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DO NOT CITE OR QUOTE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
|