.R-11/036F I Sept
United States
Environmental Protectioi
Agency
...plications of Climate Change for
Bioassessment Programs and Approaches
to Account for Effects
!
~>
.*
United States Environmental Protection Agency
Office of Research and Development, National Center for Environmental Assessment
-------
EPA/600/R-11/036F
September 2012
Implications of Climate Change for Bioassessment Programs and
Approaches to Account for Effects
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 has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
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, is anchored in comparison
to regionally established reference benchmarks of ecological condition. Climate change will
affect responses and interpretation of these indicators and metrics at both reference and
nonreference 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). (2012) 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.
ii
-------
TABLE OF CONTENTS
LIST OF TABLES VIII
LIST OF FIGURES XVIII
LIST OF ABBREVIATIONS AND ACRONYMS XXVII
PREFACE XXIX
AUTHORS, CONTRIBUTORS, AND REVIEWERS XXX
EXECUTIVE SUMMARY XXXI
1. INTRODUCTION 1-1
1.1. DECISION CONTEXT 1-2
1.2. CLIMATE CHANGE EFFECTS AND ECOLOGICAL RESPONSES 1-5
1.2.1. Expectations for Thermal Regime Changes and Associated
Biological Responses 1-9
1.2.2. Expectations for Hydrologic Changes and Associated Biological
Responses 1-10
1.3. CONCEPTUAL MODEL LINKAGES STRUCTURING THE STUDY
APPROACH 1-12
2. METHODS 2-1
2.1. DATA GATHERING 2-1
2.1.1. Exposure Data 2-1
2.1.2. Temperature and Streamflow Data 2-1
2.1.3. Biological Data 2-3
2.1.4. Site Information 2-6
2.2. DERIVATION OF INDICATORS 2-6
2.2.1. Thermal Preferences 2-6
2.2.2. Hydrologic Indicators 2-10
2.2.3. Traits-Based Indicators in Future Scenarios 2-12
2.3. LEAST-DISTURBED LONG-TERM BIOLOGICAL MONITORING SITES.... 2-14
2.4. EVIDENCE OF TRENDS AT LEAST-DISTURBED LONG-TERM
MONITORING SITES 2-15
2.4.1. Temporal Trends in Climatic and Biological Variables 2-15
2.4.2. Associations Between Biological Variables and Climatic Variables 2-17
2.4.3. Groupings Based on Climatic Variables 2-17
2.5. SENSITIVITY OF BENTHIC MACROINVERTEBRATES TO
TEMPERATURE 2-20
2.6. IMPLICATIONS FOR STATE BIOMONITORING PROGRAMS 2-21
3. UTAH 3-1
3.1. EXPOSURES 3-1
3.1.1. Regional Projections for the Southwestern United States 3-1
3.1.2. Historic Climate Trends and Climate Change Projections for Utah 3-3
3.2. DATA INVENTORY AND PREPARATION 3-12
3.3.UTAHDEQMETHODS 3-14
3.4. INDICATORS 3-14
iii
-------
3.4.1. Thermal Preference 3-14
3.4.2. Hydrologic Indicators 3-15
3.4.3. Traits-Based Indicators in a Warmer Drier Scenario 3-21
3.5. LEAST DISTURBED LONG-TERM BIOLOGICAL MONITORING SITES .... 3-21
3.6. EVIDENCE OF TRENDS AT LEAST DISTURBED LONG-TERM
MONITORING SITES 3-25
3.6.1. Weber River (UT 4927250) 3-25
3.6.1.1. Temporal Trends in Climatic and Biological Variables 3-25
3.6.1.2. Associations Between Biological and Climatic Variables 3-32
3.6.1.3. Groupings Based on Climatic Variables 3-34
3.6.2. Virgin River (UT 4951200) 3-40
3.6.2.1. Temporal Trends in Climatic and Biological Variables 3-40
3.6.2.2. Associations Between Biological Variables and Climatic
Variables 3-45
3.6.2.3. Groupings Based on Climatic Variables 3-47
3.6.3. Beaver River (UT 5940440) 3-52
3.6.3.1. Temporal trends in Climatic and Biological Variables 3-53
3.6.3.2. Associations Between Biological Variables and Climatic
Variables 3-62
3.6.3.3. Groupings Based on Climatic Variables 3-62
3.6.4. Duchesne River (UT 4936750) 3-66
3.6.4.1. Temporal Trends in Climatic and Biological Variables 3-66
3.6.4.2. Associations Between Biological Variables and Climatic
Variables 3-72
3.6.4.3. Groupings Based on Climatic Variables 3-74
3.7. SENSITIVITY OF BENTHIC MACROINVERTEBRATES TO
TEMPERATURE, PRECIPITATION, AND STREAM FLOW 3-79
3.8. IMPLICATIONS FOR UTAH DEQ'S BIOMONITORING PROGRAM 3-79
4. MAINE 4-1
4.1. EXPOSURES 4-1
4.1.1. Regional Projections for the Northeastern United States 4-1
4.1.2. Historic Climate Trends and Climate Change Projections for Maine 4-2
4.2. DATA INVENTORY AND PREPARATION 4-10
4.3. MAINE DEP METHODS 4-14
4.4. INDICATORS 4-15
4.4.1. Thermal Preference 4-15
4.4.2. Hydrologic Indicators 4-22
4.4.3. Traits-Based Indicators in a Warmer Drier Scenario 4-23
4.5. LEAST DISTURBED LONG-TERM BIOLOGICAL MONITORING SITES .... 4-23
4.6. EVIDENCE OF TRENDS AT LEAST DISTURBED LONG-TERM
MONITORING SITES 4-27
4.6.1. SheepscotRiver(ME56817) 4-27
4.6.1.1. Temporal Trends in Climatic and Biological Variables 4-27
4.6.1.2. Associations Between Biological and Climatic Variables 4-34
4.6.1.3. Groupings Based on Climatic Variables 4-34
iv
-------
4.6.2. West Branch Sheepscot (ME 57011) 4-39
4.6.2.1. Temporal Trends in Climatic and Biological Variables 4-41
4.6.2.2. Associations Between Biological Variables and Climatic
Variables 4-46
4.6.2.3. Groupings Based on Climatic Variables 4-48
4.6.3. Duck Brook (ME 57065) 4-52
4.6.3.1. Temporal Trends in Climatic and Biological Variables 4-52
4.6.3.2. Associations Between Biological Variables and Climatic
Variables 4-59
4.6.3.3. Groupings Based on Climatic Variables 4-59
4.7. SENSITIVITY OF BENTHIC MACROINVERTEBRATES TO
TEMPERATURE AND STREAM FLOW 4-63
4.8. IMPLICATIONS FOR MAINE DEP' S BIOMONITORING PROGRAM 4-67
5. NORTH CAROLINA 5-1
5.1. EXPOSURES 5-1
5.1.1. Regional Projections for the Southeastern United States 5-1
5.1.2. Historic Climate Trends and Climate Change Projections for North
Carolina 5-2
5.2. DATA INVENTORY AND PREPARATION 5-11
5.3. NORTH CAROLINA DEPARTMENT OF THE ENVIRONMENT AND
NATURAL RESOURCE (NCDENR) METHODS 5-14
5.4. INDICATORS 5-16
5.4.1. Thermal Preference 5-16
5.4.2. Hydrologic Indicators 5-21
5.4.3. Traits-Based Indicators in a Warmer Drier Scenario 5-23
5.5. LEAST-DISTURBED LONG-TERM BIOLOGICAL MONITORING SITES.... 5-24
5.6. EVIDENCE OF TRENDS AT LEAST-DISTURBED LONG-TERM
MONITORING SITES 5-28
5.6.1. New River (NC0109) 5-28
5.6.1.1. Temporal Trends in Climatic and Biological Variables 5-28
5.6.1.2. Associations Between Biological and Climatic Variables 5-33
5.6.1.3. Groupings Based on Climatic Variables 5-36
5.6.2. Nantahala River (NC0207) 5-40
5.6.2.1. Temporal Trends in Climatic and Biological Variables 5-40
5.6.2.2. Associations Between Biological Variables and Climatic
Variables 5-44
5.6.2.3. Groupings Based on Climatic Variables 5-44
5.6.3. Cataloochee Creek (NC0209) 5-48
5.6.3.1. Temporal Trends in Climatic and Biological Variables 5-48
5.6.3.2. Associations Between Biological Variables and Climatic
Variables 5-52
5.6.3.3. Groupings Based on Climatic Variables 5-52
5.6.4. Barnes Creek (NC0248) 5-55
5.6.4.1. Temporal Trends in Climatic and Biological Variables 5-55
-------
5.6.4.2. Associations Between Biological Variables and Climatic
Variables 5-59
5.6.4.3. Groupings Based on Climatic Variables 5-59
5.6.5. Little River (NC0075) 5-62
5.6.5.1. Temporal Trends in Climatic and Biological Variables 5-62
5.6.5.2. Associations Between Biological Variables and Climatic
Variables 5-66
5.6.5.3. Groupings Based on Climatic Variables 5-66
5.7. SENSITIVITY OF BENTHIC MACROINVERTEBRATES TO
TEMPERATURE AND STREAM FLOW 5-69
5.8. IMPLICATIONS FOR NORTH CAROLINA DEPARTMENT OF THE
ENVIRONMENT AND NATURAL RESOURCES (NCDENRS)
BIOMONITORING PROGRAM 5-69
6. OHIO 6-1
6.1. EXPOSURES 6-1
6.1.1. Regional Projections for the Midwestern United States 6-1
6.1.2. Historic Climate Trends and Climate Change Projections for Ohio 6-2
6.2. DATA INVENTORY AND PREPARATION 6-11
6.3. OHIO EPA METHODS 6-14
6.4. INDICATORS 6-14
6.4.1. Thermal Preference 6-14
6.4.2. Hydrologic Indicators 6-18
6.4.3. Traits-Based Indicators in a Warmer Drier Scenario 6-21
6.5. LEAST DISTURBED LONG-TERM BIOLOGICAL MONITORING SITES .... 6-21
6.6. EVIDENCE OF TRENDS AT LEAST-DISTURBED LONG-TERM
MONITORING SITES 6-22
6.6.1. Reference Sites 6-22
6.6.1.1. Temporal Trends in Climatic and Biological Variables 6-22
6.6.1.2. Associations Between Biological and Climatic Variables 6-23
6.6.1.3. Groupings Based on Climatic Variables 6-23
6.7. SENSITIVITY OF BENTHIC MACROINVERTEBRATES TO
TEMPERATURE AND STREAM FLOW 6-23
6.8. IMPLICATIONS FOR OHIO EPA'S BIOMONITORING PROGRAM 6-25
7. SYNTHESIS 7-1
7.1. EVIDENCE FOR EXISTING CLIMATE CHANGE RESPONSES 7-1
7.1.1. Existing Climate Trends Support Expectations for Biological
Responses 7-1
7.2. COMPARISON OF REGIONAL TRENDS AND INDICATORS—HOW
TO INTERPRET OBSERVED RESPONSES 7-5
7.2.1. Comparison of Indicator Responses Among States and Regions 7-5
7.2.2. Factors Contributing to Spatial Variability in Observed Biological
Responses 7-20
7.2.3. Benthic Inferred Temperature 7-22
7.2.4. Basis for Inferring Climate Change Associations 7-27
vi
-------
7.2.5. Other Sources of Potential Spatial Confounding 7-29
7.3. CHARACTERISTICS OF EXISTING BIO ASSESSMENT PROGRAMS
RELEVANT TO DISCERNING CLIMATE CHANGE TRENDS 7-31
7.3.1. Sufficiency and Limitations of Data to Define and Partition Long-
Term Trends 7-31
7.3.2. Other Biomonitoring Methods Considerations 7-34
7.4. REFERENCE STATION VULNERABILITIES 7-37
7.4.1. Vulnerabilities in Assessing Reference Condition 7-37
7.4.2. Synergistic Effects between Climate Change and Land Use 7-39
7.4.3. Future Vulnerabilities of Reference Stations to Land Use 7-47
7.5. IMPLICATIONS TO MULTIMETRIC INDICES, PREDICTIVE MODELS,
AND IMPAIRMENT/LISTING DECISIONS 7-52
7.5.1. Conclusions Across Pilot Study States 7-52
7.5.2. Recommendations for Modifying Metrics 7-57
7.6. SENTINEL MONITORING NETWORK 7-60
7.7. CLIMATE CHANGE IMPLICATIONS FOR ENVIRONMENTAL
MANAGEMENT 7-64
7.7.1. Impairment Listings and Total Maximum Daily Load (TMDL)
Development 7-65
7.7.2. Impacts on the Development of Water-quality standards and
Biocriteria 7-67
7.8. CONCLUSIONS 7-70
REFERENCES 8-1
vn
-------
LIST OF TABLES
1-1. Examples of observed changes in aquatic community structure related to climate
change that are relevant in a bioassessment framework 1-7
2-1. Future projection data from 16 GCMs were evaluated 2-2
2-2. Example of how a weighted average model temperature optimum (weighted
mean) estimate is calculated 2-7
2-3. Example taken from Utah analysis results to illustrate development of ranking for
temperature (or other environmental parameter) preference and tolerance rankings
from weighted-average or generalized linear model temperature distribution
results 2-9
2-4. Summary of IHA parameters used in the analyses 2-11
2-5. List of traits and trait modalities that were considered when developing lists of
traits-based indicator taxa for future conditions of warming temperatures and
lower flows 2-13
2-6. List of biological metrics/traits evaluated at each site considering commonly used
metrics summarized in Barbour et al. (1999) and those used by state
biomonitoring programs 2-18
2-7. Summary of types of analyses that were conducted on the Maine, North Carolina,
Utah, and Ohio data sets 2-22
3-1. Proj ections for temperature and precipitation changes in the Southwest to 2100 3-1
3-2. Change rates in Utah PRISM mean annual air temperature compared across two
time periods: 1971-2000 versus 1901-2000 3-6
3-3. Projected departure from historic (1961-1990) trends in annual and seasonal air
temperature (°C) in Utah for mid- (2040-2069) and late-century (2070-2099) for
high and low emissions scenarios 3-8
3-4. Change rates in Utah PRISM mean annual precipitation compared across two
time periods: 1971-2000 versus 1901-2000 3-9
3-5. Projected departure from historic (1961-1990) trends in annual and seasonal
precipitation (mm) in Utah for mid- (2040-2069) and late-century (2070-2099)
for high and low emissions scenarios 3-11
3-6. Distribution of reference and total stations, categorized by duration of sampling 3-12
Vlll
-------
LIST OF TABLES (continued)
3-7. List of Utah cold-water-temperature indicator taxa, sorted by order, family, then
Final ID 3-17
3-8. List of Utah warm-water-temperature indicator taxa 3-19
3-9. List of taxa that may be most and least sensitive to a warmer and drier future
scenario based on a combination of traits 3-22
3-10. Site characteristics for the long-term biological monitoring stations in Utah 3-24
3-11. Time periods for which biological data were available at the long-term monitoring
sites in Utah 3-24
3-12. Range of temperature, precipitation, and flow values that occurred at the Weber
River site (UT 4927250) during the period of biological record 3-29
3-13. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics, year, and climatic
variables at the Weber River site (UT 4927250) 3-33
3-14. Kendall tau nonparametric correlations analyses performed to examine
associations between thermal preference metrics, year, and climatic variables at
the Weber River site (UT 4927250) 3-35
3-15. Kendall tau nonparametric correlations analyses performed to examine
associations between a subset of biological metrics, year, flow, and precipitation
variables at the Weber River site (UT 4927250) 3-36
3-16. Mean metric values (±1 SD) for the Weber River site (UT 4927250) in coldest,
normal, and hottest year samples 3-37
3-17. Mean metric values (±1 SD) for the Weber River site (UT 4927250) in driest,
normal, and wettest flow year samples 3-37
3-18. Range of temperature, precipitation, and flow values that occurred at the Virgin
River site (UT 4951200) during the period of biological record 3-43
3-19. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics, year, and climatic
variables at the Virgin River site (UT 4951200) 3-46
3-20. Kendall tau nonparametric correlations analyses performed to examine
associations between thermal preference metrics, year, and climatic variables at
the Virgin River site (UT 4951200) 3-49
IX
-------
LIST OF TABLES (continued)
3-21. Kendall tau nonparametric correlations analyses performed to examine
associations between a subset of biological metrics, year, flow, and precipitation
variables at the Virgin River site (UT 4951200) 3-50
3-22. Mean metric values (±1 SD) for the Virgin River site (UT 4951200) in coldest,
normal, and hottest year samples 3-51
3-23. Mean metric values (±1 SD) for the Virgin River site (UT 4951200) in driest,
normal, and wettest flow year samples 3-51
3-24. Range of temperature, precipitation, and flow values that occurred at the Beaver
River site (UT 5940440) during the period of biological record 3-59
3-25. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics, year, and climatic
variables at the Beaver River site (UT 5940440) 3-63
3-26. Kendall tau nonparametric correlations analyses performed to examine
associations between thermal preference metrics, year, and climatic variables at
the Beaver River site (UT 5940440) 3-64
3-27. Kendall tau nonparametric correlations analyses performed to examine
associations between a subset of biological metrics, year, flow, and precipitation
variables at the Beaver River site (UT 5940440) 3-65
3-28. Mean metric values (±1 SD) for the Beaver River site (UT 5940440) in coldest,
normal, and hottest year samples 3-67
3-29. Mean metric values (±1 SD) for the Beaver River site (UT 5940440) in driest,
normal, and wettest flow year samples 3-67
3-30. Range of temperature, precipitation, and flow values that occurred at the
Duchesne River site (UT 4936750) during the period of biological record 3-70
3-31. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics, year, and climatic
variables at the Duchesne River site (UT 4936750) 3-75
3-32. Kendall tau nonparametric correlations analyses performed to examine
associations between thermal preference metrics, year, and climatic variables at
the Duchesne River site (UT 4936750) 3-76
3-33. Kendall tau nonparametric correlations analyses performed to examine
associations between a subset of biological metrics, year, flow, and precipitation
variables at the Duchesne River site (UT 4936750) 3-77
-------
LIST OF TABLES (continued)
3-34. Mean metric values (±1 SD) for the Duchesne River site (UT 4936750) in coldest,
normal, and hottest year samples 3-78
3-35. Mean metric values (±1 SD) for the Duchesne River site (UT 4936750) in driest,
normal, and wettest year samples 3-78
3-36. Summary of differences in elevation, PRISM mean annual air temperature and
precipitation, and mean number and percentage of cold and warm-water-
preference taxa across and within major ecoregions 3-80
4-1. Proj ections for temperature and precipitation changes in the Northeast to 2100 4-2
4-2. Change rates in Maine PRISM mean annual air temperature compared across two
time periods: 1971-2000 versus 1901-2000 4-4
4-3. Projected departure from historic (1961-1990) trends in annual and seasonal air
temperature (°C) in Maine for mid- (2040-2069) and late-century (2070-2099)
for high and low emissions scenarios 4-7
4-4. Change rates in Maine PRISM mean annual precipitation compared across two
time periods: 1971-2000 versus 1901-2000 4-10
4-5. Projected departure from historic (1961-1990) trends in annual and seasonal
precipitation (mm) in Maine for mid- (2040-2069) and late-century (2070-2099)
for high and low emissions scenarios 4-11
4-6. Distribution of stations that have received Class A biological condition ratings
and total stations, categorized by duration of sampling 4-13
4-7. Metrics that are used in Maine DEP's four linear discriminant models 4-16
4-8. List of Maine cold-water temperature indicator taxa, sorted by order, family, then
Final ID 4-17
4-9. List of Maine warm-water temperature indicator taxa 4-20
4-10. List of taxa that may be most and least sensitive to a warmer and drier future
scenario based on a combination of traits 4-24
4-11. Site characteristics for the long-term biological monitoring stations in Maine 4-26
4-12. Time periods for which biological data were available at the long-term monitoring
sites in Maine 4-26
4-13. Range of temperature, precipitation, and flow values that occurred at the
Sheepscot River site (ME 56817) during the period of biological record 4-31
xi
-------
LIST OF TABLES (continued)
4-14. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics, year, and climatic
variables at the Sheepscot River site (ME 56817) 4-35
4-15. Kendall tau nonparametric correlations analyses performed to examine
associations between thermal preference metrics, year, and climatic variables at
the Sheepscot River site (ME 56817) 4-36
4-16. Kendall tau nonparametric correlations analyses performed to examine
associations between a subset of biological metrics, year, flow, and precipitation
variables at the Sheepscot River site (ME 56817) 4-37
4-17. Mean metric values (±1 SD) for the Sheepscot River site (ME 56817) in coldest,
normal, and hottest year samples 4-38
4-18. Mean metric values (±1 SD) for the Sheepscot River site (ME 56817) in driest,
normal, and wettest flow year samples 4-38
4-19. Range of temperature, precipitation, and flow values that occurred at the West
Branch Sheepscot site (ME 57011) during the period of biological record 4-43
4-20. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics, year, and climatic
variables at the West Branch Sheepscot site (ME 57011) 4-47
4-21. Kendall tau nonparametric correlations analyses performed to examine
associations between thermal preference metrics, year, and climatic variables at
the West Branch Sheepscot site (ME 57011) 4-49
4-22. Kendall tau nonparametric correlations analyses performed to examine
associations between a subset of biological metrics, year, flow, and precipitation
variables at the West Branch Sheepscot site (ME 57011) 4-50
4-23. Mean metric values (±1 SD) for the West Branch Sheepscot site (ME 57011) in
coldest, normal, and hottest year samples 4-51
4-24. Mean metric values (±1 SD) for the West Branch Sheepscot site (ME 57011) in
driest, normal, and wettest flow year samples 4-51
4-25. Range of temperature, precipitation, and flow values that occurred at the Duck
Brook site (ME 57065) during the period of biological record 4-56
4-26. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics, year, and climatic
variables at the Duck Brook site (ME 57065) 4-60
xii
-------
LIST OF TABLES (continued)
4-27. Kendall tau nonparametric correlations analyses performed to examine
associations between thermal preference metrics, year, and climatic variables at
the Duck Brook site (ME 57065) 4-61
4-28. Kendall tau nonparametric correlations analyses performed to examine
associations between a subset of biological metrics, year, flow, and precipitation
variables at the Duck Brook site (ME 57065) 4-62
4-29. Mean metric values (±1 SD) for the Duck Brook site (ME 57065) in coldest,
normal, and hottest year samples 4-64
4-30. Mean metric values (±1 SD) for the Duck Brook site (ME 57065) in driest,
normal, and wettest year samples 4-64
4-31. Summary of differences in elevation, PRISM mean annual air temperature and
precipitation, and mean number and percentage of cold and warm-water-
preference taxa across and within major ecoregions 4-65
4-32. List of model input metrics from Maine DEP's linear discriminant models that
could be most affected by changing temperature and streamflow conditions 4-70
5-1. Projections for temperature and precipitation changes in the Southeast to 2100 5-1
5-2. Change rates in North Carolina PRISM mean annual air temperature compared
across two time periods: 1971-2000 versus 1901-2000 5-4
5-3. Projected departure from historic (1961-1990) trends in annual and seasonal air
temperature (°C) in North Carolina for mid- (2040-2069) and late-century (2070-
2099) for high and low emissions scenarios 5-6
5-4. Change rates in North Carolina PRISM mean annual precipitation compared
across two time periods: 1971-2000 versus 1901-2000 5-7
5-5. Projected departure from historic (1961-1990) trends in annual and seasonal
precipitation (mm) in North Carolina for mid- (2040-2069) and late-century
(2070-2099) for high and low emissions scenarios 5-10
5-6. Distribution of reference and unclassified stations, categorized by duration of
sampling 5-12
5-7. These tables are used to determine the scores for EPT taxa richness values and
NCBI values for all standard qualitative samples after seasonal corrections are
made 5-15
5-8. List of North Carolina cold-water temperature indicator taxa 5-17
xiii
-------
LIST OF TABLES (continued)
5-9. List of North Carolina warm-water temperature indicator taxa 5-19
5-10. List of taxa that may be most and least sensitive to a warmer and drier future
scenario based on the combination of traits described in Section 2.2.3 5-24
5-11. Site characteristics for the long-term biological monitoring stations in North
Carolina 5-26
5-12. Time periods for which biological data were available at the long-term monitoring
sites in North Carolina 5-27
5-13. Range of temperature, precipitation, and flow values that occurred at the New
River (NC0109) during the period of biological record 5-31
5-14. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics, year, and climatic
variables at the New River (NCO109) site 5-35
5-15. Kendall tau nonparametric correlations analyses performed to examine
associations between thermal preference metrics, year, and climatic variables at
the New River (NCO 109) site 5-37
5-16. Kendall tau nonparametric correlations analyses performed to examine
associations between a subset of biological metrics, year, flow, and precipitation
variables at the New River (NCO 109) site 5-38
5-17. Mean metric values (±1 SD) for the New River (NCO 109) site in coldest, normal,
and hottest year samples 5-39
5-18. Mean metric values (±1 SD) for the New River (NC0109) site in driest, normal,
and wettest flow year samples 5-39
5-19. Range of temperature, precipitation, and flow values that occurred at the
Nantahala River (NC0207) site during the period of biological record 5-43
5-20. Range of temperature, precipitation, and flow values that occurred at the Beaver
River site (UT 5940440) during the period of biological record 5-51
5-21. Range of temperature and precipitation values that occurred at the Barnes Creek
(NC0248) site during the period of biological record 5-58
5-22. Range of temperature, precipitation, and flow values that occurred at the Little
River (NC0075) during the period of biological record 5-65
xiv
-------
LIST OF TABLES (continued)
5-23. Summary of differences in elevation, PRISM mean annual air temperature, and
mean number and percentage of cold and warm-water taxa (based on full-scale
samples only) in the North Carolina EPA Level 3 ecoregions 5-70
6-1. Projections for temperature and precipitation changes in the Midwest to 2100 6-1
6-2. Change rates in Ohio PRISM mean annual and seasonal air temperatures
compared across two time periods: 1971-2000 versus 1901-2000 6-4
6-3. Projected departure from historic (1961-1990) trends in annual and seasonal air
temperature (°C) in Ohio for mid- (2040-2069) and late-century (2070-2099) for
high and low emissions scenarios 6-6
6-4. Change rates in Ohio PRISM mean annual and seasonal precipitation compared
across two time periods: 1971-2000 versus 1901-2000 6-7
6-5. Projected departure from historic (1961-1990) trends in annual and seasonal
precipitation (mm) in Ohio for mid- (2040-2069) and late-century (2070-2099)
for high and low emissions scenarios 6-10
6-6. Summary of Ohio EPA regional reference site network including original sites
(1980-1989) and updates via first (1990-1999) and second round resampling
(2000-2006) that were used in data analyses 6-12
6-7. Changes in ICI and mayfly influence ICI metrics related to increasing taxonomic
resolution overtime in the Ohio EPA least impacted reference data set 6-13
6-8. Table of original and recalibrated Ohio biocriteria with adjustments made to
equilibrate taxonomic advances made in the later time period 6-13
6-9. Macroinvertebrate community metrics used in the ICI for evaluating biological
condition in Ohio 6-15
6-10. Index of Biotic Integrity metrics used to evaluate wading sites, boat sites, and
headwaters stream sites in Ohio 6-16
6-11. Subcomponents of the Ohio QHEI, which were used to score a Hydro-QHEI, and
current and depth subscores 6-18
6-12. Average and range of years represented by original reference site data and
resampled (latest) data by index and stream size category 6-22
6-13. Original Ohio biocriteria (O), recalculated biocriteria (R) using similar sites, and
new biocriteria (N) using the latest data from resampling of original reference
sites 6-24
xv
-------
LIST OF TABLES (continued)
7-1. Observed and modeled future rates of change for air temperature and precipitation
for the four states analyzed in this study 7-2
7-2. Summary of results from water temperature trend analyses at 23 USGS stations
that met the screening criteria 7-3
7-3. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics and year at long-term
reference sites from three states 7-6
7-4. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics and temperature at long-
term reference sites from 3 states 7-8
7-5. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics and precipitation at long-
term reference sites from three states 7-10
7-6. Kendall tau nonparametric correlations analyses performed to examine
associations between commonly used biological metrics and year and flow at
long-term reference sites from three states 7-12
7-7. Mean metric values (±1 SD) for sites in three states in coldest, normal, and hottest
year samples 7-14
7-8. Mean metric values (±1 SD) for sites in three states in driest, normal, and wettest
flow year samples 7-16
7-9. Average distribution of reference and total stations by state, categorized by
duration of sampling 7-32
7-10. Time periods for which biological data were available at the long-term monitoring
sites in Utah (UT), Maine (ME), and North Carolina (NC) 7-33
7-11. Summary of ANOVA results for high-flow IHA metrics 7-42
7-12. Summary of ANOVA results for low-flow IHA metrics 7-43
7-13. Percentage of existing Florida reference stations (n = 58, classified as
"exceptional"), that have >20% developed land use (with 25 houses per square
mile (9.65 houses per square kilometer) or more, Categories 5-12 in the ICLUS
data set) within a 1-km buffer surrounding the station, for current and decadal
time periods through 2100 7-51
xvi
-------
LIST OF TABLES (continued)
7-14. Percentage urban and suburban development within a 1-km2 area surrounding
reference sites, for all sites and for sites at or above the impact threshold of 10% 7-51
7-15. Variables addressed in criteria and pathways through which they may be affected
by climate change (from Hamilton et al., 2010a) 7-69
xvn
-------
LIST OF FIGURES
1-1. Conceptual model of the linkages between climate forcings, climate system
changes, stream habitat changes (abiotic), and the subsequent individual-,
population- and community-level responses to these changes 1-4
2-1. Illustration of weighted average temperature distribution, where the weighted
average mean (u) is taken as the temperature optimum (preference) for the tax on,
and the magnitude of SD is taken as an estimate of the temperature sensitivity or
tolerance 2-8
3-1. Utah's temperature and precipitation patterns 3-5
3-2. Trends in annual mean air temperature in Utah from 1901-2000 3-6
3-3. Trends in seasonal mean air temperature in Utah from 1901-2000 3-7
3-4. Trends in annual mean precipitation in Utah from 1901-2000 3-9
3-5. Trends in seasonal mean precipitation in Utah from 1901-2000 3-10
3-6. Utah biomonitoring stations, coded by reference status and duration of data 3-13
3-7. Relationship between Utah cold and warm-water-preference taxa and Utah
enrichment tolerance scores 3-20
3-8. Locations of the four least disturbed long-term biological monitoring sites 3-23
3-9. Locations of the Weber River (UT 4927250) biological sampling site, USGS gage
10128500 (Weber River near Oakley) and Wanship Dam weather station 3-26
3-10. Yearly trends in annual observed air temperature (°C) at the Weber River site (UT
4927250) from 1955-2010, based on data from the Wanship Dam weather station 3-27
3-11. Yearly trends in mean annual flow (cfs) at the Weber River site (UT 4927250)
from 1904-2011, based on data from USGS gage 10128500 3-28
3-12. Yearly trends at the Weber River site (UT 4927250) in (A) O/E, (B) number of
EPT taxa and HBI; (C) mean maximum July temperature (°C) and mean
September/October/November (SON) flow (cfs) 3-30
3-13. Yearly trends at the Weber River site (UT 4927250) in (A) number of cold and
warm-water taxa; (B) percentage cold and warm-water individuals; and (C) mean
maximum July temperature (°C) and mean September/October/November (SON)
flow (cfs) 3-31
3-14. NMDS plot (Axis 1-2) for the Weber River site (UT 4927250), shown in Figure
3-8 3-38
xviii
-------
LIST OF FIGURES (continued)
3-15. NMDS plot (Axis 1-2) for the Weber River site (UT 4927250) showing which
taxa are most highly correlated with each axis 3-39
3-16. Locations of the Virgin River (UT 4951200) biological sampling site, N. Fork
Virgin River USGS gage, andZionNP weather station 3-41
3-17. Yearly trends in annual observed air temperature (°C) at the Virgin River site (UT
4951200) from 1904-2010, based on data from the Zion NP weather station 3-42
3-18. Yearly trends in mean annual precipitation (mm) at the Virgin River site (UT
4951200) from 1904-2011, based on data from the Zi on NP weather station 3-43
3-19. Yearly trends at the Virgin River site (UT 4951200) in (A) O/E, (B) number of
EPT taxa and HBI; (C) mean maximum July temperature (°C) and mean observed
September/October/November (SON) precipitation (mm) 3-44
3-20. Yearly trends at the Virgin River site (UT 4951200) in (A) number of cold and
warm-water taxa; (B) percentage cold and warm-water individuals; and (C) mean
maximum July temperature (°C) and mean observed
September/October/November (SON) precipitation (mm) 3-48
3-21. NMDS plot (Axis 1-2) for the Virgin River site (UT 4951200) 3-54
3-22. NMDS plot (Axis 1-2) for Utah Station 4951200 (Virgin) that shows which taxa
are most highly correlated with each axis 3-55
3-23. Locations of the Beaver River (UT 5940440) biological sampling site, Beaver
River USGS gage, and Beaver Canyon pH weather station 3-56
3-24. Yearly trends in PRISM annual air temperature data associated with the Beaver
River site (UT 5940440) from 1975-2005 3-58
3-25. Yearly trends in mean annual flow (cfs) from 1914-2011, based on data from
USGS gage 10234500 3-59
3-26. Yearly trends at the Beaver River site (UT 5940440) in (A) O/E, (B) number of
EPT taxa and JTBI; (C) mean maximum July temperature (°C) and mean
September/October/November (SON) flow (cfs) 3-60
3-27. Yearly trends at the Beaver River site (UT 5940440) in (A) number of cold and
warm-water taxa; (B) percentage cold and warm-water individuals; and (C) mean
maximum July temperature (°C) and mean September/October/November (SON)
flow (cfs) 3-61
XIX
-------
LIST OF FIGURES (continued)
3-28. Locations of the Duchesne River (UT 4936750) biological sampling site, W.F.
Duchesne River USGS gage and Duchesne weather station 3-68
3-29. Yearly trends in annual observed air temperature (°C) from 1906-2010, based on
data from the Duchesne weather station 3-69
3-30. Yearly trends in mean annual precipitation (mm) from 1906-2011, based on data
from the Duchesne weather station 3-70
3-31. Yearly trends at the Duchesne River site (UT 4936750) in (A) O/E, (B) number of
EPT taxa and HBI; (C) mean maximum July temperature (°C) and observed mean
September/October/November (SON) precipitation (mm) 3-71
3-32. Yearly trends at the Duchesne River site (UT 4936750) in (A) number of cold and
warm-water taxa; (B) percentage cold and warm-water individuals; and (C) mean
maximum July temperature (°C) and observed mean
September/October/November (SON) precipitation (mm) 3-73
3-33. Distribution of cold- and warm-water taxa across Strahler Orders in Utah, based
on fall (September-November) samples collected from 67 least-disturbed sites 3-81
4-1. Maine's temperature and precipitation patterns 4-4
4-2. Trends in annual mean air temperature in Maine from 1901-2000 4-5
4-3. Trends in seasonal mean air temperature in Maine from 1901-2000 4-6
4-4. Trends in annual mean precipitation in Maine from 1901-2000 4-8
4-5. Trends in seasonal mean precipitation in Maine from 1901-2000 4-9
4-6. Maine biomonitoring stations, coded by reference status and duration of data 4-12
4-7. Relationship between Maine cold and warm-water-preference taxa and Maine
enrichment tolerance scores 4-22
4-8. Locations of the three biological sampling sites that we performed long-term
trend analyses on (56817 = Sheepscot; 57011 = West Branch Sheepscot;
57065 = Duck Brook) 4-25
4-9. Locations of the Sheepscot River (ME 56817) biological sampling site, USGS
gage 1038000 (Sheepscot River at North Whitefield) and Augusta FAA AP
weatherstation 4-28
xx
-------
LIST OF FIGURES (continued)
4-10. Yearly trends in annual observed air temperature (°C) at the Sheepscot River site
(ME 56817) from 1949-2010, based on data from the Augusta FAA AP weather
station 4-29
4-11. Yearly trends in mean annual flow (cfs) at the Sheepscot River site (ME 56817)
from 1939-2009, based on data from USGS gage 1038000 4-30
4-12. Yearly trends at the Sheepscot River site (ME 56817) in (A) biological condition
class (1 = Class A; 2 = Class B; 3 = Class C; 4 = NA); (B) number of EPT taxa
and HBI; and (C) mean maximum July/August temperature (°C) and mean July-
September flow (cfs) 4-32
4-13. Yearly trends at the Sheepscot River site (ME 56817) in (A) number of cold and
warm-water taxa; (B) percentage cold and warm-water individuals; and (C) mean
maximum July/August temperature (°C) and mean July-September flow (cfs) 4-33
4-14. NMDS plot (Axis 1-2) for the Sheepscot River site (ME 56817) 4-39
4-15. Locations of the West Branch Sheepscot site (ME 57011) biological sampling
site, USGS gage 1038000 (Sheepscot River at North Whitefield) and Augusta
FAA AP weather station 4-40
4-16. Yearly trends in annual observed air temperature (°C) at the West Branch
Sheepscot site (ME 57011) from 1949-2009, based on data from the Augusta
FAA AP weather station 4-41
4-17. Yearly trends in mean annual flow (cfs) at the West Branch Sheepscot site (ME
57011) from 1939-2009, based on data from USGS gage 1038000 4-42
4-18. Yearly trends at the West Branch Sheepscot site (ME 57011) in (A) biological
condition class (1 = Class A; 2 = Class B; 3 = Class C; 4 = NA); (B) number of
EPT taxa and HBI; and (C) mean maximum July/August temperature (°C) and
mean July-September flow (cfs) 4-44
4-19. Yearly trends at the West Branch Sheepscot site (ME 57011) in (A) number of
cold and warm-water taxa; (B) percentage cold and warm-water individuals; and
(C) mean maximum July/August temperature (°C) and mean July-September
flow (cfs) 4-45
4-20. Locations of the Duck Brook site (ME 57065) biological sampling site, Bar
Harbor 3 NW weather station and AcadiaNP weather station 4-53
4-21. Yearly trends in annual observed air temperature (°C) at the Duck Brook site (ME
57065) from 1893-2009, based on data from the Bar Harbor 3 NW and Acadia
NP weather stations 4-54
xxi
-------
LIST OF FIGURES (continued)
4-22. Yearly trends in annual observed precipitation (mm) at the Duck Brook site (ME
57065) from 1893-2009, based on data from the Bar Harbor 3 NW and Acadia
NP weather stations 4-55
4-23. Yearly trends at the Duck Brook site (ME 57065) in (A) biological condition class
(1 = Class A; 2 = Class B; 3 = Class C; 4 = NA); (B) number of EPT taxa and
HBI; and (C) mean maximum July/August temperature (°C) and mean July-
September precipitation (mm) 4-57
4-24. Yearly trends at the Duck Brook site (ME 57065) in (A) number of cold and
warm-water taxa; (B) percentage cold and warm-water individuals; and (C) mean
maximum July temperature (°C) and mean July-September precipitation (mm) 4-58
4-25. Distribution of cold and warm-water taxa across Strahler Orders in Maine, based
on July-September replicates collected from sites that received Class A biological
condition ratings 4-66
5-1. North Carolina's temperature and precipitation patterns 5-3
5-2. Trends in annual mean air temperature in North Carolina from 1901-2000 5-4
5-3. Trends in seasonal mean air temperature in North Carolina from 1901-2000 5-5
5-4. Trends in annual mean precipitation in North Carolina from 1901-2000 5-8
5-5. Trends in seasonal mean precipitation in North Carolina from 1901-2000 5-9
5-6. NCDENR biomonitoring stations, coded by reference status and duration of data 5-13
5-7. Relationship between North Carolina cold- and warm-water-preference taxa and
North Carolina enrichment tolerance scores 5-21
5-8. NMDS plot of macroinvertebrate taxonomic composition and its relationship with
hydrologic parameters fora subset of North Carolina data 5-22
5-9. Locations of the five least disturbed long-term biological monitoring sites that
were examined for long-term trends 5-25
5-10. Locations of the New River (NC0109) biological sampling site, USGS gage
03164000 (New River near Galax, VA) and Sparta 2 SE weather station 5-29
5-11. Yearly trends in PRISM mean annual air temperature (°C) at the New River
(NCO109) site from 1974-2006 5-30
5-12. Yearly trends in mean annual flow (cfs) at the New River (NC0109) site from
1930-2010, based on data from USGS gage 03164000 5-31
xxii
-------
LIST OF FIGURES (continued)
5-13. Yearly trends at the New River (NC0109) site in (A) bioclassification score
(based on species-level data); (B) number of EPT taxa and HBI (based on genus-
level OTU); and (C) PRISM mean annual air temperature (°C) and mean summer
(June-September) flow (cfs) 5-32
5-14. Yearly trends at the New River (NC0109) site in (A) number of cold and warm-
water taxa; (B) percentage cold and warm-water individuals; and (C) PRISM
mean annual air temperature (°C) and mean summer (June-September) flow (cfs) 5-34
5-15. Locations of the Nantahala River (NC0207) biological sampling site, USGS gage
03504000 (Nantahala River near Rainbow Springs) and Franklin weather station 5-41
5-16. Yearly trends in observed mean annual air temperature (°C) at the Franklin
weather station from 1946-2010 5-42
5-17. Yearly trends in mean annual flow (cfs) at the Nantahala River (NC0207) site
from 1941-2010, based on data from USGS gage 03504000 5-43
5-18. Yearly trends at the Nantahala River (NC0207) site in (A) bioclassification score
(based on species-level data); (B) number of EPT taxa and HBI (based on genus-
level OTU); and (C) observed mean July/August maximum air temperature (°C)
and mean summer (June-September) flow (cfs) 5-46
5-19. Yearly trends at the Nantahala River (NC0207) site in (A) number of cold and
warm-water taxa; (B) percentage cold and warm-water individuals; and (C)
observed mean July/August maximum air temperature (°C) and mean summer
(June-September) flow (cfs) 5-47
5-20. Locations of the Cataloochee Creek (NC0209) biological sampling site, USGS
gage 03460000 (Cataloochee Creek near Cataloochee) and Cataloochee weather
station 5-49
5-21. Yearly trends in observed mean annual air temperature (°C) at the Cataloochee
weather station from 1966-2009 5-50
5-22. Yearly trends in mean annual flow (cfs) at the Cataloochee Creek (NC0209) site
from 1935-2010, based on data from USGS gage 03460000 5-51
5-23. Yearly trends at the Cataloochee Creek (NC0209) site in (A) bioclassification
score (based on species-level data); (B) number of EPT taxa and HBI (based on
genus-level OTU); and (C) observed mean July/August maximum air temperature
(°C) and mean summer (June-September) flow (cfs) 5-53
xxin
-------
LIST OF FIGURES (continued)
5-24. Yearly trends at the Cataloochee Creek (NC0209) site in (A) number of cold and
warm-water taxa; (B) percentage cold and warm-water individuals; and (C)
observed mean July/August maximum air temperature (°C) and mean summer
(June-September) flow (cfs) 5-54
5-25. Locations of the Barnes Creek (NC0248) biological sampling site and the
Albemarle weather station 5-56
5-26. Yearly trends in observed mean annual air temperature (°C) at the Albemarle
weather station from 1912-2009 5-57
5-27. Yearly trends in mean annual precipitation (mm) at the Albemarle weather station
from 1912-2010 5-58
5-28. Yearly trends at the Barnes Creek (NC0248) site in (A) bioclassification score
(based on species-level data); (B) number of EPT taxa and JTBI (based on genus-
level OTU); and (C) observed mean July/August maximum air temperature (°C)
and mean summer (June-September) precipitation (mm) 5-60
5-29. Yearly trends at the Barnes Creek (NC0248) site in (A) number of cold and
warm-water taxa; (B) percentage cold and warm-water individuals; and (C)
observed mean July/August maximum air temperature (°C) and mean summer
(June-September) precipitation (mm) 5-61
5-30. Locations of the Little River (NC0075) biological sampling site, USGS gage
02128000 (Little River near Star, NC) and the Jackson Springs 5 WNW weather
station 5-63
5-31. Yearly trends in observed mean annual air temperature (°C) at the Jackson
Springs 5 WNW weather station from 1953-2010 5-64
5-32. Yearly trends in mean annual flow (cfs) at the Little River (NC0075) site from
1955-2010, based on data from USGS gage 02128000 5-65
5-33. Yearly trends at the Little River (NC0075) site in (A) bioclassification score
(based on species-level data); (B) number of EPT taxa and JTBI (based on genus-
level OTU); and (C) observed mean July/August maximum air temperature (°C)
and mean summer (June-September) flow (cfs) 5-67
5-34. Yearly trends at the Little River (NC0075) site in (A) number of cold and warm-
water taxa; (B) percentage cold and warm-water individuals; and (C) observed
mean July/August maximum air temperature (°C) and mean summer (June-
September) flow (cfs) 5-68
xxiv
-------
LIST OF FIGURES (continued)
5-35. Distribution of cold and warm-water taxa across different stream size categories
at North Carolina reference sites 5-71
5-36. Exploratory exercise on reference station drift (degradation of assessed site
condition) over time at the three Blue Ridge stations, simulating the loss of cold-
preference EPT taxa over time due to climate change effects 5-73
5-37. 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 5-74
6-1. Ohio's temperature and precipitation patterns 6-3
6-2. Trends in annual mean air temperature in Ohio from 1901-2000 6-4
6-3. Trends in seasonal mean air temperature in Ohio from 1901-2000 6-5
6-4. Trends in annual mean precipitation in Ohio from 1901-2000 6-8
6-5. Trends in seasonal mean precipitation in Ohio from 1901-2000 6-9
6-6. Plots of macroinvertebrate taxa maximum temperature WSV values versus mean
maximum values for taxa for headwater streams (a) and wadeable streams (b) and
box and whisker plots of WSVs for maximum temperatures by Ohio EPA
macroinvertebrate tolerance values (derived for the ICI) for headwater streams (c)
and wadeable streams (d) 6-17
6-7. Scatter plots of taxa/species Hydro-QHEI WSV values versus mean Hydro-QHEI
values for macroinvertebrates taxa for headwater streams (a) and for species in
wadeable streams (b) and box and whisker plots of macroinvertebrate (c) and fish
(d) WSVs for Hydro-QHEI for these waters 6-20
7-1. Comparison of trends in benthic inferred temperature and seasonal observed air
temperatures at (A) the Sheepscot River site (ME 56817); and (B) the Weber
River site (UT 4927250) 7-25
7-2. Comparison of trends in benthic inferred temperature and PRISM mean annual air
temperatures at: (A) the Sheepscot River site (ME 56817); (B) the Weber River
site (UT 4927250); and (C) the New River site (NC0109) 7-26
7-3. Benthic macroinvertebrate inferred temperature trend for selected reference sites
in Utah 7-27
7-4. ANOVA results for high-pulse duration (days) at forested and urban sites 7-41
xxv
-------
LIST OF FIGURES (continued)
7-5. ANOVA results for 7-day minimum flow (standardized by mean annual flow) at
forested and urban sites 7-42
7-6. Relationship between richness of EPT taxa and low-pulse count of the stream for
stream types in the North Carolina Piedmont 7-44
7-7. Relationship between richness of EPT taxa and flashiness (Baker's index) of the
stream for stream types in the North Carolina Piedmont 7-45
7-8. Relationship between richness of EPT taxa and 1-day minimum flow of the
stream for stream types in the North Carolina Piedmont 7-46
7-9. Florida's biomonitoring sampling stations, including "exceptional" reference
locations (light green dots), shown in relation to current land use 7-48
7-10. 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) 7-49
7-11. Method for tracking changes in cold- and warm-water-preference taxa and
commonly used metrics 7-59
7-12. Conceptual model showing relationship between climate change trends and
reference and stressed sites with an overlay of temporal variation on the trend 7-61
7-13. 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 7-65
xxvi
-------
LIST OF ABBREVIATIONS AND ACRONYMS
ANOVA Analysis of variance
BCG Biological Condition Gradient
CCA Canonical Correspondence Analysis
cfs cubic feet per second
CWA Clean Water Act
DEP Department of Environmental Protection
DEQ Department of Environmental Quality
ECBP Eastern Corn Belt Plain
EDAS Ecological Data Application System
EMAP Environmental Monitoring and Assessment Program
ENSO El Nino/Southern Oscillation
EPA U.S. Environmental Protection Agency
EPT Ephemeroptera, Plecoptera, Trichoptera
EWH exceptional warmwater habitat
GCM general circulation models
GIS geographic information system
HAB harmful algal blooms
HBI HilsenhoffBiotic Index
IB I Index of Biotic Integrity
ICI Invertebrate Community Index
ICLUS Integrated climate and land use scenarios
MA Indicators of Hydrologic Alteration
IPCC Intergovernmental Panel on Climate Change
MBI Midwest Biodiversity Institute
Mlwb Modified Index of Well-Being
MMI multimetric index
MWH modified warmwater habitat
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 nonmetric multidimensional scaling
NPDES National Pollutant Discharge Elimination System
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
QHEI Qualitative Habitat Evaluation Index
RA relative abundance
RBI Richards-Baker flashiness Index
RIVPACS River In Vertebrate Prediction and Classification System
SD standard deviation
TMDL total maximum daily load
xxvii
-------
LIST OF ABBREVIATIONS AND ACRONYMS (continued)
USGS United States Geological Service
WSV Weighted Stressor Values
WWH warmwater habitat
xxvin
-------
PREFACE
This report was prepared by Tetra Tech, Inc. and the Global Change Research Program 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 U.S. states: Maine, North Carolina, Ohio, and Utah.
The main findings of interest to managers 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.
xxix
-------
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, MD
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
Daren Carlisle (USGS), M. Siobhan Fennessey (Kenyon College), Eric P. Smith (VA
Polytechnic Institute), R. Jan Stevenson (Michigan State Univ.), N. Scott Urquhart (Statistical
Consultant)
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 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.
XXX
-------
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 (CWA). 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, which are 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 sensitive components of
bioassessment programs include
xxxi
-------
Assessment design (e.g., multimetric 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)
Ecological traits are useful tools for these analyses because 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 sensitive to changing temperature conditions. Consequently, MMIs and
predictive models, which rely on these sensitive taxa are likely to be influenced by 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 to
conventional pollutants also have demonstrable sensitivities to climate-related changes in
temperature and flow conditions. These sensitivities remain difficult to tease apart, although
approaches to modify metrics using temperature- and possibly flow-sensitive traits show some
promise in helping separate the influence of climate change from other stressors when combined
with appropriate study designs.
The implementation of bioassessment programs often involves flexible sampling systems,
such as rotating basin designs. These systems ensure statistically adequate sampling over 5-year
periods, often at the expense of continuous monitoring of specific locations. Consequently,
states may have many reference locations but lack enough stable, long-term stations needed to
xxxn
-------
detect climate-driven changes in biotic condition. In order to account for climate change effects
in the interpretation of station conditions, consistent long-term monitoring of a site or a region
will be needed. At the least, monitoring network designs will need to consider incorporation of a
few specific locations for detection of trends over time or include more extensive probabilistic
monitoring of a watershed or region in a manner that supports climate-related trend detection.
Climate change can cause other problems for reference-based bioassessment systems.
We note that climate change can drive shifts in community composition that vary by location,
potentially further compounded by nonclimatic 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, 1-day maximum flow) tend to reflect urbanization,
swamping climate change effects; whereas low-flow metrics (e.g., short-duration minimum
flows, low pulse-count duration) respond to climate change effects more so than to land use.
Some of the long-term stations in our study showed increasing trends in benthic inferred
temperatures, though not all trends were significant. These correspond well with the magnitude
of air temperature increases observed for the period, suggesting that the estimates of benthic
invertebrate temperature optima were generally appropriate, and that using benthic invertebrate
occurrence and abundance coupled with temperature preferences provides evidence of benthic
community changes over time related to long-term changes in temperature. With a large enough
data set, this type of analysis could be informative of long-term trends that are more widely
applicable than our analyses that were limited to data from single sites. 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 water taxon) may or may not be significant despite the expectation, but it
is informative if the community as a whole is reflecting an overall progressive shift in
temperature preferences.
A synthesis of 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 may be useful to partition climatically vulnerable indicators into
new metrics that account for temperature preferences of the component taxa. Analyzing
xxxiii
-------
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. It may also be
useful to identify particular sensitive taxa by region that can be tracked for climate change
responses.
Although data limitations prevent explicit differentiation between interannual, 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 would be valuable to account for climate change effects and to further our
understanding of natural variability. 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 illustrates the 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
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
xxxiv
-------
of high quality stream reaches that define reference conditions. Protection should focus on
minimization, mitigation, and/or buffering from nonpoint 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
can affect the determination of stream reach impairment. In vulnerable watersheds, this may
lead to fewer listings of impaired stream reaches and progressive under-protection of water
resources, unless the management framework is adjusted to better account for expected climate
change effects. Adaptations that should be considered include modification of metrics so that
climate effects can be tracked, re-evaluation of thresholds for defining impairment, and actions
to document and protect reference station conditions.
Actions that are associated with the listing of a stream reach as impaired, including
stressor identification and development of total maximum daily loads (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 illustrate plausible
mechanisms through which climate change can affect many of the activities in bioassessment
xxxv
-------
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.
xxxvi
-------
1. INTRODUCTION
Water-quality agencies use biomonitoring to assess the status and health of aquatic
ecosystems as required by the Clean Water Act; however, a major environmental
driver—climate—is changing in ways that have heretofore been largely unaccounted in terms of
its influences on biomonitoring and bioassessment. There is growing information on the effects
of climate change on aquatic ecosystems (e.g., Doledec et al., 1996; Durance and Ormerod,
2007; Buisson et al., 2008; Chessman, 2009; Flenner et al., 2010; Britton et al., 2010), with the
clear potential for these to affect many activities associated with biologically based assessment
programs. It is, therefore, important to consider the influence of climate change effects on
bioassessment approaches, and to adapt these programs accordingly. This project was
implemented with the goal of contributing to the foundation for understanding how potential
climate changes affect bioassessment indicators and for advancing the development of specific
strategies to ensure the long-term effectiveness of monitoring and management plans. The study
focuses on biological responses to climate change and on biological indicators, with the main
objectives of (1) investigating whether biological response signals to climate change are
discernible within existing bioassessment data sets; (2) analyzing how responses of a variety of
biological indicators can be categorized and interpreted with regard to apparent climate
sensitivity or robustness; and (3) assessing how changes in biological responses may influence
decision-making processes that are based on comparative interpretation of combined indicator
responses.
The study objectives make this a 'data mining' study. It attempts to use existing,
long-term biomonitoring data sets, which were collected for another purpose (i.e., to monitor the
status of stream biota using reference-based comparisons) to address a new question for which
the original collection programs were not designed. While there are certainly some questions
about climate change effects that can be addressed using spatial comparisons, for the most part,
climate change is a long-term temporal question, requiring trend analysis to investigate
long-term patterns in temperature, precipitation, flow, other habitat variables, and biologic
response variables. Given that at least some state biomonitoring programs have been in place for
long periods of time (e.g., 2+ decades), and that outside of this arena, long-term biological data
1-1
-------
sets are relatively rare, it is an attractive opportunity to apply these long-term biological data set
to the climate change-related questions that are the focus of this study.
This type of postfacto analysis of historic data sets is widely used to determine whether
climate change effects are already discernible in ecosystem responses (e.g., Daufresne et al.,
2004; Durance and Ormerod, 2007; Burgmer et al., 2007; Murphy et al., 2007). However, this
data mining approach has several pitfalls. One is the contrast between the focus of most
biomonitoring designs on spatial comparisons (i.e., between reference and nonreference sites),
and the fundamentally temporal comparisons that are needed to answer climate change questions
(e.g., evaluation of long-term trends). Other new objectives with respect to existing
biomonitoring design are the need to separate climate change effects from responses to
conventional stressors, and the need to be able to apply any observed results to a regional scale
(e.g., an ecoregion, a province, or a class of stations). As a result, for mined biomonitoring data
to ideally address the goals of this study would require not only having long-term data from a
few sites, but having such data at reference sites that are minimally affected by other major
anthropogenic stressors, and having such data from a number of regionally distributed,
representative locations. As is often the case with the opportunistic use of mined data, the
existing biomonitoring data sets available for analysis in this study do not always meet the
criteria that would have allowed the most rigorous evaluation of the study questions.
1.1. DECISION CONTEXT
In order to understand the implications of climate change impacts on bioassessment
programs, it is useful to consider the regulatory framework to which bioassessment programs
contribute. The U.S. Clean Water Act (CWA) of 1972 identifies the restoration and maintenance
of physical, chemical, and biological integrity as a long-term goal (Barbour et al., 2000).
Biological assessment, or 'bioassessment,' is applied worldwide as a valuable and necessary tool
for resource managers in achieving this goal (Norris and Barbour, 2009), and one that has been
found to be more effective than sampling only chemical parameters (Karr, 2006). This is largely
due to the recognition that biological indicators reflect an integrated response to all
environmental conditions to which they are exposed over time (Moog and Chovanec, 2000;
Barbour et al., 2000) and, thus, can provide information that may not be revealed by
measurement of concentrations of chemical pollutants or toxicity tests (Barbour et al., 1999;
1-2
-------
Rosenberg and Resh, 1993; Resh and Rosenberg, 1984). Biological assessment, coupled with
multimetric or predictive modeling analyses, is a strong approach for diagnosing diminished
ecological integrity, and minimizing or preventing degradation of river systems (Karr and Chu,
2000).
In the United States, biological assessment plays a central role in numerous water quality
programs that are components of the CWA. Bioassessment data are used to assess water quality,
identify biologically impaired waters, and develop National Water Quality Inventory reports.
Bioassessment is used to develop biocriteria and set aquatic life use categories, which represent
different protection standards. Bioassessment data are used to determine whether conditions of
the waterbody support designated uses, and if not, to develop total maximum daily load (TMDL)
limitations for the pollutant(s) contributing to the impairment. Bioassessment results are used to
help identify causes of observed impairments, based on the assumption that various components
of aquatic communities will respond differently to different types of stressors. Bioassessment is
used to determine the impacts of point source discharges as well as of episodic spills, defining
the extent of damage, responses to remediations, and supporting enforcement actions. Other
CWA programs that depend on bioassessment data include permit evaluation and issuance,
tracking responses to restoration actions, and other components of watershed management.
A variety of biological metrics and indices have been developed as ecological indicators
that are mainly applied to gauge the condition of aquatic ecosystems but also to judge causes of
degradation (Niemi and McDonald, 2004). They can serve as early warnings of degradation and
often simplify extensive and complex environmental data. Biological indicators should be
selected at appropriate spatial and temporal scales, incorporate natural variability, and be
sensitive to the range of stressors expected in a system (Niemi and McDonald, 2004). The
concept of linkage between biological indicators and the stressors on a system is crucial to the
interpretation of bioassessment results. It also means that all stressors impacting a resource must
be considered to achieve valid stressor identification and attribution of causes that can lead to
effective ecosystem management. "All stressors" must now go beyond conventional pollutants
to include climate change, as well as other global changes in land and water use (Hamilton et al.,
2010a).
It is clear that if interpretations of biological response patterns are compromised by not
accounting for the potentially important stressor of climate change, this could have wide-ranging
1-3
-------
consequences. To address the goals of this study, expectations for climate change effects on
stream ecosystems are briefly outlined, and these are illustrated in a simplified conceptual model
(see Figure 1-1). A subset of effects that are relevant to bioassessment programs and that can be
tested using biomonitoring data are identified and are the focus of this study. These effects
include shifts in community composition, relative abundances of component taxa, and richness
of various taxa components, biological metrics that are typically measured and relied on in
biomonitoring programs. The mechanisms through which climate changes can translate to
changes in stream conditions and biological responses are used to define hypotheses for
responses of various biological metrics that were tested to address study objectives.
earn Abiotic Responses
Stream Biotic Response
Air temperature
Precipitation
Sea level
Ice cover/Snowmelt
Evapotranspiration
Drought
Water temperature Ice cover
Hydr ologic regime Freeze/thaw
Groundwater Habitat
Water Quality Materials cycling
Growth
Metabolism
Morphology
Demographic rates
Reproductive cycles
Evolutionary
adaptation
Selection (thermal,
hydro tolerances)
Community
composition
Abundance
Food web
Phenology
Figure 1-1. Conceptual model of the linkages between climate forcings,
climate system changes, stream habitat changes (abiotic), and the subsequent
individual-, population- and community-level responses to these changes.
1-4
-------
1.2. CLIMATE CHANGE EFFECTS AND ECOLOGICAL RESPONSES
Changing patterns of climate forcing are expected to alter spatial and temporal patterns of
air temperature and precipitation and drive changes in sea level rise, ice cover, timing and
magnitude of snow melt, evapotranspiration, drought, flooding magnitude and frequency, and
other extreme events (see Figure 1-1). Changes in air temperature and precipitation are
two principal factors that will impact stream and river ecosystems through direct effects on water
temperature and hydrologic regimes, and through indirect effects on dissolved oxygen (DO), pH,
nutrients, and other dissolved constituents, changing the assimilation capacity of pollutants into
receiving waters, sediment erosion and deposition, and habitat structure (see Figure 1-1).
Global or large-scale regional projections for changes in air temperature and precipitation
patterns are the most readily available climate change projections and are important because they
bound our expectations for overall magnitude and direction change. Multiple general circulation
models (GCMs) provide us with an ensemble of projected changes in temperature and
precipitation patterns (Intergovernmental Panel on Climate change [IPCC], 2007a). Global
average projections of temperature increases over the next century range from 1.1-2.9°C for the
lowest emissions scenario to 2.4-6.4°C for the highest emissions scenario (IPCC, 2007a). This
represents a higher rate of increase (about 0.2°C per decade) than the last 50 years (0.13°C per
decade), and further rate increases are considered possible (IPPC, 2007b; Ramstorf et al., 2007;
Hansen et al., 2006).
The ensemble of GCM results are more uncertain in their projections for precipitation
than for temperature and are variable among major geographic regions of the United States.
Details of precipitation projections for regions that correspond to study areas of this project are
presented in subsequent chapters. However, general projections include increased frequency of
heavy precipitation events, more precipitation in winter and less precipitation in summer, more
winter precipitation as rain instead of snow, earlier snow-melt, earlier ice-off in rivers and lakes,
longer periods of low flow, and more frequent droughts in summer (IPCC, 2007a; Barnett et al.,
2005; Hayhoe et al., 2007; Fisher et al., 1997). Changes in air temperature and precipitation
patterns will drive changes in stream thermal and hydrologic regimes, which in turn will directly
and indirectly influence biota (see Figure 1-1). As a result, available measures of stream thermal
and hydrologic conditions (highlighted in Figure 1-1), as well as surrogates thereof, were the
focus of climate change analyses in this study.
1-5
-------
Freshwater ecosystems are considered sensitive to climate change impacts, because of
their fundamental dependence on hydrology and thermal regimes, their dominance by
poikilotherms, and the risks of interactions with other stressors (Durance and Ormerod, 2007).
Documentation of aquatic biological responses to climate change on a basis that is meaningful to
water quality and resource managers has been slow in coming, with much early attention focused
on terrestrial ecosystems (e.g., Root et al., 2003; Thuiller, 2004; Walther et al., 2002, 2005;
Parmesan, 2006; Tobin et al., 2008; Suding et al., 2008; Zuckerberg et al., 2009). However,
there is an increasing body of information of aquatic ecosystem responses to climate change.
Table 1-1 summarizes several salient examples of observed changes in aquatic community
structure that are relevant in a bioassessment framework, though a broader range of biological
responses is included in the conceptual model (see Figure 1-1). These can have potentially major
consequences both for ecosystem function and for the interpretation of biomonitoring results
relative to an assessment of ecosystem health. Expectations for the types, direction, and
magnitude of biological responses should be linked to the magnitude and direction of climate
change projections for each region, and potentially ameliorated by local factors. That is,
biological responses are likely to be species- and/or trait-group-specific, and may vary
regionally. The literature results in Table 1-1, comprising a range of observed and expected
biological responses, are used to develop hypotheses for testing potentially sensitive trait and
taxonomic groups of invertebrates for responses to changes in stream temperature and flow
conditions.
A study of possible biological responses to climate change suggests we are not only
documenting biological responses over time or to climate-related habitat conditions (e.g., stream
temperature or flow metrics), but making causal linkages between climate change trends and
biological responses. Causal attribution requires several logical linkages: (1) that long-term
changes in climate factors (e.g., air temperature, precipitation metrics) have in fact occurred in
the regions being studied; (2) that those climate changes can be associated with changes in
stream conditions (e.g., in stream temperature and flow metrics); and (3) that observed biological
trends and responses are associated with those changes in stream temperature and/or flow. In
fact, to attribute observed biological responses to long-term climate change to the exclusion of
other potential contributing causes, such as multidecadal climate oscillations, landscape stressors
1-6
-------
Table 1-1. Examples of observed changes in aquatic community structure
related to climate change that are relevant in a bioassessment framework
Examples of aquatic community changes
Increases in abundance, species richness, and proportion of southern and
warm -water species offish in large rivers
Loss of cold-water fishes from headwater streams, but also extension of
more tolerant, thermophilic fishes from larger streams and rivers into
newly suitable habitat
Increases in fish species richness with increasing temperatures at higher
latitudes
Displacement of upstream, cold-water invertebrate taxa with
downstream, warm-water taxa
Increase in lentic and thermophilic invertebrates with increasing
temperature
Reductions of spring abundance of dominant taxa, shifts in invertebrate
assemblage composition from cooler to warmer water taxa, and possible
losses (local extinctions) of scarcer taxa with increasing temperatures
Significant long-term trends related to the thermophily and rheophily of
benthic taxa, with groups preferring cold waters and higher flows
declining
Changes in stability and persistence
Changes in species composition in lakes
Changes in structure and diversity of riverine mollusk communities with
reduction in community resilience during hot years
Reference
Daufresne and Boet, 2007
Buisson et al., 2008
Hiddink and Hofstede,
2008
Daufresne et al., 2004
Doledec et al., 1996
Durance and Ormerod,
2007
Chessman, 2009
Collier, 2008
Burgmer et al., 2007
Mouthon and Daufresne, 2006
such as urbanization, nutrient enrichment, sedimentation, habitat alteration, or others, would
require biological responses to be highly specific to climate change trends. This puts demands
on a study design that are not entirely achievable using data collected in a biomonitoring
framework, and especially using data mined from design parameters focused on dissimilar study
objectives. For example, the length of available biological records is seldom more than about
2 decades, and because this is well within the duration of a single multidecadal climate
oscillation, it is not possible to analytically separate potential contributions of such climate
cycles from those of long-term directional climate change using the bioassessment data sets
1-7
-------
analyzed in this study. In addition, the interactions between climate cycles such as the Pacific
Decadal or North Atlantic Oscillations (NAOs) can act synergistically or antagonistically with
climate change, depending on their phases (e.g., Seager and Vecchi, 2010), potentially enhancing
or obscuring the types and magnitudes of biological responses that might be expected over the
long term. In this perspective, 'climate change' can be considered the long-term, average
directional changes that span multiple climate cycle oscillations. However, recognition of the
types and directions, and with caution, the magnitudes of biological responses to changes in
climate-associated factors, and the identification of biological metrics that are sensitive to such
climate-associated changes, can be inferred from linkages between changes in climate factors,
associated changes in stream conditions, and associated changes in biological metrics, even in
the absence of the ability to partition long-term direction and cyclic climate patterns.
In the strictest sense, establishing cause and effect may require controlled experimental
design. However, it is common practice to infer probable sources of cause by clear associations
between types and sources of stressors present and responses of biota whose autecology
characteristics are known (Norris and Barbour, 2009; Cormier and Suter, 2008). This study used
various approaches to strengthen inferences drawn in relation to the main questions. When
available, we obtained long-term records of temperature, precipitation, and/or streamflows to
place data from the period of record into an historic perspective and assessed the plausibility of
the magnitude of stream temperature increases estimated for the biological periods of record in
comparison to literature study results. We cross-checked the calculation of temperature optima
for taxa used as a basis for defining thermal preference trait categories for examination and
attribution of climate change-related biological responses with other classifications when
possible, and confirmed the temperature optima through the estimation of benthic-inferred
temperature trends. Though truly pristine reference conditions are rarely available, this study
examined long-term biological trends and responses at minimally or least-impacted long-term
monitoring stations, to limit known confounding by anthropogenic factors aside from
climate-related alterations. In addition, we validated the success of this approach to the extent
possible with available data on covariates that might also explain observed trends.
-------
1.2.1. Expectations for Thermal Regime Changes and Associated Biological Responses
Changes in the thermal regimes of streams and rivers in response to climate change have
been documented from long-term river temperature data sets around the country (e.g., Kaushel
et al., 2010). Stream water temperature patterns closely follow air temperature patterns (e.g.,
Mohseni et al., 2003; Pilgrim et al., 1998; Stephan and Preudhomme, 1993). They are not
directly driven by air temperature, but rather solar radiation as the primary heat source influences
changes in stream temperature regimes (Allan and Castillo, 2007; Ward, 1985); other influences,
including variations in flow volume and snow melt, ground water influence, aspect, riparian
shading, presence of deep pools, meteorology, river conditions, and geographic setting also
influence stream temperatures (Allan and Castillo, 2007; Caissie, 2006; Mohseni et al., 2003;
Daufresne et al., 2004; Hawkins et al., 1997; Ward, 1985). These factors contribute to regional
differences in stream water temperature responses to climate change forcing. The effects of
water temperature can also interact with stream flow alterations, with higher temperatures and
higher warming rates during low flow conditions (van Vliet and Zwolsman, 2008; Zwolsman and
Van Bokhoven, 2007; Sinokrot and Gulliver, 2000). As a result, influences of stream
temperatures and flow conditions cannot always be separated in terms of their effects on biota.
It is clear that water temperature is an important ecosystem driver, affecting water quality
and the distribution of aquatic species (Caissie, 2006). Temperature regimes determine the
distribution and abundance of aquatic species through temperature tolerances and evolutionary
adaptations, along with competitive interactions, effects on food supply, and other factors (e.g.,
Matthews, 1998; Hawkins et al., 1997; Vannote and Sweeney, 1980; Sweeney and Vannote,
1978). A variety of individual-, population- and community-level changes ensue, including
altered phenology (Gregory et al., 2000; Harper and Peckarsky, 2006); changes in the number
and/or timing of reproductive periods (Hogg et al., 1995; Flanagan et al., 2003; Hampton, 2005);
and selection for new thermal or hydrological tolerances (Rahel et al., 1996; Stefan et al., 2001;
Golladay et al., 2004; Gibson et al., 2005) (see Figure 1-1).
Many of these categories of biological responses to climate change are difficult or
impossible to discern using the types of data (i.e., collection methods, timing and frequency of
collections, metrics measured) typically obtained from biomonitoring programs. Changes in
phenology, timing, or number of reproductive cycles, and altered utilization of food resources are
examples. Instead, biomonitoring programs are designed to characterize community structure,
1-9
-------
composition, abundance, and richness usually on an annual basis during a selected index period.
It is change in metrics related to these community characteristics that this study assessed for
possible responses to changing climate-related conditions.
Changes in the thermal regime of a stream/river can result in decreases in sensitive taxa
or increases in tolerant taxa. For many taxa, summer temperatures can represent upper bounds of
temperature preferences (although not necessarily true thermal maxima). However, winter
temperatures or the seasonal timing of some thermal cues or thresholds may also be important in
controlling distributions. One paradigm is that as climate change alters the spatial distribution of
the 'climate envelope' that represents the appropriate thermal regime for a taxon that the taxon
will shift its distribution accordingly. This can include range shifts northward or to higher
elevations for cold-preference taxa as, for instance, the more southerly or lower elevation
portions of their historic range become warmer, and some temperature tolerances are exceeded.
On the other hand, warm preference or more broadly tolerant taxa might increase in abundance
or extend their range abundance (e.g., Hamilton et al., 2010b). Because taxa are likely to
respond at different rates, altered abundances and distributions of temperature (or hydrologically)
sensitive and tolerant taxa will result in new species interactions and community compositions.
1.2.2. Expectations for Hydrologic Changes and Associated Biological Responses
Some of the major impacts of projected climate changes on stream systems will be to
their hydrologic characteristics. The IPCC (2007a) projects average annual runoff to increase by
10-40% at high latitudes and some tropical areas, but to decrease by 10-30% over some
midlatitudes dry regions and the dry tropics. In North America, projected changes in average
stream flow range from an increase of 10-40% at high latitudes to a decrease of about 10-30%
in midlatitude western North America by 2050 (Milly et al., 2005). In western/southwestern
snow-pack dominated regions, the combination of warming temperatures, a shift toward less
winter precipitation falling as snow, and snow-melt occurring earlier will shift the peak runoff
from spring to late-winter/early spring, accompanied a by reduced magnitude of snowpack
(Barnett et al., 2005, Clow, 2010). Typical projections are for peak runoff to shift from about
2 weeks up to 1 month earlier by the end of the century (Dettinger et al., 2004, Hayhoe et al.,
2007). Stewart et al. (2005) has already found evidence for shifts of this magnitude (1-4 week
1-10
-------
earlier timing of snow melt and runoff based on data from 1948 to 2002) for several montane
catchments in the western United States.
Numerous studies have demonstrated the importance of hydrologic changes on biological
responses (Webb et al., 2009; Dewson et al., 2007; Suren and Jowett, 2006; Lind et al., 2006;
Poff, 2002; Extence et al., 1999; Stanley et al., 1994). Significant associations between
hydrologic variables and trait modalities also have been documented in a number of other studies
(e.g., Horrigan and Baird, 2008). Dewson et al. (2007), Poff and Zimmerman (2009), and
McManamay et al. (2011) reviewed the literature documenting a broad range of biological
responses to changes in hydrologic conditions. Hydrologic regime of a stream is not a singular
variable, and the range of hydrologic alterations that can result from the combination of
increasing magnitude and variability of temperatures combined with a range of projected
changes in precipitation and drought conditions is great. As examples, these may include longer
duration and lower summer low flows, decreases in average discharge, greater incidence of
floods, greater flashiness, and many others. Dewson et al. (2007) found that invertebrate
abundances changed (increased or decreased based on flow preferences as well as habitat
availability) in response to decreases in discharge, whereas invertebrate richness decreased with
flow changes that resulted in decreased habitat diversity. Flow alterations affecting food
resources were also important in affecting invertebrate responses. Carlisle et al. (2010) found
that reduced stream discharge was the best predictor of reduced integrity of invertebrate and fish
communities. Under reduced flow conditions, fish and invertebrates that increased tended to
have traits typical of nonflowing (e.g., lake) environments, such as preferences for fine-grained
substrates and slow-moving currents, and also traits that allow escape during parts of the life
cycle.
Some of these studies are particularly relevant because they document responses to
extreme and variable hydrologic conditions, similar to those that are projected to occur as a result
of climate change. Several were conducted in streams in Mediterranean-climate regions, where
the harsh and variable climatic conditions strongly influence biological assemblages, and we
utilized this information during metric development. Results show that organisms with
resilience or resistance trait modalities, such as high dispersion and colonization capabilities,
resistance to desiccation and aerial breathing, were generally prevalent in drier, harsher climatic
conditions (Beche et al., 2006; Bonada et al., 2007b; Diaz et al., 2008). Trait modalities that
1-11
-------
confer resilience to extreme conditions were also prevalent in macroinvertebrate communities
recovering from severe drought conditions in Georgia, United States (Griswold et al., 2008).
Organisms that recovered most rapidly generally had short life cycles, resistance to desiccation,
small body size, armoring, and abundance in drift. As flow increased and habitat conditions
stabilized, larger, soft-bodied organisms that are rare in drift became more prevalent (Griswold
et al., 2008).
1.3. CONCEPTUAL MODEL LINKAGES STRUCTURING THE STUDY APPROACH
Figure 1-1 is a conceptual model that highlights the linkages between the changing
components of the climate system and various aspects of aquatic ecosystems. There are
numerous interacting pathways that link climate factors that help form regional ecosystem
characteristics with other environmental drivers, to yield a range of biological responses in
stream and river ecosystems. Projections for changes in air temperature and precipitation
patterns are the main climate drivers that are linked to stream and river habitat conditions (see
Figure 1-1). Other closely related predictions, such as for earlier snowmelt or increased drought
frequency or duration, expand the picture of climate-related alterations that are likely to
influence stream ecosystem characteristics. These changing regional climate conditions will
alter stream environments through direct and indirect processes that will lead to altered thermal
and hydrologic regimes, changes in groundwater conditions and baseflow, changes in waterborne
chemical constituents and water quality conditions, and changes in physical habitat
characteristics, such as stream morphology and substrate type (see Figure 1-1). Interactions
among these abiotic characteristics, as well as with other existing pollutants and stressors on the
stream, will contribute to changes in stream biota at every level of organization—individual,
population, community, and ecosystem. Though not intended to be exhaustive, Figure 1-1 lists
numerous anticipated and observed biotic responses that link mechanistically to the range of
climate-driven abiotic changes shown. Many of the biological responses in Figure 1-1 are not
typically measured as part of a biomonitoring program but could contribute to outcomes of
community structure that are used to characterize condition. For example, changes in number of
reproductive cycles per year for a species is not a typical bioassessment metric; however,
changes in reproductive patterns and success can alter the community composition and relative
1-12
-------
abundance measured at any time. Some other types of biological responses, such as genetic
adaptation, are only tangentially relevant to bioassessment measurements.
The strategy for this study was to consider how the multiple mechanisms through which
climate-related changes impacting streams (alterations in thermal regimes, hydrologic regimes,
water quality, nutrient status, habitat conditions, and essential interactions with food supply,
competitors, and predators) could combine to yield changes in commonly measured metrics of
community composition, abundance and richness, and then use this information to postulate what
types of biological metrics might best capture the predicted responses. Changes in the thermal
and hydrologic regimes of streams are the most direct links between the climate drivers of air
temperature and precipitation. Changes in ice cover and snowmelt patterns and expectations for
increasing drought conditions contribute to changes in hydrologic regime. Changing thermal and
hydrologic characteristics of a stream can contribute directly to alterations in community
composition, abundance, and richness through numerous direct as well as indirect mechanisms.
Mechanisms of action can be through traits related to temperature and flow preferences and
tolerances, as well as traits that confer ability to adapt to droughts or flood disturbances, or to
recover from increased stress, including greater variability in temperature and flow conditions.
Other expectations from this model include potential responses of taxonomic groups considered
sensitive to other perturbations and responses of feeding guilds through indirect effects of altered
availability of food resources. Accordingly, this study explores and develops temperature and
flow preference trait groups, and examines the responses of these, as well as trait groups related
to habit, feeding type, size, and mobility. Taxonomic groups considered sensitive or tolerant to
conventional stressors and metrics that are commonly utilized in biomonitoring programs are
also investigated. Boxes outlined in bold in Figure 1-1 identify the climate, stream condition,
and biological components on which this study focused.
The effects of global change on bioassessment programs will vary regionally. Land and
water use effects are largely driven by locations of and projected future changes in major
population and agricultural centers. Differences in the severity of climate change impacts are
instead driven by regional variability in climate, as well as regional differences in the
vulnerability of aquatic ecosystems. Differences in regional climate and disturbance regimes are
important contributors to species sensitivities to environmental changes (Helmuth et al., 2006).
Many factors can influence susceptibility to changing water temperature or hydrologic regime
1-13
-------
due to 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). To the extent possible, factors such as these that can affect
the sensitivity of biota to overarching climate change influences, including elevation, ecoregion,
and stream size, are examined in this study.
This study uses four regionally distributed state bioassessment data sets from Maine,
North Carolina, Ohio, and Utah to examine historical trends in relation to temperature,
precipitation, flow, and other environmental drivers. We use community and traits analyses to
identify potential indicators, both sensitive and insensitive (robust) to climate change effects.
Examination of climate-sensitive traits facilitates transfer of analysis results to other places.
Additional analyses focusing on the vulnerability of reference conditions and the interactions
between climate change and other landscape-level stressors, especially land use, supplement
these results. This study builds on the results of a preliminary analysis (U.S. EPA, 2008) and
feedback from a workshop convened in 2009 with state and tribal scientists and resource
managers, academic and agency experts, and decision makers to explore the following issues: the
effects of climate change on endpoints of concern; methods for integrating climate change into
existing state and tribal water quality programs; and ways to create opportunities for adaptation.
Study findings are summarized in the beginning of this report in the Summary for
Managers and Policymakers. The body of the report expands on the analyses that support these
findings. Section 2 describes methods used, including types and sources of data; data
preparation; biomonitoring station characteristics; climate conditions and climate change
projections for regions analyzed; thermal, hydrologic, and combined indicator development;
methods used for trend, categorical, and spatial analyses; and approaches for assessing impacts
to biomonitoring program decisions. Sections 3, 4, 5, and 6 apply these methods and summarize
results for each of the four state biomonitoring data sets evaluated. Section 7 integrates results
across regions, analyzes implications to environmental management, and discusses design
considerations for a monitoring network to detect climate change effects. While all primary
analysis results are summarized in the main report, some detailed results and supporting
materials are compiled in appendices.
1-14
-------
2. METHODS
2.1. DATA GATHERING
2.1.1. Exposure Data
We gathered regional and state-specific data on historic and future projected climatic
changes. Our goals were to evaluate the direction and rate of change in temperature and
precipitation patterns in each state and region, to examine differences in spatial patterns of
change within each state (i.e., identify 'hot spots'), and to compare the magnitude and direction
of future projected changes. We based summaries of regional projections on results from
literature searches. For the state-specific summaries, we obtained annual and seasonal air
temperature and precipitation data from the Climate Wizard Web site
(http://www.climatewizard.org/). We ran linear trend analyses on these data for two historic
time periods: 1901-2000 and 1971-2000. The base data for these historic trend analyses came
from the Parameter-elevation Regressions on Independent Slopes Model (PRISM) Group,
Oregon State University (http://www.prismclimate.org, Gibson et al., 2002). PRISM data are
modeled data that utilize a digital elevation model and point measurements of climate data to
generate estimates of annual, monthly, and event-based climatic elements with a 4-km
resolution. In addition to running linear trend analyses, we generated maps for each state based
on 1971-2000 averages to evaluate spatial differences in temperature and precipitation patterns.
We also used the Climate Wizard Web site to gather data on projected changes in annual
and seasonal air temperature and precipitation for high (A2) and low (Bl) emissions scenarios
for mid (2040-2069) and late (2070-2099) century compared to an historic (1961-1990) time
period. Data from 15 different GCM were evaluated (see Table 2-1). We used these data to
calculate ensemble minimum, maximum, and average values. In addition, we calculated
standard deviations (SDs) to assess levels of uncertainty across models.
2.1.2. Temperature and Streamflow Data
Water temperature is the most proximate measure of thermal change in streams. Efforts
were made to acquire all available site-specific water temperature data for the biological
monitoring sites in each state. The available data were primarily instantaneous measurements
taken at the time of each biological sampling event. In only a few instances, where the
2-1
-------
Table 2-1. Future projection data from 16 GCMs were evaluated. These
data were obtained from the Climate Wizard Web site
(http://www.climatewizard.org/)
GCM
BCCR-BCM2.0
CGCM3.1(T47)
CNRM-CM3
CSIRO-MkS.O
GFDL-CM2.0
GFDL-CM2.1
GISS-ER
INM-CM3.0
IPSL-CM4
MIROC3.2(medres)
ECHO-G
ECHAM5/MPI-OM
MRI-CGCM2.3.2
CCSM3
PCM
UKMO-HadCM3
Country
Norway
Canada
France
Australia
USA
USA
USA
Russia
France
Japan
Germany/Korea
Germany
Japan
USA
USA
UK
Institution
Bjerknes Centre for Climate Research
Canadian Centre for Climate Modelling & Analysis
Meteo-France/Centre National de Recherches
Meteorologiques
CSIRO Atmospheric Research
U.S. Department of Commerce/NOAA/Geophysical
Fluid Dynamics Laboratory
U.S. Department of Commerce/NOAA/Geophysical
Fluid Dynamics Laboratory
NASA/Goddard Institute for Space Studies
Institute for Numerical Mathematics
Institut Pierre Simon Laplace
Center for Climate System Research (The University
of Tokyo), National Institute for Environmental
Studies, and Frontier Research Center for Global
Change (JAMSTEC)
Meteorological Institute of the University of Bonn,
Meteorological Research Institute of KMA, and
Model and Data group.
Max Planck Institute for Meteorology
Meteorological Research Institute
National Center for Atmospheric Research
National Center for Atmospheric Research
Hadley Centre for Climate Prediction and
Research/Met Office
site happened to be located near a United States Geological Service (USGS) gage, we were able
to find continuous water temperature data, and even then, it was for a limited number of years.
Continuous data are preferable over instantaneous measures because they capture more aspects
of the true thermal regime, such as timing, duration, and frequency of extremes. We made
similar efforts to acquire site-specific streamflow data, but these data were only available for a
limited number of sites. If sites were colocated with USGS gages, we downloaded daily
2-2
-------
streamflow data from the USGS real-time flow data Web site (http://waterdata.usgs.gov/nwis/rt).
In other instances, state biomonitoring programs were able to provide instantaneous streamflow
measurements that were taken at the time of the biological sampling event.
2.1.3. Biological Data
Routine biomonitoring data from Maine, Utah, and North Carolina were compiled into
Ecological Data Application System (EDAS) databases, which are custom database applications
that are used with Microsoft Access. For Ohio, data were originally obtained from STORET;
however, interactions with the Ohio Environmental Protection Agency (EPA) revealed that data
generation, database development, and management, as well as ongoing analyses for Ohio are
conducted by Ed Rankin and Chris Yoder of Midwest Biodiversity Institute (MBI). Therefore,
data manipulation and analyses for Ohio were conducted by MBI under subcontract to Tetra
Tech. The Ohio database included both fish and macroinvertebrate data. The Maine, Utah, and
North Carolina databases contained macroinvertebrate data only.
Taxonomic data were screened in order to minimize the chance of detecting false trends
due to changes in field and laboratory protocols (e.g., differences in collection methods,
differences in sample processing/sub sampling methods, changes in taxonomists, and/or
taxonomic keys). In the Maine, Utah, and North Carolina data sets, preliminary iterative data
summaries, and screening procedures included
Tabulating numbers of samples by station (e.g., station name, station ID number, and/or
sample ID number) and date. Results were examined for consistent number of samples
by station/date and for breaks in sample collection at stations across years. Problems
discovered through this approach included changes over time in collection methods
and/or reporting of replicates and errors or changes in station naming that resulted in data
for the same location appearing under different station names. It also helped identify
locations with long-term data records.
Tabulating total abundance and total number of taxa by station and collection date.
Results were examined for discontinuities in magnitude or trends in values between
stations and across dates. Problems discovered through this approach included changes
in reporting of abundances (e.g., from number per sample to number per square meter);
whether replicates were averaged, summed, or reported separately and changes or errors
in whether subsampling was applied during sample analysis and how it was accounted for
in the data.
2-3
-------
Tabulating taxa (at the lowest levels reported) by collection date. For these, either taxa
abundance or occurrence was tabulated, and these were either averaged over all stations
within the state, or within each ecoregion and/or other appropriate subset (e.g., river basin
or watershed). For this purpose, the tabulations of taxa were placed in phylogenetic
order, and some higher-level phylogenetic structure (e.g., order and family names, or
others as needed) was included for reference. Results were examined for many types of
patterns, including
o changes in taxonomic naming over time (e.g., changes in genus- or higher-level
names, changes in placement within families). This not only revealed changes in
systematics over time, but also caught changes in taxonomists and/or labs used to
analyze samples.
o changes in level of attribution over time (e.g., increasing use of species names in
recent years where individuals were typically left at the genus or family level in
earlier samples);
o changes in other types of naming conventions (e.g., changes in level of placement
for taxa such as water mites).
Problems identified through these procedures included extensive changes in taxonomic
knowledge and systematics over the decades of sample analysis. For illustration, one example is
changes in the mayfly genus Ephemerella, including changes in the inclusion of various species
ofEphemerella between Ephemerella and Drunella. In addition, we found many instances of
changes in the higher-level groups under which various taxa would be reported, so that in the
database, the same genus (or species, or family) would appear in more than one place. The effect
of this was that these would act like separate taxa when a taxa ID name or number was invoked
for trend analysis. Many associated corrections were applied to the phylogenetic structuring and
naming conventions in the databases.
To address issues associated with changes in taxonomic naming of genera and/or species,
or greater prevalence of species identifications in recent years, we followed the guidelines of
Cuffney et al. (2007) to develop OTUs for the Maine, Utah, and North Carolina data sets. OTU
development involved summing species to the genus level (or similar procedures at other levels),
or combining two or more genera that could not always be reliably separated. The intent of
OTUs is to exclude ambiguous taxa from analyses and include only distinct/unique taxa.
Because a complete and correct master taxa list is required before OTUs can be established, the
master taxa lists in each of the databases were first verified through several iterative procedures.
2-4
-------
Next, three levels of OTUs were established: lowest taxonomic unit (generally species), genus,
and family. Rules were developed based on a general procedure of Remove Parent/Merge
Children (retain the Child taxa [finer level of detail] and remove the Parent tax on or merge the
Child taxa into the Parent taxon). According to Cuffney et al. (2007), this appears to be the most
robust method for retaining taxa richness and abundance information for further analysis. All
decisions were data set dependent. Rules were created on the data set as a whole and then
applied to individual samples prior to analysis. The last step in the process was to manually
review the list of OTU designations and make final corrections where necessary. Genus-level
OTUs were generally found to be most appropriate, although there were some exceptions (e.g.,
in the Utah database, a family-level OTU had to be used for Chironomidae due to inconsistencies
arising from a change in taxonomy labs).
In the Ohio data set, MBI developed a program to scan for changes in taxonomy over
time that could affect calculations of Ohio EPA's Invertebrate Community Index (ICI) (DeShon,
1995). The program provided a listing of the first and last occurrence of each taxon in the
Ohio EPA database. MBI used this to extract a list of possible taxa that could affect ICI scoring
via taxonomic refinement (splitting or lumping of taxa). MBI then conferred with senior
Ohio EPA taxonomists to determine how to best address these changes. Their efforts primarily
resulted in "lumping" individual taxa designations of mayflies back to "Baetis sp." or
"Pseudocloeon sp." Table A-l in Appendix A lists the mayfly taxa that appeared earlier and
then "disappeared" or those that "appeared" later, mostly at resampled sites.
In the Maine, North Carolina, and Utah data sets, we used Nonmetric Multidimensional
Scaling (NMDS) to evaluate whether the database 'fixes,' and in particular the taxonomic
corrections and application of OTU rules, were effective in minimizing changes over time due to
taxonomic identification procedures rather than actual community changes. NMDS is an
ordination that takes the taxa in the samples and shows in ordination space how closely related
the samples and stations are based on their species composition. Grouping variables (e.g., year,
month, collection method, taxonomy lab, ecoregion, watershed, etc.) can be overlaid to look for
trends. The NMDS ordinations were performed only on least-disturbed sites in order to
eliminate differences due to other disturbances. The NMDS ordinations were run before and
after generating genus-level OTUs. Patterns were examined for distinct shifts that might indicate
2-5
-------
changes in taxonomists or labs during the sampling period of record, as well as ineffective OTU
procedures. Section A.2 in Appendix A contains the NMDS plots.
In addition to taxonomic data, we compiled life history, mobility, morphology, habitat,
and resource acquisition traits data for North American macroinvertebrate taxa found in lotic
systems. Advantages of using traits data are that they are less susceptible to taxonomic
ambiguities or inconsistencies in long-term data sets; they can detect changes in functional
community characteristics; and they vary less across geographical areas, which allows for
larger-scale trend analyses across regional species pools. Traits data for 3,857 North American
macroinvertebrate taxa were compiled into the Freshwater Biological Traits database (U.S. EPA,
2012).
2.1.4. Site Information
In addition to water temperature, streamflow, and biological data, we gathered all
available water chemistry, habitat, and land-use data from the state biomonitoring programs.
These data allowed us to screen for potential nonclimatic factors that may have influenced trends
in the biological data over time. The amount and type of data available for each state varied.
Because of this, we used a Geographic Information System (ArcGIS 9.2) to obtain a standardized
set of parameters for each biological sampling site. These included 2001 National Land Cover
Data (Vogelmann et al., 2001) within a 1-km buffer zone, site-specific elevation, and EPA
Level 3 and 4 ecoregions. The 1-km distance for the land-use buffer was arbitrary and was
intended to provide a measure of potential anthropogenic stressors in the surrounding area. We
aggregated land-use classifications into broad categories (e.g., urban and agricultural).
2.2. DERIVATION OF INDICATORS
2.2.1. Thermal Preferences
We used weighted-average modeling or related approaches (e.g., maximum likelihood
estimates, general linear modeling) to develop lists of candidate taxa in each state that could
potentially serve as indicators of thermal change. The methods described in Yuan (2006) were
used to estimate the optima values and ranges of occurrence (tolerances) for temperature for
OTUs that had a sufficient distribution and number of observations to support the analysis.
2-6
-------
Weighted averaging is a simple, robust approach for estimating the central tendencies of
different taxa, or in our case, temperature optima and tolerance values (ter Braak and Looman,
1986). The basic approach is a straightforward weighted average—the temperature at each site
in a state at which the species is observed, multiplied by the relative abundance of the species at
that site, with the sum over all sites of the weighted temperatures divided by the sum of the
abundances of that species from all sites. This mean temperature is taken as the preferred
temperature for the taxon, and the breadth of the distribution (size of the standard deviation or
other measure of spread) represents an estimate of the tolerance or sensitivity of the taxon.
Table 2-2 and Figure 2-1 illustrate the approach.
Table 2-2. Example of how a weighted average model temperature optimum
(weighted mean) estimate is calculated
Species A temperature preference
Station ID
A
B
C
D
Sum
Relative abundance
0.10
0.02
0.02
0.04
0.18
Observed temperature
22
33
12
14
RA x temp.
2.20
0.66
0.24
0.56
3.66
Weighted average = 3.66/0.18 = 20.3333, RA = relative abundance, temp. = observed temperature.
When using weighted averages, a wide distribution of samples across the environmental
gradient results in a more robust estimate of temperatures of occurrence and, therefore, of
inferred preference. For a given state data set, weighted-average tolerance values for each OTU
are computed using the same set of environmental data; therefore, any bias arising from an
uneven distribution of data will be the same for all OTUs, and their relative placement along the
temperature gradient will generally be preserved.\
The generalized linear model is also used to estimate taxon-environment relationships for
each combination of taxon and environmental variable. In addition to providing a means of
computing tolerance values, regression estimates of the taxon-environment relationship quantify
2-7
-------
o
Temperature (°C)
Figure 2-1. Illustration of weighted average temperature distribution, where
the weighted average mean (u) is taken as the temperature optimum
(preference) for the taxon, and the magnitude of SD is taken as an estimate of
the temperature sensitivity or tolerance.
the strength of the association between a given environmental gradient and changes in the
occurrence probability or abundance of a taxon. In the case of presence/absence data, the
response variable is modeled as a binomial distribution; in the case of abundance data, a negative
binomial distribution is often assumed (maximum likelihood estimates).
Weighted-average calculations were used for the states that had absolute (noncategorical)
abundance data by taxon. If only presence/absence (categorical or qualitative abundance) data
were available, a generalized linear model was used. Calculations were made separately for each
state. For the Maine, North Carolina, and Utah analyses, stations across all ecoregions were
grouped together, and data were subset to account for seasonal variation (when needed), as well
as for variation associated with different sampling methods. For example, in Utah, only samples
collected during the fall index period were used. In North Carolina, only samples collected using
a certain method were analyzed. OTUs that occurred in fewer than 20 samples were excluded, as
low sample size affects the regression model and biases the optima and breadth values for rare
taxa, especially under extreme conditions.
Because the specific characteristics of each state data set varied (e.g., range of collection
dates, station locations, elevation), and because the methods used to derive the thermal optima
and tolerance estimates also varied in some cases, we developed an arbitrary ranking scheme to
2-8
-------
make results more comparable across data sets. For the Maine, North Carolina, and Utah data
sets, we assigned taxa rankings ranging from 1 to 7 based on percentage within each data set.
We designated taxa with rankings <3 (<40th percentile) as preliminary cold-water taxa and taxa
with rankings >5 (>60th percentile) as preliminary warm-water taxa (see Table 2-3). Thermal
optima and tolerance values were not available for all taxa, so we used literature, primarily the
traits matrix in Poff et al. (2006b) and the USGS traits database (Vieira et al., 2006), as a basis
for making some additional initial designations.
Table 2-3. Example taken from Utah analysis results to illustrate
development of ranking for temperature (or other environmental parameter)
preference and tolerance rankings from weighted-average or generalized
linear model temperature distribution results. Ranks 1-3 are cold
stenotherms; Ranks 5-7 are warm eurytherms
Rank
1
2
3
4
5
6
7
Percentage
0-0.1
0.1-0.25
0.25-0.4
0.4-0.6
0.6-0.75
0.75-0.9
0.9-1.0
Optimum
4.6-6.7
6.8-7.6
7.7-8.3
8.4-9.1
9.2-9.6
9.7-10.4
10.5-15.7
Tolerance
2.0-2.7
2.8-3.2
3.3-3.5
3.6-3.7
3.8-3.9
34.0-4.3
4.4-5.1
After making these preliminary cold- and warm-water designations, we refined the lists
based on case studies and best professional judgment from regional advisory groups. We felt
these additional considerations were necessary because some taxa occurred with greater
frequency in warm- or cold-water habitats but were not present exclusively in one or the other.
For example, some taxa initially designated as cold-water taxa also were present at sites that had
the hottest recorded water temperatures. During the refinement process, we removed these taxa
from the cold-water list.
For the Ohio data set, MBI used the same general procedures described in Yuan (2006)
when making weighted-average calculations to derive optima and tolerance values (which the
author termed weighted stressor values [WSVs]). However, there were some differences in the
2-9
-------
data they used, how they prepared their data sets for analysis and how they ranked taxa. Instead
of instantaneous water temperature measurements, MBI's calculations were based on maximum
temperature recorded from summer-fall grab samples collected during the same period within
which the biological data were collected. Before running the analyses, MBI divided the data into
different stream size categories—headwater (drainage area <20 mi2) and wadeable (drainage area
>20 to 300 mi2)—and analyzed these data sets separately. Taxa rankings, which they termed
"Taxa Indicator Values," were derived using the methodology of Meador and Carlisle (2007) and
were based on an ordinal scale of 1 (most sensitive) to 10 (most tolerant). MBI did not formally
designate lists of cold and warm-water taxa but did note which taxa occurred at the extremes of
the distributions.
In addition to estimating thermal optima and tolerance values, we also examined the
relationship between these values and organic enrichment tolerance values for each state.
Overlap between these sensitivities means that it will be difficult to tease out whether the thermal
indicator taxa are responding to changes associated with warming temperatures or whether they
are responding to other stressors, such as enrichment.
2.2.2. Hydrologic Indicators
We attempted to develop lists of candidate taxa in each state that could potentially serve
as indicators of hydrologic change. The types of analyses that were conducted for each state
varied depending on the amount and type of hydrologic data that were available. For the Maine,
North Carolina, and Utah data sets, we used a geographic information system (GIS) to associate
biological sampling sites with USGS flow gages. Sites and gages were considered to be matches
if they were located on the same stream reach and were within 500 m of one another. For the
sites that had gages, all available hydrologic data were downloaded from the USGS real-time
flow data Web site. Indicators of Hydrologic Alteration (MA) software (Version 7.0.4.0, TNC,
2007) was then used to calculate a suite of nonparametric IHA parameters for each site (see
Table 2-4). The Richards-Baker Flashiness Index (RBI, Baker et al., 2004), which uses flow
data to quantify the frequency and rapidity of short-term changes in stream flow, was also
calculated for each site. IHA and RBI data were then paired with biological data from each site.
In general, these data sets had limited sample sizes, but if sufficient data existed, we used
2-10
-------
weighted averaging to calculate taxa optima and tolerance values for hydrologic variables that
showed the strongest relationships with the biological data.
Table 2-4. Summary of IHA parameters used in the analyses. High flow
-th
events refer to flows above the 75 percentile of all flows. Low flow events
th
refer to flows less than or equal to the 50 percentile of all flows. Extreme
,th
low flow events refer to flows less than the 10 percentile of all low flows
Annual IHA parameters
Monthly
1-day min
3 -day min
1-day max
3 -day max
Date min
Date max
Lo pulse #
Lo pulse L
Hi pulse #
Hi pulse L
Description
Median discharge (cfs)
Annual minima, 1-day mean (cfs)
Annual minima, 3 -day means (cfs)
Annual maxima, 1-day mean (cfs)
Annual maxima, 3 -day means (cfs)
Julian date of each annual 1-day minimum
Julian date of each annual 1-day maximum
Number of low pulses within each water year
Median duration of low pulses (days)
Number of high pulses within each water year
Median duration of high pulses (days)
Environmental flow components
Xlowl peak
Xlowl dur
Xlowl time
Xlowl freq
Highl peak
Highl dur
Highl time
Highl freq
Baseflow index
Number of reversals
Minimum ('peak') flow (cfs) during extreme low flow event (within each year)
Duration of extreme low flow event (days)
Julian date of peak flow
Frequency of extreme low flows during water year
Maximum ('peak') flow (cfs) during extreme high flow event (within each year)
Duration of extreme high flow event (days)
Julian date of peak flow
Frequency of extreme high flows during water year
7-day minimum flow/mean flow for year
Number of hydrological reversals
For the Ohio data set, instead of using IHA and RBI data, MBI calculated
weighted-average estimates based on a subset of habitat measures from the Qualitative Habitat
Evaluation Index (QHEI), which is a visual assessment of substrate, cover, channel, riparian,
pools, riffle, and stream gradient (Rankin, 1995, 1989). Since its inception, the QHEI has been
collected by trained professionals at Ohio EPA. Recent signal/noise ratio analyses of variation
2-11
-------
from sites with multiple QHEI values indicate the index is precise, and the subcomponents are
moderately precise to precise (Miltner and Rankin, 2009). The subset of QHEI attributes that
MBI analyzed (which they termed the Hydro-QHEI), are responsive either directly (current
speed components) or indirectly (stream depth measures) to alterations of the flow regime. As
with the thermal preference calculations, MBI calculated these weighted average estimates
separately for headwater (drainage area <20 mi2) and wadeable streams (drainage area >20 to
300 mi2).
In addition to the weighted averaging, there were sufficient data in the North Carolina
and Utah data sets to further examine associations between taxonomic data and hydrologic
variables using NMDS. We performed the NMDS ordinations to evaluate which IHA
parameters had the strongest influence on taxonomic composition. We overlaid grouping
variables such as season and ecoregion to determine how much (if any) influence these variables
had on the biological assemblage. For the Utah data set, we had sufficient data to also run a
Canonical Correspondence Analysis (CCA). In Maine, we lacked sufficient data to run these
types of analyses. However, we were able to run correlations analyses to look for associations
between biological data and IHA parameters at one site that had over 20 years of data. We were
also able to do this type of analysis at seven sites in Utah.
2.2.3. Traits-Based Indicators in Future Scenarios
For the Maine, North Carolina, and Utah data sets, we conducted exploratory exercises to
develop lists of taxa that may be most and least sensitive to projected changes in temperature and
streamflow based on combinations of traits. We used relevant literature and best professional
judgment to develop lists of traits modalities likely to be "functionally" linked to projected
changes in temperature and streamflow. These included traits such as voltinism, adult ability to
exit, ability to survive desiccation, dispersal ability, adult flying strength, occurrence in drift,
swimming ability, armoring, shape, respiration, size at maturity, habit, functional feeding group,
and thermal preference.
When assessing sensitivity to future climatic changes, we focused on a generalized
scenario in which temperatures are increasing, and flows are decreasing during the low flow
periods when state biomonitoring programs typically collect their samples. These low flow
conditions can be stressful to organisms due to loss of habitat, limited food resources, and altered
2-12
-------
water chemistry. We acknowledge that using this type of generalized scenario is an
oversimplification, and that regions may experience both extreme low flow and high flow events
in a given year. A common theme across potential future scenarios is that organisms are likely to
be exposed to more extreme and unpredictable conditions. We kept this in mind when deciding
which traits and trait modalities to consider when developing the lists of candidate indicator taxa,
and when assessing whether trait modalities were favorable or unfavorable in the face of
changing climatic conditions. Table 2-5 contains a list of the traits and trait modalities that we
used. Taxa that had the most number of favorable trait modalities were placed on the least
sensitive list, while those with the most number of unfavorable trait modalities were placed on
the most sensitive list.
Table 2-5. List of traits and trait modalities that were considered when
developing lists of traits-based indicator taxa for future conditions of
warming temperatures and lower flows. The list was developed based on
relevant literature and best professional judgment and consists of trait
modalities likely to be "functionally" linked to the changes in temperature
and streamflow
Traits
Voltinism
Adult ability to exit
Ability to survive desiccation
Dispersal ability (adult)
Adult flying strength
Swimming ability
Armoring
Occurrence in drift
Respiration
Size at maturity
Rheophily
Habit (primary)
Functional feeding group
(primary)
Thermal preference
Favorable
Bi- or multivoltine
(>1 generation/yr)
Present
Present
High
Strong
Strong
Good, heavily sclerotized
Abundant, common
Plastron or spiracle (aerial)
Small
Depositional
Skater, swimmer
Collector-gatherer, predator
Warm
Unfavorable
Semivoltine (<1 generation/yr)
Absent
Absent
Low
Weak
None
None
Rare
Tegument
Large
Erosional
Clinger
Scrapers, collector-filterer
Cold
2-13
-------
We acknowledge that there are many limitations associated with this methodology.
There are many factors other than temperature and hydrological variability that control the
distribution of species in lotic environments. Also, all traits do not have the same importance
and influence on adaptation to particular environmental conditions, so focusing on just the
number of favorable or unfavorable traits is an oversimplification and will not necessarily define
a taxon's ability to adapt to climate change. Moreover, there are phylogenetic constraints to the
combination of traits (and number of "favorable" traits) that could be found in a given tax on, and
it is possible that different combinations of traits (including different numbers of traits) can
provide similar "protection," just via a different strategy. These issues are largely problematic
for the field of trait-based ecology in general. Knowing this, we included results from these
exploratory analyses as a first step towards developing more robust lists of traits-based indicator
taxa. In the future, as more data become available and we learn more about which traits are in
fact advantageous or not in the face of changing temperatures and/or hydrology, these lists
should be refined.
2.3. LEAST-DISTURBED LONG-TERM BIOLOGICAL MONITORING SITES
We focused primarily on analyses of least-disturbed sites in each state so that trends in
biological data were as free from confounding nonclimatic factors as possible. We relied upon
guidance from the respective state agencies when selecting least-disturbed sites. Of the
four states evaluated, only Ohio has a formal statewide long-term monitoring network for
least-disturbed sites, and MBI focused their analyses on this network of sites. The Maine, North
Carolina, and Utah data sets were better suited for analyses of individual least-disturbed sites that
had the longest-term biological data. In these states, we performed exploratory analyses to
evaluate whether least-disturbed sites could be grouped together to create longer term data sets,
but site-specific differences were evident within these grouped data sets, so we focused on
individual sites. At some of these sites, anthropogenic influences are higher than desired (i.e.,
>5% urban or >10% agricultural within a 1-km buffer), but the data were analyzed anyway
because they represent the best-available long-term data in each state data set.
2-14
-------
2.4. EVIDENCE OF TRENDS AT LEAST-DISTURBED LONG-TERM MONITORING
SITES
2.4.1. Temporal Trends in Climatic and Biological Variables
We examined year-to-year variability in climatic (temperature, streamflow, precipitation)
and biological variables at least disturbed biological sampling sites with the longest-term
biological data. In Ohio, MBI looked at the amount and direction of change in state
bioassessment ratings (based on ICI and Index of Biotic Integrity [IBI] scores) at a group of
about 300 least disturbed "reference" sites that were sampled at 10-year intervals. Scores from
the initial sampling period (1980-1989) were compared to data from resampling periods in
1990-1999 and 2000-2006.
The Maine, North Carolina, and Utah data sets were better suited for analyses of
individual sites. We focused our analyses on least-disturbed sites that had the longest-term
biological data. When evaluating long-term temperature trends at these sites, we lacked
sufficient water temperature data, so we used air temperature as a surrogate. While water
temperature data are obviously preferable, air temperatures can closely track water temperatures
if there are no large effects from evaporative cooling, warm-water additions, or groundwater
damping (Caissie, 2006). Stephan and Preudhomme (1993) estimated a linear relationship
(factor of 0.86 in °C) between weekly average water and air temperatures for 11 streams in the
Mississippi River Basin. While a similar linear relationship has been applied by others (e.g.,
Pilgrim et al., 1998; Eaton and Scheller, 1996), Mohseni et al. (2003) suggest the relationship
between air and water temperatures is better explained by an S-curve, such that at higher air
temperatures, stream temperature increases level off due to evaporative cooling.
For each of the selected sites, we gathered daily observed maximum and minimum air
temperature data for the full period of record from the nearest active weather reporting station (or
inactive station that had data for the biological period of record). These data were obtained for
all three states from the Utah Climate Center Web site (http://climate.usurf.usu.edu/products/
data.php). First, to screen the data, we removed missing values (recorded as 999s) and excluded
data from years for which there were 2 or more months of missing data and/or fewer than
200 total measurements. Next, we averaged maximum and minimum air temperature values to
obtain daily mean annual air temperature. We then calculated annual averages and plotted these
data against year. We fitted these data with a linear trend line and calculated r and ^-values to
test for significance. At each of the selected sites, we also determined which month was hottest
2-15
-------
(on average), based on mean monthly maximum air temperatures, and calculated the mean
maximum air temperature for the hottest month to get a sense of how much thermal stress the
organisms may have been exposed to during a given year.
In addition to the temperature data, we gathered flow data by matching the biological
sampling sites with the closest USGS gages. We performed desktop screening to assess whether
the flow data from the gages were representative of flow conditions at the biological sampling
site. To supplement the flow data (which we lacked for some sites), we gathered daily observed
precipitation data from the closest weather reporting stations, once again using the Utah Climate
Center Web site (http://climate.usurf.usu.edu/products/data.php). First, we removed the missing
values. Next we summed the daily values to obtain annual precipitation for the full period of
record. We plotted flow and precipitation data against year, fit the data with linear trend lines,
and calculated r2 and ^-values. At each of the selected sites, we also looked at hydrographs to
determine when the lowest flows typically occurred at each site. The intent was to get a sense of
how much stress the organisms may have been exposed to during a given year due to extremes in
flow conditions. We found that the low-flow periods generally corresponded with the index
periods that state biomonitoring programs use for collecting biological samples. Therefore, we
calculated mean monthly flow and precipitation values for each state's index period and
evaluated trends in these data over time.
In addition to analyzing the observed data from the nearest weather stations, we used a
GIS to obtain PRISM annual air temperature and precipitation data from 1974 to 2006 for the
selected sites in Maine, North Carolina, and Utah. We selected this time period because it
corresponds to the minimum and maximum years for which biological data were available in the
state biomonitoring databases. Where there were periods of overlap, the modeled PRISM data
were compared to the observed weather station data. Although values sometimes differed,
especially when biological sampling sites and weather stations were located in areas of differing
topography (i.e., at different elevations), there was generally good correspondence in patterns.
At each of the selected sites, we looked for temporal trends in the biological data. More
specifically, we analyzed year-to-year variability in state bioassessment scores and the following
metric values: number of EPT taxa, HBI, and thermal preference metrics. The thermal
preference metrics are based on the lists of cold and warm-water taxa that were developed for
each state (as described in Section 2.2.1). When we calculated the biological variables, if
2-16
-------
multiple samples were collected in a given year, values were averaged to derive one value per
year. We also accounted for seasonal variation by limiting samples to those collected during a
single season or index period. We plotted the biological data against year and presented the data
in a way that allows for comparison with trends in temperature and streamflow.
For each site, we reported the range of temperature, precipitation, and flow values that
occurred during the period of biological record. In addition, we reported ranges of water
chemistry values and/or habitat measures (depending on what type of data were available for
each state) for the period of biological record. We did this to evaluate whether trends in the
biological data may have been influenced by potential confounding factors that were not related
to climate.
2.4.2. Associations Between Biological Variables and Climatic Variables
In the Maine, North Carolina, and Utah data sets, we performed correlation analyses on
data from least-disturbed sites that had the longest-term biological data. We used Statistica
software (Version 10, Copyright StatSoft, Inc., 1984-2011) to run Kendall tau nonparametric
correlation analyses on state bioassessment scores, selected biological metrics, year, temperature,
flow, and precipitation variables. When deciding which biological metrics to evaluate, we
considered the list of commonly used metrics in Barbour et al. (1999) and also looked at which
metrics are most commonly used by state biomonitoring programs. We based our selection of
thermal and hydrologic indicator metrics on literature searches and best professional judgment.
Table 2-6 shows the biological metrics that were evaluated at each site. When reporting results,
we noted which biological variables had strong associations (r > 0.5) with year or climatic
parameters. We also noted whether the direction of these relationships was in keeping with
expectations, as described in Table 2-5.
2.4.3. Groupings Based on Climatic Variables
In the Maine, North Carolina, and Utah data sets, we grouped data based on extremes in
climate variables, using these groupings as proxies for future climate conditions. These analyses
were done at the least-disturbed sites that had the longest-term biological data. For the
temperature analyses, we partitioned data into years characterized by hotter (>67t percentile of
the temperature distribution during years of biological collections), colder (<33r percentile of
2-17
-------
Table 2-6. List of biological metrics/traits evaluated at each site considering
commonly used metrics summarized in Barbour et al. (1999) and those used
by state biomonitoring programs
Biological metric/trait"
Predicted response to...
Source
...increasing stress
Total number of taxa (richness)
Number of EPT taxa (Ephemeroptera
[mayflies], Plecoptera [stoneflies], and
Trichoptera [caddisflies])
Number of Ephemeroptera (mayfly) taxa
Number of Plecoptera (stonefly) taxa
Number of Trichoptera (caddisfly) taxa
Number of intolerant taxa (sensitive to
perturbation)
Percentage EPT individuals
Percentage Ephemeroptera individuals
Percentage dominant taxon
Percentage tolerant individuals
Hilsenhoff Biotic Index (tolerance toward
organic enrichment, Hilsenhoff, 1987)
Shannon-Wiener Diversity Index
Percentage noninsect individuals
Decrease
Decrease
Decrease
Decrease
Decrease
Decrease
Decrease
Decrease
Increase
Increase
Increase
Decrease
Increase
Table 7-1 in Barbour et al., 1999
(compiled from DeShon, 1995,
Barbour et al., 1996, Fore et al.,
1996, Smith and Voshell, 1997)
Commonly used, based on
inventory of multimetric indices
used by state biomonitoring
programs
...warming temperatures
Number of cold-water taxa
Percentage cold-water individuals
Number of warm-water taxa
Percentage warm-water individuals
Decrease
Decrease
Increase
Increase
Cold- and warm-water taxa
derived from weighted-average
modeling or related approaches
performed on each state data set,
with refinements literature
searches, and best professional
judgment of regional taxonomic
experts
...changing streamflow conditions
Collector filterer
Collector gatherer
Decrease during low flow
conditions
Increase during slow
velocity conditions
Bogan and Lytle, 2007
Heino, 2009
2-18
-------
Table 2-6. List of biological metrics evaluated at each site considering
commonly used metrics summarized in Barbour et al. (1999) and those used
by state biomonitoring programs (cont.)
Scraper/herbivore
Predator
Swimmer
Rheophily — depositional
Rheophily — erosional
Odonata/Coleoptera/Hemiptera (OCH)
Increase during conditions
of stable flow and habitat
availability; decrease
during drought conditions
Increase during low flow
conditions
Comprise higher
proportion of assemblage
during drier, harsher
climatic conditions
Increase during low
flow/slow velocity
conditions
Increase during high
flow/fast velocity
conditions
Expected to be more
prevalent during summer,
low flow (more pool-like)
periods
Fenoglio et al., 2007, Griswold
et al., 2008, Diaz et al., 2008
Bogan and Lytle, 2007
Beche et al., 2006, Bonada et al.,
2007a, Diaz et al., 2008
Best professional judgment
Bonada et al., 2007b
aTrait assignments were based primarily on the Poff et al. (2006b) traits matrix and Vieira et al. (2006).
temperature), and normal (33rd to 67th percentile) temperatures based on PRISM mean annual
average air temperatures. When flow data were available, a similar partitioning of high, low, and
normal flow years was applied based on mean annual flow. When flow data were not available,
we based the partitioning on PRISM mean annual precipitation. Gaps in the biological data
prevented us from designating groupings based on the full range of temperature, flow, and/or
precipitation values, which would have been preferable. For the temperature analyses,
temperatures in the hottest-year samples were generally 1-2°C higher than for the coldest-year
samples, a difference that corresponds well with future climatic projections for midcentury.
After samples were grouped based on these environmental variables, Statistica software
(Version 10, Copyright StatSoft, Inc., 1984-2011) was used to run one-way analysis of variance
(ANOVA) tests to evaluate whether significant differences existed among state bioassessment
scores, number of total taxa, number of EPT taxa, and thermal preference metrics from the
2-19
-------
different year groupings. Differences were considered significant ifp < 0.05 based on the Tukey
honest significant difference test for unequal sample size (n) (Spjotvoll/Stoline).
In addition to the ANOVAs, at the three sites that had the longest-term biological data
(15 or more years), we used PCOrd® software (Version 4.41, McCune and Mefford, 1999) to
perform NMDS ordinations. The intent was to evaluate differences in taxonomic composition
among samples collected during the different year groupings, and to determine which
environmental variables explained the greatest amount of variation on each of the ordination
axes. We examined the following environmental variables: PRISM mean annual air temperature
and precipitation, PRISM mean annual air temperature and precipitation from the previous year
(lag effects), and the absolute difference between the PRISM mean annual air temperature and
precipitation from the sampling year and the previous year (year-to-year variability).
2.5. SENSITIVITY OF BENTHIC MACROINVERTEBRATES TO TEMPERATURE
In the Maine, North Carolina, and Utah data sets, we examined the spatial distributions of
cold and warm-water taxa to gain insights into which areas in each state are likely to be most and
least sensitive to projected changes in temperature and stream flow. We based our analyses on
the premise that streams with greater numbers and abundances of cold-water taxa will be more
sensitive to warming temperatures and decreasing precipitation patterns. We performed one-way
ANOVAs to determine whether significant differences existed in the distributions of cold and
warm-water taxa across different ecoregions and stream size categories (based on Strahler order).
Although our premise makes intuitive sense, it may be that cool water taxa in transitional areas,
where species are expected to be closer to their tolerance limits, will be most sensitive and will
experience the greatest amount of change.
In the Ohio data set, MBI examined the amount and direction of change in the
bioassessment scores across different site types (stratified by stream size), habitat categories
(modified warmwater [MWH], warmwater [WWH], and exceptional warmwater [EWH]), and
ecoregions, and looked for general concordances between intolerant and sensitive species as
categorized for the IB I and ICI and species sensitive to temperature and habitat features
indicative of altered flow conditions. Trends they documented were most attributable to reduced
pollution from point sources, mostly due to municipal wastewater treatment plant upgrades after
1988 (Yoder et al., 2005), not to climate-related changes. Sensitivities in Ohio may be best
2-20
-------
monitored by tracking changes in the distributions of candidate indicator taxa that were
identified through MBI's weighted-averaging analyses and by carefully monitoring the habitats
that those taxa occur in.
2.6. IMPLICATIONS FOR STATE BIOMONITORING PROGRAMS
Our discussions of implications of climate change on state biomonitoring programs vary
depending on the type of data available for each state, and also on how each state assesses the
biological integrity of its streams. Utah uses a River In Vertebrate Prediction and Classification
System (RIVPACS) model. Maine uses linear discriminant models with over 20 model input
metrics to classify station condition. North Carolina typically calculates bioclassification scores
based on two metrics: EPT richness and the North Carolina Biotic Index (NCBI), while Ohio
uses multimetric indices for fish (IBI) and macroinvertebrates (ICI) to rate streams. For each
state, we synthesized results from the analyses of existing temperature, flow, precipitation, and
biological data. For Utah and North Carolina, in addition to analyzing existing data, we
performed exploratory analyses by manipulating data to gain further insights into how future
projected climatic changes might impact each state's assessment methods. In Utah, this involved
manipulating the climate-related predictor variables in the Utah RIVPACS model in a way that
would simulate future projected changes. We assessed how much this might affect Utah's
bioassessment scores. In North Carolina, we looked at how the loss of cold-water taxa at least-
disturbed sites in the Piedmont and Blue Ridge Mountain ecoregions would affect
bioclassification scores. Table 2-7 provides a summary of the types of analyses that were
conducted in each state.
2-21
-------
Table 2-7. Summary of types of analyses that were conducted on the Maine,
North Carolina, Utah, and Ohio data sets
Analyses
Maine
North
Carolina
Utah
Ohio
Derivation of indicators
Thermal
Weighted-average modeling or related approaches
Hydrologic
Weighted-average modeling or related approaches
NMDS
CCA
Correlation analyses
Traits-based
X
1 site
X
X
X
X
X
X
X
X
7 sites
X
X
X
Temporal trends
Temperature, flow, and/or precipitation variables
State bioassessment scores
Biological metrics (individual sites)
3 sites
3 sites
3 sites
5 sites
5 sites
5 sites
4 sites
4 sites
4 sites
statewide
Correlation analyses
Biological variables vs. temperature, flow, and/or
precipitation variables
3 sites
1 site
4 sites
Year groupings (hot/cold/normal, etc.)
ANOVA
NMDS
Future exploratory analyses
3 sites
1 site
1 site
X
4 sites
2 sites
X
2-22
-------
3. UTAH
3.1. EXPOSURES
3.1.1. Regional Projections for the Southwestern United States
In coming years, the landscape of the Southwestern United States will be impacted by
increases in temperature, drought, wildfire, and invasive species, as well as by an increased
frequency and altered timing of flooding (Karl et al., 2009). Temperature increases in the
southwest are expected to be greater than the global average (Gutzler et al., 2006), though
projections are not substantially higher than for other regions of the United States. Projections
using different models and emissions assumptions show seasonal and annual temperature
increases of 3-4°C per century (Christensen and Lettenmeier, 2006; Gutzler and Robbins, 2011)
(see Table 3-1).
Table 3-1. Projections for temperature and precipitation changes in the
Southwest to 2100
Temperature
change
3-4°C
3-4°C
Precipitation change
-2 to +1%
Decrease
No change to slight
increase
Change in precipitation
frequency
N/A
N/A
N/A
N/A
Citation
Gutzler et al., 2006
Christensen and Lettenmeier,
2006
Schoofetal., 2010
Gutzler and Robbins, 201 1
Climate model projections for precipitation show small changes, ranging from slight
decreases to slight increases for the southwestern United States among the numerous GCM
model outputs used to generate the ensemble projections (National Center for Atmospheric
Research [NCAR] Web site, http://rcpm.ucar.edu; Christensen and Lettenmeier, 2006; Schoof
et al., 2010; Gutzler and Robbins, 2011) (see Table 3-1). Many ensemble modeling results show
small decreases in summer precipitation but small increases in winter precipitation (Christensen
3-1
-------
and Lettenmeier, 2006), though some show the reverse (e.g., Gutzler and Robbins, 2011).
Schoof et al. (2010) projected an increase in the intensity of wintertime precipitation.
The impacts of projected changes in temperature and precipitation on stream hydrologic
conditions are of particular interest to the assessment of effects on freshwater systems. Changes
have already been observed (as well as modeled) in the magnitude, timing, frequency, and
duration of stream flow events, as well as in timing and amount of snow melt (e.g., Hayhoe et al.,
2007). The IPCC (2007a) projects average annual runoff to decrease by 10-30% over
midlatitude regions, including the southwestern United States. Average runoff is projected to
decrease in the Colorado River Basin by 8-11% over a century under the Bl and A2 scenarios,
respectively (Christensen and Lettenmeier, 2006). Kurd et al. (2004) modeled a greater range of
future runoff changes for the Colorado River, ranging from a 38% decrease to a 24% increase
comparing baseline and nine different combinations of future changes in temperature (+1.5, 2.5,
and 5°C) and precipitation (-10, +5, and +1%). The biggest increase in runoff corresponded
with the biggest percentage increase in precipitation combined with the smallest increase in
temperature, while the modeled decreases in average annual runoff were associated with the
largest modeled decrease in annual precipitation for all of the modeled temperature increases
(Kurd et al., 2004).
In western/southwestern snow-pack dominated regions, the combination of warming
temperatures, a shift toward less winter precipitation falling as snow, and snow-melt occurring
earlier will change peak runoff from spring to late-winter/early spring (Barnett et al., 2005;
Clow, 2010). Typical projections are for peak runoff to shift from about 2 weeks up to 1 month
earlier by the end of the century (Dettinger et al., 2004; Hayhoe et al., 2007). Stewart et al.
(2005) found evidence for shifts to earlier timing of snow melt and runoff averaging 1-4 weeks
based on evaluation of data from 1948 to 2002 for several montane catchments in the western
United States. In evaluations of snow-pack dominated streams in Colorado, Clow (2010) found
that snowmelt and the timing of peak stream runoff has shifted 2-3 weeks earlier over the
29 years from 1978-2007 (median change 4.8 days per decade). This was accompanied by a
decline in April and maximum snow-water equivalent (SWE) of 4.1 and 3.6 cm per decade,
respectively. Decreases in total snow pack also contribute to the earlier onset of snow melt and
the corresponding earlier spring runoff in western and southwestern high elevation systems.
5-2
-------
In addition, increasing air temperatures are projected to increase the likelihood of
winter/early spring precipitation as rain instead of snow. Rain-on-snow events, as well as rain
during winter months when cold to frozen ground conditions can decrease the infiltration of rain,
can increase the likelihood of severe episodic flooding (IPCC, 2007a).
An additional projection for the southwestern United States is for increased aridity,
increased severity of summer droughts, associated decreased stream discharge, and extended
periods of summer low flows (Gutzler et al., 2006; Gutzler and Robbins, 2011; Seager and
Vecchi, 2010; Seager et al., 2007). Most models for the region project increases in
evapotranspiration—due to increased temperature rather than changes in summer
precipitation—leading to a net decrease in soil moisture and a greater likelihood of late-summer
drought (NAST, 2001). This scenario of decreasing soil moisture, increasing evapotranspiration,
and higher summer temperatures leading to increasing summer dry periods was specifically
modeled in New Mexico (Gutzler et al., 2006), with expectations for decreasing summer stream
discharge. Gutzler and Robbins (2011) also projected increases in the severity of droughts over
the next century in the southwest, based on modeling of the Palmer drought index. But unlike
historic droughts, projected increases in future temperature are also expected to inhibit natural
recovery from severe droughts. Seager et al. (2007) projects more arid conditions and more
persistent drought for New Mexico, beginning in the late 20* and early 21st centuries. Seager
and Vecchi (2010) suggest this pattern is driven by reduced winter precipitation, and will reach
the amplitude of historic droughts by midcentury. Though with high uncertainty, they estimate
this pattern will be augmented by natural multidecadal oscillations of the Pacific and Atlantic,
which are currently in phases that augment drought condition in the southwestern United States.
3.1.2. Historic Climate Trends and Climate Change Projections for Utah
Utah has a semiarid to arid climate. Its diverse landscape consists of a mix of mountains,
valleys, and low lying areas. The Wasatch and Uinta Mountains, which run through the central
part of the state, are high, precipitous mountains with narrow crests and valleys flanked in some
areas by dissected plateaus and open high mountains (U.S. EPA, 2002). The Colorado Plateaus
ecoregion, which comprises much of the eastern and southern part of the state, has a mix of large
low-lying areas and rugged tableland topography with sharp changes in local relief. The Central
Basin and Range ecoregion, which makes up much of western Utah, consists of dry basins,
3-3
-------
scattered high and low mountains, and salt flats (U.S. EPA, 2002). Temperature and
precipitation patterns are influenced by topography, as shown in Figure 3-1, with the Wasatch
and Uinta Mountains having the coolest mean annual temperatures (see Figure 3-1 A) and the
greatest amount of annual precipitation (see Figure 3-1B).
There is a great deal of year-to-year variability in temperature and precipitation patterns,
but overall, temperatures in Utah have been increasing over the last century and are projected to
continue to increase. A historic trend analysis of Utah PRISM data shows that mean annual air
temperature has increased at a rate of 0.01°C/year (p < 0.01) from 1901-2000 (see Figure 3-2).
This trend has been steeper in more recent decades, with a change rate of 0.04°C/year from
1971-2000 (see Table 3-2). The long-term rate, netting a change of almost 1°C over century, is
lower than the future model projections for changes of 2.7-4.4°C over the coming century (see
Section 3.1.1 above). However, the more recent rate of increase estimate for the 1971-2000
period (approximately 4°C per century) is quite consistent with future projected rates. Seasonal
trends over the last century have been similar to the annual change rate of 0.01 °C/year (see
Table 3-2 and Figure 3-3). In recent decades, steeper trends (0.05-0.07 °C/year) have been
occurring during the winter and spring (see Table 3-2). Table 3-3 summarizes future projections
for mid- and late-century for high (A2) and low (Bl) emissions scenarios. Based on an ensemble
average across 15 models, mean annual air temperatures are projected to increase by up to 2.9°C
by midcentury and up to 4.8°C by the end of the century compared to a historic time period
(1961-1990). These future projections are consistent with the literature values summarized in
Section 3.1.1 above. The greatest increases are proj ected to occur during the summer and fall
(see Table 3-3).
Precipitation patterns in Utah have been highly variable. Overall, mean annual
precipitation has increased at a rate of 0.347 mm/year from 1901-2000 (see Figure 3-4 and
Table 3-4). In more recent decades, this rate has increased to 1.28 mm/year (see Table 3-4).
However, due to the high degree of year-to-year variability, none of the historic trends in
precipitation are significant (p > 0.05). The same holds true with seasonal change rates; the
amount and direction of change vary depending on season and time period, and no trends are
significant (see Table 3-4 and Figure 3-5). Table 3-5 summarizes future projections for mid- and
late-century for high (A2) and low (Bl) emissions scenarios. The future projections are highly
variable across models and emissions scenarios. Under the high emissions scenario, the
3-4
-------
B
Mean Temp. (C)
25 C
20
15
10
'2000
Precip (mm)
I 2500 mm
'
1500
1000
500
Figure 3-1. Utah's temperature and precipitation patterns. (A) Mean annual air temperature (°C) from 1971-2000
for the state of Utah; (B) Mean annual precipitation (mm) 1971-2000 for the state of Utah. Map produced using the
Climate Wizard Web site (http://www.climatewizard.org/). Base climate data from the PRISM Group, Oregon State
University, http://www.prismclimate.org.
-------
Table 3-2. Change rates in Utah PRISM mean annual air temperature
compared across two time periods: 1971-2000 versus 1901-2000. Entries in
bold text are significant (p < 0.05). Data were derived from the Climate
Wizard Web site (http://www.climatewizard.org/). Base climate data came
from the PRISM Group, Oregon State University,
http://www.prismclimate.org
Time period
1901-2000
1971-2000
Air temperature (°C/yr)
Annual
0.01
0.04
DJF
0.01
0.07
MAM
0.01
0.05
JJA
0.01
0.01
SON
0.01
0.02
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, and
August; SON = September, October, and November.
O
Q-
E o
ro co
o
CO
1900 1920 1940 1960
Year
1980
2000
Figure 3-2. Trends in annual mean air temperature in Utah from 1901-2000.
Change rate = 0.01°C/year,/>-value < 0.01. Figure produced using Climate
Wizard Web site (http://www.climatewizard.org/). Base climate data from the
PRISM Group, Oregon State University, http://www.prismclimate.org.
3-6
-------
1900 1920 1940 I960 'is: 2000
1900 1920 1940 1960 19SO 2000
D
Si
I
1900 1920 1940 1960 1980 2000
Y««r
•;<:: 1920 iwo i960
Y«ar
1980 2000
Figure 3-3. Trends in seasonal mean air temperature in Utah from 1901-2000. (A) DJF = December, January,
and February, change rate = 0.014°C/year,/>-value = 0.01; (B) MAM = March, April, and May, change
rate = 0.009°C/year,/?-value = 0.01; (C) JJA = June, July, and August, change rate = 0.009°C/year,/?-value < 0.01;
(D) SON = September, October, and November, change rate = 0.007°C/year,/?-value = 0.04. Figure produced using
Climate Wizard Web site (http://www.climatewizard.org/). Base climate data from the PRISM Group, Oregon State
University, http://www.prismclimate.org.
-------
oo
Table 3-3. Projected departure from historic (1961-1990) trends in annual and seasonal air temperature (°C) in
Utah for mid- (2040-2069) and late-century (2070-2099) for high and low emissions scenarios. Values represent
the minimum, average, maximum, and standard deviations from 15 different climate models. Data were derived
from the Climate Wizard Web site (http://www.climatewizard.org/)
Model
Ensemble low
Ensemble average
Ensemble high
SD
A2 (high) emissions scenario
Annual
1.7
2.9
4.1
0.7
DJF
1.4
2.6
4.1
0.9
MAM
1.2
2.6
4.6
1.0
JJA
1.8
3.3
4.4
0.8
SON
2.0
3.3
4.2
0.7
Bl (low) emissions scenario
Annual
1.0
2.3
3.5
0.7
DJF
0.5
2.1
3.7
0.9
MAM
0.7
2.1
3.9
0.8
JJA
1.3
2.6
3.6
0.7
SON
1.3
2.3
3.0
0.6
Late-Century (2070-2099) vs. historic (1961-1990)
Ensemble low
Ensemble average
Ensemble high
SD
3.0
4.8
6.7
1.1
2.3
4.3
7.1
1.3
2.1
4.3
8.0
1.6
3.4
5.4
7.1
1.2
3.5
5.5
6.9
1.1
1.8
3.0
4.4
0.8
1.6
2.8
4.8
1.0
1.4
2.8
5.4
1.0
1.8
3.4
4.4
0.8
1.6
3.0
4.0
0.8
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, August and SON = September, October, and
November.
-------
Table 3-4. Change rates in Utah PRISM mean annual precipitation
compared across two time periods: 1971-2000 versus 1901-2000. No trends
are significant (p < 0.05). Data were derived from the Climate Wizard Web
site (http://www.climatewizard.org/). Base climate data came from the
PRISM Group, Oregon State University, http://www.prismclimate.org
Time period
1901-2000
1971-2000
Precipitation (mm/yr)
Annual
0.35
1.28
DJF
-0.05
0.61
MAM
0.11
0.27
JJA
0.11
0.69
SON
0.2
-0.14
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, and
August; SON = September, October, and November.
E
E
1900
1920
1940
1960
1980
2000
Year
Figure 3-4. Trends in annual mean precipitation in Utah from 1901-2000.
Change rate = 0.347 mm/year,p-va\ue = 0.15. Figure produced using Climate
Wizard Web site (http://www.climatewizard.org/). Base climate data from the
PRISM Group, Oregon State University, http://www.prismclimate.org.
3-9
-------
g -
1900 1920
1900 1920
1940 1960
tea,
I960 2000
Figure 3-5. Trends in seasonal mean precipitation in Utah from 1901-2000. (A) DJF = December, January, and
February, change rate = -0.05 mm/year, />-value = 0.62; (B) MAM = March, April, and May, change
rate = 0.11 mm/year,/>-value = 0.27; (C) JJA = June, July, and August, change rate = 0.11 mm/year,/>-value = 0.15;
(D) SON = September, October, and November, change rate = 0.20 mm/year, p-va\ue = 0.12. Figure produced using
Climate Wizard Web site (http://www.climatewizard.org/). Base climate data from the PRISM Group, Oregon State
University, http://www.prismclimate.org.
-------
Table 3-5. Projected departure from historic (1961-1990) trends in annual and seasonal precipitation (mm) in
Utah for mid- (2040-2069) and late-century (2070-2099) for high and low emissions scenarios. Values represent
the minimum, average, maximum, and standard deviations from 15 different climate models. Data were derived
from the Climate Wizard Web site (http://www.climatewizard.org/)
Midcentury (2040-2069) vs. historic (1961-1990)
Model
Ensemble low
Ensemble average
Ensemble high
SD
A2 (high) emissions scenario
Annual
-74.7
-2.7
50.6
33.4
DJF
-12.0
10.3
50.5
16.1
MAM
-48.9
-8.5
9.9
14.8
JJA
-19.4
-6.9
6.5
7.9
SON
-10.7
2.3
24.1
11.6
Bl (low) emissions scenario
Annual
-28.9
22.3
255.9
71.8
DJF
-42.9
0.8
38.0
18.5
MAM
-26.8
3.3
92.1
28.4
JJA
-16.1
16.8
211.6
58.9
SON
-9.2
4.2
35.2
11.4
Late-century (2070-2099) vs. historic (1961-1990)
Ensemble low
Ensemble average
Ensemble high
SD
-62.1
-5.8
37.0
30.3
-14.5
14.9
66.8
18.9
-55.1
-16.0
15.4
15.1
-27.5
-4.4
24.3
13.6
-23.0
2.8
49.9
17.6
-51.5
50.7
376.9
125.4
-54.5
1.6
36.9
24.2
-45.7
10.7
127.6
44.9
-19.3
32.8
261.3
84.8
-18.5
12.3
63.1
21.8
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, August and SON = September, October, and
November.
-------
ensemble average projects that mean annual precipitation will decrease by 2.7 mm by
midcentury and 5.8 mm by the end of the century compared to a historic time period
(1961-1990). Under the high emissions scenario, the greatest changes are projected to occur
during the winter and spring (see Table 3-5).
3.2. DATA INVENTORY AND PREPARATION
The Utah database contains data for 2,337 biological samples from 615 unique stations,
with sampling dates ranging from 1977 to 2005. Water chemistry data (nutrients, metals,
alkalinity, and turbidity) and in situ measurements are available for many of these sites. No
habitat data are available. Most sites have fewer than 5 years of data, but there are 30 sites that
have 10 or more years of data (see Table 3-6). Utah Department of Environmental Quality
(DEQ) considers four of these long-term sites to be in reference (highest quality) condition.
Utah DEQ's reference designations are based on a combination of a reference scoring sheet
(multiple lines of scoring) and independent ranking of sites from field crew/scientists. Only sites
that are consistently ranked as reference are included on the list. Figure 3-6 shows the spatial
distribution of biological sampling sites.
Table 3-6. Distribution of reference and total stations, categorized by
duration of sampling
Years sampled
Ito4
5 to 9
>10
Total
Utah
Reference
61
1
4
66
Total
482
41
26
549
When preparing the biological data for long-term trend analyses, genus-level OTUs were
generally found to be most appropriate for the Utah data set. However, a family-level OTU had
be to be used for Chironomidae, as subfamily- and/or genus-level identifications only occurred in
later years in the Utah data set. "Fixes" also had to be made to OTU assignments for
Ephemerella and Drunella due to changes in taxonomic systematics. Additionally, there was
some uncertainty as to the consistency of how abundance data were recorded over the years.
3-12
-------
Utah stations
Ref Stat, Num Yrs
Ref>20(N=3)
Ref 10-19 (N=1)
Ref 5-9 (N=2)
Ref2-4(N=6)
Ref 1 (N=54)
Other > 20 (N=27)
Other 10-19 (N=22)
Other 5-9 (N=77)
Other 2-4 (N=163)
Other 1 (N=194)
O
Ecoregion
LEVEL3_NAM
Central Basin and Range
| Colorado Plateaus
| Mojave Basin and Range
Northern Basin and Range
| Southern Rockies
| Wasatch and Uinta Mountains
I Wyoming Basin
A
Figure 3-6. Utah biomonitoring stations, coded by reference status and
duration of data.
3-13
-------
These questions related to whether recorded abundances were corrected for subsampling in the
laboratory, area sampled, and/or replication. These questions could not be fully resolved based
on institutional knowledge of Utah DEQ scientists or from extant database metadata or other
documentation. Because of this uncertainty, where possible, we based our calculations and
analyses on relative abundance data. If calculations required the use of abundance data, we
interpreted results with caution.
3.3. UTAH DEQ METHODS
For the period of analyses used in this report (prior to 2006), Utah DEQ collected samples from
riffle habitats using a Hess sampler. Starting in 2006, quantitative riffle habitat samples were
collected using the Environmental Monitoring and Assessment Program (EMAP) kick method;
therefore, any future long-term trend analyses would have to examine comparability between
these sampling methods. Samples are typically collected during a September/October index
period, but the Utah data set includes samples collected throughout the year. For most analyses,
only fall samples were used to minimize variation associated with seasonal differences in
taxonomic composition.
In recent years, Utah started using a RIVPACS model (fall samples) to assess wadeable
streams (Ostermiller, unpublished presentation titled "Development of a biological assessment
framework", Appendix B). The model was calibrated based on reference data collected from
1999-2005. The random forests method was used to select predictor variables that best
discriminated among the site groups (Breiman and Cutler, 2009). The model has 15 predictor
variables, 7 of which are related to climate (e.g., temperature, precipitation, freeze dates).
Samples are scored based on the ratio of observed to expected (O/E) assemblages (expected
assemblages are established based on reference site data). If a sample receives an O/E score of
>0.74, Utah DEQ considers the beneficial use of the waterbody to be fully supported.
3.4. INDICATORS
3.4.1. Thermal Preference
As described in Section 2, we used the guidelines of Yuan (2006) to calculate thermal
optima and tolerance values. For the Utah data set, we based our calculations on a subset of data
collected during the fall season (n = 572). These data, along with weighted-average inferences
3-14
-------
derived from Idaho (Brandt, 2001), California (Herbst and Silldorff, 2007), and Oregon (Yuan,
2006; Huff et al., 2008) data sets, were used to develop lists of cold- and warm-water taxa for the
Utah data set. These lists are the basis of the region-specific thermal-preference richness and
relative-abundance metrics used in some analyses.
The Utah cold-water taxa list is composed of 33 taxa, and the warm-water taxa list is
composed of 16 taxa. The relatively low number of warm-water taxa is partially a consequence
of the need to use a family-level OTU for Chironomidae. Tables 3-6 and 3-7, respectively, list
the cold- and warm-water taxa, along with abundance and distribution information1. Ten of the
cold-water taxa are Plecopterans, eight are Dipterans, seven are Trichopterans, and six are
Ephemeropterans (see Table 3-7). Five of the warm-water taxa are Trichopterans, three are
Coleopterans, and two are Dipterans and Ephemeropterans (see Table 3-8).
The most abundant cold-water taxa are two Ephemeropterans, Ephemerella and
Cinygmula, which comprise 1.85 and 1.03% of the total individuals, respectively. Of the
cold-water taxa, Chloroperlidae occurs at the highest percentage of sites (49%), followed by
two Ephemeropterans (Ephemerella and Cinygmula), which occur at 44 and 46% of the sites,
respectively. Asellidae and Leptohyphidae are the most abundant warm-water taxa, with overall
abundances of 3.12 and 1.42%. Among the warm-water taxa, Leptohyphidae occurs at the
highest percentage of sites (31%), followed by Coenagrionidae (18%) and Cheumatopsyche
(17%). Many of the taxa on the cold- and warm-water lists have low overall abundances (less
than 0.1%) and occur at less than 10% of the sites.
Most of the taxa on the cold-water list are intolerant to enrichment, while most of the
warm-water taxa are tolerant or have intermediate tolerance to enrichment (see Figure 3-7).
Because of this, it is difficult to tease out whether organisms are responding to changes
associated with warming temperatures or whether they are responding to other stressors, such as
enrichment.
3.4.2. Hydrologic Indicators
We attempted to develop a list of candidate taxa in Utah that could potentially serve as
indicators of hydrologic change. We were able to match USGS gage data with biological data
1 There are some noteworthy genera that were excluded from the Utah cold-water taxa list. These include Zapada,
Epeorus, Drunella, Brachycentrus, wARhyacophila. These taxa were excluded because of variations in thermal
preferences among species within these genera.
3-15
-------
from 43 sampling sites, and calculated IHA parameters and the RBI per the methods described in
Section 2.2.2. The data set, which included samples from both disturbed and least-disturbed
sites, had some limitations. It had a relatively small sample size, and some sites had many more
years of data than others (i.e., one site had 19 years of data, others had 1 year of data). Despite
3-16
-------
Table 3-7. List of Utah cold-water-temperature indicator taxa, sorted by order, family, then Final ID.
Distribution and abundance information is also included. Sum_Individuals = the total number of individuals
from that taxon in the Utah database; Pct_Abund = percentage of total individuals in the database composed of
that taxon; Num_Stations = number of stations in the database that the taxon occurred at;
Pct_Stations = percentage of stations in the database at which the taxon occurred
Order
Coleoptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Dorylaimida
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Plecoptera
Plecoptera
Family
Elmidae
Blephariceridae
Ceratopogonidae
Empididae
Empididae
Empididae
Psychodidae
Tipulidae
Tipulidae
Dorylaimidae
Ameletidae
Ephemerellidae
Heptageniidae
Heptageniidae
Heptageniidae
Heptageniidae
Capniidae
Chloroperlidae
Final ID
Heterlimnius
Bibiocephala
Bezzia
Chelifera
Oreogeton
Wiedemannia
Pericoma
Dicranota
Rhabdomastix
Nematoda
Ameletus
Ephemerella
Cinygma
Cinygmula
Ironodes
Rhithrogena
Capniidae
Chloroperlidae
Sum individs
16,463.0
2,257.0
109,267.1
94,014.1
228.5
458.0
145,582.7
35,439.2
8.0
141,425.3
13,157.6
859,335.8
606.2
479,866.5
551.6
198,501.8
113,578.8
203,579.9
Pct_abund
0.0
0.0
0.2
0.2
0.0
0.0
0.3
0.1
0.0
0.3
0.0
1.9
0.0
1.0
0.0
0.4
0.2
0.4
Num_stations
50.0
15.0
232.0
261.0
13.0
13.0
210.0
220.0
1.0
249.0
137.0
292.0
6.0
278.0
6.0
243.0
228.0
309.0
Pct_stations
7.9
2.4
36.5
41.1
2.1
2.1
33.1
34.7
0.2
39.2
21.6
46.0
0.9
43.8
0.9
38.3
35.9
48.7
-------
oo
Table 3-7. List of Utah cold-water-temperature indicator taxa, sorted by order, family, then
Final ID. Distribution and abundance information is also included. Sum_Individuals = the total
number of individuals from that taxon in the Utah database; Pct_Abund = percentage of total
individuals in the database composed of that taxon; Num_Stations = number of stations in the
database that the taxon occurred at; Pct_Stations = percentage of stations in the database at
which the taxon occurred (cont.)
Order
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Family
Leuctridae
Nemouridae
Pelecorhynchidae
Peltoperlidae
Perlodidae
Perlodidae
Perlodidae
Taeniopterygidae
Apataniidae
Glossosomatidae
Hydropsychidae
Lepidostomatidae
Limnephilidae
Uenoidae
Uenoidae
Final ID
Leuctridae
Visoka
Glutops
Yoraperla
Cultus
Kogotus
Megarcys
Taenionema
Apatania
Anagapetus
Parapsyche
Lepidostoma
Ecclisomyia
Neothremma
Oligophlebodes
Sum_individs
21,176.5
50.0
91.0
72.7
20,419.7
1,288.7
7,129.9
79,949.8
20,154.3
42.0
3,552.5
353,679.8
1,262.8
129,853.8
147,256.9
Pct_abund
0.1
0.0
0.0
0.0
0.0
0.0
0.0
0.2
0.0
0.0
0.0
0.8
0.0
0.3
0.3
Num_stations
106.0
1.0
4.0
5.0
97.0
14.0
65.0
87.0
39.0
2.0
40.0
240.0
14.0
100.0
101.0
Pct_stations
16.7
0.2
0.6
0.8
15.3
2.2
10.2
13.7
6.1
0.3
6.3
37.8
2.2
15.8
15.9
-------
Table 3-8. List of Utah warm-water-temperature indicator taxa. Distribution and abundance information is
also included. Sum_Individuals = the total number of individuals from that taxon in the Utah database;
Pct_Abund = percentage of total individuals in the database composed of that taxon; Num_Stations = number of
stations in the database that the taxon occurred at; Pct_Stations = percentage of stations in the database at
which the taxon occurred
Order
Coleoptera
Coleoptera
Coleoptera
Diptera
Diptera
Ephemeroptera
Ephemeroptera
Hemiptera
Isopoda
Odonata
Plecoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Family
Elmidae
Elmidae
Psephenidae
Psychodidae
Stratiomyidae
Caenidae
Leptohyphidae
Naucoridae
Asellidae
Coenagrionidae
Perlidae
Hydropsychidae
Hydroptilidae
Leptoceridae
Leptoceridae
Psychomyiidae
Final ID
Microcylloepus
Ordobrevia
Psephenus
Maruina
Caloparyphus
Caenis
Leptohyphidae
Ambrysus
Asellidae
Coenagrionidae
Calineuria
Cheumatopsyche
Ochrotrichia
Nectopsyche
Oecetis
Tinodes
Sum_individs
114,016
360
65.8
1,140.2
9,652
567
659,670.3
25,879.7
1,450,840.4
45,144.1
245
172,233.9
6,768.2
8,434.7
28,993.3
12,774.6
Pct_abund
0.24
0
0
0
0.02
0
1.42
0.06
3.12
0.1
0
0.37
0.01
0.02
0.06
0.03
Num_stations
50
5
4
16
26
11
197
39
81
117
9
105
29
35
90
34
Pct_stations
7.87
0.79
0.63
2.52
4.09
1.73
31.02
6.14
12.76
18.43
1.42
16.54
4.57
5.51
14.17
5.35
-------
a Cold
• Warm
Intolerant Intermediate
Enrichment Tolerance
Tolerant
Figure 3-7. Relationship between Utah cold and warm-water-preference
taxa and Utah enrichment tolerance scores (tolerance scores based on the
Utah data set were not available, so scores were based on assignments used
by New Mexico Environment Department). Taxa with enrichment tolerance
scores of 0-3 were categorized as Intolerant, those with scores of 4-6 were
Intermediate, and those with scores of 7-10 were Tolerant.
these limitations, we ran several different types of analyses in search of linkages between
biological and hydrologic data.
One of the analyses run was weighted-average modeling. We found that year had a
stronger influence on taxonomic composition than the hydrologic variables. The hydrologic
variable that showed the strongest influence was the 3-day mean of the annual minima (cfs). Of
the taxa that were evaluated, Leuctridae, Asellidae, and Zapada had the lowest 3-day minima
optima values, while Hyalella and Helicopsyche had the highest. Leuctridae and Zapada had
relatively low tolerance ranges, while Hyalella and Helicopysche had large tolerance ranges.
This suggests that Leuctridae and Zapada are better adapted to low flow conditions than other
taxa in Utah, perhaps due in part to their smaller sizes. For the full set of weighted average
results for 3-day annual minima, see Table B-l in Appendix B.
In addition to the weighted-average modeling, we also performed ordinations (NMDS
and CCA analyses). Results were similar. Year had the strongest influence on taxonomic
composition. Of the hydrologic parameters evaluated, 3-day annual minima and number of high
pulses per water year were the strongest drivers. Appendix B contains ordination plots from
3-20
-------
these analyses. The last set of analyses that we ran were correlation analyses on data from
seven sites that had more than 10 years of data. Table B-2 of Appendix B lists these sites. Only
one of these sites is considered to be least disturbed based on Utah DEQ's reference criteria.
There were a number of significant correlations at each site, but none of the taxa or biological
metrics showed consistent patterns across sites, so we were unable to develop candidate indicator
taxa. Results from these analyses are available upon request.
3.4.3. Traits-Based Indicators in a Warmer Drier Scenario
• As discussed in Section 3.1.2, the climate in Utah is projected to become warmer and
drier. We developed a list of taxa that may be most and least sensitive to these projected
changes based on the suite of trait modalities considered in Section 2.2.3. The taxa in
Table 3-9 that are deemed most sensitive, or most likely to be adversely affected by these
projected climatic changes, are mostly EPT taxa.
Two taxa, a Coleopteran and a Hemipteran, were included on the least sensitive list.
These taxa have the ability to exit (as adults), have high dispersal ability, strong flying strength,
strong swimming ability, and breathe through plastron-spiracles.
3.5. LEAST DISTURBED LONG-TERM BIOLOGICAL MONITORING SITES
Utah does not have a formal statewide long-term reference network. We explored
grouping least-disturbed sites together to create a statewide data set that could be analyzed for
long-term trends, but site-specific differences were evident within the data set, and the sample
size was relatively low; therefore, we focused on individual sites. We performed trend analyses
at the four reference stations in Utah that had 10 or more years of data. Figure 3-8 shows the
locations of these stations. Table 3-10 briefly summarizes site characteristics. Two are located
in the Wasatch and Uinta Mountains ecoregion, and the others are located in the Colorado
Plateaus ecoregion. Anthropogenic influences are higher than desired (>5% urban or >10%
agricultural) at two of the sites, but data were analyzed from these sites because they represented
the best-available long-term data in the state database. Table 3-11 lists the time periods for
which biological data are available for these sites. Data used in these analyses were limited to
fall (September-November) kick-method samples.
3-21
-------
Table 3-9. List of taxa that may be most and least sensitive to a warmer and
drier future scenario based on a combination of traits
Order
Diptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Coleoptera
Hemiptera
Family
Blephariceridae
Ephemerellidae
Heptageniidae
Heptageniidae
Heptageniidae
Heptageniidae
Chloroperlidae
Perlodidae
Perlodidae
Perlodidae
Apataniidae
Hydropsychidae
Lepidostomatidae
Limnephilidae
Dytiscidae
Corixidae
Final ID
Bibiocephala
Ephemerella
Cinygma
Cinygmula
Ironodes
Rhithrogena
Chloroperlidae
Cultus
Kogotus
Megarcys
Apatania
Parapsyche
Lepidostoma
Ecclisomyia
Dytiscidae
Corixidae
Sensitivity to warmer drier scenario
most
most
most
most
most
most
most
most
most
most
most
most
most
most
least
least
3-22
-------
j^ Individual Long-term Biological Sampling Sites '
EC ore gi on
LEVEL3_NAM
| | Central Basin and Range
^^^H I
• Colorado Plateaus
| Mojave Basin and Range
] Northern Basin and Range
| Southern Rockies
| Wasatch and Uinta Mountains
I Wyoming Basin
Figure 3-8. Locations of the four least disturbed long-term biological
monitoring sites (4927250 = Weber; 4951200 = Virgin; 4936750 = Duchesne;
5940440 = Beaver).
3-23
-------
Table 3-10. Site characteristics for the long-term biological monitoring stations in Utah. Percentage urban and
percentage agricultural (Ag) apply to a 1-km buffer zone around each site and are based on 2001 National Land
Cover Data
Site ID
UT 4927250a
UT 495 1200
UT 4936750
UT 5940440
Water body
Weber
Virgin
Duchesne
Beaver
Longitude (DD)
111.37358
112.94808
110.83
112.56711
Latitude (DD)
40.75294
37.28483
40.46139
38.28
EPA Level 3 ecoregion
Wasatch and Uinta
Mountains
Colorado Plateaus
Colorado Plateaus
Wasatch and Uinta
Mountains
Elevation
(m)
1,846.6
1,369.2
2,123.5
1,904.8
Drainage
area (km2)
740.7
756.3
489.5
236.2
%
Urban
4.5
3.4
10.3
0
%Ag
21.1
0.5
1.1
0
"Site is 0.8 km above a reservoir.
' DD = decimal degrees.
to
Table 3-11. Time periods for which biological data were available at the long-term monitoring sites in Utah.
Data used in these analyses were limited to fall (September-November) kick-method samples
Station ID
UT 4927250
UT 495 1200
UT 4936750
UT 5940440
Water body
Weber
Virgin
Duchesne
Beaver
Number of years of
data analyzed
17
14
12
9
Years
1985-1995, 1998, 2000, 2001, 2003-2005
1985-1993, 1996, 2000-2002, 2004
1985-1993, 1995,2000,2001
1996-1998, 2000-2005
-------
3.6. EVIDENCE OF TRENDS AT LEAST DISTURBED LONG-TERM MONITORING
SITES
3.6.1. Weber River (UT 4927250)
The Weber River site (UT 4927250) is located approximately 0.8 km above Rockport
Reservoir in Summit County in the Wasatch Uinta Mountains/Mountain Valleys ecoregion. It
has a drainage area of 740.7 km2 and an elevation of 1,847 m. Its highest maximum monthly
temperatures occur during July, and it lowest average flows (<100 cfs) occur from September
through March. This station has 19 years of data, ranging from 1985 to 2005, with spring,
summer, and fall sampling events. When limited to fall samples only, 17 years of data are
available. Daily temperature and precipitation data from 1955 to 2010 were gathered from the
Wanship Dam weather station (SitelD 429165, Latitude: 40.7908, Longitude: 111.408), which is
located approximately 5 km northwest of the biological sampling site, below Wanship dam.
Seven months of data were missing in 1955, so we used 1956 as the start date for our analyses.
Flow data from 1904-2011 were gathered from USGS gage 10128500 (Weber River near
Oakley, Latitude: 40.7371721, Longitude: 111.247965). The gage is located 10.6 km east of the
biological sampling site. Figure 3-9 shows an aerial photograph of the site, along with the
nearest weather station and active USGS gage.
3.6.1.1. Temporal Trends in Climatic and Biological Variables
Since 1956, mean annual air temperatures at the Weber River (UT 4927250) site have
ranged from 5 to 9.5°C. There is a great deal of year-to-year variability, but overall,
temperatures have been increasing over time (when fit with a linear trend line, r2 = 0.51,
p < 0.01) (see Figure 3-10). When PRISM air temperature data are compared to observed data,
there is good correspondence in pattern, but PRISM data are generally 1-2°C lower than
observed values, perhaps because the weather station is at a slightly lower elevation than the
biological sampling site. Mean annual flow and mean annual precipitation patterns have been
highly variable over time (see Figure 3-11). Since 1904, mean annual flow values have ranged
from 77 to 417 cfs (when fit with a linear trend line, r2= 0.05, p = 0.03). Precipitation patterns
generally show good correspondence with flow patterns (see Figure 3-11).
3-25
-------
Figure 3-9. Locations of the Weber River (UT 4927250) biological sampling
site, USGS gage 10128500 (Weber River near Oakley) and Wanship Dam
weather station. Image from Google Earth.
3-26
-------
10
o
o
I
CD
Q-
I
CD
=3
I
c
CO
CD
8
Observed
PRISM
1956
1966
1976
1986
1996
2006
Figure 3-10. Yearly trends in annual observed air temperature (°C) at the
Weber River site (UT 4927250) from 1955-2010, based on data from the
Wanship Dam weather station. For comparative purposes, PRISM annual air
temperature data associated with the biological sampling site are also included
from 1975-2005. The area shaded in grey corresponds to the period of biological
record. When the observed data are fitted with a linear trend line, r2 = 0.51,
p < 0.01, and.y = 0.0412x + 5.9756.
3-27
-------
..
450
400
350
300
250
200
150
100
50
650
600 |
550 |
75
500 ;|.
450 |
400 |
350 |
300 1
1905 1915 1925 1935 1945 1955 1965 1975 1985 1995 2005
250
200
150
-a
QJ
I
(/I
-Q
O
Figure 3-11. Yearly trends in mean annual flow (cfs) at the Weber River site
(UT 4927250) from 1904-2011, based on data from USGS gage 10128500.
For comparative purposes, observed annual precipitation data from the Wanship
Dam weather station are also included from 1955-2010. The area shaded in grey
corresponds to the period of biological record. When the observed data are fitted
with a linear trend line, r2 = 0.05, p = 0.03, and.y = 242.043 - 0.4707 x x.
In addition to mean annual values, mean maximum July temperature and mean fall flow
values were also evaluated, as these are likely to be physiologically stressful time periods for the
biological organisms. During the period of biological record (1985-2005), mean maximum July
air temperatures ranged from 26.9-34.3°C, and mean fall flow values ranged from 97.8 to
373.5 cfs (see Table 3-12). O/E scores range from a low of 0.57 in 1986 to values of 1.0 or
higher during the early 1990s and 2000 (see Figure 3-12A). The number of EPT taxa was
highest in the early 1990s and dropped dramatically from 2000-2005 (see Figure 3-12B). This
decline corresponds with a period of higher than normal temperatures and lower than normal
flows (see Figure 3-12C). Fffil scores were highly variable over time (see Figure 3-12B);
because the HBI is calculated based on abundance data (vs. relative abundance data), results
should be interpreted with caution due to reasons cited in Section 3.2. The cold-water taxa
metrics also showed a sharp decline from 2000-2005 (see Figures 3-13 A and B).
3-28
-------
Table 3-12. Range of temperature, precipitation, and flow values that
occurred at the Weber River site (UT 4927250) during the period of
biological record. SON = September, October, November
Parameter
Year
PRISM mean annual air temperature (°C)
Observed mean maximum July air temperature (°C)
Mean annual flow (cfs)
Mean SON flow (cfs)
PRISM mean annual precipitation (mm)
Min
1985
4.9
26.9
276.8
97.8
35.5
Max
2005
8.1
34.3
474.4
373.5
120.0
3-29
-------
B
1.0
0.8
uj 0.6
d
0.4
0.2
0.0
22
20
18
SS 16
fl
t
LLJ
-5
14
12
10
8
6
4
35
, 34
O
^T 33
2
'CD 32
CD
E 31
£. 30
x 29
CD
,
28
26
. \
•
\
- EPT taxa
- HBI
Temperature
Flow ,1
* !•
5.5
5.0
4.5
4.0 £
3.5
3.0
2.5
130
120
110
100 |
90 |
80 z
O
70 "
60
50
40
30
1985
1990 1995 2000 2005
Figure 3-12. Yearly trends at the Weber River site (UT 4927250) in (A) O/E,
(B) number of EPT taxa and HBI; (C) mean maximum July temperature
(°C) and mean September/October/November (SON) flow (cfs).
3-30
-------
B
o
o
O
0
-1
22
20
18
16
14
12
10
8
6
4
2
0
-2
35
34
33
15 32
o
0.
s
31
30
29
23
27
26
•- Cold Water
t- Warm Water
-»-•--
' Jf
-^ Cold Water
-»- Warm Water
• Temperature
How
130
120
110
100
90
80
70
60
50
40
30
o
CO
1985
1990 1995
2000 2005
Figure 3-13. Yearly trends at the Weber River site (UT 4927250) in
(A) number of cold and warm-water taxa; (B) percentage cold and warm-
water individuals; and (C) mean maximum July temperature (°C) and mean
September/October/November (SON) flow (cfs).
3-31
-------
Confounding factors related to water chemistry were not evident. From 1985-2005,
r\
parameter values were within the following ranges:
• DO: 9 to 12.2 mg/L
• pH: 7.8 to 8.7
• Chloride: 4.4 to 24 mg/L
• Nitrite (NO2) + nitrate (NO3): 0.15 to 0.3 6 mg/L
• Total phosphorus: 0.02 to 0.09 mg/L
• Specific conductance: 306 to 444 |imho/cm
• Turbidity: 1.7 to 10.3 NTU
3.6.1.2. Associations Between Biological and Climatic Variables
Kendall tau nonparametric correlations analyses allow examination of associations
between commonly used biological metrics, year, temperature, flow, and precipitation variables
at the Weber River (UT 4927250) site. Five of the 13 biological metrics showed strong
associations (r > 0.5) with year or the environmental parameters (see Table 3-13). Three of the
metrics (number of Ephemeroptera taxa, number of Plecoptera taxa, and number of intolerant
taxa) were negatively correlated with PRISM mean annual air temperature, number of Plecoptera
taxa was negatively correlated with year (r = -0.62), and the Fffil was positively correlated with
mean fall flow (r = 0.51) (see Table 3-13). The HBI was originally developed to reflect organic
enrichment but is generally expected to increase with increasing perturbation (Barbour et al.
1999, Table 2-2). Based on this, the positive correlation of FIB I with flow was somewhat
surprising, if lower low flows are assumed to be more stressful (i.e., decreasing fall low flows
would represent increasing stress, leading to an expectation for a negative relationship with
HBI). Similarly, the responses of diversity and dominance to mean fall flow, though r < |0.5|,
were counter to expectation, assuming decreasing fall low flows are more stressful (see
Table 3-13). Per reasons cited in Section 3.2, results for the HBI and Shannon-Wiener Diversity
Up to four samples were collected per year; the values shown here represent an average of these samples.
3-32
-------
Table 3-13. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics, year, and climatic variables at the Weber River site (UT 4927250). Results
are based on 17 years of data. Entries are in bold text if r > ±0.5 and are highlighted in gray if they are in a
direction opposite of what is expected. Ranges of biological metric values are also included. SON = September,
October, November. Per reasons cited in Section 3.2, results for the HBI and Shannon-Wiener Diversity Index
should be interpreted with caution because they are calculated based on abundance data (vs. relative abundance
data)
Biological metric
Total no. taxa
No. EPT taxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. Intolerant taxa
Percentage EPT individuals
Percentage Ephemeroptera individuals
Shannon- Wiener Diversity Index
Percentage noninsect individuals
Percentage dominant taxon
Percentage tolerant individuals
HilsenhoffBiotic Index
Range of metric
values
Min
12
5
2
0
2
5
27.5
1.5
1.4
0.7
20.4
0.0
2.9
Max
33
20
8
6
9.5
15
77.7
54.7
3.4
9.4
70.7
5.5
5.1
r values (based on Kendall Tau correlations)
Year
-0.10
-0.35
-0.41
-0.62
-0.02
-0.37
0.00
-0.37
0.13
-0.06
-0.10
0.19
-0.09
Air temperature (°C)
PRISM mean
annual
-0.32
-0.45
-0.55
-0.51
-0.23
-0.51
0.07
-0.36
-0.04
-0.02
-0.04
0.00
-0.14
Observed
mean
maximum July
-0.25
-0.38
-0.38
-0.44
-0.24
-0.29
0.10
-0.35
0.03
0.04
-0.06
-0.17
-0.16
Flow (cfs)
Mean
annual
0.01
0.00
0.16
0.04
-0.08
0.05
-0.22
0.26
-0.26
-0.37
0.44
-0.17
0.34
Mean SON
-0.16
-0.05
-0.01
0.09
-0.10
-0.06
-0.25
0.26
-0.47
-0.34
0.44
-0.35
0.51
PRISM mean
annual
precipitation (mm)
-0.16
-0.18
-0.10
-0.15
-0.02
-0.17
-0.21
0.07
-0.40
-0.41
0.40
-0.04
0.18
-------
Index should be interpreted with caution because they are calculated based on abundance data
(vs. relative abundance data).
Similar analyses were performed on the thermal preference metrics. The cold-water
metrics showed strong negative associations with year, and number of cold-water taxa was
negatively correlated with PRISM mean annual air temperature (see Table 3-14). A subset of
biological metrics that have shown responsiveness to hydrologic variables in other studies (see
Section 2, Table 2-2) were also examined (see Table 3-15). Three of the metrics showed strong
associations with year. Two of these, Odonata, Coleoptera, Hemiptera (OCH) taxa, and
depositional taxa, occurred in low numbers, so results for these metrics should be interpreted
with caution. Five of the percentage individuals metrics showed strong associations with mean
fall flow. Four of these went against expectations (see Table 2-2), with collector-gatherers
showing a positive correlation with mean fall flow, and percentage scraper/herbivores and
erosional individuals having negative correlations.
3.6.1.3. Groupings Based on Climatic Variables
Samples were partitioned into hottest/coldest/normal year groups and
lowest/normal/highest flow year groups. At the Weber River site (UT 4927250), on average, the
hottest years were 1.7°C warmer than the coldest years, and highest flow years had 170 more cfs
than lowest flow years. When samples were grouped based on temperature, there were
significant (p < 0.05) differences between mean metric values for total number of taxa, number
of EPT taxa, and number of cold-water taxa in hottest and coldest years samples (see
Table 3-16). There were no significant differences in mean metric values across the flow groups
(see Table 3-17).
NMDS was used to evaluate differences in taxonomic composition among samples
collected during hottest, coldest, and normal years. "Hottest year" samples formed a distinct
cluster from the "coldest" and "normal" year samples (see Figure 3-14). PRISM mean annual air
temperature from the year the sample was collected, PRISM mean annual air temperature from
the year prior to sample collection, and the difference between PRISM mean annual precipitation
from the sample collection year and the year prior were important drivers along Axes 1 and 2.
Figure 3-15 shows which taxa are the strongest drivers along these axes. Pteronarcys,
Chloroperlidae, and Ephemerella have the strongest positive correlations with Axis 2, and
3-34
-------
Table 3-14. Kendall tau nonparametric correlations analyses performed to examine associations between
thermal preference metrics, year, and climatic variables at the Weber River site (UT 4927250). Results are
based on 17 years of data. Entries are in bold text if r > ±0.50. Ranges of biological metric values are also
included. SON = September, October, November
Biological metric
No. cold-water taxa
Percentage cold-water
individuals
No. warm-water taxa
Percentage warm-water
individuals
Range of
metric values
Min
0
0
0
0
Max
6.0
20.9
3.0
1.3
r values (based on Kendall Tau correlations)
Year
-0.50
-0.71
-0.03
-0.14
Air Temperature (°C)
PRISM mean
annual
-0.57
-0.46
-0.38
-0.15
Observed mean
maximum July
-0.36
-0.28
-0.22
-0.13
Flow (cfs)
Mean
annual
0.19
0.22
0.03
-0.07
Mean
SON
0.14
0.34
0.02
-0.01
PRISM Mean
annual
precipitation (mm)
-0.05
-0.06
-0.02
-0.08
-------
a\
Table 3-15. Kendall tau nonparametric correlations analyses performed to examine associations between a
subset of biological metrics, year, flow, and precipitation variables at the Weber River site (UT 4927250). The
subset of biological metrics were selected per the criteria outlined in Section 2 and have shown responsiveness to
hydrologic variables in other studies (see Section 2, Table 2-2). Results are based on 17 years of data. Entries
are in bold text if r > ±0.5 and are highlighted in gray if they are in a direction opposite of what is expected.
Ranges of biological metric values are also included. SON = September, October, November
Biological metric
Richness
Percentage
individuals
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Range of metric
values
Min
2.0
3.0
1.0
2.0
1.0
0.0
0.0
4.0
3.2
13.3
0.2
1.8
0.7
0.0
0.0
4.0
Max
5.0
9.0
8.0
9.0
3.0
3.0
1.0
10.0
60.2
94.2
31.0
10.9
34.2
29.5
0.9
81.6
r values (based on Kendall Tau correlations)
Year
-0.08
-0.20
0.00
-0.18
-0.12
0.56
-0.51
-0.03
0.32
-0.35
0.43
-0.15
-0.04
0.53
-0.36
0.34
Flow (cfs)
Mean
annual
-0.08
0.13
-0.06
0.07
0.24
0.09
0.25
-0.16
-0.37
0.49
-0.26
-0.28
0.12
-0.32
0.23
-0.50
Mean SON
-0.16
-0.01
-0.23
-0.04
0.03
-0.01
0.16
-0.22
-0.34
0.57
-0.53
-0.25
0.06
-0.55
0.14
-0.59
PRISM mean annual
precipitation (mm)
0.04
-0.06
-0.09
-0.12
0.03
0.36
-0.02
-0.03
-0.18
0.26
-0.10
-0.35
-0.04
-0.12
0.07
-0.28
-------
Table 3-16. Mean metric values (±1 SD) for the Weber River site (UT 4927250) in coldest, normal, and hottest
year samples. Year groups are based on PRISM mean annual air temperature values. One-way ANOVA was
done to evaluate differences in mean metric values. Groups with no superscripts are not significantly different
(p < 0.05). Entries with superscripts have significant differences across year groups; those entries with different
superscripts are significantly different from each other (e.g., coldest total no. taxa vs. normal and hottest total
no. taxa)
Year group
Coldest
Normal
Hottest
O/E
0.9 ±0.2
0.8 ±0.2
0.9±0.1
Total no. taxa
27.5±3.5A
21.5±7.8AB
17.2±3.3B
No. EPT taxa
17.4±2.1A
13.6±4.9AB
8.8±2.2B
No. cold-
water taxa
4.9±1.1A
3.4±1.1A
1.0±0.7B
No. warm-
water taxa
2.3 ±0.8
1.1 ±0.7
1.0± 1.2
% Cold-water
individuals
6. 5 ±5.4
6.7 ±7.4
1.0± 1.1
% Warm-water
individuals
0.6 ±0.5
0.4 ±0.3
0.3 ±0.4
Table 3-17. Mean metric values (±1 SD) for the Weber River site (UT 4927250) in driest, normal, and wettest
flow year samples. Year groups are based on mean annual flow values from USGS gage 10128500. One-way
ANOVA was done to evaluate differences in mean metric values. None of the year groups are significantly
different (p < 0.05)
Year group
Driest
Normal
Wettest
O/E
0.8 ±0.2
0.9±0.1
0.8 ±0.2
Total no. taxa
21.0 ±7.8
22.5 ±7.1
22.3 ±6.6
No. EPT taxa
13.6 ±5.4
12.6 ±5.0
14.0 ±4.8
No. cold-
water taxa
2.6 ± 1.3
2.9 ±2.3
4.1 ± 1.5
No. warm-
water taxa
1.0 ±0.7
1.7 ± 1.1
1.5 ± 1.2
% Cold-water
individuals
3.7 ±4.5
2.9 ±2.6
9.1 ±8.8
% Warm-water
individuals
0.4 ±0.3
0.5 ±0.4
0.4 ±0.4
-------
Utah StationID 4927250
1987
A
1994
A
2000
2001
2005^
2004
A
Cat_Temp
i\
3
Axis 1
Figure 3-14. NMDS plot (Axis 1-2) for the Weber River site (UT 4927250),
shown in Figure 3-8. Cat_Temp refers to the temperature categories, which are:
1 = coldest years; 2 = normal years; 3 = hottest years. Samples are labeled by
collection year. tmean!4 = 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 ppt!4_ab = absolute difference between the
PRISM mean annual precipitation value from the year of the sample collection
and the year prior.
3-38
-------
Utah StationID 4927250
Ctloro
Nematoda
A
Tubifici
Athenx A
lOecetis
Optioser
Axis 1
Figure 3-15. NMDS plot (Axis 1-2) for the Weber River site (UT 4927250)
showing which taxa are most highly correlated with each axis.
Optioservus, Lepidostoma, and Hyallela have the strongest negative correlations with Axis 2.
The three taxa positively associated with Axis 2 tend toward cold-water-preference—
Chloroperlidae and Pteronarcys are absent from the "hottest year" samples, and Ephemerella is
present in all the "coldest year" and "normal year" samples and is only present in one "hottest
year" sample. Some additional taxa that occurred during multiple years and were not found in
"hottest year" samples include Rhithrogena, Nematoda, and Tubificidae. Warm-water-
preference taxa that are present in the majority of "hottest year" samples include Optioservus,
Lepidostoma, and Hyallela, though they also are present in "coldest year" and/or "normal year"
samples.
3-39
-------
3.6.2. Virgin River (UT 4951200)
The Virgin River site (UT 4951200) is located near Zion National Park (NP), on the
Virgin River below Zion Narrows in the Colorado Plateaus/Escarpments ecoregion. It has a
drainage area of 756 km2 and is at an elevation of 1,369 m. Its highest maximum monthly
temperatures occur during July, and it lowest average flows (approximately 50 cfs) occur from
July through February. This station has 15 years of data, ranging from 1985-2004. Temperature
and precipitation data dating from 1904 to 2010 were gathered from the Zion NP weather station
(SitelD 429717, Latitude: 37.2083, Longitude: 112.984). The weather station is located
approximately 9 km southwest of the biological sampling site. Flow data from 1991-1994 were
gathered from USGS gage 9405490 (North Fork Virgin River above Big Bend near Springdale,
Latitude: 37.27859, Longitude: 112.94466). The gage is located on a tributary 0.8 km south of
the biological sampling site. Figure 3-16 shows an aerial photograph of the site, along with the
nearest weather station and active USGS gage.
3.6.2.1. Temporal Trends in Climatic and Biological Variables
Since 1904, observed mean annual air temperatures at the weather station closest to the
Virgin River site (UT 4951200) have ranged from 12.1 to 18.6°C. Over time, year-to-year
variability has decreased, and temperatures have shown a slight increase (when fit with a linear
trend line, r2= 0.16, p < 0.01) (see Figure 3-17). When PRISM air temperature data are
compared to observed data, the PRISM temperatures are 3.5-6°C lower, perhaps due to the
differences in the locations and elevations of the weather station versus the biological sampling
sites (the elevation of the biological sampling is about 137 m higher than the weather station).
Because flow data are not available for most of the biological sampling period,
precipitation data were used as a surrogate. Since 1904, observed mean precipitation values have
ranged from 87 to 679 mm. There is a great deal of year-to-year variability, but overall, mean
annual precipitation has increased slightly over time (when fit with a linear trend line, r2 = 0.01,
p = 0.32) (see Figure 3-18). When PRISM precipitation data are compared to observed data,
there is close correspondence (see Figure 3-18).
3-40
-------
During the period of biological record (1985-2004), mean maximum July air temperatures
ranged from 35.6-40.6°C, and mean fall precipitation values ranged from 2.5 to 80.2 mm (see
Table 3-18). O/E scores increased over time, ranging from a low of 0.42 in 1985-1986 to 0.94
in 2001 (see Figure 3-19A). The number of EPT taxa increased to a high of 18 in the early
1990s, before dropping off to a low of 4 in 2000 (see Figure 3-19B). The year 1999 was
extremely dry (see Figure 3-19C), and this may have contributed to the low numbers in 2000.
Conditions in 1989 and 2003 were hotter and drier than normal (see Figure 3-19C), and this may
also have influenced the biological assemblage.
Figure 3-16. Locations of the Virgin River (UT 4951200) biological sampling
site, N. Fork Virgin River USGS gage, and Zion NP weather station. Image
from Google Earth.
3-41
-------
19
18
o 17
o
o
0)
0)
16
15
I13
I 12
10
Observed
---- PRISM
* •
'i/
1904 1914 1924 1934 1944 1954 1964 1974 1984 1994 2004
Figure 3-17. Yearly trends in annual observed air temperature (°C) at the
Virgin River site (UT 4951200) from 1904-2010, based on data from the Zion
NP weather station. For comparative purposes, PRISM annual air temperature
data associated with the biological sampling site are also included from
1975-2005. The area shaded in grey corresponds to the period of biological
record. When the observed data are fitted with a linear trend line, r2 =0.16,
p < 0.01, and.y = -6.0313 + 0.0114 *x.
3-42
-------
E
E
800
700
§ 600
% 500
£
^ 400
<
03
300
200
100
Observed
---• PRISM
1904 1914 1924 1934 1944 1954 1964 1974 1984 1994 2004
Figure 3-18. Yearly trends in mean annual precipitation (mm) at the Virgin
River site (UT 4951200) from 1904-2011, based on data from the Zion NP
weather station. For comparative purposes, PRISM annual precipitation data
associated with the biological sampling site are also included from 1975-2005.
The area shaded in grey corresponds to the period of biological record. When the
observed data are fitted with a linear trend line, r2= 0.01,/? = 0.32, and
y = -497.0505 + 0.449 x x.
Table 3-18. Range of temperature, precipitation, and flow values that
occurred at the Virgin River site (UT 4951200) during the period of
biological record. SON = September, October, November
Parameter
Year
PRISM mean annual air temperature (°C)
Observed mean maximum My air temperature (°C)
Mean SON precipitation (mm)
PRISM mean annual precipitation (mm)
Min
1984
10.4
35.6
2.5
226.9
Max
2004
14.0
40.6
80.2
644.5
3-43
-------
B
LU
o
03
t
LU
-5
o
o
CD
3
1.0
0.8
0.6
0.4
0.2
0.0
20
18
16
14
10
8
6
4
2
41
40
39
38
37
36
35
A
D ^
"? G?
I .'' 6 7\ n
,-EJ
i; i
i -^- EPT taxa
-o- HBI
Temperature
Precipitation
5.0
4.5
4.0
3.5
3.0
2.5
90
80
70
60
50
40
30
20
10
0
o
CL
O
1984 1987 1990 1993 1996 1999 2002
Figure 3-19. Yearly trends at the Virgin River site (UT 4951200) in (A) O/E,
(B) number of EPT taxa and HBI; (C) mean maximum July temperature
(°C) and mean observed September/October/November (SON) precipitation
(mm).
3-44
-------
HBI scores were highly variable (see Figure 3-19B) and reflect large changes in the
abundances of certain taxa from year to year, in particular Chironomidae and Ephemerella.
Because the HBI is calculated based on abundance data (vs. relative abundance data), results
should be interpreted with caution due to reasons cited in Section 3.2. From 2000 onward, cold-
water taxa were absent or occurred in extremely low numbers, while the number of warm-water
taxa increased by 2-3 taxa starting in 2001 (see Figures 3-20A and B).
Confounding factors related to water chemistry were not evident during the time period
for which water chemistry data were available. From 1985-2002, parameter values3 were within
the following ranges:
• DO: 8.9 to 10.1 mg/L
• pH: 7.9 to 8.6
• Chloride: 29.0 to 43.7 mg/L
• Nitrite (NO2) + nitrate (NO3): 0.06 to 0.29 g/L
• Nitrogen, Kjeldahl: 0.10 to 1.00 mg/L
• Total phosphorus: 0.01 to 0.04 mg/L
• Specific conductance: 511 to 618 |imho/cm
• Total suspended solids (TSS): 6 to 218 mg/L
3.6.2.2. Associations Between Biological Variables and Climatic Variables
Kendall tau nonparametric correlations analyses were performed to examine associations
between 13 commonly used biological metrics, year, temperature, and precipitation variables at
the Virgin River site (UT 4951200). Five of the biological metrics (total number of taxa, number
of EPT taxa, number of Ephemeroptera taxa, number of Plecoptera taxa, and number of
intolerant taxa) had strong (r > 0.5) negative associations with PRISM mean annual air
temperature (see Table 3-19). Two metrics were strongly associated with precipitation variables.
3Up to four samples were collected per year; the values shown here represent an average of these samples.
3-45
-------
Table 3-19. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics, year, and climatic variables at the Virgin River site (UT 4951200). Results
are based on 15 years of data. Entries are in bold text if r > ±0.5 and are highlighted in gray if they are in a
direction opposite of what is expected. Ranges of biological metric values are also included. SON = September,
October, November. Per reasons cited in Section 3.2, results for the HBI and Shannon-Wiener Diversity Index
should be interpreted with caution because they are calculated based on abundance data (vs. relative abundance
data)
Biological metric
Total no. taxa
No. EPTtaxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. Intolerant taxa
Percentage EPT individuals
Percentage Ephemeroptera
individuals
Shannon-Wiener Diversity Index
Percentage noninsect individuals
Percentage dominant taxon
Percentage tolerant individuals
Hilsenhoff Biotic Index
Range of metric
values
Min
12.00
4.00
2.00
0.00
1.00
0.00
34.82
29.41
1.72
1.41
24.57
0.00
2.88
Max
32.00
18.00
10.00
4.00
7.00
13.00
89.16
86.63
3.19
9.20
53.72
6.28
4.77
r values (based on Kendall Tau correlations)
Year
-0.09
-0.32
-0.53
-0.44
0.17
-0.46
0.01
0.27
-0.16
-0.14
0.16
0.20
0.16
Air temperature (°C)
PRISM mean
annual
-0.72
-0.72
-0.79
-0.56
-0.24
-0.66
0.27
0.41
-0.43
-0.49
0.34
-0.20
-0.10
Observed mean
maximum July
-0.20
-0.34
-0.44
-0.09
-0.19
-0.32
0.08
0.12
0.03
0.05
-0.03
0.04
-0.12
Precipitation (mm)
PRISM mean
annual
0.32
0.57
0.32
0.36
0.38
0.48
0.10
0.05
0.10
0.25
-0.01
0.00
0.03
Observed
SON
-0.05
0.05
0.01
-0.19
0.40
-0.11
-0.10
-0.05
-0.36
-0.21
0.54
-0.04
0.27
a\
-------
Number of EPT taxa was positively correlated (r = 0.57) with PRISM mean annual
precipitation, and the percentage dominant taxon metric was positively correlated (r = 0.54) with
mean fall precipitation (see Table 3-19). The direction of the relationship of the taxa dominance
metric with fall flow is counter to expectation, if decreasing fall low flows are considered more
stressful, and dominance is expected to increase (and diversity decrease) as lower more stressful
flows eliminate sensitive taxa (see Table 2-2). One metric, number of Ephemeroptera taxa, was
negatively correlated with year (r = -0.53).
When similar analyses were performed on the thermal preference metrics, percentage of
warm-water individuals was positively correlated (r = 0.56) with PRISM mean annual air
temperature, and the number of warm-water taxa had a strong positive association (r = 0.67) with
year (see Table 3-20).
The biological metrics in Table 3-21 have shown responsiveness to hydrologic variables
in other studies (see Section 2, Table 2-2). None had strong associations with mean annual or
mean fall precipitation variables. One metric, percentage OCH individuals, showed a strong
positive association (r = 0.50) with year (see Table 3-21).
3.6.2.3. Groupings Based on Climatic Variables
Samples were partitioned into hottest/coldest/normal year groups and
driest/normal/wettest flow year groups. At the Virgin River site (UT 4951200), on average, the
hottest years were 2.7°C warmer than the coldest years, and wettest years had approximately
250 more millimeters of precipitation than driest years. When samples were grouped based on
temperature, there were significant (p < 0.05) differences between mean metric values for total
number of taxa, number of EPT taxa, number of cold-water taxa, and number of warm-water
taxa in hottest and coldest/normal years samples, with the lowest mean metric values occurring
in the hottest year samples for all but the number of warm-water taxa, which were higher in the
hottest years (see Table 3-22). The percentage of cold-water individuals metric was significantly
lower in hottest versus normal year samples. One metric showed significant differences across
the precipitation groups (see Table 3-23). Mean number of EPT taxa was significantly higher in
the wettest versus driest year samples.
NMDS was used to evaluate differences in taxonomic composition among samples
collected during hottest, coldest, and normal years. "Hottest year" samples formed a distinct
3-47
-------
B
9
8
7
6
| 5
"o ^
o o
z 3
2
1
0
-1
60
50
40
:f 30
£ 20
ss
10
-10
41
o
S- 40
§
1 39
C 38
f 37
I
o) 36
35
Cold
Warm
Cold
Warm
• Temperature
Fall Precipitation
1985 1988 1991 1994 1997 2000 2003
90
80
70 1
60 g
50 1
Q.
40 'JD
30 z
O
20 w
10 |
0
-10
Figure 3-20. Yearly trends at the Virgin River site (UT 4951200) in
(A) number of cold and warm-water taxa; (B) percentage cold and warm-
water individuals; and (C) mean maximum July temperature (°C) and mean
observed September/October/November (SON) precipitation (mm).
3-48
-------
Table 3-20. Kendall tau nonparametric correlations analyses performed to examine associations between
thermal preference metrics, year, and climatic variables at the Virgin River site (UT 4951200). Results are
based on 15 years of data. Entries are in bold text if r > ±0.5. Ranges of biological metric values are also
included. SON = September, October, November
Biological metric
No. cold-water taxa
Percentage cold-water
individuals
No. warm-water taxa
Percentage warm-water
individuals
Range of
metric values
Min
0.0
0.0
1.0
2.6
Max
8.0
43.5
5.0
56.6
r values (based on Kendall Tau correlations)
Year
-0.33
-0.45
0.67
0.16
Air temperature (°C)
PRISM mean
annual
-0.31
-0.36
0.42
0.56
Observed mean
maximum July
-0.29
-0.12
0.19
0.23
Precipitation (mm)
PRISM mean
annual
0.33
0.25
-0.11
-0.32
Observed
SON
-0.17
-0.16
0.29
-0.08
VO
-------
Table 3-21. Kendall tau nonparametric correlations analyses performed to
examine associations between a subset of biological metrics, year, flow, and
precipitation variables at the Virgin River site (UT 4951200). The subset of
biological metrics were selected per the criteria outlined in Section 2 and
have shown responsiveness to hydrologic variables in other studies (see
Section 2, Table 2-2). Results are based on 15 years of data. Entries are in
bold text if r > ±0.5. Ranges of biological metric values are also included.
SON = September, October, November
Biological metric
Richness
Percentage
individuals
Collector filterer
Collector
Gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Collector filterer
Collector
Gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Range of metric
values
Min
2.0
3.0
1.0
2.0
1.0
0.0
1.0
3.0
1.3
37.2
0.2
1.8
1.9
0.0
2.4
1.3
Max
4.0
11.0
7.0
9.0
3.0
3.0
2.0
8.0
40.4
95.8
50.3
21.3
47.1
8.1
53.9
57.0
r values (based on Kendall tau
correlations)
Year
0.13
0.14
-0.35
-0.02
0.06
0.44
-0.05
0.06
-0.25
0.05
0.03
-0.45
0.49
0.50
0.16
0.08
Precipitation (mm)
PRISM
mean annual
0.33
0.28
0.35
0.15
0.19
-0.01
-0.09
0.14
0.32
0.36
-0.41
0.16
-0.08
-0.17
-0.32
-0.23
Observed
SON
0.21
0.19
-0.07
-0.30
-0.06
-0.24
-0.27
-0.01
-0.01
0.30
-0.34
-0.12
0.08
0.10
-0.08
-0.16
3-50
-------
Table 3-22. Mean metric values (±1 SD) for the Virgin River site (UT 4951200) in coldest, normal, and hottest
year samples. Year groups are based on PRISM mean annual air temperature values. One-way ANOVA was
performed to evaluate differences in mean metric values. Groups with no superscripts are not significantly
different (p < 0.05). Entries with superscripts have significant differences across year groups; those entries with
different superscripts are significantly different from each other (e.g., coldest and normal total no. taxa vs.
hottest total no. taxa)
Year
group
Coldest
Normal
Hottest
O/E
0.6 ±0.2
0.6±0.1
0.8±0.1
Total no.
taxa
22.8±6.6A
19.8±3.2A
14.5 ± 1.9B
No. EPT
taxa
12.3±3.9A
9.5±2.6A
5.3 ± 1.5B
No. cold-
water taxa
4.5±2.4A
5.3 ± 1.2A
0.8±0.5B
No. warm-
water taxa
1.5±0.6A
1.5±0.8A
3.8± 1.3B
% cold-water
individuals
15.7 ± 10.9^
23.4 ± 15.6A
0.2 ± 0.2B
% warm-water
individuals
7.7 ±6.7
18.1 ± 15.3
27.8 ± 19.4
Table 3-23. Mean metric values (±1 SD) for the Virgin River site (UT 4951200) in driest, normal, and wettest
flow year samples. Year groups are based on mean annual flow values from USGS gage 10128500. One-way
ANOVA was performed to evaluate differences in mean metric values. Groups with no superscripts are not
significantly different (p < 0.05). Entries with superscripts have significant differences across year groups; those
entries with different superscripts are significantly different from each other (e.g., driest no. EPT taxa vs.
normal and wettest no. EPT taxa)
Year
group
Driest
Normal
Wettest
O/E
0.7 ±0.2
0.6 ±0.2
0.7±0.1
Total no.
taxa
16.8 ±2.5
18. 3 ±3. 9
14.5 ±7.3
No. EPT
taxa
6.3 ± 1.7A
8?±25AB
12.5±4.7B
No. cold-
water taxa
2.5 ±2.4
4.2±2.1
4.5±3.1
No. warm-
water taxa
3.0± 1.8
1.5 ±0.8
2.3 ± 1.3
% cold-water
individuals
7.1 ±8.9
20.7 ± 18.1
12.8 ± 13.4
% warm-water
individuals
17.9 ± 10.0
24.9 ±20.5
7.4 ±5.4
-------
cluster from the "coldest" and "normal" year samples (see Figure 3-21). PRISM mean annual air
temperature from the year the sample was collected, PRISM mean annual air temperature from
the year prior to sample collection, and PRISM mean annual precipitation from the year prior to
sample collection were important drivers along Axis 1, while the difference between PRISM
mean annual precipitation from the sample collection year and the year prior is a strong driver
along Axis 2. Figure 3-22 shows which taxa are the strongest drivers along these axes.
Ephemerella, Nematoda, and Heptagenia have the strongest negative correlations with Axis 1,
and appear to tend toward a cold-water-preference. Nematoda are absent from the "hottest year"
samples, and Ephemerella and Heptagenia are present in all "coldest year" samples, six of the
seven "normal year" samples and only one of the "hottest year" samples.
Forcipomyia/Probezzia, Microcylloepus, Caloparyphus, and Chimarra have the strongest
positive correlations with Axis 1, and appear to be warm tolerant. These taxa are present in at
least two of the four "hottest year" samples and are absent from the "coldest year" and/or
"normal year" samples.
3.6.3. Beaver River (UT 5940440)
The Beaver River site (UT 5940440) is located in the Wasatch Uinta Mountains/Semiarid
Foothills ecoregion. It has a drainage area of 236 km2 and is at an elevation of 1,905 m. Its
highest maximum monthly temperatures occur during July, and its lowest average flows
(<30 cfs) occur from September through March. This station has 11 years of data, ranging
from 1994-2005, with a mix of spring and fall sampling events. When limited to fall samples
only, 9 years of data are available. Precipitation data from 1939 to 2010 were gathered from the
Beaver Canyon PH weather station (SitelD 420527, Latitude: 38.2681, Longitude: 112.481).
Temperature data became available from this station starting in 1997. The weather station is
located approximately 7 km east of the biological sampling site. USGS gage 10234500 (Beaver
River near Beaver, Latitude: 38.28052, Longitude: 112.56827) is colocated with the biological
sampling site and has flow data dating back to 1914. Figure 3-23 shows an aerial photograph of
the site, along with the nearest weather station and active USGS gage.
3-52
-------
3.6.3.1. Temporal trends in Climatic and Biological Variables
Since 1974, PRISM mean annual air temperatures at the Beaver River site (UT 5940440)
have ranged from 5.8 to 9.4°C. Temperatures have varied from year to year, but overall, have
3-53
-------
Utah StationID 4951200
1990
A
1991
A
opt14_ab
2000
1987
A
1985
A
Cat_Temp
A1
A2
3
Figure 3-21. NMDS plot (Axis 1-2) for the Virgin River site (UT 4951200).
Cat_Temp refers to the temperature categories, which are 1 = cold years;
2 = normal years; 3 = hot years. Samples are labeled by collection year.
tmean!4 = 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, ppt!4_ab = absolute difference between the PRISM mean
annual precipitation value from the year of the sample collection and the year
prior, and PrevYr_p = PRISM mean annual precipitation from the year prior to
sample collection.
3-54
-------
Utah StationID 4951200
Figure 3-22. NMDS plot (Axis 1-2) for Utah Station 4951200 (Virgin) that
shows which taxa are most highly correlated with each axis.
3-55
-------
Figure 3-23. Locations of the Beaver River (UT 5940440) biological sampling
site, Beaver River USGS gage, and Beaver Canyon pH weather station.
Image from Google Earth.
3-56
-------
r\
increased over time (when fit with a linear trend line, r = 0.52, p < 0.01) (see Figure 3-24).
When observed air temperature data from the nearest weather station (available starting in 1997)
are compared to PRISM data, there is close overlap, with less than a 1°C difference in values
(see Figure 3-24). Mean annual flow and mean annual precipitation patterns have been highly
variable over time (see Figure 3-25). Since 1914, mean annual flow values have ranged from 16
to 122 cfs (when fit with a linear trend line, r2 = 0.01,/> = 0.26). Precipitation patterns generally
show good correspondence with flow patterns (see Figure 3-11).
During the period of biological record (1996-2005), mean maximum July air
temperatures ranged from 26.5-31.1°C, and mean fall flow values ranged from 20.5 to 100.7 cfs
(see Table 3-24). O/E scores have fluctuated over time, ranging from 0.73 to 1.06 (see
Figure 3-26A). HBI scores were also variable (see Figure 3-26B); because the HBI is calculated
based on abundance data (vs. relative abundance data), trends in this metric should be interpreted
with caution due to reasons cited in Section 3.2. The highest number of EPT taxa occurred in
1996, then declined and remained at lower levels through 2005 (see Figure 3-26B). The drop in
EPT taxa in 1997 and 1998 corresponded with higher than normal fall flows and only low to
average July temperatures (see Figure 3-26C). In 2002 and 2003, conditions were hotter and
drier than normal. During this time, the percentage of cold-water individuals metric dropped to
its lowest levels (3-5%) (see Figure 3-27B). No warm-water taxa were present at this site.
From 1996-2005, water chemistry parameter values4 were within the following ranges:
• DO: 9.1 to 10.8 mg/L
• pH: 8.1 to 8.6
• Chloride: 3.5 to 83.2 mg/L
• Nitrite (NO2) + nitrate (NO3): 0.07 to 0.69 mg/L
• Total phosphorus: 0.03 to 0.07 mg/L
• Specific conductance: 107 to 153 |imho/cm
• Turbidity: 2.3 to 6.3 NTU
• Aluminum: 56.4 to 500 |ig/L
4Up to four samples were collected per year; the values shown here represent an average of these samples.
3-57
-------
10
o
1
cu
Q. o
E 8
.0)
7
CD
Observed
PRISM
1974 1978 1982 1986 1990 1994 1998 2002 2006
Figure 3-24. Yearly trends in PRISM annual air temperature data
associated with the Beaver River site (UT 5940440) from 1975-2005. Also
included are annual observed air temperature (°C) data from 1997-2006 based on
data from the Beaver Canyon PH weather station. The area shaded in grey
corresponds to the period of biological record. When the PRISM data are fitted
with a linear trend line, r2 = 0.52,p< 0.01, and.y = 0.0632x + 6.6556.
3-58
-------
E
o
1
Q.
_Q
O
900
800
700
600
500
400
300
200
100
Precipitation
Flow
140
120
100
80 o
60
40
20
1915 1925 1935 1945 1955 1965 1975 1985 1995 2005
Figure 3-25. Yearly trends in mean annual flow (cfs) from 1914-2011, based
on data from USGS gage 10234500. For comparative purposes, observed
annual precipitation data from the Beaver Canyon PH weather station are also
included from 1939-2010. The area shaded in grey corresponds to the period of
biological record. When the observed data are fitted with a linear trend line,
r1= 0.0l,p = 0.26, and.y = -0.0875x + 55.707.
Table 3-24. Range of temperature, precipitation, and flow values that
occurred at the Beaver River site (UT 5940440) during the period of
biological record
Parameter
Year
PRISM mean annual air temperature (°C)
Observed mean maximum July air temperature (°C)
PRISM mean annual precipitation (mm)
Mean annual flow (cfs)
Mean SON flow (cfs)
Min
1996
7.7
26.5
16.8
189.9
20.5
Max
2005
9.4
31.1
39.1
438.1
100.7
3-59
-------
1.0
0.8
uj 0.6
d
0.4
0.2
0.0
20
18
B
x 16
I
£ 14
§ 12
10 •
o
8
32
31
o>
^
^ 30
cu ou
CL
Q)
t 29
28
X
ro
TO 27
<
26
V
&..
EPT taxa
-o-HBI
Temperature
Flow
4.5
4.0
3.5
CD
I
3.0
2.5
2.0
40
38
36
34
32 f
30 £
28 ^
26 O
24^
22 |
20 "
18
16
14
1996
1999
2002
2005
Figure 3-26. Yearly trends at the Beaver River site (UT 5940440) in (A) O/E,
(B) number of EPT taxa and HBI; (C) mean maximum July temperature
(°C) and mean September/October/November (SON) flow (cfs).
3-60
-------
CO
.3 6
2
o
O
1
22
20
18
14
12
10
8
6
4
2
32
31
I 30
29
28
S
<
27
26
40
38
36
34
32
30 |
28 I
26 I
24 I
22 5
20
18
16
14
1996
1999
2002
2005
Figure 3-27. Yearly trends at the Beaver River site (UT 5940440) in
(A) number of cold and warm-water taxa; (B) percentage cold and warm-
water individuals; and (C) mean maximum July temperature (°C) and mean
September/October/November (SON) flow (cfs).
3-61
-------
Two potential confounding factors related to water chemistry were evident during the
time period for which biological data were available. In 2004, chloride concentrations spiked to
83.3 mg/L. In years prior, on average, chloride concentrations had been approximately 5 |ig/L.
Aluminum concentrations also occurred in high levels during the biological sampling period. In
1999, the concentration hit a high of 500 |ig/L, up from 88.5 |ig/L in 1996. By 2002, aluminum
concentrations had decreased back to levels ranging from 56 to 122 |ig/L.
3.6.3.2. Associations Between Biological Variables and Climatic Variables
Kendall tau nonparametric correlations analyses were performed to examine associations
between 13 commonly used biological metrics, year, temperature, flow, and precipitation
variables at the Beaver River site (UT 5940440). Only two of the biological metrics showed
strong associations with the environmental variables. The number of Plecoptera taxa metric had
a strong negative association with PRISM mean annual air temperature (r = -0.65), and
percentage Ephemeroptera individuals was positively correlated with mean annual flow
(r = 0.56) (see Table 3-25). When similar analyses were performed on the cold-water taxa
metrics, the percentage cold-water individuals metric showed a strong positive association with
mean annual and mean fall flow (r = 0.56) (see Table 3-26).
The biological metrics shown in Table 3-27 have shown responsiveness to hydrologic
variables in other studies (see Section 2, Table 2-2). Five of the six metrics that showed strong
associations with the flow or precipitation variables at this site went against expectations. The
scraper/herbivore richness metric was negatively correlated with flow and precipitation
variables, and the percentage predator and swimmer composition metrics had strong positive
associations with mean annual flow (r = 0.61 and r = 0.50, respectively). The one metric that
showed a strong association that was in keeping with expectations was the swimmer richness
metric. However, this relationship should be interpreted with caution because only one to
two swimmer taxa occurred in the samples that were analyzed. Depositional taxa were not
present at this site.
3.6.3.3. Groupings Based on Climatic Variables
Samples were partitioned into hottest/coldest/normal year groups and
lowest/normal/highest flow year groups. At the Beaver River site (UT 5940440), on average, the
3-62
-------
Table 3-25. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics, year, and climatic variables at the Beaver River site (UT 5940440). Results
are based on 9 years of data. Entries are in bold text if r > ±0.5. Ranges of biological metric values are also
included. SON = September, October, November. Per reasons cited in Section 3.2, results for the HBI and
Shannon-Wiener Diversity Index should be interpreted with caution because they are calculated based on
abundance data (vs. relative abundance data)
Biological metric
Total no. taxa
No. EPTtaxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. intolerant taxa
Percentage EPT individuals
Percentage Ephemeroptera
individuals
Shannon-Wiener Diversity Index
Percentage noninsect individuals
Percentage dominant taxon
Percentage tolerant individuals
Hilsenhoff Biotic Index
Range of
metric values
Min
16.0
10.0
3.0
2.0
4.0
9.0
28.9
22.1
1.9
0.6
19.4
0.0
2.9
Max
31.0
19.0
7.0
6.0
6.0
19.0
78.2
76.0
3.3
18.1
55.1
2.1
4.4
r values (based on Kendall tau correlations)
Year
-0.26
-0.28
-0.23
-0.46
-0.10
-0.35
0.28
0.28
-0.06
0.11
0.28
-0.40
-0.33
Air temperature (°C)
PRISM mean
annual
-0.15
-0.40
-0.17
-0.65
-0.03
-0.35
0.00
0.00
-0.33
-0.06
0.22
-0.04
0.28
Observed mean
maximum July
0.11
-0.12
0.05
-0.27
0.05
-0.15
0.14
0.07
-0.07
0.07
0.14
0.05
-0.07
Flow (cfs)
Mean
annual
-0.20
0.15
-0.17
0.26
-0.03
-0.06
0.56
0.56
-0.22
0.39
0.33
-0.04
-0.39
Mean
SON
-0.38
0.03
-0.30
0.20
-0.17
-0.12
0.33
0.22
-0.22
0.06
0.22
-0.25
-0.39
PRISM mean
annual
precipitation
(mm)
-0.15
0.28
-0.03
0.26
0.03
0.12
0.11
0.00
-0.11
0.17
0.00
-0.11
-0.28
o\
-------
Table 3-26. Kendall tau nonparametric correlations analyses performed to examine associations between
thermal preference metrics, year, and climatic variables at the Beaver River site (UT 5940440). No warm-water
taxa were present at this site. Results are based on 9 years of data. Entries are in bold text if r > ±0.5. Ranges
of biological metric values are also included. SON = September, October, November
Biological metric
No. cold-water taxa
Percentage cold-water
individuals
No. warm-water taxa
Percentage warm-water
individuals
Range of
metric
values
Min
2.0
3.0
0.0
0.0
Max
7.0
20.6
0.0
0.0
r values (based on Kendall tau correlations)
Year
-0.46
-0.17
—
—
Air tern
PRISM
mean
annual
0.03
-0.33
—
—
perature (°C)
Observed
mean
maximum July
0.29
-0.21
—
—
Flow (cfs)
Mean
annual
-0.03
0.56
—
—
Mean
SON
-0.15
0.56
—
—
PRISM mean
annual
precipitation
(mm)
0.03
0.44
—
—
o\
-------
Table 3-27. Kendall tau nonparametric correlations analyses performed to examine associations between a
subset of biological metrics, year, flow, and precipitation variables at the Beaver River site (UT 5940440). The
subset of biological metrics were selected per the criteria outlined in Section 2 and have shown responsiveness to
hydrologic variables in other studies (see Section 2, Table 2-2). Results are based on 9 years of data. Entries are
in bold text if r > ±0.5 and are highlighted in gray if they are in a direction opposite of what is expected. Ranges
of biological metric values are also included. Depositional taxa were not present at this site. SON = September,
October, November
Biological metric
Richness
Percentage
individuals
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Range of metric values
Min
2.0
4.0
2.0
3.0
1.0
2.0
0.0
5.0
1.4
50.7
0.7
2.5
15.5
3.1
0.0
4.7
Max
4.0
10.0
5.0
7.0
2.0
3.0
0.0
9.0
25.1
78.5
19.7
19.2
55.1
25.9
0.0
43.3
r values (based on Kendall Tau correlations)
Year
-0.22
-0.22
-0.03
0.03
-0.07
0.24
-
-0.15
0.17
0.00
-0.39
0.00
0.22
-0.44
-
0.00
Flow (cfs)
Mean annual
-0.15
-0.09
-0.57
0.17
-0.45
0.00
-
-0.03
-0.22
0.17
-0.44
0.61
0.50
0.06
-
-0.17
Mean SON
-0.15
-0.34
-0.70
0.10
-0.60
-0.35
-
-0.15
-0.11
0.17
-0.33
0.39
0.28
-0.17
-
-0.28
PRISM mean annual
precipitation (mm)
-0.07
-0.09
-0.57
0.17
-0.37
-0.35
-
0.03
0.00
0.06
-0.33
0.39
0.06
-0.17
-
-0.17
o\
-------
hottest years were 1°C wanner than the coldest years, and highest flow years had approximately
55 more cfs of flow than lowest flow years. When samples were grouped based on temperature
and flow, there were differences between mean metric values. Mean metric values for total
number of taxa, number of EPT taxa, and the cold water metrics were highest in the coldest year
samples (see Table 3-28), and percentage of cold-water individuals was lowest in the driest flow
year samples (see Table 3-29). None of the differences across year groups were significant
(p > 0.05). No NMDS ordinations were conducted at this site due to insufficient sample size.
3.6.4. Duchesne River (UT 4936750)
The Duchesne River site (UT 4936750) is located in the Colorado Plateaus/Semiarid
Benchlands and Canyonlands ecoregion. It has a drainage area of 490 km2 and is at an elevation
of 2,124 m. Its highest maximum monthly temperatures occur during July, and its lowest
average flows (<20 cfs) occur from September through March. This station has 14 years of data,
ranging from 1985-2002. When limited to fall samples only, 12 years of data are available.
Temperature and precipitation data dating from 1906 to 2010 were gathered from the Duchesne
weather station (SitelD 422253, Latitude: 40.1678, Longitude: 110.395), which is located
approximately 50 km southeast of the biological sampling site. Three months of data were
missing in 1906, so we used 1907 as the start date for our analyses. Flow data from 1990-2002
were gathered from USGS gage 927660 (W. F. Duchesne River above North Fork, near Hanna,
Latitude: 40.46161, Longitude: 110.83683), which is located on a tributary approximately
0.6 km west of the biological sampling site. Figure 3-28 shows an aerial photograph of the site,
along with the nearest weather station and active USGS gage.
3.6.4.1. Temporal Trends in Climatic and Biological Variables
Since 1907, observed mean annual air temperatures at the Duchesne River site
(UT 4936750) have ranged from 4.7 to 10.6°C. There is a lot of year-to-year variability, but
overall, temperatures have increased slowly over time (when fit with a linear trend line,
r2 = 0.15,/> < 0.01) (see Figure 3-29). When PRISM air temperature data are compared to
observed data, patterns are generally similar, but the PRISM temperatures are 1.4-4.5°C lower
than the observed values, perhaps due to the differences in the locations and elevations of the
weather station and biological sampling site, which are 50 km apart.
3-66
-------
Table 3-28. Mean metric values (±1 SD) for the Beaver River site (UT 5940440) in coldest, normal, and hottest
year samples. Year groups are based on PRISM mean annual air temperature values. One-way ANOVA was
done to evaluate differences in mean metric values. No entries are significantly different (p < 0.05) across year
groups
Year group
Coldest
Normal
Hottest
O/E
0.9±0.1
0.9±0.1
0.9 ±0.2
Total no.
taxa
23.0 ±7.0
20.0 ±2.6
19.3 ±3.5
No. EPT taxa
14.3 ±4.0
12.7±2.1
ll.Oil.O
No. cold-water
taxa
4.0 ±2.6
3.3 ±0.6
3.3 ±1.2
No. warm-
water taxa
~
~
~
% Cold-water
individuals
12.1 ±6.2
10.0 ±9.2
8.4 ±5.9
% Warm-water
individuals
~
~
~
o\
Table 3-29. Mean metric values (±1 SD) for the Beaver River site (UT 5940440) in driest, normal, and wettest
flow year samples. Year groups are based on mean annual flow. One-way ANOVA was done to evaluate
differences in mean metric values. No entries are significantly different (p < 0.05) across year groups
Year group
Driest
Normal
Wettest
O/E
0.9 ±0.04
0.9 ±0.2
0.8 ±0.1
Total no.
taxa
19.3 ±0.6
23.3 ±7.5
19.7 ±2.9
No. EPT taxa
11.3±0.6
13.7±4.7
13.0±1.7
No. cold-water
taxa
3.0± 1.0
4.3 ±2.5
3.3 ±0.6
No. warm-
water taxa
~
~
~
% Cold-water
individuals
5.3±2.1
10.9 ±6.8
14.4 ±7.4
% Warm-water
individuals
~
~
~
-------
Figure 3-28. Locations of the Duchesne River (UT 4936750) biological
sampling site, W.F. Duchesne River USGS gage and Duchesne weather
station. Image from Google Earth.
3-68
-------
o
o
0
0)
CL
E
03
13
C
C
<
c
03
CD
10
9
8
7
6
5
4
3
2
Observed
PRISM
1
1907 1917 1927 1937 1947 1957 1967 1977 1987 1997 2007
Figure 3-29. Yearly trends in annual observed air temperature (°C) from
1906-2010, based on data from the Duchesne weather station (1973-1980
were excluded due to missing data). For comparative purposes, PRISM annual
air temperature data associated with the biological sampling site are also included
from 1975-2005. The area shaded in grey corresponds to the period of biological
record. When the observed data are fitted with a linear trend line, r = 0.15,
p < 0.01, and.y = 6.4434 + 0.013 * x.
Because flow data are not available for most of the biological sampling period,
precipitation data were used as a surrogate. Since 1906, observed mean precipitation values have
ranged from 120 to 462 mm. There is a great deal of year-to-year variability (when fit with a
linear trend line, r2= 0.00, p = 0.57) (see Figure 3-30). When PRISM precipitation data are
compared to observed data, patterns are generally similar, but the PRISM values are 20-175 mm
higher than the observed values (see Figure 3-30).
During the period of biological record (1985-2001), mean maximum July air
temperatures ranged from 28.6-32.5°C, and mean fall flow values ranged from 10.6 to 30.3 cfs
(see Table 3-30). O/E scores have fluctuated over time, ranging from 0.64 to 1.07. In the late
1980s, O/E scores remained around 0.7, then increased in the early 1990s, ranging from to 0.8 to
1.0 (see Figure 3-31 A). HBI scores were also variable (see Figure 3-3 IB); because the HBI is
calculated based on abundance data (vs. relative abundance data), trends in this metric should be
interpreted with caution due to reasons cited in Section 3.2. The number of EPT taxa also
3-69
-------
500
450
400
E
£
c
o
I 350
g_
| 300
| 250
C
03
Observed
PRISM
200
150
100
1907 1917 1927 1937 1947 1957 1967 1977 1987 1997 2007
Figure 3-30. Yearly trends in mean annual precipitation (mm) from
1906-2011, based on data from the Duchesne weather station (1973-1980
were excluded due to missing data). For comparative purposes, PRISM annual
precipitation data associated with the biological sampling site are also included
from 1975-2005. The area shaded in grey corresponds to the period of biological
record. When the observed data are fitted with a linear trend line, r = 0.00,
p = 0.57, and.y = 9.0483 + 0.12 x x.
Table 3-30. Range of temperature, precipitation, and flow values that
occurred at the Duchesne River site (UT 4936750) during the period of
biological record
Parameter
Year
PRISM mean annual air temperature (°C)
Observed mean maximum luly air temperature (°C)
Mean annual flow (cfs)
Mean SON flow (cfs)
PRISM mean annual precipitation (mm)
Observed mean SON precipitation (mm)
Min
1985
3.0
28.6
15.3
10.6
234.7
7.2
Max
2001
5.5
32.5
61.7
30.3
480.5
42.7
3-70
-------
1.0
K 0.8
uj 0.6
O
0.4
0.2
x
CO
29
28
6
C 33 r
O
^ 32
e
QJ
|- 31
I—
^
"1 30
-•- Temperature
•• Fall Precipitation
1985 1988 1991 1994 1997 2000
CD
X
45
40
35
so g
O
w
20 ra
0)
10 0)
^ n to
10 .0
O
Figure 3-31. Yearly trends at the Duchesne River site (UT 4936750) in
(A) O/E, (B) number of EPT taxa and HBI; (C) mean maximum July
temperature (°C) and observed mean September/October/November (SON)
precipitation (mm).
3-71
-------
fluctuated over time. Numbers dropped in 1989, which was a year during which mean maximum
July temperatures were higher than normal and precipitation levels were lower than normal (see
Figures 3-3 IB and C). EPT richness values increased to a high of 21 in 1995, before dropping
again in 2000. The year 1999 was drier than normal, which may have impacted the assemblage
in 2000. The cold-water taxa metric values also fluctuated over time. Numbers of cold-water
taxa were lowest in 2000 and 2001 (see Figure 3-32A), while percentage cold-water individuals
was variable over time (see Figure 3-32B). Very few warm-water taxa occurred at this site.
From 1985-2001, water chemistry parameter values5 were within the following ranges:
• DO: 8.6 to 11.7mg/L
• pH: 7.1 to 8.5
• Chloride: 1.6 to 30 mg/L
• Nitrite (NO2) + nitrate (NO3): 0.1 to 2.44 mg/L
• Total phosphorus: 0.01 to 0.06 mg/L
• Specific conductance: 237 to 427 |imho/cm
• Turbidity: 0.4 to 6 NTU
There may have been potential confounding factors related to water chemistry in 1993
and 1994. In 1993, concentrations of nitrite + nitrate reached a high of 2.44 mg/L. In other
years, nitrite + nitrate concentrations averaged 0.2 mg/L. In 1994, turbidity, total suspended
solids, chloride, and sulfate concentrations were higher than normal.
3.6.4.2. Associations Between Biological Variables and Climatic Variables
Kendall tau nonparametric correlations analyses were performed to examine associations
between 13 commonly used biological metrics, year, temperature, and precipitation variables at
the Duchesne River site (UT 4936750). Five metrics (total number of taxa, number of EPT taxa,
number of Trichoptera taxa, Shannon-Wiener Diversity Index, and percentage tolerant
individuals) had strong negative associations with the observed mean maximum July temperature
5Up to four samples were collected per year; the values shown here represent an average of these samples.
3-72
-------
10
B
CO
x
o
d
-2
30
25
20
_tf>
TO
-i 15
.>
T>
£ 10
o
2
s.
I
I
-5
33.0
32.5
32.0
31.5
31.0
30.5
30.0
29.5
29.0
28.5
28.0
>- Cold
f- Warm
Cold
•.Varn
• Temperature
Fall Precipitation
198S 1987 1989 1991 1993 1995 1997 1999 2001
45 „
40
c
o
35 Jo
30
25
9.
'o
I
Figure 3-32. Yearly trends at the Duchesne River site (UT 4936750) in
(A) number of cold and warm-water taxa; (B) percentage cold and warm-
water individuals; and (C) mean maximum July temperature (°C) and
observed mean September/October/November (SON) precipitation (mm).
3-73
-------
(see Table 3-31). These relationships are as would be expected except for the abundance of
tolerant taxa, which are expected to increase with increasing temperature (see Table 2-2).
Though counter to expectation, the relationship between percentage tolerant individuals and
temperature should be interpreted with caution because only 1 to 2 percentage of the assemblage
was composed of tolerant individuals. Also, because the Shannon-Wiener Diversity Index is
calculated based on abundance data (vs. relative abundance data), results for this metric should
be interpreted with caution due to reasons cited in Section 3.2. There were no strong
associations between the thermal preference metrics and temperature or precipitation variables
(see Table 3-32).
The biological metrics shown in Table 3-33 have shown responsiveness to hydrologic
variables in other studies (see Section 2, Table 2-2). Two of these metrics had strong
associations with precipitation variables, and both associations were in keeping with
expectations. The scraper/herbivore richness metric was positively correlated with PRISM mean
annual precipitation, and the percentage swimmer individual metric was negatively correlated
with mean fall precipitation. One metric, OCH richness, showed a strong positive association
with year, but this relationship should be interpreted with caution because very few (0 to 2) OCH
taxa occurred at this site.
3.6.4.3. Groupings Based on Climatic Variables
Samples were partitioned into hottest/coldest/normal year groups and driest, normal, and
wettest year groups. At the Duchesne River site (UT 4936750), on average, the hottest years
were 1°C warmer than the coldest years, and wettest years had approximately 150 more
millimeters of precipitation than driest years. When samples were grouped based on temperature
and precipitation, there were differences between mean metric values. Mean O/E values were
highest in the hottest year samples, and the mean value of the percentage of cold-water
individuals metric was highest in the coldest year samples (see Table 3-34). Mean values of the
warm-water taxa metrics were lowest in the coldest year samples, but this relationship should be
interpreted with caution because warm-water taxa occurred in very low numbers at this site (see
Table 3-34). Mean numbers of total taxa and EPT taxa were highest in the wettest year samples
(see Table 3-35). None of the differences across year groups were significant (p > 0.05). No
NMDS ordination was performed at this site due to insufficient sample size.
3-74
-------
Table 3-31. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics, year, and climatic variables at the Duchesne River site (UT 4936750). Results
are based on 12 years of data. Entries are in bold text if r > ±0.5 and are highlighted in gray if they are in a
direction opposite of what is expected. Ranges of biological metric values are also included. SON = September,
October, November. Per reasons cited in Section 3.2, results for the HBI and Shannon-Wiener Diversity Index
should be interpreted with caution because they are calculated based on abundance data (vs. relative abundance
data)
Biological metric
Total no. taxa
No. EPTtaxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. Intolerant taxa
Percentage EPT individuals
Percentage Ephemeroptera
individuals
Shannon-Wiener Diversity Index
Percentage noninsect individuals
Percentage dominant taxon
Percentage tolerant individuals
Hilsenhoff Biotic Index
Range of metric
values
Min
17.0
8.0
3.0
1.0
4.0
7.0
19.8
1.6
2.0
0.8
21.1
0.0
2.5
Max
34.0
21.0
8.0
6.0
8.0
16.0
65.0
38.3
3.7
7.8
67.7
2.0
5.0
r values (based on Kendall Tau correlations)
Year
0.34
0.05
0.13
-0.29
0.05
-0.30
-0.21
0.00
0.12
-0.06
-0.18
0.22
0.30
Air temperature (°C)
PRISM
mean annual
-0.09
-0.18
-0.19
0.19
-0.02
0.05
-0.06
-0.15
-0.03
-0.21
-0.15
-0.10
0.15
Observed mean
maximum July
-0.56
-0.53
-0.29
-0.25
-0.73
-0.14
-0.06
-0.09
-0.52
-0.21
0.39
-0.54
0.21
Precipitation (mm)
PRISM
mean
annual
0.28
0.46
0.29
0.22
0.28
0.21
0.21
0.12
0.30
-0.18
-0.24
0.35
-0.30
Observed
mean
SON
-0.09
0.08
-0.03
0.05
0.02
0.11
-0.06
-0.27
-0.21
-0.27
0.33
-0.03
-0.21
-------
Table 3-32. Kendall tau nonparametric correlations analyses performed to examine associations between
thermal preference metrics, year, and climatic variables at the Duchesne River site (UT 4936750). Results are
based on 12 years of data. Entries are in bold text if r > ±0.5. Ranges of biological metric values are also
included. SON = September, October, November
Biological metric
No. cold-water taxa
Percentage cold-water
individuals
No. warm-water taxa
Percentage warm-water
individuals
Range of metric
values
Min
4.0
6.1
0.0
0.0
Max
9.0
28.5
2.0
0.5
r values (based on Kendall Tau correlations)
Year
-0.21
-0.12
0.37
0.39
Air Temperature (°C)
PRISM
mean
annual
-0.11
-0.21
-0.10
-0.13
Observed mean
maximum July
-0.31
0.15
-0.10
-0.17
Precipitation (mm)
PRISM
mean
annual
0.31
0.12
0.13
0.17
Observed
mean SON
0.11
0.39
0.33
0.28
-------
Table 3-33. Kendall tau nonparametric correlations analyses performed to examine associations between a
subset of biological metrics, year, flow, and precipitation variables at the Duchesne River site (UT 4936750).
The subset of biological metrics were selected per the criteria outlined in Section 2 and have shown
responsiveness to hydrologic variables in other studies (see Section 2, Table 2-2). Results are based on 12 years
of data. Entries are in bold text if r > ±0.5. Ranges of biological metric values are also included.
SON = September, October, November
Biological metric
Richness
Percentage
individuals
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCR
Depositional
Erosional
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCR
Depositional
Erosional
Range of metric values
Min
1.0
3.0
2.0
6.0
0.0
0.0
0.0
4.0
1.9
21.6
4.0
2.0
0.0
0.0
0.0
6.6
Max
5.0
8.0
7.0
12.0
3.0
2.0
1.0
10.0
32.1
78.3
49.3
9.7
23.3
47.8
0.3
54.0
r values
Year
0.19
0.34
-0.03
-0.03
0.06
0.77
0.39
0.13
-0.24
-0.06
0.15
-0.03
0.12
0.33
0.40
0.06
Precipitation (mm)
PRISM mean annual
0.09
0.08
0.57
0.45
-0.42
0.09
0.26
0.30
0.18
-0.36
0.39
-0.27
-0.18
0.23
0.32
0.24
Observed mean SON
-0.26
-0.08
0.23
0.14
-0.22
0.05
0.26
0.00
-0.09
-0.03
0.18
-0.36
-0.58
0.07
0.32
0.09
-------
Table 3-34. Mean metric values (±1 SD) for the Duchesne River site (UT 4936750) in coldest, normal, and
hottest year samples. Year groups are based on PRISM mean annual air temperature values. One-way
ANOVA was done to evaluate differences in mean metric values. No entries are significantly different (p < 0.05)
across year groups
Year group
Coldest
Normal
Hottest
O/E
0.8 ±0.1
0.7±0.1
l.OiO.l
Total no.
taxa
22.3 ±6.1
25.7 ±4.0
24.3 ±8.7
No. EPT
taxa
13.7 ±3.2
15.5 ±1.4
14.0 ±6.6
No. cold-
water taxa
6.3 ±1.5
6.3 ±1.0
5.7 ±2.9
No. warm-
water taxa
0.3 ±0.6
0.7 ±0.8
0.7 ± 1.2
% Cold-water
individuals
24.3 ±4.1
14.9 ±6.8
17.7 ±8.5
% Warm-water
individuals
0.03 ±0.1
0.1 ±0.2
0.1 ±0.2
oo
Table 3-35. Mean metric values (±1 SD) for the Duchesne River site (UT 4936750) in driest, normal, and wettest
year samples. Year groups are based on PRISM mean annual precipitation values. One-way ANOVA was done
to evaluate differences in mean metric values. Groups with no superscripts are not significantly different
(p < 0.05). Entries with superscripts have significant differences across groups; those entries with different
superscripts are significantly different from each other (e.g., driest % cold-water individuals vs. normal % cold-
water individuals)
Year group
Driest
Normal
Wettest
O/E
0.8 ±0.2
0.8 ±0.1
0.8 ±0.2
Total no.
taxa
20.7 ±3.2
24.8 ±6.1
27.7±5.1
No. EPT
taxa
12.3 ±2.1
15.0 ±4.2
16.3 ±1.5
No. cold-
water taxa
5.0 ±0.8
6.8 ±2.2
6.8 ±0.9
No. warm-
water taxa
0.5 ±0.6
0.5 ± 1.0
0.8 ± 1.0
% Cold-water
individuals
11.2±6.3A
23.4±3.1B
19.3±6.6AB
% Warm-water
individuals
0.05 ±0.1
0.10±0.2
0.15 ±0.3
-------
3.7. SENSITIVITY OF BENTHIC MACROINVERTEBRATES TO TEMPERATURE,
PRECIPITATION, AND STREAM FLOW
The spatial distributions of cold and warm-water taxa were examined to gain insights into
which areas in Utah are likely to be most and least sensitive to projected changes in temperature
and stream flow. If the assumption is made that streams with greater numbers and abundances of
cold-water taxa will be most sensitive to warming temperatures and changing precipitation
patterns, then streams in the higher elevation ecoregions, such as the Wasatch and Uinta
Mountains, Southern Rockies, and Wyoming Basin, will be most sensitive. Table 3-36 shows
differences in the distributions of thermal preference taxa between ecoregions. The prevalence
and distribution of cold- and warm-water-preference taxa also vary predictably with stream
order. First- and second-order streams in Utah have slightly greater relative abundance and
richness of cold-water-preference taxa, and fewer warm-preference taxa, compared to third- or
higher-order streams (see Figure 3-33). However, the four Utah sampling stations that had
sufficient long-term data to analyze temporal trends were fourth to fifth order steams. Although
the greatest number of cold-water taxa may occur in the coldest, highest elevation streams, it
may be that the greatest amount of change will occur in transitional areas, where species are
expected to be closer to their tolerance limits. Effects are likely to vary spatially. Poff et al.
(2010) also concluded that sites will be differentially vulnerable to climate change.
3.8. IMPLICATIONS FOR UTAH DEQ'S BIOMONITORING PROGRAM
Over the last century in Utah, there has been a lot of year-to-year variability in
temperature and precipitation patterns, both statewide and at the four long-term least disturbed
biological monitoring sites that were closely examined. Air temperature has increased over time
at all the sites, but change rates have differed depending on the location and the time period
being examined. We assume that stream temperatures have followed similar patterns, although
long-term continuous stream temperature data needed to test stream temperature trends is rare.
Recognizing the value of such data, Utah is deploying some temperature data loggers, though
only a few of these correspond to sites where repeated biological sampling occurs. Precipitation
and flow patterns have been even more variable than temperature patterns over the last century.
Some sites have shown a slight overall increase in mean annual flow or precipitation; at other
sites, there has been a slight decrease. There is much uncertainty associated with future
projections for precipitation.
3-79
-------
Table 3-36. Summary of differences in elevation, PRISM mean annual air temperature and precipitation, and
mean number and percentage of cold and warm-water-preference taxa across and within major ecoregions.
Samples were not limited to a particular season
Ecoregion
Mojave basin and range
Central basin and range
Colorado plateaus
Northern basin and range
Wyoming basin
Wasatch and Uinta mountains
Southern Rockies
No.
samples
13
177
205
6
27
644
7
Elevation
(m)
736.6
1,411.7
1,729.4
1,769.7
2,002.0
2,131.1
2,535.2
Air
temperature
(°C)
16.8
10.0
9.1
8.6
5.7
5.4
6.3
Richness
Cold water
2.8 ±2.4
1.4 ± 2.0
3.8 ±2.8
4.7 ±1.0
6.1 ±4.0
5.5 ±4.0
9.1 ±0.7
Warm water
1.3 ±0.9
2.4 ± 1.4
1.2± 1.2
1.2 ±0.8
1.3 ±0.9
1.0± 1.3
0±0
Relative abundance
Cold water
6.6 ±8.9
2.1 ±7.0
9.8 ±11.5
3.2 ±2.9
13.2±13.2
13.1 ±15.4
30.6 ±14.6
Warm water
5.5 ±8.7
10.8 ±16.5
6.1 ±11.6
12 ±20.1
1.1 ±2.4
3.8 ±11.0
0±0
oo
o
-------
OJ
00
in
-------
At the four long-term biological sampling sites, O/E scores varied across years, but
generally did not show strong associations with temperature, precipitation, or flow variables. At
the Duchesne River site (UT 4936750), O/E scores were, on average, slightly higher in hottest
year samples, but this difference was not significant and did not occur at the other three sites.
Associations between certain biological metrics and climatic variables were more evident and
suggest that variables related to temperature and stream flow have influenced community
composition over time, although this cannot be proven with observational data. The Plecoptera
richness metric was negatively associated with mean annual air temperature at three of the sites,
while total number of taxa, number of EPT taxa, number of Ephemeroptera taxa, number of
intolerant taxa, and number of cold-water taxa had strong negative associations with temperature
at two sites. Other metrics showed strong associations with temperature variables as well, but
these associations only occurred at single sites. Aside from the Duchesne River (UT 4936750)
site, biological metrics showed stronger associations with mean annual temperature than mean
maximum July temperature. Although July temperatures represent a stressful period
physiologically and may impact an organism directly, these results suggest that both summer and
annual temperatures should be considered, as annual temperatures encompass a number of
critical time periods in an organism's life cycle and are likely to both directly and indirectly
impact the organisms.
When biological samples were grouped by hottest/coldest/normal years, there were
significant differences between mean metric values for total number of taxa, number of EPT taxa
and number of cold-water taxa at the Weber River site (UT 4927250) and at the Virgin River site
(UT 4951200). On average, at both sites, there were seven to eight fewer total taxa and EPT taxa
in hottest versus coldest year samples, and four fewer cold-water taxa in the hottest year samples.
Because hottest year samples were, on average, about 2°C warmer than coldest year samples,
comparisons across these year groups may provide good approximations of what types of
climate-induced changes we can expect by midcentury. Similar patterns (albeit nonsignificant),
were also evident at the Beaver River site (UT 5940440), where mean metric values for total
number of taxa, number of EPT taxa, and cold-water taxa were highest in the coldest year
samples.
Fewer biological metrics had strong associations with flow or precipitation variables, and
when a strong association did occur, it was only at a single site. The lack of consistency across
3-82
-------
sites may have been due in part to the large amount of year-to-year variability in flow and
precipitation patterns. Metrics that showed strong positive correlations with flow or precipitation
variables included number of EPT taxa (Virgin River site—UT 4951200), percentage
Ephemeroptera individuals (Beaver River site—UT 5940440), and percentage cold-water
individuals (Beaver River site—UT 5940440). When biological samples were grouped by
driest/wettest/normal years, total number of taxa, EPT taxa, and cold-water taxa were lowest in
the driest year samples at the Duchesne River site (UT 4936750).
The fact that more biological metrics had strong associations with temperature variables
than with precipitation or flow variables suggests that temperature may have a stronger influence
on the biological assemblage than precipitation or flow. However, it is important to consider
these climatic variables in combination. At various times at each of the four sites, there were
years during which temperatures were higher than normal, and flow or precipitation values were
lower than normal. These hotter drier conditions sometimes occurred over consecutive years,
and generally seemed to correspond with declines in biological metrics, mainly total number of
taxa, number of EPT taxa, and cold water metrics. This was evident at the Weber River
(UT 4927250) and Virgin River (UT 4951200) sites.
In addition to the historic trend analyses, we also performed exploratory analyses to gain
insights into how future projected climatic changes might impact Utah DEQ's assessment
methods. In these exploratory analyses, the climate-related predictor variables that are used in
the Utah fall RIVPACS models were manipulated in a way that would simulate future projected
changes. When climate-related predictor variables were altered, there was very little effect on
O/E values. The amount of change that did occur was within the range of natural variability (see
Appendix B). Similar results were obtained when the model was rerun in a way that allowed for
inclusion of rare taxa, and when extreme changes were made to the climate-related variables
(i.e., doubling temperature, halving precipitation variables).
There are a number of possible reasons that the alterations to the climate-related predictor
variables resulted in small changes to O/E values. For one, the analyses were based on reference
site data, and reference sites are typically more stable than test sites. In addition, elevation was
disregarded in the model manipulations. It might be informative to explore how manipulations
of the elevation-related predictor variables affect O/E values, especially since elevation and
temperature are linked. Another consideration is the assumption that climate-related predictor
3-83
-------
variables, which are typically based on long-term (30-year) averages, are relatively invariant
over ecologically-relevant time. If climate change is going to be an important factor in years to
come, it would be interesting to develop a second RIVPACS model that includes predictor
variables based on current climate (not just the historic benchmark climate) and to compare O/E
values across these models over time. This should allow for partitioning of climate change
effects over time. For more information on these exploratory analyses, see Appendix B.
Overall, these results suggest that changes in temperature and stream flow conditions
have influenced community composition to varying degrees at these sites over time. Impacts
were particularly evident at the Weber River site (UT 4927250) and at the Virgin River site
(UT 4951200), where consecutive years of hot dry conditions occurred from 2000-2005.
Although statistical inferences cannot be made on statewide trends based on data from
four individual sites, these analyses further our understanding of the effects that changing
temperature and stream flow conditions can have on biological assemblages, and help establish
expectations for biological responses to future climate changes.
These analyses also provide insights as to which climate change indicators might be best
to track over time in southwestern states. Results suggest that climate-induced trends are most
likely to be detected in total taxa, EPT and EPT-related metrics, and thermal preference metrics.
Some limitations of the thermal preference metrics are that they typically occur in low numbers,
and most show sensitivity to organic enrichment, which confounds the associations with
temperature. Individual cold preference taxa, including Pteronarcys, Chloroperlidae,
Ephemerella, and Heptagenia, should also be considered as good indicators that could be
targeted for tracking (forming a "watch list"), as could cold stenothermic community types (per
Poff et al., 2010). A subset of biological metrics related to rheophily, habit, and functional
feeding group that have shown responsiveness to hydrologic variables in other studies were
analyzed but were generally found to be either unresponsive to precipitation or flow variables, to
show patterns that were inconsistent across sites, or to show patterns that oftentimes went against
expectations.
3-84
-------
4. MAINE
4.1. EXPOSURES
4.1.1. Regional Projections for the Northeastern United States
Numerous indicators of climate change have been observed in the northeastern United
States over the last century, and especially the last 3 decades, and many of the trends are
projected to continue into the future. Regional average air temperatures are projected to increase
by an average of 3-5°C over the coming century (2070-2099 compared to 1961-1990) under
low to high emission scenarios (Hayhoe et al., 2007; UCS, 2006) (see Table 4-1). While greater
temperature increases have been observed in winter temperatures in recent decades, the future
projections are for slightly greater temperature increases to occur in summer (Hayhoe et al.,
2007).
Projections for precipitation are more variable than for temperature, and for the
Northeast, result in differences in expectations for both the average amount and the seasonal
distribution of precipitation changes. Hayhoe et al. (2007) project an increase in average annual
precipitation over the next century, from 5 to 8% by 2064 to 7 to 14% by 2099 (see Table 4-1).
However, their modeling approach results in even greater expected increases in winter
precipitation, but no change to slight decreases in summer precipitation (see Table 4-1). UCS
(2006) also project increases in winter precipitation, but variable projections for summer ranging
from a slight increase to slight decreases (Table 4-1). In contrast, Schoof et al. (2010) project
increasing precipitation for the Northeast in all seasons, due to both an increased frequency of
storms and an increase in the average size of each rain event (see Table 4-1). In fact, their
projected increases in warm-season precipitation are relatively large—increases of 4 to 7% in
amount and 8 to 20% in frequency by mid- and end of century, respectively (see Table 4-1). For
cold-season precipitation, Schoof et al. (2010) project increases in the frequency (4 to 9% by
mid- to end of the century), and increases in the magnitude of precipitation of 12 to 27% for the
two time periods.
A number of streamflow changes are projected in association with these temperature and
precipitation changes. Total runoff is projected to increase slightly (+0.0.2 mm/day) by the end
of the century, while the timing of spring peak centroid (i.e., the timing of spring runoff) is likely
to occur earlier, and the magnitude of the 7-day low flow is projected to decrease by
4-1
-------
Table 4-1. Projections for temperature and precipitation changes in the
Northeast to 2100
Temperature
change
2-3°C by 2064
3-5°C by 2099
Annual:
2-3°C by 2064
3-5°C by 2099
Winter:
2-3°C by 2064
3-5°C by 2099
Summer
2-4°C by 2064
3-6°C by 2099
Precipitation change
5 to 8% by 2064
7 to 14% by 2099 (12 to 14% in
winter; 0 to -2% in summer)
Winter:
11 to 16% (mid-century)
20 to 30%
Summer:
Slight increases to slight
decreases
Warm season: 4% (midcentury)
7% (end of century)
Cold season: 12% (midcentury) to
27% (end of century)
Change in precipitation
frequency
Warm season: 8%
(midcentury) to 20% (end
of century)
Cold season: 4%
(midcentury) to 9% (end
of century);
Citation
Hayhoe et al.,
2007
UCS, 2006
Schoof et al.,
2010
4-11% (Hayhoe et al., 2007). While average precipitation and runoff are both projected to
increase, the frequency of droughts are also expected to increase, reflecting the minimal change
to decrease in summer precipitation (as modeled by Hayhoe et al., 2007) in combination with
higher temperatures and increased evaporation. Other projections are for decreases in snow
water equivalent and the number of snow days (Hayhoe et al., 2007) as well as for more winter
precipitation as rain instead of snow, a 25-50% decrease in length of the snow season, and an
increase in the frequency of short-term summer and fall droughts (UCS, 2006).
4.1.2. Historic Climate Trends and Climate Change Projections for Maine
Maine's interior zone has a continental climate with cold winters and warm summers,
while its coastal zone has more moderate summer and winter temperatures (Jacobson et al.,
2009). Maine is divided into three EPA Level 3 ecoregions. The Northeastern Highlands
ecoregion is located in western Maine. It is characterized by rugged hills and mountains, a
4-2
-------
mostly forested land cover, nutrient-poor soils, and numerous high-gradient streams and glacial
lakes (Omernik, 1987; U.S. EPA, 2002). The Northeastern Coastal Zone ecoregion, which is
located in the southwestern corner of Maine and has the highest population density of the
ecoregions, has land use that consists mainly of forests, woodlands, and urban and suburban
development. The Laurentian Plains and Hills ecoregion in eastern Maine is a mostly forested
region with dense concentrations of continental glacial lakes. It is less rugged than the
Northeastern Highlands (Omernik, 1987; U.S. EPA, 2002). Temperature and precipitation
patterns vary across the state. Mean annual temperatures are highest along the coast in southern
Maine (see Figure 4-1 A). Precipitation patterns vary across the state, with northern Maine
having the lowest amount of annual precipitation (see Figure 4-1B).
There has been a great deal of year-to-year variability in temperature patterns in Maine,
but overall, temperatures have been increasing over the last century. A historic trend analysis of
Maine PRISM data shows that mean annual air temperature has increased at a rate of
0.01°C/year (p < 0.01) from 1901-2000 (see Figure 4-2). Winter temperatures have increased at
the fastest rate (0.02°C/year,/> < 0.01), while fall temperatures have increased at the slowest rate
(0.004 C/year,/?-value = 0.25) (see Table 4-2, Figure 4-3). In more recent decades (1971-2000),
winter temperatures increased at an even greater rate (0.05°C/year, p = 0.13) but none of the
seasonal or annual trends from 1971-2000 are significant (p > 0.05) due to the high degree of
year-to-year variability (see Table 4-2). Future projections for mid- and late-century for high
(A2) and low (Bl) emissions scenarios are summarized in Table 4-3. Based on an ensemble
average across 15 models, mean annual air temperatures are projected to increase by up to 3.9°C
by midcentury and up to 6.1°C by the end of the century compared to a historic time period
(1961-1990). The greatest increases are projected to occur during the winter (see Table 4-3).
Precipitation patterns in Maine have been highly variable, with the direction of change
varying by the time period being evaluated. From 1901-2000, mean annual precipitation
increased at a rate of 1.10 mm/year (p = 0.01) (see Figure 4-4 and Table 4-3). Precipitation
increased across all seasons, with the greatest increase occurring in the fall (0.43 mm/year,
p = 0.04) (see Figure 4-5). In more recent decades (1971-2000), there were no significant
(p > 0.05) annual or seasonal trends due to the high degree of year-to-year variability (see
Table 4-4). From 1971-2000, winter was the only season that showed an increase
4-3
-------
Mean Temp. (C)
125 C
20
15
10
Precip (mm)
I 2500 mm
'
2000
1500
1000
500
Figure 4-1. Maine's temperature and precipitation patterns. (A) Mean
annual air temperature (°C) from 1971-2000; (B) Mean annual precipitation
(mm) 1971-2000. Map produced using the Climate Wizard Web site
(http://www.climatewizard.org/). Base climate data from the PRISM Group,
Oregon State University, http://www.prismclimate.org.
Table 4-2. Change rates in Maine PRISM mean annual air temperature
compared across two time periods: 1971-2000 versus 1901-2000. Entries in
bold text are significant (p < 0.05). Data were derived from the Climate
Wizard Web site (http://www.climatewizard.org/). Base climate data came
from the PRISM Group, Oregon State University,
http://www.prismclimate.org
Time period
1901-2000
1971-2000
Air temperature (C/yr)
Annual
0.01
0.01
DJF
0.02
0.05
MAM
0.01
0.01
JJA
0.01
0.00
SON
0.00
0.01
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, and
August; SON = September, October, and November.
4-4
-------
o
0)
3
"ro
-------
1940
B
C »-
S -
1900 1920 1940 1960 1990 KOD
YMC
1900 1920 1940 1960 1SSO MOO
D
a ~ -
I
1940 I960 1980 2000
Figure 4-3. Trends in seasonal mean air temperature in Maine from 1901-2000. (A) DJF = December, January,
and February, change rate = 0.017°C/year,/?-value < 0.01; (B) MAM = March, April, and May, change
rate = O.OFC/year,/?-value = 0.02; (C) JJA = June, July, and August, change rate = 0.008°C/year,/?-value < 0.01;
(D) SON = September, October, and November, change rate = 0.004°C/year,/>-value = 0.25. Figure produced using
Climate Wizard Web site (http://www.climatewizard.org/). Base climate data from the PRISM Group, Oregon State
University, http://www.prismclimate.org.
-------
Table 4-3. Projected departure from historic (1961-1990) trends in annual and seasonal air temperature (°C) in
Maine for mid- (2040-2069) and late-century (2070-2099) for high and low emissions scenarios. Values
represent the minimum, average, maximum, and standard deviations from 15 different climate models. Data
were derived from the Climate Wizard Web site (http://www.climatewizard.org/)
Midcentury (2040-2069) vs. historic (1961-1990)
Model
Ensemble low
Ensemble average
Ensemble high
SD
A2 (high) emissions scenario
Annual
1.6
2.7
3.9
0.6
DJF
1.4
3.3
4.4
0.7
MAM
1.4
2.6
3.9
0.8
JJA
1.3
2.3
3.4
0.7
SON
1.4
2.6
3.7
0.7
Bl (low) emissions scenario
Annual
1.1
2.1
3.1
0.6
DJF
0.7
2.6
3.5
0.7
MAM
1.0
1.9
3.0
0.7
JJA
1.1
2.0
2.7
0.6
SON
1.0
2.1
3.2
0.7
Late-century (2070-2099) vs. historic (1961-1990)
Ensemble low
Ensemble average
Ensemble high
SD
2.8
3.6
6.1
1.0
2.6
4.3
6.9
1.2
2.3
3.4
6.5
1.3
2.5
3.2
5.3
1.0
2.9
3.6
6.0
1.0
1.3
2.8
4.3
0.8
0.5
3.4
4.7
1.0
1.1
2.5
4.0
0.9
1.5
2.6
3.7
0.8
1.6
2.7
4.7
1.0
DJF = December, January, and February; MAM = March, April and May; JJA = June, July and August and SON = September, October, and
November.
-------
o
o
o
o
CO
o °
~ CM
-
o
0) o
Q- 2
TO
o
o
o
o
oo
1900
1920
I I
1940 1960
Year
1980
2000
Figure 4-4. Trends in annual mean precipitation in Maine from 1901-2000.
Change rate = 1.103 mm/year, p-va\ue = 0.02. Figure produced using Climate
Wizard Web site (http://www.climatewizard.org/). Base climate data from the
PRISM Group, Oregon State University, http://www.prismclimate.org.
4-8
-------
A §
i
i
* i
CL i
S I-I
S
1900 1920 1940 I960 1980 2000
tt*
1
S 8-
I
1900 1WO 1940 1960 1980 2000
mm
i
1900 1910 1940 -55; 1980 MOO
D
I $
I
I.
1900 1930
1930 2000
Figure 4-5. Trends in seasonal mean precipitation in Maine from 1901-2000. (A) DJF = December, January, and
February, change rate = -0.207 mm/year,/?-value = 0.35; (B) MAM = March, April, and May, change
rate = 0.29 mm/year, p-va\ue = 0.18; (C) JJA = June, July, and August, change rate = 0.182 mm/year, p-va\ue = 0.29;
(D) SON = September, October, and November, change rate = 0.432 mm/year, p-va\ue = 0.04. Figure produced using
Climate Wizard Web site (http://www.climatewizard.org/). Base climate data from the PRISM Group, Oregon State
University, http://www.prismclimate.org.
-------
Table 4-4. Change rates in Maine PRISM mean annual precipitation
compared across two time periods: 1971-2000 versus 1901-2000. Entries in
bold text are significant (p < 0.05). Data were derived from the Climate
Wizard Web site (http://www.climatewizard.org/). Base climate data came
from the PRISM Group, Oregon State University,
http://www.prismclimate.org
Time period
1901-2000
1971-2000
Precipitation (mm/yr)
Annual
1.10
-1.13
DJF
0.21
-0.90
MAM
0.29
-0.07
JJA
0.18
-0.95
SON
0.43
0.78
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, and
August; SON = September, October, and November.
(0.78 mm/year), while annual and other seasonal precipitation patterns decreased (see Table 4-4).
Table 4-5 summarizes future projections for mid- and late-century for high (A2) and low (Bl)
emissions scenarios. The future projections are highly variable across models and emissions
scenarios. Under the high emissions scenario, the ensemble average projects that mean annual
precipitation will increase by 90.1 mm by midcentury and 125 mm by the end of the century
compared to a historic time period (1961-1990). Under the high emissions scenario, the greatest
changes are projected to occur during the winter (see Table 4-5).
4.2. DATA INVENTORY AND PREPARATION
Data for Maine were obtained from the Maine Department of Environmental Protection
(DEP). Our Maine EDAS database contains data for 1,459 biological samples (which typically
consist of 3 replicates) from 742 unique stations, with sampling dates ranging from 1974 to
2006. A mix of habitat, water chemistry, and in situ measurements are available for many of the
sites. The parameters that were most consistently reported include instantaneous water
temperature, conductivity, pH, DO, width, depth, and visual substrate estimates. Some
additional water chemistry data (i.e., nitrogen, phosphorus, total suspended solids, some metals)
became available after 2000. Most biological sampling sites in Maine have fewer than 5 years of
data, but six sites have 10 or more years of data (see Table 4-6). Two of these sites have
received Maine DEP's highest biological condition rating (Class A), which is described in more
detail in Section 4.3. Figure 4-6 shows the spatial distribution of biological sampling sites.
4-10
-------
Table 4-5. Projected departure from historic (1961-1990) trends in annual and seasonal precipitation (mm) in
Maine for mid- (2040-2069) and late-century (2070-2099) for high and low emissions scenarios. Values
represent the minimum, average, maximum, and standard deviations from 15 different climate models. Data
were derived from the Climate Wizard Web site (http://www.climatewizard.org/)
Midcentury (2040-2069) vs. historic (1961-1990)
Model
Ensemble low
Ensemble average
Ensemble high
SD
A2 (high) emissions scenario
Annual
34.4
90.1
151.3
35.7
DJF
0.1
37.3
68.8
18.3
MAM
-13.0
24.6
55.2
20.6
JJA
-19.8
15.1
65.7
22.6
SON
-20.1
12.0
58.8
21.8
Bl (low) emissions scenario
Annual
-80.5
69.0
179.4
63.2
DJF
3.6
33.5
59.4
13.8
MAM
-9.7
23.7
63.2
19.8
JJA
-167.7
-4.7
46.1
53.2
SON
-78.1
9.6
59.8
32.3
Late-century (2070-2099) vs. historic (1961-1990)
Ensemble low
Ensemble average
Ensemble high
SD
17.0
125.0
204.7
54.6
22.2
57.5
113.0
23.2
-17.9
43.9
93.1
27.6
-50.5
7.7
116.2
39.0
-32.5
18.5
61.6
26.5
-194.9
67.0
182.7
100.0
9.0
45.6
70.8
18.5
-16.6
31.7
61.9
23.9
-151.8
-8.3
55.7
62.1
-130.7
-5.1
52.3
50.4
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, August and SON = September, October, and
November.
-------
Stations (Ref Status, Nurn Years)
(N=742)
• Rel. 2Q+(N=1»
• Ref. 10-19 (N=1)
V Rtf, 5-9
-------
Table 4-6. Distribution of stations that have received Class A biological
condition ratings and total stations, categorized by duration of sampling
Years
sampled
Ito4
5 to 9
>10
Total
Maine
Class A
210
10
2
222
Total
696
40
6
742
The original Maine data set included samples collected throughout the year using
different methods. To minimize variability associated with collection method, we only analyzed
samples collected using rock baskets or rock cones. Maine DEP typically deploys three rock
baskets or cones per sample. Each rock basket is considered to be a replicate. When calculating
metrics, we averaged values across replicates to come up with a single metric value per sample.
To account for seasonal variability, we excluded samples that were collected during the winter
and spring, because these are outside Maine DEP's normal sampling period. We also considered
differences in subsampling efforts, as these can affect richness metrics, but our ability to do so
was limited because these data were not reported consistently across samples.
We used a genus-level OTU when preparing the biological data for long-term trend
analyses. Per the methods described in Section 2.1.3, we used NMDS analyses to verify the
OTU. We looked for trends associated with changes in biological condition/class, EPA Level 3
ecoregion, year (in 5, 10, and 20-year increments), and taxonomy lab and found no obvious
groupings (see Appendix A, Figures A-7 through A-14). There was, however, a subtle shift
towards finer taxonomic resolution from the early 1980s to the present (as one would assume due
to improved taxonomic keys, etc.), along with an increase in species-level identifications for
certain orders in 1990-1991.
This was particularly evident for the order Trombidiformes (water mites). Water mites
were identified to the suborder level (Prostigmata) prior to 1991, but from 1991 onwards, there
were 28 different identifications associated with the water mites, with some to the species-level.
To account for this, we grouped all taxa from the Order Trombidiformes into the suborder
Prostigmata. An increase in taxonomic resolution for Chironomidae was also evident in
4-13
-------
1990-1991. We considered grouping all Chironomidae to the family level, but decided that this
would result in the loss of too much information, and that the trends associated with this
taxonomic change were not consistent enough to warrant the change.
We also noticed a subtle change in the data in 1999. This was likely due to variability
among the taxonomic labs, because four new labs started doing taxonomic identifications for
Maine during the year. Over the 26-year period during which Maine DEP has collected
biological data, they have used 16 different taxonomy labs. The number of samples processed
by each lab varies; some labs have processed fewer than 10 samples, and others have processed
more than 100. Although we did see some variability associated with taxonomy lab, the patterns
were not clear or consistent, and the OTU "fix" resolved most of the observed differences, so no
adjustments were made.
4.3. MAINE DEP METHODS
Maine DEP typically collects macroinvertebrate samples from wadeable streams using
rock baskets and rock cones that are deployed in riffles or runs for 4 weeks from July-September
(Davies and Tsomides, 2002). Based on Maine's water classification system, rivers and streams
are divided into four classes: (1) Class A, in which aquatic life is as naturally occurs; (2) Class B,
in which there are no detrimental changes in the resident biological community, and all
indigenous species are maintained; (3) Class C, in which the structure and function of the
resident biological community is maintained; and (4) nonattainment (NA), in which minimum
aquatic life use criteria are not met.
Maine DEP uses four linear discriminant models to assign samples to classes based on
biological condition. The same models are applied to all sites. Each of the four models uses
different variables and provides independent estimates of class membership. The first model acts
as a screen and provides four initial probabilities that a given site attains a given class. Then data
are run through three subsequent models in hierarchical order (C or Better Model; B or Better
Model; and A Model) before coming to a final model determination, which is then reviewed by
Maine DEP. Table 4-7 shows a list of the input metrics used in each model. Appendix C
provides a more detailed explanation of Maine DEP's process for determining attainment class.
4-14
-------
4.4. INDICATORS
4.4.1. Thermal Preference
As described in Section 2.2.1, we used the guidelines of Yuan (2006) to calculate thermal
optima and tolerance values. For the Maine data set, we based our calculations on a subset of
616 samples that were collected from July-September. Lists of cold and warm-water taxa for
the Maine data set were developed based on these data, as well as from literature and input from
the New England regional advisory group. These lists are the basis of the region-specific
thermal-preference richness and relative-abundance metrics used in some analyses.
The Maine cold-water taxa list is composed of 41 taxa, and the warm-water taxa list is
composed of 40 taxa. Tables 4-8 and 4-9, respectively, lists the cold and warm-water taxa, along
with abundance and distribution information6. Sixteen of the cold-water taxa are Plecopterans,
10 are Trichopterans, 7 are Dipterans, and 4 are Ephemeropterans (see Table 4-8). Ten of the
warm-water taxa are Dipterans, 9 are Ephemeropterans, and 6 are Trichopterans (see Table 4-9).
The most abundant cold-water taxa are Leuctra (P\ecopteran),Epeoms (Ephemeropteran),
Eurylophella (Ephemeropteran), Perlodidae (Plecopteran), and Boyeria (Odonata). These taxa
comprise only 0.3 to 0.4% of the total individuals in the Maine database. Thirty-one of the cold-
water taxa have overall abundances of less than 0.1% and occur at less than 10% of the sites.
Boyeria occurs at the largest percentage of sites (38%), followed by Perlodidae, which occurs at
25% of the sites. Two of the taxa on the cold water list, Eurylophella and Glossosoma, are on
Maine DEP's Class A indicator list.
Stenonema and Neureclipsis are the most abundant warm-water taxa, with overall
abundances of 5.2 and 2.6%, respectively. Nine of the warm-water taxa have overall abundances
of less than 0.1%. Stenonema occurs at the highest percentage of sites (63%), followed by
Acroneuria (39%) and Neureclipsis (38%). Eight of the warm-water taxa occur at less than 10%
of the sites. Three of the taxa on the warm water list, Paragnetina, Serratella, and Leucrocuta,
are on Maine DEP's Class A indicator list.
6There are some noteworthy genera that were excluded from the Maine cold and warm water lists due to variations
in thermal preferences among species within these genera. These included Eukiefferiella and Rhyacophila from the
cold water list, and Brachycentrus, Hydropsyche, and Ceratopsyche from the warm water list. We also considered
including Antocha and Dicranota on the cold water list based on results from the weighted average inferences but
excluded them because they occurred not only at sites with cold temperatures, but also at sites which had the
warmest average water temperatures.
4-15
-------
Table 4-7. Metrics that are used in Maine DEP's four linear discriminant
models
Model
First stage
C or better
B or better
A model
Metric
Total abundance
Generic richness
Plecoptera abundance
Ephemeroptera abundance
Shannon-Wiener Generic Diversity
Hilsenhoff Biotic Index
Relative abundance chironomidae
Relative richness Diptera
Hydropsyche abundance
Probability (A + B + C) from First Stage Model
Cheumatopsyche abundance
EPT generic richness divided by diptera generic richness
Relative abundance Oligochaeta
Probability A + B from First Stage Model
Perlidae abundance
Tanypodinae abundance
Chironomini abundance
Relative abundance Ephemeroptera
EPT generic richness
Sum of mean abundances ofDicrotendipes,Micropsectra, Parachironomus, and
Helobdella
Probability A from First Stage Model
Relative generic richness Plecoptera
Sum of mean abundances of Cheumatopsyche , Cricotopus, Tanytarsus, and Ablabesmyia
Sum of mean abundances ofAcroneuria and Stenonema
Ratio Ephemeroptera and Plecoptera generic richness
Ratio of Class A indicator taxa (Brachycentrus, Serratella, Leucrocuta, Glossosoma,
Paragnetina, Eurylophella, and Psilotreta)
4-16
-------
Table 4-8. List of Maine cold-water temperature indicator taxa, sorted by order, family, then Final ID.
Distribution and abundance information is also included. Sum_Individuals = the total number of individuals
from that taxon in the Utah database; Pct_Abund = percentage of total individuals in the database composed of
that taxon; Num_Stations = number of stations in the database that the taxon occurred at;
Pct_Stations = percentage of stations in the database at which the taxon occurred. Two of the taxa, Malirekus
and Taenionema, do not occur in the Maine database, and were added based on feedback from the regional
advisory group
Order
Coleoptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Megaloptera
Odonata
Odonata
Plecoptera
Plecoptera
Family
Elmidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Ameletidae
Ephemerellidae
Heptageniidae
Heptageniidae
Corydalidae
Aeshnidae
Gomphidae
Capniidae
Capniidae
Final ID
Oulimnius
Heterotrissocladius
Larsia
Macropelopia
Natarsia
Pagastia
Prodiamesa
Pseudodiamesa
Ameletus
Eurylophella
Epeoms
Rhithrogena
Nigronia
Boyeria
Lanthus
Capnia
Paracapnia
Sum_individs
237
447
269
322
430
420
392
139
63
1,785
2,132
193
713
1,761
36
71
52
Pct_abund
0.04
0.08
0.05
0.05
0.07
0.07
0.07
0.02
0.01
0.3
0.36
0.03
0.12
0.3
0.01
0.01
0.01
Num_stations
37
73
58
43
65
96
28
12
26
170
172
23
170
321
11
5
17
Pct_stations
4.36
8.6
6.83
5.06
7.66
11.31
3.3
1.41
3.06
20.02
20.26
2.71
20.02
37.81
1.3
0.59
2
-------
Table 4-8. List of Maine cold-water temperature indicator taxa, sorted by order, family, then Final ID.
Distribution and abundance information is also included. Sum_Individuals = the total number of individuals
from that taxon in the Utah database; Pct_Abund = percentage of total individuals in the database composed of
that taxon; Num_Stations = number of stations in the database that the taxon occurred at;
Pct_Stations = percentage of stations in the database at which the taxon occurred. Two of the taxa, Malirekus
and Taenionema, do not occur in the Maine database, and were added based on feedback from the regional
advisory group (cont.)
Order
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Trichoptera
Trichoptera
Trichoptera
Family
Capniidae
Chloroperlidae
Chloroperlidae
Leuctridae
Nemouridae
Nemouridae
Nemouridae
Nemouridae
Peltoperlidae
Peltoperlidae
Perlodidae
Perlodidae
Pteronarcyidae
Taeniopterygidae
Apataniidae
Brachycentridae
Glossosomatidae
Final ID
Utacapnia
Sweltsa
Utaperla
Leuctra
Nemoura
Paranemoura
Prostoia
Zapada
Peltoperla
Tallaperla
Malirekus
Perlodidae
Pteronarcys
Taenionema
Apatania
Micrasema
Glossosoma
Sum_individs
71
640
2
2,407
17
3
6
2
9
126
0
1,775
248
0
48
405
945
pct_abund
0.01
0.11
0
0.4
0
0
0
0
0
0.02
0
0.3
0.04
0
0.01
0.07
0.16
Num_stations
3
66
2
142
4
3
1
1
4
12
0
212
80
0
23
87
119
Pct_stations
0.35
7.77
0.24
16.73
0.47
0.35
0.12
0.12
0.47
1.41
0
24.97
9.42
0
2.71
10.25
14.02
oo
-------
Table 4-8. List of Maine cold-water temperature indicator taxa, sorted by order, family, then Final ID.
Distribution and abundance information is also included. Sum_Individuals = the total number of individuals
from that taxon in the Utah database; Pct_Abund = percentage of total individuals in the database composed of
that taxon; Num_Stations = number of stations in the database that the taxon occurred at;
Pct_Stations = percentage of stations in the database at which the taxon occurred. Two of the taxa, Malirekus
and Taenionema, do not occur in the Maine database, and were added based on feedback from the regional
advisory group (cont.)
Order
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Family
Hydropsychidae
Hydropsychidae
Hydroptilidae
Limnephilidae
Limnephilidae
Limnephilidae
Phryganeidae
Final ID
Diplectrona
Parapsyche
Palaeagapetus
Hydatophylax
Limnephilus
Psychoglypha
Oligostomis
Sum_individs
1,137
398
1
114
889
329
485
pct_abund
0.19
0.07
0
0.02
0.15
0.06
0.08
Num_stations
47
27
1
49
62
37
87
Pct_stations
5.54
3.18
0.12
5.77
7.3
4.36
10.25
VO
-------
Table 4-9. List of Maine warm-water temperature indicator taxa. Distribution and abundance information is
also included Sum_Individuals = the total number of individuals from that taxon in the Utah database;
Pct_Abund = percentage of total individuals in the database composed of that taxon; Num_Stations = number of
stations in the database that the taxon occurred at; Pct_Stations = percentage of stations in the database at
which the taxon occurred
Order
Arhynchobdellida
Basommatophora
Basommatophora
Basommatophora
Basommatophora
Coleoptera
Decapoda
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Family
Erpobdellidae
Ancylidae
Physidae
Physidae
Planorbidae
Elmidae
Cambaridae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Empididae
Baetidae
Baetidae
Caenidae
Final ID
Erpobdella
Ferrissia
Physa
Physella
Helisoma
Stenelmis
Orconectes
Cardiocladius
Dicrotendipes
Labrundinia
Nilotanypus
Parachironomus
Pentaneura
Psectrocladius
Rheopelopia
Tribelos
Hemerodromia
Plauditus
Pseudocloeon
Caenis
Sum_individs
265
594
1,373
1,681
716
2,638
381
200
1,978
618
413
946
881
1,693
729
1,781
1,764
1,285
1,147
1,783
Pct_abund
0.04
0.1
0.23
0.28
0.12
0.44
0.06
0.03
0.33
0.1
0.07
0.16
0.15
0.28
0.12
0.3
0.3
0.22
0.19
0.3
Num_stations
65
102
115
155
66
280
99
52
169
137
133
83
139
161
144
78
260
125
113
169
Pct_stations
7.66
12.01
13.55
18.26
7.77
32.98
11.66
6.12
19.91
16.14
15.67
9.78
16.37
18.96
16.96
9.19
30.62
14.72
13.31
19.91
to
o
-------
Table 4-9. List of Maine warm-water temperature indicator taxa. Distribution and abundance information is
also included. Sum_Individuals = the total number of individuals from that taxon in the Utah database;
Pct_Abund = percentage of total individuals in the database composed of that taxon; Num_Stations = number
of stations in the database that the taxon occurred at; Pct_Stations = percentage of stations in the database at
which the taxon occurred (cont.)
10
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Haplotaxida
Hoplonemertea
Hydroida
Mesogastropoda
Odonata
Plecoptera
Plecoptera
Plecoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Family
Ephemerellidae
Heptagenidae
Heptageniidae
Heptageniidae
Isonychiidae
Leptohyphidae
Naididae
Tetrastemmatidae
Hydridae
Hydrobiidae
Coenagrionidae
Perlidae
Perlidae
Perlidae
Helicopsychidae
Hydropsychidae
Hydroptilidae
Leptoceridae
Leptoceridae
Polycentropodidae
Final ID
Serratella
Stenonema
Leucrocuta
Stenacron
Isonychia
Tricorythodes
Chaetogaster
Prostoma
Hydra
Amnicola
Argia
Acroneuria
Attaneuria
Paragnetina
Helicopsyche
Macrostemum
Hydroptila
Ceraclea
Oecetis
Neureclipsis
Sum_individs
2,534
30,768
3,320
6,503
5,413
2,655
342
267
483
4,589
869
4,857
172
625
2,563
4,557
1,799
876
3,390
15,523
Pct_abund
0.43
5.18
0.56
1.09
0.91
0.45
0.06
0.04
0.08
0.77
0.15
0.82
0.03
0.11
0.43
0.77
0.3
0.15
0.57
2.61
Num_stations
191
536
208
196
225
205
70
61
113
160
137
331
36
103
104
168
189
152
306
320
Pct_stations
22.5
63.13
24.5
23.09
26.5
24.15
8.24
7.18
13.31
18.85
16.14
38.99
4.24
12.13
12.25
19.79
22.26
17.9
36.04
37.69
-------
Many of the taxa on the cold water list are intolerant to enrichment. There is an even
distribution of warm-water taxa across enrichment tolerance categories (see Figure 4-7).
Because of this overlap, it is difficult to tease out whether organisms are responding to changes
associated with warming temperatures or whether they are responding to other stressors, such as
enrichment.
18
16
14
CD
X -IT
CD "
O 10
Q 8
3 6
Z
4
2 •
0
Dcold
i warm
UJ
Intolerant Intermediate Tolerant
Enrichment Tolerance
Figure 4-7. Relationship between Maine cold and warm-water-preference
taxa and Maine enrichment tolerance scores. Taxa with enrichment tolerance
scores of 0-3 were categorized as Intolerant; those with scores of 4-6 were
Intermediate and those with scores of 7-10 were Tolerant.
4.4.2. Hydrologic Indicators
We attempted to develop a list of candidate taxa in Maine that could potentially serve as
indicators of hydrologic change. We matched USGS gages with biological sampling sites per
the methods described in Section 2.2.2. There were not enough USGS gages associated with
biological sampling sites to run analyses that have statewide applicability. It is also worth noting
that about the half of the USGS gages that did match with biological sampling sites were located
in close proximity to dams and, thus, had regulated flows. We did run some analyses on
matched biological-hydrological data from a biological sampling site on the Sheepscot River
(Station ME 56817) that had over 20 years of continuous biological data. Based on correlation
analyses from this site, Hydropsyche (spotted sedge caddisfly), Promoresia (an elmid beetle),
and Rhyacophila (green sedge caddisfly) were significantly and positively correlated with 1- and
4-22
-------
3-day minima flow values. There were a mix of other significant associations as well, but results
were difficult to interpret and were too limited to draw statewide conclusions.
4.4.3. Traits-Based Indicators in a Warmer Drier Scenario
We developed a list of taxa that may be most and least sensitive to projected changes in
temperature and streamflow based on the suite of trait modalities described in Section 2.2.3.
When assessing sensitivity to future climatic changes, we focused on a generalized scenario in
which temperatures are increasing and flows are decreasing during the low flow periods when
state biomonitoring programs typically collect their samples. The taxa in Table 4-10 that are
deemed most sensitive, or most likely to be adversely affected by these projected climatic
changes, are all EPT taxa. The least sensitive taxa on our list is a Hemipteran, Belostoma, which
has the ability to exit (as adults), has high dispersal ability, strong flying strength, strong
swimming ability, and breathes through plastron-spiracles.
4.5. LEAST DISTURBED LONG-TERM BIOLOGICAL MONITORING SITES
Maine does not have a formal statewide long-term reference network. We explored grouping
sites that had received Class A biological condition ratings7 together to create a data set
statewide data set to analyze for long-term trends, but site-specific differences were evident
within the data set and the sample size was relatively low; therefore, we focused on individual
sites. We performed trend analyses on data from three sites that had received Class A biological
condition ratings and that had the longest term biological data. Figure 4-8 shows the locations of
these stations. Table 4-11 summarizes site characteristics. All three sites are located in the
Laurentian Plains and Hills ecoregion. Anthropogenic influences are higher than desired
(>5% urban and/or >10% agricultural) at all three sites, but data were analyzed from these sites
because they represented the best-available long-term data in the state database. Table 4-12 lists
the time periods for which biological data are available for these sites. Data used in these
analyses were limited to rock basket samples collected from July-September.
7It should be noted that sites that have received Class A biological condition ratings are not necessarily considered to
be reference sites by Maine DEP. At the time of this report, Maine DEP was in the process of developing strict
reference criteria based on considerations such as land use and land cover in the upstream catchment and proximity
to NPDES discharges.
4-23
-------
Table 4-10. List of taxa that may be most and least sensitive to a warmer and
drier future scenario based on a combination of traits
Order
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Hemiptera
Family
Apataniidae
Goeridae
Calamoceratidae
Limnephilidae
Limnephilidae
Hydropsychidae
Hydropsychidae
Limnephilidae
Belostomatidae
Final ID
Apatania
Goera
Heteroplectron
Onocosmoecus
Pycnopsyche
Diplectrona
Parapsyche
Psychoglypha
Belostoma
Sensitivity to warmer drier scenario
most
most
most
most
most
most
most
most
least
4-24
-------
it ates muh ihe longest continuous data Oct07 DRAFT Ecof egions in Maine
LEVEL3.NAM
| MwWNew Brunswick Rains and Hilts
f Norffieastem Coastal Zone
| Northeasstn Highlands
Figure 4-8. Locations of the three biological sampling sites that we
performed long-term trend analyses on (56817 = Sheepscot; 57011 = West
Branch Sheepscot; 57065 = Duck Brook).
4-25
-------
Table 4-11. Site characteristics for the long-term biological monitoring stations in Maine. Percentage urban
and percentage agricultural (ag) apply to a 1-km buffer zone around each site and are based on 2001 National
Land Cover Data
Site ID
ME 56817
ME 570 11
ME 57065
Water body
Sheepscot
W. Br. Sheepscot
Duck
Longitude
(°)
69.59334
69.53129
68.23461
Latitude
(°)
44.2232
44.3679
44.3934
EPA Level 3 ecoregion
Laurentian Plains and Hills
Laurentian Plains and Hills
Laurentian Plains and Hills
Elevation (m)
31.6
70.1
54.6
Drainage
area (km2)
362.8
38.1
12.8
% Urban
16.4
9.1
15.9
%Ag
23
18.5
0
to
Table 4-12. Time periods for which biological data were available at the long-term monitoring sites in Maine.
Data used in these analyses were limited to July-September rock basket samples
Site ID
ME 56817
ME 570 11
ME 57065
Water body
Sheepscot
West Branch Sheepscot
Duck
Number of years of
data analyzed
23
12
9
Years
1984-2006
1995-2006
1997-2005
-------
4.6. EVIDENCE OF TRENDS AT LEAST DISTURBED LONG-TERM MONITORING
SITES
4.6.1. Sheepscot River (ME 56817)
The Sheepscot River site (ME 56817; Maine DEP Station 74) is located in southern
Maine in the town of Whitefield. It is in the Laurentian Plains and Hills ecoregion and Central
Interior biophysical region, has a drainage area of 362.8 km2 and an elevation of 31.6 m. Its
highest maximum monthly temperatures occur during July and August, and its lowest average
flows (<85 cfs) occur from July through September. This station has 23 years of continuous
biological data, spanning from 1984 to 2006, that have been collected during Maine DEP's July
through September index period. We gathered daily temperature and precipitation data from
1949 to 2010 from the Augusta FAA AP weather station (SitelD 170275, Latitude: 44.3206,
Longitude: 69.7972), which is located approximately 20 km northwest of the biological sampling
site. Flow data from 1939-2009 were gathered from USGS gage 1038000 (Sheepscot River at
North Whitefield, Latitude: 44.22278, Longitude: 69.59389), which is colocated with the
biological sampling site. Figure 4-9 shows an aerial photograph of the site, along with the
weather station and active USGS gage.
4.6.1.1. Temporal Trends in Climatic and Biological Variables
Since 1949, mean annual air temperatures at the weather station closest to the Sheepscot
River (ME 56817) site have ranged from 6 to 9.8°C. There is a great deal of year-to-year
variability, but overall, temperatures have been increasing over time (when fit with a linear trend
r\
line, r = 0.06, p = 0.05) (see Figure 4-10). When PRISM air temperature data are compared to
observed data, PRISM data are within 1°C of the observed values, and there is good
correspondence between patterns. Mean annual flow and mean annual precipitation patterns
have been highly variable over time (see Figure 4-11). Since 1939, mean annual flow values
have ranged from 110.1 to 485.7 cfs (when fit with a linear trend line, r2 = 0.08, p = 0.02).
Precipitation patterns generally show good correspondence with flow patterns but are more
variable (see Figure 4-11).
In addition to mean annual values, mean maximum July and August temperature and
mean July-September flow values were also evaluated, as these are likely to be physiologically
stressful time periods for the biological organisms, and also correspond with Maine DEP's index
period. During the period of biological record (1984-2006), mean maximum July/August air
4-27
-------
Figure 4-9. Locations of the Sheepscot River (ME 56817) biological sampling
site, USGS gage 1038000 (Sheepscot River at North Whitefield) and Augusta
FAA AP weather station. Image from Google Earth.
4-28
-------
10.0
9.5
p 9.0
I "
CD
O.
E 8.0
?
I 7.5
TO
| 7.0
c
ra
I 6.5
6.0
5.5
Observed
- PRISM
1949
1959
1969
1979
1989
1999
2009
Figure 4-10. Yearly trends in annual observed air temperature (°C) at the
Sheepscot River site (ME 56817) from 1949-2010, based on data from the
Augusta FAA AP weather station. For comparative purposes, PRISM annual
air temperature data associated with the biological sampling site are also included
from 1975-2005. The area shaded in grey corresponds to the period of biological
record. When the observed data are fitted with a linear trend line, r2 = 0.06,
p = 0.05, and.y = -12.718 + 0.0102 x x.
4-29
-------
700
600
500
-2 400
300
200
100
Flow
- Precipitation
i
1500
1400
1300
E
E,
c
g
'
1200 |_
'o
CD
1100 £
1000
900
800
700
03
ZJ
c:
CD
CD
CD
CO
.0
O
600
1939
1949
1959
1969
1979
1989
1999
2009
Figure 4-11. Yearly trends in mean annual flow (cfs) at the Sheepscot River
site (ME 56817) from 1939-2009, based on data from USGS gage 1038000
(Sheepscot River at North Whitefield). For comparative purposes, observed
annual precipitation data from the Augusta FAA AP weather station are also
included from 1949-2009. The area shaded in grey corresponds to the period of
biological record. When the observed data are fitted with a linear trend line,
r2= O.OS,p = 0.02, and.y = -1,814.9037 + 1.0494 x x.
temperatures ranged from 24.2-27.8°C, and mean July-September flow values ranged from 17.0
to 156.7 cfs (see Table 4-13). Attainment classes based on biological condition have ranged
from Class A to B. Since 1998, samples have attained Class A status (see Figure 4-12 A). The
number of EPT taxa has varied, but overall, numbers have increased over time, ranging from 5 in
1984 to 19 in 2005 (see Figure 4-12B). HBI scores have varied from year to year and did not
show a clear trend, ranging from a low of 3.1 in 1995 to a high of 4.7 in 1984 (see
Figure 4-12B). During the period of biological record, mean maximum July/August air
temperatures and July-September flows were highly variable, with the highest maximum
July/August temperature occurring in 1999, the lowest July-September flows occurring in 1994
4-30
-------
and 2000, and the highest July-September flows occurring in 2005 (see Figure 4-12C). The
number of warm-water taxa has varied, but overall, numbers have increased over time, ranging
from 5 in 1984 to 12 in 1999 and 2001 (see Figure 4-13A). Warm-water taxa have comprised up
to 48% of the assemblage (see Figure 4-13B). Very few cold-water taxa were present at this site,
with richness values ranging from 0 to 2.
Per communications with Maine DEP, conditions at this site have been influenced by
nonpoint source pollution, with potential anthropogenic influences from urban and agricultural
land use (16% urban and 23% agricultural within a 1-km buffer). Some recent (post-2000) water
chemistry data are available for various nutrient-related parameters. The maximum nitrogen
concentration was 0.48 mg/L, the maximum ammonia concentration ammonia was 0.05 mg/L,
and the maximum total phosphorus concentration was 0.02 mg/L. Confounding factors related
to in situ measurements were not evident, with values in the following ranges:
• DO: 7.2 to 8.5 mg/L
• pH: 6.4 to 7.1
• Specific conductance: 37 to 84 |imho/cm
Table 4-13. Range of temperature, precipitation, and flow values that
occurred at the Sheepscot River site (ME 56817) during the period of
biological record
Parameter
Year
Observed mean annual air temperature (°C)
PRISM mean annual air temperature (°C)
Observed mean maximum July/August Air temperature (°C)
Mean annual flow (cfs)
Mean July-September flow (cfs)
Observed mean annual precipitation (mm)
PRISM mean annual precipitation (mm)
Min
1984
6.6
6.6
24.2
132.6
17.0
661.3
795.5
Max
2006
9.6
8.6
27.8
485.7
156.7
1,461.8
1,691.2
4-31
-------
B
D-
LU
2?
1
I
I
CO
CD
X
03
5
20
18
16
14
12
10
8
6
4
28
27
26
25
24
-•- EPT taxa
-o- HBI
Temperature
Flow
4.8
4.6
4.4
4.2
4.0 .
00
3.8 ^
3.6
3.4
3.2
3.0
180
160 _
140 ~
o
120 f
100 E
-------
B
14
12
10
8
6
4
2
0
C 28
O
27
Q.
26
en
25
CD
24
-•- Cold Water
-E>- Warm Water
a
=-a
a a
-•- Temperature
-•-• Flow
Ifl
OJ
3
TJ
C
0^-
40
30
20
10
-•- Cold Water ;•
-o- Warm Water
m \
' \ •' \ 0 CJ / >
/ \ / ( / \ ' \
• / ^ 'fe/B\.a'a
ri -/^
f
I
180
160
140
120
CD
100 E
80 g-
CO
60 f"
40
20
0
c
03
(1)
1984
1989
1994
1999
2004
Figure 4-13. Yearly trends at the Sheepscot River site (ME 56817) in
(A) number of cold and warm-water taxa; (B) percentage cold and warm-
water individuals; and (C) mean maximum July/August temperature (°C)
and mean July-September flow (cfs).
4-33
-------
4.6.1.2. Associations Between Biological and Climatic Variables
Kendall tau nonparametric correlations analyses allow examination of associations
between commonly used biological metrics, year, temperature, flow, and precipitation variables
at the Sheepscot River (ME 56817) site. None of the 13 commonly used biological metrics
showed strong associations (r > 0.5) with the environmental parameters (see Table 4-14). Four
of the metrics (total number of taxa, number of EPT taxa, number of Trichoptera taxa, and
Shannon-Wiener Diversity Index) showed strong positive associations with year. The number of
warm-water taxa metric also showed a strong positive association with year (see Table 4-15), but
none of the thermal preference metrics were strongly correlated with the environmental
parameters. The subset of biological metrics that have shown responsiveness to hydrologic
variables in other studies (see Section 2, Table 2-6) failed to show strong relationships with the
flow and precipitation variables (see Table 4-16). Four of the metrics (number of
collector-filterer taxa, number of scraper/herbivore taxa, number of erosional taxa, and
percentage scraper/herbivore individuals) had strong positive associations with year.
4.6.1.3. Groupings Based on Climatic Variables
Samples were partitioned into hottest/coldest/normal year groups and
lowest/normal/highest flow year groups. At the Sheepscot River site (ME 56817), on average,
the hottest years were 1.4°C warmer than the coldest years, and highest flow years had 168 more
cubic feet per second than lowest flow years. When samples were grouped based on
temperature, there were no significant (p > 0.05) differences between mean metric values (see
Table 4-17). Although not significant, there are some patterns worth noting. Mean numbers of
total taxa, EPT taxa and warm-water taxa, and individuals were highest in hottest year samples.
On average, cold water metrics were also higher in the hottest year samples, but so few
cold-water taxa were present that these results should be interpreted with caution. When samples
were grouped based on flow, there were also no significant (p > 0.05) differences between mean
metric values (see Table 4-18). The mean number of warm-water taxa and individuals was
highest in the driest flow year samples, and the cold-water taxa were more prevalent in the
wettest flow year samples, but as mentioned, the cold-water taxa metrics should be interpreted
with caution. NMDS was used to evaluate differences in taxonomic composition across the
4-34
-------
Table 4-14. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics, year, and climatic variables at the Sheepscot River site (ME 56817). Results
are based on 23 years of data. Entries are in bold text if r > ±0.5. Ranges of biological metric values are also
included
Biological metric
Total no. taxa
No. EPTtaxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. Intolerant taxa
Percentage EPT individuals
Percentage Ephemeroptera
individuals
Shannon-Wiener Diversity
Index
Percentage noninsect individuals
Percentage dominant taxon
Percentage tolerant individuals
Hilsenhoff Biotic Index
Range of
metric values
Min
9.7
5.2
1.7
0.5
3.0
3.0
45.6
3.9
1.8
0.0
12.8
0.3
3.1
Max
29.8
18.7
7.3
1.7
11.0
12.8
84.3
35.2
4.3
3.6
57.8
15.8
4.7
r values (based on Kendall Tau correlations)
Year
0.63
0.58
0.43
0.09
0.62
0.48
0.04
0.30
0.54
0.32
-0.36
-0.06
-0.18
Air Temperature (°C)
PRISM mean
annual
0.18
0.14
0.08
0.08
0.15
0.14
0.11
0.23
0.13
-0.05
0.00
-0.01
-0.11
Observed mean
maximum
July/August
0.24
0.19
0.22
0.12
0.11
0.11
0.16
0.31
0.20
0.14
-0.05
0.13
-0.06
Flow (cfs)
Mean
annual
0.07
0.01
0.04
0.02
0.04
0.20
-0.11
0.10
0.07
-0.06
0.00
-0.22
-0.10
Mean
July-September
0.02
-0.03
0.02
0.06
-0.03
-0.04
0.11
0.01
0.01
-0.09
-0.05
-0.04
-0.15
PRISM mean
annual
precipitation
(mm)
0.13
0.10
0.16
0.02
0.03
0.26
-0.01
0.17
0.12
-0.04
0.01
-0.15
-0.14
-------
Table 4-15. Kendall tau nonparametric correlations analyses performed to examine associations between
thermal preference metrics, year, and climatic variables at the Sheepscot River site (ME 56817). Results are
based on 23 years of data. Entries are in bold text if r > ±0.50. Ranges of biological metric values are also
included
Biological metric
No. cold-water taxa
Percentage cold-water
individuals
No. warm-water taxa
Percentage warm-water
individuals
Range of metric
values
Min
0.0
0.0
1.7
3.8
Max
2.5
5.4
11.7
47.8
r values (based on Kendall Tau correlations)
Year
0.27
0.21
0.63
0.42
Air temperature (°C)
PRISM mean
annual
0.23
0.18
0.16
0.13
Observed mean
maximum
July/August
0.07
0.05
0.26
0.25
Flow (cfs)
Mean
annual
0.14
0.08
-0.02
-0.11
Mean
July-September
0.03
-0.03
-0.04
0.11
PRISM Mean
annual
precipitation
(mm)
0.24
0.21
0.05
0.01
-^
ON
-------
Table 4-16. Kendall tau nonparametric correlations analyses performed to examine associations between a
subset of biological metrics, year, flow, and precipitation variables at the Sheepscot River site (ME 56817). The
subset of biological metrics were selected per the criteria outlined in Section 2 and have shown responsiveness to
hydrologic variables in other studies (see Section 2, Table 2-6). Results are based on 23 years of data. Entries
are in bold text if r > ±0.5. Ranges of biological metric values are also included
Biological metric
Richness
Percentage
individuals
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Range of metric
values
Min
4.0
2.3
1.0
1.0
0.5
0.3
0.2
6.0
45.6
9.1
1.2
1.2
1.2
0.2
0.2
45.7
Max
10.7
7.3
6.7
4.7
4.3
2.8
2.0
14.3
82.9
40.3
26.1
13.5
17.6
3.6
3.8
80.3
r values (based on Kendall Tau correlations)
Year
0.52
0.27
0.54
0.36
0.41
0.26
0.25
0.55
-0.23
-0.02
0.53
0.34
0.26
0.12
0.10
-0.15
Flow (cfs)
Mean
annual
0.00
0.14
0.06
0.02
0.01
0.14
0.00
0.06
-0.11
0.15
-0.02
-0.11
-0.01
0.06
-0.04
-0.12
Mean
July-September
-0.09
0.14
0.07
0.13
-0.15
0.21
0.01
-0.07
-0.13
-0.04
0.05
0.19
-0.13
0.27
0.12
-0.04
PRISM mean
annual
precipitation
(mm)
0.03
0.16
0.08
0.10
0.14
0.09
0.15
0.06
-0.10
0.18
0.08
-0.05
0.08
0.00
0.08
-0.11
-------
Table 4-17. Mean metric values (±1 SD) for the Sheepscot River site (ME 56817) in coldest, normal, and hottest
year samples. Year groups are based on PRISM mean annual air temperature values. One-way ANOVA was
done to evaluate differences in mean metric values. There were no significant (p > 0.05) differences across year
groups
Year group
Coldest
Normal
Hottest
Total no.
taxa
20.9 ±4.3
20.8 ±5.4
24.1 ±3. 8
No. EPT taxa
12.3 ±2.6
12.7 ±3.7
14.3 ±2.3
HBI
4.0 ±0.5
3.9 ±0.5
3.8 ±0.4
No. cold-
water taxa
0.5 ±0.5
0.5 ±0.8
1.0 ±0.5
No. warm-
water taxa
6.7 ±2.2
7.1 ±2.5
8.6 ±2.5
% Cold-water
individuals
0.5 ±0.6
0.8 ± 1.7
0.9 ±0.8
% Warm-water
individuals
15.1±6.9
17.7 ±8.7
23.7 ±14.4
-^
oo
Table 4-18. Mean metric values (±1 SD) for the Sheepscot River site (ME 56817) in driest, normal, and wettest
flow year samples. Year groups are based on mean annual flow values from USGS gage 1038000. One-way
ANOVA was done to evaluate differences in mean metric values. There were no significant (p > 0.05)
differences across year groups
Year group
Driest
Normal
Wettest
Total no.
taxa
22.2 ±4.4
20.9 ±2.8
22.7 ±6.9
No. EPT taxa
13.4±3.0
12.6 ±2.3
13.3±4.1
HBI
3.9 ±0.5
3.9 ±0.4
3.9 ±0.5
No. cold-
water taxa
0.7 ±0.5
0.4 ±0.4
0.9 ±0.9
No. warm-
water taxa
8.0 ±2.4
6.8 ± 1.8
7.7 ±3.3
% Cold-water
individuals
0.7 ±0.5
0.2 ±0.3
1.4 ± 1.9
% Warm-water
individuals
22.4 ±13.9
16.4 ±8.1
18.1 ±9.8
-------
temperature groups. The NMDS ordination showed no distinct clusters reflecting hottest,
coldest, and/or normal year groups (see Figure 4-14).
Maine StationID 56817
1992
A
A 1989
1987
1991
1994
A
1993
1995
1986
A
1997
1996
1999
A
1990
2004
1998
2003
2001
2006
19B5
2002
2000
2005
Cat_Temp
Al
2
3
Axis 2
Figure 4-14. NMDS plot (Axis 1-2) for the Sheepscot River site (ME 56817).
Cat_Temp refers to the temperature categories, which are 1 = 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.
4.6.2. West Branch Sheepscot (ME 57011)
The West Branch Sheepscot site (ME 57011; Maine DEP Station 268) is located in southern
Maine in the town of China. It is in the Laurentian Plains and Hills ecoregion and Central
4-39
-------
Interior biophysical region, has a drainage area of 38.1 km , and an elevation of 70.1 m. Its
highest maximum monthly temperatures occur during July and August, and its lowest average
flows (<85 cfs) occur from July through September. This station has 12 years of continuous
biological data, spanning from 1995 to 2006, that have been collected during Maine DEP's July
through September index period. We gathered daily temperature and precipitation data from
1949 to 2010 from the Augusta FAA AP weather station (SitelD 170275, Latitude: 44.3206,
Longitude: 69.7972), which is located approximately 21.8 km west/southwest of the biological
sampling site. Flow data from 1939-2009 were gathered from USGS gage 1038000 (Sheepscot
River at North Whitefield, Latitude: 44.22278, Longitude: 69.59389), which is located
approximately 17 km south of the biological sampling site, on the mainstem of the Sheepscot.
Figure 4-15 shows an aerial photograph of the site, along with the weather station and active
USGS gage.
-i'k.
Figure 4-15. Locations of the West Branch Sheepscot site (ME 57011)
biological sampling site, USGS gage 1038000 (Sheepscot River at North
Whitefield) and Augusta FAA AP weather station. Image from Google Earth.
4-40
-------
4.6.2.1. Temporal Trends in Climatic and Biological Variables
Since 1949, mean annual air temperatures at the weather station closest to the West
Branch Sheepscot site (ME 57011) site have ranged from 6 to 9.8°C. There is a great deal of
year-to-year variability, but overall, temperatures have been increasing over time (when fit with
a linear trend line, r2 = 0.06, p = 0.05) (see Figure 4-16). When PRISM air temperature data are
compared to observed data, PRISM data are within 1°C of the observed values, and there is good
correspondence between patterns. Based on the USGS gage located on the Sheepscot River
mainstem, mean annual flow and mean annual precipitation patterns have been highly variable
over time (see Figure 4-11). Since 1939, mean annual flow values have ranged from 110.1 to
485.7 cfs (when fit with a linear trend line, r2 = 0.08, p = 0.02). Precipitation patterns generally
show good correspondence with flow patterns but are more variable (see Figure 4-17).
10.0
9.5
9.0
8.5
8.0
7.5
7.0
6.5
6.0
5.5
1948
1958
1968
1978
1988
1998
2008
Figure 4-16. Yearly trends in annual observed air temperature (°C) at the
West Branch Sheepscot site (ME 57011) from 1949-2009, based on data from
the Augusta FAA AP weather station. For comparative purposes, PRISM
annual air temperature data associated with the biological sampling site are also
included from 1975-2005. The area shaded in grey corresponds to the period of
biological record. When the observed data are fitted with a linear trend line,
r2 = 0.06,/? = 0.05, and.y = -12.718 + 0.0102 x x.
4-41
-------
1500
600
1939
1949 1959
1969
1979
1989
1999
2009
Figure 4-17. Yearly trends in mean annual flow (cfs) at the West Branch
Sheepscot site (ME 57011) from 1939-2009, based on data from USGS gage
1038000 (Sheepscot River at North Whitefield). For comparative purposes,
observed annual precipitation data from the Augusta FAA AP weather station are
also included from 1949-2009. The area shaded in grey corresponds to the period
of biological record. When the observed data are fitted with a linear trend line,
r2 = 0.08,/? = 0.02, and.y = -1,814.9037 + 1.0494 x x.
In addition to mean annual values, mean maximum July and August temperature and
mean July-September flow values were also evaluated. During the period of biological record
(1995-2006), mean maximum July/August air temperatures ranged from 24.6-27.8°C, and mean
fall flow values ranged from 17.0 to 156.7 cfs (see Table 4-19). Attainment classes based on
biological condition have ranged from Class A to B. Prior to 2000, samples consistently attained
Class A status, but since that time, they have fluctuated back and forth between Class A and B
(see Figure 4-18 A). The number of EPT taxa has increased over time, ranging from 6 in 1995 to
14 in 2005 (see Figure 4-18B). HBI scores have increased as well, from a low of 3.0 in 1995 to a
high of 5.5 in 2004 (see Figure 4-18B). During the period of biological record, mean maximum
July/August air temperatures and July-September flows were highly variable, with the highest
maximum July/August temperature occurring in 1999, the lowest July-September flows
occurring in 2000, and the highest July-September flows occurring in 2005 (see Figure 4-18C).
The number of warm-water taxa has varied, but overall, numbers have increased over time,
ranging from 4 in 1996 to 11 in 2001 (see Figure 4-19 A). The percentage of warm water
4-42
-------
Table 4-19. Range of temperature, precipitation, and flow values that
occurred at the West Branch Sheepscot site (ME 57011) during the period of
biological record
Parameter
Year
Observed mean annual air temperature (°C)
PRISM mean annual air temperature (°C)
Observed mean maximum July/August air
temperature (°C)
Mean annual flow (cfs)
Mean July-September flow (cfs)
Observed mean annual precipitation (mm)
PRISM mean annual precipitation (mm)
Min
1995
6.9
6.5
24.6
141.6
17.0
661.3
756.5
Max
2006
9.6
8.4
27.8
485.7
156.7
1,461.8
1,652.4
4-43
-------
B
5
14
13
12
11
10
9
8
7
6
5
28
o
o
I
1 27
0)
Q_
,cu
26
0)
25
I
24
-•- EPT taxa
-o- HBI
Temperature
Flow
1995 1997 1999 2001 2003 2005
6.0
5.5
5.0
4.5
4.0
3.5
3.0
2.5
180
160
140
120
100
80
60
40
20
0
m
I
5=
l_
0)
.a
E
S,
a.
a>
Figure 4-18. Yearly trends at the West Branch Sheepscot site (ME 57011) in
(A) biological condition class (1 = Class A; 2 = Class B; 3 = Class C; 4 = NA);
(B) number of EPT taxa and HBI; and (C) mean maximum July/August
temperature (°C) and mean July-September flow (cfs).
4-44
-------
A
12
10
8
6
—
O
-9 4
70
B
60
50
ro 40
=6 30
35 20
10
0
C 28
O
27
CD
Q_
E
26
O)
3
_>,
3
25
o>
24
-•- Cold Water
-&• Warm Water
S3'
a.
-•a
-^ Cold Water
-E>- Warm Water
D---H
Temperature
Flow
180
160 _
140 ~
o
120 tj
CD
100 E
CD
80
60
Q.
CO
^
3
40 g
CD
20 ^
0
1995 1997 1999 2001 2003 2005
Figure 4-19. Yearly trends at the West Branch Sheepscot site (ME 57011) in
(A) number of cold and warm-water taxa; (B) percentage cold and warm-
water individuals; and (C) mean maximum July/August temperature (°C)
and mean July-September flow (cfs).
4-45
-------
individuals has varied over time, ranging from a high of 66% in 1997 to a low of 9% in 2004
(see Figure 4-19B). Very few cold-water taxa are present at this site, with richness values
ranging from 0 to 2.
It is likely that conditions at this site have been influenced by nonpoint source pollution,
with potential anthropogenic influences from urban and agricultural land use (9% urban and
18.5% agricultural within a 1-km buffer). Some recent (post-2000) water chemistry data are
available for various nutrient-related parameters. The maximum nitrogen concentration was
0.56 mg/L, the maximum ammonia concentration ammonia was 0.05 mg/L, and the maximum
total phosphorus concentration was 0.02 mg/L. Confounding factors related to in situ
measurements were not evident, with values in the following ranges:
• DO: 7.5 to 9.5 mg/L
• pH: 6.7 to 7.4
• Specific conductance: 42 to 82 |imho/cm
4.6.2.2. Associations Between Biological Variables and Climatic Variables
Kendall tau nonparametric correlations analyses allow examination of associations
between commonly used biological metrics, year, temperature, flow, and precipitation variables
at the West Branch Sheepscot site (ME 57011) site. Three of the commonly used biological
metrics (percentage EPT individuals, percentage Ephemeroptera individuals, and the Shannon-
Wiener Diversity Index) showed strong positive associations (r > 0.5) with the temperature
variables (see Table 4-20). The number of Ephemeroptera taxa metric also had a fairly strong
(r > 0.4) positive association with temperature. If we assume that higher temperatures are
associated with more stressful conditions, then the direction of these relationships is unexpected.
Only one of the biological metrics had a fairly strong (r > 0.4) association with the flow and
precipitation variables. The percentage noninsect individuals metric was negatively associated
with mean annual flow and precipitation. Four metrics (total number of taxa, number of EPT
taxa, number of Trichoptera taxa, and HBI) had strong positive associations (r > 0.5) with year
(see Table 4-20). The number of intolerant taxa and number of Ephemeropteran taxa metrics
also showed a fairly strong (r > 0.4) positive association with year. The positive trend in HBI
4-46
-------
Table 4-20. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics, year, and climatic variables at the West Branch Sheepscot site (ME 57011).
Results are based on 12 years of data. Entries are in bold text if r > ±0.5 and are highlighted in gray if they are
in a direction opposite of what is expected. Ranges of biological metric values are also included
Biological metric
Total no. taxa
No. EPTtaxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. Intolerant taxa
Percentage EPT individuals
Percentage Ephemeroptera
individuals
Shannon-Wiener Diversity Index
Percentage noninsect individuals
Percentage dominant taxon
Percentage tolerant individuals
Hilsenhoff Biotic Index
Range of metric
values
Min
12.3
6.0
2.0
0.3
2.7
5.0
15.2
6.1
2.1
0.0
15.9
2.1
3.0
Max
33.3
13.7
6.3
1.7
6.7
10.0
80.3
74.6
4.2
2.8
68.8
7.6
5.5
r values (based on Kendall Tau correlations)
Year
0.72
0.58
0.43
0.18
0.57
0.46
-0.33
-0.24
0.12
0.36
0.09
0.15
0.58
Air temperature (°C)
PRISM
mean
annual
0.05
0.12
0.46
0.11
0.05
0.09
0.55
0.52
0.45
-0.14
-0.30
-0.06
-0.24
Observed mean
maximum
July/August
0.20
0.25
0.37
0.25
0.14
0.25
0.45
0.48
0.42
-0.29
-0.33
0.21
-0.27
Flow (cfs)
Mean
annual
0.14
0.18
0.34
-0.25
0.14
-0.03
-0.09
0.18
0.00
-0.42
0.03
-0.03
0.09
Mean
July-September
-0.11
-0.12
0.06
0.04
-0.29
-0.06
0.15
0.18
0.00
-0.20
-0.15
0.21
-0.27
PRISM mean
annual
precipitation
(mm)
0.05
0.09
0.37
-0.18
0.08
-0.03
0.00
0.27
-0.03
-0.45
0.06
0.00
0.06
-------
scores suggests that the assemblage may have experienced more organic enrichment in recent
years, but this is somewhat confounded by the concurrent positive trend in the number of
intolerant taxa and EPT-related metrics.
We performed similar analyses on the thermal preference metrics. The percentage warm-
water individuals metric had a strong (r > 0.5) positive association with the temperature variables
(see Table 4-21). The number of warm-water taxa metric had a fairly strong (r > 0.4) positive
association with year, and the percentage cold-water individuals metric had a fairly strong
(r > 0.4) negative association with year. The subset of biological metrics that have shown
responsiveness to hydrologic variables in other studies (see Section 2, Table 2-6) failed to show
strong (r > 0.5) relationships with the flow and precipitation variables (see Table 4-22). The
number of deposit!onal taxa metric had a fairly strong (r > 0.4) negative association with mean
annual flow, while the number of erosional taxa metric unexpectedly had a fairly strong (r > 0.4)
negative association with mean July-September flows. Three of the metrics (number of collector
gatherer taxa, number of swimmer taxa, and percentage predator individuals) had strong (r > 0.5)
associations with year, with the percentage of predator individuals being negatively associated
with year, and the others showing a positive association with year. The number of erosional taxa
metric also had a fairly strong (r > 0.4) positive association with year, while the percentage of
OCH individuals metric had a fairly strong (r > 0.4) negative association with year.
4.6.2.3. Groupings Based on Climatic Variables
Samples were partitioned into hottest/coldest/normal year groups and
lowest/normal/highest flow year groups. At the West Branch Sheepscot site (ME 57011), on
average, the hottest years were 1.7°C warmer than the coldest years, and highest flow years had
196 more cfs than lowest flow years. When samples were grouped based on temperature, there
were no significant (p > 0.05) differences between mean metric values (see Table 4-23).
Although not significant, there are some patterns worth noting. Mean numbers of total taxa, EPT
taxa, and number of warm-water taxa were highest in hottest year samples. The percentage of
warm-water individuals metric was lowest in the coldest year samples, and the HBI was highest
in the coldest year samples. When samples were grouped based on flow, there were also no
significant (p > 0.05) differences between mean metric values (see Table 4-24). Patterns worth
noting are that, on average, the percentage of cold-water individuals metric was lowest in the
4-48
-------
Table 4-21. Kendall tau nonparametric correlations analyses performed to examine associations between
thermal preference metrics, year, and climatic variables at the West Branch Sheepscot site (ME 57011). Results
are based on 12 years of data. Entries are in bold text if r > ±0.5. Ranges of biological metric values are also
included
Biological metric
No. cold-water taxa
Percentage cold-water
individuals
No. warm-water taxa
Percentage warm-water
individuals
Range of metric
values
Min
0.3
0.2
4.0
9.0
Max
2.3
15.1
10.7
65.8
r values (based on Kendall tau correlations)
Year
-0.03
-0.42
0.47
-0.33
Air temperature (°C)
PRISM
mean annual
0.20
-0.03
0.26
0.42
Observed mean
maximum
July/August
0.20
0.12
0.32
0.52
Flow (cfs)
Mean
annual
-0.07
-0.06
0.05
-0.15
Mean
July-September
0.23
0.24
-0.11
0.09
PRISM mean
annual
precipitation
(mm)
-0.03
0.03
0.02
-0.06
-^
VO
-------
Table 4-22. Kendall tau nonparametric correlations analyses performed to examine associations between a
subset of biological metrics, year, flow, and precipitation variables at the West Branch Sheepscot site (ME
57011). The subset of biological metrics were selected per the criteria outlined in Section 2 and have shown
responsiveness to hydrologic variables in other studies (see Section 2, Table 2-2). Results are based on 15 years
of data. Entries are in bold text if r > ±0.5. Ranges of biological metric values are also included
Biological metric
Richness
Percentage
individuals
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCR
Depositional
Erosional
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCR
Depositional
Erosional
Range of metric values
Min
2.3
1.7
2.3
3.3
0.0
2.3
1.0
4.3
6.0
9.5
3.3
4.7
0.0
1.9
3.0
6.0
Max
6.7
12.3
7.0
9.3
2.7
6.7
4.0
9.0
78.4
46.2
46.4
46.0
32.6
25.5
29.2
45.9
r values (based on Kendall tau correlations)
Year
0.25
0.62
0.36
0.34
0.54
0.28
0.11
0.44
0.36
0.06
-0.30
-0.64
0.15
-0.45
-0.03
-0.21
Flow (cfs)
Mean annual
0.25
-0.06
-0.11
0.09
0.32
0.09
-0.42
0.17
0.24
0.06
0.00
-0.09
0.21
0.15
-0.15
-0.03
Mean
July-Septem her
-0.34
0.00
-0.17
-0.03
0.00
-0.12
0.02
-0.41
-0.12
0.24
0.06
0.27
0.15
0.33
0.21
0.03
PRISM mean annual
precipitation (mm)
0.13
-0.15
-0.11
0.12
0.29
0.15
-0.39
0.14
0.15
0.09
0.09
-0.12
0.24
0.24
-0.12
-0.12
-^
o
-------
Table 4-23. Mean metric values (±1 SD) for the West Branch Sheepscot site (ME 57011) in coldest, normal, and
hottest year samples. Year groups are based on PRISM mean annual air temperature values. One-way
ANOVA was performed to evaluate differences in mean metric values. There are no significant (p > 0.05)
differences across year groups
Year group
Coldest
Normal
Hottest
Total no.
taxa
21.7±4.8
24.1 ±10.4
25 .2 ±3.4
No. EPT
taxa
9.8 ±1.3
10.0±3.7
11.5±1.1
HBI
5.0 ±0.8
3.9 ±0.9
4.4 ±0.4
No. cold-water
taxa
0.8 ±0.3
1.6 ±0.6
1.2 ±0.6
No. warm-
water taxa
6.4 ±2.0
7.3 ±2.3
8.5 ±1.6
% Cold-water
individuals
3.0 ±5.2
6.1 ±6.0
1.9 ±0.4
% Warm-water
individuals
23.5 ±15.9
50.0 ±12.0
40.8 ±12.8
Table 4-24. Mean metric values (±1 SD) for the West Branch Sheepscot site (ME 57011) in driest, normal, and
wettest flow year samples. Year groups are based on mean annual flow values from USGS gage 10128500. One-
way ANOVA was performed to evaluate differences in mean metric values. There are no significant (p > 0.05)
differences across year groups
Year group
Driest
Normal
Wettest
Total no.
taxa
26.3 ±6.2
20.5 ±5.6
24.2 ±7.7
No. EPT
taxa
10.8 ±2.4
9.3 ±2.2
11.3 ±2.4
HBI
4.3 ±0.9
4.5 ± 1.2
4. 5 ±0.5
No. cold-
water taxa
1.3 ±0.3
1.1 ±0.7
1.2 ±0.8
No. warm-
water taxa
7.9 ±2.5
6.9 ±1.7
7.3 ±2.3
% Cold-water
individuals
2.3 ±1.3
4.4 ±7.2
4.2 ±4.5
% Warm-water
individuals
43.0 ±24.3
32.8 ±15.6
38.5 ±12.0
-------
driest flow year samples, and the warm water metrics were highest in the driest flow year
samples. There were insufficient data at this site to do NMDS ordinations.
4.6.3. Duck Brook (ME 57065)
The Duck Brook site (ME 57065; Maine DEP Station 322) is located in southeastern
Maine in the town of Bar Harbor. It is in the Laurentian Plains and Hills ecoregion and East
r\
Coastal Region biophysical region, has a drainage area of 12.8 km , and an elevation of 54.6 m.
Its highest maximum monthly temperatures and lowest average rainfall occur in July and August.
This station has 9 years of continuous biological data, spanning from 1997 to 2006, that have
been collected during Maine DEP's July through September index period. We gathered daily
temperature and precipitation data from 1893 to 1982 from the Bar Harbor 3 NW weather station
(SitelD 170371, Latitude: 44.4167, Longitude: 68.25), which is located on the coast,
approximately 3 km northwest of the biological sampling site. We gathered daily temperature
and precipitation data from 1982 to 2009 from the Acadia NP weather station (SitelD 170100,
Latitude: 44.3739, Longitude: 68.2592), which is located approximately 3 km southwest of the
biological sampling site. There were no USGS gages located in proximity to the biological
sampling site. Figure 4-20 shows an aerial photograph of the site, along with the weather station
and active USGS gage.
4.6.3.1. Temporal Trends in Climatic and Biological Variables
Since 1893, mean annual air temperatures at the weather stations closest to the Duck
Brook site (ME 57065) have ranged from 5.4 to 9.3°C. There is a great deal of year-to-year
variability, but overall, temperatures have been increasing over time (when fit with a linear trend
r\
line, r =0.11, p< 0.01) (see Figure 4-21). When PRISM air temperature data are compared to
observed data, PRISM data are within 1°C of the observed values, and there is good
correspondence between patterns. Flow data were not available for this site, so precipitation was
used as a surrogate. Precipitation patterns have varied a lot from year to year, but overall, mean
annual precipitation has been increasing over time (when fit with a linear trend line, r2 = 0.09,
p < 0.01) (see Figure 4-22). PRISM precipitation trends from 1975-2005 show good
correspondence with the observed precipitation patterns (see Figure 4-22). In addition to
4-52
-------
Figure 4-20. Locations of the Duck Brook site (ME 57065) biological
sampling site, Bar Harbor 3 NW weather station and Acadia NP weather
station. Image from Google Earth.
4-53
-------
o
CD
Q_
E
< 7
CO
13
c
TO
CD
Observed
---- PRISM
1893
1907
1921
1935
1949
1963
1977
1991
2005
Figure 4-21. Yearly trends in annual observed air temperature (°C) at the
Duck Brook site (ME 57065) from 1893-2009, based on data from the Bar
Harbor 3 NW and Acadia NP weather stations. For comparative purposes,
PRISM annual air temperature data associated with the biological sampling site
are also included from 1975-2005. The area shaded in grey corresponds to the
period of biological record. When the observed data are fitted with a linear trend
line, r2 = 0.11,p< 0.01, and.y = -8.6883 +0.0082 * x.
4-54
-------
2000
1800
E
E 1600
f. 1400
'o
§ 1200
1000
800
600
Observed
---- PRISM
1893
1907
1921
1935
1949
1963
1977
1991
2005
Figure 4-22. Yearly trends in annual observed precipitation (mm) at the
Duck Brook site (ME 57065) from 1893-2009, based on data from the Bar
Harbor 3 NW and Acadia NP weather stations. For comparative purposes,
PRISM mean annual precipitation data associated with the biological sampling
site are also included from 1975-2005. The area shaded in grey corresponds to
the period of biological record. When the observed data are fitted with a linear
trend line, r2= 0.09, p< 0.01, and.y = -2,852.2872 + 2.1174 x x.
mean annual values, mean maximum July and August temperature and mean July-September
flow values were also evaluated. During the period of biological record (1997-2005), mean
maximum July/August air temperatures ranged from 23.6-27.3°C, and mean July-September
precipitation values ranged from 32.8 to 97.5 mm (see Table 4-25).
Attainment classes based on biological condition have ranged from Class A to C. From
1997 to 2000, samples met Class A status, then dropped to Class C in 2001, then improved to
Class B in 2002 before returning to Class A in 2003 (see Figure 4-23 A). The number of EPT
taxa has been variable, but overall, numbers have increased over time, ranging from 5 in 1999 to
11 in 2004 (see Figure 4-23B). HBI scores have gone up and down over the period of
4-55
-------
record, hitting a high of 6.3 in 2001 before dropping back down to a score of 3.9 in 2004 (see
Figure 4-23B). During the period of biological record, mean maximum July/August air
temperatures and July-September flows were highly variable, with the highest maximum
July/August temperature occurring in 1998, the lowest July-September rainfall occurring in
2001, and the highest rainfall occurring in 1999 (see Figure 4-23 C). The number of warm-water
taxa varied from year to year, but overall, numbers have increased over time, ranging from 3 in
1999 to 8 in 2003 (see Figure 4-24A). The percentage of warm-water individuals has been
highly variable, ranging from 22 to 69% (see Figure 4-24B). Low numbers of cold-water taxa
occur at this site, with richness numbers mostly ranging from one to two, except for 2004, when
there were four cold-water taxa that comprised 15% of the assemblage.
It is possible that anthropogenic stressors associated with surrounding land use have
influenced conditions at this site. The land use within a 1-km buffer is 16% urban due to
two small roads that parallel the stream. Water chemistry data were limited to in situ
measurements, which were in the following ranges:
• DO: 7.4 to 9.0 mg/L
• pH: 6.6 to 7.0
• Specific conductance: 41 to 82 |imho/cm
Table 4-25. Range of temperature, precipitation, and flow values that
occurred at the Duck Brook site (ME 57065) during the period of biological
record
Parameter
Year
Observed mean annual air temperature (°C)
PRISM mean annual air temperature (°C)
Observed mean maximum July/August Air temperature (°C)
Observed mean annual precipitation (mm)
Observed mean July-September precipitation (mm)
PRISM mean annual precipitation (mm)
Min
1997
6.7
7.0
23.6
763.7
32.8
794.7
Max
2005
9.0
8.9
27.3
1,937.8
97.5
1,640.4
4-56
-------
B
5
12
11
10
to
X n
£ 9
CL p
LU o
'o
d 7
z
6
5
4
O
^
2
3
I
27
26
12
> 25
24
<
23
EPT taxa
- HBI
•- Temperature
E>- Precipitation
1997 1999 2001 2003 2005
6.5
6.0
5.5
5.0 !
4.5
4.0
3.5
110
100
90
80
70
60
50
40
30
§
Q.
'O
s
a
0)
I
_>,
rs
Figure 4-23. Yearly trends at the Duck Brook site (ME 57065) in
(A) biological condition class (1 = Class A; 2 = Class B; 3 = Class C; 4 = NA);
(B) number of EPT taxa and HBI; and (C) mean maximum July/August
temperature (°C) and mean July-September precipitation (mm).
4-57
-------
A
B
7
6
CO
X
|2 5
| 4
3
2
1
0
80
70
60
«j 50
I 40
~ 30
20
10
0
o
Q)
12
to
27
26
25
24
CD
23
•- Cold Water
t*- Warm Water
s—a
V
-^- Cold Water
-o- Warm Water
,ET'
•- Temperature
E>- Precipitation
110
100 I
90 |
80 f
a!
70
60
01
J2
E
.2
"o.
CD
50 IT
40 §
30
1997 1999
2001
2003 2005
Figure 4-24. Yearly trends at the Duck Brook site (ME 57065) in (A) number
of cold and warm-water taxa; (B) percentage cold and warm-water
individuals; and (C) mean maximum July temperature (°C) and mean
July-September precipitation (mm).
4-58
-------
4.6.3.2. Associations Between Biological Variables and Climatic Variables
Kendall tau nonparametric correlations analyses allow examination of associations
between commonly used biological metrics, year, temperature, and precipitation variables at the
Duck Brook site (ME 57065) site. Seven of the commonly used biological metrics showed
strong (r > 0.5) or fairly strong (r > 0.4) negative associations with mean annual and/or mean
July/August maximum air temperature (see Table 4-26). The number of Plecoptera taxa, number
of intolerant taxa, and the Shannon-Wiener Diversity Index had strong (r > 0.5) negative
associations with both temperature variables, while the total number of taxa and the number of
EPT taxa metrics had strong (r > 0.5) negative associations with the mean July/August maximum
air temperature and fairly strong (r > 0.4) negative associations with mean annual temperature.
The number of Ephemeroptera taxa metric had fairly strong (r > 0.38) negative associations with
both temperature variables, and the number of Trichoptera metric had a fairly strong (r > 0.4)
negative association with mean July/August maximum air temperature. Only one of the
commonly used biological metrics was strongly (r > 0.5) correlated with the precipitation
variables. The percentage of noninsect individuals metric was positively correlated with mean
July-September precipitation. Two of the metrics, number of intolerant taxa and percentage
noninsect individuals, had strong (r > 0.6) positive associations with year.
We performed similar analyses on the thermal preference metrics. The number of
cold-water taxa metric showed strong (r > 0.5) negative associations with both temperature
variables (see Table 4-27). The warm water metrics also had strong negative associations with
both temperature variables, which was unexpected. Only one of the biological metrics that have
shown responsiveness to hydrologic variables in other studies (see Section 2, Table 2-6) showed
a strong (r > 0.5) relationships with the precipitation variables. The percentage
collector-gatherer individuals metric had a strong (r > 0.5) negative association with mean
annual precipitation (see Table 4-28). One of the metrics, number of scraper/herbivore taxa, had
a strong (r = 0.80) positive association with year.
4.6.3.3. Groupings Based on Climatic Variables
Samples were partitioned into hottest/coldest/normal year groups and
lowest/normal/highest flow year groups. At the Duck Brook site (ME 57065), on average, the
hottest years were 1.4°C warmer than the coldest years, and wettest years had
4-59
-------
Table 4-26. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics, year, and climatic variables at the Duck Brook site (ME 57065). Results are
based on 9 years of data. Entries are in bold text if r > ±0.5. Ranges of biological metric values are also included
Biological metric
Total no. taxa
No. EPT taxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. Intolerant taxa
Percentage EPT individuals
Percentage Ephemeroptera
individuals
Shannon-Wiener Diversity
Index
Percentage noninsect
individuals
Percentage dominant taxon
Percentage tolerant
individuals
Hilsenhoff Biotic Index
Range of
metric values
Min
13.0
4.7
2.3
0.3
1.3
3.0
11.4
3.6
2.8
0.8
17.9
8.0
3.8
Max
28.0
11.0
5.3
1.7
5.3
7.3
89.2
77.4
4.1
24.6
47.6
52.5
6.2
r values (based on Kendall tau correlations)
Year
0.39
0.39
0.26
0.30
0.17
0.61
-0.11
-0.17
0.22
0.61
-0.39
-0.06
0.17
Air temperature (°C)
PRISM
mean
annual
-0.33
-0.44
-0.44
-0.50
-0.23
-0.61
-0.17
-0.22
-0.50
-0.11
0.33
0.11
0.22
Observed mean
maximum
July/August
-0.61
-0.61
-0.38
-0.50
-0.40
-0.84
0.11
0.06
-0.56
-0.39
0.39
0.06
-0.06
Precipitation (mm)
PRISM
mean
annual
-0.11
-0.22
-0.15
-0.17
-0.06
-0.03
0.06
-0.11
-0.17
0.11
-0.11
-0.33
-0.11
Observed
July-September
0.17
0.06
-0.03
0.17
0.06
0.26
-0.33
-0.39
0.33
0.50
-0.39
-0.17
0.06
o\
o
-------
Table 4-27. Kendall tau nonparametric correlations analyses performed to examine associations between
thermal preference metrics, year, and climatic variables at the Duck Brook site (ME 57065). No warm-water
taxa were present at this site. Results are based on 9 years of data. Entries are in bold text if r > ± 0.5 and are
highlighted in gray if they are in a direction opposite of what is expected. Ranges of biological metric values are
also included
Biological metric
No. cold-water taxa
Percentage cold-water
individuals
No. warm-water taxa
Percentage warm-water
individuals
Range of metric
values
Min
1.0
1.8
3.3
22.2
Max
3.7
15.2
8.3
68.9
r values (based on Kendall tau correlations)
Year
0.35
0.06
0.44
-0.17
Air temperature (°C)
PRISM
mean annual
-0.47
-0.33
-0.61
0.00
Observed mean
maximum July/August
-0.59
-0.17
-0.67
0.28
Precipitation (mm)
PRISM mean
annual
-0.18
0.00
-0.15
-0.33
Observed
July-September
0.00
-0.06
-0.03
-0.61
-------
Table 4-28. Kendall tau nonparametric correlations analyses performed to examine associations between a
subset of biological metrics, year, flow, and precipitation variables at the Duck Brook site (ME 57065). The
subset of biological metrics were selected per the criteria outlined in Section 2 and have shown responsiveness to
hydrologic variables in other studies (see Section 2, Table 2-6). Results are based on 9 years of data. Entries are
in bold text if r > ±0.5. Ranges of biological metric values are also included
o\
to
Biological metric
Richness
Percentage
individuals
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Range of metric
values
Min
0.7
4.0
3.3
4.0
0.3
1.3
0.0
3.0
1.0
11.1
8.5
12.7
0.5
2.9
0.0
9.1
Max
2.0
9.7
5.3
10.0
1.3
4.0
1.7
8.0
27.6
55.8
45.7
34.0
27.4
17.8
4.3
36.3
r values (based on Kendall tau correlations)
Year
0.15
0.38
0.80
0.20
0.13
0.15
0.19
0.20
0.11
-0.22
0.00
0.00
-0.17
0.17
0.14
0.17
Precipitation (mm)
PRISM mean
annual
0.27
-0.15
0.23
-0.31
-0.13
-0.39
-0.25
-0.08
0.28
-0.50
-0.06
0.06
0.00
-0.33
-0.14
0.11
Observed
July-September
-0.03
0.26
0.00
0.03
0.44
0.15
0.19
-0.03
0.33
-0.33
-0.11
0.00
-0.06
0.28
0.25
0.39
-------
547 more mm of precipitation than driest years. When samples were grouped based on
temperature, there were no significant (p > 0.05) differences between mean metric values (see
Table 4-29). Although not significant, it is worth noting that mean numbers of total taxa, EPT
taxa, cold-water taxa, and warm-water taxa were lowest in the hottest year samples, as was the
percentage of cold-water individuals. When samples were grouped based on precipitation, there
were also no significant (p > 0.05) differences between mean metric values (see Table 4-30).
There were insufficient data to do NMDS ordinations at this site.
4.7. SENSITIVITY OF BENTHIC MACROINVERTEBRATES TO TEMPERATURE
AND STREAM FLOW
The spatial distributions of cold and warm-water taxa were examined to gain insights into
which areas in Maine are likely to be most and least sensitive to projected changes in
temperature and stream flow. On average, there are low numbers of cold-water taxa (<2) at all
of Maine DEP's sampling locations (see Table 4-31). In all three ecoregions, on average, there
are more warm-water than cold-water taxa, with the highest numbers and abundances of
warm-water taxa occurring in the Laurentian Plains and Hills ecoregion. If the assumption is
made that streams with the highest relative abundances of cold-water taxa will be most sensitive
to warming temperatures and changing precipitation patterns, then streams in the Northeastern
Highlands ecoregion will be most sensitive (see Table 4-31).
The prevalence and distribution of cold- and warm-water-preference taxa vary
predictably with stream order. First- through third-order streams in Maine have slightly greater
relative abundance and richness of cold-water-preference taxa (see Figure 4-25 A). On average,
first- and second-order streams have fewer warm-preference taxa (see Figure 4-25B). The
three Maine biological sampling stations that we closely examined for long-term trends were
first-, third-, and fourth-order steams. Although the coldest, highest elevation streams are likely
to be sensitive to climate change effects, it may be that the greatest amount of change will occur
in transitional areas, where species are expected to be closer to their tolerance limits.
4-63
-------
Table 4-29. Mean metric values (±1 SD) for the Duck Brook site (ME 57065) in coldest, normal, and hottest year
samples. Year groups are based on PRISM mean annual air temperature values. One-way ANOVA was done
to evaluate differences in mean metric values. No entries are significantly different (p < 0.05) across year groups
Year group
Coldest
Normal
Hottest
Total no.
taxa
22.1 ±8.0
21.7±3.5
18.4±3.7
No. EPT
taxa
8.9 ±2.6
8.2 ±1.8
6.8 ±2.0
HBI
4.3 ±0.3
5.1 ±0.9
4.8 ± 1.3
No. cold-
water taxa
2.4 ±1.2
1.7 ±0.3
1.6 ±0.7
No. warm-
water taxa
6.3 ±0.6
6.8 ± 1.5
4.8 ± 1.3
% Cold-water
individuals
7.8 ±6.4
5.3 ±5.9
5.0±3.3
% Warm-water
individuals
44.0 ±22.5
32.8 ±10.8
46.6 ±17.6
Table 4-30. Mean metric values (±1 SD) for the Duck Brook site (ME 57065) in driest, normal, and wettest year
samples. Year groups are based on mean annual flow. One-way ANOVA was done to evaluate differences in
mean metric values. No entries are significantly different (p < 0.05) across year groups
Year group
Driest
Normal
Wettest
Total no. taxa
20.3 ±6.5
23.4 ±6.0
18.4 ±2.2
No. EPT
taxa
8.0 ±2.6
8.1 ±3.0
7.8 ± 1.3
HBI
4.8 ± 1.3
4.7 ±0.3
4.7 ± 1.2
No. cold-water
taxa
2.4 ± 1.2
1.6 ±0.7
1.7 ±0.3
No. warm-
water taxa
6.1 ±0.5
6.1 ±2.5
5.7 ±0.9
% Cold-water
individuals
7.5 ±6.7
3.1 ±1.0
7.6 ±5.3
% Warm-water
individuals
56.1 ±16.0
28.1±3.8
39.1 ±15.1
o\
-------
Table 4-31. Summary of differences in elevation, PRISM mean annual air temperature and precipitation, and
mean number and percentage of cold and warm-water-preference taxa across and within major ecoregions.
Samples were not limited to a particular season
Ecoregion
Northeastern Coastal Zone
Laurentian Plains and Hills
Northeastern Highlands
No.
samples
576
2,830
857
Elevation
(m)
29.3
65.2
210.4
Air temperature
(°C)
8.3
6.5
5.8
Richness
Cold water
1.7 ±1.9
1.1 ±1.4
1.7 ±2.0
Warm water
3.3 ±2.8
4.7 ±3.3
3.2 ±2.7
Relative abundance
Cold water
5.4 ±9.9
2.8 ±6.6
7.1 ±11.8
Warm water
17.0 ±20.6
22.4 ±22.0
15.1 ±17.5
-------
10
3
6
4
2
0
-2
•
o
. . T T
Class A & AA sites
o
o
T
o
n=230 n=1.49 n=273 n=284 n=95 n=32
12345
Strahler Order
6
B
Class A & AA sites
o
0
T
T
T
T
34
Order
o Median
D 25%-75%
X Non-Outlier Range
o Outliers
*: Extremes
Figure 4-25. Distribution of cold and warm-water taxa across Strahler Orders in Maine, based on
July-September replicates collected from sites that received Class A biological condition ratings. Replicates
were analyzed separately in this analysis. (A) number of cold-water taxa; (B) number of warm-water taxa. Samples
sizes are: first order = 230; second order = 149; third order = 273; fourth order = 284; fifth order = 95; and sixth
order = 32.
-------
4.8. IMPLICATIONS FOR MAINE DEP'S BIOMONITORING PROGRAM
Over the last century, there has been a lot of year-to-year variability in temperature and
precipitation patterns in Maine, both statewide and at the three long-term biological monitoring
sites that we closely examined for temporal trends. Overall, temperature and precipitation have
increased from 1901-2000. Because there has been a high degree of year-to-year variability in
more recent decades, these trends are less evident from 1971-2000. Future projections in Maine
call for a continuation of warming temperatures, especially in the winter. Changes in future
precipitation patterns are more difficult to project due to uncertainty associated with the climate
models.
When we analyzed data from three long-term biological monitoring sites in Maine, a
number of the biological variables were strongly associated with year, so temporal trends were
evident. However, few of these trends were associated with temperature, flow, and precipitation
variables, and when strong associations did occur with the climate variables, they were not
consistent across sites, and some were not in keeping with expectations. Analyses of data
grouped by hottest/normal/coldest years and lowest/normal/highest flow years also failed to
reveal consistent or significant patterns in the biological data. There was one consistent but
nonsignificant pattern that did occur at all three sites—the warm water metrics were highest in
the lowest flow/driest year samples.
The lack of strong and consistent associations between biological and climate variables
could be due in part to the large amount of year-to-year variability that occurred in the climate
variables during the period of biological record. Another possible contributing factor was that
anthropogenic influences were higher than desired at all three sites, so biological responses may
have been driven more by nonclimate-related factors. Also, the biological assemblages that we
evaluated had low numbers of cold-water taxa, which likely limits the responsiveness of the
assemblage to warming temperatures. As shown in Table 4-31, on average, there are low
numbers of cold-water taxa at sites sampled by Maine DEP. Assemblages composed of greater
numbers of cold-water taxa likely exist in Maine, but these may be limited to higher elevation
streams that are difficult to access and, thus, are sampled less frequently.
We also performed some additional analyses to try and gain more insights into how
climate change may impact Maine DEP's assessment methods. We looked into the possibility of
manipulating Maine DEP's linear discriminant models in ways that would simulate potential
4-67
-------
changes associated with climate change. However, due to the complexity of the models, we
were unable to do so. The best alternative that we could develop was to evaluate the model input
metrics individually. This type of analysis is informative but is limited by the fact that the linear
discriminant models look at multiple variables simultaneously; thus, there are no firm thresholds
or individual metric values at which a sample changes classification levels.
First, we evaluated differences in mean model input metric values across the different
classification groups. We did this by performing one-way ANOVA analyses on a data set
composed of rock basket or rock cone samples collected during Maine DEP's July-September
index period. Appendix C contains box plots showing the distributions of the model input
metrics across the different classification groups. Results show that Class A samples have, on
average:
• High generic richness
• High richness and abundance of EPT taxa
• High Shannon-Wiener diversity index values
• Low HBI scores
• Low Chironomidae abundances
• Low relative Diptera richness
• Low relative Oligochaeta abundance
• Greater presence of Class A indicator taxa
• Greater scraper relative abundance
Based on this set of results plus results from our thermal indicator analysis, we made
theoretical predictions about which model input metrics are most likely to be influenced by
increasing temperatures, as well which direction the metric values are likely to change in. When
making predictions, we also noted which model input metrics showed patterns associated with
changing streamflow conditions at the Sheepscot River site (ME 56817) site, which has over
20 years of continuous biological and hydrologic data. These considerations were based on
4-68
-------
differences in mean metric values when samples were grouped by lowest/normal/highest flow
years. Table 4-32 summarizes our predictions.
Results vary by metric, and as mentioned, are limited by the fact that Maine DEP's linear
discriminant models look at multiple variables simultaneously; thus, can only theorize how
changes in individual metrics might affect overall classifications. We predict that some metric
scores are likely to improve, which could contribute to better overall classification ratings, while
others may worsen and contribute to the lowering of overall classification ratings. A number of
the model input metrics are related to EPT taxa. Because the lists of cold and warm-water taxa
are composed of a mix of EPT taxa, in some cases, it is difficult to predict whether any
noticeable change will occur in the EPT-related metrics, and if changes do occur, in what
direction. For example, it is possible that cold-water taxa from a particular order may drop out at
a site due to warming temperatures, but then warm-water taxa from the same order could replace
these taxa, thus causing metric values to remain about the same. Another limitation in our ability
to predict and detect changes associated with changing temperatures is the fact that the
warm-water taxa in Maine are evenly distributed across enrichment tolerance categories. This
makes it difficult to tease out biological responses to warming temperatures from confounding
factors such as organic enrichment.
4-69
-------
Table 4-32. List of model input metrics from Maine DEP's linear discriminant models that could be most
affected by changing temperature and streamflow conditions. This table includes information on which
classification is associated with high metric values (for example, on average, Class A samples have the highest
EPT generic richness values), which direction we predict metric values to change in, and the reasoning behind
our assessments
Metric
Generic richness
EPT generic richness
Plecoptera abundance
Relative generic
richness Plecoptera
Perlidae abundance
Relative abundance
Ephemeroptera
Ephemeroptera
abundance
Classification
associated with highest
mean metric values
AandB
A
A
A
AandB
A
B
Predicted
change in
metric value
Increase
Variable
Decrease
Decrease
Increase
Increase
Increase
Reasoning
At ME 56817 and ME 5701 1, the mean total number of taxa was highest in the
hottest year samples; this suggests that warming temperatures could improve
scores for this metric, as well as for overall classification
There are more EPT taxa on the cold water list than on the warm water list,
which suggests that warming temperatures are most likely to decrease scores
for this metric; however, at ME 56817 and ME 5701 1, the mean number of
EPT taxa was highest in the hottest year samples, which suggest that climate
change effects on this metric will be variable
There are 16 Plecopteran taxa on the cold water list and 3 on the warm water
list; this suggests that warming temperatures are most likely to decrease scores
for this metric and lower overall classification
There are three Perlidae on the warm water list and none on cold water list; this
suggests that warming temperatures are likely to improve scores for this metric,
as well as for overall classification
There are nine Ephemeropterans on the warm water list and four on the cold
water list; this suggests that warming temperatures are likely to improve scores
for this metric, as well as for overall classification
There are nine Ephemeropterans on the warm water list and four on the cold
water list; this suggests that warming temperatures are more likely to improve
scores for this metric, which could cause more samples to receive Class B
ratings
-^
o
-------
Table 4-32. List of model input metrics from Maine DEP's linear discriminant models that could be most
affected by changing temperature and streamflow conditions. This table includes information on which
classification is associated with high metric values (for example, on average, Class A samples have the highest
EPT generic richness values), which direction we predict metric values to change in, and the reasoning behind
our assessments (cont.)
Metric
Classification
associated with highest
mean metric values
Predicted
change in
metric value
Reasoning
Relative abundance
Chironomidae
C and NA
Variable
There are seven Chironomidae on the cold water list and nine on the warm
water list; this suggests that if warming temperatures cause cold-water taxa to
be replaced by warm-water taxa, metric values are likely to remain similar
Ratio of Class A
indicator taxa
(Brachycentrus,
Serratella, Leucrocuta,
Glossosoma,
Paragnetina,
Eurylophella, and
Psilotretd)
Variable
Two of the seven Class A indicator taxa (Eurylophella and Glossosoma) are on
the cold water list, and three (Paragnetina, Serratella and Leucrocuta) are on
the warm water list; this suggests that warming temperatures will have varying
effects on metric values. In an ANOVA of data from ME 56817, on average,
more Class A indicator taxa were present in wettest years
Sum of mean
abundances of
Dicrotendipes,
Micropsectra,
Parachironomus, and
Helobdella
NA
Increase
Dicrotendipes and Parachironomus are on the warm water list; this suggests
that warming temperatures are likely to increase scores for this metric and
lower overall classification
Sum of mean
abundances of
Acroneuria and
Stenonema
B
Increase
Acroneuria and Stenonema are on the warm water list; this suggests that
warming temperatures will increase scores for this metric, which could cause
more samples to receive Class B ratings
Hilsenhoff Biotic Index
NA
Variable
Many of the cold-water taxa are intolerant to enrichment, and warm-water taxa
are evenly distributed across tolerance groups; this suggests that warming
-------
Table 4-32. List of model input metrics from Maine DEP's linear discriminant models that could be most
affected by changing temperature and streamflow conditions. This table includes information on which
classification is associated with high metric values (for example, on average, Class A samples have the highest
EPT generic richness values), which direction we predict metric values to change in, and the reasoning behind
our assessments (cont.)
Metric
Relative richness
Diptera
Tanypodinae abundance
EPT generic richness
divided by Diptera
generic richness
EPT generic richness
relative to EPT plus
Diptera
Classification
associated with highest
mean metric values
CandNA
C
A
A
Predicted
change in
metric value
Uncertain
Uncertain
Uncertain
Uncertain
Reasoning
temperatures will have variable effects on HBI scores
In an ANOVA of data from ME 56817, on average, metric values were lowest
in highest flow year samples
In an ANOVA of data from ME 56817, on average, metric values were lowest
in highest flow year samples
In an ANOVA of data from ME 56817, on average, metric values were highest
in highest flow year samples
In an ANOVA of data from ME 56817, on average, metric values were highest
in highest flow year samples
-^
to
-------
5. NORTH CAROLINA
5.1. EXPOSURES
5.1.1. Regional Projections for the Southeastern United States
There are a number of factors (e.g., convective precipitation, seasonal contributions from
hurricanes, complex moisture sources) that make current climate in the southeast regionally
variable and future climate changes challenging to model (Mearns et al., 2003). Based on a finer
scale regional climate model, average temperatures in the southeastern United States are
projected to increase 4-5°C with a doubling in CC>2 concentrations (Mearns et al., 2003) (see
Table 5-1). Spatial variability in projected temperature increases is greatest for summer
maximum temperatures, with increases of 3-4°C projected for the southwestern portion of the
region and of about 7°C in the northeastern portion of the region where the biggest decreases in
precipitation are also projected to occur.
Table 5-1. Projections for temperature and precipitation changes in the
Southeast to 2100
Temperature
change
4-5 °C
Precipitation change
27-37% (spring); -31 to -17% (summer); -7
to +3% (fall); -2 -to -19% (winter)
-1 1% (winter); -7% (summer)
Change in
precipitation
frequency
18% (winter);
-37%
(summer)
Citation
Mearns et al.,
2003
Schoofetal.,2010
Projected changes for precipitation are variable among seasons. Large increases in
precipitation are projected for the spring, while large decreases are projected for the summer
(Mearns et al., 2003) (see Table 5-1). The biggest spatial contrasts in projected precipitation
changes occur in the winter and summer, grading from smaller decreases to slight increases in
the northwestern corner of the region, to much larger decreases in the east to southeast (Mearns
et al., 2003). Schoof et al. (2010) projects an increase in the frequency of cold season
precipitation in the southeast, but decreases in the amount of precipitation. Both frequency and
5-1
-------
magnitude of warm season precipitation events are projected to decrease (Schoof et al., 2010).
Despite projections for large spring increases in precipitation, runoff in the southeast is in
general projected to decrease as a result of increases in evapotranspiration forced by increasing
temperatures (Mulholland et al., 1997). This will be most extreme during the summer when
temperature increases will be combined with projected large decreases in precipitation. In
contrast, Wolock and McCabe (1999) estimated anywhere from large decreases in runoff in the
southeast and Gulf using the Canadian Centre for Climate Prediction and Analysis GCM to
small-to-moderate increases in runoff using the Hadley Centre for Climate Prediction and
Research model, attributed mainly to projected changes in precipitation.
5.1.2. Historic Climate Trends and Climate Change Projections for North Carolina
North Carolina has a warm and wet climate, with mild winters and high humidity.
Extreme weather events, such as hurricanes and droughts, are not uncommon. When assessing
the biological integrity of streams, the North Carolina Department of the Environment and
Natural Resources (NCDENR) divides the state into three major regions: (1) Mountain (which
corresponds with the EPA Level 3 Blue Ridge ecoregion and runs along the western portion of
the state); (2) Piedmont (which corresponds with the EPA Level 3 Piedmont ecoregion in central
North Carolina); and (3) Coastal (which covers the eastern portion of the state and generally
overlaps with the Southeastern Plains and Middle Atlantic Coastal Plain EPA Level 3
ecoregions). These regions have distinct features. Topography in the Mountain region ranges
from narrow ridges to hilly plateaus to large mountainous areas with high peaks. There is a high
diversity of flora and fauna with high-gradient, cool, clear streams with rocks and boulders. The
Piedmont ecoregion is a transitional area between the mostly mountainous regions of the
Appalachians and the relatively flat coastal plain. Major land cover transformations have
occurred in this ecoregion over the past 200 years, with the landscape going from forest to farm,
back to forest, and now, in many areas, spreading urban- and suburbanization (Griffith, et al.,
2002). The Coastal ecoregion consists of low elevation, flat plains, with many swamps, marshes,
and estuaries. Streams are relatively low-gradient and sandy-bottomed (Griffith et al., 2002;
U.S. EPA, 2002). The Coastal region has the highest mean annual temperatures, while the
Mountain region has the lowest mean annual air temperatures and the greatest amount of annual
precipitation (see Figures 5-1A and B).
5-2
-------
There is large year-to-year variability in historic temperature and precipitation patterns in
North Carolina. A historic trend analysis of North Carolina PRISM data revealed that there is no
clear trend in mean annual air temperature, either annually or seasonally, from 1901-2000 (see
Table 5-2, Figures 5-2 and 5-3). In more recent decades (1971-2000), slight trends in annual and
seasonal temperatures are evident, with change rates ranging from 0.01 to 0.02°C/year. From
1971-2000, only the increasing trend associated with summer temperatures is significant. Table
5-3 summarizes future projections for mid- and late-century for high (A2) and low (Bl)
emissions scenarios. Based on an ensemble average across 15 models, mean annual air
temperatures are projected to increase by up to 2.8°C by midcentury and up to 4.9°C by the end
of the century compared to a historic time period (1961-1990). On average, the greatest
increases are projected to occur during the summer and fall seasons (see Table 5-3).
Figure 5-1. North Carolina's temperature and precipitation patterns.
(A) Mean annual air temperature (°C) from 1971-2000; (B) Mean annual
precipitation (mm) 1971-2000. Map produced using the Climate Wizard Web
site (http://www.climatewizard.org/). Base climate data from the PRISM Group,
Oregon State University, http://www.prismclimate.org.
5-3
-------
Table 5-2. Change rates in North Carolina PRISM mean annual air
temperature compared across two time periods: 1971-2000 versus 1901-
2000. Entries in bold text are significant (p < 0.05). Data were derived from
the Climate Wizard Web site (http://www.climatewizard.org/). Base climate
data came from the PRISM Group, Oregon State University,
http://www.prismclimate.org
Time period
1901-2000
1971-2000
Air temperature (C/yr)
Annual
0.00
0.01
DJF
0.00
0.02
MAM
0.00
0.00
JJA
0.00
0.02
SON
0.00
-0.01
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, and August;
SON = September, October, and November.
o^
£ u>
^ lO
2 *~
Q.
OJ
ro ±:
o>
TO
3
1900
2000
Figure 5-2. Trends in annual mean air temperature in North Carolina from
1901-2000. Change rate = 0°C/year,/7-value = 0.93. Figure produced using
Climate Wizard Web site (http://www.climatewizard.org/). Base climate data
from the PRISM Group, Oregon State University, http://www.prismclimate.org.
5-4
-------
1900 (3:0 1340 I960 1980 2000
1900 1920 1940 I960 1980 2000
D :
i
I
I
1900 1920 1940 1960 I9SO 2000
Figure 5-3. Trends in seasonal mean air temperature in North Carolina
from 1901-2000. (A) DJF = December, January, and February, change
rate = O.OOFC/year,/>-value = 0.88; (B) MAM = March, April, and May, change
rate = 0.001°C/year,/?-value = 0.77; (C) JJA = June, July, and August, change
rate = 0°C/year, p-va\ue = 0.88; (D) SON = September, October, and November,
change rate = -0.001°C/year,/>-value = 0.80. Figure produced using Climate
Wizard Web site (http://www.climatewizard.org/). Base climate data from the
PRISM Group, Oregon State University, http://www.prismclimate.org.
5-5
-------
Table 5-3. Projected departure from historic (1961-1990) trends in annual and seasonal air temperature (°C) in
North Carolina for mid- (2040-2069) and late-century (2070-2099) for high and low emissions scenarios. Values
represent the minimum, average, maximum, and standard deviations from 15 different climate models. Data
were derived from the Climate Wizard Web site (http://www.climatewizard.org/)
Midcentury (2040-2069) vs. historic (1961-1990)
Model
Ensemble low
Ensemble average
Ensemble high
SD
A2 (high) emissions scenario
Annual
1.4
2.3
2.8
0.4
DJF
1.3
2.0
3.0
0.5
MAM
1.2
2.0
3.0
0.6
JJA
1.3
2.4
3.2
0.5
SON
1.5
2.5
3.5
0.6
Bl (low) emissions scenario
Annual
1.1
1.7
2.2
0.4
DJF
0.6
1.5
2.6
0.6
MAM
1.0
1.8
2.3
0.5
JJA
1.1
1.9
3.1
0.5
SON
1.0
1.8
2.5
0.5
Late-century (2070-2099) vs. historic (1961-1990)
Ensemble low
Ensemble average
Ensemble high
SD
2.0
3.7
4.9
0.7
1.7
3.1
4.6
0.7
2.1
3.3
4.8
0.7
2.0
4.1
5.4
0.9
2.6
4.2
5.6
0.9
1.4
2.2
3.0
0.5
1.0
2.0
3.2
0.6
0.8
2.2
3.0
0.6
1.4
2.4
3.5
0.6
1.4
2.3
3.2
0.6
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, August and SON = September, October, and
November.
-------
Precipitation patterns in North Carolina have been highly variable. From 1901-2000,
mean annual precipitation increased at a rate of 0.39 mm/year (p > 0.05) (see Figure 5-4 and
Table 5-4). There were two significant (p < 0.05) trends in seasonal data over this time period.
Summer precipitation decreased at a rate of 0.68 mm/year, and fall precipitation increased at a
rate of 0.83 mm/year (see Table 5-4 and Figure 5-5). In more recent decades (1971-2000), the
trends in summer and fall precipitation were similar but not significant (p > 0.05). Compared to
1901-2000 trends in annual, winter, and spring precipitation changed direction, going from
increasing to decreasing (see Table 5-4). Table 5-5 summarizes future projections for mid- and
late-century for high (A2) and low (Bl) emissions scenarios. The future projections are highly
variable across models and emissions scenarios. Under the high emissions scenario, the
ensemble average projects that mean annual precipitation will increase by 54 mm by midcentury
and 56.9 mm by the end of the century compared to a historic time period (1961-1990). Under
the high emissions scenario, the smallest changes are projected to occur during the spring (see
Table 5-5).
Table 5-4. Change rates in North Carolina PRISM mean annual
precipitation compared across two time periods: 1971-2000 versus
1901-2000. Entries in bold text are significant (p < 0.05). Data were derived
from the Climate Wizard Web site (http://www.climatewizard.org/). Base
climate data came from the PRISM Group, Oregon State University,
http://www.prismclimate.org
Time period
1901-2000
1971-2000
Precipitation (mm/yr)
Annual
0.39
-1.47
DJF
0.04
-0.48
MAM
0.19
-2.15
JJA
-0.68
0.61
SON
0.83
1.18
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, and
August; SON = September, October, and November.
5-7
-------
c
o
o
Q>
ro
3
C
O
o
CD
o
o
in
o
o
o
o
co
o
o
CN
O
o
o
o
o
1900
1920
I
1940 1960
Year
T
1980
2000
Figure 5-4. Trends in annual mean precipitation in North Carolina from
1901-2000. Change rate = 0.39 mm/year,p-va\ue = 0.38. Figure produced using
Climate Wizard Web site (http://www.climatewizard.org/). Base climate data
from the PRISM Group, Oregon State University, http://www.prismclimate.org.
5-8
-------
1900 1920 1940 1960 1940 2000
1900 1920 1940 I960 1980 2000
1900 1920
Figure 5-5. Trends in seasonal mean precipitation in North Carolina from
1901-2000. (A) DJF = December, January, and February, change
rate = 0.035 mm/year,p-va\ue = 0.88; (B) MAM = March, April, and May,
change rate = 0.194 mm/year, p-va\ue = 0.38; (C) JJA = June, July, and August,
change rate = -0.677 mm/year, p-va\ue = .01; (D) SON = September, October,
and November, change rate = 0.83 mm/year, p-va\ue < 0.01. Figure produced
using Climate Wizard Web site (http://www.climatewizard.org/). Base climate
data from the PRISM Group, Oregon State University,
http://www.prismclimate.org.
5-9
-------
Table 5-5. Projected departure from historic (1961-1990) trends in annual and seasonal precipitation (mm) in
North Carolina for mid- (2040-2069) and late-century (2070-2099) for high and low emissions scenarios. Values
represent the minimum, average, maximum, and standard deviations from 15 different climate models. Data
were derived from the Climate Wizard Web site (http://www.climatewizard.org/)
Midcentury (2040-2069) vs. historic (1961-1990)
Model
Ensemble low
Ensemble average
Ensemble high
SD
A2 (high) emissions scenario
Annual
-230.8
54.0
171.7
115.3
DJF
-61.0
15.4
61.5
39.5
MAM
-31.6
7.0
40.9
22.3
JJA
-109.2
11.6
115.1
54.1
SON
-60.7
19.6
55.9
27.9
Bl (low) emissions scenario
Annual
-400.0
-1.0
167.4
161.6
DJF
-40.1
16.0
119.0
40.3
MAM
-91.2
-3.4
48.5
36.1
JJA
-366.2
-22.5
81.5
112.4
SON
-185.4
-6.3
47.7
59.8
Late-Century (2070-2099) vs. historic (1961-1990)
Ensemble low
Ensemble average
Ensemble high
SD
-290.3
56.9
261.2
163.6
-63.5
20.9
97.0
48.2
-62.8
6.2
53.4
36.4
-140.9
22.3
168.5
82.1
-55.1
20.8
69.0
38.6
-554.3
-16.5
179.2
222.0
-53.1
24.5
142.0
52.0
-105.5
-7.6
60.3
47.3
-391.1
-41.5
89.3
148.7
-205.1
-10.6
49.7
78.3
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, August and SON = September, October, and
November.
-------
5.2. DATA INVENTORY AND PREPARATION
Data for North Carolina were provided by NCDENR. Our North Carolina database
contains data for 5,823 biological samples from 2,786 unique stations, with sampling dates
ranging from 1978 to 2007. In situ measurements (conductivity, DO, pH, and water
temperature) were provided for some of the sites, as were habitat measurements (NCDENR
habitat index, width, depth, visual estimates of substrate composition, and canopy cover). The
NCDENR habitat index, which has scores ranging from 1 (worst) to 100 (best), is based on
assessments of channel modification, amount of instream habitat, type of bottom substrate, pool
variety, bank stability, light penetration, and riparian zone width (NCDENR, 2006). The visual
estimates of substrate composition were interpreted with caution due to observer bias (Trish
MacPherson, NCDENR, personal communication).
NCDENR records data by waterbody name, location description, latitude and longitude,
and date, but does not assign unique station IDs to its sampling sites. Sometimes we had
difficulty determining whether samples were collected from the same or different sites. This
occurred when samples had similar waterbody names but with slightly different spellings (for
example, "Creek" might be spelled out in one sample record and abbreviated as "Cr" in another);
when samples with similar waterbody names and location descriptions had slightly different
latitudes and longitudes and when sites had the same water body name but slightly different
location descriptions. To address this issue, we created unique identifiers for sites (station IDs)
based on matching a combination of waterbody name, location, and latitude-longitude.
We used a genus-level OTU when preparing the biological data for long-term trend
analyses. Per the methods described in Section 2.1.3, we used NMDS analyses to verify the
OTU. Because the same taxonomists in the North Carolina biomonitoring program have done all
the identifications for the last 25-30 years, we did not check for changes associated with
taxonomy lab, but we did look for trends associated with changes in taxonomic identification
keys, collection method, reference status, Level 3 ecoregion, and year (in 5-year increments).
We found that samples that were collected using different collection methods, in particular those
collected using the EPT method, tended to form distinct groups (see Appendix A, Figure A-18).
Because of this, we decided to limit the data sets that we analyzed to samples collected using the
standard qualitative "full-scale" method only, because the greatest number of samples were
5-11
-------
collected using this method. This correction also eliminated a spike in the total number of taxa
that occurred in 1998 when a large number of estuarine sites were sampled.
By making this limitation, we lost 4 years of data (1978-1981) and reduced the number
of unique stations, but this was a necessary and effective step in minimizing the chances of
detecting false trends. In addition to collection method, we also found that taxonomic
composition was influenced by ecoregion (see Appendix A, Figures A-22A and A-22B). We
tried to account for this in our analyses, where appropriate, by limiting samples to a particular
ecoregion. An exception is the maximum likelihood temperature optima and tolerance
calculations that are discussed in Section 5.4.1, for which sample size was an issue, and having a
wide range of temperatures was needed and appropriate.
Most of the biological sampling sites that were sampled using the full-scale collection
method have fewer than 5 years of data (see Table 5-6). There are nine sites that have 10 or
more years of data (see Table 5-6). NCDENR considers one of these long-term sites to be in
reference (least-disturbed) condition. The NCDENR reference designations were based largely
on land use/land cover in the upstream catchment area and best professional judgment.
Figure 5-6 shows the spatial distribution of all biological sampling sites (not just those sampled
using the full-scale method).
Table 5-6. Distribution of reference and unclassified stations, categorized by
duration of sampling. These numbers apply only to stations that were
sampled using the standard qualitative (full-scale) collection method
# Years
sampled
10+
5 to 9
3 to 4
2
1
Reference
stations
1
2
4
8
12
Unclassified
stations
8
146
182
237
933
5-12
-------
Q^Ti 2rnar ,-, *-* u
—! • — • ty—'fA - '' 'AT
T*T" ^~."^
rrran'? ^ L~ :• ~" ~:'A~
^•fflrf.^
NC Reference Stations NC Non-Reference Stations Ecoregion
Ref status, Yrs data
Ref 10-19 (N=3)
• Ref 5-9 (N=13)
A Ref2-4(N=53)
• Ref 1 (N=36)
Ref status, Yrs data
HH Other 20+ (N=2)
-fr Other 10-19 (N=28)
D Other 5-9 (N=210)
- Other 2-4 (N=841)
Other 1 (N=1600)
LEVEL3_NAME
Middle Atlantic Coastal Plain
Southeastern Plains
Figure 5-6. NCDENR biomonitoring stations, coded by reference status and duration of data (this includes all
sites, not just those sampled using the standard qualitative [full-scale] collection method).
-------
5.3. NORTH CAROLINA DEPARTMENT OF THE ENVIRONMENT AND NATURAL
RESOURCE (NCDENR) METHODS
NCDENR uses several different methods to collect its samples, but for reasons described
in Section 5.2, we focused our analyses on the standard qualitative "full-scale" method samples
only. The full-scale collection method is composed of two kicks, three sweeps, one leaf pack
sample, two fine mesh rock and/or log wash samples, and one sand sample. In addition, crew
members do visual collections during which they walk the stream reach, and sample habitats and
substrate types that might be missed or undersampled by the other collection techniques
(NCDENR, 2006). Abundance data were recorded as rare = 1 (1-2 specimens), common = 3
(3-9 specimens), or abundant (>10 specimens).
NCDENR assigns bioclassification scores of excellent, good, good/fair, fair or poor to
samples collected using the standard qualitative "full-scale" method, per the scoring system
outlined in Table 5-7. Different scoring criteria are applied to the Mountain, Piedmont, and
Coastal Plain regions. Two metrics, the NCBI and number of EPT taxa, are typically considered
when assigning bioclassification scores. The NCBI is calculated like the HBI, except it uses
tolerance values that are derived from the North Carolina database (see NC Standard Operating
Procedures [SOP; NCDENR, 2006] for more details). It documents the contribution of pollution
tolerant taxa to the composition of the community (Hillsenhoff, 1987). The higher the HBI, the
more strongly the community is dominated by taxa tolerant of organic pollution, and the more
impaired the site is considered. The scoring criteria for the EPT richness metric are based on
species- (or lowest) level identifications.
For most sites, when calculating the bioclassification scores, NCDENR gives equal
weight to both the NCBI value and EPT taxa richness. Exceptions are outlined in the NC SOP
(NCDENR, 2006), and include such things as pristine high altitude mountain streams, swamp
streams, and Coastal B streams. If averaging the NCBI and EPT taxa richness results in a final
score midway between two ratings, EPT abundance is taken into account when deciding whether
to round up or round down. As described in Table 5-7, due to seasonal variations in EPT taxa
(i.e., changes in winter/spring Plecoptera), corrections for nonsummer collections are also taken
into account.
5-14
-------
Table 5-7. These tables are used to determine the scores for EPT taxa
richness values and NCBI values for all standard qualitative samples after
seasonal corrections are made. EPT N refers to EPT abundance (from
NCDENR, 2006)
Score
5
4.6
4.4
4
3.6
3.4
3
2.6
2.4
2
1.6
1.4
1
NCBI values
MT
<4.00
4.00-4.04
4.05-4.09
4.10-4.83
4.84-4.88
4.89-4.93
4.94-5.69
5.70-5.74
5.75-5.79
5.80-6.95
6.96-7.00
7.01-7.05
>7.05
P
<5.14
5.14-5.18
5.19-5.23
5.24-5.73
5.74-5.78
5.79-5.83
5.84-6.43
6.44-6.48
6.49-6.53
6.54-7.43
7.44-7.48
7.49-7.53
>7.53
CP
<5.42
5.42-5.46
5.47-5.51
5.52-6.00
6.01-6.05
6.06-6.10
6.11-6.67
6.68-6.72
6.73-6.77
6.78-7.68
7.69-7.73
7.74-7.79
>7.79
EPT values
MT
>43
42-43
40-41
34-39
32-33
30-31
24-29
22-23
20-21
14-19
12-13
10-11
0-9
P
>33
32-33
30-31
26-29
24-25
22-23
18-21
16-17
14-15
10-13
8-9
6-7
0-5
CP
>29
28
27
22-26
21
20
15-19
14
13
8-12
7
6
0-5
Biotic index corrections for nonsummer data:
summer = Jun-Sep; fall = Oct-Nov; winter = Dec-Feb; spring = Mar-May
Mountain correction
Piedmont correction
Coastal Plain correction
Fall
+0.4
+0.1
+0.2
Winter
+0.5
+0.1
+0.2
Spring
+0.5
+0.2
+0.3
Rounding criteria: round down if EPT N < criterion, otherwise round up.
Bioclassification (Score)
Excellent (5) vs. good (4)
Good (4) vs. good-fair (3)
Good-fair (3) vs. fair (2)
Fair (2) vs. poor (1)
MT
191
125
85
45
P
135
103
71
38
CA
108
91
46
18
MT = Mountain, P = Piedmont, CP = Coastal Plain.
5-15
-------
5.4. INDICATORS
5.4.1. Thermal Preference
As described in Section 2.2.1, we used the guidelines of Yuan (2006) to calculate thermal
optima and tolerance values. Because the North Carolina data set is composed of categorical
abundance data, it was more appropriate to derive values using maximum likelihood calculations
instead of weighted averaging. We based our calculations on a subset of the North Carolina
biomonitoring database composed of standard qualitative "full-scale" collection method samples.
These, along with literature, primarily the traits matrix in Poff et al. (2006b) and the USGS traits
database (Vieira et al., 2006), were used as a basis for making some additional initial
designations. We refined the lists based on case studies and best professional judgment from a
regional advisory group. These lists were used to define cold and warm-water taxa for the North
Carolina data set, and are the basis of the region-specific thermal-preference richness and
relative-abundance metrics used in some analyses.
The North Carolina cold-water taxa list is composed of 32 taxa, and the warm-water taxa
list is composed of 27 taxa. Tables 5-8 and 5-9, respectively, list the cold and warm-water taxa,
along with abundance and distribution information. Ten of the cold-water taxa are Dipterans,
eight are Plecopterans, six are Ephemeropterans, and six are Trichopterans. The rest are
Coleopterans and Odonates (see Table 5-8). Seven of the warm-water taxa are Odonates, five
are Dipterans, and four are Trichopterans (see Table 5-9).
The most abundant cold-water taxa are Epeorus (Ephemeroptera), Antocha (Diptera),
Isoperla (Plecoptera), and Tallaperla (Plecoptera). These taxa comprise only 0.4 to 0.6% of the
total individuals in the North Carolina database. Seventeen of the cold-water taxa have overall
abundances of less than 0.1%. Physella (Basommatophora), Chimarra (Trichopteran), and
Macromia (Odonata) are the most abundant warm-water taxa, with overall abundances ranging
from 0.6 to 0.8%. Twelve of the warm-water taxa have overall abundances of less than 0.1%.
Of the cold-water taxa, Antocha occurs at the largest percentage of sites (25%), followed by a
Chironomidae, Eukiefferiella., and a Plecopteran, Isoperla., which occur at 18-19% of the sites.
Eighteen of the cold-water taxa occur at less than 10% of the sites. Among the warm-water taxa,
Physella occurs at the highest percentage of sites (30%), followed by Macromia (29%) and
Stenochironomus (27%). Nineteen of the warm-water taxa occur at less than 10% of the sites.
5-16
-------
Table 5-8. List of North Carolina cold-water temperature indicator taxa. Distribution and abundance
information is also included. Sum_Individuals = the total number of individuals from that taxon in the North
Carolina database; Pct_Abund = percentage of total individuals in the database composed of that taxon;
Num_Stations = number of stations in the database that the taxon occurred at; Pct_Stations = percentage of
stations in the database at which the taxon occurred
Order
Coleoptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Odonata
Family
Elmidae
Athericidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Tipulidae
Tipulidae
Baetidae
Ephemerellidae
Heptagenidae
Heptageniidae
Heptageniidae
Heptageniidae
Gomphidae
Final ID
Promoresia
Atherix
Cardiocladius
Diamesa
Eukiefferiella
Heleniella
Pagastia
Potthastia
Rheopelopia
Antocha
Dicranota
Acentrella
Drunella
Cinygmula
Epeorus
Nixe
Rhithrogena
Lanthus
Sum individs
3,020
1,236
2,300
734
2,974
95
751
757
135
5,103
1,384
2,745
2,846
247
5,226
64
725
1,174
Pct_abund
0.36
0.15
0.27
0.09
0.35
0.01
0.09
0.09
0.02
0.61
0.16
0.33
0.34
0.03
0.62
0.01
0.09
0.14
Num_stations
332
240
376
185
533
50
157
292
64
711
284
427
218
40
403
16
152
300
Pct_stations
11.81
8.54
13.38
6.58
18.96
1.78
5.59
10.39
2.28
25.29
10.1
15.19
7.76
1.42
14.34
0.57
5.41
10.67
-------
Table 5-8. List of North Carolina cold-water temperature indicator taxa. Distribution and abundance
information is also included. Sum_Individuals = the total number of individuals from that taxon in the
North Carolina database; Pct_Abund = percentage of total individuals in the database composed of that
taxon; Num_Stations = number of stations in the database that the taxon occurred at; Pct_Stations =
percentage of stations in the database at which the taxon occurred (cont.)
Order
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Plecoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Family
Nemouridae
Nemouridae
Peltoperlidae
Perlodidae
Perlodidae
Perlodidae
Perlodidae
Perlodidae
Apataniidae
Glossosomatidae
Glossosomatidae
Hydropsychidae
Hydropsychidae
Philopotamidae
Final ID
Amphinemura
Zapada
Tallaperla
Clioperla
Cultus
Diploperla
Isoperla
Malirekus
Apatania
Agape tus
Glossosoma
Arctopsyche
Parapsyche
Dolophilodes
Sum individs
1,210
3
3,337
574
296
393
4,556
753
339
247
1,755
222
280
2,905
Pct_abund
0.14
0
0.4
0.07
0.04
0.05
0.54
0.09
0.04
0.03
0.21
0.03
0.03
0.35
Num_stations
281
3
377
155
70
122
498
132
47
53
309
40
52
316
Pct_stations
10
0.11
13.41
5.51
2.49
4.34
17.72
4.7
1.67
1.89
10.99
1.42
1.85
11.24
oo
-------
Table 5-9. List of North Carolina warm-water temperature indicator taxa. Distribution and abundance
information is also included. Sum_Individuals = the total number of individuals from that taxon in the North
Carolina database; Pct_Abund = percentage of total individuals in the database composed of that taxon;
Num_Stations = number of stations in the database that the taxon occurred at; Pct_Stations = percentage of
stations in the database at which the taxon occurred
Order
Arhynchobdellida
Basommatophora
Coleoptera
Coleoptera
Decapoda
Diptera
Diptera
Diptera
Diptera
Diptera
Ephemeroptera
Hemiptera
Isopoda
Odonata
Odonata
Odonata
Odonata
Odonata
Family
Erpobdellidae
Physidae
Dytiscidae
Hydrophilidae
Palaemonidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Leptohyphidae
Belostomatidae
Asellidae
Calopterygidae
Coenagrionidae
Corduliidae
Corduliidae
Corduliidae
Final ID
Erpobdella/Mooreobdella
Physella
Lioporeus
Berosus
Palaemonetes
Nilothauma
Parachironomus
Pentaneura
Procladius
Stenochironomus
Tricorythodes
Belostoma
Caecidotea
Hetaerina
Ischnura
Epicordulia
Helocordulia
Macromia
Sum_individs
760
6,677
182
1,843
2,262
180
395
771
3,460
3,419
4,939
173
3,203
854
318
178
188
5,064
Pct_abund
0.09
0.79
0.02
0.22
0.27
0.02
0.05
0.09
0.41
0.41
0.59
0.02
0.38
0.1
0.04
0.02
0.02
0.6
Num_stations
210
853
83
277
271
124
128
154
706
750
363
99
544
153
101
78
95
813
Pct_stations
7.47
30.35
2.95
9.85
9.64
4.41
4.55
5.48
25.12
26.68
12.91
3.52
19.35
5.44
3.59
2.77
3.38
28.92
VO
-------
Table 5-9. List of North Carolina warm-water temperature indicator taxa. Distribution and abundance
information is also included. Sum_Individuals = the total number of individuals from that taxon in the North
Carolina database; Pct_Abund = percentage of total individuals in the database composed of that taxon;
Num_Stations = number of stations in the database that the taxon occurred at; Pct_Stations = percentage of
stations in the database at which the taxon occurred (cont.)
Order
Odonata
Odonata
Rhynchobdellida
Rhynchobdellida
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Unionoida
Family
Corduliidae
Corduliidae
Glossiphoniidae
Glossiphoniidae
Dipseudopsidae
Hydropsychidae
Philopotamidae
Polycentropodidae
Unionidae
Final ID
Neurocordulia
Tetragoneuria
Helobdella
Placobdella
Phylocentropus
Macrostemum
Chimarra
Neureclipsis
Elliptic
Sum_individs
1,511
687
835
677
576
1,753
5,178
2,092
1,556
Pct_abund
0.18
0.08
0.1
0.08
0.07
0.21
0.62
0.25
0.18
Num_stations
278
202
225
339
201
134
554
241
189
Pct_stations
9.89
7.19
8
12.06
7.15
4.77
19.71
8.57
6.72
to
o
-------
Most of the taxa on the cold water list are intolerant to enrichment, while most of the
warm-water taxa are tolerant or have intermediate tolerance to enrichment (see Figure 5-7).
Because of this, it may be difficult to tease out whether organisms are responding to changes
associated with warming temperatures or whether they are responding to other stressors, such as
enrichment.
Intolerant Intermediate
En rich ment tolerance
Tolerant
Figure 5-7. Relationship between North Carolina cold- and warm-
water-preference taxa and North Carolina enrichment tolerance scores.
Taxa with enrichment tolerance scores of 0-3 were categorized as Intolerant,
those with scores of 4-6 were Intermediate, and those with scores of 7-10 were
Tolerant.
5.4.2. Hydrologic Indicators
We attempted to develop a list of candidate taxa in North Carolina that could serve as
indicators of hydrologic change. We were able to match USGS gage data with data for
440 biological samples. We calculated IHA parameters and the RBI per the methods described
in Section 2.2.2, and then performed NMDS ordinations on the data set. Results showed
two hydrologic parameters, baseflow index and number of reversals, to have fairly strong
associations with taxonomic composition, but when samples were grouped by ecoregion, it
became apparent that the relationships are most likely driven by the ecoregional distribution of
5-21
-------
taxa (see Figure 5-8). Additional results from our analyses on the paired hydrologic/biological
data set are available upon request.
CM
'X
Basef low Index
No. of Reversal
A
Level III Ecoregion
A Piedmont
A Coastal Plains
TSE Plains
T Blue Ridge
Axis
Figure 5-8. 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.
We also considered results from studies conducted by NCDENR on flow permanence,
flooding, and drought. NCDENR has developed lists of indicator taxa for intermittent and
perennial streams (NCDWQ, 2005). They consider streams to be intermittent if they have water
for a significant part of an average year, but are dry for part of the year, while perennial streams
are defined as those that have water for the entire year. Based on NCDENR's findings,
amphipods, isopods, worms, small elongate Dipteran larvae, winter stoneflies, Dytiscid beetles,
and Hemipterans tend to be more dominant in intermittent conditions (many of these taxa are
5-22
-------
also found in perennial streams). Taxa that require perennial conditions (i.e., water for their
entire life cycle) include mayflies, caddisflies, nonwinter stoneflies, Megalopterans, riffle
beetles, some Dipterans, clams, fish, crayfish, salamanders, and large tadpoles (NCDWQ, 2005).
When NCDENR conducted research on responses of macroinvertebrates to hurricane
flooding that occurred in September 2004, they documented an overall decline in
bioclassification scores (NCDENR, 2005). Mayflies were reduced at all sites, and net-spinning
caddisflies declined at some sites, but the impacts were less severe than expected. Winter
stoneflies and ephemerellid mayflies, which likely hatched after the flooding, were the dominant
taxa at all the sites. In samples collected using the standard qualitative full-scale method, beetles
and odonates declined dramatically. This likely occurred because the woody debris that they
inhabit was swept away in the floods (NCDENR, 2005).
NCDENR also conducted research on responses of macroinvertebrates to drought
conditions that occurred from 1999 to 2002 (NCDENR, 2004). They documented an overall
decline in the macroinvertebrate communities. The degree of impact and speed of recovery
appeared to be influenced by baseflow, drainage area, underlying geology, and type and size of
tributary streams. Baetids and stoneflies recovered quickly, flow-dependent taxa such as
Hydropsychids, Heterocloeon, heptageniids, and Hydroptila were slower to recover, and edge
species such as Triaenodes and Nectopsyche were not present when sites were sampled in 2002.
5.4.3. Traits-Based Indicators in a Warmer Drier Scenario
We developed a list of taxa that may be most and least sensitive to projected changes in
temperature and streamflow based on the suite of trait modalities considered in Section 2.2.3.
When assessing sensitivity to future climatic changes, we focused on a generalized scenario in
which temperatures are increasing, and flows are decreasing during the low flow periods when
state biomonitoring programs typically collect their samples. The taxa in Table 5-10 that are
deemed most sensitive, or most likely to be adversely affected by these projected climatic
changes, are mostly EPT taxa. A Hemipteran, Belostoma, was included on the least sensitive
list. This tax on has the ability to exit (as adults), has high dispersal ability, strong flying
strength, strong swimming ability, and breathes through plastron-spiracles.
5-23
-------
Table 5-10. List of taxa that may be most and least sensitive to a warmer and
drier future scenario based on the combination of traits described in
Section 2.2.3
Order
Diptera
Ephemeroptera
Plecoptera
Plecoptera
Trichoptera
Trichoptera
Trichoptera
Hemiptera
Family
Athericidae
Heptageniidae
Perlodidae
Perlodidae
Hydropsychidae
Hydropsychidae
Philopotamidae
Belostomatidae
Final ID
Atherix
Rhithrogena
Cultus
Diploperla
Arctopsyche
Parapsyche
Dolophilodes
Belostoma
Sensitivity to warmer
drier scenario
most
most
most
most
most
most
most
least
5.5. LEAST-DISTURBED LONG-TERM BIOLOGICAL MONITORING SITES
North Carolina does not have a formal statewide long-term reference monitoring
network. We explored grouping least-disturbed sites together to create ecoregion-specific data
sets that could be analyzed for long-term trends, but site-specific differences were evident within
the data sets, and sample sizes were relatively low; therefore, we focused on data from individual
sites. Five least-disturbed stations (as designated by NCDENR) from the Blue Ridge and
Piedmont ecoregions with long-term biological data were identified and analyzed for temporal
trends. We focused on the Blue Ridge and Piedmont ecoregions because these ecoregions
contain the greatest number of biological sampling sites. Figure 5-9 shows locations of these
five stations. Table 5-11 summarizes site characteristics. Table 5-12 lists the time periods for
which biological data are available for these sites. Biological data were limited to samples
collected during the summer (June-September) index period using the standard qualitative
(full-scale) method.
5-24
-------
NC0109
NC0209
•'
NC0207
NC0248NC0075
""
O Piedmont Sites
0 Blue Ridge Sites
Ecoregions
LEVEL3_NAM
^\ Middle Atlantic Coastal Plain
I Southeastern Plains
Figure 5-9. Locations of the five least disturbed long-term biological
monitoring sites that were examined for long-term trends (NC0109 = New
River; NC0209 = Cataloochee Creek; NC0207 = Nantahala River;
NC0248 = Barnes Creek; NC0075 = Little River).
5-25
-------
Table 5-11. Site characteristics for the long-term biological monitoring stations in North Carolina. Percentage
urban and percentage agricultural (ag) apply to a 1-km buffer zone around each site and are based on 2001
National Land Cover Data. Reference status was designated by NCDENR
Station ID
NC0109
NC0207
NC0209
NC0248
NC0075
Water body — location
New River— SRI 345
Nantahala River — FS
RD437
Cataloochee
Creek— SR 1395
Barnes Creek— SR
1303
Little River— SR 1340
Longitude
(DD)
-81.18330
-83.61916
-83.07277
-80.00055
-79.83220
Latitude
(DD)
36.55220
35.12694
35.66722
35.43861
35.38638
EPA Level 3
ecoregion
Blue Ridge
Blue Ridge
Blue Ridge
Piedmont
Piedmont
Elevation
(m)
713.6
1,878.3
756.9
106.7
149.3
Drainage area
(km2)
2,121.6
134.4
127.4
60.3
223.8
% Urban
3.3
2.6
3
0.6
1.4
%Ag
44a
0.4
0
5.4
0.1
a99.6% pasture/hay.
-------
Table 5-12. Time periods for which biological data were available at the long-term monitoring sites in North
Carolina. Data used in these analyses were limited to samples collected during the summer (June-September)
index period using the standard qualitative (full-scale) method
Station ID
NC0109
NC0207
NC0209
NC0075
NC0248
Water body
New River
Nantahala River
Cataloochee Creek
Little River
Barnes Creek
Number of years of
data analyzed
11
8
7
6
5
Years
1983-1990, 1993, 1998, 2003
1984, 1986, 1988, 1990, 1991, 1994, 1999 and 2004
1984, 1986, 1989, 1990, 1991, 1992 and 1997
1983, 1985, 1988, 1996,2001 and 2006
1985, 1987, 1989, 1996 and 2001
to
-------
5.6. EVIDENCE OF TRENDS AT LEAST-DISTURBED LONG-TERM MONITORING
SITES
5.6.1. New River (NC0109)
The New River (NC0109) site is located in northwestern North Carolina, along State
Route 1345 in Alleghany County. It is in the Blue Ridge ecoregion, has a drainage area of
2,121.6 km2, and an elevation of 713.6 m. Its highest maximum monthly temperatures occur
during August, and lowest average flows (<1,500 cfs) occur from July through October. This
station has 11 years of biological data collected during the summer (June-September) index
period using the standard qualitative (full-scale) method. The period of biological record ranges
from 1983 to 2003. We gathered flow data from 1930-2010 from USGS gage 03164000 (New
River near Galax, VA, Latitude: 36.6473497, Longitude: 80.978969). The gage is located 21 km
northeast of the biological sampling site (as the crow flies). We also gathered daily temperature
and precipitation data from the Sparta 2SE weather station (SitelD 318158, Latitude: 36.4819,
Longitude: 81.0931), which is located approximately 11 km southeast of the biological sampling
site.
Daily precipitation data were available from 1942-2010, while air temperature data were
limited to July 2006-2010. Figure 5-10 shows an aerial photograph of the site, along with the
nearest weather station and active USGS gage.
5.6.1.1. Temporal Trends in Climatic and Biological Variables
Since 1974, mean annual air temperatures at the New River (NC0109) site have ranged
from 9.8 to 12.1°C. Overall, temperatures have shown a slight increase, but there has been a
great deal of year-to-year variability, and this trend is not significant (when fit with a linear trend
r\
line, r = 0.01,/> = 0.62) (see Figure 5-11). Mean annual flow and mean annual precipitation
patterns have also been highly variable over time, with flows ranging from 927 to 3,007 cfs (see
Figure 5-12). Overall, mean annual flows have increased slightly over time, but this trend is not
significant (when fit with a linear trend line, r2 = 0.01, p = 0.50). Precipitation patterns generally
show good correspondence with flow patterns (see Figure 5-12).
5-28
-------
Figure 5-10. Locations of the New River (NC0109) biological sampling site,
USGS gage 03164000 (New River near Galax, VA) and Sparta 2 SE weather
station. Image from Google Earth.
In addition to mean annual values, mean summer flow values were also evaluated, as this
generally corresponds with low flow and potentially physiologically stressful conditions for the
biological organisms. We would have evaluated July/August maximum temperatures from the
nearest weather station as well, but these data were not available for the biological period of
record. From 1983-2003, mean summer flows ranged from 588.7 to 3,073.3 cfs (see
Table 5-13). Bioclassification scores ranged from good (4) (1985-1990) to excellent (5)
(1983-1984, post-1990) (see Figure 5-13 A). The number of EPT taxa and HBI metrics8, which
are used to calculate the bioclassification scores, were variable, with the highest HBI scores
8Because the bioclassification scoring scheme is based on species-level data, bioclassification scores were calculated
based on the original species-level data, while the EPT taxa and HBI metrics shown in Figure 5-14 were calculated
based on the genus-level OTU that we developed for the long-term data set.
5-29
-------
o
12.4
12.2
12.0
11.8
e
| 11.6
CD
|- 11.4
11.2
CD
o:
10.8
10.6
10.4
10.2
10.0
9.8
9.6
1974
1979
1984
1989
1994
1999
2004
Figure 5-11. Yearly trends in PRISM mean annual air temperature (°C) at
the New River (NC0109) site from 1974-2006. Observed temperature data from
the Sparta 2 SE weather station are not shown because they are not available for
the period of biological record. The area shaded in grey corresponds to the period
of biological record. When the observed data are fitted with a linear trend line,
r2 = 0.0l,p = 0.62, and.y = 0.6917 + 0.005 xx.
5-30
-------
t
I
3200
3000
2800
2600
2400
2200
2000
1800
1600
1400
1200
1000
800
Flow
- Precipitation
2000
1800
1600 -
£
'g.
'o
CD
1400 ol
1200
03
ZJ
CD
CD
1000 •%
£
CD
w
800 O
600
1930 1940 1950 1960 1970 1980 1990 2000 2010
Figure 5-12. Yearly trends in mean annual flow (cfs) at the New River
(NC0109) site from 1930-2010, based on data from USGS gage 03164000
(New River near Galax, VA). For comparative purposes, observed annual
precipitation data from the Sparta 2 SE weather station are also included from
1942-2009. The area shaded in grey corresponds to the period of biological
record. When the observed data are fitted with a linear trend line, r2 = 01,
p = 0.50, and.y = -1,012.0647 + 1.4674 x x.
Table 5-13. Range of temperature, precipitation, and flow values that
occurred at the New River (NC0109) during the period of biological record.
Summer = June-September
Parameter
Year
PRISM mean annual air temperature (°C)
Mean annual flow (cfs)
Mean summer flow (cfs)
PRISM mean annual precipitation (mm)
Min
1983
10.0
927.4
588.7
707.7
Max
2003
12.1
2,744
3,073.3
1,581.3
5-31
-------
B
D.
LLJ
£ 4
'to
8 2
i
m
0
42
40
38
36
34
32
30
28
13
12
0)
a.
5 11
10
o:
0.
-•- EPT taxa
-Q- HBI
Temperature
Flow
CD
I
5.4
5.2
5.0
4.8
4.6
4.4
4.2
4.0
3.8
3.6
3.4
3.2
3200
3000
2800
2600
2400
2200
2000
1800
1600
1400
1200
1000
800
600
400
i
1
LJ.
03
E
1983 1988 1993 1998 2003
Figure 5-13. Yearly trends at the New River (NC0109) site in
(A) bioclassification score (based on species-level data); (B) number of EPT
taxa and HBI (based on genus-level OTU); and (C) PRISM mean annual air
temperature (°C) and mean summer (June-September) flow (cfs).
5-32
-------
occurring in the late 1980s and improving since the early 1990s (see Figure 5-14B). During the
period of biological record, mean annual air temperatures and summer flows were highly
variable, with the highest annual temperature occurring in 1990, the lowest summer flow
occurring in 1988, and the highest summer flow occurring in 1989 (see Figure 5-14C). More
warm-water than cold-water taxa are present at this site. The number of cold-water taxa has
increased since the early 1990s (see Figures 5-14A and B).
Anthropogenic influence is higher than desired at this site (44% agricultural, 99.6% of
this is pasture hay). Based on Fffil scores, organic enrichment may have influenced the
biological assemblage at this site in the mid- to late-1980s. Habitat index scores were not
available for this site, and confounding factors related to in situ measurements in 1998 and 2003
were not evident. In situ parameter values were within the following ranges:
• DO: 8 to 8.3 mg/L
• pH: 7.5 to 7.7
• Specific conductance: 55 to 70 |imho/cm
• Water temperature: 24.2 to 25 °C
5.6.1.2. Associations Between Biological and Climatic Variables
Kendall tau nonparametric correlations analyses allow examination of associations
between commonly used biological metrics, year, temperature, flow, and precipitation variables
at the New River (NC0109) site. None of the commonly used biological metrics were strongly
associated with PRISM mean annual air temperature, but eight showed strong associations with
flow and/or precipitation variables (see Table 5-14). The directions of the relationships were in
keeping with expectations for five of the metrics. The number of Plecoptera taxa, percentage
EPT individuals, and percentage of Ephemeroptera individuals metrics had strong positive
associations with flow and precipitation variables, while the percentage of noninsect individuals
and HBI metrics were negatively associated with flow and precipitation. If we assume that low
flows are more stressful to organisms, the total number of taxa metric and the Shannon-Wiener
Diversity Index showed unexpected negative relationships with precipitation and flow, while the
percentage of dominant taxon metric had an unexpected strong positive association with mean
summer flow. Three of the metrics showed fairly strong (r > |0.4|) relationships with year. The
5-33
-------
B
11
10
9
8
7
6
5
4
3
2
12
10
> 6
13
o
-•- Cold Water
-E> Warm Water
-•- Cold Water
•&• Warm Water
1983
1988
1993
1998 2003
Figure 5-14. Yearly trends at the New River (NC0109) site in (A) number of
cold and warm-water taxa; (B) percentage cold and warm-water individuals;
and (C) PRISM mean annual air temperature (°C) and mean summer
(June-September) flow (cfs).
5-34
-------
Table 5-14. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics, year, and climatic variables at the New River (NC0109) site. Results are
based on 11 years of data. Entries are in bold text if r > ±0.5 and are highlighted in gray if they are in a
direction opposite of what is expected. Ranges of biological metric values are also included.
Summer = June-September
Biological metric
Total no. taxa
No. EPTtaxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. Intolerant taxa
Percentage EPT individuals
Percentage Ephemeroptera
individuals
Shannon-Wiener Diversity Index
Percentage noninsect individuals
Percentage dominant taxon
Percentage tolerant individuals
Hilsenhoff Biotic Index
Range of metric
values
Min
64
30
13
1
12
1
42.7
25.6
5.3
5.1
5.5
0.3
3.4
Max
107
41
18
6
19
4
74.6
42.4
6.1
18.6
9.0
3.5
5.3
r values (based on Kendall Tau correlations)
Year
-0.49
-0.02
-0.06
0.43
-0.06
0.23
0.49
0.38
-0.27
-0.13
0.09
0.13
-0.42
PRISM mean
annual air
temperature (°C)
0.15
-0.13
-0.15
0.04
-0.02
0.23
-0.05
-0.16
0.13
0.20
-0.24
0.09
0.13
Flow (cfs)
Mean
annual
-0.67
-0.02
0.02
0.23
-0.06
0.08
0.56
0.53
-0.71
-0.71
0.31
-0.16
-0.56
Mean
summer
-0.45
-0.06
0.02
0.51
-0.26
-0.08
0.60
0.56
-0.75
-0.45
0.56
-0.05
-0.67
PRISM mean annual
precipitation (mm)
-0.60
0.17
0.11
0.51
-0.06
0.08
0.67
0.64
-0.67
-0.60
0.35
-0.20
-0.75
-------
total number of taxa metric and the HBI were negatively associated with year, while percentage
of EPT individuals was positively associated with year.
Similar analyses were performed on the thermal preference metrics. None were strongly
associated with PRISM mean annual air temperature (see Table 5-15). The cold water metrics
were positively associated with the flow and precipitation variables, and the warm water metrics
were negatively associated with flow and precipitation. The richness metrics showed stronger
(r > |0.5|) associations with the flow and precipitation variables than the percentage composition
metrics. The number of warm-water taxa metric showed a fairly strong (r > |0.4|) negative
relationship with year.
A subset of biological metrics that have shown responsiveness to hydrologic variables in
other studies (see Section 2, Table 2-5c) was also examined (see Table 5-16). Three of the
metrics—number of collector-filterer taxa, number of collector-gatherer taxa, and percentage
scraper/herbivore individuals—showed strong (r > |0.5|) associations with year. Five metrics
showed strong associations with the precipitation and flow variables. One of these, number of
scraper/herbivore taxa, went against expectations (see Table 2-5c), showing a negative
correlation with flow precipitation. The percentage of erosional individuals metric had a strong
positive association with the flow variables, and the predator metrics were negatively correlated
with flow and precipitation.
5.6.1.3. Groupings Based on Climatic Variables
Samples were partitioned into hottest/coldest/normal year groups and
lowest/normal/highest flow year groups. At the New River (NC0109) site, on average, the
hottest years were 1.5°C warmer than the coldest years, and highest flow years had 477 more cfs
than lowest flow years. When samples were grouped based on temperature, there were no
significant (p > 0.05) differences between any of the mean metric values, but the number of
warm-water taxa metric was unexpectedly highest in the coldest year samples, and the number of
cold-water taxa was highest in the normal year samples (see Table 5-17). When samples were
grouped based on mean annual flow, the number of warm-water taxa metric was significantly
higher (p < 0.05) in the driest flow years versus the normal flow years (see Table 5-18).
5-36
-------
Table 5-15. Kendall tau nonparametric correlations analyses performed to examine associations between
thermal preference metrics, year, and climatic variables at the New River (NC0109) site. Results are based on
11 years of data. Entries are in bold text if r > ±0.50. Ranges of biological metric values are also included.
Summer = June-September
Biological metric
No. cold-water taxa
Percentage cold-water individuals
No. warm -water taxa
Percentage warm-water individuals
Range of metric
values
Min
3
1.0
6
5.2
Max
8
8.4
10
11.1
r values (based on Kendall Tau correlations)
Year
0.12
0.05
-0.46
0.05
PRISM mean
annual air
temperature (°C)
-0.32
-0.27
-0.14
0.02
Flow (cfs)
Mean
annual
0.76
0.35
-0.54
-0.16
Mean
summer
0.68
0.45
-0.50
-0.42
PRISM mean
annual
precipitation (mm)
0.72
0.45
-0.54
-0.35
-------
Table 5-16. Kendall tau nonparametric correlations analyses performed to examine associations between a
subset of biological metrics, year, flow, and precipitation variables at the New River (NC0109) site. The subset
of biological metrics were selected per the criteria outlined in Section 2 and have shown responsiveness to
hydrologic variables in other studies (see Section 2, Table 2-5c). Results are based on 11 years of data. Entries
are in bold text if r > ± 0.5 and are highlighted in gray if they are in a direction opposite of what is expected.
Ranges of biological metric values are also included. Summer = June-September
Biological metric
Richness
Percentage
individuals
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Range of metric
values
Min
9
19
12
15
6
11
5
21
10.2
25.0
14.4
14.9
9.3
10.1
4.1
22.2
Max
13
31
20
34
11
20
10
29
21.3
36.2
22.7
30.8
22.4
26.6
10.6
42.5
r values (based on Kendall Tau correlations)
Year
-0.58
-0.58
-0.14
-0.19
0.12
0.09
-0.33
0.12
0.35
-0.42
0.60
-0.02
0.38
0.27
-0.24
0.45
Flow (cfs)
Mean
annual
-0.26
-0.43
-0.65
-0.72
0.00
-0.17
-0.25
-0.04
0.27
-0.05
0.24
-0.45
0.38
-0.16
-0.02
0.53
Mean
summer
-0.18
-0.31
-0.65
-0.65
-0.16
0.17
-0.25
-0.04
0.53
0.05
0.20
-0.71
0.27
-0.13
-0.05
0.56
PRISM mean annual
precipitation (mm)
-0.22
-0.31
-0.61
-0.69
-0.16
-0.02
-0.06
0.08
0.45
0.13
0.05
-0.56
0.42
-0.13
0.09
0.35
-------
Table 5-17. Mean metric values (±1 SD) for the New River (NC0109) site in coldest, normal, and hottest year
samples. Year groups are based on PRISM mean annual air temperature values. One-way ANOVA was done
to evaluate differences in mean metric values. There were no significant differences across year groups
(p > 0.05)
Year group
Coldest
Normal
Hottest
No. total
taxa
86.0 ±7.0
84.2 ±6.8
84.7 ±21.5
No. EPT
taxa
34.0 ±1.7
34.4 ±3.8
34.3 ±4.0
HBI
4.5 ±0.4
4.2 ±0.7
4.4 ±0.5
No. cold-
water taxa
4.3 ± 1.5
5.4 ± 1.7
4.0 ± 1.7
No. warm-
water taxa
8.3 ±0.6
7.4 ± 1.7
7.3 ±2.3
% Cold-water
individuals
2.3 ±0.7
3.6 ±2.9
2.2 ±1.0
% Warm-water
individuals
7.7 ±2.5
7.6 ±2.5
7.0 ± 1.3
-------
The number of EPT taxa metric and the cold-water taxa metrics were highest in the highest flow
year samples (p > 0.05).
5.6.2. Nantahala River (NC0207)
The Nantahala River (NC0207) site is located in southwestern North Carolina, along
Forest Service Road 437 in Macon County. It is in the Blue Ridge ecoregion, has a drainage area
r\
of 134.4 km , and an elevation of 1,878.3 m. Most of the upstream catchment is in the Nantahala
National Forest. The highest maximum monthly temperatures at this site occur during July and
August, and the lowest average flows (<120 cfs) occur from August through October. This
station has 8 years of biological data collected during the summer (June-September) index
period using the standard qualitative (full-scale) method. The period of biological record ranges
from 1984 to 2004.
We gathered flow data from 1941-2010 from USGS gage 03504000 (Nantahala River
near Rainbow Springs) Latitude: 35.1275, Longitude: 83.61861), which is colocated at the
biological sampling site. We also gathered daily temperature and precipitation data from
1946-2010 from the Franklin weather station (SitelD 313228, Latitude: 35.1803, Longitude:
83.61861), which is located approximately 21 km east/northeast of the biological sampling site.
Figure 5-15 shows an aerial photograph of the site, along with the nearest weather station and
active USGS gage.
5.6.2.1. Temporal Trends in Climatic and Biological Variables
Since 1946, mean annual air temperatures at the Franklin weather station have ranged
from 12.2 to 15.0°C. There has been a lot of year-to-year variability, but overall, observed
temperatures at the weather station have decreased over time (when fit with a linear trend line,
r2= 0.11 andp = 0.01) (see Figure 5-16). When PRISM air temperature data are compared to
observed data from 1974-2006, the PRISM data are 0.1-3°C lower than observed values, and
the PRISM data show an increasing trend. Because the weather station is located more than
20 km from the biological sampling site and is at a lower elevation (648 m vs. 1,878 m), the
PRISM data are likely more representative of conditions at the biological sampling. Mean
annual flow and mean annual precipitation patterns have been highly variable over time (see
Figure 5-17). Since 1941, mean annual flow values have ranged from 119.5 to 302.4 cfs (when
5-40
-------
fit with a linear trend line, r = 0.00, andp = 0.99). Precipitation patterns show good
correspondence with flow patterns (see Figure 5-17). In addition to mean annual values, mean
maximum July/August temperature and mean summer flow values were also evaluated, as these
are likely to be physiologically stressful time periods for the biological organisms. During the
period of biological record (1984-2004), mean maximum July/August air temperatures ranged
from 27.0-31.4°C, and mean summer flow values ranged from 54.4 to 299.8 cfs (see
Table 5-19).
Nantanala gage
ImageiyOate: 5/20/2009
62012 Google
Image U.S. Geological Survey
• I Image ©2012 GeoEye
35'08'50 26" N 63E29'24 07" W elev 2256
Google earth
Eye ait 1B37mi
Figure 5-15. Locations of the Nantahala River (NC0207) biological sampling
site, USGS gage 03504000 (Nantahala River near Rainbow Springs) and
Franklin weather station. Image from Google Earth.
5-41
-------
16
15
5"
V 14
TO
05
f- 13
,05
I12
11
10
Observed
PRISM
1946
1957
1967
1977
1987
1997
2007
Figure 5-16. Yearly trends in observed mean annual air temperature (°C) at
the Franklin weather station from 1946-2010. For comparative purposes,
PRISM mean annual air temperature data are also included from 1974-2006. The
area shaded in grey corresponds to the period of biological record. When the
observed data are fitted with a linear trend line, r2= 0.1 l,p = 0.01, and
y= 5.7408-0.0113 x x.
5-42
-------
320
300
2000
100
800
1941
1951
1961
1971
1981
1991
2001
Figure 5-17. Yearly trends in mean annual flow (cfs) at the Nantahala River
(NC0207) site from 1941-2010, based on data from USGS gage 03504000
(Nantahala River near Rainbow Springs). For comparative purposes, observed
annual precipitation data from the Franklin weather station are also included from
1946-2010. The area shaded in grey corresponds to the period of biological
record. When the observed data are fitted with a linear trend line, r2 = 0.00,
p = 0.99, and.y = 198.1436 + 0.0022 x x.
Table 5-19. Range of temperature, precipitation, and flow values that
occurred at the Nantahala River (NC0207) site during the period of
biological record. Summer = June-September
Parameter
Year
PRISM mean annual air temperature (°C)
Observed mean maximum July air temperature (°C)
Mean annual flow (cfs)
Mean summer flow (cfs)
PRISM mean annual precipitation (mm)
Min
1984
10
27.0
119.5
54.4
1,337.9
Max
2004
12.6
31.4
302.4
299.8
2,351.9
5-43
-------
This site has received bioclassification scores of excellent (5) over the period of record
(see Figure 5-18A), with consistently low HBI scores (the highest HBI score was a 3.65, which
occurred in 1984), and high numbers of EPT taxa (34 or more [calculated using a genus-level
OUT]) (see Figure 5-18B). During the period of biological record, mean maximum July/August
air temperatures and summer flows were highly variable, with the highest maximum July/August
temperature occurring in 1993, the lowest summer flows occurring in 1985 and 1999, and the
highest summer flows occurring in 1988 and 2004 (see Figure 5-18C). The cold-water taxa
metrics have been consistently high over time (13 or more taxa, comprising 17% or more of the
assemblage) (see Figures 5-19A and B). Very few warm-water taxa are present at this site, with
richness values ranging from 0 to 2.
Confounding factors related to land use appear to be minimal at this site (<3% urban and
<0.5% agricultural within a 1-km buffer). Habitat index scores from 1999 and 2004 are in the
range of "natural" condition (scores range from 83 to 87, with a maximum possible score of
100).
Confounding factors related to in situ measurements were not evident, with values in the
following ranges:
• DO: 7.8 to 9.0 mg/L
• pH: 6.9 to 7.0
• Specific conductance: 16 to 17 |imho/cm
• Water temperature: 16.0 to 16.9 °C
5.6.2.2. Associations Between Biological Variables and Climatic Variables
This site did not have an appropriate data set for performing Kendall tau nonparametric
correlations analyses (less than 9 years of data, gaps between data collection years).
5.6.2.3. Groupings Based on Climatic Variables
This site did not have an appropriate data set for performing analyses on biological data
grouped by extremes in temperature, flow, and/or precipitation variables (less than 9 years of
data, gaps between data collection years).
5-44
-------
5-45
-------
A
I 3
o
g
ixi
1
B
Q_
UJ
O
o
£
3
2
0)
Q.
C/3
=3
en
I
0
44
42
40
38
36
34
32
32
31
30
29
28
27
26
EPT taxa
HBI
•*- Temperature
••• Flow
3.8
3.6
3.4
3.2
3.0
2.8
2.6
2.4
320
300
280
260
240
220
200
180
160
140
120
100
80
60
40
CD
I
1984
1989
1994
1999
2004
Figure 5-18. Yearly trends at the Nantahala River (NC0207) site in
(A) bioclassification score (based on species-level data); (B) number of EPT
taxa and HBI (based on genus-level OTU); and (C) observed mean
July/August maximum air temperature (°C) and mean summer
(June-September) flow (cfs).
5-46
-------
B
18
16
14
12
TO
,8 10
o 8
6
z 6
4
2
30
25
20
15
10
5
o
32
31
30
29
28
27
26
-•- Cold Water
-•'•• Warm Water
H,
o 'B
-•- Cold Water
-E>- Warm Water
n Q-a
Temperature
Flow
320
300
280
260
240
220
200
180
160
140
120
100
80
60
40
1984
1989 1994 1999 2004
Figure 5-19. Yearly trends at the Nantahala River (NC0207) site in (A)
number of cold and warm-water taxa; (B) percentage cold and warm-water
individuals; and (C) observed mean July/August maximum air temperature
(°C) and mean summer (June-September) flow (cfs).
5-47
-------
5.6.3. Cataloochee Creek (NC0209)
The Cataloochee Creek (NC0209) site is located in the Great Smokey Mountains
National Park in western North Carolina, along State Route 1395 in Haywood County. It is in
the Blue Ridge ecoregion, has a drainage area of 127.4 km2, and an elevation of 756.9 m. The
highest maximum monthly temperatures at this site occur during July and August, and lowest
average flows (<200 cfs) occur from September through November. This station has 7 years of
biological data collected during the summer (June-September) index period using the standard
qualitative (full-scale) method. The period of biological record ranges from 1984 to 1997.
We gathered flow data from 1935-2010 from USGS gage 03460000 (Cataloochee Creek
near Cataloochee; Latitude: 35.6675, Longitude: 83.07361), which is colocated with the
biological sampling site. We also gathered daily temperature and precipitation data from the
Cataloochee weather station (SitelD 311564, Latitude: 35.6375, Longitude: 83.0958), which is
located approximately 4 km south/southwest of the biological sampling site. Precipitation data
were available from 1949-2009, while temperature data were available starting in 1966.
Figure 5-20 shows an aerial photograph of the site, along with the nearest weather station and
active USGS gage.
5.6.3.1. Temporal Trends in Climatic and Biological Variables
Since 1966, mean annual air temperatures at the Cataloochee weather station have ranged
from9.1 to 13.1°C. There has been a lot of year-to-year variability, but overall, observed
temperatures at the weather station have increased over time (when fit with a linear trend line,
r\
r = 0.06, andp = 0.12) (see Figure 5-21). When PRISM air temperature data are compared to
observed data from 1974-2006, the PRISM data are within 2°C of the observed values, and the
patterns in the PRISM data show good correspondence with patterns in the observed data. Mean
annual flow and mean annual precipitation patterns have been highly variable over time (see
Figure 5-22). Since 1935, mean annual flow values have ranged from 54.2 to 168.3 cfs (when fit
with a linear trend line, r2 = 0.00, p = 0.90). Precipitation patterns show good correspondence
with flow patterns (see Figure 5-22). During the period of biological record (1984-1997), mean
maximum air temperatures during the hottest months (July and August) ranged from
19.5—21.2°C, and mean summer flows ranged from 30.8 to 135.8 cfs (see Table 5-20).
5-48
-------
Figure 5-20. Locations of the Cataloochee Creek (NC0209) biological
sampling site, USGS gage 03460000 (Cataloochee Creek near Cataloochee)
and Cataloochee weather station. Image from Google Earth.
5-49
-------
14
13
o
o
CD
12
CD
CL
E
03
CO
0)
11
10
9
8
Observed
-•- PRISM
1966
1971
1978
1983
1988
1993
1998
2003
2008
Figure 5-21. Yearly trends in observed mean annual air temperature (°C) at
the Cataloochee weather station from 1966-2009. For comparative purposes,
PRISM mean annual air temperature data are also included from 1974-2006. The
area shaded in grey corresponds to the period of biological record. When the
observed data are fitted with a linear trend line, r = 0.06, p = 0.12, and
-21.26 + 0.0163 xx.
5-50
-------
180
160
140
I 120
03
3
ro 100
ro
80
60
40
1800
Flow
Precipitation
600
1935
1945
1955
1965
1975
1985
1995
2005
Figure 5-22. Yearly trends in mean annual flow (cfs) at the Cataloochee
Creek (NC0209) site from 1935-2010, based on data from USGS gage
03460000 (Cataloochee Creek near Cataloochee). For comparative purposes,
observed annual precipitation data from the Cataloochee weather station are also
included from 1949-2009. The area shaded in grey corresponds to the period of
biological record. When the observed data are fitted with a linear trend line,
r1= 0.00,/? = 0.90, and.y = 73.7837 + 0.0185 x x.
Table 5-20. Range of temperature, precipitation, and flow values that
occurred at the Beaver River site (UT 5940440) during the period of
biological record
Parameter
Year
Observed annual air temperature (°C)
PRISM mean annual air temperature (°C)
Observed mean maximum July/August air temperature (°C)
Mean annual flow (cfs)
Observed annual precipitation (mm)
PRISM mean annual precipitation (mm)
Mean summer flow (cfs)
Min
1984
9.8
9.7
19.5
57.7
1,003.7
1,014.4
30.8
Max
1997
12.6
11.5
21.2
168.3
1,715.3
1,664.7
135.8
5-51
-------
This site has received bioclassification scores of excellent (5) over the period of record
(see Figure 5-23 A). HBI scores have been low (less than 3.3) and variable. The number of EPT
taxa metric, which ranged from 34 to 42 (based on a genus-level OTU), also varied from year to
year, increasing from 1984 to 1989 and then decreasing from 1989 to 1992 (see Figure 5-23B).
During the period of biological record, there was a fair amount of year of year variability
in the mean maximum July/August air temperatures and summer flows (see Figure 5-23C).
From 1986-1988, conditions for the biological organisms may have been particularly stressful,
with high temperatures and very low summer flows. Biological responses to these conditions
were not evident, but this may have been due in part to gaps in the biological data. The
cold-water taxa metrics have been consistently high over time (16 or more taxa, comprising 25%
or more of the assemblage) (see Figures 5-24A and B). Very few warm-water taxa are present at
this site (richness values range from 0 to 2).
Confounding factors related to land use appear to be minimal at this site (<3% urban and
0% agricultural within a 1-km buffer). This site received a habitat index score of 93 in 1997 (out
of a possible score of 100), which is in the range of "natural" condition. Confounding factors
related to in situ measurements were not evident, with values in the following ranges:
• DO: 9.0 mg/L
• pH: 6.9
• Specific conductance: 10 to 16 |imho/cm
• Water temperature: 17.6tol8.0°C
5.6.3.2. Associations Between Biological Variables and Climatic Variables
This site did not have an appropriate data set for performing Kendall tau nonparametric
correlations analyses (less than 9 years of data, gaps between data collection years).
5.6.3.3. Groupings Based on Climatic Variables
This site did not have an appropriate data set for performing analyses on biological data
grouped by extremes in temperature, flow, and/or precipitation variables (less than 9 years of
data, gaps between data collection years).
5-52
-------
B
Q.
LLJ
M—
O
o
i*
0)
to
8 2
o
g
in
1
0
43
42
41
40
39
38
37
36
35
34
33
22
O
£
| 21
c/)
13
cn
-|- 20
—>
X.
ro
19
EPT taxa
- HBI
Temperature
Flow
3.4
3.3
3.2
3.1
3.0
2.9
2.8
2.7
2.6
2.5
2.4
160
140
120
100
80
60
40
20
I
O
1984
1988
1992
1996
Figure 5-23. Yearly trends at the Cataloochee Creek (NC0209) site in
(A) bioclassification score (based on species-level data); (B) number of EPT
taxa and HBI (based on genus-level OTU); and (C) observed mean
July/August maximum air temperature (°C) and mean summer (June-
September) flow (cfs).
5-53
-------
A
B
24
22
20
18
16
SS 14
i™
t 12
o
d 10
z 8
6
4
2
0
40
35
30
a> 25
CD
| 20
^
-E 15
6?
10
5
0
22
O
O
I 21
,o>
1/3
3
CD
>- 20
x
ro
19
-•- Cold Water
-o- Warm Water
-•- Cold Water
-*-- Warm Water
Temperature
Flow
160
140
120
100
80
0)
E
V)
60 §
40
20
1984
1988
1992
1996
Figure 5-24. Yearly trends at the Cataloochee Creek (NC0209) site in (A)
number of cold and warm-water taxa; (B) percentage cold and warm-water
individuals; and (C) observed mean July/August maximum air temperature
(°C) and mean summer (June-September) flow (cfs).
5-54
-------
5.6.4. Barnes Creek (NC0248)
The Barnes Creek (NC0248) site is located in central North Carolina, along State Route
1303 in Montgomery County. It is in the Piedmont ecoregion, has a drainage area of 60.3 km2
and an elevation of 106.7 m. The highest maximum monthly temperatures at this site occur
during July and August, and lowest rainfall occurs from October through December. This station
has 5 years of biological data collected during the summer (June-September) index period using
the standard qualitative (full-scale) method. The period of biological record ranges from 1985 to
2001.
We gathered daily temperature and precipitation data from the Albemarle weather station
(SitelD 310090, Latitude: 35.3992, Longitude: 80.1994), which is located approximately 19 km
southwest of the biological sampling site. Data were available from 1912-2010 (with some
gaps). There were no USGS gages located in proximity to the biological sampling site.
Figure 5-25 shows an aerial photograph of the biological sampling site and the Albemarle
weather station.
5.6.4.1. Temporal Trends in Climatic and Biological Variables
Since 1912, mean annual air temperatures at the Albemarle weather station have ranged
from 14.3 to 17.4°C. There has been a lot of year-to-year variability, but overall, observed
temperatures at the weather station have decreased over time (when fit with a linear trend line,
r2= 0.03 andp = 0.09) (see Figure 5-26). When PRISM air temperature data are compared to
observed data from 1974-2006, the PRISM data are within 1.1 °C of the observed values and
correspond closely with the patterns seen in the observed data. Mean annual precipitation
patterns have been highly variable over time (see Figure 5-27). Since 1912, mean annual
precipitation values have ranged from 743.8 to 1,626.1 mm (when fit with a linear trend line,
9
r = 0.00 andp = 0.72). The PRISM precipitation data correspond closely with the observed
data. During the period of biological record (1985 to 2001), mean maximum air temperatures
during the hottest months (July and August) ranged from 29.8-33.7°C, and mean summer
precipitation ranged from 162.4 to 688.6 mm (see Table 5-21).
5-55
-------
Figure 5-25. Locations of the Barnes Creek (NC0248) biological sampling
site and the Albemarle weather station. Image from Google Earth
5-56
-------
18
o
o
2> 17
35
I
CD
c
CD
CD
16
15
14
1912 1922 1932 1943 1953 1963 1973 1983 1993 2003
Figure 5-26. Yearly trends in observed mean annual air temperature (°C) at
the Albemarle weather station from 1912-2009. For comparative purposes,
PRISM mean annual air temperature data are also included from 1974-2006. The
area shaded in grey corresponds to the period of biological record. When the
observed data are fitted with a linear trend line, r = 0.03, p = 0.09, and
y = 23.6125-0.004 XJK.
5-57
-------
700
1912 1922 1932 1942 1952 1962 1972 1982 1992 2002
Figure 5-27. Yearly trends in mean annual precipitation (mm) at the
Albemarle weather station from 1912-2010. For comparative purposes,
PRISM mean annual precipitation data are also included from 1974-2006. The
area shaded in grey corresponds to the period of biological record. When the
observed data are fitted with a linear trend line, r2= 0.00, p = 0.72, and
y =1668.9664-0.2443 XJK.
Table 5-21. Range of temperature and precipitation values that occurred at
the Barnes Creek (NC0248) site during the period of biological record
Parameter
Year
Observed annual air temperature (°C)
PRISM mean annual air temperature (°C)
Observed mean maximum My/August air temperature (°C)
Observed annual precipitation (mm)
PRISM mean annual precipitation (mm)
Observed mean summer precipitation (mm)
Min
1985
14.3
14.6
29.8
743.8
750.8
162.4
Max
2001
17.4
16.8
33.7
1,546.2
1,435.4
688.6
5-58
-------
Bioclassification scores at this site have ranged from good (4) in 1987 and 1989 to
excellent (5) (see Figure 5-28A). Since 1989, the number of EPT taxa has increased from 22 to
33 (based on a genus-level OTU), and HBI scores have decreased from a high of 4.6 in 1987 to a
low of 3.9 in 2001 (see Figure 5-28B). During the period of biological record, there was a lot of
year of year variability in the mean maximum July/August air temperatures and summer
precipitation patterns. The highest maximum temperatures occurred in 1993 and 1987, the
lowest summer rainfall occurred in 1990, and the highest summer rainfall occurred in 1989 (see
Figure 5-28C). There are similar numbers of cold and warm-water taxa at this site, with richness
numbers ranging from 3 to 6, and each comprising less than 10% of the assemblage (see
Figures 5-29A and B). No clear patterns in the cold and warm water metrics are evident over
time.
There may be some confounding factors related to agricultural land use in the
surrounding area (5.4% agricultural within a 1-km buffer). Since 1996, this site has received
habitat index scores ranging from 87 to 90, which are in the range of "natural" condition.
Confounding factors related to in situ measurements were not evident, with values in the
following ranges:
• DO: 7.3 to 11.7mg/L
• pH: 7.2 to 7.6
• Specific conductance: 40 to 61 |imho/cm
• Water temperature: 16 to 25 °C
5.6.4.2. Associations Between Biological Variables and Climatic Variables
This site did not have an appropriate data set for performing Kendall tau nonparametric
correlations analyses (less than 9 years of data, gaps between data collection years).
5.6.4.3. Groupings Based on Climatic Variables
This site did not have an appropriate data set for performing analyses on biological data
grouped by extremes in temperature, flow, and/or precipitation variables (less than 9 years of
data, gaps between data collection years).
5-59
-------
A
B
0
34
32
30
i? 28
o.
ai
"5 26
6
24
22
20
34
o
o
£
33
o>
Q.
I 32
10
=1
O)
31
30
29
-•- EPTtaxa
B HBI
Temperature
Precipitation
4.7
4.6
4.5
4.4
4.3
4.2
4.1
4.0
3.9
3.8
800
700
600
500
400
300
200
100
E
c
o
•^
£
Q.
CD
(D
1985
1990
1995
2000
Figure 5-28. Yearly trends at the Barnes Creek (NC0248) site in
(A) bioclassification score (based on species-level data); (B) number of EPT
taxa and HBI (based on genus-level OTU); and (C) observed mean
July/August maximum air temperature (°C) and mean summer
(June-September) precipitation (mm).
5-60
-------
B
(0
X A
,(D 4
o
ci 3
0
10
- 7
m '
> 6
2
34
O
CD
3 33
"ro
o3
CL
.o5 32
CO
Z3
D5
31
jo 30
29
Cold Water
Warm Water
-•- Cold Water
-a- Warm Water
Temperature
Precipitation
800
700
E
600 .1
.1
500 '§"
CD
CL
400 c5
300 =
c
CO
200 5
100
1985
1990
1995
2000
Figure 5-29. Yearly trends at the Barnes Creek (NC0248) site in (A) number
of cold and warm-water taxa; (B) percentage cold and warm-water
individuals; and (C) observed mean July/August maximum air temperature
(°C) and mean summer (June-September) precipitation (mm).
5-61
-------
5.6.5. Little River (NC0075)
The Little River (NC0075) site is located in central North Carolina, along State Route
1340 in Montgomery County. It is in the Piedmont ecoregion, has a drainage area of 223.8 km2,
and an elevation of 149.3 m. The highest maximum monthly temperatures at this site occur
during July and August, and the lowest average flows (<60 cfs) occur from July through
September. This station has 6 years of biological data collected during the summer
(June-September) index period using the standard qualitative (full-scale) method. The period of
biological record ranges from 1983 to 2006.
We gathered flow data from 1955-2010 from USGS gage 02128000 (Little River near
Star, NC Latitude: 35.38722, Longitude: 79.83139), which is colocated with the biological
sampling site. We also gathered daily temperature and precipitation data from the Jackson
Springs 5 WNW weather station (SiteJD 314464, Latitude: 35.1858, Longitude: 79.6772), which
is located approximately 27 km southeast of the biological sampling site. Data were available
from 1953-2010. Figure 5-30 shows an aerial photograph of the site, along with the nearest
weather station and active USGS gage.
5.6.5.1. Temporal Trends in Climatic and Biological Variables
Since 1953, mean annual air temperatures at the Jackson Springs 5 WNW weather station
have ranged from 14.7 to 17.3°C. There has been a lot of year-to-year variability, but overall,
observed temperatures at the weather station have increased over time (when fit with a linear
trend line, r2 = 0.02 and/? = 0.31) (see Figure 5-31). When PRISM air temperature data are
compared to observed data from 1974-2006, the PRISM data are within 1°C of the observed
values, and the patterns in the PRISM data show very close correspondence with patterns in the
observed data.
Mean annual flow and mean annual precipitation patterns have been highly variable over
time (see Figure 5-32). Since 1955, mean annual flow values have ranged from 30.7 to 216.2 cfs
(when fit with a linear trend line, r2 = 0.01 and/? = 0.42). Precipitation patterns show fairly close
correspondence with flow patterns (see Figure 5-32). During the period of biological record
(1983-2006), mean maximum air temperatures during the hottest months (July and August)
ranged from 24.4-26.9°C, and mean summer flows ranged from 10.2 to 241.2 cfs (see
Table 5-22).
5-62
-------
Figure 5-30. Locations of the Little River (NC0075) biological sampling site,
USGS gage 02128000 (Little River near Star, NC) and the Jackson Springs 5
WNW weather station. Image from Google Earth.
5-63
-------
O
o
18
17
Q)
I 16
|2
:± 15
CD
=3
C
< 14
CD
CD
13
Observed
PRISM
1953
1963
1973
1983
1993
2003
Figure 5-31. Yearly trends in observed mean annual air temperature (°C) at
the Jackson Springs 5 WNW weather station from 1953-2010. For
comparative purposes, PRISM mean annual air temperature data are also included
from 1974-2006. The area shaded in grey corresponds to the period of biological
record. When the observed data are fitted with a linear trend line, r = 0.02,
p = 0.31, andy= 6.0624 + 0.0048 x x.
5-64
-------
a
03
240
220
200
180
160
140
120
3 100
80
60
40
20
Flow
Precipitation
1700
1600
1500
E
E,
o
'
1400 -
'o
1300
1200
1100
03
0>
1000 I
in
_Q
O
900
800
1953
1963
1973
1983
1993
2003
Figure 5-32. Yearly trends in mean annual flow (cfs) at the Little River
(NC0075) site from 1955-2010, based on data from USGS gage 02128000
(Little River near Star, NC). For comparative purposes, observed annual
precipitation data from the Jackson Springs 5 WNW are also included from
1953-2010. The area shaded in grey corresponds to the period of biological
record. When the observed data are fitted with a linear trend line, r2 =0.01,
p = 0.42, and.y = 644.3522 - 0.2707 x x.
Table 5-22. Range of temperature, precipitation, and flow values that
occurred at the Little River (NC0075) during the period of biological record
Parameter
Year
Observed annual air temperature (°C)
PRISM mean annual air temperature (°C)
Observed mean maximum July/August air temperature (°C)
Mean Annual flow (cfs)
Observed annual precipitation (mm)
PRISM mean annual precipitation (mm)
Mean summer flow (cfs)
Min
1983
14.8
14.9
24.4
30.7
884.2
846.8
10.2
Max
2006
17.3
17.1
26.9
216.2
1,578.4
1,541.9
241.2
5-65
-------
This site has received bioclassification scores ranging from good (4) in 1983 to excellent
(5) (see Figure 5-33 A). HBI scores have been variable over time, ranging from 4.0 to 4.7. The
number of EPT taxa also varied from year to year, ranging from a low of 22 in 1983 to a high of
32 in 1988 (based on a genus-level OTU) (see Figure 5-33B). During the period of biological
record, there was a lot of year of year variability in the mean maximum July/August air
temperatures, and mean summer flows were much higher than normal in 2003 and 1997 (see
Figure 5-33C). There are more warm-water taxa than cold-water taxa at this site (4 to
7 warm-water taxa vs. 1 to 2 cold-water taxa), but warm-water taxa only comprise a small
proportion of the assemblage (less than 6%) (see Figures 5-34A and B). The cold and warm
water metrics did not show clear trends over time.
Confounding factors related to land use appear to be minimal at this site (<1.5% urban
and 0.1% agricultural within a 1-km buffer). Habitat index scores at this site have ranged from
71 (moderate) in 2001 to 80 (natural) in 1996. Confounding factors related to in situ
measurements were not evident, with values in the following ranges:
• DO: 6.7 to 8.1 mg/L
• pH: 6.8 to 7.3
• Specific conductance: 60 to 80 |imho/cm
• Water temperature: 24.2 to 27 °C
5.6.5.2. Associations Between Biological Variables and Climatic Variables
This site did not have an appropriate data set for performing Kendall tau nonparametric
correlations analyses (less than 9 years of data, gaps between data collection years).
5.6.5.3. Groupings Based on Climatic Variables
This site did not have an appropriate data set for performing analyses on biological data
grouped by extremes in temperature, flow, and/or precipitation variables (less than 9 years of
data, gaps between data collection years).
5-66
-------
CD
I
5
CD
84
CO
1 3
t^
't/>
-------
A
B
S 5
O
o
CD
ZJ
•o
:> s
-0
C
1
0
28
27
CD
Q.
I 26
C/5
Z5
cn
25
ro 24
I
23
-•- Cold Water
-B- Warm Water
Cold Water
Warm Water
Temperature
Flow
260
240
220
200 &
o
180 ~
160 2
140 fe
120 |
100
80
60
40
20
0
CO
CD
1983 1988 1993 1998
2003
Figure 5-34. Yearly trends at the Little River (NC0075) site in (A) number of
cold and warm-water taxa; (B) percentage cold and warm-water individuals;
and (C) observed mean July/August maximum air temperature (°C) and
mean summer (June-September) flow (cfs).
5-68
-------
5.7. SENSITIVITY OF BENTHIC MACROINVERTEBRATES TO TEMPERATURE
AND STREAM FLOW
The spatial distributions of cold and warm-water taxa were examined to gain insights into
which areas in North Carolina are likely to be most and least sensitive to projected changes in
temperature and stream flow. Table 5-23 shows differences in the distributions of thermal
preference taxa between ecoregions. If the assumption is made that streams with greater
numbers and abundances of cold-water taxa will be most sensitive to warming temperatures and
changing precipitation patterns, then streams in the Blue Ridge ecoregion will be most sensitive,
and those in the Coastal region will be least sensitive. The prevalence and distribution of cold-
and warm-water- taxa also vary predictably with stream size. The median number of cold-water
taxa is highest in small and medium-sized streams (see Figure 5-35 A) while the greatest numbers
of warm-water taxa occur in the largest streams (see Figure 5-3 5B). Of the 5 least-disturbed
sites that we closely examined for long-term trends, the New River (NC0109) site is largest
(>2,000 km2), Barnes Creek (NC0248) is the smallest (<65 km2), and the three remaining sites
are medium-sized (125-225 km2). Although the greatest number of cold-water taxa may occur
in the coldest, highest elevation streams, it may be that the greatest amount of change will occur
in transitional areas, where species are expected to be closer to their tolerance limits. If this is
the case, then the greatest changes may occur in the Piedmont ecoregion.
5.8. IMPLICATIONS FOR NORTH CAROLINA DEPARTMENT OF THE
ENVIRONMENT AND NATURAL RESOURCES (NCDENRS) BIOMONITORING
PROGRAM
Over the last century, there has been a lot of year-to-year variability in temperature and
precipitation patterns in North Carolina, both statewide and at the five least disturbed biological
monitoring sites that we closely examined for temporal trends. During some years, there have
been extreme weather events, such as hurricane flooding that occurred in 2004 and drought
conditions from 1999-2002. In the future, extreme weather events are projected to occur with
greater frequency, and air temperatures are projected to increase. There is much uncertainty
associated with future projections for precipitation.
We were limited in the types of analyses that we were able to perform at four of the
five sites. This was due primarily to small sample sizes and to gaps and associated lack of
5-69
-------
Table 5-23. Summary of differences in elevation, PRISM mean annual air temperature, and mean number and
percentage of cold and warm-water taxa (based on full-scale samples only) in the North Carolina EPA Level 3
ecoregions. Relative abundances were calculated based on categorical abundance data
State
North
Carolina
Ecoregion
Middle Atlantic Coastal
Plain
Southeastern Plains
Piedmont
Blue Ridge
No
samples
173
317
1,106
631
Elevation
(m)
4.7
34.1
183.5
714.5
Air temperature
(°C)
16.7
16.3
15.0
12.1
Richness
Cold water
0.1 ± 0.2
0.1 ±0.4
1.5 ±2.0
8.0 ±4.5
Warm water
4.7±5.1
8. 8 ±3.4
5.2±3.1
2.8 ±2.4
Relative abundance
Cold water
0.1 ±0.4
0.1 ±0.4
1.8±2.7
11.4±7.9
Warm water
12.3 ±6.4
12.1±5.1
6.7 ±4.7
3.1±3.7
-------
•1 p
CD 16
TO 14
^ 12
i 10
•o p
8 o
o
•- 4
0)
£ 7
E '
5 n
z u
.-5
C
]
"I
C
r
i
"
c
"
j
•
B
S 10
03
t— o
0)
i 6
•^
o
i_ 2
0)
o
C
1 0
~z_
-2
T
0
D
n~
1
1
1 0 1
T
1
0
<90.6
90.6 to 259
>259
<90.6
90.6 to 259
>259
n Median
D 25%-75%
I Non-Outlier Range
o Outliers
+ Extremes
Figure 5-35. Distribution of cold and warm-water taxa across different stream size categories at North Carolina
reference sites (as designated by NCDENR). (A) number of cold-water taxa; (B) number of warm-water taxa.
Stream sizes categories are based on watershed areas (km2); thresholds are based on distributions of watershed areas
within the reference data set (tertiles).
-------
continuity in the biological data. At the New River site (NC0109), where we had sufficient
long-term data to run correlation and ANOVA, the biological metrics were more strongly
associated with precipitation than temperature variables. Several of the EPT-related metrics had
strong positive associations with flow and precipitation, and the HBI was negatively associated
with the flow and precipitation variables. When we grouped samples based on mean annual
flow, on average, there were about three more warm-water taxa in the lowest flow year samples
compared to the normal year samples, and there were several more EPT and cold-water taxa in
the highest versus lowest flow year samples. Although one cannot make causal inferences based
on observational data from this site, it seems evident that flow has an important influence on the
biological assemblage, and in this case, more of an influence than temperature.
At the other four sites, we did not see clear associations between the biological and
environmental variables in the temporal trend plots, but this was not unexpected due to the small
sample sizes, gaps in the biological data, and the large amount of year-to-year variability in the
climatic variables. We paid particular attention to a period during which hotter and drier than
normal conditions occurred at Cataloochee Creek (NC0209) for several consecutive years, but a
biological response to these conditions was not evident.
Because of the limitations associated with our individual site analyses, we also performed
exploratory analyses to gain insights into how future projected climatic changes might impact
NCDENR's assessment methods. We tried two techniques. In the first, we manipulated the
existing data at the three Blue Ridge sites (New River [NC0109], Nantahala River [NC0207],
and Cataloochee Creek [NC0209]) such that 50 and 100% of the cold-water EPT taxa were
removed from the assemblage. Then we recalculated the bioclassification scores based on these
two scenarios, with the intent of simulating the loss of cold-water taxa due to warming
temperatures associated with climate change. Results show that the loss of cold-water taxa has
the greatest effect on samples in the excellent and good site condition categories, with samples
generally dropping one bioclassification level (e.g., from excellent to good) (see Figure 5-36).
Sites in the "excellent" category are more likely to drop a level because this category has the
most stringent scoring criteria (i.e., it takes less of a change for a sample to drop from Excellent
to Good vs. from Good to Good-Fair).
We acknowledge that a scenario in which there is complete community replacement is
highly unlikely, especially in the near term. Nevertheless, we felt this scenario was worth
5-72
-------
North Carolina Blue Ridge Mountain ecoregion stations
• current benthic community composition
Blose 50% of co Id-p reference EPTtaxa
nlose 100% ofcold-preference EPTtaxa
Excellent
Good
Good-Fair
Classification
Fair
Poor
Figure 5-36. Exploratory exercise on reference station drift (degradation of
assessed site condition) over time at the three Blue Ridge stations, simulating
the loss of cold-preference EPT taxa over time due to climate change effects.
exploring, especially because the two metrics (EPT richness and HBI) that go into the calculation
of bioclassification scores are linked to thermal tolerance. As discussed earlier, there is a strong
association between thermal preference taxa and enrichment tolerance values, such that many of
the cold-water taxa are intolerant to enrichment, and many of the warm-water taxa are tolerant or
moderately tolerant. An increase in warm-water taxa and decrease in cold-water taxa could
result in an increase in HBI scores, which could cause a sample to drop to a lower
bioclassification level. A similar effect may be evident in future EPT richness values, because
many of the cold-water taxa are EPT taxa.
In our second exploratory analysis, we examined how bioclassification scores would
change if Mountain biocriteria were applied to biological data from the two Piedmont reference
sites (Barnes Creek [NC0248] and Little River [NC0075]). The premise of this analysis is that
biological assemblages in the Mountain region, which on average have the highest numbers of
5-73
-------
cold-water taxa, may become increasingly like Piedmont assemblages in the future. Results
show that if Mountain assemblages do indeed become like Piedmont assemblages, such as those
found at Barnes Creek (NC0248) and Little River (NC0075), and Mountain scoring criteria
remain the same, then bioclassification scores will decrease one level (from Excellent [5] to
Good [4]) (see Figure 5-37).
Mountain/Mountain
Piedmont/Piedmont
Piedmont/Mountain
20002009
Figure 5-37. 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.
Although there were limitations with the long-term trend analyses that we were able to
perform on the North Carolina data, the analyses that we were able to perform further our
understanding of the effects that changing temperature and stream flow conditions can have on
biological assemblages, and help establish expectations for biological responses to future climate
changes. Through these analyses, we were also able to provide insights as to which climate
change indicators might be best to track over time in southeastern states. Results suggest that
climate-induced trends are most likely to be detected in EPT-related metrics and thermal
preference metrics. Some limitations of the thermal preference metrics are that they typically
5-74
-------
occur in low numbers, and most show sensitivity to organic enrichment, which confounds the
associations with temperature.
5-75
-------
6. OHIO
6.1. EXPOSURES
6.1.1. Regional Projections for the Midwestern United States
Climate conditions in the Midwest are affected by its location in the middle of the
continent, removed from the moderating effects of the oceans, and by the Great Lakes (Karl et
al., 2009), and both existing conditions and future projections reflect this. Projected temperature
changes range from 1-11°C by the end of the century (Wuebbles and Hayhoe, 2004), though
several studies project changes within the lower to mid-portion of this range (Easterling and
Karl, 2001; Hayhoe et al., 2010) (see Table 6-1).
Table 6-1. Projections for temperature and precipitation changes in the
Midwest to 2100
Temperature
change
3-6°C by end of
century
2-3°C by
midcentury;
3-5°C by end of
century
1-9°C (winter);
1-1 1°C (summer)
Precipitation change
Increase
20-30% (winter, spring)
-10 to +40%
10% (midcentury) to 20%
(end of century, winter);
7% (midcentury) to 10%
(end of century, summer)
Change in precipitation
frequency
Increase
-5% (midcentury) to -9%
(end of century, winter);
-3% (midcentury) to -6%
(end of century, summer)
Citation
Easterling and
Karl, 2001
Hayhoe et al.,
2010
Wuebbles and
Hayhoe, 2004
Schoof et al.,
2010
Projections for precipitation changes are more variable and range from small decreases to
moderate increases (Easterling and Karl, 2001; Wuebbles and Hayhoe, 2004; Hayhoe et al.,
2010; Schoof et al., 2010) (see Table 6-1). Precipitation increases are expected largely from
increased occurrence of more intense storms (Easterling and Karl, 2001). This is supported by
the work of Schoof et al. (2010), showing that net increases in precipitation should occur with
decreases in the frequency of storms along with increases in the amount of precipitation.
6-1
-------
Estimates of the combined effects of changing temperatures and precipitation on
streamflow also are variable. Easterling and Karl (2001) project net decreases in stream runoff
for the midwest despite projected increases in average annual and winter precipitation amounts,
due largely to the combination of increased temperatures leading to increased evapotranspiration,
while summer precipitation is more variable and may decrease. Wuebbles and Hayhoe (2004)
project winter and spring runoff to increase but summer stream runoff to decrease. More recent
work projects variable streamflow over the near term, with net increases by the end of the
century (Hayhoe et al., 2010; Cherkauer and Sinha, 2010). End-of-the-century projections
include big increases in winter and spring flows, but variable summer flows with decreases in
low flows, increases in peak flows, decreases in the number of days with flows above the annual
mean flow, and increases in flashiness (Cherkauer and Sinha, 2010).
Several multidecadal climate changes have already been observed in the Midwest,
including increases in average annual temperature, with the largest temperature increases in
winter, along with earlier dates for last frost, reduced lake ice cover, extension of the growing
season by approximately 1 week, more severe and more frequent heat waves, above average
winter and summer precipitation, doubling of the frequency of heavy rain events, increased
frequency of large floods, and lower lake water levels resulting mainly from increased
evapotranspiration (Karl et al., 2009).
6.1.2. Historic Climate Trends and Climate Change Projections for Ohio
Ohio has a temperate climate characterized by hot, humid summers and cold winters. In
some parts of the state, the weather is influenced by the Great Lakes; these areas have increased
growing seasons, more winter cloudiness, and greater snowfall. Glaciation has played an
important role in shaping Ohio's landscape. The unglaciated areas in the southern and eastern
portions of the state are more rugged, hilly, and wooded than the glaciated areas to the north and
west. The glaciated areas consist of flat or rolling plains, low rounded hills, scattered end
moraines, kettles, wetland areas, and, in some places, relic sand dunes and beach ridges. Some
of the glaciated areas have been cleared, artificially drained, and converted to agricultural lands
(U.S. EPA, 2002). With relatively flat topography, there is not a great deal of variation in
temperature and precipitation patterns across Ohio (see Figure 6-1). The southern part of the
6-2
-------
state has the warmest mean annual temperatures (see Figure 6-1 A), and northwestern Ohio
receives the least amount of annual precipitation (see Figure 6-1B).
Over the last century, there has been a great deal of year-to-year variability in
temperature patterns in Ohio, with no clear or significant trend (see Figure 6-2). Annual
temperatures have, however, shown a more noticeable, albeit slight, increase in recent decades
(1971-2000), with the greatest increase occurring in the winter (see Table 6-2). However, even
this winter increase is not statistically significant, along with the other seasonal trends (see
Figure 6-3). Table 6-2 summarizes future temperature projections for mid- and late-century for
high (A2) and low (Bl) emissions scenarios. Based on an ensemble average across 15 models,
mean annual air temperatures are projected to increase by up to 3.6°C by midcentury and up to
5.8°C by the end of the century compared to a historic time period (1961-1990). Under the high
(A2) emissions scenario, the greatest increases are projected to occur during the fall (see
Table 6-3).
Mean Temp (C) B
I
20
Precip (mm)
12500 mm
'
1500
2000
Figure 6-1. Ohio's temperature and precipitation patterns. (A) Mean annual
air temperature (°C) from 1971-2000; (B) Mean annual precipitation (mm)
1971-2000. Map produced using the Climate Wizard Web site
(http://www.climatewizard.org/). Base climate data from the PRISM Group,
Oregon State University, http://www.prismclimate.org.
6-3
-------
o
£
O)
_ o
CD T—
1900 1920 1940 1960
Year
1980
2000
Figure 6-2. Trends in annual mean air temperature in Ohio from 1901-2000.
Change rate = 0°C/year, p-va\ue = 0.93. Figure produced using Climate Wizard
Web site (http://www.climatewizard.org/). Base climate data from the PRISM
Group, Oregon State University, http://www.prismclimate.org.
Table 6-2. Change rates in Ohio PRISM mean annual and seasonal air
temperatures compared across two time periods: 1971-2000 versus
1901-2000. No trends are significant (p > 0.05). Data were derived from the
Climate Wizard Web site (http://www.climatewizard.org/). Base climate
data came from the PRISM Group, Oregon State University,
http://www.prismclimate.org
Time period
1901-2000
1971-2000
Air temperature (C/yr)
Annual
0.00
0.02
DJF
0.00
0.06
MAM
0.00
0.01
JJA
0.00
0.02
SON
0.00
-0.01
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, and
August; SON = September, October, and November.
6-4
-------
1900 1320 1940 'SK 1980 2000
1920 1940 1960 1990 2000
1900 1920 1940 I960 1980 2000
Figure 6-3. Trends in seasonal mean air temperature in Ohio from 1901-2000. (A) DJF = December, January,
and February, change rate = 0.004°C/year,/?-value = 0.66; (B) MAM = March, April, and May, change
rate = O.OOFC/year,/>-value = 0.70; (C) JJA = June, July, and August, change rate = -O.OOFC/year,/?-value = 0.81;
(D) SON = September, October, and November, change rate = -0.002°C/year,/>-value = 0.58. Figure produced using
Climate Wizard Web site (http://www.climatewizard.org/). Base climate data from the PRISM Group, Oregon State
University, http://www.prismclimate.org.
-------
Table 6-3. Projected departure from historic (1961-1990) trends in annual and seasonal air temperature (°C) in
Ohio for mid- (2040-2069) and late-century (2070-2099) for high and low emissions scenarios. Values represent
the minimum, average, maximum and standard deviations from 15 different climate models. Data were derived
from the Climate Wizard Web site (http://www.climatewizard.org/)
Midcentury (2040-2069) vs. historic (1961-1990)
Model
Ensemble low
Ensemble average
Ensemble High
SD
A2 (high) emissions scenario
Annual
1.5
2.7
3.6
0.6
DJF
1.8
2.7
4.2
0.7
MAM
1.2
2.6
4.0
0.9
JJA
1.3
2.7
3.4
0.6
SON
1.7
3.0
4.7
0.8
Bl (low) emissions scenario
Annual
1.2
2.2
2.9
0.5
DJF
1.0
2.2
3.5
0.6
MAM
0.9
2.1
3.3
0.7
JJA
1.2
2.3
3.3
0.7
SON
0.9
2.2
3.2
0.7
Late-Century (2070-2099) vs. historic (1961-1990)
Ensemble low
Ensemble average
Ensemble high
SD
2.5
4.5
5.8
0.9
2.6
4.2
6.7
1.1
2.6
4.2
6.2
1.1
2.4
4.6
6.0
1.0
3.1
5.0
7.8
1.2
1.7
2.7
3.8
0.7
1.4
2.8
4.2
0.8
1.2
2.6
4.0
0.8
1.7
2.9
3.9
0.8
1.5
2.7
3.6
0.7
Oi
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July August, and SON = September, October, and
November.
-------
Precipitation patterns in Ohio have been highly variable. Trend direction varies
depending on the time period being evaluated. From 1901-2000, mean annual precipitation
increased at a rate of 0.22 mm/year (see Figure 6-4 and Table 6-4), while from 1971-2000, mean
annual precipitation decreased by 0.33 mm/year (see Table 6-4). Due to the high degree of
year-to year variability, none of the historic trends in precipitation are significant (p > 0.05). The
same holds true with seasonal change rates; the amount and direction of change vary depending
on season and time period, and no trends are significant (see Table 6-4 and Figure 6-5).
Table 6-5 summarizes future projections for mid- and late-century for high (A2) and low (Bl)
emissions scenarios. The future projections are highly variable across models and emissions
scenarios. Under the high emissions scenario, the ensemble average projects that mean annual
precipitation will increase by 56.7 mm by midcentury and 82 mm by the end of the century
compared to a historic time period (1961-1990) with the greatest changes during the spring (see
Table 6-5).
Table 6-4. Change rates in Ohio PRISM mean annual and seasonal
precipitation compared across two time periods: 1971-2000 versus 1901-
2000. No trends are significant (p > 0.05). Data were derived from the
Climate Wizard Web site (http://www.climatewizard.org/). Base climate
data came from the PRISM Group, Oregon State University,
http://www.prismclimate.org
Time period
1901-2000
1971-2000
Precipitation (mm/yr)
Annual
0.22
-0.33
DJF
-0.17
0.63
MAM
0.03
1.06
JJA
0.15
-0.40
SON
0.25
-1.30
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, and
August; SON = September, October, and November.
6-7
-------
o
o
o
o
CM
E o
:= °
o
—
H o
s §
TO
-value = 0.51. Figure produced using Climate
Wizard Web site (http://www.climatewizard.org/). Base climate data from the
PRISM Group, Oregon State University, http://www.prismclimate.org.
6-8
-------
i
B I-
1900 1920 1940 1960 1930 2000
1900 1920 1940 1960 1980 2000
I
1990 2000
1900 1S20 1940 I960 13SO 2000
Year
Figure 6-5. Trends in seasonal mean precipitation in Ohio from 1901-2000.
(A) DJF = December, January, and February, change rate = -0.171 mm/year,
p-value = 0.43; (B) MAM = March, April, and May, change
rate = 0.034 mm/year, p-va\ue = 0.86; (C) JJA = June, July, and August, change
rate = 0.149 mm/year, /7-value = 0.42; (D) SON = September, October, and
November, change rate = 0.248 mm/year, p-va\ue = 0.23. Figure produced using
Climate Wizard Web site (http://www.climatewizard.org/). Base climate data
from the PRISM Group, Oregon State University, http://www.prismclimate.org.
6-9
-------
Table 6-5. Projected departure from historic (1961-1990) trends in annual and seasonal precipitation (mm) in
Ohio for mid- (2040-2069) and late-century (2070-2099) for high and low emissions scenarios. Values represent
the minimum, average, maximum and standard deviations from 15 different climate models. Data were derived
from the Climate Wizard Web site (http://www.climatewizard.org/)
Midcentury (2040-2069) vs. historic (1961-1990)
Model
Ensemble low
Ensemble average
Ensemble high
SD
A2 (high) emissions scenario
Annual
-83.6
56.7
154.4
68.2
DJF
-15.6
23.0
50.6
20.8
MAM
-5.5
29.5
64.9
23.0
JJA
-49.0
-5.4
46.7
30.4
SON
-37.5
9.0
56.3
23.8
Bl (low) emissions scenario
Annual
-66.6
73.0
425.0
120.1
DJF
-22.8
28.0
141.8
39.9
MAM
-15.5
33.7
112.2
28.8
JJA
-38.3
7.6
98.1
35.6
SON
-40.3
11.1
116.2
38.4
Late-Century (2070-2099) vs. historic (1961-1990)
Ensemble low
Ensemble average
Ensemble high
SD
-146.3
82.0
297.0
122.6
-4.3
33.4
95.9
26.4
-24.7
51.3
152.3
45.6
-98.7
-5.5
54.5
50.5
-50.7
11.1
72.3
34.9
-52.3
99.7
443.8
140.3
-4.2
45.1
179.3
58.0
-13.2
43.0
101.1
32.3
-52.4
6.2
90.2
39.1
-25.4
15.0
114.8
42.0
DJF = December, January, and February; MAM = March, April, and May; JJA = June, July, August and SON = September, October, and
November.
-------
6.2. DATA INVENTORY AND PREPARATION
Ohio was one of the first states to systematically use biological assemblage data to
determine aquatic life use designations and assess the condition of those uses. Dating back to the
late 1970s, the Ohio data set represents a nearly 30-year span of standardized biological data for
fish and macroinvertebrates, and is the only state from which we looked at fish (see Section 2).
The Ohio fish assemblage database contains data from more than 10,000 unique sites and more
than 24,000 unique sampling events. Macroinvertebrate assemblage data were also collected at
most of these sites, as were habitat QHEI data (Ohio EPA, 2006; Rankin, 1995, 1989).
In the 1980s, with assistance from EPA-ORD, Ohio EPA began a focused sampling of
least-impacted reference sites in order to determine the efficacy of level III ecoregions (Omernik,
1987) as a way to account for and stratify natural variations in biological assemblages (Yoder,
1989; Ohio EPA, 1987; Whittier et al., 1987). Ohio EPA used this and other sampling data to
establish a network of "least-impacted" regional reference sites that eventually supported the
derivation of numerical biocriteria for Ohio streams and rivers. This was also accomplished
r\
across practically all wadeable and nonwadeable streams and rivers from >1 mi up to the largest
r\
inland rivers (-6,000-8,000 mi ) that could be sampled.
The initial reference data set was developed from a statewide network of about
300 reference sites that was sampled over a 10-year period (1980-1989; Table 6-6). That
reference site network was maintained and expanded with the initial resampling during
1990-1999 and a second resampling from 2000-2009 (at the time of this project, we only had
access to data through 2006). Data on habitat quality (QHEI), water quality, and other physical
data such as temperature were also collected and were based on multiple grab samples collected
during "normal" seasonal flows within a summer-fall seasonal index period (mid-June through
mid-October).
Data gathering, preparation, and analyses were conducted by MBI. Prior to running
analyses, MBI screened the data to identify any methodological differences in data collection
(environmental and biological) that could either confound or mask apparent trends. MBI also
assessed the relative contribution of taxonomic changes to trends in ICI and IB I scores at
reference sites. While fish data can be influenced by factors such as sampling efficiency, MBI
found the fish taxonomy to be comparatively stable during the period over which the Ohio
reference database was developed. Using the methods described in Section 2.1.3, MBI did,
6-11
-------
Table 6-6. Summary of Ohio EPA regional reference site network including
original sites (1980-1989) and updates via first (1990-1999) and second
round resampling (2000-2006) that were used in data analyses
Reference network
Original reference sites:
1980-1989 (sites/samples)
New reference sites:
1990-2006 (sites/samples)
Size type
Headwaters
Wadeable
Beatable
Headwaters
Wadeable
Beatable
Fish: latest (all data)
112/225
166/399
97/254
1 15(149)7150(296)
184(231)7281(539)
68(84)/127(278)
Macroinvertebrates
242
309 (525)
however, find significant changes in macroinvertebrate taxonomy over time, mostly in the form
of improved discrimination within certain genera (e.g., Baetid mayflies).
To determine how much of an impact these taxonomic changes had on ICI scores, MBI
calculated the ICI, total taxa metric, mayfly metric, and qualitative EPT metric with the original
taxon designations and then compared these values to metric and ICI scores that were calculated
using the newly "refined" taxonomy. MBI performed these calculations on data from the earliest
and most recent time periods. The recalculation of ICIs from all sites showed a 5.9 point
increase in the mean ICI score between the two time periods (see Table 6-7). MBI also
evaluated how taxonomic refinements affected ICI scores in samples from different ecoregions
in WWH and EWH. In two instances, there was a change in the biocriteria: the Huron/Erie Lake
Plain WWH biocriterion (38.5 compared to 42) and the Erie Ontario Lake Plain WWH
biocriterion (42 compared to 44) (see Table 6-8).
MBI also performed exploratory analyses to look for obvious trends related to
stream-size bias. Going into the analyses, they knew that some stream-size bias did exist in the
Ohio data set because headwater streams were less frequently sampled in the 1980s than in the
1990s and 2000s. Recognizing that the distribution of sites was different between these periods,
they tested whether bias was evident in low percentiles (1st, 5*, and 25*) for species distributions
across all sites in Ohio. Results showed that some bias between time periods existed for species
distributions. Nearly all selected sensitive species had distributions that extended further into
small streams during the later (1998-2008) compared to the earliest (1978-1989) sampling
periods. MBI performed some additional exploratory analyses. Because these analyses may
6-12
-------
Table 6-7. Changes in ICI and mayfly influence ICI metrics related to
increasing taxonomic resolution over time in the Ohio EPA least impacted
reference data set
Metric
Total taxa
Number of Mayfly taxa
QUALEPTtaxa
ICI score
Original reference sites
Standard
taxonomy
mean taxa
(mean score)
35.97(4.89)
6.95 (4.20)
11.29(3.63)
39.59
Lumped
taxonomy
mean taxa
(mean score)
35.93 (4.89)
6.90(4.17)
11.24(3.60)
39.53
New reference sites
(latest data)
Standard
taxonomy
mean taxa
(mean score)
38.36(5.18)
7.42 (4.59)
15.16(5.16)
45.35
Lumped
taxonomy
mean taxa
(mean score)
37.65 (5.04)
6.59(4.16)
14.23 (4.91)
44.56
Table 6-8. Table of original and recalibrated Ohio biocriteria with
adjustments made to equilibrate taxonomic advances made in the later time
period. Highlighted cells indicate where standardizing taxonomic resolution
would have resulted in altered criteria
Ecoregion
Huron/Erie
Lake Plain
(HELP)
Interior
Plateau (IP)
Erie Ontario
Lake Plain
(EOLP)
Western
Allegheny
Plateau (WAP)
Eastern Corn
Belt Plain
(ECBP)
Warmwater habitat
Original
reference
34
30
34
36
36
Latest
reference
42
38
44
40
42
Latest reference
w/refined
taxonomy
38.5
38
42
40
42
Exceptional warmwater habitat
Original
reference
46
Latest
reference
50
Latest reference
w/refined
taxonomy
50
6-13
-------
have been confounded by year-to-year variability in flow or temperature within each time period
being evaluated, they encourage that a more sensitive approach be used in future analyses (i.e.,
one that controls for or considers annual variation and regional variation in flows, which can be
extracted from USGS flow data using IHA flow indicators).
6.3. OHIO EPA METHODS
Ohio implemented standardized sampling methods for biological assessments in the late
1970s. Collection methods used by Ohio EPA for both fish and macroinvertebrates have been
stable over the period of the Ohio reference database. When collecting macroinvertebrate
samples, Ohio EPA uses a modified Hester-Dendy multiplate artificial substrate sampler that is
placed in-stream to colonize for 6 weeks between mid-June and late September (DeShon, 1995).
Fish sampling is conducted during the same index period and includes two or three passes. Fish
sampling gear varies depending on stream size (Ohio EPA, 1989).
Ohio calculates an ICI to evaluate biological condition based on the benthic
macroinvertebrate assemblage (DeShon, 1995) and an IBI used to evaluate fish assemblages at
wading sites, boat sites, and headwaters stream sites. Tables 6-9 and 6-10 (Ohio EPA, 1989)
show the metrics that go into the ICI and IBI.
6.4. INDICATORS
6.4.1. Thermal Preference
To evaluate which taxa could potentially serve as indicators of temperature change, MBI
used weighted-average modeling to calculate thermal optima and tolerance values (which they
termed WSVs) for fish and macroinvertebrate taxa. For more details on this methodology, see
Section 2.2.1. Separate calculations were done for headwater (drainage area <20 mi2) and
wadeable streams (drainage area >20 to 300 mi2).
MBI ordered the data by temperature optima to provide a sequential listing of sensitive
species/taxa that could potentially be used to detect temperature trends. These data are available
upon request. To visualize the distribution of the macroinvertebrate data with taxa sensitivities,
MBI plotted the means of these values versus the weighted means (WSVs) color coded by the
existing taxa tolerance rankings of Ohio EPA (see Figure 6-6). The WSVs generally track with
the "general" tolerance categories assigned by Ohio EPA for each taxon for both headwater (see
6-14
-------
Table 6-9. Macroinvertebrate community metrics used in the ICI for
evaluating biological condition in Ohio. Scoring of each metric ranges from
0 to 6 in increments of 2, and is based on drainage area (as defined in Figures
5-1 to 5-10 in Ohio EPA, 1989)
Metric
Total number of taxa
Total number of Mayfly taxa
Total number of Caddisfly taxa
Total number of Dipteran taxa
Percentage Mayfly composition
Percentage Caddisfly composition
Percentage Tribe Tanytarsini Midge composition
Percentage Other Dipteran and noninsect composition
Percentage tolerant organisms (from Table 5-2)
Total number of qualitative EPT taxa
6-15
-------
Table 6-10. Index of Biotic Integrity metrics used to evaluate wading sites,
boat sites, and headwaters stream sites in Ohio. Original metrics from Karr
(1981) are given first with substitute metrics following. Taken from
Table 4-1 in Ohio EPA's "Standardized Biological Field Sampling and
Laboratory Methods for Assessing Fish and Macroinvertebrate
Communities" (1989)
IBI metric
1 . Total number of speciesd
2. Number of Darter species
Percentage round-bodied suckersf
3 . Number of Sunfish species
Number of headwater species
4. Number of Sucker species
Number of Minnow species
5 . Number of intolerant species
Number of sensitive species
6. Percentage green Sunfish
Percentage tolerant species
7. Percentage omnivores
8. Percentage insectivorous Cyprinids
Percentage insectivorous species
9. Percentage top carnivores
Percentage pioneering species
10. Number of individuals8
1 1 . Percentage hybrids
Percentage simple Lithophils
number of simple Lithophilic species
12. Percentage diseased individuals
Percentage BELT anomalies'1
Headwaters sitesa'b
X
Xe
X
X
X
X
X
X
X
X
X
Wading sitesb
X
X
X
X
X
X
X
X
X
X
X
Boat sitesc
X
X
X
X
X
X
X
X
X
X
X
aApplies to sites with drainage areas less than 20 sq. mi.
bThese sites are sampled with wading methods.
These sites are sampled with boat methods.
dExcludes exotic species.
Includes sculpins.
Includes suckers in the genera Hypentelium, Moxostoma, Minytrema, and Erimyzon', excludes white sucker
(Catostomus commersoni).
8Excludes species designated as tolerant, hybrids, and exotics.
Includes deformities, eroded fins, lesions, and external tumors (BELT).
6-16
-------
kVSVi far Miuamun TMnfxrotarm
I
^1
15 W S3 90
Mean Maximum Temperature (C)
W$Vt lor Mjuamun fnmpo/o!in
19 » 23 JO
Mean Maximum Temperature (C]
ICI Toxa Tolerance Categories
Tdlnn
Tokr^tt
ICI Taxa T«t«ro»«* Categories
Figure 6-6. Plots of macroinvertebrate taxa maximum temperature WSV values versus mean maximum values
for taxa for headwater streams (a) and wadeable streams (b) and box and whisker plots of WSVs for maximum
temperatures by Ohio EPA macroinvertebrate tolerance values (derived for the ICI) for headwater streams (c)
and wadeable streams (d). Data for taxa represents data collected from artificial substrates where at least five
samples were represented for each stream size category.
-------
Figure 6-6C) and wadeable streams (see Figure 6-6D). A similar pattern was observed for fish
species. WSVs for temperature can be confounded with WSVs for other stressors, particularly
habitat. However, the extremes of these distributions can be useful for identifying possible
indicator taxa for future applications. For example, it was interesting to note that selected
Chironomidae taxa occurred at both extremes of the WSV for temperature. Paratany'tarsus n.sp.
1 had the lowest WSV for temperature at wadeable sites, and Parachironomus "hirtalatus" and
Tanypus neopunctipennis had among the highest WSVs (see Figure 6-6B). Additional
traits-based analyses could help in identifying rare taxa that exhibit some sensitive traits, but that
may be too rare by themselves to serve as useful indicators.
6.4.2. Hydrologic Indicators
To evaluate which taxa could potentially serve as indicators of hydrologic change, MBI
used weighted-average modeling to calculate WSVs for flow-related habitat variables for fish
and macroinvertebrate taxa. Calculations were made using the methodology described in
r\
Section 2.2.2. Separate analyses were done for headwater (drainage area <20 mi ) and wadeable
r\
streams (drainage area >20 to 300 mi ).
MBI made these calculations based on a subindex of the QHEI, which they termed the
Hydro-QHEI. The Hydro-QHEI is composed of the two QHEI subcomponents most related to
hydrology—current and depth. Table 6-11 details scoring calculations for the Hydro-QHEI.
The presence of fast current or the presence of eddies is a characteristic of permanent summer
base flows (QHEI assessments are generally conducted during summer-fall low flow periods).
Attributes related to depth (i.e., deep pool and deep runs) are also regarded as good indicators of
base flow influence. Thus, the Hydro-QHEI is expected to reflect a gradient of baseflow
stability, one of the attributes that would be expected to change with alterations in precipitation
patterns as a result of climate change.
MBI ordered the data by optima values to provide a sequential listing of sensitive
species/taxa that could potentially be used to detect changes in hydrology. These data are
available upon request. MBI plotted several examples of the WSVs for these variables versus the
simple means for these same variables (see Figure 6-7) in order to reveal the distributions of
tolerant and sensitive species along this gradient, as they did for temperature. Fish and
6-18
-------
Table 6-11. Subcomponents of the Ohio QHEI, which were used to score a
Hydro-QHEI, and current and depth subscores
Current metric
QHEI current attribute
Very fast current
Fast current
Moderate current
Slow current
Eddies
Very deep riffles
Moderate depth riffles
Interstitial flow
Intermittent flow
Score
+5
+3
+2
+1
+2
+3
+1
-1
-3
Depth metric
QHEI depth attribute
Deep pools (cover metric)
Pool depths > 1m
Pool depths 0.7-1. Om
Pool depths 0.4-0.7 m
Pool depths 0.2-0.4 m
Pool depths < 0.20
Deep riffles
Moderate riffles
Shallow riffles
Riffles absent or nonfunctional
Score
+4
+4
+3
+2
+1
-1
+3
+2
+1
-1
6-19
-------
a)
I
T5
V
b)
LU
I
l_i
i
I
Headwater Maervim-ftebratc Oats
Hydro-QHEI
InloJtnwt Metfemtily InKr-
Intolerant mtdtnK
TdtnMt
WsxteMe Siraams f >?0-a» SQ nwj
d>»
M W J5
I-vtoltrarf SoiBlnE DrternKftrtB Moderately Tole«»rt
Mean HVDRO-QHEI
Figure 6-7. Scatter plots of taxa/species Hydro-QHEI WSV values versus
mean Hydro-QHEI values for macroinvertebrates taxa for headwater
streams (a) and for species in wadeable streams (b) and box and whisker
plots of macroinvertebrate (c) and fish (d) WSVs for Hydro-QHEI for these
waters. Data from Ohio EPA.
macroinvertebrate WSVs for Hydro-QHEI and its subcomponents tracked relatively closely to
the Ohio EPA tolerance designations for macroinvertebrate taxa and fish species (see
Figure 6-7). Outlier points and variability are often associated with small sample sizes for a
given species at a given stream size. Intolerant species are frequently rarer than "sensitive"
species, especially for fish, and as such, may exhibit more variation than "sensitive" species
6-20
-------
where sample sizes are typically larger. As expected, tolerant species generally have wider
sensitivity ranges.
The identification of certain intolerant fish species in headwater streams at the "sensitive"
end of the Hydro-QHEI gradient suggests that the distribution of these species at the tails of their
preferred stream size range may reflect the degree of base flow. Fish species such as streamline
chub, variegate darter, river chub, and stonecat madtom (all with high Hydro-QHEI WSVs) are
generally found in larger wadeable streams, and their presence in headwater streams is associated
with high Hydro-QHEI scores that indicate more stable flow regimes. Year-to-year or long-term
trends of these species in headwater streams could represent a response to climate-induced
hydrologic changes. Thus, we suggest that this could be an opportunity to explore whether the
stream size "tails" of sensitivity distributions shift with hydrological change.
6.4.3. Traits-Based Indicators in a Warmer Drier Scenario
In the Maine, North Carolina, and Utah data sets, we performed exploratory exercises to
develop lists of taxa that may be most and least sensitive to projected changes in temperature and
streamflow based on combinations of traits. MBI did not perform these types of analyses on the
Ohio data set.
6.5. LEAST DISTURBED LONG-TERM BIOLOGICAL MONITORING SITES
As discussed in Section 6.2, Ohio EPA began a focused sampling of least-impacted
reference sites in the 1980s. This network of "least-impacted" regional reference sites eventually
supported the derivation of numerical biocriteria for Ohio streams and rivers. Most stations are
sampled on a regionally rotating basis, at 10-year intervals. If placed on a BCG scale (Davies
and Jackson, 2006), the Ohio reference sites generally span BCG Level 3-Level 4, with an
occasional 2 and 5. Few undisturbed sites remain in Ohio, with widespread agricultural and
development changes occurring across the landscape. Although anthropogenic influences are
higher than desired at some of the sites, the data were analyzed because they represent the
best-available long-term data in the state database. When MBI performed analyses on this
reference data set, they evaluated trends in reference condition across the entire network
(stratified by size and sometimes habitat categories) and did not perform analyses on data from
individual sites.
6-21
-------
6.6. EVIDENCE OF TRENDS AT LEAST-DISTURBED LONG-TERM MONITORING
SITES
6.6.1. Reference Sites
6.6.1.1. Temporal Trends in Climatic and Biological Variables
MBI looked at the amount and direction of change in state bioassessment scores over the
last 30 years at Ohio's network of "least-impacted" regional reference sites. Table 6-12
summarizes the ranges of years that represent the original and resampled reference sites. On
average, the latest data period was 13-16 years after the mean of the original reference sample
dates (see Table 6-12). Before making the calculations, MBI stratified the fish assemblage
r\
indices by three stream and river size strata: headwater streams (<20 mi ), "wadeable" streams
(20—300 mi2), and "boatable" (i.e., nonwadeable) rivers (>~150-200 mi2) (Yoder and Rankin,
1995). Macroinvertebrate assemblage indices were calibrated continuously across the entire
range of stream and river sizes. Samples were also divided into three habitat groups: MWH,
WWH, and EWH.
Table 6-12. Average and range of years represented by original reference
site data and resampled (latest) data by index and stream size category
Index/stream size
ICI — all sites
IBI — headwaters
IB I — wading
IBI— boat
Mean year sampled (range)
Original reference sites
1984
(1980-1988)
1984
(1978-1988)
1984
(1979-1988)
1984
(1979-1988)
Resampled sites
2000
(1989-2007)
2000
(1989-2006)
2000
(1990-2006)
1997
(1990-2005)
Table 6-13 reports the original biocriteria values and statistics, a recalculation of those
statistics using refined variables, and "new" biocriteria values based on the latest resampled
reference sites. The reason the original biocriteria statistics were recalculated was because there
are a few minor discrepancies related to uncertainties about the exact membership of the original
reference sites and gradual changes made to the database since 1990 due to changing taxonomy
6-22
-------
and a more precise calculation of drainage area (Rankin, 2009). The direction of change in
bioassessment scores between the original and latest reference site data was either positive (an
increase) or neutral (no change), with three exceptions: (1) the ICI biocriterion for the nonacidic
mine drainage modified use was four points lower (possible small sample size); (2) the IBI for
WWH headwater site type in the Erie Ontario Lake Plain ecoregion was two points lower; and,
(3) the IBI for WWH headwater site type in the Western Allegheny Plateau ecoregion was
two points lower (see Table 6-13). None of these changes are considered to be outside the range
of natural variability for each index. The largest positive changes in the biocriteria were in the
WWH boatable fish sites (IBI and the Modified Index of Weil-Being (Mlwb), a measure of the
health of the fish assemblage that is used in conjunction with the IBI), and in the WWH ICI.
6.6.1.2. Associations Between Biological and Climatic Variables
In the Maine, North Carolina, and Utah data sets, we performed correlation analyses on
individual sites to look for associations between state bioassessment scores, selected biological
metrics, year, temperature, flow, and precipitation variables. MBI did not perform these types of
analyses on the Ohio data set.
6.6.1.3. Groupings Based on Climatic Variables
In the Maine, North Carolina, and Utah data sets, we grouped biological data based on
extremes in temperature, flow, and/or precipitation variables, using these groupings as proxies
for future climate conditions. MBI did not perform these types of analyses on the Ohio data set.
6.7. SENSITIVITY OF BENTHIC MACROINVERTEBRATES TO TEMPERATURE
AND STREAM FLOW
Sensitivities in Ohio may best be monitored by tracking changes in the distributions of
the candidate thermal and hydrologic indicator taxa that MBI identified through
weighted-averaging analyses, and by carefully monitoring the habitats that those taxa occur in.
Changes in baseflow stability could be particularly important in Ohio. The identification of
certain intolerant fish species in headwater streams at the "sensitive" end of the Hydro-QHEI
gradient suggests that the distribution of these species at the tails of their preferred stream size
range may reflect the degree of baseflow. Year-to-year or long-term trends of these species in
headwater streams could represent a response to climate-induced hydrologic changes.
6-23
-------
Table 6-13. Original Ohio biocriteria (O), recalculated biocriteria (R) using
similar sites, and new biocriteria (N) using the latest data from resampling of
original reference sites. Because IBI or ICI scores based on single samples
are always even values, calculated percentage values were rounded upwards
(e.g., 41 to a 42). Sites with discrepancies between original and recalculated
criteria are shaded. Mlwb refers to the Modified Index of Well-Being, a
measure of the health of the fish assemblage that is used in conjunction with
the IBI
Eco region
MWH
Channelized
Nonacidic mine
drainage
Impounded
WWH
EWH
IBI—headwater site type
IBI—wadeable site type
N
HELP
IP
EOLP
WAP
ECBP
22
24
24
24
24
22
24
24
24
24
22
30
30
30
30
32
40
38
44
40
40
38
44
40
44
42
46
40
50
50
52
IBI—boatable site type
HELP
IP
EOLP
WAP
ECBP
20
24
24
24
20
24
24
24
24 24 24
20
24
24
24
24
24
26
22 22 26
30
30
30
28
28
28
30 28 34
34
34
34
34
38
40
40
42
30
34
38
40
40
42
30
34
47
46
40
42
48
48
52
Mlwb—wadeable site type
HELP
6-24
-------
Table 6-13. Original Ohio biocriteria (O), recalculated biocriteria (R)
using similar sites, and new biocriteria (N) using the latest data from
resampling of original reference sites. Because IBI or ICI scores based on
single samples are always even values, calculated percentage values were
rounded upwards (e.g., 41 to a 42). Sites with discrepancies between
original and recalculated criteria are highlighted in yellow. Mlwb refers to
the Modified Index of Well-Being, a measure of the health of the fish
assemblage that is used in conjunction with the IBI (cont.)
Eco region
MWH
Channelized
Nonacidic mine
drainage
Impounded
WWH
EWH
Mlwb—boatable site type
HELP
IP
EOLP
WAP
ECBP
5.7
5.8
5.8
5.8
5.8
5.7
5.7
5.7
5.7
5.7
7.5a
6.1a
6.1a
6.1a
6.1a
5.4
5.7
6.6
6.6
6.6
6.6
5.7
7.0
7.0
7.0
7.0
7.4
7.5
7.5
7.5
7.5
8.6
8.7
8.7
8.6
8.5
8.7
8.8
8.6
8.5
9.6
8.9
9.2
9.7
9.6
9.6
10.2
ICI—all site types combined
HELP
IP
EOLP
WAP
ECBP
22
22
22
22
22
22
22
22
22
22
24
24
24
24
24
34
30
34
36
36
34
30
34
36
36
42
38
44
40
42
46
46
50
aNonacidic mining influenced modified sites for headwaters combined with wading sites due to small sample size.
HELP = Huron/Erie Lake Plain, IP = Interior Plateau, EOLP = Erie Ontario Lake Plain, WAP = Western Allegheny
Plateau, ECBP = Eastern Corn Belt Plain.
Assemblages in small headwater streams (currently undersampled), streams already near the
"edge" of temperature and hydrologic thresholds, and cold water systems, exceptional systems
and areas that are "islands" of the above categories may be particularly sensitive, as they would
have difficulty recovering from episodic stressors due to lack of refugia or recolonization areas.
6.8. IMPLICATIONS FOR OHIO EPA'S BIOMONITORING PROGRAM
In Ohio and other midwestern states, climate change projections are for warmer
temperatures and slight increases in precipitation. The expectation for changes in flow are less
certain, being affected by both increasing precipitation, which may increase flows, and
6-25
-------
increasing temperatures, which can also increase evapotranspiration and contribute to decreasing
flows at least seasonally.
When MBI analyzed data from the Ohio reference data set to search for a signal or lack
of signal related to the effects of global climate change, they found that, in general, the biological
condition at Ohio's reference sites has improved over the last 30 years. Climate change effects
may be a contributing component to these observed trends, or may be decreasing the magnitude
of the positive response. However, there is evidence that the trends have been driven largely by
other environmental factors. Main contributors include reductions in point source loadings,
changes in land uses (e.g., increased urbanization), altered pollutant loadings from agricultural
lands (e.g., reductions in sediments and nutrients in response to increased conservation tillage),
and localized improvement in habitat quality due to stream restoration. There may also be
changes that are still contributing to degradation, such as loss of habitat quality due to
agricultural drainage practices and suburbanization.
The improvements in fish assemblages in large rivers are most attributable to reduced
pollution from point sources, mostly due to municipal wastewater treatment plant upgrades after
1988 (Yoder et al., 2005). While it was necessary in the derivation of the original Ohio IBI for
boatable sites to include reference sites located in effluent dominated rivers, the sites were
positioned below known recovery points. Nevertheless, the lessening of secondary impacts from
nutrient enrichment by the aforementioned controls had positive effects on fish assemblages at
these reference sites. Large river pollution reductions have also facilitated the movement of
intolerant species between watersheds (i.e., have become highways for recolonization). In
headwater and small streams, biological conditions have generally improved over time due to
better agricultural practices (e.g., conservation tillage).
In the future, it will continue to be challenging to tease out climate-related impacts from
other confounding factors in the Ohio data set, especially because a number of the
"least-impacted" regional reference sites are affected to some degree by pollution or land
alterations. If biological responses to climate change effects do become more evident, the
direction of these changes could be in a positive or negative direction. The most plausible
expectation would be for a decline in bioassessment scores due to the loss of highly intolerant
species and taxa (i.e., temperature and flow sensitive taxa/species), and an increase in
intermediate, moderately, and/or highly tolerant taxa/species. Such expectations are supported
6-26
-------
by MBI's analyses that identify a general concordance between intolerant and sensitive species
as categorized for the IB I and ICI and species sensitive to temperature and habitat features
indicative of altered flow conditions. These changes could be tracked by monitoring the
distributions of taxa that MBI identified as being sensitive to changing temperature and
hydrology.
6-27
-------
-------
7. SYNTHESIS
7.1. EVIDENCE FOR EXISTING CLIMATE CHANGE RESPONSES
7.1.1. Existing Climate Trends Support Expectations for Biological Responses
The direction and magnitude of historic trends in air temperature, precipitation, water
temperature, and flow records define whether climate-related biological responses might be
expected during the period of record in the different regions. Long-term air temperature
increases are evident from PRISM data for most states, though there is variability in the
magnitude of increases among the regions examined (see also Karl et al., 2009. North Carolina
and Ohio showed no trend in temperature over the past century, but there were modest though
nonsignificant (1-2 °C) increasing trends in temperature for these states over that past 3 decades
(see Table 7-1). Maine had similar long- and shorter-term historic net temperature increases of
about 1°C (only the long-term rate was statistically significant). Greater temperature increases
are documented in Utah, where a gradual (1 °C) but significant increase was found over that past
century, and a steeper (4 °C) and significant rate of increase occurred since 1970; this is the
largest historic temperature rise among the 4 states studied (see Table 7-1). The projected rates
of future temperature increases are generally consistent with the current documented rates of
increase in magnitudes and regional patterns. They are highest for Utah and lowest for North
Caroline, but the differences are small.
It is also reasonable to expect long-term water temperature trends to follow air
temperature trends. Previous studies (e.g., Pilgrim et al., 1998; Wehrly et al., 2009; Stephan and
Preudhomme, 1993) have established a relationship of water temperature to air temperature of
from 0.86 to 1. From this, it can be expected that an increasing trend in air temperature of 2°C
will, on the average, result in an increase in water temperature of 1.7-2 °C. Support for this also
comes from analysis of USGS gaging station records from around the United States. Stations
analyzed were screened to include gages with long-term water temperature records (30 years),
and to minimize the likelihood of confounding effects (e.g., sewage treatment plant discharges,
heavy urban/suburban development, effects of dam releases) or temporal discontinuities from
methods or data quality issues. The rate of water temperature increases averaged 0.76°C per
10-year period (see Table 7-2), but varied around the country, partly in relation to stream size.
This suggests that estimates for water temperatures increases of 1-2°C over the approximately
7-1
-------
2 decades of biological sampling are reasonable. The screening process eliminated gage data
from two (Ohio and Maine) of the four states evaluated in this study. However, the North
Carolina stream analyzed had a larger water temperature increase than the Utah stream, even
though climate change-related temperature projections are slightly greater for Utah (see
Table 7-2), suggesting that differences in stream size also are an important influence.
Table 7-1. Observed and modeled future rates of change for air temperature
and precipitation for the four states analyzed in this study. Estimated
changes based on significant increasing or decreasing trends are shown in
bold. See Tables 3-1 to 3-4, 4-1 to 4-4, 5-1 to 5-4, and 6-1 to 6-4 for sources of
data
Existing and projected rates per century
Utah
Maine
North Carolina
Ohio
Air temperature (°C)
Existing (190 1-2000)
Existing short term rate (1970-2000)
projected to century
Projected midcentury (low to high
emissions scenarios)
Projected end of century (low to high
emissions scenarios)
1
4
2.3-2.9
3.0-4.8
1
1
2.1-2.7
2.8-3.6
0
1
1.7-2.3
2.2-3.7
0
2
2.2-2.7
2.7-4.5
Precipitation — average annual (mm)
Existing (190 1-2000)
Existing short term rate (1970-2000)
projected to century
Projected midcentury (low to high
emissions scenarios)
Projected end of century (low to high
emissions scenarios)
+35
+128
+22.3 to
-2.7
+50.7 to
-5.8
+110
-113
+69.0 to
+90.1
+67.0 to
+125.0
+39
-147
-1.0 to +54.0
-16.5 to +56.9
+22
-33
+73.0 to +56.7
+99.7 to +82.0
Precipitation — summer (mm)
Existing (190 1-2000)
Existing short term rate (1970-2000)
projected to century
Projected midcentury (low to high
emissions scenarios)
Projected end of century (low to high
emissions scenarios)
+11
+69
+16. 8 to
-6.9
+32.8 to
-4.4
+18
-95
-4.7 to +15.1
-8.3 to +7.7
-68
+61
-22.5 to +11. 6
-4 1.5 to +22.3
+15
-40
+7. 6 to -5. 4
+6.2 to -5. 5
7-2
-------
Table 7-2. Summary of results from water temperature trend analyses at 23 USGS stations that met the
screening criteria. Rates of temperature (°C) change per 10-year period were evaluated at 23 of the stations.
Stations in states analyzed in this study are highlighted in grey
Site#
2423130
10339400
7086000
9169500
2266300
5474000
13340600
3354000
1600000
1021050
12363000
2077200
6338490
5056000
5058700
1466500
1428500
14138870
14372300
2160700
8123800
8181500
408000000
Stream name
Cahaba River
Martis Creek
Cache Creek
Dolores River
Reedy Creek
Skunk River
Beaver Creek
White River
North Branch Potomac River
Saint Croix River
Flathead River
Hyco Creek
Missouri River
Sheyenne River
Sheyenne River
McDonalds Branch
Delaware River
Fir Creek
Rogue River
Enoree River
Beals Creek
Medina River
Middle Branch Embarrass River
Stream order
3
3
2
5
3
6
4
5
5
6
6
3
1
4
1
1
6
2
6
5
5
5
3
NPDES
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
Land use
FOR/AG (URB)
FOR
FOR
URB
FOR
AG
URB/FOR
AG(URB)
FOR
GRASSLAND
GRASSLAND
GRASSLAND
FOR
FOR
FOR
FOR (urb)
Shrub
AG
AG
State
AL
CA
CO
CO
FL
IA
ID
IN
MD
ME
MT
NC
ND
ND
ND
NJ
NY
OR
OR
sc
TX
TX
WI
TempA/lOyear
0.73
0.28
1.48
0.93
0.3
0.25
0.4
0.32
0.5
0.39
1.36
0.7
5.09
0.41
0.43
0.33
0.42
0.38
0.16
0.5
0.46
0.7
0.96
R*
0.024
0.02
0.151
0.05
0.081
0.006
0.032
0.017
0.013
0.02
0.17
0.192
0.508
0.013
0.018
0.03
0.019
0.059
0.011
0.04
0.018
0.095
0.03
NPDES = National Pollutant Discharge Elimination System.
-------
As discussed in Sections 3.1, 4.1, 5.1, and 6.1, interannual variation in precipitation is
much greater than for temperature, making historic trends more difficult to characterize and
future projections more uncertain. This variability is reflected in the comparison among states
summarized in Table 7-1. All four states had increasing trends in precipitation from 1901-2000,
with the greatest increases in Maine and the lowest in Ohio. Only the long-term trend in average
annual precipitation in Maine was significant. But for the more recent historic period covering
the period of biological sampling, Maine, North Carolina, and Ohio had decreasing trends in
precipitation, with the largest decrease in North Carolina. Future projections for average annual
precipitation among states were variable. For Maine and Ohio, the projections are fairly
consistently for increases in average annual precipitation. This would lead to an expected
scenario of modest temperature increases and wetter conditions for these two states. But
seasonal variability must also be considered. In particular, historic trends and future projections
for summer precipitation give a somewhat different expectation. For example, in North
Carolina, average annual precipitation increased over the past century, but average summer
precipitation declined at an even higher rate (see Table 7-1). This suggests that the scenario
during the summer biological sampling period is one of warmer and drier (not wetter) conditions
(Karl et al., 2009).
Based on the relatively large historic rate of temperature increases in Utah, expectations
are that biological responses would be most readily observed for this state. However, there are
often observed interactions between temperature and flow (e.g., Yarnell et al., 2010), and though
flow data were not available from most of the long-term sites examined for biological responses,
the increasing historic trend in precipitation suggests that wetter conditions over the period of
biological sampling may have ameliorated higher temperatures to some extent. This scenario
may continue in the future, but there is a lot of uncertainty around the projections for summer
precipitation changes in Utah (Christensen and Lettenmeier, 2006; Schoof et al., 2010; Gutzler
and Robbins, 2011), which range from moderate increases to small decreases (see Table 7-1).
While Maine and Ohio both have seen declines in average summer precipitation over the
previous 3 decades, model projections are for increases in summer precipitation for the future
(Hayhoe et al., 2007; UCS, 2006; Easterling and Karl, 2001; Wuebbles and Hayhoe, 2004;
Hayhoe et al., 2010; Schoof et al., 2010). From this, expectations for Maine and Ohio should be
for warmer and wetter conditions during the summer, the season typically considered stressful to
7-4
-------
aquatic biota, and during which biological sampling usually takes place. Even with more
precipitation, increasing evapotranspiration associated with higher temperatures could still result
in lower stream flows.
7.2. COMPARISON OF REGIONAL TRENDS AND INDICATORS—HOW TO
INTERPRET OBSERVED RESPONSES
7.2.1. Comparison of Indicator Responses Among States and Regions
Biological responses were found in this study in both trend analyses with time and
climate variables, and in contrasts between years partitioned into hot and cold or wet and dry
groupings representing surrogates of future climate conditions. Such findings are reasonable, in
particular for ecological or life history trait groups that have been documented in other studies
(e.g., Gallardo et al., 2009; Beche and Resh, 2007; Bonada et al., 2007b). However, analyses
testing for relevant biological responses to climate patterns often lacked spatial consistency both
within and across states. This can be seen in the comparative results summarized in Tables 7-3
to 7-8 (original and more detailed results are presented by state in Sections 3 to 6). Temperature
preference trait groups responded to temperature and precipitation or flow changes, albeit with
regional variation (see Tables 7-5, 7-6, and 7-8). The number of warm-water taxa increased
significantly over time at lower elevation locations in both Maine (site 56187—Sheepscot, site
57011—W. Br. Sheepscot) and Utah (site 4951200—Virgin), but not at all stations, and not in
North Carolina (see Table 7-3). The response of warm-water taxa at the Maine stations appears
consistent with climate change expectations, given the predominance of warm-water taxa
coupled with the observed increasing temperatures over time. However, neither abundance nor
richness of warm-water taxa was directly correlated with temperature at this station. The
increasing temporal trend in warm-water taxa was corroborated by correlation with temperature
in Utah, but not in Maine (see Table 7-4).
Cold-water taxa decreased over time at one of the higher elevation sites in Utah (site
4927250—Weber), and were also negatively correlated with temperature, as would be expected
for a trait group responding to increases in temperature. Comparable associations with
temperature were not found in Maine or North Carolina. The longest-term station in Maine
(56817) occurred at a relatively low elevation, such that the number of cold-water taxa was very
small. Therefore, even though the long data record and low variation made trends in this trait
group significant, they are largely meaningless.
7-5
-------
Table 7-3. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics and year at long-term reference sites from three states. Entries are in bold
text if r > ±0.5 and are highlighted in gray if they are in a direction opposite of what is expected.
SON = September, October, November
Biological metric
No. cold-water taxa
Percentage cold-water individuals
No. warm-water taxa
Percentage warm-water
individuals
Total no. taxa
No. EPT taxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. intolerant taxa
Percentage EPT individuals
Percentage Ephemeroptera
individuals
Shannon- Wiener Diversity Index
Percentage noninsect individuals
Percentage dominant taxon
Percentage tolerant individuals
HilsenhoffBiotic Index
Utah
4927250
-0.50
-0.71
-0.03
-0.14
-0.10
-0.35
-0.41
-0.62
-0.02
-0.37
0.00
-0.37
0.13
-0.06
-0.10
0.19
-0.09
4951200
-0.33
-0.45
0.67
0.16
-0.09
-0.32
-0.53
-0.44
0.17
-0.46
0.01
0.27
-0.16
-0.14
0.16
0.20
0.16
4936750
-0.21
-0.12
0.37
0.39
0.34
0.05
0.13
-0.29
0.05
-0.30
-0.21
0.00
0.12
-0.06
-0.18
0.22
0.30
5940440
-0.46
-0.17
-
-
-0.26
-0.28
-0.23
-0.46
-0.10
-0.35
0.28
0.28
-0.06
0.11
0.28
-0.40
-0.33
Maine
56817
0.27
0.21
0.63
0.42
0.63
0.58
0.43
0.09
0.62
0.48
0.04
0.30
0.54
0.32
-0.36
-0.06
-0.18
57011
-0.03
-0.42
0.47
-0.33
0.72
0.58
0.43
0.18
0.57
0.46
-0.33
-0.24
0.12
0.36
0.09
0.15
0.58
57065
0.35
0.06
0.44
-0.17
0.39
0.39
0.26
0.30
0.17
0.61
-0.11
-0.17
0.22
0.61
-0.39
-0.06
0.17
North Carolina
NC0109
0.12
0.05
-0.46
0.05
-0.49
-0.02
-0.06
0.43
-0.06
0.23
0.49
0.38
-0.27
-0.13
0.09
0.13
-0.42
-------
Table 7-3. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics and year at long-term reference sites from three states. Entries are in bold
text if r > ±0.5 and are highlighted in gray if they are in a direction opposite of what is expected. SON =
September, October, November (cont.)
Biological metric
Richness
Percentage individuals
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Utah
4927250
-0.08
-0.20
0.00
-0.18
-0.12
0.56
-0.51
-0.03
0.32
-0.35
0.43
-0.15
-0.04
0.53
-0.36
0.34
4951200
0.13
0.14
-0.35
-0.02
0.06
0.44
-0.05
0.06
-0.25
0.05
0.03
-0.45
0.49
0.50
0.16
0.08
4936750
0.19
0.34
-0.03
-0.03
0.06
0.77
0.39
0.13
-0.24
-0.06
0.15
-0.03
0.12
0.33
0.40
0.06
5940440
-0.22
-0.22
-0.03
0.03
-0.07
0.24
-
-0.15
0.17
0.00
-0.39
0.00
0.22
-0.44
-
0.00
Maine
56817
0.52
0.27
0.54
0.36
0.41
0.26
0.25
0.55
-0.23
-0.02
0.53
0.34
0.26
0.12
0.10
-0.15
57011
0.25
0.62
0.36
0.34
0.54
0.28
0.11
0.44
0.36
0.06
-0.30
-0.64
0.15
-0.45
-0.03
-0.21
57065
0.15
0.38
0.80
0.20
0.13
0.15
0.19
0.20
0.11
-0.22
0.00
0.00
-0.17
0.17
0.14
0.17
North Carolina
NC0109
-0.58
-0.58
-0.14
-0.19
0.12
0.09
-0.33
0.12
0.35
-0.42
0.60
-0.02
0.38
0.27
-0.24
0.45
-------
Table 7-4. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics and temperature at long-term reference sites from three states. Entries are in
bold text if r > ±0.5 and are highlighted in gray if they are in a direction opposite of what is expected.
SON = September, October, November
Biological
metric
No. cold-water
taxa
Percentage cold-
water individuals
No. warm- water
taxa
Percentage
warm-water
individuals
Total no. taxa
No. EPT taxa
No.
Ephemeroptera
taxa
No. Plecoptera
taxa
No. Trichoptera
taxa
No. intolerant
taxa
Percentage EPT
individuals
Utah
4927250
PRISM
mean
annual
-0.57
-0.46
-0.38
-0.15
-0.32
-0.45
-0.55
-0.51
-0.23
-0.51
0.07
Obs
mean
max
Jul
-0.36
-0.28
-0.22
-0.13
-0.25
-0.38
-0.38
-0.44
-0.24
-0.29
0.10
4951200
PRISM
mean
annual
-0.31
-0.36
0.42
0.56
-0.72
-0.72
-0.79
-0.56
-0.24
-0.66
0.27
Obs
mean
max
Jul
-0.29
-0.12
0.19
0.23
-0.20
-0.34
-0.44
-0.09
-0.19
-0.32
0.08
4936750
PRISM
mean
annual
-0.11
-0.21
-0.10
-0.13
-0.09
-0.18
-0.19
0.19
-0.02
0.05
-0.06
Obs
mean
max
Jul
-0.31
0.15
-0.10
-0.17
-0.56
-0.53
-0.29
-0.25
-0.73
-0.14
-0.06
5940440
PRISM
mean
annual
0.03
-0.33
-
—
-0.15
-0.40
-0.17
-0.65
-0.03
-0.35
0.00
Obs
mean
max
Jul
0.29
-0.21
-
—
0.11
-0.12
0.05
-0.27
0.05
-0.15
0.14
Maine
56817
PRISM
mean
annual
0.23
0.18
0.16
0.13
0.18
0.14
0.08
0.08
0.15
0.14
0.11
Obs
mean
max
Jul
0.07
0.05
0.26
0.25
0.24
0.19
0.22
0.12
0.11
0.11
0.16
57011
PRISM
mean
annual
0.20
-0.03
0.26
0.42
0.05
0.12
0.46
0.11
0.05
0.09
0.55
Obs
mean
max
Jul
0.20
0.12
0.32
0.52
0.20
0.25
0.37
0.25
0.14
0.25
0.45
57065
PRISM
mean
annual
-0.47
-0.33
-0.61
0.00
-0.33
-0.44
-0.44
-0.50
-0.23
-0.61
-0.17
Obs
mean
max
Jul
-0.59
-0.17
-0.67
0.28
-0.61
-0.61
-0.38
-0.50
-0.40
-0.84
0.11
North Carolina
NC0109
PRISM
mean
annual
-0.32
-0.27
-0.14
0.02
0.15
-0.13
-0.15
0.04
-0.02
0.23
-0.05
Obs
mean
max Jul
-
-
-
—
-
-
—
-
-
-
-
oo
-------
Table 7-4. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics and temperature at long-term reference sites from three states.
Entries are in bold text if r > ±0.5 and are highlighted in gray if they are in a direction opposite of what
is expected. SON = September, October, November (cont.)
Biological
metric
Percentage
Ephemeroptera
individuals
Shannon-Wiener
Diversity Index
Percentage
noninsect
individuals
Percentage
dominant taxon
Percentage
tolerant
individuals
Hilsenhoff
Biotic Index
Utah
4927250
PRISM
mean
annual
-0.36
-0.04
-0.02
-0.04
0.00
-0.14
Obs
mean
max
Jul
-0.35
0.03
0.04
-0.06
-0.17
-0.16
4951200
PRISM
mean
annual
0.41
-0.43
-0.49
0.34
-0.20
-0.10
Obs
mean
max
Jul
0.12
0.03
0.05
-0.03
0.04
-0.12
4936750
PRISM
mean
annual
-0.15
-0.03
-0.21
-0.15
-0.10
0.15
Obs
mean
max
Jul
-0.09
-0.52
-0.21
0.39
-0.54
0.21
5940440
PRISM
mean
annual
0.00
-0.33
-0.06
0.22
-0.04
0.28
Obs
mean
max
Jul
0.07
-0.07
0.07
0.14
0.05
-0.07
Maine
56817
PRISM
mean
annual
0.23
0.13
-0.05
0.00
-0.01
-0.11
Obs
mean
max
Jul
0.31
0.20
0.14
-0.05
0.13
-0.06
57011
PRISM
mean
annual
0.52
0.45
-0.14
-0.30
-0.06
-0.24
Obs
mean
max
Jul
0.48
0.42
-0.29
-0.33
0.21
-0.27
57065
PRISM
mean
annual
-0.22
-0.50
-0.11
0.33
0.11
0.22
Obs
mean
max
Jul
0.06
-0.56
-0.39
0.39
0.06
-0.06
North Carolina
NC0109
PRISM
mean
annual
-0.16
0.13
0.20
-0.24
0.09
0.13
Obs
mean
max Jul
—
-
-
-
-------
Table 7-5. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics and precipitation at long-term reference sites from three states. Entries are in
bold text if r > ±0.5 and are highlighted in gray if they are in a direction opposite of what is expected.
SON = September, October, November
Biological metric
No. cold-water taxa
Percentage cold-water
individuals
No. warm- water taxa
Percentage warm- water
individuals
Total no. taxa
No. EPT taxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. intolerant taxa
Percentage EPT individuals
Percentage Ephemeroptera
individuals
Shannon- Wiener Diversity
Index
Percentage noninsect
individuals
Percentage dominant taxon
Utah
4927250
PRISM
mean
annual
-0.05
-0.06
-0.02
-0.08
-0.16
-0.18
-0.10
-0.15
-0.02
-0.17
-0.21
0.07
-0.40
-0.41
0.40
4951200
PRISM
mean
annual
0.33
0.25
-0.11
-0.32
0.32
0.57
0.32
0.36
0.38
0.48
0.10
0.05
0.10
0.25
-0.01
Observed
SON
-0.17
-0.16
0.29
-0.08
-0.05
0.05
0.01
-0.19
0.40
-0.11
-0.10
-0.05
-0.36
-0.21
0.54
4936750
PRISM
mean
annual
0.31
0.12
0.13
0.17
0.28
0.46
0.29
0.22
0.28
0.21
0.21
0.12
0.30
-0.18
-0.24
Observed
SON
0.11
0.39
0.33
0.28
-0.09
0.08
-0.03
0.05
0.02
0.11
-0.06
-0.27
-0.21
-0.27
0.33
5940440
PRISM
mean
annual
0.03
0.44
-
-
-0.15
0.28
-0.03
0.26
0.03
0.12
0.11
0.00
-0.11
0.17
0.00
Maine
56817
PRISM Mean
annual
0.24
0.21
0.05
0.01
0.13
0.10
0.16
0.02
0.03
0.26
-0.01
0.17
0.12
-0.04
0.01
57011
PRISM
mean
annual
-0.03
0.03
0.02
-0.06
0.05
0.09
0.37
-0.18
0.08
-0.03
0.00
0.27
-0.03
-0.45
0.06
57065
PRISM
mean
annual
-0.18
0.00
-0.15
-0.33
-0.11
-0.22
-0.15
-0.17
-0.06
-0.03
0.06
-0.11
-0.17
0.11
-0.11
Observed
SON
0.00
-0.06
-0.03
-0.61
0.17
0.06
-0.03
0.17
0.06
0.26
-0.33
-0.39
0.33
0.50
-0.39
North Carolina
NC0109
PRISM mean
annual
0.72
0.45
-0.54
-0.35
-0.60
0.17
0.11
0.51
-0.06
0.08
0.67
0.64
-0.67
-0.60
0.35
-------
Table 7-5. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics and precipitation at long-term reference sites from three states. Entries are
in bold text if r > ±0.5 and are highlighted in gray if they are in a direction opposite of what is expected.
SON = September, October, November (cont.)
Biological metric
Percentage tolerant individuals
Hilsenhoff Biotic Index
Richness
Percentage individuals
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Utah
4927250
PRISM
mean
annual
-0.04
0.18
0.04
-0.06
-0.09
-0.12
0.03
0.36
-0.02
-0.03
-0.18
0.26
-0.10
-0.35
-0.04
-0.12
0.07
-0.28
4951200
PRISM
mean
annual
0.00
0.03
0.33
0.28
0.35
0.15
0.19
-0.01
-0.09
0.14
0.32
0.36
-0.41
0.16
-0.08
-0.17
-0.32
-0.23
Observed
SON
-0.04
0.27
0.21
0.19
-0.07
-0.30
-0.06
-0.24
-0.27
-0.01
-0.01
0.30
-0.34
-0.12
0.08
0.10
-0.08
-0.16
4936750
PRISM
mean
annual
0.35
-0.30
0.09
0.08
0.57
0.45
-0.42
0.09
0.26
0.30
0.18
-0.36
0.39
-0.27
-0.18
0.23
0.32
0.24
Observed
SON
-0.03
-0.21
-0.26
-0.08
0.23
0.14
-0.22
0.05
0.26
0.00
-0.09
-0.03
0.18
-0.36
-0.58
0.07
0.32
0.09
5940440
PRISM
mean
annual
-0.11
-0.28
-0.07
-0.09
-0.57
0.17
-0.37
-0.35
--
0.03
0.00
0.06
-0.33
0.39
0.06
-0.17
--
-0.17
Maine
56817
PRISM Mean
annual
-0.15
-0.14
0.03
0.16
0.08
0.10
0.14
0.09
0.15
0.06
-0.10
0.18
0.08
-0.05
0.08
0.00
0.08
-0.11
57011
PRISM
mean
annual
0.00
0.06
0.13
-0.15
-0.11
0.12
0.29
0.15
-0.39
0.14
0.15
0.09
0.09
-0.12
0.24
0.24
-0.12
-0.12
57065
PRISM
mean
annual
-0.33
-0.11
0.27
-0.15
0.23
-0.31
-0.13
-0.39
-0.25
-0.08
0.28
-0.50
-0.06
0.06
0.00
-0.33
-0.14
0.11
Observed
SON
-0.17
0.06
-0.03
0.26
0.00
0.03
0.44
0.15
0.19
-0.03
0.33
-0.33
-0.11
0.00
-0.06
0.28
0.25
0.39
North Carolina
NC0109
PRISM mean
annual
-0.20
-0.75
-0.22
-0.31
-0.61
-0.69
-0.16
-0.02
-0.06
0.08
0.45
0.13
0.05
-0.56
0.42
-0.13
0.09
0.35
-------
Table 7-6. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics and year and flow at long-term reference sites from three states. Entries are
in bold text if r > ±0.5 and are highlighted in gray if they are in a direction opposite of what is expected.
SON = September, October, November
Biological metric
No. cold-water taxa
Percentage cold-water
individuals
No. warm- water taxa
Percentage warm- water
individuals
Total No. taxa
No. EPT taxa
No. Ephemeroptera taxa
No. Plecoptera taxa
No. Trichoptera taxa
No. intolerant taxa
Percentage EPT
individuals
Percentage
Ephemeroptera
individuals
Shannon-Wiener
Diversity Index
Percentage noninsect
individuals
Utah
4927250
Mean
annual
0.19
0.22
0.03
-0.07
0.01
0.00
0.16
0.04
-0.08
0.05
-0.22
0.26
-0.26
-0.37
Mean
SON
0.14
0.34
0.02
-0.01
-0.16
-0.05
-0.01
0.09
-0.10
-0.06
-0.25
0.26
-0.47
-0.34
4951200
Mean
annual
-
-
-
-
-
-
-
-
-
-
-
—
-
-
Mean
SON
-
-
-
-
-
-
-
-
-
-
-
—
-
-
4936750
Mean
annual
-
-
-
-
-
-
-
-
-
-
-
—
-
-
Mean
SON
-
-
-
-
-
-
-
-
-
-
-
—
-
-
5940440
Mean
annual
-0.03
0.56
-
-
-0.20
0.15
-0.17
0.26
-0.03
-0.06
0.56
0.56
-0.22
0.39
Mean
SON
-0.15
0.56
-
—
-0.38
0.03
-0.30
0.20
-0.17
-0.12
0.33
0.22
-0.22
0.06
Maine
56817
Mean
annual
0.14
0.08
-0.02
-0.11
0.07
0.01
0.04
0.02
0.04
0.20
-0.11
0.10
0.07
-0.06
Mean
SON
0.03
-0.03
-0.04
0.11
0.02
-0.03
0.02
0.06
-0.03
-0.04
0.11
0.01
0.01
-0.09
57011
Mean
annual
-0.07
-0.06
0.05
-0.15
0.14
0.18
0.34
-0.25
0.14
-0.03
-0.09
0.18
0.00
-0.42
Mean
SON
0.23
0.24
-0.11
0.09
-0.11
-0.12
0.06
0.04
-0.29
-0.06
0.15
0.18
0.00
-0.20
57065
Mean
annual
-
-
-
-
-
-
-
-
-
-
-
—
-
-
Mean
SON
-
-
-
-
-
-
-
-
-
-
-
—
-
-
North Carolina
NC0109
Mean
annual
0.76
0.35
-0.54
-0.16
-0.67
-0.02
0.02
0.23
-0.06
0.08
0.56
0.53
-0.71
-0.71
Mean
SON
0.68
0.45
-0.50
-0.42
-0.45
-0.06
0.02
0.51
-0.26
-0.08
0.60
0.56
-0.75
b
to
-------
Table 7-6. Kendall tau nonparametric correlations analyses performed to examine associations between
commonly used biological metrics and year and flow at long-term reference sites from three states. Entries are
in bold text if r > ±0.5 and are highlighted in gray if they are in a direction opposite of what is expected. SON
= September, October, November (cont.)
Biological metric
Percentage dominant
taxon
Percentage tolerant
individuals
Hilsenhoff Biotic Index
Richness
Percentage individuals
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Collector filterer
Collector gatherer
Scraper/herbivore
Predator
Swimmer
OCH
Depositional
Erosional
Utah
4927250
Mean
annual
0.44
-0.17
0.34
-0.08
0.13
-0.06
0.07
0.24
0.09
0.25
-0.16
-0.37
0.49
-0.26
-0.28
0.12
-0.32
0.23
-0.50
Mean
SON
0.44
-0.35
0.51
-0.16
-0.01
-0.23
-0.04
0.03
-0.01
0.16
-0.22
-0.34
0.57
-0.53
-0.25
0.06
-0.55
0.14
-0.59
4951200
Mean
annual
-
—
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Mean
SON
-
—
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
4936750
Mean
annual
-
—
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Mean
SON
-
—
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
5940440
Mean
annual
0.33
-0.04
-0.39
-0.15
-0.09
-0.57
0.17
-0.45
0.00
-
-0.03
-0.22
0.17
-0.44
0.61
0.50
0.06
-
-0.17
Mean
SON
0.22
-0.25
-0.39
-0.15
-0.34
-0.70
0.10
-0.60
-0.35
-
-0.15
-0.11
0.17
-0.33
0.39
0.28
-0.17
-
-0.28
Maine
56817
Mean
annual
0.00
-0.22
-0.10
0.00
0.14
0.06
0.02
0.01
0.14
0.00
0.06
-0.11
0.15
-0.02
-0.11
-0.01
0.06
-0.04
-0.12
Mean
SON
-0.05
-0.04
-0.15
-0.09
0.14
0.07
0.13
-0.15
0.21
0.01
-0.07
-0.13
-0.04
0.05
0.19
-0.13
0.27
0.12
-0.04
57011
Mean
annual
0.03
-0.03
0.09
0.25
-0.06
-0.11
0.09
0.32
0.09
-0.42
0.17
0.24
0.06
0.00
-0.09
0.21
0.15
-0.15
-0.03
Mean
SON
-0.15
0.21
-0.27
-0.34
0.00
-0.17
-0.03
0.00
-0.12
0.02
-0.41
-0.12
0.24
0.06
0.27
0.15
0.33
0.21
0.03
57065
Mean
annual
-
—
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Mean
SON
-
—
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
North Carolina
NC0109
Mean
annual
0.31
-0.16
-0.56
-0.26
-0.43
-0.65
-0.72
0.00
-0.17
-0.25
-0.04
0.27
-0.05
0.24
-0.45
0.38
-0.16
-0.02
0.53
Mean
SON
0.56
-0.05
-0.67
-0.18
-0.31
-0.65
-0.65
-0.16
0.17
-0.25
-0.04
0.53
0.05
0.20
-0.71
0.27
-0.13
-0.05
0.56
-------
Table 7-7. Mean metric values (±1 SD) for sites in three states in coldest, normal, and hottest year samples.
Year groups are based on PRISM mean annual air temperature values. One-way ANOVA was done to evaluate
differences in mean metric values. Entries with superscripts have significant differences across year groups;
those entries with different superscripts are significantly different from each other (e.g., coldest total no. taxa vs.
normal and hottest total no. taxa)
Location
Year group
No. total taxa
No. EPT taxa
HBI
No. cold-water
taxa
No. warm-water taxa
% cold-water
individuals
% warm-water
individuals
Utah
4927250
4951200
4936750
5940440
Coldest
Normal
Hottest
Coldest
Normal
Hottest
Coldest
Normal
Hottest
Coldest
Normal
Hottest
27.5±3.5A
21.5 ± 7.8AB
17.2±3.3B
22.8±6.6A
19.8±3.2A
14.5 ± 1.9B
22. 3 ±6.1
25.7 ±4.0
24. 3 ±8. 7
23.0 ±7.0
20.0 ±2. 6
19.3±3.5
17.4±2.1A
13.6 ±4.9^
8.8±2.2B
12.3 ± 3.9A
9.5±2.6A
5.3 ± 1.5B
13.7±3.2
15. 5 ±1.4
14.0 ±6.6
14. 3 ±4.0
12.7 ±2.1
ll.Oil.O
4.9±1.1A
3.4±1.1A
1.0±0.7B
4.5±2.4A
5.3 ± 1.2A
0.8±0.5B
6. 3 ±1.5
6.3 ±1.0
5.7±2.9
4.0 ±2.6
3. 3 ±0.6
3. 3 ±1.2
2.3 ±0.8
1.1 ±0.7
1.0 ±1.2
1.5±0.6A
1.5±0.8A
3.8±1.3B
0.3 ±0.6
0.7 ±0.8
0.7 ±1.2
--
--
-
6. 5 ±5.4
6. 7 ±7.4
l.Oil.l
15.7 ± 10.9AB
23.4 ± 15.6A
0.2 ± 0.2B
24. 3 ±4.1
14.9 ±6.8
17.7 ±8. 5
12.1 ±6.2
10.0 ±9.2
8.4 ±5. 9
0.6 ±0.5
0.4 ±0.3
0.3 ±0.4
7.7 ±6.7
18.1 ±15. 3
27.8 ±19.4
0.03 ±0.1
0.1 ±0.2
0.1 ±0.2
--
--
-
Maine
56817
57011
57065
Coldest
Normal
Hottest
Coldest
Normal
Hottest
Coldest
Normal
Hottest
20. 9 ±4. 3
20.8 ±5.4
24.1 ±3. 8
21.7±4.8
24.1 ± 10.4
25.2 ±3.4
22.1 ±8.0
21.7±3.5
18.4 ±3. 7
12. 3 ±2.6
12.7 ±3.7
14. 3 ±2. 3
9.8±1.3
10.0 ±3.7
11.5±1.1
8.9±2.6
8.2 ±1.8
6.8±2.0
4.0 ±0.5
3.9±0.5
3. 8 ±0.4
5.0±0.8
3.9±0.9
4.4 ±0.4
4.3±0.3
5.1 ±0.9
4.8±1.3
0.5 ±0.5
0.5 ±0.8
1.0±0.5
0.8 ±0.3
1.6±0.6
1.2 ±0.6
2.4 ±1.2
1.7±0.3
1.6±0.7
6.7 ±2.2
7.1 ±2.5
8.6 ±2. 5
6.4 ±2.0
7.3 ±2. 3
8. 5 ±1.6
6. 3 ±0.6
6. 8 ±1.5
4. 8 ±1.3
0.5 ±0.6
0.8±1.7
0.9±0.8
3.0 ±5.2
6.1 ±6.0
1.9±0.4
7.8 ±6.4
5.3±5.9
5.0 ±3. 3
15.1 ±6. 9
17.7 ±8.7
23.7 ±14.4
23.5 ±15.9
50.0 ±12.0
40. 8 ±12. 8
44.0 ±22. 5
32.8 ±10.8
46.6 ±17.6
-------
Table 7-7. Mean metric values (±1 SD) for sites in three states in coldest, normal, and hottest year samples.
Year groups are based on PRISM mean annual air temperature values. One-way ANOVA was done to
evaluate differences in mean metric values. Entries with superscripts have significant differences across year
groups; those entries with different superscripts are significantly different from each other (e.g., coldest total
no. taxa vs. normal and hottest total no. taxa) (cont.)
Location
Year group
No. total taxa
No. EPT taxa
HBI
No. cold-water
taxa
No. warm-water taxa
% cold-water
individuals
% warm-water
individuals
North Carolina
NC0109
Coldest
Normal
Hottest
86.0 ±7.0
84.2 ±6. 8
84.7 ±21.5
34.0 ±1.7
34.4 ±3. 8
34. 3 ±4.0
4. 5 ±0.4
4.2 ±0.7
4.4 ±0.5
4.3 ±1.5
5.4 ±1.7
4.0 ±1.7
8.3 ±0.6
7.4 ±1.7
7.3 ±2.3
2.3 ±0.7
3.6±2.9
2.2 ±1.0
7.7 ±2.5
7.6 ±2. 5
7.0 ±1.3
-------
Table 7-8. Mean metric values (±1 SD) for sites in three states in driest, normal, and wettest flow year samples.
Year groups are based on mean annual flow values from USGS gages. One-way ANOVA was done to evaluate
differences in mean metric values. Groups with no superscripts are not significantly different (p < 0.05). Bolded
entries with superscripts have significant differences across year groups; those entries with different
superscripts are significantly different from each other (e.g., lowest no. EPT taxa vs. highest no. EPT taxa)
Location
Year group
No. total taxa
No. EPT taxa
HBI
No. cold-water
taxa
No. warm-water
taxa
% cold-water
individuals
% warm-water
individuals
Utah
4927250
4951200
4936750
5940440
Driest
Normal
Wettest
Driest
Normal
Wettest
Driest
Normal
Wettest
Driest
Normal
Wettest
21.0 ±7.8
22.5 ±7.1
22.3 ±6.6
16.8 ±2.5
18.3 ±3.9
14.5 ±7.3
20.7 ±3.2
24.8 ±6.1
27.7 ±5.1
19.3 ±0.6
23. 3 ±7.5
19.7 ±2.9
13.6 ±5.4
12.6 ±5.0
14.0 ±4.8
6.3 ± 1.7A
8.7 ± 2.5^
12.5 ± 4.7B
12.3 ±2.1
15.0 ±4.2
16.3 ± 1.5
11.3 ±0.6
13.7 ±4.7
13.0 ± 1.7
2.6 ±1.3
2.9 ±2.3
4.1 ±1.5
2.5 ±2.4
4.2 ±2.1
4.5 ±3.1
5.0 ±0.8
6.8 ±2.2
6.8 ±0.9
3.0 ±1.0
4.3 ±2.5
3. 3 ±0.6
1.0 ±0.7
1.7±1.1
1.5 ±1.2
3.0 ±1.8
1.5 ±0.8
2.3 ±1.3
0.5 ±0.6
0.5 ±1.0
0.8 ±1.0
~
~
~
3.7 ±4.5
2.9 ±2.6
9.1 ±8.8
7.1 ±8.9
20.7 ±18.1
12.8 ±13.4
11.2±6.3A
23.4 ± 3.1B
19.3 ± 6.6^
5.3±2.1
10.9 ±6.8
14.4 ±7.4
0.4 ±0.3
0.5 ±0.4
0.4 ±0.4
17.9 ±10.0
24.9 ±20.5
7.4 ±5.4
0.05 ±0.1
0.10 ±0.2
0.15 ±0.3
-
-
-
Maine
56817
Driest
Normal
Wettest
22.2 ±4.4
20.9 ±2.8
22.7 ±6.9
13.4 ±3.0
12.6 ±2.3
13.3 ±4.1
3.9 ±0.5
3.9 ±0.4
3.9 ±0.5
0.7 ±0.5
0.4 ±0.4
0.9 ±0.9
8.0 ±2.4
6.8 ±1.8
7.7 ±3.3
0.7 ±0.5
0.2 ±0.3
1.4 ±1.9
22.4 ±13.9
16.4 ±8.1
18.1 ±9.8
-------
Table 7-8. Mean metric values (±1 SD) for sites in three states in driest, normal, and wettest flow year samples.
Year groups are based on mean annual flow values from USGS gages. One-way ANOVA was done to evaluate
differences in mean metric values. Groups with no superscripts are not significantly different (p < 0.05). Bolded
entries with superscripts have significant differences across year groups; those entries with different
superscripts are significantly different from each other (e.g., driest no. EPT taxa vs. wettest no. EPT taxa)
(cont.)
Location
57011
57065
Year group
Driest
Normal
Wettest
Driest
Normal
Wettest
No. total taxa
26.3 ±6.2
20.5 ±5.6
24.2 ±7.7
20.3 ±6.5
23.4 ±6.0
18.4 ±2.2
No. EPT taxa
10.8 ±2.4
9.3 ±2.2
11.3 ±2.4
8.0 ±2.6
8.1 ±3.0
7.8 ±1.3
HBI
4.3 ±0.9
4.5 ±1.2
4.5 ±0.5
4.8 ±1.3
4.7 ±0.3
4.7 ±1.2
No. cold-water
taxa
1.3 ±0.3
1.1±0.7
1.2 ±0.8
2.4 ±1.2
1.6 ±0.7
1.7 ±0.3
No. warm-water
taxa
7.9 ±2.5
6.9 ±1.7
7.3 ±2.3
6.1 ±0.5
6.1 ±2.5
5.7 ±0.9
% Cold-water
individuals
2.3 ±1.3
4.4 ±7.2
4.2 ±4.5
7.5 ±6.7
3.1 ±1.0
7.6 ±5. 3
% Warm-water
individuals
43.0 ±24.3
32.8 ±15.6
38.5 ±12.0
56.1 ±16.0
28.1 ±3.8
39.1 ±15.1
North Carolina
NC0109
Driest
Normal
Wettest
95.0±11.1
79.4 ±8.8
83.7 ±10.0
33.7 ±1.5
33.0 ±3.0
37.0 ±3.5
4.7 ±0.7
4.2 ±0.3
4.2 ±0.9
4.0 ±1.0
4.6 ±1.3
5.7 ±2.5
9.3 ± 1.2A
6.6 ± 0.9B
7.7 ± 1.5^
2.7 ±1.5
2.0 ±0.9
4.5 ±3. 3
8.0 ±3.0
6.7 ±1.2
8.2 ±2.4
-------
For Maine and North Carolina, abundance or richness of cold-water taxa was more often
related to precipitation (see Table 7-5), though trends were not always significant. While
long-term increasing trends in temperature already can be demonstrated for many regions (see
Sections 3.1, 4.1, 5.1, and 6.1), this is seldom the case for precipitation or flow-related variables.
Long-term data for flow (e.g., IHA) variables tend to be scarcer; and climate change projections
for precipitation are variable among regions. Nevertheless, the importance of ongoing changes
in precipitation and its effects in combination with temperature on flow regime should not be
discounted.
Several biological metrics, evaluated for differences between years partitioned based on
temperature (hottest/coldest/normal years) or precipitation (wettest/driest/normal years) regime,
showed patterns in one or another state, but only a few showed statistically significant patterns at
sites in more than one state, and none showed common patterns among all states. Overall, more
metrics were significantly associated with temperature-related variables than with precipitation
variables (see Tables 7-7 and 7-8).
Results of ANOVA testing for differences in ecological trait and scenario metrics
between year groups representing surrogates of future climate condition also varied among sites.
At Utah Stations 4927250 (Weber) and 4951200 (Virgin), hottest-year samples had significantly
fewer cold-water taxa than coldest-year samples (see Table 7-7). The greatest differences
generally occurred between hottest- and coldest-year samples, while normal-year samples were
variable. Warm-water taxa showed even fewer responses, increasing during hottest years only at
Colorado Plateau station 4951200 (Virgin) of the four reference stations tested (see Table 7-7).
Cold-water taxa were least abundant during the driest years and more abundant during wet or
normal years at Utah Station 4936750 (Duchesne), but did not respond differently among
wettest, driest, and normal years for other Utah stations (see Table 7-8). In contrast to Utah, in
Maine, there was greater response to wet/dry years than to temperature differences. Cold-water
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), though warm-water
taxa showed no response to a range of annual precipitation (see Table 7-8). No significant
responses of cold-water taxa over time or to temperature or precipitation were found at the few
other reference stations that could be tested (see Table 7-8).
7-18
-------
Other biological metrics that were sometimes responsive to climate variables include
functional feeding groups (e.g., predators, collector-filterers) or life history habits (e.g.,
swimmers, climbers). Feeding, life habit, and other functional trait groups are often included as
metrics in state multimetric indices (MMIs). It is, thus, recommended that, on a case by case
basis, the vulnerability of this class of metrics be evaluated through trend and correlation
analysis, as well as through assessment of composition by temperature sensitive taxa.
The abundance and richness of EPT taxa and of subsets of this taxonomic group also
were responsive to variations in climate variables. In Utah, the richness of EPT and some
component taxa decreased with year and/or with temperature at one or two of the stations tested,
but not at Maine or North Carolina locations (see Table 7-5). In contrast, EPT taxa increased
with increasing precipitation at one Utah station. In North Carolina, EPT taxa also increased
with increasing precipitation. At one station in Maine (5011), the abundance of EPT taxa
increased with increasing temperature. This would be counter to expectations if increasing
temperatures are equated with increasing stress; however, this followed the increase in
warm-water taxa at this location, and many abundant EPT taxa at this location are warm-water
taxa (see Table 4-9). EPT taxa are generally considered sensitive and are included in many state
condition indices and models. However, it appears possible from our study results that at least
some portions of the increases or decreases seen in richness or abundance of EPT taxa with year,
temperature, or precipitation may be related to increases or decreases in the cold or warm-water
EPT taxa that are included in the metric. Given that we consistently found a moderate
relationship between temperature sensitivity and sensitivity to organic pollution, it is also
possible that some of the observed responses are related to pollution stress. This aspect of
confounding was controlled to the extent possible through use of only reference stations.
However, a few of the stations analyzed had levels of urban and/or agricultural land uses that
would suggest possible impact (e.g., all three Maine stations).
Our spatially variable results raise the question of how widely applicable our study
results are to the regions within which the long-term stations occurred, and across regions and
states. Overall, spatial consistency in biological responses could be used as evidence that a
particular trend or relationship is real and widely occurring. It would add strength to making
regional inferences regarding particular biological responses or the value of particular indicators
or climate change responses, even though the underlying station selection process was not
7-19
-------
random, meaning that the regional representativeness of results at the individual long-term
stations cannot be determined. To decide how these results can be interpreted with regard to the
climate change vulnerabilities of biomonitoring programs and possible adaptations, it is
important to consider several contributing reasons for the inconsistencies. These include
intrinsic data limitations, other contributing or confounding factors, and differences in regional
characteristics that can alter the influence of climate variables.
The ecological traits of temperature and hydrologic preferences or sensitivities (e.g., Poff
et al., 2006b) 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; Montes-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.
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).
7.2.2. Factors Contributing to Spatial Variability in Observed Biological Responses
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 (see Section 3.7). Sites in the Colorado Plateau ecoregion and at lower
elevations had significantly more warm-water taxa, but numbers of warm-water taxa were low at
the Utah reference sites. In Maine, the Northeastern Highlands sites had the highest mean
number of cold-water taxa, followed closely by the Northeastern Coastal Zone sites (see
Section 4.7). Overall, the number of cold-water taxa in all the Maine ecoregions evaluated was
low (1 to 2 taxa). The mean number of warm-water-preference taxa at sites in the Laurentian
7-20
-------
Plains and Hills was significantly higher than at sites in other ecoregions, while the Northeastern
Highlands sites had the lowest mean number of warm-water-preference taxa. These observed
ecoregional differences appear to be driven by elevation: there are more cold-water taxa at higher
elevation (>150 m) sites and more warm-water-preference taxa at lower elevation (<150 m) sites.
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. In
North Carolina, ecoregions also vary in the predominance of cold and warm-water taxa. The
richness of cold-water taxa is, on average, higher in the Mountain ecoregion than in the other two
ecoregions (see Section 5-7). The distribution of warm-water 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. Conversely, median richness and abundance of warm-water taxa is greater at
lower elevation sites.
The prevalence and distribution of cold and warm-water taxa also varied predictably with
stream order. First- and second-order streams in Utah had slightly greater relative abundance
and richness of cold-water taxa, and fewer warm-water taxa, compared to third- or higher-order
streams. These results suggest that effects are likely to vary spatially within states, reflecting
spatial differences in vulnerabilities. Biotic assemblages in the Wasatch and Uinta Mountains
and at higher elevations may be more vulnerable to the 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.
As observed in Utah, first- and second-order streams in Maine had slightly greater
relative abundances and richness of cold-water-preference taxa, while fourth- and higher-order
streams had more warm-water-preference taxa. 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
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
7-21
-------
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.
Distribution of cold and warm-water taxa was also related to watershed size. The smaller
watersheds in North Carolina (<35 mi2) had a greater proportion of cold-water taxa (based on
both abundance and richness), while larger watersheds (>100 mi2) had a greater proportion of
warm-water taxa (see Section 5-7). 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 other North
Carolina sites 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 taxa. This may be attributable
to the more limited time series of data available from North Carolina (see Section 7.3.1), as well
as to the use of categorical rather than abundance data (though this would not affect evaluation of
richness trends).
Despite the spatial variability of results, this study supports the concept that not all
regions are equally threatened or responsive to climate change. There is regional variability in
climate combined with spatial variability in vulnerability9 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), and stream order (Cereghino et al., 2003; Minshall et al., 1985) as
observed in our study results, as well as other factors such as 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).
7.2.3. Benthic Inferred Temperature
We calculated benthic inferred temperatures for three sites—the Weber River site (UT
4927250) in Utah, the Sheepscot River site (ME 56817) in Maine, and the New River site
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 etal., 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 different scales, e.g., the biological assemblage as a whole,
individual species, particular sites, stream types, etc.
7-22
-------
(NC0109) in North Carolina. These sites had the most number of years of biological data.
Benthic inferred temperature is based on relative abundance and temperature optima data for
macroinvertebrate taxa that occur in each sample. To make the calculation, the temperature
optima values for each taxon are multiplied by the relative abundance of that tax on, then the
products of those calculations are summed over all taxa in the sample and divided by the
summed relative abundances of all the taxa in the sample. The temperature optima values used
in these calculations were derived from weighted averaging or maximum likelihood modeling on
appropriate subsets of the state biomonitoring data, as described in Section 2.2.1.
We calculated benthic inferred temperatures at these sites to evaluate how closely trends
in benthic inferred temperatures tracked observed changes in air temperature over the period of
biological record. In addition to direction and amount of change over time, actual temperature
values were also important. We compared benthic inferred temperatures to air temperature
measurements because the air temperature measurements were derived from independent data
sets10. Thus, it provided a way to "test" how well thermal optima calculations captured changes
in observed temperatures over the time periods being evaluated.
To make these comparisons, we plotted benthic inferred temperature and air temperature
measurements versus year. At the Weber River site (UT 4927250), we compared benthic
inferred temperature to observed mean September/October air temperature values from the
nearest weather station11 and to PRISM mean annual air temperature. At the Sheepscot River
site (ME 56817), we compared benthic inferred temperature to observed mean July/August air
temperature values from the nearest weather station12 and to PRISM mean annual air
temperature. At the New River site (NC0109), we were limited to comparing benthic inferred
temperature to PRISM mean annual air temperature; temperature data from the nearest weather
station was not available for the period of biological record. We also experimented with
10The thermal optima calculations are based on instantaneous water temperature measurements that were taken at the
time of the biological sampling event.
11 We chose this time period because the thermal optima calculations in Utah are based on a subset of fall data. We
did not include November air temperatures in our comparison because the November biological sampling events
took place early in the month. Plus the inclusion of November temperatures would have substantially changed the
observed air temperature values, because on average, November air temperatures are >10°C lower than September
and October temperatures.
12We chose this time period because the thermal optima calculations in Maine are based on a subset of data collected
from July-September. We did not include September air temperatures in our comparison because the September
collection events took place early in the month. Plus the inclusion of September temperatures would have
substantially changed the observed air temperature values, because on average, September air temperatures are
approximately 5°C lower than July and August temperatures.
7-23
-------
grouping data from multiple sites (Weber River—UT 4927250; Virgin River—UT 4951200;
Duchesne River—UT 4936750) on the same plot to evaluate how site-specific differences can
influence overall trends in benthic inferred temperature.
Results show that benthic inferred temperatures are less variable than seasonal and annual
air temperature; and when they do change, they vary on a much smaller scale (generally less than
1°C) (see Figure 7-1). Unless there is a large shift in the composition of the assemblage towards
cold or warm-water taxa, as reflected in the thermal optima values of the taxa, benthic inferred
temperatures will stay relatively stable over time, so this is not unexpected. In general, patterns
in the benthic inferred temperatures track fairly closely with patterns in the seasonal air
temperature data, although during some years, there appears to be a lag effect in the biological
data (see Figure 7-1). At both the Sheepscot River site (ME 56817) and the Weber River site
(UT 4927250), benthic inferred temperature values are similar to the observed seasonal
temperature values (see Figure 7-1).
The PRISM mean annual air temperature values are lower than the benthic inferred
temperatures at all three sites (see Figure 7-2). At the Sheepscot River site (ME 56817) and the
Weber River site (UT 4927250), mean annual air temperature, which is less variable than the
seasonal air temperatures, varies by greater amounts than the benthic inferred temperatures,
while at the New River site (NC0109), benthic inferred temperatures are more variable (see
Figure 7-2). Our evaluation of trends in benthic inferred temperatures at the Weber River site
(UT 4927250) and the New River site (NC0109) was hindered by gaps in the biological data. At
all sites, our trend analyses were somewhat limited by the relatively short time period for which
biological data are available.
Site-specific differences were evident (i.e., the overall trend line is not very reflective of
the trend that occurred at the Weber River site [UT 4927250]) in the three Utah sites (see
Figure 7-3). The overall benthic inferred temperature trend for these Utah sites was equivalent to
a rate of increase of approximately 3°C in 25 years. This corresponds well with the magnitude of
air temperature increases observed for the period, suggesting that the estimates of benthic
invertebrate temperature optima were generally appropriate, and that using benthic invertebrate
occurrence and abundance coupled with temperature preferences provides evidence of benthic
community changes over time related to long-term changes in temperature. With a large enough
7-24
-------
o
0)
22
21
_
CD
E 20
19
'•:
-•- Benthic Inferred
-B- Mean July/August Air
1984
1989
1994
1999
2004
B
o
o
§ 13
+-i
2
CD
| 12
,ffl
•Ai
.a
o
"o
O
Q.
ID
W
n
CD
10
-B- Mean Sept/Oct Air
-•- Benthic Inferred
10
o
o
I
-*—'
co
&
I
cu
c
CD
1985
1990
1995
2000
2005
Figure 7-1. Comparison of trends in benthic inferred temperature and
seasonal observed air temperatures at (A) the Sheepscot River site (ME
56817); and (B) the Weber River site (UT 4927250).
7-25
-------
o
o
I
2 o
aj 8
CL
I
1984
PRISM Annual
-•- Benthic Inferred
1989
1994
1999
2004
22
21
o
O>
a,
0)
CD
20
B
o
2
oj
Q- 7
E '
|2
I6
c
C
r
1985
13
o
01
5 12
,01
t 11
<
i_
< 10
C
CD
ro
Q.
T3
£
I
C
i
CD
2005
-a- PRISM Annual Air
-•- Benthic Inferred
/
\ • * f * \ / \
t*-^\ / rf
^
24
23 £
El
22 1
21
20
19
18
0)
GO
1983
1988
1993
1998
2003
Figure 7-2. Comparison of trends in benthic inferred temperature and
PRISM mean annual air temperatures at: (A) the Sheepscot River site (ME
56817); (B) the Weber River site (UT 4927250); and (C) the New River site
(NC0109).
7-26
-------
1'
2
m
\u
E 15
c
O
i^
0 n
i i i i
STATION
R:=0.1-T
• 4927250
- • 4936750 p=0-017
• 4951200
*
* . •
• . f • • . „ *-v*-;^
-.-rr • :
* »
~
i i i i
1980 1985 1990 1995 2000 2005
Year
Figure 7-3. Benthic macroinvertebrate inferred temperature trend for
selected reference sites in Utah. Only samples collected in October and
November were used in these calculations.
data set, this type of analysis could be informative of long-term trends that are more widely
applicable than our analyses that were limited to data from single sites.
7.2.4. Basis for Inferring Climate Change Associations
Biological data reflect responses to interannual variations (e.g., year-to-year variations in
temperature, precipitation regime, etc.) and to multiyear to multidecadal "cyclic" climate
variations, such as the NAO, the Pacific Decadal Oscillation (PDO), or the El Nino Southern
Oscillation (ENSO) that drive differences in water temperature and hydrologic regimes in a
manner similar to the mechanisms linking to long-term climate change responses. The NAO, for
example, affects mainly winter weather conditions on decadal time scales (Hurrell, 1995). The
PDO, which influences western and southwestern regions, is generally considered to be a much
longer term, multidecadal phenomenon (Brown and Comrie, 2004; Mantua et al., 1997), while
ENSO is found to vary in the range of multiple years to a decade or more (e.g., Brown and
Comrie, 2004).
7-27
-------
A rigorous approach, were it supported by available data, could examine what
components of observable temporal variation in biological responses are attributable to
long-term directional climate change, and then apply general linear modeling or another
comparable approach to partition the variation within the observed biological responses between
interannual or cyclic and long-term directional climatic sources. However, most state
biomonitoring data sets, even the most critically developed (sensu Yoder and Barbour, 2009) and
long term, such as those analyzed in these pilot studies, are limited in duration and frequency of
sampling. These data are not able to support linear modeling, especially of several separate
variables, because the average scope of available data is typically 20 years or less, with 10 to
fewer than 20 annual data points over that time span. We explored an alternative analysis,
examining correlations between indices of known cyclic climatic variation (e.g., the NAO, PDO,
and ENSO) with biological metrics, focusing on those that also showed long-term temporal
responses, as well as correlations with temperature or precipitation13. In general, responses
varied by state and region, but most of the results were weak or not significant. In Utah, there
were some intriguing relationships found at individual long-term reference stations between trait
groups (e.g., warm-water-preference taxa, perennial taxa) and either the ENSO or PDO annual or
monthly indices. However, none of these were consistent spatially; therefore, no particular trait
or taxonomic group is a strong indicator of PDO or ENSO responses. The complexity of the
patterns compared to the relatively short (20 years or fewer) data sets suggests the importance of
further investigation and long-term monitoring, including further study on the relative
contributions of each index.
While it is important to consider NAO, PDO, and/or ENSO when evaluating
biomonitoring (or any other) data sets for climate change effects, there are still some practical
limitations, particularly in the biomonitoring application. Analyses would require data spanning
multiple (2-3) multidecadal cycles to be able to model the cycle-associated responses and extract
the residual long-term trend on a rigorous basis. The Maine Station 56817 (Sheepscot) data
series spanned 23 years, and this is long compared to most existing available biomonitoring data.
It also is likely that variations in the effects of the NAO interact with long-term climate change
effects, potentially damping increasing temperatures in negative years and augmenting them in
positive years (Durance and Ormerod, 2007). This is important, because the composite of
13Detailed results are available upon request.
7-28
-------
climate effects may underestimate long-term climate impacts during some periods, or
overestimate them during others. It would take proportionately more (longer-term) data to
separate these and confidently define the long-term climate change component.
Because the nature of most bioassessment data limit the ability to separate the magnitude
of observed trends among interannual, cyclical, and long-term directional climate effects, the
results obtained in this study cannot be interpreted as entirely attributable to directional climate
change. However, the net response of benthic or other aquatic community metrics to climate
sensitive variables, including water temperature and hydrologic patterns, can reasonably and
effectively be used to address the primary questions of this study. The direction and nature of
the observed climate responses can be applied to characterize the types of responses that can be
expected due to climate change, to identify the most sensitive indicators to climate change, and
to understand implications to multimetric indices or predictive models and their application by
managers to characterize the condition of stream resources for decision making. These effects
may be viewed in some respects as maximum estimates of probable effects, because multiple
components of climate change could be included, though at times, the resulting estimates may
also be undervalued.
7.2.5. Other Sources of Potential Spatial Confounding
There are other potential sources of spatial confounding of temporal trends, which were
tested in this study. Land use and land cover within a 1-km buffer of individual reference sites
indicated that anthropogenic influences were higher than desired (>5% urban or >10%
agricultural) at most sites. Though stations were initially screened at 1,2, and 5% urban, and 5
and 10% agricultural land use levels, final levels applied were 5% urban/10% agricultural in
Maine and North Carolina, and 2% urban/10% agricultural in Utah, based largely on the practical
need to not eliminate all stations with data that could be used for long-term analyses. The urban
land uses surrounding these sites generally consisted of low-intensity and open-space
development, and the agricultural land uses were mostly pasture/hay, with occasional cultivated
crops. Although higher than considered desirable for reference conditions, these final land use
criteria are more conservative than those used in several states. Georgia, Alabama, and South
Carolina apply land use criteria for selecting reference stations of <15% urban/<20% agricultural
for high gradient streams, and <15% urban/<30% agricultural in low gradient streams (Barbour
7-29
-------
and Gerritsen, 2006). It will take additional analysis to determine on an objective level whether
these criteria are adequate to minimize confounding of climate change effects.
It is reasonable and sometimes necessary to use less than "natural" conditions as a
baseline for spatial comparisons. For example, accessibility of a site for frequent (e.g., annual)
long-term sampling can be an important practical consideration. For example, the longest term
reference station in Maine, 56817 (Sheepscot), is generally (though not always) categorized as an
"A" station by Maine DEP, but is surrounded by about 16 urban and 23% agricultural land uses.
Though higher than would be considered ideal for "unconfounded" analyses, the level of urban
land use was stable over time (at about 16%), although forested conditions decreased from 84 to
57%, while agricultural land use increased 0 to 23%. At Maine's Station 57065, there was an
increase from 0 to 16% urban land use, but a decrease from 4 to 0% agricultural land use. At
Maine's Station 57011, urban land use increased from about 4 to 9%, and agricultural use from 0
to 18.5% with the changes coming from both forested and wetland uses. It is possible that such
land use changes may have contributed to trends observed at these sites. It is recommended for
all sampling stations, but especially for reference stations, that quantification of land-use
categories be documented. This will support tracking changes in land uses over time (although
land-use data are often only available at infrequent intervals), which will aid in separating this
from degradation due to climate change effects (and other stressors).
We further explored these relationships by using correlation analyses to determine
whether any available chemistry and habitat variables were significantly correlated with
biological metrics. Data availability limited this pursuit. For example, Utah only had chemistry
data. At two of the Utah long-term reference stations (Stations 5940440—Beaver and
4936750—Duschesne), some of the temperature preference metrics were significantly correlated
with water chemistry variables (see Sections 3.6.3.1 and 3.6.4.1). Many of the correlations were
driven by outliers, but a few of the water chemistry variables, notably chloride, may have
influenced trends in the biological assemblage. Chloride could be an indirect indicator of human
development, as increases are sometimes associated with increasing road development and/or
increasing application of road salt overtime (TRC, 1991). However, chloride concentrations
may also vary naturally with drought conditions.
In Maine, limited chemistry and habitat information were available (mainly in situ water
quality measurements and visual substrate estimates). At Site 56187 (Sheepscot), yearly trends
7-30
-------
in the biological data were likely influenced by nonpoint-source pollution (Maine DEP, personal
communication), but we lack the long-term chemistry data necessary to confirm this possibility.
Some of the habitat variables at Site 56817 also showed trends over time. Percentage boulders
and percentage gravel were significantly correlated with some of the biological variables.
However, based on conversations with Maine DEP, it appears that this "trend" actually reflects
observer bias, and it is not considered a real change over time in substrate characteristics. A
similar example occurred in North Carolina, where visual substrate estimates for one site showed
a fairly dramatic yearly trend. Scientists at NCDENR believe this also to be observer bias. More
problematically, there were some fairly dramatic trends in canopy cover and water chemistry
found at some North Carolina sites, which turned out to be due to data entry errors. This seems a
minor but important cautionary note, as the "false" trend in canopy cover seemed feasible
(increasing cover over time would be possible if there were an earlier instance of logging), and a
(nonsignificant) trend of decreasing water temperature over time appeared to be logically
consistent with increasing canopy cover. In the end, this very "appealing" discovery was false.
7.3. CHARACTERISTICS OF EXISTING BIOASSESSMENT PROGRAMS
RELEVANT TO DISCERNING CLIMATE CHANGE TRENDS
There are some inherent qualities of biomonitoring data that limit the ability to define
long-term trends, and to consider results representative of larger regions. We discuss these
limitations in the context of existing program objectives and understanding how biomonitoring
programs are likely to be affected by climate change in the future.
7.3.1. Sufficiency and Limitations of Data to Define and Partition Long-Term Trends
State and tribal bioassessment programs establish reference stations across their
jurisdictions for reference-based comparisons to assess condition, detect impairment, and
identify causes. The main objectives of these programs focus on spatial comparisons, and
program design elements reflect this. Assessment designs generally include random sampling
within a stream reach or watershed, or a combination of random plus some targeted sampling.
Random sampling tends to maximize spatial sources of variation. Rotating basin sampling
designs are often used, which typically include sampling once every 5 years. Collections are
usually of one sample per location per year, with measurements of few covariates.
7-31
-------
In contrast to the original spatial objectives of biomonitoring designs, detection of
climate change requires evaluation of trends over time, whether at a specific location or for a
defined area or stratum. There are some commonly observed limitations of many existing
biomonitoring programs with regard to assessment of trends. Despite the relatively large
numbers and broad spatial distribution of sampling stations in the biomonitoring data sets
analyzed (see Figures 3-6, 4-6, 5-6), few are sampled in more than one or a few years over the
entire period of record (see Table 7-9). As a result, there are a limited number of stations with
long-term data from which to analyze temporal trends. In addition, samples are often not
collected from the same sites every year (see Table 7-10), so many data sets have discontinuities,
making trend detection more difficult. In addition, trend analyses should be conducted using
data from "reference" or minimally affected stations to minimize influences from conventional
stressors. This study also found that climate change responses can differ among regions (see
summary in Section 7.2), potentially making it necessary to partition analyses by ecoregion or
other classification. However, there are seldom more than one or two long-term sites within a
particular region, and many regions have no long-term reference stations.
Table 7-9. Average distribution of reference and total stations by state,
categorized by duration of sampling
Years
sampled
Ito4
5 to 9
>10
Total
Maine
Ref
57
7
2
66
Total
696
40
6
742
North Carolina
Ref
89
13
3
105
Total
2,530
223
33
2,786
Utah
Ref
61
1
4
66
Total
482
41
26
549
Average
Ref
207
21
9
237
Total
3,708
304
65
4,077
% Ref
5.6
6.9
13.8
5.8
7-32
-------
Table 7-10. Time periods for which biological data were available at the
long-term monitoring sites in Utah (UT), Maine (ME), and North Carolina
(NC). Data used in these analyses were limited to autumn
(September-November) kick-method samples in the Utah data set, summer
(July-September) rock-basket samples in the Maine data set, and summer
(July-August) standard qualitative samples in the North Carolina data set
Station ID
UT 4927250
UT 495 1200
UT 4936750
UT 5940440
ME 56817
ME 570 11
ME 57065
NC0109
Water body
Weber
Virgin
Duchesne
Beaver
Sheepscot
W. Br. Sheepscot
Duck
New
Number of
years of data
analyzed
17
14
12
9
22
12
9
11
Years
1985-1995, 1998,2000,2001,
2003-2005
1985-1993, 1996, 2000-2002,
2004
1985-1993, 1995,2000,2001
1996-1998, 2000-2005
1985-2006
1995-2006
1997-2005
1983-1990, 1993, 1998, 2003
The limited number of "reference" stations with adequate long-term data records within
each ecoregion or other stratum of interest reduces the ability (1) to confirm regional trends,
(2) assert the strength of any trends discerned, and (3) to compare biological responses between
regions. Essentially, the low number of stations with sufficient long-term data limits replication
for testing of climate change effects. The small number of reference locations with long-term
data is a surprising but important finding that likely applies to many other biomonitoring data
sets.
A related factor is the actual length of the long-term data record. Reference locations in
this study yielded some valuable results, but also many nonsignificant patterns. It appears in
several of these cases that the length of the data record along with the number of years sampled
within the period is not sufficient to detect trends given the year-to-year variability of the metrics
being tested. As examples, the longest-term reference station in North Carolina, NC0109, had
11 years of data over a 21-year time span (1983-2003); the longest-term reference station in
Maine had 23 years of data over a 23-year time span (1984-2006); and three long-term reference
7-33
-------
stations in Utah had 19 years of data over a 21-year span (1985-2005, station 4927250—Weber),
15 years of data over a 20-year span (1985-2004, station 4951200—Virgin), and 14 years of
data over an 18-year span (1985-2002, station 4936750—Duchesne).
Data durations of about 13-20 years appear in the literature as an apparent minimum.
For example, analyzing an 18-year data set from a large number of streams in the UK, Durance
and Ormerod (2008) found significantly increasing temperature trends and significant
correlations of some invertebrate variables with temperature, although they concluded that water
quality improvements confounded interpretation of results. Chessman (2009) found significant
climate change trends in benthic invertebrate taxonomic families and trait groups within a
13-year data record in New South Wales, Australia. Daufresne et al. (2004) defined aquatic
community trends in the Rhone River based on data durations of 20 (macroinvertebrates) to
21 (fish) years. Although Daufresne et al. (2004) found several meaningful community patterns
and showed statistically significant trends in temperature, trends related to flow parameters were
generally not found to be significant based on the same duration of data. Two possibilities are
(1) in the Rhone River there were no temporal trends in flow and/or no relationships between
flow and invertebrate or fish communities; or (2) given the typically high variability of
hydrologic variables, the 20 to 21-year duration of data was adequate to discern temperature
trends but not to detect flow-related responses.
7.3.2. Other Biomonitoring Methods Considerations
Each of the states analyzed in this study use different collection methods that range from
single or multihabitat kicknet samples to different types of artificial substrate samples (see
Sections 3.2 through 6.2 for the specific sampling methods employed by each state evaluated in
this study). Some methods are likely to be more effective than others for certain applications
(e.g., Flotemersch et al., 2006), but it is still unclear which sampling protocol is best suited for
detecting climate change effects. Long-term changes in climate variables are expected to
contribute to a wide range of in-stream changes that can contribute to biological responses, such
as drought or flood-related changes in flows, and associated changes in nutrient loadings,
sediment loadings, habitat availability, and other interrelated factors. Given these
considerations, the ability to examine the full spectrum of naturally occurring biological
community components may be advantageous. In-stream multihabitat sampling may be more
7-34
-------
likely to provide realistic estimates of abundance or richness of a broader spectrum of indicator
taxa.
Use of artificial substrates were favored for pollution detection on the premise that
application of a uniform substrate eliminates the substrate variation among stations as a variable
that would confound detection of community responses to a pollution discharge or other
disturbance (e.g., Barbour et al., 1999; Cairns, 1982). But with the additional objectives of
testing for climate change effects, artificial substrates may be less advantageous. For example,
in Maine, rock baskets are placed in run habitats that will have sufficient water for the entire
deployment period. If drought conditions or altered seasonal precipitation leads to reduced flows
and a loss of edge habitat, the rock baskets are less likely to reveal the potential loss of edge taxa.
Even protocols that sample only riffles may be less likely to collect edge-specialized fauna.
However, the multiple habitat protocol used in North Carolina is more likely to detect such
shifts.
On the other hand, there is a significant disadvantage to changes in sampling methods,
due to the disruption it causes in temporal patterns that might otherwise be observed. Because of
this, any consideration of changing sampling methods should at least be accompanied by a
period of time in which both methods are applied simultaneously in order to develop translation
models. Even with some "side-by-side" sampling, translational models used to correct species
abundance for sampling method may not always be effective or overcome inherent sampling
biases. For example, if rock baskets do not effectively collect edge taxa, then no factor can be
defined that would translate multiple years of near-zero results into meaningful estimates of
abundance.
Because of considerations such as these that bear on the consistency of results, states
have a vested interest in continued use of their own methods to assure that new data are
meaningful to their program. Additional sampling might be considered in representative and/or
especially vulnerable regions as an adjunct to standard biomonitoring methods. For instance, in
streams with a high likelihood of transit!oning from perennial to intermittent status, collection of
samples from edge habitats could be considered.
Another potential hindrance to effective detection of climate change trends is relatively
low sampling effort and the lack of replication in station sampling. In most biomonitoring
programs, the concept of collection of replicate samples is relinquished in favor of collecting
7-35
-------
single composite samples. The composites can be either of multiple artificial substrates (e.g., in
Ohio, 5 Hester-Dendy samples per station are composited and processed as a single unit
[DeShon, 1995]); or a single sample unit can be a composite of collections made in multiple
representative habitats (NCDENR, 2006). In general, increasing the number of samples
collected and composited for a site has been found to decrease variance among "replicate"
(similar) sites and increase the precision of characterizing the assemblage at the site (Cao et al.,
2003; Diamond et al., 1996). Multihabitat sampling, applied in many biomonitoring programs
(e.g., Utah, North Carolina) is considered to yield representative and, therefore, precise samples
(Barbour et al., 2006; Hering, 2004). Though replication is considered necessary to determine
the precision of the sampling method (Barbour et al., 2000), it is often only accomplished on
about 10% of collections (e.g., Stribling et al., 2008; Barbour et al., 2006; Flotemersch et al.,
2006). However, with regard to detecting climate change temporal trends, knowledge of spatial
variation within a station (or stream reach), as well as between sites within a watershed or
ecoregion, may be valuable.
There are some environmental variables that can be measured along with biological
samples to aid in interpretation of results. For example, a detailed assessment of substrate and
related habitat condition, as was used in EMAP (Lazorchak et al., 1998), is valuable in
differentiating habitat disturbance from other stressors. If biomonitoring programs consider
climate change as an additional stressor, it would be valuable to have good information on water
temperatures and flows from biological collection sites. Existing sampling protocols usually
include concurrent point measurements of temperature, and sometimes also of pH, DO, and
conductivity, as these values are relatively easy to obtain with portable sondes. However, the
analyses conducted in this study illustrate that point measurements of temperature are not a good
measure of the stream conditions to which an aquatic community is exposed. They tend to
include a large amount of variation from time of day as well as date during the seasonal index
period when that measurement happened to be taken.
In this study, the lack of long-term, site-specific temperature and flow data impaired the
ability to conduct weighted-average modeling (or use of related approaches) to determine
temperature or flow parameter preferences for many taxa. It also made it difficult to conduct
simple trend and correlation analyses (see Sections 3 through 5). It would be beneficial to
consider deploying in situ equipment to obtain continuous water temperature and flow
7-36
-------
measurements at as many climate change monitoring sites as possible. Though such equipment
is widely available and much less expensive than it used to be, the sometimes severe resource
limitations experienced by states and tribes may limit the extent to which this recommendation
can be applied. Priorities could be set based on regional assessments of relative vulnerability to
climate change. For example, a limited number of deployments could be done at reference
locations in higher elevations, and/or in lower-order streams. There is also high value in
continued operation of USGS long-term flow and temperature gages.
7.4. REFERENCE STATION VULNERABILITIES
Several program elements in addition to biologic indicators need to be considered for
effective program management and adaptation in relation to climate change. The use of
reference stations and comparison to reference conditions are central to bioassessment. There
are two components of reference station vulnerability to climate change that are apparent from
this study. One is the negative drift of the biologically based characterization of reference
condition over time that will likely result from climate change effects on component biota. The
other is the threat to the quality status of reference stations from other global stressors, in
particular encroaching developed land uses. We, therefore, examined potential vulnerabilities in
the definition of reference conditions, in the synergistic effects between climate change and land
use, and in the vulnerability of reference sites to encroaching developed land uses.
7.4.1. Vulnerabilities in Assessing Reference Condition
Reference station comparisons are central to bioassessment. Both in the United States
(Clean Water Act) and in Europe (Water Framework Directive), the determination of ecological
status and integrity is based on a comparative approach ("reference based comparisons")
requiring reference locations that can be used to set expectations for "natural" conditions and
associated variability (Barbour and Gerritsen, 2006; Stoddard et al., 2006; Verdonschot. 2006;
Nijboer et al., 2004; Wallin et al., 2003). Impairment in the regulatory context represents an
unacceptable level of departure from this "expected" reference condition. Climate change can
alter the biological conditions at reference stations, and thereby influence reference-based
comparisons and the decisions that are based on those comparisons.
7-37
-------
The exploratory analyses conducted using North Carolina protocols (see Section 5.8)
illustrate expectations for drift over time in the biological status of reference stations that could
impact impairment decisions. As cold-water taxa are lost from North Carolina biomonitoring
stations due to warming temperatures, and possibly also to related decreasing flows especially
during the summer, the percentage of stations that are characterized as excellent or good
decreases (see Figure 5-36). The net effect, even with only 50% loss of cold-water taxa, is that
the average condition of reference stations has drifted down the condition scale to be closer to
test stations. The implications of this reference station condition drift is that in reference-based
comparisons used to judge the status of test locations, the condition of test locations will be more
similar to and, therefore, more difficult to differentiate from reference conditions. From this, it
should be less likely to characterize a test location as being impaired, and more difficult to
recognize sources of impairment.
This analysis was based on our study findings showing that cold-water taxa decrease, and
also that warm-water taxa increase over time and/or with increasing temperature or decreasing
flow or precipitation, at some, though not all, long-term biomonitoring stations. We
acknowledge that this basic finding was not universal, and so this threat to reference status may
be more important to consider in more vulnerable regions where the benthic communities are
composed of a greater proportion of cold-water taxa. The study supports the inference that
temperature-preference taxa can be expected to respond as climate changes progress in the
future, because when the responses of temperature preference trait groups were observed, they
were consistent with expectations based on the direction and magnitude of temperature or
flow/precipitation changes that occurred at the stations tested, they occurred at locations that
based on elevation, stream size, and/or ecoregion were composed of sufficient cold-water taxa
for responses to be testable, and were not limited in trend detection by shorter data durations.
We also acknowledge that a 50 to 100% loss/replacement of cold-water taxa may be more
extreme than what will occur at existing biomonitoring sites in the near term. This approach is
only intended to show the direction and extent of alterations that can be expected from the types
of biological responses occurring as a result of climate change effects.
Given this expected effect of climate change in altering reference baseline conditions and
its implications to reference-based comparisons, it would be valuable to be able to characterize
reference conditions on a more complete and objective condition scale than is represented by the
7-38
-------
"impaired/not impaired" decision approach. The BCG (Davies and Jackson, 2006) captures the
full range of biological conditions, from natural/undisturbed to completely impaired. The more
numerous, subtle and well-defined levels captured in the BCG delineate a meaningful and scaled
framework for characterizing existing reference conditions, and within which changes in
reference condition attributable to climate change could be judged. A BCG would allow
reference stations to be more accurately characterized, would support evaluation of reference
station condition or drift over time, and would similarly support characterization of nonreference
station changes over time.
7.4.2. Synergistic Effects between Climate Change and Land Use
Though slightly different in geographic scale, both climate and land-use change can be
considered large-scale impacts (Hamilton et al., 2010a). Global climate change drivers are well
described (TPCC, 2007c). Land-use change is generally considered a landscape-scale stressor,
but is driven by global population growth (Nakicenovic and Swart, 2000). Land-use changes,
such as urban/suburban land development, have encroached on and impaired reference stations
across the United States. However, documentation of such problems has been sparse and likely
has been handled on a local, case-by-case basis.
The successful use of biomonitoring data for evaluating pollution impairment in the
context of climate change is in part related to understanding synergistic effects between climate
change and conventional stressors, and how they can be separated. These synergistic effects can
impact approaches used for attributing causes through the stressor identification process (see
U.S. EPA, 2000). Synergistic effects between climate change and other stressors are
increasingly documented (Clement et al., 2008; Collier, 2008; Kaushal et al., 2008).
We examined the relative responses to climate change compared to land-use change
(urbanization) through analyses of existing biomonitoring data. Hydrologic response variables
play important roles in defining habitat conditions and structuring aquatic communities (e.g.,
Poff et al., 1997) and are responsive to both climate change and urbanization.
Flow data from USGS gages in the Baltimore-Washington, DC area (Mid-Atlantic
region) were used in this case study. The main question that was addressed was how hydrologic
response to climatic change in the Mid-Atlantic would compare with land use impacts. Data
preparation involved gathering historical flow and precipitation data for urban and forested sites,
7-39
-------
calculating Baker's Flashiness Index (Baker et al., 2004) and IHA parameters for these sites, and
identifying which historical years of data had conditions that most resembled those that are
projected to occur in the future. Data were analyzed using ANOVA analyses.
ANOVA results are shown for one example high flow metrics in Figure 7-4 and for one
low flow metric in Figure 7-5. Tables 7-11 and 7-12 summarize complete results. All plots of
the ANOVA results for the IHA parameters are available on request. Results show differences in
the types of hydrologic variables (IHA, sensu Richter et al., 1996) that are likely to be most
responsive to either climate change or urbanization effects. High flow metrics, such as
flashiness, high-pulse-count duration, 1-day maximum flow, and others, tend to strongly reflect
urbanization, swamping inputs from climate change effects. In comparison, several low-flow
metrics, such as 1-, 3-, and 7-day minimum flows and low-pulse count, show responses to
climate change effects more so than to land use (see Table 7-12). Where future climate change
effects are small compared to land use, expectations are for more frequent, shorter, higher flows
in urban-affected streams. Where future climate change effects are large compared to land-use
effects, expectations are for more frequent, longer, lower flows. Accordingly, low-flow
parameters should be selected as sensitive climate change indicators, and low-flow effects on
biota are correspondingly expected to be most influential.
We further evaluated the relative effects of climate change and urbanization on stream
condition through benthic invertebrate responses, using the sampling results from the Piedmont
regions of North Carolina as a test case. The study area has undergone rapid population growth
and urbanization since 1945, which has contributed to flashier streams and altered habitat. Data
preparation for the study involved developing OTUs, calculating taxa richness-based metrics,
calculating IHA parameters (Richter et al., 1996) and Baker's Flashiness Index (Baker et al.,
2004) for 67 biological sampling sites that were associated with USGS gage stations, and
dividing the sites into natural, urban, agricultural and other land use categories based on
examination of the watersheds in Google Earth.
The main objective of this study was to assess the response of macroinvertebrates in
urban and nonurban streams to hydrologic changes. We used number of EPT taxa as the
principal response metric and flashiness (the sum of daily flow changes divided by total flow),
low pulse count (number of events per year where flow is below the 25* percentile), and 1-day
minimum flow as the hydrologic indicators. Flashiness is predicted to increase with urbanization
7-40
-------
4.5
4.0
3.5
S30
«
2.5
2.0
1.5
1.0
Forest
Urban
LU
Climate
_ NORM
ESE Climate
FREQ
Figure 7-4. ANOVA results for high-pulse duration (days) at forested and
urban sites.
7-41
-------
0.20
0.18
0.16
0.14
0.12
0.10
0.08
0.06
0.04
Forest
Urban
LU
Climate
_NORM
J3E Climate
FREQ
Figure 7-5. ANOVA results for 7-day minimum flow (standardized by mean
annual flow) at forested and urban sites.
Table 7-11. Summary of ANOVA results for high-flow IHA metrics. Land
use effects are greater than climate effects for most high-flow metrics tested
High flow metrics
Flashiness
High-pulse count/duration
1-day maximum
3 or 7-day maximum
Rise rate/fall rate
Reversals
High flood peak/frequency/duration
Small flood peak/duration
Land use
Y
Y
Y
N
Y
Y
Y
Y
Climate
N
N
N
N
N
N
N
N
7-42
-------
Table 7-12. Summary of ANOVA results for low-flow IHA metrics. Climate
effects are greater than land use effects for most low-flow metrics tested
High flow metrics
Low pulse count
Low pulse duration
1, 3, or 7-day minimum
Extreme low peak
Extreme low frequency/duration
Land use
Y
Y
N
N
Y
Climate
Y
N
Y
N
Y
but not with climate change, while low-pulse count and 1-day minimum flow are predicted to
increase with climate change.
EPT taxa respond to both high-flow metrics (flashiness) and to low-flow metrics. For
example, extreme increases in frequency of low-flow pulses (>20/y) are associated with EPT
taxa loss (see Figure 7-6), though low-pulse count did not differ much between the natural and
urban streams in this analysis. There was a strong association of decreasing richness of EPT taxa
with increasing flashiness (see Figure 7-7), as well as confirmation of the greater flashiness of
urban streams. The flashiest urban streams had poorer conditions than the moderately flashy
urban streams. In the plots, it appears that there may be a possible threshold at 0.5 (sites that had
flashiness values of less the 0.5 generally showed no relationship, while sites with flashiness
values greater than 0.5 generally showed strong relationships).
Natural and urban streams did not differ greatly in low-pulse count, although the Smith
River is an important exception. This site is dominated by natural land cover but has extremely
high low-pulse counts (28-44 per year) because it is regulated by a peaking hydropower dam.
Overall results show that there was not a strong relationship between low-pulse count and
number of EPT taxa (see Figure 7-6). Low-pulse count was most strongly associated with EPT
taxa loss when there was an extreme increase in frequency of low pulses (>20 per year).
In this component of the study, urban conditions were compared with natural stream
conditions, and the urban streams had lower 1-day minimum flows than natural streams (see
Figure 7-8). However, within the urban sites, there was no association between number of EPT
taxa and minimum flow. There is an apparent threshold response below minimum flows of
7-43
-------
50
40
CO 30
X
CO
1—
Q_
LU 20
10
0
All Streams
'<
•
I
* ,a
1 * * A
* ***g\» *
A /pr\ *
1 ^|£^. *
i
* &^^^A /
• " v •
^A
.
* *
A *
A
.
'
,
•
•
:
•
• m J m m
0 10 20 30 40 5
• Natural
A Agriculture
3 • Urban
Low pulse count + other
Figure 7-6. Relationship between richness of EPT taxa and low-pulse count
of the stream for stream types in the North Carolina Piedmont.
7-44
-------
oc
(D <*u
X
CO
1— 25
£
LU 20
15
10
5
0
All Streams
'
•
•
/«*•
%** ". « *
A\ A ^( • .A
A A ZO ^ ••' . I
A Zl A • • A A
.*- v- ~3>
£&OS 0AA A 4E * *
A ^ * • •
^VA A-$^t A ^ /»V\. AA •
• u
0 0.2 0.4 0.6 0.8 1
Flashiness
A
'
0 1
•
2 1.
• Natural
A Agriculture
• Urban
* Other
Figure 1-1. Relationship between richness of EPT taxa and flashiness
(Baker's index) of the stream for stream types in the North Carolina
Piedmont.
7-45
-------
40
co 30
x
CO
Q_
LJJ
10
0
All sites
• **
>
A
A
0.0 0.1 0.2 0.3 0.4 0.5
1-day minimum flow
0.6
0.7
• Natural
A Agriculture
• Urban
4 Other
Figure 7-8. Relationship between richness of EPT taxa and 1-day minimum
flow of the stream for stream types in the North Carolina Piedmont.
about 15% in natural streams, where richness of EPT taxa is lower and less variable compared to
higher flows, but this is confounded by the association of minimum flows with flashiness.
There were several conclusions that were drawn from this study, and also several
questions that remained unanswered. We are aware the flow regime is a causal link that changes
habitat, but we are uncertain as to whether or not it is a direct stressor. In this study,
intermediate-term changes in flow were not associated with taxa change within streams, but this
analysis has low power. The biological responses that are seen indicate that natural stream
communities are highly resilient within the range of natural hydrologic variability. Because of
this resilience, effects from hydrologic changes associated with climate change are unlikely
unless these changes are truly extreme, such as those that occurred in the regulated river in this
study. Future climatic changes are likely to be beyond the variability observed in the recent past.
7-46
-------
Therefore, we have not seen anything as extreme as is predicted to occur, and this makes it
difficult to predict future impacts. These results suggest that natural streams are more resilient to
hydrologic changes within the range of recent past climate. Large changes in minimum or low
flows may take much longer to become biologically meaningful, and in the shorter term,
temperature effects may be more important.
7.4.3. Future Vulnerabilities of Reference Stations to Land Use
References stations are vulnerable to human-induced changes to the surrounding
landscape. We evaluated current and future vulnerabilities of existing reference stations to
urban/suburban development for three study states (Maine, Utah, and North Carolina), as well as
for Florida as a case study representing a high level of population growth. Data on current and
future land uses comes from the Integrated Climate and Land Use Scenarios (ICLUS) project
(Bierwagen et al., 2010). Future land-use scenarios are consistent with the IPCC Special Report
on Emissions Scenarios social, economic, and demographic storylines used in global climate
models (U.S. EPA 2009; Nakicenovic and Swart, 2000). The ICLUS scenarios consider
different levels of population growth, with different assumptions about development patterns
(U.S. EPA 2009). The two most extreme scenarios are A2, which has high population growth
rates and business-as-usual development patterns and Bl, which has low population growth rates
and compact development patterns. The base case uses medium growth and migration rates,
along with a business-as-usual development pattern. We used a total of 248 reference sites
compiled from Maine, Utah, and North Carolina to examine their vulnerability to current and
future land use. The number and distribution of reference stations for these states are discussed
in Sections 3, 4, and 5 of this report. Florida DEP has about 308 sampling locations, with
58 reference sites designated as "exceptional" (see Figure 7-9).
Urbanization affects stream conditions through alterations in hydrology and
geomorphology, with typically increased loading of nutrients, metals, pesticides, and other
contaminants; these effects are associated with increases in impervious surface (Paul and
Meyers, 2001). To estimate the degree of urbanization representing a threshold of impairment
for the Florida case study, the relationship between human population density and
Ephemeroptera (mayfly) taxon richness developed from analyses in New England were used (see
Figure 7-10) (Snook et al., 2007). At low population densities, up to approximately 50 persons
7-47
-------
(O t/) rt
— c m
^£ C fB
O "Z C
Legend
Sampling Locations (FL DEP)
RATING
o 0- Unrated
O 1 -Exceptional
• 2 -Healthy
A 3 -Impaired
|^^| Commercial /Industrial
^^| >1 0 units/acre
|| || 5-9.9 units/acre
ji J 2-4 .9 units/acre
|| || 0.5-1 .6 acreAjnit
| | 1.7-4 .9 acreAjnit
20-39 .9 acre Ainit
40-79.9 acreAinit
80-1 59.9 acreAjnit
>1 60 acreAjnit
Private undeveloped
Figure 7-9. Florida's biomonitoring sampling stations, including "exceptional" reference locations (light green
dots), shown in relation to current land use.
-------
New England
35
05
X
05
30
25
0s
05
(D
"a.
O
CD
E
CD
.C
Q.
LJJ
20
15
10
0
0.5
5.0
50.0
500.0
Population Density, /mi:
CT
Rl
VT
ME
NH
MA
Figure 7-10. 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).
(-25 houses) per square mile, there are few detectable biological responses. From
50-500 people (25-250 houses) per square mile corresponds to a degradation gradient, and
above 500 people (250 houses) per square mile, New England streams are degraded. Therefore,
a threshold of housing density >25 houses per square mile was selected to indicate potential
degradation. Using the land use composition within a 1-km (0.62-mi) radius buffer around each
reference station, vulnerability was defined as >20% of the buffer with a land use at or above the
threshold of housing density.
For the analysis conducted for Maine, Utah, and North Carolina, urban and suburban
(>0.6 units/acre, or about 384 per square mile) was used. However, a threshold of 10% of
development within a 1-km buffer was used to reflect expectations for impacts to the biological
7-49
-------
communities from urbanization (Schueler, 1994; Booth and Jackson, 1997; Wang et al., 2001).
These differences in thresholds may account for some of the differences in results between the
evaluation of the three study state reference stations and the Florida case study. Given the low
threshold of development used and the high population growth rates for Florida, we take the
Florida results to represent a worst-case scenario.
Among the 58 "except!onal"-grade reference stations in Florida under year 2000
conditions, 19% of the stations can be classified as vulnerable to land-use impacts (see
Table 7-13). That is, nearly 1/5 of Florida reference stations may already exhibit impacts from
urbanization. Within the next 2 decades, more than one third of existing reference stations will
be vulnerable, and by 2100, nearly half of current reference stations may be impacted by
urbanization under the base case and A2 scenarios. This level of vulnerability is significant. The
spatial distribution of this vulnerability is broad. In Florida, most sampling stations are in the
northern half of the state. Future projections of urbanization generally follow current patterns of
development, with particularly dense future development projected for the northern half of the
Florida peninsula. The only reference locations that appear to be protected from future land
development are those largely surrounded by water, and/or those within government-owned or
protected lands that cannot be developed. In Florida, this represents about 17% of existing
reference locations.
The results for Maine, North Carolina, and Utah show a somewhat lesser degree of
vulnerability. Under current (2000) conditions, 22% reference locations in these three states
have greater than 10% urban/suburban densities within a 1-km2 neighborhood (see Table 7-14).
Under the worst case (A2) scenario, future housing development increased to 34% by 2100. The
r\
maximum amount of suburban and urban development within the 1-km neighborhood in 2000
was 58%; this increased to 99% by 2050. The average amount of development increased from
22% in 2000 to 28% in 2050 and 34% in 2100 using the A2 scenario, while it leveled off at 26%
using a lower population growth and higher development density scenario (Bl) (see Table 7-14).
The results for Utah are difficult to interpret, and the projections not very meaningful, as the
number of reference sites falling within the 10% development threshold as calculated for a 1-km2
neighborhood was very small.
7-50
-------
Table 7-13. Percentage of existing Florida reference stations (« = 58,
classified as "exceptional"), that have >20% developed land use (with
25 houses per square mile (9.65 houses per square kilometer) or more,
Categories 5-12 in the ICLUS data set) within a 1-km buffer surrounding the
station, for current and decadal time periods through 2100
Year
2000
2010
2020
2030
2040
2050
2060
2070
2080
2090
2100
Scenario
BC (%)
19.0
36.2
36.2
37.9
41.4
44.8
44.8
44.8
44.8
44.8
44.8
A2 (%)
19.0
34.5
36.2
37.9
39.7
44.8
44.8
44.8
44.8
44.8
48.3
Bl (%)
19.0
36.2
36.2
36.2
36.2
36.2
36.2
36.2
36.2
36.2
36.2
Table 7-14. Percentage urban and suburban development within a 1-km
area surrounding reference sites, for all sites and for sites at or above the
impact threshold of 10%. Number of sites is shown in parentheses.
Scenario A2 has high population growth and a business-as-usual
development pattern; Scenario Bl has low population growth and a compact
development pattern (U.S. EPA, 2009)
Mean of
reference sites
threshold)
Area
Combined
Maine
North Carolina
Utah
2000
22% (35)
23% (26)
20% (9)
0% (0)
A2 2050
28% (37)
24% (26)
27% (9)
87% (2)
A2 2100
34% (45)
30% (32)
40% (10)
64% (3)
Bl 2050
26% (37)
23% (26)
24% (9)
77% (2)
Bl 2100
26% (37)
23% (26)
24% (9)
77% (2)
The specific patterns of reference station distribution and vulnerability to land
development will vary among states, although there are widely applicable lessons from these
results. The high level of current vulnerability to urbanization (about 20% in all states tested
7-51
-------
except Utah) highlights the difficulties in siting reference locations in many areas and the
probability of encountering substantial existing urban influences, which impact baseline
reference conditions. This evidence suggests that protection of reference stations is of
substantial importance. Options for protection may differ regionally and include zoning changes,
limitations to development within buffer zones of selected stream reaches, incorporation into
land protection programs (U.S. EPA, 2011), or other sociological, economic, and/or political
solutions. If alternatives for protecting reference locations are limited or costly, it may be that
reference stations in already protected areas, such as national parks, other government lands, or
in otherwise inaccessible areas may represent the only "protected" references. This is likely to
leave many watersheds and regional ecotypes without good reference conditions for comparison.
In Florida, this would reduce the ratio of reference sites to total sampling sites from 19 to 3%. If
reference sites are too scarce, they will be unrepresentative.
The need to protect reference locations is an important issue for the future of
bioassessment. If reference stations become urbanized, the ability to detect climate change, and
separate climate responses from conventional stressors in order to continue to manage resources,
set permit limits, and meet CWA requires, may be hampered. It may become important to
consider and promote more broad-based alternatives than just local or state-specific protections,
such as regional cooperation in the establishment and monitoring of long-term fixed "sentinel"
locations.
7.5. IMPLICATIONS TO MULTIMETRIC INDICES, PREDICTIVE MODELS, AND
IMPAIRMENT/LISTING DECISIONS
7.5.1. Conclusions Across Pilot Study States
Among the four states evaluated in this study, three of them—Maine, North Carolina, and
Ohio—use some form of MMI. Utah uses a predictive model, RTVPACS, for assessing
wadeable streams. These states are representative of major regions of the United States,
encompassing large-scale variations in climate, climate change projections, geography,
topography, geology, and hydrology. State-specific analysis results also inform a regional view
of climate change implications to commonly used MMIs and predictive models.
MMIs are generally structured as a composite of biological metrics selected to capture
ecologically important community structural or functional characteristics and have been applied
to fish and benthic macroinvertebrate communities (Norris and Barbour, 2009; Bohmer et al.,
7-52
-------
2004; Sandin and Johnson, 2000; Barbour et al, 1995; Yoder and Rankin, 1995; DeShon, 1995;
Karr, 1991). Component metrics are selected based on their responsiveness to the environmental
impacts most often evaluated. Results of this study suggest that climate changes in temperature
and flow conditions can elicit responses in these commonly use metrics through their
temperature preference traits, and potentially through their flow preferences as well, in ways that
can influence the outcomes of MMI condition scores. This means that the commonly used
biological indicators of environmental condition are not only linked to the conventional stressors
usually evaluated, but also to changing climate variables. At least in the most vulnerable regions
(e.g., higher elevations, ecoregions composed of a high proportion of cold-water taxa, smaller
watersheds or stream sizes), the scoring of station condition that relies on MMIs can be altered
by climate change effects. The importance of this finding is that the scoring of stations
according to their apparent biological status becomes the basis for impairment decisions and
associate management actions.
There is much variation among states and tribes in the particular components included in
MMIs or predictive models, because, as a rule, they are calibrated to the state, or more often, to
regions within a state to account for predictable (natural) variability (Barbour and Gerritsen,
2006). Added to this index variability is the regional variability in both climate change
projections and associated biological responses. These sources of variability make
generalizations about the implications of climate change for bioassessment indices challenging.
However, there are some commonalities among states, such as the categories of metrics used,
which we use to investigate vulnerabilities of these approaches to climate change.
There are a variety of regional differences in biological responses evident from this study.
More and stronger trends and responses were found in Utah, largely related to temperature
changes. Fewer significant trends were found in North Carolina, and more were related to
precipitation or flow (see also Section 7.2). There is much spatial variation in these patterns, in
part due to ecoregional, geographic, and climatological variations, and in part attributable to
limitations of the available data. The results point to several conclusions. One is the importance
of categorizing taxa based on ecological traits, especially temperature sensitivities, in order to
evaluate responses to climate change variables and to estimate future vulnerabilities to climate
change. It is a relatively consistent finding that biological metrics and indices used by states and
tribes are either composites of cold and warm-water taxa, or are dominated by one or the other.
7-53
-------
This composition defines the nature of responses and, therefore, the vulnerability of the metric or
index to climate change effects. The richness of cold-water taxa is a metric that was often
responsive, especially at higher elevations, where high-elevation communities tend to have more
cold-water taxa. Metrics using cold-water taxa will help identify climate change "sensitive" or
vulnerable areas. Such information would assist in detecting climate change effects and in
identifying sites to monitor these changes.
For example, in Maine, several EPT metrics (e.g., EPT richness, Plecoptera abundance
and richness, Ephemeroptera abundance and relative abundance) are incorporated into their
linear discriminant model. We have found these taxonomic metrics are composed of varying
combinations of both cold and warm-water taxa, and in relationship to this, predicting their
responses to increasing temperatures and changing hydrologic regimes resulting from climate
change is complex. As summarized in Section 4.8, due to the greater prevalence of warm-water
taxa at the low-elevation long term site evaluated, increases in abundance of warm-water EPT
taxa could results in increasing values of some of these metrics, while losses of the cold-water
EPT taxa could reduce the abundance or richness of other EPT metrics. In addition, taxa
replacements could have variable results. An additional factor is that not all of the EPT metrics
that are components of the linear discriminant model have a simple linear relationship with site
class condition. For example, Ephemeroptera abundances increase initially as station condition
degrades from Class A to B, and then declines again with further reduction in station condition
status. Through this mechanism, increases or decreases in EPT taxa through temperature or flow
preferences could have either positive or negative effects on the final station condition decision.
Another example is that 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.
Predictive models use regional reference conditions to develop relationships between
environmental predictor variables and macroinvertebrate taxon occurrence from which
predictions for an "expected" (E) community are based. A commonly applied model for
macroinvertebrate communities is RIVPACS (Wright, 2000). An important assumption is that
7-54
-------
the predictor variables are minimally affected by human disturbance and are relatively invariant
over an ecologically relevant time (USU, 2009; Tetra Tech, 2008; Hawkins et al., 2000; Wright,
2000; Wright et al., 1984). The E community is then compared to various "observed" (O)
communities at nonreference locations. A basis for comparison is that any differences between
O and E communities reflect biological responses to the range of environmental pollutants or
alterations that are intended to be evaluated. This is similar to the MMI approach.
In Utah, evidence of responses of temperature trait and taxonomic groups to temperature
increases, and to a lesser extent with changes in precipitation, was somewhat stronger and more
widespread (though not consistent at all stations). In particular, responsive groups that should be
tracked in the future include total taxa, EPT and EPT-related metrics, and thermal preference
metrics. But our examination of corresponding impacts to the Utah RIVPACS model responses
showed minimal changes in O/E ratios (see Section 3.8), suggesting that predictive models used
by states may be more resilient to climate change than MMIs. This is in part because they
incorporate long-term (e.g., 30-year) averages of environmental predictor variables, including
climate parameters.
On the other hand, the Utah results on trends in biological trait and taxa groups can also
reflect on potential vulnerabilities of MMIs used by other southwestern states. The EPT trends at
some (though not all) higher elevation stations in Utah indicate fairly predictable losses of EPT
taxa over time in response to increasing temperatures. These losses are in the magnitude of up to
a 25% loss of EPT taxa with current scenarios of temperature increases by 2050, attributable to
the loss of cold-water EPT taxa components. Changes of this magnitude could result in notable
responses in MMIs. Note that over the long term, it is also possible that increases in warm-water
EPT taxa could result in taxa replacements, as both decreases in cold-water EPT taxa and
increases in warm-water EPT taxa were observed at some stations (see Section 3.5).
In North Carolina, even more than temperature, expected climate changes in flow had
important influences on the biological assemblage, including biological metrics that are used in
the MMI to assess condition status. The very limited long-term data mean that we cannot
conclude that the climate change responses are widespread. However, as examples of the types
of biological response that could be expected in the future with continued climate changes in
flow and temperature, we show that losses of cold-water EPT taxa can lead to changes in
bioclassification scores, with the highest quality stations (those currently classified as excellent
7-55
-------
to good) being the most vulnerable (see Section 5.8). In addition, the linkage between
temperature preferences and tolerance to organic enrichment means that increases in warm-water
taxa, and/or losses of cold taxa can alter HBI scores, and this will also alter bioclassification
scoring. The maximum effect appears to be decreases of one bioclassification level (e.g., from
excellent to good, or from good to fair).
In Ohio, the MMI and the determination of the final station rating are also potentially
vulnerable to climate change because of the positive association between temperature sensitivity
and pollution tolerance. Percentage 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 potential vulnerability of the Ohio MMI
through the relative contribution of cold-water taxa within these groups, with the most plausible
expectation being for a decline in bioassessment scores due to losses of sensitive taxa and/or
increases in tolerant taxa. However in Ohio, the biological condition of reference sites has
improved over the last 30 years (see Section 6.5). Climate change effects may be a contributing
component to these observed trends, or may be decreasing the magnitude of the positive
response. However, there is evidence that the trends have been driven largely by other
environmental factors, and in particular, management efforts that have reduced pollutant
loadings and better agricultural practices (see Section 6.8). Thus, in Ohio, the relative
vulnerability of the bioassessment process and the MMI in particular is difficult to assess, as are
approaches that could be applied to adapt metrics to assist in tracking climate change and
partition its effects from other sources.
Overall, the vulnerabilities of MMIs used to estimate bioassessment station condition
scores appear directly related to responses of thermal preference trait groups to both temperature
and flow changes. Responses mediated through hydrologic preference traits may be equally
important, but due especially to limited availability of associated flow and biological data, we
were unable to sufficiently develop hydrologic indicator groups to examine these responses. In
addition, MMIs appear indirectly vulnerable to climate change influences through the
correspondence between the general biological sensitivities to pollution and temperature
preferences. Because many metrics commonly used in MMIs can be comprised predominantly
7-56
-------
of cold or warm-water taxa, or of both, the changes in these metrics alter MMIs through shifts in
the proportion of cold to warm water-preference taxa.
Another widespread and related finding is the moderate but significant relationship
between temperature sensitivity and sensitivity to organic pollution. Metrics selected because
the composite taxa were considered to be generally sensitive, such as EPT taxa, or generally
tolerant, such as Diptera taxa, or to represent responses to conventional pollutants (e.g., organic
pollution as in the HBI), also have demonstrable sensitivities to climate-related changes in
temperature and flow conditions. We have shown these sensitivities to be related, at least in part,
to the predominance of cold and/or warm-water taxa at a location. Assemblage composition by
cold and warm-water taxa may be related to ecoregion, latitude, watershed size, and/or stream
order, and is also clearly affected by elevation. This association between temperature and
pollution sensitivities will affect how indices are interpreted with regard to the conventional
stressors for which the indices were originally developed.
From more limited evidence, it also appears that the ability to categorize taxa according
to flow preferences and requirements could be useful. However, there are generally fewer data
available for this analysis. We augmented the approach of grouping taxa by traits responsive to
one climate variable (temperature) through consideration of a suite of traits. This was useful in
some cases, though it produced fewer significant results. This was probably due to the fact that
fewer taxa were included when categorized by a suite of several traits, resulting in more limited
and/or more variable data and smaller sample sizes with which to test responses. Still, this is
potentially a useful approach to apply as more data become available.
7.5.2. Recommendations for Modifying Metrics
In general, biological metrics (indicators) are selected for their diagnostic value
(Verdonschot and Moog, 2006). However, the effects of global climate changes in temperature
and precipitation on biological metrics have, until now, been largely untested, because climate
change was not considered a "stressor of concern" until recently (Hamilton et al., 2010a). Given
our demonstrations of the vulnerabilities of traditional metrics to climate change, and associated
impacts to the classification of station conditions, it is important that state and tribal
biomonitoring programs consider adopting modified metrics with the purpose of tracking
climate-associated changes in MMI outputs (Hamilton et al., 2010b). This will support making
7-57
-------
inferences about causes, helping differentiate climate change from other stressors as part of a
weight-of-evidence evaluation. It will allow resource managers to more effectively make
management and regulatory decisions on the basis of biomonitoring results in the face of climate
change impacts (Hamilton et al., 2010a).
The focus here is on the relative contribution of cold- and warm-water ecological trait
groups to the composition of traditional metrics. The general recommendation is that cold and
warm water components of traditional metrics be documented and tracked separately. A
recommended approach for incorporating modified metrics into a biomonitoring data analysis
regime is to continue calculating the traditional metric (e.g., EPT richness, HBI), while adding
new cold and warm water metrics. Proportional changes in cold and warm-water taxa would
provide a basis for estimating how much of the change in the traditional metric can be accounted
for by changes in temperature trait groups. This provides a basis for comparing potential climate
change effects to those of other stressors in a weight-of-evidence assessment. Comparisons
could be made over time and among locations or groups of sites (both reference and
nonreference). An option for tracking climate-related changes is to put traditional and modified
metrics on the same plot and compare their trends over time (i.e., Figure 7-11). Another option
that requires further testing is to track the ratio of the cold- or warm-modified metric to the
traditional metric. For example, separate tracking of cold-to-total EPT and warm-to-total EPT
richness metrics was able to account for trends in total EPT richness over time in circumstances
where changes in total EPT richness were caused by losses of cold-water taxa, and where
changes include both losses of cold-water taxa plus gains of warm-water taxa (i.e., taxon
replacements) (Hamilton et al., 2010b).
We examined evidence in this study for the value of adopting temperature-modified
metrics for diversity and total taxa richness metrics; for EPT-related metrics; and for pollution
tolerance metrics, such as the HBI or related indices. However, the principle of partitioning
metrics to separate component taxa based on cold or warm water should be considered for other
biological metrics (Hamilton et al., 201 Ob). These could include trait metrics related to
functional feeding groups (e.g., predators, collector-filterers) or life history habits (e.g.,
swimmers, climbers). Such metric modification should be considered on a state or
region-specific basis, in particular for climate-vulnerable regions (e.g., high elevations,
low-order streams, small watersheds). In addition, an OCH taxa metric may be valuable to track
7-58
-------
taxa that are robust to warmer conditions and/or more intermittent flows. This may be especially
valuable in regions at lower elevations, where temperature increases may be large, and/or where
summer flow conditions are likely to be especially vulnerable to climate change effects.
35
30
25
20
15
10
5
0
-5
Q 0 D Q-
D u D " D
Total Taxa
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 ^^ C0kj Water Taxa
Year ^K Warm Water Taxa
Figure 7-11. 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).
We cannot yet make strong suggestions for metrics related to hydrologic sensitivity, in
part because the lack of flow data corresponding to biological collections has limited ability to
calculate flow metric preferences by taxon. However, hydrology-related trait characterizations
can be based on known life history traits coupled with regional observations and literature
information, as with the intermittent taxa metric used in North Carolina. A metric that accounts
for tolerance to intermittent flows, requirement for perennial flows, or some similar
hydrologic-preference metric, may become valuable as changes in flow conditions are more
evident. Such a metric would have to be calibrated by region.
Calculation of modified metrics for incorporation into biomonitoring data evaluation will
require designation of cold- and warm-water ecological trait groups. Cold and warm-water taxa
lists must be developed on a state- or region-specific basis, which is a substantial undertaking.
The efforts initiated in this study, including the process of applying weighted average or
7-59
-------
maximum likelihood modeling in concert with literature information and best professional
judgment to estimate temperature preferences by taxon from biomonitoring data, and the
development of a traits database that documents the temperature preferences and tolerance
results calculated for the three states analyzed in this study (see Stamp et al., 2010; U.S. EPA,
2012), can be used as a starting point for future state efforts.
7.6. SENTINEL MONITORING NETWORK
Results of this study have demonstrated the importance of accounting for climate change
effects in order to maintain sound bioassessment decision making. The next step is to consider
possibilities for augmenting existing programs to address this need. Section 7.3 discusses
characteristics of a biomonitoring program and their inherent limitations with regard to detecting
trends that might be associated with climate change. Approaches to address some of those
limitations are discussed here.
A monitoring network designed to detect climate change effects needs to account for
regional variations in numerous factors, including climate, geology (including soils), topography,
elevation, latitude, vegetation, etc. Such conditions often cross state and tribal boundaries.
Therefore, this kind of monitoring network may require collaboration among states and tribes
with regard to technical considerations (e.g., site selection, sampling methods) and funding.
Regional and national coordination will be important to facilitate this process.
Thorough coverage across ecoregions and other environmental variants would require a
large network of sites. A modest initial effort for sentinel site monitoring could focus on highly
vulnerable areas and watershed types. Because not all watersheds or community types would be
represented by such selective establishment of a sentinel site monitoring network, the
classification of conditions and transferability of bioassessment results will be integral for
extrapolation to other areas (e.g., Allan et al., 1997; Gerritsen et al., 2000; Wu and Li, 2006).
In order to separate climate change effects from other stressors, both reference and some
portion of impaired sites should be measured over time. Thus, an ideal network of sentinel sites
would be established along the BCG and be anchored in reference conditions. This would
support an analysis approach in which temporal trends at reference sites could be compared to
temporal trends at impaired sites, in order to differentiate between climate effects and
7-60
-------
conventional stressors, as illustrated in Figure 7-12. Different levels of stressor effects could
also be compared, and synergistic effects could be considered (see Figure 7-12).
05
u
ro
u
'no
_o
o
CO
Ideal reference
sites (MDC)
Stressed
sites
Effect of climate
change
CC attenuates stress
o relative effect
CCincreases stress
Time
Figure 7-12. 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.
It is possible that in a monitoring context, as opposed to a controlled study, synergisms
between climate change and conventional stressor responses could not be fully partitioned.
Inference using literature studies, especially through use of CADDIS and the stressor
identification process (Suter et al., 2002; U.S. EPA, 2000) would contribute to data interpretation
in a weight-of-evidence approach. The efficacy of conducting long-term sampling along the
BCG should be considered through interactions with state and tribal biomonitoring managers,
consideration of avenues of funding support, and finally, through practical evaluation of existing
opportunities for establishing such a sentinel site monitoring network in representative and
vulnerable regions.
If a sentinel site monitoring network along the BCG is infeasible, a less
resource-intensive alternative would be to establish long-term sentinel sites only at high-quality
reference locations. Lack of trend data from nonreference sentinel locations would present some
limitations to separating climate change from other stressors responses. Selection of such
7-61
-------
locations would face some of the same difficulties as any reference selection effort conducted by
individual states. However, the larger spatial scale and regional perspective necessary for
implementation would offer opportunities to search for and select least-affected locations from a
larger area and share results across jurisdictional boundaries.
While typical bioassessment approaches include sampling watersheds on a (typically)
5-year rotating basis, biomonitoring at sentinel sites should be considered on a regular, repeating
basis, annually if possible. With less frequent data, temporal variations from interannual and
cyclic climatic sources would greatly extend the time frame needed to describe climate change
responses.
Another consideration for sentinel site monitoring for climate change is the inclusion of
continued monitoring at targeted locations, even if initial site selection is probability-based,
rather than only application of a probability-based sampling approach in which all sites are
reselected each year. Probability sampling has important strengths in capturing the (often large)
range of variability within a defined stratum, such as low-order stream reaches (Barbour and
Gerritsen, 2006; Hughes et al., 2000). It also provides valuable data about the status of our
nation's waters at any given time (Hughes et al., 2000; Paulsen et al., 1998). This is important
for defining the range of conditions within the stratum at any one time, but it requires replication
(multiple reference sites) within the stratum. Reference conditions are often established based on
a population of reference locations that together reflect the range of natural variability for a
region (Barbour and Gerritsen, 2006). Combining reference stations across major physiographic,
geomorphic or climatological regions inflates the range of measured variation in biological
parameters from predictable, natural sources (Barbour and Gerritsen, 2006). It is, thus, important
to account for predictable, natural sources of variation. This will affect how many reference
stations within a defined area must be sampled, how frequently they must be sampled, and the
sampling duration needed to have the power to detect climate change response trends. In the
current study, groups of reference stations analyzed were typically not of sufficient duration to
define statistically significant trends within the context of natural spatial and interannual
variation.
We found high among-site variability within ecoregions despite the expectation that
partitioning by ecoregion should control major predictable sources of variation. This maximizes
the effects of "natural" site (spatial) variability on the detection of temporal trends and greatly
7-62
-------
extends the time it will take to discern climate change effects. This suggests a trade-off between
gaining knowledge about regional status and knowledge about long-term trends.
Climate change trends observed from single fixed locations may not be transferable to
corresponding regions or strata because they do not account for the real range of conditions that
defines the stratum. However, replication of targeted locations within a region or stratum would
account for natural spatial variability. Combining some fixed with random sites in a
predetermined sampling pattern may be the most likely design that accomplishes both trend
detection and representation (Urquhart et al., 1998). One observation that stands out regarding
the Maine, North Carolina, and Utah reference locations is that most of these have more frequent
annual sampling than would be the case if they were only sampled on a "rotating basin" basis.
Utah adopted a rotating basin sampling scheme as well as a probability-based station selection
approach within the last decade (Utah DEQ, 2006). However, they maintain regular annual
sampling at a small number of fixed locations with long-term historic records. Whether by
formal decision or historic happenstance, some other states also have regularly sampled stations
outside of rotating and/or probabilistic designs.
The selection of an index period will also be affected by climate change. Projected
climate changes are likely to impact seasonal patterns through changes in flow conditions as well
as in temperature regimes. These will influence a variety of biological processes, including rates
of development, timing of emergence, and other components of reproduction (Seebens et al.,
2009; Harper and Pecarsky, 2006; Poff et al., 2002; Vannote and Sweeney, 1980). This may
have several ramifications to biomonitoring designs. If samples are collected at a fixed time
during the year, then in the future, sampling may yield lower abundances of some species,
different species composition, or different relative abundances. This impacts temporal
comparisons. Also, spatial comparisons may now be based on communities of more limited
seasonal diversity. More extreme or extended summer low flows may, over the long term,
become an impediment to sampling for states that use summer or fall index periods. This may be
a particular concern in perennial streams vulnerable to a shift to intermittent conditions in the
future. Biological responses to reductions in flow can represent legitimate responses to climate
change. However, the eventual inability to sample during a late-season index period in some
stream locations must be considered and planned for. Though highly unlikely due to resource
limitations, sampling more than once per year, including once during the spring/high flow index
7-63
-------
period, could provide valuable information on components of the benthic community that
emerge early in summer.
Many different groups are considering, or have already started, monitoring for climate
change effects. If possible, collaboration among at least some groups, particularly among
bordering states, would have many potential benefits. Some duplication of effort could be
avoided, results could be integrated in a more meaningful way, and resources could potentially
be saved. Collaboration would foster consistency across groups in types of data collected, as
well as potential use of a common database. Efforts to discuss and establish a sentinel
monitoring network might facilitate collaboration among existing efforts. A common vision of
sampling and agreement on types of data that could be incorporated into a common database
related to a potential climate change monitoring network could have a better chance of success.
7.7. CLIMATE CHANGE IMPLICATIONS FOR ENVIRONMENTAL MANAGEMENT
The components of bioassessment programs that may be affected by climate change
include assessment design, implementation, and environmental management (see Figure 7-13).
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 assertions
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.
One of the central objectives of state programs for establishing a reference condition baseline
and conducting ongoing biomonitoring at reference and nonreference locations is to detect
locations, or stream reaches, that are sufficiently different from the established baseline to be
considered impaired. The approach and specific criteria used to make impairment decisions are
established by states and tribes, and vary among regions to reflect the appropriate range of
natural variability (Barbour and Gerritsen, 2006). But the assumptions inherent in the almost
universally applied reference comparison approach include that the stressors likely to impair
7-64
-------
streams and rivers within a region are accounted for within the sampling and analysis scheme
applied, and that if a real impairment exists, it can be detected with a reasonable level of
confidence. The concept that all stressors must be accounted for presents an unusual problem
with regard to climate change effects, because climate change effects are "global," so reference
stations are equally at risk. This threatens the reference comparison paradigm.
Bioassessment Program Activities
Change
- Precipitation
Climate
- Temperature
*
Assessment Design
- Select reference sites - Select communities
- Select sampling design - Select indicators
- Determine reference - Create indices,
condition predictive models
{
Implementation
- Collect biotic samples - Collect abiotic data
- Analyze data
{
Environmental Management
- Develop biocriteria - Determine impairment
- Determine water - Develop TMDLs
quality standards
Figure 7-13. 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.
7.7.1. Impairment Listings and Total Maximum Daily Load (TMDL) Development
Results of this study reveal changes in biological indicators and within specific ecological
traits groups that are reasonably attributable to climate change effects and are likely to interfere
with impairment determinations. Trends in cold- and warm-water trait groups result in
corresponding changes in biological metrics used by states, such as EPT taxa richness or
abundance in the HBI index. The observed and projected changes in biological metrics are
7-65
-------
sufficient to downgrade reference station condition. Degradation of reference station condition
is essentially causing references stations to become more similar to nonreference stations, and
diminishes the ability to detect impairment. These findings imply that unless metrics are
modified so that climate effects can be tracked and thresholds for defining impairment
re-evaluated, degraded reference conditions will cause fewer stream reaches to be defined as
impaired, at least in the most climate-vulnerable watersheds. Where this occurs, fewer corrective
actions would be taken, and greater long-term degradation of stream conditions could result (see
also Hamilton et al., 2010a).
When a stream segment is found to be impaired, TMDLs of pollutants are developed by
states, and the cause(s) of the impairment are identified through the stressor identification
process (U.S. EPA, 2000; Suter et al., 2002). In permitting (e.g., the National Pollutant
Discharge Elimination System [NPDES]), discharge limits must be set considering any existing
TMDLs. Beyond the possibility of underprotection with fewer impairment listings and fewer
requirements for TMDLs, there may be other climate change implications to TMDL
development. Climate change scenarios show greater variability in runoff and flow, which may
result in greater uncertainty in loadings expected from nonpoint sources. Critical low flows also
drive TMDLs, and these may become uncertain and more difficult to predict. The identification
of culpable stressors is also complicated by the effects of climate change on biological
indicators.
The main approaches pertinent to preserving the ability to detect impairment include
adopting climate change-related modifications of biological metrics, associated re-evaluation of
impairment thresholds, and reference station classification and protection. These actions are
directed at improving the ability to track effects of climate variables, compare these between
reference and nonreference locations, and, thus, increase the information brought to bear on
differentiating climate change from other stressors and detecting conventional stressor
impairment. The stressor identification process, tailored to include detailed climate change
information, would facilitate partitioning biological responses between climate change and other
stressors.
The paradigm for conventional stressor identification is based on spatial
(reference/nonreference) comparisons, combined with a weight-of-evidence evaluation of
potential causes, augmented by research and other literature-based knowledge of major
7-66
-------
cause-effect expectations (Suter et al., 2002; U.S. EPA, 2000). The need to partition climate
change effects could add a relatively extensive time component to this framework if the process
were to rely primarily on site-specific, long-term field data. However, it is impractical and
undesirable from a decision-maker's point of view to obtain this degree of detailed, long-term
sampling for every case of impairment assessment. From a practical perspective, it also is likely
to be outside of the level of resources available to most states or tribes for routine bioassessment
sampling. An alternative approach includes monitoring a more limited network of sentinel sites
(see Section 7.6). Documentation of trends from monitoring data, other aspects of
weight-of-evidence evaluation of potential causes, and an expanded knowledge database on
biological responses to climate change could be included in an expanded stressor identification
process.
With regard to other vulnerabilities in the TMDL development process, there is a need
for watershed-specific modeling to predict how flow dynamics change with climate, to provide
support for estimating future changes in low flows, and to modify loading calculations and
limitations accordingly.
7.7.2. Impacts on the Development of Water-quality standards and Biocriteria
Biological responses to climate change will likely impact water-quality standards and
biocriteria through shifts in baseline conditions. This study illustrates several avenues through
which climate change is affecting stream communities in ways that have implications for
biocriteria programs. The cascading effects of climate change-related trends in temperature and
precipitation on watershed conditions, water quality, and aquatic biological communities, will
lead to shifting, most often degrading, baseline conditions. Decreases in mean abundances
and/or species richness of cold water or other sensitive taxa and trait groups, increases in warm
water or other tolerant taxa and groups, and also increases in variability of these indicators drive
reference sites to greater similarity with nonreference areas, as well as greater difficulty in
establishing statistical differentiation (see also U.S. EPA, 2008). As a result, reference-based
standards will be liable to progressive underprotection.
Given the types of biological responses observed in this study, climate change can be
expected to alter some uses and their attainability, especially in vulnerable streams or regions.
For example, some cold-water streams could take on cool-water characteristics, with declining
7-67
-------
abundances and/or richness of sensitive cold-water taxa, possible increases in warm-water taxa,
and other changes potentially related to altered hydrologic patterns. Regulated parameters such
as temperature, DO, ammonia, and pH may also be sensitive to climate change effects, and their
values may need to be adjusted relative to revised designated uses.
There are numerous criteria, both biological and chemical, that are addressed in
water-quality standards, and which may be affected by climate change (see Table 7-15).
Biocriteria are of particular interest, as they tie closely to the indices and thresholds used to
determine condition and impairment. The climate-related causes of drifting (degrading) baseline
conditions cannot be directly controlled, but can be assessed, at least to the degree resources
allow. Necessary steps would include documentation of reference conditions, tracking of
changes in reference conditions over time, and to the extent possible, protection of reference
conditions from other encroaching impacts, particularly land-use changes. This may be extended
to include repetitive regional monitoring of sentinel sites, carefully chosen to represent the best
conditions of the most vulnerable regional watersheds. Further efforts to address climate change
impacts on standards would require examination of which water-quality standards are resilient to
climate change impacts and will remain protective, and identification of susceptible standards
that may need adjustment.
For watersheds that are found to be particularly vulnerable to climate change effects and
are characterized by particularly vulnerable trait groups, more refined aquatic life uses should be
considered for application. Refinement of aquatic life uses can be applied to guard against
lowering of water quality protective standards. Uses are designated for a stream segment based
on conditions at similar reference stream segments, using information on habitat characteristic
and associated biological communities, and potentially also consideration of economics and
human-related conditions. Criteria are set to protect designated uses, and often differ between
use levels. Application of refined aquatic uses could provide a greater number of more narrowly
defined categories, which could accommodate potentially "irreversible" changes (e.g., increased
temperatures driven by long-term climate change), but with sufficient scope to maintain
protection, and also support antidegradation from regulated causes.
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
7-68
-------
Table 7-15. Variables addressed in criteria and pathways through which
they may be affected by climate change (from Hamilton et al., 2010a)
Criteria
Pathogens
Sediments
Temperature
Nutrients
Chemical
Biological
Flow
Salinity
pH
Climate change impacts
Increased heavy precipitation and warming water temperatures may require the
evaluation of potential pathogen viability, growth, and migration.
Changing runoff patterns and more intense precipitation events will alter sediment
transport by potentially increasing erosion and runoff.
Warming water temperatures from warming air temperatures may directly threaten the
thermal tolerances of temperature-sensitive aquatic life and result in the emergence of
HABs, invasion of exotic species, and habitat alteration.
Warming temperatures may enhance the deleterious effects of nutrients by decreasing
oxygen levels through eutrophication (hypoxia), intensified stratification, and extended
growing seasons.
Some pollutants (e.g., ammonia) are made more toxic by higher temperatures, and also
by pH, which may be altered as a result of climate change.
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.
Changing flow patterns from altered precipitation regimes are projected to increase
erosion, sediment and nutrient loads, pathogen transport, and stress infrastructure.
Depending on the region, climate change is also projected to change flood patterns
and/or drought and associated habitat disturbance.
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.
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.
7-69
-------
vulnerable to climate change degradation, making application of antidegradation policies in
vulnerable water bodies important. Management approaches and special considerations for
implementation of anti degradation policies may need attention. In addition, the application of
use attainability analyses on vulnerable water bodies may be pertinent for characterizing climate
change effects.
7.8. CONCLUSIONS
Climate change will affect many of the components of bioassessment programs,
including assessment design, implementation, and environmental management. Implementing
the recommendations derived from the results in this study can improve the resilience of
bioassessment programs and ensure that management goals can be met under changing climatic
conditions. These steps can help manage the risks associated with not meeting goals, even
though the magnitude and timing of climate change effects on aquatic resources is uncertain.
There are four main sets of recommendations from this study specific to adaptations of
biomonitoring programs:
1. Multimetric indices should be revised to reflect the sensitivity of taxa and trait groups to
climate change effects; predictive models should also reflect these changes in indicators
and periodically revise the expected community composition used in the analysis. At
present, the most accessible information relates to temperature sensitivities and
preferences; however, sensitivities to changing hydrologic conditions should be pursued
in the future.
2. A monitoring network to detect climate change effects should be set up, at least for the
most climate-vulnerable regions. This network will need to be more comprehensive
spatially and sampled more frequently than current bioassessment sites. Detecting
climate change at these monitoring sites requires that they are protected from other
stressors.
3. Abiotic data needs to be collected more frequently and at more sites; a monitoring
network to detect climate change effects should incorporate abiotic data collection as
well, including water temperature and flow. The value of better water temperature and
flow data is great, and consideration should be given to deploying in situ temperature and
flow meters.
4. TMDLs and water-quality standards should be examined to ensure that these remain
protective of aquatic life uses under changing climatic conditions.
7-70
-------
We have some additional recommendations for further study and collaboration that
would enhance our ability to track climate change effects and separate these from other stressor
responses in the context of biomonitoring:
1. The use of thermal-preference metrics for detecting climate-related trends should be
further explored. Monitoring of thermal-preference metrics will increase the probability
of detecting community responses to warming trends and reduce the likelihood that they
will be obscured by taxonomic variability.
2. The lists of cold and warm-water taxa developed in this study should be refined and
extended to more states and regions. Refinements can be made by using continuous
water-temperature data instead of instantaneous water-temperature data, by calculating
propensity scores to help improve the robustness of the analyses (Yuan, 2010), and by
using species-level OTUs for genera in which differences in which species-level thermal
preferences are known to occur.
3. Continue to further our knowledge of traits and how they relate to climate change. More
information is needed about which traits are most important in the context of climate
change, the influence of each trait on an organism's ability to adapt, and which
combinations of traits are most adaptive to particular environmental conditions (Stamp
et al., 2010). A key component of furthering the traits-based framework will be
expansion and unification of existing trait databases (Statzner and Beche, 2010).
7-71
-------
-------
8. REFERENCES
Allan, JD; Castillo, MM. (2007) Stream ecology: structure and function of running waters. Second edition. The
Netherlands: Springer. ISBN 978-1-4020-5582-9.436 pp.
Allan, JD; Erickson, DL; Fay, J. (1997) The influence of catchment land use on stream integrity across multiple
spatial scales. Freshw Biol 37:149-161.
Baker, DB; Richards, PP; Loftus, TT; et al. (2004) A new flashiness index: characteristics and applications to
midwestern rivers and streams. J Am Water Resour Assoc 40(2):503-522.
Barbour, MT; Gerritsen, J. (2006) Key features of bioassessment development in the United States of America. In:
Ziglio, G; Siligardi, M; Flaim, G; eds. Biological monitoring of rivers: applications and perspectives. Chichester,
England: John Wiley and Sons, Ltd.; pp. 351-366.
Barbour, MT; Stribling, JB; Karr, JR. (1995) Multimetric approach for establishing biocriteria and measuring
biological condition. In: Davis, WS; Simon, TP; eds. Biological assessment and criteria: tools for water resource
planning and decision making. Boca Raton, FL: CRC Press; pp. 63-76.
Barbour, MT; Gerritsen, J; Griffith, GE; et al. (1996) A framework for biological criteria for Florida streams using
benthic macroinvertebrates. J N AmBenthol Soc 15(2):185-211.
Barbour, MT; Gerritsen, J; Snyder, BD; et al. (1999) Rapid bioassessment protocols for use in streams and wadeable
rivers: periphyton, benthic macroinvertebrates and fish. Second Edition. U.S. Environmental Protection Agency;
Office of Water, Washington, DC; EPA/841/B-99/002. Available online at
http://water.epa.gov/scitech/monitoring/rsl/bioassessment/index.cfm.
Barbour, MT; Swietlik, WF; Jackson, SK; et al. (2000) Measuring the attainment of biological integrity in the USA:
A critical element of ecological integrity. Hydrobiologia 422/423:453-464.
Barbour, MT; Stribling, JB; Verdonschot, PFM. (2006) The multihabitat approach of USEPA's rapid bioassessment
protocols: benthic macroinvertebrates. Limnetica 25(3):839-850.
Barnett, TP; Adams, JC; Lettenmaier, DP. (2005) Potential impacts of a warming climate on water availability in
snow-dominated regions. Nature 438(17):303-309.
Beche, LA; McElravy, EP; Resh, VH. (2006) Long-term seasonal variation in the biological traits of benthic-
macroinvertebrates in two Mediterranean-climate streams in California, U.S.A. Freshw Biol 51(l):56-75.
Beche, LA; Resh, VH. (2007) Biological traits of benthic macroinvertebrates in California Mediterranean-climate
streams: long-term annual variability and trait diversity patterns. Fund Appl Limnol Archiv fur Hydrobiol 169(1): 1-
23.
Bierwagen, BG; Theobald, DM; Pyke, CR; et al. (2010) National housing and impervious surface scenarios for
integrated climate impact assessment. Proc Nat Acad Sci 107(49):20887-20892.
Bohmer, J; Rawer-Jost, G; Zenker, A. (2004) Multimetric assessment of data provided by water managers from
Germany: assessment of several different types of stressors with macrozoobenthos communities. Hydrobiologia
516(l-3):215-228.
Bogan, MT; Lytle, DA. (2007) Seasonal flow variation allows "time-sharing" by disparate aquatic insect
communities in montane desert streams. Freshw Biol 52(2):290-304. doi: 10.1111/j. 1365-2427.2006.01691.x.
Bonada, N; Rieradevall, M; Prat, N. (2007a). Macroinvertebrate community structure and biological traits related to
flow permanence in a Mediterranean river network. Hydrobiologia 589(1):91-106.
8-1
-------
Bonada, N; Doledec, S; Statzner. B. (2007b). Taxonomic and biological trait differences of stream
macroinvertebrate communities between Mediterranean and temperate regions: implications for future climatic
scenarios. Glob Change Biol 13(8):1658-1671.
Booth, DB; Jackson, CR. (1997). Urbanization of aquatic systems: degradation thresholds, stormwater detection,
and the limits of mitigation. J Am Water Resour Assoc 33(5): 1077-1090.
Brandt, D. (2001) Temperature preferences and tolerances for 137 common Idaho macroinvertebrate taxa. Idaho
Department of Environmental Quality, Coeur d'Alene, ID.
Breiman, L; Cutler, A. (2009) Random forests. Available online at
http://www.stat.berkeley.edu/~breiman/RandomForests/cc_home.htm#intro. Accessed April 23, 2009.
Britton, JR; Cucherousset, J; Davies, GD; et al. (2010) Non-native fishes and climate change: predicting species
responses to warming temperatures inatemperate region. FreshwBiol 55(5):1130-1141.
Brown, DB; Comrie, AC. (2004) A winter precipitation "dipole" in the Western United States associated with
multidecadal ENSO variability. Geophys Res Lett 31(9):L09203, doi: 10.1029/2003GL018726.
Buisson, L; Thuiller, W; Lek, S; et al. (2008) Climate change hastens the turnover of stream fish assemblages. Glob
Change Biol 14(10):2232-2248.
Burgmer, T; Hillebrand, H; Pfenninger, M. (2007) Effects of climate-driven temperature changes on the diversity of
freshwater macroinvertebrates. Oecologia 151(1):93-103.
Cairns, J, Jr. (1982) Artificial Substrates. Ann Arbor, Michigan: Ann Arbor Science Publishers, Inc.
Caissie, D. (2006) The thermal regime of rivers: a review. Freshw Biol 51:1389-1406.
Cao, Y; Hawkins, CP; Vinson, MR. (2003) Measuring and controlling data quality in biological assemblage surveys
with special reference to streambenthic macroinvertebrates. Freshw Biol 48(10):1898-1911.
Carlisle, DM; Wolock, DM; Meador, MR. (2010). Alteration of streamflow magnitudes and potential ecological
consequences: a multiregional assessment. Front Ecol Environ doi: 10.1890/100053.
Cereghino, R; Park, Y-S; Compin, A; et al. (2003) Predicting the species richness of aquatic insects in streams using
a limited number of environmental variables. J N Am Benthol Soc 22(3):442-456.
Cherkauer, KA; Sinha, T. (2010) Hydrologic impacts of projected future climate change in the Lake Michigan
region. J Great Lakes Res 36(Suppl 2):33-50.
Chessman, B. (2009) Climatic changes and 13-year trends in stream macroinvertebrate assemblages in New South
Wales, Australia. Glob Change Biol 15(11): 2791-2802, doi: 10.1111/j.l365-2486.2008.01840.x.
Christensen, N; Lettenmaier, DP. (2006) A multimodel ensemble approach to assessment of climate change impacts
on the hydrology and water resources of the Colorado River basin. Hydrol Earth Syst Sci Discuss 3(6):3727-3770.
Clements, WH; Brooks, ML; Kashian, ML; et al. (2008) Changes in dissolved organic material determine exposure
of stream benthic communities to UV-B radiation and heavy metals: implications for climate change. Glob Change
Biol 14:2201-2214.
Clow, D. (2010) Changes in the timing of snowmelt and streamflow in Colorado: a response to recent warming. J
Climate 23(9):2293-2306.
Collier, KJ. (2008) Temporal patterns in the stability, persistence and condition of stream macroinvertebrate
communities: relationships with catchment land-use and regional climate. Freshw Biol 53(3):603-616.
8-2
-------
Cormier, SM; Suter, GWII. (2008) A framework for fully integrating environmental assessment. Environ Manage
42:543-556. DOI 10.1007/s00267-008-9138-y.
Cuffney, TF; Bilger, MD; Haigler, AM. (2007) Ambiguous taxa: effects on the characterization and interpretation of
invertebrate assemblages. J N Am Benthol Soc 26(2):286-307.
Daufresne, M; Boet, P. (2007) Climate change impacts on structure and diversity offish communities in rivers.
Glob Change Biol 13(12):2467-2478.
Daufresne, M; Roger, MC; Capra, H; et al. (2004) Long term changes within the invertebrate and fish communities
of the upper Rhone River: effects of climate factors. Glob Change Biol 10(1):124-140.
Davies, SP; Jackson, SK. (2006) The biological condition gradient: a descriptive model for interpreting change in
aquatic ecosystems. Ecolog Appl 16(4): 1251-1266.
Davies, SP; Tsomides, L. (2002) Methods for biological sampling and analysis of Maine's rivers and streams.
Maine Department of Environmental Protection; Bureau of Land and Water Quality, Augusta, Maine; DEP
LW0387-B2002. Available online at
http://www.maine.gov/dep/water/monitoring/biomonitoring/materials/finlmethl.pdf. Accessed July 2009.
DeShon, JE. (1995) Development and application of the Invertebrate Community Index (ICI). In: Davis, WS;
Simon, TP; eds. Biological assessment and criteria: tools for water resource planning and decision making. Boca
Raton, FL: CRC Press; pp. 217-243.
Dettinger, MD; Cayan, DR; Meyer, MK; et al. (2004) Simulated hydrologic responses to climate variation and
changes in the Merced, Carson, and American River basins, Sierra Nevada, California, 1900-2099. Clim Change
62:283-317.
Dewson, ZS; James, ABW; Death, RG. (2007) Invertebrate community responses to experimentally reduced
discharge in small streams of different water quality. J N Am Benthol Soc 26(4):754-766.
Diamond, JM; Barbour, MT; Stribling, JB. (1996) Characterizing and comparing bioassessment methods and their
results: a perspective. J N Am Benthol Soc 15(4):713-727.
Diaz, AM; Suarez Alonso, ML; Vidal-Abarca Gutierrez, MR. (2008) Biological traits of stream macroinvertebrates
from a semi-arid catchment: patterns along complex environmental gradients. Freshw Biol 53(1):1-21.
Doledec, S; Dessaix, J; Tachet, H. (1996) Changes within the Upper Rhone River macrobenthic communities after
the completion of three hydroelectric schemes: anthropogenic effects or natural change? Archiv fur Hydrobiol
136(1): 19-40.
Durance, I; Ormerod, SJ. (2007) Climate change effects on upland stream macroinvertebrates over a 25-year period.
Glob Change Biol 13(5):942-957.
Durance, I; Ormerod, SJ. (2008) Trends in water quality and discharge confound long-term warming effects on river
macroinvertebrates. Freshw Biol(2) 54:388-405. doi:10.1111/j.l365-2427.2008.02112.x.
Easterling, DR; Karl, TR. (2001) Potential consequences of climate variability and change for the Midwestern
United States. In: Climate change impacts on the United States: the potential consequences of climate variability
and change. Report of the National Assessment Synthesis Team for the US Global Change Research Program.
Cambridge, UK: Cambridge University Press; 167-189.
Eaton, JG; Scheller, RM. (1996) Effects of climate warming on fish habitat in streams of the United States. Limnol
Oceanogr41(5):1109-1115.
Extence, CA; Balbi, DM; Chadd, RP. (1999) River flow indexing using British benthic macroinvertebrates: a
framework for setting hydroecological objectives. Regul River 15(6):543-574.
8-3
-------
Fenoglio, S; Bo, T; Cucco, M. (2007) Response of benthic invertebrate assemblages to varying drought conditions
in the Po River (NW Italy). Ital J Zool 74(2): 191-201.
Fisher, AE; Barren, B; Yarnal, CG; et al. (1997) Climate change impacts in the Mid-Atlantic region, a workshop
report. Workshop held at the Pennsylvania State University, September 9-11, 1997. Sponsored by U.S.
Environmental Protection Agency and the Pennsylvania State University. Available online at
http://www.usgcrp.gov/usgcrp/nacc/mara-workshop-report-1997.pdf.
Flanagan, KM; McCauley, E; Wrona, F; et al. (2003) Climate change: the potential for latitudinal effects on algal
biomass in aquatic ecosystems. Can J Fish Aqua Sci 60(6):635-639.
Flenner, I; Richter, O; Suhling, F. (2010) Rising temperature and development in dragonfly populations at different
latitudes. FreshwBiol 55:397-410.
Flotemersch, JE; Stribling, JB; Paul, MJ. (2006) Concepts and approaches for the bioassessment of non-wadeable
streams and rivers. U.S. Environmental Protection Agency, Natural Exposure Research Laboratory, Cincinnati,
Ohio; EPA/600/R-06/127. Available online at http://www.epa.gov/eerd/rivers/non-wadeable_full_doc.pdf.
Fore, LS; Karr, JR; Wisseman, RW. (1996) Assessing invertebrate responses to human activities: evaluating
alternative approaches. J N Am Benthol Soc 15(2):212-231.
Gallardo, B; Gascon, S; Cabezas, A; et al., (2009) Relationship between macroinvertebrate traits and environmental
gradients in a large regulated floodplain. Fund Appl Limnol Archiv fur Hydrobiologie 173(4):281-292.
Gerritsen, J; Barbour, MT; King, K. (2000) Apples, oranges, and ecoregions: on determining pattern in aquatic
assemblages. J N Am Benthol Soc 19(3):487-496.
Gibson, CA; Meyer, JL; Poff, NL; et al. (2005) Flow regime alterations under changing climate in two river basins:
implications for freshwater ecosystems. River Res Appl 21:849-864.
Gibson, WP; Daly, C; Kittel, T; et al. (2002) Development of a 103-year high-resolution climate data set for the
conterminous United States. In: Proceedings of the 13th AMS Conference on Applied Climatology, American
Meteorological Society, Portland, OR, May 13-16; pp. 181-183.
http://www.prism.oregonstate.edu/pub/prism/docs/appclim02-103yr_hires_dataset-gibson.pdf.
Golladay, SW; Gagnon, PG; Kearns, M; et al. (2004) Response of freshwater mussel assemblages
(Bivalvia:Unionidae) to a record drought in the Gulf Coast Plain of southwestern Georgia. J N Am Benthol Soc
23(3):494-506.
Gregory, JS; Beesley, SS; Van Kirk, RW. (2000) Effect of springtime water temperature on the time of emergence
and size of Pteronarcys californica in the Henry's Fork catchment, Idaho, U.S.A. Freshw Biol 45(l):75-83.
Griffith, G; Omernik, J; Comstock, J. (2002) Ecoregions of North Carolina. U.S. Environmental Protection Agency,
National Health and Environmental Effects Research Laboratory, Corvallis, OR.
Griswold, MW; Berzinis, RW; Crisman, TL; et al. (2008) Impacts of climatic stability on the structural and
functional aspects of macroinvertebrate communities after severe drought. Freshw Biol 53(12):2465-2483.
doi: 10.1111/j. 1365-2427.2008.02067.X.
Gutzler, D; Robbins, TO. (2011) Climate variability and projected change in the western United States: regional
downscaling and drought statistics. ClimDynam 37(5-6):835-849. DOI 10.1007/s00382-010-0838-7.
Gutzler, D; Garfin, G; Zak, B. (2006) Observed and predicted impacts of climate change on New Mexico's water
supplies. In: Watkins, A; ed. The impact of climate change on New Mexico's water supply and ability to manage
water resources. The New Mexico Office of the State Engineer, Interstate Stream Commission, Santa Fe, NM; pp.
4-32. Available online at http://www.nmdrought.state.nm.us/ClimateChangeImpact/completeREPORTfinal.pdf.
8-4
-------
Hamilton, AT; Stamp, JD; Bierwagen, BG. (2010a) Vulnerability of biological metrics and multimetric indices to
effects of climate change. J N Am Benthol Soc 29(4): 1379-1396.
Hamilton, AT; Barbour, MT; Bierwagen, BG. (2010b) Implications of global change for the maintenance of water
quality and ecological integrity in the context of current water laws and environmental policies. Hydrobiologia
657:263-278.
Hampton, SE. (2005) Increased niche differentiation between two Conochilus species over 33 years of climate
change and food web alteration. Limnol Oceanogr 50(2):421-426.
Hansen, J; Sato, M; Ruedy, R; et al. (2006) Global temperature change. Proc Nat Acad Sci 103(39): 14288-14293.
Hargrave, CW; Gary, KP; Rosado, SK. (2009) Potential effects of elevated atmospheric carbon dioxide on benthic
autotrophs and consumers in stream ecosystems: a test using experimental stream mesocosms. Glob Change Biol
15(11):2779-2790. doi: 10.1111/J.1365-2486.2009.01897.X.
Harper MP; Peckarsky, BL. (2006) Emergence cues of a mayfly in a high-altitude stream ecosystem: Potential
response to climate change. Ecol Appl 16(2):612-621.
Hawkins, CP; Hogue, JN; Decker, LM; et al. (1997) Channel morphology, water temperature, and assemblage
structure of stream insects. J N Am Benthol Soc 16(4):728-749.
Hawkins, CP; Norris, RH; Hogue, JN; et al. (2000) Development and evaluation of predictive models for measuring
the biological integrity of streams. Ecol Appl 10(5): 1456-1477.
Hayhoe, K; VanDorn, J;. Croley II, E; et al. (2010) Regional climate change projections for Chicago and the U.S.
Great Lakes. J Great Lakes Res 36(Suppl 2):7-21.
Hayhoe, K; Wake, CP; Huntington, TG; et al. (2007) Past and future changes in climate and hydrological indicators
in the U.S. Northeast. Clim Dynam 28(Suppl):381-407.
Heino, J. (2009) Species co-occurrence, nestedness and guild- environment relationships in stream
macroinvertebrates. FreshwBiol 54(9):1947-1959. doi:10.1111/j.l365-2427.2009.02250.x.
Helmuth, B; Mieszkowska, N; Moore, P; Hawkins, SJ. (2006) Living on the edge of two changing worlds:
forecasting the responses of rocky intertidal ecosystems to climate change. Ann Rev Ecol Evol Syst 37:373-404.
Herbst, D; Silldorff, EL. (2007) Development and evaluation of tolerance values for Lahontan Region invertebrates-
preliminary analysis summary. Sierra Nevada Aquatic Research Laboratory, University of California. Available
online at http://traits.tetratech-ffx.corn/DataSource/Herbst_Silldorff_2007/Herbst_CA_Report.doc.
Hering, D; Moog, O; Sandin, L; et al. (2004) Overview and application of the AQEM assessment system.
Hydrobiologia 516:1-20.
Hiddink, JG; ter Hofstede, R. (2008) Climate induced increases in species richness of marine fishes. Glob Change
Biol 14:453-460.
Hilsenhoff, WL. (1987) An improved biotic index of organic stream pollution. Great Lakes Entomol 20:31-39.
Hogg, ID; Williams, DD; Eadie, JM; Butt, SA. (1995) The consequences of global warming for stream
invertebrates: afield simulation. J Therm Biol 20(1-2): 199-206.
Horrigan, N; Baird, DJ. (2008) Trait patterns of aquatic insects across gradients of flow-related factors: a
multivariate analysis of Canadian national data. Can J Fish Aquat Sci 65(4):670-680.
Huff, DD; Hubler, AL; Pan, Y et al. (2008) Detecting shifts in macroinvertebrate community requirements:
implicating causes of impairment in streams. Oregon Department of Environmental Quality, Watershed Assessment,
8-5
-------
and Portland State University, Portland, OR; DEQ06-LAB-0068-TR. Available online at
http://fwcb.cfans.umn.edu/coop/people/huff/Stressor%20ID_ODEQ%20report_Dec08_FINAL.pdf.
Hughes, RN; Paulsen, SG; Stoddard, JL. (2000) EMAP-surface waters: a multiassemblage, probability survey of
ecological integrity in the U.S.A. Hydrobiologia 422-423:429-443.
Hurd, BH; Callaway, M; Smith, J; et al. (2004) Climatic change and U.S. water resources: from modeled watershed
impacts to national estimates. J Am Water Resour Assoc 40(1): 129-148.
Hurrell, JW. (1995) Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation.
Science 269(5224):676-679.
IPCC (Intergovernmental Panel on Climate Change). (2007a) Climate change 2007: impacts, adaptation and
vulnerability. Contribution of Working Group II to the fourth assessment report of the Intergovernmental Panel on
Climate Change. Parry, ML; Canziani, OF; Palutikof, PJ; et al; eds. Cambridge, UK: Cambridge University Press.
Available online at
http://www.ipcc.ch/publications_and_data/publications_ipcc_fourth_assessment_report_wg2_report_impacts_adapt
ation_and_vulnerability.htm.
IPCC (Intergovernmental Panel on Climate Change). (2007b) Climate change 2007: the physical science basis.
Summary for policy makers. Contribution of Working Group I to the fourth assessment report of the
Intergovernmental Panel on Climate Change. Geneva, Switzerland: IPCC Secretariat. Available online at
http://www.pnud.cl/recientes/IPCC-Report.pdf.
IPCC (Intergovernmental Panel on Climate Change). (2007c) Climate change 2007: the physical science basis. .
Contribution of Working Group I to the fourth assessment report of the Intergovernmental Panel on Climate Change.
Solomon, S; Qin, D; Manning, M; et al.; eds. Cambridge, UK: Cambridge University Press. Available online at
http://www.ipcc.ch/publications_and_data/publications_ipcc_fourth_assessment_report_wgl_report_the_physical_s
cience_basis.htm.
Jacobson, GL; Fernandez, IJ; Mayewski, PA; eds. (2009) Maine's climate future: an initial assessment. Orono, ME:
University of Maine.
Karl, TR; Melillo, JM; Peterson, TC. eds. (2009) Global Climate Change Impacts in the United States. Cambridge,
UK: Cambridge University Press.
Karr, JR. (1981). Assessment of biotic integrity using fish communities. Fisheries 6(6):21-27.
Karr, JR. (1991) Biological integrity: a long-neglected aspect of water resource management. Ecol Appl 1:66-84.
Karr, JR. (2006) Seven foundations of biological monitoring and assessment. Biol Ambien 20(2):7-18.
Karr, JR; Chu, EW. (2000) Sustaining living rivers. Hydrobiologia 422-423:1-14.
Kaushal, SS; Groffman, PM; Band, LE; et al. (2008) Interaction between urbanization and climate variability
amplifies watershed nitrate export in Maryland. Environ Sci Technol 42(16):5872-5878.
Kaushal, SS; Likens, GE; Jaworski, NA; et al. (2010) Rising stream and river temperatures in the United States.
Front Ecol Environ 8(9):461^66.doi:10.1890/090037.
Lamouroux, N; Doledec, S; Gayraud, S. (2004) Biological traits of stream macroinvertebrate communities: effects
of microhabitat, reach, and basin filters. J N Am Benthol Soc 23(3):449-466.
Lazorchak, JM; Klemm, DJ; Peck, DV; eds. (1998) Environmental monitoring and assessment program -surface
waters: field operations and methods for measuring the ecological condition of wadeable streams. U.S.
Environmental Protection Agency, National Exposure Research Laboratory, Cincinnati, OH; EPA/620/R-94/004F.
http://www.epa.gov/bioiwebl/pdf/EPA-620-R-94-
004FFieldOperationsandMethodsforMeasuringtheEcologicalConditionofWadeableStreams.pdf
8-6
-------
Lind, PR; Robson, BJ; Mitchell, BD. (2006) The influence of reduced flow during a drought on patterns of variation
in macroinvertebrate assemblages across a spatial hierarchy in two lowland rivers. Freshw Biol 51(12):2282-2295.
Mantua, NJ; Hare, SR; Zhang, Y; et al. (1997) A Pacific interdecadal climate oscillation with impacts on salmon
production. Bull Am Meteorol Soc 78(6): 1069-1079. Available online:
http://www.atmos.washington.edu/~mantua/abst.PDO.html.
Matthews, WJ. (1998) Patterns in Freshwater Fish Ecology. New York, NY: Chapman & Hall.
McCune, B; Mefford, MJ. (1999) PC-ORD: multivariate analysis of ecological data: version 4.0 for Windows. MjM
Software Design, Gleneden Beach, Oregon.
McManamay, RA; Orth, DJ; Kauffman, J. (2011) Ecological responses to flow alteration in the South Atlantic
region: a literature review and meta-analysis. Report to the Southeastern Aquatic Resources Partnership. 62 pp.
Available online at
http://filebox.vt.edu/users/dorth/Ecological%20Responses%20to%20Flow%20Alteration/Ecological%20Responses
%20to%20Flow%20Alteration%20in%20the%20South%20Atlantic%20Region%202011 .pdf.
Meador, MR; Carlisle, DM . (2007) Quantifying tolerance indicator values for common stream fish species of the
United States . Ecol Indie 7(2):329-338
Mearns, LO; Giorgi, F; McDaniel, L; et al. (2003) Climate Scenarios for the Southeastern U.S. based on GCM and
regional model simulations. Clim Change 60:7-35.
Milly, PCD; Dunne, KA; Vecchia, AV. (2005) Global pattern of trends in streamflow and water availability in a
changing climate. Nature 438:347-350.
Miltner, RJ; Rankin, ET. (2009) Analysis of national wadeable stream survey results in Region V: 2004-5.
Prepared by Midwest Biodiversity Institute for U.S. EPA, Region V, Water Division, Chicago, IL and U.S. EPA,
Office of Wetlands, Oceans, and Watersheds, Washington, DC. Yoder, CO (ed.), MBI Technical Report MBI/2009-
03-1, Columbus, OH. 88pp.
Minshall, GW; Petersen, RC; Nimz, CF. (1985) Species richness in streams of different size from the same drainage
basin. Am Natural 125(1): 16-38.
Mohseni, O; Stephan, HG; Eaton, JG. (2003) Global warming and potential changes in fish habitat in U.S. streams.
Clim Change 59(3):389-409.
Moline, MA; Claustre, H; Frazer, TK; et al. (2004) Alteration of the food web along the Antarctic Peninsula in
response to a regional warming trend. Glob Change Biol 10:1973-1980.
Montes-Hugo, M; Doney, SC; Ducklow, HW; et al. (2009) Recent changes in phytoplankton communities
associated with rapid regional climate change along the western Antarctic Peninsula. Science 323(5920): 1470-
1473.
Moog, O; Chovanec, A. (2000) Assessing the ecological integrity of rivers: walking the line among ecological,
political and administrative interests. Hydrobiologia 422-423:99-109.
Mouthon, J; Daufresne, M. (2006) Effects of the 2003 heatwave and climatic warming on mollusc communities of
the Saone: a large lowland river and of its two main tributaries (France). Glob Change Biol 12(3):441-449.
Mulholland, PJ; Best, GR; Coutant, CC; et al. (1997) Effects of climate change on freshwater ecosystems of the
southeastern United States and the Gulf of Mexico. HydrolProc 11(11):949-970.
Murphy, JF; Davies, CE; Corbin, TA. (2007) Developing indicators to climate-induced change in UK river
ecosystems. In: Kernan, M; Battarbee, RW; Binneu, HA eds. Proceedings of climate change and aquatic ecosystems
in the UK: science, policy and management. Environmental Change Research Centre, University College London,
May 16, 2007; pp. 8-14.
8-7
-------
Nakicenovic, N; Swart, R; eds. (2000) Special report on emissions scenarios. Intergovernmental Panel on Climate
Change (IPCC). Cambridge UK: Cambridge University Press. Available online at
http://www.grida.no/publications/other/ipcc_sr/.
NAST (National Assessment Synthesis Team). (2001) Climate change impacts on the United States - The potential
consequences of climate variability and change. Report of the National Assessment Synthesis Team for the US
Global Change Research Program. Cambridge, UK: Cambridge University Press. ISBN 0-521-00075-0.
NCDENR Memo by Kathy Herring. (2004) Effects of Long Term Drought on Benthic Macroinvertebrate
Communities in NC Streams and Tracking Their Recovery, 2002-2004.
NCDENR Memo by Trish MacPherson and Bryn Tracy. (2005) Post Hurricanes Frances, Ivan, and Jeanne
Biological Monitoring (French Broad and Watauga River Basins) Biological Sampling, November 30-December 2,
2004.
NCDENR (North Carolina Department of Environment and Natural Resources). (2006) Standard operating
procedures for benthic macroinvertebrates. Biological Assessment Unit, Division of Water Quality, Environmental
Sciences Section, Raleigh, NC.
NCDWQ (North Carolina Division of Water Quality). (2005) Identification methods for the origins of intermittent
and perennial streams, version 3.1. North Carolina Department of Environment and Natural Resources, Division of
Water Quality, Raleigh, NC. Available online at http://pbadupws.nrc.gov/docs/ML0914/ML091470581.pdf.
Niemi, GJ; McDonald, ME. (2004) Application of ecological indicators. Annu Rev Ecol Evol Syst 35:89-111.
Nijboer, RC; Johnson, RK; Verdonschot, PFM; et al. (2004) Establishing reference conditions for European streams.
Hydrobiologia 516:91-105.
Norris, RH; Barbour, MT. (2009) Bioassessment of aquatic ecosystems. In: Likens, GE; ed. Encyclopedia of inland
waters, volume 3. Oxford: Elsevier; pp. 21-28.
Odum, EP; Finn, JT; Franz, EH. (1979) Perturbantion theory and the subsidy-stress gradient. BioScience
29(6):349-352.
Ohio Environmental Protection Agency. (1987) Biological Criteria for the Protection of Aquatic Life. Vol. II. Users
Manual for Biological Field Assessment of Ohio Surface Waters. Division of Water Quality Monitoring and
Assessment, Surface Water Section, Columbus, OH.
Ohio Environmental Protection Agency. (1989) Biological Criteria for the Protection of Aquatic Life. Vol. III.
Standardized Biological Field Sampling and Laboratory Methods for Assessing Fish and Macroinvertebrate
Communities. Division of Water Quality Monitoring and Assessment, Surface Water Section, Columbus, OH.
Ohio EPA (Environmental Protection Agency). (2006) Methods for assessing habitat in flowing waters: using the
qualitative habitat evaluation index (QHEI). Division of Surface Water, Columbia OH; Technical Bulletin
EAS/2006-06-1. Available online at http://epa.ohio.gov/portals/35/documents/QHEIManualJune2006.pdf.
Omernik, JM. (1987) Ecoregions of the conterminous United States. Map (scale 1:7,500,000). Ann Assoc Am
Geogr 77(1): 118-125.
Parmesan, C. (2006) Ecological and evolutionary responses to recent climate change. Ann Rev Ecol Evol Syst
37:637-669.
Paul, MJ; Meyers, JL. (2001) Streams in the urban landscape. Ann Rev Ecol Syst 32:333-365.
Paulsen, SG; Hughes, RM; Larsen, DP. (1998) Critical elements in describing and understanding our nation's
aquatic resources. J Am Water Resour Assoc 34(5):995-1005.
-------
Pilgrim, JM; Fang, X; Stefan, HG. (1998) Stream temperature correlations with air temperatures in Minnesota:
implications for climate warming. J Am Water Resour Assoc 34(5): 1 109-1 121.
Poff, NL. (1997) Landscape filters and species traits: towards mechanistic understanding and prediction in stream
ecology. J N Am Benthol Soc 16(2):391-409.
Poff, ML. (2002) Ecological response to and management of increased flooding caused by climate change. Phil
Trans R Soc Lond A 360:1497 -1510.
Poff, NL; Zimmerman, JZH. (2009) Ecological responses to altered flow regimes: a literature review to inform the
science and management of environmental flows. Freshw Biol 55(1): 194-205. doi:10.1111/j.l365-
2427.2009.02272.x.
Poff, NL; Allan, JD, Bain, BM; et al. (1997) The natural flow regime: a paradigm for river conservation and
restoration. BioScience 47(ll):769-784.
Poff, LN; Brinson, MM; Day, JW, Jr. (2002) Aquatic ecosystems and global climate change: potential impacts on
inland freshwater and coastal wetland ecosystems in the United States. Prepared for the Pew Center on Global
Climate Change. Available online at http://www.pewtrusts.org/our_work_report_detail.aspx?id=30677.
Poff, NL; Bledsoe, BP; Cuhaciyan, CO. (2006a). Hydrologic variation with land use across the contiguous United
States: geomorphic and ecological consequences for stream ecosystems. Geomorphology 79(3-4) :264-285.
Poff, NL; Olden, JD; Vieira, NKM; et al. (2006b) Functional trait niches of North American lotic insects: traits-
based ecological applications in light of phylogenetic relationships. J N Am Benthol Soc 25(4):730-755.
Poff, NL; Pyne, MI; Bledsoe, BP; et al. (2010) Developing linkages between species traits and multiscaled
environmental variation to explore vulnerability of stream benthic communities to climate change. J N Am Benthol
Soc 29(4): 1441-1458.
PRISM Group. (2009) PRISM data sets. Oregon State University, Corvallis, OR. Available online at
http://www.prismclimate.org. Accessed April 23, 2009.
Rahel, FJ; Keleher; CJ; Anderson, JL. (1996) Potential habitat loss and population fragmentation for coldwater fish
in the North Platte River drainage of the Rocky Mountains: response to climate warming. Limnol Oceanogr
Ramstorf, S; Cazenave, A; Church, JA; et al. (2007) Recent climate observations compared to projections. Science
316(5825):709.
Rankin, ET. (1989) The Qualitative Habitat Evaluation Index (QHEI), rational, methods and application. Ohio
Environmental Protection Agency, Division of Surface Water Quality Planning and Assessment, Ecological
Assessment Section, Columbus, OH. Available online at
http://www.epa.state.oh.us/portals/35/documents/BioCrit88_QHEIIntro.pdf.
Rankin, ET. (1995) Habitat indices in water resource quality assessments. In: Davis, WS; Simon, TP; eds.
Biological assessment and criteria, tools for water resource planning and decision making. Boca Raton, FL: Lewis
Publishers; pp. 181-208.
Rankin, ET. (2009) Recalibration of the Ohio biocriteria: initial steps and recommendations. Prepared for Ohio
EPA, Division of Surface Water by Institute for Local Government Administration and Rural Development
(ILGARD), Ohio University, Athens, Ohio.
Resh, V; Rosenberg, D. (1984) The Ecology of Aquatic Insects. New York, NY: Praeger.
Richter, BD; Baumgartner, JV; Powell, J; et al. (1996) A method for assessing hydrologic alteration within
ecosystems. Conserv Biol 10(4): 1 163-1 174.
8-9
-------
Root, TL; Price, JT; Hall, KR; et al. (2003) Fingerprints of global warming on wild animals and plants. Nature
421:57-60.
Rosenberg, D; Resh, V; eds. (1993) Freshwater Biomonitoring and Benthic Macroinvertebrates. New York:
Chapman & Hall Publishers.
Sandin, L; Johnson, RK. (2000) The statistical power of selected indicator metrics using macroinvertebrates for
assessing acidification and eutrophication of running waters. Hydrobiologia 422-423:23 3-243.
Schoof, JT; Pryor, SC; Surprenant, J. (2010) Development of daily precipitation projections for the United States
based on probabilistic downscaling. J Geophy Res 115(D13106):1-13. doi:10.1029/2009JD013030.
Schueler, T. (1994) The importance of imperviousness. Watersh Protect Tech 1:100-111.
Seager, R; Vecchi, GA. (2010) Greenhouse warming and the 21st century hydroclimate of southwestern North
America. Proc AcadNatl Sci 107(50):21277-21282.
Seager, R; Ting, M; Held, I; et al. (2007) Model projections of an imminent transition to a more arid climate in
southwestern North America. Science 316(5828):1181-1184. [DOI:10.1126/science.ll39601.
Seebens, H; Cheinsle, U; Straile, D. (2009) Copepod life cycle adaptations and success in response to phytoplankton
spring bloom phenology. Glob Change Biol 15(6): 1394-1404. doi: 10.1111/j.l365-2486.2008.01806.x.
Shannon, CE; Wiener, W. (1963) The Mathematical Theory of Communication. Champaign, IL: University of
Illinois Press.
Sinokrot, BA; Gulliver, JS. (2000) In-stream flow impact on river water temperatures. J Hydraul Res 38(5):339-
349.
Smith, EP; Voshell, JR, Jr. (1997) Studies of benthic macroinvertebrates and fish in streams within EPA Region 3
for development of biological indicators of ecological condition. Virginia Polytechnic Institute and State
University, Blacksburg, VA.
Snook, H; Davies, SP; Gerritsen, J; et al. (2007) The New England Wadeable Stream Survey (NEWS): development
of common assessments in the framework of the biological condition gradient. Prepared for U.S. EPA Office of
Science and Technology, and U.S. EPA Office of Watersheds, Oceans, and Wetlands Washington, DC Washington,
DC. Available online at http://www.epa.gov/regionl/lab/pdfs/NEWSfinalReport_August2007.pdf.
Stamp, JD; Hamilton, AT; Zheng, L; et al. (2010) Use of thermal preference metrics to examine state biomonitoring
data for climate change effects. J N Am Benthol Soc 29(4): 1410-1423.
Stanley, EH; Buschman, DL; Boulton, AJ; et al. (1994) Invertebrate resistance and resilience to intermittency in a
desert stream. Am Midi Nat 131(2):288-3000.
Statzner, B; Beche, LA. (2010) Can biological invertebrate traits resolve effects of multiple stressors on running
water ecosystems? Freshw Biol 55(Suppl 1):80-119.
Stefan, HG; Fang, X; Eaton, JG. (2001) Simulated fish habitat changes in North American lakes in response to
projected climate warming. Trans Am Fish Soc 130(3):459-477.
Stephan, HG; Preudhomme, EB. (1993) Stream temperature estimation from air temperature. Water Resour Bull
29(l):27-45.
Stewart, IT; Cayan, DR; Dettinger, MD. (2005) Changes toward earlier streamflow timing across western North
America. J Clim 18(8):1136-1155.
Stoddard, JL; Larsen, DP; Hawkins, CP; et al. (2006) Setting expectations for the ecological condition of streams:
the concept of reference condition. Ecol Appl 16(4): 1267-1276.
8-10
-------
Stribling, JB; Jessup, BK; Feldman, DL. (2008) Precision of benthic macroinvertebrate indicators of stream
condition in Montana. J N Am Benthol Soc 27(l):58-67.
Suding, KN; Lavorel, S; Chapin, FS, III; et al. (2008) Scaling environmental change through the community-level: a
trait-based response-and-effect framework for plants. Glob Change Biol 14:1125-1140. doi: lO.llll/j.1365-
2486.2008.01557.x.
Suren, AM; Jowett, IG. (2006) Effects of floods versus low flows on invertebrates in a New Zealand gravel-bed
river. FreshwBiol 51(12):2207-2227.
Suter, GW, II; Norton, SB; Cormier, SM. (2002) A methodology for inferring the causes of observed impairments in
aquatic ecosystems. EnvironToxicol Chem21(6):1101-llll.
Sweeney, BW; Vannote, RL. (1978) Size variation and the distribution of hemimetabolous aquatic insects: two
thermal equilibrium hypotheses. Science 200(4340):444-446.
terBraak, CJF; Looman, CWN. (1986) Weighted averaging, logistic regression and the Gaussian response model.
Vegetation 65:3-11.
Tetra Tech, Inc. (2008) Draft predictive bioassessment models for New Mexico streams. Prepared for New Mexico
Environment Department, Santa Fe, NM by Tetra Tech, Inc. Owings Mills, MD.
Thuiller, W. (2004) Patterns and uncertainties of species' range shifts under climate change. Glob Change Biol
10(12):2020-2027.
TNC (The Nature Conservancy). (2007) Indicators of hydrologic alteration version 7 user's manual. Arlington, VA:
TNC.
Tobin, PC; Nagarkatti, S; LOEB, G; et al. (2008) Historical and projected interactions between climate change and
insect voltinism in a multivoltine species. Glob Change Biol 14(5):951-957, doi: lO.llll/j.1365-
2486.2008.01561.x.
TRC (Transportation Research Board). (1991). Road salt use in the United States. In: Highway deicing: comparing
salt and calcium magnesium acetate, Chapter 2. Road salt use in the United States. Special report 235. National
Research Council. Washington, DC: The National Academies Press.
Tuchman, NC; Wetzel, RG; Rier, ST; et al. (2002) Elevated atmospheric CO2 lowers leaf litter nutritional quality
for stream ecosystem food webs. Glob Change Biol 8(2): 163-170.
UCS (Union of Concerned Scientists). (2006) Climate Change in the U.S. Northeast. A report of the Northeast
Climate Impact Assessment. UCS Publications, Cambridge, MA. 35pp. Available online at
http://www.climatechoices.org/assets/documents/climatechoices/NECIA_climate_report_final.pdf
U.S. EPA (Environmental Protection Agency). (2000) Stressor identification guidance document. Office of Water
and Office of Research and Development, Washington DC. EPA-822-B-00-025. Available online at
http://cfpub.epa. gov/ncea/cfm/recordisplay.cfm?deid=20685.
U.S. EPA (Environmental Protection Agency). (2002) Primary distinguishing characteristics of Level III ecoregions
of the continental United States. Available online at
http://www.hort.purdue.edu/newcrop/cropmap/ecoreg/descript.html. Accessed July 30, 2009.
U.S. EPA (Environmental Protection Agency). (2008) Climate change effects on stream and river biological
indicators: a preliminary analysis. Global Change Research Program, National Center for Environmental
Assessment, Washington, DC; EPA/600/R-07/085. Available online at
http://cfpub.epa. gov/ncea/cfm/recordisplay.cfm?deid=190304.
U.S. EPA (Environmental Protection Agency). (2009) Land-use scenarios: national-scale housing-density scenarios
consistent with climate change storylines (final report). National Center for Environmental Assessment,
8-11
-------
Washington, DC; EPA/600/R-08/076F. Available online at
http://nepis.epa.gov/Exe/ZyPURL.cgi?Dockey=P100D08U.txt.
U.S. EPA (United States Environmental Protection Agency). (2011). An Assessment of Decision-Making Processes:
Evaluation of Where Land Protection Planning Can Incorporate Climate Change Information (Final Report). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-09/142F, 2011.
U.S. EPA (Environmental Protection Agency). (2012) Freshwater biological traits database. (External Review
Draft). National Center for Environmental Assessment, Global Change Research Program, Washington, DC.
Available online at http://cfpub.epa.gov/ncea/global/recordisplay.cfm?deid=234484.
Urquhart, NS; Paulsen, SG; Larsen, DP. (1998) Monitoring for policy-relevant regional trends over time. Ecol Appl
8(2):246-257.
Utah DEQ (Department of Environmental Quality). (2006) Monitoring manual. Division of Water Quality, Salt
Lake City, Utah. Available online at
http://www.waterquality.utah.gov/Monitoring/06_DWQ_monitoring_manual.pdf. Accessed July 2009.
USU (Utah State University). (2009) Predictive models. The western center for monitoring and assessment of
freshwater ecosystems. Department of Aquatic, Watershed, and Earth Resources, Logan, UT. Available online at
http://www.cnr.usu.edu/wmc/. Accessed April 23,2009.
van Vliet, MTH; Zwolsman, JJG. (2008) Impact of summer droughts on the water quality of the Meuse river. J
Hydrol 353(1-2): 1-17.
Vannote, RL; Sweeney, BW. (1980) Geographic analysis of thermal equilibria: a conceptual model for evaluating
the effects of natural and modified thermal regimes on aquatic insect communities. Am Nat 115(5):667-695.
Verberk, WCEP; Siepel, H; Esselink, H. (2008a) Life-history strategies in freshwater macroinvertebrates. Freshw
Biol 53(9): 1722-1738.
Verberk, WCEP; Siepel, H; Esselink, H. (2008b) Applying life-history strategies for freshwater macroinvertebrates
to lentic water. Freshw Biol 53(9): 1739-1753.
Verdonschot, PFM. (2006) Evaluation of the use of Water Framework Directive typology descriptors, reference sites
and spatial scale in macroinvertebrate stream typology. Hydrobiologia 566(l):39-58.
Verdonschot, PFM; Moog, O. (2006) Tools for assessing European streams with macroinvertebrates: major results
and conclusions from the STAR project. Hydrobiologia 566(1):299-309.
Vieira, NKM; Poff, NL; Carlisle, SM; et al. (2006) A database of lotic invertebrate traits for North America. U.S.
Geological Survey Data Series 187. U.S. Department of the Interior, U.S. Geological Survey, Washington, DC.
Available online at: http://pubs.water.usgs.gov/dsl87.
Vogelmann, JE; Howard, SM; Yang, L; et al. (2001) Completion of the 1990's national land cover data set for the
conterminous United States. Photogram Eng Rem S 67:650-662.
Wallin, M; Wiederholm, T; Johnson, RK. (2003) Guidance on establishing reference conditions and ecological
status class boundaries for inland surface waters (final). Produced by CIS Working Group 2.3 - REFCOND. Final
version 7.0, 2003-03-05. Available online at http://www.minenv.gr/pinios/00/odhgia/7th_draft_refcond_fmal.pdf.
Walther, GR; Post, E; Convey, P; et al. (2002) Ecological responses to recent climate change. Nature 416:389-395.
Walther, GR; Berger, S; Sykes, MT. (2005) An ecological "footprint" of climate change. Proc Royal Soc B Biol Sci
272(1571):1427-1432.
Wang, L; Lyons, J; Kanehl P; et al. (2001) Impacts of urbanization on stream habitat and fish across multiple spatial
scales. Environ Manage 28(2):255-266.
8-12
-------
Ward, JV. (1985) Thermal characteristics of running waters. Hydrobiologia 125(l):31-46.
Webb, JA; Stewardson, MJ; Koster, WM. (2009) Detecting ecological responses to flow variation using Bayesian
hierarchical models. FreshwBiol 55(1): 108-126. doi:10.1111/j.l365-2427.2009.02205.x.
Wehrly, KE; Brenden, TO; Wang, L. (2009) A comparison of statistical approaches for predicting stream
temperatures across heterogeneous landscapes. J Am Water Resour Assoc 45(4):986-997.
Whittier, TR; Larsen, DP; Hughes, RM; et al. (1987) The Ohio stream regionalization project: a compendium of
results. U.S. Environmental Protection Agency, Environmental Research Laboratory, Corvallis, OR; EPA/600/3-
87/025, 66 p. Available online at http://www.epa.gov/bioiwebl/pdf/EPA-600-3-87-
025OhioStreamRegionalizationProjectCompendiumOfResults.pdf.
Wolock, DM; McCabe, GJ. (1999) Estimates of runoff using water-balance and atmospheric general circulation
models. J Am Water Resour Assoc 35(6): 1341-1340.
Wright, JF. (2000) An introduction to RIVPACS. In: Wright, JF; Sutcliffe, DW; Furse, MT; eds. Assessing the
biological quality of freshwaters: RIVPACS and other techniques. Ambleside, Cumbria, United Kingdom:
Freshwater Biological Association; pp. 1-24.
Wright, JF; Moss, D; Armitage, PD; et al. (1984) A preliminary classification of running-water sites in Great Britain
based on macroinvertebrate species and the prediction of community type using environmental data. Freshw Biol
14(3):221-256.
Wu, J; Li, H. (2006) Perspectives and methods of scaling. In: Wu, J; Jones, KB; Li, H; et al; eds. Scaling and
uncertainty analysis in ecology: methods and applications. Dordrecht, The Netherlands: Springer; pp. 17-44.
Wuebbles, DJ; Hayhoe, K. (2004) Climate change projections for the United States Midwest. Mit Adapt Strat Glob
Change 9(4):335-363.
Yarnell, SM; Viers, JH; Mount, JF. (2010) Ecology and management of the spring snowmelt recession. BioScience
60(2): 114-127.
Yoder, CO. (1989) The development and use of biological criteria for Ohio surface waters. In: Water-quality
standards for the 21st century. Proceedings of a national conference, March 1-3, 1989, Dallas, Texas. Office of
Water, U.S. Environmental Protection Agency, Washington, DC; EPA/906/R-89/103; pp. 139-146. Available online
at http//nepis.epa.gov/Exe/ZyPURL.cgi?Dockey=20014EQF.txt.
Yoder, CO; Barbour, MT. (2009) Critical technical elements of state bioassessment programs: a process to evaluate
program rigor and comparability. Environ Monit Assess 150(l-4):31-42.
Yoder, CO; Rankin, ET. (1995) Biological criteria program development and implementation in Ohio. In: Davis,
WS; Simon, T; eds. Biological assessment and criteria: tools for water resource planning and decision making. Boca
Raton: Lewis Publishers; pp. 109-144.
Yoder, CO; Rankin, ET; Smith, MA; et al. (2005) Changes in fish assemblage status in Ohio's nonwadeable rivers
and streams over two decades. In: Rinne, JN; Hughes, RM; eds. Historical changes in large river fish assemblages
of the Americas. American Fisheries Society Symposium Series Vol 45. BethesdaMD: American Fisheries Society;
399-129.
Yuan, L. (2006) Estimation and application of macroinvertebrate tolerance values. U.S. Environmental Protection
Agency, Office of Research and Development, Washington, DC; EPA/600/P-04/116F. Available online at
http://ofmpub.epa. gov/eims/eimscomm.getfile?p_download_id=457196.
Yuan, L. (2010) Estimating the effects of excess nutrients on stream invertebrates from observational data. Ecol
Appl 20:110-125.
8-13
-------
Zuckerberg, B; Woods, AM; Porter, WF. (2009) Poleward shifts in breeding bird distributions in New York State.
Glob Change Biol 15(8):1866-1883. doi: 10.1111/j.l365-2486.2009.01878.x.
Zwolsman, JJG; Van Bokhoven, AJ. (2007) Impact of summer droughts on water quality of the Rhine River - a
preview of climate change? Water Sci Technol 56(4):45-55.
Selected Sourcers from Climate Change Information
Climate Wizard - Accessed 2009
http://www.climatewizard.org/
PRISM Group - Accessed 2009
http://www.prismclimate.org
USGS
http://waterdata.usgs.gov/nwis/rt
Utah Climate Center
http://climate.usurf.usu.edu/products/data.php
National Center for Atmospheric Research
http://rcpm.ucar.edu
8-14
-------
APPENDIX A
TAXONOMIC CORRECTIONS AND EVALUATION
A-l
-------
A.l. TAXONOMIC CORRECTIONS AND EVALUATIONS PERFORMED ON THE
OHIO DATA SET
The Midwest Biodiversity Institute (MBI) developed a list of possible taxa that could
affect the Invertebrate Condition Index (ICI) scoring via taxonomic refinement (splitting or
lumping of taxa). MBI then conferred with senior Ohio Environmental Protection Agency
(EPA) taxonomists (Mike Bolton and Jack Freda) to determine how to best address these
changes. Their efforts primarily resulted in "combining of the" individual taxa designations of
mayflies back into "Baetis sp." or "Pseudocloeon sp." as described in Table A-l. This process
assured that changes found in the ICI calculated at reference sites for the historic and current
periods would be reflecting biological responses to changing conditions and not changes in
taxonomy. See results in Tables 6-7 and 6-8 of the main report for a summary of the impact of
these taxonomic fixes on index values.
A.2. EVALUATION OF TAXA CORRECTIONS—NONMETRIC
MULTIDIMENSIONAL SCALING (NMDS)
In the Maine, North Carolina, and Utah data sets, we used NMDS to evaluate whether the
database 'fixes,' and in particular the taxonomic corrections and application of operational
taxonomic unit (OTU) rules, were effective in minimizing changes over time due to taxonomic
identification procedures rather than actual community changes. For the Ohio data set,
taxonomic fixes were conducted by Ed Rankin and Chris Yoder of MBI and were
straightforward, mainly recombining mayfly taxa for which refinements resulted in renaming or
splitting of taxa since the historic time period during which reference communities were
evaluated using the ICI. Postfacto NMDS evaluation was not deemed necessary for that
application (see results in Tables 6-7 and 6-8 of the main report for a summary of the impact of
these taxonomic fixes on index values). For the Maine, North Carolina, and Utah data sets, the
NMDS ordinations were run before and after generating genus-level OTUs. Various grouping
variables (i.e., year, month, collection method, taxonomy lab, ecoregion, watershed, etc.) were
overlaid to look for trends. Figures A-l A through A-14B and Figures A-18 through A-22B show
the NMDS plots that were generated as part of this exercise. Figures A-15 through A-17 show
more details about number of identifications by species, genera, and families, as well as
differences in total taxa identifications by laboratory. Table A-2 lists the laboratories references
in Figure A-17.
A-2
-------
Table A-l. Mayfly taxa from reference sites in Ohio that abruptly appeared
(Later) or disappeared (Earlier) in the Ohio data set and explanation of
change. Explanations were provided by Mike Bolton and Jack Freda of OH
EPA
Taxa
code
11010
11014
11015
11018
11020
11110
11115
11118
11119
11120
11125
11130
11150
11155
11175
Taxon name
Acentrella sp.
Acentrella turbida
Acerpenna sp.
Acerpenna
macdunnoughi
Acerpenna pygmaea
Acentrella parvula
Baetis tricaudatus
Plauditus dubius
Plauditus dubius or P.
virilis
Baetis flavistriga
Pseudocloeon
frondale
Baetis intercalaris
Pseudocloeon
propinquum
Plauditus
punctiventris
Plauditus virilis
Appearance
Later
Later
Later
Later
Later
Later
Later
Later
Later
Later
Later
Later
Later
Later
Later
Explanation of change
Advancements in taxonomy allow this taxa to
be distinguished from Pseudocloeon sp.
Advancements in taxonomy allow this taxon
to be distinguished from Pseudocloeon sp.
Advancements in taxonomy allow this taxon
to be distinguished from Baetidae sp.
Advancements in taxonomy allow this taxon
to be distinguished from Baetidae sp.
Advancements in taxonomy allow this taxon
to be distinguished from Baetidae sp.
Advancements in taxonomy allow this taxon
to be distinguished from Pseudocloeon sp. or
was renamed from Pseudocloeon parvulum
Advancements in taxonomy allow this taxon
to be distinguished from Baetidae sp.
Advancements in taxonomy allow this taxon
to be distinguished Pseudocloeon sp.
Advancements in taxonomy allow this taxon
to be distinguished Pseudocloeon sp.
Advancements in taxonomy allow this taxon
to be distinguished from Baetidae sp.
Advancements in taxonomy allow this taxon
to be distinguished from Baetidae sp.
Advancements in taxonomy allow this taxon
to be distinguished from Baetidae sp.
Advancements in taxonomy allow this taxon
to be distinguished from Baetidae sp.
Advancements in taxonomy allow this taxon
to be distinguished Pseudocloeon sp.
Advancements in taxonomy allow this taxon
to be distinguished Pseudocloeon sp.
A-3
-------
Table A-l. Mayfly taxa from reference sites in Ohio that abruptly appeared
(Later) or disappeared (Earlier) in the Ohio data set and explanation of
change. Explanations were provided by Mike Bolton and Jack Freda of OH
EPA (continued)
Taxa
code
11250
11400
11430
11503
11600
11625
11645
11650
11651
11670
11700
13010
13030
14501
14900
14950
Taxon name
Centroptilum sp. (w/o
hindwing pads)
Centroptilum sp. or
Procloeon sp.
(formerly in Cloeon
Diphetor hageni
Heterocloeon
curiosum
Paracloeodes sp. 1
Paracloeodes sp. 3
Procloeon sp.
Procloeon sp.
(w/hindwing pads)
Procloeon sp. (w/o
hindwing pads)
Procloeon irrubrum
Acentrella sp. or
Plauditus sp.
(formerly in Pseudoc}
Leucrocuta hebe
Leucrocuta
maculipennis
Leptophlebiidae
Leptophlebia sp.
Leptophlebia sp. or
Paraleptophlebia sp.
Appearance
Later
Earlier
Later
Later
Later
Later
Later
Later
Later
Later
Earlier
Earlier
Earlier
Earlier
Later
Later
Explanation of change
Advancements in taxonomy allow this taxon
to be distinguished Cloeon sp.
Advancements in taxonomy allow this taxon
to be distinguished Cloeon sp.
Advancements in taxonomy allow this taxon
to be distinguished from Baetidae sp.
Renamed Heterocloeon (H.) sp.,
Heterocloeon sp.
Advancements in taxonomy allow this taxon
to be distinguished from Paracloeodes sp.
Advancements in taxonomy allow this taxon
to be distinguished from Paracloeodes sp.
Was earlier classified as Centroptilum sp. or
Cloeon sp.
Was earlier classified as Cloeon sp.
Was earlier classified as Centroptilum sp.
Advancements in taxonomy allow this taxon
to be distinguished from Cloeon sp.
Renamed as Pseudocloeon sp.
Renamed as Heptagenia hebe
Renamed as Heptagenia maculipennis
Now coded as Leptophlebia sp.
Leptophlebia sp.
Small specimens lumped
A-4
-------
Utah (pre-OTU)
Lab
i pre-1989
post-1989
Axis 1
Figure A-1A. Pre-OTU (genus) NMDS plot when lab is used as the grouping
variable.
Utah (post-OTU)
M * A
^A^AV^V^A
A IfA. A A^ A
Lab
Apre-1989
post-1989
Axis 1
Figure A-1B. Post-OTU (genus) NMDS plot when lab is used as the grouping
variable.
A-5
-------
Utah (pre-OTU)
^.v
Axis 1
Ecoregion (Level 3)
> 13 (Central Basin and Range)
14 (Mojave Basin and Range)
18 (Wyoming Basin)
' 19 (Wasatch and Uinta Mountains)
> 20 (Colorado Plateaus)
21 (Southern Rockies)
> 80 (Northern Basin and Range)
Figure A-2A. Pre-OTU (genus) NMDS plot when Level 3 ecoregion is used
as the grouping variable.
Utah (post-OTU)
Axis 1
Ecoregion (Level 3)
A 13 (Central Basin and Range)
4 14 (Mojave Basin and Range)
18 (Wyoming Basin)
T 19 (Wasatch and Uinta Mountains)
o 20 (Colorado Plateaus)
21 (Southern Rockies)
a 80 (Northern Basin and Range)
Figure A-2B. Post-OTU (genus) NMDS plot when Level 3 ecoregion is used
as the grouping variable.
A-6
-------
Utah (pre-OTU)
Ref Status
ARef
» So-So
^ Trash
Axis 1
Figure A-3A. Pre-OTU (genus) NMDS plot when reference status is used as
the grouping variable.
Utah (post-OTU)
Ref Status
A Ref
So-So
Trash
Axis 1
Figure A-3B. Post-OTU (genus) NMDS plot when reference status is used as
the grouping variable.
A-7
-------
Utah (pre-OTU)
HUC04
a1403
1404
1405
'1406
01407
1408
= 1501
•1601
n1602
• 1603
x1704
Axis 1
Figure A-4A. Pre-OTU (genus) NMDS plot when Hydrologic Unit Code
(HUC)-04 is used as the grouping variable.
Utah (post-OTU)
HUC04
A 1403
A 1404
1405
T1406
«1407
1408
°1501
•1601
n1602
• 1603
X1704
Axis 1
Figure A-4B. Post-OTU (genus) NMDS plot when HUC-04 is used as the
grouping variable.
A-8
-------
Utah (pre-OTU)
Latitude
Axis 1
r= .139tau= .107
Axis 2
r = -.010tau = -.003
^via ' :7~ ••~
Ref Status
A Ret
A So-So
Trash
TNA
Figure A-5A. Pre-OTU (genus) NMDS plot when reference status is used as
the grouping variable. Trends related to latitude are also evaluated.
Utah (post-OTU)
37
39 41
Latitude
Axis 1
r= .119tau= .078
Axis 2
r= .087tau = .061
41 •
39-
37-
Ref Status
A Ref
A So-So
V Trash
TNA
Figure A-5B. Post-OTU (genus) NMDS plot when reference status is used as
the grouping variable. Trends related to latitude are also evaluated.
A-9
-------
Utah (pre-OTU)
(N
(A
Longitude
Axisl
r= -.220 tau = -.188
Axis 2
r = .008 tau = .036
Axis 1
IS'i^tt',1.
Ref Status
ARef
A So-So
V Trash
TNA
Figure A-6A. Pre-OTU (genus) NMDS plot when reference status is used as
the grouping variable. Trends related to longitude are also evaluated.
-115
-113
-111
-109
Longitude
Axisl
r = -.218tau = -.208
Axis 2
r = .076 tau = .045
-111 •
-113-
-115-1
Utah (post-OTU)
Axis 1
Ref Status
ARef
A So-So
Trash
TNA
Figure A-6B. Post-OTU (genus) NMDS plot when reference status is used as
the grouping variable. Trends related to longitude are also evaluated.
A-10
-------
Maine (pre-OTU)
"O V
O^>««W&VT "?V«At°l« T»v
^V^jS A:*:
pjTf_t^ ^0^-40,, \ 0
v^-3 . n
Year Group (5)
a 1970-4
1980-4
1985-9
T 1990-4
» 1995-9
2000-4
a 2005-6
Axis 1
Figure A-7A. Pre-OTU (genus) NMDS plot using sample years (5-year
increments) as the grouping variable.
Maine (post-OTU)
o o o c
Year Group (5)
A 1970-4
1980-4
1985-9
T1990-4
»1995-9
2000-4
o 2005-6
Axis 1
Figure A-7B. Post-OTU (genus) NMDS plot using sample years (5-year
increments) as the grouping variable.
A-ll
-------
Maine (pre-OTU)
Year Group (10)
A 1970-9
1980-9
1990-9
T 2000-6
Axis 1
Figure A-8A. Pre-OTU (genus) NMDS plot using sample years (10-year
increments) as the grouping variable.
Maine (post-OTU)
* v WFgfeS> S* ^
, *^S$K£&'^
.
Year Group (10)
A 1970-9
1980-9
1990-9
' 2000-6
Axis 1
Figure A-8B. Post-OTU (genus) NMDS plot using sample years (10-year
increments) as the grouping variable.
A-12
-------
Maine (pre-OTU)
i
*7 A
Year Group (20)
ANA
1970-1989
1990-2006
Axis 1
Figure A-9A. Pre-OTU (genus) NMDS plot using sample years (20-year
increments) as the grouping variable.
Maine (post-OTU)
Year Group (20)
ANA
1970-1989
1990-2006
Axis 1
Figure A-9B. Post-OTU (genus) NMDS plot using sample years (20-year
increments) as the grouping variable.
A-13
-------
Maine (pre-OTU)
*
Wtr'-^
; V -0%,^^% * ^TTT T
Ref Status
»NA
Axis 1
Figure A-10A. Pre-OTU (genus) NMDS plot when reference status is used as
the grouping variable.
Maine (post-OTU)
SSltXlf^^v
' ^'vfe^,|ji^L< vK^w/
\«s®3!5Si^'A^ -
->\
-r
/Tl^T^^/'
i*r>: ^- r:
T" "/» TT'
V
Ref Status
Axis 1
Figure A-10B. Post-OTU (genus) NMDS plot when reference status is used
as the grouping variable.
A-14
-------
Maine (pre-OTU)
'T
>" ^
*• 'y ^V>^
^USI^T T
T^1%V
Axis 1
Ecoregion (Level 3)
ANA
58 (NE Highlands)
59 (NE Coastal Zone)
* 82 (Laurentine Plaines and Hills)
Figure A-11A. Pre-OTU (genus) NMDS plot when Level 3 ecoregion is used
as the grouping variable.
Maine (post-OTU)
Axis 1
Ecoregion (Level 3)
ANA
58 (NE Highlands)
v 59 (NE Coastal Zone)
* 82 (Laurentine Plains and Hills)
Figure A-11B. Post-OTU (genus) NMDS plot when Level 3 ecoregion is used
as the grouping variable.
A-15
-------
Maine (pre-OTU)
Latitude
Axisl
r= .014tau = -.171
Axis 2
r= .009tau= .029
w
Ref Status
AAA
AA
B
TC
ONA
Figure A-12A. Pre-OTU (genus) NMDS plot when reference status is used as
the grouping variable. Trends related to latitude are also evaluated.
Maine (post-OTU)
Latitude
Axis 1
r = -.022tau= -.178
Axis 2
r = -.009tau=-.036
' M
I
40-
20-
Ref Status
AAA
AA
B
TC
ONA
Axis 1
Figure A-12B. Post-OTU (genus) NMDS plot when reference status is used
as the grouping variable. Trends related to latitude are also evaluated.
A-16
-------
Maine (pre-OTU)
Longitude
Axis 1
r = -.109tau = -.095
Axis 2
r= .000 tau= .046
Axis 1
Ref Status
AAA
AA
B
TC
ONA
Figure A-13A. Pre-OTU (genus) NMDS plot when reference status is used as
the grouping variable. Trends related to longitude are also evaluated.
: CM
CO
I
Longitude
Axis 1
-.072tau = -.084
Axis 2
-.007tau = -.056
0
-20
Maine (post-OTU)
•STO
\$L.
• ,""* %
A*'"1-^'*.^ V. • "
^"^p^'
W\iV^V*-T »3f.» •
T
i
Axis 1
Ref Status
AAA
AA
B
TC
ONA
Figure A-13B. Post-OTU (genus) NMDS plot when reference status is used
as the grouping variable. Trends related to longitude are also evaluated.
A-17
-------
v7 A A
^_ *TAA
LabCode2
42
A4
6
T8
»9
11
°16
• 17
°18
• 20
x99
Axis 1
Figure A-14A. Pre-OTU (genus) NMDS plot for Maine data when lab is used
as the grouping variable.
LabCode2
A2
*4
6
»8
»9
11
016
• 17
D18
• 20
X99
Axis 1
Figure A-14B. Post-OTU (genus) NMDS plot for Maine data when lab is
used as the grouping variable.
A-18
-------
16 i
1970
2005 2010
Figure A-15A. Average number of species-level identifications per replicate
sample per year in the Maine database (using original data; not adjusted for
OTUs).
1970
2005
2010
Figure A-15B. Average number of genus-level identifications per replicate
sample per year in the Maine database (using original data; not adjusted for
OTUs).
A-19
-------
Year
Figure A-16A. Average number of species-level identifications per replicate
sample per year for selected families in the Maine database (using original
data; not adjusted for OTUs).
Figure A-16B. Average number of genus-level identifications per replicate
sample per year for selected families in the Maine database (using original
data; not adjusted for OTUs).
A-20
-------
Maine, GTU
b
I
40 -
30 -
20 -
10 -
o H
H
B n.
B
Lab
Figure A-17. Distribution of the total number of taxa (average per replicate)
among laboratories in Maine.
A-21
-------
Table A-2. List of 16 different individuals or labs that performed
taxonomic analyses on Maine benthic samples during the study period
1983-2006. Per communication with Leon Tsomides Maine
Department of Environmental Protection (ME DEP), some
adjustments were made to taxonomy produced from different sources
to assure consistency
Lab
BILLIE BESSIE
DAVID COURTEMANCH
B.A.RENVIRONM
WOODWARD CLYDE
Unknown
BBL SCIENCES
CF RABENI
QST ENVIRONMENTAL (BOWATER)
CHRIS PINNUTO
NORMANDEAU
SUSAN DAVTES
NEW BRUNSWICK
IDAHO ECO ANALYSTS
TERRY MINGO
LOTIC
MICHAEL WINNELL
Year_min
1996
1983
1994
1981
1995
2004
1974
1994
2000
1989
1981
1999
1999
1983
1988
1983
Year_max
1996
1983
1994
1981
1995
2004
1974
1996
2000
1999
1989
2001
2005
1987
2006
2006
#Samp
2
5
6
6
7
9
10
20
22
45
74
84
100
254
743
2,509
LabNum
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
A-22
-------
CM
W
"x
Collection Method
A Full Scale
EPT
Boat
T 2Kicks
o EQUAL
QuaM
o Qual 5
• Qual 7
nSwamp
Axis 1
Figure A-18. Preliminary North Carolina NMDS plot (genus-level OTU)
using collection method as the grouping variable.
A-23
-------
NC - Taxonomy - GTU IDs
pnn
A nn
4UU -
inn
1 nn -
* --"^ V*NW*~*
7 T ~*4
7
_^ I
1978 1982 1986 1990 1994 1998 2002 200I
—•— Num Taxa
• Taxa First
Taxa Last
Num Stations
Figure A-19A. North Carolina genus-level OTU (GTU) data using all
collection methods. "Num Taxa" refers to the total number of taxa recorded in a
particular year; "Taxa First" refers to the number of taxa that appear in the
database for the first time in a particular year; "Taxa Last" refers to the number of
taxa that appear in the database for the last time in a particular year; "Num
Stations" refers to the number of stations sampled in a particular year.
NC - Taxonomy - GTU IDs - (Full-scale method only)
finn -i
ADO .
onn
100 -
>-**y***-«L
7 ** *^**^>*v*^V*
jm
4 .
/ H
1978 1982 1986 1990 1994 1998 2002 200!
5
— *— Num Taxa
• Taxa First
Taxa Last
Nurn Stations
Figure A-19B. North Carolina GTU using data from only the Full-scale
collection method. "Num Taxa" refers to the total number of taxa recorded in a
particular year; "Taxa First" refers to the number of taxa that appear in the
database for the first time in a particular year; "Taxa Last" refers to the number of
taxa that appear in the database for the last time in a particular year; "Num
Stations" refers to the number of stations sampled in a particular year.
A-24
-------
«*^£A rw
^:%t
* ? T^,
YrGrpOS
A3
Axis 1
Figure A-20A. Pre-OTU (genus) NMDS plot for North Carolina data when
year (5-year increments) is used as the grouping variable, and only full-scale
collection method data are used.
L *W^WA,AA A^
S«^fe4
YrGrpOS
A3
*4
5
»6
»7
8
Axis 1
Figure A-20B. Post-OTU (genus) NMDS plot for North Carolina data when
year (5-year increments) is used as the grouping variable, and only full-scale
collection method data are used.
A-25
-------
CM
(0
Reference Status
A Reference
Unknown
Axis 1
Figure A-21A. Pre-OTU (genus) NMDS plot for North Carolina data using
reference status as the grouping variable, and only full-scale collection
method data are used.
1 : *" .
* * 4*. » .* . i *
Reference Status
A Reference
Unknown
Axis 1
Figure A-21B. Post-OTU (genus) NMDS plot for North Carolina data using
reference status as the grouping variable, and only full-scale collection
method data are used.
A-26
-------
:'#sfiafi$£
Ecoregion
A Unknown
45, Piedmont
63, Mid Atlantic CP
» 65, SE Plains
o 66, Blue Ridge
Axis 1
Figure A-22A. Pre-OTU (genus) NMDS plot for North Carolina data using
Level 3 ecoregion as the grouping variable, and only full-scale collection
method data are used.
Ecoregion (Level 3)
A Unknown
45, Piedmont
63, Mid Atlantic CP
» 65, SE Plains
° 66, Blue Ridge
Axis 1
Figure A-22B. Post-OTU (genus) NMDS plot for North Carolina data using
Level 3 ecoregion as the grouping variable, and only full-scale collection
method data are used.
A-27
-------
-------
APPENDIX B
ADDITIONAL ANALYSES PERFORMED ON UTAH DATA
B-l
-------
B.I. HYDROLOGIC ANALYSIS PERFORMED ON THE UTAH DATA SET
Figure B-l shows the locations of the 43 Utah biological sampling stations that we
associated with United States Geological Service (USGS) stream gages.
n
2®
Figure B-l. Locations of the 43 Utah biological sampling stations (red
triangles) and associated USGS stream gages (yellow circles). Stations that are
highlighted in blue are classified as reference sites by Utah DEQ Division of
Water Quality. The numbers next to the sites are the number of years of data that
were available for each station.
B-2
-------
Table B-l shows results from the weighted-average modeling for the 3-day annual
minima indicators of hydrologic alteration (IHA) parameters.
Table B-l. Weighted-average indicator values for annual minima, 3-day
means
3-Day annual minima
Taxa
Pisidium
Ambrysus
Mayatrichia/Neotrichia
Neotrichia
Leuctridae
Asellidae
Lymnaea
Zapada
Neothremma
Physella
Skwala
Petrophila
Coenagrionidae
Bibiocephala
Cultus
Serratella
Dytiscidae
Pelecypoda
Hesperoperla
Epeorus
Physa
Claassenia
Podmosta
Tipula
Capniidae
Apatania
Oecetis
Optimum
0.030
0.041
0.045
0.046
0.049
0.050
0.056
0.057
0.059
0.060
0.061
0.062
0.064
0.065
0.066
0.067
0.068
0.069
0.069
0.070
0.071
0.072
0.072
0.072
0.073
0.073
0.073
Tolerance
0.04
0.05
0.03
0.04
0.03
0.06
0.04
0.04
0.04
0.06
0.02
0.05
0.07
0.01
0.04
0.04
0.04
0.06
0.05
0.04
0.06
0.03
0.03
0.05
0.05
0.02
0.04
Rank_opt
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
Rank tol
2
o
J
2
2
1
4
3
3
3
5
1
4
6
1
o
J
2
2
5
4
2
5
1
1
4
4
1
2
Count
16
17
16
12
24
45
15
35
19
13
31
36
36
17
20
11
10
44
33
92
54
12
10
31
38
20
45
B-3
-------
Table B-l. Weighted-average indicator values for annual minima, 3-day means
(continued)
3-Day annual minima
Taxa
Baetidae
Heptagenia
Pteronarcella
Ephemerella
Chloroperlidae
Hemerodromia
Antocha
Ostracoda
Lepidostoma
Paraleptophlebia
Arctopsyche
Rhithrogena
Simuliidae
Chelifera
Isoperla
Cheumatopsyche
Rhyacophilidae
Cinygmula
Optioservus
Glossosoma
Acarina
Zaitzevia
Planaria
Leptohyphidae
Ameletus
Hydroptila
Nematoda
Hexatoma
Hydropsyche
Taenionema
Copepoda
Optimum
0.073
0.075
0.076
0.076
0.076
0.076
0.077
0.077
0.077
0.078
0.078
0.078
0.079
0.079
0.080
0.080
0.080
0.080
0.080
0.081
0.081
0.081
0.082
0.082
0.082
0.082
0.082
0.082
0.083
0.083
0.084
Tolerance
0.06
0.05
0.04
0.05
0.04
0.07
0.05
0.06
0.05
0.04
0.05
0.04
0.06
0.06
0.04
0.07
0.05
0.05
0.06
0.05
0.06
0.05
0.07
0.07
0.05
0.06
0.06
0.03
0.06
0.04
0.07
Rank_opt
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
Rank tol
6
4
2
4
2
6
o
J
5
4
2
3
3
5
5
3
6
4
o
J
5
4
5
4
7
6
4
6
6
2
5
o
J
6
Count
277
58
91
149
105
103
126
96
88
96
99
127
234
98
105
55
98
90
148
60
268
97
90
133
26
97
125
88
232
29
35
B-4
-------
Table B-l. Weighted-average indicator values for annual minima, 3-day means
(continued)
3-Day annual minima
Taxa
Microcylloepus
Leucotrichia
Chironomidae
Euparyphus
Isogenoides
Drunella
Dicranota
Tubificidae
Pteronarcys
Atherix
Planorbidae
Alisotrichia/Leucotricia
Micrasema
Brachycentms
Hirudinea
Oligophlebodes
Forcipomyia/Probezzia
Agapetus/Culoptila/Protoptila
Pericoma
Bezzia
Helicopsyche
Hyalella
Traverella
Hesperophylax
Gammams
Optimum
0.085
0.085
0.085
0.086
0.086
0.087
0.089
0.090
0.090
0.091
0.091
0.091
0.092
0.093
0.094
0.094
0.094
0.097
0.100
0.103
0.110
0.111
0.116
0.159
0.170
Tolerance
0.04
0.06
0.07
0.10
0.04
0.05
0.05
0.06
0.03
0.05
0.08
0.06
0.05
0.06
0.09
0.05
0.08
0.03
0.07
0.08
0.08
0.09
0.03
0.08
0.07
Rank_opt
5
5
5
5
6
6
6
6
6
6
6
6
6
6
6
6
7
7
7
7
7
7
7
7
7
Rank tol
3
5
6
7
2
4
4
5
1
4
7
6
4
5
7
4
7
1
6
7
7
7
1
7
6
Count
10
23
291
12
19
119
32
107
27
81
37
32
55
145
75
35
20
12
47
53
68
62
10
12
15
B-5
-------
Figures B-l to B-4 show the ordination plots from the Nonmetric Multidimensional
scaling (NMDS) and canonical correlation analysis (CCA).
Utah I HA
1980
1990
2000
Year
Axis 1
r= .313 tau = .197
Axis 2
r = -.566 tau = -.405
2000 -
1990 •
1980
Season
A1
A2
T3
T4
Axis 1
Figure B-2. Taxonomical trends in the Utah data set were examined using
NMDS. Year had the strongest influence on taxonomical composition. However,
when NMDS ordinations were run on a selected subset of data that only contained
data from sites with multiple years of samples, the year trend was not as strong.
-------
Utah IHA
CM
U)
PSroph
+•
+
OOTioptef
Sa-rald Ostraood
Capnida +
"*" ^Qr^grrul AmeWus
+•
t-
-i- Aflajtfys
Me
+
ft)dmosla
Cimgma Polyeant
-t- " +-
I
+
»«"
-(- Na=icps» PfysOa
CoriKida
+
OigopH
•f
Brachyce
t-
Foropcm
+ + „.
PlffOfiar T
Axis 1
AsHUdJ
-I- .
"op-, Potncp,
Figure B-3. Species trends along year. These were derived from the CCA
analysis.
B-7
-------
SPECIES
A
ENV. VARIABLES
CCA - Utah
iaevapex
A
Hipulse
Siilis
Bit
Amiuc&ftcra
PisuSum
Calapary typfitifqft
Culicoid ,,&aj,jB$Btcta'a „
Ofalvtxa A Patamopy
Fossma oaanK
Afarmna
-1.0
1.0
Figure B-4. CCA plot of a selected subset of the Utah biological-hydrological
data.
B.2.
Table B-2 shows a list of the Utah sites at which we ran correlation analyses.
'EXTREME' ALTERATIONS OF UTAH FALL RIVPACS MODEL
CLIMATE-RELATED PREDICTOR VARIABLE VALUES
We also ran some 'extreme' scenarios (i.e., doubling temperature, dividing precipitation
values by two, changing freeze dates by 30 days, etc.) to explore how much the climate-related
predictor variables would have to change in order to result in substantial changes to
observed/expected (O/E) scores. Tables B-3 through B-8 show which scenarios were run and
what the results were.
-------
Table B-2. Data that were used in the Utah correlation analyses were gathered from these biological sampling
stations/USGS gages. %URB = % urban, %AGR = % agricultural and %FOR = % forested land use within a
1-km buffer of the sites
BioStationID
4926350
4934100
4937900
4954380
4996690
4998400
5940440
USGS gage
10131000
9302000
9261000
9330000
10163000
10154200
10234500
# Years of data
14
12
14
19
17
18
11
Elevjt
5,573.3
4,762.6
4,766.1
6,940.5
4,521.3
6,971.4
6,249.3
Eco_L3
Wasatch and Uinta
Mountains
Colorado Plateaus
Colorado Plateaus
Wasatch and Uinta
Mountains
Central Basin and Range
Wasatch and Uinta
Mountains
Wasatch and Uinta
Mountains
Eco_L4
Mountain Valleys
Uinta Basin Floor
Uinta Basin Floor
Semiarid Foothills
Moist Wasatch Front
Footslopes
Mid-elevation Uinta
Mountains
Semiarid Foothills
Ref status
TRASH
UNKNOWN
SO-SO
TRASH
TRASH
SO-SO
REF
%URB
32.5
3.9
0
6.9
73.2
5.7
3.9
%AGR
27.9
18.4
20.3
30.3
15.8
0.7
0
%FOR
30.2
24
65.1
56
5.3
93.6
96.1
td
-------
Table B-3. Descriptions of how the climate-related predictor variables were altered in the 'extreme alteration'
RIVPACS analyses
Run#
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Category
Baseline
Temperature
Precipitation
Temperature and
precipitation
Freeze date
Altered predictor variables
None — used original values
TMEAN.WS + 2 and TMEAN.NET + 2
TMEAN.WS + 4 and TMEAN.NET + 4
TMEAN.WS + 10 and TMEAN.NET + 10
TMEAN.WS + 20 and TMEAN.NET + 20
MEANP.PT - 0.05
MEANP.PT-0.1
MEANP.PT - Minimum PRISM ppt!4
MEANP.PT/2
MINP.PT/2
MEANP.PT/2 and MINP.PT/2
MINWD.WS/2
TMEAN.WS + 2 and TMEAN.NET + 2 and MEANP.PT - 0.05
TMEAN.WS + 4 and TMEAN.NET + 4 and MEANP.PT -0.1
LST32AVE - 2
LST32AVE - 5
FST32AVE + 5
LST32AVE - 5 and FST32AVE + 5
LST32AVE - 10
FST32AVE + 10
LST32AVE - 10 and FST32AVE + 10
LST32AVE - 15
LST32AVE - 15 and FST32AVE + 15
Rationale
Get baseline values and quality control
National Center for Atmospheric Research (NCAR) annual
temperature predictions (2050)
NCAR annual temperature predictions (2090)
Curiosity
Curiosity
NCAR annual precipitation predictions (2050)
NCAR annual precipitation predictions (2090)
Based on Parameter- elevation Regressions on Independent Slopes
Model (PRISM) ppt!4 minimum values (1975-2006)
Curiosity
Curiosity
Curiosity
Curiosity
NCAR annual temperature and precipitation predictions (2050)
NCAR annual temperature and precipitation predictions (2090)
Best professional judgment
Best professional judgment
Best professional judgment
Best professional judgment
Curiosity
Curiosity
Curiosity
Curiosity
Curiosity
td
o
-------
Table B-3. Descriptions of how the climate-related predictor variables were altered in the 'extreme alteration'
RIVPACS analyses (continued)
Run#
24
25
Category
Combine all
Altered Predictor variables
LST32AVE - 1, MINP.PT - 1, MEANP.PT - 1,
TMEAN.NET + 1, TMEAN.WS + 1, FST32AVE + 1,
MINWD.WS - 1
LST32AVE - 2, MINP.PT - 2, MEANP.PT - 2,
TMEAN.NET + 2, TMEAN.WS + 2, FST32AVE + 2,
MINWD.WS - 1
Rationale
Best professional judgment
Best professional judgment
td
-------
Table B-4. Results for the scenarios in which temperature predictor
variables were altered
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
TMEAN.WS + 2 and
TMEAN.NET + 2
O
14
10
15
8
E
14.92
9.56
14
8.74
O/E
0.94
1.05
1.07
0.92
Dif ce O/E
0.01
0
0
0
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
0
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
TMEAN.WS + 4 and
TMEAN.NET + 4
0
14
10
15
7
E
14.8
9.6
14
8.25
O/E
0.95
1.04
1.07
0.85
Dif ce O/E
0.02
0
0
-0.07
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
TMEAN.WS + 10 and
TMEAN.NET + 10
O
14
10
15
7
E
14.65
9.61
13.89
8.24
O/E
0.96
1.04
1.08
0.85
Dif ce O/E
0.03
0
0.01
-0.07
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
0
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
TMEAN.WS + 20 and
TMEAN.NET + 20
0
13
10
15
7
E
14.08
9.63
13.44
8.24
O/E
0.92
1.04
1.12
0.85
Dif ce O/E
0
-0.01
0.05
-0.07
B-12
-------
Table B-5. Results for the scenarios in which precipitation predictor
variables were altered
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
MEANP.PT - 0.05
O
14
10
15
8
E
15.1
9.59
14
8.75
O/E
0.93
1.04
1.07
0.91
Dif ce O/E
0
0
0
0
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
MEANP.PT - 0.1
O
14
10
15
8
E
15.08
9.58
14.01
8.74
O/E
0.93
1.04
1.07
0.92
Dif ce O/E
0
0
0
0
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
0
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
MEANP.PT - Min ppt!4
PRISM
0
14
10
15
8
E
14.78
9.51
13.79
8.71
O/E
0.95
1.05
1.09
0.92
Dif ce O/E
0.02
0.01
0.02
0
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
MEANP.PT/2
O
14
10
15
8
E
14.79
9.43
13.8
8.68
O/E
0.95
1.06
1.09
0.92
Dif ce O/E
0.02
0.02
0.02
0.01
B-13
-------
Table B-5. Results for the scenarios in which precipitation predictor
variables were altered (continued)
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
0
14
10
15
8
E
15.09
9.58
14.04
8.74
0/E
0.93
1.04
1.07
0.92
MINP.PT/2
0
13
10
15
8
E
13.92
9.46
13.58
8.69
O/E
0.93
1.06
1.1
0.92
Dif ce O/E
0.01
0.01
0.04
0.01
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
MEANP.PT/2 and
MINP.PT/2
O
13
10
15
8
E
13.69
9.33
13.38
8.16
O/E
0.95
1.07
1.12
0.98
Dif ce O/E
0.02
0.03
0.05
0.07
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
0
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
MINWD.WS/2
0
13
10
15
7
E
13.81
9.53
13.47
7.63
O/E
0.94
1.05
1.11
0.92
Dif ce O/E
0.01
0.01
0.05
0
B-14
-------
Table B-6. Results for the scenarios in which both temperature and
precipitation predictor variables were altered
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
TMEAN.WS + 2 and TMEAN.NET + 2
and MEANP.PT - 0.05
O
14
10
15
7
E
14.93
9.56
14.01
8.24
O/E
0.94
1.05
1.07
0.85
Dif ce O/E
0.01
0
0
-0.07
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
0
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
TMEAN.WS + 4 and TMEAN.NET + 4
and MEANP.PT - 0.1
0
14
10
15
7
E
14.83
9.58
14.02
8.26
O/E
0.94
1.04
1.07
0.85
Dif ce O/E
0.02
0
0
-0.07
B-15
-------
Table B-7. Results for the scenarios in which freeze date predictor variables
were altered
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
0
14
10
15
8
E
15.09
9.58
14.04
8.74
0/E
0.93
1.04
1.07
0.92
LST32AVE - 2
0
14
10
15
7
E
15.05
9.58
14.01
8.25
0/E
0.93
1.04
1.07
0.85
Dif ce O/E
0
0
0
-0.07
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
0
14
10
15
8
E
15.09
9.58
14.04
8.74
0/E
0.93
1.04
1.07
0.92
LST32AVE - 5
0
14
10
15
7
E
14.733
9.5648
13.999
8.2433
0/E
0.95
1.05
1.07
0.85
Dif ce O/E
0.02
0
0
-0.07
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
FST32AVE+5
O
15
10
15
8
E
15.374
9.5875
14.028
8.7184
O/E
0.98
1.04
1.07
0.92
Dif ce O/E
0.05
0
0
0
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
LST32AVE - 5 and
FST32AVE+5
O
13
10
15
7
E
14.128
9.5647
13.992
8.224
O/E
0.92
1.05
1.07
0.85
Dif ce O/E
-0.01
0
0
-0.06
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
LST32AVE - 10
O
13
10
15
7
E
14.02
9.56
13.7
8.23
O/E
0.93
1.05
1.09
0.85
Dif ce O/E
0
0
0.03
-0.07
B-16
-------
Table B-7. Results for the scenarios in which freeze date predictor variables
were altered (continued)
Baseline (original)
Group Site Sample O E O/E
1 5940440 127636 14 15.09 0.93
7 4951200 120184 10 9.58 1.04
1 4936750 118524 15 14.04 1.07
6 4927250 127718 8 8.74 0.92
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
FST32AVE + 10
O E O/E Dif ce O/E
14 14.713 0.95 0.02
10 9.6097 1.04 0
15 13.797 1.09 0.02
7 8.1843 0.86 -0.06
LST32AVE - 10 and
FST32AVE + 10
O
13
10
15
7
E
13.743
9.6115
13.532
8.1706
O/E
0.95
1.04
1.11
0.86
Dif ce O/E
0.02
0
0.04
-0.06
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
0
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
LST32AVE - 15
0
13
10
15
7
E
13.945
9.5818
13.454
8.2214
O/E
0.93
1.04
1.11
0.85
Dif ce O/E
0
0
0.05
-0.06
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
0
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
LST32AVE - 15 and
FST32AVE + 15
0
13
10
14
7
E
13.415
9.6052
12.787
8.1713
O/E
0.97
1.04
1.09
0.86
Dif ce O/E
0.04
0
0.03
-0.06
B-17
-------
Table B-8. Results for scenarios in which combinations of all climate-related
predictor variables were altered simultaneously
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
Changed by 1
O
13
10
15
8
E
14.04
9.51
14.03
8.71
O/E
0.93
1.05
1.07
0.92
Dif ce O/E
0
0.01
0
0
Group
1
7
1
6
Site
5940440
4951200
4936750
4927250
Sample
127636
120184
118524
127718
Baseline (original)
O
14
10
15
8
E
15.09
9.58
14.04
8.74
O/E
0.93
1.04
1.07
0.92
Changed by 2
O
13
10
15
7
E
13.81
9.49
14.03
8.23
O/E
0.94
1.05
1.07
0.85
Dif ce O/E
0.01
0.01
0
-0.06
B-18
-------
APPENDIX C
MAINE DECISION MODEL AND ANALYSES ON COMPONENT METRICS
C-l
-------
C.I. OVERVIEW OF MAINE'S DEPARTMENT OF ENVIRONMENTAL
PROTECTION (ME DEP) AQUATIC LIFE DECISION MODELS AND SAMPLE
VARIABLES (PROVIDED BY MAINE DEP)
ME DEP's aquatic life decision models are four statistical models that use 30 variables of
the macroinvertebrate community to determine the strength of association of a sample
community to Maine's water quality classes. Each of the four linear discriminant models uses
different variables, providing independent estimates of class membership. Association values
are computed for each classification using one 4-way model and three 2-way models. The
protocol is outlined in the ME DEP methods manual (Davies and Tsomides, 2002).
C.I.I. First-Stage Model and Variables
The first-stage model acts as a screen and gives the strength of association of the sample
to each of the different water quality classes. This model provides four initial probabilities that a
given site attains one of three classes (A, B, or C) or is in nonattainment (NA) of the minimum
criteria for any class. These probabilities have a possible range from 0.0 to 1.0 and, after
transformation, they are used as variables in each of the three subsequent second-stage or final
decision models. See the section below on second-stage models.
The variables used in the first-stage model are variables important to the evaluation of all
classes. Of the nine variables used in the first modeling stage, five measure abundance,
two measure richness, and two variables are biotic indices involving tolerance to pollution and
abundance. The first-stage model uses the following nine variables:
1. Total Mean Abundance—Total mean abundance (the mean number of individuals in a
sample, usually based on 3 replicates) is a basic measure of community structure and is a
strong predictor of both Class A and nonattainment. Total abundance values for the
water quality classes appear to follow a curve shaped like the Odum et al. (1979)
subsidy-stress gradient. Values for Class A are relatively low, due to low nutrients in
natural Maine waters. Values for Class B and C communities tend to be high, indicating
increased resources that might be available in a waterbody with increased loadings of
materials from human alterations. Abundance values in nonattainment waters tend to be
low but can also be highly variable.
2. Generic Richness—Richness (total number of taxa in a sample) is a good measure of
water quality impact, declining as water quality declines. Low richness is a good
predictor of nonattainment. Like abundance, richness follows the generalized
subsidy-stress curve.
C-2
-------
3. Plecoptera Mean Abundance—Plecoptera, or stoneflies, are very intolerant of even
mild levels of pollution. Abundance is highest for Class A and declines with the classes
to be nearly absent from the nonattainment class. The Maine water quality classification
requires that Class A and B waters support all indigenous species, so it is expected that
Plecoptera numbers will be maintained in those classes. Stoneflies function as predators
and shredders.
4. Ephemeroptera Mean Abundance—Ephemeroptera, or mayflies, are intolerant of
many pollutants, so abundances are distinctly lower for nonattainment samples than the
other classes. Mayflies function as scrapers and collectors. Together with the stoneflies,
these two groups represent highly sensitive orders that fulfill the major functional feeding
roles in the community. These orders are important components of a Class A or B
community.
5. Shannon-Wiener Generic Diversity (Shannon and Wiener, 1963)—Diversity is
composed of a richness factor and an eveness factor. Richness distributes between the
classes along a subsidy-stress curve. Diversity shows a decline in value from Class A to
the nonattainment class as certain pollution-tolerant taxa gain advantages, due to
increasing pollution load or other activities. As both diversity and richness decline, the
stability of most natural communities usually declines.
6. Hilsenhoff Biotic Index (Hilsenhoff, 1987)—The biotic index provides a measure of the
general tolerance level of the sample community toward organic (nutrient) enrichment.
The index increases in value from Class A to the nonattainment class, indicating that
increases in abundance may be attributable to increases among the tolerant taxa (a change
allowed in Class B or C), or that there may be a decline in the taxa pool of intolerant
organisms (a change allowed in Class C).
7. Relative Chironomidae Abundance—Chironomidae, a Family of flies in the Order
Diptera that includes Nonbiting Midges and Midges, consist of a great number of taxa
with wide-ranging tolerances and adaptations. Many tend to increase with increasing
pollution load, probably as a response to reduced competition and predation, and to
increased organic matter supply. Many have very short generation times and are, thus,
capable of quickly colonizing areas where these conditions exist. The taxa that cause
these increases are the collector types adapted to feeding on fine organic matter; some are
primarily predators. These genera have been observed to increase in relative abundance
presumably because of tolerance to reduced water quality, particularly the presence of
some toxic substances, and the availability of other pollution tolerant prey.
8. Relative Diptera Richness—Many Diptera, or true flies, are pollution tolerant
organisms. Relative Diptera richness increases from Class A to the nonattainment class.
Increases in Diptera, particularly Chironomidae, have been observed with increasing
pollution and sedimentation and loss of Ephemeroptera, Plecoptera, and Trichoptera.
9. Hydropsyche Mean Abundance—The genus Hydropsyche., one of the common
net-spinning Caddisflies, provides some added discrimination to the model. Higher
values for Hydropsyche abundance are found for Class B and are nearly absent from
C-3
-------
nonattainment samples. Hydropsyche is a filter feeder and prospers under conditions of
mild enrichment of suspended organic particles, conditions that might naturally be found
below a lake outlet or might be found in Class B waters below a treatment plant outfall or
in the presence of nutrient enrichment from nonpoint source pollution activities (e.g.,
agriculture). Relative to other genera of the Hydropsychidae family, Hydropsyche is
usually less tolerant of low dissolved oxygen or toxic substances.
C.1.2. Second-Stage Models and Variables
The final decision models (the three, two-way models) are designed to distinguish
between a given class and any higher classes as one group and any lower classes as another
group (e.g., Classes A + B + C vs. NA; Classes A + B vs. Class C + NA; Class A vs.
Classes B + C + NA). The equations for the final decision models use the predictor variables
relevant to the class being tested. The process of determining attainment class using the
association value is outlined in Appendix F of the ME DEP methods manual (Davies and
Tsomides, 2002). Application of the three second-stage models or two-group tests is
hierarchical:
"C or better" model: The first second-stage model determines the probability that an
unknown sample belongs in the cluster of samples A + B + C versus the probability that it
belongs in the cluster of nonattainment of Class C samples. This is referred to as the "C or
better" model, which determines if the sample is at least a Class C, using the following variables:
1. Probability (A + B + C) from First-stage Model
2. Cheumatopsyche Mean Abundance—The abundance of Cheumatopsyche, one of the
common net-spinning Caddisflies, generally increases with declining water quality and is
usually the last of the Ephemeroptera-Plecoptera-Trichoptera genera found in abundance
as water quality declines because Cheumatopsyche are generally found to be the most
pollution tolerant genera within the family Hydropsychidae, among the order Trichoptera.
3 Ephemeroptera, Plecoptera, Trichoptera (EPT)—Diptera Richness Ratio—(uses all
Diptera rather than just the Chironomidae.). Ephemeroptera-Plecoptera-Trichoptera are
usually poorly represented in communities where water quality is poor. These orders
provide considerable functional variety to aquatic communities, and when severely
depleted, or replaced by Diptera, signal dysfunction of the community. Maine data show
distinct separation of values for this variable between Class A, B, and C communities and
the nonattainment communities.
C-4
-------
4. Relative Oligochaeta Abundance—Proliferation of Oligochaeta, aquatic worms, has
long been recognized as an indication of polluted waters, because many taxa are highly
tolerant of low oxygen conditions and certain toxic substances, feed on fine organic
particles and can colonize quickly in the absence of predators. Communities dominated
by Oligochaeta are found when pollution loads are excessive. These organisms are
usually the last to be eliminated by pollutant overloading and as the relative abundance of
Oligochaeta increases, community structure, and function are usually diminished.
"B or better" model: The second two-way model is the "B or better" model, which
determines if the unknown sample attains at least Class B standards. It discriminates between
the cluster of A + B samples and the cluster of C + nonattainment of Class C samples. Family
functional groups are important in this second two-way model. Changes in functional feeding
group composition indicate the energy pathways through the aquatic ecosystem have been
significantly altered. The major functional groups in the Maine data are as follows:
collector-filterer, collector-gatherer, piercer, predator, scraper, and shredder. The "B or better"
model uses the following variables:
1. Probability (A + B) from First-stage Model
2. Perlidae Mean Abundance (Family Functional Group)—Greater abundance of this
family functional group is expected to occur in higher quality waters. This family of
stoneflies encompasses large predators and usually occurs in waters of good quality.
Generation time for some of these taxa is greater than 1 year; therefore, populations will
persist only where water quality is consistently good for long periods of time.
3. Tanypodinae Mean Abundance (Family Functional Group)—This subfamily functional
group is usually found in greater abundance in waters of lower quality. This
Chironomidae subfamily is also a predator group, but these organisms are small in
comparison to the Perlidae, and feed on small Oligochaeta and other Chironomidae that
can also tolerate lower water quality.
4. Chironomini Mean Abundance (Family Functional Group)—Greater abundance of this
Chironomidae subfamily group indicates increased availability of organic matter. Many
taxa in this group are known to tolerate lower water quality. These organisms are
collector-gatherers favoring fine, settled organic particles. Many of these taxa are
multivoltine, capable of quickly colonizing favorable habitats and recolonizing after
disturbances.
5. Relative Ephemeroptera Abundance—The Ephemeroptera, or mayflies, are generally
an intolerant order and tend to be indicators of good to excellent water quality. While
total Ephemeroptera abundance was used as a discriminating variable in the second-stage
C-5
-------
discriminant model to separate the four classes, relative abundance is used to separate
these two groups, particularly between the Class B and C waters. While Ephemeroptera
abundance may not decline appreciably in Class C waters, there is an expectation for
other non-Ephemeroptera taxa to increase.
EPT Generic Richness—EPT richness has been a common measure to identify waters of
good quality. Of the three orders, Ephemeroptera and Plecoptera are considered the more
intolerant. Many of the Trichoptera are also intolerant of low water quality.
Collectively, these orders have a wide array of functional characteristics (feeding
strategies and preferred resources, reproductive and life cycle strategies, habitat
preferences). Higher values for EPT richness are indicative of a structurally and
functionally diverse community. As EPT richness diminishes, it is presumed that this
functional diversity also declines.
Sum of Mean Abundances of Dicrotendipes, Micropsectra, Parachironomus, and
Helobdella—The sum of the abundance of four indicator taxa (three Chironomidae
genera and one leech genera) is also used. All four are detritivores and generally occur in
abundance only when water quality is diminished. A high abundance of this group is
indicative of conditions of Class C or nonattainment.
"Class A" model: Class A is the highest quality water and is expected to be supportive of
natural populations with the expectation that the community include many pollutant-intolerant
organisms. The Class A decision model relies on the probability score from the second-stage
linear discriminant function and many indicator taxa to ascertain Class A quality. The third
two-way model is the "Class A" model and discriminates Class A samples from the cluster of
samples in Classes B + C + Nonattainment of Class C using the following variables:
1. Probability of Class A from First-stage Model
2. Relative Plecoptera Richness—Plecoptera are well known as an intolerant order,
showing great intolerance to a variety of pollutants. Their reproductive strategies render
them slow to recolonize areas where they have been eliminated. Water quality, therefore,
needs to be consistently good for the Plecoptera to be present. Relative richness of
Plecoptera is expected to be greatest in the highest quality waters.
3. Sum of Mean Abundance of Cheumatopsyche, Cricotopus, Tanytarsus, and
Ablabesmyia—These four taxa (a net-spinning Caddisfly and three Chironomidae
genera) are considered pollution-tolerant and are not expected to occur in abundance in
Class A waters. All four taxa occur most commonly in lower quality waters and may
replace functions of less tolerant organisms when those populations decline.
C-6
-------
4. Sum of Mean Abundances ofAcroneuria and Stenonema—Acroneuria (a stonefly
genera of the Perlidae Family) and Stenonema (a mayfly genera) are two of the most
common and abundant taxa in their respective orders and indicators of good water
quality. The sum of their abundance provides a good discriminating variable.
5. Ratio of Ephemeroptera and Plecoptera (EP) Generic Richness—EPT richness is a
good discriminating variable to identify Class A and B waters, but of this group, the
Ephemeroptera and Plecoptera were usually the less tolerant taxa of the three orders.
EPT richness is, thus, used as a variable for Class A waters.
6. Ratio of Class A Indicator Taxa—The number of Class A indicators divided by 7
(which is the total number possible). Seven indicator taxa were identified for Class A
communities. Class A indicator taxa were present in 100% of Class A communities,
<26% of Class B communities, <16% of Class C communities, and <1% of
nonattainment communities. Class A indicator taxa were rarely found to be dominant
taxa except in Class A communities. Values of zero for this variable (# of Class A
indicator taxa among 5 most dominant taxa) were found in sample communities that were
not determined to support Class A conditions. Class A communities had one or more
indicator taxa among the five most dominant taxa for 54% of the samples. The Class A
indicators are Brachycentrus (Trichoptera: Brachycentridae), Serratella (Ephemeroptera:
Ephemerellidae), Leucrocuta (Ephemeroptera: Heptageniidae), Glossosoma (Trichoptera:
Glossosomatidae), Paragnetina (Plecoptera: Perlidae), Eurylophella (Ephemeroptera:
Ephemerellidae), and Psilotreta (Trichoptera: Odontoceridae).
Figure C-l shows a flow chart that depicts Maine DEP's decision criteria. The protocol
is also outlined in the Maine DEP methods manual (i.e., Davies and Tsomides, 2002).
C-7
-------
Process for Determining Attainment Class Using Association Values
Is the sample appropriate for LDM?
YES
C
3PJ
J>
Is the sample class C or better?
Ai least C
1
4
Is the sample class B or better?
!
+
A1 least B ~] ( AtleasiB Y
•
1
ide:ermicjiE /
1
c ;
1
:' c
i
trLe
:ample cla^ A?
I
Figure C-l. Flow chart that outlines the process that Maine DEP uses for
determining attainment class using association values from its four linear
discriminant models (chart by Thomas J. Danielson, taken from ME DEP
2002 monitoring manual).
-------
C.2. BOX PLOTS SHOWING THE DISTRIBUTIONS OF THE MODEL INPUT
METRICS ACROSS THE DIFFERENT CLASSIFICATION GROUPS
Figures C-2 through C-24 show categorical box-and-whisker plots showing distributions
of mean model input metric values across the classification groups based on a data set composed
of rock-basket or rock-cone samples collected during Maine DEP's July-September index
period.
Maine Model Input Metrics
1400
1200
1000
< 300
_i
O
5 600
400
200
T
B C
Class
NA
n Mean
D MeaniSE
I Mean±1.96*SE
Figure C-2. Differences in total taxa abundance by class showing mean and
standard error (SE).
C-9
-------
Maine Model Input Metrics
44
42
40
CO
CO
z M
rr 36
o
CC ^A
LLI •"
LU
0 32
OJ
0
28
26
-T -r-
^ L^
L^
h
IE
D
in
-
T
D
-
A B C NA
n Mean
D MeaniSE
I Mean±1.96*SE
Class
Figure C-3. Differences in richness of genera by class.
Maine Model Input Metrics
24
22
LJJ -,q
0 15
-------
Maine Model Input Metrics
BUNDANCE
•0.
o:
O
LJJ
1
HI
4
o
160
140
120
100
80
fin
An
on
~r
D
X
I D
~L
c3&
B C
Class
NA
n Mean
0 Mean±SE
I Mean±1.96*SE
Figure C-5. Differences in Ephemeroptera abundance by class.
Maine Model Input Metrics
4.0
w
cc
w 3.6
D
O
o: 3.2
LU
3.0
< 2.8
CO
2.6
2.4
NA
n Mean
D MeantSE
I Mean±1.96*SE
Class
Figure C-6. Differences in Shannon-Wiener diversity of genera by class.
C-ll
-------
Maine Model Input Metrics
6.2
6.0
3iO ' O
5.6
X
LU 54
'l
o 5'2 L
O 5.0
m
t 4.8
O
I 4.6
LU
co 44
X
to 42
o
4.0
3.S
3.6 ,
n Mean
3-4 ' ' ' ' D Mean±SE
A K NA lMean±1.96*SE
Class
Figure C-7. Differences in Hilsenhoff Biotic Index by class.
Maine Model Input Metrics
0.38
LU _._
O
< 0.3
Q
z:
| 0.32
LU
- 030 -L
° 1
£ 0.2
I
O
^ 0.26
D
O
0.22
0.20
Mean
J Mean±SE
NA lMean±1.96*SE
Class
Figure C-8. Differences in relative Chironomid abundance by class.
C-12
-------
Maine Model Input Metrics
0.43
0.42
co 0.41
CO
LJJ
m °'39
w °-38
5 °-37
cr
oS
- 0.36
0.35
0.34
T
B C
Class
NA
° Mean
D Mean+SE
I Mean±1.96*SE
Figure C-9. Differences in relative Diptera richness by class.
Maine Model Input Metrics
450
400
LU 35°
O
z
Q 300
z
CD
250
O
to 200
a
o
a:
9 150
100
50
0
NA
n Mean
D MeaniSE
I Mean±1.96*SE
Class
Figure C-10. Differences in Hydropsyche abundance by class.
C-13
-------
Maine Model Input Metrics
220
200
LU
- 180
Q
Z>
<
LU 140
I
O
a
O -inn
2
D 80
m
9 60
AC\
n
_
-
l -g- 1
T
T
T
D
I
1
D
1
° Mean
B C
Class
NA lMean±1.96*SE
Figure C-ll. Differences in Cheumatopsyche abundance by class.
Maine Model Input Metrics
LU 1.6
Q.
D
% 1.4
LU
Q
Q co -I o
«w
w ty
^x
i y
o o: 1.0
E
o
K
^ 0.3
Q.
Ill
0.6
0.4
B C
Class
NA
n Mean
0 Mean±SE
lMean±1.96*SE
Figure C-12. Differences in EPT richness over diptera richness by class.
C-14
-------
Maine Model Input Metrics
0-M
LU
O
^
>
H
tr
0.02
n nn
n
- ^ ^ ^
B C
Class
NA
n Mean
D MeaniSE
lMean±1.96*SE
Figure C-13. Differences in relative Oligochete abundance by class.
Maine Model Input Metrics
14
12
10
LU
O
8
CD
LLl
Q_
A
-2
B C
Class
Mean
D Mean+SE
NA lMean±1.96*SE
Figure C-14. Differences in Perlidae abundance by class.
C-15
-------
Maine Model Input Metrics
LU
LU
32
30
28
26
24
22
20
18
S 16
Q_
12
10
s
6
4
I
B C
Class
NA
° Mean
0 Mean±SE
I Mean±1.96*SE
Figure C-15. Differences in Tanypodinae abundance by class.
Maine Model Input Metrics
160
140
uj 120
O
z
i 100
DO
5 80
O
1 60
S
- 40
20
T
T
T
B C
Class
n Mean
0 Mean±SE
NA lMean±1.96*SE
Figure C-16. Differences in Chironomid abundance by class.
C-16
-------
Maine Model Input Metrics
0
Q
-z.
CQ
<
<
cr
t
o
IT
LU
111
T
Q_
LJJ
>
P
118-RELA
0.28
n OR
n 9A
n on
01 A
fi IK
0.14
01 9
0.06
0.04
n no
n on
LJjG
_r
r°i
-L
— |—
n
1
: ^
B C
Class
NA
° Mean
0 Mean±SE
I Mean±1.96*SE
Figure C-17. Differences in relative Ephemeroptera abundance by class.
Maine Model Input Metrics
22
20
18
16
14
12
o
D£
111
U
CD
ill 10
CD
B C
Class
Mean
J Mean±SE
NA lMean±1.96*SE
Figure C-18. Differences in EPT richness by class.
C-17
-------
Maine Model Input Metrics
90
|| SO
o o
w w 70
ri < 60
si50
ir
m (^
< 2; 30
Ig2o
13 W
10
T
ABC
Class
Mean
JQMean±SE
NA lMean±1.96*SE
Figure C-19. Differences in total abundances of Dicrotendipes, Micropsectra,
Pamchironomus, and Helobdella by class.
Maine Model Input Metrics
0.09
co
CO
LU
~r n n&
o °'08
OL
<
tr
LLI n m
Q.
O
O
LU
-r1 n AC
3 u"
3 C
3' c
d 3AI1V1
Of
(N
0.04
n n^
-
D
-L-
A
i
B
n
C
[
N
:
A
•
D Mean
0 Mean±SE
lMean±1.96*SE
Class
Figure C-20. Differences in relative Plecoptera richness by class.
C-18
-------
Maine Model Input Metrics
220
**
MATOPSYC
\BLABESMX
A _1 K>
•> CO O
3 O O
LU Q
0 < 140
o =>
LU 2 120
il
g> 100
z <
1 ^ 80
< w
125-SUMME
CRICOTC
hO -£i C
D 0 0 C
T
n
^ ^ T
T
1 D
1
*
ABC
Class
NA
Mean
Mean±SE
lMean±1.96*SE
Figure C-21. Differences in total abundances of Cheumatopsyche, Cricotopus,
Tanytarsus, and Ablabesmyia by class.
Maine Model Input Metrics
60
LU
O
-z.
LU
I—
W
OL LU
Sit
feS
2^
° s
§0
il
=)
CQ
a
LU
6-SU
30
K>
0
10
-10
^
B C
Class
n Mean
J Mean±SE
NA lMean±1.96*SE
Figure C-22. Differences in total abundances of Acroneuria, Stenonema, and
Maccaffertium by class.
C-19
-------
0.9
0.7
Q 0.6
w
w
LU
i 0.5
o
CL
O
E 0.4
LU
LU
O
o. 0.3
UJ
ci
0.2
0.1
Maine Model Input Metrics
B C
Class
IMA
a Mean
D Mean±SE
lMean±1.96*SE
Figure C-23. Differences in EP richness by class.
0.42
0.40
0.38
< 0.36
P 0.34
CL
g 0.32
<
9 0.30
D
- 0.28
fe 0.26
O 0.24
1 0.22
LU
8; 0.20
§ 0.18
0.16
0.14
0.12
Maine Model Input Metrics
B C
Class
NA
P Mean
D Mean±SE
lMean±1.96*SE
Figure C-24. Differences in presence of indicator taxa by class.
C-20
-------
C.3. DISTRIBUTION OF INDICATOR TAXA BY YEARS GROUPED AS CLIMATE
SURROGATES
Figure C-25 shows indicator taxa grouped by driest-, normal-, and wettest-year samples,
while Figure C-26 shows indicator taxa grouped by lowest-, normal-, and highest-flow year
samples.
0.8
0.7
1 0.6
I 0.5
T3
0 0.4
0)
0.3
0.2
0.1
Driest Normal
Year Groupings
Wettest
n Median
D 25%-75%
I Non-Outlier Range
o Outliers
* Extremes
Figure C-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 Parameter-elevation Regressions on Independent Slopes
Model (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.
C-21
-------
0.44
042
0.40
0.38
8
| 036
0= 034
to 32
_
? 0.30
§028
025
024
022
n •?TI
U^AJ
074
i»
a. °72
3 0.70
Ui
t 068
Relative to
o o
2 8
£ 062
.C Q CQ
n J^d
A
*
T"
•
1
— r~
c
P
0
0
--
C __
T .
-|—
D
1
D
o
M
22
.':
ia
16
14
i 2°
1
12
1.0 I '
Highest Lowest
Year Groupings
I
n
T
a Median
Q 25%-75%
I Non-Outlier Range
Highest
* Extremes
Figure C-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.
C-22
-------
United States
Environmental Protection
Agency
National Center for Environmental Assessment
Office of Research and Development
Washington, DC 20460
------- |