EPA 842-R-16-001
A Practitioner's Guide to the Biological
Condition Gradient: A Framework to Describe
Incremental Change in Aquatic Ecosystems
February 2016
3 f ^^"^^
•*
-"TX'
»• 6
*1 '
A
* it
• j.
? / r" ^
-------
&EPA
A Practitioner's Guide to the Biological
Condition Gradient: A Framework to Describe
Incremental Change in Aquatic Ecosystems
Front cover sources:
1. Vermont Department of Environmental Conservation
2. Christophe Quintin, flickr
3. ©istockphoto.com
4. USEPA
5. Vermont Department of Environmental Conservation
6. ©istockphoto.com
7. USDA NRCS
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
A Practitioner's Guide to the Biological Condition
Gradient: A Framework to Describe Incremental
Change in Aquatic Ecosystems
Contact Information
For more information, questions, or comments about this document, please contact Susan Jackson,
U.S. Environmental Protection Agency, at Office of Science and Technology, Office of Water,
U.S. Environmental Protection Agency, 1200 Pennsylvania Avenue, Mail Code 4304T, Washington, DC
20460 or by email at jackson.susank@epa.gov.
Citation
USEPA. 2016. A Practitioner's Guide to the Biological Condition Gradient: A Framework to Describe
Incremental Change in Aquatic Ecosystems. EPA-842-R-16-001. U.S. Environmental Protection Agency,
Washington, DC.
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Contributing Authors
Susan Jackson, USEPA Office of Water, Office of Science and Technology, Washington, DC
Jeroen Gerritsen, PhD, Tetra Tech, Inc., Owings Mills, Maryland
Susan Davies, Midwest Biodiversity Institute, Columbus, Ohio*
Lucinda Johnson, PhD, Associate Director and Senior Research Associate, Natural
Resources Research Institute, University of Minnesota Duluth, Minnesota
Appendix B Case Study Authors
Bl. Upper Mississippi River: Development of a Biological Condition Gradient for Fish Assemblages of the
Upper Mississippi River and a "Synthetic" Historical Fish Community
Ed Rankin, Midwest Biodiversity Institute, Columbus, Ohio
B2. Narragansett Bay: Development of a Biological Condition Gradient for Estuarine Habitat Quality
Emily Shumchenia, PhD, E&C Enviroscape, LLC; Giancarlo Cicchetti, PhD, and Marguerite C. Pelletier,
PhD, USEPA Office of Research and Development, National Health and Environmental Effects Research
Laboratory-Atlantic Ecology Division, Narragansett, Rhode Island
B3. Caribbean Coral Reefs: Benchmarking a Biological Condition Gradient for Puerto Rico Coral Reefs
Patricia Bradley** and Deborah Santavy, PhD, USEPA Office of Research and Development, National
Health and Environmental Effects Research Laboratory-Gulf Ecology Division, Gulf Breeze, Florida
B4. New England: Using the Biological Condition Gradient and Fish IBI to Assess Fish Assemblage
Condition in Large Rivers
Chris Yoder, Research Director, Midwest Biodiversity Institute, Columbus, Ohio
Technical Editors
Susan B. Norton, PhD, USEPA Office of Research and Development, National Center for Environmental
Assessment, Alexandria, Virginia
Douglas Norton, USEPA Office of Water, Office of Wetlands, Oceans and Watersheds, Washington, DC
Clair Meehan, Tetra Tech, Inc., Fairfax, Virginia
Formerly with the Maine Department of Environmental Protection and state co-chair for the expert workgroup
that developed the conceptual Biological Condition Gradient Framework. See Davies and Jackson 2006.
" Patricia Bradley retired from USEPA in October 2015. Since that time, she has joined Tetra Tech, Inc.
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Acknowledgements
Thank you to the following scientists and state, territorial, county and tribal agencies for their
leadership and support with development and piloting of the biological condition gradient.*** There
were additional experts involved for each of the case examples presented in this document.
State, Territorial, and Tribal BCG Workgroup Members and Pilots (2000-2015)
Alabama Department of Environmental Management- Lisa Huff
Alabama Geological Survey - Patrick O'Neill
Arizona Department of Environmental Quality - Patti Spindler
California Department of Fish and Game -Jim Harrington
Colorado Department of Public Health and Environment - Robert McConnell, Paul Welsh
Connecticut Department of Energy and Environmental Protection - Chris Bellucci
Florida Department of Environmental Protection - Russ Frydenborg, Ellen McCarron, Nancy Ross
Idaho Department of Environmental Quality- Mike Edmondson
Kansas Department of Health and Environment - Robert Angelo, Steve Haslouer, Brett Holman
Kentucky Department for Environmental Protection -Tom VanArsdall
Maine Department of Environmental Protection - David Courtemanch, Susan Davies, Leon Tsomides,
Tom Danielson, Jeanne DeFranco
Maryland Department of the Environment - Richard Eskin, George Harmon, Matthew Stover
Maryland Department of Natural Resources - Scott Stranko
Minnesota Pollution Control Agency - Will Bouchard, Greg Gross
Mississippi Department of Environmental Quality- Leslie Barkley, Natalie Guerdon
Montana Department of Environmental Quality - Randy Apfelbeck, Rosie Sada
Montgomery County, Maryland, Department of the Environment - Kenneth Mack, Jennifer St John,
Keith Van Ness
Narragansett Bay National Estuary Program -Tom Borden, Program Director
New Jersey Department of Environmental Protection - Kevin Berry, Thomas Belton
Nevada Division of Environmental Protection - Karen Vargas
North Carolina Department of Environment and Natural Resources - David Lenat, Trish MacPherson
Ohio Environmental Protection Agency-Jeff DeShon, Dan Dudley
Ohio River Valley Water Sanitation Commission - Erich Emery
Oregon Department of Environmental Quality - Doug Drake, Rick Hafele
Pennsylvania Department of Environmental Protection - Dustin Shull, Gary Walters
Pyramid Lake Paiute Tribe - Dan Mosley
Rhode Island Department of Environmental Management - Chris Deacutis
Texas Commission on Environmental Quality- Mark Fisher, Charles Bayer, William Harrison, Ann Rogers
Upper Mississippi River Basin Association - David Hokanson, Deputy Director
Vermont Department of Environmental Conservation - Doug Burnham, Steve Fisk
Virginia Department of Environmental Quality - Alexander Barren, Larry Willis
Washington State Department of Ecology- Robert Plotnikoff
Wisconsin Department of Natural Resources-Joseph Ball, Edward Emmons, Robert Masnado, Greg
Searle, Michael Talbot, Lizhu Wang
Affiliations of each individual are those at the time of participation in the BCG development and/or pilots. Since
that time, a number of these individuals have changed affiliations or have retired.
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
U.S. Environmental Protection Agency
Office of Water: Chris Faulkner, Thomas Gardner, Susan Holdsworth, Susan Jackson, Kellie Kubena,
Douglas Norton, Christine Ruff, Robert Shippen, Treda Grayson, William Swietlik, Lester Yuan
Regional (R) Offices: Peter Nolan, Margherita Pryor, Hilary Snook, Diane Switzer (Rl), Jim Kurtenbach,
Thomas Belton, Eveylyn Huertas, David Cuervas (R2), Maggie Passmore, Gregory Pond, Louis Reynolds
(R3), Ed Decker, Jim Harrison, David Melgard, Eve Zimmerman (R4), Ed Hammer, David Pfeifer (R5),
Philip Crocker, Charlie Howell (R6), Gary Welker (R7), Tina Laidlaw, Jill Minter (R8), Gary Wolinsky, Terry
Fleming (R9), Gretchen Hayslip (RIO)
Office of Environmental Information: Wayne Davis
Office of Research and Development: Laurie Anderson, Britta Bierwagen, Patricia Bradley, Karen
Blocksom, Susan Cormier, Giancarlo Cicchetti, William Fisher, Joseph Flotermirsch, Timothy
Gleason, Philip Larsen, Frank McCormick, Susan Norton, Marguerite Pelletier, Kenneth Rocha,
Deborah Santavy, Danielle Tillman
U.S. Geological Survey
Evan Hornig, Ken Lubinski, Caroline Rogers
Scientific Community (Academic. Non Governmental, and Private Sector)
David Allan, University of Michigan
David Braun, The Nature Conservancy
Don Charles, Philadelphia Academy of Natural Sciences at Drexel University
Jan Ciborowski, University of Windsor
Michael Barbour, Jeroen Gerritsen, Benjamin Jessup, Clair Meehan, Michael Paul,
Jennifer Stamp, Tetra Tech, Inc.
Richard Hauer, University of Montana
Charles Hawkins, Utah State University
Donald Huggins, Central Plains Center for Bioassessment
Bob Hughes, Oregon State University
Lucinda Johnson, University of Minnesota Duluth
James Karr, University of Washington
Dennis Mclntyre, Great Lakes Environmental Center
Gerald Niemi, University of Minnesota
Jan Stevenson, Michigan State University
Denice Wardrop, Pennsylvania State University
Edward Rankin and Chris Yoder, Midwest Biodiversity Institute
IV
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Disclaimer
The discussion in this document is intended solely to provide information on advancements in the field
of biological assessments and on use of biological assessments to support state water quality
management programs. The statutory provisions and the U.S. Environmental Protection Agency (EPA)
regulations described in this document contain legally binding requirements. This document is not a
regulation itself, nor does it change or substitute for those provisions or regulations. The document does
not substitute for the Clean Water Act, a National Pollutant Discharge Elimination System permit, or EPA
or state regulations applicable to permits; nor is this document a permit or regulation itself. Thus, it
does not impose legally binding requirements on EPA, states, tribes, or the regulatory community. This
document does not confer legal rights or impose legal obligations on any member of the public.
While EPA has made every effort to ensure the accuracy of the discussion in this document, the
obligations of the regulated community are determined by statutes, regulations, and other legally
binding requirements. In the event of a conflict between the discussion in this document and any
statute or regulation, this document will not be controlling.
The general descriptions provided here might not apply to a situation depending on the circumstances.
Interested parties are free to raise questions and objections about the substance of this document and
the appropriateness of the application of the information presented to a situation. EPA and other
decision makers retain the discretion to adopt approaches on a case-by-case basis that differ from those
described in this document where appropriate.
Mention of any trade names, products, or services is not and should not be interpreted as conveying
official EPA approval, endorsement, or recommendation.
This is a living document and might be revised periodically. EPA could revise this document without
public notice to reflect changes in EPA policy, guidance, and advancements in field of biological
assessments. EPA welcomes public input on this document at any time. Send comments to Susan
Jackson, Office of Science and Technology, Office of Water, U.S. Environmental Protection Agency, 1200
Pennsylvania Avenue, Mail Code 4304T, Washington, DC 20460.
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Contents
Executive Summary xviii
What is the Biological Condition Gradient? xviii
Who Will Use the Biological Condition Gradient and For What Purpose? xviii
Why Now? xix
Biological Condition Gradient Development: Decision Rules xix
The Stress Axis xx
Document Organization xx
Chapter 1. Introduction to the Biological Condition Gradient 1
1.1 Document Purpose 1
1.2 Background: When and Why? 2
1.3 The Biological Condition Gradient: Brief Overview 4
1.4 Use of the Biological Condition Gradient to Support Water Quality Standards and
Condition Assessments 7
1.4.1 Use of the Biological Condition Gradient to Support Aquatic Life Use
Assessments 8
1.4.2 Use of the Biological Condition Gradient to Define Levels of Condition 9
Chapter 2. The Biological Condition Gradient: Fundamental Concepts 11
2.1 The Scientific Foundation of the Biological Condition Gradient 12
2.2 The Biological Condition Gradient Attributes 13
2.3 The Biological Condition Gradient Levels of Biological Condition 17
2.3.1 Bringing the Biological Condition Gradient Levels and Attributes Together 21
2.4 How the Conceptual Biological Condition Gradient was Developed, Tested, and
Evaluated 27
2.5 Conclusion 30
Chapter 3. Calibration of Biological Condition Gradient Models 31
3.1 Overview 32
3.1.1 Case Studies and Applications 35
3.2 Step One: Assemble and Organize Data 36
3.2.1 Data Requirements: Understanding the Quality of the Data Set 38
3.3 Step Two: Preliminary Data Analysis and Data Preparation 43
3.3.1 Data Preparation: Characterize Stress Gradients 44
3.3.2 Data Preparation: Analyze Taxon Stressor-Response 49
3.3.3 Data Preparation: Organize Data for Expert Panel 53
3.4 Step Three: Convene an Expert Panel 55
3.4.1 Expert Panel 55
3.4.2 Assign Taxa to Attributes 56
3.4.3 Assign Sites to Condition Levels 58
3.5 Biological Condition Gradient Decision Rules 65
Chapter 4. Quantitative Rules and Decision Systems 66
4.1 Quantitative Rule Development and Application 66
4.1.1 Elicitation of Numeric Decision Criteria 67
4.1.2 Codification of Decision Criteria: Multiple Attribute Decision Criteria Approach 68
VI
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
4.2 Calibrating Indices to the Biological Condition Gradient 85
4.2.1 Biological Condition Gradient Thresholds for Multimetric Indices and
Multivariate Models 88
4.3 Statistical Models to Predict Expert Decisions: Multivariate Discriminant Model Approach 90
4.3.1 Approach 90
4.4 Automation of Decision Models 98
4.5 Conclusion 99
Chapter 5. The Generalized Stress Axis 100
5.1 The Conceptual Foundation of the Generalized Stress Axis 101
5.1.1 Technical Issues in Developing a Generalized Stress Axis 106
5.2 Development of a Generalized Stress Axis 109
5.2.1 Using Land Cover Measures as Stressor Indicators 110
5.2.2 Ranking Sites by Summing Stressor Indicators 113
5.2.3 Using Statistical Approaches to Combine Stressor Indicators 113
5.3 Linking the Science with Management Actions 115
5.4 Conclusions 119
Chapters. Case Studies 120
6.1 Montgomery County, Maryland: Using the Biological Condition Gradient to
Communicate with the Public and Inform Management Decisions 121
6.1.1 Key Message 121
6.1.2 Background: Early County Policy 121
6.1.3 Development of the Biological Condition Gradient 124
6.1.4 Use of the Biological Condition Gradient Model in County Planning Decisions 132
6.1.5 Lessons Learned 133
6.2 Pennsylvania: Using Complementary Methods to Assess Biological Condition of Streams 134
6.2.1 Key message 134
6.2.2 Using Index of Biological Integrity to Assess Aquatic Life Uses 134
6.2.3 Use of the Biological Condition Gradient to Complement Aquatic Life Use
Assessments 135
6.2.4 Potential Application to Support Aquatic Life Use Assessments and Protection
of High Quality Waters 141
6.3 Alabama: Using the Biological Condition Gradient to Communicate with the Public and
Inform Management Decisions 142
6.3.1 Key Message 142
6.3.2 Program Development 142
6.3.3 Index Development 143
6.3.4 The Biological Condition Gradient 147
6.3.5 Application of the Biological Condition Gradient to Support Aquatic Life Use
Assessments 150
6.3.6 Future Applications 152
6.3.7 Conclusion 155
6.4 Minnesota: More Precisely Defining Aquatic Life Uses and Developing Biological Criteria 156
6.4.1 Key Message 156
6.4.2 Background 156
6.4.3 Tiered Aquatic Life Uses and Biological Criteria Development 158
6.4.4 Benefits of the Biological Condition Gradient 162
VII
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.4.5 Conclusion 164
6.5 Maine: Development of Condition Classes and Biological Criteria to Support Water
Quality Management Decision Making 165
6.5.1 Key Message 165
6.5.2 Background 165
6.5.3 Maine's Numeric Biological Criteria and Tiered Aquatic Life Uses 167
6.5.4 Goal-based Management Planning to Optimize Aquatic Life Conditions 168
6.5.5 Early Detection and Management of an Emerging Problem 170
6.5.6 Monitoring and Assessment to Determine Current Condition: Using Biological
Condition Gradient Concepts to Integrate Biological Information from Multiple
Assemblages and Water Body Types 172
6.5.7 Using Maine's Tiered Aquatic Life Uses and Biological Assessment Methods to
Evaluate Wetland Condition 173
6.5.8 Conclusion 176
6.6 Ohio: Use of Biological Gradient to Support Water Quality Management 177
6.6.1 Key Message 177
6.6.2 Background 177
6.6.3 Determining Appropriate Levels of Protection 183
6.6.4 Setting Attainable Goals for Improvements 186
6.6.5 Protecting High Quality Water Bodies 188
6.6.6 Conclusion 190
References 191
Glossary 218
Abbreviations and Acronyms 225
Appendix A A-l
Appendix B B-l
VIII
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Figures
Figure 1. Stream and wadeable river 1
Figure 2. Conceptual model of the BCG. Although in reality the relationship between stressors and
their cumulative effects on the biota is likely nonlinear, the relationship is presented as
such to illustrate the concept 4
Figure 3. Model illustrating the multiple pathways through which human activities may exert
pressure on an aquatic system by altering fundamental environmental processes and
materials, creating stressors that may adversely affect the aquatic biota (Source: Modified
from figure courtesy of David Allen, University of Michigan) 6
Figure 4. Biologists conducting stream and lake assessments 11
Figure 5. Response of mayfly density to stress in Maine streams as indicated by a gradient of
increasing conductivity 19
Figure 6. Hypothetical examples of biological response to the cumulative impact of multiple
stressors 21
Figure 7. Benthic macroinvertebrate and fish experts developing decision rules for freshwater
streams in Alabama 31
Figure 8. Steps in a BCG calibration 34
Figure 9. Scatterplots of number of total taxa (upper) and number of EPTtaxa (lower) versus %
impervious surface in the macroinvertebrate data set for streams in the Northern
Piedmont of Maryland. Plots are fit with a linear trend line 45
Figure 10. Number of Plecoptera (stonefly) taxa and dissolved copper (Cu) concentration,
Connecticut sites. The screening criterion, (0.008 mg/L Cu) was estimated by eye from the
presence of stoneflies at low Cu concentrations, and their near absence above 0.008 mg/L
Cu. In the calibration, least stressed sites were required to have Cu < 0.008 mg/L (among
other criteria). The screening criterion separates sites with no detectable influence of
copper from those where copper may be a factor (among others) in loss of Plecoptera 48
Figure 11. The frequency of occurrence and abundances of attribute II, III, IV, and V taxa are
expected to follow these patterns in relation to the stressor gradient. Attribute II taxa
have a high relative abundance and high probability of occurrence in minimally-disturbed
sites. Attribute III taxa occur throughout the disturbance gradient, but with higher
probability in better sites. Attribute IV taxa also occur throughout the disturbance
gradient, but with roughly equal probability throughout, or with a peak in the middle of
the disturbance range. Attribute V taxa occur throughout the disturbance gradient, but
with higher probability of occurrence, and higher abundances, in more stressed sites 51
Figure 12. Examples of attribute II (highly sensitive), III (intermediate sensitive), IV (intermediate
tolerant), and V (tolerant) taxon-response plots for the Northern Piedmont of Maryland
and Minnesota lakes. The plots on the left show responses of four macroinvertebrate taxa
from the Northern Piedmont of Maryland to impervious surface (the x-axis is log-
transformed). The plots on the right show responses of five fish taxa from Minnesota lakes
to urban/agricultural/mining land use 52
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Figure 13. Example data table for site assessment, showing how site data may be arranged for a
panel's assessment. Attribute summary information is included at the bottom. Note that
stressor information is blank—the panel rates sites without knowledge of stressors 54
Figure 14. Decline in geographical distribution of black sandshell mussel in Kansas (after Angelo et
al. 2009) 59
Figure 15. Box plots of HDSfor Minnesota streams, grouped by nominal BCG level (panel majority
choice) for fish (upper) and macroinvertebrate (lower) samples. HDS scores range from 0
(most disturbed) to 81 (least disturbed) (Gerritsen et al. 2013) 63
Figure 16. Distribution of individual panelists BCG assignments, as deviations from group sample
median, Maryland Piedmont BCG workshop. Percentages above each bar. Data from
Stamp et al. 2014 64
Figure 17. Fuzzy set membership functions assigning linguistic values to defined ranges for Total
Taxa (top) and Sensitive Taxa (bottom). Shaded regions correspond to example rules for
BCG level 3: "Number of total taxa is high," and "number of sensitive taxa is low-moderate
to moderate." 69
Figure 18. Flow chart depicting how rules work as a logical cascade in the BCG model, from Upper
Midwest cold and coolwater example (Source: Modified from Gerritsen and Stamp 2012).
For convenience, midpoints of membership functions (50% value) only are shown. For
complete rules, see Table 15 and Table 16 72
Figure 19. Benthic Macroinvertebrate Taxa: Box plots of sensitive (attribute l+ll+lll) and tolerant
(attribute V) BCG attribute metrics, grouped by nominal BCG level (panel majority choice).
These metrics were used in the macroinvertebrate BCG model for coldwater streams in
the Upper Midwest 74
Figure 20. Connecticut MMI by BCG levels, estimated from decision analysis model. Number of
samples given below boxes 89
Figure 21. Schematic of four-way and two-way model relationships used by Maine DEP to refine
the discrimination among classes (Source: MEDEP 2014) 97
Figure 22. The five major factors that determine the biological condition of aquatic resources
(modified from Karr and Dudley 1981). Four of the five factors, flow regime, water quality,
energy source, and physical habitat structure, are the basis for the conceptual GSA as
described in this document. The fifth factor, biotic interaction, is incorporated as part of
the BCG y-axis levels and attributes 102
Figure 23. Human activities can cause disturbances in the environment that exceed the range of
natural variability, generating pressure upon an aquatic system that results in altered
environmental processes and materials, which, in turn, create stressors that adversely
impact biological condition 103
Figure 24. Hierarchical effects of disturbance. When assessing the relationship between stressors
and biological effects, one of two implicit models is assumed. Model 1—the biota at a site
are determined by the environmental covariates characteristic of the habitat. The
stressors associated with a human-related disturbance directly influence biota. Model 2—
the biota at a site are determined by the environmental characteristics of the site.
However, the stressors associated with a human-related disturbance influence both the
physical habitat structure and the biota itself. Consequently, the biological effects reflect
the combined direct effects of the stress and the disturbance-mediated habitat alteration
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
(From: Ciborowski et al. unpublished). Comprehensive and integrated monitoring data
(biological, chemical, physical) coupled with causal assessment will help distinguish direct
from indirect effects (USEPA 2013a) 105
Figure 25. Cumulative stress within the St. Louis River watershed, a tributary to Lake Superior.
Darker shading indicates increased stress. The stress score is based on the cumulative sum
of % agricultural land use, population density, road density, and point source density.
Values were each normalized to a 0-1 scale before summation. This index was used to
calibrate water quality responses to stress in the St. Louis River Area of Concern (Bartsch
et al. 2015). (Map by Tom Hollenhorst, EPA, Mid-Continent Ecology Division) 107
Figure 26. LDI applied to St. Croix watersheds and associated coral stations (Source: Oliver et al.
2011). Top figure shows land use/land cover and EPA coral reef stations. Land use/land
cover used in the analysis is shown at 2.4 m resolution. Bottom figure show the watershed
LDI values on a green- yellow-red continuum, where green indicates the lowest human
disturbance and red indicates the highest. Watershed abbreviations: Bl: Buck Island; NC:
North Central; NE: Northeast; SC: South Central; SE: Southeast; SW: Southwest; W: West 112
Figure 27. The first principal component of the agricultural variables for the U.S. Great Lakes basin.
Darker shading indicates greater amounts of agriculture (Source: Danz et al. 2005) 114
Figure 28. Flow diagram detailing the steps used by GLEI researchers in quantifying their stressor
gradient (modified from Danz et al. 2005) 115
Figure 29. The specific stressor(s) and their intensity (the BCG x-axis—termed the GSA) are created
by pressure(s) acting through specific mechanisms. BMPs can be implemented to prevent
or reduce effect on the biota through restoration, remediation, and/or mitigation 116
Figure 30. Conceptual Models (CM) A-B: Human activities can generate pressures, ultimately
producing stressors (BCG x-axis) that adversely affect the aquatic biota (BCG y-axis). CM C-
D: Implementation of a BMP can dampen the translation of pressures into the expression
of stress and reduce the adverse effects on the biota 118
Figure 31. Ten Mile Creek, Maryland 121
Figure 32. Important aquatic species in Maryland's Piedmont headwater streams. Salamanders
(Long-tailed, Northern Dusky, and Northern Red); fish (Potomac Sculpin, Rosyside Dace,
American Eel); insects (Sweltsa, Paraleptophlebia, Ephemerella) 122
Figure 33. Clarksburg Area and Ten Mile Creek Subwatershed 123
Figure 34. Box plots of sensitive (attribute ll+lll) and tolerant (attribute V) percent taxa and
percent individual metrics for macroinvertebrate calibration samples, grouped by nominal
BCG level (expert consensus) (Source: Stamp et al. 2014) 130
Figure 35. Comparative BCG ratings of macroinvertebrate community data from the county
monitoring data set for streams in the TMC watershed and comparable county streams in
other watersheds. Data from streams in the State of Maryland Piedmont Sentinel data set
were also rated by the experts. The sites were mapped on the gradient according to the
expert-derived decision rules for assigning sites to BCG levels 131
Figure 36. Relationship between average BCG level assignments (left) and % Sensitive Taxa (right)
versus % impervious cover. This analysis included sites from throughout the Piedmont
Region in Maryland. Ten Mile Creek sites are indicated (red dots) 132
Figure 37. Top: Carbaugh Run, Adams County; Bottom: Rock Run, Lycoming County (Photos
courtesy of PA DEP) 135
XI
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Figure 38. Topographic Map of Pennsylvania 136
Figure 39. Pennsylvania Land Use 136
Figure 40. Box plots of BCG metrics, by nominal level (group majority choice). Sensitive taxa are
the sum of both attribute II (highly sensitive) and attribute III taxa (intermediate sensitive)
(Source: Gerritsen and Jessup 2007c) 137
Figure 41. Comparison of calibrated BCG level assignments (mean value) and IBI scores for
freestone streams representing range of conditions from minimal to severely stressed 139
Figure 42. Multi-tiered benchmark decision process for wadeable, freestone, riffle-run streams in
Pennsylvania (Modified from PA DEP 2013a). The ratio of BCG attributes for sensitive to
tolerate taxa (i.e., attributes I, II, and III to attributes IV, V, and VI) are included as part of
attainment determination (see yellow box). Rules have not been defined for attribute I
and IV but these attributes are included in the assessment protocol if decision rules are
developed in the future and determined to be appropriate to include 140
Figure 43. Left: Macroinvertebrate site classes in Alabama; Right: Fish site classes in Alabama 144
Figure 44. Alabama land use/land cover map 144
Figure 45. Frequencies of sites in ranked WDG categories (x-axis), distinguishing reference and
non-reference sites in each site class. Distributions are based on sites monitored in
ADEM's biological assessment program. WDG categories are numerically ranked with
increased levels of stress. ADEM converted the WDG scores to ranks 1-8, with lower
numbers representing less disturbance 146
Figure 46. Frequency distribution of fish IBI condition categories for sites in the three
ichthyoregions discussed in this case study: the (A) Plateau; (B) Piedmont, Ridge, and
Valley; and (C) Tennessee Valley site classes. The x-axis is divided into five condition
categories: excellent, good, fair, poor, and very poor 147
Figure 47. Example of range in typical northern Alabama streams with riffle-run habitat. Top:
Hendriks Mill Branch; Bottom: Hatchet Creek 148
Figure 48. Taxa relative abundance and the GAM slope based on capture probabilities for
Acroneuria (Plecoptera: Perlidae; attribute III) and Ferrissia (Gastropoda: Ancylidae;
attribute V) 148
Figure 49. Frequencies of sites (y-axis) in each BCG level (x-axis) in each northern Alabama site
class, showing reference sites as the blue portions of the bars. Distributions are based on
sites monitored in ADEM's biological assessment program 151
Figure 50. BCG scores and corresponding WDG scores for Northern Alabama. Distributions are
based on sites monitored in ADEM's biological assessment program 152
Figure 51. Alabama macroinvertebrate MMI distributions in site classes and BCG levels 153
Figure 52. Alabama fish IBI distributions in site classes and BCG levels 153
Figure 53. Distributions of Healthy Watershed Index (HWI) scores by macroinvertebrate BCG level
and site class 154
Figure 54. Left: St. Louis River; Right: Beaver Creek 157
Figure 55. Left: West Branch Little Knife River; Right: Judicial Ditch 7 157
Figure 56. Frequency distributions of IBI scores by BCG level for macroinvertebrate stream types
using data from natural channel streams sampled 1996-2011. Symbols: upper and lower
XII
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
bounds of box = 75th and 25th percentiles, middle bar in box = 50th percentile, upper and
lower whisker caps = 90th and 10th percentiles. MPCA also did a calibration offish index
scores with BCG levels assigned to sites 161
Figure 57. BCG illustrating the location of proposed biological criteria (black dotted line) for
protection of Minnesota's TALL) goals (Exceptional, General, Modified) (Source: MPCA
2014b) 162
Figure 58. Relation between Maine TALUs, the BCG, and Maine's other water quality standards
and criteria. Class AA/A is approximately equivalent to BCG levels 1 and 2. Classes B and C
approximate BCG levels 3 and 4, respectively. Non-attainment conditions below Class C
are approximately equivalent to BCG levels 5 and 6 167
Figure 59. Distribution of Maine water quality classifications in 1987 prior to WQS revisions 169
Figure 60. Distribution of Maine water quality classifications in 2012 following 25 years of water
quality improvements and classification upgrades 169
Figure 61. Box-and-whisker plot of % 1C of samples grouped by biological assessment results for
(A) macroinvertebrates and (B) algae with number of samples in parentheses. The NA
group includes samples that do not attain biological criteria for Classes AA/A, B, or C
(Source: Danielson et al. In press) 171
Figure 62. Pleasant River sites with attained water quality class and BCG level for different
assemblages and water body types 172
Figure 63. Comparison of reference and mitigation sites for the Maine Tolerance Index and
sensitive/tolerant taxa metrics (reference site N=51; mitigation site N=9) (DiFranco et al.
2013) 175
Figure 64. Numeric biological criteria adopted by Ohio EPA in 1990, showing stratification of
biological criteria by biological assemblage, index, site type, ecoregion for warmwater and
modified warmwater habitat (WWH and MWH, respectively), and statewide for the
exceptional warmwater habitat (EWH) use designations 180
Figure 65. An initial mapping of the Ohio TALUs to the BCG relating descriptions of condition along
the yl-axis and ranges of condition encompassed by the numerical biological criteria for
each of four tiered use subcategories and the highest antidegradation tier (ONRW) along
the y2-axis. ONRW - Outstanding National Resource Waters; EWH - Exceptional
Warmwater Habitat; WWH -Warmwater Habitat; MWH - Modified Warmwater Habitat;
LRW-Limited Resource Waters 181
Figure 66. Descriptive model of the response offish and macroinvertebrate assemblage metrics
and characteristics to a quality gradient and different levels of impact from stressors in
Midwestern U.S. warmwater rivers and streams (modified from Ohio EPA 1987 and Yoder
andRankin 1995b) 182
Figure 67. The number of individual stream and river segments in which ALU designations were
revised during 1978-1992, 1993-2001, and 2002-2016. Cases where the use was revised
to a higher use are termed "upgrades" and cases where a lower use was assigned are
termed "downgrades." Previously undesignated refers to streams that were not listed in
the 1985 WQS, but which were added as each was designated as a result of systematic
monitoring and assessment. The number of waters previously undesignated in the first
interval is unknown 185
XIII
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Figure 68. The flow of information from biological and water quality assessments to support for
major water quality management programs in Ohio 186
Figure 69. Key steps showing how a TALL) based framework can be used to organize and guide a
TMDL development and implementation process 187
Figure 70. The Mohican River in northeastern Ohio—a candidate for OSW classification because of
its high quality ecological and recreational attributes 188
Figure 71. Mapping the Ohio antidegradation tiers to the BCG relating descriptions of condition
along the yl-axis and ranges of condition encompassed by the numerical biological criteria
for each of four tiered use subcategories and the four antidegradation tiers along the y2-
axis. ONRW - Outstanding National Resource Waters; OSW - Outstanding State Waters;
SHOW - Superior High Quality Waters; GHQW - Generally High Quality Waters; LQW -
Low Quality Waters; EWH - Exceptional Warmwater Habitat; WWH - Warmwater Habitat;
MWH - Modified Warmwater Habitat; LRW - Limited Resource Waters 189
XIV
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Tables
Table 1. Ecological characteristics (i.e., attributes) used to develop the BCG 5
Table 2. BCG: Ecological Attributes 22
Table 3. BCG Matrix: Taxonomic Composition and Structure Attributes I-V 26
Table 4. BCG calibration and testing projects 35
TableS. Definitions of the technical elements (USEPA2013a) 38
Table 6. Examples of quantitative stressor variables that have been used for BCG projects 44
Table 7. Input variables for Minnesota Pollution Control Agency's (MPCA's) HDS (MPCA 2014a) 46
Table 8. Input variables for Alabama's HDG (Source: Lisa Huff, ADEM, personal communication) 46
Table 9. Example screening thresholds for stressor gradient (Connecticut) 49
Table 10. Distribution of macroinvertebrate and fish taxa across the BCG attributes in northern
Alabama 57
Table 11. Description of transitional cold-cool assemblages (benthic macroinvertebrate and fish
taxa) in each assessed BCG level, Upper Midwest coldwater streams. Definitions are
modified after Davies and Jackson (2006) (Source: Gerritsen and Stamp (2012)) 61
Table 12. Example of Narrative rules for transitional cold-cool assemblages in Upper Midwest
streams (Source: Gerritsen and Stamp (2012)) 73
Table 13. Benthic macroinvertebrate taxa: Ranges of attribute metrics in cold-cool transitional
macroinvertebrate samples. BCG levels by panel consensus, in the Upper Midwest BCG
data set (Gerritsen and Stamp 2012) 75
Table 14. Fish taxa: Ranges of attribute metrics in cold-cool transitional fish samples. BCG levels by
panel consensus 76
Table 15. Benthic macroinvertebrate taxa: Decision rules for macroinvertebrate assemblages in
coldwater and coolwater (transitional cold-cool) streams; samples with > 200 organisms.
Rules show the midpoints of fuzzy decision levels, followed by the range of the
membership function. The midpoint is where membership in the given BCG level is 50%
for that metric 78
Table 16. Fish taxa: Decision rules for fish assemblages in coldwater and coolwater (cold-cool
transitional) streams. Rules show the midpoints of fuzzy decision levels, where
membership in the given BCG level is 50% for that metric 79
Table 17. Benthic macroinvertebrate and fish taxa: Model performance—cold and coolwater
samples 81
Table 18. Narrative description of diatom assemblages in six BCG levels for streams of northern
New Jersey. Definitions are modified after Davies and Jackson (2006) 83
Table 19. BCG quantitative decision rules for diatom assemblages in northern New Jersey streams.
The numbers in parentheses represent the lower and upper bounds of the fuzzy sets. BCG
level 6 is not shown, because there are no specific rules for level 6: If a site fails level 5, it
falls to level 6. Shaded rules under BCG level 3 are alternate rules, that is, at least one
must be true for a site sample to meet BCG level 3 84
xv
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Table 20. Model performance for calibration and confirmation samples. "Yi better" indicates
models scored the sample VT. BCG level higher than the panel; e.g., Panel score was 4 and
model score was 3-4 tie. Half-level mismatches are counted half the value of full matches.
No mismatches exceeded 1/£ BCG level 85
Table 21. Cross referencing the 10 BCG attributes with selected fish IBI and macroinvertebrate
MMI metrics for streams and wadeable rivers 86
Table 22. Correlations (Pearson r) among Connecticut MMI index metrics 88
Table 23. Scoring thresholds for the Connecticut MMI to correspond to BCG levels 89
Table 24. Maine Biologists' Relative Findings Chart Using Macroinvertebrates (Source: Davies et al.
In press) 93
Table 25. Measures of community structure used in linear discriminant models for Maine (from
MEDEP 2014; State of Maine 2003). Means refer to the mean of three rock baskets
sampled at each site 96
Table 26. Classification of stream and river sites by two-way linear discriminant models for three
classifications. Numerical entries represent the percent of sites classified from a priori
classes (row) into predicted classes (columns). Therefore, diagonals are % correct
classification 97
Table 27. Land use classification and intensity factor (LDI coefficient) for Florida landscapes
(modified from Brown and Vivas 2005) Ill
Table 28. Description offish, salamander, and macroinvertebrate assemblages in each assessed
BCG level. Definitions are modified after Davies and Jackson (2006) 125
Table 29. BCG quantitative decision rules for macroinvertebrate assemblages. The numbers in
parentheses represent the lower and upper bounds of the fuzzy sets 128
Table 30. BCG quantitative decision rules for fish assemblages in small (0.5-1.4 mi2), medium (1.5-
7.9 mi2) and larger streams (> 8 mi2). The numbers in parentheses represent the lower and
upper bounds of the fuzzy sets. The mid-water cyprinid taxa metric is comprised of
notropis, luxilus, clinostomus, and cyprinella, minus swallowtail shiners 129
Table 31. Potential narrative decision rules for invertebrate samples from Pennsylvania high
gradient streams (modified from Gerritsen and Jessup 2007c) 138
Table 32. Characterization of Reference Conditions Using WDG and the Alabama
Macroinvertebrate MMI for streams. WDG scores increase with level of land use activity 144
Table 33. Example of narrative and quantitative rules from Northern Alabama BCG: BCG level 2
narrative and quantitative rules for macroinvertebrates and quantitative rules for fish in
northern Alabama. Macroinvertebrate rules apply in all northern Alabama streams. Fish
rules are applied by site class (PLA, RVP, and TV) and stream size (Small and Large) 149
Table 34. Decision rules for fish assemblages in two classes of Minnesota rivers. Rules show the
ranges of fuzzy membership functions. N indicates the number of sites for a given BCG
level and stream class in the calibration data set 160
Table 35. Criteria for Maine river and stream classifications and relationship to antidegradation
policy 166
Table 36. Examples of how numeric biological criteria results determine whether or not a water
body attains designated ALUs in Maine 168
XVI
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Table 37. Measured values of chemical and watershed stressors, attained water quality classes,
and corresponding BCG levels of reference wetlands and mitigation wetlands (DiFranco et
al. 2013) 176
Table 38. Descriptive summary of Ohio's tiered aquatic life use designations 178
Table 39. Narrative biological criteria (fish) for determining ALL) designations and attainment of
CWA goals (November, 1980; after Ohio EPA 1981) 179
Table 40. Narrative biological criteria (macroinvertebrates) for determining ALL) designations and
attainment of CWA goals (November 1980; after Ohio EPA 1981) 179
Table 41. General guidelines for nominating OSW, SHQW, and GHQW categories in Ohio.
Attributes are considered both singly and in the aggregate 189
XVII
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Executive Summary
The Clean Water Act (CWA) established a long-term objective to restore and protect the biological
integrity of the nation's waters. In the more than 40 years since the passage of the CWA, there has been
considerable progress in the science of aquatic ecology and in the development of biological monitoring
and assessment techniques to support implementation of the Act. The U.S. Environmental Protection
Agency (EPA) published its first guidance document on biological assessments and criteria in 1990. Since
then, aquatic science and its application in state water quality programs has advanced significantly.
States, territories, and authorized tribes (herein identified as "states") now routinely use biological
information to directly assess the biological condition of their aquatic resources, track changes in their
condition, and develop biological criteria to set expectations for maintaining biological integrity.
This document is designed for scientists engaged in biological assessments of water bodies. It outlines a
conceptual framework, the Biological Condition Gradient (BCG), for states to use to more precisely
define and interpret baseline biological conditions, help evaluate potential for improvement in degraded
waters, and measure and document incremental changes in condition along a gradient of anthropogenic
stress. The conceptual framework can be populated with state or regional data to develop a quantitative
model and establish numeric thresholds. The BCG is intended to complement existing biological
assessment and criteria methods and approaches.
What is the Biological Condition Gradient?
The BCG is a conceptual, scientific framework for interpreting biological response to increasing effects of
stressors on aquatic ecosystems. The framework was developed based on common patterns of
biological response to stressors observed empirically by aquatic biologists and ecologists from different
geographic areas of the United States. Scientists from 21 states, one interstate basin association, and
one tribe were involved in BCG development, in addition to scientists from EPA, the U.S. Geological
Survey, universities, and the private sector. The framework describes how 10 characteristics of aquatic
ecosystems change in response to the increasing levels of stressors, from an "as naturally occurs"
condition (e.g., undisturbed/minimally disturbed condition) to severely altered conditions. The
characteristics, defined in this document as "attributes," include aspects of community structure,
organism condition, ecosystem function, and connectivity. The BCG framework can be considered
analogous to a field-based dose-response curve where the dose (x-axis) represents increasing level of
anthropogenic stress, and the response (y-axis) represents biological condition.
Who Will Use the Biological Condition Gradient and For What Purpose?
Currently most states are using biological assessment information to support their water quality
management programs. The BCG contributes to the EPA biological assessment and criteria "toolbox,"
which includes biological indices, models, statistical approaches, and guidance. The BCG builds upon and
complements these approaches to provide a more refined and detailed measure of biological condition
and can help water quality management programs to:
• More precisely define and measure biological condition for specific waters;
• Identify and protect high quality waters;
• Evaluate potential for improvement in degraded waters;
• Track changes in condition;
XVIII
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
• Develop biological criteria; and
• Clearly communicate the likely impact of water quality management decisions to stakeholders.
These applications support CWA programs such as 305(b) assessments and reports, 303(d) listing of
impaired waters, and the Total Maximum Daily Load program implementation. The document includes
examples of how states are using, or are considering using, the BCG to support their water quality
management programs.
Why Now?
As the first BCG projects have been completed, there has been increasing interest in the BCG by other
state water quality management programs. Based on informal discussion with state water quality
managers and scientists who have been directly engaged in BCG development, their primary motivation
for using a BCG has been to more precisely define baseline conditions, better understand the quality of
their reference sites, identify high quality waters as candidates for additional protection, help evaluate
the potential for restoration of degraded waters, and document incremental improvements as best
management practices are implemented. In all cases, the states have emphasized the value of the BCG
to help communicate to the public the biological condition of their waters in context of the CWA
integrity objectives and the likely outcomes of water quality management decisions.
Because of the interest in BCGs, it is important now to document the status of model development,
discuss current strengths and limitations, and provide examples of how states are developing and
applying the BCG. This document provides a template and step-by-step process for constructing robust
BCGs, drawing from the lessons learned during a decade of testing by interstate, state, territorial, and
local government water quality management programs. As BCG development and calibration continues,
it is expected that the BCG process will be refined and improved.
Biological Condition Gradient Development: Decision Rules
This document describes the steps that entail convening an expert panel in order to construct narrative
descriptions and quantitative rules for assigning sites to BCG levels. Different approaches to developing
quantitative rules are discussed (e.g., mathematical set theory, derivation and calibration of biological
indices, and multivariate statistical and/or predictive modeling approaches). The core objective of the
panel process is to elicit expert judgment on defining ecologically significant change in the biotic
community and to document the underlying rationale for the judgments. By using a process to elicit
expert judgment, first narrative and then quantitative rules emerge and are tested and refined based on
the current state of the science, expert knowledge, and available data. The intended end product is a set
of well-vetted and transparent decision rules that can be readily understood and implemented by state
water quality program managers and scientists. Routine use of a quantitative BCG model by state water
quality management programs requires well documented and transparent decision rules so that
assessments can be made for newly sampled water bodies without reconvening the expert panel.
Specifically, the document presents:
• An approach to quantify the conceptual BCG framework and develop a numeric model. This
approach is based on elicitation of the experts' decision criteria and incorporation those of
criteria into a numeric decision model using a mathematical set theory approach (e.g., fuzzy
logic). This approach has been tested and refined in most of the BCG projects to date.
XIX
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
• Considerations and approaches for relating the BCG with the state's existing biological
assessment methods and tools such as multimetric biological indices. To date, most states have
developed biological indices.
• An example of a state approach to quantify the conceptual BCG. This approach involves
development of statistical models that predict (or simulate) the expert decisions and may or
may not use elicited expert reasoning or rules.
Building on these initial efforts, it is expected that additional methods to quantify the conceptual BCG
will be identified and tested.
The Stress Axis
The x-axis of the BCG framework, the Generalized Stress Axis (GSA), conceptually describes the range of
anthropogenic stress that may adversely affect aquatic biota in a particular area. It is a theoretical
construct. As multiple stressors are usually present in a system, the GSA seeks to represent the
cumulative stress that may influence biological condition. Typically, states have defined a stress gradient
using single or a combination of known, measurable stress gradients that in reality represent a portion
of the stressors impacting a water body. The conceptual GSA provides a framework to assist in
development of as comprehensive and robust a quantitative stress gradient as possible to support BCG
development. A well-defined, quantitative GSA, and the underlying data used to develop it, may serve as
a nexus between biological and causal assessments, thereby linking management goals and selection of
management actions for protection or restoration. However, a systematic testing of technical
approaches to define and apply a GSA to BCG development has not been conducted. This document
discusses technical issues to consider and provides examples of approaches to quantify a GSA.
Opportunities in the future may include piloting methods for application of national, regional, or basin
scale databases and methods to support state efforts to quantify a GSA for a specific geographic region
and water body type.
Document Organization
Chapters 1 and 2 explain the purpose and scientific underpinnings of the BCG. Chapters 3 and 4 present
methods on how to define and quantify the BCG biological axis, the biological levels of condition that
span undisturbed to severely altered conditions. Chapter 5, supported by Appendix A, provides an
overview, framework, and examples to describe the stress axis of the BCG model, the GSA. Examples of
how states have developed and applied the BCG are presented in Chapter 6. To date, use of the BCG to
support water quality management has primarily been for fresh water, perennial streams. However,
work underway is presented in Appendix B on BCG development for large rivers, estuaries, and coral
reefs.
xx
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Chapter 1. Introduction to the Biological Condition Gradient
1.1 Document Purpose
The Clean Water Act (CWA) established a long-term objective to, among other things, restore and
protect the biological integrity of the nation's waters (Figure 1). In the more than 40 years since the
passage of the CWA, there has been considerable progress in the science of aquatic ecology and in the
development of biological monitoring and assessment techniques to support implementation of the Act
(USEPA 2011a, 2013a). Since the U.S. Environmental Protection Agency (EPA) published its first guidance
document on biological assessments and criteria, aquatic science and its application in state water
quality programs has advanced (USEPA 1990, 2002, 2011a, 2013a). States, territories, and authorized
tribes (herein referred to as "states") now routinely use biological information to directly assess the
condition of their aquatic resources, track changes in biological condition, and develop biological criteria
to set expectations for maintaining biological integrity.
Figure 1. Stream and wadeable river.
Under the CWA, states have the primary authority to implement their water quality programs with EPA
review for consistency with the CWA requirements, which include implementing regulations. As a
consequence, states have independently developed technical approaches to assess biological condition
and establish thresholds (Hawkins 2006; USEPA 2002). Although these different approaches have
fostered innovation, they have complicated a nationally consistent approach to interpreting the
condition of aquatic resources. A consistent approach to interpreting biological condition will allow
scientists, water resource managers, and stakeholders to share a common understanding and language
to describe the condition of their waters, as well as share data and information across jurisdictional
boundaries (Davies and Jackson 2006).
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
In addition to using a variety of approaches for assessing and interpreting biological condition, states
have created a range of different aquatic life use (ALL)) classes to describe the expected biological
condition of their waters. At one end of the spectrum, states have adopted a general narrative
statement that replicates the ALL) goal identified in the CWA (e.g., protection and propagation offish,
shellfish, and wildlife). At the other end are more detailed approaches that describe the expected
species, assemblages, or habitats (e.g., salmonids, warmwater habitat, coldwater fisheries) or that
specify levels of condition (e.g., excellent, good, fair). Currently, most states have established one
general ALL) class, with a single threshold for assessing attainment. A limitation of a single ALL) class is
that the full range of biological conditions along a human disturbance gradient is limited to only two
categories: pass and fail. Water bodies assigned to a single ALL) class could include a range of biological
conditions found in undisturbed to moderately disturbed landscapes, or, in some cases even include
highly disturbed conditions where anthropogenic impacts are widespread and pervasive. As a result, a
water body supporting biological conditions characteristic of higher quality waters could degrade to a
lower level of water quality yet still be categorized as meeting its ALL). In contrast, for water that is
severely degraded, the designated ALL) might not be achievable in the short term, and therefore
incremental improvements due to management actions will not be measured or acknowledged. A
scientific framework that describes incremental biological changes along the full gradient of human
disturbance helps water quality managers identify and protect high quality waters and track incremental
improvements in degraded waters.
This document outlines a conceptual framework, the Biological Condition Gradient (BCG), that states
can use to more precisely describe existing, or baseline, biological condition; help evaluate potential for
improvement in condition; and measure incremental changes in condition along a gradient of human
disturbance, i.e., anthropogenic stress. The conceptual framework can be populated with state or
regional data to develop a quantitative model. It is intended to complement existing biological
assessment and criteria methods and approaches.
This document reports on the current status of quantitative model development and application. As BCG
development and calibration continues, it is expected that the BCG process will be further refined and
improved.
1.2 Background: When and Why?
In 2000, EPA convened a technical expert workgroup to identify scientifically sound and practical
approaches that would help states use biological assessments to better determine existing conditions
and potential for improvement, more precisely define ALUs, and develop biological criteria. The
workgroup consisted of scientists from federal, state, and tribal water programs, an interstate basin
association, the academic research community, and the private sector (see Davies and Jackson 2006 for
a list of workgroup members). The overarching objective of this effort was to develop a common
framework and language for interpreting biological condition. In the subsequent four years, the
workgroup met annually with drafts of the framework undergoing review and preliminary testing
between meetings. The effort was primarily guided by the practical experience of scientists and water
quality program managers from the 21 states, the interstate basin association and tribe participating in
the workgroup. The workgroup developed the conceptual BCG framework to describe levels, or tiers, of
biological response to increasing levels of stressors. The conceptual BCG was developed and tested
through a series of data exercises using a diverse array of data sets with initial focus on freshwater
perennial streams and wadeable rivers.
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
The workgroup activities coincided with a National Research Council (NRC) review of EPA's Total
Maximum Daily Load (TMDL) program and publication of its report Assessing the TMDL Approach to
Water Quality Management (NRC 2001). Among other recommendations, the NRC recommended the
use of biological assessments to better understand water quality and the establishment of a more
precise, descriptive approach to goal-setting as a step towards improving decision making and
establishing appropriate ALL) goals. For example, rather than stating that a water body needs to be
"fishable," the ALL) would ideally describe the expected fish assemblage or population (e.g., salmonid,
coldwater fishery, warmwater fishery), as well as the other biological assemblages necessary to support
that fish population. Additionally, levels of expected condition would be defined based on potential of a
water body to achieve a higher level of condition (e.g., salmonid spawning versus migration;
undisturbed and minimally disturbed conditions versus moderately or highly disturbed). The NRC
recommendation to more precisely define designated ALUs was taken into account by the BCG
workgroup as they developed the BCG framework. Since completion of the conceptual BCG framework
(Davies and Jackson 2006), many states have further developed and refined quantitative BCG models
(see Table 4, Chapter 3). In conjunction with other water quality management technical tools, the state
programs that have developed and applied the BCG have done so to help:
• Set scientifically defensible, ecologically-based aquatic life goals based on existing conditions
and potential for improvement;
• Determine baseline conditions and measure impacts of multiple stressors or system altering
conditions (e.g., climate change) on aquatic life;
• Further the use of monitoring data for the assessment of water quality standards (WQS) and
tracking changes in biological condition;
• Identify high quality waters for protection (e.g., Tier III antidegradation); and
• Communicate to stakeholders the likely impact of decisions on protection and management of
aquatic resources.
When asked about the most immediate, value-added benefits to their water quality management
programs from the development of a quantitative BCG model, state water quality program managers
and scientists cited the ability to measure and document incremental improvements due to
management actions and better identify and protect high quality waters.
The BCG conceptual framework, quantitative model development, and implementation reflects an
improved understanding of aquatic ecosystems and their biota resulting from more than 40 years of
assessment data and advances in use of these data in state water quality management programs. This
document represents the culmination of four years of workgroup deliberations, including four
workgroup meetings and two workshops to "road test" the conceptual BCG framework, followed by ten
years of development and application of quantitative BCG models in state programs. Over the past ten
years, the BCG has been developed for perennial streams, including headwater streams, using expert
consensus to develop narrative and numeric decision rules to assign sites to BCG levels. The use of the
BCG to complement or refine existing state measures such as Indices of Biotic Integrity (IBIs) is being
explored. Application of the BCG to water bodies other than perennial streams is underway for large
rivers, estuaries, and coral reefs. These latter efforts show promise for expanding the application of the
BCG beyond streams to more complex systems.
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
1.3 The Biological Condition Gradient: Brief Overview
The conceptual BCG is a scientific framework for interpreting biological response to increasing effects of
stressors on aquatic ecosystems (Figure 2). The framework was developed based on common patterns
of biological response to stressors observed empirically by aquatic biologists and ecologists from
different geographic areas of the United States (Davies and Jackson 2006). It describes how
characteristics of aquatic ecosystems that are typically measured by state water quality management
programs change in response to increasing levels of stress (see Table 1). The characteristics, defined as
attributes, include properties of the communities (e.g., tolerance, rarity, native-ness) and organisms
(e.g., condition, function) and are more fully described in Chapter 2.
The Biological Condition Gradient:
Biological Response to Increasing Levels of Stress
Levels of Biological Condition
Level 1. Natural structural, functional,
and taxonomic integrity is preserved.
Level 2. Structure & function similar
to natural community with some
additional taxa & biomass; ecosystem
level functions are fully maintained.
Level 3. Evident changes in structure
due to loss of some rare native taxa;
shifts in relative abundance; ecosystem
level functions fully maintained.
Level 4. Moderate changes in structure
due to replacement of some sensitive
ubiquitous taxa by more tolerant
taxa; ecosystem functions largely
maintained.
Level 5. Sensitive taxa markedly
diminished; conspicuously unbalanced
distribution of major taxonomic groups;
ecosystem function shows reduced
complexity & redundancy.
Level 6. Extreme changes in structure
and ecosystem function; wholesale
changes in taxonomic composition;
extreme alterations from normal
densities.
Watershed, habitat, flow regime
and water chemistry as naturally
occurs.
Chemistry, habitat, and/or flow
regime severely altered from
natural conditions.
Figure 2. Conceptual model of the BCG. Although in reality the relationship between stressors and their
cumulative effects on the biota is likely nonlinear, the relationship is presented as such to illustrate the concept.
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
The BCG can be considered analogous to a field-based dose-response curve where the dose (x-axis)
represents increasing levels of stressors, and the response (y-axis) represents biological condition.
Stressors are physical, chemical, or biological factors that induce an adverse response from aquatic biota
(USEPA 2000b). For example, high concentrations of certain metals, nutrients, or sediment can adversely
impact, or stress, aquatic biota. Loss of suitable aquatic habitat or presence of aquatic invasive species
can also adversely impact the aquatic biota expected for a specific water body. These stressors can
cause aquatic ecosystems to change from natural conditions and exhibit altered compositional,
structural, and functional characteristics. The degree to which stressors affect the biota depends on the
magnitude, frequency, and duration of the exposure of the biota to the stressors. Developing a BCG for
a given system characterizes the general relationship between its stressors in total and a water body's
overall biological condition. Multiple stressors are usually present, and thus, the stress x-axis of the BCG
seeks to represent their cumulative influence as a Generalized Stress Axis (GSA),1 much as the y-axis
generalizes biological condition. The x and y axes of the BCG serve as a framework to organize, relate,
and help reconcile the mosaic of factors and interactions that exist, parts of which will be characterized
and measured using biological, chemical, physical, and/or land use/land cover indicators.
Table 1. Ecological characteristics (i.e., attributes) used to develop the BCG
Attribute
1
II
III
IV
V
VI
VII
VIII
IX
X
Description
Historically documented, sensitive, long-lived, or regionally endemic taxa
Highly sensitive taxa
Intermediate sensitive taxa
Intermediate tolerant taxa
Tolerant taxa
Non-native or intentionally introduced species
Organism condition
Ecosystem function
Spatial and temporal extent of detrimental effects
Ecosystem connectance
*Note: Identified as Sensitive-rare taxa in Davies and Jackson 2006.
The BCG differs from the standard dose-response curve in that the BCG does not represent the
laboratory response of a single species to a specified dose of a known chemical, but rather the in-situ
response of the resident biotic community to the sum of stressors to which that community is exposed.
Thus, it is an outcome-based measure and something that can express complex water quality goals such
as biological integrity. In this document EPA proposes a BCG that is divided into six levels of biological
condition along a generalized stressor-response curve, ranging from observable biological conditions
found at no or low levels of stressors (level 1) to those found at high levels of stressors (level 6). States
may propose to consolidate or aggregate these levels into fewer levels or further refine and increase the
number of levels. Regardless of how many levels a quantitative BCG may ultimately include, it can be
crosswalked with the conceptual model. Chapter 6 and Appendix B illustrate examples of ecoregional or
state-specific BCGs and how they may be "mapped" onto the conceptual BCG.
Between 2000 and 2005, the original framework was tested at annual workgroup meetings and then at
two regional workshops in the Great Plains and in the Arid Southwest. It was tested by determining how
consistently the scientists assigned samples of benthic macroinvertebrates or fish to the different levels
1 For more information on the Generalized Stress Axis, see Chapter 5.
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
of biological condition in freshwater streams. Workgroup members identified similar sequences of
biological response to increasing levels of stressors regardless of geographic area and predicted that the
framework in principal should be applicable to other water body types. These results support the
development and application of the BCG as a nationally applicable framework for interpreting the
biological condition of aquatic systems (Davies and Jackson 2006).
Understanding the links between stressors (and their sources) with the response of the aquatic biota will
help water quality managers to more accurately determine both the existing and potential conditions of
the aquatic biota in a specific water body and help predict the stressors that affect that condition (Figure
3). This information will assist water quality program managers in determining the most effective
recourse to address biological impairment. There are different approaches and new studies, methods,
and large data sets that can assist states to better define and quantify the causal sequence between
stressors and their sources and biological responses once biological impairment is identified.2
Ultimately, the goal of the EPA biological criteria program is to build a stronger technical bridge between
biological condition assessments, causal assessments, and the actions taken to protect and restore
biological condition. A well-defined BCG x-axis, the GSA, and the science underlying it may help achieve
this objective. In Chapter 5, information on approaches and technical challenges to define the GSA are
discussed, with examples of a conceptual GSA framework and potential stress indicators included in
Appendix A.
Human activity:
"the drivers"
Altered environmental
processes and materials
Change in
biological
condition
Figure 3. Model illustrating the multiple pathways through which human activities may exert pressure on an
aquatic system by altering fundamental environmental processes and materials, creating stressors that may
adversely affect the aquatic biota (Source: Modified from figure courtesy of David Allen, University of Michigan).
2 See http://www3.epa.gov/caddis/ and
http://water.epa.gov/lawsregs/lawsguidance/cwa/tmdl/recovery/overview.cfm. Accessed February 2016.
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
1.4 Use of the Biological Condition Gradient to Support Water Quality Standards
and Condition Assessments
The full objective of section 101(a) of the CWA is to restore and maintain the chemical, physical, and
biological integrity of the Nation's waters. In the scientific literature, an aquatic system with chemical,
physical, and biological integrity has been described as being capable of "supporting and maintaining a
balanced, integrated, adaptive community of organisms having a composition and diversity comparable
to that of the natural habitats of the region" (Frey 1977).
Over the intervening years, the understanding of how to define and measure the integrity of aquatic
systems has advanced considerably. The term "integrity" has been further refined in the literature to
mean a balanced, integrated, adaptive system having a full range of ecosystem elements (e.g., genera,
species, assemblages) and processes (e.g., mutation, demographics, biotic interactions, nutrient and
energy dynamics, metapopulation dynamics) expected in areas with no or minimal human disturbance
(Karr 2000). The aquatic biota residing in a water body are the result of complex and interrelated
chemical, physical, and biological processes that act over time and on multiple scales (e.g., instream,
riparian, landscape) (Karr et al. 1986; Yoder 1995). By directly measuring the condition of the aquatic
biota, one is able to more accurately define the aquatic community that is the outcome of all these
factors.
To help achieve the integrity objective, the CWA also established an interim goal for the protection and
propagation offish, shellfish, and wildlife and recreation in and on the water. EPA has interpreted the
"protection and propagation" interim goal for aquatic life to include the protection of the full
complement of aquatic organisms residing in or migrating through a water body. As explained in EPA's
Water Quality Standards Handbook (USEPA 2014a), the protection afforded by WQS includes the
representative aquatic community (e.g., fish, benthic macroinvertebrates, and periphyton):
The fact that sport or commercial fish are not present does not mean that the water may not be
supporting an aquatic life protection function. An existing aquatic community composed entirely of
invertebrates and plants, such as may be found in a pristine tributary alpine stream, should be protected
whether or not such a stream supports a fishery. Even though the shorthand expression
'fishable/swimmable' is often used, the actual objective of the Act is to restore the chemical, physical and
biological integrity of our Nation's waters (section 101(a)). The term 'aquatic life' would more accurately
reflect the protection of the aquatic community that was intended in section 101(a)(2) of the Act.
The representative community of aquatic organisms residing in, or migrating through, a water body will
vary depending on the water body type. For example, fish, benthic macroinvertebrates, and periphyton
are aquatic assemblages measured by states and tribes when assessing the biological condition of most
streams and rivers. However, in headwater streams and many wetlands, amphibians are an important
component of the biotic community, and fish may be absent. Large river and estuarine assessments
typically include both benthic invertebrates and fish community measures. In coral reefs, coral, sponge,
and fish communities are key assemblages to measure and assess. The BCG offers a framework to
provide more detailed and descriptive statements of the aquatic community expected in an undisturbed
or minimally disturbed aquatic community, as well as potential incremental changes that might be
expected in community characteristics with increasing levels of anthropogenic stress.
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
1.4.1 Use of the Biological Condition Gradient to Support Aquatic Life Use
Assessments
While section 101(a) of the CWA establishes the objective to restore and maintain the chemical,
physical, and biological integrity of the nation's waters, other sections of the CWA establish the
programs and authorities for implementation of this objective. Section 303(c) provides the basis of the
WQS program. WQS are components of state (or, in certain instances, federal) law that define the water
quality goals of a water body, or parts of a water body, by designating the use or uses of the water body
and by setting criteria necessary to protect the uses (in addition to antidegradation requirements).
Although the CWA gives EPA an important role in determining appropriate minimum levels of protection
and providing national oversight, it also gives considerable flexibility and discretion to state water
quality managers to design their own programs and establish levels of protection above the national
minimums. CWA section 303 directs states to adopt WQS to protect the public health and welfare,
enhance the quality of water, and serve the purposes of the CWA. "Serve the purposes of the Act" (as
defined in sections 101(a), 101(a)(2), and 303(c) of the CWA) means that WQS should include provisions
for restoring and maintaining chemical, physical, and biological integrity of state waters; provide,
wherever attainable, water quality for the protection and propagation of fish, shellfish, and wildlife and
recreation in and on the water (i.e., "fishable/swimmable"); and consider the use and value of state and
tribal waters for public water supplies, propagation offish and wildlife, recreation, agricultural and
industrial purposes, and navigation. Further requirements for WQS can be found at 40 Code of Federal
Regulations (CFR) Part 131.
State WQS provide the foundation for water quality-based pollution control programs. With the public
participating in their adoption (see 40 CFR 131.20), such standards serve the dual purposes of (1)
establishing the water quality goals for a specific water body and (2) providing the regulatory basis for
the establishment of water quality-based treatment controls and strategies beyond the technology-
based levels of treatment required by sections 301(b) and 306 of the CWA. The WQS serve as, among
other things, the basis for ALL) attainment decisions, National Pollutant Discharge Elimination System
(NPDES) permit limits, and the targets for TMDLs.3
40 CFR Part 131.10(a) of the WQS regulation requires that states specify appropriate water uses to be
achieved and protected. A water body's designated uses are those uses specified in WQS, whether or
not they are being attained (40 CFR 131.3(f)). The designated use of a water body is the most
fundamental articulation of the water body's role in the aquatic environment as defined by society. All of
the water quality protections established by the CWA follow from the water body's designated use. As
designated uses are critical in determining the water quality criteria that apply to a given water body,
determining and clearly defining the appropriate designated use is of paramount importance in
establishing criteria that are appropriately protective of that designated use. In addition, the regulations
establish a rebuttable presumption that the uses of protection and propagation of fish, shellfish, and
wildlife and recreation in and on the water are attainable and must apply to a water body, unless it has
been affirmatively demonstrated that such uses are not attainable.
3 For more information about Water Quality Standards, see the WQS Regulation at
http://water.epa.gov/lawsregs/lawsguidance/wqs index.cfm (Accessed February 2016) and EPA's Water Quality
Standards Handbook at http://water.epa.gov/scitech/swguidance/standards/handbook/ (Accessed February
2016).
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Biological assessments can be effectively used to help subcategorize the ALL) designations. For example,
states may adopt subcategories of a use and set the appropriate criteria to reflect varying needs of such
subcategories of uses to differentiate between coldwater and warmwater fisheries (see 40 CFR
131.10(c)). States may also adopt seasonal uses, such as the use of streams or rivers for migratory or
spawning purposes (40 CFR 131.10(f)). One major challenge in assigning designated uses for aquatic life
to surface waters is separating the natural differences inherent in aquatic ecosystems and appropriately
classifying them by type (e.g., naturally coldwater vs. warmwater streams) and location (e.g., ecoregion)
from the differences that result from exposure to anthropogenic stressors. Natural or "naturally
occurring" conditions can be interpreted as comparable to the range of physical, biological, and
chemical conditions observed in undisturbed to minimally disturbed reference sites (Stoddard et al.
2006). When developed using reference data sets from long term biological monitoring and assessment
programs, the boundaries for the upper BCG levels can be described in a narrative form and quantified
to document the observed natural conditions. The BCG thus provides a descriptive framework to help
biologists and water quality managers interpret their aquatic life goals relative to natural conditions. By
more fully accounting for natural differences in aquatic ecosystems, designating more specific ALUs
helps to reduce a major source of uncertainty and error in water quality management.
The BCG can be used by state programs not only to develop detailed narrative descriptions of ALL) goals
in terms of the expected biological community, but also to help develop numeric biological criteria for
measuring attainment of the goals (USEPA 1990, 2011a). Water quality criteria are elements of state
WQS expressed as constituent concentrations, levels, or narrative statements representing a quality of
water that supports a particular use. When criteria are met, water quality is expected to protect the
designated use (40 CFR 131.3). Once adopted into standards, criteria can serve as the basis for (1)
controls on point and nonpoint source pollution concentrations to protect aquatic life, (2) statements of
expectations for the condition of aquatic life in a water body, and (3) guidelines helpful in water quality
planning (e.g., tracking of cumulative loads of point and nonpoint source pollutants). Biological criteria
have been defined as narrative expressions or numeric values of the biological characteristics of aquatic
communities based on appropriate reference conditions.
1.4.2 Use of the Biological Condition Gradient to Define Levels of Condition
By designating uses and articulating narrative and numeric criteria, states can establish environmental
goals for their water resources and measure attainment of these goals. When designating uses, a state
may weigh the environmental, social, and economic consequences of different use designations. Water
quality regulations allow the state, with public participation, flexibility in weighing these considerations
and adjusting designated uses over time. Clearly defining the uses that appropriately reflect the current
and potential future uses for a water body, determining the attainability of those goals, and
appropriately evaluating the consequences of a designation can be a challenging task.
A principal function of designated uses in WQS is to communicate the desired condition of surface
waters to water quality managers, the regulated community, and the public. For designating ALUs, an
effective approach is one that readily and transparently translates narrative biological descriptions of
the ALL) into quantitative measures, such as biological index values. The index values can be adopted
into the WQS as biological criteria and thresholds established for assessing attainment. The indices
should respond in predictable ways to stress so that degradation can be detected early and incremental
improvements tracked. States that have developed robust biological assessment programs typically
strive to distinguish different levels of biological condition. States have either made these levels explicit
in their WQS by adopting detailed biological descriptions of ALUs, or they have implicitly done so by
recognizing levels of condition in their monitoring protocols for assessing attainment of ALU.
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Although the benefits of specificity might apply to any of the designated uses described in CWA section
303, the benefits are particularly relevant for ALUs, because a broad range of biological conditions can
be interpreted as supporting an ALL). For example, biological conditions in a minimally disturbed stream
in a wilderness area would likely support a biotic community close to what would naturally be expected,
whereas the biological condition in a stream in a more developed watershed might be measurably
impacted relative to the wilderness stream, the degree of impact dependent upon effectiveness of best
management practices (BMPs) that have been implemented. Under non-specific ALL) classification with
a single ALL) threshold, both streams might be judged as meeting the designated ALL), and a threshold
might be set that does not protect the higher biological conditions in the wilderness stream from
degrading. By specifically articulating ALL) goals for systems with different levels of human disturbance,
deterioration can be detected and preventive management actions can be triggered earlier in the
process prior to serious and irretrievable degradation. The BCG provides a framework for defining
management goals and designated uses for water bodies having different levels of biological condition.
10
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Chapter 2. The Biological Condition Gradient: Fundamental Concepts
The BCG is a scientific framework that supports more refined interpretation of biological condition even
when assessment approaches may differ. The BCG combines scientific knowledge with the practical
observations and experience of biological assessment practitioners (Figure 4) with the needs of resource
managers. In conjunction with other environmental data and information, it can be used by
environmental practitioners to help:
• Determine the environmental conditions that exist, relative to naturally-derived conditions—
The BCG provides a common language with which to interpret and communicate current
ecological conditions relative to baseline conditions that are anchored in level 1 of the BCG, "as
naturally occurs."
• Decide what environmental conditions are desired—The BCG can be used with expert groups
and stakeholders to set easily communicated environmental goals.
• Plan for how to achieve these conditions—The BCG provides a scientific basis for planning,
restoration, protection, and monitoring by providing a common language and a pathway to
shared quantitative goals.
Figure 4. Biologists conducting stream and lake assessments.
The BCG translates the theoretical and empirical work of researchers and practitioners to create a
nationally-applicable model that helps to link management goals for resource condition with the
quantitative measures used in biological assessments. As discussed in Chapter 1, the conceptual BCG
was developed and tested by an expert workgroup that included scientists from 21 states, an interstate
basin association, and a tribe. The BCG was designed to describe ecological response to anthropogenic
stressors in sufficient detail so that a site can be placed into a level4 along the BCG continuum through
use of the core data elements collected by most state monitoring programs (USEPA 2013a). This
framework can be used to organize biological, chemical, physical, and land cover data and information
to interpret changes in assemblage composition and structure, spatial and temporal size of disturbance,
and declines in function and connectivity relative to a baseline of undisturbed or minimally disturbed
conditions.
A full description of the BCG levels is provided in section 2.3.
11
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
The BCG provides an interpretative framework explicitly linking science and monitoring information to
goals in water quality standards and criteria and, thus, aids in management decision making (Davies and
Jackson 2006). Each of the proposed six levels of the BCG is described via a detailed narrative that
communicates ecological characteristics associated with that condition level. In this way, the descriptive
gradient can be used to interpret numeric metric scores into a fuller understanding of their ecological
meaning and importance. Once calibrated to local data, the BCG creates a bridge between biological
metric scores and the condition levels with which they are commonly associated.
2.1 The Scientific Foundation of the Biological Condition Gradient
The practice of using biological indicators to assess water quality is over a century old, and the scientific
foundation of the BCG is based on many decades of biologists' accumulated experience with biological
assessment and monitoring. The Saprobien System is a concept based on organism tolerance proposed
by Lauterborn in 1901 and further developed by Kolkwitz and Marsson (Davis 1995). This system uses
benthic macroinvertebrates and planktonic plants and animals as indicators of organic loading and low
dissolved oxygen (DO). It has been updated since its initial development and is currently used in several
European countries. The limnologists Thienemann and Naumann developed the concept of trophic state
classification for lakes in the 1920s (Cairns and Pratt 1993; Carlson 1992). Both the Saprobien System
and lake trophic state classifications describe a response gradient (or response classes for lakes) to
nutrient pollution. The Saprobien System was explicitly developed to assess human pollution in rivers,
but the trophic state concept was originally developed to describe natural conditions in lakes and only
later became a concept to describe pollution-induced eutrophication (e.g., Vollenweider 1968). The
1950s marked the development of Beck's biotic index in the U.S. and Pantle and Buck's Saprobic Index in
Europe, both of which were directly based on the Saprobien System (Beck 1954; Pantle and Buck 1955).
The Saprobic Index, which led to the development of the widely used Hilsenhoff Index (e.g., Hilsenhoff
1987a, 1987b) in the U.S., could be considered the predecessor of today's biotic indices (Davis 1995).
Later studies used diversity indices based on information theory to describe changes in community
structure, richness, and dominance (evenness) as a measure of pollution effects (e.g., Wilhm and Dorris
1966).
Biological information from monitoring programs has been frequently synthesized by constructing biotic
indices, such as the IBI (Karr 1981; Karr et al. 1986). The IBI integrates the concept of anchoring the
measurement system in undisturbed reference conditions with the measurement of several indicators
intended to reflect ecological components of composition, diversity, and ecosystem processes. It thus
combines a conceptual model of ecosystem change in response to increasing levels of stressors with a
practical measurement system. The BCG is also grounded in the concepts of stress ecology articulated
by Odum et al. (1979), Odum (1985), Rapport et al. (1985), and Cairns et al. (1993), describing "natural"
conditions and the change in biological condition caused by stressors. To achieve maximum potential
application nationwide, the BCG levels were developed based on state biologists' experiences with
water quality management (Courtemanch et al. 1989; Yoder and Rankin 1995a), as well as the practical
experience of a diverse group of aquatic scientists from different bio-geographic areas (Davies and
Jackson 2006). The BCG:
• Describes a scale of six condition levels, from undisturbed (level 1) to highly disturbed conditions
(level 6).
• Synthesizes existing field observations and generally accepted interpretations of patterns of
biological change within a common framework.
12
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
• Incrementally measures how a system may have departed from undisturbed condition, based on
observable, ecological attributes.
In its initial development, the description of biological attributes that make up the model applied best to
permanent, hard-bottom streams that are exposed to increases in temperature, nutrients, fine
sediments, and other pollutants. This is the stream-type and stressor regime originally described by the
model and the one most developed to date, for example, in Alabama, Connecticut, Maine, Maryland,
Minnesota, New Jersey, Ohio, and Vermont. The model has been further tested with states in different
parts of the country and increasingly in different water body types (e.g., headwater streams, coastal
plains freshwater streams, rivers, wetlands, estuaries, and coral reefs) to evaluate the national
applicability of the model (see Appendix B for examples). Results have shown good correlation with
some necessary refinement of the model attributes to accommodate regional and water body
differences. For example, for the southern great plains region, attribute II, originally defined as sensitive-
rare taxa, was redefined as highly sensitive taxa because rarity of a taxon in the region was not
associated with sensitivity to stress. In this region, many rare, native taxa might be highly tolerant to
stressors, such as low DO and high temperature. Through similar developmental processes, the BCG, as
initially developed and tested, is applicable to other aquatic ecosystems and stressors with appropriate
modifications. The BCG should be viewed as a scientific framework that can readily incorporate future
advances in scientific understanding. The model building was initially based on expert consensus and
then further tested and refined following procedures detailed in Chapter 3. Quantitative approaches for
translating the narrative model into numeric values are discussed in Chapter 4.
The value of a conceptual framework such as the BCG is not only that it documents experimentally
established knowledge, but that it also promotes a more rigorous testing of empirical observations by
clearly stating them in a provisional model (Davies and Jackson 2006). Conceptual models formalize the
state of knowledge and guide research. Empirically-based generalizations have led to conceptual models
that describe the behavior of biological systems under stress (Brinkhurst 1993; Fausch et al. 1990; Karr
and Dudley 1981; Margalef 1963, 1981; Odum et al. 1979; Rapport et al. 1985; Schindler 1987). For
example, Brinkhurst (1993) observed that "Everyone knew [in 1929] that increases in numbers and
species could be related to mild pollution, that moderate pollution could produce changes in taxa so
that diversity remained similar but species composition shifted, and that eventually species richness
declined abruptly and numbers of some tolerant forms increased dramatically." Such ecosystem
responses to stressor gradients have been portrayed as a progression of stages that occur in a generally
consistent pattern (Cairns and Pratt 1993; Odum 1985; Odum et al. 1979; Rapport et al. 1985).
Establishing and validating quantifiable thresholds along that progression with empirical data is a
priority need for resource managers (Cairns 1981).
2.2 The Biological Condition Gradient Attributes
The BCG framework depicts ecological condition in terms of observable or measurable changes in an
aquatic system in response to anthropogenic stress. The characteristics, described as "attributes" in this
document, were selected because they corresponded to the characteristics used by state workgroup
members to measure biological condition and develop biological criteria. The 10 attributes are discussed
below and listed in Table 1. In biological assessments, most information is collected at the spatial scale
of a site or reach and the temporal scale ranging from a season to as short as a single sampling event.
Many of the attributes that make up the BCG are based on these scales. Site scale attributes include
aspects of taxonomic composition and community structure (attributes I-V), organism condition
(attribute VI), and organism and system performance (attributes VII and VIII). At larger temporal and
13
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
spatial scales, physical-biotic interactions (attributes IX and X) are also included because of their
importance to state water quality management programs in evaluating longer-term impacts,
determining restoration potential, and tracking recovery in specific water bodies.
Information used to assess the ten attributes may be acquired from two sources. Sample-based data
from instream monitoring using standardized sampling protocols can produce the most reliable,
reproducible form of information and are best used for attributes II-V. Knowledge-based information,
such as evidence from natural history surveys, agency records and reports (e.g., stocking reports),
academic studies and journal publications, expert observations, and so on, can contribute significantly to
BCG development even when methods are inconsistent. Since many of the attributes rely on the
positive observation (i.e., presence) of an organism and its relative occurrence in the community, any
reliable sources of information can be used to develop and calibrate the BCG for a specific water body
and/or region. Attributes I-X are described below (from Davies and Jackson 2006).
Attribute I: Historically Documented, Sensitive, Long-lived, or Regionally Endemic Taxa
Attribute I can be developed using both sample-based and knowledge-based sources. Taxa that are
historically documented refer to those known to have been supported in a water body or region
according to historical records. This attribute was derived to cover taxa that are sensitive or regionally
endemic that have restricted, geographically isolated distribution patterns (occurring only in a locale as
opposed to a region), often due to unique life history requirements. They may be long-lived and late
maturing and have low fecundity, limited mobility, multiple habitat requirements as with diadromous
species, or require a mutualistic relationship with other species. They may be among listed Endangered
or Threatened (E/T) or special concern species. Predictability of occurrence is often low, and therefore
requires documented observation. The presence or absence of a population might provide significant
information in an assessment, but there are typically insufficient data to develop the stress response
relationships needed to assign these taxa to attributes II through V (as discussed below). Recorded
occurrence may be highly dependent on sample methods, site selection, and level of effort, thus
requiring use of knowledge-based sources in addition to actual instream sampling. The taxa that are
assigned to this category require expert knowledge of life history and regional occurrence of the taxa to
appropriately interpret the significance of their presence or absence. Long-lived species are especially
important as they provide evidence of multi-annual persistence of habitat condition. For example, many
species of freshwater mussels in the Southeast U.S. are highly endemic and have been extirpated in
many areas. The presence of freshwater mussels in a stream might signify high quality conditions, but
their absence does not necessarily indicate poor conditions if overharvesting of the mussels is the cause.
Attribute II: Highly Sensitive Taxa
Highly sensitive taxa typically occur in low numbers relative to total population density, but they might
make up a large relative proportion of richness. In high quality sites, they might be ubiquitous in
occurrence or might be restricted to certain micro-habitats. Many of these species commonly occur at
low densities, so their occurrence is dependent on sample effort. They are often stenothermic
(i.e., having a narrow range of thermal tolerance) or cold-water obligates, and their populations are
maintained at a fairly constant level, with slower development and a longer life-span. They might have
specialized food resource needs, feeding strategies, or life history requirements, and they are generally
intolerant to significant alteration of the physical or chemical environment. They are often the first taxa
lost from a community following moderate disturbance or pollution.
In earlier descriptions of the BCG, highly sensitive taxa were called sensitive-rare taxa (Davies and
Jackson 2006), but experience with calibrating the BCG showed that some highly sensitive species are
14
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
found at many exceptional sites, and some were occasionally highly abundant (e.g., Snook et al. 2007).
The distinguishing characteristic for this attribute category was found to be sensitivity and not relative
rarity, although some of these taxa might be uncommon in the data set (e.g., very small percent of
sample occurrence or sample density)
Attribute III: Intermediate Sensitive Taxa
Intermediate sensitive taxa were formerly labeled sensitive-ubiquitous taxa (Davies and Jackson 2006),
but subsequent development revealed that the experts relied upon the sensitivity of a species to stress
rather than whether it was "ubiquitous," though intermediate sensitive taxa are ordinarily common and
abundant in natural communities. They tend to have a broader range of tolerances than highly sensitive
taxa, and they usually occur in reduced abundance and reduced frequencies at disturbed or polluted
sites. These taxa often comprise a substantial portion of natural communities.
Attribute IV: Intermediate Tolerant Taxa
Attribute IV taxa commonly comprise a substantial portion of an assemblage in undisturbed habitats, as
well as in moderately disturbed or polluted habitats. They exhibit physiological or life-history
characteristics that enable them to thrive under a broad range of thermal, flow, or oxygen conditions.
Many have generalist or facultative feeding strategies enabling utilization of diverse food types. These
species have little or no detectable response to moderate stress, and they are often equally abundant in
both reference and moderately stressed sites. Some intermediate tolerant taxa may show an
"intermediate disturbance" response, where densities and frequency of occurrence are relatively high at
intermediate levels of stress, but they are intolerant of excessive pollution loads or habitat alteration.
Attribute V: Tolerant taxa
Tolerant taxa are those that typically comprise a low proportion of natural communities. These taxa are
more tolerant of a greater degree of disturbance and stress than other organisms and are, thus,
resistant to a variety of pollution or habitat induced stress. They may increase in number (sometimes
greatly) under severely altered or stressed conditions. They may possess adaptations in response to
organic pollution, hypoxia, or toxic substances. These are the last survivors in severely disturbed systems
and can prevail in great numbers due to lack of competition or predation by less tolerant organisms, and
they are key community components of level 5 and 6 conditions.
Attribute VI: Non-native or Intentionally Introduced Taxa
With respect to a particular ecosystem, species fitting attribute VI are any species not native to that
ecosystem. Species introduced or spread from one region of the U.S. to another outside their normal
ranges are non-native, or non-indigenous. This category also includes species introduced from other
continents and referred to as "alien" species. Attribute VI can also include introduced disease or
parasitic organisms. This attribute represents both an effect of human activities and a stressor in the
form of biological pollution. Although some intentionally introduced species are valued by large
segments of society (e.g., gamefish), these species might be as disruptive to native species as
undesirable opportunistic invaders (e.g., zebra mussels). Many rivers in the U.S. are dominated by non-
native fish and invertebrates (Moyle 1986), and the introduction of non-native species is the second
most important factor contributing to fish extinctions in North America (Miller et al. 1989). The BCG
identifies the presence of native taxa expected under undisturbed or minimally disturbed conditions as
an essential characteristic of BCG level 1 and 2 conditions. The BCG only allows for the occurrence of
non-native taxa in these levels if those taxa do not displace native taxa and do not have a detrimental
effect on native structure and function. Condition levels 3 and 4 depict increasing occurrence of non-
15
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
native taxa. Extensive replacement of native taxa by tolerant or invasive, non-native taxa can occur in
levels 5 and 6. Attribute VI may rely on either sample-based or knowledge-based sources.
Attribute VII: Organism Condition
Organism condition is an element of ecosystem function, expressed at the level of anatomical or
physiological characteristics of individual organisms. Organism condition includes direct and indirect
indicators such as fecundity, morbidity, mortality, growth rates, and anomalies (e.g., lesions, tumors,
and deformities). Some of these indicators are readily observed in the field and laboratory, whereas the
assessment of others requires specialized expertise and much greater effort. Organism condition can
also change with season or life stage, or occur as short-term events making assessment difficult. The
most common approach for state programs is to forego complex and demanding direct measures of
organism condition (e.g., fecundity, morbidity, mortality, disease, growth rates) in favor of indirect or
surrogate measures (e.g., percent of organisms with anomalies, age or size class distributions) (Simon
2003). Organism anomalies in the BCG vary from naturally occurring incidence in levels 1 and 2 to higher
than expected incidence in levels 3 and 4. In levels 5 and 6, biomass is reduced, the age structure of
populations indicates premature mortality or unsuccessful reproduction, and the incidence of serious
anomalies is high. This attribute has been successfully used in stream indices based on the fish
assemblage (Sanders et al. 1999; Yoder and Rankin 1995a). Incidence of disease is being evaluated as an
indicator of organism condition for the coral reef BCG (see Appendix B-3).
Attribute VIII: Ecosystem Function
Ecosystem function refers to any processes required for the performance of a biological system
expected under naturally occurring conditions. Naturally occurring conditions have been typically
interpreted as those conditions found in undisturbed to minimally disturbed conditions but some
processes can be sustained under moderate levels of disturbance. Examples of ecosystem functional
processes are primary and secondary production, respiration, nutrient cycling, and decomposition.
Assessing ecosystem function includes consideration of the aggregate performance of dynamic
interactions within an ecosystem, such as the interactions among taxa (e.g., food web dynamics) and
energy and nutrient processing rates (e.g., energy and nutrient dynamics) (Cairns 1977).
Additionally, ecosystem function includes aspects of all levels of biological organization (e.g., individual,
population, and community condition). Altered interactions between individual organisms and their
abiotic and biotic environments might generate changes in growth rates, reproductive success,
movement, or mortality. These altered interactions are ultimately expressed at ecosystem-levels of
organization (e.g., shifts from heterotrophy to autotrophy, onset of eutrophic conditions) and as
changes in ecosystem process rates (e.g., photosynthesis, respiration, production, decomposition).
At this time, the level of effort required to directly assess ecosystem function is beyond the means of
most state monitoring programs. Instead, in streams and wadeable rivers, most programs rely on
taxonomic and structural indicators to make inferences about functional status (Karr et al. 1986). For
example, shifts in the primary source of food might cause changes in trophic guild indices or indicator
species. Although direct measures of ecosystem function are currently difficult or time consuming, they
might become practical in the future (Gessner and Chauvet 2002). The BCG conceptual model includes
ecosystem function for future application.
16
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Attribute IX: Spatial and Temporal Extent of Detrimental Effects
The spatial and temporal extent of stressor effects includes the near-field to far-field range of
observable effects of the stressors on a water body. Such information can be conveyed by biological
assessments provided the spatial density of sampling sites is sufficient to convey changes along a
pollution continuum (USEPA 2013a). Use of a continuum provides a method for determining the severity
(i.e., departure from the desired state) and extent (i.e., distance over which adverse effects are
observed) of an impairment from one or more sources. Yoder et al. (2005) detailed this approach in
their historical assessment of large rivers in Ohio. As with attribute VIM above, attribute IX has not yet
been developed and applied in BCG models for specific streams and wadeable rivers. It is included for
future development and application. State scientists involved in the development of the BCG conceptual
model stated that this attribute was important to include for future testing and development. Some
state biological monitoring and assessment programs document the spatial and temporal extent of
stressor effects and use the information to predict the recovery potential of a degraded stream, as well
as the risk of degradation in high quality streams. This information informs water quality management
decisions on prioritization of actions. The National Hydrography Dataset (NHD) (USGS 2014), together
with biological assessment information from attributes I-VIII can be an important tool to help evaluate
position and extent of condition and stressors in a water body or watershed by mapping the locations
(i.e., spatial distribution) of the biological samples.
Attribute X: Ecosystem Connectance
Attribute X refers to the access or linkage (in space/time) to materials, locations, and conditions
required for maintenance of interacting populations of aquatic life. It is the opposite of fragmentation
and is necessary for persistence of metapopulations and natural flows of energy and nutrients across
ecosystem boundaries. Ecosystem connectance can be indirectly expressed by certain species that
depend on the connectance, or lack of connectance, within an aquatic ecosystem to fully complete their
life cycles and thus maintain their populations. Diadromous fish species are one such example—their
absence or presence can provide information on the presence or absence of critical habitats to support
different life stages. However, the inverse of connectance, isolation, is important for some species (e.g.,
amphibians, which are negatively impacted by fish that gain access to amphibian habitat via artificial or
natural connections). This difference dependence upon connectance underscores the importance of
well-defined BCG levels 1 and 2 as the benchmark for interpreting change in the BCG attributes. The
NHD can be an important tool to evaluate the extent of connections (or occurrence of barriers or habitat
disconnects) in a water body or watershed. A habitat mosaic measure is being evaluated as an indicator
of ecosystem connectance in the estuarine BCG (see Appendix B-2).
2.3 The Biological Condition Gradient Levels of Biological Condition
The BCG has been divided into six levels along a generalized stressor-response continuum to provide
discrimination of different levels of condition that are detectable, given current assessment methods
and well-designed monitoring protocols. Since the BCG is a continuum, in principle it is possible to
determine more or fewer levels depending upon the discriminatory power of a state water quality
management program (USEPA 2013a). The six levels are proposed as a hypothetical framework for
which the practical concerns of the state would determine the number of levels that can be
implemented. For example, in most forested perennial stream ecosystems it may be technically possible
to discriminate six classes in the condition gradient, ranging from undisturbed to highly disturbed
conditions (Davies and Jackson 2006). However, some states or regions may only be capable of
discriminating two or three levels, given current technical program capabilities, while others might be
capable of discerning six or more levels based on highly proficient programs and robust data sets (USEPA
17
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
2013a). In addition, some regions of the country may not currently support level 1 water conditions.
Regardless of the number of levels a state can detect, the BCG framework is to be a starting point for a
state to think about how to use biological information to better determine existing conditions and
potential for improvement and how to use the information to better communicate biological condition
and to set water quality objectives.
The six levels of the BCG are described as follows (modified from Davies and Jackson 2006).
• Level 1, Natural or native condition—Native structural, functional, and taxonomic integrity is
preserved; ecosystem function is preserved within the range of natural variability. Level 1
represents biological conditions as they existed (or still exist) in the absence of measurable
effects of stressors and provides the basis for comparison to the next five levels. The level 1
biological assemblages that occur in a given biogeophysical setting are the result of adaptive
evolutionary processes and biogeography. For this reason, the expected level 1 assemblage of a
stream from the arid southwest will be very different from that of a stream in the northern
temperate forest. The maintenance of native species populations and the expected natural
diversity of species are essential for levels 1 and 2. Non-native taxa (attribute VI) might be
present in level 1 if they cause no displacement of native taxa, although the practical
uncertainties of this provision are acknowledged (see section 2.2). Attributes I and II (i.e.,
historically documented and sensitive taxa) can be used to help assess the status of native taxa
when classifying a site or assessing its condition.
• Level 2, Minimal changes in structure of the biotic community and minimal changes in
ecosystem function—Most native taxa are maintained with some changes in biomass and/or
abundance; ecosystem functions are fully maintained within the range of natural variability.
Level 2 represents the earliest changes in densities, species composition, and biomass that occur
as a result of slight elevation in stressors (e.g., increased temperature regime or nutrient
pollution). There might be some reduction of a small fraction of highly sensitive or specialized
taxa (attribute II) or loss of some endemic or rare taxa as a result. The occurrence of non-native
taxa should not measurably alter the natural structure and function and should not replace any
native taxa. Level 2 can be characterized as the first change in condition from natural, and it is
most often manifested in nutrient-polluted waters as slightly increased richness and density of
either intermediate sensitive and intermediate tolerant taxa (attributes III and IV) or both. These
early response signals have been observed in many state programs as illustrated in Figure 5,
which shows slight to moderate increases of mayfly density in response to increases in
conductivity in Maine streams. Mayfly taxa typically have been identified in Maine as sensitive
ubiquitous taxa and show an increase to initial levels of some stress (e.g., an increase in
conductivity or nutrient pollution), followed by a decrease in abundance as stress levels
continue to rise.
18
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
uw
500
o 400
(0
^^
o
c
3 300
n
<
£ 200
i
100
0
>
^
0
0
o
o
E
D
D
D
-
-
0-30 30-60 60-100 100-200 >200
I Non-Outlier Max
Non-Outlier Min
CZI 75%
25%
D Median
o Outliers
- Extremes
Conductivity
Figure 5. Response of mayfly density to stress in Maine streams as indicated by a gradient of increasing
conductivity.
• Level 3, Evident changes in structure of the biotic community and minimal changes in
ecosystem function—Evident changes in structure due to loss of some highly sensitive native
taxa; shifts in relative abundance oftaxa, but sensitive-ubiquitous taxa are common and
relatively abundant; ecosystem functions are fully maintained through redundant attributes of
the system. Level 3 represents readily observable changes that, for example, can occur in
response to organic pollution or increased temperature. The "evident" change in structure for
level 3 is interpreted to be perceptible and detectable decreases in highly sensitive taxa
(attribute II), and increases in sensitive-ubiquitous taxa or intermediate organisms (attributes III
and IV). Attribute IV taxa (intermediate intolerance) might increase in abundance as an
opportunistic response to nutrient or organic inputs.
• Level 4, Moderate changes in structure of the biotic community with minimal changes in
ecosystem function—Moderate changes in structure due to replacement of some intermediate
sensitive taxa by more tolerant taxa, but reproducing populations of some sensitive taxa are
maintained; overall balanced distribution of all expected major groups; ecosystem functions
largely maintained through redundant attributes. Moderate changes of structure occur as
stressor effects increase in level 4. A substantial reduction of the two sensitive attribute groups
(attributes II and III) and replacement by more tolerant taxa (attributes IV and V) might be
observed. A key consideration is that some attribute III sensitive taxa are maintained at a
reduced level, but they are still an important functional part of the system (i.e., function is
19
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
maintained). While total abundance (density) of organisms might increase, no single taxa or
functional group should be overly dominant.
• Level 5, Major changes in structure of the biotic community and moderate changes in
ecosystem function—Sensitive taxa are markedly diminished or missing; conspicuously
unbalanced distribution of major groups from those expected; organism condition shows signs of
physiological stress; ecosystem function shows reduced complexity and redundancy; increased
build-up or export of unused materials. Changes in ecosystem function (as indicated by marked
changes in food-web structure and guilds) are critical in distinguishing between levels 4 and 5.
This could include the loss of functionally important sensitive taxa and keystone taxa (attribute I,
II, and III taxa), such that they are no longer important players in the system, though a few
individuals may be present. Keystone taxa control species composition and trophic interactions,
and are often, but not always, top predators. As an example, removal of keystone taxa by
overfishing has greatly altered the structure and function of many coastal ocean ecosystems
(Jackson et al. 2001). Additionally, tolerant non-native taxa (attribute VI) may dominate some
assemblages, and changes in organism condition (attribute VII) may include significantly
increased mortality, depressed fecundity, and/or increased frequency of lesions, tumors, and
deformities.
• Level 6, Severe changes in structure of the biotic community and major loss of ecosystem
function—Extreme changes in structure; wholesale changes in taxonomic composition; extreme
alterations from normal densities and distributions; organism condition is often poor; ecosystem
functions are severely altered. Level 6 systems are taxonomically depauperate (i.e., low diversity
and/or reduced number of organisms) compared to the other levels. For example, extremely
high or low densities of organisms caused by excessive organic pollution, severe toxicity, and/or
severe habitat alteration may characterize level 6 systems. Non-native taxa may predominate.
20
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
2.3.1 Bringing the Biological Condition Gradient Levels and Attributes Together
The BCG narrative portrays general patterns of biological and ecological response common across
regions, as measured by the BCG attributes. Table 2 organizes the ten BCG attributes into six categories:
community structure, non-natives, condition, function, landscape, and connectivity. Attributes I through
V have been combined in one category in Table 2—structure and compositional complexity. This
category typically includes measures of the number, type, and proportion of individual taxa within an
assemblage (e.g., benthic macroinvertebrates, fish, and algal assemblages). These attributes are the
foundation of most state biological assessment programs for streams and wadeable rivers. The five
taxonomic attributes characterize biological sensitivity to the cumulative impact of stressors (e.g.,
highly, intermediate, or tolerant taxa). In addition to the sensitivity-based attributes, biologists have also
used assemblage richness and balance, assemblage abundance or biomass, and keystone or habitat-
structuring species (e.g., reef-building corals) to define attributes and distinguish levels of condition
along a stress gradient. Attributes respond to stressors in distinctly different ways so that there are
predictive, quantitative measures along the full range of stress levels (Figure 6, Table 3). Defining and
quantifying these changes along the full gradient of stress effects is necessary for developing reliable,
predictable measures for incremental changes in biological condition. For example, highly sensitive taxa
might disappear from a community in early, or low, levels of stress. Tolerant taxa might become more
dominant as stress increases, not only because they might thrive, but also because there are fewer
sensitive species and the proportion of tolerant taxa in the entire community increases. Intermediate
tolerant taxa might not provide a
significant signal under most
conditions if they are present under
a wide range of stress. However, the
absence of this group of taxa in
highly stressed conditions can help
document highly disturbed
conditions, and their reappearance
may indicate initial response to
management actions for restoration.
As work proceeds on applying the
BCG to other water body types and
developing approaches for including
additional assemblages (e.g.,
periphyton, amphibians, birds) and
new methods for sampling and
analyzing aquatic life (e.g., DNA
analysis), it is expected that these
attributes will be refined and
comparable detailed descriptions for
the remaining attributes will
emerge.
HIGH
re
u
m
0
I
O
CQ
Abundance
LOW
Stress Gradient
HIGH
Figure 6. Hypothetical examples of biological response to the
cumulative impact of multiple stressors.
21
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 2. BCG: Ecological Attributes
LLI
ce.
•3
t
^
fe
Attribute
Grouping
Structure and
Compositional
Complexity
(Attributes I-V)
See Table 3 for
detailed
descriptions
for these
attributes.
Description
Community or habitat
structure and complexity.
May also recognize loss of
habitats or species due to
human activities.
Examples include
macroinvertebrate or fish
indices, phytoplankton or
zooplankton community
measures, epifaunal
measures, biotope
mosaics,
presence/quantity of
sensitive taxa or biotopes,
wetland vegetative
indices, etc.
Examples of BCG
1
Community
composition is as
naturally occurs,
except for global
extinctions based
on observations
from water bodies
with similar habitat
and ecoregion
without
measurable human-
caused stressors
(this includes
chlorophyll a levels,
biotope mosaics,
species
composition
including large,
long-lived, and
sensitive species;
patterns of
vegetation are as
naturally occurs)
2
Minor changes in
natural occurrences
of biotopes or
patterns of
vegetation, slight
decreases in
sensitive species,
and slight increases
in tolerant species
3
Evident changes in
biological metrics
(decreases in
sensitive species
and increases in
more tolerant
species, evident
changes in
vegetation
patterns); may be
slight decreases in
biotope or habitat
area; biotope
mosaic basically
intact
4
Significant changes
in biological metrics
(marked decreases
in sensitive species
[including large or
long-lived taxa] and
increases in tolerant
species, evident
changes in
vegetation
patterns); biotope
mosaic slightly
altered with
replacement of
natural
habitats/biotopes
with tolerant or
non-natu rally
occurring
components;
detectable loss in
some biotope types
or habitat area
5
Most sensitive,
large and/or long-
lived taxa are
absent, with a
dominance in
abundance of
tolerant taxa;
significant shifts in
species diversity,
size, and densities
of remaining
species; biotope
mosaic significantly
altered with many
natural
habitats/biotopes
lost with
replacement by
tolerant or non-
natu rally occurring
components;
evident loss in
biotope or habitat
area
6
Sensitive, large,
and/or long-lived
taxa largely absent;
possible high or low
extremes in
abundance of
remaining taxa;
marked reduction in
species diversity
and in size spectra
of remaining
organisms; near
complete loss or
alteration of natural
biotope mosaic with
marked loss in
biotope or habitat
area
22
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
ft
E
<
z
Z
o
H
Q^
Z
o
u
Attribute
Grouping
Non-Native
Taxa
(Attribute VI)
Organism
Condition
(Attribute VII)
Description
Status of non-native
species. May include
measures of the impact of
invasive and non-native
species.
Examples include
estimated numbers of
species or individuals,
relative density or
biomass measures of
natives and non-natives,
or replacement of native
species
Measures condition of
individual organisms,
including anomalies and
diseases.
Examples include external
anomalies, lesions,
disease outbreaks (local or
widespread), coral
bleaching, seagrass
condition, fish pathology,
and frequency of diseased
or affected organisms
Examples of BCG
1
Non-native taxa, if
present, do not
significantly reduce
native taxa or alter
structural or
functional integrity
Diseases and
anomalies are
consistent with
naturally occurring
incidents and
characteristics
2
Non-native taxa
may be present, but
occurrence has a
non-detrimental
effect on native
taxa
Diseases and
anomalies are
consistent with
naturally occurring
incidents and
characteristics
3
Non-native taxa
may be prominent
in some
assemblages (e.g.,
crustaceans,
bivalves, fish) and
some sensitive
native taxa may be
reduced or replaced
by equivalent non-
native species (e.g.,
replacement of
native trout with
introduced
salmonids)
Incidences of
diseases and
anomalies may be
slightly higher than
expected conditions
4
Increased
abundance of
tolerant non-native
species (e.g.,
Common Carp, non-
native centrarchids,
Common Reed) or
native species (e.g.,
salmonids) only
maintained by
regular stocking
Incidences of
diseases and
anomalies are
slightly higher than
expected. For
example, coral
bleaching events
may occur
sporadically and
result in slightly
elevated mortality.
Anomalies in fish
occur in a small
fraction of a
population
5
Some assemblages
(e.g., mollusks,
fishes,
macrophytes) are
dominated by
invasive non-native
taxa (e.g., Silver
Carp, Zebra
Mussels, Eurasian
Watermilfoil); or
increasing
dominance by
tolerant non-native
species such as
Common Carp
Disease outbreaks
are increasingly
common, anomalies
are increasingly
common,
particularly in long-
lived taxa where
biomass may also
be reduced (e.g.,
bleaching events
are frequent
enough to cause
mortality of corals).
Anomalies, such as
deformities,
erosion, lesions, and
tumors in fish, occur
in a measurable
fraction of a
population
6
Same as level 5; not
distinguishable
based on non-native
species alone
Host species in
which diseases and
anomalies have
been observed are
now absent, so
diseases might be
difficult to detect.
Anomalies, disease,
etc. may occur
across multiple
species or taxa
groups
23
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
z
o
h
u
D
u.
UJ
Q.
U
O
z
5
Attribute
Grouping
Function
(Attribute VIII)
Spatial and
Temporal
Extent of
Detrimental
Effects
(Attribute IX)
Description
Measures of energy flow,
trophic linkages and
material cycling. They may
include proxy or snapshot
structural metrics that
correlate to functional
measures.
Examples include
photosynthesis:
respiration ratios, benthic:
pelagic production rates,
chlorophyll a
concentrations,
macroalgal biomass,
bacterial biomass and
activity
Measures of a landscape's
capacity, contributing
surface water to a single
location, to maintain the
full range of ecological
processes and function
that support a resilient,
naturally occurring
aquatic community. The
functions and processes to
be measured include
hydrologic regulation,
regulation of water
chemistry and sediments,
hydrologic connectivity
(see also attribute X),
temperature regulation,
and habitat provision
Examples of BCG
1
Energy flows,
material cycling,
and other functions
are as naturally
occur;
characterized by
complex
interactions and
long-lived links
supporting large,
long-lived
organisms
N/A— A natural
disturbance regime
is maintained
2
Energy flows,
material cycling,
and other functions
are within the
natural range of
variability;
characterized by
complex
interactions and
long-lived links
supporting large,
long-lived
organisms
Limited to small
pockets and short
duration
3
Virtually all
functions are
maintained through
operationally
redundant system
attributes, minimal
changes to export
and other indicative
functions. Some
functions increased
due to pollution or
low level
disturbance (e.g.,
production,
biomass,
respiration)
Limited to a local
area or within a
season
4
Most functions are
maintained through
operationally
redundant system
attributes, though
there is evidence of
loss of efficiency
(e.g., increased
export or decreased
import, there may
be shifts in benthic:
pelagic production
rates
Mild detrimental
effects may be
detectable beyond
the local area and
may include more
than one season
5
Loss of some
ecosystem functions
are manifested as
changed export or
import of some
resources and
changes in energy
exchange rates
(photosynthesis:
respiration ratios,
benthic: pelagic
production rates,
respiration or
decomposition
rates)
Detrimental effects
extend far beyond
the local area
leaving only a few
islands of adequate
conditions; effect
extends across
multiple seasons
6
Most functions
show extensive and
persistent
disruption, shifts to
primary production,
microbial
dominance, fewer
and shorter-length
trophic links and
highly simplified
trophic structure,
marked shifts in
benthic: pelagic
production rates
Detrimental effects
may eliminate all
refugia and
colonization sources
within a region or
catchment and
affect multiple
seasons
24
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
t
t~
LLJ
z
8
Attribute
Grouping
Ecosystem
Connectance
(Attribute X)
Description
Observations of exchange
or migrations of biota
between adjacent water
bodies or habitats.
Important measures
within the area being
studied may be strongly
affected by factors
adjacent to or larger than
the immediate study area.
Metrics may include
dams, causeways,
fragmentation measures,
hydrological measures, or
proxies such as
characteristic migratory
species
Examples of BCG
1
System is naturally
connected, or
disconnected, in
space and time,
exchanges,
migrations, and
recruitment from
adjacent water
bodies or habitats
are as naturally
occurs
2
System is naturally
connected, or
disconnected, in
space and time,
exchanges,
migrations, and
recruitment from
adjacent water
bodies or habitats
are as naturally
occurs
3
Slight loss, or
increase, in
connectivity
between adjacent
water bodies or
habitats (e.g.,
between upstream
and downstream
water bodies), but
colonization
sources, refugia,
and other
mechanisms mostly
compensate. May
also be increase in
connectivity due to
canals, interbasin
transfers
4
Some loss, or
increase, in
connectivity
between adjacent
water bodies or
habitats (e.g.,
between upstream
and downstream
water bodies), but
colonization
sources, refugia,
and other
mechanisms
prevent complete
disconnects or
other failures
5
Significant loss, or
increase, in
ecosystem
connectivity
between adjacent
water bodies or
habitats (e.g.,
between upstream
and downstream
water bodies or
habitats) is evident;
recolonization
sources do not exist
for some taxa, some
near-complete
disconnects or
connect exist
6
For many groups, a
complete loss in
ecosystem
connectivity in at
least one dimension
(either spatially or
temporally) lowers
reproductive or
recruitment success
or prevents
migration or
exchanges with
adjacent water
bodies or habitats,
frequent
disconnects or
other failures. For
other groups, a
complete loss in
ecosystem
disconnect in at
least one dimension
lowers reproductive
or recruitment
success (e.g.,
predation of
amphibians by fish
in once isolated
headwater streams)
25
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 3. BCG Matrix: Taxonomic Composition and Structure Attributes I-V
Ecological
Attributes
1
Historically
documented,
sensitive, long-
lived or
regionally
endemic taxa
II
Highly
sensitive taxa
III
Intermediate
sensitive taxa
IV
Intermediate
tolerant taxa
V
Tolerant taxa
BCG Levels
1
Natural or native
condition
Native structural,
functional, and
taxonomic integrity
is preserved;
ecosystem function
is preserved within
the range of natural
variability
As predicted for
natural occurrence
except for global
extinctions
As predicted for
natural occurrence,
with at most minor
changes from
natural densities
As predicted for
natural occurrence,
with at most minor
changes from
natural densities
As predicted for
natural occurrence,
with at most minor
changes from
natural densities
As naturally occur,
with at most minor
changes from
natural densities
2
Minimal changes in
the structure of the
biotic community
and minimal
changes in
ecosystem function
Virtually all native
taxa are maintained
with some changes
in biomass and/or
abundance;
ecosystem functions
are fully maintained
within the range of
natural variability
As predicted for
natural occurrence
except for global
extinctions
Most are
maintained with
some changes in
densities
Present and may be
increasingly
abundant
As naturally present
with slight increases
in abundance
As naturally present
with slight increases
in abundance
3
Evident changes in
structure of the
biotic community
and minimal
changes in
ecosystem function
Some changes in
structure due to loss
of some rare native
taxa; shifts in
relative abundance
of taxa but
sensitive-
ubiquitous taxa are
common and
abundant-
ecosystem functions
are fully maintained
through redundant
attributes of the
system
Some may be
marginally present
or absent due to
global extinction or
local extirpation
Some loss, with
replacement by
functionally
equivalent sensitive-
ubiquitous taxa
Common and
abundant; relative
abundance greater
than sensitive-rare,
taxa
Often evident
increases in
abundance
May be increases in
abundance of
functionally diverse
tolerant taxa
4
Moderate changes
in structure of the
biotic community
and minimal
changes in
ecosystem function
Moderate changes
in structure due to
replacement of
some sensitive-
ubiquitous taxa by
more tolerant taxa,
but reproducing
populations of some
sensitive taxa are
maintained; overall
balanced
distribution of all
expected major
groups; ecosystem
functions largely
maintained through
redundant
attributes
Some may be
marginally present
or absent due to
global, regional, or
local extirpation
May be markedly
diminished
Present with
reproducing
populations
maintained; some
replacement by
functionally
equivalent taxa of
intermediate
tolerance.
Common and often
abundant; relative
abundance may be
greater than
sensitive- ubiquitous
taxa
May be common
but do not exhibit
significant
dominance
5
Major changes in
structure of the
biotic community
and moderate
changes in
ecosystem function
Sensitive taxa are
markedly
diminished;
conspicuously
unbalanced
distribution of major
groups from that
expected; organism
condition shows
signs of
physiological stress;
system function
shows reduced
complexity and
redundancy;
increased build- up
or export of unused
materials
Usually absent
Usually absent or
only scarce
individuals
Frequently absent
or markedly
diminished
Often exhibit
excessive
dominance
Often occur in high
densities and may
be dominant
6
Severe changes in
structure of the
biotic community
and major loss of
ecosystem function
Extreme changes in
structure; wholesale
changes in
taxonomic
composition;
extreme alterations
from normal
densities and
distributions;
organism condition
is often poor;
ecosystem functions
are severely altered
Absent
Absent
Absent
May occur in
extremely high or
extremely low
densities; richness
of all taxa is low
Usually comprise
the majority of the
assemblage; often
extreme departures
from normal
densities (high or
low)
26
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
2.4 How the Conceptual Biological Condition Gradient was Developed, Tested,
and Evaluated
The conceptual BCG model was developed and tested by an expert workgroup primarily composed of
scientists from government and the research community (Davies and Jackson 2006). This section
summarizes how the BCG conceptual model was tested to the satisfaction of the expert workgroup and
peer reviewers (from Davies and Jackson 2006). For examples on constructing BCG models and
quantitative decision rules applied to specific assemblages and habitats, please see Chapters 3 and 4.
A matrix was created that summarized biologists' experience and knowledge about how biological
attributes change in response to stress in aquatic ecosystems (Davies and Jackson 2006). In building the
model, the workgroup followed an iterative, inductive approach, similar to means-end analysis
(Martinez 1998). The workgroup understood that the primary value of the model is as a tool for shared
learning and as a framework for communication.
The workgroup began by testing whether biologists from different parts of the country would draw
similar conclusions regarding the condition of a water body using simple lists of organisms and their
counts. This approach was initially based on Maine's experience, in which expert biologists
independently assigned samples of macroinvertebrates to a priori defined levels of biological condition
defined by differences in assemblage attributes (Davies et al. 1995; Davies et al. In press; Shelton and
Blocksom 2004).
To provide a functional framework for practitioners, the workgroup described how each of the 10
attributes varies across six levels of biological condition along a gradient of increasing anthropogenic
stress (i.e., human disturbance). The general model was then described in terms of the biota of a specific
region (Maine). Based on 20 years of monitoring data, the Maine BCG describes how the relative
densities of specific taxa, with varying sensitivities to stress, change across the BCG levels of condition
(Davies and Jackson 2006).
To test the general applicability of the BCG to sampling data taken from other stream systems across the
country, the workgroup evaluated how consistently individual biologists classified samples of aquatic
biota based on the attributes incorporated into the BCG. Government, field, and research biologists
participated in the data exercise. The full workgroup was divided into breakout groups according to
region (northeast, south-central, northwest, arid southwest/great plains) and assemblage (fish or
invertebrates) expertise. Samples were selected from invertebrate and fish data sets to span as many of
the BCG levels as possible (i.e., to span the full gradient of conditions). The invertebrate samples and fish
samples used in the tests were collected from six different regions within the U.S. (northeast, mid-
Atlantic, southeast, northwest, southwest, central) and included only basic descriptors of stream
physical characteristics (e.g., substrate, velocity, width, depth), taxonomic names, densities, and in some
cases, metric values. These data represent the basic core elements common to nearly all biological
monitoring programs. Participants were asked to place each sample into one of the six condition levels,
and they were cautioned not to apply a simple relative quality ranking since all six levels did not
necessarily occur within the data sets. Biologists relied primarily on differences in relative abundances
and sensitivities of taxa (i.e., attributes I-VI) to make level assignments, because this was the
information typically collected in state monitoring programs and the data needed to evaluate the status
of the other attributes were not available. Percent concurrence among the individuals was calculated to
assess the level of agreement among biologists when applying the BCG to raw data. Perfect concurrence
was set to equal the product of the number of raters by the number of streams.
27
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
In the first stage of the data exercise, between-biologist differences were evaluated by asking all
workgroup participants to rate a single data set of 6-8 samples. The breakout groups were then asked
to classify samples from larger and more variable data sets. The groups were also instructed to
summarize their interpretations and to identify biological responses to changes in conditions not
captured by the BCG. Finally, the workgroup participants identified how, from their perspectives, the
BCG levels corresponded to the CWA biological integrity objective and interim goal for protection of
aquatic life (e.g., protection and propagation offish, shellfish, and wildlife).
Overall, workgroup members independently agreed on placement of sites in the same BCG levels for
82% of the benthic macroinvertebrate samples and 74% of the fish samples. When assignments differed,
the range of variation among workgroup members was within one level in either direction for all
samples with a few exceptions. BCG levels were revised following full workgroup discussion so that
transitions were more distinct.
Each of the breakout groups independently reported that the ecological characteristics corresponding to
BCG levels 1, 2, 3 and either some or all of BCG level 4 characteristics were generally compatible with
how they assess the CWA's interim goal for protection of aquatic life. The experts unanimously agreed
that BCG levels 1, 2, and 3 attained the CWA goal and BCG levels 5 and 6 did not. Opinions differed
among the experts on whether all or some aspects of BCG level 4 characteristics were compatible with
attaining this goal. For example, the workgroup extensively discussed what constituted an acceptable
degree of replacement of sensitive taxa by tolerant taxa. However, experts united in a clear consensus
that the BCG process provided detailed, readily transparent documentation of the expert logic and
underlying science for establishing BCG levels. Additionally, expert discussion on implementation of the
BCG framework to interpretation of condition included the following programmatic considerations:
• The technical rigor of the monitoring program that produced the condition assessments—
Conceptually, a less rigorous monitoring program produces assessments with a greater degree
of uncertainty, or precision, and potentially lower accuracy. In lieu of improving the program's
technical rigor, or to compensate for uncertainty associated with monitoring programs of lower
technical rigor, some experts recommended that a more protective, e.g., conservative, BCG level
be used to measure attainment of the CWA ALL) goal.
• Protection of high quality conditions—The experts identified the characteristics described by
BCG levels 1 and 2 as consistent with their understanding of the CWA "biological integrity"
objective. Concern was expressed that a single threshold comparable to BCG level 4 is not
protective of high ecological quality and that water bodies comparable to BCG levels 1, 2, or 3
would likely decline significantly before action would be triggered to address sources of
degradation. Experts noted that restoration and remediation costs are typically much higher
than costs for prevention. Experts recommended that multiple thresholds protective of existing
ALL) conditions be established (e.g., thresholds comparable to BCG levels 2, 3, or 4).
Alternatively, if only a single threshold is established, some experts recommended that the
threshold should be protective of higher level conditions comparable to BCG level 3.
Workgroup members reported that key concepts were important with respect to classifying samples
into levels and identifying the boundaries in between. For levels 1 and 2, biologists identified the
maintenance of native species populations as essential to their understanding of biological integrity.
Although many participants noted that methods for distinguishing differences between levels in
attribute VIM (ecosystem function) were poorly defined, most nevertheless identified ecosystem
28
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
function changes (as inferred by marked changes in food-web structure and guilds) as critical in
distinguishing between levels 4 and 5.
Discussion following the data exercise revealed that participants readily agreed on some of the BCG
attributes, but not others. For example, participants indicated they mostly used attributes I-V
(taxonomic composition, pollution sensitivity), attribute VI (non-native taxa, for levels 2-6 only), and
attribute VII (organism condition) to evaluate biological conditions in streams and wadeable rivers. In
contrast, because attributes VIII-X (ecosystem function and scale-dependent features) are rarely directly
assessed by biologists, the evaluation of these attributes was accompanied by relatively high
uncertainty. Even so, workgroup members strongly advocated retaining these attributes in the BCG
because of the importance of this information in making restoration decisions. There was considerable
discussion regarding to which axis, the biological or stress axis, the attributes for ecosystem
connectance and spatial and temporal extent of detrimental effect should be assigned. As an interim
measure, the workgroup members recommended including these attributes as components of the
biological axis primarily because of the importance state biologists placed on this information in
predicting restoration or protection success. The BCG, thus, serves as a guide to interpret condition and
is expected to be further refined as development and application continues.
The presence of non-native taxa in level 1 was also the subject of considerable discussion. Knowledge of
the extensive occurrence of some non-native taxa in otherwise near-pristine systems conflicted with the
desire by many to maintain a conceptually pure and natural level. Further discussion resulted in
agreement that the presence of non-native taxa in level 1 is permissible only if they cause no
displacement of native taxa, although the practical uncertainties of this provision were acknowledged.
The resulting level descriptions, which allow for non-native species in the highest levels as long as there
is no detrimental effect on the native populations, has practical management implications. For example,
introduced European brown trout (Salmo trutta) have replaced native brook trout (Salvelinusfontinalis)
in many eastern U.S. streams. In some catchments, brook trout only persist in stream reaches above
waterfalls that are barriers to brown trout. The downstream reaches can be nearly pristine except for
the presence of brown trout. In these places, if society decided to remove the introduced brown trout,
and if stream habitat is preserved throughout the catchment, brook trout can potentially repopulate
downstream reaches. In the use designation process, recognizing that the entire catchment has the
potential to attain level 1 condition will inform the public that a very high quality resource exists.
Critical gaps in knowledge and scientific literature were uncovered during the development of the BCG.
For example, the workgroup identified the need for regional evaluations of species tolerance to
stressors. Tolerance information presented in the current version of the BCG tends to be based on
generalized taxa responses to a non-specific stressor gradient. At that time, tolerance information was
not available for most taxa and for many common stressors (temperature, nutrients, and sediments). In
some cases, tolerance values are based on data collected in other geographic regions or for other
purposes (e.g., van Dam's European diatom tolerances are used for North American taxa) (van Dam et
al. 1994). In the future, availability of improved tolerance value information can be used to refine the
BCG and improve its precision (e.g., Cormier et al. 2013; Whittier and Van Sickle 2010).
Additionally, taxa that are considered tolerant to stressors in one region of the country may not be
similarly classified in another region. For example, long-lived taxa have generally been characterized as
sensitive to increasing pressure and tend to be replaced by short-lived taxa in stressed systems. As such,
the presence of long-lived taxa in a water body has been used to indicate high quality conditions,
whereas the predominance of short-lived taxa may indicate degradation. However, in streams in the
29
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
arid western U.S., extreme changes in hydrology might define the natural regime and an opposite trend
might be observed: short-lived taxa can dominate the biological community in natural settings. In these
systems, a shift to long-lived taxa may be an indicator of altered, less variable flow regimes due to flow
management.
When the expert workgroup was initially developing the conceptual BCG framework (2000-2004),
attributes VIII-X were not routinely measured as part of a state biological monitoring and assessment
program. However, the state scientists participating in the workgroup deemed these attributes as
ecologically important because the extent of ecosystem alteration has important environmental
implications in terms of an individual water body's vulnerability to further effects from stressors, as well
as potential for mitigation (Davies and Jackson 2006). The state scientists explained that they informally
estimated ecosystem function, connectance, and extent of detrimental effects using different surrogate
measures (e.g., shift in functional feeding groups) and/or measures of watershed condition (e.g.,
presence and connection of wetlands and streams, intact forests). This information was used to inform
decisions on recovery potential for a water body and prioritize actions to protect high quality conditions.
Additionally, attributes IX and X might play an important role in evaluating longer term impacts,
restoration potential, and recoveries. For example, ecosystem connectivity is fundamental to the
successful recruitment into and maintenance of organisms in any environment. A single impacted
stream reach in an otherwise intact watershed has far more restoration potential than a similar site in a
basin that has undergone extensive landscape alteration.
A critical gap that was not discussed in 2005, but is now an area of intensive work, is predicting the
impacts of climate change on aquatic systems. Gaining an understanding of how the BCG attributes (I-X)
will behave under future climate scenarios, and developing approaches and indicators to measure these
impacts, will be important future work for improving the BCG.
2.5 Conclusion
The conceptual BCG framework is a tool to help state water quality management programs better
describe their ALL) goals and measure increments of change in biological condition along a full gradient
of stress—and to use that information to interpret existing conditions, identify high quality waters, and
track progress towards achieving desired improvements. The BCG provides a common interpretative
framework to assist in comparability of results across jurisdictional (e.g., county, state, national) and
program (e.g., water quality and natural resource agencies) boundaries and to communicate this
information to the public. In order to use the BCG, states will need to calibrate it to their own habitats
and monitoring data and develop a numeric model. Although the BCG is a universal conceptual
framework, quantitative calibrations are regionally data set-specific. Additionally, as an added benefit,
state water quality management programs have reported that using expert consensus in developing
BCGs has proven to be a valuable training tool for their technical staff and field crews. The panel
interactions and development of consensus in interpreting data directly contribute to a more uniform
approach and shared understanding of the aquatic ecosystems for which the state is responsible.
Chapters 3 and 4 describe how a quantitative BCG model can be developed using expert panels and
different approaches for quantification of the conceptual framework.
30
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Chapter 3. Calibration of Biological Condition Gradient Models
The purpose of calibrating the BCG is to populate the conceptual model with quantitative data, develop
quantitative decision rules to assign sites to BCG levels, and build a bridge from that model to
management goals and endpoints. A calibrated BCG has both a narrative and a quantitative scientific
description applicable to specific ecological regions orsubregions. The BCG level descriptions can be used
to describe the biological conditions associated with specific management goals and to support
biological criteria development. The scientific description of the BCG can help make the management
goals transparent to both decision makers and stakeholders. It can be used to assess baseline conditions
and track incremental changes in condition.
This chapter proposes an approach to develop detailed narrative descriptions of BCG levels and
attributes. Description and calibration of the BCG are achieved through consensus of expert opinion
(Figure 7). The experts define the attributes, and the changes in those attributes, that characterize BCG
levels and signal shifts to a different level. The outcome is a multiple attribute decision model that
simulates the consensus expert decisions based on a set of quantitative rules. The next chapter provides
three approaches to quantify the narrative BCG and develop numeric thresholds for site assignments.
Figure 7. Benthic macroinvertebrate and fish experts developing decision rules for freshwater streams in
Alabama.
Use of professional expert consensus has a long pedigree in the medical field, including the National
Institutes of Health (NIH) Consensus Development Conferences to recommend best practices for
diagnosis and treatment of diseases.5 In addition to the NIH consensus conferences, other researchers,
institutes, and countries develop medical consensus statements, using both the NIH methods and others
(Nairetal. 2011).
Recent environmental assessments developed using professional judgment have shown that experts are
highly concordant in their ratings of sites, including marine benthic invertebrate communities in
California bays (Weisberg et al. 2008). Another example is in nearshore marine environments assessed
by an international panel covering European Atlantic, Mediterranean, American Atlantic, and American
5 The program ran from 1977 to 2013. For more information, see: http://consensus.nih.gov/. Accessed February
2016.
31
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Pacific habitats and experts (Teixeira et al. 2010). The approach has also been demonstrated effective
for developing assessments of sediment quality (Bay et al. 2007; Bay and Weisberg 2010) and a decision
model for fecal contamination of beaches (Cao et al. 2013). Likewise, in BCG development, aquatic
biologists have come to very tight consensus on the descriptions of individual levels of the BCG, as well
as very close agreement on the BCG level assigned to individual sites (e.g., Danielson et al. 2012; Davies
and Jackson 2006; Gerritsen and Jessup 2007a; Gerritsen and Leppo 2005; Gerritsen and Stamp 2012;
Gerritsen et al. 2013; Jessup and Gerritsen 2014; Kashuba et al. 2012; Snook et al. 2007).
All scientific and technical products, including biological indices used for assessment, include results of
professional judgment and assumptions throughout (Scardi et al. 2008; Steedman 1994). The BCG expert
consensus approach asks the experts to make judgments on the biological significance of changes in the
attributes identified in Chapter 2. For this approach to be credible and valid, the panel should be
comprised of experts with a wide and deep breadth of knowledge and expertise and not be constrained
to a single agency in order to minimize internal bias. Additionally, it is essential that the expert logic in
developing the decisions be fully documented so the rules will be transparent and understandable to
those that were not engaged in the expert panel. The objective is to develop a set of decision rules that
can be implemented by others not engaged in the expert panel.
3.1 Overview
The first step in calibration of the BCG is to develop detailed narrative descriptions of BCG levels and
attributes specific to the water body type and region. Experts are given assemblage species composition
and abundance data sets from the region for which they are developing the BCG. In order to minimize
pre-conceived judgments, they are also given physical information about the sites (e.g., catchment area,
slope, elevation, ecoregion, habitat type) but not the precise locations, land uses, sources, and stressor
information. Following discussion of the conceptual model of the BCG, including detailed presentation
on the description of the BCG levels and attributes, the experts are asked to put each sample site into
one of the BCG levels. Each sample is discussed by the group, and facilitators elicit the reasoning used by
the experts in their ratings. The median of the expert ratings is taken as the final BCG level for a sample
(Gerritsen and Jessup 2007; Gerritsen and Leppo 2005; Gerritsen and Stamp 2012; Gerritsen et al. 2013;
Jessup and Gerritsen 2014; Snook et al. 2007).
After an initial rating of at least 30 samples, the experts are asked to begin to articulate rules or
guidelines that they use to make their decisions, starting with the highest level (BCG level 1) and
working through level 6. Data evaluations and site assignments continue as rules are articulated and
then revisited and further tested. In some situations, it may be necessary for the experts to use
historical data and information to develop rules for the highest levels of the BCG when there are no or
few samples in the data set that are representative of undisturbed or minimally disturbed conditions.
Following the expert meetings, organizers and analysts examine the distributions of the quantitative
data with respect to the initial proposed guidelines stated by the experts and the experts' actual BCG
decisions. The distribution analysis forms the basis of quantitative boundaries around the experts'
proposed rules, and analysts in turn develop quantitative rule-based models. Quantitative rules and
performance are in turn reviewed by the expert panels to adjust rules or thresholds as necessary.
Reviews and iterative recalibration are typically carried out by webinar and conference call. The panels
also rate an independent set of test samples that were excluded from the calibration process.
32
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
The outcome of a full BCG calibration, including development of a quantitative model, includes:
• A current state-of-knowledge description of the biological assemblage of water bodies under
pre-development, undisturbed condition to serve as a fixed, historic baseline (the level 1
prototype). If there are no BCG level 1 sites available, then this description may be based on
historical observations, records, and/or data.
• Descriptions of each identified level of the BCG.
• A set of transparent rules for assigning sample sites to levels of the BCG.
• A quantitative model of the rules, or other technical approach, to assign new samples to levels
of the BCG, without reconvening an expert panel.
• A set of BCG condition levels that can serve as management goals for classes of water bodies
and as thresholds for biological criteria, if the state so chooses.
There are several key steps to the calibration process (Figure 8):
• Assemble and organize data—The BCG is developed using information and data from the
state's existing biological monitoring program and/or other data sources (e.g., different data
sets or regional pooled data from other states and federal agencies). The data should cover the
entire range of conditions and stress within at least one ecological region. The data set should
be sufficiently large with a well-defined approach for classification, identification of natural
conditions, and criteria for reference site selection. Usually, the BCG cannot be calibrated within
small jurisdictions or within urban or agricultural regions only—it requires data from outside the
jurisdiction to ensure that the least stressed reference, as well as the full range of other
stressors, are represented.
• Conduct preliminary data analysis/data preparation—Prior to the calibration workshop, the
data must be put in a format that can be readily used by workshop participants. In addition,
stressor-response relationships are examined to describe the responses of the assemblages and
of individual taxa to the stress gradients represented in the data.
• Convene expert panel—The key component of BCG calibration is expert consensus of aquatic
biologists on qualitative and quantitative descriptions of the BCG levels. Experts selected should
be familiar with the water bodies, identities of species, and species and assemblage responses
to stress in the regions of concern. The panel should include experts from not only the state
biological assessment program but other state and federal natural resource agencies and
research scientists from the academic community. Additionally, experts who regularly work with
the regulated community can offer a level of assurance and interpretive assistance about the
purpose and value of using the BCG in water quality assessments.
• Develop quantitative BCG model—Following the development of decision rules, one of several
approaches can be applied to automate assigning water bodies to condition levels in the state
database. Approaches discussed in Chapter 4 of this document include multiple attribute
decision models, multivariate discriminant models, and development of thresholds for
commonly used biological indices (e.g., multimetric indices (MMIs) or predictive model indices
(e.g., observed over expected taxa [O/E])).
• Test models, adjust, and recalibrate—The development process is iterative and may require
several passes through the process to converge on a consistent, locally calibrated BCG that is
scientifically defensible.
33
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
BCG Calibration
Stepl
Assemble data
Step 2
Analyze and prepare data
Identify
expert
panel
Step 3
Convene expert panel
Step 4
Develop decision model
StepS
Test and
review model:
adequate
performance?
Yes
Calibrated BCG model with
quantitative decision rules for
assigning sample sites to BCG levels
Figure 8. Steps in a BCG calibration.
34
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
3.1.1 Case Studies and Applications
Since 2005, several states or other entities (e.g., river basin associations, counties) have either
calibrated, or are in the process of calibrating, the BCG (Table 4). Most of the BCG models that have
been calibrated to date apply to perennial streams that are exposed to increases in temperature,
nutrients, toxic substances, and fine sediments. This is the stream-type and stressor regime originally
described by the conceptual model. Nevertheless, the model has been extended and calibrated to large
rivers (Appendices Bl and B4; Bradley et al. 2014), estuaries (Appendix B2), coral reefs (Appendix B3;
Shumchenia et al. 2015), and lakes (Gerritsen and Stamp 2014). Refinement of the model attributes to
accommodate regional and water body type differences for water bodies other than streams and
wadeable rivers has occurred without loss of model integrity. Thus, the BCG can be applicable to other
aquatic ecosystems and stressors with appropriate modifications.
Section 3.2 below provides a detailed description of the step-by-step process that has been used to
calibrate BCG models. Chapter 4 provides approaches to quantify the expert-derived BCG model, and
case studies drawn from Table 4 illustrate different components of the process.
Table 4. BCG calibration and testing projects
State/Region
Alabama
California
Connecticut
Illinois
Indiana
Maine
Maryland,
Montgomery
County
Water body type
(if applicable)
Highland streams and
wadeable rivers
Coastal plains streams
Streams
High gradient streams
and wadeable rivers
Streams
Streams and rivers
Streams and wadeable
rivers
Wetlands
Streams
Biological Assemblages
Benthic macroinvertebrates
and fish in high gradient
streams
Benthic macroinvertebrates
and fish in low gradient
streams
Algae
Benthic macroinvertebrates
Fish
Benthic macroinvertebrates
and fish
Fish
Algae
Benthic macroinvertebrates
Benthic macroinvertebrates
and fish (quantitative),
salamanders (qualitative)
Objective
Calibrated BCG model and
automated decision model for
invertebrates (all streams) and
fish (3 regions)
Calibrated BCG model and
automated decision model for
invertebrates and fish
Calibrated BCG model and
decision model for stream algae
Calibrated BCG model and
automated decision model; also
calibrated to Connecticut's
macroinvertebrate MMI
Calibrated BCG model and
automated decision model; also
calibrated to Connecticut's fish
MMI
Calibrated BCG model and
automated decision model
Calibrated BCG model and
automated decision model
Calibrated BCG model to assign
ALUs per Maine's 3 designated
use classes and technical
approach for benthic
macroinvertebrates
Calibrating automated decision
model to assess tiered
designated ALU classes
Calibrated BCG model to
communicate monitoring
information on condition
Status
(Citation)
Complete
(Jessup and Gerritsen
2014)
In progress
In progress
Complete. (Gerritsen
and Jessup 2007b)
Complete
(Stamp and Gerritsen
2011)
In progress
In progress
Complete
(Davies and Tsomides
2002; Davies et al. In
press; Danielson et
al. 2012)
In progress
Stamp etal. 2014
35
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
State/Region
Minnesota
New England
New England
New Jersey
Pennsylvania
Puerto Rico and
U.S. Virgin
Islands
Rhode Island
Upper
Mississippi River
Basin
Vermont
Water body type
(if applicable)
Streams and wadeable
rivers
Lakes
High gradient streams
and wadeable rivers
Large rivers
High and low gradient
streams and wadeable
rivers
Streams and wadeable
rivers
High gradient streams
and wadeable rivers
Stony coral reefs
Estuaries
Large rivers
Streams and wadeable
rivers
Biological Assemblages
Benthic macroinvertebrates
and fish
Fish
Benthic macroinvertebrates
Fish
Benthic macroinvertebrates
Diatoms
Benthic macroinvertebrates
Stony corals and resident
reef fish
Seagrass extent, benthic
community, shellfish
production, primary
productivity in Greenwich
Bay, Rhode Island
Habitat mosaic indicator as
measure of whole system
condition for Narragansett
Bay
Fish
Benthic macroinvertebrates
Objective
Calibrated BCG model and
automated decision model for
nine stream types; also
incorporates Region 5 coldwater
results
Calibrated BCG model and
automated decision model for
four lake types
Cross-calibrated BCG model and
automated decision model for
multiple sampling methodologies
Calibrated BCG model and
automated decision model
Calibrated BCG model and
automated decision model
Calibrated BCG model and
automated decision model
Conceptual model and verbal
description of BCG levels,
calibrated to Pennsylvania's MMI
Calibrated BCG model and
automated decision model
Conceptual BCG model anchored
in natural conditions prior to
1850 and showing changes
In progress
Calibrated BCG model and
automated decision model
Calibrated BCG model and
biological criteria
Status
(Citation)
Complete (Gerritsen
etal. 2012)
Complete (Gerritsen
and Stamp 2014)
Complete
(Snook etal. 2007)
In progress
Complete
(Gerritsen and Leppo
2005)
In progress
Complete (Gerritsen
and Jessup2007a)
In progress
(Bradley etal. 2014)
Complete
(Shumchenia etal.
2015)
In progress
preparation,
Shumchenia et al. in
review)
In progress
VT DEC 2004
3.2 Step One: Assemble and Organize Data
Evaluating data quality and preparing it for rule development is critical for an efficient and effective
expert panel meeting. The data should cover the entire range of conditions and stress within at least
one ecological region. Typically state databases have been used in the stream BCGs developed to date,
but large regional databases, either a single data set or pooled data sets, have also been used (e.g.,
Upper Mississippi Basin BCG and New England River BCG (see Appendices Bl and B4)). Combining data
sets presents a unique set of challenges for experts in interpreting site data and detecting consistent
patterns of biological change in response to increasing stress. If different data sets are combined,
decisions on how the data sets are reconciled must be well documented for the experts to successfully
use the data in rule development. When BCG rules are developed for more than one assemblage,
36
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
typically different data sets are used for each assemblage and the rules are developed and applied
independently. The rules for the different assemblages are tested jointly as a later step in the BCG
model development.
There are three tasks required for assembling and organizing data prior to convening an expert panel:
1. Obtain Data—In preparation for the calibration process, relevant data are extracted from the
database. Data should include the biological survey (taxonomic identification and counts) and all
related data on the geo-referenced sampling site: locations and characteristics; catchment data
including area, slope, land use characteristics, and physical habitat; chemical water quality data;
and field observations by sampling personnel. Evaluation and documentation of the quality of
the data set is an essential component of the BCG approach, including documentation of
technical issues and concerns that should be further addressed through additional data
collection and analysis. Section 3.2.1 discusses elements of a data set and monitoring program
that should be evaluated and documented.
2. Determine Natural Classification—In order to prevent natural variability from confounding
responses to stress, it is necessary to determine a natural classification system for the water
bodies under consideration (if not already complete) (USEPA 2013a). If there is only a collection
of data, and no agreed-upon classification system, substantial analytical effort might be needed
to develop it. Classification is beyond the scope of this document; see Barbour et al. (1999),
Hawkins et al. (2000a), USEPA (2013a), Olivero and Anderson (2008), and Olivero Sheldon et al.
(2015) for references to classification approaches for freshwater streams. Selection of a
classification method was one of the first tasks the coral reef expert panel undertook prior to
successful rule development (see Appendix B3). The classification decision has implications for
statistical sampling design and monitoring protocols.
3. Organize Data Tables—A comprehensive and relational database is a requirement for a high
quality monitoring program (USEPA 2013a). Data can be organized in spreadsheets for the panel
workshops (see Figure 12 for an example of datasheet used in BCG development to date). For
permanent storage, retrieval, archiving, and to maintain a quality record, a relational database
will be necessary (e.g., Oracle®, MS-Access®, Sequel Server®).
Quantitative assessment within the BCG framework requires consistent, high quality biological, physical,
chemical, and geographic monitoring information. The technical foundation of monitoring determines
the degree of confidence with which the information can be used to support water quality management
decision making, including calibration of the BCG. This section describes data requirements consistent
with EPA's recommended program review of biological assessment programs (USEPA 2013a).
All BCG developments to date have used existing state or federal agency monitoring data. There have
been no monitoring programs specifically designed for BCG development. However, recommendations
on the technical elements of a monitoring program that would produce good data for BCG development
are not different from the requirements for a high-quality program specified by EPA (2013a) and are
discussed below. This document is not guidance for monitoring design, optimal effort, or sampling
methods. Instead, it focuses on minimum requirements for BCG development, including consistently
sampled aquatic biota; water quality and habitat observations adequately matching the biological
sampling; and land use/land cover information (e.g., from the NHD coverage). Consistency and
adequacy of a data set are evaluated by the expert panel, analysts, and facilitators, and documentation
of BCG development includes recommendations on specific technical areas where further development
would strengthen or refine the quantitative BCG model and underlying decision rules.
37
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
3.2.1 Data Requirements: Understanding the Quality of the Data Set
EPA described 13 technical elements contributing to quality of biological assessment programs (USEPA
2013a). These elements are listed below and constitute the technical underpinnings important for a
biological assessment program to be able to discriminate levels of condition along a gradient of
disturbance (Table 5). Selected elements of biological assessment program design and data collection,
compilation, and interpretation important for BCG development are discussed below. For a more
complete description, see EPA's Biological Assessment Program Review: Assessing Level of Technical
Rigor to Support Water Quality Management (USEPA 2013a). It is recommended that these elements be
considered when assembling a data set for BCG calibration. Understanding the technical strengths and
limitations of the data sets to be used in the calibration will help guide development of the BCG and its
application.
Table 5. Definitions of the technical elements (USEPA 2013a)
Biological Assessment Design
Data Collection and
Compilation
Analysis and Interpretation
Technical Element
Index Period
Spatial Sampling Design
Natural Variability
Reference Site Selection
Reference Conditions
Taxa and Taxonomic
Resolution
Sample Collection
Sample Processing
Data Management
Ecological Attributes
Discriminatory Capacity
Stressor Association
Professional Review
Definition
A consistent time frame for sampling the assemblage to characterize and account for
temporal variability.
Representativeness of the spatial array of sampling sites to support statistically valid
inference of information over larger areas (e.g., watersheds, river and stream segments,
geographic region) and for supporting WQS and multiple programs.
Characterizing and accounting for variation in biological assemblages in response to
natural factors.
Abiotic factors to select sites that are least impacted, or ideally, minimally affected by
anthropogenic stressors.
Characterization of benchmark conditions among reference sites, to which test sites are
compared.
Type and number of assemblages assessed and resolution (e.g., family, genus, or species)
to which organisms are identified.
Protocols used to collect representative samples in a water body including procedures
used to collect and preserve the samples (e.g., equipment, effort).
Methods used to identify and count the organisms collected from a water body, including
the specific protocols used to identify organisms and subsample, the training of personnel
who count and identify the organisms, and the methods used to perform quality
assurance/quality control checks of the data.
Systems used by a monitoring program to store, access, and analyze collected data.
Measurable attributes of a biological community representative of biological integrity and
that provide the basis for developing biological indices.
Capability of the biological indices to distinguish different increments, or levels, of
biological condition along a gradient of increasing stress.
Relationship between measures of stressors, sources, and biological assemblage response
sufficient to support causal analysis and to develop quantitative stressor-response
relationships.
Level to which agency data, methods, and procedures are reviewed by others.
38
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
3.2.1.1 Biological Assessment Design Elements
The first four technical elements are particularly critical aspects of sampling design to consider when
evaluating data for BCG calibration, and they involve selection of sites and times for sampling to obtain
representative and statistically valid information (USEPA 2013a). The fifth element, reference condition,
is also discussed below but in relation to its role relative to the BCG benchmark, BCG level 1 (e.g.,
anthropogenically undisturbed reference condition).
Index Period
Sampling index periods are selected based on known ecology to minimize or account for natural
variability, maximize sampling gear efficiency, and maximize the information gained on the assemblage
(Barbour et al. 1999; USEPA 2013a). For temperate fresh water bodies, index periods are typically a span
of 3-6 months during the growing season.
Spatial Sampling Design: Representative of Stress Gradient
The objective of BCG calibration is to characterize the biological response across a generalized stress
gradient from undisturbed to highly disturbed conditions. The BCG should be developed for specific
natural classes, such as ecoregions or physiographic provinces. Sample coverage must be representative
of the ecoregion(s), as well as the stress gradients that can occur. Case examples of characterizing stress
gradients are given later in this chapter (section 3.3.1) and discussed more generally in Chapter 5.
Achieving representativeness might require using data from outside of the jurisdictional boundaries of a
state so that ecoregional expectations are as fully sampled and defined as possible. In addition to
representativeness, the data should have sufficient sample size to support the calibration. As a rule-of-
thumb, 30 or more samples for each water body class (at a minimum for levels 2-5, and levels 1 and 6, if
regionally available) are generally required (see natural classification below). If samples are not
sufficient, then BCG development should be delayed until enough data are acquired.
Calibration of the BCG model requires data points (samples) along the stress gradient from low to high
levels of stress. An expert panel examines the sites and assigns the sites to BCG level based solely on the
biological information. Having the stress information ensures that the expert panel sees sites that are
representative of the stress gradient. Ideally, the data set needs to include the full gradient of conditions
and complement of stressors (e.g., pollution sources, invasive species, habitat disturbances) that are
common to a state or region, such that the full gradient of assemblage response is included in the model
development.
Natural Variability: Classification
Biological assessment based on knowledge of the biota under undisturbed or minimally disturbed
reference sites forms expectations for natural conditions. Many natural regional and habitat characteristics
(e.g., stream size, slope, dominant natural substrate) also affect the species composition of undisturbed
water bodies. Accordingly, a critical step in developing a biological assessment program is to classify the
natural conditions to the extent that they affect the biological indicators (e.g., Barbour et al. 1999; Gibson
et al. 1996; Hawkins et al. 2000a). The term classification includes development of continuous models
that explain natural variability of biological assemblages. For example, fish species richness is strongly
dependent on catchment area or average flow. Modeling approaches that combine both discrete and
continuous variables (e.g., general linear models) may be especially powerful if the data support them.
Failure to properly classify sites can cause the BCG calibration to fail, yielding assessment errors that can
undermine confidence in results. Classification of natural conditions should be complete and satisfactory
to experts. If not, time and resources will be necessary to develop the classification system.
39
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Reference Site Selection
Obtaining a representative stress gradient requires that the data set is large enough to include the full
range of disturbance, from undisturbed to highly disturbed conditions. Data owners and field personnel
should document the comprehensiveness of the data set with respect to coverage of the full range, or
not, of disturbance. It might be necessary to obtain data from neighboring states or regions to ensure
that the gradient is represented in the data set. A minimum of 30 to 40 sites might be sufficient to
calibrate the BCG depending upon both the characteristics of the natural system and the quality of the
data set. Characterization of the quality of reference sites is essential to defining the range of conditions
in the data set the experts will be using to develop decision rules. The criteria used by states to select
their reference sites inform this determination.
Reference Condition
In this document, the terms "undisturbed," "minimally disturbed," and "least disturbed" conditions are
used when referring to the level of anthropogenic stress to which a water body and its surrounding
watershed may be subject. These terms are well defined by Stoddard et al. (2006). The level of stress
associated with the reference sites used by the state to define reference condition is the critical
information needed for BCG calibration. In many cases, the state's reference condition is not comparable
to the BCG benchmark for undisturbed or minimally disturbed conditions. This is important information,
not only for the BCG calibration but also for water quality program managers and the public.
BCG calibration is not based on least disturbed reference sites, because least disturbed sites are typically
the "best of what is left" and may mistakenly be perceived by the public as the best that can be because
undisturbed conditions no longer exist (e.g., Dayton et al. 1998; Papworth et al. 2008; Pauly 1995). In
this case, expectations for improvements might be set lower than the potential for a water body to
improve. Part of the BCG process can include developing a description of undisturbed conditions that
may include consideration of contemporary, empirically least stressed sites and historical descriptions;
paleolimnological investigations; and museum records. The description of an undisturbed condition may
be narrative and perhaps incomplete, but its documentation helps provide a transparent and clear
framework for the public to understand what biologically may have already been lost from their waters
as well as potential for what could be restored. In many of the BCGs that have been developed,
undisturbed and minimally disturbed conditions have been combined for practical purposes and
categorized as representing BCG levels 1 and 2.
3.2.1.2 Methodological Elements
The second set of technical elements are aspects of quality in sampling, processing, and data
management. Data used for calibration must be consistent, or be made consistent in post-processing. It
is especially important to examine methods when biological assessment data from multiple sources are
to be pooled. For information on combining data derived from multiple sources, see Gerritsen et al.
(2015). Elements of sampling methodology include:
Taxa and Taxonomic Resolution
The biological response data should be the taxonomic composition and related information from one or
more biological assemblages in water bodies: benthic macroinvertebrates, fish, periphyton, aquatic
macrophytes, phytoplankton (lakes and estuaries), and zooplankton (lakes and estuaries). To develop
the model, a knowledgeable panel of experts is required for each assemblage.
40
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Experience has shown that "lowest practical" identification, to species when possible, is superior for
BCG calibration, because species differ in their characteristics within genera. Species identification is
necessary for fish assemblages, but genus-level identifications are adequate for BCG calibration using
benthic macroinvertebrates. When pooling data, the taxonomic resolution must be standardized to the
lowest common level among the data sets.
Sample Collection
Field methods should be consistent and well-documented. The objective of the sampling methods
should be to obtain consistent samples that are representative of the target biological assemblage (see
Barbour et al. (1999) and USEPA (2013a) for discussions of sample collection methods). The BCG has
been cross-calibrated for several sampling methods used in New England and in the Upper Midwest
(Gerritsen and Stamp 2012; Snook et al. 2007). Where possible, initial BCG development in a new region
is done with data from a single sampling methodology and then can be calibrated and tested with data
generated using different sampling methods. Level of effort is a key consideration in sample design.
Many of the BCG attributes (attributes I, II, VI, VII, IX, and X) are particularly sensitive to level of effort.
Certain key taxa may be sparse, seasonal, or patchy in their distribution and easily missed by a
standardized field collection method. In making a site assessment, other supplemental information (e.g.,
natural history surveys, fishery agency reports and observations, academic studies), beyond just the
collected samples, should be included in making a level determination. This will lend an additional layer
of confidence and improve the result.
Sample Processing
Laboratory processing of samples (except fish) is recommended (USEPA 2013a; Yoder and Barbour
2009). Macroinvertebrate and diatom samples are typically processed to a standardized count
representing a constant sampling effort. In some cases, if subsampling efforts are mixed, it is possible to
randomly subsample larger efforts to smaller efforts (e.g., 300-count subsamples randomly subsampled
further to match 100-count subsamples).
Data Management
Identification of reference and stressed sites requires that the monitoring database be comprehensive,
including watershed and site characteristics, habitat measurement, and physical and chemical water
quality measurements. Physical and chemical measurements should be made at the same time and
place that the biological community information is collected. Non-biological data, including catchment
area, slope, land use, site, habitat, and physical and chemical water quality data are used to determine a
site's natural classification and stressor status, such as whether it is a reference site or a stressed site,
and where it is located along the stress gradient.
Data should be stored in a relational database so that queries can retrieve relevant information (e.g.,
biological data, chemical data, physical measurements) on site, geo-referenced location, multiple
measurements from multiple sampling times, and catchment data. Data stored in spreadsheets or
warehoused in such a way that physical, chemical, and accurate geo-reference cannot be located are of
limited use and might require substantial costs to fill in missing information and for quality assurance
(Gerritsen et al. 2015). Exceptions should be made for historic information and data that may not be
amenable to spreadsheets. These data may not be suitable for a relational database but should be
retained because they may provide important qualitative information and context that can be used to
inform BCG development.
41
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
3.2.1.3 Analysis and Interpretation Elements
Ecological Attributes
These are the measurable attributes of a biological community that are representative of biological
integrity and which provide the basis for developing a BCG model. The BCG attributes (Table 1) are the
basis for this technical element. The selection of attributes might depend on the spatial scale and
specific water body being assessed. Each attribute provides some information about the biological
condition of a water body. Combined into a conceptual model comparable to the BCG, the attributes
can offer a more complete picture about current water body conditions and also provide a basis for
comparison with naturally expected water body conditions. All states that have applied a BCG for
streams, rivers, and wetlands have used the first seven attributes that describe the composition and
structure of the biotic community on the basis of the tolerance of species to stressors. Where available,
they have included information on the presence or absence of native and nonnative species, and, for
fish and amphibians, used measures of overall condition (e.g., size, weight, abnormalities, and tumors).
Though not measured directly in state or tribal stream biological assessment programs, the last three
BCG attributes of ecosystem function and connectedness and spatial and temporal extent of stressors
can provide valuable information when evaluating the potential for a stream, river, or wetland to be
protected or restored.
Discriminatory Capacity
This technical element addresses the degree of sensitivity of the BCG model in distinguishing
incremental change along a continuous gradient of stress. Detailed descriptions of biological change
along a gradient of stress can provide the conceptual basis for refined ALUs for specific ecotypes and
regions leading to biological criteria development. Additionally, depending on the sensitivity, or
discriminatory capacity, of the BCG model, the information can be used to help identify high quality
waters and establish incremental goals for improving degraded waters. Six general increments of change
can be described for each of the BCG's ecological attributes (for example, see Table 3). These
incremental changes can serve as a template for developing biological criteria that represent aspects of
biological integrity and which show a predictable, measurable response to increasing levels of stress.
The number of increments that can realistically be distinguished in a BCG model is dependent not only
on the water body ecotype and natural classification factors that define biological assemblage
characteristics, but also on the effect of anthropogenic stressors. For example, the sensitivity of an index
developed for a forested, high-gradient stream might support distinguishing five or even six increments
of quality along a continuous stressor gradient, while an intermittent, seasonal, or desert stream may
yield fewer increments. Some of this difference is due to inherent natural characteristics of the
assemblages, and some might be due to current limitations of science and practice.
Stressor Associations
Stressor association refers to the use of biological assessment data at appropriate levels of taxonomy to
develop relationships between measures of biological response and anthropogenic stressors, including
both stressors and their sources (Huff et al. 2006; Miller et al. 2012; Yuan 2010; Yuan and Norton 2003).
This element includes examination of biological assessment data for patterns of response to categorical
stressors (Riva-Murray et al. 2002; Yoder and DeShon 2003; Yoder and Rankin 1995a). A capability for
developing these relationships extends the use of biological assessments from assessing condition to
informing identification of possible causes and sources of a biological impairment at multiple scales.
42
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Stressor association is directly dependent on a high level of technical development of other elements,
particularly the elements for spatial sampling design, taxa and level of taxonomic resolution, database
management, and discriminatory capacity. These elements are important building blocks for the data
collection and analysis needed to more confidently identify stressors and their sources and to estimate
stressor-response relationships. For example, the ability to estimate these relationships relies on paired
stressor and response sampling at appropriate spatial and temporal scales and a level of taxonomic
resolution and index sensitivity sufficient to detect incremental biological changes along a stress
gradient. Also, a relational database that supports complex queries enables efficient and full utilization
of data. A high level of technical development for each of these elements and others provides the
foundation for stressor association.
Professional Review
Professional review and testing of the BCG quantitative decision rules should be conducted by experts
outside of the panel to evaluate and improve model objectivity and scientific defensibility and to refine
and improve the model. Review by outside experts can be used to refine and improve the model.
Because of the specific knowledge of the expert panel for any given BCG, discussion with the outside
peer reviewers is essential. Technical expert review across expert panels has not yet been conducted for
the BCGs developed to date, but it is planned as a pilot.
3.3 Step Two: Preliminary Data Analysis and Data Preparation
Before an expert panel is convened to describe the BCG levels, it is necessary to reduce and prepare the
data for the panel's use during the workshops and webinars. In addition, it is useful to conduct
exploratory data analyses to visualize empirical relationships of the biotic assemblages. Analysis and
data preparation include:
1. Characterizing Stress Gradients—Identifying stress gradients to select sites for BCG calibration
that are representative of stress gradients in the region, from undisturbed to highly disturbed
levels of condition. If undisturbed conditions do not exist, the level of disturbance should be
recorded and efforts made to collect historical data and records that may help the panel
develop a conceptual, descriptive condition level absent of anthropogenic influence (BCG
level 1). See Chapter 5 for further discussion.
2. Analyzing Taxon Response Relationships—Using the stress gradient(s) to examine stressor-
response of individual taxa to augment known or surmised species tolerances and traits with
empirical information on responses observed in the field, as well as to develop species
distribution maps of species observed in the data set. This step also ensures that all panelists
have the same information available to them, as some panelists may be more familiar with the
monitoring data set than others. See Chapter 5 for further discussion.
3. Preparing Data Work Sheets—Identifying and formatting a calibration data set for the
workgroup's calibration exercise.
This section discusses the preliminary analysis and data manipulation prior to calibration workshops.
43
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
3.3.1 Data Preparation: Characterize Stress Gradients
Water bodies are subject to a wide variety of anthropogenic stressors, and multiple stressor situations
are common. However, few state data sets are sufficiently large and complete to be able to analytically
separate the effects of individual stressors. To help select sites for the calibration exercises, a practical
approach is to consider all stressors together without regard for interactions among them (Smith et al.
2001), or to use aggregate land cover as a summation of sources of potential stressors to streams (e.g.,
Landscape Development Intensity Index [LDI]; Brown and Vivas 2005). Stressor-response analysis with
multiple, independent stressor gradients is currently an area of active research (e.g., Baker and King
2010; Norton et al. 2015), but it is beyond the scope of this document.
Quantitative Gradients
Identifying stress gradients relevant to the data sets at hand will be facilitated by some exploratory data
analysis to identify biological responses to the stressors. Scatter plots are generally the most useful and
efficient, but more detailed analysis can be done if desired, including regression analyses, quantile
regression, and classification and regression tree (CART) analysis (Death and Fabricius 2000), and other
models. The purpose of these analyses is not diagnostic, as BCG calibration does not include identifying
the most likely causes for biological impairment. The purpose is to develop a database suitable for
discerning patterns of biological response to increasing levels of stress.
Scatter plots can be examined for every stressor that will be included in a stress gradient, as well as for
aggregated sources of stress such as land use/land cover. For this purpose, scatter plots are simple
graphical displays of a response variable on the y-axis, against a presumed correlated parameter on the
x-axis (e.g., Figure 9). Examples of stress variables examined for some of the BCG applications are shown
in Table 6.
Table 6. Examples of quantitative stressor variables that have been used for BCG projects
Project
Minnesota streams
Connecticut streams (fish)
Minnesota lakes
Maine stream algae
Maine stream benthic macroinvertebrates
Northern Piedmont region of Maryland
Alabama
Illinois
Indiana
Quantitative Disturbance gradient
Human Disturbance Score (HDS)
% Developed area
% Urban + Agricultural + Mining land use
Trophic State Index
Total phosphorus
% Impervious surface
% Impervious surface
Habitat index
Human Disturbance Gradient (HDG)
Habitat index
% Impervious surface
Total nitrogen
% Impervious surface
44
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
3.3.1.1 Example—Using Land Use/Land Cover Indicators to Develop a Quantitative Stress Gradient
(Minnesota, Alabama, Maryland Piedmont)
Measures of land use and land cover have been used as surrogate indicators of stressor effects. Table 6
contains a list of these type of indicators that have been used for GSA development (For more
information on the GSA, see Chapter 5). In the Northern Piedmont of Maryland, the workgroup selected
imperviousness as a primary stress indicator (Stamp et al. 2014, see Chapter 6). The percent
imperviousness in a watershed or a catchment was available for all sites in the data set. Based on
scatterplots like the one shown in Figure 9, the level of imperviousness has a clear impact on biological
assemblages. Imperviousness was considered during the sample selection process to ensure that the full
stress gradient was represented in the BCG model calibration data set. Imperviousness was also used to
generate the taxon-response plots that helped inform BCG attribute assignments (see section 3.3.2).
ro
CL
LU
.0
E
Z!
20
16
12
8
4
0
CO
0 0
n>
n
0 0
0
n on
f
•\
O
HO
OO O ^
O 00 OOO^\^ O O O
(HDD U>«!i O OO O
ODD OCOD O O O dDODSJODTSUCJOnD O
O O OO O OOCTXCD) ODD O
OGGDO O O QDO G(n)
10
40
50
20 30
% Impervious surface
Figures. Scatterplots of number of total taxa (upper) and number of EPTtaxa (lower) versus % impervious
surface in the macroinvertebrate data set for streams in the Northern Piedmont of Maryland. Plots are fit with a
linear trend line.
45
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
In some instances, quantitative stress gradients have been developed to capture multiple stressors in
one integrated score. Examples include Minnesota's Human Disturbance Score (HDS) and Alabama's
Human Disturbance Gradient (HDG). Input variables for the Minnesota HDS and the Alabama HDG are
listed in Table 7 and Table 8, respectively. The Alabama HDG utilizes the LDI (developed by Brown and
Vivas (2005)), which associates land uses with a scale of disturbance intensity and weights the index
score based on land uses in the upstream catchment. Alabama Department of Environmental
Management (ADEM) has used the HDG to assign its stream reaches to one of eight HDG categories
based on the percentile of its overall HDG score, with categories 1-3 representing the top 25th percentile
of watershed condition.
Table 7. Input variables for Minnesota Pollution Control Agency's (MPCA's) HDS (MPCA 2014a)
HDS Metric
Animal unit density
Feedlot density
Feedlot proximity
Point source density
Point source proximity
Percent disturbed riparian habitat
Riparian condition rating
Percent agricultural land use
Percent agricultural land use within 100-m riparian buffer
Percent agricultural land use on > 3% slope
Percent impervious surface
Urban land use proximity
Percent of stream distance modified by channelization
Site channelization rating
Road/stream intersection (road crossing) density
Scale
watershed
watershed
local
watershed
local
watershed
local
watershed
watershed
watershed
watershed
local
watershed
local
watershed
Score
10
adjust
adjust
10
adjust
10
10
10
adjust
adjust
10
adjust
10
10
adjust
Table 8. Input variables for Alabama's HDG (Source: Lisa Huff, ADEM, personal communication)
The LDI associates land uses with a scale of disturbance intensity and weights the index score based on land uses in
the upstream catchment, such that land uses that produce higher levels of disturbance receive higher LDI
coefficients (Brown and Vivas 2005).
Variable
Population density/km2
% Urban
% Barren
% Pasture
% Cropland
Road density
# Stream/road crossings
LDI coefficient
1
8
8.6
3.1
4.7
8.3
8.3
Source
2000 U.S. Census
2006 National Land Cover
Database
2010 Census TIGER/Line
Shapefiles
46
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Ordinal Stress Gradient(s)
Where there are many measured stressors, it is possible to develop an ordinal, generalized stress
gradient by summing and ranking the number of different stressors observed at distinct sites. A site is
given a score of 1 for each stressor observed there (e.g., copper above a chronic screening threshold,
excess nutrients, poor habitat score, upstream discharge), and the site score is the sum of all stressor
scores. Sites with scores of 0 are candidates for "least stressed" within the context of the region.
Categories of stress can be defined using measured stressors (e.g., contaminants and habitat condition)
from the monitoring data, with watershed information that identifies sources of stress. The categories
are a mixture of both sources and measured stressors and will inevitably be correlated to some extent.
The categorization can identify a gradient of stress levels comparable to levels of disturbance (e.g.,
undisturbed, minimally disturbed, highly disturbed conditions (Stoddard et al. 2006)).
Stressors, whether individual or categories, can be screened by examining the response of individual
taxa to the stressor or source (Figure 10)—if there is no response, the stressor should not be used in
developing the ordinal gradient.
After relevant stressors and sources have been categorized, sites can be identified according to the
number of stressors and sources in low, medium, or high categories. Sites where all stressors and
sources are "low" qualify as least stressed, and sites where many stressors are "high" qualify as most
stressed. Depending on the number of sites and stressors, intermediate categories can also be
identified. Depending on the level of stressors detected in the "least stressed" category, undisturbed or
minimally disturbed conditions may not be included in the data set. If this is the case, expert judgment
on undisturbed or minimally disturbed conditions can be elicited based on historical observations,
records, and data. Although a qualitative assessment, this information provides context for the
quantitative information (e.g., "least stressed" conditions do not present undisturbed or minimally
disturbed conditions).
3.3.1.1 Example: Connecticut Ordinal Stress Gradient
To identify sites to use in a BCG calibration exercise for Connecticut, analysts developed an ordinal stress
gradient to apply to sample sites. The approach was to screen measured stressors for association with
biological measurements and identify thresholds (stressor concentrations) below which no effects or
association could be detected and screening thresholds above which association was strong. This was
not an attempt to do a causal analysis (Norton et al. 2015), but simply a screening based on pairwise
associations.
Connecticut DEP had sampled dissolved metals and several other water quality parameters simultaneously
with each stream biological sample. For example, Figure 10 shows the number of Plecoptera (stonefly)
taxa and dissolved copper concentration in Connecticut stream sites. High numbers of stonefly taxa (> 4)
only occur when copper is less than 0.008 mg/L, and nearly all samples where copper was greater than
0.008 mg/L had fewer than 4 stonefly taxa (Figure 10). For the stressor gradient, the threshold for low
copper stress was set at < 0.008 mg/L, and the threshold for high copper stress was set at > 0.008 mg/L.
Note that there is not inference of causality, only screening of associations.
Stress categories were identified for Connecticut monitoring sites based on land use and water
chemistry parameters in the database. Urban land use, natural land cover, population density, and
chloride concentration were all good predictors of biological condition. Connecticut Department of
Energy and Environmental Protection (CT DEEP) defined six stress categories for streams, based on the
distribution of stressor parameters in the database.
47
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
CT DEEP's thresholds for the "least stressed" category (Table 9) were determined from stressor-
response scatterplots of sensitive taxa in the samples versus the stressor parameters (dashed line in
Figure 10). Screening thresholds for metals (Table 9) were determined from stress-response scatterplots
of number of mayfly or stonefly taxa in the samples vs. metal concentrations (Figure 10). These two
orders are generally considered highly sensitive to metal contamination (e.g., Buchwalter and Luoma
2005). Metals not included in Table 7 (aluminum, cadmium, mercury, lead, selenium) were either not
associated with biological responses (no observable stress-response), or they were not detected in most
observations in the data set. Using the criteria of Table 9, least disturbed sites were identified as sites
with all eight stressor values in the "least stressed" category, and highly disturbed sites were identified
as sites with four or more stress values in the "high" category. The screening allowed selection of sites
for calibration to cover the range from "least disturbed" to putative "highly disturbed."
ro
x
2
Q.
O
u
y
8
7
6
5
4
3
2
1
0
-1
A A
A <
• J^ JL JL •
T T TrJ
ss --ea ea •
ii : El li •
H»IH 1
-••••I
|
f --*
k
'
tJL ,
T ^
»- •
•^
i+^
!••«
^--
.
Bl
•
^-- Screen
IJL
nr
ing criteric
iH
--•'-•
>n
j
L
T
• Least stressed
• Highly stressed
4- Other
0.000 0.005 0.010 0.015 0.020
Copper, Dissolved (mg/L)
0.025
0.030
Figure 10. Number of Plecoptera (stonefly) taxa and dissolved copper (Cu) concentration, Connecticut sites. The
screening criterion, (0.008 mg/L Cu) was estimated by eye from the presence of stoneflies at low Cu
concentrations, and their near absence above 0.008 mg/L Cu. In the calibration, least stressed sites were
required to have Cu < 0.008 mg/L (among other criteria). The screening criterion separates sites with no
detectable influence of copper from those where copper may be a factor (among others) in loss of Plecoptera.
48
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 9. Example screening thresholds for stressor gradient (Connecticut)
Parameter
Stress Category
Least Stress
Slight Stress
Moderate Stress
High Stress
Catchment parameters
Natural land cover*
Developed land
Impervious surface
>80%
70%-80%
<10%
<4%
60%-70%
10%-25%
4%-10%
Severe Stress
<60%
>25%
>10%
Water quality, non-metals
Chloride
<15 mg/L
15-20 mg/L
20-30 mg/L
>30
mg/L
Water quality, metals
Copper
Iron
Nickel
Zinc
Decision criteria for
stress level
< 0.008 mg/L
< 0.4 mg/L
< 0.01 mg/L
< 0.02 mg/L
All parameters
lowest stress
category
Land cover or
chloride Slight
category; All
others lowest
category
Anyone nonmetal
allowed High
category; All others
Moderate or lower
> 0.008 mg/L
> 0.4 mg/L
>0.01 mg/L
>0.02 mg/L
Up to three non-
metals High; Any
metals High
All non-metals
High-Severe; Any
metals High
*defined as the sum of deciduous, conifer, open water, and all wetland categories
3.3.2 Data Preparation: Analyze Taxon Stressor-Response
An early task of the expert panel is to assign taxa to the attributes I through VI for development of
stream and river BCGs. These are the primary attributes that are used to assess sites among BCG levels 2
through 6 for streams and rivers. Attribute VII, which provides information on organism condition
(especially of long-lived organisms), is a general indicator of organism health, such as deformities,
anomalies, lesions, tumors, or excess parasitism. This attribute has been used with great success in
indices based on the fish assemblage. To date, attributes VIM through X have not been consistently
applied to biological assessment and BCG development for streams and rivers. These attributes are also
being explored for application in larger, more complex systems such as large rivers, estuaries, and coral
reefs (see Appendix B). Additionally, these attributes may be more easily assessed, quantifiable, and
amenable to rule development using spatial analysis.
Attribute assignment uses both empirical data analysis and expert judgment. Typically, tolerances of
many genera or species are available from well-known compendia on macroinvertebrates (e.g., Barbour
et al. 1999; Hilsenhoff 1982; Merritt et al. 2008). The published tolerances are broad and might not
apply to species or genera in the data set at hand, but they provide a convenient initial value for the
panel to consider. To augment the published tolerances and traits information, local data are also
evaluated empirically to determine whether the published values, or the expert's opinions, are
supported by the local data.
While it is tempting to rely only on the empirical analysis and "let the data tell the story," in practice,
many data sets are not sufficient to determine tolerance of all taxa. For example, a taxon that occurs in
five samples is too infrequent in the data set to estimate its tolerance. Nevertheless, it may be a taxon
where the tolerance is well-established; for example, Limnodrilus hoffmeisteri is a worm characteristic of
severe organic enrichment associated with untreated sewage discharge, and Brook Trout is a highly
sensitive fish species in streams of northeastern North America. Both of these organisms are relatively
uncommon in regions of the country with a high degree of development and with regulated discharges.
49
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
However, their biology and tolerance are well-known. Similarly, there are likely to be other taxa for
which the assembled experts have substantial experience, but that might be insufficiently represented
in the data set. Presentation of the stress-response analysis ensures that all experts in the workgroup
are aware of, and familiar with, the data set at hand and associations that exist in that data set.
Empirical analysis of the data set being used in the calibration can greatly assist the attribute
assignment. After developing a stressor gradient, it becomes possible to support assignment of taxa to
attributes based on biological responses to the stressor gradient. This is similar to the analysis often
used to identify tolerance groups (e.g., Yuan 2004, 2006).
Several different statistical approaches can be applied to examine individual species' response to
stressors: (1) correlation tables and simple scatter plots, (2) central tendencies, (3) environmental limits,
(4) optima, and (5) curve shapes (Yuan 2006). Correlations and scatter plots show the strength and
shape of a stress-response. Tolerance values expressed in terms of central tendencies attempt to
describe the average environmental conditions under which a species is likely to occur; tolerance
expressed in terms of environmental limits attempt to capture the maximum or the minimum level of an
environmental variable under which a species can persist; and tolerance expressed in terms of optima
define the environmental conditions that are most preferred by a given species. These types of
tolerances are expressed in terms of locations on a continuous numerical scale that represent the
environmental gradient of interest. Both abundance-based and presence/absence-based models can be
built using these statistical approaches. See Yuan (2006) for analytical methods.
3.3.2.1 Example: Stressor-response of Macroinvertebrates (Maryland Piedmont) and Fish (Minnesota
Lakes)
When panelists assign taxa to attribute groups I-VI, they rely on a combination of empirical examination
of taxon occurrences at sites that span a human disturbance, or stress, gradient, as well as professional
experience as field biologists who have sampled water bodies in the areas of interest. During the
attribute assignment process, panelists are provided with taxon-response plots in which the frequency
and abundance of the taxa are plotted over the range of the disturbance gradient (Yuan 2006). Several
different statistical models can be used to generate these plots, including:
• Weighted averaging to estimate optima and tolerance values (abundance based).
• Cumulative distribution function median and extreme limits (presence/absence).
• Logistic regression (linear, nonlinear, generalized additive model) median and extreme limits
(presence/absence).
Taxon-response plots can be used to infer central tendencies (average environmental conditions under
which a species is likely to occur), tolerance limits (maximum or minimum levels of an environmental
variable under which a species can persist), and optima (environmental conditions that are most
preferred by a given species).
The panelists use these plots to help inform BCG attribute assignments, particularly for attributes II
(highly sensitive), III (intermediate sensitive), IV (intermediate tolerant), and V (tolerant). Taxa in these
attribute categories are expected to follow the response patterns shown in Figure 11.
50
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Prior to generating the plots, stressor variables are selected based on considerations such as availability
of quantitative field-collected data and responsiveness of the biological assemblage to the stressor, or a
stressor index such as Minnesota's HDS (Table 6). In one example, taxon-response plots were generated
for two stressor variables—imperviousness and habitat index scores—based on data from the Northern
Piedmont of Maryland (Stamp et al. 2014). For Minnesota lakes, the group examined taxon-response
plots for urban/agricultural/mining land use in the contributing watershed and the trophic state index
(Gerritsen and Stamp 2014). Examples of taxon-response plots from these two projects can be found in
Figure 12. In these examples, there was good agreement between the taxon-response plots and
attribute assignments, but this does not always happen. In cases of disagreement, the group relies on
consensus professional opinion, unless contradicted by an overwhelming response in the data analysis.
To interpret the graphs in Figure 12, the points are actual data of relative abundance, the curve
represents the capture probability (logistic regression generalized additive model fit and confidence
interval following Yuan 2006), and the red vertical dashed lines represent the optimum (50%) and
tolerance (95%) values. Curves are smoothed to facilitate comparison to the "ideal" plots of Figure 11.
§
03
T3
.Q
03
CD
CT
CD
Att II: Highly sensitive
Att IV: Intermediate tolerant
Att III: Intermediate-sensitive
Increasing Stress
Figure 11. The frequency of occurrence and abundances of attribute II, III, IV, and Vtaxa are expected to follow
these patterns in relation to the stressor gradient. Attribute II taxa have a high relative abundance and high
probability of occurrence in minimally-disturbed sites. Attribute III taxa occur throughout the disturbance
gradient, but with higher probability in better sites. Attribute IV taxa also occur throughout the disturbance
gradient, but with roughly equal probability throughout, or with a peak in the middle of the disturbance range.
Attribute V taxa occur throughout the disturbance gradient, but with higher probability of occurrence, and
higher abundances, in more stressed sites.
51
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Epeorus
Cisco
I"
Is
<3 r
£ 8 .£"
o c 3
* 4 Q^
. 8
- 8
13 10 30
Imperviousness Surface (%)
0 20 40 60 80
"oUrban * Agr * Mine
Amphinemura
Logperch
r
£ -.
£ °
I? -I
o e
8 °
a. „
go
3 ~ -,
r
r &
13 10 30
Imperviousness Surface (%)
0 20 40 80 80
"oUrban + Agr * Mine
Antocha
Golden shiner
o .
fa-I
Is
O
° -f~-
£ 8
o c
•
° J
!j
o o:
s
Q.
I
_ •'
13 10 30
Imperviousness Surface (%)
0 20 40 60 80
%Urban + Agr * Mine
Physa
Fathead minnow
S .
|s,
2 e
a
IJ
o
8 -
fS^
i
8i
8
C
e» .
o
*
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
3.3.3 Data Preparation: Organize Data for Expert Panel
The expert panel will need to work with a taxon list for the database and with sample data. The taxon
list should include the taxonomic hierarchy for each genus or species in the database, and it should be
sorted taxonomically for ease of use. Information to be included for each species should include known
tolerance/sensitivity from other sources (e.g., published Hilsenhoff tolerances, trophic guild, spawning
guild, habit, habitat preference). For lists of taxa that include some of these characteristics, see Barbour
et al. (1999) and Merritt and Cummins (1996).
The panel will also need to work with data sheets from individual sites. Figure 13 is an example of a data
sheet that has been used in stream BCG development. These sheets should include all taxa, counts, and
the panel-assigned attribute for each taxon, sorted taxonomically. Attribute assignments (left-hand
column, Figure 13) are finalized during the expert panel meeting, and they are entered into the tables at
that time.
In a typical workshop, the expert panel should have data available from approximately 20 to 40 sites
from a single water body class, which are selected (by data analysts, not panelists) from the entire range
of the stressor gradient. There should be good representation of least stressed sites, as well as most
stressed sites, and all categories of stress in between. The sites selected are typically a subset of sites
used to develop the stress-response curves (Figure 12).
Although the data analysts have selected cover the range of disturbance, stress information on
individual sites is not provided to the expert panel. In BCG development, the rating should be done
"blind" without knowledge of stressors or levels of disturbance to minimize preconceived perceptions
and bias.
53
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
BCG SampID
StationID
Station Name
WMA
Gradient
CollDate
BCG Attribute
4
2
5
5
5
5
5
5
2
2
3
3
3
2
2
3
2
4
4
5
3
4
HA11
High
05-05-1994
FinallD
Psephenus herricki
Diamesa nivoriunda
Dicrotendipes neomodestus
Orthociadius dorenus
Orthodadius obumbratus
Orthociadius rivulorum
Micropsectta
Tanytarsus
Acentrella turbida
Drunella cornutella
Ephemerella dorothea
Ephemerella rotunda
Eurylophella temporalis
Epeorus
Ameletus
Amphinemura delosa
Isoperla transmarina
Ceratopsyche slossonae
Cheuinatopsyche
Hydropsyche betteni
Pycnopsyche
Polycentropus
Assigned Level
Individuals
1
2
1
6
1
2
1
1
2
12
21
3
12
2
3
25
6
1
1
1
4
1
Area (km2)
Pet Urban
PctAgr
Pet Forest
Pet Wetlands
Habitat Score
Order
Coleoptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Diptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Ephemeroptera
Plecoptera
Plecoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
Trichoptera
7.68
Family
Psephenidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Chironomidae
Baetidae
Ephemerellidae
Ephemerellidae
Ephemerellidae
Ephemerellidae
Heptageniidae
Siphlonuridae
Nemouridae
Perlodidae
Hydropsychidae
Hydropsychidae
Hydropsychidae
Limnephilidae
Polycentropodidae
Summary
BCG Attribute
1
2
3
4
5
6
X
Total
Taxa
0
6
5
4
7
0
0
22
Individuals
0
27
65
4
13
0
0
109
Figure 13. Example data table for site assessment, showing how site data may be arranged for a panel's
assessment. Attribute summary information is included at the bottom. Note that stressor information is blank-
the panel rates sites without knowledge of stressors.
54
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
3.4 Step Three: Convene an Expert Panel
The expert workshop to calibrate the BCG is central to BCG development. Calibrating a BCG requires
refining the generalized conceptual model to reflect regional conditions (Davies and Jackson 2006). The
process has several steps:
• An expert panel of ecologists and field biologists is assembled.
• The panel assigns taxa to attributes I-VI. This step makes use of the taxon-response analysis
described in section 3.3, combined with the experience and judgment of panel members.
• The panel assigns a set of sites to levels of the BCG. In this step, the panel also develops a
general description of the native aquatic assemblages under natural, undisturbed conditions.
The description of natural conditions requires biological knowledge of the region, a natural
classification of the assemblages, and, if available, historical descriptions of habitats and
assemblages.
• The panel develops narrative and quantitative decision rules to assign sites to BCG levels.
3.4.1 Expert Panel
An expert panel provides specific technical descriptions of each BCG level through the process of
assigning sites to the levels. The panel should consist of (1) ecologists with strong field and identification
experience with organisms represented in the monitoring data; (2) ecologists with knowledge of the
natural history of the organisms and organism tolerances; (3) water quality experts; and, if possible, (4)
one or more persons familiar with the historical background and context of water bodies of the region.
This expertise could include knowledge of historic vegetation cover of the region and changes to the
present or past distributions from museum records and old accounts of the taxa in the species list. Past
experience with panels suggests that an ideal number of participants for each assemblage group is
between 8 and 12; fewer than 8 results in a narrow diversity of expertise and viewpoints represented,
yet a panel with more than 12 participants can become unwieldy and slow in identifying individual
opinions. Panel meetings should also include a facilitator familiar with the BCG calibration process; staff
familiar with the data and analysis already done (section 3.3); and recorder(s) to record decisions, expert
logic, and important discussion points.
In the introductory session of the workshop, the panel is introduced to the BCG concept and ground
rules for assessing sites and developing decision rules. Panel members must have sufficient time to
digest and discuss the process and feel comfortable with it. This requires one or more introductory
sessions to familiarize them with the conceptual BCG model, applications, calibration, and the data and
procedures to be used. These sessions may be done as webinars to save time in the face-to-face panel
meetings. For several of the BCG development efforts, two to three webinars have been conducted with
the expert panel and have proven to be very effective in educating the panelists about the BCG. These
webinars have also been useful in addressing questions and issues ahead of the workshop that would
otherwise have sidetracked the work of the panel during the face to face meeting. Additionally, a dry
run with the panelist in use of data spreadsheets and evaluating the data can result in new information
and insight from the panelists that can be incorporated into developing the BCG. A very useful initial
exercise is a "practice run" to rate approximately three sites that the facilitation team has reason to
believe might be relatively good condition, mediocre condition, and poor condition, respectively. This
allows panelists to experience the process on which they will be spending considerable time. Upon
completion of the introductory session, the panel begins work, as explained in the following sections.
55
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
3.4.2 Assign Taxa to Attributes
Prior to calibrating BCG levels, the panel assigns taxa in the database to the taxonomic attributes
(attributes I to VI). Assignments of taxa to attributes rely on examination of empirical stress-response
relations, as well as professional experience of field biologists who have sampled the water bodies of
the region. In this way, the professional opinions of the workgroup can be tested with the empirical data
(Figure 12). Several taxa may have insufficient data within the statewide data set. The wider collective
experience of the workgroup can enhance the empirical database with experience with under-
represented taxa, and knowledge of natural history.
In cases of disagreement between empirical analyses and professional opinion, the group can employ a
weight of evidence approach, including consensus professional opinion and strong and consistent
response shown in the data analysis (Figure 12). To save time in the face-to-face panel workshop,
attributes and assignment of taxa to the (taxonomic) attributes can be introduced in the pre-workshop
webinars, and each expert is asked to assign taxa to attributes as homework. Experts are also given
results of the stressor-response analyses of individual taxa. The facilitation team compiles the experts'
taxon assignments prior to the workshop, and participants discuss each taxon to develop consensus
assignments.
After the taxa are assigned to the attributes, the attribute numbers should be entered into the site-
specific data sheets (Figure 12).
3.4.2.1 Example: Alabama Taxon Assignments
Prior to the face-to-face BCG workshop for northern Alabama streams, panelists received taxa lists from
the facilitation team and were asked to make preliminary attribute assignments based on (1) relevant
literature and (2) taxon-response plots showing relationships between the frequency and abundance of
the taxa over the range of the Alabama HDG. The facilitation team compiled the results and used them
as a starting point for the attribute assignment component of the workshop, during which panelists
made assessments based on consensus professional opinion. Once the attribute assignments were
made, the facilitator entered them into a master taxa worksheet, which automatically updated the
attribute assignments in the sample worksheets (Figure 13). Table 10 shows the distribution of
macroinvertebrate and fish taxa across attribute categories for northern Alabama streams.
56
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 10. Distribution of macroinvertebrate and fish taxa across the BCG attributes in northern
Alabama
BCG Attribute
1
II
III
IV
V
Va
VI
X
—
Historically
documented, sensitive,
long-lived, or regionally
endemic taxa
Highly sensitive taxa
Intermediate sensitive
taxa
Taxa of intermediate
tolerance
Tolerant native taxa
Opportunistic tolerant
taxa
Non-native taxa
Migrating fish (surrogate
for ecosystem
connectance)
No attribute assignment
(insufficient
information)
Totals
Macroinvertebrates
#of
taxa
1
110
136
173
67
2
171
660
%of
individuals
0.2
16.7
20.6
26.2
10.2
0.3
25.9
100
Examples
Gastropods: Fontigens
Beetles: Optioservus,
Mayflies: Heptagenia,
Leucrocuta,
Caddisflies:
Brachycentrus,
Glossosoma,
Stoneflies/Leuctra,
Tallaperla
Beetles: Macronychus,
Mayflies: Stenonema,
Isonychia, Midges:
Tvetnia, Brillia,
Caddisflies: Chimarra,
Odonata: Macromia
Midges: Polypedilum,
Tanytarsus,
Rheotanytarsus,
Thienemannimyia,
Beetles: Stenelmis,
Dragonflies: Boyeria,
Mayflies: Baetidae
Caddisflies:
Cheumatopsyche,
Worms: Oligochaeta,
Midges: Ablabesmyia,
Dicrotendipes,
Dragonflies: Argia,
Flies: Simulium,
Gastropods: Physella
Corbiculo and
Coarse identifications
and uncommon
occurrences
Fish
#of
taxa
6
15
38
76
29
9
5
2
39
219
%of
individuals
2.7
6.8
17.4
34.7
13.2
4 1
2.3
0.9
17.8
100
Examples
Bankhead Darter, Crown
Darter, Holiday Darter, Sipsey
Darter
Burrhead Shiner, Cahaba
Shiner, Bigeye Shiner, Goldline
Darter, Warpaint Shiner,
Blenny Darter
Shadow Bass, Black Redhorse,
Rock Bass, Northern Studfish,
Southern Studfish, Bigeye
Chub, Tuskaloosa Darter,
Rainbow Shiner
Longear Sunfish, Alabama Hog
Sucker, Banded Sculpin,
Alabama Shiner, Silverstripe
Bluegill, Blackbanded Darter,
Largemouth Bass, Striped
Shiner, Spotted Bass, Blacktail
Shiner, Blackspotted
Topminnow
Creek Chub, Bluntnose
Minnow, Redbreast Sunfish,
Western Mosquitofish, Eastern
Mosquitofish, Green Sunfish,
Largescale Stoneroller, Yellow
Bullhead
Common Carp, Fathead
Minnow, Goldfish, Grass Carp,
Red Shiner
American Eel, Atlantic
Needlefish
Uncommon occurrences
57
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
3.4.3 Assign Sites to Condition Levels
Working from a description of undisturbed communities and the species composition data from
example sites, the panel assigns sites to the levels of the BCG. These site assignments are used to
describe changes in the aquatic communities for lower levels of biological condition, leading to a
complete descriptive model of the BCG for the region. Throughout this process, the panel makes use of
the prepared data (Figure 12 and Figure 13) to examine species composition and abundance data from
sites with different levels of cumulative stress, from least stressed to severely stressed.
3.4.3.1 Description of Natural, Undisturbed Conditions
First, the panel attempts to reconstruct the native aquatic assemblages under natural, undisturbed
conditions. This is an application of historical ecology (McClenachan et al. 2015), and if resources are
available, a formal effort should be made to describe the historical conditions. The description of natural
conditions requires biological knowledge of the region, a natural classification of the assemblages, and,
if available, historical descriptions of the habitats and assemblages. A useful exercise is to ask each
panelist to describe the community of an undisturbed, natural system. This develops a best professional
judgment description of undisturbed communities for the region that is, at best, qualitative.
Descriptive studies of historic and prehistoric distributions of species can be useful in developing a
description of pre-settlement or pre-industrial conditions. For example, most classic fish distribution
monographs draw heavily on early descriptions and collections by 19th century naturalists (e.g.,
descriptions in The Fishes of Ohio; Trautman 1981) to develop estimates of pre-settlement distributions
for as many species as possible. Fish and mollusks have also been investigated from native and early
settler middens to derive distributions of harvested species, and these can be combined with other
studies to develop more comprehensive descriptions (e.g., Angelo et al. 2002, 2009).
For example, in Kansas, few streams have completely escaped the effects of large-scale agricultural and
livestock practices implemented over the past 150 years (Angelo et al. 2009). Although many of the
biological surveys from the mid-1800s were performed after the start of intensive agriculture, they still
provide valuable documentation of the occurrence of several freshwater species that soon disappeared
from specific watersheds or the region as a whole. Museum collections and other historical records
indicate that many creeks and smaller rivers in the Great Plains supported a variety of predominately
eastern fish and shellfish species, most requiring clear water and relatively stable stream bottoms. In
fact, Kansas was once home to more than 50 Unionid mussel species. Today, several mollusk species are
no longer found in most of their original habitats (Figure 14). Over the past 150 years, at least 11 aquatic
molluscan taxa have become extinct in Kansas, and an additional 23 species are currently designated as
endangered, threatened, or vulnerable.
58
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Historical populations
Known extant population
Figure 14. Decline in geographical distribution of black sandshell mussel in Kansas (after Angelo et al. 2009).
A description of undisturbed conditions may also be developed more quantitatively if databases,
expertise, and resources are available. With the growth of biological monitoring, there have been
several recent attempts to develop predictive statistical models of biological composition (typically
metrics, but also taxa) using multiple regression (e.g., Waite et al. 2010) or other modeling approaches
(e.g., random forests [DeWalt et al. 2009]; Threshold Indicator Taxa ANalysis (TITAN) [Baker and King
2010]). These model approaches can be used to extrapolate to undisturbed conditions and predict
relevant metrics (Waite et al. 2010), composition, or individual species ranges (DeWalt et al. 2009)
under undisturbed conditions. They are especially useful if museum records, paleolimnological
investigations, or historical descriptions do not apply (e.g., invertebrates were typically of less interest
than fish to early explorers and many naturalists).
There are challenges and drawbacks when using historical data to reconstruct natural stream conditions.
It takes a great deal of time and commitment to piece together numerous bits of information, especially
considering the limitations and inconsistencies inherent in historical data. Much of the information is
not directly comparable to modern assessment data, largely because results from previous studies and
observations are often based on different sampling methodologies. Sometimes the data are not
applicable because they were obtained after settlers significantly impacted the land, but often such
physical habitat data are missing or incomplete. Finally, some regions settled early in the history of the
nation may simply lack definitive historical data on the baseline biological condition.
As an example, Shumchenia et al. (2015) constructed the first estuarine BCG framework that examines
changes in habitat structure through time. Using historical data and descriptions, including maps,
navigational charts, land use descriptions, sediment cores, and shellfish landings, they described a
minimally disturbed range of conditions for the ecosystem, anchored by observations before 1850. Like
59
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
many estuaries in the U.S., the relative importance of environmental stressors changed over time, but
even qualitative descriptions of the biological indicators' status provided useful information for defining
condition levels. In addition to helping conceptually define the biotic community expected in an
undisturbed or minimally disturbed environment, the BCG was used to show that stressors rarely acted
alone and that declines in one biological indicator influenced the increase or decline of others.
3.4.3.2 Assignment of Current Sites
The panel works with data tables showing the species and attributes for each site (Figure 13). In
developing assessments, the panel works "blind," that is, no stressor information is included in the data
table. Only non-anthropogenic classification variables are shown (in Figure 13, watershed area and
gradient). Sites are selected by the facilitation team to span the range of stress that occurs in the region,
from the least stressed to the most stressed. Panel members discuss the species composition and what
they expect to see for each level of the BCG.
A typical site assignment proceeds as follows: The facilitator projects the data onto a screen (Figure 13)
and calls out some salient data on the site, including area, gradient, total taxa, and possibly some
summary metrics. Panelists take several minutes to look at the data, and each panelist proposes a BCG
level for the site, along with principal reasons for the decision. The site and decision reasons are
discussed by the panel, and panelists are allowed to change their decisions, if desired.
Following assignment of 20 or more sites to levels of the BCG, the panel develops a description of each
level, along with rules that are expected to be met by each level, starting from the highest quality
condition observed in the data set (e.g., level 1) and working down to the most severely altered
condition (e.g., level 6). The description and rules can be as quantitative as the panel cares to make
them. Examples of water bodies that might have low resolution include intermittent and ephemeral
streams, wetlands, and tidal fresh portions of estuaries. Also, BCG levels might be absent from the data
set. In most developed states, there is general recognition that BCG level 1 is exceedingly rare or absent.
BCG level 6 is often absent from data sets because the most egregious pollution has been remediated,
leaving level 5 as the poorest quality observed. Level 6 may sometimes be observed in older data (pre-
1985). If a panel determines that two or more levels cannot be discriminated, then they are typically
combined into one; for example "levels 3-4" or "levels 5-6." This should only be done when the panel
determines that the levels cannot be discriminated, not simply because one or more levels happen to be
absent from the given data set.
Assessing biological condition and assigning sites to a level of the BCG are based on the detailed
attribute descriptions developed earlier for the water body and region for which the model is being
developed, plus other taxonomic attributes the panel agrees are important. It is entirely possible to
determine biological condition with a subset of the attributes. For example, biological assessment in
streams and rivers is currently carried out with indicators very similar to taxonomic and condition
attributes I through VII of the BCG, all derived from species composition. However, a measure of the
spatial distribution of estuarine habitats for assessing whole estuary condition is under development in
Narragansett Bay based on a spatial habitat measure and on the "historic balance" of critical estuarine
habitats in Tampa Bay (Cicchetti and Greening 2011; Shumchenia et al. 2015). This indicator is under
development as a surrogate for attribute X (ecosystem connectivity), and would provide information on
the presence and spatial relationship of habitats critical to a functioning estuarine system. The
importance of individual attributes depends on the system being assessed, and information or indicators
for all attributes may not be necessary.
60
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
As an example, a panel of aquatic biologists from three states (Michigan, Wisconsin, and Minnesota) and
four tribal water quality agencies calibrated BCG models for coolwater wadeable streams of the Upper
Midwest (Gerritsen and Stamp 2012). Prior to performing site assessments, the group discussed their
expectations for sites spanning the different BCG levels. Table 11 contains the narrative descriptions of
each of the BCG levels (modified after Davies and Jackson (2006)), as well as lists of fish and
macroinvertebrate taxa that the group expected to commonly find in samples from each BCG level. The
overall relationship between BCG level and Minnesota's disturbance score is shown in Figure 15.
Table 11. Description of transitional cold-cool assemblages (benthic macroinvertebrate and fish taxa)
in each assessed BCG level, Upper Midwest coldwater streams. Definitions are modified after Davies
and Jackson (2006) (Source: Gerritsen and Stamp (2012)).
Definition: Natural or native condition—native structural, functional, and taxonomic integrity is preserved;
ecosystem function is preserved within the range of natural variability
BCG
level 1
Fish: If the stream is in a location where brook trout are native, native brook trout must be present. Non-native
salmonids must be absent. Up to twelve additional taxa, including highly sensitive (attribute I, II, & III) species such
as slimy sculpin and brook lamprey, are also be present. If tolerant taxa are present, they occur in very low
numbers.
Macroinvertebrates: There is a lack of sufficient information to know what the historical undisturbed
macroinvertebrate assemblage looked like.
Definition: Minimal changes in structure of the biotic community and minimal changes in ecosystem function—
virtually all native taxa are maintained with some changes in biomass and/or abundance; ecosystem functions are
fully maintained within the range of natural variability
BCG
level 2
Fish: Overall taxa richness and density is as naturally occurs. Non-native salmonids may be present. If the stream is
in a location where brook trout are native, native brook trout must be present and must not be negatively
impacted by non-native salmonids such as brown trout. Other highly sensitive (attribute II) and intermediate
sensitive (attribute III) taxa such assculpins (mottled or slimy), dace (pearl, finescale, northern red belly, longnose)
and brook lamprey are also present. Tolerant taxa may be present but in low numbers.
Macroinvertebrates: Overall taxa richness and density is as naturally occurs. Most sensitive (attribute II) taxa (e.g.,
Trichoptera'. Glossosoma, Rhyacophila, Lepidostoma, Dolophilodes; Ephemeroptera: Ephemerella, Epeorus;
Plecoptera: Leuctridae) and other taxa must be present. These plus intermediate sensitive (attribute III) taxa (e.g.,
Ephemeroptera: Paraleptophlebia; Plecoptera: Acroneuria, Isoperla, Paragnetina; Trichoptera: Brachycentrus,
Chimarra) occur in higher relative abundances than in BCG level 3 samples. Tolerant taxa occur in low numbers.
Definition: Evident changes in structure of the biotic community and minimal changes in ecosystem function-
Some changes in structure due to loss of some rare native taxa; shifts in relative abundance of taxa but
intermediate sensitive taxa are common and abundant; ecosystem functions are fully maintained through
redundant attributes of the system
BCG
level 3
Fish: Overall taxa richness and density is as naturally occurs. Sensitive taxa such as dace (pearl, finescale, northern
red belly, longnose) and northern hog suckers must outnumber tolerant taxa such as central stonerollers and
bluegill. Taxa of intermediate tolerance (attribute IV) such as white suckers, blacknose dace, common shiners,
darters (johnny, fantail), and creek chub are common, and some tolerant (attribute V) taxa such as northern pike,
yellow perch, and stonerollers may be present. If extra tolerant taxa such as green sunfish and bluntnose and
fathead minnows are present, they occur in very low numbers.
Macroinvertebrates: Overall taxa richness and density is as naturally occurs. Similar to BCG level 2 assemblage
except sensitive taxa (e.g., Ephemeroptera: Paraleptophlebia; Plecoptera: Acroneuria, Isoperla, Paragnetina;
Trichoptera: Brachycentrus, Chimarra; Diptera: Diamesa, Eukiefferiella, Tvetenia) occur in lower relative
abundance and the most sensitive (attribute II) taxa may be absent. Taxa of intermediate tolerance (attribute IV)
(e.g., Gammarus, Oligochaeta, Simulium; Coleoptera: Optioservus, Stenelmis; Ephemeroptera: Baetis, Stenonema;
Trichoptera: Hydropsyche, Cheumatopsyche) are common, and some tolerant taxa (attribute V) occur in low
numbers.
61
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Definition: Moderate changes in structure of the biotic community and minimal changes in ecosystem function—
Moderate changes in structure due to replacement of some intermediate sensitive taxa by more tolerant taxa, but
reproducing populations of some sensitive taxa are maintained; overall balanced distribution of all expected major
groups; ecosystem functions largely maintained through redundant attributes
BCG
level 4
Fish: Sensitive taxa such as dace (pearl, finescale, northern red belly, longnose) and northern hog suckers are
present but occur in very low numbers. Taxa of intermediate tolerance (attribute IV) such as white suckers,
blacknose dace, common shiners, darters (johnny, fantail) and creek chub are common, and some tolerant
(attribute V) taxa such as northern pike, yellow perch and stonerollers are present. When compared to BCG level 3
samples, highly tolerant taxa such as green sunfish and bluntnose and fathead minnows are present in greater
numbers.
Macroinvertebrates: Overall taxa richness is slightly reduced. Sensitive taxa (including EPT taxa) are present but
occur in low numbers. Taxa of intermediate tolerance (attribute IV) (e.g., Gammarus, Oligochaeta, Simulium;
Coleoptera: Optioservus, Stenelmis; Ephemeroptera: Baetis, Stenonema; Trichoptera: Hydropsyche,
Cheumatopsyche) are common, as are tolerant (attribute V) taxa (e.g., Diptera: Cricotopus, Dicrotendipes,
Paratanytarsus; Hyalella; Physa; Turbellaria).
Definition: Major changes in structure of the biotic community and moderate changes in ecosystem function—
Sensitive taxa are markedly diminished; conspicuously unbalanced distribution of major groups from that
expected; organism condition shows signs of physiological stress; system function shows reduced complexity and
redundancy; increased build-up or export of unused materials.
BCG
level 5
Fish: Overall taxa richness may be reduced. Sensitive taxa drop out. Taxa of intermediate tolerance (attribute IV)
such as white suckers, blacknose dace, common shiners, darters (johnny, fantail), and creek chub are common.
There is an influx of tolerant and highly tolerant taxa such as bluegill, yellow perch, largemouth bass, northern
pike, central stonerollers, bluntnose minnows, fathead minnows, and green sunfish.
Macroinvertebrates: Overall taxa richness is slightly reduced. Sensitive taxa may be absent. Taxa of intermediate
tolerance (attribute IV) (e.g., Gammarus, Oligochaeta, Simulium; Coleoptera: Optioservus, Stenelmis;
Ephemeroptera: Baetis, Stenonema; Trichoptera: Hydropsyche, Cheumatopsyche) and tolerant (attribute V) taxa
(e.g., Diptera: Cricotopus, Dicrotendipes, Paratanytarsus; Hyalella; Physa; Turbellaria) are common. Tolerant taxa
occur in higher abundances than in BCG level 4 samples.
62
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
90
80
70
60
V) 50
Q
1 40
30
20
10
0
90
80
70
60
2 50
1 40
30
20
10
n
" T
D
I
a
o
.
J
[
a
*•
a
0
1
1
D
(
J -
•
-
-
n '
•
.
T
n
1
j T
i
a
-1-
D
I
n
D
o
D
•
-
•
.
o •L
n Median
25%-75%
2345
Nominal BCG Level
o Outliers
* Extremes
Figure 15. Box plots of HDS for Minnesota streams, grouped by nominal BCG level (panel majority choice) for
fish (upper) and macroinvertebrate (lower) samples. HDS scores range from 0 (most disturbed) to 81 (least
disturbed) (Gerritsen et al. 2013).
63
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
3.4.3.3 Variability in Panelist Biological Condition Gradient Calls
Consistency among panelists is important. In addition to integer BCG levels (e.g., levels 2, 3, 4), panelists
also aim to identify sites somewhat better or somewhat worse than the integer levels, up to and
including samples that are borderline between adjacent BCG levels. In calibration exercises,
intermediate levels have been assigned (+) and (-). This information has been used to help define the
threshold where an expert would assign a site to a different BCG level. An expert assigning a site to a
BCG level with a (+) or (-) caveat would be asked what additional change in the site data would lead to a
different level assignment, and why.
For the BCG project in the Northern Piedmont of Maryland, the macroinvertebrate workgroup assessed
46 calibration samples. Panelists rated samples in the six BCG levels, and modified those with (+) and (-)
as desired. Median BCG level assignments were calculated for each sample as the group nominal level.
Deviations of each panelist's assignments from the group median call were estimated, where deviations
were assumed to be in quantities of ]4 BCG level. Deviations are shown in Figure 16. On average, 62% of
BCG level assignments matched exactly with the median, 32% were within ±]4 BCG level, 5% were within
±% BCG level, and 1% differed by one BCG level (Figure 16).
400
350
CD
O)
Cfl
CD
"c/5
"CD
CD
.9- 150
.0
E
100
0
62%
3%
2%
0%
-1.00 -0.67 -0.33 0 0.33 0.67 1.00
Difference from median
Figure 16. Distribution of individual panelists BCG assignments, as deviations from group sample median,
Maryland Piedmont BCG workshop. Percentages above each bar. Data from Stamp et al. 2014.
64
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
3.5 Biological Condition Gradient Decision Rules
This chapter described steps to develop narrative descriptions and rules for assigning sites to BCG levels.
The core objective of the panel process is to elicit expert judgment on what the experts consider
ecologically significant change in the biotic community—and to document the underlying rationale.
Through development of expert consensus, first narrative and then quantitative rules emerge, and they
are tested and refined based on the current state of the science. Additionally, where gaps in information
are identified, the development of decision rules is comparable to formulating a hypothesis, thereby
setting up opportunities for applied research that clearly articulate water quality management
information needs for goal setting and condition assessments.
The chapter concludes with development of narrative descriptions of BCG levels for specific water
bodies within a region or basin. Chapter 4 addresses how to convert these narrative descriptions into
narrative then quantitative decision rules for a numeric BCG model. There is no bright line between
development of the narrative description and numeric decision rules. In all BCG development efforts to
date, preliminary quantitative decision rules have emerged early as part of developing the narrative
description and rules. In the first round of data analysis and interpretation, the experts typically
formulate their reasoning in the following manner: "I expect more (or fewer) species because ...." or
"the presence of two or more taxa of attribute III signifies this condition level to me because ...." By the
second or third round of the data exercise assigning sites to BCG levels, increasingly quantitative
statements are provided when experts are asked to explain their logic for assigning sites to BCG levels.
These preliminary quantitative statements provide a template for building quantitative decision rules
through an iterative, interactive process with the expert panel. Encapsulation of expert judgment
provides the transparency and clarity for decision makers and stakeholders to understand the logic and
science underpinning ALL) goal descriptions and assessments.
65
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Chapter 4. Quantitative Rules and Decision Systems
Routine use of a quantitative BCG model requires a way to automate application of the decision rules so
that assessments can be made for newly sampled water bodies without reconvening the expert panel.
This chapter discusses approaches to quantify the narrative BCG model and to test and validate the
numeric model, corresponding to Steps 4 and 5 of the BCG Calibration process (Figure 8). Quantitative
rules rely on sample data using standardized protocols (i.e., most applicable to attributes II-VI). This
chapter presents:
• An approach to quantify the conceptual BCG framework and develop a numeric model. This
approach is based on elicitation of the experts' decision criteria and incorporation into a
numeric decision model using a mathematical set theory approach (e.g., fuzzy logic) (See section
4.1). This approach has been tested and refined in most of the BCG projects to date.
• Considerations and approaches for relating the BCG with the state's existing biological
assessment methods and tools (e.g., biological indices such as MMIs and O/E models) (See
section 4.2). To date, most states have developed biological indices.
• An additional approach to quantify the BCG narrative decision rules that has been implemented
by a state, multivariate linear discriminant modeling. This approach involves development of
statistical models that "predict" (or imitate) the expert decisions and may or may not use
elicited expert reasoning or rules (See section 4.3). As BCG development and calibration
continues, it is expected that the BCG process will be refined and expanded and alternate
methods identified and tested.
4.1 Quantitative Rule Development and Application
This approach assumes that because the expert panelists largely agree on BCG ratings for water bodies,
they use a common set of decision criteria to achieve the ratings. The approach consists of deriving
narrative and numeric decision rules based on expert logic and consensus, including testing of the rules
with the expert panel and then with experts outside of the panel. Application of the decision criteria—a
set of quantitative rules—can then be applied to any relevant data set or sample.
Quantitative rule and direct decision model development is comprised of the following steps:
• Elicitation of numeric decision criteria—During the expert panel meeting, experts are asked for
their reasoning behind the decisions. The reasoning is the basis for the BCG level descriptions
(Table 11), and also for decision criteria (narrative rules) that the experts use. The narrative
rules are elicited from the panel and then quantified.
• Quantification and testing—Quantitative rules in turn form the basis of a decision model. A
methodology to apply the elicited rules is through a mathematical set theory approach, fuzzy
logic (Zadeh 1965, 2008), which mimics human thinking and decision making. Results of the
quantitative decision model are compared to the panel's decision, and mismatches are further
discussed by the panel to resolve ambiguous or incomplete rules. Ideally, the final model should
be tested with an independent data set that was assessed by the panel but not used to calibrate
the model. Other approaches to rule elicitation and development include reproducing the
expert panel results (but not necessarily their reasoning) with an empirical discriminant analysis
model (section 4.2; Davies et al. In press; Shelton and Blocksom 2004), or developing a Bayesian
predictive model from the elicitation of reasoning (e.g., Kashuba et al. 2012).
66
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
4.1.1 Elicitation of Numeric Decision Criteria
Level descriptions in the BCG conceptual model are intentionally general (e.g., reduced richness,
increased dominance, loss or replacement of specific assemblages), which allows for different methods,
sources of information, and interpretations to be used in rule development. To allow for consistent
assignments of sites to levels, it is necessary to formalize the expert knowledge by codifying level
descriptions into a set of rules (e.g., Droesen 1996). If formalized properly, water quality management
program scientists with adequate data can follow the rules to obtain the same level assignments as the
group of experts. This replicability makes the actual decision criteria transparent to stakeholders.
Rules are logic statements that experts use to make their decisions (e.g., "If plecoptera richness is high,
then biological condition is high."). Rules on attributes can also be combined (e.g., "If the proportion of
highly sensitive taxa (attribute II) is high, the proportion of tolerant individuals (attribute V) is low, and
so on, then assignment is BCG level 2.").
Numeric rule development requires discussion and documentation of level assignment decisions and the
reasoning behind the decisions. During this discussion, it is necessary to record each participant's level
decision (i.e., vote) for the site, the critical or most important information for the decision (e.g., the
number of taxa of a certain attribute, the abundance of an attribute, the presence of indicator taxa), and
any confounding or conflicting information and how this information was reconciled for the eventual
decision.
As the panel assigns example sites to BCG levels, the panel members are polled on the critical
information and criteria they used to make their decisions. These form preliminary, narrative rules that
explain how panel members make decisions. For example, "For BCG level 2, sensitive taxa must make up
at least half of all taxa in a sample." The decision rule for a single level of the BCG does not always rest
on a single attribute (e.g., highly sensitive taxa) but may include other attributes as well (intermediate
sensitive taxa, tolerant taxa, indicator species, organism condition), so these are termed "Multiple
Attribute Decision Rules." With data from the sites, the rules can be checked and quantified. For
mathematical fuzzy set modeling, quantification of rules will allow the agency to consistently assess sites
according to the same rules used by the expert panel, and it will allow a computer algorithm, or other
persons, to obtain the same level assignments as the panel.
Rule development requires discussion and documentation of BCG level assignment decisions and the
reasoning behind the decisions. During this discussion, the facilitators record:
• Each participant's decision for the site:
o The critical or most important information for the decision—for example, the number or
abundance of taxa of a certain attribute, the presence of indicator taxa, the absence of
certain taxa, and explanation why this information is ecologically important.
o Any confounding or conflicting information and how this was resolved for the eventual
decision.
• Iteration
o Rule development is iterative, and it usually requires at least two panel sessions.
o Building from the initial site assignments, preliminary narrative rules are developed.
Descriptive statistics of the attributes and other biological indicators for each BCG level
67
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
determined by the panel are then developed for testing. These statistical descriptions will
be used for testing and refinement as numeric decision rules are developed and vetted.
o Following the initial development phase, the draft rules are tested by the panel with new
data to ensure that new sites are assessed in the same way. The new test sites should not
have been used in the initial rule development and also should span the range of
anthropogenic stress. Any remaining ambiguities and inconsistencies from the first
iterations are also resolved at this stage.
4.1.2 Codification of Decision Criteria: Multiple Attribute Decision Criteria Approach
The expert rules can be automated in Multiple Attribute Decision Models. These models replicate the
decision criteria of the expert panel by assembling the decision rules using logic and set theory, in the
same way the experts used the rules. In the case studies presented later in this chapter, the models
replicated expert panel's decisions at greater than 90% accuracy, including tied or intermediate
decisions between adjacent BCG levels (e.g., between level 3 and level 4).
Instead of a statistical prediction of expert judgment, this approach directly and transparently converts
the expert consensus to automated site assessment. The method uses modern mathematical set theory
and logic (called "fuzzy set theory") applied to rules developed by the group of experts. Mathematical
fuzzy set theory is directly applicable to environmental assessment, it has been used extensively in
engineering applications worldwide (e.g., Demicco and Klir 2004), and environmental applications have
been explored in Europe and Asia (e.g., Castella and Speight 1996; Ibelings et al. 2003).
Mathematical fuzzy set theory allows degrees of membership in sets, and degrees of truth in logic,
compared to all-or-nothing in classical set theory and logic. Membership of an object in a set is defined
by its membership function, a function that varies between 0 and 1. One can compare how classical set
theory and fuzzy set theory treat the common classification of sediment, where sand is defined as
particles less than or equal to 2.0 mm diameter, and gravel is greater than 2.0 mm (Demicco and Klir
2004). In classical "crisp" set theory, a particle with diameter of 2.00 mm is classified as "sand," and one
with 2.01 mm diameter is classified as "gravel." In fuzzy set theory, both particles have nearly equal
membership in both classes (Demicco 2004). Measurement error of 0.005 mm in particle diameter
greatly increases the uncertainty of classification in classical set theory, but in fuzzy set theory a particle
near the boundary would have nearly equal membership in both sets "sand" and "gravel." Fuzzy sets,
thus, retain the understanding and knowledge of measurements close to a set boundary, which is lost in
classical sets.
Demicco and Klir (2004) proposed four reasons why mathematical fuzzy sets and logic enhance scientific
methodology, and these are applicable to BCG development:
• Fuzzy set theory has greater capability to deal with "irreducible measurement uncertainty," as in
the sand/gravel example above.
• Fuzzy set theory captures vagueness of linguistic terms, such as "many," "large," or "few."
• Fuzzy set theory and logic can be used to manage complexity and computational costs of control
and decision systems.
• Fuzzy set theory enhances the ability to model human reasoning and decision making, which is
critically important for defining thresholds and decision levels for environmental management.
68
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
4.1.2.1 Rule-based Inference Model
People tend to use strength of evidence in defining decision criteria, and in allowing some deviation
from their ideal for any individual attributes, as long as most attributes are in or near the desired range.
For example, the definitions of "high," "moderate," "low," etc. are quantitative and can be interpreted
and measured to mean different things. An important step in the BCG process is development of expert
consensus defining these, or other, general terms and documenting the expert logic that is the basis for
the decisions. The decision rules preserve the collective professional judgment of the expert group and
set the stage for the development of models that can reliably assign sites to levels without having to
reconvene the same group. In essence, the rules and the models capture the panel's collective decision
criteria.
An inference model is developed to replicate the panel decision process, and this section describes an
inference model that uses mathematical fuzzy logic to mimic human reasoning. Each linguistic variable
(e.g., "high taxon richness") must be defined quantitatively as a fuzzy set (e.g., Klir 2004). A fuzzy set has
a membership function, and example membership functions of different classes of taxon richness are
shown in Figure 17. In this example (Figure 17), piecewise linear functions (functions consisting of line
segments) are used to assign membership of a sample to the fuzzy sets. Fuzzy membership functions
were assumed to be adequately defined by piecewise linear functions. Metric values below a lower
threshold have membership of 0; values above an upper threshold have membership of 1, and
membership is a straight line between the lower and upper thresholds. For example, in Figure 17 (top), a
sample with 20 taxa would have a membership of approximately 0.5 in the set "Low to Moderate Taxa"
and a membership of 0.5 in the set "Moderate Taxa."
S2 o
n
E
10 15 20 25 30
Number of Total Taxa
35 40
10
15
20
Number of Sensitive Taxa
Figure 17. Fuzzy set membership functions assigning linguistic values to defined ranges for Total Taxa (top) and
Sensitive Taxa (bottom). Shaded regions correspond to example rules for BCG level 3: "Number of total taxa is
high," and "number of sensitive taxa is low-moderate to moderate."
69
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
How are inferences made? Suppose there are two rules for determining whether a water body is BCG
level 3 (using definitions of Figure 17):
• The number of total taxa is high.
• The number of sensitive taxa is low-moderate to moderate.
In classical set theory, the boundaries between the categories would be vertical lines at the intersections
of the membership functions in Figure 17. The rules would then be:
• Total taxa > 30
• Sensitive taxa > 4 and sensitive taxa < 15
If the two rules are combined with an "AND" operator, that is, both must be true, then under classical
set theory, if total taxa = 30 and sensitive taxa = 5, the sample would be judged not to be in the set of
BCG level 3, because the rule specifies total taxa must be greater than 30. Finding a single additional
taxon would result in assessment of BCG level 3. In fuzzy set theory, an AND statement is equivalent to
the minimum membership given by each rule:
Level 3 = MIN (total taxa is high, sensitive taxa is low to moderate)
For 30 total taxa, fuzzy membership in "total taxa is high" = 0.5 (Figure 17), and fuzzy membership in
"Sensitive taxa is low-moderate to moderate" = 1.0 (Figure 17). Membership of level 3 is then 0.5. In the
fuzzy set case, a single additional taxon raises the membership in BCG level 3 from 0.5 to 0.6.
If the two rules are combined with an "OR" operator, then either can be true for a site to meet BCG level
3, and both conditions are not necessary. Crisp set theory now yields a value of "true" if total taxa = 32
and sensitive taxa = 4 (total taxa > 27, therefore it is true). Fuzzy set theory yields a membership of 1
(maximum of 0.5 and 1). Using the fuzzy set theory model, finding a single additional taxon in a sample
does not cause the assessment to flip to another level, unlike crisp decision criteria.
Output of the inference model may include membership of a sample in a single level only, ties between
levels, and varying memberships among two or more levels. The level with the highest membership
value is taken as the nominal level.
4.1.2.2 Quantitative Model Development
Rules identified by the panel, whether quantitative or qualitative, are compared to data summaries of
the panel decisions. In particular, if the panel identified a moderate number of sensitive taxa for BCG
level 3, then the analyst (i.e., the individual who develops the quantitative decision model) examines the
number of sensitive taxa in samples the panel assigned to BCG level 3. The analyst selects a reasonable
minimum of the distribution of sensitive taxa in BCG level 3, say the minimum or a 10th quantile, as the
decision threshold. This is repeated for all rules and attributes identified by the panel members as being
important to their decisions. As a starting point, a plot of the attribute or metric values as box plots by
the panel-designated BCG level can be helpful (see section 4.1.2.3 for an example). This type of graphic
shows minimum, maximum, median, and selected quantiles for each metric and BCG level. Sample sizes
for each BCG level might be small, especially for the highest and lowest levels (BCG levels 1 and 2, and 6,
respectively), and might require more professional judgment from the panel to develop rules.
70
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
For a particular attribute or metric, the threshold identified by the panel will typically be the 50%
membership value in a fuzzy membership function. For example, if the panel identifies "5 or more"
sensitive taxa as a requirement for BCG level 3, then 5 taxa would correspond to 50% membership;
3 taxa may correspond to 0% membership, and 7 taxa to 100%. Because number of taxa are always
whole numbers, the membership function is not continuous. Some rules are non-fuzzy: if a rule requires
"at least 1" or "presence," then presence receives a membership of 100% and absence receives 0%.
A spreadsheet is convenient for developing the rule-based model. Membership functions and rules for
each level and each relevant attribute or metric are laid out in the top row, and data for each sample are
arrayed in rows. Sample data are called by the rule formulas and the final decision logic is applied to
determine membership in each BCG level for each sample.
In models developed up to now, rules work as a logical cascade from BCG level 1 to level 6. A sample is
first tested against the level 1 rules; if a required rule fails, then the level fails, and the assessment
moves down to level 2, and so on (Figure 18). Depending on how the expert panel makes decisions and
rates samples, component rules for a single level may be (1) all-or-nothing (i.e., all rules must be met);
(2) some rules have alternate rules (e.g., a very low percentage of tolerant individuals may substitute for
a high percentage of sensitive individuals); or (3) any number n of, say, n + 1 rules must be met.
Required rules must be true for a site to be assigned to a level. BCG levels 1 and 2 represent minimally-
disturbed, natural conditions, hence the rules tend to be the most restrictive. As assemblages change
with increasing anthropogenic influence, the changes may manifest in different effects (decline of
sensitive species; and/or increases in abundance or dominance of tolerant individuals), and the rules for
the middle levels may have more alternative situations. In the more degraded levels (especially BCG
level 5), the rules tend to be simple, reflecting a degraded and simplified assemblage. In the cascading
logic from BCG level 1 to 6 (Figure 18), there are no rules for level 6 because it is the bottom "bin" that
catches sites that fail rules from levels 1 to 5. Examples of these are shown in the case studies that
follow.
Two examples on development of numeric decision rules for streams and wadeable rivers follow. The
first example shows development of numeric decision rules for benthic macroinvertebrates and fish for
cold- and cool-water streams in the Upper Midwest. The second example highlights use of diatom
assemblage data from Northern New Jersey in developing a numeric BCG. Both examples illustrate the
BCG development process. Macroinvertebrates follow the classic paradigm that overall species richness
is higher in the higher BCG levels (levels 1 and 2), but coldwater fish and diatoms are nearly opposite:
overall richness is low in pristine coldwater streams, and diatom richness is low in undisturbed
oligotrophic streams. Both are dominated by a small number of highly sensitive taxa. As streams
become more disturbed, richness and abundance of intermediate and tolerant taxa increase for both
fish and diatoms. In the fish assemblage, sensitive taxa disappear in the most disturbed sites, but
sensitive taxa may hang on in highly-disturbed diatom assemblages.
71
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
How does the BCG model work? Like a cascade..
Example: coldwater sample from site where watershed size is < 10 mi2 and
brook trout are native
Does the sample
meet ALL BCG
Level 1 criteria?
NO
• # Total taxa <4
• Sensitive taxa (Att 1 + II) - present
• Native brook trout - present
• % Sensitive taxa (Att II + III) > 50% YES
• % Sensitive individuals (Att II + III) > 60%
• % Tolerant (Att V + Va + Via) individuals < 5%
• Non-native salmonids (Att VI) - absent
>
Assigned to
BCG LEVEL 1
i j
^p
Does the sample
meet ALL BCG
Level 2 criteria?
NO
• # Total taxa <8
• Sensitive taxa (Att II + III) - present
• Native brook trout- present
• % Sensitive taxa (Att II + III) > 40% Y Eb
• % Native brook trout: total salmonid individuals > 40%
• % Tolerant non-salmonid (Att V + Va + Via)
individuals < 10%
¥
Assigned to
BCG LEVEL 2
[J
Does the sample
meet ALL BCG
Level 3 criteria?
NO
• # Total individuals < 20
• Sensitive taxa (Att 1 + II) - present
• Salmonids - present
• % Sensitive (Att II + III) and salmonid taxa > 25% YES
• % Sensitive (Att II + III) and salmonid individuals > 20%
• % Non-native trout: total salmonid individuals < 70%
• %Tolerant (Att V + Va + Via) individuals < 40%
>
Assigned to
BCG LEVEL 3
And so on...
* In some situations, alternate rules had to be developed. For example, more taxa naturally occur in large vs. small streams, so total taxa richness rules
were adjusted for watershed size. Some rules also had to be adjusted for streams in which brook trout are not native.
Figure 18. Flow chart depicting how rules work as a logical cascade in the BCG model, from Upper Midwest cold
and coolwater example (Source: Modified from Gerritsen and Stamp 2012). For convenience, midpoints of
membership functions (50% value) only are shown. For complete rules, see Table 15 and Table 16.
4.1.2.3 Example #1: Quantitative Rules and Decision System for Benthic Macroinvertebrates and Fish,
Upper Midwest
Panelists from Indian Nations and the states of Michigan, Wisconsin, and Minnesota calibrated BCG
models for fish and macroinvertebrate assemblages in cold and cold-cool transitional wadeable streams
of the Upper Midwest (Gerritsen and Stamp 2012). The cool-transitional water macroinvertebrate BCG
model was calibrated based on assessments of 37 samples. Panelists made the site assessments using
worksheets that contained lists of taxa, taxa abundances, BCG attribute levels assigned to the taxa, BCG
attribute metrics, and limited site information, such as watershed area, stream size, average July
temperature, and percent forest.
72
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Study Sites
Panelists assigned fish and macroinvertebrate samples from cool-transitional streams to four BCG levels
(BCG levels 2-5). Samples were not assigned to BCG level 1 because panelists did not feel that there was
enough information to know what the historical undisturbed macroinvertebrate assemblage in this
region looked like. Only two of the 37 calibration samples were assigned to BCG level 5 (many of the
coolwater sites in this region are in the Northern Lakes and Forests ecoregion). A detailed verbal
description of each level is given above in Table 11 (Chapter 3).
Decision rules were initially derived from discussions with the panelists on why individual sites were
assessed at a certain level. Panelists made statements such as "BCG level 2 samples should have both a
moderate abundance and richness of sensitive taxa (attributes I, II, and III)." These statements were
compiled into a set of narrative rules (Table 12).
Table 12. Example of Narrative rules for transitional cold-cool assemblages in Upper Midwest streams
(Source: Gerritsen and Stamp (2012))
Definition: Minimal changes in structure of the biotic community and minimal changes in ecosystem function —
virtually all native taxa are maintained with some changes in biomass and/or abundance; ecosystem functions are
fully maintained within the range of natural variability
BCG
level 2
Fish
Taxa richness is low to moderate
Brook Trout, if native, are present
Total sensitive taxa are one third of taxa richness
Abundance of sensitive individuals is low to moderate
Brook Trout (if native) are nearly half of all Salmonidae individuals
Tolerant individuals may be a small fraction of total
Macroinvertebrates
Taxa richness is moderate to high
Highly sensitive (attribute I and II) taxa make up a very small fraction (or more) of total richness and total
abundance
All sensitive taxa (attributes I + II + III) make up moderate fraction of richness and abundance
Sensitive EPT taxa make up at least a small fraction of total richness
Definition: Evident changes in structure of the biotic community and minimal changes in ecosystem function —
Some changes in structure due to loss of some rare native taxa; shifts in relative abundance of taxa, but
intermediate sensitive taxa are common and abundant; ecosystem functions are fully maintained through
redundant attributes of the system
BCG
level 3
Fish
Taxa richness is moderate but not high
Total number of sensitive taxa is greater than tolerant taxa, OR number of sensitive individuals is twice greater
than number of tolerant individuals
Single most dominant intermediate taxon (attribute III) is less than half of all individuals
Extremely tolerant individuals are a very small fraction of total
Macroinvertebrates
Taxa richness is moderate to high
Highly sensitive (attribute I and II) taxa are present
Total sensitive taxa (attributes I + II + III) make up small fraction of richness and abundance
Most dominant tolerant taxon is less than a small fraction of abundance
Sensitive EPT taxa make up at least a small fraction of total richness
73
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
IUU
sn
s
a
— fin
T
CD
•t± ,
C
13
*c
0
0
D
0
D
I T
0
o n
T 1 1 "
e -•-
2345
IUU
CO
CD an
p
T
~ 40
22
ja
•" on
5?
0
•inn
;o
1 an
Q)
£ &n
i
co 40
0
"to
o
S? 0
n
o
D T
f.
1
D T
j_
0 D
T T
D ^ 1— [—
e —
2345
n Median
25%-75%
I Non-Outlier Range
o Outliers
* Extremes
Nominal BCG Level
Figure 19. Benthic Macroinvertebrate Taxa: Box plots of sensitive (attribute l+ll+lll) and tolerant (attribute V)
BCG attribute metrics, grouped by nominal BCG level (panel majority choice). These metrics were used in the
macroinvertebrate BCG model for coldwater streams in the Upper Midwest.
Using the narrative rules, data were examined for numerical ranges and relationships. For example,
examination of the data ranges of attribute I, II, and III taxa for macroinvertebrates (Table 13; Figure 19)
showed that the median percent abundance of attribute I, II, and III taxa from BCG level 2 was 75%. The
decision rules were adjusted by the empirical distributions of the attribute metrics shown in Table 13
and Figure 19, so that the model would replicate the panel's actual decisions as closely as possible. For
the macroinvertebrates, the most important considerations expressed by the experts were percent
individuals and percent taxa metrics for attribute II, ll+lll, IV, and V taxa, and metrics pertaining to three
sensitive orders of aquatic insect taxa (e.g., Ephemeroptera, Plecoptera, and Trichoptera (EPT)).
Panelists expected BCG level 2 samples to have a moderate presence of highly sensitive (attribute II)
taxa, moderate to high total taxon richness, and a low proportion of tolerant (attribute V) taxa. BCG
level 3 samples had similar numbers of total taxa but slightly reduced numbers of highly sensitive
(attribute II) taxa. Total sensitive taxa (attribute ll+lll) were still required to be present in BCG level 4
samples, but with reduced richness and abundance. Higher proportions of tolerant (attribute V)
individuals occurred in BCG level 4 samples, but could not comprise more than 60% of the assemblage.
74
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
BCG level 5 samples were discriminated from BCG level 4 samples by complete loss of sensitive taxa and
a further increase in the percent tolerant (attribute V) individuals.
Table 13. Benthic macroinvertebrate taxa: Ranges of attribute metrics in cold-cool transitional
macroinvertebrate samples. BCG levels by panel consensus, in the Upper Midwest BCG data set
(Gerritsen and Stamp 2012).
Attributes
0 General
II Highly sensitive taxa
III Intermediate
sensitive taxa
II + III All sensitive taxa
IV Intermediate
tolerant taxa
V Tolerant taxa
Metric
Total Taxa
Total Individuals
ft Taxa
% Taxa
% Individuals
ft Taxa
%Taxa
% Individuals
ft Taxa
%Taxa
% Individuals
SensEPTttTaxa
SensEPT_% Individuals
ft Taxa
% Taxa
% Individuals
% Most Dom Individuals
ft Taxa
%Taxa
% Individuals
% Most Dom Individuals
BCG Level (Panel Consensus)
2 (n=19)
20-63
91-359
3-11
8-28
6-42
6-19
19-61
13-55
10-26
30-71
31-76
6-20
18-71
7-28
26-49
23-53
6-31
0-10
0-17
0-22
0-17
3 (n=13)
20-64
134-407
0-7
0-15
0-7
7-19
18-49
17-54
10-24
22-57
20-56
6-14
14-47
7-29
35-53
43-71
8-34
1-11
3-22
0-12
0-6
4 (n=7)
13-58
138-336
0-1
0-3
0-1
4-17
9-31
3-83
4-17
11-31
3-83
1-6
2-17
8-32
50-65
17-87
5-27
0-9
0-16
0-59
0-57
5(n=2)
31-56
294-321
0
0
0
2-6
6-11
1-9
2-6
6-11
1-9
2-4
1-2
16-29
52
26-30
5-15
9-11
20-29
40-72
17-59
6 (n=l)
31
192
4
13
34
16
52
44
20
65
78
13
60
9
29
21
7
0
0
0
0
Observations of the attribute metrics from the fish assemblage are shown in Table 14. No attribute I
species were identified in the coldwater fish assemblage. The fish assemblage in undisturbed or
minimally disturbed coldwater streams typically has few species: native trout, sculpins, and possibly a
minnow species. Increases in fish taxa richness in true coldwater is an indicator of degradation. BCG
levels 1 and 2 required native trout (Brook Trout), but the native trout could be replaced by non-native
salmonids in BCG levels 3 and 4. As with the invertebrates, there was increasing abundance and
dominance of tolerant species, both native and non-native, in the poorer condition levels (BCG levels 4
and 5). No BCG level 6 sites were observed in the cold and cool data set. Panelists identified level 5 rules
(governing the transition from level 5 to level 6) from their experience with BCG level 6 in warmwater
streams.
75
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 14. Fish taxa: Ranges of attribute metrics in cold-cool transitional fish samples. BCG levels by
panel consensus.
Attributes
0 General
II Highly sensitive taxa
III Intermediate sensitive
taxa
II + III All sensitive taxa
IV Intermediate tolerant
taxa
V Tolerant taxa
Va Highly tolerant native
taxa
VI Non-native or
intentionally introduced
taxa
Via Highly tolerant non-
native taxa
Metric
Total Taxa
Total Individuals
ft Taxa
% Taxa
% Individuals
ft Taxa
%Taxa
% Individuals
ft Taxa
%Taxa
% Individuals
ft Taxa
% Taxa
% Individuals
% Most Dom Individuals
ft Taxa
% Taxa
% Individuals
% Most Dom Individuals
ft Taxa
%Taxa
% Individuals
% Most Dom Individuals
ft Taxa
%Taxa
% Individuals
% Most Dom Individuals
ft Taxa
% Taxa
% Individuals
% Most Dom Individuals
BCG Level (Panel Consensus)
1 (n=l)
9
470
2
22
7
3
33
68
5
56
75
4
44
25
14
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2 (n=13)
1-15
11-207
0-2
0-100
0-100
0-5
0-67
0-72
1-5
33-100
14-100
0-9
0-60
0-83
0-39
0-1
0-17
0-13
0-13
0-1
0-11
0-1
0-1
0-1
0-20
0-25
0-25
0
0
0
0
3 (n=14)
4-18
8-598
0-2
0-25
0-20
0-5
0-36
0-60
0-6
0-50
0-60
1-10
18-63
14-88
8-63
0-5
0-36
0-20
0-16
0-2
0-13
0-1
0-1
0-3
0^3
0-41
0-41
0
0
0
0
4 (n=9)
10-24
109-534
0-1
0-7
0-1
1-4
4-22
0-44
1-4
4-22
0-44
4-12
40-60
39-83
18-68
3-8
20^0
4-30
2-18
0-3
0-13
0-18
0-18
0-1
0-6
0-7
0-7
0-1
0^
0
0
5(n=7)
10-17
102-1483
0
0
0
0-1
0-10
0-4
0-1
0-10
0-4
3-8
29-55
13-93
7^8
3-7
19-70
1-43
1-19
0-5
0-36
0-85
0-56
0-1
0-9
0-2
0-2
0-1
0-6
0-3
0-3
76
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
BCG Rule Development
For the Upper Midwest, BCG quantitative rule development can be followed by comparing Table 12
(narrative rules), Table 13 (metric distributions), and Table 15 (quantitative rules). In Table 12, the
narrative rule for BCG level 2, macroinvertebrate taxa richness is: "Taxa richness is moderate to high"
(Table 12). In Table 13, total taxa in BCG level 2 sites ranged from 20 to 63 invertebrate taxa (Table 13),
so 20-63 is "moderate to high." The rule for total taxa (Figure 17, BCG level 2, Coolwater) was set at a
midpoint of > 20 taxa, with the fuzzy boundaries defined as 16 to 24. The fuzzy boundary of 16-24
defines the lower end of the "moderate" membership function for total taxa in Figure 17; membership
functions were assumed to be described by straight-line segments (Figure 17). For the total taxa rule, a
site with 20 invertebrate taxa would then have a membership of 50% in BCG level 2; a site with 16 taxa
would have a membership of 0 (zero), and a site with 18 taxa would have a membership of 25%. A site
with 24 or more taxa would have full (100%) membership in BCG level 2 for the total taxa rule. Note that
the total taxa rule is the same for BCG levels 2 and 3; these BCG levels cannot be distinguished based on
total taxa. Other rules must be used.
The panel's discrimination between levels 2 and 3 was primarily from richness and abundance of
sensitive taxa. Attribute II taxa were always present in BCG level 2, but they were allowed to be absent
in BCG level 3 (Table 13). The rules for level 2 required highly sensitive taxa (attribute II) to make up
more than 5% of taxon richness and 8% of the individuals, while in level 3 the attribute II taxa were only
required to be present (e.g., one taxon, one individual; Table 15). Similarly, total sensitive taxa (sum of
attributes II and III) were required to comprise 30% or more of both richness and abundance in BCG
level 2, but only 20% of richness, and 10% of abundance in BCG level 3. Here the panel also allowed an
exception or alternative in the rules: if sensitive attribute III individuals were particularly abundant
(> 40% of the community), then attribute II taxa were allowed to be absent (Alternate rule in Table 15).
The quantitative rules of Table 15 and Table 16 were developed in the same way: panel members
expressed why decisions were made, with statements of what they would require to rate a higher BCG
level, or what would be lost for them to rate the sample lower. These statements were later compared
to the distributions of the metrics in the panel's assessed sites to yield first-iteration quantitative rules
and model. The panel would then review the quantitative rules and their assessments and make
adjustments to the rules (or assessments) as needed. The final quantitative rules typically emerge after
two or three iterations.
Decision rules follow the patterns observed in the distributions of the metrics among BCG levels
assigned by the panel. BCG level 2 requires a strong presence of sensitive (attribute II and III) taxa and,
for invertebrates, sensitive EPTtaxa. Other level 2 rules include minimum numbers of total taxa for
invertebrates, maximum number of total taxa for fish, and low dominance of tolerant taxa in both
assemblages. It is important here to emphasize that whenever absolute values are used, the sampling
effort should be specified.
BCG level 3 decision rules allow slight reductions in sensitive taxa and individuals and increases in
tolerant taxa. Total number of taxa requirements are the same as BCG level 2. Since metrics do not
decline in lockstep with each other, the panels occasionally allowed alternative rules where an
exceptionally good value in one metric could be balanced by a poor value of another. Typically, these
were tradeoffs of number of sensitive taxa for number of sensitive individuals. For example, in the
invertebrate rules (Table 15), the percent sensitive (attributes I, II, and III) taxa and individuals—were
subject to alternate rules: If the value of the percent sensitive taxa metric is > 20%, then the percent
77
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
sensitive individuals must be > 10%. Alternatively, if the value of the percent sensitive taxa metric is
> 40%, then the percent sensitive individuals metric need only be > 5%.
BCG level 4 is characterized by decreased richness and abundance of sensitive taxa. However, sensitive
taxa must still be present above a minimum floor. The disappearance of sensitive taxa is what typically
discriminates level 5 from level 4, as well as an increase in the percent tolerant (attribute V) individuals
(Table 12, Table 16, Table 17).
Table 15. Benthic macroinvertebrate taxa: Decision rules for macroinvertebrate assemblages in
coldwater and coolwater (transitional cold-cool) streams; samples with > 200 organisms. Rules show
the midpoints of fuzzy decision levels, followed by the range of the membership function. The
midpoint is where membership in the given BCG level is 50% for that metric.
BCG
Level
2
3
4
5
Metrics
# Total taxa
% Most sensitive taxa (Att 1 + II)
% Most sensitive individuals (Att 1 & II)
% Sensitive taxa (Att II + 111)
% Sensitive individuals (Att II + III)
% Most dominant tolerant taxa (Att V)
% Sensitive EPT taxa (Att 1 + II + III)
# Total taxa
# Most sensitive (Att 1 + II) taxa
% Sensitive taxa (Att II + 111)
% Sensitive individuals (Att II + III)
% Most dominant intermediate tolerant
taxa (Att IV)
% Tolerant (Att V) individuals
% Most dominant tolerant taxa (Att V)
% Sensitive EPT taxa (Att 1 + II + III)
# Total taxa
% Sensitive taxa (Att II + 111)
% Sensitive individuals (Att II + III)
% Tolerant (Att V) individuals
Number of sensitive EPT taxa (Att 1 + II +
IN)
# Total taxa
% Tolerant (Att V) individuals
% Most dominant tolerant taxa (Att V)
Coldwater
Rule
> 14 (11-16)
> 10% (7%-13%)
—
> 30% (25%-35%)
> 30% (25%-35%)
< 5% (3%-7%)
> 10% (7%-13%)
Rule Alt Rule
> 14 (11-16)
Coolwater
Rule
> 20 (16-24)
> 5% (3%-7%)
> 8% (6%-10%)
> 30% (25%-35%
> 30% (25%-35%)
—
> 10% (7%-13%)
Rule Alt Rule
> 20 (16-24)
— present NA
,££» (35«S%, >«<»«»•
> 10% > 5%
(7%-13%) (3%-7%)
< 50% (45%-55%)
< 20% (15%-25%)
—
> 10% (7%-13%)
Rule
> 8 (6-10)
> 10% (7%-13%)
> 5% (3%-7%)
< 40% (35%-45%)
present
Rule
> 8 (6-10)
< 60% (55%-65%)
—
> 10% > 40%
(7%-13%) (35%-45%)
—
—
< 10% (7%-13%)
> 10% (7%-13%)
Rule
> 14 (11-16)
> 10% (7%-13%)
> 6% (4%-8%)
< 60% (55%-65%)
present
Rule
> 14 (11-16)
—
< 60% (55%-65%)
78
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 16. Fish taxa: Decision rules for fish assemblages in coldwater and coolwater (cold-cool
transitional) streams. Rules show the midpoints of fuzzy decision levels, where membership in the
given BCG level is 50% for that metric.
BCG
Level
1
Metrics
# Total taxa
% Most sensitive
taxa (Att II)
% Brook trout
individuals
% Sensitive taxa
(Att II + 111)
% Sensitive
individuals (Att II
+ 111)
% Tolerant (Att
V + Va + VIa)
individuals
% Non-native
salmonids (Att
VI)
Coldwater
Brook Trout (BT) Native BT Non-native
<4 (2-5)
Present
Present Absent
> 50% (45%-55%)
> 60% (55%-65%)
< 5% (3%-7%)
Absent
Coolwater
BT Native BT Non-native
Meets Coldwater level 1,
OR Coolwater rules below:
> 3 and < 14 (2-5 and 11-16)
Present
Present Absent
> 40% (35%-45%)
> 40% (35%-45%)
< 5% (3%-7%)
Absent
2
Metrics
# Total taxa
% Most sensitive
taxa (Att II)
% Brook trout
individuals
% Sensitive taxa
(Att II + 111)
% Sensitive
individuals (Att II
+ 111)
% Brook trout:
total salmonid
individuals
% Tolerant non-
salmonid (Att V
+ Va + Vla)
individuals
BT Native
Altl
Alt 2
BT Non-native
Altl
Alt 2
If watershed size < 10 mi2, < 8 (6-10)
If watershed size > 10 mi2, > 3 and < 14 (2-4 and 11-16)
Present
Present
>40%
(35%^5%)
>20%(15%-
25%)
NA
> 40% (35%^5%)
< 10% (7%-
13%)
Absent
NA
NA
> 20% (15%-25%)
> 70% (65%-
75%)
NA
NA
NA
< 10% (7%-
13%)
BT Native
BT Non-native
< 20 (16-24)
Present
Present
NA
NA
> 30% (35%-45%)
> 12% (9%-15%)
> 40% (35%^5%)
NA
< 20% (15%-25%)
79
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
BCG
Level
3
Metrics
# Total taxa
% Salmonid individuals
% Sensitive & non-native salmonid (Att 1 + II +
111+ VI) taxa
% Sensitive & non-native salmonid (Att 1 + II +
111+ VI) individuals
% Non-native salmonid (Att VI): total sensitive
(Att 1 + II + 111+ VI) individuals
% Sensitive taxa (Att II + 111)
% Sensitive individuals (Att II + III)
% Most dominant intermediate tolerant taxa
(Att IV)
% Extra tolerant individuals (Att Va + Via)
Coldwater
Rule Alt Rule
Coolwater
Rule Alt Rule
(brook trout native/non-native status not used)
If watershed size > 10 mi2, > 5
(3-7)
Present
> 25% (20%-30%)
> 20% (15%-25%)
< 70% (65%-75%)
-
-
-
-
< 20 (16-24)
-
-
-
-
>% Tolerant
(Att V + Va + NA
Via) taxa
>2*Tolerant(AttV
NA + Va + Via) %
individs
If watershed size > 10 mi2, < 40%
(35%-45%)
< 5% (3%-7%)
Metrics
% Sensitive & salmonid taxa (Att II + III + VI)
% Sensitive & salmonid individuals (Att II + III +
VI)
% Tolerant taxa (Att V + Va + Via)
% Extra tolerant individuals (Att Va + Via)
(no alternate rules)
> 5% (3%-7%)
> 5% (3%-7%)
< 45% (40%-50%)
< 10% (7%-13%)
> 5% (3%-7%)
> 5% (3%-7%)
-
< 20% (15%-25%
5
Metrics
# Total taxa
% Intermediate tolerant taxa (Att IV)
(no alternate rules)
> 2 (1-3)
> 10% (7%-13%)
> 3 (2^)
> 10% (7%-13%)
Model Performance
In general, the fuzzy model identified 75%-80% of samples as primarily a single BCG level (75%
membership or greater). Approximately 10%-15% of samples had a large minority membership in an
adjacent BCG level to the "nominal" level (25%-40% membership), and approximately 10%-15% of
assessments are ruled ties or near-ties between adjacent BCG levels (minority membership > 40%).
To measure model performance with the calibration data sets, two matches in BCG level choice were
considered: an exact match, where the BCG decision model's nominal level matched the panel's
majority choice; and a "minority match," where the model predicted a BCG level within one level of the
majority expert opinion. When model performance was evaluated in this calibration data set, the
coldwater macroinvertebrate model matched exactly with the regional biologists' BCG level assignments
on 97.6% of the coldwater samples (Table 17). In the single sample without agreement, the model
assignment was one level better than the majority expert opinion.
80
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
In order to confirm the model, panelists made BCG level assignments on additional samples. When
nominal level assignments from the BCG decision model were compared to the panelists' nominal level
assignments in the confirmation data set, the model matched exactly with the regional biologists' BCG
level assignments on 80% or more of the samples (Table 17). In both cold and coolwater, three
confirmation samples were rated differently by model and panel, where the model rated the samples as
being one BCG level better than the majority expert opinion. Based on the combined results, in 89% of
cases, the macroinvertebrate model predicts the same BCG level as the majority expert opinion.
Table 17. Benthic macroinvertebrate and fish taxa: Model performance—cold and coolwater samples
Model
Difference
2 better
1 better
same
1 worse
2 worse
Total #
Samples
% Correct
Benthic macroinvertebrates
Coldwater
Calib.
0
2
39
1
0
42
98%
Conf.
0
3
13
0
0
16
81%
Cool-transitional
Calib.
0
1
31
2
3
34
91%
Conf.
0
3
15
0
0
18
83%
Fish
Coldwater
Calib.
0
3
47
2
0
52
90.4%
Conf.
0
3
21
1
0
25
84%
Cool-transitional
Calib.
0
3
38
1
0
42
90%
Conf.
1
5
17
2
0
25
68%
4.1.2.4 Example #2: Quantitative Rules and Decision System for Diatoms, New Jersey
New Jersey DEP developed and calibrated a BCG model for sampled diatoms in northern New Jersey
streams (Gerritsen et al. 2014). The models were developed using data collected by the Academy of
Natural Sciences for New Jersey DEP. Workshop participants included scientists from around the United
States. The calibrated BCG models will allow New Jersey to express and assess goals for classes of water
bodies in terms of their biological condition.
Study sites
The data set consisted of 42 samples collected from streams and rivers in northern New Jersey. Sites
were located in the Northern Piedmont (25), the Northern Highlands (6), the Ridge and Valley (7), the
Atlantic Coastal Pine Barrens (3), and the Middle Atlantic Coastal Plain (1) ecoregions (Omernik 1987;
Woods et al. 2007). Land-use in the Piedmont is primarily urban and agriculture, whereas in the
Highlands and the Ridge and Valley it is predominantly forest and agriculture (USEPA 2000a). Within
ecoregions, the study sites had relatively similar natural environmental conditions (e.g., geology,
geomorphology), but with a wide range of nutrient concentrations.
A narrative description was derived from discussions with the panelists about why individual sites were
assessed at a certain level (Table 18). The rules were calibrated from the narrative description and the
30 calibration samples rated by the group, and the rules were adjusted so that the model would
replicate the panel's decisions as closely as possible. Panel members were highly quantitative in their
thinking and deliberations, and they developed the first iteration of quantitative rules based on the
narrative descriptions.
81
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Rule Development
Rules adopted for the quantitative decision model are listed in Table 19. BCG level 1 has five rules: one
on taxa richness, two rules on abundance of sensitive taxa, and two rules on abundance of tolerant taxa.
For BCG level 1, sensitive taxa are required to be dominant, and tolerant taxa are very minor
constituents of the community. The rules for BCG level 2 are similar to level 1, but all have been relaxed
to some extent. The largest relative difference between levels 1 and 2 is that attribute II individuals are
required to be highly abundant in level 1 (roughly 35% or more), but they are subdominant in level 2
(10% or more).
In BCG level 4, sensitive individuals are greatly diminished, but still present (9% or more), and tolerant
taxa can occur at higher abundances. There are only three rules for BCG level 5: tolerant taxa may not
exceed 40% of taxa or 80% of individuals. Samples that fail to meet the BCG level 5 requirements would
be assigned to BCG level 6, but no such samples were encountered in this data set.
Model Performance
To evaluate the performance of the 40-sample calibration data set and the 10-sample confirmation data
set, the number of samples where the BCG decision model's nominal level exactly matched the panel's
majority choice ("exact match"), and the number of samples where the model predicted a BCG level that
differed from the majority expert opinion ("anomalous" samples) were assessed. Then, for the
anomalous samples, the degree of differences among the BCG level assignments, and also whether
there was a bias was examined (e.g., did the BCG model consistently rate samples better or worse than
the panelists?).
Two types of ties were taken into account: (1) BCG model ties, where there is nearly equal membership
in two BCG levels (e.g., membership of 0.5 in BCG level 2 and membership of 0.5 in BCG level 3); and (2)
panelist ties, where the difference between counts of panelist primary and secondary calls is less than or
equal to 1 (e.g., 4-4 or 4-3 decisions). If the BCG model assigned a tie, and that tie did not match with
the panelist consensus, it was considered to be a difference of half a BCG level (e.g., if the BCG model
assignment was a BCG level 2/3 tie and panelist consensus was a BCG level 2, the model was considered
to be "off" by a half BCG level; or more specifically, the model rating was a half BCG level worse than the
panelists' consensus). The BCG model was also considered to differ by a half level if the panelists
assigned a tie and the BCG model did not.
Results show that the diatom BCG model performed well (Table 20). The models assigned scores that
are within a half BCG level or better on 100% of the samples in both the calibration and confirmation
data sets (Table 18). When half levels were considered, the BCG model rated three of the calibration
samples a half level worse than the panelists, and five confirmation samples (two better, three worse).
Based on results from the calibration data set, the model has a slight bias towards rating samples a half
level worse than the panel consensus.
82
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 18. Narrative description of diatom assemblages in six BCG levels for streams of northern New
Jersey. Definitions are modified after Davies and Jackson (2006).
Definition: Natural or native condition— native structural, functional, and taxonomic integrity is preserved;
ecosystem function is preserved within the range of natural variability
BCG
level 1
Narrative: BCG level 1 streams in northern New Jersey highlands are oligotrophic, with a mature forested
watershed. Unlike macroinvertebrates, the diatom community is relatively depauperate, with typically 15-20 taxa
in a 500-count sample. The top dominant taxa are extreme low-nutrient adapted taxa of attributes II and III (e.g.,
Achnanthes subhudsonis orAchnanthidium rivulare). Subdominants (up to 10% abundance) may include attribute IV
taxa. Tolerant taxa (attribute V) make up a very small fraction of the community.
BCG
level 2
Definition: Minimal changes in structure of the biotic community and minimal changes in ecosystem function—
virtually all native taxa are maintained with some changes in biomass and/or abundance; ecosystem functions are
fully maintained within the range of natural variability
Narrative: BCG level 2 streams are very similar to level 1, however, a slight increase in disturbance or enrichment
has allowed more diatom taxa to colonize (20^40 total taxa). Richness is slightly higher than level 1, but low
nutrient taxa (attribute II and III) are dominant. There may be several tolerant taxa, but their abundance is low.
BCG
level 3
Definition: Evident changes in structure of the biotic community and minimal changes in ecosystem function-
Some changes in structure due to loss of some rare native taxa; shifts in relative abundance of taxa but intermediate
sensitive taxa are common and abundant; ecosystem functions are fully maintained through redundant attributes of
the system
Narrative: Richness is higher than level 2 (> 30 taxa). Dominant taxon may or may not be sensitive (attribute II or
III). Tolerant taxa have increased to more than 10% of the assemblage, and some of the tolerant taxa are now in the
subdominant category.
BCG
level 4
Definition: Moderate changes in structure of the biotic community and minimal changes in ecosystem function—
Moderate changes in structure due to replacement of some intermediate sensitive taxa by more tolerant taxa, but
reproducing populations of some sensitive taxa are maintained; overall balanced distribution of all expected major
groups; ecosystem functions largely maintained through redundant attributes
Narrative: BCG level 4 sites tend to have the highest taxa richness as more diatom niches open up with increased
enrichment, light penetration (from canopy loss), and moderate sedimentation. Sensitive species and individuals
are still present but in reduced numbers. The persistence of some sensitive species indicates that the original
ecosystem function is still maintained albeit at a reduced level. Intermediate and tolerant taxa may be dominant,
sensitive taxa are often still subdominant.
BCG
level 5
Definition: Major changes in structure of the biotic community and moderate changes in ecosystem function —
Sensitive taxa are markedly diminished; conspicuously unbalanced distribution of major groups from that expected;
organism condition shows signs of physiological stress; system function shows reduced complexity and redundancy;
increased build-up or export of unused materials
Narrative: Overall diversity is still high, but may be slightly reduced from level 4. Sensitive species may be present
but their functional role is negligible within the system. The most abundant and dominant taxa are tolerant or have
intermediate tolerance, and there may be relatively high diversity within the tolerant organisms.
BCG
level 6
Definition: Major changes in structure of the biotic community and moderate changes in ecosystem function —
Sensitive taxa are markedly diminished; conspicuously unbalanced distribution of major groups from that expected;
organism condition shows signs of physiological stress; system function shows reduced complexity and redundancy;
increased build-up or export of unused materials
Narrative: Heavily degraded from urbanization and/or industrialization. No level 6 samples were encountered by
the panel.
83
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 19. BCG quantitative decision rules for diatom assemblages in northern New Jersey streams. The
numbers in parentheses represent the lower and upper bounds of the fuzzy sets. BCG level 6 is not
shown, because there are no specific rules for level 6: If a site fails level 5, it falls to level 6. Shaded
rules under BCG level 3 are alternate rules, that is, at least one must be true for a site sample to meet
BCG level 3.
Attribute metric
Threshold
BCG Level 1
# Total taxa
% Attribute ll+lll individuals
% Attribute II individuals > % Attribute III individuals; expressed as (% Att ll-% Att III)
% Attribute V+VI individuals
% Most dominant Attribute V or VI taxon
< 20 (15-25)
> 65% (60%-70%)
> 0% (-10% to 10%)
< 2.5% (l%-4%)
< 1% (0%-2%)
BCG Level 2
# Total taxa
% Attribute II individuals
% Attribute ll+lll individuals
% Attribute ll+lll taxa
% Attribute V+VI individuals
% Most dominant Attribute V or VI taxon
< 40 (35-45)
> 10% (5%-15%)
> 50% (45%-55%)
> 15% (10%-20%)
< 10% (5%-15%)
< 5% (3%-7%)
BCG Level 3
# Attribute ll+lll taxa
# Attribute II taxa
Most dominant taxon*
Alt 1:% Attribute ll+lll taxa
Alt II: % Attribute ll+lll individuals
% Attribute V+VI individuals
% Most dominant Attribute V or VI taxon
> 5 (2-8)
> 1 (0-1)
Att II or 3
> 15% (10%-20%)
> 15% (10%-20%)
< 30% (25%-35%)
< 10% (5%-15%)
BCG Level 4
% Attribute ll+lll individuals
% Attribute V+VI individuals
% Most dominant Attribute V or VI taxon
> 9% (5%-13%)
< 65% (60%-70%)
< 40% (35%-45%)
BCG Level 5
% Attribute V+VI taxa
% Attribute V+VI individuals
< 40% (35%-45%)
< 80% (75%-85%)
: Dominant taxon must be sensitive (Att II or III); membership = 0 if rule fails
84
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 20. Model performance for calibration and confirmation samples. "% better" indicates models
scored the sample % BCG level higher than the panel; e.g., Panel score was 4 and model score was 3-4
tie. Half-level mismatches are counted half the value of full matches. No mismatches exceeded % BCG
level.
Difference (model vs. panel
consensus call)
model 1 level better
model K level better
exact match
model 1/2 level worse
model 1 level worse
Total # Samples
Calibration
Number
0
0
27
3
0
30
Percent
0
0
90
10
0
95
Confirmation
Number
0
2
7
3
0
12
Percent
0
17
58
25
0
79
4.2 Calibrating Indices to the Biological Condition Gradient
Most states have developed biological indices for their streams and wadeable rivers (USEPA 2002). In
the initial development of BCGs, common questions asked by states included:
• What is the relationship between the BCG and the state's existing biological index, or indices?
• Does the BCG replace the existing biological index, or indices?
• How can the BCG and the existing biological index, or indices, be used together to better assess
ALUs?
The linkage between a biological index and the BCG could be addressed in a state program review
(USEPA 2013a) and/or as a topic of discussion within the expert panel. Existing indices could be
evaluated for how extensively they include attributes of the BCG or how the BCG decision criteria match
up with the metrics that comprise the index. If needed, recommendations for specific technical
improvements and analyses can then be made to guide the redevelopment of an index and/or refine the
BCG model.
As in section 4.1., the objective is to calibrate a BCG model with a quantitative model, or in this case, an
index that will duplicate the expert panel BCG assessments for new samples and water bodies, without
having to reconvene the panel. In this approach, a conventional IBI (e.g., Karr 1986) or predictive
biological index model (e.g., Hawkins et al. 2000b; Wright 2000) could be calibrated to the expert-
derived BCG. While the seminal works about these indices preceded the BCG, they are based on parallel
ecological concepts, and to varying degrees each incorporates BCG attributes. As an example of this,
Table 21 illustrates the overlap between the 10 BCG attributes and a selection offish and
macroinvertebrate indices for freshwater streams and wadeable rivers. For the fish indices, the metrics
used for each capture the more commonly measured attributes I-VI (taxa composition and effects of
non-native taxa), but they also address attributes VII (organism condition), VIII (ecosystem function),
and X (ecosystem connectance). The routine inclusion of the deformities, erosions, lesions, and tumors
(DELT) anomalies metric (e.g., measure of deformities, erosion, lesions, and tumors) in all fish indices
contains attribute VI. Functional feeding and reproduction guilds that are routinely included in fish
indices might provide a surrogate for attribute VIII. The inclusion of diadromous metrics provides for the
direct inclusion of species that depend on access to and from coastal rivers for completing their life
85
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
cycles. Other metrics that include species that are dependent on free access to a drainage network can
illustrate the concept of connectivity in inland streams and rivers. Attribute IX (spatial and temporal
extent of detrimental effects) can be accounted for by the spatial extent of the sampling design and is
independent of the composition offish IBIs. For the macroinvertebrate metrics in Table 21, coverage of
attributes I-V is provided by most biological indices used by states. It is also possible to develop non-
native taxa metrics for attribute VI (presence and effect of non-native taxa) and metrics for attribute X
(ecosystem connectance). Biological metrics could serve as a surrogate for attribute X—Unionid mussels
might be a good choice given their dependency on fish hosts for dispersal and to sustain their
populations. The key point is that (MMIs) have been developed from the same or parallel concepts as
the BCG.
Table 21. Cross referencing the 10 BCG attributes with selected fish IBI and macroinvertebrate MMI
metrics for streams and wadeable rivers
BCG Attribute
1. Historically documented,
sensitive, long-lived, or
regionally endemic taxa
II. Highly sensitive taxa
III. Intermediate sensitive taxa
IV. Intermediate tolerant taxa
V. Tolerant taxa
VI. Non-native or intentionally
introduced species
VII. Organism condition
VIII. Ecosystem function
IX. Spatial and temporal extent
of detrimental effects
X. Ecosystem connectance
Fish IBI Metrics
Great River species
Sensitive sucker species
Native salmonid species
American eel numbers & size classes
Selected diadromous species
Highly intolerant species
Sensitive species
Temperate stenotherms
Native salmonids
Moderately intolerant species sensitive species
Round-bodied suckers
Included in native species richness
Number of minnow species
Number of sunfish species
Highly tolerant species
Exotic and introduced species of intracontinental origin
Non-native species
DELT anomalies
Total native species biomass
Proportion in functional feeding groups
Specialist metrics, i.e., fluvial specialists & dependents
Accounted for in spatial sampling design
Diadromous species
Native Salmonids
Non-indigenous species
Macroinvertebrate Metrics
Unionid mussels
# of Pteronarcys species
Mayfly & EPT metrics
Mayfly, caddisfly, Tanytarsini
midge, EPT metrics
Taxa richness, caddisfly, Dipteran
taxa, Non-insect & Other Dipteran
taxa
Tolerant taxa
% Abundance tolerant Taxa
%Corbicula; Dreissenid mussels
Head capsule deformities
%0ther Dipteran & non-insects
%Filterers
%Grazers/scrapers
%Clingers
(Same as fish)
Unionid mussel
86
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Indices that are currently in widespread use are of two basic types:
• Indices comprised of metrics that are the aggregation of species/taxa abundance data based on
taxonomy, environmental tolerance, functional role, assemblage condition, and organism
condition. Each metric is calibrated on a range from best to poorest conditions and also with
respect to natural factors such as watershed size. The index development process usually
includes an examination of tens to hundreds of candidate metrics and reducing this list to the
most relevant and/or responsive 8-12 metrics (approximately). The metrics can be somewhat
independent in response to each other and, when summed together, can either dilute or
amplify an interpretation. They are useful in observing trajectory, but they may require
recalibration to the BCG attributes before they can produce a BCG assessment. Most of this
class of indices have been developed for fish, macroinvertebrates, and algae although
development for other groups such as Unionid mussels have been attempted (Barbour et al.
1999). Within this broad class of indices are the classic IBIs that follow the seminal guidance of
Karr et al. (1986), most of which have been developed for fish assemblages, but some for
macroinvertebrates. While the original IBI was developed for central Illinois fish assemblages,
Karr et al. (1986) provided guidelines about the possible application to other regions and other
aquatic assemblages. This was done knowing that different metrics would be needed, but the
goal was to maintain the essential attributes and ecological content of an IBI. Other multimetric
approaches have been developed and applied for macroinvertebrates that, while utilizing a
generally similar process, are somewhat distinctive from IBIs in having metrics that are
predominantly based on taxa attributes (Plafkin et al. 1989; Barbour et al. 1999).
• Predictive models, where the observed species composition at a site is compared to an idealized
reference site predicted from a multivariate statistical model. These models develop an
expected taxon list and use the O/E ratio (e.g., the River Invertebrate Prediction and
Classification System, RIVPACS, e.g., Wright (2000) and the Australian RIVer Assessment System,
AUSRIVAS, Simpson and Norris (2000)). A second approach has been to use a multivariate
similarity index between a specific sample and a centroid defined by undisturbed reference sites
(e.g., Percent Model Affinity, Novak and Bode 1992; BEAST, Reynoldson et al. 1995; dissimilarity,
Van Sickle 2008). Predictive approaches have also been applied in a multimetric framework, in
which expectations for the metrics are based on environmental variables (Chen et al. 2014;
Esselman et al. 2013; Moya et al. 2011; Oberdorff et al. 2002; Pont et al. 2006; Pont et al. 2009).
Ideally, the calibration of MMIs are based on minimally disturbed reference sites and with respect to
natural classification strata such as bioregions, thermal gradients, and other factors that determine the
baseline expectations of a regional aquatic fauna (Stoddard et al. 2006). Some have used all the data
assuming that the best, or least disturbed, sites reflect the highest possible condition (Blocksom 2003;
Stoddard et al. 2006). Such an assumption should be evaluated by expert opinion before it is accepted that
the best condition found in a data set reasonably represents the highest expected condition. Calibration
techniques have also evolved from the ordinal approach of Fausch et al. (1984) to continuous calibration
techniques (Blocksom 2003; Mebane et al. 2003) that could be applied to BCG development. The
expectations for achieving a high level of rigor in this process are described in EPA's Biological
Assessment Program Review document (USEPA 2013a). As such, the level of technical rigor achieved in
these important calibration steps can also affect the ability to measure condition along the BCG.
As with the development of the BCG, it is also necessary to test an index or model across a gradient of
different environmental stressors. The ability to quantify departures from reference-derived thresholds
is an important step in evaluating any assessment model.
87
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
4.2.1 Biological Condition Gradient Thresholds for Multimetric Indices and
Multivariate Models
Indices and models as generally described herein should accurately translate to a position along the
BCG. However, the proficiency of a particular index or model to actually accomplish this, at a particular
level of resolution, is dependent on the level of detail and rigor applied in construction of the index or
model and the calibrated BCG model. EPA (2013a) provides a standardized way to evaluate the technical
strengths and gaps in a biological assessment program and to determine how well a particular biological
assessment protocol discriminates incremental changes in biological condition (i.e., the higher the level
of rigor, the more precision is achieved in incremental measurement along a gradient of stress).
However, simply stratifying an index scoring range along the BCG is neither sufficient nor recommended,
especially if an index has not been explicitly developed within the conceptual framework of the BCG or
the BCG attributes have not be reconciled with the metrics that comprise the index. For example,
metrics in a MMI may have been selected because of strong known response to current or selected
stressors and may not comprehensively characterize the full range of biological conditions, while the
BCG decision rules are based on benchmarks for undisturbed or minimally disturbed conditions. This has
been a challenge, especially with the upper BCG levels where reference analogs to BCG levels 1, 2, 3, or
sometimes even 4 either do not exist or have not been identified. If this is the case, it will be necessary
to revisit the existing index derivation and BCG model calibration and possibly revise either one, or both,
for better correspondence. This task can be accomplished by the state biological assessment and criteria
program, but it should be done in collaboration with the full expert panel that developed the BCG model
and the underlying quantitative decision rules. As described in Chapter 3, through an iterative process,
scoring criteria can be developed for new or refined indices that correspond with biologists' consensus
about narrative descriptions of the levels in the BCG.
4.2.1.1 Calibrating Index Scores: Connecticut Stream Example
The set of sites that have been assigned to levels of the BCG are used to calibrate index scores. Index
scores for the sites are examined, and, if separation of the index scores among levels is good, then index
thresholds can be selected to maximize the ability to discriminate among the levels. This is
demonstrated in the Connecticut case example below and by the Minnesota case study where IBI
thresholds for refined ALUs were based on the correspondence between their IBIs and BCG levels
(section 6.4). In the Connecticut example, BCG calibration and a macroinvertebrate MMI were
developed at the same time. The MMI consisted of seven metrics (Table 22; Gerritsen and Jessup
2007b), including an abundance-weighted average of BCG attributes II through VI.
Table 22. Correlations (Pearson r) among Connecticut MMI index metrics
#
1
2
3
4
5
6
7
Metric
Ephemeroptera taxa (adj.)
Plecoptera taxa
Trichoptera taxa
% sensitive EPT (adj.)
Scraper taxa
BCG Taxa Biotic Index
% dominant genus
1
•
0.58
0.57
0.69
0.67
-0.76
-0.61
2
•
0.50
0.54
0.50
-0.76
-0.54
3
•
0.52
0.75
-0.68
-0.62
4
•
0.52
-0.74
-0.59
5
•
-0.69
-0.60
6
•
0.66
7
•
Note: Adj. = Metric scoring was adjusted for catchment size.
88
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
The Connecticut stream MMI uses metrics that are similar in objective to the BCG attributes, but which
are calculated somewhat differently (e.g., EPT taxa metrics in the MMI include taxa considered to be
attributes II, III, IV; and attribute II includes taxa from the EPT orders, as well as a few dipteran and
beetle taxa). The total MMI score is based on the average of all metrics, while BCG decisions are based
on decision-specific critical attributes (e.g., attributes II and III for the higher levels and attribute V for
lower levels). Concordance of the two assessment endpoints is strong (Figure 20). Figure 20 shows the
predicted results of the BCG inference model.
1 UU
on
yu
pn
ou
Xyn
/ u
CD
~a
c. fin
_ DU
O
±= 5D
QJ
-i in
^j 41 1
"5
^ °n
OU
9D
1 n
n
T
i
n
~-
N T
— *— n
-L
n
I i q i
32 89 41 74 2
345
BCG Level (nominal)
a Median
D 25%-75%
I Non-Outlier Range
° Outliers
:: Extremes
Figure 20. Connecticut MMI by BCG levels, estimated from decision analysis model. Number of samples given
below boxes.
In spite of these differences, MMI scores could be used to separate levels (Figure 20). Potential MMI
scoring thresholds are given in Table 23.
Table 23. Scoring thresholds for the Connecticut MMI to correspond to BCG levels
BCG Level
Levels 1, 2
Level 3
Level 4
Level 5
Level 6
MMI Scoring Range
>75
60-74.9
43-59.9
2CM12.9
>20
89
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
The BCG decision model and the MMI were in overall concordance on the assessments from the two
methods. The scoring range of the MMI was broken into categories corresponding to BCG levels. This
resulted in disagreement of 32% of multimetric scores compared to the BCG decision model, but
disagreements were never by more than a single level. There was no bias in the direction of
disagreement among models, determined by the similar number of MMI assessments that were better
or worse than the corresponding BCG assessments.
An additional example of an approach to reconcile an existing index to the BCG is included in Appendix
B. This example involves an innovative technique to "back calculate" a historically representative IBI
(Appendix Bl). In this case it helped to clarify the position of an IBI based on current-day stressors for
the Upper Mississippi River.
4.3 Statistical Models to Predict Expert Decisions: Multivariate Discriminant
Model Approach
Another approach to quantify expert consensus and develop a BCG model is use of multivariate
statistical models to predict expert judgment. For example, Maine DEP developed a set of multivariate
linear discriminant models to simulate the expert consensus and predict a site assessment (Danielson et
al. 2012; Davies et al. In press), and the United Kingdom Environmental Agency defined ranges of scores
of two indices (their RIVPACS index and a tolerance index) that correspond to expert consensus
(Hemsley-Flint 2000). Both of these approaches utilize one or more multivariate statistical models to
predict the expert judgment in assessments. The following section describes Maine's use of linear
discriminant models to discern levels of biological condition.
4.3.1 Approach
The objective of the discriminant model approach is the same as that of the quantitative rule
development approach described in section 4.1: to develop a predictive model that will duplicate the
decisions of the expert panel, so that new water bodies can be assessed without having to reconvene
the panel. As with the rule development, the discriminant model (a multivariate statistical model) uses
the same data available to the expert panel.
Discriminant analysis can be used to develop a model that will divide, or discriminate, observations
among two or more groups whose membership characteristics have been defined a priori (i.e., in
advance) of the construction of the model. This is accomplished through use of a model-building or
"learning" data set in which samples have been assigned into the groups of interest, for example by
expert consensus much like the expert panel process discussed in section 4.1.1. In short, for purposes of
calibrating a BCG model, a discriminant function model can be developed from a biological data set
where sites in a training data set have previously been assigned to BCG levels. A discriminant function
model is a linear function combining those input variables that most successfully contribute to group
definition and discrimination among groups. The resulting model yields the maximum separation
(discrimination) among the groups (e.g., levels of the BCG). The analysis objectively identifies the best
discriminatory variables and weights their relative contribution to the discriminatory model using
coefficients. Selection of input variables is aided by initial exploratory data analysis to investigate
relationships between biological response variables and physical stream characteristics (width, depth,
velocity, elevation, temperature), and by data reduction techniques to eliminate highly correlated
variables.
90
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
The linear discriminant model (LDM) approach may reveal subtle discriminatory variables within the
data set that the biologists might not have recognized as important. This feature of statistical selection
of variables contributes to building a highly discriminatory model. In construction of an LDM, input
variables can also be included in the model on the basis of the judgment of experts that the variable
provides an important link to assessment of the specific biological values that are stated in narrative
biological criteria. Once constructed, the model can be used to objectively and consistently determine
membership in a BCG level for new observations where the level is unknown. Maine uses this method to
determine whether streams are meeting biological criteria for the state's tiered ALUs.
Although it requires statistical expertise to develop, another advantage of discriminant analysis is that it
uses established and well-documented statistical methodology, with known confidence limits, and it
reports group membership of a sample as probability statements, providing an understanding of the
degree of certainty of the reported result. While LDMs require a relatively large set of assigned sites to
calibrate the model (approximately 20 per group due to dependence upon having a suitable number of
degrees of freedom, Manly 1991; Wilkinson 1989), accuracy of the model to the expert-assigned
calibration and test sites can be as high as 89%-97%6 (Davies et al. In press; Shelton and Blocksom
2004).
Using a discriminant model to develop biological criteria requires both a set of model-building data to
develop the model and confirmation data to test the model. If a sufficient number of samples are
available, the training and confirmation data may be from the same biological database, randomly
divided into two sets (60% to 70% of data for calibration), or they may be drawn from two or more years
of survey data. All sites in each data set are assigned to BCG levels by the expert workgroup.
Depending upon the required precision of the model, one or more discriminant function models that
function in a hierarchical fashion may be developed from the model-building set to predict level
membership from biological data. Building a set of nested, hierarchical models is an effective way of
improving overall predictive accuracy (Davies et al. In press). Once developed, the model is applied to
the confirmation data set to determine how well it can assign sites to levels using independent data not
used to develop the model. More information on discriminant analysis can be found in many available
textbooks on multivariate statistics (e.g., Jongman et al. 1987; Legendre and Legendre 1998; Ludwig and
Reynolds 1998; Rencher 2003).
4.3.1.1 Example—Maine Discriminant Model for Benthic Macroinvertebrate Assemblages (Source:
Shelton and Blocksom 2004)
Maine has four designated use classifications for its streams, AA, A, B, and C, with three corresponding
ALUs. Classes AA (Maine's outstanding natural resource waters) and A correspond to BCG levels 1 and 2
(per Maine's narrative criteria, "as naturally occurs"), and they are not distinguishable based on Maine's
biological assessment method. Class B ("no detrimental change") corresponds approximately to BCG
level 3, and Class C ("maintain structure and function") corresponds approximately to BCG level 4.7
Streams in poorer condition than Class C, comprising BCG levels 5 and 6, are not in attainment (NA) of
minimum state ALU standards. Section 6.5 provides details of implementation and application of
6 Based on jack-knife tests of the combined nested LDMs in Maine's two-stage hierarchy of LDM analysis. Results
for a new test data set, not used to build the model were 75%-100% accuracy (Davies et al. In press).
7 The percentage of river and stream miles assigned to each ALU classification in Maine is: Class AA/A-49%; Class
B-51%; Class C- 0.4%.
91
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Maine's biological criteria models. After testing multiple statistical modeling techniques (e.g., k-means
clustering, Two-Way Indicator Species Analysis, multivariate ordination), the use of best professional
judgment of expert aquatic biologists and construction of a set of hierarchical linear discriminant models
was selected as the most promising approach to accomplish both technical and regulatory policy goals.
Maine's tiered ALUs and calibration process for benthic macroinvertebrate samples utilizing professional
judgment actually predated the formalization of the BCG, and development of the BCG was in fact
based, in part, on Maine's approach to biological assessment and biological criteria (Davies and Jackson
2006). The calibration approach in Maine was similar to that described in section 4.1, except that
professional judgment was used to place streams into Maine's designated ALL) classes (Class A, Class B,
Class C) instead of into BCG levels. Maine's tiered ALUs provide an ecologically descriptive gradient of
condition tiers, with detailed definitions, to express the expected goal condition for each class. These
clearly articulated goals provided the "guiding image" (Poikane et al. 2014; Willby 2011) for biologists to
assign samples to classes. Maine DEP developed a set of multivariate linear discriminant models to
predict the expert site assessment (Davies et al. 1995; Shelton and Blocksom 2004; State of Maine 2003;
Davies et al. In press). The description of the model-building data set below is modified from Shelton
and Blocksom (2004):
The MEDEP [MDEP] originally developed the linear discriminant models based on 145 rock basket samples
collected from across the state and representing a range of water quality during 1983-1989. They
recalibrated the models in 1998 using a much larger macroinvertebrate database with a total of 376
sampling events (Davies et al. 1999). The final step involved assigning each of the 376 sites in the
database to one of four o priori groups using the quantifiable measures.
MEDEP also conducts biological assessments of stream algal, wetland macroinvertebrate, and wetland
phytoplankton and epiphytic algal assemblages (Danielson et al. 2011, 2012). MEDEP used Maine's
narrative biological criteria and the BCG as the foundation of biological assessment models for stream
algae, also using the LDM approach outlined here (Danielson et al. 2012). A first step in model-building
was to empirically compute tolerance values for algal and macroinvertebrate species that had been
collected in Maine's monitoring program. After computing tolerance values, the species were grouped
into the BCG framework's sensitive, intermediate, and tolerant attribute groups. MEDEP then modified
the model BCG framework for stream macroinvertebrates for stream algae and wetland
macroinvertebrates, describing how those assemblages empirically respond to anthropogenic stressor
gradients. MEDEP used those modified BCG frameworks and tolerance metrics along with the narrative
biological criteria and other metrics to build predictive biological assessment models for the additional
assemblages. MEDEP has completed LDM statistical models to predict ALL) attainment for both stream
algal and wetland macroinvertebrate community data. These models currently are used to help
interpret narrative biological criteria. Following adequate testing and standard public review protocols,
MEDEP will amend the Maine Biological Criteria Rule8 to include the stream algal and wetland
macroinvertebrate models as numeric biological criteria.
8 See Code of Maine Rules, MEDEP, Chapter 579, http://www.maine.gov/dep/water/rules/index.html. Accessed
February 2016.
92
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
To define a priori groups for stream macroinvertebrates, biologists were given data from a set of sites
and asked to place the sites into Maine's use classes based on the biological data only (Willby 2011).
This set of sites was then used as the calibration data (or "learning" data) for an LDM. The objective of
the discriminant model is to replicate ("predict") the professional judgment of the panel of biologists.
The excerpt below describes how MEDEP biologists assigned calibration sites to Maine's three classes
and to NA (from Davies et al. In press):
Maine's statutory classes are goal-based and thus do not necessarily correspond to actual biological
condition of streams in Maine so legislatively assigned classes could not be used to define groups ... As an
alternative approach to defining stream classes, we used "expert knowledge/prior experience" to identify
response signals (to different levels of human disturbance) for 30 quantifiable measures of
macroinvertebrate community structure (Table 24 below). This classification process was then followed
by validation using objective methods to confirm that the o priori groupings were, in fact, statistically
distinguishable. This approach has been well developed (Anderson 1984; Press 1980). Discriminant
analysis and function derivation does not have to rely on classes that only occur in nature. As long as
classes are statistically distinct and their members possess a Gaussian distribution within a class, then
most assumptions are met (Anderson 1984). To establish o priori groups, MDEP biologists, along with
independent biologists from other states, and the private stakeholder sector, evaluated benthic
macroinvertebrate community data for each stream sample (without knowing site locations or pollution
influences) and assigned samples to an aquatic life condition category. The methodology was based on
the degree to which each biologist found the sampled community conformed to one of the narrative
aquatic life criteria (Class AA/A, B, C; or NA if the community assemblage did not conform to the narrative
criteria of the lowest class) as described in the statute and accompanying definitions (Shelton and
Blocksom 2004). The panel of biologists received limited habitat data (e.g., depth, water velocity,
substrate composition, temperature) in order to evaluate the intrinsic biotic potential of the sampled
habitat, but biologists had no knowledge of the site locations, or degree of human disturbance.
Biologist's Classification Criteria
Each biologist reviewed the sample data for the values of a list of measures of community structure and
function. Criteria used by biologists to evaluate each measure are listed in Table 24. In 64% of the cases,
there was unanimous agreement among the independent raters, and in an additional 34% of the
samples, two of the raters were in agreement and one had assigned a different classification. In three of
the rated samples, there was disagreement among all three raters (2%).
Table 24. Maine Biologists' Relative Findings Chart Using Macroinvertebrates (Source: Davies et al. In
press)
Measure of Community
Structure
Total Abundance of
Individuals
Abundance of
Ephemeroptera
Abundance of Plecoptera
Proportion of
Ephemeroptera
Proportion of
Hydropsychidae
Proportion of Plecoptera
Proportion of Glossoma
Relative Findings by Water Body Class
A
often low
high
highest
highest
intermediate
highest
highest
B
often high
high
some present
variable, depending
on dominance by
other groups
highest
variable
low to intermediate
C
variable to high
low
low to absent
low
variable
low
very low to absent
NA
variable: often very
low or high
low to absent
absent
zero
low to high
zero
absent
93
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Measure of Community
Structure
Proportion of
Brachycentrus
Proportion of Oligochaetes
Proportion of Hirudinea
Proportion of Gastropoda
Proportion of
Chironomidae
Proportion of
Conchapelopia &
Thienemannimyia
Proportion of Tribelos
Proporation of Chironomus
Genus Richness
Ephemeroptera Richness
Plecoptera Richness
EPT Richness
Proportion Ephemeroptera
Richness
Proportion Plecoptera
Richness
Proporation Diptera
Richness
Proporation
Ephemeroptera &
Plecoptera Richness
EPT Richness divided by
Diptera Richness
Proporation Non-EPT or
Chronomid Richness
Percent Predators
Percent Collectors,
Filterers, & Gatherers
divided by Percent
Predators & Shredders
Number of Functional
Feeding Groups
Represented
Shannon-Weiner Generic
Diversity
Hilsenhoff Biotic Index
Relative Findings by Water Body Class
A
highest
low
low
low
lowest
lowest
low to absent
low to absent
variable
highest
highest
high
highest
highest
low to variable
highest
high
lowest
low
high
variable
low to intermediate
lowest
B
low to intermediate
low
variable
low
variable, depending
on the dominance of
other groups
low to variable
low to absent
low to absent
highest
high
variable
highest
high
high
variable
high
highest
low
low
highest
highest
highest
low
C
very low to absent
low to moderate
variable
variable
highest
variable
low to variable
low to variable
variable
low
low to absent
variable
low
low
highest
low to variable
low to variable
intermediate to high
high to variable
low
variable
variable to
intermediate
intermediate
NA
absent
highest
variable to highest
variable to highest
variable
variable to highest
variable to highest
variable to highest
lowest
very low to absent
absent
low
zero
low to zero
variable to high
low to absent
lowest to zero
highest
high to variable
lowest
lowest
lowest
highest
94
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Once these groups were determined subjectively and independently by three biologists, univariate and
multivariate analysis of variance (ANOVA and MANOVA, respectively) confirmed that the assigned
groups were in fact statistically distinct. Following establishment and statistical validation of the groups,
MEDEP applied additional analyses to evaluate the necessity to develop stratified models to account for
natural factors, such as geographic location and stream size. The uni- and multivariate analyses (cluster
analysis, multidimensional scaling, and principle components analysis, in part) suggested that a
physically or geographically stratified model for Maine was not warranted. To determine variability in
expert judgment assignments, a new test data set was assigned to a priori groups by two non-MEDEP
biologists, yielding an average concurrence with MEDEP biologists' assignments of 80%. Furthermore, as
a check against potential circularity in the model (i.e., "this site looks good, so this must be what good
sites look like"), MEDEP chose 27 minimally disturbed sites based on non-biological criteria. These sites
were not originally used in the expert assessment or to build the model. This reference data set was
used to determine the success of the model to assign them to Class A conditions. These sites had no
known point sources and land uses were characterized as 97% forested (3% logged); 2% crop; and 1%
residential, industrial, or commercial.
Next, statistical methods and expert judgment were used to identify 26 biological community variables
from a list of over 400 variables using stepwise discriminant analysis and iterative backward selection
procedures to best assess attainment of the biological goals in the state's ALUs, and to best predict
membership of an unknown stream sample to one of the four water quality classes (A, B, C, and NA).
These were the methods used by Maine; for alternative approaches to variable selection and optimizing
group separation, see Van Sickle et al. (2006). The 26 variables are in Table 25 (four original variables
were discontinued following recalibration of the model). Linear discriminant functions were developed
from the 26 quantitative macroinvertebrate variables. The discriminant functions determine the
probability that a site belongs to a given water quality class. Using a linear optimization algorithm to
calculate the discriminant function coefficients, multivariate space distance was minimized between
sites within a class, while the distance between classes was maximized. Note that three variables used
as predictors in the second-stage models were not calculated directly from the biological data, but
instead were probabilities of group membership reported by the First Stage (four-way) discriminant
model (see below).
The final, overall discriminant function is calculated using one four-way model and three two-way
models. First, using only nine variables and calculated coefficients, the four-way model calculates the
probability (range 0.0-1.0) that a site fits into each of the three attainment classes (AA/A, B, or C) and
the non-attainment class (NA). The resultant probabilities are then transformed and used as variables in
the three two-way models (Table 25). Use of the second stage, two-way models significantly improves
the predictive accuracy of the overall model.
95
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 25. Measures of community structure used in linear discriminant models for Maine (from
MEDEP 2014; State of Maine 2003). Means refer to the mean of three rock baskets sampled at each
site.
Model
First Stage
(four-way)
model
Class C or Better
model
Class B or Better
model
Class A model
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
21
22
23
25
26
28
30
Measure
Total mean abundance
Generic richness
Plecoptera mean abundance
Ephemeroptera mean abundance
Shannon-Wiener generic diversity (Shannon and Weaver 1963)
Hilsenhoff Biotic Index (Hilsenhoff 1987a, 1987b)
Relative Chironomidae abundance
Relative Diptera richness (Diptera richness/generic richness)
Hydropsyche mean abundance
Probability (A+B+C) from First Stage Model
Cheumatopsyche mean abundance
EPT:Diptera richness ratio
Relative Oligochaeta abundance
Probability (A+B) from First Stage Model
Perlidae mean abundance
Tanypodinae mean abundance
Chironomini mean abundance
Relative Ephemeroptera abundance
EPT generic richness
Sum of mean abundances of: Dicrotendipes, Microspectra, Pamchironomus, and Helobdella
Probability of Class A from First Stage Model
Relative Plecoptera richness (Plecoptera richness/generic richness)
Sum of mean abundances of Cheumatopshyche, Cricotopus, Tanytarsus, and Ablabesmyia
Sum of mean abundances of Acroneuria and Stenonema
Ratio of EP generic richness (EP richness/14; 14 is maximum)
Ratio of Class A indicator taxa (Class A taxa/7)
Note: Variable numbers are not sequential; variables 20, 24, 27, and 29 were discontinued following re-parameterization of the
model.
The three two-way models further refine the discrimination among classes AA/A, B, or C. These models
distinguish between a given class plus any higher classes as a group and any lower classes as a group
(i.e., Classes AA/A + B + C vs. NA; Classes AA/A + B vs. Class C + NA; Class AA/A vs. Classes B + C + NA) as
depicted in Figure 21, and model performance is shown in Table 26 below (MEDEP 2014; State of Maine
2003; Davies et al. In press). The two-way models are not strictly independent of the four-way model,
because they use output probabilities of the four-way model as predictor variables.
96
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Four-way model
Two-way models
Class A model
Class B or better model
Class C or better model
B, C, NA
Figure 21. Schematic of four-way and two-way model relationships used by Maine DEP to refine the
discrimination among classes (Source: MEDEP 2014).
Table 26. Classification of stream and river sites by two-way linear discriminant models for three
classifications. Numerical entries represent the percent of sites classified from o priori classes (row)
into predicted classes (columns). Therefore, diagonals are % correct classification.
Final A Classification
Model Predicted Class
A priori class
Class A
Classes B, C, NA
Class A
90.00% (108)
10.28% (26)
Classes B,C, or NA
10.00% (12)
89.72% (227)
Final B or Better Classification
Model Predicted Class
A priori class
Class B or better
Classes C, NA
Class B or better
96.57% (225)
11.43% (16)
Classes C or NA
3.43% (8)
88.57% (124)
Final C or Better Classification
Model Predicted Class
A priori class
Class C or better
NA
Class C or better
96.07% (293)
14.71% (10)
NA
3.93% (12)
85.29% (58)
Note: Number in parentheses indicates the number of sites.
97
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Once the probability that a site belongs to a certain class is calculated, the Maine Biocriteria Rule
describes the assessment process the Department follows to conclude whether the site attains the
minimum standards of its assigned classification (MEDEP 2014; State of Maine 2003). In order to
determine whether a site attains at least Class C or is in non-attainment, the probability outcome using
the "Class C or better model" is used. If the probability is greater than 60%, then the sample attains
Class C or higher, but if it is less than 40% then the site is in non-attainment. If a site falls within 40%-
60%, then best professional judgment is used to determine whether the site attains Class C, does not
attain Class C, or is indeterminate of Class C. For any site found to be indeterminate, additional
monitoring is scheduled in order to make a decision.
Those samples that attain Class C are then tested for Class B attainment using the probability of Class B
outcome from the "Class B or better model." If the probability is greater than 60%, then the sites are
deemed to attain at least Class B status. Those values below 40% are now considered to be sites that
attain to Class C. If a value falls between 40% and 60%, then the outcome is indeterminate of Class B. If
the site designated ALL) is Class A or Class B, then additional monitoring is conducted to determine to
which attainment class the site belongs.
When the probability outcome for a site is 60% or greater using the Class B or better model, it is then
tested using the "Class A Model." If the probability of Class A is 60% or greater, then the site attains class
A standards. If the value is 40% or less, then the site attains to Class B. If the value is between 40% and
60%, the finding is indeterminate of Class A (though it does attain Class B). Additional sampling will be
required if the designated use of the site is Class A. Maine's WQS state that sites determined to attain
the standards of the next higher class must be reviewed and considered for re-classification to the next
higher class in order to maintain the higher water quality conditions that are being achieved (State of
Maine 2004).
The LDM provides a probability of membership result. It explains model performance on a particular
sample and can be used to assess the strength of the model decision. Additionally, each of variables can
be examined to determine the strength of their contribution to the decision. After the LDM predicts the
class attained by a site, a provision in MEDEP regulations (State of Maine 2003) allows for professional
judgment to make an adjustment to the evaluation. Any adjustment may be made using analytical,
biological, and habitat data. Professional judgment also may be employed when the condition of the
stream does not allow for the accurate use of the linear discriminant models. Such factors may include
habitat influences (e.g., lake outlets, impounded waters, substrate characteristics, tidal waters),
sampling issues (e.g., disturbed samples, unusual taxa assemblages, human error in sampling), or
analytical and sample processing issues (e.g., subsample vs. whole sample analysis or human error in
processing) (MEDEP 2014; State of Maine 2003).
4.4 Automation of Decision Models
Any of the BCG decision models described above (sections 4.1-4.3) can be automated in databases,
spreadsheets, or other commonly available software. Multimetric models have been incorporated into
spreadsheet formulas and relational databases (e.g., Environmental Data Acquisition System [EDAS] and
many state databases). Discriminant models and other statistical tools can also be coded in R and
combined with a database or interactive web pages. More recently, several BCG multiple attribute
decision models have been incorporated into MS-Access® applications.
98
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
For example, user-friendly automated models have been developed in Microsoft Excel® for the Upper
Midwest (Gerritsen and Stamp 2012) and Northern Piedmont region of Maryland (Stamp et al. 2014).
Additionally, the Little River Band of Ottawa Indians (LRBOI) has been using the Excel spreadsheets for
the Upper Midwest BCG models to obtain BCG level assignments for all of their fish and
macroinvertebrate samples from the lower Big Manistee watershed.
Geospatial database technology has advanced in recent years and shows promise for application in
water quality management programs, including condition assessments. For example, Maine's
discriminant model is incorporated into Maine's Oracle® relational database that is fully georeferenced
and linked to the state's spatial database. The state's spatial database and selected, quality assured
environmental data, including biological criteria assessment results, are publicly accessible via Google
Earth.9 Linkage between traditional databases that report biological assessment outcomes, and geo-
spatial databases connected to natural bio-geophysical factors and disturbance parameters at multiple
spatial scales, represent the growing edge of the emerging science of biological assessment.
4.5 Conclusion
A core objective of BCG calibration, from conceptualization to quantification, is to explicitly and
transparently link science with management decisions in using biology to interpret AW goals. This
linkage can lead to enhanced stakeholder understanding and engagement in public decision making on
goal setting and in assessing current conditions in relation to the ALU goals. However, information on
stressors, their sources, and mechanism will be needed to identify actions to restore degraded waters
and protect current conditions. Chapter 5 provides a conceptual framework, or template, to assist states
in identifying the primary stressors and their sources and mechanisms of action, that impact their
waters. This framework can be used by the states to organize data and information on watershed
characteristics, hydrologic modifications, and stressors related to ALU goals.
9 http://www.maine.gov/dep/gis/datamaps/index.htmltfblwq. Accessed February 2016.
99
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Chapter 5. The Generalized Stress Axis
The x-axis of the BCG, the GSA, conceptually describes the full range, or gradient, of anthropogenic
stress that may adversely affect aquatic biota in a particular geographic area. It is a theoretical construct
that in application has been defined by states using known, quantitative stress gradients typically
representing a portion of the stressors impacting a water body. The GSA provides a template for
development of a quantified stress axis using available databases. Since the BCG curve represents the in-
situ response of the resident biota to the sum of the stressors to which they are exposed, the GSA
should be developed for the same geographic area and water body type for which the BCG is to be
developed.
Once quantified, a GSA can serve several purposes. First and foremost, it can be used in development of
decision rules for BCG model calibration. Second, the GSA and its underlying data can be used to inform
management decisions and assess outcomes. Key applications of a GSA include:
Guide to selection of samples to be used in BCG decision rule development:
• Guide the selection of sites from a data set to ensure that the assessed sites cover as wide and
full a range of stressors as possible, within the limits of the data set (see Chapter 3, section
3.3.1).
• Guide the assignment of different taxa to the different tolerance categories specified in the BCG
(see Chapter 3, section 3.3.2).
Better link management decisions and outcomes:
• The data collected for developing a stress gradient might be used to help identify and rank
sources and stressors within a region, watershed (e.g., 8- or 12-digit hydrologic unit code (HUC8
or HUC12, respectively)), and/or catchment10 and improve the linkage between biological goals
and management actions. Ideally, an improved connection between biological condition and
stressors will assist state agencies in prioritizing sources and stressors for action, select effective
BMPs, and track improvements. This application will likely occur after BCG development and
require causal analysis (e.g., CADDIS; Suter et al. 2002; Norton et al. 2015).
• The data collected in development of the GSA might also be repurposed to inform additional
management tools. For example, field-based stressor-response relationships can be used to help
develop benchmarks for ALL) (protective thresholds for contaminants or excess nutrients or
conductivity; e.g., Cormier and Suter 2013; Cormier et al. 2013; USEPA 2011a). In addition, data
analyses that describe the distribution of stressors that occur naturally can be repurposed to
define background conditions.
This chapter describes the conceptual foundation of the GSA; discusses technical issues to be considered
in developing a GSA for specific geographic areas and water body types; and, provides an overview of
some approaches for quantifying a GSA.
10 Catchment is defined as an incremental watershed that drains directly into a stream reach and excludes
upstream areas. See: http://nhd.usgs.gov/. Accessed February 2016.
100
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
To date, GSAs have been used to develop decision rules to assign sites to BCG levels using known stress
gradients and available regional, state, and/or county data (as described in first two bullets above).
Some of these GSA applications were explained in the case studies in Chapter 3; they include
quantitative gradients based on use of land cover indicators as surrogates for stressors (Minnesota,
Alabama; see section 3.3.1.1), and an ordinal gradient based on the sum of cumulative stressors present
at a site (Connecticut; see section 3.3.1.2). However, a systematic review and testing of the full suite of
potential technical approaches to define and apply a GSA to BCG development has not been conducted.
Opportunities in the future may include piloting methods for application of national, regional, or
watershed scale data and methods to support state efforts to define and quantify the GSA. Examples of
sources of data include EPA's National Aquatic Resource Surveys,11 the StreamCat data set12 (Hill et al.
2015), and EPA Office of Research and Development's watershed integrity indicators and map of the
ecological condition of watersheds across the country (Flotemersch et al. 2015). Examples of methods
that are currently available include the Healthy Watershed Methodology,13 the Recovery Potential
Screening tool (Norton et al. 2009),14 the Analytical Tools Interface for Landscape Assessments
(ATtiLA),15 and the National Land Cover Database (NLCD).16 Sources for both data and methods include
the Watershed Index Online (WSIO)17 and EnviroAtlas.18
5.1 The Conceptual Foundation of the Generalized Stress Axis
The purpose of this section is to provide a broad conceptual framework and terminology that describes
the effects of human activities on biological communities and forms the basis for constructing a GSA.
This framework can also be used to facilitate application of research to advance the development and
application of the GSA as part of a quantitative BCG model.
The intent of the GSA is to reflect the cumulative degree of anthropogenic stress experienced by aquatic
biota. Five major ecological factors that reflect environmental processes and materials determine the
biological condition of freshwater aquatic resources: flow regime, water quality, energy source, physical
habitat structure, and biotic interactions (Figure 22) (Karr and Dudley 1981). The first four of these
factors (flow regime, water quality, energy source, and physical habitat structure) form the construct for
a GSA. Appendix A-l provides an organizing framework for a GSA and illustrates how a GSA might
classify sites as high, medium, or no/low levels of stress for two general regions of the U.S., humid
temperate and arid, based on these major factors.
11 http://www.epa.gov/national-aquatic-resource-surveys. Accessed February 2016.
12 http://www.epa.gov/national-aquatic-resource-surveys/streamcat. Accessed February 2016.
13 http://www.epa.gov/hwp. Accessed February 2016.
14 http://www.epa.gov/rps. Accessed February 2016.
15 http://www2.epa.gov/eco-research/analytical-tools-interface-landscape-assessments-attila-landscape-metrics.
Accessed February 2016.
16 http://landcover.usgs.gov/. Accessed February 2016.
17 http://www.epa.gov/watershed-index-online. Accessed February 2016.
18 http://www.epa.gov/enviroatlas. Accessed February 2016.
101
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Solubilities
Adsorption
Nutrients
Organics
Hardness
Nutrients
Sunlight
Organic Matter Inputs
1°and 2
Production
Alkalinity
\ I /
Velocity
High/Low
Extremes
Precipitation
& Runoff
Non-native
Species
Hatcheries
Width/Depth
Bank Stability
Channel
Morphology
Riparian Vegetation
Siltation \
Sinuosity
Current
Substrate
t N
Canopy Instream Gradient
Cover
t t
Reproduction Feeding
Parasitism
Competition
Disease
Predation
Figure 22. The five major factors that determine the biological condition of aquatic resources (modified from
Karr and Dudley 1981). Four of the five factors, flow regime, water quality, energy source, and physical habitat
structure, are the basis for the conceptual GSA as described in this document. The fifth factor, biotic interaction,
is incorporated as part of the BCG y-axis levels and attributes.
An event or activity that alters one or more of these five factors is called a disturbance. Disturbances can
occur outside of the stream and riparian zone (e.g., land use changes within the watershed, climate) or
within it (e.g., dams, point source discharges). Ecosystems normally have some level of disturbances that
occurs within a range of natural variability (e.g., Berger and Hodge 1998; White and Pickett 1985).
Anthropogenic activities can cause disturbances that exceed the range of natural variability, and they
are said to exert pressure19 upon an aquatic system, or state, by altering ecosystem processes and
materials, ultimately generating stressors that adversely impact biological condition (Niemi and
McDonald 2004). The term pressure conceptually and mechanistically links larger scale landscape and
hydrological alterations to the in-stream stressors that affect aquatic biota (Grain and Bertness 2006;
Rapport and Friend 1979; Samhouri et al. 2010; Villamagna et al. 2013). Though different terminology is
employed, the Stressor-Exposure-System Response paradigm (e.g., Barnthouse and Brown 1994)
typically employed in water quality criteria development is comparable in that both conceptual models
ultimately help accomplish the same objective—linking human activities to stressors to changes in
biological condition (Figure 23) so action can be taken to protect or restore aquatic resources.
The use of the word pressure in this context has a well-established history in the European environmental
literature. Pressure is a term originally proposed by the Organisation for Economic Co-operation and Development
(OECD 1998) and used by the European Union in its Water Framework Directive (European Environment Agency
1999).
102
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Human Activity Can
Generate Pressures
i
that alter
I
Ecosystem Processes
and Elements
_
and create
Stressors
which cause
I
Change in
Biological Condition
Figure 23. Human activities can cause disturbances in the environment that exceed the range of natural
variability, generating pressure upon an aquatic system that results in altered environmental processes and
materials, which, in turn, create stressors that adversely impact biological condition.
103
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Stressors are the proximate causes of biological effects. They are the link between human activities and
the change in biological condition (Figure 23). Stressors can co-occur in time and space when they are
generated by the same human activity or source and/or any overlapping activity or source. Stressors
may affect more than one aspect of biological condition, and a particular change in biological condition
can also be the result of multiple Stressors acting simultaneously. Since multiple Stressors are usually
present, the x-axis is intended to reflect their cumulative spatial/temporal co-occurrence in a GSA, much
as the y-axis generalizes biological condition.
Point source discharges of pollutants were the dominant pressures to fresh waters addressed in the
initial implementation of the CWA. While this pressure still exists today, water quality managers also
face additional challenges stemming from in-stream hydrological modifications, forest harvest,
agriculture, and urbanization, as well as emerging pressures associated with the inadvertent or
deliberate introduction of invasive species (Ricciardi and Maclsaac 2000), the consequences of
greenhouse gas emissions (e.g., Bierwagen et al. 2012), use of pharmaceutical products (Rosi-Marshall
and Royer 2012; Rosi-Marshall et al. 2013), and even recreation (Bryce et al. 1999; Poff et al. 2002;
Richter et al. 1997). Additionally, Stressors can exert both direct effects on the biota and indirect effects
through modification of habitat and interactions with other Stressors (Karr and Dudley 1981; Karr et al.
1986; Poff et al. 1997; Slivitzky 2001) (Figure 24).
For example, a GSA that considers flow regime changes would consider many Stressors and their
interactions. Stream flows directly influence stream biota, but they also interact in multiple ways with
other in-stream factors including water quality parameters, such as DO and temperature. Altered stream
flows are strongly associated with many habitat variables such as channel structure, erosion, bank
instability, and lower base flows (Poff et al. 1997; Richter et al. 2003; Poff et al. 2010). All of these
factors associated with the flow regime have the capability of affecting species distributions,
abundances, life history traits, and competitive interactions (Greenberg et al. 1996; Kennen et al. 2008;
Poff and Allan 1995; Poff et al. 1997; Robson et al. 2011; Walters and Post 2011).
Many of the changes to the natural flow regime can be attributed to human activities, such as dam
creation, channelization, and impervious surfaces, along with associated removal of natural vegetation,
water extraction, and loss of surface water storage capacity (e.g., wetlands) (Poff et al. 1997). Altered
flow regimes are also the result of changing climate, with changes observed in precipitation and runoff
amounts, seasonal patterns, and timing, frequency, and intensity of large storms (Frich et al. 2002; Karl
and Trenberth 2003; Poff et al. 2002). Still, flows vary naturally, and it can be difficult to distinguish
anthropogenic disturbance from the range of variation produced by natural processes (e.g., see review
by Berger and Hodge 1998). All of these issues should be considered when developing a GSA that
reflects the stress associated with flow regime changes.
104
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Human Activity Can
Generate Pressures
I
that alter
Human Activity Can
Generate Pressures
I
that alter
I
I
Ecosystem Processes
and Materials
and create
Stressors
which cause
I
Physical Habitat
Structure
Change in
Biological Condition
\l
Ecosystem Processes
and Materials
and create
which cause
I
which cause
I
Change in Physical
Habitat Structure
Change in
Biological Condition
I
Model I
J
Model II
Figure 24. Hierarchical effects of disturbance. When assessing the relationship between stressors and biological
effects, one of two implicit models is assumed. Model 1—the biota at a site are determined by the
environmental covariates characteristic of the habitat. The stressors associated with a human-related
disturbance directly influence biota. Model 2—the biota at a site are determined by the environmental
characteristics of the site. However, the stressors associated with a human-related disturbance influence both
the physical habitat structure and the biota itself. Consequently, the biological effects reflect the combined
direct effects of the stress and the disturbance-mediated habitat alteration (From: Ciborowski et al.
unpublished). Comprehensive and integrated monitoring data (biological, chemical, physical) coupled with
causal assessment will help distinguish direct from indirect effects (USEPA 2013a).
105
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
5.1.1 Technical Issues in Developing a Generalized Stress Axis
This section discusses some of the technical issues to be considered in defining a GSA, including
temporal and spatial scales, multiple stressors, legacy effects, and predicted impacts of climate change
on aquatic systems. The concepts of spatial and temporal scale are critical issues in adequately defining
the GSA. Pressures, stressors, and their effects on biota (e.g., biotic response) operate at different
spatial and temporal scales (Glasby and Underwood 1996). Stressors are expressed over temporal and
spatial scales ranging from a one-time, localized event (pulse event; Bender et al. 1984) to long-term
chronic exposures occurring continuously (press events) over vast landscapes. Additionally, stressors
may be introduced through diffuse or point sources delivered from upstream in the channel or
watershed, or laterally from riparian, floodplain, or upland sources. Pollutants can also be delivered to a
stream, river, lake or wetland from above through atmospheric sources, or below from groundwater
sources. Activities in the watershed or along the water body corridor will influence the connectivity and
integrity of the water resource. Additionally, climate change can exacerbate the intensity of local
stressors (e.g., more heavy rainfalls can produce increased runoff and sediment load).
As discussed previously, human activities can produce multiple stressors, which in turn will affect
biological condition. Stressors can interact with one another to create a synergistic response, behaving
in an additive or multiplicative manner; they also may counteract one another. The steady accumulation
of small pressures in watersheds results in cumulative effects, which add to the challenges of
characterizing, evaluating, and managing stressors.
The influence of individual stressors on biological condition in specific water bodies can be particularly
difficult to disentangle because each stressor potentially exerts indirect and direct forces. The
complexity of interactions among stressors makes it difficult to identify single stressor-single biological
effect relationships (Hodge 1997; Noss 1990; Vander Laan et al. 2013). Stressor identification is one
causal assessment approach useful for identifying the stressors that cause biological effects (USEPA
2000; Norton et al. 2015).20
However, when sufficient data are available, quantitative modeling approaches can be used to describe
the complex relationships between pressures, stressors, and their effects on the biota. Niemeijer and
deGroot (2008a, 2008b) advocated summarizing the interactions among stressors to create causal
networks as a means of better understanding the complex relationships between pressures and their
ultimate effect on the biota, and this approach has been applied to streams with qualified success. Allan
et al. (2012) used Bayesian Belief Network analysis to characterize the effects of sedimentation on
macroinvertebrates in agricultural streams in the U.S. Midwest and in New Zealand affected by
sedimentation due to grazing and forestry practices. Riseng et al. (2010, 2011) used Structural Equation
Modeling to document relationships between stress and stream biota. They determined that land use
effects in total were more important influences on metrics offish and invertebrate biota than effects of
point source discharges.
The concept that human activities produce multiple stressors provides the foundation for one common
approach to describing an overall gradient of stress using land cover information as a surrogate for
stressor information. In this approach, the GSA is developed using broadly defined, relatively easily
measured factors that produce many stressors simultaneously (e.g., amount of urban development or
road density in a catchment). Mapping the distribution of pressures, for example land uses associated
20 See also http://www3.epa.gov/caddis/. Accessed February 2016.
106
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
with particular human activities, has proven to be an effective way of documenting the location of
possible sources that produce the stressors that lead to biological degradation (Allan et al. 2013; Brooks
et al. 2009; Danz et al. 2005, 2007).
Stressor indicators can be developed from such measures as population density, proportion of land
devoted to agriculture or urban development, total miles of roadway, or quantities of water
used/released (e.g., Allan et al. 2013; Host et al. 2005, 2011; Hunsaker et al. 1992; Jones et al. 1999,
2001; O'Neill et al. 1988, 1997; Riitters et al. 1995, 1996, 1997). The advent of improved remote sensing,
digital technology, and the ability to map land uses has provided an important tool for documenting the
location and extent of pressures on the landscape. This approach has been used effectively to assess
watershed and coastal conditions such as in the Laurentian Great Lakes for decades where Danz et al.
(2005, 2007) and Allan et al. (2013) documented the distribution of the composite stress contributed by
human activity throughout the Great Lakes (Figure 25). A simplified form of the Danz et al. (2005)
system, the Watershed Stress Index (Host et al. 2011), is currently used to report on the condition of
Great Lakes watershed, including tracking progress towards achieving the overall purpose of the
binational Great Lakes Water Quality Agreement "to restore and maintain the physical, chemical and
biological integrity of the Great Lakes Basin Ecosystem."21 Allan et al. (2013) used expert assessment to
delineate threats to the biological integrity of the Great Lakes themselves. Host et al. (2011) mapped the
distribution of watersheds in which specific groups of biota were at least and at greatest risk of
degradation due to urban and agricultural pressures.
Figure 25. Cumulative stress within the St. Louis River watershed, a tributary to Lake Superior. Darker shading
indicates increased stress. The stress score is based on the cumulative sum of % agricultural land use, population
density, road density, and point source density. Values were each normalized to a 0-1 scale before summation.
This index was used to calibrate water quality responses to stress in the St. Louis River Area of Concern (Bartsch
et al. 2015). (Map by Tom Hollenhorst, EPA, Mid-Continent Ecology Division)
https://www.ec.gc.ca/grandslacs-greatlakes/default.asp?lang=En&n=70FFEFDF-l. Accessed February 2016.
107
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
However, although land use can be a useful general pressure indicator, practices within a given land use
category can change over time, which may reduce or increase the stressors that are produced by that
land use. Local variables can exert important influences on biological conditions that are not captured by
remote sensing or other land cover data alone. For example, the incidence of tile drainage is generally
not mapped; drainage intensity has increased in some areas of the Midwest resulting in increased
annual flows in ditches related to reduced evaporation off of land surfaces (Blann et al. 2009). Miltner
(2015) used extensive biological, stressor, and pressure (agricultural practices) data in Ohio and
demonstrated that conservation measures have contributed to improved environmental conditions in
Ohio headwater streams. Miltner (2015) concluded "that stream physical habitat clearly influences
water quality, and therefore structural measures that improve habitat function in channelized streams
and drainage ditches are a necessary component of efforts to combat eutrophication." Analyses such as
these would not be possible without the accumulation of substantial monitoring data collected at a
higher spatial resolution (Blann et al. 2009; Miltner 2015). Additionally, documenting biological
conditions at the local reach and watershed scale makes it apparent that broad scale use of indicators
such as land cover are not in themselves adequate predictors of biological impairment in specific water
bodies. The scale of application is a critical factor—important stressors that act at the local reach and
watershed scale can be missed.
An additional caveat in using land cover as a sole basis for GSA development is that the indicators are
typically based on current land uses although some types of past land use patterns are available as
mapped information. Many human activities in watersheds leave permanent or semi-permanent
changes, termed "legacy effects." For example, persistent contaminants such as DDT, PCBs, PAHs,22 and
metals can end up in sediments, and they may be resuspended or buried permanently, depending on
the depositional environment. Excess phosphorus may be buried in lake or pond sediments. In eastern
U.S. Piedmont and Appalachian highlands, stream valley morphology has changed permanently in many
places due to historic land use changes from the colonial period to the present: from initial clearing, to
colonial and early American hydropower development, early agriculture, subsequent agricultural
abandonment and forest regrowth, followed by recent suburban development (e.g., Maizel et al. 1998;
Walter and Merritts 2008). These legacies may account for intermittent stressors in the form of
contaminants, nutrients, and sediments that can be eroded and resuspended from historic
sedimentation during storm events, or permanent stressors in the form of hydrological modifications or
sedimentation. Documenting previous land use and expanding monitoring programs to include
appropriate parameters will assist in detection of these stressors.
Regardless of the information used in defining a GSA, the impact of climate change will increasingly
need to be taken into account. Climate change is a widespread disturbance that is capable of moving the
system outside its natural range of variation, even in the absence of other anthropogenic disturbances,
by elevating air and water temperatures, altering flow regimes through changes in the seasonality of
precipitation, altering soil moisture regimes, and through changes in the frequency and intensity of
storm events and fires (IPCC 2014; Melillo et al. 2014). The effects of changing climatic conditions,
whether considered naturally or anthropogenically driven, are superimposed on other anthropogenic
stressors generally leading to an exacerbated effect (c.f. Comte et al. 2013; Palmer et al. 2009; Hoegh-
Guldberg et al. 2007; Arnell 1999). In general, water quality is likely to be negatively impacted by effects
of climate change through altered flow regimes leading to higher peak flows and lower base flows.
Altered flow regimes in turn influence extremes in water temperature, DO concentrations, changes in
22 DDT: dichlorodiphenyltrichloroethane; PCB: polychlorinated biphenyl; PAH: polycyclic aromatic hydrocarbon
108
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
biogeochemical processing, and biotic assemblage structure and function that these factors regulate
(Melillo et al. 2014). The effects of heavy downpours are exacerbated by impervious surfaces, leading to
greater sediment, contaminant, and nutrient loading. Appendix A-2 provides examples of stressors and
potential indicators of climate change under low, medium, and high stress scenarios for humid and arid
regions. The BCG with well-defined biological indicators (y-axis) and stress indicators (x-axis) can be used
to determine current baseline conditions and track changes in parameters that are associated with
climate change, such as flow and temperature.
5.2 Development of a Generalized Stress Axis
In preparation for BCG development (see Chapter 3, sections 3.2 and 3.3), the process to develop a GSA
for a specific geographic area and water body type includes a series of steps: classifying sites to reduce
natural variability; identifying undisturbed or minimally disturbed conditions; and identifying indicators
and the data that will be used to define the gradient of stress.
The first step in GSA development is to classify the aquatic resource (e.g., biogeographic regions, basins,
biological considerations) (Herlihy et al. 2008; McCormick et al. 2000; Van Sickle and Hughes 2000;
Waite et al. 2000). Classification is also an important component of biological assessment program
development (see section 3.2.1.1). The purpose of classification is to reduce variability in natural
conditions that can contribute to or influence stressors and biological assemblages. Features such as
latitude, climate, geology, and landforms can explain the dominant patterns of variation in stressors
across large regions (e.g., Herlihy et al. 2008). These broad-scale classification systems can be
supplemented by local-scale features (e.g., slope, groundwater seeps) that can contribute to site-scale
patterns in biotic assemblages (Hawkins and Vinson 2000; Pyne et al. 2007; Snelder et al. 2004, 2008;
Van Sickle and Hughes 2000).23
A second step in GSA development is characterizing undisturbed or minimally disturbed conditions for a
particular area. This characterization is the benchmark against which areas to be evaluated will be
compared (as discussed in section 3.2.1.1), allowing for development and calibration of indices such as
the mlBI and O/E assessment models. For most state biological assessment programs for streams, this
step involves use of the state's reference site database. An important consideration when selecting
reference sites is whether the reference sites represent undisturbed, minimally disturbed, or least
disturbed conditions (Hawkins et al. 2010; Herlihy et al. 2008; Hughes 1985, 1994; Hughes et al. 1986;
Moss et al. 1987; Stoddard et al. 2008). In BCG development, descriptions of undisturbed and minimally
disturbed reference conditions (e.g., BCG levels 1 and 2) are critical components of model calibration. In
some places, calibration may be based solely on historic records or other sources of information. Like
level 1 of the BCG, the "low stress" end of the stress axis is anchored in the "as naturally occurs" or
undisturbed or minimally disturbed, condition (i.e., no/minimal anthropogenic stressors).
The third step is to identify indicators and data sets that will be used to define the GSA. The major
environmental factors shown in Figure 22 can be used as prompts to identify indicators (e.g., Appendix
A-3). When evaluating data sets to develop a GSA, it is important to bear in mind that the biological
conditions will reflect effects of unknown sources and unmeasured stressors, as well as incorrectly
23 A comprehensive review of recent classification systems is beyond the scope of this document. There is still
much to be learned about how biotic effects from local vs. catchment scale disturbances differ between
catchments that are largely disturbed, and those that are relatively undisturbed (see review by Johnson and Host
2010).
109
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
characterized data sets. In this regard, the GSA is only as robust as the data upon which it is based.
Characterizing to the extent possible the degree of uncertainty around the stressor-response (i.e.,
effect) relationships is important. There will always be some level of unexplained variation. But, where
relationships between stress, or stressors, and biological response are poorly predicted, further
assessments should be conducted. For example, as mentioned above, legacy contaminants from long-
defunct industrial activities are typically invisible to remote imaging, yet may wash out periodically in
storm events. A water quality assessment conducted for screening purposes is unlikely to capture such
rare events. Intensive, directed sampling is more likely to detect the contamination, possibly after
determination that a downstream location is biologically impaired from unknown causes and historical
land use records are researched.
As explained earlier, this document does not comprehensively review or evaluate the approaches
available to define a GSA. The examples discussed below represent several approaches that have been
used to define stress gradients and are intended to prompt ideas and enhancements.
5.2.1 Using Land Cover Measures as Stressor Indicators
One approach to quantify a GSA relies upon land cover data. The land cover indicators serve as
surrogate indicators for stressors, typically multiple stressors associated with a specific land use. Many
human activities that cause stress in aquatic systems can be summarized in land cover delineations.
Because land cover can be expressed as a fraction or percent of a watershed, catchment, or zone within
the catchment (e.g., riparian corridors), using land cover data provide an obvious initial approach for
summing land uses for an overall index of pressure. Land cover data generally do not include
information on legacy sources and stressors unless intentionally mapped, nor do the data usually
incorporate in-stream measures of water quality or habitat quality. Thus, the methods that rely solely on
land cover should be regarded as the "first cut" tool in a toolbox that may contain multiple approaches.
If stress-response relationships are poorly predicted by land cover data, subsequent analyses should
include a more complete portfolio of stressors that contain both local habitat and water quality
variables, as well as potential legacy pressures. Although remote sensing is a useful coarse focus,
stressors and their effects on the biota can vary substantially.
The simplest land cover-based GSA is comprised of one, or the sum of several, land covers calculated for
the catchment of each aquatic sampling point in the database being used. For example, in the Maryland
Piedmont, percent impervious surface was used as a single stressor gradient because of the extent of
urban and suburban land use throughout the mid-Atlantic Piedmont (see Chapter 3, sections 3.3.1.1 and
3.3.2.1). As another example, developers of a BCG for fish assemblages in Minnesota lakes used a GSA
composed of a simple sum of percentages of urban, agricultural, and mining lands (section 3.3.2.1).
The above land cover-based GSAs do not differentially weight various land uses (as measured by land
cover) in terms of their effects on aquatic biota. For example, impervious surface strongly affects stream
hydrology, habitat quality, and biology (e.g., Stranko et al. 2008) and effects of agricultural land use
depend on its intensity and local agricultural practices. An alternative method, the landscape
development intensity index (LDI), weighs the intensity of multiple land uses in a study area (Brown and
Vivas 2005). The LDI is a measure of human activity based on a development intensity measure derived
from non-renewable energy use in the surrounding landscape. The LDI is calculated using all
nonrenewable forms of energy (e.g., electricity, fuels, fertilizers, pesticides, and water (both public
water supply and irrigation) (Brown and Vivas 2005)) used directly or implicitly in various land use
classifications. Land uses are classified, and an intensity factor is assigned to each land use type (Table
27).
110
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 27. Land use classification and intensity factor (LDI coefficient) for Florida landscapes (modified
from Brown and Vivas 2005)
Land Classification
Natural system
Natural open water
Pine plantation
Recreational/open space - low intensity
Woodland pasture (with livestock)
Improved pasture (without livestock)
Improved pasture - low intensity (with livestock)
Citrus
Improved pasture- high intensity (with livestock)
Row crops
Single-family residential - low density
Recreational/open space - high intensity
Agriculture - high intensity
Single-family residential - medium density
Single-family residential - high density
Mobile home (medium density)
Highway (2-lane)
Low intensity commercial
Institutional
Highway (4-lane)
Mobile home (high density)
Industrial
Multi-family residential (low-rise)
High-intensity commercial
Multi-family residential (high-rise)
Central business district (average 2-stories)
Central business district (average 4-stories)
Intensity Factor (LDI coefficient)
1.00
1.00
1.58
1.83
2.02
2.77
3.41
3.68
3.74
4.54
6.9
6.92
7.00
7.47
7.55
7.70
7.81
8.00
8.07
8.28
8.29
8.32
8.66
9.18
9.19
9.42
10.00
The LDI has been used as a human disturbance gradient for wetlands (Brown and Vivas 2005; Chen and
Lin 2011; Lane 2003; Mack 2006, 2007; Reiss 2004, 2006; Reiss and Brown 2005, 2007; Surdick 2005;
Vivas 2007; Vivas and Brown 2006), streams (Brooks et al. 2009; Fore 2003, 2004; Harrington 2014;
Stanfield and Kilgour 2012), and lakes (Fore 2005). It has also been used for coral reefs (Oliver et al.
2011). Figure 26 shows application of the LDI for coral reefs. Land use indices similar to the LDI were
used to develop BCG calibrations for Minnesota and Alabama (see section 3.3.1.1).
Ill
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
64°50'0"W
64°40'0"W
8 km
NC1 Christiansted
-17°50'0"
N
-5m
-10m
-20m -500m
Water Current Flow
Land Cover Class: 2.4m resolution
H Forest / Wetland
| | Developed Open Space / Bare Lands
^H Grassland / Herbaceous
! Pasture / Hay
^H Cultivated Crops
| Impervious Surface
-17°40'0"
64°50'0"W
64°40'0"
0 1.5 3
6km
W1
W2
Bl
NE2
SE2
SW2
SC1
Water Current Flow
Watershed LDI
•• 1.73
^H 1.76
•I 2.15
229
235
241
264
H 2.55
H 2.59
H 282
^B 3.48
17°50'0"
N
17°40'0"
Figure 26. LDI applied to St. Croix watersheds and associated coral stations (Source: Oliver et al. 2011). Top
figure shows land use/land cover and EPA coral reef stations. Land use/land cover used in the analysis is shown
at 2.4 m resolution. Bottom figure show the watershed LDI values on a green-yellow-red continuum, where
green indicates the lowest human disturbance and red indicates the highest. Watershed abbreviations: Bl: Buck
Island; NC: North Central; NE: Northeast; SC: South Central; SE: Southeast; SW: Southwest; W: West.
Nationally, the LDI has been mapped at HUC12 watershed as part of the WSIO data library using publicly
available data from 2001. The WSIO contains mean, median, standard deviations, and sum of values for
empower density (derivation of LDI) for a HUC12 watershed, its riparian zone, and hydrologic connected
zone. Currently the WSIO data set is being updated nationally with the most recent NLCD data and
should be available for use in near future.
112
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
5.2.2 Ranking Sites by Summing Stressor Indicators
Another approach to develop a GSA is to tally the number of stressor indicators observed at a particular
site and establish a method to score the results. Many examples of this approach have been used across
different regions, spatial extents, and ecosystem types (Chow-Fraser 2006; Uzarski et al. 2005). This
approach entails identifying observed human activities and observed stressors (and their sources if
information is available) and summing them to produce an overall index that can then be used to place
sites in order from least to most stress.
The first step for the ordinal approach involves identifying and quantifying, for each site in a biological
monitoring database, the relevant data available, including data on sources, in-stream measured water
quality, riparian condition, land cover, riverscape alterations, known point source discharges, and
observed nonpoint sources. For instream measures, it is important to distinguish non-detects (known
and effectively absent) from not sampled (unknown; no data). A conceptual diagram of sources,
stressors, mechanisms, and effects is helpful in organizing the information (e.g., Norton et al. 2015).
In the simplest implementation, each stressor indicator is evaluated as being present (1) or absent (0) at
a site. The results are added to produce a score for each site. In the Connecticut case example (section
3.3.1.2), stressor indicators included reduced natural land cover, developed land, impervious surface,
total chloride (a measure of total point source discharge), and four metals (copper, iron, nickel, zinc).
Scoring in the case example was not simply 0-1; some stressor scores could range on an ordinal scale of
0-3, depending on the concentration or intensity of a given stressor. The results were used to divide
sites into five overall stress categories ranging from "least stressed" to "severe stressed." The resultant
gradient helped identify potential most-stressed, least stressed, and intermediately stressed sites in the
BCG development data set. It is important to reiterate that the stress information was hidden from the
expert panel during its deliberations.
For development of the BCG in Minnesota, MPCA developed a disturbance index (the HDS) that
combined scores associated with land use metrics with additional indicators. The index includes eight
primary metrics, which include measures of watershed land use, stream alteration, riparian condition,
and known permitted discharges. The disturbance index scores can range from 1, representing
completely altered and heavily stressed streams, to 81, representing nearly pristine watersheds. The
HDS is described by MPCA (2014e) (see section 3.3.1.1, Table 7). Alabama DEM developed a similar
index (see section 3.3.1.1, Table 8).
5.2.3 Using Statistical Approaches to Combine Stressor Indicators
In the U.S. Great Lakes coastal region, principal components analysis (PCA) was used by a team of
researchers and investigators participating in the Great Lakes Environmental Indicators (GLEI) Project24
(Niemi et al. 2007) to reduce over 200 variables into a single gradient, applying measures of
anthropogenic pressures as surrogate measures of stressors (Danz et al. 2005). The Danz approach
individually considered six different indicators of pressure: agriculture, atmospheric deposition, land
cover, human population, point sources, and shoreline alteration. The GLEI team used a watershed-
based approach to reflect the premise that the environmental effects of these activities in coastal
watersheds can influence environmental conditions in downstream coastal ecosystems. The first
principal component from the analysis explained 73% of the variance in the agricultural-chemical (Ag-
Chem) variables (reflecting land use, agricultural chemical use, and agricultural-influenced nutrient and
24 http://glei.nrri.umn.edu/default/default.htm. Accessed February 2016.
113
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
sediment loading) and was interpreted as an overall gradient in stressors across the basin (Figure 27).
Environmental effects such as changes in water quality, fish assemblage metrics, and bird abundances
were strongly correlated with scores of this stressor gradient, providing verification that the statistically
extracted PCA was biologically meaningful (see description of this project by Niemi et al. 2007). The GLEI
researchers created a flow diagram (Figure 28) that details their steps for quantifying a stressor gradient
(modified from Danz et al. 2005).
PC-based Agricultural Stress Gradient
300 Kilometers
Figure 27. The first principal component of the agricultural variables for the U.S. Great Lakes basin. Darker
shading indicates greater amounts of agriculture (Source: Danz et al. 2005).
While the pressure-stressor model eventually developed for the Great Lakes coastal region was
visualized as a single gradient from low to high levels of stressors, different individual and combinations
of stressors are expected to dominate in different regions. Furthermore, disaggregating the PCA into
individual categories of stress could provide important information about potential mechanisms
affecting the state of the system.
In addition to PCA, there are other statistical approaches to consider. For example, the use of non-
metric multidimensional scaling (NMDS) provides a robust analysis. Unlike PCA, NMDS can deal with
non-normal data, data of varying scales, and outliers in the data. Like PCA, NMDS is a multivariate
statistical analysis that one can use to look at multiple stressors at the same time to create the GSA.
Biological data can also be used to statistically combine stressor indicators into a GSA. For example,
Wang et al. (2008) used Canonical Correspondence Analysis (CCA) to derive the relationship among the
biota and stressor and land use data and weight their disturbance index. They then plotted the
calculated disturbance index against fish IBI scores and percentages of intolerant individuals, dividing
the disturbance index values into five tiers. The first tier was the maximum disturbance index value at
which the fish measures did not show an obvious decline. The remaining four tiers were determined by
dividing the remainder of the disturbance index values into even categories. Use of biological data
ensures that the stressor indicators will be biologically relevant. However, this approach can introduce
some circularity into the analysis if the indicators of biological quality are the same as those used to
develop the BCG.
114
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Define sampling units
*
Compile & summarize
environmental variables
*
Evaluate & categorize
variables
*
Apply regional classification
system
*
Organize
Data
Generate
Remove redundancy with PCA ^IUma"
Disturbance
Gradient
* i
Compute overa
gradient using PC
each stressor
1 stressor ^ Map disturbance gradient using
A axes from color-coded scheme
Figure 28. Flow diagram detailing the steps used by GLEI researchers in quantifying their stressor gradient
(modified from Danz et al. 2005).
Stressor gradients like that developed by GLEI, or others as referenced above, can be developed at
different spatial scales. The GLEI study assessed 5,971 watersheds comprising the Great Lakes basin.
Watershed sizes (areas) were lognormally distributed, with a median watershed area of 4.3 km2 and a
mean watershed size of approximately 86.7 km2 (Ciborowski et al. 2011). However, the gradient can be
applied and scaled as needed to other geospatial units. For example, Nieber et al. (2013) conducted this
same analysis for watersheds of the north shore of Lake Superior, and Bartsch et al. (2015) scaled their
analysis to watersheds of the St. Louis River estuary to assess relationships between stressors and water
chemistry.
5.3 Linking the Science with Management Actions
A quantitative BCG model provides a framework for assessing baseline biological condition and, with
systematic monitoring, can be used to track changes in biological condition. Ideally, a well-defined GSA
and the stressor effects and biotic response models underlying it can be used in conjunction with causal
assessments to better link biologically-defined management goals to the actions taken to protect or
restore the biological conditions.
A stressor can be traced back to its source or tracked forward to the biological effect via a causal
pathway (Figure 29). For example, stream banks that become destabilized due to removal of riparian
plants could be the source of excess fine sediment to a stream. Erosion by high flows is the mechanism
by which the excess fine sediments are generated, and the resulting in-stream siltation is the stressor.
Smothering of bottom substrate habitat and organism gills by these fine sediments are two mechanisms
by which biota are exposed and adversely affected. Invertebrate mortality and fish emigration could be
115
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
some of the environmental outcomes or changes in biotic condition. Further, degradation or loss of
recreational fishing could be societal impact of these changes and may prompt a conservation or
restoration effort depending upon the circumstances.
Human Activity Can
Generate Pressures
I
that alter
Removal of
Riparian Habitat
Ecosystem Processes
and Materials
and create
Stressors
which cause
which cause
Change in
Physical Habitat
Structure
Change in
Biological Condition
Unstable Stream Banks
and Altered Flow
Erodes Fine Sediments
Creating
In Stream Siltation
Smothers Gills
Reduces Predation
Efficiency
Covers Larger Substrate
\
1
Eliminates Pore
Space Degrades
Habitat
Fish and
Invertebrate Condition
(Emigration and
Mortality)
Figure 29. The specific stressor(s) and their intensity (the BCG x-axis—termed the GSA) are created by
pressure(s) acting through specific mechanisms. BMPs can be implemented to prevent or reduce effect on the
biota through restoration, remediation, and/or mitigation.
Actions can be taken that insulate the aquatic biota from the effects of anthropogenic pressures, helping
to maintain or restore the ecological potential of an aquatic system. In the example above, re-
establishing the riparian zone would stabilize the banks and prevent further erosion and unchecked flow
into the stream. Appendix A-4 and MPCA (2015) provide examples of pressures linked to mechanisms
and potential management actions that can mitigate the effect on biota.
116
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Mechanistic processes operate between pressures and stressors and between stressors and their effects
on biological condition (Figure 30; Appendices A-3 and A-4). Understanding these mechanisms and how
they operate helps in predicting the potential effects of a particular management action. The BCG
provides a framework for tracking and documenting incremental improvements in biological conditions
resulting from implementation of a single BMP or combination of BMPs.
Integrating monitoring programs with frameworks like the BCG can improve understanding of how
human activities, stressors, biological responses, and management actions are linked, providing
feedback to guide management decisions. For example, Yoder et al. (2005) reviewed changes to fish
assemblages over 25 years based on an intensive pollution survey designed to assess non-wadeable
rivers in Ohio. They used the linkages between changes in point source pollution loadings,
improvements in instream water quality, and reductions in the extent and severity of biological
impairments to document the effectiveness of advanced wastewater treatment on a statewide scale
beginning in the late 1970s. At that time the documented improvements in biological condition across
all rivers and streams were almost solely in response to water quality-based NPDES permitting for point
sources. Rivers predominantly impacted by nonpoint sources showed improvement over a longer
timeframe where there was a concerted effort to apply BMPs over a wide enough region. Miltner (2015)
was able to document widespread improvements in stream biota and water quality in smaller
headwater streams in Ohio. Both of these studies were based on the state's routine biological
monitoring and assessment of rivers and streams.
A well-defined GSA, and the underlying data set, can serve as a nexus between biological and causal
assessments and provide a link between management goals and selection of management actions for
protection or restoration. The basis of the BCG framework is that greater pressures can generate
increased levels of stressors, and in turn, increased stressors are associated with reduced biological
condition (Figure 30A and B). Typically, the stressors on aquatic systems increase as pressures increase,
which results in a consequent decrease in biological condition. Effective management practices can
target any point in the web of causal events, mitigating the effects of pressures and reducing stressors
with resulting protection or improvement in biological condition (Figure 30C and D).
117
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
A
HIGH
VI
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
5.4 Conclusions
Anthropogenic activities exert pressures on aquatic systems by altering ecosystem processes and
materials and generating stressors that adversely impact biological condition. Many of these stressors
co-occur in time and space, and effects on the biota are cumulative. The relationships between stressors
and effects are complicated—stressors may affect more than one aspect of biological condition, and a
particular change in biological condition can also be the result of multiple stressors acting
simultaneously.
The conceptual GSA describes the range of anthropogenic stress experienced by aquatic biota in a
particular geographic area. Once quantified, it is used in the development of the decision rules to assign
sites to BCG levels (Chapter 3, section 3.3.1) and ensures that the BCG encompasses the full range of
condition along a stress gradient. There is much complexity of interactions and effects from multiple
stressors with varying effects on different biotic components of any aquatic system. The GSA represents
the sum total of stressors and their sources in concept, but in implementation it is composed of multiple
known, quantitative stress gradients that each represent a portion of the actual stress gradient to which
the aquatic biota are exposed. The usefulness of the conceptual framework is to provide a template for
as thorough and comprehensive a technical approach as possible to develop the BCG x-axis and relate
level of stress to the BCG levels and attributes.
Additionally, developing a GSA that reflects the human activities and stressors in a particular
geographical area helps in understanding how specific stressors are generated and how they affect
biotic condition. The data generated in developing a GSA can be used to help identify and rank sources
and their stresses in a particular area and inform management decisions on appropriate actions to
protect or improve a water body. The case examples discussed in this chapter and in Chapter 3 illustrate
how state and local governments have quantified a GSA as part of developing a BCG model for their
specific region or watershed area. As further experience is gained and approaches to define and quantify
the GSA evolve, EPA may supplement this document with additional information.
119
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Chapter 6. Case Studies
The BCG can provide critical information to state water quality management programs at the watershed,
statewide, and ecoregional scales. A comprehensive monitoring and assessment program is a critical
aspect of implementation of the BCG to support water quality management programs. The same data
and information that provide baseline condition assessments over time also can provide information to
inform trend assessments and track incremental changes in condition. In conjunction with monitoring
data, a BCG can be used to help address watershed-specific management needs such as detailed
biological descriptions of designated ALUs, identification of high quality waters and impaired waters,
and documentation of incremental improvements due to controls and BMPs. This information can also
inform TMDL development. This chapter presents six case examples of how states, counties, or
municipalities are using, or considering using, the BCG to support water quality management decision
making.
The six case examples are:
• 6.1 Montgomery County, Maryland: Using the Biological Condition Gradient to Communicate
with the Public and Inform Management Decisions
• 6.2 Pennsylvania: Using Complementary Methods to Assess Biological Condition of Streams
• 6.3 Alabama: Using the Biological Condition Gradient to Communicate with the Public and
Inform Management Decisions
• 6.4 Minnesota: More Precisely Defining Aquatic Life Uses and Developing Biological Criteria
• 6.5 Maine: Development of Condition Classes and Biological Criteria to Support Water Quality
Management Decision Making
• 6.6 Ohio: Tiered Aquatic Life Use Classes and Comprehensive Water Quality Management
Program Support
120
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
6.1 Montgomery County, Maryland: Using the Biological Condition Gradient to
Communicate with the Public and Inform Management Decisions
6.1.1 Key Message
Montgomery County helped to develop a BCG to better inform the public and county decision makers
about a high quality watershed (e.g., undisturbed/minimally disturbed conditions) and the potential
outcome of planned development. Local government decision makers were able to understand how
these high quality streams compared to other streams in Montgomery County and Maryland.
Development plans were modified to protect the streams and watershed and reduce environmental
impacts, while allowing development to proceed.
6.1.2 Background: Early County Policy
In 1994, the Maryland-National Capital Park and Planning Commission (M-NCPPC) adopted the
Clarksburg Master Plan & Hyattstown Special Study Area. The Plan established goals for development of
Clarksburg, Maryland, at that time a mostly undeveloped area along a six to eight lane highway corridor
outside the Washington, DC metropolitan area. The Plan's goals included development of the town with
emphasis on maintaining farmland and open space and promotion of transit-oriented neighborhoods
(M-NCPPC 1994). One critical objective of the plan was the protection of environmental resources while
accommodating development, such as affording special protection to high quality stream systems,
including tributaries to the streams and associated wetlands. The plan specified that development occur
in four phases, with requirements that must be met in order for development to proceed from one
phase to the next. This staging allowed for consideration of new data and information on the impacts of
development on streams and rivers, as well as improvements in mitigation technology and changes in
county, state, or federal policies
or regulations that might affect
implementation of the 1994
plan. For example, in 2008, the
County revised the 1994 plan to
meet the newly adopted state
law requiring the use of
Environmental Site Design (BSD)
practices to minimize
stormwater runoff throughout
the county.
Development in one of the high
quality areas slated for
development, Ten Mile Creek
(TMC) (Figure 31), was afforded
special protection under the
Master Plan. TMC, a subwatershed25 of the Little Seneca Creek watershed, was assigned to stage four to
ensure that the 1994 development plan could be reviewed and potentially adjusted based on relevant
new data and information. This case example shows how the BCG was used to provide information on
current conditions in TMC relative to other county subwatersheds and streams in excellent, good, fair,
Figure 31. Ten Mile Creek, Maryland.
' A subwatershed is the topographic perimeter of a stream catchment.
121
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
or poor condition. Information from the BCG was used in conjunction with other data to help inform the
County Council in its deliberation on whether or not to adjust the stage four development plan.
6.1.2.1 Ten Mile Creek Subwatershed, Stream, and Tributaries
The TMC subwatershed, stream, and tributaries comprise a headwater stream system in which the
majority of tributaries are small and spring fed. Abundant springs and seeps supply cold and clean water
that supports a diverse community offish, benthic macroinvertebrates, and amphibians (Boucher,
personal communication, 2014) (Figure 32). The area is highly forested with a low level of impervious
surface, < 1% to 3%. TMC is one of three reference watersheds remaining in the county and has
supported good to excellent conditions based on a long term county data set using IBIs for benthic
macroinvertebrates and for fish that were developed by the county (MCDEP 2012). TMC and its
tributaries are adjacent to both Little Bennett Creek, a natural resource conservation management area,
and to the county's agricultural reserve. The location of TMC provides not only a bridge between these
two protected areas, but also a cost efficient opportunity to maintain natural flows, clean water, and
high biological diversity, and provide for recreational use and appreciation by the public (Figure 33).
Figure 32. Important aquatic species in Maryland's Piedmont headwater streams. Salamanders (Long-tailed,
Northern Dusky, and Northern Red); fish (Potomac Sculpin, Rosyside Dace, American Eel); insects (Sweltsa,
Paraleptophlebia, Ephemerella).
122
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Legend
Interstate 270
Clarkesburg Boundary
Agricultural Reserve
Little Seneca Lak
Ten Mile Creek
NAD_1983_StatePlane_Maryland_FIPS_1900_Feet
Map Produced 07-28-2015
Figure 33. Clarksburg Area and Ten Mile Creek Subwatershed.
123
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.1.2.2 Monitoring the Impacts of Development
Beginning in 1994, the Montgomery County Department of Environmental Protection (MCDEP)
monitored conditions throughout the Clarksburg development area as construction progressed. Analysis
included evaluating the effectiveness of BMPs and regulations to minimize both the immediate impacts
from construction and the longer term impacts from the subsequent development. Annual monitoring
reports were published beginning in 2001 (e.g., MCDEP 2009, 2012). Initial monitoring found stream
conditions in the Clarksburg development area ranged from good to excellent in most sensitive, high
quality areas such as the TMC subwatershed. However, by the mid-2000s, the water quality at several
good quality streams in the urbanizing areas began to degrade from good to fair (MCDEP 2009, M-
NCPPC 2014a). In October 2012, the Montgomery County Council directed the County Planning Board to
undertake a limited amendment of the 1994 Clarksburg Master Plan. Monitoring of earlier Clarksburg
developments showed uncertainty about the ability to protect the sensitive environmental resources
found in the stage four development area, such as TMC subwatershed, if full development were to occur
according to the original 1994 plan.
A number of scientific analyses informed the development of the Ten Mile Creek Area Limited
Amendment to the Clarksburg Master Plan and Hyattstown Special Study Area. County staff sought to
use their extensive monitoring data to further characterize the watershed and to identify analytical ways
to present information on the environmental status of County waters. Specifically, staff wanted to
assess the current conditions in those waters and the expected changes that would occur in relation to
further development in the area. In an effort to further characterize and assess incremental changes in
local biological conditions, in 2013 the County embarked on the process of developing a BCG model for
the Piedmont region of Maryland using both county and state data for fish and benthic
macroinvertebrate assemblages (USEPA 2013b). Observations on the presence of salamanders were also
incorporated where data were available. The presence of stream salamander species such as the
northern dusky salamander, long tailed salamander, northern two-lined salamander, and the northern
red salamander aided in confirming the high quality of streams.
6.1.3 Development of the Biological Condition Gradient
The County saw the BCG as one way to provide more detailed information on streams and their
response to land use change. In 2013, scientists from agencies within the state, Delaware, Pennsylvania,
Virginia, EPA, consulting groups, and academia convened as an expert panel to develop a BCG for the
Northern Piedmont. The goal of this effort was to use data collected primarily from Montgomery County
to develop a BCG model to describe changes in the biota in response to increasing stress in the
landscape. For example, a BCG level 2 stream would be minimally disturbed and include the presence of
native top predator fish (e.g., brook trout) as well as mayflies, stoneflies, and caddisflies. A BCG level 3
or 4 stream would include incrementally higher loss of sensitive species and an increased abundance of
tolerant species (e.g., blacknose dace and northern two-lined salamander). A BCG level 5-6 stream
would show an abundance of highly tolerant species (e.g., brown bullhead, tubificid and naidid worms).
Experts at the workshop were able to distinguish five distinct levels of biological condition for the
Piedmont region within Montgomery County (BCG levels 2-6). There were no BCG level 1 sites. Most
TMC sites ranged from a level 3+ to a level 4, although several sites (e.g., primarily headwater streams)
were judged as very good quality (a level 2 rating). Narrative and numeric decision rules to consistently
describe and quantify site assessments were developed based on mathematical set theory using the
fuzzy logic method (Table 28, Table 29, Table 30) and taxa response relationships derived from the
county data sets (Figure 34).
124
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 28. Description offish, salamander, and macroinvertebrate assemblages in each assessed BCG level. Definitions are modified after
Davies and Jackson (2006).
Definition: Natural or native condition—native structural, functional and taxonomic integrity is preserved; ecosystem function is preserved within the range of
natural variability
BCG level
1
Narrative from expert panel: There are no BCG level 1 sites within the Piedmont. All sites have some degree of disturbance, including legacy effects from
agriculture and forestry from 100 to 200 years ago. Conceptually, BCG level 1 sites would have strictly native taxa for all assemblages evaluated (fish, salamander,
benthic macroinvertebrates), some endemic species, and evidence of connectivity in the form of migratory fish.
Fish: Examples of endemic species that might be present (depending on the size of the stream) include: Bridle Shiner, Brook Trout, Chesapeake Logperch, Maryland
Darter, Trout Perch
Macroinvertebrates: Sensitive-rare, coldwater indicator taxa such as the mayfly Epeorus, and stoneflies Sweltsa and Talloperla are expected to be present
Definition: Minimal changes in structure of the biotic community and minimal changes in ecosystem function—virtually all native taxa are maintained with some
changes in biomass and/or abundance; ecosystem functions are fully maintained within the range of natural variability
BCG level
2
Narrative from expert panel: Overall taxa richness and density is as naturally occurs (watershed size is a consideration). These sites have excellent water quality
and support habitat critical for native taxa. They have many highly sensitive taxa and relatively high richness and abundance of intermediate sensitive-ubiquitous
taxa. Many of these taxa are characterized by having limited dispersal capabilities or are habitat specialists. If tolerant taxa are present, they occur in low numbers.
There is connectivity between the mainstem, associated wetlands and headwater streams.
Fish: Highly sensitive (attribute II) and intermediate sensitive (attribute III) taxa such as yellow perch, northern hog sucker, margined mad torn, fallfish and fantail
darter are present, as are native top predators (e.g., brook trout). Migratory fish and amphibians (e.g., eel, lamprey, salamanders) are present or known to access
the site. Long-tailed and Dusky salamanders are also good indicators, given a complimentary fish community. Non-native taxa such as brown trout or rainbow
trout, are absent or, if they occur, their presence does not displace native trout or alter structure and function.
Macroinvertebrates: Highly sensitive taxa are present—especially coldwater indicator mayflies, stoneflies, and caddisflies (e.g., Epeorus, Paraleptophlebia,
Sweltsa, Tallaperla, and Wormaldia)—and occur in higher abundances than in BCG level 3 samples.
125
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Definition: Evident changes in structure of the biotic community and minimal changes in ecosystem function—Some changes in structure due to loss of some rare
native taxa; shifts in relative abundance oftaxa but intermediate sensitive taxa are common and abundant; ecosystem functions are fully maintained through
redundant attributes of the system
Narrative from expert panel: Generally considered to be in good condition. Similar to BCG level 2 assemblage except the proportion of total richness represented
by rare, specialist and vulnerable taxa is reduced. Intermediate sensitive-ubiquitous taxa have relatively high richness and abundance. Taxa with intermediate
tolerance may increase but generally comprise less than half total richness and abundance. Tolerant taxa are somewhat more common but still have low
abundance. Taxa with slightly broader temperature or sediment tolerance may be favored.
BCG level
3
Fish: Intermediate sensitive (attribute III) taxa such as fallfish and fantail darter are common or abundant. Taxa of intermediate tolerance (attribute IV) such as
channel catfish, least brook lamprey, pumpkinseed and tessellated darter are present in greater numbers than in BCG level 2 samples. Some tolerant (attribute V)
taxa such as mummichog and white suckers may be present, but highly tolerant taxa are absent. Pioneering species such as blacknose dace, creek chubs and white
suckers may be naturally common in smaller streams. Migratory species such as American Eel may be absent. Two-lined salamanders may occur.
Macroinvertebrates: Similar to BCG level 2 assemblage except sensitive taxa (e.g., Sweltsa, Tallaperla and Wormaldia) occur in lower numbers. Level 3 indicator
taxa include the caddisfly Diplectrona, the mayfly Ephemerella and the stonefly Amphinemura.
Definition: Moderate changes in structure of the biotic community and minimal changes in ecosystem function—Moderate changes in structure due to
replacement of some intermediate sensitive taxa by more tolerant taxa, but reproducing populations of some sensitive taxa are maintained; overall balanced
distribution of all expected major groups; ecosystem functions largely maintained through redundant attributes
Narrative from expert panel: Sensitive species and individuals are still present but in reduced numbers (e.g., approximately 10%-30% of the community rather
than 50% found in level 3 streams). The persistence of some sensitive species indicates that the original ecosystem function is still maintained albeit at a reduced
level. Densities and richness of intermediate tolerance taxa have increased compared to BCG level 3 samples.
BCG level
4
Fish: 2 or 3 sensitive taxa may be present but occur in very low numbers (e.g., Blue Ridge Sculpin, Fantail Darter, Potomac Sculpin, Fallfish, Rosy-side Dace, River
Chub). Taxa of intermediate tolerance (attribute IV) such as tesselated darter, least brook lamprey, longnose dace are common, as well as tolerant taxa like yellow
bullhead, red-breast sunfish and bluntnose minnow. Level 4 streams may harbor two to three salamander species (Dusky, Red, and Two-lined).
Macroinvertebrates: Sensitive taxa (including EPT taxa) are present but occur in low numbers. Taxa such as Diplectrona and Dolophilodes may occur, but other key
taxa such as Ephemerella and Neophylax are absent. Taxa of intermediate tolerance (e.g., Baetis, Stenonema, Caenis, Chimarra, Cheumatopsyche, Hydropsyche)
occur in greater numbers. Tolerant taxa such as Chironomini and Orthocladiinae are present but do not exhibit excessive dominance.
126
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Definition: Major changes in structure of the biotic community and moderate changes in ecosystem function— Sensitive taxa are markedly diminished;
conspicuously unbalanced distribution of major groups from that expected; organism condition shows signs of physiological stress; system function shows reduced
complexity and redundancy; increased build-up or export of unused materials
BCG level
5
Narrative from expert panel: Overall abundance of all taxa reduced. Sensitive species may be present but their functional role is negligible within the system.
Those sensitive taxa remaining are highly ubiquitous within the region and have very good dispersal capabilities. The most abundant organisms are typically
tolerant or have intermediate tolerance, and there may be relatively high diversity within the tolerant organisms. Most representatives are opportunistic or
pollution tolerant species.
Fish: Facultative species reduced or absent. Tolerant taxa like yellow bullhead, red-breast sunfish, and bluntnose minnow are common. Blacknose dace, creek
chubs and white suckers may dominate. Two-lined salamanders might be the only salamander present.
Macroinvertebrates: Highly sensitive macroinvertebrate taxa are usually absent and Chironomid midges (mostly tolerant Orthocladiinae and Chironomini) often
comprised > 50% of the community in level 5 streams.
Definition: Major changes in structure of the biotic community and moderate changes in ecosystem function—Sensitive taxa are markedly diminished;
conspicuously unbalanced distribution of major groups from that expected; organism condition shows signs of physiological stress; system function shows reduced
complexity and redundancy; increased build-up or export of unused materials
BCG level
6
Narrative from expert panel: Heavily degraded from urbanization and/or industrialization. Can range from having no aquatic life at all or harbor a severely
depauperate community composed entirely of highly tolerant or tolerant invasive species adapted to hypoxia, extreme sedimentation and temperatures, or other
toxic chemical conditions.
Fish: Fish are low in abundance or absent, represented mainly by blacknose dace, green sunfish, bluntnose minnow, or creek chub.
Macroinvertebrates: May be dominated by tolerant non-insects (Physid snails; Planariidae; Oligochaeta; Hirudinea; etc.)
127
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 29. BCG quantitative decision rules for macroinvertebrate assemblages. The numbers in
parentheses represent the lower and upper bounds of the fuzzy sets.
BCG Level 2
# Total taxa
% Attribute II taxa
% Attribute ll+lll taxa
% Attribute II individuals
% Attribute ll+lll individuals
% Attribute V individuals
rule
> 17 (13-22)
> 8% (5-10)
> 50% (45-55)
> 3% (2-5)
> 60% (55-65)
< 15% (10-20)
BCG Level 3 alt 1
# Total taxa
% Attribute ll+lll individuals
# Attribute II taxa
% Attribute ll+lll taxa
% Attribute V individuals
% Most dominant Attribute V individual
BCG Level 4
# Total taxa
% Attribute ll+lll taxa
% Attribute ll+lll individuals
% Attribute V individuals
% Most dominant Attribute V individual
BCG Level 5
# Total taxa
% Attribute V individuals
% Most dominant Attribute V individual
alt 2
> 17 (13-22)
> 40% (35-45)
—
> 25% (20-30)
< 40% (35-45)
< 20% (15-25)
> 1 (0-2)
> 45% (40-50)
< 50% (45-55)
—
rule
> 15 (10-20)
> 20% (15-25)
> 10% (5-15)
< 70% (65-75)
< 60% (55-65)
rule
> 8 (6-10)
< 85% (80-90)
< 70% (65-75)
128
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 30. BCG quantitative decision rules for fish assemblages in small (0.5-:
numbers in parentheses represent the lower and upper bounds of the fuzzy
luxilus, clinostomus, and cyprinella, minus swallowtail shiners.
1.4 mi2), medium (1.5-7.9 mi2) and larger streams (> 8 mi2). The
sets. The mid-water cyprinid taxa metric is comprised of notropis,
BCG Level 2
# Attribute 1 taxa
# Attribute l+ll taxa
# Attribute l+ll+lll taxa
# Sensitive salamander taxa (if surveyed)
% Attribute l+ll+lll taxa
% Attribute l+ll+lll individuals
# Attribute Vlt taxa
% Attribute Vlt individuals
# Attribute X taxa
BCG Level 3
# Attribute l+ll taxa
# Attribute l+ll+lll taxa
% Attribute l+ll+lll taxa
% Attribute l+ll+lll individuals
% Attribute V individuals
# Attribute Vlt taxa
% Attribute Vlt individuals
# Mid-water cyprinid taxa
BCG Level 4
# Attribute l+ll+lll taxa
% Attribute l+ll+lll individuals
% Most dominant Attribute Va or Vlt individual
BCG Level 5
# Total taxa
# Total individuals
% Attribute V+VIt taxa
% Attribute V+VIt individuals
Small
rule alt rule
>0 (present)
-
> 1 (0-3)
>0
> 35% (30-40)
-
< 2 (1-3)
< 5% (3-7)
-
Small
-
> 2 (0-4)
-
> 25% (20-30)
-
< 2 (1-4)
< 15% (10-20)
>0
Small
> 1 (0-3)
> 5% (3-7)
< 65% (60-70)
Small
>4(3-6)
> 100 (90-110)
-
-
Medium
rule alt rule
>0 (present)
>2(1^)
-
>0
> 35% (30^0)
> 50% (45-55)
<2(l-3)
< 5% (3-7)
>0
Medium
-
-
> 25% (20-30)
> 25% (20-30)
-
<2(1^)
< 15% (10-20)
>1
Medium
> 1 (0-3)
> 10% (7-13)
< 65% (60-70)
Medium
>4(3-6)
> 100 (90-110)
< 65 (60-70)
< 90 (85-95)
Large
rule
-
> 4 (2-6)
-
-
> 35% (30-40)
> 50% (45-55)
< 2 (1-3)
< 5% (3-7)
>0
Large
> 1 (0-2)
-
> 25% (20-30)
> 25% (20-30)
< 40% (35-45)
-
< 15% (10-20)
>1
Large
> 1 (0-3)
> 10% (7-13)
< 65% (60-70)
Large
> 4 (3-6)
> 100 (90-110)
< 65 (60-70)
< 90 (85-95)
129
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
100
•= 80
co
CM
.Q
'i_
<
60
40
20
D
1
_
03
T3
ID
.0
I
IUU
80
60
40
20
0
H
0
• T
I •
} ! 1
i : i
23456
Figure 34. Box plots of sensitive (attribute ll+lll) and tolerant (attribute V) percent taxa and percent individual
metrics for macroinvertebrate calibration samples, grouped by nominal BCG level (expert consensus) (Source:
Stamp et al. 2014).
Additional expert panel findings include:
• One headwater site within the TMC watershed (King Spring) was identified as a high quality
stream (BCG level 2) with taxa comparable to streams in the adjacent regional park (Little
Bennett Regional Park) and with State of Maryland Sentinel Sites for the Piedmont region
(Figure 35). Impervious cover for these BCG level 2 sites was at 3% or less. Three other TMC sites
with impervious cover ranging between 4% and 11% were rated between BCG levels 3 and 4
(lower condition but considered comparable to "good to fair" conditions). The sites that were
approaching BCG level 4 were considered by the experts as candidates for cost effective
restoration.
• Sites within the TMC watershed having higher levels of impervious surface were assessed as
lower quality. These more degraded sites had elevated levels of specific conductance, an
indicator of urban runoff. However, tributaries in excellent to good condition, like King Spring,
diluted specific conductance in the lower mainstem TMC.
• Sites within the Piedmont with levels of impervious surface typically higher than 4% showed
increasingly degraded aquatic communities. Figure 36 shows average BCG level assignment for
benthic macroinvertebrate sampling sites with % sensitive species plotted against % impervious
surface. Increased level of impact on the aquatic biota can also be caused by confounding and
synergistic effects of other stressors. Additionally, the degree of degradation can be moderated
by implementing BMPs. These two considerations likely account for the observed scatter.
• Across Montgomery County both fish and benthic macroinvertebrate assemblages are assessed
and may show divergent ratings of condition because of different responses to type and
mechanistic pathway of stressors. In some instances, the experts assigned lower condition
ratings for the fish community, because there were no or fewer than expected native species.
This result was generally attributed to prevention of native fish migration due to dams and other
obstacles. Additionally, there was evidence of intrusion of lake fish species from reservoirs so
that lake species were dominant over the expected stream species. However, there was
sufficient fish habitat and food supply (the benthic macroinvertebrates) to support re-
130
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
introduction of native species or migration of other species, such as eel. Depending upon
existing temperature regimes, these sites might be excellent sites for re-introduction of native
and migratory species.
The decision rules were considered by experts to be applicable to the larger Piedmont region and with
minor modification to reflect climate and other latitudinal gradients, useful for assessing biological
condition in Piedmont regions in Virginia, Delaware and Pennsylvania.
O
O
u
_o
O
in
Natural structural, functional, and taxonomic integrity is preserve
Minimal changes in both structure and function
Bennett Creek
TMC: King Spring
Evident (e.g., measurable) changes in structure,
minimal changes in function
Moderate changes in structure
\and evident changes in function
Major changes in structure an
moderate changes in function
Turkey Bridge
Severe changes in structure and function
Upper Sligo and Breewood Tributary
Level of Stressors
Low-
High
Figure 35. Comparative BCG ratings of macroinvertebrate community data from the county monitoring data set
for streams in the TMC watershed and comparable county streams in other watersheds. Data from streams in
the State of Maryland Piedmont Sentinel data set were also rated by the experts. The sites were mapped on the
gradient according to the expert-derived decision rules for assigning sites to BCG levels.
131
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
01
"c
'in
C/3
gj
i-
o
o
m
a)
8 o
L« o°
1
\
_L
10
50 60
20 30 40
% Impervious Cover
20 30 40
% Impervious Cover
Figure 36. Relationship between average BCG level assignments (left) and % Sensitive Taxa (right) versus %
impervious cover. This analysis included sites from throughout the Piedmont Region in Maryland. Ten Mile
Creek sites are indicated (red dots).
6.1.4 Use of the Biological Condition Gradient Model in County Planning Decisions
Based on the findings in the environmental analyses associated with the proposed Limited Amendment,
the County planning staff and MCDEP scientists concluded that there was significant uncertainty
whether high quality aquatic resources assigned special protection, such as TMC subwatershed and
streams, would be protected under the 1994 plan. The county planning and MCDEP staff provided
several possible development scenarios with predicted outcomes and recommended one option that
would modify development in the TMC area while maintaining good environmental conditions (M-
NCPPC 2014b). The County Council accepted the recommended option, and it was adopted on April 1,
2014.
The BCG was used in conjunction with expert testimony, peer reviewed literature, research, modeling,
and the environmental analysis to inform the County's decision to adopt the 2014 Limited Amendment
for Clarksburg. This amendment revised zoning restrictions outlined in the 1994 Master Plan to reduce
the impact of development on TMC. The 1994 Master Plan allowed a total impervious cap of 9.8%, while
the Limited Amendment proposed a 6.3% impervious surface cap for new development in the most
sensitive subwatersheds but allowed a maximum of 15% impervious cover in the Town Center District.
The amendment also included a recommendation for increasing forest cover to 65% of the watershed
and increasing the size of riparian buffers to better protect the streams and tributaries (M-NCPPC
2014b).
In 2014, the Montgomery County Council adopted the Limited Amendment to the 1994 Clarksburg
Master Plan, which focused on TMC. The 2014 Limited Amendment concluded that TMC "warrants
extraordinary protection," and offered recommendations for additional zoning restrictions that would
allow for continued development, while continuing to study how development and mitigation activities
(e.g., implementation of BSD) might affect sensitive water resources in the TMC watershed (M-NCPPC
2014a). The most sensitive streams or tributaries in the TMC system, such as King Spring, are currently
132
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
at less than 1% impervious cover, so a cap of 6% will likely result in loss of some sensitive species and
change from excellent to good, or potentially fair, condition depending on what other development
activities occur or protective measures are put in place. For example, the amendment provides for
consideration of additional measures (e.g., expanded stream buffer protections) and technology (e.g.,
BSD) that might minimize these changes (M-NCPPC 2014a). The use of the BCG in conjunction with other
data, information, and expert testimony, successfully brought scientific information into the decision-
making process and provided for informed decision making that balanced multiple public and private
concerns and priorities.
6.1.5 Lessons Learned
Montgomery County found that the BCG framework was a good tool to better articulate current
conditions in TMC and illustrate how water quality could be impacted by future development as
outlined in the 1994 Master Plan. The 2014 Limited Amendment will allow for continued development
with some restrictions on impervious cover. Because the BCG can be used in conjunction with
monitoring data to detect incremental changes in stream health, county scientists will be able to closely
monitor the effects of using BSD and other BMPs to mitigate the impacts of development on sensitive
waters. County officials found that the BCG gave experts and the public a common understanding of
water quality issues and allowed for informed policy making.
In the future, the County plans to use the BCG as an interpretative framework to examine restored sites
and identify incremental improvements or declines in biological condition. Future use of this
information might also include using county data for restoration modeling. In addition, the BCG might
be used as one way to reconcile databases maintained at the County-level with those at the state level.
Ultimately, one goal of such an effort could be to have county-level data used by the state when
classifying streams.
133
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.2 Pennsylvania: Using Complementary Methods to Assess Biological Condition
of Streams
6.2.1 Key message
Pennsylvania Department of Environmental Protection (PA DEP) implements a multi-tiered benchmark
decision process for assessing attainment of ALL) for wadeable, freestone, riffle-run streams in
Pennsylvania. This multi-tiered approach incorporates stream size and sampling season as factors for
determining ALL) attainment based on benthic macroinvertebrate sampling. A BCG calibrated for
freestone, riffle-run streams is used to supplement the state's primary screening tool, the IBI for benthic
macroinvertebrates (PA DEP 2013a).
6.2.2 Using Index of Biological Integrity to Assess Aquatic Life Uses
PA DEP has developed a multimetric benthic macroinvertebrate IBI for the wadeable, high gradient,
freestone26 streams in Pennsylvania using the reference condition approach (PA DEP 2012). These
streams are non-calcareous, or softwater, free flowing streams and comprise the majority of the state's
streams. PA DEP has alternative assessment methods in place for other stream types (i.e., low-gradient
pool-gliders, karst- [limestone]-dominated). The IBI provides an integrated measure of the overall
condition of a benthic macroinvertebrate community in a water body by combining multiple metrics into
a single index value. A number of different metric combinations were evaluated during IBI development.
Based on discrimination efficiencies, correlation matrix analyses, and other index performance
characteristics, PA DEP selected the following six metrics for inclusion as core metrics in the MMI (PA
DEP 2012):
1. Total Taxa Richness—This taxonomic richness metric is a count of the total number of taxa in a
subsample. Generally, this metric is expected to decrease with increasing anthropogenic stress
to a stream ecosystem, reflecting loss of taxa and increasing dominance of a few pollution-
tolerant taxa.
2. Ephemeroptera + Plecoptera + Trichoptera Taxa Richness (EPT)—This taxonomic richness
metric is a count of the number of taxa belonging to the orders Ephemeroptera, Plecoptera, and
Trichoptera in a sub-sample—common names for these orders are mayflies, stoneflies, and
caddisflies, respectively. The aquatic life stages of these three insect orders are generally
considered sensitive to, or intolerant of, many types of pollution (Lenat and Penrose 1996),
although sensitivity to different types of pollution varies among specific taxa in these insect
orders. This metric is expected to decrease in value with increasing anthropogenic stress to a
stream ecosystem, reflecting the loss of taxa from these largely pollution-sensitive orders.
3. Beck's Index—This taxonomic richness and tolerance metric is a weighted count of taxa. The
name and conceptual basis of this metric are derived from the water quality work of William H.
Beck in Florida (Beck 1955). This metric is expected to decrease in value with increasing
anthropogenic stress to a stream ecosystem, reflecting the loss of pollution-sensitive taxa.
4. Shannon Diversity—This community composition metric measures taxonomic richness and
evenness of individuals across taxa in a sub-sample. This metric is expected to decrease in value
with increasing anthropogenic stress to a stream ecosystem, reflecting loss of pollution-sensitive
26 Freestone is a term familiar to fly-fisherman, denoting streams with little groundwater influence showing high
annual variation in flow (spring freshet, summer drought).
134
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
6.
taxa and increasing dominance of a few pollution-tolerant taxa. The name and conceptual basis
for this metric are derived from the information theory work of Claude Elwood Shannon
(Shannon 1948).
Hilsenhoff Biotic Index—This community composition and tolerance metric is calculated as an
average of the number of individuals in a sub-sample, weighted by pollution tolerance values.
Developed by William Hilsenhoff, the Hilsenhoff Biotic Index (Hilsenhoff 1977, 1987a, 1987b,
1988; Klemm et al. 1990) generally increases with increasing ecosystem stress, reflecting
increasing dominance of pollution-tolerant organisms.
Percent Sensitive Individuals—This community composition and tolerance metric is the
percentage of individuals in a sub-sample and is expected to decrease in value with increasing
anthropogenic stress to a stream ecosystem, reflecting loss of pollution sensitive organisms.
PA DEP determined that these six metrics all exhibited a strong ability to distinguish between reference
and stressed conditions in testing with benthic invertebrate assemblage data from riffle run habitats in
wadeable, freestone streams. When used together in an MMI, these metrics provide PA DEP with a
consistent and defensible index for assessing the biological condition of these streams (PA DEP 2012).
6.2.3 Use of the Biological Condition Gradient to Complement Aquatic Life Use
Assessments
PA DEP is exploring use of a BCG to describe the
biological characteristics of wadeable, freestone
streams along a gradient of stress. More than 75%
of Pennsylvania is in the hills and low mountains of
the Appalachian Highlands, so streams throughout
the state are predominantly relatively high gradient
(> 1% slope) (Figure 37 and Figure 38). Pennsylvania
is largely forested, but there are significant areas
where agricultural land use, including row-crops
and pasture, is dominant (Figure 39). Limestone
and spring-dominated streams occur in parts of
southeast, south-central and east-central
Pennsylvania. The BCG assessments and model
discussed in the case study do not apply to this
subset of streams.
Between 2006 and 2008, PA DEP conducted a
series of expert workshops to calibrate a BCG along
a gradient from minimally to heavily stressed
conditions (PA DEP 2013b). To develop the BCG for
the wadeable, freestone streams, biologists from
PA DEP, in conjunction with external taxonomic
experts and scientists (e.g., the Delaware River
Basin Commission, Western Pennsylvania Rgure 37 Top. Carbaugh Run, Adams County.
Conservancy, and EPA), used the BCG attributes Bottom: Rock Run, Lycoming County (Photos courtesy
that characterize specific changes in community Of p^ DEP).
taxonomic composition (PA DEP 2013b). For
example, in the highest levels of the BCG, locally
135
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
endemic, native, and sensitive taxa are well represented, and the relative abundances of pollution-
tolerant organisms are typically lower. With increasing stress, more pollution-tolerant species may be
found with concurrent loss of pollution-sensitive species. At the beginning of the expert workshop, the
participants assigned a BCG attribute for sensitivity to stress (i.e., attributes I-V) to each
macroinvertebrate taxon based on expert knowledge and biological response data. The data used was
from sampling sites that spanned a range of condition from reference quality (e.g., at or close to
minimally disturbed conditions) to heavily stressed sites (PA DEP 2013b). Using the BCG level
descriptions of predicted changes in the attributes as a guide, the expert panel then assigned each site
to one of the six BCG levels and developed candidate decision rules (Figure 40, Table 31).
Figure 38. Topographic Map of Pennsylvania.
Legend
2011 NLCD Landcover Categories f
^m Barren Land |
j^B Cultivated Oops
^^| Deciduous Fores?
HI Developed Higti Intensity
^^ Developed Low Intensity
^^| Developed. Medium Intensi
] Developed. Open Space
Herbaceuous Wetlands
[ | Hay.'Pasture
| Hefbaceuous
| Mi*ed Forest
^^| Open Water
| Snrutj.'Scnjb
j wooay Wetlands
Figure 39. Pennsylvania Land Use.
136
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
3456
Nominal Tier
(0
X
ffl
-I-*
~
0
3
-Q
T
s
IH
12
10
8
4
2
0
T
r^n ^
- |D|
T
n
T fi
-D- D •
3456
Nominal Tier
18
ro 16
x 14
^ 12
= 10
(D Q
_i_j O
s
1
3456
Nominal Tier
Sensitive taxa
ou
25
20
15
10
5
0
O
R
T 1
R
@
5 o .
3456
Nominal Tier
ro
X
"0
"^
_Q
1
it
12
10
8
6
4
2
o
.1 T I JL .
.[ U
nH ^ Hi
1
n
0
. . i . .
23456
Nominal Tier
23456
Nominal Tier
Figure 40. Box plots of BCG metrics, by nominal level (group majority choice). Sensitive taxa are the sum of both
attribute II (highly sensitive) and attribute III taxa (intermediate sensitive) (Source: Gerritsen and Jessup 2007c).
137
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 31. Potential narrative decision rules for invertebrate samples from Pennsylvania high gradient
streams (modified from Gerritsen and Jessup 2007c)
Attributes
All Taxa
1. Historically
documented,
sensitive, long-
lived, or regionally
endemic taxa
II. Highly sensitive
taxa
III. Intermediate
sensitive taxa
IV. Intermediate
tolerant taxa
V. Tolerant taxa
Indicator taxa
BCG Level
2
> 25 taxa
3
> 20 taxa
4
5
> 10 taxa
No single taxon >
50% of abundance
> 50 individuals in
sample
6
Low richness or
low abundance
No rules determined for attribute 1
Taxa II > 33% of
Taxa III
Taxa (11 + III) >
50% of all taxa
lndiv(ll + lll) >
50% of all indiv
Taxa II present (>
0)
Taxa (11 + III) >
30% of all taxa
lndiv(ll + lll)>
30% of all indiv
May be absent
(no rule)
Taxa (II + 111)
present (> 10% of
taxa, or 2 taxa)
Indiv (11 + III) >
15%-20%of all
indiv
No rules determined for attribute IV
Few tolerant taxa;
Tolerants are
small % of total
abundance (< 5%)
Many EPT taxa;
EPT > 15
Tolerant
individuals < 20%
of total
abundance
Tolerant
Caddisflies <
20% abundance
EPT > 12
Tolerant
individuals < 40%
of total
abundance
Tolerant
Caddisflies <
40% abundance
EPT>8
Tube worms not
dominant; < 50%
of abundance
Tolerant
individuals may
dominate
Mayflies maybe
absent;
Tube worms may
dominate
Each sampling site used to develop and test the BCG decision rules had corresponding IBI scores. The IBI
uses metrics that are similar in objective to the BCG attributes, but which are calculated differently (PA
DEP 2013a). The total IBI score is based on the sum or average of all metrics, while BCG decision rules
are based on specific attribute groups and patterns of change along a gradient of stress (e.g., attributes
II and III for the higher levels and attribute V for lower levels).
For all the evaluated samples, PA DEP biologists analyzed the relationship between a sample's BCG level
assignment with its corresponding IBI score (PADEP 2013b). A strong correlation existed between the
calibrated BCG level assignments and the IBI scores (Figure 41). On the basis of this comparative
analysis, PA DEP determined that with further testing and evaluation, the IBI scores could potentially be
used to discriminate BCG levels. PA DEP is evaluating using the BCG to describe the biological
characteristics of streams assessed based on the IBI scores; for example, the reference sites clustered at
IBI scores near 80 and above would be interpreted as primarily comparable to BCG levels 1-2. On the
basis of taxonomic information, and without knowledge of the IBI scores, the experts assigned these
sites to BCG levels 1.5 to 2.5. BCG level 2 represents close to natural conditions (e.g., minimal changes in
structure and function relative to natural conditions; supports reproducing populations of native species
of fish and benthic macroinvertebrates). This information can meaningfully convey to the public the
138
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
biological characteristics of waters in the context of the CWA and the goal to protect aquatic life. PA DEP
is evaluating use of the BCG to complement the IBI in assessing ALL) attainment and to help identify
potential high-quality (HQ) or exceptional value (EV) streams. As a first step in application of the BCG, PA
DEP has incorporated BCG attributes for taxa sensitivity to stress as part of its protocol for wadeable,
freestone streams (Figure 42) (PA DEP 2013a).
(D
0
0
m
100 —
90 —
80 —
70 —
60 —
50 —
40 —
30 —
20 —
10 —
0 —
o Q, o A: reference
^ooa,pD
D A D
* ^
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Stream size
i
1st, 2nd, or 3rd Order
Drainage Area < 25 mi2
i
r
3rd, 4th, or 5th Order
Drainage Area 25 to 50 mi2
i
5th Order or Larger
Drainage Area > 50 mi2
Calculate Small-Stream IBI Score
Calculate Large-Stream IBI Score
Sample date
*
June Se
stember
October
v
r
Novemt
Aquatic life use
>er May
Exceptional Value
High Quality
s
Resample
Nov-May
Cold Water Fishes (CWF)
Trout Stocking (TSF)
Warm Water Fishes (WWF)
i
Cold Water Fishes (CWF)
Trout Stocking (TSF)
Warm Water Fishes (WWF)
IBI score
r \
Exceptional Value
High Quality
r >
r
Impaired
yes
Are mayflies, stoneflies, or caddisflies
absent from the sub-sample?
yes
Is the Beck's Index standardized score < 33.3 with the
Percent Sensitive Individuals standardized score < 25.0?
lno
Is the IBI score more
than 11 points
lower than the site
baseline IBI score?
yes
Is the ratio of BCG attribute I, II, III taxa to BCG attribute IV, V, VI taxa < 0.75 with the
ratio of BCG attribute I, II, III individuals to BCG attribute IV, V, VI individuals < 0.75?
Impaired
yes
Does the sub-sample show signatures of acidification year-round
(i.e., < 3 mayfly taxa with mayflies representing < 5% and Leuctra +
Amphinemura representing > 25% of individuals in the sub-sample)?
no
no
Attaining
Figure 42. Multi-tiered benchmark decision process for wadeable, freestone, riffle-run streams in Pennsylvania
(Modified from PA DEP 2013a). The ratio of BCG attributes for sensitive to tolerate taxa (i.e., attributes I, II, and
III to attributes IV, V, and VI) are included as part of attainment determination (see yellow box). Rules have not
been defined for attribute I and IV but these attributes are included in the assessment protocol if decision rules
are developed in the future and determined to be appropriate to include.
140
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.2.4 Potential Application to Support Aquatic Life Use Assessments and Protection of
High Quality Waters
Pennsylvania's regulations define waters of EV that are of unique ecological or geological significance.
EV streams are given the highest level of protection and constitute a valuable subset of Pennsylvania's
aquatic resources. To support protection of these waters, PA DEP is considering the use of a discriminant
analysis model to evaluate the relationship between condition of the watershed, a stream, and its
aquatic biota (e.g., the connection of riparian areas with a stream and the floodplain or the spatial
extent of stressors and their sources in the watershed). PA DEP is evaluating the use of a discriminant
model that incorporates measures of land use and physical habitat along with IBI scores and the BCG to
make distinctions between EV and HQ waters. PA DEP is also evaluating how to consider effects of
habitat fragmentation and spatial and temporal extent of stress. The results of this effort could
potentially support state water quality management decisions on where to target resources for
sustainable, cost-effective protection of EV waters and healthy watersheds. Through this work, PA DEP
can provide EPA valuable feedback on the technical development and potential program application for
BCG with specific focus on defining indicators for BCG attributes IX (spatial and temporal extent of
detrimental effects) and X (ecosystem connectance).
141
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.3 Alabama: Using the Biological Condition Gradient to Communicate with the
Public and Inform Management Decisions
6.3.1 Key Message
ADEM has strategically built a comprehensive biological monitoring program over the past four decades
and has, more recently, invested in developing BCGs for streams in all regions of the state. ADEM has
identified reference conditions in order to better characterize current water quality condition, and it has
built increasing capability in terms of data management. As ADEM's capabilities have evolved, it is
applying biological data, biological indices, and the BCG for a variety of management purposes, including
identification of high quality waters and waters that need restoration. As part of this process, ADEM has
improved its ability to communicate to the public on the condition of streams and to measure
incremental improvements in condition. Though the state is developing and applying the BCG and
biological assessments on a statewide basis, this case study reports on the development and application
of a BCG for the high gradient streams of Northern Alabama.
6.3.2 Program Development
Since 1974, ADEM has been monitoring its surface water quality, and the capabilities of the monitoring
program have evolved overtime. In 1997, ADEM first formalized a coordinated monitoring strategy to
outline its surface water quality monitoring efforts. Today, ADEM collects biological, chemical, and
physical data and uses those data to inform management decisions, including assessing the health of
state waters, determining whether those waters are meeting their designated uses, and identifying
impacts from a variety of sources (ADEM 2012).
ADEM continues to build its monitoring program to meet emerging data needs, and it is currently
evaluating the use of its biological data in new ways. ADEM conducted a preliminary critical elements
review27 of its biological assessment program in 2006 to assess the strengths of the technical program.
The review highlighted ADEM's efforts to that point, and it included recommendations for
enhancements relative to design, methodology, and execution for credible data as a basis of making
informed decisions regarding the ecological condition of Alabama's streams. The review resulted in a
recommendation that ADEM fully implement its monitoring strategy to accomplish a variety of goals,
including more complete development of reference conditions and site criteria, and development
and/or refinement of macroinvertebrate, fish, and diatom community assessment methods; ecological
attributes; response patterns; and indices along a continuous BCG scale. The review also highlighted the
need for an improved and enhanced database management system; improved technical capabilities to
carryout survey needs; statewide completion of monitoring unit delineation; and incorporation of up-to-
date land cover data sources.
Since the 2006 review, ADEM has continued to make improvements in the technical capabilities of its
biological assessment program. In 2008, ADEM used data collected in 1994-2005 to develop MMIs for
high and low gradient streams. The indices were used for site assessments with thresholds derived from
the reference distribution. At the same time, the biological database was updated to a new platform,
integrated into ADEM's centralized surface water database, Alabama Water-Quality Assessment and
27 For more information about Critical Element Review, see Biological Assessment Program Review: Assessing Level
of Technical Rigor to Support Water Quality Management (USEPA 2013, http://www.epa.gov/wqc/biological-
assessment-technical-assistance-documents-states-and-tribes. Accessed February 2016.)
142
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Monitoring Data Repository (ALAWADR), which houses chemical, physical, and biological data. In 2009,
the database was modified to calculate macroinvertebrate metrics and indices. Incorporating these tools
into ALAWADR assisted greatly in the development and testing of ecological attributes, stress-biological
response patterns, and indices along continuous BCG and stressor scales. In 2013, ADEM expanded the
effort to use data from the 1994-2011 period to incorporate additional reference site data to refine the
site classes, and MMIs (Jessup 2013). In these efforts, ADEM considered regional differences in
biological habitat and species distribution, and it found that variability was best explained using
ecoregions28 for classification. ADEM calibrated the MMIs to categorize water quality on a scale from
Very Good to Very Poor (Jessup 2013).
In a similar effort spear-headed by the Geological Survey of Alabama, ADEM and the Alabama
Department of Conservation and Natural Resources collaborated to develop statewide multimetric fish
community indices. In 2004, the Geological Survey of Alabama completed refinement of collection
methods developed by the Tennessee Valley Authority and established five site classes, or
ichthyoregions, primarily based on ecoregions and basins. Statewide MMIs were completed in 2011-
2012.
6.3.3 Index Development
As a result of the work and collaboration among state agencies discussed above, ADEM developed
biological indices for both macroinvertebrates and fish statewide. Assessment thresholds were
established for both assemblages using similar analytical methods though there were differences in site
classification and threshold delineation. First, similar regions for classification were identified for each
assemblage, but they were not identical (Figure 43). For site classification, the similarity of species
composition relative to ecoregions, drainage basins, and other natural site characteristics was analyzed.
Shared environmental variables associated with the ecoregional distinctions for both assemblages
included elevation, temperature, and percent cobble and boulder substrate. However, differences in
classification for the two assemblages were attributed to the dependence on drainage continuity for fish
migrations, whereas macroinvertebrates (especially insects) can move among drainages by flying during
adult stages.
Second, for benthic macroinvertebrates, candidate reference sites were identified based on
measurements of disturbances both at the site and in the landscape. A watershed disturbance gradient
(WDG) was calculated using land use coverage (e.g., percent urban, row crop, and/or pasture in the
catchment) and road density (Brown and Vivas 2005; ADEM 2005). Figure 44 shows broad land cover
patterns throughout the state. The 25th percentile of the WDG was used as the threshold for selecting
candidate reference streams. These reference streams experienced minimal to moderate levels of stress
and are considered "least disturbed" conditions (Stoddard et al. 2006). However, for some regions, land
use intensity as measured by the WDG was considerably higher and more widespread, reflecting
regional patterns in agricultural and urban land use. Reference streams in the regions with more
intensive development (e.g., higher WDG scores) generally had lower biological scores (e.g., benthic
macroinvertebrate scores) (Table 32). Figure 45 shows the range of land intensity scores in sites
assessed by ADEM, including reference sites.
28 Ecoregions describe areas with similar features related to geology, physiography, vegetation, climate, soils, land
use, wildlife, and hydrology.
143
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
. Southwestern Appalachians (68)
^B Piedmont (45) and Ridge & Valley (67)
i i Interior Plateau (71) & Transition Hills (65j)
^^| SE Plains (plains) (65 a.D.t.g.p) & Southern Coastal Plain (75}
| | SE Plains (hills) (65 dj.q)
Ichthyoregions
|^| Hills & Coastal Terraces
Plateau
^H Ridge & Val ey / Piedmont
^H Southern Plains
Tennessee Valley
Figure 43. Left: Macroinvertebrate site classes in Alabama; Right: Fish site classes in Alabama.
NLCD Land Cover Classification Legend
t Open Water
Perennial Ice/ Snow
Developed, Open Space
! Developed, Low Intensity
Developed. Medium Intensity
Developed, High Intensity
Barren Land (Rock/Sand/Clay)
i Deciduous Forest
rest
Evergreen
Mixed Forest
i Dwarf Scrub*
Shrub/Scrub
Grassland/Herbaceous
Sedge/Herbaceous*
Lichens*
Moss1
Pasture/Hay
Cultivated Crops
Woody Wetlands
| Emergent Herbaceous Wetlands
aska only
71 Interior Plateau
68 Southwestern Appalachians
67 Ridge and Valley
45 Piedmont
65 Southeastern Plains
Figure 44. Alabama land use/land cover map.
Table 32. Characterization of Reference Conditions Using WDG and the Alabama Macroinvertebrate
MMI for streams. WDG scores increase with level of land use activity.
Macroinvertebrate Site Class
Median Reference WDG Score
Benthic Macroinvertebrate MMI Score:
25th Quantile of Reference
144
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Macroinvertebrate Site Class
Interior Plateau
Southeastern Plains-Hills
Piedmont, Ridge & Valley
Southwest Appalachians
Southeastern Plains-Plains
Median Reference WDG Score
61
64
46
31
90
Benthic Macroinvertebrate MMI Score:
25th Quantile of Reference
43
47
69
58
45
Additionally, there are differences in how the two assemblage indices were scored and benchmarks
established. As described above, the benchmark for the macroinvertebrate index was based on a
reference condition approach. Reference sites were selected based on abiotic parameters that met
predetermined selection criteria and a 25% threshold was established (Table 32). However, for fish, the
range of index scores from all sites was divided into five condition categories: excellent, good, fair, poor,
and very poor (Figure 46). The thresholds between fish categories were selected to create a balanced
distribution of conditions among the sampled sites, with most samples in the fair category, and similar
numbers of excellent and good samples compared to poor and very poor samples (Figure 46; O'Neil and
Shephard 2011). Thus, the reference condition for macroinvertebrates and the excellent and good
categories for fish are not a one for one match. ADEM wanted to develop the BCG model and numeric
decision rules so that benthic macroinvertebrate and fish assemblage data could be mapped on the BCG
and interpreted against a uniform standard despite differences in sample collection and analysis.
145
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
25
i/>
01
±; 20
:. is
01
_Q
E 10
-
2°
01
-Q 10
14
12
.i 10
(7i
L.
OJ
SW Appalachians
(Reference • Non-reference
Piedmont, Ridge & Valley
• Reference • Non-reference
Interior Plateau
• Reference • Non-reference
.11
0
Figure 45. Frequencies of sites in ranked WDG categories (x-axis), distinguishing reference and non-reference
sites in each site class. Distributions are based on sites monitored in ADEM's biological assessment program.
WDG categories are numerically ranked with increased levels of stress. ADEM converted the WDG scores to
ranks 1-8, with lower numbers representing less disturbance.
146
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
60
50
40
30
20
10
0
Excellent ' Good ' Fair ' Poor ' Vitypoor
Poor ' Very poor
tallenl ' Good ' [lit ' Aw ' Very poor '
Figure 46. Frequency distribution of fish IBI condition categories for sites in the three ichthyoregions discussed in
this case study: the (A) Plateau; (B) Piedmont, Ridge, and Valley; and (C) Tennessee Valley site classes. The x-axis
is divided into five condition categories: excellent, good, fair, poor, and very poor.
6.3.4 The Biological Condition Gradient
In 2014, ADEM and Geological Survey of Alabama convened an expert panel of scientists from the state,
outside agencies, academia, and other research organizations. The charge to the expert panel was to
develop a quantitative BCG and to use the BCG to calibrate BCG-based indices for fish and
macroinvertebrate assemblages for wadeable streams in Alabama. The first phase of BCG development
was on wadeable streams in Northern Alabama in three ecoregions: the Interior Plateau, Southwest
Appalachian, and the Piedmont Ridge Valley ecoregions. This case study reports on these results. The
second phase of BCG development is underway for the coastal plain streams in central and southern
Alabama.
Wadeable streams in northern Alabama are higher gradient relative to streams in the coastal plains of
southern Alabama and tend to have a riffle habitat (Figure 47). Experts developed numeric decision
rules to predict the BCG level of a stream based on site classes for fish and macroinvertebrates (Jessup
and Gerritsen 2014). Models were then developed to replicate the expert decisions for assigning new
samples to BCG levels 2-6 without having to reconvene the expert panel. There were no sites in the data
set used to develop the BCG that the experts considered comparable to BCG level 1 (undisturbed), so
the experts conceptually described the expected biological community for a BCG level 1. The conceptual
description provided a shared, narrative starting point for assessing incremental changes from BCG level
1 to BCG level 6. The final modeled BCG levels correctly predicted the expert ratings of actual site data
for BCG levels 2-6 in 94% and 96% of cases for macroinvertebrates and fish, respectively.
147
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Figure 47. Example of range in typical northern Alabama streams with riffle-run habitat. Top: Hendriks Mill
Branch; Bottom: Hatchet Creek.
As the first step in developing the BCG model for northern Alabama streams, the benthic
macroinvertebrate and fish species were assigned BCG attributes corresponding to their prevalence and
sensitivity to disturbance. These characteristics were analyzed using abundance of individuals and
general additive models (GAMs) based on the capture probability of each taxon along the WDG scale.
Experts in the workgroup used the model results and their own experience to assign attributes to each
taxon. Taxa with steeply descending model slopes were sensitive to disturbance and were assigned
attributes II or III (e.g., highly and intermediate pollution sensitivity) based on the slope of the response
curve (e.g., capture probability) (e.g., Acroneuria in Figure 48). Taxa with flat slopes were found in a
variety of disturbance conditions and were assigned to BCG attribute IV (taxa of intermediate tolerance).
Taxa with increasing capture probabilities with increasing disturbance were assigned to BCG attribute V
(tolerant taxa) (e.g., Ferrissia in Figure 48). In the second step of the BCG process, the experts used the
attribute assignments in developing the decision rules for assigning sites to BCG levels (Table 33).
Acroneuria
Fem'ss/a|
CO
o
re
O
.
re
O
o
o
H ni| irq
10 30 100 300 1000
Human Disturbance Score
8
O
_
O
o
I
re
-o
.a
a:
- o
- o
T
10 30 100 300 1000
Human Disturbance Score
Figure 48. Taxa relative abundance and the GAM slope based on capture probabilities for Acroneuria
(Plecoptera: Perlidae; attribute III) and Ferrissia (Gastropoda: Ancylidae; attribute V).
148
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 33. Example of narrative and quantitative rules from Northern Alabama BCG: BCG level 2
narrative and quantitative rules for macroinvertebrates and quantitative rules for fish in northern
Alabama. Macroinvertebrate rules apply in all northern Alabama streams. Fish rules are applied by
site class (PLA, RVP, and TV) and stream size (Small and Large).
Narrative Macroinvertebrate Rules for BCG Level 2
The sample is considered a level 2 condition if:
The number of all taxa in the sample is greater than 50-60 taxa and
The number of highly sensitive (attribute II) taxa is greater than 6-10 taxa and
The percentage of sensitive (attribute ll+lll) taxa is greater than 35%-40% of all taxa and
The number of sensitive (attribute ll+lll) EPT taxa is greater than 10-18 taxa and
The percentage of individuals in the 5 most abundant taxa is less than 60%-70% and
The percentage of individuals in the most abundant 5 tolerant (attribute IV, V, VI) taxa is less than 45%-55% OR the number
of all taxa in the sample is greater than 70-80 taxa.
If any of these rules is not met at least half-way, the sample is level 3-6, depending on rules for those levels.
Macroinvertebrates: BCG Level 2
# Total taxa
# Attribute II taxa
% Attribute ll+lll taxa
# Attribute ll+lll EPT taxa
% individuals in the most dominant 5 taxa
% individuals in the most dominant 5 tolerant taxa
Quantitative Rule
> 55 (50-60)
> 8 (6-10)
> 40% (35-45)
> 14 (10-18)
< 65% (60-70)
< 50% (45-55) or Total Taxa > 75 (70-80)
Narrative Fish Rules for BCG Level 2
The sample is considered a level 2 condition if:
The number of all taxa in the sample is greater than 10-25 taxa in the PLA and RVP and
The number of highly sensitive (attribute l+ll) taxa is greater than 0-4 taxa and
The number of sensitive (attribute l+ll+lll) taxa is greater than 5-10 in large TV sites and
The percentage of sensitive (attribute l+ll+lll) taxa is greater than 10%-25% and
The percentage of sensitive (attribute l+ll+lll) individuals is greater than 5%-30% and
The percentage of tolerant (attribute V+Va+VI) individuals is less than 15%-30% in the PLA and RVP and
The percentage of the most abundant Va or VI individuals is less than 30%-40% in the TV.
If any of these rules is not met at least half-way, the sample is level 3-6, depending on rules for those levels.
Fish: BCG Level 2
# Total taxa
# Attribute l+ll taxa
# Attribute l+ll+lll taxa
% Attribute l+ll+lll taxa
% Attribute l+ll+lll individuals
% Attribute V+Va+VI
individuals
% Most dominant Attribute Va
or VI individuals
PLA
Small Large
> 15 (10-20) > 20 (15-25)
>2(1^)
—
> 20% (15-25)
> 25% (20-30)
< 25% (20-30)
—
RVP
Small Large
>15 (10-20) > 20 (15-25)
> 0(0-1) >2(1^)
—
> 15% (10-20)
> 20% (15-25)
< 20% (15-25)
—
TV
Small Large
—
> 1 (0-3) > 2 (1-4)
— > 7 (5-10)
> 20% (15-25)
> 10% (5-15)
—
< 35% (30-40)
149
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.3.5 Application of the Biological Condition Gradient to Support Aquatic Life Use
Assessments
Because biotic assemblages may respond to stressors differently depending on the mechanism of action,
information from two or more assemblages provides more comprehensive insight into condition of the
water, possible sources of stress, and potential for improvements. For example, the presence of small
dams along streams and rivers alter natural flow and in stream habitat. These barriers prevent migration
of some native species from rivers into streams. Likewise, presence of large reservoirs can introduce
lake species into adjacent streams. Both of these impacts could result in a lower rating of biological
condition using fish community data. An assessment of the benthic macroinvertebrate community of
the same stream might result in a better biological condition rating if there are no additional stressors
and physical habitat "as naturally occurs." This information would indicate that habitat and food source
for fish exist and inform ADEM or other state agency decision makers that the stream may be a prime
candidate for restocking of native species.
The BCG can be used by ADEM to characterize and communicate the biological conditions in the "least
disturbed" reference reaches, aiding the interpretation of reference site quality relative to the absolute
definitions of the BCG levels. "Least disturbed" reference sites are the best observable landscape and
stream sites within a region. They can differ across regions of Alabama because development can be
ubiquitous across entire regions of the state. The BCG can be used to interpret biological conditions in
the "least disturbed" reference sites based on expert consensus in a manner that is transparent as long
as expert judgment and the resulting decision rules are documented. For example, 57% and 44% of sites
from ADEM's reference data set for two macroinvertebrate site classes, the Piedmont, Ridge, and Valley
and the Southwest Appalachian regions, were assigned as BCG level 2 based on the benthic
macroinvertebrate decision rules with the remainder of the sites primarily assigned as BCG level 3
(Figure 49). In contrast, only 13% of reference sites in the Interior Plateau were modeled as BCG level 2
and the majority of sites were assigned to BCG level 3. The differences in BCG levels among the
reference sites of the three site classes illustrates how the "least disturbed" reference condition can
have different biological meaning. BCG level 2 conditions support an aquatic community comparable to
what would be expected under naturally occurring conditions with no or minimal anthropogenic
impacts. The biological community characteristic of BCG level 3 includes loss of some native taxa and
shifts in relative abundance of taxa relative to BCG level 2. Integration of the reference information and
the BCG scale can be used to more clearly communicate to the public the quality of the reference
condition for each region. In addition, existing indicators could be calibrated to the BCG scale to refine
attainment thresholds. Despite the differences in reference site quality within the ADEM reference data
set, there is a comparable relationship with the WDG in all three regions (Figure 50). The scatter
observed with increasing WDG could, in part, be attributed to confounding effects and different
mechanisms of action of multiple stressors as well as mitigating influence of BMPs that have been
implemented.
150
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
40
20
15
10
Piedmont, Ridge & Valley
345
• Reference • Non-reference
SW Appalachians
345
• Reference • Non-reference
Interior Plateau
I Reference • Non-reference
Figure 49. Frequencies of sites (y-axis) in each BCG level (x-axis) in each northern Alabama site class, showing
reference sites as the blue portions of the bars. Distributions are based on sites monitored in ADEM's biological
assessment program.
151
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
2000
1800
1600
1400
„. 1200
>_
o
o 1000
(O
S3 800
Q uvjvj
600
400
200
o
one uiasses
| Interior Plateau
[n] Piedmont, Ridge & Valley
ill
-ti Ml
0
o
f O
K
*:
T 1 1
III fl
1 71 T
TTf 1 J
D
D
23456
BCG Level
Figure 50. BCG scores and corresponding WDG scores for Northern Alabama. Distributions are based on sites
monitored in ADEM's biological assessment program.
6.3.6 Future Applications
With the BCG model now available to characterize multiple levels of biological conditions, goals for
protection of high quality waters and for improvements in degraded waters can be better defined.
Currently, monitoring, assessment, and restoration focus on the most degraded watersheds throughout
Alabama, leaving fewer resources to prevent threatened waters from degrading and becoming listed as
impaired. Additionally, because success has typically been defined as a single threshold (i.e.,
attaining/nonattaining), incremental improvements in water quality and watershed conditions are not
effectively measured and documented. Information that conditions are incrementally improving is
valuable feedback to management, and stakeholders, including the public. Incremental changes can be
observed with a shift in BCG levels or in index values associated with the BCG levels (Figure 51 and
Figure 52).
152
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
100
80
V
I 60
v
1
I 40
20
Piedmont,
Ridge & Valley
I
T
I
T
SW Appalachians
I
T
I
Interior Plateau
3 4
BCG Level
2345
BCG Level
3 4
BCG Level
Figure 51. Alabama macroinvertebrate MMI distributions in site classes and BCG levels.
60
55
50
45
5 40
£35
il
-L D
Piedmont, Ridge & Valley
Plateau
Tennessee Valley
23456
BCG Level
23456
BCG Level
23456
BCG Level
Figure 52. Alabama fish IBI distributions in site classes and BCG levels.
With the BCG, multiple condition levels can be recognized, and each can be associated with different
resource status and management goals. For example, sites with BCG level 4, 5, or 6 conditions might be
targeted for incremental improvements with interim milestones set based on next BCG level. Streams
that score close to the next BCG level could be further prioritized for management actions. Such
incremental improvements would document successful management strategies and actions and support
adaptive management approaches. For sites supporting BCG level 2 conditions, the management goal
might be protection so that the water body continues to support exceptional biological communities.
BCG level 2 conditions could be identified using the predictive BCG models and/or the MMI and IBI
scores.
153
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
As part of its Healthy Watersheds Program/9 in 2011 EPA acknowledged the need to increase protection
of U.S. waters and provided states with a framework and tools. In 2013, ADEM completed the Alabama
and Mobile Bay Basin integrated assessment of watershed health (USEPA 2014b). The purpose of this
project was to characterize the relative health of catchments across Alabama and the Mobile Bay Basin
for the purpose of guiding future initiatives to protect healthy watersheds. The assessment synthesized
disparate data sources and types to depict current landscape and aquatic ecosystem conditions
throughout the Alabama/Mobile Bay Basin assessment area. The assessment included six distinct, but
interrelated attributes of watersheds and the aquatic ecosystems within them, including landscape
condition, habitat, hydrology, geomorphology, water quality, and biological condition. A total of 12
indicators were used to characterize the relative health of Alabama's watersheds. By integrating
information on multiple ecological attributes at several spatial and temporal scales, it provided a
systems perspective on watershed health. To compare the Healthy Watersheds Index (HWI) to BCG
assessments, ADEM recalculated the HWI after removing the biological components from the
calculation. The comparison showed a clear association between the non-biological HWI scores and the
BCG scores (Figure 53). The ranges of HWI scores in each BCG level were similar among site classes,
indicating that the BCG reflects differences in watershed integrity despite differences in landscape
stressor intensity among site classes.
90
80
70
g> 60
0
O
W 50
I
40
30
20
m
o
J 1
.
!-j
o 8
*T
H
*
Site Classes
• Interior Plateau
• Piedmont, Ridge & Valley
• SW Appalachians
h
o
'i
D
i ;
23456
BCG Level
Figure 53. Distributions of Healthy Watershed Index (HWI) scores by macroinvertebrate BCG level and site class.
29 More information on the Healthy Watersheds Program is available at: http://www.epa.gov/hwp. Accessed
February 2016.
154
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
The most pervasive changes to watershed condition are predicted to come from population increase
(changes in land and water use) and climate change (USEPA 2014b). Watershed vulnerability can be
defined as a combination of an ecological system's exposure, sensitivity, and adaptive capacity to cope
with changes in population and climate (IPCC 2007). The adaptive capacity of a watershed to cope with
such changes is enhanced by connectivity of habitats and maintenance of floodplain, wetland, and other
landscape features in their natural conditions to support natural hydrology and sediment supply.
Vulnerability was characterized for Alabama watersheds using indicators of projected changes in
precipitation, temperature, impervious cover, and water use (USEPA 2014b). Estimates of watershed
health and vulnerability combined with the BCG level scores can potentially be used together to inform
management decisions and priorities for protection and restoration.
6.3.7 Conclusion
ADEM developed a BCG model to expand the technical capability of its biological monitoring and
assessment program, with four key results. First, ADEM has been able to use the BCG to more accurately
characterize the quality of reference sites relative to natural conditions (e.g., no or minimal
anthropogenic disturbance). Second, in conjunction with biological indices, ADEM has used the BCG as a
tool to help identify high quality streams, evaluate recovery potential of degraded streams, propose
incremental biological goals for improvements, and track improvements. Third, ADEM is better able to
convey to the public and decision makers more detail about the aquatic community to assist both the
public and water quality managers in prioritizing areas for protection and restoration.
Finally, ADEM has found that adding fish community assessments to its biological assessment program
produces more robust and comprehensive assessments of aquatic life (USEPA 2013a). Fish assessments
are the primary biological indicator used to assess the status of threatened and endangered aquatic
species within the state. Macroinvertebrate and fish assessments are generally conducted at different
sites to make the most of limited resources and enable ADEM and partner agencies to assess biological
conditions at more sites throughout the state. The two assemblages are sensitive to different stressors
because of differences in the life cycles and motility offish and benthic macroinvertebrates. The
potential for different kinds of stress, the presence of threatened and endangered species, watershed
area, and depth are all factors used to determine which assemblage will be assessed at each site. The
BCG provides a common interpretive framework for benthic and fish assemblage data so both sets of
information could be mapped on a common assessment scale and the information used to inform
management decisions.
155
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.4 Minnesota: More Precisely Defining Aquatic Life Uses and Developing
Biological Criteria
6.4.1 Key Message
Most surface waters in Minnesota are protected for aquatic life and recreation to meet the objectives
set forth in CWA section 101(a). In the state, there are two primary sub-classes of streams protected for
aquatic life, including a cold water stream class (2A) and a warm water stream class (2B). While the
current system of beneficial uses and WQS has served Minnesota well, advances in the fields of
biological assessment have led to the recognition that among the diversity of water body types there are
variable biological conditions. For example, within rivers and streams, factors such as water body size,
geographic location, hydrology, water temperature, and stream gradient influence chemical, physical,
and biological composition. The Minnesota Pollution Control Agency (MPCA) recognized that effective
water quality management requires a more comprehensive approach in which goals for water quality
protection are tailored to specific water body types and uses. In response to these challenges, MPCA is
proposing to modify its beneficial use framework for aquatic life. The new framework will allow for
better goal-setting processes through the application of a framework that recognizes tiers, or levels, of
aquatic life-use based on a stream's type and potential. MCPA is using the BCG to describe existing
biological conditions and help provide the technical basis for assigning streams to ALL) classes.
6.4.2 Background
MPCA's collection of biological water quality information began in the 1960s as part of an effort to
monitor the conditions of state waters and since that time the state has developed a robust biological
assessment program (USEPA 2013a). Over the past two decades, MPCA has routinely monitored both
fish and benthic macroinvertebrates in streams, and, in combination with assessment of chemical and
physical parameters, has used this information to assess the integrity of streams (MPCA 2014b). In the
mid-1990s MPCA developed IBIs for fish (F-IBI) and benthic macroinvertebrates (M-IBI) to characterize
the health of biological communities, identify stressors, select management actions to protect and
restore water bodies, and determine how effective management actions are in meeting those goals. The
initial IBIs developed were supported by narrative statements in the state's regulatory language that
identified how to calculate an IBI. In 2003 and 2004, IBIs were developed for streams in specific basins of
the state, and subsequently MPCA developed IBIs that could be applied statewide (MPCA 2014c, 2014d).
Both the M-IBI and F-IBI used today are calibrated for a number of stream environments (e.g., large
rivers, moderate-sized streams, headwaters, low-gradient streams, and cold water streams) (MPCA
2014c, 2014d). The IBIs for different stream types minimize the effects of natural differences between
streams in order to enhance the signal from anthropogenic stressors. For example, the St. Louis River, a
large river in northern Minnesota, naturally has a very different fish fauna compared to a small cold
water stream in southern Minnesota such as Beaver Creek (Figure 54). Because the fish communities are
naturally different in these habitats, IBI models need to be specific to the stream type so that
appropriate expectations for healthy communities can be established. Since 2007, MPCA has monitored
the state's rivers and lakes using a 10-year rotating watershed approach.
Minnesota's WQS classify state waters according to their designated beneficial uses (e.g., aquatic life,
recreation, drinking water), and the state applies chemical, physical, and biological criteria to protect
designated uses. Currently, the majority of surface waters in Minnesota are classified as Class 2,
156
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Figure 54. Left: St. Louis River; Right: Beaver Creek.
protection of aquatic life and recreation30 (i.e., the "General Use" goal). For streams and rivers, class 2
waters are further distinguished as Class 2A (aquatic life cold water habitat) or Class 2B (aquatic life
warm water habitat). Despite the application of chemical, physical, and biological criteria, state
scientists determined that a single biological threshold does not reflect existing conditions in high
quality waters, nor set attainable restoration goals for degraded waters. For example, the West Branch
of the Little Knife River (Figure 55) in the Lake Superior drainage in Minnesota supports fish and
macroinvertebrate assemblages that would be expected in environments comparable to BCG level 1 or
2. A contrasting example is Judicial Ditch 7 in southeastern Minnesota (Figure 55). Fish and
macroinvertebrate assemblages in this stream do not meet the stream's current aquatic life goal, which
is estimated to be comparable to BCG level 4, because it is maintained for drainage. The activities
associated with maintaining this ditch for drainage remove the habitat necessary to support natural
aquatic assemblages and might limit attainment of the designated ALL). A use attainment analysis (UAA)
will support determination of the highest attainable use for these types of streams, and the BCG could
provide the basis for setting incremental restoration targets and tracking improvements.
Figure 55. Left: West Branch Little Knife River; Right: Judicial Ditch 7.
A full definition of Class 2 water can be found in Minnesota Administrative Rule 7050.0140, Subp. 3.
https://www.revisor.leg.state.mn.us/rules/?id=7050.0140. Accessed February 2016.
157
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.4.3 Tiered Aquatic Life Uses and Biological Criteria Development
Over the past ten years, state scientists have sought an approach that would capitalize on the state's
wealth of biological monitoring data and more specifically define the ALUs of rivers and streams in
Minnesota. MPCA is revising the state WQS to more accurately designate ALUs and establish multiple
levels (or goals) for aquatic life conditions in the WQS (in Minnesota this is known as the tiered aquatic
life use (TALU) framework). Using this framework, Minnesota is proposing to classify rivers and streams
based on the best attainable biological condition for a water body. The state is also proposing to
subcategorize its designated ALU categories to best reflect a stream or river's current conditions and its
ecological potential. This approach requires knowledge of the current condition of water bodies and the
stressors affecting them (MPCA 2012). In order to develop TALUs and associated biological criteria,
MPCA has capitalized on a variety of past work, including stream classification, IBI development, an HDS,
and the BCG (MPCA 2014b). The BCG was used to interpret current conditions and set expectations for
biological communities across the state. IBIs are used to determine the biological conditions of state
rivers and streams and to determine which ALU best describes the highest attainable biological
conditions in a specific water body.
MPCA's application of TALUs will subdivide Class 2 streams into three designated use class tiers (MPCA
2014e):
• Exceptional uses—"High quality waters with fish and invertebrate communities at or near
undisturbed conditions."
• General uses—"Waters with good fish and invertebrate communities that meet minimum
restoration goals."
• Modified uses—"Waters with legally altered habitat that prevents fish and invertebrate
communities from meeting minimum goals."
For each designated use class tier, MPCA has developed biological criteria using biological, chemical,
physical, and land use data collected during the 1995-2010 period. MPCA used a multiple lines of
evidence approach that included use of the BCG and the reference condition.
In order to identify reference streams, MPCA first calculated an HDS, an index that measures the degree
of human activity upstream of and within a stream. MPCA defined stream reference sites as those with
an HDS score of 61 or greater; this is a defined least disturbed condition (the upper 25% of the HDS
distribution). The reference streams are least influenced by stressors within the context of the current
landscape condition of Minnesota (Stoddard et al. 2006), as far as practical from urban areas, point
sources, feedlots, and other sources. MPCA also identified a subset of reference streams that satisfied
"minimally disturbed" in the northern part of the state where widespread and long-term human
disturbance is much less than in the south. MPCA compared the IBI scores for reference and non-
reference sites. While MPCA identified some concerns with applicability of the reference condition
approach in southern Minnesota due to widespread, high levels of land use and development, the
agency determined that reference data sets were sufficient to develop biological criteria in the northern
regions and in cold water classes (MPCA 2014b). Reference conditions for the southern region might
require an alternate approach to more precisely characterize least disturbed conditions.
158
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
During 2009-2012, expert panels were assembled to develop BCG models for both macroinvertebrate
and fish assemblages (Gerritsen et al. 2013). The conceptual BCG model (Davies and Jackson 2006) was
calibrated by these expert panels using regional data for each of the two assemblages. The narrative
descriptions for the different BCG condition levels were used by MPCA to describe each of the three
designated use class tiers proposed in the revision to its WQS regulation:31
• Exceptional Use—"Evident changes in structure due to loss of some rare native taxa; shifts in
relative abundance; ecosystem level functions fully maintained."
• General Use—"Overall balanced distribution of all expected major groups; ecosystem functions
largely maintained through redundant attributes."
• Modified Use—"Sensitive taxa markedly diminished; conspicuously unbalanced distribution of
major taxonomic groups; ecosystem function shows reduced complexity & redundancy."
The MPCA expert panels characterized and calibrated the BCG for both benthic macroinvertebrates and
fish for seven classes of warm water streams and two classes of cold and coolwater streams. A summary
of the narrative rules includes:
• Taxa richness declined from BCG level 1 to level 6. All level 1 sites were large water bodies
(rivers), and might be more influenced by size than by condition
• Attribute I taxa were characteristic of BCG level 1, occurred occasionally in BCG level 2, and
were generally absent in levels 3-6
• All sensitive taxa (attributes I, II, and III combined) are common and abundant in levels 1 and 2,
somewhat reduced in level 3, decline markedly in level 4, and have almost disappeared from
levels 5 and 6.
• Intermediate taxa (attribute IV) are nearly constant throughout the gradient, but are reduced in
level 6.
• MPCA divided the tolerant fish category into two: tolerant taxa (attribute V), and highly tolerant
taxa (attribute V-a), as well as highly tolerant nonnative (attribute Vl-a). The highly tolerant
subgroups increased in abundance, dominance and variability at BCG levels 4 to 6, although the
natives are represented at all levels.
An example of quantitative BCG rules derived for fish in the two river classes is shown in Table 34.
31 Information about Minnesota's WQS process is available at:
http://www.pca.state.mn.us/index.php/water/water-permits-and-rules/water-quality-standards.html. Accessed
February 2016.
159
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 34. Decision rules for fish assemblages in two classes of Minnesota rivers. Rules show the ranges
of fuzzy membership functions. N indicates the number of sites for a given BCG level and stream class
in the calibration data set.
Metric
BCG Level 1
Total taxa
Endemic taxa (Att 1)
Attl+ll taxa
Attl+ll+lll %taxa
Attl+ll+lll %ind
Att Va or Via Dominance
Tolerant % ind (V + Va + Via)
Highlytol%ind(Va + Vla)
BCG Level 2
Total taxa
Attl+ll taxa
Att 1+11+111% taxa
Att 1+11+111% Ind
Att Va or Via Dominance
Highlytol%ind(Va + Vla)
BCG Level 3
Total taxa
Att 1+11+111% taxa
Att 1+11+111% Ind
Tol% ind(V + Va + Vla)
Att Va or Via Dominance
Highly tol% ind (Va + Via)
Total taxa
Att 1+11+111% taxa
Att 1+11+111% Ind
I+II+III+IV% Ind
Att Va or Via Dominance
Tol% ind(V + Va + Vla)
HighlyTol%ind(Va + Vla)
BCG Level 5
Total taxa
Att l+l 1+111+4% Taxa
Att Va or Via Dominance
Highly tol% ind (Va + Via)
BCG Level 6 (no rules)
Prairie Rivers
N=2
> 25-35
Present
>2-5
> 45%-55%
> 25%-35%
< 3%-7%
N=6
> 16-24
Present
> 35%-45%
> 15%-25%
< 7%-13%
N=25
> 11-16
> 15%-25%
> 7%-13%
-
< 7%-13%
< 25%-35%
> 11-16
10%-20%
0%-1%
< 35%-45%
< 45%-55%
N=12
> 11-16
< 65%-75%
N=l
Northern Forest Rivers
N=3
> 16-24
Present
> 1
-2
> 35%-45%
> 45%-55%
<7%-
<7%-
-13%
-13%
N=15
>6-10
> 25%-35%
> 25%-35%
<7%-
<7%-
-13%
-13%
N=ll
>6-10
> 15%-25%
>7%-
-13%
< 25%-35%
< 10%-20%
N=16
Altl
>6-10
> 15%-25%
> 3%-7%
< 25%-35%
n/a
< 35%^5%
Alt 2
= alt I1
> 7%-13%
present
= alt I1
< 30%-40%
= alt I1
N=2
6-10
< 35%-45%
< 55%-65%
N=0
1 "= alt 1" the rule is the same as given under Alt 1 for this metric
MPCAthen calibrated the BCG with the state's index for biological assessment of Minnesota's warm
water and cold water streams for both the fish and macroinvertebrate assemblages (Figure 56). MPCA
has used this information to develop draft numeric biological criteria that would be applied to each
designated use class tier—thus directly linking the ALL) goal with the state's assessment method (Figure
57). In December 2015, MPCA held a formal public comment period on a proposed revision to the state
WQS that would include TALUs.
160
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
CD
100
90
80
70
. 60
: 50
40
30
20
10
0
100
90 -
80 -
70 -
. 60 -
50 -
40 -
30 -
20 -
10 -
0
CD
Northern Forest Rivers
Prairie Rivers
±
X
T
High Gradient
Northern Forest Streams
Low Gradient
Northern Forest Streams
High Gradient
Southern Streams
Low Gradient
Southern Forest Streams
100
90 -
80 -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
0
Low Gradient
Prairie Streams
1
X
T
Northern
Coldwater Streams
3 4
BCG Level
1
3 4
BCG Level
Southern
Coldwater Streams
—i 1 1 1—
234
BCG Level
1
Figure 56. Frequency distributions of IBI scores by BCG level for macroinvertebrate stream types using data from natural channel streams sampled 1996-
2011. Symbols: upper and lower bounds of box = 75th and 25th percentiles, middle bar in box =
percentiles. MPCA also did a calibration of fish index scores with BCG levels assigned to sites.
2011. Symbols: upper and lower bounds of box = 75th and 25th percentiles, middle bar in box = 50th percentile, upper and lower whisker caps = 90th and 10th
161
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Natural structural, functional, and taxonomic integrity is preserved
Minimal changes in both structure and function
Exceptional Use Goal
Evident (e.g., measurable) changes in structure,
minimal changes in function
Moderate changes in structure and
evident changes in function
General Use Goal
Major changes in structure and
moderate changes in function
Modified Use Goal
Severe changes in structure and function
Level of Stressors
Low-
High
Figure 57. BCG illustrating the location of proposed biological criteria (black dotted line) for protection of
Minnesota's TALU goals (Exceptional, General, Modified) (Source: MPCA 2014b).
6.4.4 Benefits of the Biological Condition Gradient
Because the BCG provides a common framework to interpret changes in biological condition regardless
of geography or water resource type, Minnesota will be able to make more accurate determinations and
classifications of its aquatic resources on a statewide basis. The state will be in position to make
decisions on aquatic life designations based on robust and detailed ecological data and information.
Another advantage of the BCG is that it provides a means to communicate with the public about existing
conditions and potential for improvement for specific water bodies. BCGs were developed for each of
Minnesota's aquatic resource classes for streams (e.g., cold water and warm water streams). The
development of warm water BCG models involved input from biological experts familiar with biological
communities in Minnesota from the MPCA and Minnesota Department of Natural Resources. BCG
models were developed for fish and macroinvertebrates for each of the seven warm water stream
classes. A cold water BCG involved experts from Minnesota, Wisconsin, Michigan, and several tribes
located in those states. In Minnesota this effort included two classes each for fish and
macroinvertebrates. Model development for each class involved reviewing biological community data
from monitoring sites and then assigning that community to a BCG level. A sufficient number of samples
were assessed to develop a model that can duplicate the panel's BCG level assignments. Using the BCG
and reference conditions permits MPCA to provide more detailed descriptions of the expected biota for
each ALL) and to develop biological criteria that are protective, consistent, and attainable across the
state (MPCA 2012). These accomplishments will help Minnesota achieve several key goals described
below.
162
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Refinement of Biological Standards
Numeric water quality criteria that are codified in the Minnesota WQS are currently based on chemical
and physical criteria such as DO, temperature, and pH. These criteria do not directly measure the
condition of biological communities that include fish, insects, mussels, aquatic plants, and algae.
Biological communities can be monitored as a direct measure of the response of the biota to a wide
range of physical and chemical stressors and provide a quantitative measure of the cumulative and
synergistic impacts of multiple stressors over time. A major goal of Minnesota's water quality
management program is to protect the fish, invertebrates, and other aquatic organisms in the state's
waters. Therefore, it is sensible that a direct measurement of these communities is used to monitor
their condition.
Ability to Address Natural Variation
One of the strengths of Minnesota's approach is the ability to address the natural variation in water
resources across the state. Minnesota's diverse water resources mean that refined biological monitoring
tools are needed to reduce errors in assessment and management. For example, streams along the
shore of Lake Superior in northern Minnesota are very different from streams in southern Minnesota
such that, under natural conditions, the biological communities in streams in each location are expected
to be different. The Minnesota BCG framework takes into account these natural differences and
requires that comparisons be made between streams with naturally similar biological communities. As
the state's database is built through long term monitoring, Minnesota will be able to define current, or
baseline, conditions and be in a better position to discern shifts in species composition and structure
due to climate change impacts.
Identification of Reference Condition Quality
The biological monitoring program in Minnesota relies on BCG models and the reference condition
approach to set expectations for water bodies. The BCG provides a common "yardstick" of biological
condition that is rooted in the natural condition. As a result, the BCG can be used to develop biological
criteria that are consistent across regions and stream types in Minnesota—particularly important for a
state where the range of existing quality is regionally distinct and extreme (i.e., undisturbed to highly
disturbed conditions). The reference condition approach identifies water bodies that are least disturbed
and uses them to establish the reference condition. Once this reference condition has been established,
water bodies with unknown condition can be compared to this baseline. If the condition of the water
body is lower than that of the reference condition, it would be considered impacted or stressed. The use
of a reference condition relies on the development of accurate expectations for least disturbed sites.
The BCG provides a framework for assessing the quality of reference sites relative to undisturbed
conditions and can be used to interpret the quality of reference sites, including reference sites in regions
where the least disturbed conditions include sites with moderate to higher levels of stress. In these
regions, such as in southern Minnesota, the BCG was used to help develop protective ALL) goals (MPCA
2014b, 2014e).
Protection of High Quality Water Resources
Minnesota's classification framework and BCGs will be applied in conjunction with another element of
states' antidegradation policy. This policy requires:
• Maintenance of existing uses;
• Prevention of degradation of water quality that exceeds levels necessary to support the
protection and propagation of aquatic life and recreation unless the state finds that lowering of
163
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
water quality is necessary to accommodate important economic or social development (Tier 2
protection); and
• Maintenance of water quality needed to protect outstanding resource waters (Tier 3
protection).
Minnesota is planning to propose a higher tier of ALL) (i.e., exceptional use goal) to protect high quality
biological communities. Once it has been established that a water body is meeting the requirements
associated with an exceptional water resource, the resource needs to be protected to maintain that
status. The BCG provides a framework with which to identify candidate high quality streams and rivers
for designation as exceptional resources.
Setting Expectations for Modified Water Resources
There are water resources in Minnesota that will not in the near future meet the CWA interim goals due
to historical or legacy impacts. These legacy impacts include streams under drainage maintenance or
other irreversible hydromodification that preclude attainment of water body goals (e.g., channelized
streams and ditches). The BCG provides a framework to monitor and help set realistic expectations for
waters that are unlikely to meet ALL) goals due to legacy impacts and have been designated as modified
water resources. Additionally, as conditions improve, the BCG provides a framework to document and
acknowledge these improvements to reflect existing conditions.
6.4.5 Conclusion
In conjunction with numeric biological indices developed for macroinvertebrates, the BCG allows
Minnesota to set consistent and protective ALL) goals and numeric biological criteria across the state
despite the heterogeneity of its water bodies. This heterogeneity is due both to natural conditions and
human disturbance, and the BCG provides a framework to characterize and communicate these
differences. The BCG described in this case study is applicable to streams and wadeable rivers.
Minnesota is currently developing a BCG and biological criteria for lakes using fish assemblage
information.
164
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.5 Maine: Development of Condition Classes and Biological Criteria to Support
Water Quality Management Decision Making
6.5.1 Key Message
Clear, technically rigorous goal statements have provided Maine with an effective framework to improve
biological condition of streams and rivers. Maine has established four ALL) classes (Classes AA/A/B/C)
with different ecological expectations. The classes span the range from Maine's interpretation of the
CWA interim goal to the ultimate CWA objective "to restore and maintain chemical, physical and
biological integrity" (Class AA/A). All rivers and streams in Maine are assigned to one of the four classes
in Maine's WQS for planning and management purposes. These TALUs and numeric biological criteria
have enabled Maine to inject critical biological information into all aspects of water quality
management. Along with the practical experience and scientific advancements demonstrated by other
states with strong biological assessment programs, Maine's approach to classification and biological
criteria development provided the template for the conceptual BCG (Davies and Jackson 2006). In turn,
Maine continues to strengthen and develop its biological assessment program to address other water
bodies and include measures of the algal communities in its assessments. The BCG is being incorporated
as part of its "toolbox" to accomplish these tasks.
6.5.2 Background
Since the 1960s, prior to adoption of the CWA, Maine water quality law has had a tiered structure based
on observations of gradients of water quality conditions. In 1986, Maine revised its water classification
law and added TALUs to maintain and restore the structure, function, and biological integrity of aquatic
life communities. Maine's TALUs were based on concepts of John Cairns, H.T. Odum, and others who
observed declines in biological condition in response to gradients of increasing stressors (Ballentine and
Guarraia 1977; Odum et al. 1979, Cairns et al. 1993; Karr and Chu 2000). The four narrative TALU
standards in Maine's water classification law describe conditions across a biological gradient ranging
from "as naturally occurs" (Classes AA and A) to "maintenance of structure and function" (Class C). Class
C is the lowest ALU designation allowed in the state and consistent with Maine's interpretation of the
CWA fishable/swimmable interim goal (Table 35; M.R.S.A Title 38 Article 4-A § 464-466). Maine's TALUs
for fresh surface waters apply to streams, rivers, and wetlands. Maine has similar TALUs for coastal
marine waters (SA, SB, SC). Maine has established a single class for lakes that is equivalent to Class A.
Maine's TALUs are based on tiers of biological condition along observed human disturbance gradients.
Such stressor-response relationships are also the foundation of the later development of the BCG.
Maine's TALUs are supported by ecologically-based definitions in the law. The narrative definitions in
Maine law establish the biological characteristics that are required to attain the standards of each class
(Table 35). Class AA and Class A have the same "as naturally occurs" aquatic life goals and will hereafter
be referred to as Class AA/A; Class AA is more restrictive in allowable permitted activities. For example,
no dams or discharges are allowed in Class AA waters. Maine's assessed streams and rivers are
predominantly classified as either Class AA/A or B waters, 48.6% and 51%, respectively. Class A/AA
waters have been interpreted by Maine as comparable to BCG levels 1 and 2 and class B waters are
equivalent to BCG level 3. Less than 1% of Maine's streams and rivers are classified as Class C waters,
which have been deemed as comparable to BCG level 4. These waters are primarily in urbanizing areas
or downstream of significant point sources. Figure 58 summarizes relationships between Maine's
narrative biological, chemical, and physical standards and shows Maine's TALUs in relation to the BCG.
165
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 35. Criteria for Maine river and stream classifications and relationship to antidegradation policy
Class
AA
A
B
C
DO criteria
As naturally
occurs
7 ppm; 75%
saturation
7 ppm; 75%
saturation
5 ppm; 60%
saturation;
and
6.5 ppm
(monthly avg.)
when
temperature
is < 24 °C
Bacteria
criteria
As
naturally
occurs
As
naturally
occurs
64/100 mg
(g.m.) or
236/100
ml (inst.)*
125/100
mg(g.m.)
or
236/100
(inst.)*
Habitat
narrative
criteria
Free-flowing
and natural
Natural**
Unimpaired**
Habitat for
fish and other
aquatic life
Aquatic life narrative criteria***
and management
limitations/restrictions
As naturally occurs**; no direct
discharge of pollutants; no dams or
other flow obstructions.
Discharges permitted only if the
discharged effluent is of equal to or
better quality than the existing
quality of the receiving water; before
issuing a discharge permit the
Department shall require the
applicant to objectively demonstrate
to the department's satisfaction that
the discharge is necessary and that
there are no reasonable alternatives
available. Discharges into waters of
this class licensed before 1/1/1986
are allowed to continue only until
practical alternatives exist.
Discharges shall not cause adverse
impact to aquatic life** in that the
receiving waters shall be of sufficient
quality to support all aquatic species
indigenous** to the receiving water
without detrimental changes to the
resident biological community.**
Discharges may cause some changes
to aquatic life**, provided that the
receiving waters shall be of sufficient
quality to support all species of fish
indigenous** to the receiving waters
and maintain the structure** and
function** of the resident biological
community. **
2012
Percentage of
Maine waters
designated in
class ****
3.6%
45%
51%
0.4%
Corresponding
federal
antidegradation
policy tiers
3 (Outstanding
National
Resource Water
[ONRW])
2/2
2 to 2 1/2
Ito2
Source: Maine DEP (modified), http://www.maine.gov/dep/water/monitoring/classification/index.html. Accessed February 2016.
Notes:
* g.m. = geometric mean; inst. = instantaneous level.
** Terms are defined by statute (Maine Revised Statutes Title 38, §466).
*** Numeric biological criteria in Maine regulation Chapter 579, Classification Attainment Evaluation Using Biological Criteria for Rivers
and Streams http://www.maine.gov/dep/water/rules/index.html. Accessed February 2016.
**** Source: 2012 Maine Integrated Water Quality Report,
http://www.maine.gov/dep/water/monitoring/305b/2012/report-final.pdf. Accessed February 2016.
166
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Maine's Aquatic Life Management Classes
1 '
V
2 2
O 2
§
s 3
o
U 4
•3
3
*5o
5
o
5
As naturally
occurs.
Habitat: "natural"
No detrimental change;
support all indigenous
species.
Habitat: "unimpaired"
Maintain structure
and function: support
all indigenous fish
(salmonids).
Habitat for fish and
aquatic life
Non-
Attainment
of minimum
standards
I
CLASS AA
Zero discharge;
No hydrologic
alteration; DO
and bacteria as
natural
Maine's Water Quality Management Classes
»
CLASS C
CLASS A CLASS B
No alternatives; D/C with ample dilution;
D/C Equal to or DO: 7pprn/75% saturation;
better; hydro 9ppin for salmonid
allowed; DO: 7ppm/ spawning; Bacteria:
75% saturation: 64/100 mil- in the summer
bacteria as natural
DO: 5ppin/60% saturation;
Water quality sufficient to
ensure salmouid
spawning/survival;
Bacteria:126/100 mil
\
\
NTA
Maine's
interpretation
of failure of
CWA
minimum
Figure 58. Relation between Maine TALUs, the BCG, and Maine's other water quality standards and criteria.
Class AA/A is approximately equivalent to BCG levels 1 and 2. Classes B and C approximate BCG levels 3 and 4,
respectively. Non-attainment conditions below Class C are approximately equivalent to BCG levels 5 and 6.
6.5.3 Maine's Numeric Biological Criteria and Tiered Aquatic Life Uses
In 2003, Maine adopted numeric biological criteria in rule for rivers and streams, based on assessment
of benthic macroinvertebrates (State of Maine 2003; Shelton and Blocksom 2004; Davies et al. In press).
Technical details describing development of the statistical biological criteria models are found in Chapter
4 of this document and in Davies et al. (In press). In short, MEDEP utilized expert consensus to establish
four a priori groups corresponding to Maine's TALUs, and developed and tested a linear discriminant
model (LDM) to predict the probability of a sample attaining ALL) goal conditions (Class AA/A, Class B,
and Class C). The fourth group, termed "non-attainment" (NA) represents samples that are in poorer
condition than Class C. The LDM and accompanying provisions for application are codified in rule and
constitute Maine's numeric biological criteria.32 When confirmed (e.g., by re-sampling and review of
data results) that a stream reach fails to attain its assigned water quality goal, the water body segment is
listed as impaired on the state's 303(d) list (Table 36). State law requires that water bodies be
considered for upgrade to a higher class if they are found to be consistently attaining the standards of
that higher classification.
http://www.maine.gov/dep/water/rules/index.html. Accessed February 2016.
167
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Table 36. Examples of how numeric biological criteria results determine whether or not a water body
attains designated ALUs in Maine
Legislative Class
A
C
A
B
Monitoring Result
A
B
B
NA
Attains Class?
Yes
Yes
No
No
Next Step
--
Review for upgrade
303(d) list as impaired if confirmed
303(d) list as impaired if confirmed
MEDEP also conducts biological assessments of stream algal, wetland macroinvertebrate, and wetland
phytoplankton and epiphytic algal assemblages (Danielson et al. 2011, 2012). MDEP used Maine's
narrative biological criteria and the BCG as the foundation of biological assessment models for stream
algae and wetland macroinvertebrates. A first step in model-building was to empirically compute
tolerance values for algal and macroinvertebrate species that had been collected in Maine's monitoring
program. After computing tolerance values, the species were grouped into the BCG framework's
sensitive, intermediate, and tolerant attribute groups. MEDEP then modified the BCG framework for
stream macroinvertebrates for stream algae and wetland macroinvertebrates, describing how those
assemblages empirically respond to anthropogenic stressor gradients. MEDEP used the BCG and
tolerance metrics along with the narrative biological criteria and other metrics to build predictive
biological assessment models for the additional assemblages. MEDEP has completed LDM statistical
models to predict TALU attainment for both stream algal and wetland macroinvertebrate community
data. These models currently are used to help interpret narrative biological criteria. Following adequate
testing and standard public review protocols, MEDEP intends to amend the Maine Biological Criteria
Rule33 to include the stream algal and wetland macroinvertebrate models as numeric biological criteria.
In summary, numeric biological assessment models, when codified in the MEDEP biological criteria rule
(as for stream macroinvertebrates), or when used as an objective corroboration of expert judgment (as
for stream algae and wetlands), provide a transparent and standardized quantitative means for
determining attainment of TALUs in Maine WQS. Numeric biological criteria have enabled Maine to use
biological information to support multiple water quality management information needs and decision
making. Examples of applications follow.
6.5.4 Goal-based Management Planning to Optimize Aquatic Life Conditions
As described in section 6.5.2, the Maine State Legislature revised Maine's WQS and classification law in
1986 (M.R.S.A Title 38 Article 4-A § 464-466) establishing narrative biological criteria for four ALL) classes
for rivers and streams. This law set in motion a process involving the public, the state environmental
agency, and the Maine legislature to assign all Maine waters to an appropriate goal classification. All
available monitoring data and information about then-current biological and/or water quality conditions
were used to initially propose the statutory classes for stream and river segments for the 1986 law.
Many waters that lacked current monitoring data retained their previous water quality goals (generally
Class B, except for some urban or industrialized areas, which were Class C) until monitoring data or
other evidence was found to recommend a different (and in most cases higher) class.
33 See Code of Maine Rules, MEDEP, Chapter 579, http://www.maine.gov/dep/water/rules/index.html. Accessed
February 2016.
168
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Maps spanning the period between 1987 (Figure 59) and 2012 (Figure 60) show the past and present
distribution of water quality classifications. Approximately 99% of Maine's rivers and streams have been
designated for classes of protection equal to or higher than Maine's interpretation of the CWA Interim
Goal (i.e., Class C). Reclassification upgrades have been implemented with strong public and legislative
support. The state has designated water bodies into higher classes to protect waters currently
demonstrating high quality and to retain improvements in lower quality waters that had been restored
to higher conditions due to wastewater treatment successes. During the nearly three decades since
1987, the Maine State Legislature has assigned 13,955 river and stream miles to a Class A or Class AA
management goal, an increase of 25.5%34. Numeric biological criteria and articulation of the gradient of
aquatic life management classes facilitated the recognition of both the presence of high quality waters
and improvements in condition due to remediation. As shown in Figures 21 and 22, these classification
upgrades have mostly been drawn from Class B and Class C waters where biological monitoring data
demonstrated either the potential, or the actual achievement of the standards of Class A or Class AA.
Without their ALL) classification approach, TALUs, and criteria, these gains in condition would likely have
gone un-detected and unprotected. Additionally, the state's ecologically descriptive condition classes
have enhanced public understanding of existing conditions, problems, and restorable target conditions.
They provide an important tool in building public and stakeholder support for the often substantial
investment that is required to restore aquatic resources.
• AA
• A
B
C
Classification of Maine
Waters 1987
13.471
34,515
6,552
1.7%
M.3%
62.2%
I
Figure 59. Distribution of Maine water quality
classifications in 1987 prior to WQS revisions.
• AA
A
B
C
•
*
Classification of Maine
Waters 2012
Sticani Mitel Percent otTotal
3.4IH d.2%
25.007 45.2%
26 JH 475%
614 1.1%
AA
i
Figure 60. Distribution of Maine water quality
classifications in 2012 following 25 years of water
quality improvements and classification upgrades.
See State of Maine Water Quality Standards Docket, http://www.maine.gov/dep/water/wqs/docket/index.html
(Accessed February 2016) and USEPA, State Tribal and Territorial Standards
http://water.epa.gov/scitech/swguidance/standards/wqslibrary/me index.cfm (Accessed February 2016).
169
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.5.5 Early Detection and Management of an Emerging Problem
When Maine's Biological Monitoring Program was initiated, a primary concern was management of
point source discharges. Implementation of Best Available Technology for point sources eliminated
many of these causes of biological impairment with the result that the aquatic life in receiving waters
throughout the State rebounded to significantly improved conditions (Davies et al. 1999; Davies et al. In
press). More recently, however, biological assessment of smaller streams has revealed impairment
caused by changes in physical stream conditions (e.g., increased impervious surfaces in the watershed,
hydrologic and stream channel shape alteration). Chemical assessments in these smaller streams have
documented increased nutrients and toxic constituent concentrations, salt runoff, increased
temperature, and decreased DO.
In 2006, Maine became one of the first states to issue TMDLs based on the percent of a stream
watershed covered by impervious surfaces such as roads and parking lots (% 1C) (Meidel and MEDEP
2006a, 2006b). Narrative and numeric biological criteria in Maine's WQS were used as the TMDL end
point, goal, and ultimate numeric water quality compliance measure for the impaired portions of the
streams in order to address non-attainment of ALUs. The restoration pathway described in the TMDL
focused on realistic approaches to minimizing the biological, physical, and chemical effects of
impervious cover, rather than direct elimination of 1C. Expanding on the success of the 2006 % 1C TMDL,
in 2012, MEDEP completed a statewide % 1C TMDL for 30 urban impaired streams and 5 associated
wetlands (MEDEP 2012). As in 2006, the 2012 TMDL also included aquatic life restoration targets based
on the relationship of % 1C in the stream watersheds and target improvements in macroinvertebrate
community condition.
In 2015, MEDEP conducted a fine-scale geospatial analysis of % 1C in watersheds upstream of algal and
macroinvertebrate biological assessment sites and determined attainment of TALL) for each assemblage
at those sites (Danielson et al. In press). Watershed % 1C estimates were computed in ArcMap with 1-
meter, high-resolution spatial data from 2004 and 2007. Results, shown in Figure 61, revealed that in
general, streams become vulnerable to no longer attaining Class AA/A biological criteria when % 1C in
upstream watersheds is in the range of l%-3% 1C. The risk of not attaining Class B biological criteria
increases in the range of 3%-6% 1C. Finally, the transition from low risk to high risk of attaining Class C
criteria is in the range of 10%-15% 1C.
The % 1C study revealed that small streams are at risk of impairment at lower levels of watershed % 1C
than previously recognized. Recognizing the difficulty, expense, and extended lag times associated with
urban stream restoration, environmental managers and urban planners in Maine increasingly realize the
importance and cost-effectiveness of preventing impairment of urban streams. TALL) and BCG concepts,
along with rigorous biological assessment data, helped MDEP raise awareness about the vulnerability of
biological assemblages to urbanization and other human-caused stressors. This information is used in
Maine at both the state and local level to inform water quality management decisions and local land use
planning and design of development.
170
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
o
O
o
'£:
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
6.5.6 Monitoring and Assessment to Determine Current Condition: Using Biological
Condition Gradient Concepts to Integrate Biological Information from Multiple
Assemblages and Water Body Types
u
n
ti
[A]
[A]
WATERBODY
WETLAND
WETLAND
b STREAM
STREAM
STREAM
STREAM
ASSEMBLAGE
MACROINVERTEBRATE
MACROINVERTEBRATE
MACROINVERTEBRATE
ALGAE
ALGAE
ALGAE
CLASS
A
B
B
B
B/C
C
BCG
LEVEL
2
3
3
3
3/4
4
Figure 62. Pleasant River sites with attained water quality
class and BCG level for different assemblages and water body
types.
BCG concepts provide Maine with a common
assessment framework for comparing
biological integrity among different types of
water bodies (wetlands, rivers, and streams),
regardless of the assemblage assessed or the
sampling methods used. This enables MEDEP
to evaluate condition and threats to aquatic
resources on a watershed basis. The
integrated assessment also contributes
important information for design of
remediation activities, even in the absence
of formally promulgated numeric biological
criteria. For example, MEDEP evaluated the
condition of the Pleasant River watershed
using multiple biological assessment models,
water quality class attainment, expert
judgment, the BCG, and supporting chemical
and physical information. Located in
southern Maine, the Pleasant River
watershed is primarily forested with some
agriculture, as well as increasing amounts of
urbanization in the downstream portions of
the watershed. The Pleasant River has a
TALU goal of Class B. MEDEP sampled algae
and macroinvertebrates in several locations
on the Pleasant River and sampled
macroinvertebrates in several headwater
wetlands (MEDEP 2006, 2009, 2014;
Danielson etal. 2011). Biological assessment
showed that the headwater stream and
wetland samples attained Class A or B
biological criteria using macroinvertebrate
data (Figure 62).
However, further downstream, the stream
macroinvertebrate samples attained Class B
biological criteria, but stream algal samples
were mixed, attaining Class B or C. MEDEP
used water chemistry data, habitat
evaluations, diagnostic algal and
macroinvertebrate metrics, expert
judgment, and the BCG concept to
determine that nutrient pollution was the
172
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
probable stressor to which the algal community was responding. A watershed survey identified potential
sources of nutrients in the lower part of the watershed. The combination of biological assessments for
two water body types and taxonomic groups allowed MEDEP to complete a more holistic and
meaningful evaluation of the Pleasant River watershed than what could have been accomplished with
only one biological assessment method. MEDEP now has a tool to detect early signals of nutrient
pollution before the full aquatic community is detrimentally impacted.
Findings from biological assessments of multiple assemblages and water body types have also been used
to improve and strengthen Maine's statewide impervious cover TMDL report.35 For example, in Maine's
2010 Integrated Water Quality Report, Capisic Brook in Portland and Westbrook, Maine was 303(d)-
listed for stream benthic macroinvertebrate impairment based on MEDEP's numeric biological criteria
rule. Although numeric biological criteria for Maine wetlands had not yet been formally promulgated,
Capisic Pond was also listed for wetland macroinvertebrate impairments based on interpretation of
quantitative data showing that narrative ALUs were not attained. The state's multivariate biological
assessment models for wetland macroinvertebrates and stream algae enabled results to be compared to
Maine's TALL) classes and macroinvertebrate numeric biological criteria. Stream algal and wetland
macroinvertebrate biological assessments helped biologists determine that Capisic Pond and Capisic
Brook were not attaining narrative biological criteria, resulting in biological impairment listing for
multiple causes.
6.5.7 Using Maine's Tiered Aquatic Life Uses and Biological Assessment Methods to
Evaluate Wetland Condition
The MEDEP Biological Monitoring Program assesses the health of inundated emergent and aquatic bed
freshwater wetlands. Samples consist of aquatic macroinvertebrates, planktonic and epiphytic algae,
and physical and chemical data related to trophic state and habitat condition (MEDEP 2006; MEDEP
2009). Sampling typically occurs in freshwater marshes and fringing wetlands associated with rivers,
streams, lakes, and ponds. The biological assessment statistical model for wetlands provides an
objective means of assessing condition.
Maine has found that wetland biological assessment provides a complementary approach to
assessments of wetland function and value. Under the definitions established by the USEPA Wetland
Core Elements of an Effective State and Tribal Wetlands Program36 Maine conducts a "level 3" biological
assessment of wetlands. According to EPA, "level 3 or intensive site assessments provide a more
thorough and rigorous measure of wetland condition by gathering direct and detailed measurements of
biological taxa and/or hydro-geomorphic functions." Maine's wetland macroinvertebrate biological
assessment program can detect incremental differences in aquatic resource condition utilizing a locally
calibrated statistical model consistent with the BCG concepts (MDEP 2006; MDEP 2009). Additional
applications of wetland biological assessments include determining whether wetlands attain designated
ALUs, tracking trends over time, and, in conjunction with chemical and physical assessments, diagnosing
stressors, and assessing impacts or threats related to land use practices (e.g., point source discharges,
toxic contaminants, hydropower, and water withdrawal projects).
In 2013, the MEDEP Biological Monitoring Program evaluated the biological condition of wetland
compensatory mitigation projects using wetland biological assessment methods (DiFranco et al. 2013).
35 See http://www.maine.gov/dep/water/monitoring/tmdl/tmdl2.html. Accessed February 2016.
36 See http://water.epa.gov/grants funding/wetlands/cefintro.cfm. Accessed February 2016.
173
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Mitigating adverse environmental impacts of development is an integral part of Maine's Natural
Resources Protection Act,37 a state law regulating land use activities and administered by MEDEP. The
State of Maine or federal agencies administering resource protection regulations might require
appropriate and practicable compensatory mitigation as a condition of granting a permit to alter or
destroy wetlands. Compensation is defined in the NRPA as "replacement of a lost or degraded wetland
function with a function of equal or greater value." If ecologically appropriate compensation is not
available or otherwise practicable, a permit applicant may request to pay an in-lieu compensation fee to
be used for the purpose of restoring, enhancing, creating or preserving other resource functions or
values that are environmentally equal or preferable to the functions and values being lost. Upon
authorization the In-Lieu Fee is placed in a "Natural Resource Mitigation Fund" administered by The
Nature Conservancy's (TNC's) Maine office.
For this study, MEDEP wanted to determine whether compensatory mitigation projects supported
aquatic life communities comparable to minimally disturbed reference sites. The MEDEP Biological
Monitoring Program evaluated quantitative biological data, biological assessment model results, expert
judgment, and the BCG, to compare the biological condition of 9 wetland compensation sites to that of
51 minimally disturbed reference sites. The mitigation sites in the study represented a cross section of
available Maine "permittee-responsible" compensation projects that used restoration, creation,
enhancement, and preservation techniques, and were completed between 1995 and 2007. The
compensation projects varied in age and encompassed a range of freshwater wetland types, including
forested, scrub-shrub, emergent, wet meadow, aquatic bed, and open water marsh.
Figure 63 illustrates comparisons of reference and mitigation sites for sensitive versus tolerant taxa
metrics using box and whisker plots and quantile (cumulative distribution) plots. In general, mitigation
sites had fewer numbers and types of sensitive taxa and a higher proportion of eurytopic taxa (i.e., taxa
that are adapted to a wide range of environmental conditions). Table 37 shows estimated BCG condition
based on data analysis, expert judgment and the provisional wetland biological assessment model
(DiFranco et al. 2013). Results of this study indicated that community structure is significantly different
between a set of 51 reference wetlands and nine mitigation wetlands based on taxa tolerance metrics
and BCG level. This type of information can improve monitoring and assessment of mitigation sites.
37 See NRPA, http://www.maine.gov/dep/land/nrpa/index.html (Accessed February 2016), 38 M.R.S.A. § 480 A-BB.
174
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Sensitive Taxa Relative Richness Sensitive Taxa Relative Abundance Maine Tolerance Index
POOPOPO 00000000 hi U ft ID
o-trooj.fe.cfta> o-»roG>ifetno>->j °
i
' ' | _
t
1
MtigaNon Reference
Site Type
|~~
0
k
_
F^=|
1.0
09
08
BjO.7
QQ6
°Q5
lo.4
E
U-0.3
-0.2
01
00
i
09
O.S
.0.7
Q0.6
I05
tjO.4
IJL0.3
0.2
0.1
f 0
/ o
f °
I °
f 0
K* 1 1
Site Type
o Mitigation
x Reference
!0 M 40 SO
Maine Tolerance Index
i i i i i iw*
0 f x
X
o jjT
- / :
0 /
I
Site Type
o Mitigation
x Reference
Mtigrtioo Ri*erence 00 Q1 Q2 03 Q4 05 06 Q?
site Type Sensitive Taxa Rel. Abundance
«
•
T
-
-
Mtigation Reference
Site Type
1.U
0.9
as
«o.7
ID
O0.6
]os
t)0.4
£
U-0.3
0.2
0.1
i i i i rj
t*
a jf
a
-of
"o f
' /, I , "
Site Type
o Mitigation
x Reference
00
00 01 0.2 0.3 04 0.5 Q6
Rencitiup Tava Rpl Rirhnpcc
Figure 63. Comparison of reference and mitigation sites for the Maine Tolerance Index and sensitive/tolerant
taxa metrics (reference site N=51; mitigation site N=9) (DiFranco et al. 2013).
175
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 37. Measured values of chemical and watershed stressors, attained water quality classes, and
corresponding BCG levels of reference wetlands and mitigation wetlands (DiFranco et al. 2013)
Mitigation Site
Station
Number
Reference site
range
Reference site
mean
W-171
W-173
W-174
W-175
W-179
W-180
W-181
W-182
W-184
Specific
Conductance
uS/cm
9-95
30.6
98
141
57
25
265
76
163
1120
234
Total
Phosphorus
(mg/L)
.005-097
.017
0.15
0.22
0.071
0.013
0.051
0.032
0.091
0.069
0.027
MEDEP Human
Disturbance
Score
1-10
5
26
20
10
23
23
22
24
40
22
% Watershed
Alteration
0-5.5
1.9
24.1
74.7
37.6
16.7
84.0
21.9
39.9
100
73.3
Assigned
Legislative
Class
B
B
C
B
B
B
C
B
B
BCG
Level
2.5-4.5
2.8
5.23
5.5
4.2
4.2
5.5
4.2
4.8
4.5
4.5
1 Reference site classification attainment: Class AA/A or Class B: 78%; Class C: 8%; Non-attainment: 0
2 Non-attainment of Class C (i.e., lower than the lowest Maine ALU standards)
3 MEDEP assigns BCG scores utilizing digits to the right of the decimal point to indicate the strength of association, e.g., level 3.2
means "Leans toward level 2"; level 3.5 means "Solid level 3", level 3.8 means "Leans toward level 4".
6.5.8 Conclusion
For Maine, their approach to classifying waters based on current ecological condition provides a direct
linkage to CWA biological integrity objectives and ALU goals. This direct linkage facilitates effective
communication with stakeholders and water quality management decision makers on current conditions
and the likelihood for improvements. As sustained and significant improvements in biological condition
were observed based on systematic monitoring of streams, these improvements were documented and
class assignments for specific streams were upgraded (e.g., Class C to B; Class B to A as appropriate). As
Maine further develops and applies biological assessment tools and data to water bodies other than
streams (e.g., wetlands, estuaries, lakes, large rivers), the BCG is included as part of their toolbox.
176
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
6.6 Ohio: Use of Biological Gradient to Support Water Quality Management
6.6.1 Key Message
Ohio has used biological assessment information in conjunction with chemical water quality and physical
habitat assessments to support water quality management decisions since the late 1970s. While the
Ohio ALL) classification framework pre-dated the BCG by 25 years, it is based on concepts that are
parallel to the BCG, highlighting the relationship between biology, habitat, and the potential for water
quality improvements. Ohio's ecological based approach contributed both technical and implementation
"lessons learned" to conceptualization of the BCG (Davies and Jackson 2006). The state's biological
monitoring and assessment program has provided timely information about the status of individual
water bodies and the data to support water quality management program information needs for more
than 35 years. This includes when biological conditions improve and when revisions of designated uses
are warranted. A systematic process to determine which use(s) is (are) appropriate and attainable for a
stream or river has been and remains the key first step in using biological assessment data to support
water quality management.
6.6.2 Background
A major aspect of the development of the Ohio biological assessment program and tiered ALL)
framework is the experience gained through the sustained development of systematic biological
assessments beginning in the late 1970s and through the 1980s. This is where the methods, concepts,
and theories were tested, applied, and refined, resulting in a tractable system for measuring biological
quality at appropriate spatial scales and through time. Qualitative, narrative guidelines were initially
used to assess biological status via systematic watershed monitoring and assessment. The data and
experiences gained in this early assessment process provided the raw materials for incorporating the
concepts of biological integrity that emerged later. Further refinements were also made to the biological
assessment tools and the tiered uses including how they are assigned and assessed. Keys to the success
of this approach were the initial decisions about indicator assemblages and methods. These have
remained stable through time with no major modifications that could have resulted in disconnections
within the statewide database that is more than 35 years old.
Ohio EPA formally adopted numeric biological criteria into the Ohio Water Quality Standards (Ohio
WQS; Ohio Administrative Code 3745-1) in 1990. The biological criteria have been used to guide and
enhance water quality management programs and assess their environmental outcomes. As a result, the
state refined definitions of some ALUs, adopted new ones, and added numerical biological criteria to
support a tiered approach to water quality management within the Ohio WQS (Table 38). The numeric
biological criteria are an outgrowth of an existing framework of TALUs and narrative biological
assessment criteria that had been in place since the late 1970s (Table 39 and Table 40). Ohio's approach
to biological assessment evolved from an initial reliance on best professional judgment guided by the
narrative biological criteria for determining the quality of fish and macroinvertebrate assemblages to a
more quantitative and independent approach based on calibrated indices and numeric biological
criteria. While the early narrative descriptions of four levels of quality ranging from excellent to poor
(Table 39 and Table 40) predated the BCG, the narrative attributes and the rating of multiple levels of
condition are consistent with the attributes and scaling of the current BCG. These concepts were
retained and further refined with the development of the fish IBI and invertebrate community integrity
index (ICI) and the derivation of numeric biological criteria for the current Ohio TALUs (Figure 64) which
were initially mapped to the BCG as part of the early BCG development workshops hosted by EPA
(Figure 65).
177
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 38. Descriptive summary of Ohio's tiered aquatic life use designations
Aquatic Life
Use
Key Attributes
Why a Water body Would Be
Designated
Practical Impacts
(compared to a baseline of WWH)
Warmwater
Habitat (WWH)
Balanced assemblages of
fish/invertebrates comparable
to least impacted regional
reference condition
Either supports biota consistent
with numeric biological criteria for
that ecoregion or exhibits the
habitat potential to support
recovery of the aquatic fauna
Baseline regulatory requirements
consistent with the CWA "fishable"
and "protection & propagation"
goals; criteria consistent with EPA
guidance with state/regional
modifications as appropriate
Exceptional
Warmwater
Habitat (EWH)
Unique and/or diverse
assemblages; comparable to
upper quartile of statewide
reference condition
Attainment of the EWH biological
criteria demonstrated by both
organism groups
More stringent criteria for DO,
temperature, ammonia, and nutrient
targets; more stringent restrictions
on dissolved metals translators;
restrictions on nationwide dredge &
fill permits; may result in more
stringent wastewater treatment
requirements
Coldwater
Habitat (CWH)
Sustained presence of
Salmonid or non-salmonid
coldwater aquatic organisms;
bonafide trout fishery
Biological assessment reveals
coldwater species as defined by
Ohio EPA (2014); put-and-take trout
fishery managed by Ohio
Department of Natural Resources
Same as above except that common
metals criteria are more stringent;
may result in more stringent
wastewater treatment requirements
Modified
Warmwater
Habitat (MWH)
Warmwater assemblage
dominated by species tolerant
of low DO, excessive nutrients,
siltation, and/or habitat
modifications
Impairment of the WWH biological
criteria; existence and/or
maintenance of hydrological
modifications that cannot or will
not be reversed or abated In the
foreseeable future so that WWH
biological criteria can be attained; a
UAA is required
Less stringent criteria for DO,
ammonia, and nutrient targets; less
restrictive applications of dissolved
metals translators; Nationwide
permits apply without restrictions or
exception; may result in less
restrictive wastewater treatment
requirements
Limited
Resource
Waters (LRW)
Highly degraded assemblages
dominated exclusively by
tolerant species; should not
reflect acutely toxic conditions
Extensive physical and hydrological
modifications that cannot be
reversed, are essentially
irretrievable and which preclude
attainment of higher uses; a UAA is
required
Chemical criteria are based on the
prevention of acutely lethal
conditions; may result in less
restrictive wastewater treatment
requirements
178
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Table 39. Narrative biological criteria (fish) for determining ALU designations and attainment of CWA
goals (November, 1980; after Ohio EPA 1981)
Evaluation
Class
Category
1.
2.
3.
4.
5.
6.
"Exceptional"
Class 1
(EWH)
Exceptional or unusual
assemblage of species
Sensitive species
abundant
Exceptionally high
diversity
Composite index
> 9.0-9.5
Outstanding recreational
Fishery
Rare, endangered, or
threatened species
present
"Good"
Class II
(WWH)
Usual association of
expected species
Sensitive species present
High diversity
Composite index
> 7.0-7.5; < 9.0-9.5
"Fair"
Class III
Some expected species
absent, or in very low
abundance
Sensitive species absent,
or in very low abundance
Declining diversity
Composite index
> 4.5-5.0; < 7.0-7.5
Tolerant species
increasing, beginning to
dominate
"Poor"
Class IV
Most expected species
absent
Sensitive species absent
Low diversity
Composite index
< 4.0^.5
Tolerant species dominate
Conditions: Categories 1, 2, 3, and 4 (if data are available) must be met and 5 or 6 must also be met in order to be
Table 40. Narrative biological criteria (macroinvertebrates) for determining ALU designations and
attainment of CWA goals (November 1980; after Ohio EPA 1981)
Evaluation
Class
Category
1.
2.
3.
4.
5.
"Exceptional"
Class 1
(EWH)
Pollution sensitive species
abundant
Intermediate species
present in low numbers
Tolerant species present
in low numbers
Number of taxa > 301
Exceptional diversity
Shannon index < 3.5
"Good"
Class II
(WWH)
Pollution sensitive species
present in moderate
numbers
Intermediate species
present in moderate
numbers
Tolerant species present
in low numbers
Number of taxa 25-30
High diversity
Shannon index 2.9-3.5
"Fair"
Class III
Pollution sensitive species
present in low numbers
Intermediate species
abundant
Tolerant species present
in moderate numbers
Number of taxa 20-25
Moderate diversity
Shannon index 2.3-2.9
"Poor"
Class IV designated in a
particular class.
Pollution sensitive species
absent
Intermediate species
present in low numbers or
absent
Tolerant species abundant
(all types may be absent if
extreme toxic conditions
exist)
Number of taxa < 20
Low diversity
Shannon index < 2.3
1Number of quantitative taxa from artificial substrates.
179
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Ohio Biological Criteria: Adopted May 1990
(OAC 3745-1-07; Table 7-14)
Huron Erie Lake Plain (HELP)
Use Size IBI Mlwb ICI
WWH H 28
W 32
B 34
MWH-C H 20
W 22
B
MWH-I B
20
30
NA
7.3
8.6
NA
5.6
5.7
5.7
34
34
34
22
22
22
NA
Erie Ontario Lake Plain (EOLP)
Use Size IBI Mlwb ICI
Eastern Corn Belt Plains (ECBP)
Use Size IBI Mlwb
WWH H 40 NA
W 40 8.3
B 42 8.5
MWH-C H 24 NA
W 24 6.2
B 24 5.8
MWH-I B 30 6.6
Interior Plateau (IP)
Use Size IBI Mlwb
WWH H 40 NA
W 40 8.1
B 38 8.7
MWH-C H 24 NA
W 24 6.2
B 24 5.8
MWH-I B 30 6.6
ICI
36
36
36
22
22
22
NA
ICI
30
30
30
22
22
22
NA
WWH H
W
-, _ B
/^
~S /
Erie-Ontario
Lake Plain
(EOLP)
^—
/ MWH-C H
^ W
B
MWH-I B
40
38
40
24
24
24
30
NA
7.9
8.7
NA
6.2
5.8
6.6
34
34
34
22
22
22
NA
Western Allegheny Plateau (WAP)
Use Size IBI Mlwb ICI
WWH H
W
B
MWH-C H
W
B
MWH-A H
W
B
MWH-I B
44
44
40
24
24
24
24
24
24
30
NA
8.4
8.6
NA
6.2
5.8
NA
5.5
5.5
6.6
34
34
34
22
22
22
30
30
30
NA
Statewide Exceptional Criteria
Use .Size IBI Mlwb ICI
EWH H 50 NA 46
W 50 9.4 46
B 48 9.6 46
Figure 64. Numeric biological criteria adopted by Ohio EPA in 1990, showing stratification of biological criteria by
biological assemblage, index, site type, ecoregion forwarmwater and modified warmwater habitat (WWH and
MWH, respectively), and statewide for the exceptional warmwater habitat (EWH) use designations.
Developed and adopted by Ohio EPA in 1978, the original tiered aquatic life use narratives represented a
major revision to a general use framework that was adopted in 1974. Ohio's tiered uses recognized the
different types of warmwater aquatic assemblages that corresponded to the mosaic of natural features
of the landscape and nearly two centuries of human-induced changes. The eventual development of
more refined tiered uses and numeric biological criteria that are in place today was the result of
sustained state support to develop a biological monitoring and assessment program with technical
capability to discriminate incremental changes in biological condition with increasing stress. The
empirical evidence used to develop the initial concepts for tiered uses can be found in comprehensive
works on the natural history and zoogeography of the Midwest such as Fishes of Ohio (Trautman 1957,
1981). This and other natural history texts documented the natural and human-caused variations in the
distribution, composition, and abundance of biological assemblages over space and through time
including before and after European settlement. Trautman (1957) not only provides a detailed narrative
of Ohio's natural history, but describes the biological evidence that was used to formulate the initial
concepts about biological integrity that emerged in the late 1970s and early 1980s and which were later
incorporated in the BCG. Such works also described the key features of the landscape that influence and
determine the potential aquatic fauna of water bodies and were the forerunners of the regionalization
frameworks that appeared soon after. As an alternative to a "one-size-fits-all" approach, these provided
180
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
an important foundation for the development of Ohio's tiered uses. The articulation of a practical
definition of biological integrity by Karr and Dudley (1981) provided a theoretical framework for the
development of Ohio's numeric biological criteria (Figure 65). Key components of this framework are:
(1) using biological assemblages as a direct measure of ALL) attainment status (Herricks and Schaeffer
1985; Karr et al. 1986), (2) the development and use of IBIs as assessment tools (Karr 1981; Karr et al.
1986), (3) derivation of regional reference condition to determine appropriate and attainable ALL) goals
and assessment endpoints (Hughes et al. 1986), and (4) systematic monitoring and assessment of the
state's rivers and streams using a pollution survey design. These represented a major advancement over
previous attempts (Ballantine and Guarria 1975) to define and develop a workable framework to
address the concept of biological integrity. Embedded in this framework is the recognition that water
quality management must be approached from an ecological perspective that is grounded in sound
ecological theory and which is validated by empirical observation. This means developing monitoring
and assessment and WQS to encompass the five factors that determine the integrity of a water resource
Figure 22; Karr et al. 1986).
Natural
Minimally
Disturbed
c
.0
? Least
c Impacted
O
U
"re
1
— Degraded
ffi
Severely
DdororloH
[ONRW]
^^^r 2 EWH
^3
WWH
4
MWH
5
LRW
6
1
•
State-
wide
WAP
ECBP
IP
EOLP
HELP
Non-
HELP
HELP
State-
wide
NA
Low
Stressor Gradient
High
Figure 65. An initial mapping of the Ohio TALUs to the BCG relating descriptions of condition along the yl-axis
and ranges of condition encompassed by the numerical biological criteria for each of four tiered use
subcategories and the highest antidegradation tier (ONRW) along the y2-axis. ONRW - Outstanding National
Resource Waters; EWH - Exceptional Warmwater Habitat; WWH - Warmwater Habitat; MWH - Modified
Warmwater Habitat; LRW - Limited Resource Waters.
The understanding offish and macroinvertebrate assemblage responses to stressor gradients ranging
from minimally disturbed to severely altered conditions was affirmed by repeated empirical
observations of assemblage responses which are depicted in Figure 66. This graphic represents
measured assemblage abundance (y-axis) against assemblage indices (fish IBI, macroinvertebrate ICI;
x-axis) with the response of selected metrics and other assemblage attributes at increments along what
181
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
|I
rf 2
^ !Q
o .
u>>.
Cw ^E
1 =
E o
reference reference
"Degraded";
highly tolerant
taxa pre-
dominate;
reduced
abundance;
anomalies
increasing
"Degraded"
quality; toxics >
chronic; low
D.O., nutrients
» reference
Physical Habitat & Flow Regime
Excellent
quality habitat
& flow regime;
recovered from
human-made
modifications
Point sources
present, do not
dominate flows;
NPS impacts
buffered by
extensive
riparian system
Good quality
habitats flow
regime; de
minimis human
modifications
Fair quality
habitats flow
regime; active
human modifi-
cations;
incomplete
recovery
Poor quality
habitats flow
regime; active
human modifi-
cations; no
recovery
Examples of Sources and Activities
Point sources
may dominate
flows; NPS
impacts
buffered by
good riparian
zones
PS/N PS enrich-
ment impacts;
NPS unbufferd;
channel modifi-
cations; im-
poundments
Gross PS/NPS
enrichment
impacts inc.
CSOs;NPS
unbufferd; chan-
nel modifications;
urbanization
"Severely
degraded"; very
low numbers;
few taxa; very
high %
anomalies; toxic
signatures
"Extremely poor"
quality; toxics >
acute; very low
D.O., nutrients »
reference;
contaminated
sediments
Severe modifi-
cations; ephemeral
flows; active human
modifications; no
recovery potential
Severe PS/NPS
toxic impacts;
extreme channel
modifications;
urbanization; acid
mine drainage,
severe thermal
Figure 66. Descriptive model of the response of fish and macroinvertebrate assemblage metrics and
characteristics to a quality gradient and different levels of impact from stressors in Midwestern U.S. warmwater
rivers and streams (modified from Ohio EPA 1987 and Yoder and Rankin 1995b).
182
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
is in reality a continuum. Biological descriptions correspond to the six levels of the then emerging BCG
model and include descriptions of key assemblage characteristics, chemical water quality conditions,
physical habitat and flow regime, and sources of stress that are typically associated with each. This was
modified from the original conceptual model of Ohio EPA (1987a) and Yoder and Rankin (1995b), and it
includes the probable upper limits of Ohio's fish and macroinvertebrate indices. It demonstrates that
understanding the relationship between assemblage responses and stressors is a fundamental aspect of
using biological assessments to support condition assessments and water quality management
programs. It also demonstrates the pre-BCG concepts that eventually merged in the formal
development and description of the current BCG.
6.6.3 Determining Appropriate Levels of Protection
By merging the ALL) framework with systematic monitoring and assessment, Ohio has been able to
determine attainable levels of condition for streams and rivers and also to set protection levels for high
quality waters. This framework is consistently applied within a rotating basin sequence of "biological
surveys" that address the following questions:
1) Is the current designated ALL) appropriate and attainable and if not, what is the appropriate use
for a water body?
2) Are the biological criteria for the most appropriate and attainable use tier attained?
3) Have there been any changes through time and what do they portend for water quality
management?
The scale of monitoring and assessment is sufficiently detailed so that designations of individual water
bodies or segments of a water body can be made based on scientific information and data. Getting this
task done correctly affects everything that follows including assessments of condition and which WQS
will guide water quality management actions such as permitting and TMDLs. The data gathered by a
biological survey is processed, evaluated, and synthesized in a biological and water quality report. The
report serves as the rationale for justifying recommended changes to a currently assigned ALL). The
report also identifies sources of pollutants and/or pollution contributing to impairment(s) of the
recommended designated uses. The recommendations for use designation revisions are a direct output
of the biological and water quality assessment. Recommended revisions to the WQS are based on a UAA
framework that emphasizes the demonstrated potential to attain a particular use tier based on the
following information (and in order of importance):
1) Attainment of the numeric biological criteria for WWH38 or EWH results in designation of that
use; or,
2) If the WWH biological criteria are not attained, the habitat determined by the Qualitative
Habitat Evaluation Index (QHEI; Rankin 1995) based on an assessment of habitat attributes is
used to determine the potential to attain WWH.
38 WWH - Warmwater Habitat is the minimum condition that meets the 101[a][2] goal of the Clean Water Act under the Ohio
WQS. A UAA is required to designate a river or stream to a lower use (e.g., MWH or LRW).
183
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
For uses below WWH (i.e., MWH or LRW), a UAA is performed and includes consideration of the
restorability of the water body and of the factors that may preclude WWH attainment. This process
requires the following information:
1) The current attainment status of the water body based on a biological assessment performed in
accordance with the requirements of the biological criteria, the Ohio WQS, and the Five-Year
Monitoring Strategy;
2) A habitat assessment to evaluate the potential to attain WWH; and,
3) A reasonable relationship between the impaired status and the precluding human-caused
activities based on an assessment of multiple indicators used in their most appropriate indicator
roles and a demonstration consistent with 40 CFR Part 131.10[g].
Since 1978 Ohio EPA has used a consistent process to validate and, if necessary, revise uses in the Ohio
WQS. The codified uses for approximately 2,000 streams and rivers have been revised using this process
(Figure 67) and information from a biological and water quality assessment. This became a routine
practice once the assessment criteria and decision making process for UAAs were established in the mid-
1980s. It required the parallel development of reliable tools, particularly for determining status,
assessing habitat, and determining causal associations, all of which is part of the developmental process
described in several documents and publications (Ohio EPA 1987; 2006; Rankin 1989; 1995; Yoder 1995).
The terms "upgrade" and "downgrade" are used only as descriptions of the direction of change from the
current codified use to that derived from systematic monitoring and assessment. The vast majority of
these changes are from the baseline of original designations that were made in 1978 without the benefit
of systematic monitoring and assessment data, numerical biological criteria, and refinements in the
process that occurred in the mid-1980s. Hence, these original designations are merely being replaced by
the most appropriate use designation based on consistently applied criteria and assessments.
Undesignated streams are almost always smaller watersheds of < 5-10 mi2 drainage area that were
missed by the default stream naming format that was employed when stream and river specific
designations were originally adopted in 1985. Prior to that time, smaller tributaries were
"automatically" assigned the use tier of the parent mainstem river or stream, a practice that resulted in
numerous erroneous use designations. The more frequent monitoring of these smaller streams and
watersheds in the 1990s and 2000s was partially the result of a shift in emphasis to watershed based
TMDLs which resulted in numerous undesignated streams being monitored and hence designated for
the first time. A detailed fact sheet is prepared for each use designation rulemaking to communicate the
types of proposed changes to the WQS, the rationale for the changes, and which rivers and streams are
affected by the proposed changes. When use designation rulemakings are underway, fact sheets specific
to affected river basins can be found on Ohio EPA's website.39
39 See http://epa.ohio.gov/dsw/dswrules.aspxtfl20473212-early-stakeholder-outreach. Accessed February 2016.
184
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Ohio Aquatic Life Use Revision History 1978-2016
LLI
^
O
LLI
(f>
cc.
LLI
LLI
o:
_
O
a:
LLI
m
S
D "UPGRADES'
D "DOWNGRADES'
'_ D PREVIOUSLY UNDESIGNATED
8
(I)
:« 1
0) il
•c o>
re p
S o
? P
?
"(A
0)
re
i_
O)
Q.
D
"(A
0)
re
O)
^
o
Q
*J
Sre |
sSSS: ^ •»:
/Undesigr
^— I ;,
S 3 §
tiSS O ^
1 1
M9-..I
"(A
m
y/
m
£
O)
Q.
D
"(A
0)
re
i_
O)
c
o
0)
13
fffffi
Undesign
^ >S
(A
iH 3 g
jlgj Q Stt
1
Q.
1978-1992
1993-2001
YEAR INTERVAL
2002-2016
Figure 67. The number of individual stream and river segments in which ALL) designations were revised during
1978-1992,1993-2001, and 2002-2016. Cases where the use was revised to a higher use are termed "upgrades"
and cases where a lower use was assigned are termed "downgrades." Previously undesignated refers to streams
that were not listed in the 1985 WQS, but which were added as each was designated as a result of systematic
monitoring and assessment. The number of waters previously undesignated in the first interval is unknown.
The Ohio tiered use and biological criteria framework and their application to Ohio rivers and streams
were first tested in the Ohio court system in 1989 and were validated by a lower court and upheld in
appeals up to, and including, the Ohio Supreme Court (NEORSD vs. Shank No. 89-1554, Supreme Court
of Ohio, Feb. 27, 1991). The application of the biological criteria to justify additional pollution controls in
response to a biological impairment was likewise validated by a lower court and upheld in subsequent
appeals (City of Salem vs. Korleski No. 09AP-620, Tenth District Court of Appeals, March 23, 2010; Ohio
Supreme Court 2010-0818; appeal not accepted, August 25, 2010).
185
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
6.6.4 Setting Attainable Goals for Improvements
Ecologically-based tiered uses, a systematic approach to monitoring and assessment, and a tractable
UAA process can provide substantial benefits for water quality management programs related to guiding
efforts to improve conditions and assessing the effectiveness of those efforts in protecting and restoring
an ALL). The identification of the recovery potential for aquatic life in a water body using a systematic
approach can help set attainable goals for improvements and support evaluation of environmental risks.
The Ohio case example illustrates the role of tiered ALUs using a BCG approach for interpretation of
conditions, systematic monitoring and assessment, and a consistent process for conducting UAAs in
support of TMDLs. The UAA process is routinely applied as a result of the systematic monitoring and
assessment of Ohio rivers and streams (Figure 68). The data are used to support recommendations for
revisions to the Ohio WQS on an annual basis.
Functional Support Provided by Annual
Rotating Basin Assessments
Individual
Basin
Assessment
WQS/Use
Attainability
Analyses
NPDES
Permits
Permit
Development
Watershed
Specific Issues
• TMDL develop-
ment
• Local water-
shed groups
• 319 projects
• 404/401 dredge
& fill permits
• Problem
discovery
• Special
Investigations
Integrated
Report
Annual
WQS Rule
Revisions
303d List of
Impaired/Threat
ened Waters
Permit
Defense/
Fact Sheets
Enforcement
Support
Goals
Tracking
(GPRA, Ohio
2010)
Figure 68. The flow of information from biological and water quality assessments to support for major water
quality management programs in Ohio.
Ohio's tiered ALL) designation procedures were incorporated into the TMDL process beginning in 1999
(Figure 69; Ohio EPA 1999). Figure 69 illustrates the steps for validating the most appropriate tiered ALL)
and then basing a TMDL on the criteria embodied by that use tier and the attendant assessment of the
receiving streams and rivers. It also illustrates the delineation of the severity and extent of impairments,
the most probable causes of the impairments, and follow-up assessments to validate TMDL
effectiveness. Because the Ohio EPA monitoring and assessment strategy includes chemical, physical,
and biological indicators which are used in their most appropriate roles as indicators of stress, exposure,
and response (Yoder and Rankin 1998), support for the development of TMDLs can go beyond
addressing singular pollutants to addressing the combination of pollution and pollutants that impair an
ALL).
186
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
TMDL Process Under a TALU Framework
Watershed Assessment Process
•Are uses appropriate & attainable?
'Determine & quantify attainment status.
•Characterize extent & severity of impairments.
'Delineate associated causes & sources.
Technical Reports
•WQS use revisions (UAA).
•Use attainment status.
•Permit support document.
•Other specialized reports.
Follow-up
Assessment
305b/303d
•Assessment Data Base (ADB),
•Integrated report
303d Listings
TMDL Development
TMDL Implementation
Figure 69. Key steps showing how a TALU based framework can be used to organize and guide a TMDL
development and implementation process.
187
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
6.6.5 Protecting High Quality Water Bodies
Ohio's antidegradation rule (Ohio Administrative Code 3745-1-05) incorporates levels of protection
between the minimum required under the CWA and the maximum protection afforded by federal
regulations. The most stringent application of antidegradation is to disallow any lowering of water
quality in waters listed as ONRWs. The minimum requirement allows for a lowering of water quality to
the minimum WQS applicable to the water body if a determination is made that lowering water quality
is necessary to accommodate important social and economic development. However, lowering of water
quality below that which is necessary to protect an existing use is prohibited. Ohio has two intermediate
levels of protection for certain ecologically important water bodies that permanently reserve a portion
of the unused pollutant assimilative capacity, thereby assuring maintenance of a water quality that is
better than that prescribed by the prevailing designated use tier. The two intermediate levels are: (1)
Outstanding State Water (OSW; Figure 70), and (2) Superior High Quality Water (SHQW) which fall in
between ONRW and General High Quality Waters (GHQWs; Figure 71). High quality water bodies are
valued public resources because of their ecological and human benefits. Their biological components act
as an early warning system that can indicate potential threats to human health, degradation of aesthetic
values, reductions in the quality and quantity of recreational opportunities, and other ecosystem
Figure 70. The Mohican River in northeastern Ohio—a candidate for OSW classification because of its high
quality ecological and recreational attributes.
188
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Natural
Minimally
Disturbed
.0
:§ Least
c Impacted
3
s
00
— Degraded
CO
Severely
Hocrr^rlarl
- [ONRW] _
1 ^
[OSW] 4
2 EWH
3 [SHQW]H
[GHQWT-
\ 4
5 [LQW]-
6
WWH
MWH
LRW
-
•
(State-
wide
WAP
ECBP
IP
EOLP
HELP
Non-
HELP
HELP
State-
wide
NA
Low
Stressor Gradient
High
Figure 71. Mapping the Ohio antidegradation tiers to the BCG relating descriptions of condition along the yl-axis
and ranges of condition encompassed by the numerical biological criteria for each of four tiered use
subcategories and the four antidegradation tiers along the y2-axis. ONRW - Outstanding National Resource
Waters; OSW - Outstanding State Waters; SHQW - Superior High Quality Waters; GHQW - Generally High
Quality Waters; LQW - Low Quality Waters; EWH - Exceptional Warmwater Habitat; WWH - Warmwater
Habitat; MWH - Modified Warmwater Habitat; LRW - Limited Resource Waters.
benefits, or services. The ability of streams and rivers to provide these beneficial services and to act as
environmental sentinels is reduced whenever their integrity is degraded. Under the Ohio
antidegradation rule, a portion of the remaining assimilative capacity is reserved for water bodies
classified as OSW or SHQW in order to preserve an already existing high quality.
Ohio uses a number of biological and physical attributes to place river and stream segments into the
OSW, SHQW, and GHQW antidegradation tiers (Table 41). Included are the presence of state or federally
listed endangered and threatened species, declining fish species (as defined in the antidegradation
rules), the fish and macroinvertebrate assemblage indices (IBI and ICI), the QHEI, the vulnerability of the
river or stream to increased stressors, the relative abundance of fish species sensitive to pollution and
habitat destruction, and the accumulation of multiple attributes. Adjustments are also made for the
Lake Erie drainage to account for the fewer endemic fish and mussel species. Additional considerations
include other designations, such as state and national scenic river status, outstanding biodiversity
among all aquatic assemblages, exceptionally high quality habitat, and the presence of unique landforms
along geological and geomorphological boundaries.
Table 41. General guidelines for nominating OSW, SHQW, and GHQW categories in Ohio. Attributes
are considered both singly and in the aggregate
Attribute
OSW
SHQW
GHQW
189
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Attribute
Endangered &
Threatened
Species
Declining Fish Species
IBI and ICI
QHEI
Vulnerability
Relative Abundance of Fish
Species Sensitive to Pollution
and Habitat Destruction
Multiple Attributes
osw
Multiple species; large
populations; include the
most vulnerable species
>4 declining fish
species/segment; large
populations
High mean scores; very
high max scores (> 56)
High percentage of QHEI
scores > 80
Little wastewater
effluent; high
vulnerability
Relative abundance is >
3 standard deviations
compared to statewide
collections of similar
sized streams
High co-occurrence of
above attributes
SHQW
Present, smaller
populations; may include
less vulnerable species
2-4 declining fish
species/segment;
moderate populations
Lower mean scores; fewer
high max scores or, if more
high scores, few other
attributes
Fewer QHEI scores > 80,
many above 70
May be more wastewater
effluent; moderate
vulnerability
Relative abundance is > 2
standard deviations
compared to statewide
collections of similar sized
streams
Lower co-occurrence of
above attributes or
individual attributes more
marginal
GHQW
Absent or, if present, small
populations or of low vulnerability
< 2 declining fish species/segment;
typically small populations
Lower mean scores; few or no very
high max scores
Few or no QHEI scores > 80, fewer
above 70
Lower vulnerability; for vulnerable
components, antidegradation
application may still be denied
Relative abundance is < 2 standard
deviations compared to statewide
collections of similar sized streams
Little co-occurrence of above
attributes, individual attributes
often marginal if present
6.6.6 Conclusion
The Ohio approach to classifying waters based on current ecological condition and potential for
improvement provides a direct linkage to the CWA biological Integrity objective and ALL) goals. This
direct linkage enables more effective communication with stakeholders and water quality management
decision makers on current conditions and likelihood for improvements. The BCG-like approach enables
Ohio EPA to account for biological expectations relative to ecoregion and drainage area and provides a
numeric value that synthesizes everything that is being experienced by the biota that can be tracked,
monitored, and compared over time to determine if conditions are improving, stabilizing, or
deteriorating. As chemical, physical, and biological monitoring has been coordinated and the database
expanded, critical information for investigating cause and source of biological impairments has been
built and has enabled water quality managers to target sources of stressors and their mechanism of
action on the aquatic ecosystem. Because of this database, the state has been able to develop water
quality goals for some parameters less well-suited to the classic dose-response relationship for DO and
many toxicants. Ohio's ecologically-based approach to classifying waters combined with a robust
monitoring program has provided a scientifically defensible method to categorize waters into
designated uses and antidegradation tiers. The process has generated UAAs and justification documents
as an accepted and routine rulemaking process, primarily resulting in incremental upgrades as controls
and BMPs were implemented and improvements observed.
190
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
References
ADEM. 2005. ADEM Surface Water Quality Monitoring Strategy. Alabama Department of Environmental
Management.
http://www.adem.state.al.us/programs/water/waterforms/SurfaceWaterMonitoring.pdf.
Accessed February 2016.
ADEM. 2012. State of Alabama Water Quality Monitoring Strategy. Alabama Department of
Environmental Management.
http://www.adem.state.al.us/programs/water/wqsurvev/2010WQMonitoringStrategy.pdf.
Accessed February 2016.
Allan, J.D., L.L. Yuan, P. Black, T.O.M. Stockton, P.E. Davies, R.H. Magierowski, and S.M. Read. 2012.
Investigating the relationships between environmental stressors and stream condition using
Bayesian belief networks. Freshwater Biology 57:58-73. Retrieved from 10.1111/J.1365-
2427.2011.02683.x.
Allan, J.D., P.B. Mclntyre, S.D.P. Smith, B.S. Halpern, G.L. Boyer, A. Buchsbaum, G.A. Burton, Jr., LM.
Campbell, W.L Chadderton, J.J.H. Ciborowski, P.J. Doran, T. Eder, D.M. Infante, L.B. Johnson, C.A.
Joseph, A.L. Marino, A. Prusevich, J.G. Read, J.B. Rose, E.S. Rutherford, S.P. Sowa, and A.D.
Steinman. 2013. Joint analysis of stressors and ecosystem services to enhance restoration
effectiveness. Proceedings of the National Academy of Science 110:372-377.
Anderson, T.W. 1984. An Introduction to Multivariate Statistical Analysis. John Wiley & Sons, New York
675 pp.
Angelo, R.T., M.S. Cringan, and J.E. Fry. 2002. Distributional revisions and new and amended occurrence
records for prosobranch snails in Kansas. Transactions of the Kansas Academy of Science 105(3-
4):246-257.
http://www.kdheks.gov/befs/download/bibliography/ProsobranchSnails RTA 2002.pdf. Accessed
February 2016.
Angelo, R.T., M.S. Cringan, E. Hays, C.A. Goodrich, E.J. Miller, M.A. VanScoyoc, and B.R. Simmons. 2009.
Historical changes in the occurrence and distribution of freshwater mussels in Kansas. Great Plains
Research 19:89-126.
http://www.kdheks.gov/befs/download/bibliography/Angelo et al 2009.pdf. Accessed February
2016.
Arnell, N.W. 1999. Climate change and global water resources. Global Environmental Change 9:531-549.
Baker, M.E., and R.S. King. 2010. A new method for detecting and interpreting biodiversity and
ecological community thresholds. Methods in Ecology and Evolution 1:25-37.
Ballentine, L.K., and L.J. Guarraia (eds.). 1977. Integrity of Water. EPA 055-001-010-01068-1. U.S.
Environmental Protection Agency, Office of Water and Hazardous Materials, Washington, DC.
191
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Barbour, M.T., J. Gerritsen, B.D. Snyder, and J.B. Stribling. 1999. Rapid Bioassessment Protocols for Use
in Streams and Wadeable Rivers: Periphyton, Benthic Macroinvertebrates and Fish, Second Edition.
EPA/841-B-99-002. U.S. Environmental Protection Agency, Office of Water, Washington, DC.
http://www.waterboards.ca.gov/water issues/programs/tmdl/docs/303d policvdocs/161.pdf.
Accessed February 2016.
Barnthouse, L.W., and J. Brown. 1994. Issue paper on conceptual model development. Chapter 3 in
Ecological Risk Assessment Issue Papers. EPA/630/R-94/009. U.S. Environmental Protection
Agency, Office of Research and Development, Risk Assessment Forum. DIANE publishing
Company. ISBN 0788119591, 9780788119590.
https://books.google.com/books?id=QatzcSsidzkC&dq=Ecological+risk+Assessment+lssue+Papers
&source=gbs navlinks s. Accessed February 2016.
Bartsch, W.M., R.P. Axler, and G.E. Host. 2015. Evaluating a Great Lakes scale landscape stressor index to
assess water quality in the St. Louis River Area of Concern. Journal of Great Lakes Research 41:99-
110.
Bay, S.M., and S.B. Weisberg. 2010. Framework for interpreting sediment quality triad data. Integrated
Environmental Assessment and Management 8(4):589-596.
Bay, S.M., W. Berry, P.M. Chapman, R. Fairey, T. Gries, E. Long, D. MacDonald, and S.B. Weisberg. 2007.
Evaluating Consistency of Best Professional Judgment in the Application of a Multiple Lines of
Evidence Sediment Quality Triad. Integrated Environmental Assessment and Management
3(4):491-497.
Beck, W.M., Jr. 1954. Studies in stream pollution biology: I. A simplified ecological classification of
organisms. Quarterly Journal of the Florida Academy of Sciences 17:211-227.
http://www.biodiversitvlibrary.org/page/41493954tfpage/235/mode/lup. Accessed February
2016.
Beck, W.H., Jr. 1955. Suggested method for reporting biotic data. Sewage and Industrial Waste
27(10):1193-1197.
Bender, E.A., T.J. Case, and M.E. Gilpin. 1984. Perturbation experiments in community ecology: Theory
and practice. Ecology 65:1-13.
Berger, A.R., and R.A. Hodge. 1998. Natural change in the environment: A challenge to the pressure-
state-response concept. Social Indicators Research 44:255-265.
Bierwagen, B.G., AT. Hamilton, J. Stamp, M.J. Paul, J. Gerritsen, L. Zheng, and E.W. Leppo. 2012.
Implications of Climate Change for Bioassessment Programs and Approaches to Account for
Effects. EPA/600/R-11/036F. U.S. Environmental Protection Agency, Office of Research and
Development, Washington, DC. http://cfpub.epa.gov/ncea/global/recordisplay.cfm?deid=239585.
Accessed February 2016.
Blann, K.L., J.L. Anderson, G.R. Sands, and B. Vondracek. 2009. Effects of agricultural drainage on aquatic
ecosystems: A review. Critical Reviews in Environmental Science and Technology 39(11):909-1001.
192
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Blocksom, K.A. 2003. A performance comparison of metric scoring methods for a multimetric index for
mid-Atlantic highlands streams. Journal of Environmental Management 31:670-682.
ftp://ftp.chesapeakebav.net/Monitoring/Foreman/2009%20Benthic%20Database/review/Literatu
re/EnvironmentalManagement comparescoring.pdf. Accessed February 2016.
Boucher, K. 2014. Letter from Kathleen Boucher, Maryland Department of Environmental Protection, to
Keith Levchenko dated January 13, 2014.
Bradley, P., D.L. Santavy, and J. Gerritsen. 2014. Workshop on Biological Integrity of Coral Reefs. August
21-22, 2012 Caribbean Coral Reef Institute, Isla Magueyes, La Parguera, Puerto Rico.
EPA/600/R13/350. U.S. Environmental Protection Agency, Office of Research and Development,
Washington, DC. http://cfpub.epa.gov/si/si public file download.cfm?p download id=522578.
Accessed February 2016.
Brinkhurst, R. 1993. Future directions in freshwater biomonitoring. In Freshwater Biomonitoring and
Benthic Macroinvertebrates, D.H. Rosenberg and V. H. Resh (eds.), pp. 442-460. Chapman and
Hall, New York.
Brooks, R., M. McKenney-Easterling, M. Brinson, R. Rheinhardt, K. Havens, D. O'Brien, J. Bishop, J.
Rubbo, B. Armstrong, and J. Hite. 2009. A Stream-Wetland-Riparian (SWR) index for assessing
condition of aquatic ecosystems in small watersheds along the Atlantic slope of the eastern U.S.
Environmental Monitoring and Assessment 150:101-117.
Brown, M.T., and M.B. Vivas. 2005. Landscape development index. Environmental Monitoring and
Assessment 101:289-309.
Bryce, S.A., D.P. Larsen, R.M. Hughes, and P.R. Kaufmann. 1999. Assessing relative risks to aquatic
ecosystems: A mid-Appalachian case study. Journal of the American Water Resources Association
35:23-36.
Buchwalter, D.B., and S.N. Luoma. 2005. Differences in dissolved cadmium and zinc uptake among
stream insects: Mechanistic explanations. Environmental Science and Technology 39:498-504.
http://www.ephemeroptera-galactica.com/pubs/pub b/pubbuchwalterd2005p498.pdf. Accessed
February 2016.
Cairns, J., Jr. 1977. Quantification of Biological Integrity. In The Integrity of Water, Proceedings of a
Symposium, ed. R.K. Ballentine and L.J. Guarraia, U.S. Environmental Protection Agency,
Washington, DC, March 10-12, 1975, pp. 171-187.
Cairns, J. Jr. 1981. Biological monitoring part Vl-future needs. Water Research 15:941-952.
Cairns, J., Jr., and J.R. Pratt. 1993. A History of Biological Monitoring Using Benthic Macroinvertebrates.
In Freshwater Biomonitoring and Benthic Macroinvertebrates, ed. D.M. Rosenberg and V.H. Resh,
pp. 10-27. Chapman & Hall, New York.
Cairns, J. Jr., P.V. McCormick, and R.R. Niederlehner. 1993. A proposed framework for developing
indicators of ecosystem health. Hydrobiologia 263:1-44.
193
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Cao, Y., C. Hagedorn, O.C. Shanks, D. Wang, J. Ervin, J.F. Griffith, B.A. Layton, C.D. McGee, I.E. Reidel,
and S.B. Weisberg. 2013. Towards establishing a human fecal contamination index in microbial
source tracking. International Journal of Environmental Science and Engineering Research 4(3):46-
58.
Carlson, R.E. 1992. Expanding the trophic state concept to identify non-nutrient limited lakes and
reservoirs. In Proceedings, National Conference on Enhancing the States' Lake Management
Programs, Chicago, IL, 1991, pp. 59-71. North American Lake Management Society.
https://www.researchgate.net/publication/246134025 Expanding the trophic state concept to
identify non-nutrient limited lakes and reservoirs. Accessed February 2016.
Castella, E., and M.C.D. Speight. 1996. Knowledge representation using fuzzy coded variables: An
example based on the use of Syrphidae (Insecta, Diptera) in the assessment of riverine wetlands.
Ecological Modelling 85:13-25.
Chen, K., R.M. Hughes, S. Xu, J. Zhang, D. Cai, and B. Wang. 2014. Evaluating performance of
macroinvertebrate-based adjusted and unadjusted multi-metric indices (MMI) using multi-season
and multi-year samples. Ecological Indicators 36:142-151.
Chen, T., and H. Lin. 2011. Application of a Landscape Development Intensity Index for Assessing
Wetlands in Taiwan. Wetlands 31:745-756.
Chessman, B.C. 2014. Predicting reference assemblages for freshwater bioassessment with limiting
environmental difference analysis. Freshwater Science 33:1261-1271.
Chow-Fraser, P. 2006. Development of the Wetland Water Quality Index for Assessing the Quality of
Great Lakes Coastal Wetlands. Chapter 5 in Coastal Wetlands of the Laurentian Great Lakes:
Health, Habitat and Indicators, ed. T.P. Simon and P.M. Stewart, pp. 137-166. Indiana Biological
Survey, Bloomington, IN.
Ciborowski, J.J.H., G.E. Host, T.A. Brown, P. Meysembourg, and L.B. Johnson. 2011. Linking Land to the
Lakes: The Linkages Between Land-based Stresses and Conditions of the Great Lakes. Background
Technical Paper prepared for Environment Canada in support of the 2011 State of the Lakes
Ecosystem Conference (SOLEC), Erie, PA. 47 p + Appendices.
Comte, L., L. Buisson, M. Daufresne, and G. Grenouillet. 2013. Climate-induced changes in the
distribution of freshwater fish: Observed and predicted trends. Freshwater Biology 58:625-639.
Cormier, S.M, and G.W. Suter. 2013. A method for assessing causation of field exposure-response
relationships. Environmental Toxicology and Chemistry 32:272-276.
Cormier, S.M, G.W. Suter II, and L. Zheng. 2013. Derivation of a benchmark for freshwater ionic strength.
Environmental Toxicology and Chemistry 32(2):263-271.
Cicchetti, G., and H. Greening. 2011. Estuarine biotope mosaics and habitat management goals: An
application in Tampa Bay, Florida, USA. Estuaries and Coasts 34:1278-1292.
Courtemanch, D.L., S.P. Davies, and E.B. Laverty. 1989. Incorporation of biological information in water
quality planning. Environmental Management 13:35-41.
194
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Grain, CM., and M.D. Bertness. 2006. Ecosystem Engineering across environmental gradients:
Implications for conservation and management. BioScience 56(3):211-218. doi:10.1641/0006-
3568(2006)056[0211:EEAEGI]2.0.CO;2.
Danielson, T.J., C.S. Loftin, L. Tsomides, J.L DiFranco, and B. Connors. 2011. Algal bioassessment metrics
for wadeable streams and rivers of Maine, USA. Journal of the North American Benthological
Society 30(4): 1033-1048.
Danielson, T.J., C.S. Loftin, L. Tsomides, J.L. DiFranco, B. Connors, D.L. Courtemanch, F. Drummond, and
S.P. Davies. 2012. An algal model for predicting attainment of tiered biological criteria of Maine's
streams and rivers. Freshwater Science 31(2):318-340.
Danielson, T.J., L. Tsomides, D. Suitor, J.L. DiFranco, B. Connors. In press. Relationship of Impervious
Cover and Attainment of Aquatic Life Criteria for Maine Streams. Maine Department of
Environmental Protection.
Danz, N.P., R.R. Regal, G.J. Niemi, V.J. Brady, T. Hollenhorst, LB. Johnson, G.E. Host, J.M. Hanowski, C.A.
Johnston, T. Brown, J. Kingston, and J.R. Kelly. 2005. Environmentally stratified sampling design for
the development of Great Lakes environmental indicators. Environmental Monitoring and
Assessment 102:41-65.
Danz, N.P., G.J. Niemi, R.R. Regal, T. Hollenhorst, L.B. Johnson, J.M. Hanowski, R.P. Axler, J.J.H.
Ciborowski, T. Hrabik, V.J. Brady, J.R. Kelly, J.A. Morrice, J.C. Brazner, R.W. Howe, C.A. Johnston,
and G.E. Host. 2007. Integrated measures of anthropogenic stress in the U.S. Great Lakes Basin.
Environmental Management 39:631-647.
Davies, S.P., and S.K. Jackson. 2006. The Biological Condition Gradient: A descriptive model for
interpreting change in aquatic ecosystems. Ecological Applications 16:1251-1266.
Davies, S.P., L. Tsomides, D.L. Courtemanch, and F. Drummond. 1995. Maine Biological Monitoring and
Biocriteria Development Program. Maine Department of Environmental Protection, Bureau of
Water Quality Control. DEP-LW108.
Davies, S.P., L. Tsomides, J.L. DiFranco, and D.L. Courtemanch. 1999. Biomonitoring Retrospective:
Fifteen Year Summary for Maine Rivers and Streams. Maine Department of Environmental
Protection, Augusta, ME.
http://www.maine.gov/dep/water/monitoring/biomonitoring/retro/romans.pdf. Accessed
February 2016.
Davies, S.P., and L. Tsomides. 2002. Revised April 2014. Methods for the Biological Sampling and
Analysis of Maine's Rivers and Streams. DEP LW0387-C2014, Augusta, ME.
http://www.maine.gov/dep/water/monitoring/biomonitoring/materials/sop stream macro met
hods manual.pdf. Accessed February 2016.
Davies, S.P., F. Drummond, D.L. Courtemanch, L. Tsomides, T.J. Danielson. In press. Biological water
quality standards to achieve optimal biological condition in Maine rivers and streams: Science and
policy. Maine Agricultural and Forest Experiment Station Technical Bulletin 111 pp.
195
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Davis, W.S. 1995. Biological Assessment and Criteria: Building on the Past. In Biological Assessment and
Criteria: Tools for Water Resource Planning and Decision Making, ed. W.S. Davis and T.P. Simon,
pp. 15-29. Lewis Publishers, Boca Raton, FL
https://www.researchgate.net/publication/235792912 Biological Assessment and Criteria Buil
ding on the Past.Accessed February 2016.
Dayton, P.K., M.J. Tegner, P.B. Edwards, and K.L Riser. 1998. Sliding baselines, ghosts, and reduced
expectations in kelp forest communities. Ecological Applications 8(2):309-322.
Death, G., and K.E. Fabricius. 2000. Classification and regression trees: A powerful yet simple technique
for ecological data analysis. Ecology 81:3178-3192.
https://www.google.com/url?sa=t&rct=i&q=&esrc=s&source=web&cd=3&cad=ria&uact=8&ved=0
CC8QFiACahUKEwiS7ZDliJ IAhWRUplKHWOsA8A&url=http%3A%2F%2Fmoodle.epfl.ch%2Fpluginf
ile.php%2F161301%2Fmod folder%2Fcontent%2FO%2FDe Ath2000 Ecology.pdf%3Fforcedownlo
ad%3Dl&usg=AFQiCNGWIwZvLiZwBvvlfzedfx3bHd6s5g. Accessed February 2016.
Demicco, R.V. 2004. Fuzzy Logic and Earth Science: An Overview. In Fuzzy Logic in Geology, ed. R.V.
Demicco and G.J. Klir, pp. 11-61. Elsevier Academic Press, San Diego, CA.
Demicco, R.V., and G.J. Klir. 2004. Introduction. In Fuzzy Logic in Geology, ed. R.V. Demicco and G.J. Klir,
pp. 1-10. Elsevier Academic Press, San Diego, CA.
DeWalt, R.E., Y. Cao, L. Hinz, and T. Tweddale. 2009. Modelling of historical stonefly distributions using
museum specimens. Aquatic Insects: International Journal of Freshwater Entomology 31(Suppl
l):253-267. Special Issue: Proceedings of the 12th International Conference on Ephemeroptera
and the 16th International Symposium on Plecoptera, Stuttgart, 2008.
DiFranco, J.D., B. Connors, T.J. Danielson, L. Tsomides. 2013. Evaluating Alternative Wetland
Compensatory Mitigation Assessment Techniques. MEDEP document DEPLW-1258, 39 pp.
Droesen, W.J. 1996. Formalisation of ecohydrological expert knowledge applying fuzzy techniques.
Ecological Modelling 85:75-81.
Esselman, P.C., D.M. Infante, L. Wang, A.R. Cooper, D. Wieferich, Y-P. Tsang, D.J. Thornbrugh, and W.W.
Taylor. 2013. Regional fish community indicators of landscape disturbance to catchments of the
conterminous United States. Ecological Indicators 26:163-173.
European Environment Agency. 1999. Environmental Indicators: Typology and Overview. European
Environment Agency, Technical Report No. 25, Copenhagen, 19 pp.
Fausch, K.D., J. Lyons, P.L. Angermeier, and J.R. Karr. 1990. Fish communities as indicators of
environmental degradation. American Fisheries Society Symposium 8:123-144.
https://www.researchgate.net/publication/248554980 Fish communities as indicators of envir
onmental degradation. Bioindicators of stress in fish. Accessed February 2016.
Fausch, K.D., J.R. Karr, and P.R. Yant. 1984. Regional application of an index of biotic integrity based on
stream fish communities. Transactions of the American Fisheries Society 113:39-55.
https://www.researchgate.net/publication/248814396 Regional Application of an Index of Bi
otic Integrity Based on Stream Fish Communities. Accessed February 2016.
196
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Flotemersch, J.E., S.G. Leibowitz, R.A. Hill, J.L Stoddard, M.C. Thorns, and R.E. Tharme. 2015. A
watershed integrity definition and assessment approach to support strategic management of
watersheds. River Research and Applications. doi:10.1002/rra.2978.
Fore, L. 2003. Development and Testing of Invertebrate Biomonitoring Tools for Florida Streams.
Unpublished report submitted to Florida Department of the Environment. 72 pp.
Fore, L.S. 2004. Development and Testing of Biomonitoring Tools for Macroinvertebrates in Florida
Streams. Statistical Design, Seattle, Washington. A report for the Florida Department of
Environmental Protection, Tallahassee, Florida, USA. 62 p.
http://fwcg.myfwc.com/docs/Stream invertebrate assessment protocol.pdf. Accessed February
2016.
Fore, L.S. 2005. Assessing the Biological Condition of Florida Lakes: Development of the Lake Vegetation
Index (LDV). Statistical Design, Seattle, Washington. A report for the Florida Department of
Environmental Protection, Tallahassee, Florida, USA. 29 pp. & Appendices.
http://publicfiles.dep.state.fl.us/dear/sas/sopdoc/lvi final05.pdf. Accessed February 2016.
Frey, D.G. 1977. Biological Integrity, a Historical Approach. In The Integrity of Water, Proceedings of a
Symposium, ed. R.K. Ballentine and L.J. Guarraia, U.S. Environmental Protection Agency,
Washington, DC, March 10-12, 1975, pp. 127-140.
Frich, P., LV. Alexander, P. Della-Marta, B. Gleason, M. Haylock, A.M.G. Klein Tank, T. Peterson. 2002.
Observed coherent changes in climatic extremes during the second half of the twentieth century.
Climate Research 19:193-212.
Gerritsen, J., and E. Leppo. 2005. Biological Condition Gradient for Tiered Aquatic Life Use in New Jersey.
Prepared for U.S. Environmental Protection Agency, Office of Science and Technology,
Washington, D.C. http://www.ni.gov/dep/wms/bears/docs/FINAL%20TALU%20NJ%20RPT 2.pdf.
Accessed February 2016.
Gerritsen, J., and B. K. Jessup. 2007a. Identification of the Biological Condition Gradient for Freestone
(non-calcareous) Streams of Pennsylvania. Prepared for U.S. Environmental Protection Agency,
Office of Science and Technology and Pennsylvania Department of Environmental Protection.
Gerritsen, J., and B.K. Jessup. 2007b. Identification of the Biological Condition Gradient for High Gradient
Streams of Connecticut. Prepared for U.S. Environmental Protection Agency, Office of Science and
Technology and Connecticut Department of Environmental Protection.
Gerritsen, J., and B. Jessup. 2007c. Identification of the Biological condition Gradient for Freestone (Non-
calcareous) Streams of Pennsylvania. U.S. Environmental Protection Agency, Washington, DC.
Gerritsen, J., E.W. Leppo, L. Zheng, and C.O. Yoder. 2012. Calibration of the Biological Condition Gradient
for Streams of Minnesota. Prepared for Minnesota Pollution Control Agency, St. Paul, MN.
197
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Gerritsen, J., and J. Stamp. 2012. Calibration of the Biological Condition Gradient (BCG) in Cold and Cool
Waters of the Upper Midwest: BCG-based indexes (BCG-I) for Fish and Benthic Macroinvertebrate
Assemblages. Prepared for U.S. Environmental Protection Agency, Office of Science and U.S.
Environmental Protection Agency Region 5. https://www.uwsp.edu/cnr-
ap/biomonitoring/Documents/pdf/USEPA-BCG-Report-Final-2012.pdf. Accessed February 2016.
Gerritsen, J., and J. Stamp. 2014. Biological Condition Gradient (BCG) Assessment Models for Lake Fish
Communities of Minnesota. Prepared for Minnesota Pollution Control Agency and Minnesota
Department of Natural Resources, St. Paul, MN.
Gerritsen, J., J. Stamp, D. Charles, and S. Hausmann. 2014. Biological Condition Gradient (BCG)
Assessment Models for Diatom Communities of Upland Streams in New Jersey. Prepared for U.S.
Environmental Protection Agency and New Jersey Department of Environmental Protection.
Gerritsen J., L. Zheng, E. Leppo, and C.O. Yoder. 2013. Calibration of the Biological Condition Gradient for
Streams of Minnesota. Prepared for the Minnesota Pollution Control Agency, St. Paul, MN.
Gerritsen, J., L. Yuan, P. Shaw-Allen, and D. Farrar. 2015. Regional Observational Studies: Assembling and
Exploring Data. In Ecological Causal Assessment, ed. S.B. Norton, S.M. Cormier, and G.W. Suter II,
pp. 155-168. CRC Press, Boca Raton, FL
Gessner, M.O., and E. Chauvet. 2002. A case for using litter breakdown to assess functional stream
integrity. Ecological Applications 12:498-510. https://hal.archives-ouvertes.fr/hal-
00870744/document. Accessed February 2016.
Gibson, G.R., M.T. Barbour, J.B. Stribling, J. Gerritsen, and J.R. Karr. 1996. Biological Criteria: Technical
Guidance for Streams and Small Rivers (revised edition). EPA/822/B/96/001. U.S. Environmental
Protection Agency, Office of Water, Washington, DC. http://www.epa.gov/wqc/biological-
assessment-technical-assistance-documents-states-and-tribes#streams. Accessed February 2016.
Glasby, T.M., and A.J. Underwood. 1996. Sampling to differentiate between pulse and press
perturbations. Environmental Monitoring and Assessment 42:241-252.
Greenberg L, P. Svendsen, and A. Harby. 1996. Availability of microhabitats and their use by brown trout
(Salmo trutta) and grayling (Thymallus thymallus) in the River Vojman, Sweden. Regulated Rivers:
Research & Management 12:287-303.
Harrington, J.W. 2014. Quantifying Land Use Disturbance Intensity (LDI) in the Skokomish River
Watershed: Salmonid Habitat Implications. Master's Thesis, Evergreen State College.
Hawkins, C. 2006. Quantifying biological integrity by taxonomic completeness: Its utility in regional and
global assessments. Ecological Applications 16(4):1277-1294.
Hawkins, C.P., and M.R. Vinson. 2000. Weak correspondence between landscape classifications and
stream invertebrate assemblages: Implications for bioassessment. Journal of North American
Benthological Society 19:501-517.
198
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Hawkins, C, R. Morris, J. Gerritsen, R. Hughes, S.K. Jackson, R.K. Johnson, and R.J. Stevenson. 2000a.
Evaluation of the use of landscape classifications for the prediction of freshwater biota: Synthesis
and recommendation. Journal of the North American Benthological Society 19:541-556.
Hawkins, C, R. Norris, J.N. Hogue, and J.W. Feminella. 2000b. Development and valuation of predictive
models for measuring the biological integrity of streams. Ecological Applications 10(5):1456-1477.
http://www.auburn.edu/academic/cosam/facultv/biology/feminella/lab/documents/Hawkins eta
I 2000.pdf. Accessed February 2016.
Hawkins, C.P., J.R. Olson, and R.A. Hill. 2010. The reference condition: Predicting benchmarks for
ecological and water-quality assessments. Journal of the North American Benthological Society
29:312-343.
Hemsley-Flint, B. 2000. Classification of the Biological Quality of Rivers in England and Wales. In
Assessing the Biological Quality of Fresh Waters, ed. J.F. Wright, D.W. Sutcliffe, and M.T. Furse, pp.
55-70. Freshwater Biological Association, Ambleside, UK.
Herlihy, AT., S.G. Paulsen, J. Van Sickle, J.L Stoddard, C.P. Hawkins, and L.L Yuan. 2008. Striving for
consistency in a national assessment: The challenges of applying a reference-condition approach
at a continental scale. Journal of the North American Benthological Society 27(4):860-877.
Herricks, E.E., and D.J. Schaeffer. 1985. Can we optimize biomonitoring? Environmental Management
9:487-492.
Hill, R.A., M.H. Weber, S.G. Leibowitz, A.R. Olsen, and D.J. Thornbrugh. 2015. The Stream-Catchment
(StreamCat) Dataset: A database of watershed metrics for the conterminous USA. Journal of the
American Water Resources Association.
Hilsenhoff, W.L. 1977. Use of Arthropods to Evaluate Water Quality of Streams. Technical Bulletin
Number 100. Wisconsin Department of Natural Resources. 15 pp. Madison, Wisconsin.
http://dnr.wi.gov/files/PDF/pubs/ss/SS0100.pdf. Accessed February 2016.
Hilsenhoff, W.L. 1982. Aquatic Insects of Wisconsin: Keys to Wisconsin Genera and Notes on Biology,
Distribution, and Species. University of Wisconsin - Madison, Publication of the Natural History
Council, 60 pp. https://www.uwsp.edu/cnr-
ap/UWEXLakes/Documents/programs/LakeShoreTraining/5.2 challenges created from unsound
lakeshore/G3648 aquatic insects of wi.pdf. Accessed February 2016.
Hilsenhoff, W.L. 1987a. Rapid field assessment of organic pollution with a family level biotic index.
Journal of the North American Benthological Society 7(l):65-68.
Hilsenhoff, W.L. 1987b. An improved biotic index of organic stream pollution. The Great Lakes
Entomologist 20(l):31-39.
Hilsenhoff, W.L. 1988. Rapid field assessment of organic pollution with a family-level biotic index.
Journal of the North American Benthological Society 7(l):65-68.
Hodge, R.A. 1997. Toward a conceptual framework for assessing progress toward sustainability. Social
Indicators Research 40:5-98.
199
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Hoegh-Guldberg, O., P.J. Mumby, A.J. Hooten, R.S. Steneck, P. Greenfield, E. Gomez, C.D. Harvell, P.P.
Sale, A. Edwards, K. Caldeira, N. Knowlton, CM. Eakin, R. Iglesias-Prieto, N. Muthiga, R.H.
Bradbury, A. Dubi, and M.E. Hatziolos. 2007. Coral reefs under rapid climate change and ocean
acidification. Science 318:1737-1742.
Host, G.E., J.A. Schuldt, J.J.H. Ciborowski, L.B. Johnson, T.P. Hollenhorst, and C. Richards. 2005. Use of
GIS and remotely sensed data for a priori identification of reference areas for Great Lakes coastal
ecosystems. International Journal of Remote Sensing (Special Issue on Estuarine Ecosystem
Analysis) 26(23):5325-5342.
Host, G.E., T. Brown, T.P. Hollenhorst, L.B. Johnson, and J.J.H. Ciborowski. 2011. High-resolution
assessment and visualization of environmental stressors in the Lake Superior basin. Aquatic
Ecosystem Management and Health 14(4):376-385.
Huff, D.D., S. Hubler, Y. Pan, and D. Drake. 2006. Detecting Shifts in Macroinvertebrate Community
Requirements: Implicating Causes of Impairment in Streams. DEQ06-LAB-0068-TR. Oregon
Department of Environmental Quality, Hillsboro, OR.
http://www.deq.state.or.us/lab/techrpts/docs/10-LAB-005.pdf. Accessed February 2016.
Hughes, R.M. 1985. Use of watershed characteristics to select control streams for estimating effects of
metal mining wastes on extensively disturbed streams. Environmental Management 9:253-262.
Hughes, R.M. 1994. Defining Acceptable Biological Status by Comparing with Reference Conditions. In
Biological Assessment and Criteria: Tools for Water Resource Planning and Decision Making, ed.
W.S. Davis and T.P. Simon, pp. 31-47. CRC Press, Boca Raton, FL
Hughes, R.M., D.P. Larsen, and J.M. Omernik. 1986. Regional reference sites: A method for assessing
stream potential. Environmental Management 10:629-635.
Ibelings, B.W., M. Vonk, H.F.J. Los, D.T. Van Der Molen, and W.M. Mooij. 2003. Fuzzy modeling of
Cyanobacterial surface waterblooms: Validation with NOAA-AVHRR satellite images. Ecological
Applications 13:1456-1472.
IPCC. 2007. Climate Change 2007: Synthesis Report. Intergovernmental Panel on Climate Change.
Geneva, Switzerland.
https://www.ipcc.ch/publications and data/publications ipcc fourth assessment report synthe
sis report.htm. Accessed February 2016.
IPCC. 2014. Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the
5th Assessment Report of the Intergovernmental Panel on Climate Change (Core Writing Team,
ed. R.K. Pachauri and L.A. Meyer). Intergovernmental Panel on Climate Change, Geneva,
Switzerland. https://www.ipcc.ch/pdf/assessment-report/ar5/syr/SYR AR5 FINAL full.pdf.
Accessed February 2016.
200
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Jackson, J.B.C., M.X. Kirby, W.H. Berger, K.A. Bjorndahl, LW. Botsford, B.J. Bourque, R.H. Bradbury, R.
Cooke, J. Erlandson, J.A. Estes, T.P. Hughes, S. Kidwell, C.B. Lange, H.S. Lenihan, J.M. Pandolfi, C.H.
Peterson, R.S. Steneck, M.J. Tegner, and R.R. Werner. 2001. Historical overfishing and the recent
collapse of coastal ecosystems. Science 293:629-638.
http://faculty.washington.edu/stevehar/JacksonETAI2001-overfishing.pdf. Accessed February
2016.
Jessup, B.K. 2013. Multimetric Stream Macroinvertebrate Bioindicator Development in Alabama. DRAFT
report prepared by Tetra Tech, Inc. Prepared for Alabama Department of Environmental
Management.
Jessup, B.K., and J. Gerritsen. 2014. Calibration of the Biological Condition Gradient (BCG)for Fish and
Benthic Macroinvertebrate Assemblages in Northern Alabama. Prepared for Alabama Department
of Environmental Management, Montgomery, AL.
Johnson, L.B., and G.E. Host. 2010. Recent developments in landscape approaches for the study of
aquatic ecosystems. Journal of the North American Benthological Society 29(l):41-66.
Jones, E.B.D., G.S. Helfman, J.O. Harper, and P.V. Bolstad. 1999. Effects of riparian forest removal on fish
assemblages in southern Appalachian streams. Journal of Environmental Management 13:1454-
1465.
Jones, K.B., A.C. Neale, M.S. Nash, R.D. Van Remortel, J.D. Wickham, K.H. Riitters and R.V. O'Neill. 2001.
Predicting nutrient and sediment loadings to streams from landscape metrics: A multiple
watershed study from the United States Mid-Atlantic Region. Landscape Ecology 16:301-312.
Jongman, R.H.G., C.J.F. ter Braak, and O.F.R. van Tongeren, ed. 1987. Data Analysis in Community and
Landscape Ecology. Pudoc, Wageningen, Netherlands.
Karl, T.R., and K.E. Trenbreth. 2003. Modern climate change. Science 302:1719-1723.
Karr, J.R. 1981. Assessment of biotic integrity using fish communities. Fisheries 6(6):21-27.
Karr, J.R. 2000. Health, Integrity, and Biological Assessment: The Importance of Whole Things. In
Ecological Integrity: Integrating Environment, Conservation, and Health, ed. D. Pimentel, L.
Westra, and R.F. Noss, pp. 209-226. Island Press, Washington, DC.
Karr, J.R., and E.W. Chu. 2000. Sustaining living rivers. Hydrobiologia 422:1-14.
Karr, J.R., and D.R. Dudley. 1981. Ecological perspective on water quality goals. Environmental
Management 5:55-68.
https://www.researchgate.net/publication/227272834 Ecological Perspective on Water Qualit
y Goals. Accessed February 2016.
Karr, J.R., K.D. Fausch, P.L Angermeier, P.R. Yant, and I.J. Schlosser. 1986. Assessing Biological Integrity
in Running Waters: A Method and its Rationale. Illinois Natural History Survey, Special Publication
5, Champaign.
http://www.nrem.iastate.edu/class/assets/aecl518/Discussion%20Readings/Karr et al. 1986.pdf
Accessed February 2016.
201
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Kashuba, R., G. McMahon, T.F. Cuffney, S. Qian, K. Reckhow, J. Gerritsen, and S. Davies. 2012. Linking
Urbanization to the Biological Condition Gradient (BCG) for Stream Ecosystems in the Northeastern
United States Using a Bayesian Network Approach. U.S. Geological Survey Scientific Investigations
Report 2012-5030, 48 p. http://pubs.usgs.gov/sir/2012/5030/. Accessed February 2016.
Kennen, J.G., L.J. Kauffman, M.A. Ayers, D.M. Wolock, and S.J. Colarullo. 2008. Use of an integrated flow
model to estimate ecologically relevant hydrologic characteristics at stream biomonitoring sites.
Ecological Modeling 211:57-76.
Klemm, D.J, P.A. Lewis, F. Fulk, and J.M. Lazorchak. 1990. Macroinvertebrate Field and Laboratory
Methods for Evaluating the Biological Integrity of Surface Waters. EPA-600-4-90-030. U.S.
Environmental Protection Agency, Environmental Monitoring Systems Laboratory. Cincinnati, OH.
Klir, G.J. 2004. Fuzzy Logic: A Specialized Tutorial. In Fuzzy Logic in Geology, ed. R.V. Demicco and G.J.
Klir, pp. 11-61. Elsevier Academic Press, San Diego, CA.
Lammort, M., and J.D. Allan. 1999. Assessing Biotic Integrity of Streams: Effects of scale in measuring
the influence of land use/cover and habitat structure on fish and macroinvGrtobratoG.
Environmental Management 23(2):257-270.
Lane, C.R., 2003. Development of Biological Indicators of Freshwater Wetland Condition in Florida. Ph.D.
Dissertation. University of Florida, Gainesville, FL, USA.
Legendre, P., and L. Legendre. 1998. Numerical Ecology: Second English Edition. Elsevier, New York.
Lenat, D.R. and D.L. Penrose. 1996. History of the EPTtaxa richness metric. Bulletin of the North
American BenthologicalSociety 13(2):305-306.
Ludwig, J.A., and J.F. Reynolds. 1988. Statistical Ecology. John Wiley and Sons, New York, New York.
Mack, J.J. 2006. Landscape as a predictor of wetland condition: An evaluation of the landscape
development index (LDI) with a large reference wetland dataset from Ohio. Environmental
Monitoring and Assessment 120:221-241.
Mack, J.J. 2007. Developing a wetland IBI with statewide application after multiple testing iterations.
Ecological Indicators 7:864-881.
Maizel, M., R.D. White, R. Root, S. Gage, S. Stitt, L. Osborne, and G. Muehlbach. 1998. Historical
Interrelationships between Population Settlement and Farmland in the Conterminous United
States, 1790 to 1992. Chapter 2 In Perspectives on the Land Use History of North America: A
Context for Understanding Our Changing Environment, ed. T.D. Sisk, Biological Science Report
USGS/BRD/BSR 1998-0003. U.S. Geological Survey, Biological Resources Division, Reston, VA.
Accessed February 2016. http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA362077.
Manly, B.F.J. 1991. Multivariate Statistical Methods, A Primer. Chapman & Hall, London. 159 pp.
Margalef, R. 1963. On certain unifying principles in ecology. American Naturalist 97:357-374.
http://isites.harvard.edu/fs/docs/icb.topic281447.files/Unifying Principles.pdf. Accessed February
2016.
202
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Margalef, R. 1981. Stress in Ecosystems: A Future Approach. In Stress Effects on Natural Ecosystems, ed.
G.W. Barrett and R. Rosenberg, pp. 281-289. Wiley, London, UK.
Martinez, M.E. 1998. What is problem-solving? Phi Delta Kappan 79:605-610.
McClenachan, L, A.B. Cooper, M.G. McKenzie, and J.A. Drew. 2015. The importance of surprising results
and best practices in historical ecology. Bioscience 65(9):932-939.
McCormick, F.H., D.V. Peck, and D.P. Larsen. 2000. A comparison of ecological classification hierarchies
for Mid-Atlantic stream fish assemblages. Journal of North American Benthological Society
19:385-404.
MCDEP. 2009. Special Protection Area Program Annual Report 2007. Montgomery County Department
of Environmental Protection.
https://www. montgomervcountvmd.gov/DEP/Resources/Files/ReportsandPublications/Water/Sp
ecial%20Protection%20Areas/Special-protection-area-program-annual-report-07.pdf. Accessed
February 2016.
MCDEP. 2012. Special Protection Area Program Annual Report 2010. Montgomery County Department
of Environmental Protection.
http://www.montgomervcountvmd.gov/DEP/Resources/Files/downloads/water-
reports/spa/2010 spa report.pdf. Accessed February 2016.
Mebane, C.A., T.R. Maret, and R.M. Hughes. 2003. Development and testing of an index of biotic
integrity (IBI) for Columbia River basin and western Oregon. Transactions of the American
Fisheries Society 132:239-261.
MEDEP. 2006. Protocols for Sampling Aquatic Macroinvertebrates in Freshwater Wetlands. DEPLW0640.
Maine Department of Environmental Protection, Portland, ME.
http://www.dep.wv.gov/WWE/getinvolved/sos/Documents/SOPs/Maine.pdf. Accessed February
2016.
MEDEP. 2009. Quality Assurance Project Plan for Biological Monitoring of Maine's Rivers, Stream and
Freshwater Wetlands. DEP-LW-0638B-2009. Maine Department of Environmental Protection,
Augusta, ME.
http://www.maine.gov/dep/water/monitoring/biomonitoring/material.htmltfQAandSOPs.
Accessed February 2016.
MEDEP. 2012. Maine Impervious Cover Total Maximum Daily Load Assessment (TMDL)for Impaired
Streams. DEPLW-1239. Maine Department of Environmental Protection, Augusta, ME.
http://www.maine.gov/dep/water/monitoring/tmdl/2012/IC%20TMDL Sept 2012.pdf. Accessed
February 2016.
MEDEP. 2014. Methods for Biological Sampling and Analysis of Maine's Waters. DEP LW0387-C2014.
Maine Department of Environmental Protection, Augusta Maine.
http://www.maine.gov/dep/water/monitoring/biomonitoring/materials/sop stream macro met
hods manual.pdf. Accessed February 2016.
203
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Meidel, S., and MEDEP. 2006a. Barberry Creek Total Maximum Daily Load (TMDL). DEPLW0712. Maine
Department of Environmental Protection, Augusta, ME.
http://www.maine.gov/dep/water/monitoring/tmdl/tmdl2.html. Accessed February 2016.
Meidel, S., and MEDEP. 2006b. Trout Brook Total Maximum Daily Load (TMDL). DEPLW0714. Maine
Department of Environmental Protection, Augusta, ME.
http://ofmpub.epa.gov/waterslO/attains impaired waters.show tmdl document?p tmdl doc b
lobs id=72410. Accessed February 2016.
Melillo, J.M., T.C. Richmond, and G.W. Yohe, ed. 2014. Highlights of Climate Change Impacts in the
United States: The Third National Climate Assessment. U.S. Global Change Research Program,
Washington, DC. Accessed February 2016. http://nca2014.globalchange.gov/.
Merritt, R.W., and K.W. Cummins. 1996. An Introduction to the Aquatic Insects of North America, Third
Edition. Kendal/Hunt Publishing Company, Dubuque, IA.
Merritt, R.W., K.W. Cummins, and M.B. Berg, (editors) 2008. An Introduction to the Aquatic Insects of
North America. Fourth Edition. Kendall/Hunt Publishing Co., Dubuque, IA.
Miller, R.R., J.D. Williams, and J.E. Williams. 1989. Extinctions of North American fishes during the past
century. Fisheries 14:22-38.
http://www.ndwr.state.nv.us/hearings/past/spring/browseable/exhibits%5CUSFWS/FWS-
2068.pdf. Accessed February 2016.
Miller, M.P., J.G. Kennen, J.A. Mabe, and S.V. Mize. 2012. Temporal trends in algae, benthic
macroinvertebrates, and fish assemblages in streams and rivers draining basins of varying land use
from the south-central United States, 1993-2007. Hydrobiologia 684(l):15-33.
Miltner, R.A. 2015. Measuring the contribution of agricultural conservation practices to observed trends
and recent condition in water quality indicators in Ohio, USA. Journal of Environmental Quality
44:1821-1831. doi:10.2134/jeq2014.12.0550.
M-NCPPC. 1994. Clarksburg Master Plan & Hyattstown Special Study Area. Maryland-National Capitol
Park and Planning Commission, Silver Spring, MD.
http://www.montgomeryplanning.org/communitv/plan areas/rural area/master plans/clarksbur
g/toc clark.shtm. Accessed February 2016.
M-NCPPC. 2014a. Ten Mile Creek Area Limited Amendment to the Clarksburg Master Plan and
Hyattstown Special Study Area. Maryland-National Capitol Park and Planning Commission, Silver
Spring, MD.
http://www.montgomeryplanning.org/communitv/plan areas/1270 corridor/clarksburg/clarksbur
g lim amendment.shtm. Accessed February 2016.
M-NCPPC. 2014b. Ten Mile Creek Area Limited Amendment to the Clarksburg Master Plan and
Hyattstown Special Study Area: Appendix 9. Maryland-National Capitol Park and Planning
Commission, Silver Spring, MD.
http://www.montgomeryplanning.org/communitv/plan areas/1270 corridor/clarksburg/docume
nts/appendix 9 materials-for couonty council.pdf. Accessed February 2016.
204
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Moss, D., M.T. Furse, J.F. Wright, and P.O. Armitage. 1987. The prediction of the macroinvertebrate
fauna of unpolluted running-water sites in Great Britain using environmental data. Freshwater
Biology 17:41-52.
Moya, N., R.M. Hughes, E. Domfnguez, F-M. Gibon, E. Goitia, and T. Oberdorff. 2011. Macroinvertebrate-
based multimetric predictive models for evaluating the human impact on biotic condition of
Bolivian streams. Ecological Indicators 11:840-847.
Moyle, P.B. 1986. Fish Introductions into North America: Patterns and Ecological Impact. In Ecology of
Biological Invasions of North America and Hawaii, ed. H. A. Mooney and J. A. Drake, pp. 27-43.
Springer, New York.
MPCA. 2012. Framework and Implementation Recommendations for Tiered Aquatic Life Uses: Minnesota
Rivers and Streams. Document number wq-s6-24. Minnesota Pollution Control Agency,
Environmental Analysis and Outcomes Division, St. Paul, MN.
http://www.pca.state.mn.us/index.php/view-document.html?gid=18309. Accessed February
2016.
MPCA. 2014a. Development of a Human Disturbance Score (HDS)for Minnesota Streams [draft].
Minnesota Pollution Control Agency, St. Paul, MN.
MPCA. 2014b. Development of Biological Criteria for Tiered Aquatic Life Uses: Fish and
Macroinvertebrate Thresholds for Attainment of Aquatic Life Use Goals in Minnesota Streams and
Rivers. Document number wq-bsm4-02. Minnesota Pollution Control Agency, Environmental
Analysis and Outcomes Division, St. Paul, MN.
MPCA. 2014c. Development of a Macroinvertebrate-based Index of Biological Integrity for Assessment of
Minnesota's Rivers and Streams. Document number wq-bsm4-01. Minnesota Pollution Control
Agency, Environmental Analysis and Outcomes Division, St. Paul, MN.
https://www.pea.state.mn.us/sites/default/files/wq-bsm4-01.pdf. Accessed February 2016.
MPCA. 2014d. Development of a Fish-based Index of Biological Integrity for Assessment of Minnesota's
Rivers and Streams. Document number wq-bsm2-03. Minnesota Pollution Control Agency,
Environmental Analysis and Outcomes Division, St. Paul, MN.
https://www.pea.state.mn.us/sites/default/files/wq-bsm2-03.pdf. Accessed February 2016.
MPCA. 2014e. Tiered Aquatic Life Uses Overview. Document number wq-s6-33. Minnesota Pollution
Control Agency, St. Paul, MN.
MPCA. 2015. The Aquatic Biota Stressor and Best Management Practice Selection Guide. Technical
Report. Minnesota Pollution Control Agency, Watershed Division, Detroit Lakes, MN.
Nair, R., R. Aggarwal, and D. Khanna. 2011. Methods of formal consensus in classification/diagnostic
criteria and guideline development. Seminars in Arthritis and Rheumatism. 41(2):95-105.
doi:10.1016/j.semarthrit.2010.12.001.
205
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Nieber, J.L, C. Arika, B. Hansen, K. Evans, and G. Johnson. 2013. Lower Poplar River Watershed Sediment
Source Assessment. Report prepared for Minnesota Pollution Control Agency, February, 2013.
http://www.pca.state.mn.us/index.php/view-document.html?gid=19265. Accessed February
2016.
Niemi, G.J., and M. McDonald. 2004. Application of ecological indicators. Annual Review of Ecology,
Evolution, andSystematics 35:89-111.
Niemi, G.J., J.R. Kelly, and N.P. Danz. 2007. Environmental indicators for the coastal region of the North
American Great Lakes: Introduction and prospectus. Journal of Great Lakes Research 33(special
issue 3):1-12.
Niemeijer, D., and R. de Groot. 2008a. Framing environmental indicators: Moving from causal chains to
causal networks. Environmental Development and Sustainability 10:89-106.
Niemeijer D., and R.S. de Groot. 2008b. A conceptual framework for selecting environmental indicator
sets. Ecological Indicators 8:14-25.
Noss, R.G. 1990, Indicators for monitoring biodiversity: a hierarchical approach. Conservation Biology
4:355-364.
Norton, S.B., S.M. Cormier, and G.W. Suter II, ed. 2015. Ecological Causal Assessment. CRC Press, Boca
Raton, FL
Novak, M.A., and R.W. Bode. 1992. Percent model affinity: A new measure of macroinvertebrate
community composition. Journal of the North American Benthological Society 11:80-85.
NRC. 2001. Assessing the TMDL Approach to Water Quality Management. National Research Council,
Water Science and Technology Board, Division on Earth and Life Studies. National Academy Press,
Washington, DC.
Oberdorff, T., D. Pont, B. Hugueny, and J.P. Porcher. 2002. Development and validation of a fish-based
index (FBI) for the assessment of "river health" in France. Freshwater Biology 47:1720-1735.
Odum, E.P. 1985. Trends expected in stressed ecosystems. BioScience 35:419-422.
http://www.life.Illinois.edu/ib/451/Odum%20(1985).pdf. Accessed February 2016.
Odum, E.P., J.T. Finn, and E.H. Franz. 1979. Perturbation theory and the subsidy-stress gradient.
BioScience 29:349-352.
OECD. 1998. Towards Sustainable Development: Environmental Indicators. Organisation for Economic
Co-operation and Development. 132 pp.
Ohio EPA. 1981. 5-year Surface Water Monitoring Strategy, 1982-1986. Ohio Environmental Protection
Agency, Office of Wastewater Pollution Control, Division of Surveillance and Standards, Columbus,
Ohio. 52 pp. + appendices.
Ohio EPA. 1987. Biological Criteria for the Protection of Aquatic Life: Volumes /-///. Ohio Environmental
Protection Agency, Columbus, Ohio.
206
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Ohio EPA. 1999. Total Maximum Daily Load. TMDLteam. Division of Surface Water, Columbus, OH. 142
pp. http://www.epa.ohio.gov/portals/35/tmdl/FinalTMDLReport.pdf. Accessed February 2016.
Ohio EPA. 2006. Methods for Assessing Habitat in Flowing Waters: Using the Qualitative Habitat
Evaluation Index (QHEI). Ohio Environmental Protection Agency, Division of Surface Water,
Ecological Assessment Section, Columbus, OH. 23 pp.
http://www.epa.state.oh.us/portals/35/documents/QHEIManualJune2006.pdf. Accessed
February 2016.
Ohio EPA. 2014. Updates to Biological Criteria for the Protection of Aquatic Life: Volume II and Volume II
Addendum. User's Manual for Biological Field Assessment of Ohio Surface Waters. Ohio
Environmental Protection Agency, Division of Surface Water, Columbus, OH. 11 pp. + appendices.
http://www.epa.ohio.gov/dsw/bioassess/BioCriteriaProtAqLife.aspx. Accessed February 2016.
Olivero, A., and M. Anderson. 2008. Northeast Aquatic Habitat Classification. The Nature Conservancy,
Eastern Regional Office, Boston, MA, 88 pp.
http://www.conservationgateway.org/ConservationBvGeographv/NorthAmerica/UnitedStates/ed
c/reportsdata/freshwater/habitat/Pages/Northeast-Stream-Classification.aspx. Accessed February
2016.
Olivero Sheldon, A., A. Barnett and M.G. Anderson. 2015. A Stream Classification for the Appalachian
Region. The Nature Conservancy, Eastern Conservation Science, Eastern Regional Office. Boston,
MA, 86 pp.
http://www.conservationgatewav.org/ConservationBvGeography/NorthAmerica/UnitedStates/ed
c/reportsdata/freshwater/habitat/Pages/Appalachian-Stream-Classification.aspx. Accessed
February 2016.
Oliver, L.M., J.C. Lehrter, and W.S. Fisher. 2011. Relating landscape development intensity to coral reef
condition in the watersheds of St. Croix, US Virgin Islands. Marine Ecology Progress Series
427:293-302.
Omernik, K.M. 1987. Ecoregions of the conterminous United States. Annals of the Association of
American Geographers 77:118-125.
O'Neil, P.E., and T.E. Shepard. 2011. Calibration of the Index ofBiotic Integrity for the Plateau
Ichthyoregion In Alabama. Open-file Report 1111. Prepared by the Geological Survey of Alabama
in cooperation with the Alabama Department of Environmental Management and the Alabama
Department of Conservation and Natural Resources, Tuscaloosa, AL.
http://www.gsa.state.al.us/gsa/eco/pdf/OFR 1111.pdf. Accessed February 2016.
O'Neill, R.V., B.T. Milne, M.G. Turner, and R.H. Gardner. 1988. Resource utilization scales and landscape
pattern. Landscape Ecology 2(l):63-69.
O'Neill, R.V., C.T. Hunsaker, K.B. Jones, K.H. Riitters, J.D. Wickham, P.M. Schwartz, I.A. Goodman, B.L
Jackson, and W.S. Baillargeon. 1997. Monitoring environmental quality at the landscape scale.
BioScience 47(8):513-519.
207
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
PA DEP. 2012. A Benthic Macroinvertebrate Index of Biological Integrity for Wadeable Freestone Riffle-
Run Streams in Pennsylvania. Pennsylvania Department of Environmental Protection, Division of
Water Quality Standards.
PA DEP. 2013a. An Index ofBiotic Integrity for Benthic Macroinvertebrate Communities in Pennsylvania's
Wadeable, Freestone, Riffle-run Streams. Pennsylvania Department of Environmental Protection,
Bureau of Point and Non-Point Source Management. Harrisburg, PA.
PA DEP. 2013b. Instream Comprehensive Evaluation Surveys. Pennsylvania Department of Environmental
Protection, Bureau of Point and Non-Point Source Management. Harrisburg, PA.
http://files.dep.state.pa.us/Water/Drinking%20Water%20and%20Facility%20Regulation/WaterQu
alitvPortalFiles/Methodologv/2013%20Methodology/ICE.pdf. Accessed February 2016.
Palmer, M.A., D.P. Lettenmaier, N.L Poff, S.L Postel. B. Richter, and R. Warner. 2009. Climate Change
and River Ecosystems: Protection and Adaptation Options. Environmental Management 44:1053-
1068. doi:10.1007/s00267-009-9329-l.
Pantle, R. and H. Buck. 1955. Biological monitoring of water quality and the presentation of results. Gas
und Wasserfach 96:604.
Papworth. S.K., J. Rist, L. Coad, and E.J. Milner-Gulland. 2008. Evidence for shifting baseline syndrome in
conservation. Conservation Letters 2(2):93-100.
Pauly, D. 1995. Anecdotes and the shifting baseline syndrome of fisheries. Trends in Ecology and
Evolution 10(10):430.
Plafkin, J.L., M.T. Barbour, K.D. Porter, S.K. Gross, and R.M. Hughes. 1989. Rapid Bioassessment
Protocols for Use in Streams and Rivers: Benthic Macroinvertebrates and Fish. EPA/440/4-89/001.
U.S. Environmental Protection Agency, Office of Water, Washington, DC.
Poff, N.L., and J.D. Allan. 1995. Functional organization in stream fish assemblages in relation to
hydrologic variability. Ecology 76:606-627.
Poff, N.L., J.D. Allan, M.B. Bain, J.R. Karr, K.L Prestegaard, B.D. Richter, R.E. Sparks, and J.C. Stromberg.
1997. The natural flow regime: A paradigm for river conservation and restoration. BioScience
47:769-784.
Poff, N.L., M. Brinson, and J.B. Day. 2002. Freshwater and Coastal Ecosystems and Global Climate
Change: A Review of Projected Impacts for the United States. Pew Center on Global Climate
Change, Arlington, VA.
Poff, N.L., B.D. Richter, A.H. Arthington, S.E. Bunn, R.J. Naiman, E. Kendy, and A. Warner. 2010. The
ecological limits of hydrologic alteration (ELOHA): A new framework for developing regional
environmental flow standards. Freshwater Biology 55:147-170. doi:10.1111/j.l365-
2427.2009.02204.x.
Poikane, S., N. Zampoukas, S. Davies, W. Van de Bund, S. Birk, and A. Borja. 2014. Intercalibration of
aquatic ecological assessment methods in the European Union: Lessons learned and way forward
Environmental Science and Policy 44:237-246.
208
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Pont, D., B. Hugueny, B. Beier, D. Goffaux, A. Melcher, R. Noble, C. Rogers, N. Roset, and S. Schmutz.
2006. Assessing river biotic condition at the continental scale: A European approach using
functional metrics and fish assemblages. Journal of Applied Ecology 43:70-80.
Pont, D., R.M. Hughes, T.R. Whittier, and S. Schmutz. 2009. A predictive index of biotic integrity model
for aquatic vertebrate assemblages of Western U.S. streams. Transactions of the American
Fisheries Society 138:292-305.
Press, S.J. 1980. Multivariate Group Judgments. In Multivariate Analysis V, ed. P.R. Krishnaiah, pp. 581-
592. North-Holland Publishing Company, Amsterdam.
Pyne, M.I., R.B. Rader, and W.F. Christensen. 2007. Predicting local biological characteristics in streams:
A comparison of landscape classifications. Freshwater Biology 52:1302-1321.
Rankin, E.T. 1989. The Qualitative Habitat Evaluation Index (QHEI), Rationale, Methods, and Application.
Ohio Environmental Protection Agency, Division of Water Quality Planning and Assessment,
Ecological Assessment Section, Columbus, OH.
http://www.epa.ohio.gov/portals/35/documents/BioCrit88 QHEIIntro.pdf. Accessed February
2016.
Rankin, E.T. 1995. The Use of Habitat Indices in Water Resource Quality Assessments. In Biological
Assessment and Criteria: Tools for Water Resource Planning and Decision Making, ed. W.S. Davis
and T.P. Simon, pp. 181-208. Lewis Publishers, Boca Raton, FL
Rapport, D., and A. Friend. 1979. Towards a Comprehensive Framework for Environmental Statistics: A
Stress-response Approach. Statistics Canada Report. Catalogue 11-510 Occasional. Ottawa,
Ontario, Canada. 90 p.
Rapport, D. J., H. A. Regier, and T. C. Hutchinson. 1985. Ecosystem behavior under stress. American
Naturalist 125:617-640.
Reiss, K.C., and M.T. Brown. 2005. The Florida Wetland Condition Index (FWCI): Preliminary Development
of Biological Indicators for Forested Strand and Floodplain Wetlands. Report submitted to the
Florida Department of Environmental Protection under contract #WM-683. Howard T. Odum
Center for Wetlands, University of Florida, Gainesville, Florida, USA. 94 p.
http://ufdc.ufl.edu/AA00004283/00001. Accessed February 2016.
Reiss, K.C., and M.T. Brown. 2007. Evaluation of Florida palustrine wetlands: application of USEPA levels
1, 2, and 3 assessment methods. EcoHealth 4:206-218.
Reiss, K.C. 2004. Developing biological indicators for isolated forested wetlands in Florida. Ph.D.
Dissertation, University of Florida, Gainesville, Florida, USA.
http://etd.fcla.edu/UF/UFE0004385/reiss k.pdf. Accessed February 2016.
Reiss, K.C. 2006. Florida Wetland Condition Index for depressional forested wetlands. Ecological
Indicators 6:337-352.
Rencher, A.C. 2003. Methods of Multivariate Analysis, Second edition. John Wiley & Sons. 738 pages.
209
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Reynoldson, T.B., R.C. Bailey, K.E. Day, and R.H. Norris. 1995. Biological guidelines for freshwater
sediments based on BEnthic Assessment of SedimenT (the BEAST) using a multivariate approach
for predicting biological state. Australian Journal of Ecology 20:198-219.
Ricciardi, A., and H.J. Maclsaac. 2000. Recent mass invasion of the North American Great Lakes by
Ponto-Caspian species. Trends in Ecology and Evolution 15(2):62-65.
Richter, B.D., D.P. Braun, M.A. Mendelson, and L.L. Master. 1997. Threats to imperiled freshwater fauna.
Conservation Biology 11:1081-1093.
Richter, B.D., R. Matthews, D.L Harrison, and R. Wigington. 2003. Ecologically sustainable water
management: Managing river flows for ecological integrity. Ecological Applications 13:206-224.
Riseng, C.M., M.J. Wiley, P.W. Seelbck, and R.J. Stevenson. 2010. An ecological assessment of Great
Lakes tributaries in the Michigan Peninsulas. Journal of Great Lakes Research 36:505-519.
Riseng, C.M., M.J. Wiley, R.W. Black, and M.D. Munn. 2011. Impacts of agricultural land use on biological
integrity: a causal analysis. Ecological Applications 21:3128-3146.
Riitters, K.H., R.V. O'Neill, C.T. Hunsaker, J.D. Wickham, D.H. Yankee, S.P. Timmins, K.B. Jones, and B.L
Jackson. 1995. A factor analysis of landscape pattern and structure metrics. Landscape Ecology
10:23-39.
Riitters, K.H., R.V. O'Neill, J.D. Wickham, and K.B. Jones. 1996. A note on contagion indices for landscape
analysis. Landscape Ecology 11(4): 197-202.
Riitters, K.H., R.V. O'Neill, and K.B. Jones. 1997. Assessing habitat suitability at multiple scales: A
landscape-level approach. Biological Conservation 81(1):191-202.
Riva-Murray, K., R.W. Bode, and P.J. Phillips. 2002. Impact source determination with biomonitoring
data in New York State: Concordance with environmental data. Northeastern Naturalist 9(2):127-
162.
Robson, B.J., E.T. Chester, and C.M. Austin. 2011. Why life history information matters: Drought refuges
and macroinvertebrate persistence in son-perennial streams subject to a drier climate. Marine
and Freshwater Research 62:801-810.
Rosi-Marshall, E.J., and T.V. Royer. 2012. Pharmaceutical Compounds and Ecosystem Function: An
Emerging Research Challenge for Aquatic Ecologists. Ecosystems 15(6):867-880.
doi:10.1007/s!0021-012-9553-z.
Rosi-Marshall, E.J., D.W. Kincaid, H.A. Bechtold, T.V. Royer, M. Rojas, and J.J. Kelly. 2013.
Pharmaceuticals suppress algal growth and microbial respiration and alter bacterial communities
in stream biofilms. Ecological Applications 23(3):583-593.
Samhouri, J.F., P.S. Levin, and C.H. Ainsworth. 2010. Identifying thresholds for ecosystem-based
management. PLoS ONE 5(l):e8907. doi:10.1371/journal.pone.0008907.
210
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Sanders, R.S., R.J. Miltner, C.O. Yoder, and E.T. Rankin. 1999. The Use of External Deformities, Erosions,
Lesions, and Tumors (DELT anomalies) in Fish Assemblages for Characterizing Aquatic Resources: A
Case Study of Seven Ohio Streams. In Assessing the Sustainability and Biological Integrity of Water
Resources Using Fish Communities, ed. T.P. Simon, pages 225-248. CRC Press, Boca Raton, FL.
Scardi, M., S. Cataudella, P. Di Dato, E. Fresi, and L Tancioni. 2008. An expert system based on fish
assemblages for evaluating the ecological quality of streams and rivers. Ecological Informatics
3:55-63.
Schindler, D.W. 1987. Detecting ecosystem responses to anthropogenic stress. Canadian Journal of
Fisheries and Aquatic Sciences 44:6-25.
Shannon, C.E. 1948. A mathematical theory of communication. Bell System Technical Journal 27:379-
423, 623-656.
Shannon, C.E., and W. Weaver. 1963. The Mathematical Theory of Communication. University of Illinois
Press, Urbana.
http://monoskop.Org/images/b/be/Shannon Claude E Weaver Warren The Mathematical The
ory of Communication 1963.pdf. Accessed February 2016.
Shelton, A.D., and K.A. Blocksom. 2004. A Review of Biological Assessment Tools and Biocriteria for
Streams and Rivers in New England States. EPA/600/R-04/168. U.S. Environmental Protection
Agency, National Exposure Research Laboratory.
Shumchenia, E.J., M.C. Pelletier, G. Cicchetti, S. Davies, C.E. Pesch, C.F. Deacutis, and M. Pryor. 2015. A
biological condition gradient model for historical assessment of estuarine habitat structure.
Environmental Management 55:143-158.
Shumchenia, E.J., M.L. Guarinello, and J.W. King. In review. A re-assessment of Narragansett Bay benthic
habitat quality between 1988 and 2008. Estuaries and Coasts.
Simon, T.P., ed. 2003. Biological Response Signatures: Indicator Patterns Using Aquatic Communities.
CRC Press, Boca Raton, FL.
Simpson, J.C., and R.H. Norris. 2000. Biological Assessment of River Quality: Development of AusRivAS
Models and Outputs. In Assessing the Biological Quality of Fresh Waters: RIVPACS and Other
Techniques, ed. J.F. Wright, D.W. Sutcliffe, and M.T. Furse, pp. 125-142. Freshwater Biological
Association, Ambleside, UK.
Slivitsky, M. 2001. A Literature Review on Cumulative Ecological Impacts of Water Use and Changes in
Levels and Flows. The Great Lakes Commission, October 15, 2001. 63 pp.
Smith, R.W., M. Bergen, S.B. Weisberg, D. Cadien, A. Dalkey, D. Montagne, J.K. Stull, and R.G. Velarde.
2001. Benthic response index for assessing infaunal communities on the Southern California
mainland shelf. Ecological Applications 11(4):1073-1087.
Snelder, T.H., F. Cattaneo, A.M. Suren, and B.J. Biggs. 2004. Is the river environment classification an
improved landscape-scale classification of rivers? Journal of the North American Benthological
Soc/ety23(3):580-598.
211
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Snelder, T.H., H. Pella, J.G. Wasson, and N. Lamouroux. 2008. Definition procedures have little effect on
performance of environmental classifications of streams and rivers. Environmental Management
42(5):771-788.
Snook, H., S.P. Davies, J. Gerritsen, B.K. Jessup, R. Langdon, D. Neils, and E. Pizutto. 2007. The New
England Wadeable Stream Survey (NEWS): Development of Common Assessments in the
Framework of the Biological Condition Gradient. U.S. Environmental Protection Agency and New
England Interstate Water Pollution Control Commission.
Stamp, J., and J. Gerritsen. 2011. A Biological Condition Gradient (BCG) Assessment Model for Stream
Fish Communities of Connecticut. Prepared for USEPA Office of Science and Technology and
Connecticut DEEP.
Stamp, J., J. Gerritsen, G. Pond, S.K. Jackson, and K. Van Ness. 2014. Calibration of the Biological
Condition Gradient (BCG) for Fish and Benthic Macroinvertebrate Assemblages in the Northern
Piedmont region of Maryland. Prepared for US EPA Office of Water, Office of Science and
Technology.
Stanfield, L.W., and B.W. Kilgour. 2012. How proximity of land-use affects stream fish and habitat. River
Research and Applications 29:891-905.
State of Maine. 2003. Code of Maine Rules 06-096. Chapter 579: Classification and Attainment
Evaluation Using Biological Criteria for Rivers and Streams. Office of the Secretary of State of
Maine, Augusta, ME. http://www.maine.gov/sos/cec/rules/06/chaps06.htm. Accessed February
2016.
State of Maine. 2004. Maine Revised Statutes Annotated, Title 38, Section 464-470. Protection and
Improvement of Waters, Maine State Legislature, Office of the Revisor of Statutes, State House,
Augusta, Maine 04333-0007.
http://www.mainelegislature.org/legis/statutes/38/title38sec464.html. Accessed February 2016.
Steedman, R.J. 1994. Ecosystem health as a management goal. Journal of the North American
Benthological Society 13(4):605-610.
Stoddard, J.L., AT. Herlihy, D.V. Peck, R.M. Hughes, T.R. Whittier, and E. Tarquinio. 2008. A process for
creating multimetric indices for large-scale aquatic surveys. Journal of the North American
Benthological Society 27:878-891.
Stoddard, J.L., D.P. Larsen, C.P. Hawkins, R.K. Johnson, and R.H. Norris. 2006. Setting expectations for
the ecological condition streams: The concept of reference condition. Ecological Applications
16:1267-1276.
Stranko, S.A., R.H. Hildebrand, R.P. Morgan, E.S. Paerry, and P.T. Jacobson.2008. Brook trout declines
with land cover and temperature changes in Maryland. North American Journal of Fisheries
Management 28:1223-1232.
212
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Surdick, A.J. 2005. Amphibian and avion species composition of forested depressional wetlands and
circumjacent habitat: The influence of land use type and intensity. Ph.D. Dissertation, University of
Florida, Gainesville, Florida, USA.
http://ufdcimages.uflib.ufl.edu/UF/EO/01/07/45/00001/surdick j.pdf. Accessed February 2016.
Suter, G. II, S.B. Norton, and S.M. Cormier. 2002. A methodology for inferring the causes of observed
impairments in aquatic ecosystems. Environmental Toxicology and Chemistry 21:1101-1111.
Teixeira, H., and 19 others. 2010. Assessing coastal benthic macrofauna community condition using best
professional judgement—Developing consensus across North America and Europe. Marine
Pollution Bulletin 60:589-600.
Trautman, M.B. 1957. The Fishes of Ohio. The Ohio State Univ. Press, Columbus, OH. 683 pp.
Trautman, M. 1981. The Fishes of Ohio. Revised edition. The Ohio State University Press, Columbus.
USEPA. 1990. Biological Criteria: National Program Guidance for Surface Waters. EPA-440-5-90-004. U.S.
Environmental Protection Agency, Washington, D.C.
USEPA. 2000a. Mid-Atlantic Highlands Streams Assessment. EPA/903/R-00/015. U.S. Environmental
Protection Agency, Philadelphia, PA.
USEPA. 2000b. Stressor Identification Guidance Document. EPA/822/B-00/025. U.S. Environmental
Protection Agency, Office of Water, Washington, DC.
USEPA. 2002. Summary of Biological Assessment Programs and Biocriteria Development for States,
Tribes, Territories, and Interstate Commissions: Streams and Wadeable Rivers. EPA-822-R-02-048.
U.S. Environmental Protection Agency, Office of Environmental Information and Office of Water,
Washington, D.C.
USEPA. 2011a. A Primer on Using Biological Assessments to Support Water Quality Management. EPA
810-R-11-01. U.S. Environmental Protection Agency, Washington, DC.
USEPA. 2011b. A Field-based Aquatic Life Benchmark for Conductivity in Central Appalachian Streams.
EPA/600/R-10/023F. U.S. Environmental Protection Agency, Office of Research and Development,
National Center for Environmental Assessment, Washington, DC.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=233809. Accessed February 2016.
USEPA. 2013a. Biological Assessment Program Review: Assessing Level of Technical Rigor to Support
Water Quality Management. EPA820-R-13-001. U.S. Environmental Protection Agency,
Washington, DC.
USEPA. 2013b. Biological Condition Gradient: A Headwater Steam Catchment in the Northern Piedmont
Region, Montgomery County, Maryland. Technical Expert Workshop Report. U.S. Environmental
Protection Agency, Washington, DC.
http://www.montgomervplanningboard.org/agenda/2013/documents/20130411 Clarksburg Att
achments for Staff Report OOO.pdf. Accessed February 2016.
213
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
USEPA. 2014a. Water Quality Standards Handbook. U.S. Environmental Protection Agency, Washington,
DC. http://water.epa.gov/scitech/swguidance/standards/handbook/index.cfm. Accessed February
2016.
USEPA. 2014b. Alabama and Mobile Bay Basin Integrated Assessment of Watershed Health: A Report on
the Status and Vulnerability of Watershed Health in Alabama and the Mobile Bay Basin. EPA 841-
R-14-002. U.S. Environmental Protection Agency, Healthy Watersheds Program.
http://www.mobilebavnep.com/images/uploads/librarv/ALMB HW Report Final Assessment (1
).pdf. Accessed February 2016.
USGS. 2014. National Hydrography Dataset Watershed Boundary Dataset. U.S. Geological Survey.
http://nhd.usgs.gov/. Accessed February 2016.
Uzarski, D.G., T.M. Burton, M.J. Cooper, J.W. Ingram, and S.T.A. Timmermans. 2005. Fish habitat use
within and across wetland classes in coastal wetlands of the five Great Lakes: Development of a
fish-based index of biotic integrity. Journal of Great Lakes Research 31(Supplement 1):171-187.
van Dam, H., A. Mertenes, and J. Sinkeldam. 1994. A coded checklist and ecological indicator values of
freshwater diatoms from the Netherlands. Netherlands Journal of Aquatic Ecology 28:117-33.
Vander Laan, J.J., C.P. Hawkins, J.R. Olson, and R.A. Hill. 2013. Linking land use, in-stream stressors, and
biological condition to infer causes of regional ecological impairment in streams. Freshwater
Science 32:801-820.
Van Sickle, J. 2008. An index of compositional dissimilarity between observed and expected
assemblages. Journal of the North American Benthological Society 27(2):227-235.
Van Sickle, J., and R.M. Hughes. 2000. Classification strengths of ecoregion, catchments and geographic
clusters for aquatic vertebrates in Oregon. Journal of the North American Benthological Society
19:370-384.
Van Sickle, J., D.D. Huff, and C.P. Hawkins. 2006. Selecting discriminant function models for predicting
the expected richness of aquatic macroinvertebrates. Freshwater Biology 51(2):359-372.
Villamagna, A.M., P.L. Angermeier, E.M. Bennett. 2013. Capacity, pressure, demand, and flow: A
conceptual framework for analyzing ecosystem service provision and delivery. Ecological
Complexity 15:114-121.
Vivas, M.B. 2007. Development of an Index of Landscape Development Intensity for Predicting the
Ecological Condition of Aquatic and Small Isolated Palustrine Wetland Systems in Florida. Doctoral
Thesis, University of Florida. http://ufdc.ufl.edu/AA00003991/00001. Accessed February 2016.
Vivas, M.B., and M.T. Brown. 2006. Areal Empower Density and Landscape Development Intensity (LDI)
Indices for Wetlands of the Bayou Meto Watershed, Arkansas. Report Submitted to the Arkansas
Soil and Water Conservation Commission under the Sub-grant Agreement SGA 104.
214
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Vollenweider, R.A. 1968. Water Management Research. Scientific Fundamentals of the Eutrophication of
Lakes and Flowing Waters with Particular Reference to Nitrogen and Phosphorus as Factors in
Eutrophication. Organisation for Economic Co-operation and Development. Directorate for
Scientific Affairs. Paris. Mimeographed. 159 p. + 34 Figs. + 2 separately paged annexes:
Bibliography, 61 p; Current status of research on eutrophication in Europe, the United States and
Canada, 20 p.
VT DEC. 2004. Biocriteria for Fish and Macroinvertebrate Assemblages in Vermont Wadeable Streams
and Rivers—Development Phase. Vermont Department of Environmental Conservation, Water
Quality Division, Biomonitoring and Aquatic Studies Section, Waterbury VT.
http://www.watershedmanagement.vt.gov/bass/docs/bs wadeablestreamlb.pdf. Accessed
February 2016.
Waite, I.R., AT. Herlihy, D.P. Larsen, and D.J. Klemm. 2000. Comparing strengths of geographic and
nongeographic classifications of stream benthic macroinvertebrates in the Mid- Atlantic
Highlands, USA. Journal of North American Benthological Society 19(3):429-441.
Waite, I.R., L.R. Brown, J.G. Kennen, J.T. May, T.F. Cuffney, J.L Orlando, and K.A. Jones. 2010.
Comparison of watershed disturbance predictive models for stream benthic macroinvertebrates
for three distinct ecoregions in western US. Ecological Indicators 10:1125-1136.
Walter, R.C., and D.J. Merritts. 2008. Natural streams and the legacy of water-powered mills. Science
319:299-304.
Walters, A.W., and D.M. Post 2011. How low can you go? Impacts of a low-flow disturbance on aquatic
insect communities. Ecological Applications 21:163-174.
Wang, L.T. Brenden, P. Seelbach, A. Cooper, D. Allan, R. Clark Jr. M. Wiley. 2008. Landscape based
identification of human disturbance gradients and reference conditions for Michigan Streams.
Environmental Monitoring and Assessment 141(1):1-17.
Weisberg, S.B., B. Thompson, J.A. Ranasinghe, D.E. Montagne, D.B. Cadien, D.M. Dauer, D. Diener, J.
Oliver, D.J. Reish, R.G. Velarde, and J.Q. Word. 2008. The level of agreement among experts
applying best professional judgment to assess the condition of benthic infaunal communities.
Ecological Indicators 8:389-394.
White, P.S., and S.T.A. Pickett. 1985. Natural Disturbance and Patch Dynamics: An Introduction. In The
Ecology of Natural Disturbance and Patch Dynamics, ed. S.T.A. Pickett and P.S. White, pp. 3-13.
Academic Press, Orlando, FL.
Whittier, T.R., and J. Van Sickle. 2010. Macroinvertebrate tolerance values and an assemblage tolerance
index (ATI) for western USA streams and rivers. Journal of the North American Benthological
Society 29(3):852-866.
Wilhm, J.L, and T.C. Dorris. 1966. Species diversity of benthic macroinvertebrates in a stream receiving
domestic and oil refinery effluents. American Midland Naturalist 76:427-449.
Willby, N.J. 2011. From metrics to Monet: The need for an ecologically meaningful guiding 543 image.
Aquatic Conservation: Marine and Freshwater Ecosystems 21:601-603.
215
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Wilkinson, L. 1989. SYSTAT: The System for Statistics. Systat, Inc., Evanston, II. 638 p.
Woods, A.J., J.M. Omernik, and B.C. Moran. 2007. Level III and IV Ecoregions of New Jersey. Map and
text description. U.S. Environmental Protection Agency, Western Ecology Division, Corvallis, OR.
http://archive.epa.gov/wed/ecoregions/web/html/ni eco.html . Accessed February 2016.
Wright, J.F. 2000. An Introduction to RIVPACS. In Assessing the Biological Quality of Fresh Waters:
RIVPACS and Other Techniques, ed. J.F. Wright, D.W. Sutcliffe, and M.T. Furse, pp. 1-24.
Freshwater Biological Association, Ambleside, UK.
Yoder, C.O. 1995. Policy Issues and Management Applications for Biological Criteria. In Biological
Assessment and Criteria: Tools for Water Resource Planning and Decision Making, ed. W.S. Davis
and T.P. Simon, pp. 327-343. Lewis Publishers, Boca Raton, FL
http://www.epa.state.oh.us/portals/35/volunteermonitoring/references/Yoderl995.pdf.
Accessed February 2016.
Yoder, C.O., and M.T. Barbour. 2009. Critical elements of state bioassessment programs: A process to
evaluate program rigor and comparability. Environmental Monitoring and Assessment 150(1):31-
42.
Yoder, C.O., and J.E. DeShon. 2003. Using Biological Response Signatures within a Framework of Multiple
Indicators to Assess and Diagnose Causes and Sources of Impairments to Aquatic Assemblages in
Selected Ohio Rivers and Streams. In Biological Response Signatures: Indicator Patterns Using
Aquatic Communities, ed. T.P. Simon, pp. 23-81. CRC Press, Boca Raton, FL.
Yoder, C.O., and E.T. Rankin. 1995a. Biological Response Signatures and the Area of Degradation Value:
New Tools for Interpreting Multimetric Data. In Biological Assessment and Criteria: Tools for
Water Resource Planning and Decision Making, ed. W.S. Davis and T.P. Simon, pp. 263-286. Lewis
Publishers, Boca Raton, FL.
Yoder, C.O., and E.T. Rankin. 1995b. Biological Criteria Program Development and Implementation in
Ohio. In Biological Assessment and Criteria: Tools for Water Resource Planning and Decision
Making, ed. W.S. Davis and T.P. Simon, pp. 109-144. Lewis Publishers, Boca Raton, FL.
Yoder, C.O., and E.T. Rankin. 1998. The role of biological indicators in a state water quality management
process. Environmental Monitoring and Assessment 51:61-68.
Yoder, C.O., E.T. Rankin, M.A. Smith, B.C. Alsdorf, D.J. Altfater, C.E. Boucher, R.J. Miltner, D.E. Mishne,
R.E. Sanders, and R.F. Thoma. 2005. Changes in Fish Assemblage Status in Ohio's Non-Wadeable
Rivers and Streams over Two Decades. In Historical Changes in Fish Assemblages of Large Rivers of
the Americas, American Fisheries Society Symposium 45, ed. J.N. Rinne, R.M. Hughes, and B.
Calamusso, pp. 399-429. American Fisheries Society, Bethesda, MD.
Yuan, L.L. 2004. Assigning macroinvertebrate tolerance classifications using generalised additive models.
Freshwater Biology 49:662-677.
Yuan, L.L., 2006. Estimation and Application of Macroinvertebrate Tolerance Values.
EPA/600/P-04/116F. US Environmental Protection Agency, Office of Research and Development,
Washington, DC.
216
-------
A Practitioner's Guide to the Biological Condition Gradient February 2016
Yuan, L.L. 2010. Estimating the effects of excess nutrients on stream invertebrates from observational
data. Ecological Applications 20(1):110-125.
Yuan, LL, and S.B. Norton. 2003. Comparing responses of macroinvertebrate metrics to increasing
stress. Journal of the North American Benthological Society 22(2):308-322.
Zadeh, LA., 1965. Fuzzy sets. Inform. Control 8:338-353.
Zadeh, LA., 2008. Is there a need for fuzzy logic? Information Sciences 178:2751-2779.
217
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Glossary
aquatic assemblage
aquatic community
aquatic life use
attribute
benthic macroinvertebrates or
benthos
best management practice
biological assessment or
bioassessment
biological criteria or biocriteria
biological indicator or bioindicator
biological integrity
biological monitoring or
biomonitoring
biological survey or biosurvey
An association of interacting populations of organisms in a
given water body; for example, fish assemblage or a benthic
macroinvertebrate assemblage.
An association of interacting assemblages in a water body, the
biotic component of an ecosystem.
A beneficial use designation in which the water body provides,
for example, suitable habitat for survival and reproduction of
desirable fish, shellfish, and other aquatic organisms.
The measurable part or process of a biological system.
Animals without backbones, living in or on the sediments, of a
size large enough to be seen by the unaided eye and which can
be retained by a U.S. Standard no. 30 sieve (28 meshes per
inch, 0.595-mm openings); also referred to as benthos, infauna,
or macrobenthos.
An engineered structure or management activity, or
combination of those, that eliminates or reduces an adverse
environmental effect of a pollutant.
An evaluation of the biological condition of a water body using
surveys of the structure and function of a community of
resident biota.
Narrative expressions or numeric values of the biological
characteristics of aquatic communities based on appropriate
reference conditions; as such, biological criteria serve as an
index of aquatic community health.
An organism, species, assemblage, or community characteristic
of a particular habitat, or indicative of a particular set of
environmental conditions.
The ability of an aquatic ecosystem to support and maintain a
balanced, adaptive community of organisms having a species
composition, diversity, and functional organization comparable
to that of natural habitats in a region.
Use of a biological entity as a detector and its response as a
measure to determine environmental conditions; ambient
biological surveys and toxicity tests are common biological
monitoring methods.
Collecting, processing, and analyzing a representative portion
of the resident aquatic community to determine its structural
and/or functional characteristics.
218
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
biotope
catchment
Clean Water Act
Clean Water Act section 303(d)
Clean Water Act section 305(b)
Clean Water Act section 304(a)
criteria
criteria
designated uses
disturbance
ecological integrity
An area that is relatively uniform in physical structure and that
is identified by a dominant biota.
An incremental watershed that drains directly into a stream
reach and excludes upstream areas.
The act passed by the U.S. Congress to control water pollution
(formally referred to as the Federal Water Pollution Control Act
of 1972). Public Law 92-500, as amended. 33 U.S.C. 1251 etseq.
This section of the act requires states, territories, and
authorized tribes to develop lists of impaired waters for which
applicable WQS are not being met, even after point sources of
pollution have installed the minimum required levels of
pollution control technology. The law requires that the
jurisdictions establish priority rankings for waters on the lists
and develop TMDLs for the waters. States, territories, and
authorized tribes are to submit their lists of waters on April 1 in
every even-numbered year.
Biennial reporting requires description of the quality of the
nation's surface waters, evaluation of progress made in
maintaining and restoring water quality, and description of the
extent of remaining problems.
EPA-published, recommended water quality criteria that
consist of scientific information regarding concentrations of
specific chemicals or levels of parameters in water that protect
aquatic life and human health. The States may use these
contents as the basis for developing enforceable water quality
standards.
Elements of state water quality standards, expressed as
constituent concentrations, levels, or narrative statements,
representing a quality of water that supports a particular use.
When criteria are met, water quality will generally protect the
designated use.
Those uses specified in WQS for each water body or segment
whether or not they are being attained.
Human activity that alters the natural state and can occur at or
across many spatial and temporal scales.
The condition of an unimpaired ecosystem as measured by
combined chemical, physical (including physical habitat), and
biological attributes. Ecosystems have integrity when they have
their native components (plants, animals and other organisms)
and processes (such as growth and reproduction) intact.
219
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
ecoregion
function
guild
historical data
index of biological/biotic integrity
invasive species
least disturbed condition
maintenance of populations
metric
minimally disturbed condition
multimetric index
A relatively homogeneous ecological area defined by similarity
of climate, landform, soil, potential natural vegetation,
hydrology, or other ecologically relevant variables.
Processes required for normal performance of a biological
system (may be applied to any level of biological organization).
A group of organisms that exhibit similar habitat requirements
and that respond in a similar way to changes in their
environment.
Data sets from previous studies, which can range from
handwritten field notes to published journal articles.
An integrative expression of site condition across multiple
metrics; an IBI is often composed of at least seven metrics.
A species whose presence in the environment causes economic
or environmental harm or harm to human health. Native
species or nonnative species can show invasive traits, although
that is rare for native species and relatively common for
nonnative species. (Note that this term is not included in the
biological condition gradient [BCG].)
The best available existing conditions with regard to physical,
chemical, and biological characteristics or attributes of a water
body within a class or region. Such waters have the least
amount of human disturbance in comparison to others in the
water body class, region, or basin. Least disturbed conditions
can be readily found but can depart significantly from natural,
undisturbed conditions or minimally disturbed conditions.
Least disturbed condition can change significantly over time as
human disturbances change.
Sustained population persistence; associated with locally
successful reproduction and growth.
A calculated term or enumeration that represents some aspect
of biological assemblage, function, or other measurable aspect
and is a characteristic of the biota that changes in some
predictable way with increased human influence.
The physical, chemical, and biological conditions of a water
body with very limited, or minimal, human disturbance.
An index that combines indicators, or metrics, into a single
index value. Each metric is tested and calibrated to a scale and
transformed into a unitless score before being aggregated into
a multimetric index. Both the index and metrics are useful in
assessing and diagnosing ecological condition. See index of
biological/biotic integrity (IBI).
220
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
narrative biological criteria
native
nonnative or intentionally
introduced species
numeric biological criteria
periphyton
rapid bioassessment protocols
rebuttable presumption
recovery potential
Written statements describing the structure and function of
aquatic communities in a water body that support a designated
aquatic life use.
An original or indigenous inhabitant of a region; naturally
present.
With respect to an ecosystem, any species that is not found in
that ecosystem; species introduced or spread from one region
of the United States to another outside their normal range are
nonnative or non-indigenous, as are species introduced from
other continents.
Specific quantitative measures of the structure and function of
aquatic communities in a water body necessary to protect a
designated aquatic life use.
A broad organismal assemblage composed of attached algae,
bacteria, their secretions, associated detritus, and various
species of microinvertebrates.
Cost-effective techniques used to survey and evaluate the
aquatic community to detect aquatic life impairments and their
relative severity.
In the context of water quality standards, the concept that the
CWA 101(a)(2) uses are attainable and therefore must be
assigned to a water body, unless a State or Tribe affirmatively
demonstrates, with appropriate documentation, that such uses
are not attainable.
In the context of water quality management, the likelihood that
an impaired water body can be restored so that it ultimately
meets water quality standards. Consideration of ecological,
stressor, and social factors are involved in the consideration of
recovery potential.
221
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
reference condition (biological
integrity)
reference site
refugia
sensitive taxa
sensitive or regionally endemic
taxa
The condition that approximates natural, unaffected conditions
(biological, chemical, physical, and such) for a water body.
Reference condition (biological integrity) is best determined by
collecting measurements at a number of sites in a similar water
body class or region undisturbed by human activity, if they
exist. Because undisturbed conditions can be difficult or
impossible to find, minimally or least disturbed conditions,
combined with historical information, models, or other
methods can be used to approximate reference condition as
long as the departure from natural or ideal is understood.
Reference condition is used as a benchmark to determine how
much other water bodies depart from this condition because of
human disturbance.
See definitions for minimally and least disturbed condition
A site selected for comparison with sites being assessed. The
type of site selected and the types of comparative measures
used will vary with the purpose of the comparisons. For the
purposes of assessing the ecological condition of sites, a
reference site is a specific locality on a water body that is
undisturbed or minimally disturbed and is representative of the
expected ecological integrity of other localities on the same
water body or nearby water bodies.
Accessible microhabitats or regions in a stream reach or
watershed where adequate conditions for organism survival
are maintained during circumstances that threaten survival; for
example, drought, flood, temperature extremes, increased
chemical stressors, habitat disturbance.
Taxa intolerant to a given anthropogenic stress; first species
affected by the specific stressor to which they are sensitive and
the last to recover following restoration.
Taxa with restricted, geographically isolated distribution
patterns (occurring only in a locale as opposed to a region),
often because of unique life history requirements. Can be long-
lived, late-maturing, low-fecundity, limited-mobility, or require
mutualist relation with other species. Can be among listed
endangered/threatened or special concern species.
Predictability of occurrence often low; therefore, requires
documented observation. Recorded occurrence can be highly
dependent on sample methods, site selection, and level of
effort.
222
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
sensitive-rare taxa
sensitive-ubiquitous taxa
stressors
structure
taxa
taxa of intermediate tolerance
Taxa that naturally occur in low numbers relative to total
population density but can make up large relative proportion of
richness. Can be ubiquitous in occurrence or can be restricted
to certain micro-habitats, but because of low density, recorded
occurrence is dependent on sample effort. Often stenothermic
(having a narrow range of thermal tolerance) or coldwater
obligates; commonly K-strategists (populations maintained at a
fairly constant level; slower development; longer life span). Can
have specialized food resource needs or feeding strategies.
Generally intolerant to significant alteration of the physical or
chemical environment; are often the first taxa observed to be
lost from a community.
Taxa ordinarily common and abundant in natural communities
when conventional sample methods are used. Often having a
broader range of thermal tolerance than sensitive or rare taxa.
These are taxa that constitute a substantial portion of natural
communities and that often exhibit negative response (loss of
population, richness) at mild pollution loads or habitat
alteration.
Physical, chemical, and biological factors that adversely affect
aquatic organisms.
Taxonomic and quantitative attributes of an assemblage or
community, including species richness and relative abundance
structurally and functionally redundant attributes of the system
and characteristics, qualities, or processes that are represented
or performed by more than one entity in a biological system.
A grouping of organisms given a formal taxonomic name such
as species, genus, family, and the like.
Taxa that compose a substantial portion of natural
communities; can be r-strategists (early colonizers with rapid
turnover times; boom/bust population characteristics). Can be
eurythermal (having a broad thermal tolerance range). Can
have generalist or facultative feeding strategies enabling
utilization of relatively more diversified food types. Readily
collected with conventional sample methods. Can increase in
number in waters with moderately increased organic resources
and reduced competition but are intolerant of excessive
pollution loads or habitat alteration.
223
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
tolerant taxa
total maximum daily load
water quality management
(nonregulatory)
water quality standard
whole effluent toxicity
Taxa that compose a small proportion of natural communities.
They are often tolerant of a broader range of environmental
conditions and are thus resistant to a variety of pollution- or
habitat-induced stresses. They can increase in number
(sometimes greatly) in the absence of competition. Commonly
r-strategists (early colonizers with rapid turnover times;
boom/bust population characteristics), able to capitalize when
stress conditions occur; last survivors.
The sum of the allowable loads of a single pollutant from all
contributing point and nonpoint sources; the calculated
maximum amount of a pollutant a water body can receive and
still meet WQS and an allocation of that amount to the
pollutant's source.
Decisions on management activities relevant to a water
resource, such as problem identification, need for and
placement of best management practices, pollution abatement
actions, and effectiveness of program activity.
A law or regulation that consists of the designated use or uses
of a water body, the narrative or numerical water quality
criteria (including biological criteria) that are necessary to
protect the use or uses of that water body, and
antidegradation requirements.
The aggregate toxic effect of an aqueous sample (e.g., whole
effluent wastewater discharge) as measured by an organism's
response after exposure to the sample (e.g., lethality, impaired
growth or reproduction); WET tests replicate the total effect
and actual environmental exposure of aquatic life to toxic
pollutants in an effluent without requiring the identification of
the specific pollutants.
224
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
Abbreviations and Acronyms
ADEM
AIS
ALAWADR
ALU
ANOVA
aRPD
ATtiLA
AUSRIVAS
BCG
BEAST
BMP
BT
CADDIS
CART
CBP
CCA
CFR
CIBI
CNMI
CRW
CT DEEP
Cu
CWA
CWH
DELT
D-IBI
DO
ED AS
EMAP
EPA
EPT
ESD
E/T
EV
FACI
F-IBI
GAM
GHQW
Alabama Department of Environmental Management
aquatic invasive species
Alabama Water-Quality Assessment and Monitoring Data Repository
aquatic life use
univariate analysis of variance
apparent redox potential discontinuity
Analytical Tools Interface for Landscape Assessments
Australian RIVer Assessment System
biological condition gradient
BEnthic Assessment of SedimenT
best management practice
brook trout
Causal Analysis/Diagnosis Decision Information System
classification and regression tree (statistical analysis)
Chesapeake Bay Program
Canonical Correspondence Analysis
Code of Federal Regulations
Continuous Index of Biological Integrity
Commonwealth of the Northern Mariana Islands
Coral Reef Watch, NOAA
Connecticut Department of Energy and Environmental Protection
copper
Clean Water Act
coldwater habitat
deformities, erosion, lesions, and tumors
diadromous index of biotic integrity
dissolved oxygen
Environmental Data Acquisition System
Environmental Monitoring and Assessment Program
U.S. Environmental Protection Agency
ephemeroptera, plecoptera, trichoptera taxa
environmental site design
endangered/threatened
exceptional value
Fish Assessment Community Index
fish index of biological/biotic integrity
general additive model
General High Quality Water
225
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
GLEI
GRE
GRFIn
GSA
HDG
HDS
HQ
HUC
HWI
IBI
1C
ICI
LDI
LDM
LRBOI
LRW
LWD
MANOVA
MCDEP
MEDEP
M-IBI
mlBI
Mlwb
MMI
M-NCPPC
MPCA
MWH
NA
NBEP
NCRMP
NELP
NHD
NIH
NJDEP
NLCD
NMDS
NOAA
NPDES
NRC
0/E
OM
Great Lakes Environmental Indicators
Great Rivers Evaluation
Great River Fish Index
generalized stress axis
human disturbance gradient
human disturbance score
high-quality
hydrologic unit code
Healthy Watershed Index
index of biological/biotic integrity
impervious cover
invertebrate community integrity index
landscape development intensity index
linear discriminant model
Little River Band of Ottawa Indians
limited resource water
large, woody debris
multivariate analysis of variance
Montgomery County Department of Environmental Protection
Maine Department of Environmental Protection
macroinvertebrate index of biological/biotic integrity
modified index of biological integrity
modified index of well-being
multimetric index
Maryland-National Capital Park and Planning Commission
Minnesota Pollution Control Agency
modified warmwater habitat
non-attainment
Narragansett Bay Estuary Program
National Coral Reef Monitoring Program
New England large rivers
National Hydrography Dataset
National Institutes of Health
New Jersey Department of Environmental Protection
National Land Cover Database
non-metric multidimensional scaling
National Oceanic and Atmospheric Administration
National Pollutant Discharge Elimination System
National Research Council
observed over expected
organic matter
226
-------
A Practitioner's Guide to the Biological Condition Gradient
February 2016
ONRW
OSI
OSW
PADEP
PAR
PCA
QHEI
POM
REMAP
RIDEM
RIVPACS
RM
RPS
SHQW
SPI
SST
STORE!
TALU
TBEP
TITAN
TIV
TMC
TMDL
TNC
UAA
UMRBA
UMR
USVI
WDG
WSIO
WQS
WQTF
WQV
WWH
WWTF
Outstanding National Resource Water
Organism-Sediment Index
Outstanding State Water
Pennsylvania Department of Environmental Protection
photosynthetic active radiation
Principal Component Analysis
qualitative habitat evaluation index
particulate organic matter
Regional Environmental Monitoring and Assessment Program
Rhode Island Department of Environmental Management
River Invertebrate Prediction and Classification System
river mile
Recovery Potential Screening
Superior High Quality Water
sediment profile imagery
sea surface temperature
STOrage and RETrieval
tiered aquatic life use
Tampa Bay Estuary Program
Threshold Indicator Taxa ANalysis
Tolerance Indicator Value
Ten Mile Creek
Total Maximum Daily Load
The Nature Conservancy
use attainability analysis
Upper Mississippi River Basin Association
Upper Mississippi River
U.S. Virgin Islands
watershed disturbance gradient
Watershed Index Online
water quality standards
Water Quality Task Force
Weighted Stressor Value
warmwater habitat
wastewater treatment facility
227
-------
4>EPA
EPA 842-R-16-001
photos courtesy of Maine Department of Environmental Protection
. /
------- |