0.21
N=12
p=.504
0.46
N=12
p=. 135
0.29
N=12
p=.363

4951200
0.21
N=15
p=.452
-0.25
N=15
p=.361
0.26
N=15
p=.353
-0.18
N=15
p=.517

5940440
0.01
N=9
p=.975
0.00
N=9
p=1.00
0.25
N=9
p=.512
-0.14
N=9
p=.723

WU SF
0.06
N=20
p=.803
-0.32
N=20
p=. 171
0.03
N=20
p=.890
-0.19
N=20
p=.426

WU ME
0.40
N=12
p=.201
-0.50
N=12
p=.097
0.18
N=12
p=.584
-0.70
N=12
p=.010

CP
0.00
N=14
p=.996
-0.04
N=14
p=.896
0.00
N=14
p=.991
0.11
N=14
p=.703

NC0109 (BR)
0.85
N=ll
p=.001
-0.65
N=ll
p=.029
0.63
N=ll
p=.036
-0.83
N=ll
p=.002

NC0207 (BR)
0.39
N=9
p=.305
-0.29
N=9
p=.445
-0.04
N=9
p=.921
-0.78
N=9
p=.013
S-H
u
NC0209 (BR)
0.31
N=7
p=.498
-0.56
N=7
p=. 189
0.30
N=7
p=.518
-0.75
N=7
p=.053

NC0075 (P)
0.34
N=7
p=.457
0.20
N=7
p=.663
0.40
N=7
p=.369
0.03
N=7
p=.942
£
NC0248 (P)
0.00
N=7
p=.992
-0.43
N=7
p=.333
-0.60
N=7
p=.154
0.37
N=7
p=.408
1-24

-------
Table 13-6. Pearson product moment correlations of temperature-sensitive % individual metrics versus PRISM mean annual
precipitation for individual sites and site groups in Maine, Utah and North Carolina. Highlighted correlations were significant
(p<0.05). The 'long-list' metrics were based on the original list of cold and warm-water taxa (which were derived from weighted
average and maximum likelihood calculations and literature - see Appendix K for more information). The 'short-list' metrics were
based on a subset of the 'long-list' taxa that are referred to (in other sections of this report) as temperature indicator taxa. Data from
other states, case studies (i.e. evaluation of taxa lists at the coldest and warmest sites), and best professional judgment from the
regional climate change working groups were taken into account when developing the temperature indicator lists (see Attachments in
Appendices E, F and G for more information on temperature indicator taxa for each state). The 'short-list' metrics were developed
Temperature % Individual Metrics - PRISM mean annual precipitation


% Cold-water Individs
% Warn Water Individs
% Cold-water Individs
% Wann-water Individs
State
Site/ Site Group

(Short-list)

(Short-list)

(Long-list)

(Long-list)


r
N
P
r
N
P
r
N
P
r
N
P

56817
0.58
N=23
p=.003
0.04
N=23
p=.852
0.27
N=23
p=.218
-0.25
N=23
p=.247
£
57011
0.03
N=12
p=.932
-0.10
N=12
p=.764
-0.44
N=12
p=. 154
-0.18
N=12
p=.585
"3
57065
-0.02
N=9
p=.968
-0.44
N=9
p=.234
0.26
N=9
p=.497
-0.39
N=9
p=.298

Laur
0.13
N=8
p=.756
0.00
N=8
p=.996
-0.24
N=8
p=.566
-0.32
N=8
p=.446

NEHigh
0.36
N=8
p=.381
-0.43
N=8
p=.286
0.12
N=8
p=.768
-0.08
N=8
p=.846

4927250
0.08
N=17
p=.748
-0.14
N=17
p=.596
0.17
N=17
p=.524
0.05
N=17
p=.849

4936750
0.30
N=12
p=.345
0.33
N=12
p=.290
0.40
N=12
p=.203
-0.54
N=12
p=.072

4951200
0.10
N=15
p=.720
-0.34
N=15
p=.210
0.11
N=15
p=.701
0.01
N=15
p=.968
"to
5940440
0.54
N=9
p=. 137
0.00
N=9
p=1.00
0.40
N=9
p=.291
-0.41
N=9
p=.275

WU SF
-0.46
N=20
p=.042
-0.17
N=20
p=.485
-0.27
N=20
p=.241
0.15
N=20
p=.529

WU ME
-0.20
N=12
p=.529
-0.24
N=12
p=.446
-0.60
N=12
p=.039
0.15
N=12
p=.635

CP
-0.09
N=14
p=.763
-0.16
N=14
p=.586
-0.09
N=14
p=.761
0.21
N=14
p=.481
23
NC0109 (BR)
0.63
N=ll
p=.038
-0.57
N=ll
p=.065
0.82
N=ll
p=.002
-0.83
N=ll
p=.001
.g
NC0207 (BR)
0.71
N=9
p=.032
-0.32
N=9
p=.402
0.75
N=9
p=.020
-0.41
N=9
p=.267
1
u
NC0209 (BR)
0.54
N=7
p=.209
-0.58
N=7
p=. 168
0.55
N=7
p=.202
-0.27
N=7
p=.552

NC0075 (P)
0.64
N=7
p=. 119
0.10
N=7
p=.824
0.54
N=7
p=.214
-0.52
N=7
p=.229
£
NC0248 (P)
-0.29
N=7
p=.527
0.25
N=7
p=.581
-0.79
N=7
p=.035
0.56
N=7
p=.194
1-25

-------
Attachment 14
Climate Variable -EPT Metric Trends
In this Attachment, we show results for a selected subset of EPT metrics, which were examined
for yearly trends and trends related to PRISM air temperature and precipitation variables.
1-26

-------
ATTACHMENT 14. CLIMATE VARIABLE-EPT METRIC TRENDS
Results for Ephemeroptera/Plecoptera/Trichoptera (EPT) metrics (richness and percent
individuals) are shown in Tables 14-1 through 14-3. There were a few significant associations
between EPT metrics, year, and PRISM mean annual air temperature and annual precipitation.
At 3 of the Maine sites, EPT richness was positively (and significantly) correlated with year. At
3 of the Utah sites, EPT richness was negatively (and significantly) correlated with PRISM mean
annual air temperature. Two of the North Carolina stations were significantly correlated with
PRISM mean annual precipitation (one was positive, the other negative). Overall there was a
lack of consistent patterns, which makes it difficult to project how EPT metrics may change as a
result of climate change. Developing EPT metrics that are geared more specifically towards
detecting climate change effects may be worth exploring (i.e. one that detects replacement of
cold-water EPT taxa with warm-water EPT taxa).
Table 14-1. Pearson product moment correlations of EPT richness (EPT Tax) and % individual
(EPT Pct) metrics versus year for individual sites and site groups in Maine, Utah and North
EPT Metrics - Year
State
Site/ Site Group
EPT Pet
EPT Tax
r
N
P
r
N
P

56817
0.06
N=23
p=.801
0.75
N=23
p=.000
Maine
57011
-0.52
N=12
p=.082
0.76
N=12
p=.004
57065
-0.36
N=9
p=.342
0.51
N=9
p=. 156
Laur
0.39
N=8
p=.337
0.71
N=8
p=.050

NEHigh
-0.67
N=8
p=.067
-0.60
N=8
p= .117

4927250
0.06
N=17
p=.812
-0.59
N=17
p=.014

4936750
-0.26
N=12
p=.416
-0.21
N=12
p=.520
Utah
4951200
0.00
N=15
p=.992
-0.49
N=15
p=.066
5940440
0.44
N=9
p=.232
-0.65
N=9
p=.058
WU SF
0.14
N=20
p=.570
0.42
N=20
p=.068

WU ME
-0.57
N=12
p=.052
-0.34
N=12
p=.273

CP
-0.49
N=14
p=.077
0.30
N=14
p=.303
O
NC0109 (BR)
0.74
N=ll
p=.010
0.30
N=ll
p=.364
NC0207 (BR)
0.27
N=9
p=.485
-0.18
N=9
p=.651
o
NC0209 (BR)
0.60
N=7
p=.157
0.16
N=7
p=.724
ts
NC0075 (P)
-0.60
N=7
p=.153
0.15
N=7
p=.747
£
NC0248 (P)
0.36
N=7
p=.434
0.54
N=7
p=.210
1-27

-------
Table 14-2. Pearson product moment correlations of EPT richness (EPTTax) and % individual
(EPTPct) metrics versus PRISM mean annual air temperature for individual sites and site
EPT Metrics - PRISM mean annual air temperature
State
Site/ Site Group
EPT Pet
EPT Tax
r
N
P
r
N
P

56817
0.08
N=23
p=.714
0.17
N=23
p .444
Maine
57011
0.64
N=12
p=.025
0.25
N=12
p=.428
57065
-0.07
N=9
p=.868
-0.64
N=9
p=.062
Laur
0.16
N=8
p=.698
-0.50
N=8
p=.211

NEHigh
0.58
N=8
p=.135
0.29
N=8
p=.493

4927250
0.03
N=17
p=.912
-0.57
N=17
p=.017

4936750
0.04
N=12
p=.899
-0.09
N=12
p=.772

4951200
0.27
N=15
p=.335
-0.73
N=15
p=.002

5940440
0.07
N=9
p=.864
-0.43
N=9
p=.248
WU SF
0.18
N=20
p=.449
0.31
N=20
p=. 186

WU ME
-0.27
N=12
p=.396
-0.77
N=12
p=.004

CP
-0.43
N=14
p=. 125
0.37
N=14
p=. 187

NC0109 (BR)
-0.09
N=ll
p=.796
0.00
N=ll
p=.988
o
NC0207 (BR)
-0.22
N=9
p=.574
0.35
N=9
p=.359
o
NC0209 (BR)
0.03
N=7
p=.943
0.39
N=7
p=.381

NC0075 (P)
0.14
N=7
p=.764
-0.15
N=7
p=.754
£
NC0248 (P)
0.24
N=7
p=.610
0.41
N=7
p=.365
1-28

-------
Table 14-3. Pearson product moment correlations of EPT richness (EPTTax) and % individual
(EPTPct) metrics versus PRISM mean annual precipitation for individual sites and site groups
EPT Metrics - PRISM mean annual preei
)itation
State
Site/ Site Group
EPT Pet
EPT Tax
r
N
P
r
N
P
Maine
56817
0.01
N=23
p=.973
0.20
N=23
p=.365
57011
-0.05
N=12
p=.884
0.24
N=12
p=.449
57065
0.17
N=9
p=.662
-0.12
N=9
p=.756
Laur
-0.45
N=8
p=.262
-0.34
N=8
p=.411
NEHigh
0.06
N=8
p=.895
0.34
N=8
p=.416
Utah
4927250
-0.29
N=17
p=.265
-0.245
N=17
p=.343
4936750
0.32
N=12
p=.303
0.4497
N=12
p=.142
4951200
-0.18
N=15
p=.527
0.4483
N=15
p=.094
5940440
0.32
N=9
p=.396
0.1728
N=9
p=.657
WU SF
-0.03
N=20
p=.909
-0.0528
N=20
p=.825
WU ME
0.43
N=12
rf
II*
Ph
-0.2622
N=12
p=.410
CP
-0.19
N=14
p=.521
-0.1404
N=14
p=.632
North Carolina
NC0109 (BR)
0.82
N=ll
p=.002
0.36
N=ll
p=.275
NC0207 (BR)
0.44
N=9
p=.234
0.26
N=9
p=.502
NC0209 (BR)
0.59
N=7
p=.162
0.54
N=7
p=.213
NC0075 (P)
0.15
N=7
p=.743
0.39
N=7
p=.382
NC0248 (P)
-0.70
N=7
p=.081
-0.80
N=7
p=.033
1-29

-------
Attachment 15
Climate Variable -HBI Trends
In this Attachment, we show results for the HBI metric, which we examined for yearly trends
and trends related to PRISM air temperature and precipitation variables.
1-30

-------
ATTACHMENT 15. CLIMATE VARIABLE-HBI TRENDS
Results for the Hilsenhoff Biotic Index (HBI) are shown in Tables I5-lthrough 15-3. We
evaluated long-term HBI trends because HBIs are used as part of water quality assessments in
many states (including Maine and North Carolina, analyzed in this study). It was also valuable
because we lacked long-term nutrient data for most sites, and the HBI provided us with some
insight as to whether or not a site had been influenced by organic enrichment.
There were a few significant associations between HBI values, year, PRISM mean annual air
temperature, and PRISM annual precipitation. At Maine Station 57011, HBI values were
positively (and significantly) correlated with year, which suggests that long-term trends at this
site may have been influenced by organic enrichment (higher HBI scores suggest greater organic
enrichment). One site in Utah and 1 site in North Carolina were negatively (and significantly)
correlated with year. HBI values at one of the Utah site groups (Wasatch Uinta Semiarid
Foothills) was negatively (and significantly) correlated with PRISM mean annual air
temperature. Two of the Blue Ridge North Carolina sites had strong negative correlations
between HBI values and PRISM mean annual precipitation. Overall there was a lack of
consistent patterns, which makes it difficult to project how HBI values may change as a result of
climate change.
It should be noted that the tolerance values that were used in the HBI calculations for Maine and
North Carolina sites/site groups were derived from their respective state datasets. Utah HBIs
were calculated using tolerance values from the New Mexico traits database because we did not
have access to state-specific ones at the time of the analyses.
1-31

-------
Table 15-1. Pearson product moment correlations of HBI versus year for individual sites and site
groups in Maine, Uta
1 and North Carolina. Highlighted correlations were significant (p<0.05).
HBI - Year
State
Site/ Site Group
r
N
P

56817
-0.13
N=23
p=.544
c
*c3
57011
0.75
N=12
p=.005
57065
0.00
N=9
p=.992

Laur
-0.03
N=8
p=.951

NEHigh
0.37
N=8
p=.363

4927250
-0.19
N=17
p=.466

4936750
0.32
N=12
p=.313
Utah
4951200
0.27
N=15
p=.329
5940440
-0.46
N=9
p=.213
WU SF
-0.64
N=20
p=.002

WU ME
-0.32
N=12
p=.312

CP
-0.06
N=14
p=.841
£
NC0109 (BR)
-0.63
N=ll
p=.038

NC0207 (BR)
-0.34
N=9
p=.374
s
u
NC0209 (BR)
-0.50
N=7
p=.251
NC0075 (P)
0.23
N=7
p=.614
o
£
NC0248 (P)
-0.17
N=7
p=.710
1-32

-------
Table 15-2. Pearson product moment correlations of HBI versus PRISM mean annual air
temperature for individual sites and site groups in Maine, Utah and North Carolina. Highlighted
HBI - PRISM mean annual air temperature
State
Site/ Site Group
r
N
P

56817
-0.07
N=23
p=.760
Maine
57011
-0.21
N=12
p=.512
57065
0.12
N=9
p=.761
Laur
-0.44
N=8
p=.275

NEHigh
-0.15
N=8
p=.725

4927250
-0.32
N=17
p=.208

4936750
0.09
N=12
p=.773
Utah
4951200
0.00
N=15
p=.993
5940440
0.09
N=9
p=.816
WU SF
-0.69
N=20
p=.001

WU ME
-0.38
N=12
p=.227

CP
-0.26
N=14
p=.376
cd
NC0109 (BR)
0.13
N=ll
p=.705

NC0207 (BR)
-0.07
N=9
p=.855
1
u
ts
NC0209 (BR)
-0.12
N=7
p=.790
NC0075 (P)
-0.56
N=7
p=.193
o
£
NC0248 (P)
0.13
N=7
p=.789
1-33

-------
Table 15-3. Pearson product moment correlations of HBI versus PRISM mean annual
precipitation for individual sites and site groups in Maine, Utah and North Carolina. Highlighted
HBI - PRISM mean annual precipitation
State
Site/ Site
Group
r
N
P

56817
-0.22
N=23
p=.309
Maine
57011
0.25
N=12
p=.434
57065
-0.51
N=9
p=.165
Laur
0.42
N=8
p=.299

NEHigh
-0.47
N=8
p=.235

4927250
0.16
N=17
p=.533

4936750
-0.55
N=12
p=.064

4951200
0.32
N=15
p=.246
Utal
5940440
-0.37
N=9
p=.322
WU SF
0.18
N=20
p=.442

WU ME
0.68
N=12
p=.016

CP
0.38
N=14
r-
II*
Oh

NC0109 (BR)
-0.86
N=ll
p=.001

NC0207 (BR)
-0.72
N=9
p=.030
1
u
NC0209 (BR)
-0.57
N=7
Jl
<1
VO
NC0075 (P)
-0.14
N=7
p=.770
o
£
NC0248 (P)
0.60
N=7
p=.154
1-34

-------
Attachment 16
Climate Variable -OCH Metric Trends
In this Attachment, we show results for a selected subset of OCH metrics, which were examined
for yearly trends and trends related to PRISM air temperature and precipitation variables.
1-35

-------
ATTACHMENT 16. CLIMATE VARIABLE-OCH METRIC TRENDS
Results for Odonata/Coleoptera/Hemiptera (OCH) trait metrics (richness and percent
individuals) are shown in Tables 16-1 through 16-3. OCH metrics may be useful as 'hydrologic
indicator' metrics because these Orders are generally expected to do better during drier, more
intermittent conditions (Bonada et al. 2007a). Results in Maine, North Carolina and Utah show
that there were 4 significant correlations between OCH metrics and PRISM mean annual air
temperature, and all of them were positive and all occurred at site groups (Maine Laurentian
Plains and Hills site group and all 3 Utah site groups). None of the OCH metrics were
significantly correlated with PRISM mean annual precipitation. Most of the significant
correlations were with year and occurred at the Utah sites and site groups. None of the OCH
metrics at the North Carolina sites/site groups showed significant trends.
It should be noted that the lack of consistent patterns may be due in part to sampling methods.
There are probably not many state biomonitoring programs that target Hemipterans for capture
or that record data on them consistently. Collection methods are likely also a factor in the capture
of Odonata. There tends to be greater Odonate abundance and diversity in edge habitats, and
many state biomonitoring programs target riffle habitats only.
1-36

-------
Table 16-1. Pearson product moment correlations of OCH (Odonata/Coleoptera/Hemiptera)
richness (OCH Tax) and % individual metrics (OCH Pct) versus year for individual sites and
site groups in Maine, Utah and North Carolina. Highlighted correlations were significant
(p<0.05).	
OCH Metrics - Year
State
Site/ Site Group
OCH Pet
OCH Tax
r
N
P
r
N
P

56817
0.28
N=23
p=.204
0.43
N=23
p=.038
Maine
57011
-0.52
N=12
p=.087
0.43
N=12
p=.162
57065
0.37
N=9
p=.329
0.28
N=9
p=.468
Laur
-0.80
N=8
p=.016
-0.16
N=8
p=.709

NEHigh
0.06
N=8
p=.895
0.34
N=8
p=.416

4927250
0.66
N=17
p=.004
0.61
N=17
p=.010

4936750
0.32
N=12
p=.313
0.83
N=12
p=.001

4951200
0.38
N=15
Jl
C\
o
0.45
N=15
p=.091

5940440
-0.59
N=9
p=.097
0.28
N=9
p=.471
WU SF
0.30
N=20
p=.201
0.80
N=20
p=.000

WU ME
0.75
N=12
p=.005
0.88
N=12
p=.000

CP
0.84
N=14
p=.000
0.66
N=14
p=.010

NC0109 (BR)
0.14
N=ll
p=.676
0.10
N=ll
p=.775
&
NC0207 (BR)
-0.59
N=9
p=.091
-0.25
N=9
p=.509
c
(J
NC0209 (BR)
-0.07
N=7
p=.875
0.06
N=7
p=.902

NC0075 (P)
0.69
N=7
p=.083
0.48
N=7
p=.272
o
£
NC0248 (P)
-0.27
N=7
p=.560
-0.40
N=7
p=.375
1-37

-------
Table 16-2. Pearson product moment correlations of OCH (Odonata/Coleoptera/Hemiptera)
richness (OCH Tax) and % individual metrics (OCH Pct) versus PRISM mean annual air
temperature for individual sites and site groups in Maine, Utah and North Carolina. Highlighted

OCH Metrics -
PRISM mean annual air temperature

State
Site/ Site Group
OCH Pet
OCH Tax
r
N
P
r
N
P

56817
0.01
N=23
p=.977
0.13
N=23
p=.541
Maine
57011
-0.09
N=12
p=.787
0.35
N=12
p=.259
57065
-0.33
N=9
p=.379
-0.10
N=9
p=.802
Laur
0.89
N=8
p=.003
0.55
N=8
Jl
C\
o

NEHigh
0.17
N=8
p=.679
0.38
N=8
p=.349

4927250
0.44
N=17
p=.074
0.11
N=17
p=.684

4936750
-0.11
N=12
p=.741
0.27
N=12
p=.395
Utah
4951200
0.27
N=15
p=.328
0.13
N=15
p=.656
5940440
-0.01
N=9
p=.981
0.59
N=9
p=.092
WU SF
-0.06
N=20
p=.812
0.64
N=20
p=.003

WU ME
0.51
N=12
p=.087
0.58
N=12
p=.047

CP
0.71
N=14
p=.005
0.35
N=14
p=.219
fl
"o
NC0109 (BR)
0.30
N=ll
p=.369
0.14
N=ll
p=.679
NC0207 (BR)
0.05
N=9
p=.891
-0.03
N=9
p=.942
a
cj
NC0209 (BR)
0.28
N=7
p=.543
-0.11
N=7
p=.818

NC0075 (P)
0.18
N=7
p=.696
0.16
N=7
p=.728
o
£
NC0248 (P)
0.09
N=7
p=.855
-0.42
N=7
p=.344
1-38

-------
Table 16-3. Pearson product moment correlations of OCH (Odonata/Coleoptera/Hemiptera)
richness (OCH Tax) and % individual metrics (OCH Pct) versus PRISM mean annual
precipitation for individual sites and site groups in Maine, Utah and North Carolina. No
OCH Metrics - PRISM mean annual precipitation
State
Site/ Site Group
OCH Pet
OCH Tax
r
N
P
r
N
P
Maine
56817
0.25
N=23
p=.244
0.28
N=23
p=. 198
57011
0.29
N=12
p=.359
0.07
N=12
p=.819
57065
-0.33
N=9
p=.390
-0.44
N=9
p=.237
Laur
0.17
N=8
p=.692
-0.31
N=8
p=.455
NEHigh
-0.29
N=8
p=.493
-0.25
N=8
p=.546
Utah
4927250
-0.04
N=17
p=.886
0.46
N=17
p=.063
4936750
0.25
N=12
p=.430
0.15
N=12
p=.640
4951200
-0.33
N=15
p=.230
0.23
N=15
p=.415
5940440
-0.28
N=9
p=.470
-0.30
N=9
p=.440
WU SF
0.23
N=20
p=.328
-0.23
N=20
p=.336
WU ME
-0.45
N=12
-t
ll'
Qh
-0.27
N=12
p=.392
CP
0.08
N=14
p=.783
0.28
N=14
p=.328
North Carolina
NC0109 (BR)
-0.22
N=ll
p=.520
-0.06
N=ll
II
00
On
NC0207 (BR)
-0.23
N=9
p=.551
-0.50
N=9
p=. 169
NC0209 (BR)
-0.48
N=7
p=.274
0.52
N=7
p=.236
NC0075 (P)
-0.24
N=7
p=.597
0.09
N=7
p=.853
NC0248 (P)
0.26
N=7
p=.572
-0.13
N=7
p=.783
1-39

-------
Attachment 17
Climate Variable -'Hydrologic' Metric Trends
In this Attachment, we show results for a selected subset of hydrologic metrics, which were
examined for yearly trends and trends related to PRISM air temperature and precipitation
variables.
1-40

-------
ATTACHMENT 17. CLIMATE VARIABLE-HYDROLOGIC METRIC TRENDS
The perennial and intermittent 'hydrologic indicator' metrics are based on literature (NCDWQ,
2005; Del Rosario and Resh, 2000). If these taxa, which require water for their entire life cycle,
are found at a site in a later instar larval stage, they are considered indicators of perennial stream
features. The list of intermittent taxa was based on interpretation of NCDWQ (2005) and
includes amphipods, isopods, small elongate Dipteran larvae (Ceratopogonidae, Blephariceridae,
Chironomidae, Deuterophlebiidae, Psychodidae) winter stoneflies (Capniidae,
Taeniopterygidae), Dytiscidae, Helichus larvae and Dasyhela (family Dolchopodidae). These
taxa tend to be more dominant in numbers in intermittent conditions (probably due in part to loss
of predators), but are (aside from Helichus larvae and Dasyhela) also found in perennial streams.
Results for the perennial and intermittent metrics (richness and percent individuals) are shown in
Tables 17-1 through 17-3. There were a few significant associations between these metrics, year,
PRISM mean annual air temperature, and PRISM mean annual precipitation in each of the states.
All of the significant correlations with PRISM mean annual air temperature occurred at the Utah
sites/site groups: at 3 sites/site groups, the perennial richness metric was negatively correlated
with annual air temperature, and at one of the site groups (Wasatch Uinta Semiarid Foothills),
the intermittent richness metrics was positively correlated with annual air temperature. The
intermittent metrics were significantly correlated with PRISM mean annual precipitation at 4
sites/site groups (3 in Utah, 1 in North Carolina), while % perennial individuals was positively
correlated with annual precipitation at 1 of the North Carolina sites (Station NC0109). In terms
of yearly trends, all 4 metrics were significantly correlated with year at Maine Station 57011.
The metrics also showed various yearly trends (more with richness metrics than with %
individuals) at 3 sites/site groups in Utah and 1 site in North Carolina (Station NC0109).
1-41

-------
Table 17-1. Pearson product moment correlations of 'hydrologic' richness ( Tax) and % individual (_Pct) metrics versus year for
individual sites and site groups in Maine, Utah and North Carolina. Highlighted correlations were significant (p<0.05). The perennial
taxa require water for their entire life cycle and the intermittent taxa tend to be more dominant in numbers in intermittent conditions.
Perennial and intermittent taxa lists were derived from NCDWQ 2005 and Del Rosario et al. 2000.	
Perennial/Intermittent Metrics - YEAR
State
Site/ Site Group
Perennial Pet
Intermit Pet
Perennial
Tax
Intermit Tax
r
N
P
r
N
P
r
N
P
r
N
P

56817
0.16
N=23
p=.466
-0.22
N=23
p=.323
0.75
N=23
p=.000
0.62
N=23
p=.001
Maine
57011
-0.68
N=12
p=.016
0.58
N=12
p=.046
0.66
N=12
p=.020
0.71
N=12
p=.010
57065
-0.38
N=9
p=.314
0.06
N=9
p=.877
0.53
N=9
p=.146
0.26
N=9
p=.493
Laur
-0.09
N=8
p=.841
0.18
N=8
p=.677
0.70
N=8
p=.054
0.16
N=8
p=.706

NEHigh
-0.56
N=8
p=. 145
0.46
N=8
p=.252
-0.69
N=8
p=.058
0.46
N=8
p=.248

4927250
0.18
N=17
p=.499
-0.39
N=17
p=. 119
-0.26
N=17
p=.319
0.41
N=17
p=. 105

4936750
0.26
N=12
p=.411
0.07
N=12
p=.833
0.11
N=12
p=.734
0.48
N=12
p=. 115
Utah
4951200
-0.11
N=15
p=.688
-0.30
N=15
p=.277
-0.43
N=15
p=. 114
0.15
N=15
p=.601
5940440
0.31
N=9
p=.424
-0.42
N=9
p=.255
-0.55
N=9
p=.123
-0.73
N=9
p=.027
WU SF
-0.04
N=20
p=.860
-0.39
N=20
p=.094
0.48
N=20
p=.033
0.69
N=20
p=.001

WU ME
0.34
N=12
p=.281
-0.29
N=12
p=.362
-0.21
N=12
p=.508
0.58
N=12
p=.050

CP
0.03
N=14
p=.922
0.08
N=14
p=.778
0.46
N=14
p=.098
0.67
N=14
p=.009
a
"o
NC0109 (BR)
0.68
N=ll
p=.023
-0.70
N=ll
p=.016
-0.01
N=ll
p=.968
-0.78
N=ll
p=.005
NC0207 (BR)
0.39
N=9
p=.296
-0.35
N=9
p=.349
-0.47
N=9
p=.199
-0.44
N=9
p=.238
a
cj
NC0209 (BR)
0.60
N=7
p=.157
-0.70
N=7
p=.083
0.46
N=7
p=.300
-0.29
N=7
p=.527

NC0075 (P)
-0.48
N=7
p=.279
0.34
N=7
p=.462
-0.10
N=7
p=.830
0.06
N=7
p=.893
o
£
NC0248 (P)
-0.07
N=7
p=.886
0.49
N=7
p=.261
-0.34
N=7
p=.450
0.58
N=7
p=. 175
1-42

-------
Table 17-2. Pearson product moment correlations of 'hydrologic' richness ( Tax) and % individual (_Pct) metrics versus PRISM
mean annual air temperature for individual sites and site groups in Maine, Utah and North Carolina. Highlighted correlations were
significant (p<0.05). The perennial taxa require water for their entire life cycle and the intermittent taxa tend to be more dominant in
numbers in intermittent conditions. Perennial and intermittent taxa lists were derived from NCDWQ 2005 and Del Rosario et al. 2000.
Perennial/Intermittent Metrics - PRISM mean annual air temperature
State
Site/ Site Group
Perennial Pet
Intermit Pet
Perennial Tax
Intermit Tax
r
N
P
r
N
P
r
N
P
r
N
P
Maine
56817
0.08
N=23
p=.726
-0.11
N=23
p=.604
0.11
N=23
p=.620
0.36
N=23
p=.087
57011
0.52
N=12
p=.084
-0.52
N=12
p=.082
0.45
N=12
p=.143
-0.09
N=12
p=.774
57065
-0.11
N=9
p=.768
0.23
N=9
p=.547
-0.54
N=9
p=.135
-0.27
N=9
p=.479
Laur
0.61
N=8
p=.106
-0.69
N=8
p=.056
-0.28
N=8
p=.498
-0.58
N=8
p=.134
NEHigh
0.61
N=8
p=.108
-0.57
N=8
p=.139
0.32
N=8
p=.437
-0.47
N=8
p=.240
Utah
4927250
-0.09
N=17
p=.741
-0.20
N=17
p=.430
-0.52
N=17
p=.034
0.09
N=17
p=.721
4936750
-0.12
N=12
p=.709
0.13
N=12
p=.695
-0.05
N=12
p=.869
0.14
N=12
p=.668
4951200
-0.12
N=15
p=.666
-0.42
N=15
p=.121
-0.68
N=15
p=.005
-0.13
N=15
p=.656
5940440
0.20
N=9
p=.600
-0.10
N=9
p=.799
-0.21
N=9
p=.591
-0.39
N=9
p=.304
WU SF
-0.28
N=20
p=.235
-0.29
N=20
p=.217
0.35
N=20
p=.129
0.51
N=20
p=.023
WU ME
0.34
N=12
p=.275
-0.29
N=12
p=.353
-0.68
N=12
p=.014
0.23
N=12
p=.467
CP
-0.36
N=14
p=.211
0.22
N=14
p=.456
0.29
N=14
p=.312
0.12
N=14
p=.684
North Carolina
NC0109 (BR)
0.07
N=ll
p=.845
-0.21
N=ll
p=.543
0.24
N=ll
p=.473
-0.08
N=ll
p=.809
NC0207 (BR)
-0.07
N=9
p=.862
0.14
N=9
p=.724
0.05
N=9
p=.892
0.24
N=9
p=.534
NC0209 (BR)
0.29
N=7
p=.522
-0.25
N=7
p=.593
0.41
N=7
p=.360
-0.30
N=7
p= .511
NC0075 (P)
0.08
N=7
p=.869
-0.26
N=7
p=.566
-0.11
N=7
p=.819
-0.19
N=7
p=.687
NC0248 (P)
-0.07
N=7
p=.887
0.30
N=7
p=.518
-0.27
N=7
p=.565
0.21
N=7
p=.656
1-43

-------
Table 17-3. Pearson product moment correlations of 'hydrologic' richness ( Tax) and % individual (_Pct) metrics versus PRISM
mean annual precipitation for individual sites and site groups in Maine, Utah and North Carolina. Highlighted correlations were
significant (p<0.05). The perennial taxa require water for their entire life cycle and the intermittent taxa tend to be more dominant in
numbers in intermittent conditions. Perennial and intermittent taxa lists were derived from NCDWQ 2005 and Del Rosario et al. 2000.
Perennial/Intermittent Metrics - PRISM mean annual precipitation
State
Site/ Site Group
Perennial Pet
Intermit Pet
Perennial Tax
Intermit Tax
r
N
P
r
N
P
r
N
P
r
N
P
Maine
56817
0.02
N=23
p=.939
-0.01
N=23
p=.950
0.22
N=23
p=.324
-0.01
N=23
p=.978
57011
0.21
N=12
p=.506
-0.03
N=12
p=.918
0.41
N=12
p=.188
0.13
N=12
p=.696
57065
0.05
N=9
p=.891
-0.17
N=9
p=.654
-0.38
N=9
p=.317
-0.30
N=9
p=.440
Laur
-0.36
N=8
p=.382
0.22
N=8
p=.597
-0.42
N=8
p=.305
0.07
N=8
p=.874
NEHigh
0.06
N=8
p=.886
-0.26
N=8
p=.530
0.06
N=8
p=.893
-0.09
N=8
p=.831
Utah
4927250
-0.32
N=17
p=.204
0.33
N=17
p=.198
-0.11
N=17
p=.679
0.59
N=17
p=.013
4936750
0.34
N=12
p=.281
-0.53
N=12
p=.075
0.37
N=12
p=.234
0.43
N=12
p=.167
4951200
0.06
N=15
p=.845
0.63
N=15
p=.012
0.30
N=15
p=.272
0.15
N=15
p=.590
5940440
0.22
N=9
p=.562
-0.34
N=9
p=.366
0.16
N=9
p=.687
0.19
N=9
p=.616
WU SF
0.35
N=20
p=.134
0.12
N=20
p=.609
0.04
N=20
p=.856
-0.08
N=20
p=.748
WU ME
-0.14
N=12
p=.654
0.39
N=12
p=.206
-0.34
N=12
p=.278
-0.07
N=12
p=.819
CP
0.09
N=14
p=.753
0.03
N=14
p=.918
0.08
N=14
p=.784
0.54
N=14
p=.045
North Carolina
NC0109 (BR)
0.66
N=ll
p=.026
-0.39
N=ll
p=.242
0.03
N=ll
p=.941
-0.76
N=ll
p=.006
NC0207 (BR)
0.35
N=9
p=.354
-0.48
N=9
p=.187
-0.13
N=9
p=.746
-0.44
N=9
p=.237
NC0209 (BR)
0.31
N=7
p=.495
-0.44
N=7
p=.326
0.60
N=7
p=.159
-0.36
N=7
p=.431
NC0075 (P)
-0.31
N=7
p=.495
0.29
N=7
p=.529
0.34
N=7
p=.451
0.50
N=7
p=.252
NC0248 (P)
-0.30
N=7
p=.510
-0.02
N=7
p=.974
-0.29
N=7
p=.524
-0.05
N=7
p=.907
1-44

-------
Attachment 18
Climate Variable -Scenario Metric Trends
In this Attachment, we show results for a selected subset of 'scenario' metrics, which were
examined for yearly trends and trends related to PRISM air temperature and precipitation
variables.
1-45

-------
ATTACHMENT 18. CLIMATE VARIABLE-SCENARIO METRIC TRENDS
In addition to looking at individual trait metrics, we also developed metrics that were based on
combinations of traits. The first step was to select traits and trait states likely to be "functionally"
linked to the projected changes in temperature and precipitation associated with climate change.
We used the available information (some literature, some best professional judgment) to develop
composite lists of favorable and unfavorable traits and trait states under two different generalized
scenarios (thus we termed these metrics 'scenario' metrics): 1. conditions become drier and
warmer (i.e. interrupted flows, more pool-like conditions, maybe some perennial streams become
intermittent, conditions become more unpredictable and organisms experience more
disturbances); and 2.conditions become warmer and wetter (i.e. more frequent and severe flood
events, more winter rains (instead of snow), more high flows, conditions become more
unpredictable and organisms experience more disturbances). Lists of the traits and trait states that
were deemed favorable and unfavorable are shown in Tables 18-1 (drier-warmer scenario) and
18-2 (wetter-warmer scenario). Taxa that possessed the most number of favorable traits states
formed the basis of the 'robust' metrics. Those that had the most number of unfavorable trait
states formed the basis of the 'vulnerable' metrics.
There are too many results to show and easily summarize, but all are available upon request.
Results for the warmer-drier-vulnerable, drier-vulnerable and drier-robust scenario metrics
(richness and % individuals) are shown in Tables 18-3 through 18-8. The drier scenario metrics
were chosen because it seems likely that drier conditions will impact the biota more than wetter
conditions. There was at least one significant association between at least one of these metrics,
year and PRISM mean annual air temperature and annual precipitation in each of the states. As
expected, the drier-vulnerable/warmer-drier-vulnerable metrics tended to follow similar patterns.
Drier-vulnerable richness metrics were significantly and negatively correlated with PRISM
mean annual air temperature at Maine Station 57065 and at 3 Utah sites/site groups. The drier-
robust richness metric was negatively correlated with annual air temperature at the Utah
Colorado Plateau site group. The drier-vulnerable % individuals metrics did not show consistent
patterns across sites within or across states (i.e. warmer-drier-vulnerable was negatively
correlated with annual air temperature at Maine Laurentian Plains and Hills site group and
1-46

-------
positively correlated at the Utah Wasatch Uinta Mid-elevation Mountain site group). Only a few
of the metrics were significantly correlated with PRISM mean annual precipitation and these
metrics sometimes showed unexpected patterns (i.e. % drier-vulnerable individuals was
negatively correlated with mean annual precipitation with a Utah site group). Both the richness
and % individual metrics showed some significant but mixed yearly trends in each state, with the
most number of significant correlations occurring in Utah and the least number in North
Carolina.
There are limitations associated with the scenario metrics:
These metrics are essentially exploratory. It was difficult to know which groups of traits
to use in the metrics (see Appendix K for more information on traits and trait selections),
and these results should be viewed as a first step that motivates additional investigation.
As more information becomes available about which combinations of traits and trait
states are most strongly linked to climate change effects, these metrics should be further
refined. More experimental data would be very helpful.
Some traits are likely more important than others and should probably be weighted
differently. However, we had insufficient information on which to base such decisions at
this time.
The different scenarios are not mutually exclusive (i.e., there could be both wetter (more
flood events) and drier (drought or more frequent and severe low flow events) conditions
occurring in the same year in some regions.
Climate models consistently project that temperatures will increase but there is more
uncertainty regarding the potential changes to hydrologic regimes. It is tough to make
generalizations about favorable/unfavorable trait states because the characteristics of the
hydrologic events can vary so much (severity, timing, duration and frequency).
Intuitively, it would seem that taxa that are best suited to surviving unpredictable
conditions/more frequent disturbance will fare better (i.e. can reproduce quickly, develop
quickly, small size). More experimental data on which taxa do best under disturbance
conditions also would be very helpful.
1-47

-------
Table 18-1. SCENARIO: Drier Warmer, Drier Warmer Conditions (more pool-like, maybe some
s go intermittent, unpredictable, more disturbance)
Traits
Favorable
Unfavorable
Life history


Voltinism
bi multi
semi (<1 generation/yr
Development


Synchronization of emergence


Adult life span


Adult ability to exit
present
absent
Ability to survive desiccation
present
absent
Mobility


Dispersal (adult)
high
low
Adult flying strength
strong
weak
Occurrence in drift
rare
abundant
Maximum crawling rate


Swimming ability
strong
none
Morphology


Attachment


Armoring
good
none
Shape
not stream

Respiration
plastron spir
tegument
Size at maturity
small

Resource acquisition/preference


Rheophily
depo
eros
Habit (primary)
SK, SW

Functional feeding group (primary)
CG
CF, SH
Temperature Indicator
warm
cold
1-48

-------
Table 18-2. SCENARIO: Wetter Warmer, More frequent and severe flood events, more winter
snow) (more high flows, unpredictab
e, more disturbance]

Traits
Favorable
Unfavorable
Life history


Voltinism
bi multi
semi
Development


Synchronization of emergence


Adult life span


Adult ability to exit
present
absent
Ability to survive desiccation


Mobility


Dispersal (adult)
high
low
Adult flying strength


Occurrence in drift
abundant

Maximum crawling rate
high
very low
Swimming ability


Morphology


Attachment


Armoring


Shape
stream
not stream
Respiration


Size at maturity
small
large
Resource acquisition/preference


Rheophily
eros
depo
Habit (primary)

SK, SW
Functional feeding group (primary)
CF, SH
CG
Temperature Indicator
warm
cold
1-49

-------
Table 18-3. Pearson product moment correlations of drier scenario richness metrics versus year
for individual sites and site groups in Maine, Utah and North Carolina. Highlighted correlations
were significant (p<0.05).	
Drier Scenario Richness Metrics - YEAR
State
Site/ Site Group
WarmDrier VulnerableTax
Drier VulnerableTax
Drier WinTax
r
N
P
r
N
P
r
N
P

56817
0.42
N=23
p=.047
0.71
N=23
p=.000
0.10
N=23
p=.659
Maine
57011
0.29
N=12
p=.352
0.60
N=12
p=.040
0.25
N=12
p=.443
57065
0.38
N=9
p=.313
0.43
N=9
p=.253
-0.19
N=9
p=.618
Laur
0.39
N=8
p=.345
0.58
N=8
p=.135
-0.78
N=8
p=.021

NEHigh
-0.37
N=8
p=.373
-0.33
N=8
p=.422
0.24
N=8
p=.574

4927250
-0.64
N=17
p=.006
-0.57
N=17
p=.018
0.66
N=17
p=.004

4936750
-0.08
N=12
p=.815
-0.01
N=12
p=.964
0.00
N=12
p=1.00

4951200
-0.62
N=15
p=.014
-0.35
N=15
p=.205
-0.05
N=15
p=.858

5940440
-0.51
N=9
cn
\Q
II*
Ph
-0.58
N=9
rf
o
II*
Ph
-0.47
N=9
Jl
VO
WU SF
0.39
N=20
p=.087
0.42
N=20
p=.062
0.35
N=20
p=.128

WU ME
-0.50
N=12
p=.098
-0.45
N=12
II*
Ph
0.23
N=12
p=.468

CP
0.15
N=14
p=.607
0.26
N=14
p=.367
-0.13
N=14
p=.665
£
NC0109 (BR)
-0.16
N=ll
p=.637
-0.19
N=ll
p=.573
0.34
N=ll
p=.313

NC0207 (BR)
-0.21
N=9
p=.579
-0.94
N=9
p=.000
0.00
N=9
p=1.00
u
NC0209 (BR)
0.12
N=7
p=.799
-0.26
N=7
p=.567
0.00
N=7
p=1.00
NC0075 (P)
-0.09
N=7
p=.854
-0.08
N=7
p=.871
0.41
N=7
p=.356
o
£
NC0248 (P)
-0.32
N=7
p=.481
-0.65
N=7
p=.116
-0.06
N=7
p=.901
1-50

-------
Table 18-4. Pearson product moment correlations of drier scenario % individual metrics versus
year for individual sites and site groups in Maine, Utah and North Carolina. Highlighted
correlations were significant (p<0.05).	
% Drier Scenario Metrics - YEAR
State
Site/ Site Group
Drier WinPct
Drier VulnerablePct
WarmDrier VulnerablePct
r
N
P
r
N
P
r
N
P

56817
0.16
N=23
p=.476
0.19
N=23
p=.389
0.49
N=23
p=.018
c
"3
57011
0.19
N=12
p=.561
-0.53
N=12
p=.078
-0.35
N=12
p=.270
57065
-0.22
N=9
p=.566
-0.17
N=9
p=.659
0.36
N=9
p=.336

Laur
-0.51
N=8
p=. 199
0.62
N=8
p=. 101
0.82
N=8
p=.012

NEHigh
-0.04
N=8
p=.930
-0.54
N=8
p=. 167
-0.20
N=8
p=.635

4927250
0.60
N=17
p=.011
0.09
N=17
p=.719
-0.72
N=17
p=.001

4936750
0.00
N=12
p=1.00
-0.46
N=12
p=. 137
0.01
N=12
p=.985
Utah
4951200
0.07
N=15
p=.793
-0.17
N=15
p=.534
-0.58
N=15
p=.023
5940440
-0.06
N=9
p=.879
0.30
N=9
p=.426
-0.10
N=9
p=.790
WU SF
0.34
N=20
p=. 143
0.15
N=20
p=.533
-0.16
N=20
p=.488

WU ME
0.23
N=12
p=.468
0.45
N=12
p=. 145
0.79
N=12
p=.002

CP
-0.35
N=14
p=.220
-0.16
N=14
p=.583
-0.02
N=14
p=.957
3
NC0109 (BR)
0.33
N=ll
p=.317
0.77
N=ll
p=.005
0.65
N=ll
p=.030

NC0207 (BR)
0.00
N=9
p=1.00
0.04
N=9
p=.921
0.20
N=9
p=.605
s
u
NC0209 (BR)
0.00
N=7
p=1.00
0.31
N=7
p=.493
0.26
N=7
p=.568
NC0075 (P)
0.09
N=7
p=.853
-0.54
N=7
p=.208
0.13
N=7
p=.783
£
NC0248 (P)
-0.22
N=7
p=.635
-0.60
N=7
p=. 153
-0.09
N=7
p=.845
1-51

-------
Table 18-5. Pearson product moment correlations of drier scenario richness metrics versus
PRISM mean annual air temperature for individual sites and site groups in Maine, Utah and
North Carolina. Highlighted correlations were significant (p<0.05).	
Drier Scenario Richness Metrics - PRISM mean annual air temperature
State
Site/ Site Group
WarmDrier VulnerableTax
Drier VulnerableTax
Drier WinTax
r
N
P
r
N
P
r
N
P

56817
0.30
N=23
p=.159
0.13
N=23
p=.547
0.21
N=23
p=.329
Maine
57011
-0.01
N=12
p=.975
0.29
N=12
p=.354
-0.11
N=12
p=.738
57065
-0.71
N=9
p=.033
-0.73
N=9
p=.026
0.02
N=9
p=.962
Laur
-0.70
N=8
p=.054
-0.48
N=8
p=.231
0.70
N=8
p=.051

NEHigh
-0.06
N=8
p=.883
0.09
N=8
p=.826
-0.64
N=8
p=.090

4927250
-0.56
N=17
p=.019
-0.66
N=17
p=.004
0.16
N=17
p=.537

4936750
0.01
N=12
p=.973
-0.08
N=12
p=.810
0.00
N=12
p=1.00

4951200
-0.62
N=15
p=.013
-0.61
N=15
p=.016
-0.32
N=15
p=.239

5940440
0.09
N=9
p=.820
-0.08
N=9
p=.833
0.09
N=9
p=.828
WU SF
0.43
N=20
p=.058
0.38
N=20
p=.095
0.12
N=20
p=.626

WU ME
-0.40
N=12
p=. 198
-0.60
N=12
p=.040
0.18
N=12
p=.580

CP
0.14
N=14
p=.624
0.27
N=14
p=.359
-0.57
N=14
p=.032
£
NC0109 (BR)
-0.23
N=ll
p=.501
-0.06
N=ll
p=.853
0.26
N=ll
p=.449

NC0207 (BR)
-0.41
N=9
p=.279
-0.47
N=9
p=.205
0.00
N=9
p=1.00
u
NC0209 (BR)
0.30
N=7
p=.519
-0.27
N=7
p=.560
0.00
N=7
p=1.00
NC0075 (P)
-0.24
N=7
p=.601
0.15
N=7
p=.750
-0.44
N=7
p=.323
o
£
NC0248 (P)
-0.02
N=7
p=.975
-0.29
N=7
p=.522
-0.71
N=7
p=.072
1-52

-------
Table 18-6. Pearson product moment correlations of drier scenario % individual metrics versus
PRISM mean annual air temperature for individual sites and site groups in Maine, Utah and
North Carolina. Highlighted correlations were significant (p<0.05).	
% Drier Scenario Metrics - PRISM mean annual air temperature
State
Site/ Site Group
Drier WinPct
Drier VulnerablePct
WarmDrier VulnerablePct
r
N
P
r
N
P
r
N
P

56817
0.17
N=23
p=.425
0.18
N=23
p=.407
0.17
N=23
p=.428
Maine
57011
-0.05
N=12
p=.876
0.64
N=12
p=.026
-0.14
N=12
p=.661
57065
0.29
N=9
p=.447
-0.29
N=9
p=.442
-0.37
N=9
p=.329
Laur
0.68
N=8
p=.062
-0.12
N=8
p=.781
-0.71
N=8
p=.050

NEHigh
-0.13
N=8
p=.755
0.30
N=8
p=.472
0.16
N=8
p=.707

4927250
0.52
N=17
p=.032
0.31
N=17
p=.231
-0.30
N=17
p=.243

4936750
0.00
N=12
p=1.00
-0.06
N=12
p=.862
-0.29
N=12
p=.357

4951200
-0.06
N=15
p=.819
-0.17
N=15
p=.538
-0.48
N=15
p=.071

5940440
0.38
N=9
p=.311
-0.38
N=9
p=.314
-0.27
N=9
p=.481
WU SF
0.19
N=20
p=.433
0.11
N=20
p=.647
0.03
N=20
p=.914

WU ME
0.18
N=12
p=.580
0.27
N=12
p=.394
0.58
N=12
p=.048

CP
-0.68
N=14
p=.007
-0.37
N=14
p=. 189
0.04
N=14
p=.891
£
NC0109 (BR)
0.01
N=ll
p=.984
0.04
N=ll
p=.910
-0.24
N=ll
p=.476

NC0207 (BR)
0.00
N=9
p=1.00
-0.40
N=9
p=.284
0.03
N=9
p=.936
u
NC0209 (BR)
0.00
N=7
p=1.00
0.20
N=7
p=.675
0.44
N=7
p=.325
NC0075 (P)
-0.51
N=7
p=.244
0.17
N=7
p=.719
-0.13
N=7
p=.786
o
£
NC0248 (P)
-0.49
N=7
p=.262
-0.75
N=7
p=.052
0.28
N=7
p=.549
1-53

-------
Table 18-7. Pearson product moment correlations of drier scenario richness metrics versus
PRISM mean annual precipitation for individual sites and site groups in Maine, Utah and North
Carolina. Highlighted correlations were significant (p<0.05).	
Drier Scenario Richness Metrics - PRISM mean annual precipitation
State
Site/ Site Group
WarmDrier VulnerableTax
Drier VulnerableTax
Drier WinTax
r
N
P
r
N
P
r
N
P

56817
0.45
N=23
p=.032
0.27
N=23
p=.217
-0.04
N=23
p=.865
Maine
57011
0.07
N=12
p=.833
0.49
N=12
p=.105
0.37
N=12
p=.237
57065
-0.51
N=9
Jl
C\
o
-0.49
N=9
p=. 181
-0.08
N=9
p=.848
Laur
-0.52
N=8
p=.183
-0.42
N=8
p=.298
-0.08
N=8
p=.853

NEHigh
0.09
N=8
p=.832
0.27
N=8
p=.513
-0.49
N=8
p=.217

4927250
-0.19
N=17
p=.474
-0.12
N=17
p=.656
0.42
N=17
p=.090

4936750
0.60
N=12
p=.037
0.31
N=12
p=.334
0.00
N=12
p=1.00

4951200
0.24
N=15
p=.392
0.24
N=15
p=.384
0.47
N=15
p=.078

5940440
-0.11
N=9
p=.785
0.22
N=9
p=.578
-0.04
N=9
p=.928
WU SF
-0.12
N=20
p=.622
0.03
N=20
p=.885
0.10
N=20
p=.676

WU ME
0.16
N=12
p=.631
-0.18
N=12
p=.569
0.31
N=12
p=.331

CP
-0.01
N=14
p=.972
-0.03
N=14
p=.910
0.45
N=14
p=. 103
£
NC0109 (BR)
0.24
N=ll
p=.477
-0.53
N=ll
p=.091
-0.33
N=ll
p=.324

NC0207 (BR)
0.04
N=9
p=.920
-0.22
N=9
p=.577
0.00
N=9
p=1.00
u
NC0209 (BR)
0.10
N=7
p=.838
-0.02
N=7
p=.963
0.00
N=7
p=1.00
NC0075 (P)
0.01
N=7
p=.977
0.18
N=7
p=.692
0.19
N=7
p=.690
o
£
NC0248 (P)
0.07
N=7
p=.876
0.23
N=7
p=.618
0.15
N=7
p=.744
1-54

-------
Table 18-8. Pearson product moment correlations of drier scenario % individual metrics versus
PRISM mean annual precipitation for individual sites and site groups in Maine, Utah and North
Carolina. Highlighted correlations were significant (p<0.05). The Blue Ridge Drier WinPct
entry was NA (=not available) because of a low sample size/lack of drier-robust individuals in
% Drier Scenario Metrics - PRISM mean annual precipitation
State
Site/ Site Group
Drier WinPct
Drier VulnerablePct
WarmDrier VulnerablePct
r
N
P
r
N
P
r
N
P

56817
-0.03
N=23
p=.877
0.06
N=23
p=.780
0.59
N=23
p=.003
Maine
57011
0.48
N=12
p=. 115
0.31
N=12
p=.322
-0.21
N=12
p=.517
57065
0.05
N=9
p=.904
-0.01
N=9
p=.979
-0.06
N=9
p=.869
Laur
0.09
N=8
p=.832
0.16
N=8
p=.703
-0.35
N=8
p=.390

NEHigh
-0.63
N=8
p=.095
-0.01
N=8
p=.987
0.39
N=8
p=.338

4927250
0.33
N=17
Jl
VO
VO
0.23
N=17
p=.372
0.08
N=17
p=.763

4936750
0.00
N=12
p=1.00
0.52
N=12
p=.082
0.23
N=12
p=.465

4951200
0.42
N=15
p=. 119
-0.46
N=15
p=.085
0.13
N=15
p=.634

5940440
-0.28
N=9
p=.463
0.28
N=9
p=.460
0.53
N=9
p=.141
WU SF
0.04
N=20
p=.853
-0.03
N=20
p=.909
-0.27
N=20
p=.242

WU ME
0.31
N=12
p=.331
-0.62
N=12
p=.030
-0.19
N=12
p=.563

CP
0.26
N=14
p=.379
0.13
N=14
p=.657
-0.09
N=14
p=.747
M
o
NC0109 (BR)
-0.11
N=ll
p=.748
0.16
N=ll
p=.639
0.57
N=ll
p=.067
NC0207 (BR)
0.00
N=9
p=1.00
0.10
N=9
p=.806
-0.20
N=9
p=.614
a
o
NC0209 (BR)
0.00
N=7
p=1.00
0.55
N=7
p=. 196
0.45
N=7
p=.310

NC0075 (P)
0.23
N=7
p=.624
-0.22
N=7
p=.636
0.47
N=7
p=.283
£
NC0248 (P)
0.16
N=7
p=.727
0.40
N=7
p=.368
-0.32
N=7
p=.479
1-55

-------
APPENDIX J
Case Studies
Part of this project involved doing 3 case studies. One was a case study on the combined effects
of climate change and urbanization on stream condition in the North Carolina Piedmont
physiographic region. In the second study, we compared hydrologic response to fluctuating
climate with land use effects in the Mid-Atlantic region. In the third, data from Florida reference
sites were analyzed to assess the vulnerability of reference condition and biological monitoring
to climate change and increasing population densities. This Appendix contains summaries of
each of the case studies.
Jl. Combined effects of climate change and urbanization on stream
condition (North Carolina Piedmont physiographic region)
J2. Another face of the changing climate: comparing hydrologic response to
fluctuating climate with land use effects (Mid-Atlantic region)
J3. Shifting Baselines of Perception: Vulnerability of Reference Condition to
Climate and Land Use Change (Florida Reference Sites)
J-l

-------
Jl. Combined Effects of Climate Change and Urbanization on Stream Condition
Data from the North Carolina Piedmont physiographic region were used in this case
study. Locations of the sampling sites are shown in Figure Jl-1. The study area has undergone
rapid population growth and urbanization since 1945, which has contributed to flashier streams
and altered habitat. Data preparation for the study involved developing operational taxonomic
units (OTUs), calculating taxa richness-based metrics, calculating Indicator of Hydrologic
Alteration (IHA) parameters (Richter et al., 1996) and Baker's Flashiness Index (Baker et al.,
2004) for 67 biological sampling sites that were associated with USGS gage stations, and
dividing the sites into natural, urban, agricultural and other land-use categories based on quick
examination of the watersheds in Google Earth.
The main objective of this study was to assess the response of macroinvertebrates in
urban and non-urban streams to hydrologic changes. We used number of EPT taxa as the
principal response metric and flashiness (sum of daily flow changes divided by total flow), low
pulse count (number of events per year where flow is below the 25th percentile ) and 1-day
minimum flow as the hydrologic indicators. Flashiness is predicted to increase with urbanization
but not with climate change, while low pulse count and 1-day minimum flow are predicted to
increase with climate change.
Results showed that the number of EPT taxa was strongly associated with flashiness. As
expected, the urban streams were flashier than the non-urban streams (Figures Jl-2 through Jl-
4). The flashiest urban streams had poorer condition than the moderately flashy urban streams
(Figure Jl-4). In the plots it appears that there may be a possible threshold at 0.5 (sites that had
flashiness values of less the 0.5 generally showed no relationship, while sites with flashiness
values greater than 0.5 generally showed strong relationships).
Natural and urban streams did not differ greatly in low pulse count, although the Smith
River is an important exception. This site is dominated by natural land cover but has extremely
high low pulse counts (28-44 per year) because it is regulated by a peaking hydropower dam.
Overall results show that there was not a strong relationship between low pulse count and
number of EPT taxa (Figures Jl-5 through Jl-7). Low pulse count was most strongly associated
with EPT taxa loss when there was an extreme increase in frequency of low pulses (>20 per
year), which occurred at the Smith River site as mentioned above.
J-2

-------
The urban streams had lower 1-day minimum flows than natural streams (Figures Jl-8
through Jl-10). However, within the urban sites, there was no association between number of
EPT taxa and minimum flow. In the plots, it appears that there may be a possible threshold at
15%, but this is confounded by the association of minimum flows with flashiness.
There were several conclusions that were drawn from this study, and also several
questions that remained unanswered. We are aware the flow regime is a causal link that changes
habitat, but we are uncertain as to whether or not it is a direct stressor. In this study,
intermediate-term changes in flow were not associated with taxa change within streams, but this
analysis had low power. The biological responses that were seen indicate that natural stream
communities are highly resilient within the range of natural hydrologic variability. Because of
this resilience, we may be unlikely to see effects from hydrologic changes associated with
climate change unless these changes are truly extreme, such as those that occurred in the
regulated river in this study, or occur in concert with rising temperatures that cross biological
thresholds. Future climatic changes are likely to be beyond the variability observed in the recent
past. Therefore, historic patterns are likely not as extreme as projected variability, and this makes
it difficult to predict future impacts. A powerpoint presentation of this case study is available
upon request.
Figure Jl-1. Locations of the North Carolina Piedmont stream sites that were used in the case
study.
J-3

-------
50
All Streams
45
40
35
05 30
x
05
n*
Q.
LU 20
a **;
A ^ •*
\r*. . ~ ~
6 ~
^4'i® a/4." "
1 V.^ V
¦ tfA
0.0
0.2
0.4
00 OS
Flashiness
1.0
Stable
12
1.4 .
Natural
Agriculture
Urban
Other
Flashy
Figure Jl-2. Number of EPT taxa is negatively associated with flashiness.
Natural Land Cwer Dominant
r = 0.0974
05 30
LLI 20
OA 0J5
Flashiness
Stable
"O. Oari R
C., Jaoob Fk
Unville R
"*v First Broad R
Roaring R
*.. Smith R
Hy:c Cr
, Mayt> R
"IX Other
Flashy
Figure Jl-3. Association between number of EPT taxa and flashiness at sites dominated by
natural land cover.
J-4

-------
Urban Dominated Streams
50 										
40
r = 0.2043
£ 30
to
Q_
UJ 20
10
0.0 0.2 0.4 DJ8 OS 10
Flashiness
Stable
"XX Abotts Cr
X TarR
*». Marsh Cr
Qabtree Cr
N Buffalo Cr
	 Morgan Q-
, , , AN Second Cr
pother
Flashy
Figure Jl-4. Association between number of EPT taxa and flashiness at sites dominated by
urban land cover.
All Streams
50
05 30 -
Q_
uj :o
10
• J.

* *
20	30
Lew pulse count
50
*	Natural
&	Agriculture
¦	Urban
~	Other
Stable
Unstable
Figure Jl-5. Association between number of EPT taxa and low pulse count at all streams.
J-5

-------
Natural Cwer Dominant
r - 0.2557
nj 30
W 20
2D	30
Low pulse count
s
Dan R
"V
Jao&b Fk

LinvtUe R

First Broad R

Roaring R

Sfrith R

Hycocr
X
Mayo R
X
Other
Stable
Unstable
Figure Jl-6. Association between number of EPT taxa and low pulse count at sites dominated
by natural land cover.
Urban Dominant
(- = 0.1292
. fbbots Or
Tar ft
> Wtsrsh Cr
Cratrtree Cr
N Buffalo R
Morgan Cr
~ s+^ N Second Cr
Other
Stable
Unstable
Figure Jl-7. Association between number of EPT taxa and low pulse count at sites dominated
by urban land cover.
J-6

-------
All sites
50
40
m w
%
S 30
H
Q.
LU 20
*
&
«?* ft A
«&A %	A ^
**«». v f ^ &
ffl
2?
&r-a ~
10 A/ f V* ¦*'
_ A ¦. *.
ft** *
" * \
0.0 0.1 0.2 0.3 0.4 0.5
1-day minimum flow
0.6 0.7
Wetter
•	Ha tural
fi	Agricultwe
¦	Utban
~	Oth«i
Figure Jl-8. Association between number of EPT taxa and 1-day minimum flow at all streams.
Natural Cover Dominant
|r= 0.0543
• A
"J 30
^ *P
02 03 0.4 05
1-day minimum flow
"Cw. Dan R
"a. Jacob Fk
UfiviBe R
"V First 0ro3d R
Rearing R
"*. Smith R
_ Hvco Cr
D1 Kl, May> R
\X Other
Figure Jl-9. Association between number of EPT taxa and 1-day minimum flow at sites
dominated by natural land cover.
J-7

-------
Urban Dominant
r = 0.0881
Ga _ .
3 +
db & •€>
~Kk Abbotts Cr
"4, Tar R
fofersh Cr
Crabtree cr
V H Buffalo Cr
sJ Morgan Cr
n 7 N Second Cr
"0. Other
00
0.2
0.3
0.1
0.4
0.5
0.6
1-day minimum flow
Drier	Wetter
Figure Jl-10. Association between number of EPT taxa and 1-day minimum flow at sites
dominated by urban land cover.
J2. Comparing Hydrologic Respone to Fluctuating Climate with Land-Use Effects
Flow data from USGS gages in the Baltimore-Washington D.C. area (Mid-Atlantic
region) were used in this case study. The main question that was addressed was how hydrologic
response to climatic change in the Mid-Atlantic would compare with land-use impacts. Data
preparation involved gathering historical flow and precipitation data for urban and forested sites,
calculating Baker's Flashiness Index (Baker et al., 2004) and IHA parameters for these sites, and
identifying which historical years of data had conditions that most resembled those that are
projected to occur in the future. Data were analyzed using ANOVA analyses.
Results are summarized in Figures J2-1 and J2-2. They show that for high flow metrics,
climate effects were small relative to land use change, while for low flow metrics, climate
change effects were large relative to land use. Plots of the ANOVA results for some of the IHA
parameters are also included. Figure J2-3 provides guidance on how to interpret these plots,
Figures J2-4 through J2-7 show results for high flow IHA parameters and Figures J2-8 through
J2-13 show results for low flow IHA parameters. Overall conclusions were that climate will
affect stream flow. This will be happening over an ongoing dramatic change in land use, and the
J-8

-------
effects of climate change will be felt to differing degrees relative to land use change. A
powerpoint presentation of this case study is available upon request.
High Flow Metrics	Land Use Climate
Flashmess
Y
N
High Pulse Count/Duration
Y
N
1 day max
Y
N
3 day max/7 day max
H
N
Rise rate/Fall rate
Y
N
Reversals
Y
N
High Flood Peak/Frequency/Duration
Y
N
Small Flood Peak/Duration
Y
N
Land Use Swamps Climate Effects
Climate; Magnitude NA; Frequency NA; Duration NA; Timing IMA; Rate of Change NA
Land Use: Magnitude 7; Frequency t; Duration ]; Timing NA; Rate of Change t
Figure J2-1. Summary of ANOVA results for high flow IHA metrics.
Low Flow Metrics
Land Use
Climate
Low Pulse Count
Y
Y
Low Pulse Duration
Y
N
1 day/3 day/7 day min
N
Y
Extreme Low Peak
N
N
Extreme Low Frequency/Duration
Y
Y
Climate Swamps Land Use Effects

Climate: Maanitude i: Freauencv t: Duration Timino 1: Rale of Chance NA
Land Use: Maanitude NA: Freauencv f: Duration uTimtnaNA: Rate of ChanoeNA
Figure J2-2. Summary of ANOVA results for low flow IHA metrics.
J-9

-------
Normal
Urban
Land Use-Yes
Climate-No
Land Use-No
Climate-Yes
Land Use-Yes
Climate-Yes
Normal
Forest Urban
Normal
Forest
Urban
Figure J2-3. Aid for interpreting the A NOVA plots.
c
-
vn
& 05
=£ Climate
mm
-T- Citrate
FREQ
Forest
Urban
Figure J2-4. ANOVA results for flashiness at forested and urban sites.
J-10

-------
Forest
LH>an
IU
3E Climae
_ NORM
31 Climae
FREQ
Figure J2-5. ANOVA results for high pulse count at forested and urban sites.
Forest
Utan
LU
Cfirrste
NORM
51 Cfirrate
FREQ
Figure J2-6. ANOVA results for high pulse duration at forested and urban sites.
J-ll

-------
Forest
Lrt>an
IU
^ Climate
NORM
i5Z Climate
FREQ
Figure J2-7. ANOVA results for 1-day maximum flow at forested and urban sites.
Forest
Ufcan
*
5E Climate
NO RM
31 Climate
FREQ
Figure J2-8. ANOVA results for low pulse count at forested and urban sites.

-------
11
Figure J2-9. ANOVA results for low pulse duration at forested and urban sites.
Forest
Urtan
•i?' Ctirrete
FREQ
3E Climate
NOFM
Forest
Urban
LU
~
^ Climate
NORM
3: Ciitnae
FREQ
Figure J2-10. ANOVA results for 1-day minimum flow at forested and urban sites.
J-13

-------
020
0.18
016
0.14
| 0.12
fw
0.10
0.08
0.06
004
Forest
Oban
LU
~
Climate
NO RM
33 Climate
FREQ
Figure J2-11. ANOYA results for 7-day minimum flow at forested and urban sites.
Forest
titan
LU
*
3E Clirrsrte
NO FM
3Z Climate
FREQ
Figure J2-12. ANOVA results for extreme low flow frequency at forested and urban sites.
J-14

-------
2
I
Fonest
Ut>an
LU
CKmste
_ NORM
-T- Climate
~ FRGQ
Figure J2-13. ANOYA results for extreme low flow duration at forested and urban sites.
J3. Shifting Baselines of Perception: Vulnerability of Reference Condition to Climate and
Land Use Change
The vulnerability of reference locations to climate change and the need to protect
reference locations is recognized as an important issue for the future of bioassessment. Our
concept of the natural condition of streams is based on these reference locations, yet we also
recognize that they have been subject to industrial anthropogenic influences and disturbance for
up to 2 centuries. While reference sites may often be located in remote and less developed
regions, they are nevertheless vulnerable to human development and urbanization.
We examined actual and potential reference sites for aquatic biological monitoring, and
examined both their regional vulnerability to future climate changes, as well as vulnerability to
land-use changes. Where possible, we also examined the degree of change from pre-European
settlement in North America to current reference condition. Florida was chosen for this case
study because its historical pollution has been less than in other eastern states, but its current-era
growth and urbanization has been extraordinary.
J-15

-------
In this study we examined 54 Florida reference sites under future growth scenarios: A1
(IPCC) (rapid global economic growth and lower population growth), A2 (IPCC) (slower
economic growth and higher population growth) and Base (closest to current conditions)
(Nakicenovic and Swart, 2000; USEPA, 2009a). Projections for population growth under each
scenario are shown in Figure J3-1. To assess vulnerability, land use within a 1 km buffer around
each reference site was calculated, land use for each decade was projected from the A2 and Base
Case scenario outputs, and the fraction of buffer in categories of increasing housing density
were estimated.
The link between population density and biota has been previously examined in New
England, as shown in Figure J3-2. Results showed that effects begin but are not universal or
severe when densities reach 50 people per square mile (25 houses). A degradation gradient
becomes evident at densities of 50-500 people per square mile (25-250 houses). Once densities
exceed 500 people per square mile (>250 houses), streams in New England are generally
degraded.
Results of the housing density and fraction suburban projections for sites in Florida are
shown in Figures J3-3 through J3-5. Based on the New England results, the average site
(statewide) in Florida will approach the 'complete degradation' point by 2100. The average
reference site will exceed the 'effects threshold' around 2020 but will not reach the 'complete
degradation' point. Seventeen percent of the reference sites appear to be protected in that they
are surrounded by government land or water and approximately 25% of the reference sites
appear to be completely unprotected from development. A PowerPoint presentation of this case
study is available upon request.
J-16

-------
US Population Growth
800 0
700 0
o
| 6000
g 3000 +- '•
3 200 0
a
£ 1000
00
2 900 0
4300


Base Case
2000 2020 2D4D
2060
Yesr
2080 2100 2120
Figure J3-1. Projections of future population growth under the 3 scenarios: Al, A2 and Base.
05
X
TO
35
30
25
o
20
03
8
o. 15
o
i_
0)
E 10
. ^.
• ~,
¦	CT
¦	Rl
*	VT
" ME
*	NH
*	MA
50	90.0	500.0
Population Density, /mi2
Figure J3-2. The link between population density and biota in New England.
J-

-------
All Sites
v> 200
o> 150
Base Case
A2
2000
2020
2080
2100
2O40 2060
Year
Figure J3-3. A2 and Base Case projections for mean housing density per square mile at all
Florida sites.
Reference Sites
2000 2020
2080 2100
Base Case
A2
2043 2090
Year
Figure J3-4. A2 and Base Case projections for mean housing density per square mile at Florida
reference sites.
J-18

-------
Reference Sites
cu 0.20
¦ Base Case
1 A2
2040 2060 2080 2100
Year
J3-5. A2 and Base Case projections for fraction suburban land use at Florida reference sites.
2000 2020
J-19

-------
APPENDIX K
2
3	Explorations of relationships between
4	hydrological and biological data
5
6	The intent of this appendix is to provide more comprehensive and detailed information on the
7	analyses that were performed on the Utah and North Carolina hydrological data. Some of the
8	analyses that are covered in this appendix are also referenced in the main body of the report.
9	When this occurred, attempts were made to reduce any overlap or duplication in the reporting of
10	results.
11
12	Kl. Overview
13	K2. Utah Analyses
14	K3. North Carolina Analyses
K-l

-------
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
K1 Overview
Changes in hydrology are projected to occur as a result of climate change. In this study,
we attempted to gain a better understanding of the link between hydrology and biology by
creating and analyzing datasets comprised of paired hydrological and biological data. To derive
these datasets, we used the following criteria to match USGS gages with biological sampling
sites:
If a biological sampling site fell within 500-meters of the gage, the gage was retained and
matched with the biological sampling site (this was done using ArcGIS).
Stream gages were excluded if they were not on the same stream reach as the biological
sampling site, or if tributaries entered between the gage and site
All available data from the following time period was downloaded: 1940-01-01 to 2007-
12-31
Hydrologic data for the matched stations were downloaded from the USGS real-time
flow data website: http://waterdata.usgs.gov/nwis/rt. Indicators of Hydrologic Alteration (1HA)
software (version 7.0.4.0) was then used to calculate a suite of IHA parameters for each site. The
Richards-Baker Flashiness Index (RBI) (Baker et al., 2004) was also analyzed (the R code that
was used to calculate the RBI is available upon request). The RBI uses flow data to quantify the
frequency and rapidity of short-term changes in stream flow. The IHA and RBI data was then
matched with the biological data from the site. These merged datasets were then used in our
analyses. Descriptions of the analyses that were performed on the Utah and North Carolina
datasets are described in Sections L2 and L3 \ Only the subset of IHA parameters that were
believed to have greatest relevance to this study was used in our analyses. A list of these
parameters is shown in Table Kl-1.
1 Similar types of analyses were attempted in Maine, but there were not enough USGS gages associated with
biological sampling sites to make weighted average and ordination analyses worthwhile.
K-2

-------
1 Table Kl-1. Summary of IHA parameters used in biological analyses
Annual IHA parameters
Description
Conversion (to standardize)
monthly
median discharge (cfs)
divided by median value for entire period of gage data
1-day min
annual minima, 1-day mean (cfs)
divided value for each year by mean annual flow
3-day min
annual minima, 3-day means (cfs)
divided value for each year by mean annual flow
1-day max
annual maxima, 1-day mean (cfs)
divided value for each year by mean annual flow
3-day max
annual maxima, 3-day means (cfs)
divided value for each year by mean annual flow
Date min
Julian date of each annual 1-day minimum
none
Date max
Julian date of each annual 1-day maximum
none
Lo pulse #
Number of low pulses within each water year
none
Lo pulse L
Median duration of low pulses (days)
none
Hi pulse #
Number of high pulses within each water year
none
Hi pulse L
Median duration of high pulses (days)
none
Environmental Flow Components (EFC)
Xlowl peak
minimum ('peak') flow (cfs) during extreme low flow event (within
each year)
divided value for each year by mean annual flow
Xlowl dur
duration of extreme low flow event (days)
none
Xlowl time
Julian date of peak flow
none
Xlowl freq
frequency of extreme low flows during water year
none
K-3

-------
Highl peak
Highl dur
Highl time
Highl freq
Baseflow index
Number of reversals:
maximum ('peak') flow (cfs) during extreme high flow event (within
each year)
duration of extreme high flow event (days)
Julian date of peak flow
frequency of extreme high flows during water year
7-day minimum flow/mean flow for year
Number of hydrological reversals
divided value for each year by mean annual flow
none
none
none
none
none
2
3
Definitions-
high flow events
low flow events
extreme low flow
All flows above the 75th percentile of all flows are classified as high flow events
All flows less than or equal to the 50th percentile of all flows are classified as low flow events
10th percentile of all low flows
K-4

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Hydrographs were also generated for selected reference sites in Maine (see Attachment
Kl), Utah (see Attachment K2) and North Carolina (see Attachment K3). The R code that was
used to create these plots is available upon request. The hydrographs were used in the initial
phases of our analyses to gain a better understanding of the natural flow regimes in each of the 3
states. In addition, they can provide information on how strong the groundwater influence is at a
site. During discussions at the 2008 Workshop on Bioindicators and Climate Change in Crystal
City, VA,2 the importance of learning more about groundwater influence came up on several
occasions, so we attempted to gather groundwater data and incorporate it into our analyses.
Unfortunately we could not find the type of data that we needed. However, we were able to find
a number of valuable resources (i.e., NCDENR, 2004, NCDWQ, 2005, NCDENR, 2005,
Borwick et al., 2006, Douglas, 2006). Another potential lead, which was suggested to us by
Maine DEP, was to use water temperature data. If summer low flow temperatures were less than
20°C, there were generally believed to be at least some groundwater influence.
K2. Utah
A number of different analyses were run on a subset of Utah IHA-biological data that
was derived from 43 biological sampling sites (locations of these stations are shown in Figure
K2-1). The dataset was somewhat limited by sample size and by the fact that some sites had
many more years of data than others (i.e. one site had 19 years of data, others had 1 year of data).
One analysis involved examining taxonomical trends using Canonical Correspondence Analysis
(CCA) and Nonmetric Multidimensional Scaling (NMDS). In another analysis, a subset of data
that only had sites with multiple years of data was evaluated. A third analysis involved
calculating weighted average (WA) indicator values for the parameters that showed the strongest
influence on taxonomic composition. For the final analysis, correlation analyses were performed
on data from the 7 Utah stations that had the most number of years of biological-hydrological
data.
2 A report on the workshop is available online at: oaspub.epa.gov/eims/eimscomni.getfile?p_do\vn]oad_id=4861.53.
Additional information on the workshop can be found at: http://www .epa. gov/ncea/workshop/.
K-5

-------
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Results of the CCA are shown in Section 2 of the main report. Additional CCA results
(species trends) are shown in Figure K2-2. Results of the NMDS analyses are shown in Figure
K2-3. Both analyses indicated that year had the strongest influence on taxonomic composition.
However, when only data from sites with multiple years of data were used, year showed a
weaker effect in the NMDS analysis. Results from the WA model are shown in Tables K2-1 and
K2-2. Of all the models tested, most low-flow parameters performed better than high-flow
parameters. The WA model for year had very strong performance (r2=0.6). The next best
parameter was the IHA parameter for annual minima, 3-day means. Optima and tolerance results
for taxa that had more than 20 occurrences in the dataset (which is generally regarded as an
adequate sample size) show that Leuctridae, Asellidae and Zapada had the lowest values, while
Hyalella and Helicopsyche had the highest. Leuctridae and Zapada had relatively low tolerance
ranges, while Hyalella and Helicopysche had large tolerance ranges. These results suggest that
Leuctridae and Zapada are better adapted (perhaps partly due to their smaller sizes) to lower flow
conditions than other taxa.
There are too many results from the correlation analyses to show in this report, but results
are available upon request. There were a number of significant correlations, but none of the taxa,
trait metrics or IHA parameters showed consistent patterns across the 7 sites, which makes the
results very difficult to summarize. Site information for the sites is summarized in Table K2-3.
K-6

-------
46

£
<£>
14
i1
a a
Q
a
S&1®®5

a
a
o
a
2®
a
a

a
9 11
a *
a
a
a
a
,a
a
2
a
a'
a
a
a
1
a 1
3
47
48	Figure K2-1. Locations of the 43 Utah biological sampling stations (red triangles) and
49	associated USGS stream gages (yellow circles). Stations that are highlighted in blue are
50	classified as reference sites by Utah DWQ. The numbers next to the sites are the number of
51	years of data that were available for each station.
K-7

-------
Utah I HA
C\l
w
X
<
+
+
Leuctrid
+
Amphinem
Cultus
+
Taenione
+
+
Oemopter
Coiydalu
+
+ Agapetus
Holorusi
+
+
Petrophi
Maya trie
Cheumato
Neotrich
Traverel
+
Capniida
+ +,
Cinygmul
Zapada
Cinyqma
Ostracod	_j_
+	Daphnia
Ameletus
Heptagen
Isogeruji.
Polycent
Leptohyp
+
+ +
perop
Podmosta
+
+
Epeorus
Megarcys
-I- Hespefop
j Glossoso
Hydrophi
+

+
+
Asellida
+ +
Eupaiyph
+
+
+
^Vt
/~
+
++ +
+
ear
+
Hesperop
+
+
Coenagri
Hyalella +
+
+
-H-+"
Calopaiy
+
+
Potamopy
Nectopsy Physella
++
+
Wiedeman
+
Brachyce
+
Corixida
+
Gammarus
Forcipom
Helicops
+
Prostoia
Oligophl
Apatania
Heterlim
+
+
Pteronar
Clinocer
+
Lepidost
r Bibiocep
Narpus
Cleptelm,ria
^ t ^
-H-
Pisidium
Halipl us
Season
A 1
t 2
t 3
A 4
Axis 1
Figure K2-2. Species trends along year. These were derived from the CCA analysis.

-------
1
£
A
i
i
1
A
C\l
i
A
T	n	n	n	n	r
1980	1990	2000
Year
Axis 1	2000
r = .313 tau = .197
Axis 2
r = -.566 tau = -.405
1990
1980
Utah I HA
T T
T
T
Axis 1
A
A
AA
A
Figure K2-3. Taxonomical trends in the Utah dataset were examined using Nonmetric
Multidimensional Scaling (NMDS). Year had the strongest influence on taxonomical
composition. However, when NMDS ordinations were run on a selected subset of data that
only contained data from sites with multiple years of samples, the year trend was not as
strong.
K-9

-------
63	Table K2-1. Weighted average indicator values for year, which had very strong
64	performance (r2=0.6). Sorted by optimum values. Ranks are on a scale of 1 to 7 and are
65	based on the following percentiles: 0, 0.1, 0.25, 0.4, 0.6, 0.75, 0.9 and 1.	
YEAR
Taxa
Optimum
Tolerance
RankOpt
RankTol
Count
Traverella
1989.4
3.59
1
2
10
Taenionema
1989.5
4.38
1
3
29
Cultus
1989.8
3.66
1
2
20
Leuctridae
1990.1
4.50
1
4
24
Zapada
1990.2
4.55
1
4
35
Planaria
1990.9
3.11
1
2
90
Apatania
1991.3
3.71
1
2
20
Serrate 11a
1991.4
4.70
1
4
11
Nematoda
1991.5
3.44
1
2
125
Hesperoperla
1991.7
5.08
2
5
33
Ostracoda
1991.7
4.20
2
3
96
Cinygmula
1991.8
4.62
2
4
90
Copepoda
1991.9
3.46
2
2
35
Pelecypoda
1992.0
3.02
2
2
44
Capniidae
1992.0
5.28
2
5
38
Ameletus
1992.1
6.25
2
7
26
Mayatrichia/Neotrichia
1992.2
1.89
2
1
16
Alisotrichia/Leucotricia
1992.2
2.66
2
2
32
Heptagenia
1992.2
2.67
2
2
58
Neotrichia
1992.2
2.04
2
1
12
Micrasema
1992.3
5.61
2
6
55
Glossosoma
1992.4
4.13
3
3
60
Podmosta
1992.6
3.78
3
3
10
Dicranota
1992.6
5.57
3
6
32
Cheumatopsyche
1992.7
5.20
3
5
55
Agapetus/Culoptila/Protoptila
1992.7
3.15
3
2
12
Ephemerella
1992.7
4.53
3
4
149
Epeorus
1992.9
4.99
3
4
92
Dytiscidae
1993.0
6.80
3
7
10
Euparyphus
1993.1
2.43
3
1
12
Skwala
1993.2
5.27
3
5
31
Neothremma
1993.3
7.28
3
7
19
Leucotrichia
1993.4
1.99
3
1
23
Paraleptophlebia
1993.5
5.07
4
4
96
Pericoma
1993.6
5.41
4
6
47
Simuliidae
1993.6
5.13
4
5
234
Chloroperlidae
1993.6
5.82
4
7
105
K-10

-------
66 Table K2-1. Continued
Taxa
Optimum
Tolerance
RankOpt
RankTol
Count
Leptohyphidae
1993.7
4.89
4
4
133
Drunella
1993.8
5.61
4
6
119
Hemerodromia
1993.8
3.84
4
3
103
Atherix
1993.9
5.85
4
7
81
Baetidae
1994.0
5.27
4
5
277
Bezzia
1994.0
2.66
4
2
53
Pteronarcella
1994.0
5.73
4
6
91
Isoperla
1994.0
4.87
4
4
105
Isogenoides
1994.1
5.74
4
6
19
Hydroptila
1994.1
4.49
4
4
97
Physa
1994.1
5.00
4
4
54
Antocha
1994.2
5.27
4
5
126
Acarina
1994.3
5.47
4
6
268
Oligophlebodes
1994.3
4.80
5
4
35
Tubificidae
1994.4
2.16
5
1
107
Hydropsyche
1994.5
5.38
5
5
232
Planorbidae
1994.5
4.04
5
3
37
Lymnaea
1994.5
3.85
5
3
15
Chironomidae
1994.5
5.63
5
6
291
Rhyacophilidae
1994.5
5.41
5
6
98
Rhithrogena
1994.5
5.69
5
6
127
Petrophila
1994.6
4.04
5
3
36
Chelifera
1994.6
3.62
5
2
98
Oecetis
1994.7
4.72
5
4
45
Hirudinea
1994.7
5.14
5
5
75
Arctopsyche
1994.8
5.22
6
5
99
Hexatoma
1995.2
5.18
6
5
88
Brachycentrus
1995.3
5.39
6
6
145
Asellidae
1995.4
4.84
6
4
45
Hyalella
1995.4
4.39
6
3
62
Lepido stoma
1995.7
4.68
6
4
88
Ambrysus
1995.8
3.80
6
3
17
Helicopsyche
1995.9
4.33
6
3
68
Gammarus
1995.9
6.89
6
7
15
Claassenia
1996.0
6.40
6
7
12
Hesperophylax
1996.1
6.31
6
7
12
Coenagrionidae
1996.1
5.52
6
6
36
Bibiocephala
1996.3
5.31
7
5
17
Optioservus
1996.5
4.45
7
4
148
Zaitzevia
1996.6
4.46
7
4
97
K-ll

-------
67 Table K2-1. Continued
Taxa
Optimum
Tolerance
RankOpt
RankTol
Count
Pteronarcys
1996.7
5.75
7
7
27
Tipula
1998.1
3.80
7
3
31
Physella
2000.5
1.55
7
1
13
Forcipomyia/Probezzia
2001.4
1.80
7
1
20
Microcylloepus
2001.6
2.32
7
1
10
Pisidium
2002.3
1.40
7
1
16
68
69
70	Table K2-2. Weighted average indicator values for annual minima, 3-day means, which
71	had relatively strong performance	
3-DAY ANNUAL MINIMA
Taxa
Optimum
Tolerance
RankOpt
RankTol
Count
Pisidium
0.030
0.04
1
2
16
Ambrysus
0.041
0.05
1
3
17
Mayatrichia/Neotrichia
0.045
0.03
1
2
16
Neotrichia
0.046
0.04
1
2
12
Leuctridae
0.049
0.03
1
1
24
Asellidae
0.050
0.06
1
4
45
Lymnaea
0.056
0.04
1
3
15
Zapada
0.057
0.04
1
3
35
Neothremma
0.059
0.04
1
3
19
Physella
0.060
0.06
2
5
13
Skwala
0.061
0.02
2
1
31
Petrophila
0.062
0.05
2
4
36
Coenagrionidae
0.064
0.07
2
6
36
Bibiocephala
0.065
0.01
2
1
17
Cultus
0.066
0.04
2
3
20
Serratella
0.067
0.04
2
2
11
Dytiscidae
0.068
0.04
2
2
10
Pelecypoda
0.069
0.06
2
5
44
Hesperoperla
0.069
0.05
2
4
33
Epeorus
0.070
0.04
2
2
92
Physa
0.071
0.06
2
5
54
Claassenia
0.072
0.03
3
1
12
Podmosta
0.072
0.03
3
1
10
Tipula
0.072
0.05
3
4
31
Capniidae
0.073
0.05
3
4
38
Apatania
0.073
0.02
3
1
20
Oecetis
0.073
0.04
3
2
45
K-12

-------
Table K2-2. Continued
Taxa
Optimum
Tolerance
RankOpt
RankTol
Count
Baetidae
0.073
0.06
3
6
277
Heptagenia
0.075
0.05
3
4
58
Pteronarcella
0.076
0.04
3
2
91
Ephemerella
0.076
0.05
3
4
149
Chloroperlidae
0.076
0.04
3
2
105
Hemerodromia
0.076
0.07
3
6
103
Antocha
0.077
0.05
4
3
126
Ostracoda
0.077
0.06
4
5
96
Lepidostoma
0.077
0.05
4
4
88
Paraleptophlebia
0.078
0.04
4
2
96
Arctopsyche
0.078
0.05
4
3
99
Rhithrogena
0.078
0.04
4
3
127
Simuliidae
0.079
0.06
4
5
234
Chelifera
0.079
0.06
4
5
98
Isoperla
0.080
0.04
4
3
105
Cheumatopsyche
0.080
0.07
4
6
55
Rhyacophilidae
0.080
0.05
4
4
98
Cinygmula
0.080
0.05
4
3
90
Optioservus
0.080
0.06
4
5
148
Glossosoma
0.081
0.05
4
4
60
Acarina
0.081
0.06
4
5
268
Zaitzevia
0.081
0.05
4
4
97
Planaria
0.082
0.07
4
7
90
Leptohyphidae
0.082
0.07
5
6
133
Ameletus
0.082
0.05
5
4
26
Hydroptila
0.082
0.06
5
6
97
Nematoda
0.082
0.06
5
6
125
Hexatoma
0.082
0.03
5
2
88
Hydropsyche
0.083
0.06
5
5
232
Taenionema
0.083
0.04
5
3
29
Copepoda
0.084
0.07
5
6
35
Microcylloepus
0.085
0.04
5
3
10
Leucotrichia
0.085
0.06
5
5
23
Chironomidae
0.085
0.07
5
6
291
Euparyphus
0.086
0.10
5
7
12
Isogenoides
0.086
0.04
6
2
19
Drunella
0.087
0.05
6
4
119
Dicranota
0.089
0.05
6
4
32
Tubificidae
0.090
0.06
6
5
107
Pteronarcys
0.090
0.03
6
1
27

-------
74 Table K2-2. Continued
Taxa
Optimum
Tolerance
RankOpt
RankTol
Count
Atherix
0.091
0.05
6
4
81
Planorbidae
0.091
0.08
6
7
37
Alisotrichia/Leucotricia
0.091
0.06
6
6
32
Micrasema
0.092
0.05
6
4
55
Brachycentrus
0.093
0.06
6
5
145
Hirudinea
0.094
0.09
6
7
75
Oligophlebodes
0.094
0.05
6
4
35
Forcipomyia/Probezzia
0.094
0.08
7
7
20
Agapetus/Culoptila/Protoptila
0.097
0.03
7
1
12
Pericoma
0.100
0.07
7
6
47
Bezzia
0.103
0.08
7
7
53
Helicopsyche
0.110
0.08
7
7
68
Hyalella
0.111
0.09
7
7
62
Traverella
0.116
0.03
7
1
10
Hesperophylax
0.159
0.08
7
7
12
Gammarus
0.170
0.07
7
6
15
75
K-14

-------
Table K2-3. Data that was used in the Utah correlation analyses was gathered from these biological sampling stations/USGS
BioStationID
USGS
gage
# Yrs of
data
Elevft
Eco_L3
Eco_L4
Ref Status
%URB
%AGR
%FOR
4926350
10131000
14
5573.3
Wasatch and Uinta
Mountains
Mountain Valleys
TRASH
32.5
27.9
30.2
4934100
9302000
12
4762.6
Colorado Plateaus
Uinta Basin Floor
UNKNOWN
3.9
18.4
24
4937900
9261000
14
4766.1
Colorado Plateaus
Uinta Basin Floor
SO-SO
0
20.3
65.1
4954380
9330000
19
6940.5
Wasatch and Uinta
Mountains
Seiniarid Foothills
TRASH
6.9
30.3
56
4996690
10163000
17
4521.3
Central Basin and
Range
Moist Wasatch Front
Footslopes
TRASH
73.2
15.8
5.3
4998400
10154200
18
6971.4
Wasatch and Uinta
Mountains
Mid-elevation Uinta
Mountains
SO-SO
5.7
0.7
93.6
5940440
10234500
11
6249.3
Wasatch and Uinta
Mountains
Seiniarid Foothills
REF
3.9
0
96.1
K-15

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
K3. North Carolina
A number of different analyses were run on a subset of North Carolina IHA-biological
data.
One analysis involved examining taxonomical trends using NMDS. One set of results is
shown in Section 2 of the report. Additional results are shown in Figures K3-1 through K3-3.
They show that baseflow index (a parameter representing low-flow influence) had the strongest
correlation with macroinvertebrate species composition, though this relationship may be mostly
due to ecoregional distributions of taxa. A number of covariates, such as elevation, temperature,
and other factors may co-affect the observed pattern. The second IHA parameter that related to
taxonomical compositions was number of reversals, which is a measurement of flashiness. The
RBI had weaker correlation with species axes. Other factors that showed correlations were low
pulse and high pulse parameters. Selected results from the Pearson and Kendall Correlations with
Ordination Axes are shown in Table Kl-1.
IHA parameter inference models. According to the NMDS ordination, the most important
parameters associated with species compositions are baseflow index, number of reversals, and
RBI (which is much weaker compared to the previous two). Inference models were developed
for these three parameters using both R and C2 (Table K3-1).
Additional analyses were performed on this dataset to generate species response curves
for baseflow index (Attachment K4), number of reversals (Attachment K5) and RBI (Attachment
KL6). These were derived from a generalized linear model (GLM) output (Yuan 2006). The y-
axis shows the probability of capture for a single taxon, and the gradient of environmental
variables is represented on the x-axis. The curve is the GLM fitting into the dataset.
K-16

-------
27	Table K3-1. According to the NMDS ordination, the most important parameters
28	associated with species compositions are baseflow index, number of reversals, and RBI
29	(which is much weaker compared to the previous two). Inference models were developed
30	for these three parameters using both R and C2. The final reported indicator values were
31	based on R results.

R2
RMSE

Baseflow
index
Number of
reversal
RBI
Baseflow
index
Number of
reversal
RBI
C2
0.556
0.413
0.437
0.149
0.135
0.227
Bootstrap
0.492
0.245
0.369
0.154
0.141
0.219
32
33
K-17

-------
34
35
36
37
38
39
C\l
CO
X
<
~ 0 0 0 0 0 0 0 ~
0.0 0.2 0.4 0.6 0.6
Baseflow
Axis 1
r = -.510 tau = -.382
Axis 2
r = .416 tau = .317
0.6
0.4
0.2
0.0
Level3

Axis 1
Figure K3-1. NMDS of macroinvertebrate taxonomical composition and its relationship
with the baseflow index. Samples are grouped by level 3 ecoregion. Only samples collected
using the standard qualitative/full-scale method were used in this analysis.
K-18

-------
100	200
Reversals
Axis 1
r = -.512 tau = -.358
Axis 2
r = .039 tau = .011
41
42
43
44
45
46
47
C\l
CO
X
<
Level3

Axis 1
200
100
0 ¦

Figure K3-2. NMDS of macroinvertebrate taxonomical composition and its relationship
with the number of reversals index. Samples are grouped by level 3 ecoregion. Only
samples collected using the standard qualitative/full-scale method were used in this
analysis.
K-19

-------
49
50
51
52
53
54
0.0
0.4
1.2
r =
RBI
Axis 1
.046 tau
Axis 2
.391 tau
.017
-.211
1.2
O.J
0.4
0.0
Level3
T_ ^ ~
t
Axis 1
'	i A
Figure K3-3. NMDS of macroinvertebrate taxonomical composition and its relationship
with the Richards-Baker Flashiness Index (R-B Flashiness Index). Samples are grouped by
level 3 ecoregion. Only samples collected using the standard qualitative/full-scale method
were used in this analysis.
K-20

-------