vvEPA
United States Off ice of Water
Environmental Protection (WH-586)
Agency Washington, DC 20460
EPA-440/5-91-OD5
July 1991
Biological Criteria:
Research and
Regulation
Proceedings of a Symposium
Printed on Recycled Paper
-------
-------
Proceedings of symposium
Biological Criteria:
Research and Regulation
December 12-13, 1990
Hyatt Regency Crystal City, Arlington, Virginia
Sponsored by the
Office of Water
U.S. Environmental Protection Agency
1991
-------
This proceedings was compiled from presentations made at a symposium convened by Dr.
Suzanne Marcy, U.S. Environmental Protection Agency. Mr. David Penrose, North Caro-
lina Department of Environment, Health, and Natural Resources, significantly contributed
to this document and his work is gratefully acknowledged.
Any questions, inquiries, or requests for copies of these proceedings should be addressed
to Dr. George R. Gibson, Jr., Office of Science and Technology, U.S. Environmental
Protection Agency, 401 M Street, SW (WH-585), Washington, DC 20460,. •
Prepared by JT&A, inc. for the U.S. Environmental Protection Agency. Publication does not
signify that the contents necessarily reflect the views and, policies of the Environmental .
Protection Agency, nor does mention of trade names or commercial products constitute .
endorsement or recommendation for use.
-------
CONTENTS
Opening Remarks 1
Martha Prothro
Establishing Biological Criteria: Functional Views of Biotic Community Organization 3
Kenneth W. Cummins
Biocriteria in Regulations: the EPA Headquarters NPDES Permitting View 9
James F. Pendergast
Biocriteria in the Water Quality, Management Process . . .............. ; 12
Bruce J. Newton . . • •:•»- pw;-.," -.-• . ••- •
Biological Criteria: A Regional Perspective 13
Ronald Preston and Linda Hoist
A State Perspective on Biological Criteria in Regulation .. ...,............;... .15
Daniel R. Dudley
Utilization of Biological Information in North Carolina's Water Quality Regulatory
Program 19
Jimmie Overton
Use of Habitat Assessment in Evaluating the Biological Integrity of Stream Communities .... 25
Michael T. Barbour and James B. Stribling
Variability in Lakes and Reservoirs 39
John Magnuson
Biological Criteria Development for Marine and Estuarine Habitats ...,;... .40
JackQ. Word , .•,.......
North American Wetlands: Habitats Shaped by Dynamic Hydrologic Fluctuations '.". ..... .41
Frederic A. Reid ' - •
Reference Ecosystems of the Upper Mississippi River—Past, Present and Future 45
KennethS. Lubinski
Biocriteria for Lacustrine Systems: A Case History from the Laurentian Great Lakes 46
John E. Gannon
*
The Development of Biocriteria in Marine and Estuarine Waters in Delaware 47
John R. Maxted
The Puget Sound Wetland Restoration Monitoring Protocol 55
Ronald M. Thorn
Relationships Among Water and Sediment Contamination Toxicity and Community
Responses in the Trinity River, Texas 61
James H. Kennedy, Kenneth L Dickson, William T. Waller, and Ray Arnold
Designing Surveys to Assess Biological Integrity in Lakes and Reservoirs 62
James R. Karr and Michele Dionne
" : ! fir
-------
Fish Assemblages as Indicators of Environmental Quality in Chesapeake Bay 73
Stephen J. Jordan
Selection of Biological Indicators for Integrating Assessments of Wetland, Stream, and
Riparian Habitats -,-.;.: ^ ............... .81
Robert P. Brooks, Mary Jo Croonquist, Elizabeth T. D'Silva, and Joseph E, Gallagher
Biological Survey Study Design Considerations When Representing Biointegrity and
Evaluating Non-Attainment of Designated Uses .90
James M. Lazorchak
Significance of Change in Community Structure: A New Method for Testing Differences . . . . .91
James R. Pratt
Errors in Errors in Hypothesis Testing . . . 104
BrockB. Bernstein i'.: •" . .
The Integrated Biosurvey as a Tool for Evaluation of Aquatic Life Use Attainment and
Impairment in Ohio Surface Waters . .... . . 110
Chris O. Yoder
Using Machine Learning Techniques to Visualize and Refine Criteria for Biological
Integrity ; . . 123
Kenneth Anderson, Albert Boulanger, Herbert Gish, James Kelly, and Jeffrey Morrill
Poec/7/ops/s: A Fish Model for Evaluating Genetically Variable Responses to
Environmental Hazards . . . , . 1129
Lawrence E. Hightower and R. Jack Schultz . . . '
Assessing Biological Integrity Using, EPA Rapid Bioassessmeht Protocol II —
The Maryland Experience 131
Niles L Primrose, Walter L. Butler, and Ellen S. Friedman
The Use of the Qualitative Habitat Evaluation Index for Use Attainability Studies in .
Streams and Rivers in Ohio . . . . . . . .'. • • • 133
Edward T. Rankin
The Use of the Amphipod Leptocheirus Plumulosus to Determine Sediment Toxicity in
Chesapeake Bay: Development and-Field Applications 134
C,E. Schlekat, B.M. McGee, and E. Reinharz
Compliance Monitoring of the Aquatic Biota in Vermont 135
Doug Burnham, Steve Fiske, and Rich Langdon
A Method for Rapid Bioassessment of Streams in New Jersey Using Benthic
Macroinvertebrates • • • 138
James Kurtenbach
Development of Biological Impairment Criteria for Streams in New York State 139
Robert W. Bode
Development of Sediment Criteria for the Protection and Propagation of
Salmonid Fishes '. 142
Timothy A. Burton, William H. Clark, Geoffrey W. Harvey, and Terry R. Maret
Biomonitoring Methods in the Tennessee Valley 145
Anne £ Keller
-
-------
Development of Biological Criteria, for Use in Maine's-Water Quality Classification
Program 147
Susan P. Davies, Leonidas Tsomides, and David L Courtemanch
The Relationship of Chironomidae (Diptera) Community Structure to Chlordane Levels
in Sediments of Streams Near St. Louis, Missouri . ... . . . . . . 151
Christopher A. Wright
Regional Standardization of Taxonomy . . . . . 155
Lawrence L. Love//
Integrated Chemical and Biological Monitoring of Sun Creek, McPhearson County,
Kansas, U.S.A 158
N.H. Crisp, LC. Ferrington, and L Cowles
The Use of Biocriteria in the Ohio EPA Biological Monitoring and Assessment Program .... 159
Chris O. Voder
Water Quality Indicators for Rivers and Streams: Selection, Stratification and
Aggregation for Decisionmaking 160
J. Harrison
Aquatic Macroinvertebrates as Biological Indicators of Water Pollution in Arizona 161
J.F. Boggs, C. Olson, M. Lowry, M. Longsworth, R. Williams,'J. Clayton, £ Swanson, and F.
Woodwick
Development of Diagnostic Procedures, to Evaluate Aquatic Resources in Regional
Watersheds -
John W. Arthur
162
Conference Attendees List 163
Index of Authors .'....' 171
v
-------
-------
Biological Criteria: Research and Regulation, 1991
Opening Remarks
Martha Prothro
Office of Water
U.S. Environmental Protection Agency
Washington, D.C.
This symposium is in part the result of a se-
ries of developments within EPA which
have progressively led to a more expanded
and comprehensive role by the Agency in the pro-
tection and restoration of our nation's surface water
resources.
These developments particularly,received im-
petus from the mandate of the 1987 Clean Water Act
and from a report written by our Administrator,
William Reilly. He wrote "Meeting Environmental
Challenges" earlier this year which summarized
EPA's progress and innovations in environmental
protection. As head of our agency, Bill Reilly urged
our renewed commitment to ecology and natural
resources conservation. The Agencies' mission goes
beyond protection of human environmental inter-
ests. EPA is equally responsible for protecting fish
and wildlife habitats and other ecological systems.
In his report and in many other reports and direc-
tives, Bill Reilly has set Agency policy and given
EPA a clear mandate to truly protect the environ-
ment in the broadest sense of the phrase.
Contributing to Bill Reilly's strong statement
was a report by the EPA Science Advisory Board
which stressed much the same breadth of concern.
That 1987 report, "Unfinished Business", also advo-
cated that we assume an expanded environmental
responsibility and it identified many vital environ-
mental risks requiring attention.
The basic, and perhaps obvious, conclusion
therefore is that many environmental problems still
exist even after 20 years of progress. EPA must now
view its responsibilities for broad environmental
protection as a proactive policy. It should do so as
an expansion of our earlier agenda of pinpointing
environmental strategies to react to preexisting
problems. Thus, EPA recognizes that we should pro-
vide more guidance to state and federal authorities
as to what is necessary to protect as well as to restor
the environment. With relative success in the latter,
we can now emphasize the former.
Finally, EPA must lead the nation in under-
standing that national environmental policy jnust
include new initiatives focusing more on opportuni-
ties for improvement through nonregulatory and
nonlegislative means. There is definitely now an
emphasis on outreach; on training and education; as
well as on identifying economic incentives to en-
courage the public to help protect the environment
through responsible consumerism such as energy
and water conservation, safe product disposal, and
recycling.
More specific to our meeting here, the Science
Advisory Board also recommended that the EPA at-
tach as much importance to reducing ecological risk
as it does to reducing human health risk. This focus
is especially relevant to this conference because nat-
ural reproductive ecosystems are essential to
human health and to sustaining long-term eco-
nomic growth. These natural systems are, of course,
also intrinsically valuable for their own sake.
The Board has also recommended that EPA im-
prove the data gathering, handling procedures,
sampling methodologies, and assessment methods
to identify and evaluate environmental risks. They
urged us to develop more and better scientific and
technical measures to improve our understanding
of primary ecological risk problems and to identify
solutions.
In his subsequent report, Administrator Reilly
focuses on four areas of risk the Board identified as
critical. These include: habitat alteration and de-
struction which has always been a prime concern of
EPA, particularly with respect to wetlands; species
extinction and overall loss of biological diversity,
also very important to our Office; and ozone deple-
tion and global climate change both of which affect
many EPA initiatives.
We feel that work on ecological criteria is essen-
tial to several of the objectives of the Office of Water.
These include not only the regulation of discharges,
but also the control of nonpoint source pollution
-------
M. PROTHRO
and storm water runoff and the determination of
overall surface water quality problems and man-
agement priorities. In pursing this direction, we are
concerned about biological community quality, in-
tegrity, and diversity of the collection of species and
organisms. Thus, we aim to develop biological cri-
teria that can be used independently, but to comple-
ment not replace existing physical and chemical
water quality criteria and standards.
Several states have already begun to implement
such ecological criteria as will be documented by
the presentations at this conference. We are depend-
ing on you, the scientists, citizens, and state agency
managers here to help us meet the challenge pre-
sented to the Office of Water by the Administrator.
We want to consult with top federal, state and pri-
vate organizations to help us design our biological
criteria.
Thank you all for being here. We look forward
very much to your participation and contributions,
and I hope this is just a part of the beginning as we
continue through the development and the imple-
mentation of biological criteria.
-------
Biological Criteria: Research and Regulation, 1991
Establishing Biological Criteria: Functional
Views of Biotic Community Organization
Kenneth W. Cummins
Department of Biological Sciences and
the Pymatuning Laboratory of Ecology
University of Pittsburgh
Pittsburgh, Pennsylvania
As we enter the 90s, public awareness of en-
vironmental issues is at an all-time high,
because the planet's environmental
"health" is at an all time low. Since water is prob-
ably the most critical limiting resource, the status of
surface waters is often at the forefront of concern.
Terminology such as "drinkable," "swimmable,"
fishable," and "biological integrity" in the 1972
Water Quality Act, and its subsequent amend-
ments, indicated the broad nature of this concern.
Over the past few years, the Nation has ad-
dressed the relatively easy, short-term problems of
point source contamination with reasonable suc-
cess. In many cases, chemical analyses based on the
single grab sample technique were sufficient to
identify a problem and verify the remediation.
However, it soon became clear that the real long-
term problems stemmed from nonpoint sources.
Degradation over ever larger spatial and longer
temporal scales was the result of agriculture (row
and field crops, grazing), timber harvest (clear cut-
ting), surface mining, landscape pesticide applica-
tions, general urbanization, and regional watershed
fallout such as acid rain.
Monitoring strategies such as EPA's MAP
(Environmental Monitoring Assessment Program)
and the U.S. Geological Survey NAWQA Program
(National Water Quality Assessment Program) are
attempting to assess the severity of the problem.
Yet, as before, significant questions still remain re-
garding how to evaluate the condition of our sur-
face waters, and what role biological criteria should
play in its assessment.
The Gap in Technology
Transfer
The fundamental problem is that the process of
establishing biological criteria for water quality as-
sessment has not been coupled to the major ad-
vances in lotic ecology of the last 25 years. Some ex-
amples are: the River Continuum Concept
(Vannote et al. 1980; Minshall et al. 1983, 1985;
Cummins et al. 1984; Cummins, 1988), including
the related paradigms of the Intermediate Distur-
bance and the Serial Discontinuity Hypotheses
(Ward and Stanford, 1983); Nutrient Spiralling (El-
wood et al. 1983); Riparian Control (Cummins,
1988); Watershed Budgets (Fisher and Likens, 1973;
Cummins et al. 1983); and Functional Groups
(Cummins and Klug, 1979; Merritt and Cummins,
1984; Cummins and Wilzbach, 1985). Furthermore,
few of the researchers who have played major roles
in developing the above paradigms in running
water ecology have been involved in the develop-
ment or implementation of the assessment criteria.
Symptomatic of this schism is the almost com-
plete dependence of the biocriteria development on
taxonomic procedures, while basic researchers have
moved into questions or process and function.
Thus, an identification and quantification approach
formed the basis for biological water quality assess-
ment and remains fundamental today. The pattern
can be seen in EPA workshops: indicator species in
the late 1960s, diversity indices in the late 1970s, and
biocriteria in the 1980s.
The reliance on taxonomically dependent nu-
merical indices, most frequently applied to inverte-
brates, has severely limited the development of
bioassessment indices. The main problem is that the
taxonomic definition of North American freshwater
invertebrates is an ongoing, long-term process, with
all groups remaining incompletely described.
Probably the most extreme example among the
macroinvertebrates is the dipteran insect family, the
Chionomidae. This family is, almost without excep-
tion, the most diverse species complex in any run-
ning water system (Merritt and Cummins, 1984),
-------
K.W. CUMMINS
and yet it typically is the least known. In a Pennsyl-
vania stream that has been studied for more than 25
years, over 180 species of Chionomidae have been
found on an annual basis (Coffman et al. 1971; Coff-
man 1973, 1974). Many of these species are un-
described and keys for their separation await
monographic treatment. It is clear that the assign-
ment of a taxonomic value of 1 to the Chionomidae
in an assessment of a running water system will
usually subsume a species complex in excess of 50
(on an annual basis).
Given the embryonic state of taxonomic resolu-
tion of freshwater organisms, significant support for
taxonomic research efforts on the part of agencies
that need the information, such as EPA, might be
expected. Unfortunately, such support is not forth-
coming. Thus, derived indices that purport to detect
biological redundancy in a given system usually
find redundancy in the taxonomy. Because of the
unavailability of sufficient taxonomic keys to sepa-
rate running water organisms, process- or function-
related measurements should be prime criteria for
incorporation into bioassessment and biomonitor-
ing protocols. Agencies such as EPA would greatly
benefit from research devoted to moving some of
the time-tested methods in the area of process and
function in basic running water ecology into the
field of application.
Reference Sites
The concept of reference sites within a given eco-
logical region (Omernik, 1987) has been a welcome
addition to strategies used in the evaluatibn of
fresh Waters (Hughes et al. 1986), particularly by*
state Environmental Protection Agencies. Evalua-
tion of the biological condition of a given freshwa-
ter system relative to an appropriate -reference
("control") site should be fundamental. However,
the site to be evaluated must be placed not only in
the appropriate spatial context (e.g., ecoregion,
basin, watershed), but also within the relevant tem-
poral (historical) perspective (Sedell and Frogatt,
1984; Cummins et al. 1984; Cummins, 1988).
• Furthermore, the selection of reference sites
needs careful attention. Any comparison must be
made between systems of similar scale because
there are fundamental differences between running
water ecosystems that transcend ecoregioris. For
example, paramount in the River Continuum Con-
cept is the idea that relative position of a stream in
its watershed, designated by stream order, stream
size, or drainage area, confers some similarities in
ecosystem structure and function independent of
the local setting of geology, soil, and stream-side ri-
parian vegetation (Vannote et al. 1980; Minshall et
al. 1983, 1985). Riparian vegetation can also exert
influences that override such parameters as alti-
tude, stream gradient, of bottom sediment type.
First-, second-, and many third-order streams
in the Cascade Mountains of the Pacific Northwest
have dominant populations of invertebrates,
termed shredders, that feed on litter derived from
the riparian zone. Such headwater streams of vari-
ous gradients and lithology share this functional
component and differ from streams of higher order
(e.i. < 3 or 4) that are in the same region. However,
if the. riparian vegetation is predominantly decidu-
ous, especially red alder (Alnus rubra), the normal
recovery tree following a disturbance (such as clear
cutting), the shredder populations are active in the
streams largely in the fall-winter period. If the, ri-
parian vegetation is primarily coniferous, the
shredder populations are most active in the spring-
summer period (Cummins, et al. 1989). Thus,
stream size is linked to the presence of shredders
through the .direct influence of the riparian zone,
but the timing of the growth period of the shredder
populations is keyed to the type of riparian plants
and the timing of the litter inputs (Cummins, 1974;
Cummins et al. 1989). The override is, clear because
both alder and conifer-dominated headwater
streams can be found in all of the major ecoregions
of the Pacific Northwest.
j,. Given .the nee,d for. carefully studied and ^well-
documented reference sites, it is quite surprising
that most of the best running water research sites
have not been used for comparison. Much of the in-
formation on these potential reference sites is well
documented in the literature.-The tendency of
workers involved in bioassessment to ignore this
natiprial data resource, which exists for virtually
every major ecoregion, is further evidence of the
gap between basic research and assessment. Many
of the potential reference sites are associated with
ongoing national research initiatives such as the
National Science Foundation's Long-Term Ecologi-
cal Research Program (OTER), the U.S. Geological
Survey's National Water Quality Assessment Pro-
gram (NAWQA), and projects at many of the 200
inland biological field stations. There can be no
doubt that incorporation of these sites, arid collabo-
ration with their resident researchers, would help
fill EPA's need for an extensive and carefully docu-
mented matrix of reference sites.
Function/Process
Measurements as Biocriteria
Although many' examples of procedures designed
to incorporate functional and process-related data
as biocriteria for bioassessment and biomonitoiirtg
-------
Biological Criteria: Research and Regulation, 1991
could be discussed, only two are given here. Both
are based on the belief that the macroinvertebrates .
are often the most suitable lotic ecosystem compo-
nents for such assessments. The small, rapidly re-
producing microorganisms (algae, bacteria, etc.)
that are quantitatively most significant at the level
of nutrient cycling require specialized techniques
for analysis. However, some methods that monitor
the general activity levels of these organisms, such
as the measurement of P/R (the ratio of gross pri-
mary production to total community respiration)
certainly could be employed as biological criteria
of ecosystem condition (Minshall et al. 1983).
At the other end of the scale, fish, often the bio-
logical components of most direct human interest,
can also serve as effective subjects to establish
biocriteria for the assessment process (Karr, 1981,
1987, 1991). However, analysis of fish populations
requires special equipment due to their low densi-
ties and species complexity relative to microorga-
nisms and invertebrates, and their great mobility.
Thus, the invertebrates constitute both a biolog-
ical and operational link between the microorga-
nisms and fish. That is, most invertebrates feed on
microorganisms and in turn serve as food for fish,
and their size, abundance, species diversity, ease of
capture, and annual life cycle enhance their suitabil-
ity for observation and interpretation relevant to
ecosystem function and processes (Cummins and
Klug,1979).
The example of a functional analysis to be dis-
cussed here was first described 18 years ago
(Cummins, 1973) and has been modified in some
details since then (Cummins and Klug, 1979; Merritt
and Cummins, 1984; Cummins and Wilzbach, 1985).
The macroinvertebrate Functional Feeding Group
(Cummins and Wilbach, 1985) method is based on
the association between a limited set of feeding ad-
aptations found in freshwater invertebrates and
their basic nutritional resource categories. These
food resources are categorized as detritus (coarse
CPOM, or fine FPOM, particulate organic matter
and the associated microbiota), periphyton (at-
tached algae and associated entrained material),
live macrophytes, and prey (Cummins, 1974).
The level of morphological and behavioral ad-
aptation of the invertebrates that allows them to ex-
ploit these resource categories can be obligate or
facultative (Cummins and Klug, 1979). The obligate
specialist forms are more readily displaced and the
facultative generalists are more tolerant under con-
ditions of disturbance. The presence and abundance
of the various functional feeding groups, and the
dominance of obligate or facultative representa-
tives, is a direct reflection of the availability of the
required food resources and the condition of the re-
lated environmental parameters.
The invertebrate functional groups are:
• Shredders feeding on CPOM (primarily litter
of terrestrial origin from the riparian zone) ;
, or live macrophytes; ' _ . .
* Collectors feeding on FPOM either by
filtering from the water column (filtering
collections) or by "mining" the sediments of
browsing surface deposits (gathering
collectors);
• Scrapers feeding on periphyton;
• Piercers feeding on macroalgae by piercing
individual cells; and
• Predators feeding on prey (Cummins and
, Wilzbach, 1985).
The analysis is made on a hierarchical basis of
increasing levels of resolution. The first (lowest)
level of resolution allows separation of live inverte-
brate collections in the field at an efficiency of 80 to
85 percent. The second level of resolution increases
the efficiency another 5 to 10 percent. When com-
parisons are to be made between sites on a regional
basis the level of resolution must be set so that all
workers involved in the assessment can accomplish
the task. Levels of greater resolution allow groups
having the appropriate expertise to produce more
detailed analyses. An example of the method from
Cummins and Wilzbach (1985) is given in Figure 1.
Assignments into functional groups of most of the
North American genera of aquatic insects can be
found in the ecological tables in Merritt and
Cummins (1984). .'...- ,
The figure shows two levels of resolution are in-
dicated; the first level allows general separation, the
second level allows correction of the assignments of
some of the more common exceptions. The third
level of resolution would rely on the taxonomic .res-
olution (essentially at the generic level) given in
Merritt and Cummins (1984). •;...••
. The Leaf Pack Bioassay method is an example
of a procedure that would be suitable for establish-
ing biocriteria to be used in freshwater ecosystem
assessment and monitoring. The technique was first
described 17 years ago (Peterson and Cummins,
1974) and has undergone some modification since
(Merritt et al. 1979; Hanson et al. 1985). The objec-
tive of the method is to use the rate of leaf litter pro-
cessing, defined as the total weight loss from
simulated litter accumulations from all causes, as a
general measure of ecosystem structure and func-
tion. The method is suitable primarily for low order
-------
K. W. CUMMINS
FIRST LEVEL OF RESOLUTION
IN PORTABLE CASE
Caddisflies (order Trichoptera)
1 CASES ORGANIC
«^-' Leaf, stick, needle, bark
Families Limnephilidae (in part),
Lepidostomatidae (in part),
Phryganeidae, Leptoceridae (in part)
CASES MINERAL
Sand, fine gravel
Families Glossosomatidae, Limne-
philidae (in part), Helicopsychidae
SHBEDDEBS
SECOND LEVEL OF RESOLUTION considers a few fairly common caddisflies that would be misclassified
above on the basis of case composition alone.
CASES ORGANIC CASES MINERAL
Cases square in cross section and
tapered, with no bark or flat leaf
pieces Included. Front attached to
substrate. Larvae extend legs and
filter the current.
Foreleg with
filtering hairs
Cases long, slender, and tapered,
made of plant material
Cases long, slender, and tapered
"(mostly fine sand) or cases ovoid
and very flat in cross section
Family Leptoceridae (in part)
Family Leptoceridae (in part)
Figure 1.—An example of functional group separations In the case-bearing caddisflies (Cummins and Wllzbach, 1985).
-------
Biological Criteria: Research and Regulation, 1991
. (1-3 or 4), headwater streams where litter inputs
and invertebrate shredder activity will be maxi-
mized per unit of habitat.
Processing rates have been determined for a
wide variety of riparian plant species in a range of
stream orders (Webster and Benfield, 1986). In gen-
eral, the method involves the preparation of packs
of leaves of a given species, or species combination,
that are fastened together with plastic Tees of the
type used to fasten buttons or price labels to cloth-
ing. The packs of preweighted leaves (usually 5
grams dry weight) are attached to elastic bands,
slipped around common masonry bricks, and
placed on the stream bottom with the pack facing
into the current to simulate a natural accumulation
of litter at the leading edge of an obstruction in the
water. After 24 to 28 hours, the initial sets of packs
are removed to determine leaching and handling
weight loss; additional sets of packs are removed at
intervals, determined by temperature (usually
every 150 to 300. degree days) to follow the rate of
processing (per degree day).
The invertebrates are removed and the recov-
ered packs are reweighed. The invertebrate func-
tional groups can be noted, especially shredders.
Other analyses such as microbial population com-
position, densities, and biochemical changes from
initial conditions can also be performed. The leaf
pack bioassay method has been shown to accurately
simulate natural rates of processing of unconfined
leaves in exposed, aerobic stream sites, which was
not true for litter held in mesh bags (Gummins et al.
1980).
As indicated previously/the examples given are
only two among many that are well documented in
the literature and have'significant potential for use
in establishing biocriteria in freshwater ecosystems.
The point is that very few such methods have been
incorporated into bioassessment and biomonitoring
and, given the present policy of agencies such as
EPA, it is unlikely that they will be. For example, the
original papers describing the two methods used as
examples above have both been designated "citaT
tion classics." Thus, it is clear that such techniques
are well known (i.e., adequately documented) but,
with few exceptions (Plafkin et al. 1989), they have
not made the technological transfer to the problem-
solving arena within which agencies such as EPA
operate.
References
Coffman, W.P. 1973. Energy flow in a woodland stream eco-
system: II. The taxonomic composition and phenology
of the Chironomidae as determined by the collection of
pupal exuviae. Arch. Hydrobiol. 71:281-322.
—• . 1974. Seasonal differences in the diel emergency of a
lotic chironomid community. Ent. Tidskr. 95:42-48.
Coffman, W.P., K.W. Cummins, and J.C.. Wuycheck. 1971, En-
ergy flow in a woodland stream ecosystem: I. Tissue
; •',' support-trophic structure of the autumnal community.
' .Arch. Hydrobiol. 68:232-76.
Cummins, K.W. 1973. Trophic relations of aquatic insects.
Ann. Rev. Ent. 18:183-206.
-. 1974. Structure and function of stream ecosystems.
BioScience 24:631-41.
. 1988. The study of stream ecosystems: a functional
view. In Pomeroy, L.R. and J.J. Alberts, eds. Concepts of
ecosystem ecology. Springer-Verlag, New York.
Cummins, K.W. and M.J. Klug. 1979. Feeding ecology of
stream invertebrates. Ann. Rev. Ecol. Syst. 10:147-72.
Cummins, K.W., G.W. Minshall, J.R. Sedell, and R.C.
Petersen. 1984. Stream ecosystem theory. Verh. Int. Ver-
''ein. Limnol, 22:1818-27.
Cummins, KW. and M.A. Wilzbach. 1985. Field procedures
for the analysis of functional feeding groups in stream
ecosystems. Pymatuning Laboratory of Ecology, Univ.
Pittsburgh, Linesville, PA.
Cummins, K.W., M.A. Wilzbach, D.M. Gates, J.B. Perry, and
W.B. Taliaferro. 1989. Shredders and riparian vegeta-
tion. BioScience. 39:24-30.
Cummins, K.W., J.R. Sedell, F.J. Swanson, G;W, Minshall,
S.G. Fisher, C.E. Gushing, R.C. Peterson, and R.L.
Vannote. 1989. Organic matter budgets for stream eco-
systems: problems in their evaluation. Pages 229-353 in
Barnes, J.R. and G.W. Minshall, eds. Stream ecology. Ple-
num Press, NY.
Cummins, K.W., G.L. Spengler, G.M. Ward, R.M. Speaker,
R.W. Ovink, and D.C. Mahan. 1980. Processing of con-
fined and naturally entrained leaf litter in a woodland
. stream ecosystem. Limnol. Oceanogr. 25:952-57.
Elwood, J.W., J.D. Newbold, R.V. O'Neill, and W. VanWinkle.
1983. Resource spiralling: an operational paradigm for
1 analyzing lotic ecosystems. Pages 3-27 in Fontaine, T.D.
and S.M. Bartell, eds. The dynamics of lotic ecosystems.
Ann Arbor Sci., Ann Arbor, ML
Fisher, S.G. and G.E, Likens! 1973. Energy flow in Bear Brook,
New Hampshire: an integrative approach to stream eco-
system metabolism. Ecol. Monogr. 43:421^39.
Hanson, B.J., K,W. Cummins, J.B. Barnes, and M.W. Carter.
1985. Leaf litter processing in aquatic systems: a two
variable model. Hydrobiologia 111:21-29.
Hughes, R.M., D.P. Larsen, and J.M. Omernik. 1986. Regional
reference sites: a method for assessing stream pollution.
Environ. Manage. 10:629-35.
Karr, J.R. 1981. Assessment of biotic integrity using fish com-
munities. Fisheries 6:21-17.
. 1987. Biological monitoring and environmental as-
sessment: a conceptual framework. Environ. Manage;
11:249-56. • ... ':;'''. : , ,,;
. In press. Biological integrity: a long neglected aspect
of water resource management. Ecol. Applications 1: in
press.
Merritt, R.W. and K.W. Cummins, eds. 1984. An introduction
to the aquatic insects of North America. Kendell/Hunt,
Dubuque,IL.
Merritt, R.W., K.W. Cummins, and J.R. Barnes. 1979. Demon-,
stration of stream watershed community processes with
some simple bioassay techniques. Pages 101-13 in Resh,
V.H. and D.M. Rosenberg, eds. Innovative teaching in
aquatic entomology. Can. Spec. Publ. 43. Fish Aqua. Sci.
-------
K. W. CUMMINS
Minshall, F.W., R.C. Peterson, K.W. Cummins, T.L. Bott, J.R.
Sedell, C.E. Gushing, and R.L. Vannpte. 1983. Inter-,
biome comparison of stream ecosystem dynamics. Ecol.
Monogr.53:1-25. '.,.•-.. •
Minshall, G.W., K.W. Cummins, R.C. Peterson, C.E. Gushing,
D.A. Bruins, J.R. Sedell, and R.L. Vannote, 1985. Devel-
opments in stream ecology. Can. J. Fish Aquat. Sci.
43:1045-55.
Omemik, J.M. 1987. Ecoregions of the coterminous United
States. Ann. Ass. Am. Geogr. 77:118-25.
Plafkin, J.L., M.T. Barbour, K.D. Porter, and S.K. Gross. 1989.
Rapid bioassessment protocols for use in streams and
rivers. Benthic macroinvertebrates and fish. EPA-444/4-
89-001. U.S. Environ. Prot. Agency, Washington, DC.
Sedell, J,R. and J.L. Frogatt. 1984. The importance of streams-
ide forests to large rivers: the isolation of the Willamette
River, Oregon, U.S.A., from its floodplain. Verh. Int.
Verein. Limnol. 22:1828-34. , , ,
Vannote, R.L., G.:W~. Minshall, K.W. Cummins, J.R. Sedell,
C.E. Gushing. 1980, The river continuum concept. Can-
J. Fish. Aquat. Sci. 37:130-37.
Ward, J.V. and J.A. Stanford. 1983. The serial discontinuity
concept of lotic ecosystems: Pages 29-42 in Fontaine,
T.D. and S.M. Bartell, eds. Dynamics of lotic ecosystems.
Ann Arbor Sci., Ann Arbor, MI. ,
Webster, J.R. and E.F. Benfield. 1986. Vascular plant break-
down in freshwater ecosystems. Ann. Rev. Ecol. Syst.
17:567-94.
8
-------
Biological Criteria: Research and Regulation,
Biocriteria in Regulations: the EPA
Headquarters NPDES Permitting View
James F. Pendergast
Chief, Water Quality Permitting Section
U.S. Environmental Protection Agency
Washington, D.C.
ABSTRACT
Sections 303 and 304 of the Clean Water Act now require States to develop biological criteria
as part of their water quality standards. Biological criteria are presently used in the NPDES
permitting program as a tool to identify waters that are not achieving their designated use,
and therefore may be impacted by point source discharges. In the future, there are two areas
in which biological criteria can be used in an NPDES permitting context. The first way is to
verify that NPDES permit limits are indeed resulting in achievement of State water quality
standards. A second approach that would require considerable development is to establish
NPDES effluent limits that directly assure compliance with biological criteria. Once biologi-
cal criteria become part of State water quality standards, NPDES regulations require that
permit limits assure compliance with these standards. To accomplish this, a permitting au-
thority must develop a protocol that can demonstrate the relationship of the biological cri-
teria to effluent characteristics. This goal has already been accomplished for toxicity.
However, EPA takes the position that biological criteria must not supersede chemical-spe-
cific numerical standards or toxicity tests now used to achieve compliance with the narrative
standards.
Introduction
Sections 303 and 304 of the Clean Water Act now
require States to develop biological criteria as part
of their water quality standards. As States begin to
adopt biological criteria, NPDES permitting au-
thorities are obligated by federal regulations to es-
tablish effluent limitations to ensure that these
criteria are maintained. This paper describes the
ways in which NPDES permits now assure compli-
ance with water quality standards, and explores
ways in which biological criteria could be used in
concert with NPDES permits to achieve full imple-
mentation of water quality standards.
NPDES Program Overview
National Point Discharge Elimination System per-
mits are the instruments by which EPA and States
control the types and amounts of pollutants dis-
charged into ambient receiving waters. The permits
contain effluent limits that set the maximum level
of discharge for each pollutant. The limits require
the use of best-treatment technology to maintain
State water quality standards. Many NPDES per-
mits are issued with technology-based limits that
are sufficient to protect the quality of the receiving
water. For those permits where technology-based
limits are insufficient, more stringent water quality
based limits are required. These are set at the maxi-
-------
J. F. PENDERGAST
mum level of a pollutant that, after mixing with the
receiving water, will not violate a water quality
standard.
EPA and States are now issuing a number of
NPDES permits that contain new requirements to
control the discharge of toxic substances that may
endanger aquatic life or human health. These per-
mits are issued to comply with EPA's national pol-
icy (Federal Register, 1984) stating that
... to control pollutants beyond Best Available
Technology Economically Achievable (BAT),
secondary treatment, and other Clean Water
Act technology-based requirements in order to
meet water quality standards, the EPA will use
an integrated strategy consisting of both
biological and chemical methods to address
toxic and nonconventional pollutants from
industrial and municipal sources. Where State
standards contain numerical criteria for toxic
pollutants, NPDES permits will contain limits
as necessary to assure compliance with these
standards. In addition to enforcing specific
numerical criteria, EPA and the States will use
biological techniques and available data on
chemical effects to assess toxicity impacts and
hitman health hazards based on the standard of
'no toxic materials in toxic amounts.'
Existing Uses of Biological
Criteria
The most effective use of biological criteria in the
NPDES permitting program is as a screening and
identification tool. Biological criteria provide the
ability to identify waters that are not achieving
their designated use, and therefore may be im-
pacted by point source discharges.
The best example of the use of biological criteria
in this fashion is in Arkansas, a State authorized to
write NPDES permits. The State has developed a
program to identify facilities with the potential for
causing ambient toxicity through the use of whole-
effluent toxicity (WET) testing. This program was
initially established to apply to "major" NPDES fa-
cilities—i.e, those facilities where initial information
suggested that toxics might be discharged. Arkan-
sas has also used biological criteria and rapid
bioassessments to identify other facilities that need
WET evaluation. At several of these sites, the State
found that a point source -facility was discharging
an effluent with sufficient toxicity to cause a biolog-
ical effect. Two of these were Superfund sites in the
remedial action stage.
In using biological criteria, EPA takes the posi-
tion that biological criteria must not supersede
chemical-specific numerical standards or toxicity
tests used to achieve compliance with the narrative
standards. Biological criteria and biosurveys should
be fully integrated with toxicity testing and chemi-
cal-specific assessment methods in State water qual-
ity programs. Whenever any one of the three types
of assessments demonstrates that the standard is
not attained, appropriate action should be taken by
the regulatory authority. However, since each
method has unique as well as overlapping charac-
teristics, sensitivities, and program applications, no
single approach for detecting impacts should be
considered inherently superior to any other ap-
proach. The inability to detect receiving water im-
pacts using a biosurvey alone is insufficient
evidence to waive or relax a permit limit established
using either of the other methods. The most protec-
tive results from each assessment conducted should
be used to establish the necessary NPDES permit
limits. This concept is fully discussed in the 1990
Technical Support Document for Water Quality-based
Toxics Control (U.S. Environ. Prot. Agency, 1990).
Potential Future Uses of
Biological Criteria
In the future, there are two areas in which biologi-
cal criteria can be used in an NPDES permitting
context. The first way is to verify that NPDES per-
mit limits are indeed resulting in achievement of"
State water quality standards. The biosurvey data
used for assessment against the criteria could be
collected by the regulatory authority or by the
NPDES permittee. EPA and States have the author-
ity under section 308 of the Clean Water Act to re-
quire information necessary to establish permit
limitations. In some instances this authority could
be used to require facilities to conduct the
biosurvey.
A second way in which biological criteria could
be used to assess water quality is to establish
NPDES effluent limits that directly assure compli-
ance with biological criteria. This approach requires
considerable development before it can be generally
implemented. Once biological criteria become part
of State water quality standards, NPDES regula-
tions require that permit limits assure compliance
with the standards. To accomplish this, a permitting
authority needs to demonstrate the relationship of
biological criteria to effluent characteristics.
This goal has already been accomplished for
toxicity. EPA conducted eight stream surveys to cor-
10
-------
Biological Criteria: Research and Regulation, 1991
relate effluent toxicity tests to actual ambient bio-
'logical impairment. The positive results of this re-
search enabled EPA to use effluent toxicity controls
in NPDES permits as a control mechanism. In addi-
tion/EPA developed a toxicity identification evalua-
tion (TIE) process to identify the causes of toxicity in
effluents and to suggest possible remediation mea-
sures. An analogous procedure is needed to fully
implement biological criteria in an NPDES context.
References
Federal Register. 1984. National Policy for Development of
Water Quality-based Permit Limitations for Toxic pollu-
tants. Off. Water, U.S. Environ. Prot. Agency, Washing-
ton, DC.
U.S. Environmental Protection Agency. 1990. Technical Sup-
port Document for Water Quality-based Toxics Control.
; Off. Water Enforce. Permits and Off. Water Reg. Stand.
Washington, DC.
11
-------
Biocriteria in the Water Quality
Management Process
Bruce J. Newton
Assessment and Watershed Protection Division
U.S. Environmental Protection Agency
Washington, D.C.
Recent EPA management directives have stressed the need to attach as much importance to reducing
ecological risk as to reducing human health risk. EPA and the States are beginning to address non-
traditional problems such as nonpoint source pollution, habitat degradation, the effects of land use
practices, and stormwater discharges. Many of these problems may not involve chemical pollution as the
principal stressor. Biological criteria (biocriteria) will be necessary to identify these types of problems and to
develop the tools needed to devise mitigation strategies. Biocriteria fit well into the existing regulatory ap-
proach. Substantial progress is being made to institute monitoring programs and resolve issues of consis-
tency in application. EPA has adopted a policy of independent application to govern the interpretation of
information from biosurveys, chemical analyses, and toxicity tests.
If you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
12
-------
Biological Criteria: Research and Regulation, 1991
Biological Criteria: A Regional Perspective
Ronald Preston
Linda Hoist
U.S. Environmental Protection Agency., Region III
Philadelphia, Pennsylvania
The Water Management and Environmental
Services Divisions in Region III have initi- -
ated various activities in preparation for
the integration of biological criteria into the existing
state water quality programs. The Divisions have
drafted a strategy which identifies both short-; and
long-term goals for incorporating biological criteria
and standardized bioassessment protocols into the
state programs and defines the responsibilities of
each Division to implement the strategy. The
Region's short-term goals include the adoption of
narrative biological criteria and the use of EPA's
Rapid Bioassessment Protocols by the state pro-
grams, and further definition and refinement of
ecoregions in support of numeric biocriteria. The
Region's long-term goals include the adoption of
numeric biological criteria by the states, establish-
ment of comprehensive data sets for ecoregion ref-
erence sites, and improvement of long-term water
quality assessments through the use of environ-
mental indicators.
The Water Management Division (WMD) has
the lead in providing program guidance and assis-
tance to the states for the incorporation of both nu-
meric and narrative biological criteria into their
water quality standards regulations. WMD has in- ••
formed all Region III states that in order to satisfy
EPA water program requirements, they are expected ,
to adopt narrative biological criteria for all surface
waters into their state water quality standards dur-
ing the FY 1991-1993 triennium. In addition, WMD
has begun working with individual States to deter-
mine what revisions are necessary to the existing
state water quality standards to meet the program.
requirement for biological criteria.
The Environmental Services Division (ESD) has
the lead in providing technical assistance to the
states in the areas of bioassessment methodologies
and their applications. During 1990, ESD, in cooper-
ation with EPA Headquarters, held four Rapid
Bioassessment Protocol workshops for the state
aquatic biology staffs. These workshops outlined:
and,demonstrated standard assessment procedures.
that are utilized to ensure consistent and valid data
collection in support of biological criteria.,Each ses-
sion consisted of: 1) description and discussion of,
major concepts, 2) field trip to demonstrate habitat
assessment and sampling techniques and 3) data
analysis and discussion of assessment results. Ap-
proximately 30 representatives from state, interstate
and federal agencies were in attendance at each
workshop. The Region received positive response,
and the state agencies are proceeding to adopt the
rapid bioassessment protocols.
In addition, ESD coordinated a workshop on a
specific habitat type (i.e., nontidal coastal streams)
at which data collection procedures utilized by the
Region III coastal states were described, standard-
ization of methods was outlined, and the develop-
ment of biocriteria for that system was initiated. At
the workshop, the attendees decided to organize
four subgroups whose purpose would be to de-
velop standard operating procedures for the follow-
ing areas: 1) field collection, 2) laboratory
processing and data analysis, 3) habitat assessment
and coastal stream characterization and 4) reference
site selection. These work groups,reviewed the on-
going standard operating procedures used by each
individual state and'developed common standard
operating procedures suitable for -all the states. The
final draft of the procedures will be presented at the
Region III Biology Workshop in March 1991. ESD
will continue to provide training and technical sup-
port to the states regarding the assessment proce-
dures and their application.
In addition, ESD is developing a cooperative
program that will assist the states in accumulating
baseline reference site data to support the biological
criteria program. This program will refine
ecoregions and subregions that cross state bound-
13
-------
R. PRESTON and L HOLST
aries and will provide coordination of regional and
interstate data collection activities for the develop-
ment of biological criteria. In 1991 the Region, in co-
operation with the EPA Office of Research arid
Development-Corvallis laboratory, will hold a Ref-
erence Site Workshop with the state agencies to
identify reference watersheds and to select reference
sampling sites for a pilot ecoregion (i.e., the Central
Appalachian Ridge and Valley). Following the
workshop, the reference sites will be field evaluated
in cooperation with the states. Once the reference
sites are finalized, comprehensive biological sam-
pling and habitat assessments will be periodically
performed to provide the necessary baseline data.
EPA Region JII strongly supports the expansion
of the use of biological parameters in the state pro-
grams. The Region's future role concerning the
biocriteria implementation will bp to provide tech-
nical assistance and reference site biological data.
All of the Region III states have strong biological as-
sessment programs and are capable of meeting the
challenge for the expanded use of biological data.
14
-------
Biological Criteria: Research and Regulation, 1991
A State Perspective on Biological Criteria
in Regulation
Daniel R. Dudley
Ohio Environmental Protection Agency
Columbus, Ohio
ABSTRACT
The State of Ohio has been a leader in developing numerical biological criteria and exploring their use
in aquatic resource assessment. More recently Ohio has adopted state water quality standards which in-
extricably link aquatic life beneficial use designations and numerically expressed biological criteria.
With this foundation Ohio has shown that biological criteria incorporated into state regulations can
have a powerful impact on a wide range of water resource issues. This paper will highlight the role bio-
logical criteria have played in (1) section 401 water quality certifications for stream channel modifica-
tions; (2) the nonpoint source assessment process and the section 319 management plan for Ohio; (3)
stream use attainability issues; and (4) the NPDES permit system. Although Ohio has realized extensive
benefits through the biological criteria program, the development and application process has been con-
troversial. Ironically, the focus of biocriteria on water resource quality, as opposed to just water quality,
can, in some situations, pose a real dilemma for managers charged with both narrowly defined regula-
tory responsibilities and the obligation to carry out reasonable actions in the public's interest. For exam-
ple, our propensity to "solve" water quality problems on a cost-effective, regional basis has led to
dewatering and habitat alteration of small headwater streams in Ohio and elsewhere. Is this furthering
the goal of biological integrity? In other situations, small communities have spent large sums of money
to upgrade wastewater treatment to meet water quality standards on waterways that in subsequent
years may be subjected to routine "channel maintenance" (i.e., habitat destruction and loss of biological
integrity) associated with the petition ditch laws of many midwestern states. These are just two exam-
ples that illustrate the need for major re-drafting of much of our water resource legislation in this coun-
try if we are to realize the biological integrity goal of the Clean Water Act. For the environment, as well
as the public's sake, we must begin to better manage the whole of water resource quality. The present
programmatic emphasis on developing biological criteria affords us the opportunity to forcefully
re-examine and extend the focus of federal, state and local regulations from just water quality to water
resource quality.
Introduction
Ohio has developed a fairly extensive biological
survey program and biological criteria stream per-
formance standards, based on various attributes of
the resident fish and macroinvertebrate communi-
ties. Three separate numerical biological criteria
were incorporated into the Ohio administrative
code in a water quality standard rulemaking in
February 1990.
Before describing Ohio's regulatory experience
with biological criteria though, it is valuable to
focus on the fairly common perception of water pol-
lution control regulation as an impossible goal and
constantly moving target.
The same scenario plays out time and time
again. First technology standards are issued, then
revised downward to reflect best available technol-
ogy, new water quality standards are adopted, fol-
lowed by new or revised water quality based
permitting procedures—all of which results in new
15
-------
D. a DUDLEY
or more stringent permit limitations. When faced
with typically hostile reactions from the permittee
over this moving target and the need for investment
in pollution abatement, a common reply from the
regulatory official is that in keeping with the philos-
ophy behind the Clean Water Act regulators are
seeking incremental steps in pollution abatement
and—step by step—just racheting permit limits to-
ward zero discharge.
This cycle has recurred several times in Ohio,
and probably in most other states. In fact, this
racheting syndrome has become so pervasive in our
regulation that it is fair to compare the fate of the
permit holder to that of Sisyphus, a figure from
Greek Mythology. Sisyphus was the king of Corinth
and was noted for his' trickery and wicked nature.
He was condemned to forever push a huge stone up
a mountain in Hades only to have it roll down again
just as he almost reached the peak. The stone repre-
sents the permit, the top of the slope compliance.
A myth represents how people perceive the
world around them and their place in it. The regu-
lated community perceives itself in the role of,
Sisyphus. Scientists and technicians of the permit-
ting process don't necessarily believe or share that
perception. There are very complete and defensible
reasons for writing new and more stringent permit
limitations. However, technical explanations of the
new permit limitations will do little to change the
perception of the present situation by the regulated
community, and increasingly, some members of the
informed public.
Thus, it is important to shape positive percep-
tions of the permitting process and its goals in the
minds of the regulated community and the public,
because, ultimately, they shape the lawmaker's
opinion of our regulations. The mandate to regulate
was given by the public through the nation's elected
officials, and is subject to review and change. The
challenge of rational and understandable regulation
development can possibly be met through use of bi-
ological criteria in the regulatory process.
Five different examples of how biological cri-
teria have been applied in Ohio should make this
clear. These examples are: (1) water quality stan-
dard use designations; (2) section 319 - the non-
point source program; (3) section 401—water
quality certifications; (4) section 305(B) - water
quality inventory report; and (5) NPDES discharge
permits.
These examples contain a common characteris-
tic; in each, biocriteria provide a direct measure of
water quality, protecting the chemical, physical, and
biological integrity of our nation's waters. Such a
direct and objective assessment of the biological in-
tegrity goal fosters a positive perception of regula-
tion.
Stream Use Designations
Stream use designation is the longest running regu-
latory application of biological criteria. It began as
a system of tiered aquatic life use designations
launched in narrative language, and has evolved to
incorporate numerical biological criteria. The link-
ing of biological criteria to stream use designations
was particularly important to Ohio because the
State Water Quality Standard Rules in 1978 classi-
fied over 60 highly polluted stream segments as
less than fishable/swimmable uses. The task of re-
evaluating the potential aquatic life use designa-
tions for these and other streams was based upon
biological and water quality survey results and the
biological criteria.
Although some appeals are still pending, to
date the upgrade of all these stream segments to
fishable/swimmable uses has withstood legal chal-
lenges. Part of the credit for this must be given to
the strong theoretical and technical foundation pro-
vided by the biological criteria. It was clear to
Ohio's environmental Board of Review and the ap-
peal courts that the Ohio Environmental Protection
Agency had done its homework. More important,
water resource quality goals were clearly articulated
and measured, and. the goals were in fact attainable
given the installation of reasonably advanced treat-
ment technologies.
The Nonpoint Source Program
The state of Ohio has developed a nonpoint source
program consisting of two parts, an assessment
and a management program. The development of
the assessment program was coordinated by the
Ohio EPA, while the development of the manage-
ment program was coordinated by the Ohio De-
partment of Natural Resources. Both components
of the program have been approved by U.S. EPA.
Biological survey data and Ohio's biological cri-
teria for streams and rivers played a key role in the
assessment process. A total of 48 stream segments
were targeted for nonpoint source implementation
projects based upon aquatic life use impairment
that was documented in biosurvey results. This was
done by comparing the measured Index of Biotic In-
tegrity (IBI) values to the IBI criteria values for the
appropriate ecoregion specified in the Ohio water
quality standards. For the purpose of the assess-
ment, only IBI values that departed from ecoregion
standards by more than 10 percent were considered
significant. ,
16
-------
Biological Criteria: Research and Regulation, 1991
Streams and rivers with severe habitat prob-
lems caused by channel modification also were
setaside in this analysis because the impairment
problems were considered to be long term and in-
tractable through land application of best manage-
ment practices. Next, a cumulative score for the.
severity of pollution over the length of the stream
segment was calculated by graphing the measured
IBI values versus river mile and connecting the
points. The area below the biological criteria for IBI
was determined and used as a ranking score.
The end result was a list of 48 stream segments
with reasonably good habitat characteristics that
should respond in a relatively short time period to
accelerated implementation of best management
practices. The management program has given pri-
ority to funding section 319 projects in these water-
sheds. In summary, monies were targeted toward
stream segments that were selected based upon a
direct and objective measurement of extent and se-
verity of biological use impairment.
Finally, Ohio also established a special setaside
of 10 percent of section 319 project funds to be di-
rected toward the protection of threatened waters.
These were defined as waterbodies that were
known through the biological survey program to be
of high quality. These setaside funds help preserve
high quality waters in the face of mounting non-
point source pollution threats caused by' rapid
changes in land use patterns.
Section 401 Water Quality
Certifications
The placement of fill in navigable waters of the
United States is regulated through section 404 per-
mits issued by the Army Corps of Engineers. Pur-
suant to section 401, all permits for such activities
must include a water quality certification from the
appropriate state authority. The purpose of this
program is to ensure the permitted activities will
comply with state water quality standards.
For purposes of example, assume a proposed
404 permit is drafted to allow the placement of fill in
a stream channel and the relocation of several thou-
sand yards of stream bottom. Because Ohio has
adopted narrative and numerical biological criteria
associated with aquatic life use designations, sec-
tion 401 issues can be reviewed based on the tradi-
tional chemical water quality criteria, plus the
expected impact of the proposed channel work
upon the direct biological measurement of use at-
tainment. The state has been very successful in
achieving major adjustments in the scope and de-
sign of stream channel work, and when such ac-
commodations are not forthcoming, the Ohio EPA
has denied the section 401 Water Quality Certifica-
tion. •
The Section 305(B) Report
A biennial water quality report is required of each
state by the U.S. Environmental Protection Agency.
Whenever available, Ohio uses biocriteria as the
final arbitrator for determining the status of stream
use attainment or impairment. It is estimated that
50 percent of impaired waters would be rnisclassi-
fied as attaining Clean Water Act goals if biocriteria
were not available. Although this is unfortunate,
the protocol allows identification of problems so
reasonable corrective measures can be formulated.
The NPDES Permit Program
A direct use of biological criteria for permit limita-
tions or for specific design of wastewater treatment
in the point source program is not possible, and
given the present regulatory structure, such a use
should not really be expected. However, biosurvey
results and comparisons of stream performance as
measured by the biological criteria have been help-
ful in the permit reissuance process in Ohio. Some
of the typical applications and their advantages in-
clude:
1. Documentation of aquatic life use impair-
ment under present pollutant loading.
These data provide strong evidence that ih-
' vestment in pollution abatement is justified.
However, evidence of aquatic life use im-
pairment should not be a prerequisite for
the appropriate implementation of limits
based upon a reasonable expectation that
water quality standards may be violated.
2. Documentation of aquatic life impact, either
near by or far from the discharge point, can
help determine appropriate regulatory re-
sponse to whole effluent toxicity testing pro-
grams that are inconclusive regarding the
potential for a hazard to resident aquatic
life. Basecl on this data, the permittee might
be required to begin a toxicity identification
evaluation or toxicity reduction evaluation. '
3. Documenting aquatic life impact can detect
previously unknown or "under regulated"
sources. Nearly every biological and water
quality survey conducted by the Ohio EPA
17
-------
a a DUDLEY
has detected a problem previously un-
known to the paper world of permits and
compliance. Such impacts include the dis-
covery of contaminated nonprocess waters,
unreported episodes of chemical spillage,
generally poor housekeeping at the facility,
or nonchemicals at the point of discharge. In
urban areas bioassessment can also better
define the need for attention to combined
sewer overflows and stormwater controls.
The concept of a reasonable potential to violate
water quality standards is open to various interpre-
tations. When biological criteria are violated down-
stream from a point source, regulators need to react
quickly and with environmentally conservative
methods and assumptions to control the source.
However, when biological criteria are attained
downstream from a point source, .regulators need to
react in a slightly different manner if those same en-
vironmentally conservative methods predict possi-
ble water quality standards violations. That reaction
should consider the documented biological criteria ,
attainment as feedback that triggers a reassessment
of the modeling techniques that predicted the water
quality standard violations. These situations require
stepping up from simplistic mass balance models to,
more complex applications such as fate and trans-
port, dynamic, or probabilistic type modeling. The
outcome may still be the same—a prediction that
water quality standards have been violated-—or
have not. The point is that biological criteria were
used to help define what a reasonable showing is.
In summary, there is no doubt that the public
perception of water pollution regulation is becom-
ing increasingly important. Biocriteria can play a
valuable role in creating positive, reasonable, and
environmentally. protective approaches to water
pollution regulation.
18
-------
Biological Criteria: Research and Regulation, 1991
Utilization of Biological Information in
North Carolina's Water Quality Regulatory
Program
Jimmie Overton
North Carolina Department of Environment, Health
and Natural Resources
Raleigh, North Carolina
ABSTRACT
The utility and effectiveness of biological assessments in water quality programs is intrinsically
linked to the criteria and standards within a state's regulations. Realizing the scope and limita-
tions of this information, as well as seeking innovative and expanded uses, can enhance pro-
gram capabilities. Environmental surveys assess existing in-stream conditions and identify
source and extent of impairments to biological integrity. This information is transferred to the
regulatory process by having aquatic life standards (narrative or numerical) within state regu-
lations. North Carolina Water Quality Regulations .contain narrative standards for aquatic life
protection. Biological information is used in assigning appropriate designations for supple-
mental classifications including Nutrient Sensitive Waters, High Quality Water, Outstanding
Resource Waters, and Trout Waters. Development of biological assessment capabilities and reg-
ulatory use of information gathered should not be independent processes. Classifications and
standards can be written to use biological information, and choice of biological information
gathered and presented should be influenced by its regulatory use. Qualified ecologists with
strong interpretive capabilities are the most necessary element of good programs. Integrating
science and regulations is an ongoing challenge for state and federal programs.
Introduction
In-stream biota has been utilized to assess water
quality in North Carolina since the mid-1970s. Ini-
tially, work was conducted in free-flowing streams
to identify impaired waters of the State using colo-
nization of macroinvertebrates onto artificial sub-
strata. However, beginning in 1982, criteria in three
major ecoregions were developed to establish, rela-
tive degrees of impairment in flowing streams. Im-
pairments in lakes and large rivers due to cultural
eutrophication led to development of expertise in
limnology and phytoplankton ecology within the
North Carolina program. Assessments of these wa-
ters and changes in regulations provided specific
means of protecting waters from impairment due
to enrichment.
Work has begun in developing an environmen-
tal integrity index similar to Karr's IBI and based oh
fisheries community habitat and community struc-
ture (Karr, 1981). Development of this tool would
provide information from another major group of
biota, and should facilitate assessment of impacts
resulting from sedimentation and other nonpoint
source pollutants.
Refining biological criteria to account for all of
the variables resulting from natural and human
causes is a goal of many programs but has yet to be
completely achieved. Attempts to develop these cri-
teria have resulted in many indices, some useful,
but all with limitations. The presence of gross pollu-
tants is fairly easy to identify with any assessment
method. Identifying subtle impairments requires
19
-------
J. OVERTON
species/genus-level analyses involving the ecology
of specific organisms and groups of organisms.
Careful selection of baseline sites and subse-
quent acquisition of baseline information is impera-
tive when dealing with the inherent variability of
stream size, as well as seasonal and regional differ-
ences among samples. Selection of defensible, re-
producible, and efficient sampling methodology
and analyses by competent, well-trained ecologists
provide the basis for acquiring assessments that
meet the immediate needs of the program, as well
as long-term needs such as development of biocrite-
ria. Biological information must be integrated with
all phases of water quality assessment, including
chemical, physical, and toxicological information.
The use of all information must be reflected and de-
fined in the state regulations to maximize its utility.
Biological Information and
Standardized Methods
Information derived through biological monitoring
is useful only if it is collected and analyzed using
standard scientific methods. These methods must
be adequately tested for reproducibility indepen-
dent of the collector, provide a good census of
stream community, and produce information that
can be related to water quality. Method selection
also requires consideration of program needs and
available resources. Statistical indices and other an-
alytical tools are in constant review and develop-
ment. Therefore the data base generated should be
consistent, additive, and flexible to be utilized for
immediate and long-term needs.
The North Carolina Program
These requirements faced macroinvertebrate ecolo-,
gists in North Carolina when they were employed
to assess impacts to the aquatic community in free-,
flowing streams. Initial sampling was conducted
with the U.S. Environmental Protection Agency-ac-
cepted methodology, using artificial substrata. Re-
sults were quantitative and amenable to statistical
analyses, but several problems were identified with
this methodology. The substrata required multiple
visits, were often missing because of high flow or
vandalism, were habitat specific, and were costly in
time, money, and technician attrition.
A standard qualitative sampling technique was,
developed for wadable streams in North Carolina
(Lenat, 1988). The method samples all available
habitats; samples are processed in the field; one visit
to the site is required; and a good census of the
macroinvertebrate community is taken. Most im-
portantly, this community changes consistently
with variations in water quality. Several years were
required to address within-rnethod variability, in-
cluding level of effort, collector, and site selection.
Natural variability, including regional differences,,
watershed size, and seasonality continue to be ana-
lyzed with a data set of over 1,500 collections.
Phytoplankton ecologists also selected sam-
pling and analytical methods best suited to pro-
gram needs, which are capable of initiating and
maintaining a quality data base. Logistical consider-
ations mandated preservation of samples. An in-
verted microscope technique with species-level
taxonomy and measured biovolumes produced es-
timates of density and biovolume for each sample,
as well as each species within the sample. Resulting
information (over 5,000 samples) has been used to
identify nutrient-sensitive surface waters in North
Carolina requiring additional management strate-
gies. '
Program review suggested the need for pro-
gram expansion. Multiple community components
and a variety of metrics enhance the capability of
evaluating biological integrity in surface waters.
Work has begun on method development of a
North Carolina biotic index relative to fisheries
community structure and habitat. Approximately
22 baseline stations will be used to test the method
arid metrics used for analysis. Acquisition of quali-
fied ecologists and selection of standard methodol-
ogy are the first steps toward providing biological
information useful to a regulatory program. Devel-
opment of criteria is dependent on the consistency
with which these tools measure water quality im-
pairment.
Statutory Authority
Biological information is an assessment tool. Im-
pacts to biological integrity should reflect some
measurable deviation from baseline or reference
conditions. Establishing and maintaining a net-
work of baseline stations capable of accounting for
seasonal and regional variability is ideal, but often
not feasible for each survey. In those instances up-
stream or paired watershed reference collections
maybe used.
The measurement of biological impairment as a
part of the regulatory process is driven by the sig-
nificance of that information relative to defined use.
North Carolina adjusted its water quality regula-
tions to take the degree of biological impairment
into account (N. Carolina Dep. Environ. Health Nat.
Resour. 1990). Code revisions included the anti-
20
-------
Biological Criteria: Research and Regulation, f&fft
degradation statement, supplemental classifications
of High Quality Waters and Outstanding Resource
Waters, and inclusion of protection for aquatic life
propagation and maintenance in all best-use defini-
tions. Changes also included addition of whole ef-
fluent toxicity standards and an addendum to the
chlorophyll a standard that provides authority to
prohibit or limit any discharge of waste into surface
waters that would result in violations caused by nu-
trient enrichment. A brief overview of these adjust-
ments follows.
• Antidegradation Policy. The policy includes a
statement of purpose in maintaining, protecting,
and enhancing water quality within North Caro-
lina. Existing uses shall be protected by properly
classifying waters and having standards sufficient
to protect these uses. The policy also includes a def-
inition for High Quality Waters (HQW) and spe-
cific protection for waters with quality higher than
the standards to prevent degradation below the
quality necessary to maintain existing and antici-
pated uses,' including uses not specified by the as-
signed classification. HQW are identified by the
Division on a case-by-case basis. Use attainability
analyses are conducted to identify waters with ex-
cellent water quality as. determined by chemical
and biological information. Outstanding Resource
Waters (ORW) are a special subset of High Quality
Waters with unique and special characteristics.
Classification and protection of ORW will be dis-
cussed with other supplemental classifications.
Specific point and nonpoint protection measures
include:
• All new National Pollutant Discharge Elimi-
nation Systems (NPDES)- wastewater dis-
charges will have effluent limitations of BODS
= 5 mg/L, NH3-N = 2 mg/L and DO = 6
mg/ L. More stringent limits may be set, if nec-
essary to ensure that the discharge will not re-
sult in a drop of more than .5 mg/L of DO in
the receiving waters below background levels.
The total volume of treated wastewater for all
discharges combined will not exceed 50 per-
cent of the total in-stream flow under 7Q10
conditions. In general only the discharge of
domestic or nonprocess wastewater will be
permitted into HQW. Whole effluent toxicity
is allocated to protect for chronic toxicity at an
effluent concentration equal to twice that
which is acceptable under design conditions.
• All expanded NPDES wastewater discharges
will be required to meet the treatment de-
scribed, except for those existing discharges
that expand with no increase in permitted pol-
lutant loading.
• Development activities that drain to and are
, within one mile of HQW are required to con-
trol runoff from a 1-inch design storm. Two
options are provided related to density of de-
velopment. The "low density option" for de-
velopment limits single-family residences to
one acre or larger lots and other, developments
to 12 percent. This option does not require
stormwater collection systems but built-upon
areas must be at least 30 feet from surface wa-
ters. The "high density option" requires
stormwater control systems utilizing wet de-
tention ponds. These systems must be in-
stalled, operated, and maintained to control
runoff generated from 1-inch of rainfall from
all developed areas. Systems must be sized to
control runoff from all pervious Surfaces
draining to them. More stringent requirements
, , may be required on a case.-by-case basis with
either option.
• Fresh Surface Waters Classifications and Stan-
dards. Changes were made in "best usage" require-
ments for all fresh surface waters allowing use of
ecological surveys in establishing protection strate^
gies. Conditions related to best usage include, suit-
ability for aquatic life propagation and main-
tenance and restrictions on sources of water pollu-
tion that preclude any of these uses on either a
short-term or long-term basis. The chlorophyll a
standard now prohibits or limits discharge of waste
into surface waters if so doing would exacerbate or
result in growths of microscopic or macroscopic
vegetation that would violate standards or impair
best usage.
• Tidal Salt Water Classifications and Stan-
dards. Standards for these waters use language
similar to that for fresh waters. In, addition, added
protection is included for shellfishing waters, re-
quiring resource and water quality necessary to
provide shellfishing for market purposes. No sew-
age is allowed in shellfishing waters, which are by
definition included as High Quality Waters. All wa-
ters in the 20 coastal counties have requirements
for nonpoint source controls similar to those re-
quired for protection of HQW.
• Standards for Toxic Substances and Tempera-
ture. Aquatic life standards limit the concentration
of toxic substances to less than that which would
21
-------
J. OVERTON
sessment of fisheries integrity through community
structure analyses. .... '•' . >i'
Summary, and the Future
North Carolina currently incorporates ecological
information throughout its water quality program.
Fiscal realities dictate that expanded use within the
program will probably be through increased effi-
ciency in data analysis, criteria development, and
basin management. All NPDES permits within a
given river basin are scheduled to be issued within
the same year. Plans are well underway to adjust
the ambient chemical and physical monitoring net-
work to produce information necessary for model
needs. Intensive surveys including time of travel
studies, long-term BOD, and sediment oxygen.de-
mands have also changed dramatically in scope to
address basin management. Benthic macroinverte-
brate surveys, limnological work, and other ecolog-
ical tools will shift spatial emphasis to provide
assessment throughout basins prior to manage-
ment plans and permit decisions. Conducting this
work while meeting the other needs of the pro-
gram, continued development of analytical tools,
and collection of baseline information to address
natural variability provide a significant challenge.
The single largest need for state programs in-
volves data management. Accessing digitized data
layers in a usable format is critical to efficient prepa-
ration of study plans that effectively analyze results
to include cumulative impacts and can retrieve re-
sultant information in formats appropriate to meet
state arid federal needs. Accessible geographic data
layers must become a reality for maximum effi-
ciency in state programs. • ':''.':•••
' Biological information is obviously of greater
utility to a regulatory agency if numerical criteria
are used directly or indirectly in regulations; Bene-
fits .exist for both approaches. Administratively, nu-
merical biological criteria simplify interpretation
and enforcement. However, natural variability >in
aquatic environments has frustrated most attempts,
to standardize regulation with a numerical index.
Well-constructed narrative criteria are not necessar-
ily a perfect solution, but can provide immediate
protection to surface waters. Numerical biological
criteria, however used, will need to be modified
with new information; caution should be taken that
administrative simplicity does not circumvent the
ability of qualified ecologists to assess changing sit-
• uations and recommend effective action to protect
water quality.
References
Karr, J.R. 1981. Assessment of biotic integrity using fish com-
munities. Fisheries 6:21-27.
Lenat, D. L. 1988. Water quality assessment of streams using
'a qualitative "collection method- for benthic
macroinvertebrates. J. N. Am. Benthol. Soc. 7(3):222-33.
North Carolina Department of Environment, Health and
Natural Resources. 1990. Admin. Code Sec. ISA NCAC
2B .0100, Procedures for Assignment of Water Quality
Standards and ISA NCAC 2B .0200 Classification and
. Water Quality Standards Applicable to Surface Waters
of North Carolina. Raleigh. ,
24
-------
Biological Criteria: Research and Regulation, 1991
Use of Habitat Assessment in Evaluating
the Biological Integrity of Stream
Communities
Michael T. Barbour
James B. Stribling
EA Engineering, Science, and Technology, Inc.
Sparks, Maryland
ABSTRACT
An evaluation of habitat quality is critical to any assessment of ecological integrity (Plafkin et al.
1989). For streams, a holistic approach to assessing habitat quality would include an evaluation of va-
riety and quality of substrate, channel morphology, bank structure, and riparian vegetation. Biologi-
cal potential is limited by the quality of the habitat. Three general relationships between habitat"
quality and biological condition can be expected: (1) a direct response of the biological community to
variation in the habitat quality in the absence of water quality problems; (2) a degradation of the bio-
logical community greater than habitat quality would predict, when combined with toxicant or or-
ganic pollution loadings to the stream; and (3) an artificial elevation of the biological condition
beyond that predicted by habitat quality due to organic enrichment. Studies from different areas of
the United States have shown that a knowledge of the habitat quality has enhanced an assessment of
biological impairment caused by water quality problems. The establishment of parameters delimit-
ing the relationship between habitat quality and biological integrity improves the ability to set envi-
ronmental goals and evaluate program results. This relationship between habitat quality and,
biological integrity may vary among physiographic regions or ecoregions, but is determined by ref-
erence databases. Once confidence limits have been established, the reference database can be moni-
tored to adjust for changes in habitat quality or the condition of the biological communities.
Habitat Quality and Biological
Condition
Habitat assessment supports understanding of the
relationship between habitat quality and biological
conditions. Such assessment identifies, obvious
constraints on the attainable potential of the site,
assists in the selection of appropriate sampling sta-
tions, and provides basic information for interpre-
ting biosurvey results.
Assessment of biological potential is recom-
mended as a correlate of community analysis. Vari-
ability of environmental conditions directly affects
patterns of life, population, and micro- and
macrogeographic distribution of organisms (Price,
1975; Smith, 1974; Cooper, 1984). Physical habitat
quality is a major factor influencing the biological
condition of aquatic communities. Bioassessment
procedures such as the Rapid Bioassessment Proto-
cols (RBPs) (Plafkin et al. 1989) stress the impor-
tance of this variable as a major determinant of the
biological potential of a particular habitat. This po-
tential relates to the structure and composition of
the biota and must be recognized before habitat
evaluations can be made.
To estimate biological potential, reference con-
ditions are used to identify parameters of the assess-
ment. An understanding of the expected charac-
teristics or conditions is inherent in the judgment of
impairment or degradation.
25
-------
M. T. BARBOUR and J. B. STRIBLING
For most surface waters, baseline data were not
collected prior to an impact; thus, impairment must
be inferred from differences between the impact site
and established references (U.S. Environ. Prot.
Agency, 1990). This approach is also critical to the
assessment because stream characteristics will vary
dramatically across different regions (Plafkin et al.
1989). Furthermore, wide variability among streams
and rivers across the country resulting from cli-
matic, landform, and other geographic differences
prevents the development of nationwide reference
conditions (U.S. Environ. Prot. Agency, 1990): A
range of parameters representing "best attainable"
condition in terms of habitat quality and aquatic
communities will be developed for ecosystems with
similar physical and chemical dimensions, individ-
ual watersheds, or individual streams. A decision as
to which of these reference points will be accepted
as the attainable standard for the ecoregion is cru-
cial to the bioassessment process.
Assuming that water quality remains constant,
the predictable relationship between habitat quality
and biological condition can be a sigmoid curve, as
illustrated in Figure 1. On the x-axis, habitat is
shown to vary from poor to optimal, relative to the
reference conditions. Therefore, the quality of the
habitat can range from zero to 100 percent of the ref-
erence, and can be categorized as nonsupporting,
partially supporting, supporting, or comparable, re-
ferring to the support of well-balanced biological
communities.
The curve is divided into three parts. The first,
or upper right hand corner, reflects a situation with
good habitat quality and good biological condition.
Some variability in habitat quality is possible
without affecting the condition of the biological
communities. As the habitat quality decreases
within some range of "good to excellent," the bio-
logical condition will remain high, and subtle differ-
ences will be difficult to detect. However, in the
second, or midseetional part of the curve, the de-
crease in biological condition is proportional to a
decrease in habitat quality. This situation occurs
when habitat quality decreases, and the biological
community responds with a concomitant decrease.
In the lower left hand section of the curve, habitat
quality is poor, and further degradation may result
in relatively little difference in biological condition.
Communities in this region of the curve are pollu-
tion tolerant, opportunistic, thrive in areas 6f re-
"'•' • ' ',„"""<:
Partially I ' Corn-
Supporting ' Supporting parable
40 50 60 70
Habitat Quality (% of Reference)
Figure 1.—The relationship between habitat and biological condition.
90
100
-------
Biological Criteria: Research and Regulation, 1991
duced competition, and are able to withstand
highly variable conditions.
Following the establishment of reference condi-
tions for determination of water quality, habitat as-
sessment that accounts for the various habitat
parameters influencing the structure and, function
of communities should be conducted to enhance the
interpretation of biological data (Plafkin et al. 1989).
The actual orientation of the relationship line be-
tween habitat quality and biological condition is not
fixed and may differ in the degree of linearity, slope,
and y-intercept, depending on the physiographic re-
gion of the country. The collection of substantial ref-
erence data would allow for the development of this
empirical line along with statistical parameters.
From this information, the expected biological rela-
tionship can be determined from a known range of
habitat quality conditions (Fig. 2). In this manner,
estimates of water quality effects beyond those ex-
pected from habitat constraints are possible.
As depicted in Figure 2, three general outcomes
are possible when comparing ambient stream sta-
tions to a reference: (1) no biological effects, or ef-
fects due to habitat degradation; (2) effects due to
water quality; or (3) an artificial elevation of the per-
ceived condition of the community beyond the ex-
pected relationship because of mild enrichment
effects. A fourth outcome is where it is not possible
to separate the combined effects of habitat and
water quality degradation.
Biological
Condition
Poor Habitat Quality Good
Figure 2.—Combined influence of habitat and water quality
on biological condition.
The accurate determination of these possible
outcomes is supported by a reference database ade-
quate to defining the expected relationship between
habitat quality and biological integrity. The theoreti-
cal regression line between habitat quality and bio-
logical condition should be substantiated with a
larger database than is currently available. To date,
habitat assessment results are not available in the
historical database, with the possible exception of
the U.S. Forest Service and the Ohio Environmental
Protection Agency. Data analysis should be con-
ducted to produce guidance on data variability ex-
pectations and the slope of the line to be used for
predictions.
Establishing the reduction of habitat quality
may be all that is needed to judge impairment. The
quantification of habitat quality may be as impor-
tant as measuring instream communities in docu-
menting nonpoint source impact. Guidance for this
type of definitive assessment needs to be devel-
oped. The following discussion provides guidance
for establishing a minimum level habitat assess-
ment.
Habitat Parameters
The habitat parameters designed to assess habitat
quality are separated into three main categories:
primary, secondary, and tertiary parameters. Pri-
mary parameters are those that characterize the
stream "microscale" or specific niche habitats and
have the greatest direct influence on the structure
of the indigenous communities (Plafkin et al. 1989).
The secondary parameters measure the "macro-
scale" habitat such as channel morphology charac-
teristics. Tertiary parameters evaluate riparian and
bank structure, features often ignored in
biosurveys. These three categories are weighted ac-
cording to their influence on the biota, with pri-
mary parameters having more weight than
secondary or tertiary characteristics.
Although the streams across the country exhibit
a wide range in variability, generalizations can be
made about the types and similarities. The gradient
of the streams is perhaps the most influential factor
in categorizing a water body, because it is related to
topography and landform, geological formations,
and elevation, which in turn influence vegetation
types. Four generic stream categories related to gra-
dient can be identified: mountain, piedmont, val-
ley/plains, and coastal. From these four categories,
two sets of habitat parameters can be developed to
conduct a holistic habitat assessment; these are
roughly equivalent to evaluation of high gradient
(riffle/run prevalence) and low gradient streams
(glide/pool prevalence).
These two categorical approaches are intended
to provide guidance in assessing habitat quality of
two very different stream/river types based on gra-
dient. Further subsets are possible, depending on
regional specifications. However, the evaluation of
habitat quality takes into consideration reference
conditions that will automatically adjust for some
regional differences. A mountain trout stream
should not be used as a benchmark for a lowland
27
-------
M. T. BARBOUR and J. B. STRIBLING
plains stream. Habitat parameters, which have been
selected to support the assessment approach for the
two general stream type categories, include pri-
mary, secondary, and tertiary characteristics. These
are:
RIFFLE/RUN PREVALENCE
PRIMARY — Substrate, Instream Cover, and Canopy
1. Substrate variety/ instream cover
2. Embeddedness
3. Flow or velocity/depth
4. Canopy cover (shading)
SECONDARY— Channel Morphology
5. Channel alteration
6. Bottom scouring and deposition
7. Pool/riffle, run/bend ratio
8. Lower bank channel capacity
TERTIARY—Riparian and Bank Structure
9. Upper bank stability
10. Bank vegetative stability (grazing/disruptive
pressure)
11. Streamside cover
12. Riparian vegetative zone width
GLIDE/POOL PREVALENCE
PRIMARY — Substrate, Instream Cover, and Canopy
1. Substrate variety/instream cover
2. Bottom substrate characterization
3. Pool variability
4. Canopy cover (shading)
SECONDARY— Channel Morphology
5. Channel alteration
6. Deposition
7. Channel sinuosity
8. Lower bank channel capacity
TERTIARY — Riparian and Bank Structure
9. Upper bank stability
10. Bank vegetative stability (grazing/disruptive
pressure)
11. Streamside cover
12. Riparian vegetative zone width
The categorical approach is intended to provide
a refined framework for increased accuracy in, and
applicability of, habitat assessment. The main dif-
ferences between these two habitat assessment ma-
trices are found in the primary parameter grouping.
These parameters relate directly to the specific niche
characteristics and will need to be altered depend-
ing on the stream type being evaluated. The second-
ary parameters differ only slightly in the specific
parameter characteristics between the two matrices,
and the tertiary parameters are identical. Some
modification of the decision criteria might be useful
to refine for regional purposes.
The Matrices
The original habitat assessment matrix presented
by Plafkin et al. (1989) is based on Ball (1982) and
Platts et al. (1983). Although these still make up the
primary foundation for the RBP habitat assessment
matrix, additional sources provide information for
refinement of the habitat assessment approach.
Habitat parameters are categorized as primary,
secondary, or tertiary, relating to the degree of influ-
ence exerted on the biological community. Through
field observation and measurement, scores are as-
signed to each parameter ranging from 0 (poor) to
20 (excellent) for primary, 0 to 15 for secondary, and
0 to 10 for tertiary parameters. A more complete ex-
planation of the scoring procedures for performing
a habitat assessment is provided in Plafkin et al.
(1989). A description of the parameters for the two
categorical approaches is presented in the following
section.
Riffle/Run Prevalence
The first matrix (Fig. 3) is similar to the original de-
scribed in Plafkin et al. (1989), but has been modi-
fied to be more appropriate for wadable streams
and rivers having a prevalence of riffles and runs.
These primary habitat parameters are weighted the
highest to reflect their degree of importance to the
biological community. The primary parameters re-
late to substrate and instream characteristics and
are:
• 1. Bottom substrate/instream cover. This char-
acteristic refers to the availability of habitat for the
support of aquatic organisms and cover for nest-
ing, oviposition sites, or avoidance behavior. A va-
riety of substrate materials and habitat types is
desirable (U.S. Environ. Prot. Agency, 1983; Ball,
1982; Hamilton and Bergersen, 1984). The presence
of broad variability in particle size of a rock/gravel
substrate is considered to be the optimal habitat for
benthic macroinvertebrate communities. However,
instream materials such as logs and snags, tree
roots, submerged and emergent vegetation, and
28
-------
Biological Criteria: Research and Regulation, 1991
HABITAT ASSESSMENT FIELD DATA SHEET
RIFFLE/RUN PREVALENCE
Category
Habitat Parameter
Optima!
Sub-Optimal
Marginal
Poor
1. Bottom substrate/
, instream cover (a)
Greater than 50% mix
of rubble, gravel,
submerged logs,
undercut banks, or other
stable habitat.
16-20
30-50% mix of rubble,
gravel, or other stable
habitat. Adequate
habitat.
11-15
10-30% mix of rubble,
gravel, or other stable
habitat. Habitat
availability less than
desirable.
6-10
Less than 10% rubble,
gravel, or other stable
habitat. Lack of habitat
is obvious.
0-5
2. Embeddedness (b)
Gravel, cobble, and
boulder particles are
between 0-25%
surrounded by fine
sediment.
Gravel, cobble, and
boulder particles are
between 25-50%
surrounded by fine
sediment.
Gravel, cobble, and
boulder particles are
between 50-75%
surrounded by fine
sediment.
Gravel, cobble, and
boulder particles are
over 75% surrounded
by fine sediment.
16-20
11-15
6-10
0-5
3. =s0.1 5 cms (5 cfsH
Flow at rep. low
Cold >0.05 cms (2 cfs)
Warm >0.15 cms
(5 cfs)
0.03-0.05 cms
(1-2 cfs)
0.05-0.15 cms
(2-5 cfs)
0.01-0.03 cms
(.5-1 cfs)
0.03-0.05 cms (1-cfs)
<0.01 cms (.5 cfs)
<0.03 cms (1 cfs)
OR
>0.15 cms
(5 cfsH>
velocity/depth
16-20
Slow (<0.3 m/s), deep
(>0.5 m): slow, shallow
(<0.5 m); fast (>0.3
m/s), deep; fast, shallow
habitats all present.
16-20
11-15
Only 3 of the 4 habitat
categories present
(missing riffles or runs
receive lower score than
missing pools).
11-15
6-10
Only 2 of the 4 habitat
categories present
(missing riffles or runs
receive lower score).
6-10
0-5
Dominated by 1
velocity/depth category
(usually pools).
0-5
4. Canopy cover
(shading) (c) (d) (g)
A mixture of conditions
where some areas of
water surface fully
exposed to sunlight, and
other receiving various
degrees of filtered light.
16-20
Covered by sparse
canopy; entire water
surface receiving filtered
light.
11-15
Completely covered by
dense canopy; water
surface completely
shaded OR nearly full
sunlight reaching water
surface. Shading limited
to <3 hours per day.
6-10
Lack of canopy, full
sunlight reaching water
surface.
0-5
5. Channel alteration
(a)
Little or no enlargement
of islands or point bars,
and/or no
channelization.
12-15
Some new increase in
bar formation, mostly
from coarse gravel; and/
or some channelization
present.
8-11
Moderate deposition of
new gravel, coarse sand
on old and new bars;
and/or embankments on
both banks.
4-7
Heavy deposits of fine
material, increased bar
development; and/or
extensive
channelization.
0-3
6. Bottom scouring and
deposition (a)
Less than 5% of the
bottom affected by
scouring and/or
deposition.
5-30% affected. Scour
at constrictions and
where grades steepen.
Some deposition in
pools.
12-15 8-11
30-50% affected.
Deposits and/or scour at
obstructions,
constrictions, and
bends. Filling of pools
prevalent.
4-7
More than 50% of the
bottom changing
frequently. Pools almost
absent due to
deposition. Only large
rocks in riffle exposed.
0-3
Pool/riffle, run/bend
ratio (a) (distance
between riffles
divided by stream
width)
Ratio: 5-7. Variety of
habitat. Repeat pattern
of sequence relatively
frequent.
12-15
7-15. Infrequent repeat
pattern. Variety of
macrohabitat less than
optimal.
15-25. Occasional riffle
or bend. Bottom
contours provide some
habitat.
4-7
>25. Essentially a
straight stream;
Generally all flat water
or shallow riffle. Poor
habitat.
0-3
8. Lower bank channel
capacity (b)
Overbank (lower) flows
rare. Lower bank W/D
ratio <7. (Channel width
divided by depth or
height of lower bank.)
12-15
Overbank (lower) flows
occasional. W/D ratio
8-15.
8-11
Overbank (lower) flows
common. W/D ratio
15-25.
4-7
Peak flows not
contained or contained
through channelization.
W/D ratio >25.
0-3
Figure 3.—Habitat assessment field data sheets for riffle/run prevalent situations.
29
-------
M. T. BARBOUR and J. B. STRIBLING
Category
Habitat Parameter
9. Upper bank stability
(a)
Optimal
Upper bank stable. No
evidence of erosion or
bank failure. Side
slopes generally <3Q°.
Little potential for future
problems.
9-10
Sub-Optimal
Moderately stable.
Infrequent, small areas
of erosion mostly healed
over. Side slopes up to
40° on one bank. Slight
potential in extreme
floods.
6-8
Marginal
Moderately unstable.
Moderate frequency and
size of erosional areas.
Side slopes up to 60°
on some banks. High
erosion potential during
extreme high flow.
3-5
Poor
Unstable. Many eroded
areas. "Raw" areas
frequent along straight
sections and bends.
Side slopes >60°
common.
0-2
10. Bank vegetative
protection (d)
OR
Grazing or other
disruptive pressure
(b)
Over 90% of the
streambank surfaces
covered by vegetation.
9-10
Vegetative disruption
minimal or not evident.
Almost all potential plant
biomass at present
stage of development
remains.
9-10
70-89% of the
streambank surfaces
covered by vegetation.
6-8
Disruption evident but
not affecting community
vigor. Vegetative use is
moderate, and at least
one-half of the potential
plant biomass remains.
6-8
50-79% of the
streambank surfaces
covered by vegetation.
3-5
Disruption obvious;
some patches of bare
soil or closely cropped
vegetation present. Less
than one-half of the
potential plant biomass
remains.
3-5
Less than 50% of the
streambank surfaces
covered by vegetation.
0-2
Disruption of
streambank vegetation
is very high. Vegetation
has been removed to 2
inches or less in
average stubble height.
0-2
11. Streamslde cover
(b)
Dominant vegetation is
shrub.
Dominant vegetation is
of tree form.
Dominant vegetation is
grass or forbes.
9-10
6-8
3-5
Over 50% of the
streambank has no
vegetation and
dominant material is
soil, rock, bridge
materials, culverts, or
mine tailings.
0-2
12. Riparian vegetative
zone width (least
buffered side) (e) (f)
(9)
>18 meters.
Between 12 and 18
meters.
Between 6 and 12
meters.
<6 meters.
9-10
6-8
3-5
0-2
Column Totals
Score.
(a) From Ball 1982.
(b) From Plattsetal. 1983.
(c) From EPA 1983.
(d) From Hamilton and Bergersen 1984.
(e) From Lafferty 1987.
(f) From Schueler 1987.
(g) From Bartholow 1989.
Figure 3.—Habitat assessment field data sheets for riffle/run prevalent situations (continued).
undercut banks provide exceptional habitat for a
diversity of organisms, particularly fish. This pa-
rameter is evaluated by visual observation.
• 2. Embeddedness. This rating is a consideration
of how much of the surface area of the larger sub-
strate particles are surrounded by fine sediment
(Plaits et al. 1983). This parameter should allow
evaluation of the substrate as a habitat for benthic
macroinvertebrates and for fish spawning and egg
incubation. Higher levels of embeddedness are
thought to correlate with lower biotic productivity.
Two aspects of embeddedness are of concern: (a)
the degree to which the primary substrate, i.e., cob-
ble, rubble, is buried in the finer sediments, and (b)
the covering of the cobble with a layer of silt or or-
ganic floe. Both aspects of embeddedness will elim-
inate niche space and reduce attachment viability.
Heavy silting (resulting in embeddedness) is
known to cause a reduction in insect species diver-
sity and productivity (Minshall, 1984). The degree
of embeddedness can vary, depending on whether
the riffle, run, or pool is being rated. Emphasis
should be placed on the sampling area, which will
normally be the riffle or run.
• 3. Stream flow and/or stream velocity. The size
of the stream or river will influence the structure
30
-------
Biological Criteria: Research and Regulation, 1991
and function of the aquatic communities. This hab-
itat parameter rates the quality of the stream size
with respect to the amount of water in small
streams and the variety of velocity/depth regimes
in larger streams and rivers. A particular water-
body being rated must be assigned one of these
two conditions before a rating can be ascertained.
The flow parameter (water quantity) indicates the
ability of a stream to produce and maintain a stable
environment in the substrate (Ball, 1982). This pa-
rameter is most critical to the support of aquatic
communities when the representative low flow is &
0.15 cubic meters per second (cms) (=5 cubic feet a
second [cfs]). The evaluation is based on flow
rather than velocity, since in small streams flow is
the predominating constraint. In these small
streams, flow should be estimated in a straight
stretch of run area where banks are parallel and
bottom contour is relatively flat.
In larger streams and rivers, i.e., those > 0.15
cms, velocity and depth are more important to the
maintenance of aquatic communities (Osborne and
Herricks, 1983; Oswood and Barber, 1982). The
quality of the aquatic habitat can therefore be evalu-
ated in terms of a velocity and depth relationship.
Both factors are used to establish this parameter,
with values patterned after Oswood and Barber
(1982). Four general categories of velocity, and
depth are optimal for benthic and fish communities:
(1) slow (< 0.3 m/s), shallow (< 0.5 m); (2) slow (<
0.3 m/s), deep (> 0.5 m); (3) fast (> 0.3 m/s), deep (>
0.5 m); and (4) fast (> 0.3 m/s), shallow (< 0.5 m)
(Oswood and Barber, 1982). Habitat quality is re-
duced in the absence of one or more of these four
categories. Characteristics of water current make up
the major determining factors of substrate quality
and, by implication, the structure and composition
of benthic communities (Minshall, 1984).
• 4. Canopy cover (shading). The shading aspect
provided by canopy cover is important in consider-
ation of water temperature and its effect on biologi-
cal processes in general, and as a mediating factor
in the solar energy available for photosynthetic ac-
tivity and primary production (U.S. Environ. Prot.
Agency, 1983; Bartholow, 1989; Platts et al. 1983).
Diversity of shade conditions is considered opti-
mal, with different areas of the sampling station re-
ceiving direct sunlight, complete shade, and
filtered light.
The secondary parameters are weighted less
heavily than the primary, with a maximum score of
15 points. These characteristics relate directly to
channel morphology or macrohabitat features.
Local geological features, including soil character
and human activities (Platts et al. 1983) influence
these parameters. The sediment movement along
the channel, as influenced by the tractive forces of
flowing water and the sinuosity of the channel, also
affects habitat conditions. The secondary character-
istics are as follows:
• 5. Channel alteration. Sediment deposits trans-
ported from upstream and forming bars are an in-
dication of watershed erosion or other more acute
disturbances. This characteristic potentially allows
crude estimation of stream system stability (Platts
et al. 1983) and relates primarily to above-water de-
posits. Channelization involves reduction in sinu-
osity and results in increased velocity and
subsequent intensification of erosional effects (U.S.
Environ. Prot. Agency, 1983; Plafkin et al. 1989;
Newbury, 1984; Schueler, 1987). Channel alteration
may be caused by dredging activities, cementing or
rip-rapping of banks, or natural watershed erosion.
Channel alteration also results in deposition, which
may occur on the inside of bends, below channel
constrictions, and where stream gradient flattens
out (Plafkin et al. 1989).
• 6. Bottom scouring and deposition. This pa-
rameter specifically targets disruption of instream
habitat as a result of the channel altering factors
discussed. With increases in velocity, there is more
likelihood for scouring and streambed erosion.
Also, scouring tends to occur during periods of in-
creased discharge (floods), and these same areas of
scour are refilled by material from further up-
stream (Hynes, 1970). The potential for scouring is
increased by channelization. Characteristics to ob-
serve are scoured substrate and the degree of silt-
ation in pools and riffles.
Deposition of sediment from large-scale water-
shed erosion can smother the instream habitat. The
degree to which pools and run areas are filled with
silt is an indication of the severity of this parameter.
Deposition will also affect embeddedness (a pri-
mary habitat parameter). However, deposition is
rated on a macrohabitat scale, and the focus is on
the filling of runs and pools.
Evaluation of this parameter is by estimation of
percentage of substrate affected by scouring or de-
position; this should be evident from the observed
degree of substrate stability in the reach of the
stream being evaluated. The deposition and scour-
ing parameter is rated by estimating the percentage
of an evaluated reach that is scoured or silted (i.e.,
50-m silted in a 100-m stream length equals 50 per-
cent).
31
-------
r
M. T. BARBOUR and J. B. STRIBLING
HABITAT ASSESSMENT FIELD DATA SHEET
GLIDE/POOL PREVALENCE
Habitat Parameter
1. Bottom substrate/
instream cover (a)
Optimal
Greater than 50% mix
of rubble, gravel,
submerged logs,
undercut banks, or other
stable habitat.
16-20
Category
Sub-Optimal Marginal Poor
30-50% mix of rubble, 10-30% mix of
gravel, or other stable gravel, or other
habitat. Adequate habitat. Habitat
habitat. availability less
desirable. •
11-15
rubble, Less than 10% rubble,
stable gravel, or other stable
habitat. Lack of habitat
than is obvious.
6-10 0-5
2. Pool substrate
characterization (c)
Mixture of substrate
materials with gravel
and firm sand prevalent;
root mats and
submerged vegetation
common.
16-20
mud, or clay; mud may
be dominant; some root
mats and submerged
vegetation present.
11-15
channelized with sand
bottom; little or no root
mat, no submerged
vegetation.
6-10
bedrock; no root mat or
vegetation.
0-5
3. Pool variability (b) (c)
Even mix of deep/
shallow/large/small
pools present.
16-20
Majority of pools large
and deep; very few
shallow.
11-15
Shallow pools much
more prevalent than
deep pools.
6-10
Majority of poo s small
and shallow or pools
absent.
0-5
I ar>k nf rnnnnv full
4. Canopy cover
(shading) (c) (d) (g)
A mixture of conditions
where some areas of
water surface fully
exposed to sunlight, and
other receiving various
degrees of filtered light.
16-20
canopy; entire water
surface receiving filtered
light.
11-15
dense canopy; water
surface completely
shaded OR nearly full
sunlight reaching water
surface. Shading limited
to <3 hours per day.
6-10
sunlight reaching water
surface.
0-5
5. Channel alteration
(a)
Little or no enlargement
of islands or point bars,
and/or no ;
channelization.
12-15
Some new increase in
bar formation, mostly
Moderate deposition of
new gravel, coarse sand
from coarse gravel; and/ on old and new bars;
or some channelization and/or embankments on
present.
both banks.
8-11
4-7
Heavy deposits of fine
material, increased bar
development; and/or
extensive
channelization.
0-3
6. Deposition (c)
Less than 5% of bottom
affected; minor
accumulation of coarse
sand and pebbles at
snags and submerged
vegetation.
12-15
5-30% affected;
moderate accumulation
of sand at snags and
submerged vegetation.
8-11
5-30% affected; major
deposition of sand at
snags and submerged
vegetation; pools
shallow, heavily silted.
Channelized; mud, silt,
and/or sand in braided
or nonbraided channels;
pools almost absent due
to deposition. . '
4-7 0-3
7. Channel sinuosity (b)
Instream channel length
3 to 4 times straight line
distance.
12-15
Instream channel length
2 to 3 times straight line
distance.
8-11
Instream channel length
1 to 2 times straight line
distance.
4-7
Channel straight;
channelized waterway.
0-3
8. Lower bank channel
capacity (b)
Overbank (lower) flows
rare. Lower bank W/D
ratio <7.
12-15
Overbank (lower) flows
occasional. W/D ratio
8-15.
8-11
Overbank (lower) flows
common. W/D ratio
15-25.
4-7
Peak flows not
contained or contained
through channelization.
W/D ratio >25.
0-3
9. Upper bank stability
(a)
Upper bank stable. No
evidence of erosion or
bank failure. Side
slopes generally <30°.
Little potential for future
problems.
9-10
Moderately stable.
Infrequent, small areas
of erosion mostly healed
over. Side slopes up to
40° on one bank. Slight
potential in extreme
floods. •
6-8
Moderately unstable.
Moderate frequency and
size of erosional areas.
Side slopes up to 60°
on some banks. High
erosion potential during
extreme high flow.
3-5
Unstable. Many eroded
areas. "Raw" areas
frequent along straight
sections and bends.
Side slopes >60°
common.
0-2
Figure 4.—Habitat assessment field data sheets for glide/pool prevalent situations.
34
-------
Biological Criteria: Research and Regulation, 1991
Category
Habitat Parameter
10. Bank vegetative
protection (d)
OR
Grazing or other
disruptive pressure
(b)
Optimal
Sub-Optimal
Marginal
Poor
Over 90% of the
streambank surfaces
covered by vegetation.
9-10
Vegetative disruption
minimal or not evident.
Almost all potential plant
biomass at present
stage of development
remains.
9-10
70-89% of the
streambank surfaces
covered by vegetation.
6-8
Disruption evident but
not affecting community
vigor. Vegetative use is
moderate, and at least
one-half of the potential
plant biomass remains.
6-8
50-79% of the
streambank surfaces
covered by vegetation.
3-5
Disruption obvious;
some patches of bare
soil or closely cropped
vegetation present. Less
than one-half of the
potential plant biomass
remains.
3-5
Less than 50% of the
streambank surfaces
covered by vegetation.
0-2
Disruption of
streambank vegetation
is very high. Vegetation
has been removed to 2
inches or less in
average stubble height.
0-2
11. Streamside cover
(b)
Dominant vegetation is
shrub.
Dominant vegetation is
of tree form.
Dominant vegetation is
grass or forbes.
9-10
6-8
3-5
Over 50% of the
streambank has no
vegetation and
dominant material is
soil, rock, bridge
materials, culverts, or
mine tailings.
0-2
12. Riparian vegetative
zone width (least
buffered side) (e) (f)
(9)
>18 meters.
Between 12 and 18
meters.
Between 6 and 12
meters.
<6 meters.
9-10
6-8
3-5
0-2
Column Totals
Score.
(a) From Ball 1982.
(b) From Plattsetal. 1983.
(c) From EPA 1983.
(d) From Hamilton and Bergersen 1984.
(e) From Lafferty 1987.
(f) From Schueler 1987.
(g) From Bartholow 1989.
Figure 4.—Habitat assessment field data sheets for glide/pool prevalent situations.
A similar nonpoint source evaluation was con-
ducted on Rock Creek, Idaho, in September 1988,
using the RBPs. As in Texas, no habitat limitations
were detected (Fig. 6). However, the biological com-
munity at one station (S-3) was classified as moder-
ately impaired when compared to the reference
(S-6). This level of biological condition is attributed
to water quality effects.
The assessment of a point source influence
wastewater treatment plant (WWTP) to Little Mill
Creek, Kansas, indicated a highly degraded benthic
community immediately downstream of the plant;
but a recovery of the condition of the community
was noted at Station 3 located approximately 1 mile
downstream of the facility. In this study (Fig. 7), the
habitat quality was highly comparable among all
stations because of a riparian protection program
implemented in Johnson County, Kansas.
The point source discharge being assessed on
the North Nashua River, Massachusetts, was a
small paper mill and a wastewater treatment plant.
An additional complication at this site was the pres-
ence of urban runoff. A combination of habitat and
water quality effects was noted from the bioassess-
ment conducted in June 1989. Station 3 was influ-
enced dramatically by a severe habitat degradation
due to construction activities. Station 2, located less
than a half mile downstream of the paper mill and
treatment plant, was judged to be moderately im-
paired and having a supporting habitat quality. A
recovery, both in terms of habitat quality and bio-
logical condition, was observed at Station 4, located
approximately 6 miles downstream of the point
source discharges (Fig. 8).
In these case studies, a knowledge of the vari-
ability to be expected in the relationship between
35
-------
M. T. BARBOUR and J. B. STRIBLING
100
Reference
(Clear Creek)
Habitat Quality (% of Reference)
Figure 5.—Benthlc bloassessment of the Trinity River, Texas.
Reference
(S-6)
10
Habitat Quality (% of Reference)
Figure 6.—Benthlc bloassessment of Rock Creek, Idaho.
100
36
-------
Biological Criteria: Research and Regulation, 1991
Reference
(Station 1)
'v/^t/s' £•* -',, Z,#&M,,"',
/ Supporting ,, Supporting |;, parable
0 10 20 30 40
Habitat Quality (% of Reference)
Figure 7.—Benthlc bloassessment of Little Mill Creek, Kansas.
100
habitat quality and biological integrity would en-
hance interpretation of the results. An understand-
ing of the impact of habitat degradation is critical to
the assessment of the potential of a biological sys-
tem. In situations where habitat has deteriorated,
mitigation or improvement of the habitat through
stream restoration activities should be evaluated.
The implementation of water quality improvements
can be independent of the habitat quality, but judg-
ment of the improvement in biological integrity
cannot.
References
Ball, J. 1982. Stream Classification Guidelines for Wisconsin.
Wisconsin Dep. Nat. Resour. Tech. Bull. In Water Qual-
ity Standards Handbook. 1983. U.S. Environ. Prot.
Agency, Off. Water Reg. Standards. Washington, DC.
Bartholow, J.M. 1989. Stream Temperature Investigations:
Field and Analytic Methods. Instream flow info, paper
13. Biological Report 89(17). Fish Wildl. Serv. U.S. Dep.
Inter., Washington, DC.
Cooper, J.E. 1984. Vanishing species: the dilemma of re-
sources without price tags. Pages 7-32 in A.W. Norden,
D.C. Forester, and G.H. Fenwick, eds. Threatened and
Endangered Plants and Animals of Maryland. Spec.
Publ. 84-1. Maryland Nat. Heritage Progr., Annapolis.
Hamilton, K and E.P. Bergersen. 1984. Methods to Estimate
Aquatic Habitat Variables. Environ. Eval. Proj. N. DPTS-
35-9. Bur. Reclam., Denver Federal Center, Denver, CO.
Hynes, H.B.N. 1970. The Ecology of Running Waters. Univ.
Toronto Press. Ontario, Canada.
Lafferty, B. 1987. A procedure for evaluating buffer strips for
stream temperature protection under the Forest Practi-
ces Act. Pages 70-77 in Managing Oregon's Riparian
Zone for Timber, Fish, and Wildlife. Tech. Bull. No. 514.
Natl. Counc. Air Stream Improve. New York.
Minshall, G.W. 1984. Aquatic insect—substratum relation-
ships. Pages 358-400 in V.H. Resh and D.M. Rosenberg,
eds. The Ecology of Aquatic Insects. Praeger Publ., New
York.
Newbury, R. W. 1984. Hydrologic determinants of aquatic in-
sect habitats. Pages 323-356 in V.H. Resh and D.M. Ro-
senberg, eds. The Ecology of Aquatic Insects. Praeger
Publ. New York.
Osborne, L.L. and E.E. Herricks. 1983. Streamflow and Veloc-
ity as Determinants of Aquatic Insect Distribution and
Benthic Community Structure in Illinois. Rep. No.
UILU-WRC-83-183. Water Resour. Center, Univ. Illinois.
Bur. Reclam. U.S. Dep. Interior, Washington, DC.
37_
-------
M. T. BARBOUR and J. B. STRIBLING
Reference
(Whitman River)
Partially
Supporting H Supporting m parable
Habitat Quality (% of Reference)
Figure 8.—Benthlc bloassessment of North Nashua River, Massachusetts.
Oswood, M.E., and W.E. Barber. 1982. Assessment of fish
habitat in streams: goals, constraints, and a new tech-
nique. Fisheries 7(3):8-ll.
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. Off. Water. EPA/444/4-89-001. U.S. Environ. Prot.
Agency, Washington, DC.
PJatts, W.S., W.F. Megahan, and G.W. Minshall. 1983. Meth-
ods for Evaluating Stream, Riparian, and Biotic Condi-
tions. Gen. Tech. Rep. INT-138. Forest Serv., U.S. Dep.
Agric., Ogden, UT.
Price, P.W. 1975. Insect Ecology. John Wiley and Sons. New
York.
Schueler, T.R. 1987. Controlling Urban Runoff: A Practical
Manual for Planning and Designing Urban BMPs. Publ.
No. 87703. Metro. Washington Counc. Govts., Washing-
ton, DC.
Smith, R.L. 1974. Ecology and Field Biology. 2nd ed. Harper
and Row Publ. New York.
U.S. Environmental Protection Agency. 1983. Technical Sup-
port Manual: Waterbody surveys and assessments for
conducting use attainability analyses. Off. Water Reg.
Standards, Washington, DC.
. 1990. Biological Criteria: National Program Guid-
ance for Surface Waters. EPA-440/5-90-004. Off. Water
Reg. Standards, Washington, DC.
38
-------
Biological Criteria: Research and Regulation, 1991
Variability in Lakes and Reservoirs
John Magnuson
University of Wisconsin
Madison, Wisconsin
Variability occurs at all time and space scales and the extent and grain of the observation system de-
termines many of the properties of the apparent variation. Grain and extent also are relevant when
considering the choice of taxonomic breadth and level for biological analyses as well as when
choosing the organisms with the most appropriate life span for the issue of concern. An analysis of variabil-
ity, such as with coefficients of variation, can be used to help choose criteria or perhaps be criteria, in them-
selves, for surface waters. At the North Temperate Lake Long-term Ecological Research Site in Wisconsin,
observed differences in interyear variation within lakes is related to the lake's position in the landscape
rather than to its proximity to the other lake. Temporal coherence in the interyear variation among lakes is
related more to their similarity in exposure to climate, i.e. surface area to mean depth ratios, than to their
physical proximity. Adjacent lakes have incoherent patterns of interyear variability especially of biological
properties. Knowledge of these patterns of variability can help in the design of measurements systems for
evaluating surface water quality. Community analysis of fishes can be used as a water quality criterion and
can be tracked through time to detect major changes in lake ecosystems.
If you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
39
-------
F.A. REID
in the world, with average net primary production
reaching 2,500 g/m2/yr (Whittaker and Likens,
1973), and it is the perpetual destruction and cre-
ation of individual wetlands within a general re-
gion that maintains long-term productivity of these
systems.
In addition to high organic production in
wetlands, these systems may remove inorganic, or-
ganic, and toxic substances from flowing waters,
thus improving water quality. The high rate of pro-
ductivity can lead to high rates of mineral uptake
by vegetation and subsequent burial in sediments
when the plants senesce (Sather and Smith, 1984).
Denitrification, methane production, and chemical
precipitation remove certain chemicals from wet-
land waters because of aerobic and anaerobic con-
ditions. Reduction of stream velocity as water
enters wetlands causes sediments to fall out of the
water column. Because of shallow water condi-
tions, a substantial water-sediment exchange oc-
curs.
Unfortunately, productivity and water quality
of our national wetland resources have been se-
verely impacted through impacts on natural hy-
drology and alternative land use practices.
Development of dams for flood control and hydro-
power, levees for flood protection, wetland drain-
age for urban, industrial, and agricultural
developments, and dredging for marinas or ports
have all modified wetlands across the continent
(Fredrickson and Reid, 1990).
Although direct wetland conversion has been
slowed in most regions, degradation continues
through the alteration of flooding regimes.
Changes in timing, depth, duration, or frequency
of flooding causes alteration in the hydrologic cycle
of wetlands. The four general categories of hydro-
logic alterations include: (1) stabilization, (2) shift
in flood timing, (3) increased flooding, and (4) de-
creased flooding (Klimas, 1988). Prolonged inunda-
tion of substrates that were periodically exposed
corresponds to stabilization and may involve mod-
ification of temporary, seasonal, annual, or multi-
year flooding patterns.
Shifts in flood timing occur when natural flood
periodicity and chronology change. Increased
flooding is certainly a result of additions in flood-
ing depth or duration, but may result from changes
in timing or frequency as well. Although flood con-
trol reservoirs, levees, and drainage tiles generally
decrease flooding, severe floods may still occur
(Belt, 1975; Klimas, 1988; Reid et al. 1989). Any pro-
gram that intends to evaluate the chemical, physi-
cal, or biological integrity of the nation's wetlands
must be able to detect impacts caused by hydro-
logic alteration, as well as water quality modifica-
tions.
Evaluation at the System Level
As assessment criteria are developed for biological
integrity of wetland waters (U.S. Environ. Prot.
Agency, 1990), several habitat components need
consideration. The broadest consideration is wet-
land type at the system level (Cowardin et al. 1979).
The "system level" is defined as a complex of
wetlands and deepwater habitats that share the in-
fluence of similar hydrologic, geomorphologic,
chemical, or biological factors. These broad catego-
ries include:
1. Marine: Open ocean overlying the continental
shelf and its associated high-energy coastal
line.
2. Estuarine: Tidal wetlands that are usually
semi-enclosed by land but have open, par-
tially obstructed, or sporadic access to the
ocean and in which ocean water is at least oc-
casionally diluted by freshwater runoff from
the land.
3. Riverine: Wetlands and deepwater habitats
contained within a channel, with the excep-
tions of wetlands dominated by trees, shrubs,
persistent emergents, emergent mosses, or li-
chens, or habitats with water containing
ocean-derived salts in excess of 0.5 percent.
4. Lacustrine: Wetlands and deepwater habitats
situated in a topographic depression or a
dammed river channel; lacking trees, shrubs,
persistent emergents, emergent mosses, or li-
chens with greater than 30 percent areal cover-
age; and with a total area exceeding 8 ha.
5. Palustrine: Nontidal wetlands dominated by
trees, shrubs, persistent emergents, emergent
mosses of lichens, and tidal wetlands where
salinity resulting from ocean-derived salts is
less than 5 percent (Cowardin et al. 1979).
Regional and Watershed
Influences
Other habitat components that need consideration
include regional and watershed influences. Certain
wetlands may be confined within a single basin,
42
-------
Biological Criteria: Research and Regulation, 1991
such as prairie potholes or southern playas,
whereas other wetlands, such as lowland hard-
wood swamps, occur across an elevational gradient
or mosaic. Differences in animal response occur
across hydrophyte zones or plant types. One of the
earliest recognized habitat relationships for aquatic
invertebrates was that with aquatic plants. Hydro-
phyte leaf shape, structure, and surface are related
to invertebrate abundance (Wieser, 1951; Rosine,
1955). Several investigators (Krecker, 1939; An-
drews and Hasler, 1943; Krull, 1970) have found
higher density of insects associated with aquatic
plants containing highly dissected leaves. Commu-
nity composition is dependent on plant condition
and food habits of the invertebrates (Reid, 1985).
Seasonal senescence of emergent vegetation en-
courages colonization by detritivore communities
(Danell and Sjoberg, 1979). Annual periphyton
shifts (Millie, 1979) undoubtedly influence grazer
community composition. Water depth or soil mois-
ture, water flow (stream or tidal), and type of water
input (rain, headwater or backwater flooding) may
all influence biological assessment. The timing of
sampling should consider diurnal, tidal, or sea-
sonal patterns.
Habitat Structure and
Biological Response
Several patterns of biological response and habitat
structure may be interrelated. A conceptual model
has been presented for prairie wetlands, in which a
dense emergent zone (Typha and Scirpus domi-
nated) is compared to a deeper submergent zone
(Nelson and Kadlec, 1984). In that example, differ-
ences existed in chemical and biological compo-
nents across both spatial and temporal frame-
works. For example, production:respiration (P:R),
dissolved oxygen, and invertebrate production dif-
fered between these two plant zones, both in spring
and summer periods.
Perhaps the potential of biological criteria as-
sessment for wetlands can be illustrated by recent
investigations in lowland hardwood swamp habi-
tat in southeastern Missouri. Limnological investi-
gations suggested that alteration of natural
flooding regime may result in nutrient export from
the wetland system (Wylie and Jones, 1986). Inves-
tigations of tree, invertebrate, and fish distributions
suggested that flooding regime altered species re-
sponses and community assemblage structure
(Batema et al. 1985: Finger and Stewart, 1988;
Heitmeyer et al. 1989). Biological assessment iden-
tified community shifts in relation to human-in-
duced hydrologic alterations.
If the United States is to truly "enhance and re-
store" wetlands, mechanisms must be found to
identify degraded habitats and protect or replicate
the natural hydrologic regimes for complexes of
quality wetlands. Sites of historic wetlands that
have been altered by other land practices should
also be identified for potential restoration.
References
Andrews, J.D. and A.D. Hasler. 1943. Fluctuations in the ani-
mal populations of the littoral zone in Lake Mendota.
Trans. Wis. Acad. Sci. Arts Lett. 35:175-85.
Batema, D.L., G.S. Henderson, and L.H. Fredrickson. 1985.
Wetland invertebrate distribution in bottomland hard-
woods as influenced by forest type and flooding regime.
Pages 196-202 in J.O. Dawson and K.A. Majeros, eds.
Proc. 5th Central Hardwoods Forest Conf. Dep. Forest.,
Univ. Ilinois, Champaign-Urbana.
Belt, C.B. Jr. 1975. The 1973 flood and man's constriction of
the Mississippi River. Science 189:681-84.
Cowardin, L.M., V. Carter, F.C. Golet, and E.T. LaRoe. 1979.
Classification of Wetlands and Deepwater Habitats of
the United States. FWS/OBS 79/31. U.S. Fish WildL
Serv., Washington, DC.
Danell, K. and K. Sjoberg. 1979. Decomposition of Carer and
Equisetum in a northern Swedish lake: dry weight loss
and colonization by macroinvertebrates. J. Ecol. 67:191-
200.
Finger, T.R. and E.M. Stewart. 1988. Response of fishes to
flooding regime in lowland hardwood wetlands. In W.J.
Matthews and D.C. Heins, eds. Community and Evolu-
tionary Ecology of North American Stream Fishes. Univ.
Oklahoma Press, Norman.
Fredrickson, L.H. and F.A. Reid. 1990. Impacts of hydrologic
alteration on management of freshwater wetlands.
Pages 71-90 in J.M. Sweeney, ed. Management of Dy-
namic Ecosystems. N. Central Sect., The Wildl. Soc.
West Lafayette, IN.
Heitmeyer, M.E., L.M. Fredrickson, and G.F. Krause. 1989.
Water and habitat dynamics of the Mingo Swamp in
southeastern Missouri. U.S. Dep. Inter. Fish Wildl. Res.
G:l-26.
Klimas, C.V. 1988. River, regulation effects on floodplain hy-
drology and ecology. Pages 40-49 in D.D. Hock, W.H.
McKee, Jr., H.K. Smith, J. Gregory, R. Banks, L.H. Stoley,
C. Brooks, T.D. Matthews, and T.H. Shear, eds. The Ecol-
ogy and Management of Wetlands. Vol. 1. Ecology of
Wetlands. Timber Press, Portland, OR.
Krecker, F.H. 1939. A comparative study of the animal popu-
lations of certain submerged aquatic plants. Ecology
20:553-62.
Krull, J.N. 1970. Aquatic plant macroinvertebrate associa-
tions and waterfowl. J. Wildl. Manage. 34:707-18.
Likens, G.E. and F.H. Bormann. 1974. Linkages between ter-
restrial and aquatic systems. Bioscience 24:447-56.
Millie, D.F. 1979. The epiphytic diatom flora of three Lake
Erie marshes. M.S. Thesis. Bowling Green Univ. Bowl-
ing Green, OH.
Mitsch, W.J. and J.G. Gosselink 1986. Wetlands. Van
Nostrand Reinhold Co., New York.
Nelson, J.W. and J.A. Kadlec. 1984. A conceptual approach to
relating habitat structure and macroinvertebrate pro-
duction in freshwater wetlands. Trans. N. Am. Wildl,
Nat. Resour. Conf. 49:262-70.
43
-------
F.A. REID
Reid, EA. 1985. Wetland invertebrates in relation to hydrol-
ogy and water chemistry. Pages 72-79 in M.D. Knighton,
ed. Water Impoundments for Wildlife: a Habitat Man-
agement Workshop. Forest Serv. U.S! Dep. Agric. St.
Paul, MM.
Reid, F.A., J.R. Kelley Jr., T.S. Taylor, and L.H. Fredrickson.
1989. Upper Mississippi Valley wetlands—refuges and
moist-soil impoundments. Pages 181-202 in L.M. Smith,
R.L. Pedersen, and R.M. Kaminski, eds. Habitat Man-
agement for Migrating and Wintering Waterfowl in
North America. Texas Tech. Univ. Press, Lubbock.
Rosine, W.N. 1955. The distribution of invertebrates on sub-
merged aquatic plant surfaces in Muskee Lake, Colo-
rado. Ecology 36:308-14.
Sather, J.H. and R.D. Smith. 1984. An Overview of Major
Wetland Functions and Values. FWS/OBS 84/18.
Western Energy and Land Use Team. U.S. Fish Wildl.
Serv., Washington, DC.
Tiner, R.W. Jr. 1984. Wetlands of the United States: Current
Status and Recent Trends. U.S. Fish Wildl. Serv.
Wetlands Inventory, Washington, DC.
U.S. Environmental Protection Agency. 1990. Biological Cri-
teria: National Program Guidance for Surface Waters.
EPA-440/5-90-004. Washington, DC.
Whittaker, R.H. and G.E. Likens. 1973. Primary production:
the biosphere and man. Human Ecol. 1:357-69.
Wieser, W. 1951. Uber die quantitative bestimmung der al-
genbewchnenden mikrofauna felsiger meeraskusten.
Oikos 3:124-31.
Wylie, G.D. and J.R. Jones. 1986. Limnology of a wetland
complex in the Mississippi alluvial valley of southeast
Missouri. Arch. Hydrobiol. (Suppl.) 74:288-314.
44
-------
Biological Criteria: Research and Regulation, 1991
Reference Ecosystems of the Upper
Mississippi River—Past, Present and Future
Kenneth S. Lubinski
Environmental Management Technical Center
• U.S. Fish and Wildlife Service
Onalaska, Wisconsin
The Upper Mississippi River System (UMRS) includes the commercially navigable reaches of six large
midwestern rivers. The Long Term Resource Monitoring Program for the System was established in
1986 to provide information needed by decisionmakers to maintain a balance between the System's
multiple uses. Program objectives require that we know enough about the past and present ecological status
of the system to be able to predict where it will be at selected points in the future. We used ideas and infor-
mation from the field of large river ecology and publications on river history, populations and habitats to es-
tablish an ecological perspective of the UMRS. Five spatial scales are addressed in the perspective:
watershed, stream network, floodplain reach, navigation pool, and aquatic area. The perspective describes
the major abiotic and biotic factors, and natural and human-induced disturbances that operate at each scale.
Each disturbance is in part defined by the time period over which it re-occurs. The perspective is being used
to establish areas for research emphasis, identify required products, and develop research strategies. For in-
stance, historical vegetation and land-use data at the floodplain reach scale are being digitized and mapped
to visualize the ecological structure of the system during pre-settlement and pre-dam periods. Predictions of
future UMRS ecological status will require an understanding of the relative contribution of each disturbance
to the whole.
// you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
45
-------
Biocriteria for Lacustrine Systems: A Case
History from the Laurentian Great Lakes
John E. Gannon
U.S. Fish and Wildlife Service
Ann Arbor, Michigan
/°
Comprehensive, long-term research and monitoring of the Laurentian Great Lakes has been under-
way only since the 1970's, although limnological and fisheries surveys over the past 100 years-have
provided important benchmarks on changes in water quality and biota. Historically, water quality
monitoring has been the primary, indicator of the need for water pollution control efforts, and continues to
be an indicator in evaluating the effectiveness of implemented pollution control programs. Phytoplankton,
zooplankton, benthos, fish, and fish-eating wildlife have been variously used as indicators with emphasis on
biological function, community structure, and toxicity testing. Most recently, the Great Lakes community
has been attempting to shift away from traditional "one-chemical-at-a-time" water quality objectives, cri-
teria, and standards to integrative ecosystem objectives. Broadly based ecosystem objectives are being estab-
lished for nearshore and offshore aquatic communities, wildlife communities dependent upon aquatic food
chains, habitat, human health, and stewardship. Quantitative indicators are being developed to determine
whether the ecosystem objectives are being met. This approach was first developed on Lake Superior using
the lake trout (Salvelinus namaycush) as the indicator species. The criteria used in selecting the lake trout as
the indicator of ecosystem quality are pertinent to selecting suitable indicators and reference sites elsewhere.
// you loould like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
46
-------
Biological Criteria: Research and Regulations, 1991
The Development of Biocriteria in Marine
and Estuarine Waters in Delaware
John R. Maxted
Delaware Department of Natural Resources
and Environmental Control
Dover, Delaware
ABSTRACT
Delaware is focusing its initial effort on developing biological criteria for marine and estuarine
waters in the three inland bays within the State; Rehoboth, Indian River, and Little Assawo-
man bays. These areas have been selected because of high development pressure, consistently
high salinities, and relatively stable biota. Benthic organisms are used as the indicator of envi-
ronmental quality. The Department is evaluating the utility of this information for managing
development activities in the Inland Bays. By rating or scoring the quality of the benthic com-
munity, the Department can direct development away from high quality areas and encourage
development in low quality areas. This approach is similar to the Clean Water Act Section 404
Advance Identification Program used by the Environmental Protection Agency and the Corps
of Engineers to identify and direct dredged and fill activities away from high quality
wetlands. Two of the most important tasks in the development of biological criteria are (1) the
selection of a biological collection method that is sensitive to pollution and (2) defining the
condition and variability of reference sites. Data collected in the Lower Chesapeake Bay and in
Rehoboth Bay shows that the selected method is sensitive to pollution, and shows greater sen-
sitivity than other methods (e.g., biomass or diversity). The State is currently evaluating minor
changes to the collection method before undertaking extensive sampling throughout the bays.
Introduction
Every two years the States must report on the sta-
tus of their waters in attaining the fishable/swim-
mable goals of the Clean Water Act. The reporting
requirements are met by determining, for each
waterbody, whether State water quality standards
are currently being attained. As in most States, Del-
aware does this by comparing water quality moni-
toring data with numeric water quality criteria
(Del. Dep. Nat. Resour. Environ. Control, 1990a).
Recently, this task has become more complex with
the added emphasis on toxic pollutants in sections
303(c)(2)(B) and 304(1). of the Clean Water Act. The
ultimate purpose of these assessments is to answer
the simple question: "Is the water healthy enough
for human consumption and aquatic life protec-
tion?
Assessments that use chemical criteria are
based on the presumption that if these criteria are
not exceeded, then the uses are attained. As toxics
are increasingly controlled through additional
chemical criteria and whole effluent toxicity testing,
regulatory agencies and the public wonder if these
controls have resulted in a healthy indigenous bio-
logical community of plants and animals.
Water chemistry data and criteria are powerful
tools in regulating water quality. They are used to
measure the pollutant removal effectiveness of
treatment technologies and quality assessments of
surface and ground waters. These techniques have
been and will continue to be fundamental to pollu-
tion control for point sources through discharge
. permits.
However, our ability to determine the overall
health of natural systems is limited. As the U.S. En-
• vironmental Protection Agency (EPA) and selected
47
-------
J. R. MAXTED
States have made clear through guidance (U.S. En-
viron. Prot. Agency, 1990) and regulations (Ohio En-
viron. Prot. Agency, 1988), the best approach to
assessment is an integrated one in which the
strengths of each assessment tool are emphasized.
Biological tools are most effective in .assessing bio-
logical integrity. Where water quality problems are
detected, chemical criteria are best at controlling
pollution sources. Biology should not be used as the
sole basis for controls, nor should water chemistry
be considered the sole basis for assessment.
Numeric criteria provide a quantitative mea-
sure of performance. In a society that is driven by
numbers in everything from speed limits to school
grades, they seem necessary. However, the quantita-
tive approach raises a particular dilemma for both
freshwater and marine biologists—how to charac-
terize the quality of the aquatic community numeri-
cally while recognizing the inherent complexity of
natural systems. The issue is the degree to which bi-
otic integrity can be quantified while still retaining
scientific validity.
Jim Karr, who developed the Index of Biotic In-
tegrity (IBI) (Karr et al. 1986), and others have dem-
onsbrated that numerical interpretation of natural
systems can be done without sacrificing scientific
validity. The IBI concept does not constitute a new
. approach to biological assessment. Rather, it has
provided a new way of reporting the results that
make it easier for biologists to communicate scien-
tific information to regulatory agencies, the regu-
lated community, and the public. The IBI provides a
vehicle for bringing biology out of the file drawer
and into the hands of decisionmakers.
Many numerically based assessment tools have
been developed for marine and estuarine environ-
ments. It is up to the States to apply these tools to
the management of marine and estuarine waters so
that they can better answer the question: Is the
water healthy?
Biocriteria Program —
Delaware
Delaware is testing a numerically based biological
assessment tool. This program is designed to ad-
dress all types of surface waters in the State, in-
cluding rivers, ditches, ponds, estuaries, and
wetlands, both tidal and nontidal. Initially, it has
been focused on the use of benthic invertebrates as
indicators of biotic integrity.
To manage this complex task, Delaware's sur-
face waters have been divided into four major cate-
gories that are relatively homogeneous with regard
to biological conditions. This division is based oh
three factors: physiographic characteristics or
ecoregions (Omernik, 1987), tidal influence, and
sampling equipment.
These regions and the assessment strategies to
be applied to them are described as follows:
• Freshwater/nontidal—piedmont ecoregion:
Kick net in riffles using EPA Rapid - .,
Bioassessment Protocol III (Plafkin et al.
1989); salinity 0 ppt. • . .
• Freshwater/nontidal—coastal plain .
ecoregion: D-frame net swept along banks
(under development); salinity 0 ppt.
• Freshwater/tidal (under development).,
Salinity less than 5 ppt.
• Marine/estuarine—Depth stratified sample
using box or tube cores; salinity greater than
5 ppt.
Marine and Estuarine
Biocriteria Program
The program to develop biocriteria for estuarine
and marine waters is initially based in the Inland
Bays region of southern Delaware: the Indian
River, Rehoboth, and Little Assawoman bays. This
focus is in large part the result of intense develop-
ment pressure in these areas as evidenced by their
designation as a National Estuary Program; a 40
percent increase in population over the last 10
years; the development in 1990 of a water Use plan
to help manage the multiple uses of water within
the watershed and the designation of the region as
an outstanding water resource in State water qual-
ity standards. These designations have focused
State efforts in the Inland Bays region, including
nonpoint source activities under section 319 and
regulated activities, including those permits for.
point source discharges, marina projects, and activ-
ities affecting subaqueous lands and wetlands.
The recently adopted State marina regulation
Pel. Dep. Nat. Resour. Environ. Control, 1990b) has
spurred the development of biological indicators in
marine and estuarine systems. The regulation re-
quires marina developments to address several liv-'
ing resource components: wetlands, subaqueous
lands, shellfish beds, submerged aquatic vegetation,
and benthic resources. The latter component re-
quires assessment of benthic invertebrate commune
ties using a method developed by Luckenbach,
Diaz, and Schaffner (Luckenbach et al. 1988) (Fig; 1).
48
-------
Biological Criteria: Research and Regulations, 1991
MARINA REGULATIONS
Benthic Resources
"Benthic resources are protected as a matter of
policy because of their importance in the food
chain and their value as commercial and
recreational food sources.
The status of the benthic community must be
assessed by the applicant using frequency,
diversity and abundance measures approved by
the Department. As a part of this determination,
the rapid bioassessment techniques of
Luckenbach, Diaz and Schaffner (1989) will be
used by the Department to characterize benthic
communities. Taxonomic and biomass data
specific to this methodology shall be collected.
Only areas scoring 0-3, on a relative scale of
0-8, will be considered for marina siting. The
Department may modify this methodology as
'experience is gained in applying these
techniques in Delaware waters."
Total Score
0-1,
2-3'
Figure 1.—Delaware Department of Natural Resources and
Environmental Control marina regulations.
— See also Figure 2.
Delaware is in the process of testing and modi-
fying this methodology in State estuaries. These
data will be evaluated with regard to establishing
numeric biocriteria in State water quality standards.
Methods
The rapid assessment technique developed by
Luckenbach, Diaz, and Schaffner is based on the
premise that a healthy benthic community is char-
acterized by large, deep-dwelling organisms, pri-
marily animals from the Annelida (worms) and
Mollusca (clams) orders. A benthic community that
is dominated by small animals from families that
are 'characteristic of unstable environments is an in-
dicator of impact or stress.
The method has been tested in the lower Chesa-
peake Bay and been shown to be an indicator of bi-
otic integrity (Luckenbach et al. 1988). Sampling
requires recovery of a sediment sample intact to
allow sectioning with depth. The fraction in the top
5 centimeters is processed separately from the sam-
ple from 5 to 15 cm. The sample collection is rapid,
requiring no more than 30 minutes at each station.
The cost of lab processing is approximately $100 to
$200 for each sample (both top and bottom). Nu-
merical scores are calculated from these data and
the benthic community is defined according to Fig-
ure 2.
Benthic Community Character
"Poor" health, highly disturbed,
early successional, poor water
quality or other severe disturbance
"Poor"-to "Fair" health, moderately
disturbed, perhaps recovering
community, suggestion of poor
water quality
"Moderate" to "Good" health, mid-
successional stage
"Good" health, undisturbed, late •
successional community
Figure 2.—Benthic community scoring system.
The method uses a mula'-variate approach
based upon three pieces of information to derive a
numerical score:
• Size determination—number of animals
greater than 2 cm in length;
• Taxonomic composition—number of
families characteristic of stable conditions;
and
• Biomass—percent of the total biomass
contained below the surface of the sediment
(below 5 cm).
The physical habitat quality of the sediments is
also evaluated. Measurements of percent sand and
percent volatile residue are made along with quali-
tative information on the color and texture of the
sediments and the presence of submerged aquatic
vegetation. Generally, the procedure is most appli-
cable to unvegetated bottoms. Sites with submerged
aquatic vegetation may require a different scoring
approach. Detailed water chemistry data are not
collected. Scoring is performed according to the
procedures presented in Figure 3.
Data Collection — Rehoboth
Bay
Three types of data were considered most impor-
tant for the development of biocriteria focused on
benthos: benthic community, sediment type, and
salinity. A review of historical data indicated that
benthic resource and sediment type data have not
been collected in the Delaware's inland bays since
1970 (Maurmeyer and Carey, 1986). Because of de-
velopment that has occurred in the bays over the
last 20 years, additional data collection was
deemed necessary. The review of historical salinity
data indicates that all of Rehoboth Bay is polyhal-
ine (greater than 25 ppt). Therefore, the benthic
data collected in Rehoboth Bay will not be affected
49
-------
J. R. MAXTED
Phase I Scores
Fauna present below five cm?
Fauna below five cm greater
two cm in maximum
dimension?
Yes
No
Yes
No
Score
1
0
Phase II Scores
Species present below five cm "
Only surface dwellers present
(Spionidae, Capitelidae
Ollgochaeta)
Small burrowers and commensals,
(Mactridae, Nereldae, Glyceridae
Nephytiidae, Polynoidae,
Syllidae, Cirratulidae,
Phyllodocidae, Hesionidae,
Pilargidae), but not those listed
below.
Long-lived, large fauna
(Tellinidae, Veneridae,
Solenidae, Chaetopteridae,
Onuphidae, Maldanidae,
Terebellidae, Ophioroida)
Score
0
Phase III Scores
% Blomass below five cm
0-1
1-10
10-30 -'
30 - 60
60 -100
Score
0
1
2
3
4
Figure 3.—Bonthlc community scoring metrics.
by changes in salinity. Benthic resource data were
collected at four stations in Rehoboth Bay in July
1990 (Fig. 4).
This initial sampling had two objectives. First,
the sampling tested the sensitivity of the method.
Two stations were chosen in areas of intense human
activity and two in areas protected from human ac-
tivity. The second objective was to define the spatial
heterogeneity of the data and the variability of the
unit sampling effort (250 sq. cm of bottom). To ad-
dress this objective, three replicates were collected
at each station.
Results and Discussion
The results of the scoring are presented in Table 1.
The biomass and size data are presented in Table 2,
while the taxonomic composition data are pre-
sented in Table 3. Several conclusions can be drawn
from the data.
• Differences between impacted and unimpacted
stations were not clearly distinguished. These dif-
ferences would be more clearly defined by adjust-
ing the calculation procedures. The method may
need to be regionally customized.
k
• Numerical scores ranged from 5 to 8, or all in
the "good" to "excellent" range. Station 4, Sally's
Cove, was significantly better in quality with re-
gard to the criteria calculations, number of sensi-
tive families, and percent of biomass in the bottom
fraction than the other sites.
• There is insufficient data on sediment type. Ad-
ditional data on sediment type throughout the bay
are needed to interpret the biological data.
• For percent biomass calculations (Table 2), there
was good correlation between annelids and whole
samples, except large clams were present (Station
3). Future sampling will be focused in nonshellfish
areas, and biomass calculations will be made usirig
Annelids only.
Table 1.—Rehoboth Bay scores (Stations 1-4)
(as revised 9/28/90). .
PHASES
STATIONS
III'
SCORE
State Park (sand)
1
1-A
1-B
Composite3
Marina (mud)
2
2-A
2-B
Composite
L&R Canal (mud)
3
3-A
3-B
Composite
Sally's Cove (sand)
4
4-A
4-B
Composite
2
2 • -
2
2
2
2
2
2
2.
2
2
2
2
2
2
2
1
•• ' "1
1 -
, 1
2
1
1
2
1
1
0
1
2
2
2
2
. 4
4
3
4
4
3
4
3
3
3
3
3
4
4
4
4
7
7
6 •
x = 6.6
7
8
6
7
x '= 7.0
7
'6
6
5
x = 5.6
6
8
8
8
x = 8.0
-••••'••': 8'.
Note: Based on Luckenbach/Diaz/Shaffner Rapid Assessment Procedure
(Luckenbach et al. 1988).
1 Families represented by the data that resulted in a one point score in-
cluded four Annelids (Cirratulaidae, Nereidae, Phyllodocidae, and Sylli-
dae) and one Mollusc (Mactridae). Families represented by the data that
resulted in a 2 point score included three Annelids (Chaetoptaridae, Mal-
donidae, and Onuphidae) and two molluscs (Tellenidae and Veneridae).
2Phase III biomass calculations were based upon Annelids only due to
dominance of one Mollusc in Station 3-B sample.
Calculation of a single composite value for each station, based upon'
composite of the data for each station. . .,, v j,.
50
-------
Biological Criteria: Research and Regulations, 1991
O«LAWA»« 1AT
SCALE IM un.cs
3TUOT AHEA (OUNOAmCS
WOHCIXTXn COUMTT - UAMVLAMO
Figure 4.—Delaware Inland Bays and Rehoboth Bay sampling locations. (1) State Park; (2) Marina; (3) L&R Canal; (4) Sally's
Cove. ,
51
-------
J.R. MAXTED
Table 2.— Rehoboth
Bay biomass data (as revised 9/28/90).
Macroinfauna biomass as gross wet weight, and size
distribution, Rehoboth Bay, July 1990
NO. 2 cm % BIOMASS-BOTTOM
State Park
Marina
L&R
Canal
Sally's
Cove
STATION
1
1
1
1-A
1-A
1-A
1-A
1-B
1-B
1-B
2
2
2
2-A
2-A
2-A
2-B
2-B
2-B
2-B
3
3
3
3
3-A
3-A
3-A
3-A
3-B
3-B
3-B
3-B
4
4
4
4-A
4-A
4-A
4-B
4-B
4-B
4-B
4-B
DATE
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
90/07/12
TAXON
Annelida
Mollusca
Miscellaneous
Annelida
Arthropoda
Mollusca i
Miscellaneous
Annelida
Mollusca '
Miscellaneous
Annelida
Arthropoda
Mollusca
Annelida .
Arthropoda
Mollusca
Annelida
Arthropoda
Mollusca
Echinodermata
Annelida
Arthropoda
Mollusca
Miscellaneous
Annelida
Arthropoda
Mollusca
Miscellaneous
Annelida
Arthropoda
Mollusca
Miscellaneous
Annelida
Arthropoda
Mollusca
Annelida
Arthropoda
Mollusca
Annelida
Arthropoda
Mollusca
Chironomidae
Miscellaneous
BOTTOM
0.712
0.000
0.070
1.645
0.000
0,349
0.057
0.501
0.000
0.000
0.748
0.000
0.000
0.439
0.000
0.000
0.508
0.002
0.000
0.000
0.169
0.002
0.000
0.000
0.246
0.002
0.000
0.001
0.194
0.003
0.000
, 0.000
1.322
0.001
0.225
0.658
0.001
0.112
0.818
0.000
0.020
0.001
0.000
TOP
0.330
0.097
0.007
0.317
0.001
0.024
0.001
0.425
0.022
0.004
0.450
0.002
0.017
0.539
0.001
0.005
0.188
0.012
0.013
0.001
0.114
0.065
0.002
0.001
0.246
0.039
0.078
0.000
0.188
0.066
2.022
0.007
0.231
0.050
0.000
0.149
0.022
0.002
0.147
0.035
0.021
0.000
0.004
BOTTOM TOP ANNELIDS WHOLE COMPOSITE*
9
5
8
22
15
5
11
31
7
1
3
11
6
11
9
. 26
0 68 64
1 83 86
68
2 54 -53
3
4 62 61
5 45 45
59
1 73 70
10
0 60 48
1 50 41 53
2 51 8
3
llyanassa obsoleta (1 spec.)
1 85 85
0 81 • 82
84
0 85 80
1
Source: DNREC, Div. ol Water Resources, Dover, 1990.
•Annelids, only.
• There was a fair degree of spatial heterogeneity
in the biomass and size distribution data. Surveys
using a 3-replicate design at 250 sq. cm per repli-
cate will continue to be conducted.
• The method allows comparison with historical
data using straight grab sampling by combining
the top and bottom fractions. Therefore, the data
are easily comparable with other studies using a
straight grab sampling method.
Reference Conditions
It is easy to score biotic integrity numerically as
shown above. It is more difficult to set the thresh-
old or criteria for water quality standards. Criteria
are needed to determine whether actions should be
taken to restore degraded conditions or maintain
existing quality.
The process of setting criteria in freshwater
streams has used two basic approaches: regional
reference streams that are determined to be "least
impacted" and upstream-downstream compari-
sons. Clearly, an upstream-downstream approach is
not applicable to marine and estuarine systems.
Therefore, establishing a set of regional references is
necessary.
This approach may be problematic in that it
may simply define the "best of what is left" rather
than what is attainable. In other words, the "best of
52
-------
Biological Criteria: Research and Regulations, 1991
what is left" may be impacted when compared to
conditions within a larger region. This is especially
true when assessing small systems with a limited
pool of reference conditions from which to choose.
For example, it is difficult to say if Station 4 (Sally's
Cove) in Rehoboth Bay is impacted because of
large-scale development in the region.
This type of sampling bias could drastically af-
fect the derivation of biocriteria in estuaries and
alter the technical and political decisions made to
manage these resources. Unfortunately, the behav-
ior of ambient biological systems is difficult to pre-
dict. Otherwise, we could crank coefficients into a
model to tell us the biological community that is at-
tainable under various scenarios. Clearly, an empiri-
cal or observed approach is therefore necessary.
Blindly implementing controls and observing
what is attainable is costly, time-consuming, and
wasteful. To date, the use of "least impacted" natu-
ral systems to derive biocriteria has worked in those
States (Ohio and Maine) that have developed
biocriteria. When dealing with complex natural sys-
tems, we may have no choice but to strive to attain
"the best of what is left." The only question that re-
mains is the spatial scale that is used. The pool pf es-
tuaries within Delaware is clearly not large enough,
while using all the estuaries in the United States
does not recognize major differences in estuaries on
the Atlantic, Pacific, and Gulf coasts.
The selection of references for estuaries will re-
quire a regionally coordinated approach, not only in
the selection of "least impacted" sites but also in the
development and use of standard data collection
methods. Unfortunately, coordinating the many di-
verse groups involved (States, estuary programs,
local governments, researchers, and academics) will
not.be easy. ,
EPA can play a vital role in facilitating this coor-
dination. Ongoing EPA programs that could con-
tribute include the Biocriteria Development
Program, the Environmental Monitoring and As-
sessment Program (EMAP) (U.S. Environ. Prot.
Agency, 1990b) and local programs such as the Na-
tional Estuary Program and the Chesapeake Bay
Program. The provinces used in EMAP, as shown in •
Figure 5, may provide a framework for managing
the development of biocriteria for estuaries on a re-
gional scale.
Columbian
Acadian
Virginian
Callfornlan '&££$
West Indian
Figure 5.—EMAP Physiographic provinces.
53
-------
J. R. MAXTED
Table 3.—Rehoboth Bay taxonomic data summary
(Indicators of good/excellent quality).
RESULTS-ALL STATIONS
(BELOW 5 CM)
FOUND IN
REHOBOTH BAY
Annelida
Polychaeta
1. Chaetopteridae
2. Cirratuladae
3. Glyceridae
4. Hesionidae
5. Maldonidae
6. Nephytidae
7. Nereidae
8. Onuphidae
9. Phyllodocidae
10. Pilargldae
11. Polynoidae
12. Syllidae
13. Terebellidae
Mollusca
Petecypoda
* 14. Mactridae
" 15. Teliinidae
* 16. Solenldae
" 17. Veneridae
Echinodermata
Ophiuroida
•• 18. All Families
(Segmented worms)
X
X
X
X
X
(Bivalves)
X
X
(Brittle stars)
Total 9
RESULTS BY STATION (TOTAL NUMBER, NUMBER OF FAMILIES)
Stalion 1 — 7, 2
Station 2 — 8, 3
Stations— 2, 2
Station 4 — 23. 4
Source: DNREC, Div. of Water Resources, Dover; 1990,
•1 pi, score ' !
"8pt. score
The first step in this process is to draw together
representatives from government, research, and ac-
ademia to help standardize the collection methods
and select sites for data collection, including the se-
lection of references. In this way, data can be col-
lected over the next several years to support the
derivation of biocriteria in the future. The develop-
ment of biocriteria requires a long term commit-
ment. Through a coordinated effort, we can produce
quantitative biocriteria for estuaries to help answer
the question, is the estuary healthy?
References
Delaware Department of Natural Resources and Environ-
mental Control. 1990a. Delaware Water Quality Inven-
tory, Vol. I, II, and III. Dover.
. 1990b. Marina Regulations. Dover.
Karr, J.R. et al. 1986. Assessing Biological Integrity of Run-
ning Waters — A Method and Its Rationale. Spec. Pub.
5. 111. Nat. History Surv., Champaign.
Luckenbach, M.W., R.J. Diaz, and L.C. Schaffner. 1988. Ben-
thic Assessment Procedures. Va. Inst. Mar. Sci., Glouces-
ter Point.
Maurmeyer, E.M. and W.L. Carey. 1986. A Preliminary Re-
search Master Plan for the Delaware Inland Bays. Del.
Dep. Nat. Resour. Environ. Control, Dover.
Ohio Environmental Protection Agency. 1988.
Biological Criteria for the Protection of Aquatic Life.
Vol. I. Columbus.
Omernik, J.M. 1987. Aquatic Ecoregions of the Contermi-
nous United States. Ann. Ass. Am. Geogr. 77:118-25.
Plafkin, J.L. et al. 1989. Rapid Bioassessment Protocols for
Use in Streams and Rivers. EPA 444/4-89-001. U.S. Envi-
ron. Prot. Agency, Washington, DC.
U.S. Environmental Protection Agency. 1990a. Biological Cri-
teria—National Program Guidance for Surface Waters.
EPA 440/5-90-004. Off. Water Reg. Stand., Washington,
DC.
. 1990b. Environmental Monitoring and Assessment
Program—Near Coastal. Program Plan for 1990. Off.
Res./Dev., Narragansett, Rl.
54
-------
Biological Criteria: Research and Regulation, 1991
The Puget Sound Wetland Restoration
Monitoring Protocol
Ronald M. Thorn
Battelle Marine Sciences Laboratory
Sequim, Washington
Charles A. Simenstad
Wetland Ecosystem Team, Fisheries Research Institute
University of Washington
Seattle, Washington ,
Curtis Tanner
Wetland Ecosystem Team, Fisheries Research Institute
University of Washington
Seattle, Washington
ABSTRACT
A systematic approach for measuring estuarine wetland function, particularly in wetland
restoration and mitigation projects, has been lacking; consequently, the development of the
"ecotechnology" of estuarine wetland restoration and creation has proceeded haphazardly.
To remedy this situation, the Urbanized Estuary Mitigation Working Group (UEMWG) de-
veloped a protocol to quantitatively assess the function of estuarine wetlands and associ-
ated habitats for fish and wildlife. The goal of the protocol is to initiate systematic, on-site
measurement of estuarine wetland function for fish and wildlife utilization by assessing at-
tributes of the habitats identified as being functionally important to fish and wildlife. The
information gathered is added to the data base on the ecotechnology of estuarine wetland
construction. The protocol specifies parameters, measurement methods, and statistical eval-
uation criteria for assessing the level of functioning of habitats. This information provides
the groundwork for development of biological criteria for evaluating the quality of estua-
rine wetlands in the Pacific Northwest and can be used as a benchmark for gauging effects
of development and mitigation on wetlands.
Introduction
The demand for wetland restoration and creation is
increasing at a rate beyond the ability of present
"ecotechnology" to effectively implement or man-
age (Zedler, 1986). In a recent analysis of 35 projects
receiving wetland development permits requiring
mitigation (< 5 percent of all permits) in Washing-
ton, Kunz et al. (1988) documented that only 68
percent of the lost wetland types were replaced.
This level of mitigation is in line with that occur-
ring on a national basis (Kusler et al. 1988).
Failure to mitigate wetlands loss and damage
may be attributed to two principal problems: (1) a
lack of technical knowledge of wetlands structure
and function; and (2) an inability of regulators and
managers to uniformly assess mitigation projects
and their outcome (Cooper, 1987; Kunz et al. 1988).
To advance the technology of wetland construction,
a relevant and scientifically sound data base that in-
cludes samples from both natural and developed
systems is needed.
The protocol outlined here describes and rec-
ommends techniques for quantitatively measuring
55
-------
Ft. M. THOM, C. A SIMENSTAD, and C. TANNER
of habitats, species, or attributes could be sampled
than might have originally been recommended.
Comparison between restored and natural systems
must be made with caution, because natural sys-
tems have developed over long periods of time
(e.g., hundreds of years for salt marshes), and even
the best restored systems may not reach the refer-
ence level in a lifetime (Frenkel and Morlan, 1990).
The quality of habitats could be assessed rela-
tive to criteria (e.g., mean prey densities, seasonal
dynamics in the mean prey densities) established
for reference conditions in these systems. The proto-
col utilizes a selected set of measurable parameters
that are indicators of habitat quality. These parame-
ters are based on recommendations for assessing
pollution impacts (Gray, 1981) and for determining
the important members of a biological assemblage
that are responsible for its structure (i.e., Paine,
1966). The procedure used to develop the present
regional protocol was efficient and involved re-
gional scientists. This latter fact increases the proba-
bility of reaching a large data base for a region and
enhances the likelihood of development of a credi-
ble protocol. Because the species guilds and the lev-
els of attributes are regionalized, development of a
similar protocol for other regions would require a
process similar to that used for the Pacific North-
west.
References
Cooper, J.W. 1987. An overview of estuarine habitat mitiga-
tion projects in Washington State. Northw. Environ. J.
3:113-27.
Forman, R.T.T. and M. Godron. 1986. Landscape Ecology.
John Wiley and Sons, New York. • '.,"•' :.
Frenkel, R.E. and J.C. Morlan. 1990. Restoration of the
Salmon River Salt Marshes: Retrospective and Prospect.
Final Rep. to the U.S. Environ. Prot. Agency, Seattle,
WA. Oregon State Univ., Corvallis. : ::
Gray, J.S. 1981. The Ecology of Marine Sediments; Cam7
bridge Univ. Press, MA. , .
Kusler, J.A., M.L. Quammen, and G. Brooks, eds. 1988. Pro-
ceedings of the National Wetland Symposium: Mitiga-
tion of Impacts and Losses. Ass. State Wetland
Managers,Berne,NY. _•.'..:.».-•..•.'—'.,
Kunz, K., M. Rylko, and E. Sommers. 1988. An assessment of
wetland mitigation practices pursuant to section, 404
permitting activities in Washington State. Pages 515-531
in Proc. First Annu. Meet. Puget Sound Res; Vol. 2,
Puget Sound Water Qual. Auth., Seattle, WA.
Paine, R.T. 1966. Food web complexity and species diversity.
Am. Nat. 100:65-75.
Simenstad, C.A., C.D. Tanner, and R.M. Thorn. 1990. Estua-
rine wetland restoration monitoring protocol. Final
Draft Rep. to U.S. Environ. Prot. Agency, Region 10, Se-
attle, WA.
Zedler, J.B. 1986. Wetland restoration: trials and errors in
ecotechnology? Pages 11-16 in R. Strickland, ed., Wet-
land Functions, Rehabilitation and .Creation in(the Pa-
cific Northwest: The State of Our Understanding. Proc.
Conf., April 30-May 2, 1986, Fort Worden State Park,
Port Townsend, WA. Publ. No. 86-14. Wash. Dep. Ecol.,
Olympia. '•••••. : :• -
60
-------
Biological Criteria: Research and Regulation. 1991
Relationships Among Water and Sediment
Contamination Toxicity and Community
Responses in the Trinity River, Texas
James H. Kennedy
Kenneth L. Dickson
William T-Waller
Ray Arnold ;
University of North Texas . , ,
Denton, Texas . . '....,. . '
- - ( ' •• • '
The community structure of the fish and benthic macroinvertebrates was measured quarterly over an
18 month period along a 200 mile section of the Trinity River in Texas. Chemical analyses of the
water and sediment were conducted at each sampling station during each survey. Water and sedi-
ments were evaluated for toxicity via chronic bioassay using Ceriodaphnia dubia (water), fathead minnow
(water and sediment), Chironomus tentans (sediment), Microtox and Corbicula fluminea (in situ water). A syn-
thesis of the taxonomic analyses, water/sediment chemical data, and toxicity tests results provided insight
into factors regulating faunal distributions. Probably the single greatest complicating factor in, establishing
associations between these parameters is the lack of habitat equality between the stations. In the absence of
stress from either point or nonpoint pollutants, one might anticipate that the community structure would be
the same. However, given the distance over which the stations were distributed in this study and the likeli-
hood of finding equal habitat, in the system uncomplicated by anthropogenic sources is unrealistic and,
therefore, the likelihood of rinding significant associations is also reduced. A rank sum analysis approach
was used to evaluate the relationships between the biological, physical, and chemical metrics collected dur-
ing the study. Data from this study suggest that the ranking scheme technique can provide insights on tem-
poral and spatial relationships of point and nonpoint source impacts.
// you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
61
-------
Designing Surveys to Assess Biological
Integrity in Lakes and Reservoirs
James R. Karr
Michele Dionne
Department of Biology
Virginia Polytechnic Institute and State University
Blacksburg, Virginia
ABSTRACT
Successful approaches have been developed during the past decade for monitoring biological
conditions in streams. Much less progress has been made in assessing biological conditions in
lakes and reservoirs. Development of assessment approaches for these water bodies presents
several challenges. These include the need to (1) identify the important physical and biological
variation within and between water bodies; (2) develop field sampling methodologies that pro-
vide quality field data useful in defining ambient conditions; (3) specify reference conditions to
provide expectations against which sample locations can be evaluated; and (4) develop methods
of data analysis and synthesis that best reflect ambient environmental conditions. Assessment
methods should be grounded in ecological principles, and be sensitive to the full range of
human influences (pollution) on aquatic ecosystems; that is, they should be sensitive to degrada-
tion caused by both chemical contaminants and other impacts of human society. Using fish sam-
pling data from the Tennessee Valley Authprity, this study sought parameters that reflect basic
ecological relationships in reservoirs. Our first series of metrics developed as an index of biotic
integrity includes species richness and composition; fish health; reproductive guilds; individual
biology; population structure; and community trophic structure. Although much work remains
to be done in the development of biological assessment approaches in lakes and reservoirs, this
research demonstrates that sufficient knowledge is currently available to make more informed
decisions about the protection and management of these important water resources.
Background
The goal of monitoring is to accurately depict the
conditions of the sample environment in an effort
to assess the degree and causes of biological degra-
dation, if any. If a balanced biological community is
the expectation, then deviation from that condition
could result from pollution, defined in the Clean
Water Act of 1977 (PL 95-217) as "the manmade or
man-induced alteration of the chemical, physical,
biological, or radiological integrity of water." Thus,
pollution is not narrowly defined as chemical con-
tamination, but also includes any human action
that degrades a water resource.
Chemical analysis of water samples has long
been used for evaluation of water resources because
of the ease of sampling and the apparent rigor con-
ferred by sample analysis in a controlled laboratory
environment. However, rigorous analytical quality
does not compensate for the weaknesses of chemi-
cal sampling. Samples are representative of condi-
tions at the sampling site for only a brief period.
Transitory divergence from those conditions, such
as during a runoff event or intermittent industrial
release may be missed. Finally, the high level of nat-
ural variation in chemical characteristics requires
that substantial changes occur before statistical in-
ferences about change can be made.
The addition of biological monitoring does not
entirely avoid these problems. However, judicious
use of biological monitoring can, because of the
long life cycles of individual organisms, provide a
62
-------
Biological Criteria: Research and Regulation, 1591
more integrative view of the cumulative impact of
many human influences. In addition, knowledge of
the age, size, and trophic structure of sampled pop-
ulations integrates biological conditions over peri-
ods that exceed the life of individual organisms.
Historically, studies of lake systems have in-
cluded aspects of biological dynamics more often
than have surveys of streams. Connections between
nutrients and phytoplankton abundances in lakes
have been explored by limnologists for decades,
and attempts to link nutrients, phytoplankton activ-
ity (e.g., chlorophyll a) and lake morphoedaphic in-
dexes are common (Hutchinson, 1967; Wetzel,
1983). Others have explored connections between
abiotic factors and fish communities of lakes (John-
son et al. 1977; Tonn and Magnuson, 1982;
Matuszek and Beggs, 1988). All emphasize the bot-
tom-up determination of biological conditions. Rer
cent evidence suggests that aquatic ecosystem
dynamics are determined by a complex interaction
of bottom-up and top-down regulation (i.e., limita-
tion of target species by processes at lower trophic
levels or higher trophic levels, respectively) of pop-
ulations and variation in the physical environment
(Tonn and Magnuson, 1982; Northcote, 1988; Karr et
al. in press). Thus, knowledge of nutrient or phyto-
plankton abundances may not be sufficient to en-
sure understanding of biological condition.
Fortunately, this improved understanding of
the dynamics of lakes :comes at a time when society
and the regulatory agencies are calling for a revolu-
tion in monitoring programs. Continuing declines
in the quality and quantity of water resources de-
spite extensive regulatory efforts demonstrate the
inadequacies of existing programs (Gen. Ace. Off.
1977; Karr and Dudley, 1981; Nat. Res. Counc. 1987;
U.S. Environ. Prot. Agency, 1987, 1988a,b, 1989,
1990a; Karr, 1987, 1991). Calls for restructuring of
existing monitoring programs and acceleration of
the development and application of promising bio-
logical monitoring techniques are common (U.S.
Environ. Prot. Agency, 1987), and substantive prog-
ress is being made in defining the conceptual under-
pinnings of biological assessment (Karr, 1987; U.S.
Environ. Prot. Agency, 1990a) and developing better
methods for assessment (Karr, 1991, Karr et al. 1986;
Ohio Environ. Prot. Agency, 1988; Plafkin et al. 1989;
Davis, 1990).
During the past decade, successful approaches
for monitoring ambient biological conditions have
been developed for streams. These new assessment
methodologies are based on long-established princi-
ples. Kolkwitz and Marsson (1908) demonstrated
that pollution-tolerant forms replaced less tolerant
species in degraded environments. Richardson
(1928) noted that degradation was better indicated
by tracking species composition and relative abun-
dances than by the often dramatic changes in abso-
lute abundances of individual species. Patrick
(1950) and Cairns (1974) called for use of biological
communities in assessment of water resources. Ad-
vances in streams may be useful in guiding the de-
velopment of biomonitoring in lakes.
Four factors have contributed to rapid advances
in biomonitoring in the last decade (Karr, 1991): (1)
recognition that past approaches have not protected
water resources; (2) development of integrative eco-
logical indexes; (3) development of regional ap-
proaches to establishing ecological expectations;
and (4) assessment of cumulative impacts of numer-
ous, often small, societal activities.
An. index of biotic integrity (IBI) developed
nearly 10 years ago (Karr, 1981) was designed to
measure the extent to which a stream fish commu- .
nity approximates an excellent natural community.
IBI was specifically developed for biological moni-
toring of small- to medium-sized streams. As a re-
sult, the sampling protocol was established to
account for patterns of variation in stream commu-,
nities at several spatial and temporal scales, a goal
made easier by the linear nature of habitat condi-
tions—pools, riffles, and raceways—in small
streams. Although the situation is more complicated
in large rivers with floodplains that may include
braided channels, log piles, islands, floodplain
lakes, and other side channel habitats, IBI has been
used successfully in large rivers as well (Hughes
and Gammon, 1987; Gammon et al. 1990; Ohio En-
viron, Prot. Agency, 1988; Rankin and Yoder, 1990).
Effective biomonitoring depends on a sampling
scheme that covers the local mosaic of habitats.
With fish sampling, coverage requires sample dis-
_tances that vary with stream size (Karr et al. 1986;
Ohio Environ. Prot. Agency, 1988). In contrast to
fish sampling, invertebrate sampling protocols
often specify sampling from riffle areas only (Plaf-
kin et al. 1989). However, some invertebrate special-
ists believe that sampling of invertebrates should
extend to at least three habitats in locations where
conservation issues predominate (Jenkins et al.
1984). Use of riffle-only collections has been chal-
lenged by Brooker (1984), and Cuff and Coleman
(1979) also preferred a multihabitat sampling de-
sign. Both fish and invertebrate-based indexes are
robust only so far as the sampling protocol specified
provides a sample that is representative of the local
biological community.
The Ohio Environmental Protection Agency
(1988) adopted a protocol that modifies fish sam-
pling methods according to stream size. This ap-
63
-------
J. R. KARR and M. DIONNE
proach calls for use of backpack electrofishing or
seining in small streams, small boat-mounted gen-
erators or land-based generators with long electric
lines for mid-sized streams, and electrofishing
equipment mounted in 12-foot to 16-foot boats for
larger rivers. Specific criteria were established for
each sampling method following detailed testing to
ensure that representative samples are collected in-
dependent of river size and sampling method.
The general IBI approach has now been used in
over 35 states, in several provinces of Canada, and
in France. In addition to state efforts (Ohio Environ.
Prot. Agency, 1988), a number of federal agencies
have adopted IBI and its derivatives as tools for as-
sessment of running-water resources (e.g., Plafkin et
al. 1989; Hunsaker and Carpenter, 1990; Hirsch et al
1988; Saylor and Scott, 1987).
Establishing biological criteria for lakes and res-
ervoirs presents several challenges. These include
1. gaining knowledge of the. environmental
factors that determine the characteristics of
the resident biotic community;
2. defining the attributes of relatively undis-
turbed systems, especially as a function of
basin size and morphometry, in much the
same way that stream communities vary
with stream size and valley (channel) type;
and
3. developing efficient and reliable metrics to
determine how and to what extent human
activities influence the structure of the sam-
pled communities.
Classifying Aquatic
Ecosystems
Gaining an understanding of the ecological dy-
namics that characterize the water resource system
is the,first step in planning a sampling program to
evaluate biological integrity. Two major classes of
aquatic ecosystems exist: lotic (flowing water) or
lentic (standing water); that is, streams and lakes.
In flowing water systems, stream morphometry
and size are the dominant variables of ecological
significance. At the scale of square decimeters, the
upper or lower, upstream or downstream surfaces
of rocks experience significant heterogeneity of
flow. Along a few hundred meters of stream chan-
nel, the habitat may alternate among three major
habitats—pools, raceways, and riffles. Finally, lon-
gitudinal changes occur as water flows, downhill
and streams fuse to form larger and larger rivers
(stream order). A similar gradation occurs within
standing water systems. Beginning with temporary
ponds, it extends to the deepest and longest lived
lakes. Basin shape and size influence the relative
areas of (1) wetlands with semiaquatic and emer-
gent macrophytes; (2) littoral zone with emergent
and submerged macrophytes; and (3) pelagic or
Open water areas.
On long time scales, surface waters are dynamic
as streams meander and lakes accumulate sedi-
ments. At shorter time scales, identifiable character-
istics can be associated with a community's position
along a lake or stream habitat gradient. A promi-
nent difference between streams and lakes is that
habitat gradients extend in every direction from a
lake's center, while stream habitats occur in linear
sequences as water flows downhill. Surface water
connections are present along the size gradient in
streams, while lakes of different size within a region
need not share such connections. The first steps in
designing an appropriate lake monitoring program
are to (1) define the position along the morphomet-
ric size gradient; (2) determine the biological attri-
butes to be sampled; and (3) select the techniques
that will ensure the collection of quality data, while
avoiding artifacts resulting from inconsistencies in
sampling methods.
Establishing Biological
Criteria for Lakes and
Reservoirs
Planning for ambient biological monitoring in
lakes and reservoirs requires integrating knowl-
edge of biological patterns within those systems
with understanding of the limitations of different
sampling techniques. Moreover, effective evalua-
tion of human impacts on these systems requires
some knowledge of how human actions might af-
fect those biological communities. ' ''"••''
Preliminary application of biological monitor-
ing to lake, reservoir, and even estuarine environ-
ments . has already provided useful insights
regarding monitoring programs (Greenfield and
Rogner, "1984; Dionne and Karr, in press; 'Miller,
1988). Although superficial similarities exist among
these types of water bodies, differences require
unique monitoring approaches.
Lakes
Lakes are natural environments, so reference data
are usually available for determining the expected
condition of biological communities. As many re-
64
-------
Biological Criteria: Research and Regulation, 1991
searchers have shown, lakes exhibit variation in
physical and biological characteristics regionally
and as a function of size and depth configuration,
pH, bibgeographic context, and other factors (John-
son et al. 1977; Tonn and Magnuson, 1982;
Matuszek and Beggs, 1988). Size and basin
morphometry are especially important because
they determine the extent of wetland, littoral, and
pelagic areas. Morphometry, productivity, and
presence of tributary streams influence winterkill
frequency, and thus duration of numerous aspects
of community structure. The complex of habitats
that are present affect the biotic community and
success of different sampling gears.
Because habitat heterogeneity is a strong corre-
late of basin morphometry, lake volume, depth, and
other factors embodied in the lake morphoedaphic
index are critical in determining the attributes of the
lake biota. Habitat heterogeneity in lakes contrib-
utes to ecosystem dynamics through the addition of
species complexes associated with each habitat; het-
erogeneity also supports taxa that depend on the
juxtaposition of two or more habitats.
Reservoirs
Examples of biological assessment of lentic waters
presented here come from recent applications with
reservoirs/because our experience in lakes is lim-
ited. Reservoirs differ from lakes in a number of
ways that can be attributed, directly or indirectly,
to control of flow regime. The primary uses of im-
pounded waters—hydropower, flood control, and
navigation (Counc. Environ. Qual. 1987; Voightlan-
der and Poppe, 1989)—all produce unnatural varia-
tion in flow rates and water levels that have major
impacts on the biota. Rapid, often extreme, changes
in water level create a barren "intertidal zone" in
place of the littoral vegetation and woody debris
that normally provide habitat structure in lakes
(Aggus, 1971; Groen and Schroeder, 1978).
The loss of littoral habitat structure also results
in the loss of the plant- and debris-colonizing inver-
tebrates that are important prey for many fish spe-
cies (June, 1976; Strange et al. 1982; Crowder and
Copper, 1982; Killgore et al. 1989; Schramm and
Jirka, 1989; O'Brien, 1990). Unprotected banks in
turn create turbid water conditions. High discharge
rates result in passage of great numbers of plank-
tonic organisms (including pelagic fish eggs and lar-
vae) out of the system (Cowell and Hudson, 1967;
Walburg, 1971). Patterns of discharge also influence
temperature and oxygen gradients both within and
downstream of reservoirs. As a result, the normal
seasonal stratification and turnover that typically
drive natural lake processes can be highly per-
turbed (Wunderlich, 1971; Cole and Hannan, 1990;
Ford, 1990). :
Because reservoirs are created by flooding river
floodplains and uplands rather than by filling a nat-
ural depression or basin, shorelines are often highly
dendritic, with a high ratio of shore length to water
volume in contrast to that of many lakes. This factor
contributes to high turbidity in reservoirs (Kirrimel
et al. 1990; Mafzolf, 1990; O'Brien, 1990), and may
also result in greater influence of the runoff associ-
ated with various human land uses. Permanent or
inconsistent flooding may destroy the tributary
spawning habitats often used by riverine fishes
(Richards et al. 1986; Walburg, 1977).
The contrast between reservoirs and lakes pro-
vides complementary systems for the development
of monitoring schemes. In some regkms, natural
and impounded lakes occur in the same biogeo-
graphic areas. In these situations natural lakes may
provide reference data for assessment of the overall
impact of flow management on reservoir health.
Recommendations for changes in water level man-
agement based on these comparisons could then be
implemented and reservoir response measured. By'
the same token, assessment of reservoir health by
comparison with healthy lakes can help identify
common impacts on lakes. For example, bank ero-
sion similar to that found in reservoirs can result
from lakeside forestry, agriculture, housing, urban
development, and recreation. Loss of lake habitat
structure through removal of aquatic plants and
woody debris to improve aesthetic or recreational
appeal also creates reservoir-like conditions.
Reservoir IBI
Initial efforts to develop an IBI for Tennessee Valley
Authority (TVA) reservoirs use a reservoir classifi-
cation based on fish communities (McDonough
and Barr, 1977) and reflect the geographic region,
elevation, size, and function of the impoundments;
The small, high-elevation Appalachian storage res-
ervoirs used for flood control in the Blue-Ridge
Mountains and in the Upper Holston River Valley
form two distinguishable classes or groups/The
lower elevation "Large Storage/Upper Mainstem"
(hydropower and either navigation or flood con-
trol) reservoirs form a third group. The "Lower
Mainstem" reservoirs (hydropower and naviga-
tion) extending westward for 400 river miles to the
confluence of the Tennessee and Ohio rivers form
the final group. ... :
The TVA has maintained a cove fdtenorie pro-
gram since the 1940s to sample fish (Dibrine arid
Karr, in press). The practice has been standardized
65
-------
J. R. KARR and M. DIONNE
in recent years to provide reliable quantitative data.
Coves are blocked off with nets, rotenone is dis-
persed throughout the cove, and fish are collected,
identified, counted, and weighed. The result is a de-
tailed record of species composition, abundances,
and biomass for the entire fish community. This
study is based on these samples for the initial explo-
ration of reservoir assessment procedures.
Metric Development
The selection process for reservoir metrics follows
the same concepts that guided the selection of IBJ
metrics for streams. To create an index that is sensi-
tive to many causes of degradation, metrics must
represent the extent to which fish communities di-
verge from an optimal condition. Efforts to reach
this goal began with analyses of the TVA's cove ro-
tenone data. The choice of metrics was based on
presumed important features of the population:
community and trophic structure of the reservoir
fishes (Table 1). At the population and community
levels, metrics were defined to measure the total
numbers of individuals (Metric 13) and species
(Metric 1), as well as important taxonomic (Metrics
2-3) and functional groups (Metrics 4-9). Some met-
rics are likely to change as the reservoir IBI is tested,
and refined. Species designations for tolerance, in-
tolerance, and trophic guild required for metrics 4-
8 were based on information provided by the TVA
(Saylor, 1990). Details of the metrics are described
in Dionne and Karr (in press).
Our selection of metrics allows evaluation of
the species richness, dominant taxa, population
structure, reproductive habits of reservoir fish, indi-
vidual biology, fish health, and community trophic
structure. For the stream IBI, most metrics are not
sensitive over the entire range of degradation (Karr
et al. 1986; Karr, 1991). For example, darters are
found only in streams of intermediate to high qual-
ity, so they cannot reflect different levels of degrada-
tion at the low end of the quality scale. Conversely,
fish with external signs of disease occur only in
highly degraded systems, so the proportion of dis-
' eased fish in streams with low and intermediate lev-
els of degradation will typically be 0 percent.
The final reservoir IBI will contain the set of
metrics from each category (Species Richness and
Composition, Trophic Composition, Fish Abun-
dance and Condition, and Reproductive Composi-
tion; Table 1) that most effectively assesses the
health of the ecosystem on a scale from very poor to
excellent. The effectiveness of the final index will be
a funttion of its biological sensitivity and its relative
cost.
The presence of shad is an important feature of
fish assemblages in reservoirs. Young-of-year (YOY)
gizzard shad (Dorosomo. cepedianuni) and threadfin
shad (Dorosoma petenense) often dominate both fish
numbers and biomass, frequently comprising well
above 50 percent of total fish biomass (Zeller and
Wyatt, 1967; Jenkins, 1967; Noble, 1981; Downey
and Toetz, 1983). In the Tennessee River mainstem
reservoirs, young-of-year bluegill can be equally
Table 1.—Preliminary metrics for reservoir index of biotic integrity based on cove rotenone sampling.
RATING CRITERIA
Species Richness and Composition
1. Total species number
2. Number of small cyprinid and darter species
3. Number of sucker species
4. Number of intolerant species
; 5. Percent individuals as tolerant species
Trophic Composition
6. Percent individuals as specialized benthic insectivores*
7. Percent individuals as omnivores*
8. Percent individuals as piscivores* '
9. Percent individuals as YOY shad and bluegill
10. Percent individuals as adult shad and bluegill
Reproductive Composition
11. Percent of species as plant and rock substrate spawners
12. Number of migratory spawning species
Fish Abundance and Individual Health
13. Total number of individuals'*
14. Fish health score (TVA)
YOY « Young-of-year.
'Excluding YOY shad and bluegill.
"Excluding shad and bluegill.
Source: Dionne and Karr, In press.
5
s=33
£ 6
£ 5
a 5
<27
>8.4
<27
' '£18
<33
<26
£ 14,000
FOREBAY
3
23-32
4-5
3-4
3-4
27-53
4.2-8.3
27-53
9-17
33-63
26-51
Will be
7-13,999
1
<23
,£ 3
£ 2
£ 2
' st54
£4.1
£54
< 9
s64
£52
included
<7,000
Data not availble
5
£29
a 5
£ 5
£ 4
<33
a 2
<27
£28
<40
<27
INFLOW
3
22-28
3-4
3-4
, 3
33-65
1
27-53
14-27
40-69
27-53 •
1
<22
£ 2
^ 2
£ 2
£66
• 0
£54
<14
, £70
£54
when data available.
£20,000
in historical
10-19,999
samples.
< 10,000
66
-------
Biological Criteria: Research and Regulation, 1991
important in cove fish assemblages, although they
do not share the dominant status of young-of-year
shad in open water. All three species depend on
zpoplankton as a food source at this stage in their
life cycles, although the shad can also consume phy-
toplankton (Scott and Grossman, 1973; DeVries and
Stein, 1991). These forage species comprise a large
proportion of the available prey for piscivores in
some reservoirs, but large populations do not con-
sistently lead to enhanced populations of their pred-
ators (Jenkins and Morais, 1978; Ziebell et al. 1986;
DeVries and Stein, 1991).
Given their great tendency to dominate reser-
voir standing stock, knowledge of shad ecology is
essential to understanding reservoir fish commu-
nity and trophic structure, as well as to recognizing
differences in the ecology of reservoirs and lakes.
Recent work on ecological interactions between
shad, bluegill, and largemouth bass in Ohio reser-
voirs containing one or the other shad species (DeV-
ries et al. 1991; DeVries and Stein, 1991) indicate that
bottom-up (competition between shad and
zooplankton for phytoplankton, competition be-
tween shad and bluegill for zooplankton) and top-
down processes (predation by largemouth bass on
bluegill) combine to determine reservoir commu-
nity structure.
The relative roles of these two processes is dic-
tated by the timing of spawning of shad and blue-
gill, which determines their sequence of appearance
as larvae in the limnetic zone to feed on plankton
food resources, and their subsequent growth and
survival. This same complex of ecological processes
most likely drives systems where both gizzard shad
and threadfin shad occur together. In these systems,
competition for zooplankton and phytoplankton
between the two shad species, size-selective preda-
tion on the two species by piscivores, and relatively
frequent winterkills of threadfin shad must be
added to the interactions observed by 'DeVries and
colleagues (DeVries and Stein, 1990,1991; DeVries et
al. 1991).
If little is known about the role of shad in reser-
voir community structure, even less is known about
their function in natural lakes. This may in part be
explained by the fact that few natural lakes exist
through much of these two species ranges, espe-
cially for the cold-sensitive threadfin shad. Never-
theless, the gizzard shad is found well into the
latitudes of natural glacial lakes, as far north as the
Great Lakes region of North America. In spite of
this, we are not aware of evidence that shad regu-
larly dominate fish standing stock in lakes. The po-
tentially important variations in reservoir and lake
ecosystems responsible for this contrast may be ex-
plained by differences in habitat structure on a scale
discernable to individual fish.
Physical structure in the form of aquatic vegeta-
tion or woody debris in the littoral zone is often
greatly reduced in reservoirs. This may lead to re-
duced success of piscivores (e.g., largemouth bass,
pike, muskellunge) known to associate with struc-
tured littoral habitat (Scott and Grossman, 1973;
Diana et al. 1977; Fish and Savitz, 1983; Chapman
and MacKay, 1984; Savino and Stein, 1989), which in
turn prevents this trophic level from achieving the
numbers required to exert top-down control of res-
ervoir shad populations. DeVries et al. (1991) sug-
gest that competition for zooplankton between
larval shad and bluegill in the limnetic zone of
Stonelick Reservoir led to low survival and limited
return migration of juvenile bluegill to the littoral
zone, and subsequent poor growth in their large-
mouth bass predators. In addition to these docu-
mented interactions in the limnetic zone, loss of
littoral benthic and plant-associated invertebrate
communities may have a negative influence on po-
tential shad predators and competitors.
The complex interactions between piscivores,
shad, and other planktivores and invertebrate-feed-
ing fish in natural lakes may well be mediated by
littoral habitat structure and associated invertebrate
productivity. Bluegill in lakes use vegetated littoral
habitats as a predator refuge, and use both the litto-
ral and open water as foraging habitats, depending
on their body size, piscivore density, and habitat-
specific foraging success (Mittelbach, 1981; Werner
et al. 1983a,b). Small shad forage on both phyto-
plankton and zooplankton in open water, and large
gizzard shad feed on detritus. The timing of spawn-
ing for all these species is influenced by the spring
rise in water temperature, and this can determine
the outcome of interactions between new year
classes (DeVries et al. 1991). Perhaps the lack of lit-
toral structure and associated invertebrate produc-
tivity in reservoirs, as well as the influence of
fluctuations in water level on water temperature
and available spawning habitat, all contribute to the
dynamics of shad, their predators, and their com-
petitors in these ecosystems.
Metric Rating
Defining a reference site is a major challenge in the
ecological monitoring of reservoirs. No objectively
defined, "healthy" (i.e., free from human distur-
bance) reference systems are available to provide
benchmark values for index metrics. As a class, res-
ervoirs differ in important ways from natural rivers
and lakes, precluding the use of natural surface wa-
ters as reference sites. In the Tennessee River sys-
tem, all reservoirs are subject to artificial
fluctuations in water level and flow regime, and
67
-------
J. R. KARR and M. DIONNE
many have been exposed to substantial point and
nonpoint sources of pollution, especially on the
mainstem. Thus, no set of "pristine" reservoirs is
available within the system to serve as undisturbed
reference sites. As understanding of Tennessee Val-
ley impoundments develops, it may be possible to
establish hypothetical reference values. For the
present our approach is to generate metric values
for samples from a number of ecologically similar
reservoirs, and use the range of values obtained to
assign high, intermediate, and low ratings (5, 3 and
1, respectively) to metric values from the study
samples. This approach (Fig. l).is modeled after the
stream IBI "maximum species richness line" devel-
oped by Fausch et al. (1984). To generate IBI scores
for a number of mainstem impoundments, data
from the "Lower Mainstem" reservoirs were
pooled. The scores are based on data collected since
1970 with standardized techniques on the lower
mainstem.
A number of metrics from cove fotenone sam-
ples exhibited trends according to distance of the
sample site from the dam. For example, the "total
number of individuals per hectare" metric declines
with distance from the dam in Chickamauga Reser-
voir (Dipnne and Karr, in press), reflecting the gra-
dient in physico-chemical characteristics from the
more lacustrine region near the dam to the more
riverine region upstream (Siler et al. 1986; Thornton,
1990). These regions are classified as forebay, inflow,
and an intermediate transition zone, defined as the
region where suspended sediment settles out of the
water column. To account for longitudinal heteroge-
neity, sites were rated based on their position within
the reservoir.
Implementing the Reservoir IBI
Information necessary for inclusion of the two re-
productive metrics and the fish health metric are
not yet available. Hence, these results are based on
only 11 of the 14 proposed reservoir IBI metrics. An
example of reservoir IBI scores is presented for a
niainstem impoundment, Wheeler Reservoir (Fig.
2). IBI scores are more variable from year-to-year
than is typical of stream systems, This variability
may stem from the sampling design, and may also
reflect a basic difference in temporal variation of
ecological parameters between streams and im-
pounded rivers, parallel to the trend in streams of
_
*G)
CJ
3
S
TJ
C
C3
•d
ctf
O
^
^J
c
u
U
o>
100-
90-
~
80-
70-
60-
50-
40-
30-
•
20-
.
10-
f
t * »• * •
^ ' A ft
0 • • 1
A ' *
•*•«.*:**
i ****** . . »*
* ^1 * ** **
^ ° t * 1 •- * -, ••* A •
* < * . § : t J * * i f * * * ! %
* t * r
* * « * * i *
t • * i • *
* • * , t
* * * » * * * * * 5
* • * o * * 1 * * *
. . . • ., '—. _ ^-^ $ . . . , . • r— — i 1 1
68 70 72 74 76 78 80 82 84 86 88 90
Year
Flflur* 1.—An Index of blotlc Integrity metric scatter plot for all coves sampled from 1970 In lower mainstem reservoirs of the
Tennessee River. The range of possible values for the "percent of Individuals as young-of-year (YOY) shad and blueglll" metric
It divided Into three equal Intervals. Metric values within these Intervals are rated as 1,3, or 5, representing low, Intermediate,
and high ecological health, respectively. Plots such as this were constructed for all reservoir metrics, and used to rate metrics
for Individual cove samples. (Dlonne and Karr, In press).
68
-------
Biological Criteria: Research and Regulation, 1991
greater variability for disturbed than undisturbed
sites (Karr et al. 1986; Rankin and Yoder, 1990). Per-
haps the repeated (annual) destructive sampling at
sites using rotenone produces perpetually unstable
"young" communities; that is, these samples may
represent transient assemblages analogous to an
early successional plant community maintained by
frequent disturbance. For these and other reasons
(Diorine arid Karr, in press), cove rotenone sam-
pling does not provide the best approach for moni-
toring the effects of increasing human use of
Tennessee River resources. '•'.•
TVA Reservoir Biomonitoring
Recognizing the need for more representative sam-
pling methods, TVA recently (since 1988) imple-
mented a fish community assessment program
using boat electroshocking, experimental gill nets,
and hydroacoustics in all. mainstern and selected
tributary reservoirs. The electroshocking sampling
regime consists of 10 timed runs (10 minutes each)
along shore in the forebay, transition zone, and in-
flow of each impoundment, for a total of 30 sam-
ples, collected each autumn. The 10 samples are
distributed among the major habitat types in each
area (rip rap, rock, bluff, gravel bank, mud bank,
submerged brush, and vegetation). Because elec-
troshocking is nondestructive, samples can be col-
lected along developed shoreline, and direct
assessment of the influence of local human distur-
bance on the aquatic resource is possible.
The IBI metrics developed from the cove rote-
none data should be readily modified for use with
the new electroshocking data. However, because
qualitative and quantitative differences exist be-
tween cove rotenone and electroshocking samples,
the IBI scores from the two sampling methods are
not directly comparable. Electroshocking samples
are biased because some fish species and size classes
are more effectively shocked than others. However,
the maximum water depth that can be sampled
without specialized equipment is less than 2 m. Fish
species with small body size and benthic habit are
the most seriously under-represented, followed by
the young-of-year of all species. Electroshocking
samples are less frequently dominated by young-of-
year shad because many sampling runs do not in-
tercept their schools. Finally, electroshocking data
based on timed runs (as opposed to measured runs
of a set distance) do not produce fish density esti-
mates, but only the relative abundance of fish spe-
cies. Thus, all metrics based upon fish densities
must be adapted for use with relative abundances.
For this reason, • researchers may be advised to
switch.to distance-based samples. Future metrics
may also include information from experimental
gill-netting (relative abundance of larger deep water
fish species) and hydroacoustics (offshore shad den-
sities). . .
As mentioned earlier, many reservoirs are char-
acterized by a long,'irregular shoreline composed of
a main river channel and numerous embayments
formed by flooding of tributary streams. Yet, little
progress has been made in understanding hydro-
cu
O
u
tt
55
50
45-:
40:
35-:
30:
25-.
20-.
15-
10
Wheeler Reservoir
69
71.
73
75
—i—
81
—i—
83
—i—
85
—i—
87
—i—
89
77 79
' , . ''. '-. . • - Year . •' • ' • : '.-'•• • , -'• '•••••''
Figure 2.—Temporal variation In Index of blotic Integrity (IBI) scores for coves sampled repeatedly with rotenone In Wheeler
69
-------
J. R. KARR and M. DIONNE
logical and ecological interactions between main
channel and embayments. Because a large share of
the human use of reservoir water resources occurs
in embayments, it is important to know whether the
impact of these uses remains localized, or extends
from embayments into the main channel.
In the mainstem of the Tennessee River, im-
poundments exhibit "plug" flow, where the water
of the main channel travels through the reservoir as
a cohesive unit. Marked differences in temperature,
chlorophyll, and turbidity levels often occur be-
tween embayments and the main channel (Baxter,
1977; Kennedy et al. 1982; Butkus, 1989). However,
little is known about the contribution of each em-
bayment to overall discharge, or about the active
and passive movements of invertebrates and verte-
brates between the main channel and embayments.
Mixing of embayment and channel water and biota
change with the seasonal hydrologic cycle. Are
there periods during the year when high tributary
discharge flushes water, animals, and sediments
(along with the existing pollutant load) into the
main channel? How often and under what condi-
tions do water, animals, and sediment from the
main channel back up into embayments?
These mixing processes influence the impact of
human use on overall reservoir health. Similarly,
knowledge of the interactions of wetland, littoral,
and pelagic zones in lakes is basic to understanding
how human activities affect lake or reservoir eco-
logical health.
Development of Lake IBI
An index designed to measure the biotic integrity
of lakes can be constructed according to the general
scheme outlined for reservoirs. Because the ecology
of a reservoir is in some ways like that of a dis-
turbed lake, the type of metrics included in a lake
IBI would be similar to those developed for reser-
voirs. However, the details of the lake IBI metrics
necessarily depend upon lake size and shape, taxa
of fish present, and other factors such as presence
of wetland and littoral vegetation, and chemical
composition. Lake and reservoir biomonitoring
would benefit from a comprehensive approach
combining major taxa such as fish, invertebrates,
and plankton. Unfortunately, teams assessing bio-
logical integrity rarely include a broad range of tax-
onomic perspectives. Rather, experts in each group
too frequently debate the merits of their taxon in-
stead of working cooperatively for more integra-
tive assessments.
As the reservoir IBI is developed and tested, ex-
tension of this approach to the assessment of natu-
ral lake systems would be useful. Much of what is
being learned about reservoirs can be applied di-
rectly to lakes, and what is learned about lakes
would provide an important perspective on water
resource conditions created by reservoir impound-
ment. Following impoundment, reservoirs typically
experience a rapid growth in production followed
by a decline in fishery and other biotic potential
(June, 1976; Kimmel and Groeger, 1986, O'Brien,
1990). IBI should be useful in tracking and identify-
ing the factors responsible for this cycle. Such infor-
mation would be instrumental in efforts to extend
the period of increased production by control of
water levels and flow. This is relevant because the
construction of new reservoirs continues (Counc.
Environ. Qual. 1987), and the future of existing res-
ervoirs is debated at the time of relicensing.
More generally, both natural and impounded
' lakes are increasingly affected by human activities.
For the most part, the biological consequences of
these impacts remain unquantified until they reach
critical levels. To determine an effective course of
water resource management, development of meth-
ods for monitoring the biological response of reser-
voirs and lakes to human influence is essential.
Only then is it possible to detect the consequences
of chronic impacts and the onset of degradation in
healthy systems; likewise, effective monitoring ap-
proaches are required to assess the response of de-
graded systems to management strategies.
References
Agguss, L.R. 1971. Summer benthos in newly flooded areas
of Beaver Reservoir during the second and third years
of filling 1965-1966. Spec. Publ. 8:219-31. Am. Fish. Soc.,
Washington, DC.
Baxter, R.M. 1977. Environmental effects of dams and im-
poundments. Annu. Rev. Ecol. Syst. 8:255-83.
Brooker, M.P. 1984. Biological surveillance in Welsh rivers for
water quality and conservation assessment. Pages 25-33
in D. Pascoe and R.W. Edwards, eds. Freshwater Biolog-
ical Monitoring. Pergamon Press, Oxford, England
Butkus, S.R. 1989. Reservoir Embayments as Potential Units
for Water Quality Management. TVA/WK/WQ 98/7.
Term. Valley Author., Chattanooga.
Cairns, J., Jr. 1974. Indicator species vs. the concept of com-
munity structure as an index of pollution. Water Resour.
Bull. 10:338-47.
Chapman, C.A. and W.C. MacKay. 1984. Direct observation
of habitat utilization by northern pike. Copeia 1984:255-
58.
Cole, T.M. and H.H. Hannan. 1990. Dissolved oxygen dy-
namics. Pages 71-108 in KW. Thornton, B.L. Kimmel,
and F.E. Payne, eds. Reservoir Limnology. John Wiley
and Sons, New York.
Council on Environmental Quality. 1987. Developed water
resources. Pages 89-147 in Environmental Quality: The
Annual Report of the Council on Environmental Qual-
ity. Washington, DC.
70
-------
Biological Criteria: Research and Regulation, tffffl
Cowell, B.C. and P.L. Hudson. 1967. Some environmental
factors influencing benthic invertebrates in two Mis-
souri River reservoirs. Pages 541-555 in Reservoir Fish-
ery Resources: Symposium. Reserv. Comm. Southern
Div., Am. Fish. Soc., Washington, DC.
Crowder, L.B. and WE. Cooper. 1982. Habitat structural
complexity and the interaction between bluegills and
their prey. Ecology 63:1802-13.
Cuff, W. and N. Coleman. 1979. Optimal survey design: les-
sons from a stratified random sample of macrobenthos.
J. Fish. Res. Board Can. 36:351-61.
Davis, W.S. ed. 1990. Proceedings of the 1990 Midwest Pollu-
tion Control Biologists Meeting. EPA-905/9-90-005. Re-
gion V, Environ. Sci. Div., U.S. Environ. Prot. Agency,
Chicago, IL.
DeVries, D.R. and R.A. Stein. 1990. Manipulating shad to en-
hance sport fisheries in North America: an assessment.
N. Am. J. Fish. Manage. 10:209-23.
DeVries, D.R. and R.A. Stein. 1991. Complex interactions
among fish, zooplankton, and phytoplankton as influ-
enced by an open-water planktivore. Mss.
DeVries, D.R., R.A. Stein, J.G. Miner and G.G. Mittelbach.
1991. Threadfin shad as supplementary forage: conse-
quences for young-bf-year fishes. Trans. Am. Fish. Soc.
In press.
Diana, J.S., W.S. MacKay, and M. Ehrman. 1977. Movements
and habitat preference of northern pike (Esox lucius) in
Lac St. Anne, Alberta, Trans. Am. Fish. Soc. 106:560-65.
Dionne, M. and J.R. Karr. In press. Ecological monitoring of
Tennessee River reservoirs. Ecological Indicators. Proc.
Int. Symp. Ecological Indicators. Elsevier Applied Sci-
ence, Essex, England.
Downey, P. and D. Toetz. 1983. Distribution of larval gizzard
shad (Dorosoma cepidianum) in Lake Carl Blackwell,
Oklahoma. Am. Midland Nat. 109:23-33.
Fausch, K.D., J.R. Karr, and P.R. Yant. 1984. Regional applica-
tion of an index of biotic integrity based on stream-fish
communities. Trans. Am. Fish. Soc. 113:39-55.
Fish, P.A. and J. Savitz. 1983. Variation in home ranges of
largemouth bass, yellow perch, bluegills, and pumpkin-
seeds in an Illinois lake. Trans. Am. Fish. Soc. 112:147-53.
Ford, D.E. 1990. Reservoir transport processes. Pages 15-42 in
K. W. Thornton, B. L. Kimmel, and F. E. Payne, eds. Res-
ervoir Limnology. John Wiley and Sons, New York
Gammon, J.R., C.W. Gammon, and M.K. Schmid. 1990. Land
use influence on fish communities in Central Indiana
streams. Pages 111-120 in W. S. Davis, ed. Proc. 1990
Midwest PoUut. Control Biologist Meet. EPA-905-9-
. 90/005. Region V, Environ. Sci. Div., U.S. Environ. Prot.
Agency, Chicago, IL.
General Accounting Office. 1987. The Nation's Water: Key
Unanswered Questions about the Quality of Rivers and
Streams. GAO/PEMD-86-6. Washington, DC.
Goede, R.W. 1988. Fish health/condition assessment proce-
dures. Utah Div. Wildl. Resour. Fish., Exp. Sta. Logan,
Greenfield, D.W. and J.D. Rogner. 1984. An assessment of the
fish fauna of Lake Calumet and its adjacent wetlands,
Chicago, Illinois: past, present, and future. Trans. 111.
Acad. Sci. 77: 77-93.
Groen, C.L. and T.A. Schroeder. 1978. Effects of water level
management on walleye and other coolwater fishes in
Kansas reservoirs. Am. Fish. Soc. Spec. Publ. 11:278-83.
Hirsch, R.M., W.M. Alley, and W.G. Wilber. 1988. Concepts
for a national water-quality assessment program. Circu-
lar 1021. U.S. Geol. Surv., Denver, CO.
Hughes, R.M. and J.R. Gammon. 1987. Longitudinal changes
in fish assemblages and water quality in the Willamette
River, Oregon. Trans. Am. Fish. Soc. 116:196-209.
Hunsaker, C.T. and D.E. Carpenter, eds. 1990. Ecological in-
dicators for environmental monitoring and assessment
program. EPA-600/3-90/060. Off. Res. Dev., U.S. Envi-
ron. Prot. Agency, Research Triangle Park, NC.
Hutchinson, G.E. 1967. A Treatise on Limnology. II. Introduc-
tion to Lake Biology and the Limnoplankton. J. Wiley
and Sons, New York.
Jenkins, R.M. 1967. The influence of some environmental fac-
tors on standing crop and harvest of fishes in U.S. reser-
voirs. Pages 298-312 in Reserv. Fish. Resour.: Symp.
Southern Div. Am. Fish. Soc., Washington, DC.
Jenkins, R.A., K.R. Wade, and E. Pugh. 1984.
Macroinvertebrates-habitat relationships in the River
Teifi catchment and the significance to conservation.
Freshw. Biol. 14:23-42.
Jenkins, R.M., and D.I. Morais. 1978. Prey-predator relations
in the predator-stocking-evaluation reservoirs. Proc.
Southeastern Ass. Fish Wildl. Agencies 30:141-57.
Johnson, M.G., J.H. Leach, C.K. Minns, and C.H. Olvef. 1977.
Limnological characteristics of Ontario lakes in relation
to associations of walleye (Stizostedion vitrewn), north-
ern pike (Esox lucius), lake trout (Salvelinus namaycush),
and smallmouth bass (Micropterus dolomieui). J. Fish.
Res. Board Can. 34:1592-1601.
June, F.C. 1976. Changes in young-of-the-year fish stocks
during and after filling of Lake Oahe, an upper Missouri
River storage reservoir. Tech. Pap. 87. Fish Wildl. Serv.,
Washington, DC.
Karr, J.R. 1981. Assessment of biotic integrity using fish com-
munities. Fisheries (Bethesda) 6(6)31-27.
. 1987. Biological monitoring and environmental as-
sessment: a conceptual framework. Environ. Monitor.
11:249-256.
. 1991. Biological integrity: a long-neglected aspect of
water resource management. Ecol: Applica. 1:66-84.
Karr, J.R. and D.R. Dudley. 1981. Ecological perspective on
water quality goals. Environ. Manage. 5:55-68.
Karr, J.R., M. Dionne, and I.J. Schlosser. In press. Bottom-up
vs. top- down regulation of vertebrate populations: les-
son from birds and fish. In M.D, Hunter, T. Ohgushi,
and P.W. Price/eds. Effects of Resource Distribution on
Animal-plant Interactions. Academic Press, Orlando,
FL. •
Karr, J.R., K.D. Fausch, P.L. Angermeier, P.R. Yant, and I.J.
Schlosser. 1986. Assessing biological integrity in run-
ning waters: a method and its rationale. Spec. Publ. 5. Il-
linois Nat. Hist. Surv., Urbana.
Kennedy, R.H., K.W. Thornton, and R.C. Gunkel. 1982. The
establishment of water quality gradients in reservoirs.
Can. Water Resour. J. 7:71-87.
Killgore, K.J., R.P. Morgan, and N.B. Rybicki. 1989. Distribu-
tion and abundance of fishes associated with sub-
mersed aquatic plants in the Potomac River. N. Am. J.
Fish. Manage. 9:101-11.
Kimmel, B.L. and A.W. Groeger. 1986. Limnological and eco-
logical changes associated with reservoir aging. Pages
103-109 in G.E. Hall and M.J. Van Den. Avyle, eds. Reser-
voir Fisheries Management: Strategies for the 80's.
Reserv. Comm., Southern Div. Am. Fish. Soc., Bethesda,
MD. • . " .
Kimmel, B.L., O.T. Lind, and L.J. Paulson. 1990. Reservoir
primary production. Pages 133-193 in K. W. Thornton,
B. L. Kimmel, and F. E. Payne, eds. Reservoir Limnol-
ogy. John Wiley and Sons, New York ,
Kolkwitz, R. and M. Marsson. 1908. Okplogie der
pflanzlichen Saprobien. Berichte der Deutschen
Botanischen Gesellschaft 26:505-19.
Marzolf, G.R. 1990. Reservoirs as environments for
zooplankton. Pages 195-208 in K.W. Thornton, B.L. Kim-
71
-------
J. R. KARR and M. DIONNE
mel, and F.E. Payne, eds. Reservoir Limnology. John
Wiley and Sons, New York. . , ' t.
Matuszek, J.E. and G.L. Beggs. 1988. Fish species richness in
relation to lake area, pH, and other abiotic factors in On-
tario lakes. Can J. Fish. Aquat. Sci. 45:1931-41.
McDonough, T. A. and W.C. Barr. 1977. An analysis of fish as-
sociations in Tennessee and Cumberland drainage im-
poundments. Proc. Southeastern Ass. Fish Wildl.
Agencies 31:555-63.
Miller, D.L. et al. 1988. Regional applications of an index of
biotic integrity for use in water resource management.
Fisheries (Bethesda) 13(5): 12-20.
Mittelbach, G.G. 1981. Foraging efficiency and body size: a
study of optimal diet and habitat use by bluegills. Ecol-
ogy 62:1370-1406.
National Research Council. 1987. National water quality
monitoring and assessment. Nad. Acad. Press, Wash-
ington, DC.
Noble, R.L. 1981. Management of forage fishes in impound-
ments of the southern United States. Trans. Am. Fish.
Soc. 110:738-50.
Northcote, T.G. 1988. Fish in the structure and function of
freshwater ecosystems: a "top-down" view. Can. J. Fish.
Aquat. Sci. 45:361-79.
O'Brien, W.J. 1990. Perspectives on fish in reservoir limnol-
ogy. Pages 209-225 in K.W. Thornton, B.L. Kimmel, and
F.E Payne, eds. Reservoir Limnology. John Wiley and
Sons, New York.
Oberdorff, T. and R.M. Hughes. 1991. Modification of an
index of biotic integrity based on fish assemblages to
characterize rivers of the Seine-Normadie Basin, France.
Hydrobiologie. In press.
Ohio Environmental Protection Agency. 1988. Users manual
for biological field assessment of Ohio surface waters. 3
Vok. Div. Water Qual. Monitor. Assess. Surface Water
Sec., Columbus, OH.
Omemik, J.M. 1987. Ecoregions of the conterminous United
States. Ann. Ass. Am. Geographers 77:118-25.
Patrick, R. 1950. Biological measure of stream conditions.
Sewage Ind. Wastes 22:926-38.
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/444/4-89-001. U.S. Environ. Prot. Agency,
Washington, DC.
Rankin, ET. and C.O. Yoder. 1990. The nature of sampling
variability in the index of biotic integrity (IBI) in Ohio
streams. Pages 9-18 in W. S. Davis, ed. Proc. 1990 Mid-
west Pollut. Control Biologists Meet. EPA 905/9-90-005.
Environ. Sci. Div., U.S. Environ. Prot. Agency, Chicago,
IL.
Richards, K.R., R.J. Dent, Jr., and W.H. Dieffenbach. 1986.
Fisheries problems associated with the1 Truman Dam
pumped storage hydroelectric project in West Central
Missouri. Pages 247-254 in G.E. Hall and M.J. Van den
Avyle, eds. Reservoir Fisheries Management: Strategies
for the 80's. Reserv. Comm., Southern Div. Am. Fish.
Soc., Bethesda, MD.
Richardson, R.E. 1928. The bottom fauna of the Middle Illi-
nois River, 1913-1925: its distribution, abundance, valu-
ation, and index value in the study of stream pollution.
Bull. Illinois Nat. Hist. Surv. 17:387-472.
Savino, J.F., and R.A. Stein. 1989. Behavior of fish predators
and their prey: habitat choice between open water and
dense vegetation. Environ. Biol. Fishes 24:287-93.
Saylor, C. 1990. Personal communication. Tennessee Valley
Authority, Norris.
Saylor, C. and E.M. Scott, Jr. 1987. Application of the index of
biotic integrity to. existing : ,TVA data. (
' TVA'/ONRED/AWR 87/32. Tenn. Valley Author., Chat-
tanooga.
Schramm,, H.L., Jr. and , K. Jirka. 1989. ..Epiphytic
rhacroinvertebrates as a food resource for bluegills in
Florida lakes. Trans. Am. Fish. Soc. 118:416-26.
Scott, W.B, and E.J. Grossman. 1973. Freshwater fishes of
Canada. Bull. Fish. Res. Board Can. 184:1-966.
Siler, J.R., W.J. Foris and M.C. Mclnerny. 1986. Spatial hetero-
geneity in fish parameters within a reservoir. Pages 122-
136 in G.E. Hall and M.J. Van Den Avyle, eds. Reservoir
Fisheries Management: Strategies for the 80's. Reserv.
Comm. Southern Div. Am. Fish. Soc., Bethesda, MD.
Steedman, R.J. 1988. Modification and assessment of an
index of biotic integrity to quantify stream quality in
southern Ontario. Can. J. Fish. Aquat. Sci. 45:492-501.
Strange, J.R., W.B. Kittrell, and t.D. Broadbent. 1982. Effects
of seeding reservoir fluctuation zones on young-of-the-
year black bass and associated species. N. Am. J. Fish.
Manage. 2:307-15.
Thornton, K.W. 1990. Sedimentary processes. Pages 43-70 in
K.W. Thornton, B.L. Kimmel and F.E. Payne, eds. Reser-
voir Limnology. Wiley-Interscience, New York.
Tonn, W.M. and J.J. Magnuson. 1982. Patterns in the species
composition and richness of fish assemblages in north-
ern Wisconsin lakes. Ecology 63:1149-66.
U.S. Environmental Protection Agency. 1987. Surface Water
Monitoring: a Framework for Change. Off. Water, Off.
Policy Plann. Eval., Washington, DC.
. 1988a. WQS draft framework for the water quality
standards program. Draft 11-8-88. Off. Water, Washing-
ton, DC.
. 1988b. Future risk: research strategies for the 1990s.
SAB-EC-88-040. Science Advisory Board, Washington,
DC.
. 1989. Water Quality Standards for the 21st Century.
r Off. Water, Washington, DC. ,
1990a. Biological Criteria: National Program Guid-
ance for Surface Waters. EPA-440/5-90-004.Off. Water
Reg. Stand., Washington, DC. }:
Voigtlander, C.W. and W.L. Poppe. 1989. The Tennessee
River. Can. Spec. Publ. Fish. Aquat. Sci. 106:372-84.
Walburg, C.H. 1971. Loss of young fish in reservoir, discharge
and year- class survival, Lewis and Clark Lake, Mis-
souri River. Am. Fish. Soc. Spec. Publ. 8:441-48.
. 1977. Lake Francis Case, a Missouri River reservoir:
changes in the fish population in 1954-1975, and sugges-
tions for management. Tech. Pap. 95. U.S. Fish Wildl.
Serv. Dep. Inter., Washington, DC.
Werner, E.E., G.G. Mittelbach, D.J. Hall, and J.E Gilliam.
1983a. Experimental tests of optimal habitat use in fish:
the role of relative habitat profitablility. Ecology
64:1525-39.
Werner, E.E., J.F. Gilliam, D.J. Hall, and G.G. Mittelbach.
1983b. An experimental test of the effects of predation
risk on habitat use in fish. Ecology 64:1540-48.
Wetzel, R.G. 1983. Limnology. Saunders, New York;
Wunderlich, W.O. 1971. The dynamics of density-stratified
reservoirs. Am. Fish. Soc. Spec. Publ. 8:219-31.
Zeller, H.D. and H.N. Wyatt. 1967. Selective shad removal in
southern reservoirs. Pages 405-414 in Reservoir Fishery
Resources: Symposium. Reserv. Comm. Southern Div.
Am. Fish. Soc., Athens, GA.
Ziebell, CD., J.C. Tash, and R.L. Barefield. 1986. Impact of
threadfin shad on macrocrustacean zooplankton in two
Arizona lakes. J. Freshw. Ecol. 3:399-406. ,
72
-------
Biological Criteria: Research and Regulation, 1991
Fish Assemblages as Indicators of
Environmental Quality in Chesapeake Bay
Stephen J. Jordan
Maryland Department of Natural Resources
Chesapeake Bay Research and Monitoring Division
Annapolis, Maryland
Pauline A. Vaas
School of Forestry and Environmental Science
Duke University
Durham, North Carolina
James Uphoff
Maryland Department of Natural Resources
Chesapeake Bay Research and Monitoring Division
Annapolis, Maryland
ABSTRACT
Standard indicators of ecological integrity have not been developed for general use in estuaries.
One promising indicator is based on fish assemblages, which can be sampled, identified, and
interpreted with relative ease and moderate cost. By focusing on multispecies analysis, it may
be possible to diminish the problems associated with great temporal variations in the occur-
rence of individual species, and to provide more comprehensive ecological information. An
Index of Biotic Integrity (IBI) developed from long-term beach seine survey records is applica-
ble to a wide range of salinity. The index appears responsive to temporal and spatial patterns of
contaminant loads and water quality. Long-term trends and covariation in the relative abun-
dance of 19 species of fish that are captured consistently in seine surveys were analyzed, in com-
bination with established management goals for water quality and fisheries to examine the
potential of forecasting ecosystem recovery in the bay. Sampling was also expanded to several
previously unmonitored tidal tributaries to further evaluate the IBI. The project examined a va-
riety of approaches to clear interpretation, presentation, and use of fish assemblage information
in environmental management of estuaries. .
Introduction
The value of biological indicators and ecological
measures of environmental quality is well-estab-
lished, both conceptually and in practice (Karr,
1987; U.S. Environ. Prot. Agency, 1990). In nontidal
aquatic systems, community properties such as di-
versity, relative abundance of functional or trophic
groups, numerical dominance, and so forth, long
have been used as evidence for anthropogenic deg-
radation or its absence. Until recent years, the bulk
of this type of analysis was confined to communi-
ties of invertebrates and microorganisms.
In the past decade, fish communities have been
recognized as excellent indicators of environmental
quality in freshwater streams, to the point where at
least one fish community measure, the Index of Bi-
otic Integrity, or IBI (Karr, 1981, 1987; Fausch et al.
1984; Angermeier and Karr, 1986) has been incorpo-
rated into some State water quality regulatory pro-
73
-------
S. J. JORDAN, P. A. VAAS, and J. UPHOFF
grams (U.S. Environ. Prot. Agency, 1990). In estuar-
ies and marine waters, much of the effort in sam-
pling biotic communities has been devoted to
plankton and benthic invertebrates, and a large part
of the analysis appears to have been devoted to un-
derstanding processes, or to comparative descrip-
tions. Apparently, there has not been a widespread
effort to evaluate community-based measures, espe-
cially of fish, and to apply them to questions of an-
thropogenic impacts in tidal waters.
Recognition that estuaries and coastal waters
have suffered degradation of water quality, habitat
loss, and concurrent losses of fisheries and other
aquatic resources, has resulted in greatly increased
attention to understanding and managing these sys-
tems. National and regional research and manage-
ment programs have been initiated with the goal of
restoring the quality of large bodies of water such as
Chesapeake Bay, and protecting living aquatic re-
sources from pollution, habitat destruction, and
overharvesting. Biologically based indicators will
be essential to these programs for (1) determining
priority areas for management, (2) measuring the ef-
fectiveness of management actions and progress to-
ward restoration goals, and (3) predicting the
ecological consequences of management scenarios
(Karr, 1987). Although States are required by Fed-
eral law to incorporate biological criteria into their
water quality standards, a guidance document for
estuaries probably will not be published before 1995
(U.S. Environ. Prot. Agency, 1990).
Clearly stated estuarine biological criteria are
obviously essential to carry out regulatory man-
dates. It would also be quite useful to represent the
condition of these complex ecosystems by means of
a composite index or simple graphics, so that man-
agers and nonspecialists can readily evaluate and
compare information, establish goals, and set priori-
ties for remediation. The problem is to develop con-
cise, understandable statistics that also are
ecologically meaningful, representative, reproduci-
ble, and can be generated routinely without massive
investments in data collection. For estuaries and
coastal waters, which are open systems with large
physical, chemical, and biological gradients on vari-
ous spatial and temporal scales, this is a difficult
challenge.
Biological indicators of water quality have been
developed based on fish assemblages of northern
Chesapeake Bay. Goals for the use of these indica-
tors are: (1) to identify ecological degradation in
specific areas of the bay and its tidal tributaries, and
(2) to focus environmental management programs
on specific areas based on validated ecological con-
cern.
It is hoped that this project will be helpful in es-
tablishing numerical biological criteria for estuaries.
Specific goals are to: (1) explore the potential of fish
assemblages as biological indicators in tidal waters
of Chesapeake Bay and tributaries, (2) develop and
test rapid, cost-effective techniques for biological
assessments in the estuary, (3) define the reference
condition and regional scales over which direct
comparisons can be made; (4) develop techniques
for identifying habitat factors associated with deg-
radation; and (5) develop techniques for predicting
multispecies responses to attainment of water qual-
ity and ecological restoration goals.
Methods
The area for this study is the tidal tributaries of
Maryland's Chesapeake Bay (Fig. 1). The Chesa-
peake Bay is the focus for a multijurisdictional res-
toration and protection program dedicated to
comprehensive management of the estuary and its
living resources (Chesapeake Exec. Counc. 1989).
the Chesapeake watershed supports a full range of
land uses, including large urban areas, intensive
agriculture, and extensive forests. Major concerns
for the condition of the estuary include rapid popu-
lation growth and development, eutrophication,
and declining fisheries.
Salinity in northern Chesapeake Bay ranges
from freshwater (< 0.5 ppt) at the head of the bay
and in the upper tributaries to about 20 ppt near the
Virginia border. The mainstream of the bay and the
major tributaries include large, tidal, fresh oligohal-
ine (0.5-5.0 ppt) and mesohaline (5.0-18 ppt) areas.
Sampling programs are conducted in all of these sa-
linity zones.
Data were analyzed from two somewhat differ-
ent kinds of habitats: (1) major tributary areas
(Nanticoke, Choptank, and Potomac Rivers and the
head of the bay) sampled by the Estuarine Juvenile
Finfish Survey (EJFS) (Cosden and Schaefer, 1988)
and (2) small subestuaries (Severn, South, Magothy,
and Wicomico Rivers) sampled by the Small Tribu-
tary Finfish Monitoring Project (STFMP). Both sam-
pling programs are conducted by the Maryland
Department of Natural Resources.
The EJFS began in 1954; sampling methods and
most of the sites have been consistent since 1958.
The sampling gear is a 30.5 m x 1.2 m, 6.35 mm
mesh bagless seine. Two seine hauls are taken at
each of 22 fixed, permanent stations (Fig. 1) in the
major tributaries once a month during July, August,
and September. All fish captured are identified and
counted. The EJFS was designed to measure striped
bass (Morone saxcitilis) recruitment, but over the long
term, the survey has provided consistent data on at
74
-------
Biological Criteria: Research and Regulation, 1991
EJFS SITES
STFMP SITES
Figure 1.—Sampling locations In northern Chesapeake Bay. EJFS: Estuarlne Juvenile Finflsh Survey. STFMP: Small Tributary
Flnflsh Monitoring Project Five stations, evenly spaced along the tributary axis, are monitored In each of the STFMP tributar-
ies.
75
-------
S. J. JORDAN, P. A. VAAS. and J. UPHOFF
least 19 species of fish. Salinity, temperature, and
physical habitat characteristics are recorded at each
site during each sampling period.
The STFMP employs the same methods as the
EJFS (30.5 m beach seine, fixed stations) with the ad-
dition of midwater and bottom trawls (in channel
areas) and near-shore bottom trawls at each station.
Sampling areas are shown in Figure 1; five evenly
spaced stations have been established along the axis
of each tributary from its mouth to near the head of
tide. Bottom nets are box trawls with a 3.05 m head-
rope; bodies and cod ends of midwater trawls have
a 1.53 x 1.53 m square opening; both nets have 12.7
mm stretch mesh. All fish captured are identified
and counted and the number of fish with obvious
physical anomalies (lesions, damage, parasites, and
so forth) is recorded. Temperature, salinity, pH, and
dissolved oxygen are measured at surface, midwa-
ter, and bottom depths of the channel at each sta-
tion; physical habitat characteristics are recorded at
each site. A detailed habitat assessment protocol is
under development for the STFMP
The data collected have been analyzed to pro-
duce five types of information: (1) development of a
prototype IBI, (2) analysis of long-term trends in
species abundance and the IBI, (3) interpretation of
interspecies covariation in terms of groups of spe-
cies with similar sensitivities to water quality and
habitat conditions, (4) graphic community "snap-
shots" at decade-long intervals, and (5) the potential
of predicting community composition from knowl-
edge of long-term trends and established manage-
ment goals. •"'•:•
Index of Biotic Integrity
A prototype IBI, comprised of 12 metrics describing
the fish community was developed from a 23-year
subset of the EJFS data. Rather than defining a ref-
erence station, the reference condition was taken to
be the upper third of the long-term distributions of
each metric from seven stations in the Potomac
River estuary, after adjusting for the effects of salin-
ity. Salinity adjustments made for metrics that cor-
related significantly with salinity were made by
removing regression equations from the data and
scoring the distributions of residuals. Fish abun-
dance data were summed over each year's sam-
pling period (three "rounds") before computing
the IBI. The IBI was evaluated for long-term trends
at all 22 permanent EJFS stations, and tested by
comparing STFMP data from a tributary with a rel-
atively undeveloped watershed (Wicomico River)
to data from a more degraded tributary (Severn
River).
Long-term Trends
In addition to evaluating trends in the IBI, the
study examined 30-year trends in relative abun-
dance of 19 species of fish caught routinely (i.e.,
there were few seine samples that failed to capture
at least one individual of the species) by the EJFS.
Linear, quadratic, and cubic regressions of log-
transformed annual mean catch per unit effort
were computed against time for the whple upper
bay area. The focus was on long-term trends over a
large geographical area, and in similarities in tenv
poral patterns in the abundance of species groups.
The expectation was that common trends among
species would reflect common responses to envi-
ronmental changes related to anthropogenic
stresses.
Interspecies Covariation
In addition to evaluating common trends, correla-
tion and cluster analysis were used to estimate sim-
ilarities among species. The working hypothesis
was that the 19 species would be grouped into a
few categories that would reflect, qualitatively, dif-
ferent types of tolerances. It was also expected that
strong correlations between species would reflect
common responses to future changes and provide a
basis for predictions.
Graphic Analysis
Histograms and star charts were used to illustrate
community composition at different points in time.
For example, three-year means of logio catch per
seine haul for the 19 common species from the EJFS
were graphed as histograms. For the reference pe-
riod (1958-60), the species were arrayed from left to
right in order of declining abundance. Later peri-
ods were shown in the same way, except that the
reference period species order was maintained.
This technique was also used to compare degraded
areas to a reference area. Star charts were used to
display essentially the same information as the his-
tograms, except that species were color coded for
greater visual impact.
Prediction
The project explored the potential for predicting
community composition from knowledge of
trends, species covariation, and management goals.
Long-term management goals have been estab-
lished for two of the species in the analysis: striped
bass and American shad (Alosu sapidissima). Attain-
76
-------
Biological Criteria: Research and Regulation, 1991
ment of these goals was assumed, and a combina-
tion of interspecies linear regressions and interven-
tion analysis was used to develop a view of the fish
community 10 years hence,
Results
A combination of two variants of the prototype IBI
performed reasonably well over all of the perma-
nent EJFS stations. It was necessary to substitute
two metrics in the IBI for the freshest tidal areas
(salinity < 2 ppt) because of the near absence of ma-
rine- and estuarine-spawning fish. Otherwise, the
salinity adjustment approach appeared to work
well. The IBI showed significant difference be-
tween the Wicomico and Severn Rivers (Fig. 2).
Nine of 12 IBI metrics received higher scores in the
Wicomico River "reference" tributary. The great
variability of dissolved oxygen in the Severn River
(Fig. 2) probably is symptomatic of the tributary's
water quality problems, and may be related di-
rectly to the poorer fish community and low IBI.
The IBI showed significant long-term trends at sev-
eral EJFS stations; these trends appeared to be con-
sistent with our knowledge of pollution loads over
this period; however, quantitative information on
loads generally was not available.
Significant 30-year trends were found for 15 of
the 19 common EJFS species. Examples of trend pat-
terns are shown in Figure 3. Pairwise species corre-
lations and cluster analysis also identified groups of
species that had similar or inverse patterns of abun-
dance. Anadromous and, semianadromous species.
(striped bass, white perch (Morone americano), ale-
wife (Alosa pseudoharengus), blueback herring (Alosa
aestivalis), and American shad) were grouped to-
gether strongly both by clustering and correlation.
This group is characterized by similar life histories
and sensitive early life stages, which are spawned in
fresh, or nearly fresh water. The juveniles of these
species (the life stage captured in these surveys) ap-
pear, to be diagnostic of less degraded conditions.
More anadromous species and more individuals of
these species are found in the Wicomico River (the
STFMP reference tributary) than in the other STFMP
tributaries.
A second strong grouping, based upon correla-
tion analysis, included gizzard. shad (Dorosoma
cepedianum), a freshwater plankton feeder, menha-
den (Brevoortia tyrannus), a marine-spawning plank-
ton feeder, mummichog (Fundulus heteroditus), an
estuarine resident species, and spot (Leiostomus
xanthurus), a marine-spawning benthic feeder.
Trend and cluster analysis grouped these latter spe-
cies somewhat differently, but at least two of these
14
12
10
X
I 2
WICOMICO RIVER
I
I
60
36 £
24
5 7 ' 11 13 15 17 19 21
KILOMETERS FROM RIVER MOUTH
1.2
B 12
E
z 10
x
O
a
UJ
>
SEVERN RIVER
60
48
36 i
24
12
IBI
5 7 9 11 13 15 17 19 21
KILOMETERS FROM RIVER MOUTH
DISSOLVED OXYGEN
Figure 2.—Prototype Index of Blotlc Integrity and dissolved
oxygen (circles are means and bars are ranges) at STFMP
stations In the Wicomico and Severn Rivers. Data are from
the summer of 1989. .• ' '
species occurred in the same group in all analyses.
These species also correlated negatively with spe-
cies in the anadromous group.
Graphic analysis of community structure at
three periods in time showed great differences be-
tween the 1958-60 reference period and later de-
cades (Fig. 4). The 1978-80 pattern appeared to be
the most disrupted. The forecast of the community
in the year 2000 (Fig. 4) showed a return to a condi-
tion more similar to the reference period than the
1978-80 or the 1987-89 periods.
Discussion
The results presented here are preliminary, and are
intended only to show the potential of fish commu-
nity analysis for representing temporal and spatial
patterns of the quality of estuarine habitats. Con-
77
-------
S. J. JORDAN, P. A. VMS, and J. UPHOFF
0.45
g 0.39
< 0.33
I 0.27
* 0.21
O
3j 0.09
aos
- A
MUMMICHOG
55
60
65
85 90
1.3
LU
I "
g 0.9
=
< 0.7
< 0.3
0.1
1.0
O 0.8
i
== 0.6
ATLANTIC MENHADEN
55
60
65
70 75
YEAR
80
85
90
S "•«
<
0.2
0.0
BAY ANCHOVY
55
60
65
70 75
YEAR
SO
85
90
Figure 3.—Examples of trend patterns for common species
captured by the EJFS. The "average abundance" Is the Bay-
wide (22 stations) annual mean of logio catch per seine
haul. A- tolerant species; B - sensitive (anadromous) spe-
cies; C - pattern that may Indicate eutrophlcatlon trends; D
- species with no trend.
elusions to date support the idea that sampling fish
with methods commonly used in Chesapeake Bay
will provide very useful data for establishing estu-
arine biological criteria, for monitoring, and for set-
ting management priorities.
The prototype IBI needs additional evaluation
arid testing, although the general approach appears
to be sound. The data and field experience gained
over the past two years of the STFMP will be used
to fine-tune the IBI.
The analysis of species covariation points to at
least two groups of species that appear to be indica-
tive of water quality conditions. The anadromous
group is indicative of good water quality, while the
menhaden-mummichog-spot-gizzard shad group
appears to represent pollution-tolerant species. The
similar long-term trends in this latter group suggest
that they have functioned as opportunists in Chesa-
peake Bay, perhaps filling niches vacated by more
sensitive species that have declined in response to
declining water quality and overfishing.
The presence, absence, or abundance of indica-
tor species often has been used in water quality as-
sessments to simplify sampling and interpretation.
This analysis suggests a few indicator species. The
species with the largest number of significant asso-
ciations with other species (correlation analysis)
were the blueback herring (anadromous group) and
the menhaden (tolerant group). It would be better,
however, to identify indicator species that are not
subject to fishing pressure. The mummichog is a
strong candidate to represent the tolerant group.
Both trend and correlation analysis suggested the
Atlantic needlefish (Strongylum marina) and the
banded killifish (Fundulus-diaphanus) as possible in-
dicators of the anadromous> or sensitive group, but
this judgement was not supported by cluster analy-
sis. Given the natural variability of these systems,
drawing conclusions about biotic integrity entirely
from the presence or abundance of one or two spe-
cies are not recommended.
The design of fish-based biological monitoring
programs for estuaries will require different ap-
proaches than currently are used in fresh water sys-
tems. The requirement for complete sampling of the
fish community (Karr, 1981) will have to be relaxed,
for example. Electrofishing or exhaustive seining or
trawling of large, tidal, saline systems is not practi-
cal. The gear used in the surveys discussed here is
selective for small species and juvenile fish of larger
species. The beach seine does not adequately repre-
sent benthic or pelagic species, but is selective for
species that prefer neritic habitats. The use"'"of small
trawl nets as in the STFMP, in combination with
seines, appears to give a better, but still incomplete,
representation of the total assemblage of fish. How-
ever, the sampling biases that are inherent to work
in estuaries do not necessarily preclude community
analysis as long as sampling methods remain con-
sistent over time. It must be recognized that only a
78
-------
Biological Criteria: Research and Regulation, 199}
1.0
0.8
X
ui
Q
o
Q 0.4
0.2
0.0
1958-1960
III...,
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
SPECIES
1
2
3
4
' 5
" 6
7
8
9
10
11
12
13
14
IS-
16
17
18
19
White perch
Atlantic Silverside
Striped Bass
Blueback Herring
Bay Anchovy
Atlantic Needlefish
Spdttail Shiner
Tidewater Silverside
Banded Killifish
American 'Shad
Alewife , . .'
Striped Killifish
Atlantic Menhaden
Rough Silverside
Mummichbg
Spot
Hogchoker
Silvery Minnow
Gizzard Shad
1.0
0.8
X
uj
Q
5 0.6
in
o
Q 0.4
z
3 0.2
0.0
1978-1980
hl.ll. ••IliL.i
1 2 3 45 6 78 9 10 11 12 13 14 15 16 17 18 19
SPECIES
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
SPECIES
12345678 910111213141516171819
SPECIES
Figure 4.—Community structure at three points in time
(three-year means of Baywide logio catch per seine haul),
and community structure projected to the year 2000. Spe-
cies numbers are keyed to the 1958-60 reference period.
few "slices" of the total fish
assemblage are sampled at
any one time and place.
This study indicates that
these slices contain suffi-
ciently consistent informa-
tion about environmental
quality not only to identify
degraded waters, but even
to begin to make some pre-
dictions about the ecologi-
cal future of areas, given
some knowledge of former
conditions, present condi-
tions, and quantitative
management goals.
Fish sampling in estuaries, with the designs
presented here, is rapid and inexpensive. Sampling
for the STFMP, for example, generally requires two
work days for two biologists with a small boat to
sample each of the monitored tributaries (each with
five stations). Because fish are identified and
counted in the field, no laboratory processing of
samples is required and data return is immediate.
Those who use the data, however, must be willing
to wait for a full season's sampling before obtaining
final results, except in extreme cases (for example,
where no fish are found during sampling).
Arguments against the use of fish as definitive
biological indicators in estuaries include (1) the in-
complete sampling problem discussed above, (2)
the fact that many of the fish captured are migratory
and do not necessarily reflect immediate conditions
at the location sampled, (3) incomplete knowledge
of the life histories and environmental tolerances of
many of the species captured, and (4) the confound-
ing effects of environmental variation and fisheries
on the abundance of species. For those familiar with
these problems, any one of these constraints might
appear to be fatal to consistent interpretation of this
data. However, this study suggests that this type of
analysis is robust because the interpretation de-
pends on patterns of communities, or at least consis-
tent portions of communities, rather than
individual species!
References
Angermeier, P. L. and J. R. Karr. 1986. Applying an index of
biotic integrity based on stream-fish communities: con-
siderations in sampling and interpretation. N. Am. J.
Fish. Manage. 6:418-29.
Chesapeake Executive Council. 1989. The Second Progress
Report under- the 1987 Chesapeake Bay Agreement. De-
cember 1989. U.S. Govt. Printing Office: 1990—720-
080/06332. Available from U.S. Environ. Prot. Agency,
Chesapeake Bay Program Liaison Off., Annapolis, MD.
79
-------
S. J. JORDAN. P. A. VMS, and J. UPHOFF
Cosden, D. and R. K Schaefer. 1988. Maryland striped bass
survey. In Investigation of Striped Bass in Chesapeake
Bay. U.S. Fish. Wildl. Serv. Federal Aid Proj. F-42-R-1
1987-1988. Maryland Dep. Nat. Resour. Annapolis, MD.
Fausch, K D., J. R. Karr, and P. R. Yant. 1984. Regional appli-
cation of an index of biotic integrity based on stream
fish communities. Trans. Am. Fish. Soc. 113:39-55
Karr, J. R. 1981. Assessment of biotic integrity using fish com-
munities. Fisheries 6:21-27.
. 1987. Biological monitoring and environmental as-
sessment: a conceptual framework. Environ. Manage.
2:249-56.
U.S. Environmental Protection Agency. .1990. Biological Cri-
teria: National Program Guidance for Surface Waters.
EPA-440/5-90-004. Off. Water Reg. Standards. U.S. En-
viron. Prot. Agency, Washington, DC.
80
-------
Biological Criteria: Research and Regulation, 1991
Selection of Biological Indicators for
Integrating Assessments of Wetland,
Stream, and Riparian Habitats1
Robert P. Brooks
Mary Jo Croonquist
Elizabeth T. D'Silva
Joseph E. Gallagher
School of Forest Resources
Dean E. Arnold
Pennsylvania Cooperative Fish and Wildlife Research
Pennsylvania State University
University Park, Pennsylvania
ABSTRACT
Biological indicators were compared to physical and chemical parameters for assessing the effects of
human disturbance in wetlands, streams, and riparian habitats. Two watersheds were studied in cen-
tral Pennsylvania, one relatively undisturbed and one disturbed by agricultural and residential devel-
opment in the lower sections. Methods based primarily on the structure and functional groupings of
biological communities were used to compare the intensity of impacts. Avian similarity indices and re-
sponse guilds reflected differences in habitat condition within the wetland and riparian components of
watersheds. Neotropical migrants and species that have specific habitat requirements were more
abundant in the reference watershed. Edge and exotic species occurred more frequently in disturbed
areas. Fish and benthic macroinvertebrate communities varied between lentic and lotic waters, and be-
tween disturbed and undisturbed reaches of streams. More warmwater fish and omnivorous species
were present in the disturbed watershed. Wetlands with flowing water supported macroinvertebrate
taxa similar to streams, whereas wetlands with standing waters contained more pollution-tolerant spe-
cies. The forested watershed provided habitat for four functional feeding groups of stream inverte-
brates (scrapers, shredders, collectors, and predators), whereas streams of the agricultural watershed
contained primarily herbivores (scrapers and collectors). Biological monitoring, using a variety of
community-based indicators, may be useful for detecting the degree of habitat disturbance and identi-
fying areas in need of restoration.
1 Contribution No. 316 of the Pennsylvania Cooperative Fish and Wildlife Research Unit. The Unit is jointly sponsored by the U.S.
Fish and Wildlife Service, the Pennsylvania State University, the Pennsylvania Fish Commission, and the WUdlife Management
Institute.
81
-------
R.P. BROOKS, M.J. CROONQUIST, E.T. D'SILVA,
J. £. GALLAGHER, and D. E. ARNOLD
Introduction
The ultimate objective of the Clean Water Act is to
restore and maintain the chemical, physical, and bi-
ological integrity of the nation's waters (section
101). The initial actions designed to achieve that
objective focused primarily on monitoring the
quality of surface waters as defined by their chemi-
cal constituents. More recently, greater attention
has been directed toward a- more balanced ap-
proach that adds biological and physical parame-
ters to the assessment tool box (e.g., Karr and
Dudley, 1981; Plafkin et al. 1989).
As the agency with primary responsibility for
implementing provisions of the Clean Water Act,
the U.S. Environmental Protection Agency (EPA)
has begun to emphasize the importance of using bi-
ological criteria to assess the health of surface wa-
ters. EPA is directing states to adopt narrative
biological criteria as part of their water quality stan-
dards (U.S. Environ. Prot. Agency, 1990). The pri-
mary advantage of biological indicators is that they
presumably integrate the impacts of water pollution
and habitat disturbance over time. This continuous
record typically is not available from chemical sam-
pling protocols. Whereas chemical parameters have
proved useful for monitoring point discharges into
surface waters, biological and physical measures
appear to be better for assessing the effects of more
dispersed impacts such as nonpoint source runoff,
incremental losses of wetlands, and changes in land
use along riparian corridors and throughout water-
sheds. Thus, as the Clean Water Act recognized ini-
tially, a multifaceted assessment using chemical,
physical, and biological parameters will provide a
comprehensive measure of ecological integrity for
the nation's waters (U.S. Environ. Prot. Agency,
1990).
Given the importance of monitoring multiple
parameters, it should be equally obvious that
streams, rivers, and lakes should be studied simul-
taneously with their wetland and upland surround-
ings. The interface between surface waters,
wetlands, and adjacent uplands is usually referred
to as a riparian zone (Hunt, 1985). This zone, de-
fined here as the soil, flora, and fauna within the
100-year floodplain, serves as critical habitat for a
great diversity of species, and buffers the stream
channel from point and nonpoint sources of pollu-
tion. This study is based on the assumption that the
linkages among the biota, land use, and hydrology
can be used to monitor the environmental health of
watersheds. Consequently, we studied the biologi-
cal, physical, and chemical characteristics of both
in-stream and wetland-riparian components of two
watersheds in an effort to understand these link-
ages. This paper documents the different responses
of biotic communities to anthropogenic distur-
bances that affect both wetland-riparian areas and
in-stream conditions.
Methods
ti
Study Area
This study compared two watersheds in central
Pennsylvania from 1987 to 1990 to investigate how
biotic communities are altered when a watershed is
disturbed. One of these was relatively undisturbed
and was used as a reference watershed; the other
had been disturbed by agricultural and residential
development in the lower sections. Methods based
primarily on the structure and functional group-
ings of biological communities were used to com-
pare the intensity of impacts. The degree of
disturbance occurring within watersheds was de-
termined by analyzing land use patterns and hy-
drologic changes regardless of the specific origins
.of those impacts.
The undisturbed, or reference, watershed was
White Deer Creek and its associated tributaries, lo-
cated in rural areas of Centre and Union Counties,
Pennsylvania. White Deer Creek has a drainage
area of 117 km2 and flows into the West Branch of
the Susqiiehanna River (the first 89 km2, within
Bald Eagle State Forest, were used as the study re-
gion). Limited forestry operations as well as sea-
sonal fishing and hunting are the only major
activities within the watershed. Forested habitat
covered 94 percent of the watershed, 1 percent of
the area was wetland, 4 percent was partially dis-
turbed (shrub/brush and old field), and 1 percent
was disturbed area (gravel pit, barren and minor
agriculture; Fig. 1; Croonquist, 1990).
Little Fishing Creek and its tributaries (109
km2), located within agriculturally-dominated por-
tions of Centre and Clinton Counties, Pennsylvania,
constituted the disturbed watershed. Little Fishing
Creek drains into Fishing Creek and eventually into
the West Branch of the Susquehanna River. Little
Fishing Creek flows into agricultural and residential
areas along midreach and mainstem channels
where the riparian and wetland zones have been al-
tered substantially. Livestock freely roamed in and
out of the stream, causing much bank erosion and
siltation that had degraded both terrestrial and
aquatic habitats. Forested habitats covered 71 per-
cent of the watershed, 57 percent of which was in
the upper, protected regions. Undisturbed wetlands
comprised < 1 percent of the watershed, and 13 ha
of the 28 ha (46 percent) of wetlands were in head-
82
-------
Biological Criteria: Research and Regulation, 1991
UJ
CO
:D
Q
100 -i
80 -
60 -
40 -
20 -
UNDISTURBED
PARTIALLY DISTURBED
DISTURBED
WDC
LFC
WATERSHED
Figure 1.—Land use and cover for White Deer Creek and Little Fishing Creek.
water (undisturbed) regions. Partially disturbed
habitats (shrub/brush, old fields, and partially dis-
turbed wetlands) covered 4 percent. Over 25 per-
cent of the watershed consisted of disturbed
habitats (agriculture, residential, commercial); 94
percent of these disturbed habitats were in the mid-
dle and mainstem sections (Fig. 1; Croonquist,
1990).
Because of the large size of both watersheds and
the variability of habitat from headwaters to
mainstems, each watershed was divided into four
hierarchical .sections based on mean annual dis-
charge and stream order. Hierarchical sections were
headwater, second order tributary, midreach chan-
nel, and lower mainstem channel, with associated
wetlands within each section. Three study sites
were selected within each section, providing 12
study sites per watershed, 24 sites in total
(Croonquist, 1990).
A geographic information system (GIS) was
used to characterize land use of each watershed and
each hierarchical section; however, headwaters and
small tributaries could not be differentiated with .the
GIS, so they were grouped together. Land uses were
assigned to three categories—undisturbed, partially
disturbed, and disturbed. Undisturbed land use
types included all forested and wetland types, ex-
•cept disturbed emergent and lacustrine wetlands
along the middle and mainstem areas of Little Fish-
ing Creek. Partially disturbed land use types in-
cluded old fields, shrub/brush, and the previously
mentioned emergent and la-
custrine wetlands. Disturbed
types included agriculture and
development (either residen-
tial or industrial (Croonquist,
1990). The sites contained in
the two middle hierarchies of
Little Fishing Creek were
modified to give greater
weight to land disturbance
rather than discharge alone.
Stream Habitat
Assessment
A standardized rapid bio-
assessment protocol compiled
by Plafkin et ,al. (1989) was
used to characterize physical
stream habitat. This method
estimates general land use
and physical stream charac-
teristics such as stream width,
depth, flow, and substrates to
arrive at a relative assessment
score ranging from severely degraded (0) to excel-
lent (132). A total score was obtained for each study
site and mean scores were compared among hierar-
chies within each watershed and between water-
sheds during the fall of 1989. In addition, we
quantified the degree of sediment embeddedness
in stream substrates using a 0.1 m -ring made of
white electrical cable. The ring was tossed ran-
domly five times at each site. Sediment area and
depth were visually estimated (nearest 5 percent).
Biological and Water Chemistry
Sampling
Biological and water quality sampling were con-
ducted at each of the 24 study sites from October
1987 through September 1989 (see Brooks et al.
1990; Brooks and Croonquist, 1990; and Croonqu-
ist, 1990 for detailed descriptions of sites and meth-
ods). Each of the 24 study sites contained 3, 100-m
transects. The riparian transect was located along
the riparian zone (0 - 2 m from the bank) parallel
with the stream channel. A second, the wetland
transect, began at one end of the riparian transect
and extended 100 m from and perpendicular to the
channel through the adjacent wetland/upland
zone. A third transect for aquatic sampling was lo-
cated in the stream channel parallel to the riparian
transect
83
-------
R. P. BROOKS, M. J. CROONQUIST, E. T. D'SILVA,
J. £ GALLAGHER, and D. E. ARNOLD
Water chemistry sampling and censusing for
birds were done periodically for two years to esti-
mate seasonal changes. Conductivity, alkalinity, pH,
and temperature were used to characterize the
water quality of streams and wetlands. Once each
year total nitrogen (NO3 mg/L) and total phospho-
rus (PO4 mg/L) were measured in stream samples
at all 24 study sites. Bird censuses consisted of 5-
minute point counts at every other sample plot
(every 50 m), totaling five point counts per sample
site (Croonquist, 1990). All species heard and seen
within a 25-m radius were recorded, which created
an effective sampling area of 0.2 ha per plot and 1.0
ha for each site. Fish (electrofishing) and benthic
macroinvertebrates (collected by Surber sampler in
sbream and Ekman dredge in wetlands) were sam-
pled during the spring and summer of 1989. Fish
were sampled in nine sites from White Deer Creek
and 11 sites from Little Fishing Creek because of the
absence of sufficient water in some headwater sites.
Additional sampling of mammal, herpetile, and
plant communities was done simultaneously, but
are reported elsewhere (Brooks et al. in press;
Croonquist, 1990; Croonquist and Brooks, in press).
To detect changes in biological communities
among sites and between watersheds, we examined
the number of species present, the amount of over-
lap between communities (Jaccard's Coefficient of
Community; Jaccard, 1912), and the kinds of species
present based on response guilds (see Brooks and
Croonquist, 1990), a modified Index of Biological
Integrity (Karr, 1981; Fausch et al. 1984), and other
functional groupings (Plafkin et al. 1989) (Table 1).
Results and Discussion
Stream Habitat Assessment
Stream habitat assessment values provided infor-
mation as to the quality of riparian and in-stream
habitat within and between watersheds (Table 2).
There was a negative correlation between land use
disturbance and habitat quality within Little Fish-
ing Creek (r = -0.652,2 = 0.05) and between water-
sheds (r = -0.771, £ = 0.05; Croonquist, 1990). Along
White Deer Creek, the headwater section had the
lowest mean score (92 ± 7.2) and mainstem had the
highest (130 ± 2.1); but all sections had relatively
high scores. Sites with greatest anthropogenic dis-
turbance, the mainstem sites of Little Fishing
Creek, had the lowest mean score (45 ± 5.8;
Croonquist, 1990). Student's t-test showed that the
mean score of White Deer Creek mainstem was sig-
nificantly higher (i.e., higher quality habitat) than
the mean score of Little Fishing Creek mainstem
Table 1—Criteria of response guilds for avian ,
communities.
RESPONSE GUILDS SCORES
Wetland Dependency
Obligate species (> 99% in wetlands)
Facultative wet (usually in or near wetlands)
Facultative (wetlands not essential)
Facultative dry (occasional or no use)
Upland (> 99% in uplands)
Habitat Specificity
Alpha species—stenotypic, specialist
Gamma species—landscape dependent
Beta species—generalist, edge
Trophic Level
Carnivore, specialist (restricted diet)
Carnivore, generalist
Herbivore1, specialist (e.g., nuts, nectar)
Herbivore, generalist
Omnivore (plants or animals)
Species Status
Endangered, endemic, of concern
Commercial, recreational value
Other native species
Exotic
Seasonality
Neotropical migrant
Short-distance migrant
Year-round resident
Nonbreeding season resident only
Migratory transient
Occasional
4
3
2
1
5
3
1
0
5
4
3
2
1
0
Source: Brooks and Croonquist, 1990.
(p_ = 0.05, d.f. = 10). Sediment embeddedness de-
creased down the hierarchy of White Deer Creek,
presumably because of increasing water velocity
and lack of sediment inputs from the forested wa-
tershed. Sediment embeddedness increased sub-
stantially in the lower reaches of Little Fishing
Creek where cropping and grazing were the domi-
nant land uses (Table 3). Thus, as physical distur-
bance increased, habitat quality decreased
(Croonquist, 1990).
Water Chemistry
During the two years of study, White Deer Creek
had a mean pH of 6.3 ± 0.7, mean conductivity of
34.6 ± 44.8 u.S/cm, and mean water temperature of
9.2 ± 5.4 °C. Little Fishing Creek had a mean pH of
7.1 ± 0.8, mean conductivity of 153.8 ± 458.4 \iS/cm,
and mean water temperature of 9.8 ± 6.1 °C (Table
4). Water quality comparisons between the two wa-
tersheds were confounded somewhat because of a
change in substrate type within Little Fishing
Creek. Both watersheds had sandstone substrates
along upper regions. As they flowed into the val-
leys, however, the substrate of Little Fishing Creek
changed to limestone, whereas White Deer Creek
remained sandstone. Mean water temperatures in-
creased slightly through both watersheds, but did
84
-------
Biological Criteria: Research and Regulation, 1991
Table 2—Stream habitat assessment scores for each hierarchy in White Deer Creek (reference) and Little Fishing
Creek (impacted).
WHITE DEER CREEK
STUDY SITE
Headwater
Kemmerer Trail
Sand Spring Run
Mile Run
Mean ± S.D.
Tributary
Camp Site
Black Gap
Kettle Hole
Mean ± S.D.
Midreach
Beaver Dam
McCall Dam
Clark Trail '
Mean ± S.D.
Mainstem
Gauging Station
White House
Dump Site
Mean ± S.D.
SCORE
100 •
90
86
92 ±7.21
110
115
107
1 1 1 ± 4.04
106
127
121
118 ±10.82
131 '
132
128
130 ± 2.08a
LITTLE FISHING CREEK
STUDY SITE
Headwater
Dismal Swamp
'Fulton Gap
Camp Kill
Undisturbed Middle
Dam Site
Hecla Gap ,
Kfislund Camp
' : i. ,
, Disturbed Middle
Lee's Gap
" Mingoville
Deitrich's Trib.
Mainstem ..,..•
Hublersburg
Syndertown
N.J. Farm
79
61
112
84
78
121
113
104
108
88
47
81
42
42
•52
'45
SCORE
± 25.87
± 22.87
± 31.1
±, 5.77a'b
Source: Croonquist, 1990. • ' .. . ... . :- . • .
Note: Mean score and standard deviation are given for each watershed section (Croonquist, 1990).'
"Comparisons of mean scores between watersheds are.significantly different (Student's t-test, p = 0.05, d.f. = 10). ' '" '
"Pearson correlation test results show negative correlation between assessment score .and land-use disturbance within Little Fishing Creek (r = - 0.652,
p = 0.05), and negative correlation between assessment score and land-use disturbance between undisturbed sites at the lower half of White Deer Creek
and disturbed sites at the lower half of Little Fishing Creek (r = -0.771, p = 0.05). '• . '• . -
Table 3—Percent sediment embeddedness for each hierarchy in White
Deer Creek (reference) and Little Fishing Creek (impacted) watersheds.
HIERARCHY
Headwater
Tributary/Undis. Mid.
Midreach/Dist. Mid.
Mainstem
WHITE DEER
CREEK
PERCENT
68
37
10
3
LITTLE FISHING
CREEK
PERCENT
65
37
37
77
CHI-SQUARED
VALUE
0.1
0.0
73a
1 ,825a
ap < 0.05
not differ significantly between watersheds (Table
4). Concentrations of nitrates and phosphates did
not vary significantly within nor between water-
sheds during the two years (Table 5). The lower
reaches of Little Fishing Creek did not have the
high concentrations of nitrate and phosphate that
were expected from the surrounding farms. How-
ever, samples were taken only once in the fall of
each year, and therefore the effects of runoff of
chemical fertilizers in agricultural areas probably
were not adequately represented (Croonquist,
1990).
Biological Communities
Watersheds had similar communities in the upper
regions where land use patterns were similar. Com-
munity similarities and functional guilds began to
diverge in the lower reaches of the
disturbed watershed. Birds were
more indicative of these changes
than in-stream fauna (Tables 6, 7,
and, 8). Overall, the reference wa-
tershed had somewhat fewer spe-
cies than the disturbed watershed.
The numbers of vertebrate species
found in the undisturbed water-
shed versus the disturbed water-
shed were, respectively: 94 vs. 110
birds and 16 vs. 20 fish. The
greater number ,o£ species ob-
served in the disturbed watershed probably was
due to the abundance of edge (for birds) and
warmer water temperatures (for fish) in the lower
reaches of Little Fishing Creek. Few differences
were observed between headwaters of each water-
shed because conditions were similar for both for-
ested habitat (Fig. 1), and streams (Tables 2, 3, 4,
and 5).
Birds were indicative of changes in wetlands
and riparian areas (Fig. 2, Table 8). Avian response'
guilds, when used individually or in combination,
reflected the disturbance patterns of the landscape.
The following avian guilds provided the most infor-
mation for characterizing differences in the bidtic
communities between the undisturbed arid dis-
turbed watersheds: habitat specificity, trophic level,
and seasonally (Brooks et al. 1990). Resident and
85
-------
H. P. BROOKS, M. J. CROONQUIST, E. T. D'SILVA,
J. £. GALLAGHER, and D. E. ARNOLD
Table 4—Mean water quality and quantity values, and their standard deviations, from 1937-89 for White Deer
Creek and Little Fishing Creek watersheds.
WATERSHED/HIERARCHY
White Deer Creek
Headwater
Tributary
Midreach
Mainstem
Little Fishing Creek
Headwater
Undisturbed middle
Disturbed middle
Mainstem
pH
6.0 ± 0.6
6.5 ± 0.2
6.6 ± 0.2
6.6 ± 0.2
6.7 ± 0.5
6.9 ± 0.6
7.4 ± 0.3
7.7 ± 0.3
CONDUCTIVITY
(|iS/CM)
34.9
41.3
37.3
52.9
90.0
92.0
159.3
50.3
54.7
50.3
70.0
123.9
103.7
131.7
199.8 74.6
TEMP.
(°C)
8.2
8.9
10.7
10.9
9.6
10.1
11.2
4.2
4.5
5.8
6.2
5.6
5.7
6.0
12.5 7.2
STAGE
(m)
-1.6± 5.9
-3.4 ± 8.4
-0.8 ± 8.1
2.7 ±11.7
0.8 ± 9.9
-2.7 ± 10.7
-2.3 ± 9.3
-1.4 ± 14.7
Table 5—Mean values (n=3), and standard deviation, of total nitrate (NO3 mg/L) and total phosphate (PO4 mg/L)
from water samples of each hierarchy of White Deer Creek and Little Fishing Creek watersheds.
TOTAL NITRATE
HIERARCHY
9/88
12/89
TOTAL PHOSPHATE
9/88
12/89
White Deer Creek
Headwater
Tributary
Midreach
Mainstem,
Little Fishing Creek
Headwater
Undisturbed middle
Disturbed middle
Mainstem
0.67
0.97
0.40
0.72
0.48
0.23
0.40
0.19
0.12
0.05
0.36
0.46
0.06
0.17
0.72 0.38
0.68 ±0.71
0.75 ± 0.77
0.28 ± 0.03
1.35 ± 0.61
1.32 ± 0.35
,1.13 ± 0.39
0.65 ± 0.39
1.03 ± 0.33
0.024
0.008
0.014
0.068
0.025
0.015
6.025
0.016
0.006
0.005
0.108
0.017
0.009
0.012
0.052 0.025
0.005 ± 0.006
0.004 ± 0.004
0.005 ± 0.002
0.004 ± 0.001
0.068 ± 0.095
0.010 ± 0.003
0.012 ± 0.007
0.017 ± 0.003
Table 6—Metrics used to assess fish communities in White Deer Creek and Little Fishing Creek.
SCORING CRITERIA
CATEGORY
Species Richness and Composition
Trophic Composition
1.
2.
3.
4.
5.
6.
7.
8.
9.
METRIC
Total no. of fish species
Proportion of sculpins
Number cyprinid species
Number adult trout spp.
Number intolerant spp.
Proportion suckers
Proportion individuals as omnivores
Proportion total insectivores
Proportion Salmonids
5
<5
<20
>80-100
>5
3
5-20
20-45
>40-80
5-1
1
. >20
>45
0-40
<1
Key: Integrity classes for total IBI scores are: excellent (58-60); good (48-52); fair (40-44); poor (28-34); and very poor (12-22). Expectations for metrics
1-5 vary with stream size and region (Karr, 1981; Fausch et al. 1984).
neotropical-migrant breeders that had specific habi-
tat requirements and/or were carnivorous (e.g.,
woodland warblers, such as cerulean warbler,
Dendroica cerulea, black-throated blue warbler,
Dendroica caerukscens, and northern waterthrush,
Seiurus noveboracensis) decreased in percentage only
down the hierarchy of the disturbed stream. Edge
(blue jay, Cyanocitta cristata) and exotic (house spar-
row, Passer domesticus) species were found in greater
abundance in the disturbed watershed. Neotropical
migrants with high guild scores for habitat specific-
ity formed about 25 percent of the community in
upper regions of both watersheds. These species
formed only 5 percent of the mainstem community
of the disturbed watershed, but remained a major
component of the undisturbed watershed commu-
nity.
Aquatic communities paralleled the trends ob-
served for physical habitat assessments of in-stream
and riparian areas despite the modest sampling ef-
fort. The results of the IBI were more indicative of
changes in fish communities than results from the
community coefficients (Fig. 3> Tables 6 and 7). The
IBI suggested that fish communities of the lower
86
-------
Biological Criteria: Research and Regulation, 1991
O
8.
i.o
0.8-
0.6-1
0.4-
0.2-
0.0
Headwater Trib./Undisturb. Mid. Mid./Disturb, Mid. Mainstem
Subwatershed Section
Figure 2.—Avian community coefficient values between watershed sections of White Deer Creek and Little Fishing Creek, for
the riparian transect.
Table 7—Mean Index of Biotic Integrity (IBI) for fish in each hierarchy in White Deer Creek (reference) and Little
Fishing Creek (impacted) watersheds.
HIERARCHY
WHITE DEER CREEK
SCORE (CLASS)
LITTLE FISHING CREEK
SCORE (CLASS)
CHI-SQUARED VALUE
Headwater
Tributary/Undis. Mid.
Midreach/Dis. Mid.
Mainstem
48 ± Oa (good)
48 ± 0 (good)
47 ± 1 (good)
42 ± 2 (good)
20 ± 28b (very poor)
41 ± 2 (fair)
45 ± 9 (fair to good)
35 ± .1 (poor to fair)
2.67
2.99
3.63
9.77°
"Mean ± SD.
"Large standard deviation because one site contained no fish
=p < 0.05. ' ' ' . •
Key: Integrity classes for total IBI scores are: Excellent (58-60); good (48-52), fair (40-44); poor (28-34); and very poor (12-22).
Table 8—Percent composition (%) of selected response guilds of birds. Results given by hierarchical section of
White Deer Creek and Little Fishing Creek.
HEADWATER
Exotic
(Status = 0)
Edge
(Habitat Specificity = 1 )
Permanent Resident + Edge
• (Seasonality = 3 & Hab. Spec = 1 )
Habitat Specific
(Habitat Specificity = 5 or 3)
Neotropical Migrant
(Seasonality = 5)
Wetland Dependent
(Wetland Dependency = 5, 3, or 1)
Neotropical Migr. + Habitat Specific
(Season. = 5 & Hab. Spec = 5 or 3)
WDC
2
• 39
19
61
41
36 ,;
29
LFC
4
51
29
49
40
37 '
23
TRIBUTARY"
WDC
0
42
21
' 58
42
32
28
LFC
3
47
28
53
32
35
18
MIDREACH"
WDC
0
46
25
54
32
33
19
LFC
7
66
£L
34
24
30
9
MAINSTEM
WDC
0
55
31
45
41 .
33
22
LFC
8
68
45
32
20
31
6
"Tributary section of White Deer Creek and undisturbed middle section of Little Fishing Creek.
"Midreach section of White Deer Creek and disturbed middle section of Little Fishing Creek.
Note: Numbers underlined represent large differences in percent composition between watersheds (Croonquist and Brooks, 1991).
87
-------
R P. BROOKS, M. J. CROONQUIST, E. T. D'SILVA,
J. E. GALLAGHER, and D. E. ARNOLD
O
o
O
o
O
1.0
0.8-
0.6-
0.4-
0.2-
0.0
Headwater Trib./Undisturb. Mid. Mid./Disturb. Mid.. Mainstem
Subwatershed .Section
Figure 3.—Fish community coefficient values between watershed sections of White Deer Creek and Little Fishing Creek.
reaches of Little Fishing Creek, where disturbance
was greatest, were negatively impacted (Tables 6
and 7). The lower regions of the disturbed water-
shed, which had degraded water quality, severely
disturbed stream banks, and increased sediment
loads on the stream bottom, supported more
warmwater fish species (centrarchids) and pollu-
tion-tolerant omnivores (white sucker, Catostomus
commersoni, and horned chub, Nocomis biguttatus).
More sensitive species, such as salmonids and other
insectivores (blacknose dace, Rhinidhys atratulus),
were more abundant in the reference watershed.
Storage and sorting problems with the
macroinvertebrate samples prevented the use of a
quantitative assessment method such as those sug-
gested by Plafkin et al. (1989); however, some quali-
tative trends were apparent. Wetlands with flowing
water supported macroinvertebrate taxa similar to
those of streams (ephemeropterans, plecopterans,
tricopterans), whereas wetlands with standing wa-
ters contained hydracarinids, dipterans, annelids,
and pelecypods. The forested watershed provided
habitat for four functional feeding groups (scrapers,
shredders, collectors, and predators), whereas
streams of the agricultural watershed contained pri-
marily herbivores (scrapers and collectors).
In summary, the subtle impacts of habitat dis-
turbance in the riparian zone were apparent
through investigations of physical, chemical, and
biological parameters. Differences between refer-
ence and disturbed watersheds were more obvious
in the riparian corridor (e.g., stream habitat assess-
ment and avian response guilds) than for in-stream
conditions (e.g., water chemistry, fish, and in-stream
macroinvertebrate communities). However, the in-
stream parameters were sampled at much lower fre-
quencies. Changes in fish and macroinvertebrate
communities were most obvious where physical
conditions of the stream were severely degraded, as
in the mains tern of Little Fishing Creek.
Protection of wetlands, streams, and riparian
corridors is critical for maintaining biological diver-
sity of terrestrial and aquatic species. Biological
monitoring, using a variety of community-based in-
dicators, may be useful for detecting the degree of
habitat disturbance and identifying areas in need of
restoration when used in conjunction with physical
and chemical indicators. This three-parameter ap-
proach can be used to target restoration efforts to-
ward portions of watersheds where recovery is
feasible.
ACKNOWLEDGMENTS: Funding for this study was
provided by the School of Forest Resources of The Penn-
sylvania State University, the Pennsylvania Wild Re-
source Conservation Fund, the Pennsylvania Game
Commission, and the U. S. Army Corps of Engineers,
Waterways Experiment Station, P.O. No. DACW-89-M-
2335. Special thanks go to J. Hassinger of the PGC and C.
Klimas of WES for providing advice and technical assis-
tance. We appreciate the cooperation of landowners who
gave permission to conduct field work on their proper-
ties and of the many field assistants who helped collect
•and analyze data.
88
-------
Biological Criteria: Research and Regulation, 1991
References
Brooks, R. P., D. E. Arnold, E. D. Bellis, C. S. Keener, and M. J.
Croonquist. In press. A methodology for biological
monitoring of cumulative impacts on wetland, stream,
and riparian components of watersheds. J. A. Kusler
and G. Brooks, eds. Proc. Int. Symp. Wetlands and River
Corridor Management. Ass. Wetland Managers, Berne,
NY.
Brooks, R. P. and M. J. Croonquist. 1990. Wetland, habitat,
and trophic response guilds for wildlife species in Penn-
sylvania. J. Pa. Acad. Sci. 64(2):93-102.
Brooks, R. P., M. J. Croonquist, D .E. Arnold, E. D. Bellis, and
C. S. Keener. 1990. Analysis of Wetland-Riparian Corri-
dors. Final Rep. Contr. No. DACW-89-M-2335. U.S.
Army Corps Eng., Waterways Exp. Sta., Vicksburg, MS.
Croonquist, M J. 1990. Avian and mammalian community
comparisons between protected and altered watersheds
- a landscape approach. M. S. Thesis. Pennsylvania
State Univ., University Park.
Croonquist, M. J. and R. P. Brooks. In press. Use of avian and
mammalian guilds as indicators of cumulative impacts ,
in riparian-wetland areas. Environ. Manage.
Fausch, K. D., J. R. Karr, and P. R. Yant. 1984. Regional appli-
cation of an index of biotic integrity based on stream
fish communities. Trans. Am. Fish. Soc. 113:39-55.
Hunt, C. 1985. The need for riparian habitat protection. Natl.
Wetlands Newsl. 7:5-8.
Jaccard, P. 1912. The distribution of the flora in the alpine
zone. New Phyto. 11:37-50.
Karr, J. R. 1981. Assessment of biotic integrity using fish com-
munities. Fisheries 6(6):21-27.
Karr, J. R. and D. R. Dudley. 1981. Ecological perspective on
water quality goals. Environ. Manage. 5:55-68.
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/444/4-89-001. Assess. Watershed Prot.
Div., U.S. Environ. Protect. Agency, Washington, DC.
.U. S. Environmental Protection Agency. 1990. Biological Cri-
teria: National Program Guidance for Surface Waters.
EPA-440/5/90-004. Off. Water Reg. Stand., Washington,
DC.
89
-------
Biological Survey Study Design
Considerations When Representing
Biointegrity and Evaluating Non-Attainment
of Designated Uses
James M. Lazorchak
U.S. Environmental Protection Agency
Environmental Monitoring Systems Laboratory
Cincinnati, Ohio
Biological survey study design is critically important in biocriteria development. The design must be
scientifically rigorous to provide for legally defensible data and be biologically relevant to detect
problems of regulatory concern. It is not financially nor technically feasible to completely evaluate
an entire ecosystem at all times, selecting community components, the time, season, station location, meth-
ods to measure the community of interest, and a quality assurance and quality control program are impor-
tant to the success of a biocriteria program. When using biological surveys to establish what the state of
biointegrity is and to determine if designated uses are being attained there are several considerations that
should be taken into account. An introduction to quality assurance and quality control considerations as
well as the role data quality objectives have in helping focus biological survey design will be discussed. An
overview will be given on things to keep in mind when designing biological surveys like selecting aquatic
community components, designing biological surveys to measure these aquatic community components,
metric selection and sampling design. The discussion of these topics will follow the Biological Criteria Na-
tional Program Guidance for Surface Waters.
If you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
90
-------
Biological Criteria: Research and Regulation, 1991
Significance of Change in Community
Structure: A New Method for Testing
Differences
James R. Pratt
School of Forest Resources and Graduate Program in Ecology
Pennsylvania State University
University Park, Pennsylvania
E.P. Smith
Department of Statistics
Virginia Polytechnic Institute and State University
Blacksburg, Virginia
ABSTRACT
Structural changes in biotic communities often precede detectable adverse effects on ecosystem pro-
cess rates. Management concerns may focus on changes in abundance, the loss of important species,
or changes in the composition or diversity of groups of taxa. Community structure data are high in in-
formation content, but may be difficult to analyze and interpret. Bioassessment methods have recom-
mended use of multivariate procedures for examining differences in community similarity, but the
number of species (variables) in community samples and the limited number of replications make ap-
plication of multivariate methods problematic; the number of degrees of freedom will usually be
fewer than the number of variables and the covariance matrix will be singular. Additionally, some
species present at one site will be absent at other sites, especially those that are impacted, thus invali-
dating assumptions of normality. An analytical method is needed that can use community informa-
tion for inferential analysis of environmental effects without violating assumptions of statistical
models. To use community structure information, species data must first be reduced to measures of
similarity (or distance) among replicates. Measures (indices) for assessing community similarity can
be based on presence-absence, ranked or rated abundance, and relative or absolute abundance of
taxa. A permutation procedure involving a large number of random switches of similarity measures
is used to build a probability distribution of the ratio of mean between treatment or location analysis
of variance procedures. Assuming that a null hypothesis of no difference among locations is rejected,
follow-up analyses can indicate the locations or treatments that are different and can identify the in-
fluence of species whose presence, absence, or abundance greatly affects analyses. This research has
used the permutation method and binary, ranked, and continuous data to examine community
changes in multispecies laboratory experiments and field evaluations of periphyton and invertebrate
communities. Unlike descriptive multivariate methods or rating/score methods, the permutation
method provides an inferential test of hypothesized differences between reference sites to suspected
impact sites. The technique is an objective means of determining differences and identifying the taxa
indicative of community differences.
91
-------
J. R. PRATT and E. P. SMITH
Introduction
Evaluating biological changes in stressed ecosys-
tems is based on the hypothesis that human influ-
ence alters the sustainability of ecosystem services.
Controversy over the conceptual basis for ecosys-
tem protection has followed the topical, favoring
paradigms of ecology: studies that once focused on
describing the structure of ecosystems gave way to
process studies examining sources and flows of en-
ergy and materials. Increasing evidence suggests,
however, that aquatic ecosystem structure changes
before process accommodations are detectable
(Odum, 1985,1990; Schindler, 1987; Schaeffef et al.
1988; Pratt, 1990). Process measures are robust (anti.
change primarily with the availability of substrates
(nutrients, dead organic matter) rather than the bi-
ological machinery that processes these substrates.
Where stresses do not significantly alter the supply
of substrates, process measures show little impact
(Levine, 1989).
Evaluations of aquatic community structure
have re-emerged as an important facet bf environ-
mental impact analysis and have already been in-
corporated into the regulatory framework in several
States. The establishment of biological criteria for
waters implies that methods are available to detect
significant changes in community structure. Taxo-
nomic lists incorporating the presence, absence, or
abundance of species at particular locations of inter-
est have high information content. The means for
dealing with this information vary considerably de-
pending on the questions under study. This paper
presents methods for examining community struc-
ture information using inferential procedures for
testing hypotheses of community change.
Early in this century, the ubiquity of organic
pollution and its effects on aquatic communities led
to the classification of organisms by enrichment (or
low oxygen) tolerance, the classic Saprobian system
(Kolkwitz and Marsson, 1908). In the early 1970s,
questions focused on changes in biotic diversity,
and information theory (Shannon and Weaver,
1949) was applied to the comparisons of commu-
nity taxa abundance. However, information theory
indices were often sensitive to the number,of taxa
(Green, 1979) and aroused considerable debate
about the relationship between biotic diversity and
community stability (Hurlbert, 1971).
Current impact analysis makes use both of im-
proved knowledge of stream ecology arid our abil-
ity to deal with complex data sets. Understanding
of the distribution, tolerance, and habits of aquatic
species is now much clearer (e.g., Lowe, 1974;
Hilsenhoff, 1982; Cummins and Klug, 1979; Karr et
al. 1986). A variety of systematic procedures for
evaluating taxonomic structure is available
(Metcalfe, 1989; Cairns and Pratt, in press). Other
descriptive procedures allow comparison of collec-
tions according to taxonomic composition and
abundance using community similarity or multi-
variate distance measures (e.g., Gauch, 1982; Pielou,
1984; Digby and Kempton, 1987). Such procedures
have been used to examine community differences
along environmental gradients that include habitat
alterations and pollutants (e.g., Pratt et al., 1985;
Whittier et al. 1988). However, these descriptive
procedures do not provide rigorous, inferential
methods for testing differences among communi-
ties. ;' .......
In 1986 the Environmental Protection Agency
moved to standardize and improve biological meth-
ods for assessing and monitoring surface waters. An
important product for this effort was a guidance
document on rapid biological assessment focusing
on community structure of fish, macroinvertebrates,
and algae (Plafkiri et al. 1989). Methods for assess-
ing fish communities were based oh the work of
Karr and colleagues (1986), while those' for
macroinvertebrates were based both on historical
methods and the development of additional meth-
ods analogous to those for fish. Much of this work
was incorporated into methods now used by'the
state of Ohio in assessing biological water quality
(Ohio Environ. Prot. Agency, 1987). s
Rapid bioassessment procedures are systematic
means for collecting and evaluating community
and habitat structure information. The procedures
have various uses, including determining attain-
ability of water uses and characterizing the degree
of use impairment. The procedures recognize" the
potential for regional variation in ecosystem condi-
tion and performance (Omernik, 1987). Site compar-
isons are made across' limited environmental
gradients to either an upstream reference site com-
munity, or to a regional reference community. Meth-
ods for assessing macroinvertebrate communities
use the extent of similarity to the reference commu-
nity as one of several community assessment met-
rics. While the rapid bioassessment procedures are
rational means for evaluating community data, they
are not inferential. The indices developed usually
lack any estimate of variance and, therefore, cannot
provide a statistical comparison among sites. Addi-
tionally, the use of multivariate procedures to com-
pare communities violate assumptions of normality
when taxa are present at some sites but not others.
Regardless of the mechanism by which biologi-
cal criteria are developed, community comparisons
and inferential procedures'will be needed to detect
changes in biological structure. The methods de-
92
-------
Biological Criteria: Research and Regulation, 1991
scribed in this paper can be used to compare com-
munities to regional reference communities or to
upstream reference sites and can provide an inferen-
tial test of hypotheses of no difference among com-
munities. Studies of biological quality will need to
demonstrate rigor in the quality of sampling, the
identification of species, and the analysis of results.
Methods
The analytical method presented here tests for dif-
ferences in community structure among sites or
treatments (see Smith et.al. 1990). The analyses re-
quire the selection of a measure of community sim-
ilarity-dissimilarity or distance. Following the
construction of a matrix comparing community
measures by sites or treatments, a permutation pro-
cedure is used to repeatedly and randomly switch
measures among the site or treatment categories.
Switching similarity measures is equivalent to
switching data vectors (i.e., switching all data for
one treatment replicate). Statistical comparisons are
based on the relationship of between treatment
similarity to within treatment similarity, analogous
to .analysis of variance procedures. Follow-up anal-
yses examine pairwise comparisons of treatment
categories and identify the influence of particular
species on the chosen community measure.
Community Similarity—an Overview
The array of methods for estimating community
similarity is diverse and will not be reviewed in de-
tail here. Legendre and Legendre (1983) list 27 com-
munity metrics. With reference to ecological
studies, indices and methods for comparing com-
munities are succinctly reviewed by Pielou (1984)
and Digby and Kempton (1987). Certain indices are
, recommended by Plafkin et al. (1989), but most in-
dices have uncertain or unknown statistical proper-
ties and few have been rigorously studied for their
sensitivity in detecting community change.
Two aspects of the assessment of community
similarity are worth noting. First, both similarity
measures (comparisons of species overlap between
physical samples) and distance measures (distance
between samples in multidimensional species
space, often a complement of similarity) are avail-
able to examine the relationship among replicates
from different sites. Second, presence-absence (bi-
nary), relative abundance, and actual or absolute
abundance are simply scales that weight the impor-
. tance of a species. In binary data the weightings are
,1 (present) and 0 (absent). Relative abundance data
may be presented as abundance rankings (usually
scaled as integer values between 0 and 10) or as pro-
portional abundance (scaled by the total number of
individuals counted in each replicate). Actual or ab-
solute abundance data weight individual species ac-
cording to the actual number of individuals counted
and weight different replicates by the total number
of individuals. Other measures of abundance may
be used and include such estimators as biomass,
biovolume, cover, and importance value.
Binary data underestimate the importance of
dominance changes, but are more closely related to
expectations of falling species numbers in stressed
communities. Binary data are comparatively easier
to obtain because time is not consumed enumerat-
ing individuals,.a laborious task for small taxa (e.g.,
algae, protozoa, micrometazoa). Where the scale of
sampling is large compared to the size of organisms,
relatively complete sampling can be assumed. This
is usually the case in sampling microorganisms.
However, when communities of larger taxa such as
macroinvertebrates or fishes are used, evidence of
the adequacy of sampling (e.g., a species-area or
species-effort curve) is needed to determine the ap-
propriate scale and number of replicates. When the
number of taxa is large, binary data are sufficient to
detect changes in community structure.
The example analyses presented here use bi--
nary data or rated abundance, although the meth-
ods are applicable to indices derived from
weightings with approximately continuous data
such as individuals per species or .continuous data
such as biomass or biovolume per species. Obvi-
ously, enumeration of individuals ignores size dif-
ferences among species; however, these estimates of
population sizes within replicate physical samples
are appropriate for comparisons among sites or
treatments. The analyses are appropriate for the
analysis of multispecies laboratory experiments, arid
for comparisons of replicate samples among field
sites. . - . . •
Data Requirements
The starting point for analyses is the development
of a taxonomic summary of species presence (or
abundance) in replicate physical samples by treat-
ment or site. Replicate physical samples of the
within-sife or within-trdatment conditions are,re-
quired. The number of replicates may vary, but at
least three or four replicate samples per site are
necessary to obtain adequate estimates of variabil-
ity. Taxa may be identified to any practical taxo-
nomic level, although the rigor of the analyses and
the conclusions that can be drawn will depend on
the level of taxonomic precision chosen. Family-
93
-------
J. tf. PRATT and E P. SMITH
level classification will likely increase similarity
while species-level taxonomy will decrease similar-
ity among replicates. Two example data sets are
shown in Tables 1 and 2.
Selection of Community Measure
As previously mentioned, the selection of commu-
nity association measures will, in part, influence
the outcome of analyses. Similarity measures are
based on comparisons of shared and unique taxa
between pairs of samples (Fig 1). Examples of com-
monly used measures of both similarity and dis-
tance are shown in Table 3. Negative matches (taxa
failing to occur in both samples) are often ignored
in similarity measures because the absence of a par-
ticular taxon may not be judged important
(Roback, 1974). However, when taxa have particu-
lar indicator value, the absence of a taxon in two
samples may make the association between the
samples stronger. This is especially useful when
important taxa are expected at study sites. For ex-
ample, the absence of red oak trees at two sites
might be considered important in associating those
Sample j
Sample i
Species Species
present absent
Species
present
Species
absent
a
- C
b
d
Figure 1.—Association matrix used to compute similarity
measures. .
sites, and this importance should be reflected in the
choice of the community similarity measure. Bi-
nary, ranked, and continuous data can be used to
form many of the indices.
Some measures or their complements are met-
ric; that is, they satisfy the triangle inequality princi-
ple. Other measures that fail to satisfy the triangle
inequality are termed semimetric (Legendre and
Legendre, 1983). Semimetric measures may have
less predictable statistical properties. A difference of
a given magnitude between semimetric coefficients
may not have the same meaning for all values of the
Table 1.—Sample data set showing ranked abundance of taxa (scaled 1-5) by treatment and replicate.
TREATMENT
TAXON
A
A
C
C
C
C
C
O
F
F
6
G
G
G
M
M
N
N
N
N
N
N
N
P
S
S
S
S
S
S
Tola!
sp1
sp2
sp1
sp2
sp3
sp4
sp5
spl
spl
sp2
spl
SP2
sp3
sp4
SP1
sp2
spl
Sp2
sp3
sp4
sp5
sp6
sp?
spl
spl
sp2
sp3
sp4
SP5
sp6
laxa
1
0 0 1
.555
1 1 1
1 1 1
000
1 1 1
1 1 1
000
1 1 1
1 1 1
0 0 1
1 1
1 1
1 1
1 1
1 1
555
1 1 1
111
000
1 0 0
1 1 1
1 1 1
0 1 0
100
0 1 0
0 1 0
1 1 1
1 1 1
555
22 22
24
2
000
444
1 1 1
1 1 1
1 0 0
0 1 1
1 1 1
000
1 1 1
1 1 1
000
0 1 1
0 1 1
1 1 1
111'
1 1 0
444
0 1 0
011
000
1 1 1
1 1 1
111
0 1 0
100
1 0 1
1 1 0
0 1 -1
1 1 0
444
19 20 -
23
3
010
333
0 1 1
1 1- 0
000
1 0 1
1 1 0
000
0 1 1
1 1 1
0 0 1
0 1 0
1 0 0
0 0 1
1 0 1
110
333
111 ,
001
1 0 0
1 0 1
111
0 0 1
1 0 0
1 1 0
0 1 1
000
101
1 1 0
333
19 17
16
4
0 1 0
222
1 1 0
001
• 1 1 1
0 1 1
1 1 0
110
0 1 1
1 1 1
022
1 0 1
1 1 0
0 1 0
100
101
222
1 0 0
0 1 1
1 0 1
000
0 0 1
1 0 1
010
000
000
1 1 0
0 1 1
1 1 0
222
17, 16
19
5
333
1 1 1
1 0 1
0 1 0
100
033
000
1 1 0
1 1 1
0 1 0
333
1 0 0
0 0 1
000
o. b 1
1 1 1
1 1 1
0 0 1
1 0 0
0 1 1
000
1 1 1
000
101
000
100
001
010
1 1 0
110
16 15
.15
6
044
1 1 1
0 1 0
000
1 1 1
555
1 0 0
'101
1 0 1
1 1 0
044
0 0 1
0 1 0
1 0 1
100
0 1 1
1 1 1
0 1 1
000
101
00 0
0 0 1
1 1 1
1 0 0
1 1 0
000
000
1 1 1
1 1 0
1 01
17 17
15
Note: Tho ordered treatment values correspond to,increasing concentrations of copper.
94
-------
Biological Criteria: Research and Regulation, 1991
Table 2.—Data from Table 1 rearranged by frequency of occurrence of taxa in replicates.
TREATMENT
TAXON
A sp2
N sp1
S sp6
F sp2
F sp1
C sp4
M sp2
N sp6
C sp1
S sp4
S sp5
N . sp7
C sp5 .
M sp1
N sp2
C sp2
G sp4
G sp2
G sp3
G sp1
N sp3
A sp1
C sp3
N sp4
P sp1
D sp1
N sp5
S sp1
S sp3
S sp2
1
555
555
555
1 1 1
1 1 1
1 1 1
1 1 1
•111
1 -1 1
1 1 1
1 ,1 1
1 1 1
111
1 1 1
1 1 1
1 1 1
1 1 1
1 f 1
1 1 1
0 0 1
1 1 1
0 0 1
000
0 0.0
0 1 0
000
1 0 0
1 0 0
0 1 0
0 1 0
Table 3. — Exemplary
COEFFICIENT
Simple matching
Jaccard
Czekanowski
Margalef
Community loss
4
4
4
1
1
0
1
1
. . 1
0
1
1
1
1
0
1
1
0
0
0
0
0
1
0
0
0
1
1
1
' 1
measures
2
4 4
4 4
4 4
1 1
1 1
1 1
1 0
1 1
1 1
1 1
1 0
1 1
1 1
1 1
1 0
1 1
1 1
1 1
1 1
0 0
1 1
0 0
0 0
0 0
1 0
0 0
1 1
0 0
1 0
01
3
3
3
1
0
1
1
1
0
. 1
1
0
1
1
1
1
0
0
1
0
b
0
0
1
1
0
1
1
0
0
of community
a +
a
2a
FORM
a + d
b + c + d
a
+ b + c
2a
+ b + c
3
3 3.
3 3
3 3
1 1
1 1
0 1
1 0 ,
11
•1...1
0 ,1 . .
1 0
0 1
1 0
0 1
1 -1 " .
1 0
0 1
1 0
0 0
0 1
0 1
1 d
0 0
0 0
0 0
00
01
1 0'
0 0
11'"
similarity.
t
4 ,
222
222
222
1 1 .1
0 1 1
.011
101
:0 0 1
110
0 1- 1
110
1 0 1
1 1 0
1 0 0
'1,0 0
0 0 1
0 1 0
1 0 1
110
0 2'2
0 1 1
0'1 0
1 1 1
1 0 1
0 1 0
1 1 0
000
000
110
000
t
. 1
1
. 0
1
b
1
1
: - - 1
0
1
•..-:-•• 0
0
0
0
0
0
1
0
, -. . 3
1
3
1
0
1
1
0
0
0
1
5
1 1
1 1
1 0
1 0
1 1
3 3
1 1
1 1
0 1
1 0
1 0
0 0
0 0
0 1
0 1
1 0
0 0
0 0
0 1
3 3
0 0
3 3
0 0
1 1
0 1
1 0
0 0
0 0
0 1
o d
Formulas are based on
a(a + b + c + d)
(a+b)(a + c)
c
a + b
CLASS
metric
metric
semimetric
semimetric
nonmetric
6
1 1
1 1
1 0
1 r
1 0
5 5
0 1
0 0
0 1
1 1
1 1
1 1
1 0
1 0
0 1
00
1 0
0 0
0 1
0 4
0 0
0 4
1 1
1 0
. 1-0
1 0
00
1 1
0 0
00
FREQUENCY
1 -
1
1
0
.1 •
5
1
1
0
1
0
1
0
0
1
0
1
1
0
4
0
4
1
1
0
1
o • •
0
0
0
terminology shown in
18
18
16"
15
15
14
14
14
13
13 '
13
12 '
. 11
11
11
10
10
10
10
9
9
8
8
7
7
6
6
6
6 '
6
Fig.1.
REFERENCE
Digby & Kerhpton, 1987
Jaccard, 1901
Czekanowski, 1909
Sorensen, 1948
Margalef, 1958
Courtemanch
1987
& Davies,
Note: Measures whose complements satisfy the triangle inequality are termed metric. Those that do not are semimetric. Measures whose complements
can take negative values are termed nonmetric. • '
coefficients. Still others are nonmetric because their
complements can be negative, and the application
of such coefficients may be problematic.
Many other measures of both community simi-
larity or distance can be used (Legendre and
Legendre, 1983; Digby and Kempton, 1987; Plafkin
et al. 1989). Among the more familiar to ecologists
and pollution biologists are Euclidian distance,
Manhattan (city block) distance, Bray-Curtis mea-
sure, Canberra measure, correlation coefficient com-
plement, Morisita's index, and Pinkham and
Pearson measure. A general form for community
measures (and their complement distance mea-
sures), is given by Gower (1971).
Analysis Method
A similarity matrix for all possible replicate pairs is
constructed using a chosen metric. A test statistic is
computed comparing the mean similarity of repli-
cate objects within "treatments" to the between
treatment similarity. If the test statistic for data is
L(data) = B /W
(1)
where_B is the mean between treatment similarity
and W is the mean within treatment similarity, then
this statistic can be compared to one derived from a
permutation procedure in which coefficients in the
95
-------
J. R. PRATT and E. P. SMITH
similarity matrix are randomly switched a large
number of times (1,000). A test statistic L (permute)
is recalculated as above after each permutation of
the matrix. The distribution of L (permute) can be
used to determine if L (data) can be differentiated
from L (permute) at a given alpha (a). For example,
if 1000 random switches (permutations) are made
then one would reject a null hypothesis of no dif-
ference in community similarity at a = 0.05 if L
(data) were more extreme than 950 (95%) of the L
(permute) values. Because the total similarity (T) is
a constant for any given matrix, the component B
and W similarities may also be used as test statis-
tics.
Similar arguments can be made for the use of
distance measures rather than community similar-
ity. However, in multidimensional space, the loca-
tion of similar samples is associated with a small
distance; dissimilar samples would be more widely
separated. Therefore, the principles of using dis-
tance measures as measures of similarity would
apply to testing community differences, but the ex-
pected direction of change in the randomly
switched matrix would be opposite to expectations
for similarity measures.
Follow-up Analyses
Assuming that a hypothesis of community similar-
ity among sites or treatments is rejected, several
follow-up analyses are possible. One approach
would be to conduct multiple comparisons using
the permutation procedure on treatment pairs. This
approach would test differences between individ-
ual treatments or sites. However, if the number of
replicates is small (as it often is in field studies), de-
tection power is limited by the fact that only a
small number of unique permutations of the simi-
larity matrix are possible.
A second useful follow-up analysis is to deter-
mine the relative contribution of each taxon to com-
munity similarity. The effect of taxa on similarity
can be determined by computing the effect on simi-
larity of removing each taxon; In this analysis, re-
moving common taxa will reduce total similarity.
Taxa adding heterogeneity to the matrix (and so de-
creasing similarity) will increase total similarity
when removed. Identification of these taxa permits
an inspection of the data matrix to identify taxa that
may appear or disappear in ceratin treatments or
sites. When coupled with a data matrix sorted by
the total frequency of occurrence of taxa, these iden-
tifications become easier (e.g., Table 2). Additional
follow-up analyses are summarized by Smith et al.
(1990).
Community Data Analysis
The hypothetical data set in Table 1 was con1
structed (based on actual data of effects of copper
on periphyton algae) to demonstrate a method for
detecting changes in community composition be-
tween sites or treatments. The presence of taxa was
determined from 500 cell counts of preserved sam-
ples. The ordered treatments (1-6) ranged from con-
trols (<10 ng Cu/L) to 80 ng Cu/L. The data matrix
has been edited to include only the 30 most com-
mon taxa and to magnify abundance differences
among taxa.
Results
Example Analysis—Similarity
Based on presence-absence data shown in Table 1,
a similarity matrix was constructed using the
Jaccard measure. Communities have higher simi-
larity if they share a greater proportion of their
total species. A portion of this matrix is shown in
Figure 2. Rectangular blocks in the table mark simi-
larity coefficients comparing samples between
treatments. Triangular areas lying along the matrix
•diagonal are coefficients showing the within treat-
ment similarity. Computation of within and be-
tween treatment similarity was followed by 1,000
permuta- tions randomly switching matrix coeffi-
cients to produce a hypothetical distribution of
within-treatment similarity (Fig. 3).
Thesfe analyses showed that the critical within-
treatment similarity based on a = 0.05 was 0.5155.
This compared to the actual with-treatment similar-
ity of 0.5271. Further, the number of permutation
similarity values that were more extreme than this
observed similarity was 17, corresponding to a p-
value of 0.018 (p = [17+1] /1000)). Therefore, the hy-
pothesis of no difference in similarity among
treatments is rejected.
Follow-up analyses showed that similarity of
treatments to controls (group 1) decreased with in-
creasing copper levels (Fig. 4) and that common
species strongly affected community similarity. Spe-
cies with moderate negative or positive influence
are typically those that are either eliminated in
higher treatments (e.g., C. sp2) or that become more
frequent in replicates at high copper levels (e.g., C.
sp3, D. spl). Based on multiple pairwise compari-
sons, only treatments 2 and 5 could be differenti-
ated.
96
-------
Biological Criteria: Research and Regulation, 1991
Treatment/Replicate , ;
1A . 1B 1C 2A 2B 2C
1 .80 .83 .64 .88 .78
1 .80 .62 .92' :75
1 .52 .80 .71
1 .62 .58
• . - . _
1 .75
' " ' • 1
3A
.67
.58
.54
.48
.64
.48
1
3B 3C
.58 .63
.56 .60
.58 ..63.
.59 . .50
.50 .60 :
,46 . -71
.48 .40
1 .38
1
... 6A
... .50
... .48
.44
.'..-' .50
. . . .48
... . .44
.52
... .32
.42
6B
.48
.41
.54
.36
.41
.31
'.43
.41
,:39
6C
.44
.43 '
.56
.33
.43 '
.38
:35
.38
. ' -i ;
. .48 ,
1A ...,,
1B '-
• r •
1C
2A .
2B '."'
2C,
3 A
3B
3C- ,
Figure 2.—Part of the matrix of Jaccard similarity coefficients from the examination of data In Tables 1 and 2.
Effect of Measure
Comparable analyses to those presented were done
using only presence-absence data to determine Eu-
clidian distance among treatments. Distance analy-
ses provide estimates of community dissimilarity:
dissimilar communities are more widely separated
in multidimensional space. These analyses are es-
sentially identical to those using presence—absence
based similarity (Fig. 5). The within-treatment dis-
tance estimate is less extreme than 7 of 1,000 per-
mutation-based within-treatment values (p =
7+1/1000 = 0.008), so the hypothesis of no differ-
ence in distance among treatments is rejected.
Follow-up analyses identified the same patterns
found using analysis with Jaccard's coefficient (Fig.
6). Evaluation of these data using several other pres-
ence-absence based metrics revealed essentially
similar patterns, although other metrics weight
common species more heavily and so produce coef-
ficients of greater magnitude.
Effect of Data Type
While presence-absence data were effective in de-
tecting differences among treatments, data evalu-
ated by rank abundance resulted in greater
detection power. For example, the rankings shown
in Table 1 weight only 6 of 30 species; three become
less abundant across treatments, and three become
more abundant. Examination of the weighted data
using Euclidian distance showed clear treatment
differences (Fig. 7) and allowed detection of addi-
tional differences between treatment pairs (Fig. 8).
Discussion
The structure of biological communities is vulnera-
ble to a host of potential adaptations in response to
environmental stressors ranging from pollutants, ,t.o
physical stresses, to habitat modification. The anal-
ysis of populations within communities often pres-
ents equivocal data: some species increase : in
abundance while others decrease. In extreme cases,
some species are locally extirpated, reducing com-
munity heterogeneity. In other cases, intolerant
species are replaced by stress-tolerant forms. •-,
• The application of community similarity and re-
lated analyses provides a means for comparing sites
under differing conditions. The above methods for
critically examining replicate communities at differ-
ing sites or in multispecies experiments provide a
means of inferentially locating and evaluating com-
munity change, i and identifying the species influ-
encing the change in community structure.
Data Type and Effect of Measure
Measures based on binary (presence-absence) data
limit the ability of investigators to detect commu-
nity differences. Analyses that place more weight
on common species in community pairs (for exam-
ple, Czekanowski) increase the absolute value of
the similarity index, but provide no greater detec-
tion power for comparing communities. Addition-
ally, the complements of these measures are
typically semimetric and are, therefore, less useful
indicators of relative difference (distance) among
communities. Measures such as the Coefficient of
97
-------
J. R. PRATT and E. P. SMITH
A
frequency
50.
45.
40.
35.
30.
25.
20.
15.
10.
5. '
n
u.
Total similarity
*
, . * ' * *.
1 ***** * * *
************* ,
* **************
***************** **
******************** * , ,
* ************************* *
*****************************
2 1 424534********************************44**5
4_ _i_ _i_ •
.4269 , .4537 , .4828 .5096
Jaccard similarity
73.7955 mean 0.4823
321
_|_
i
.5386
Between similarity 64.3072 mean 0.4763
Within similarity
Notes: The critical
The critical
9.4883 mean , 0.5271 .
value is the '950th value.
mean within similarity is 0.51 545.
Reject hypothesis of no effect if this value is smaller than the observed mean
Number >
observed within 17.
within similarity.
B
•-
1
2
3
4
5
6
Note
near
1
1.0000
1.0051
.9690
.8659
.7022
.7608
1
1
1
2
.0051
.0000 1
.0334 1
.8996
.6832
.7421
: Lambda values for pairwise comparisons,
1 indicate little difference between groups.
Treatment
3
.9690
.0334
.0000
.9900
.9607
.9552
ratios of mean
1
1
1
4
.8659
.8996
.9900
.0000
.0414
.1360
between to
5
.7022
.6832
.9607
1.0414
1.0000
.9639
mean within similarity.
6
.7608
.7421
.9552
1.1360
.9639
1 .0000
Values
Figure 3.—(A) Distribution of permutations of within similarity. (B) Pairwise comparisons (Lamda values) showing progressive
differences between treatments.
Community Loss are nonmetric and require addi-
tional investigation before they can be recom-
mended for detecting community differences.
Measures that place emphasis on missing species
are useful only when the base species list includes
important indicator species.
Measures based on binary data require species
loss and gain to detect differences in community
structure and are not sensitive to changes in abun-
dance. However, when the number of species exam-
ined is high (as for microbial communities), detec-
tion of community differences is greater that for less
diverse assemblages such as fishes. Schindler (1987)
has recently recommended small, rapidly reproduc-
ing species with poor dispersal capabilities as effi-
cient environmental monitors. Additional evidence
suggests that structural changes in ecosystems con-
sistently process changes (Pratt, 1990).
Measures that incorporate ranked, relative, or
absolute abundance as weightings for taxa in com-
98
-------
Biological Criteria: Research and Regulation, 1991
Species removed
A
C
C
D
F
F
N
N
N
N
N
P
S
S
S
S
sp2
sp2
sp3
sp1
sp1
sp2
spl
Sp3 ;
sp4
sp5
sp6
sp1
sp1
sp2
sp3
sp6
Influence
"-4.675
1.087
1.417
1.459
- 1 .956
- 1 .885
-4.675
1 .332
1.331
1.774
-1.109
1.330
1 .492
1.508
1 .544
-2.735
B
Trt. Trt.
. 1 2
3
1
1
1
1
2
2
2
2
3
3
3
4
4
5
4
5
6
3
4
5
6
4
5
6
5
6
6
Critical value
.38724
.39258
.39516
.38613
.39156
.39058
.39742
.38703
.38590
.38267
.39143
.39180
.39739
.38616
.38786
Between similarity
.73406
.59541
.52651
.43858
.47645
.55150
.47437
.37148 *
.40479
^40726
.41105
.41033
.43892
.48076
.42378 .
Note: The critical values below are for the multiple
comparisons on the mean between similarity.
Reject if the mean between similarity is smaller
than the critical value. Asterisk (*) denotes
detected difference.
Figure 4.—(A) Species influence scores for select species
from Tables 1 and 2. Scores indicate the magnitude and di-
rection of effect on community similarity when the species
is removed from the analysis. (B) Pairwise tests of treat-
ment differences based on between-treatment similarity.
Differences are detected at p = 0.1.
munity samples have improved the ability to detect
differences in structure. In these cases, measures are
sensitive to changes in abundance, not simply to
gain or loss of taxa. Distance measures are effective
in detecting the separation of samples in multidi-
mensional (multispecies) space; however, no rigor-
ous analysis comparing the relative powers or
sensitivities of these measures has been used to de-
tect biotic community differences that result from
stress.
Competing Analytical Methods
Several other methods for-evaluating community
structure data have been used to determine ecosys-
tem health. These methods include calculation of
diversity indices or indices of biotic integrity, and
the comparison of communities using cluster anal-
ysis.
Diversity indices compare sites based on spe-
cies abundance or some other measure of the distri-
bution of individuals or biomass for species present.
The most commonly used indices are based in infor-
mation theory (for example, Shannon and Weaver,
1949). While these indices provide numerical values
for diversity, including values for replicate samples
at a site, they do not account for the identity of the
component taxa. That is , two samples might have
the same information theory diversity and share no
species in common. Diversity indices have value in
comparing sites, and the computed index is often
strongly correlated with the number of taxa (Green,
1979), suggesting that species richness alone is a suf-
ficient measure of diversity. The concept of species
diversity is often ignored in investigations where
problem taxa are pooled (for example, chironomid
midge larvae) so that the resulting index is com-
puted from a mixture of taxa that are comprised of
species, genera, and families.
Indices of biotic integrity provide a score for
each of several biological criteria. Unlike diversity
indices, where the identity of the taxa is ignored, the
identity of collected taxa is of primary importance.
Indices such as the fish-based Index of Biotic Integ-
rity (IBI), (Karr et al. 1986) or the Invertebrate Com-
munity Index (ICI) (Ohio Environ. Prot. Agency,
1987) are based on a combination of concepts. These
indices are computed by assigning integer scores to
community metrics that are conceptually catego-
rized. •.-.'.
For example, the IBI separates metrics accord-
ing to species richness (all taxa and some indica-
tors), trophic composition, and abundance and
condition of specimens. Similar concepts are used in
the ICI, although there is more reliance on indicator
taxa. The scores assigned are derived from a subjec-
tive rating system .based on determination of re-
99
-------
J. R PRATT and E P. SMITH
frequency
50.
45.
40.
35.
30.
25.
20.
15.
10.
5.
0.
*
'•:'-• *
.*.•* **** *
• ************ *
' **************
* ***************
** ** **************** *
*********************** **
** ***************************
4**4*********************************4*4323
+—
.3233
Total distance
Between distance
Within distance
.3369 .3517
.Euclidian distance
53.9703 mean 0.3527
48.0703 mean 0.3561
5.9000 mean 0.3278
.3654
-+
.3801
Notes: The critical value is the 51st value.
The critical mean within distance is 0.33505.
Reject hypothesis of no effect if this value is smaller than the observed mean within distance.
Number < observed within 7. '
B
1
2
3
4
5
6
1
1.0000
.9830
1.0712
1.1715
1.3650
1.2963
Note: Lambda values for
Values
2
.9830
1 .0000
.9841
1.0835
1.2646
1.2149
Treatment
', . 3 .
• 1.0712
.9841
1.0000
1.0057
1.0219
1 .0236
pain/vise comparisons, ratios of the
near 1 indicate little difference
between groups.
1.
1.
1.
1.
mean
4
1715
0835
0057
0000
9789
9296
between
1
1
1
1
1
5
.3650
.2646
.0219
.9789
.0000
.0208
to the mean within
6 •
1 .2963
1.2149
. 1 .0236
.9296
1 .0208
1 .0000
distance.
Figure 5.—
-------
Biological Criteria: Research and Regulation, 1991
Species removed
A
C
C
D
F
F
N
N
N
N
N
P
S
S
S
S
sp2
sp2
sp3 •
sp1
sp1
sp2
sp1
sp3
sp4
sp5
sp6
sp1
sp1
sp2.
sp3
sp6'
Influence
.000 •
-2.157
-2:209 , ' ••'•-•
- 1 .896
-i.137
-1.155
.000
-2.142
- 1 .997
-2.048
-1.486
- 1 .954
- 1 .909 ,
-1.928
-1.921
-.788
B
Trt. Trt. Critical value *
1 2 .38717
.38667
.38612
.38813
.38776
.39018
.38646
.38768
.38789
.38773
.38783
.39041
.38657
.39088
1
1
1
1
2
2
2
2
3
3
3
4
4
5
3
4
5
6
3
4
5
6
4
5
,6
5
6
6
.38875
Between distance
.24979
.31570 •
.35068
.38352
.37036
.32776
.36588
.40385 *
-,'. .39370 *
.38046
.36781
.37327
.35689
.34328
.35822
Note: The critical values below are for the multiple
comparisons of the mean between distance.
Reject if the mean between distance is larger than
the critical value. Asterisk (*) denotes detected
difference.
Figure 6.—Follow-up analyses. (A) Species influence
scores based on Euclidian distance. (B) Pairwise tests of
treatment differences .based on between-treatment similar-
ity. Differences are detected at p = 0.1.
when the number of samples is fewer than the num-
ber of variables (species), as is often the case in envi-
ronmental analysis. Assumptions of normal
distribution of variables are often violated when
species abundances are zero at some sites.
Summary and Conclusions
Choice of data type and measure is important and
should be based on the hypothesis being tested.
The similarity measure used affects the outcome of
analyses. The numerical value of the similarity
measure is influenced by the functional form of the
measure, transfprmations applied to the data prior
to analysis, and the type of data. Choice of data
limits the type of changes that can be detected. For
example, presence-absence data are used primarily
to assess changes in species composition. Abun-
dance; data, in contrast, reveal decreases or in-
creases in the relative abundance of species.
With some measures, dominant species may
, strongly influence the measure. Changes in the rela-
tive abundance of a dominant species may not re-
sult in the loss of a species; thus, measures based on
presence-absence may not reflect changes in the
dominant species. In fact, if a species undergoes a
large change but is not absent, the species may have
a positive (stabilizing) effect in a presences-absence
measure. On the other hand, changes that result in
loss of species may not be reflected in measures
based on proportional abundance unless they are
accompanied by strong changes in the relative
abundance (the total relative abundance of the spe,-
cies absent must be moderate when they are pres-
; eht). .. - - ,
Biological criteria for evaluating individual and
cumulative effects of stresses might be based on
comparisons with regional or upstream reference
communities, although often no acceptable up-
stream corollary can be found. The usefulness of bi-
ological criteria for evaluating ecosystem health will
be.determined by the scientific adequacy of the
analyses applied for detecting community change.
At the present time, critical evaluations of analytical
tools are needed to determine the appropriateness
pf available measures for detecting community
change. Critical detections cannot be made by indi-
ces that are only descriptive and lack measures of
variability.
ACKNOWLEDGEMENTS: Portions of this work were
supported by the U.S. Army, Biomedical Research and
Development Laboratory, Fort Detrick, Maryland. The
opinions and conclusions are those of the authors and do
hot necessarily represent those of the U.S. Army.
References
Cairns, J. and J.R. Pratt. In press, A history pf biomdnitoring
using invertebrates. In V. Resh and ,D: Rosenberg, eds.
Freshwater Biomonitoring and Benthic Macroinverte-
•brates. Chapman and Hall, New York.
Courtemanch, D.L. and S.P. Davies. 1987. A coefficient of
community loss to assess detrimental change in aquatic
communities. Water Res. 21:217-22.
101
-------
J. R. PRATT and £ P. SMITH
frequency
50.
45.
40.
35.
30.
25.
20.
15.
10.
5.
0.
* *
* **
********
** ***********
***************
** ******************
* ***********************
****************************
11 1 3215**31*********************************421
'
.4459
.5054 .5698
Euclidian distance
.6293
.6938
Total distance
Between distance
Within distance
92.8750
86.1072
6.7679
mean
me.an
mean
0.6070
0.6378
0.3760
Note: The critical value is the 51st value.
The critical mean within distance is 0.52928.
Reject hypothesis of no effect if this value is smaller than the observed mean within distance.
Number < observed within 0. •
B
1
2
3
4
5
6
Note:
1
1.0000
1.2057
1.5910
2.0498
2.8655
2.5655
Treatment
2 3 4
1.2057 1.5910
1.0000 1.1144
1.1144 1.0000
1.4945 1.1155
2.2802 1.6803
2.1458 1.7064
Lambda values for pairwise comparisons, ratios of the
Values near 1 indicate
little difference between groups.
2.0498
1.4945
1.1155
1 .0000
1 .3238
1 .4395
mean between
• 5
2.8655
2.2802
1 .6803
1 .3238
1 .0000
1.1116.
to the mean within
6
2.5655
2.1458
1 .7064
1 .4395
1.1116
1.0000
distance.
Figure 7.—(A) Distribution of permutations of within-treatment distance for the example using rank abundance data. (B) Pair-
wise comparisons.
Cummins, K.W. and M.J. Klug. 1979. Feeding ecology of
stream invertebrates. Annu. Rev. Ecol. Sys. 10:147-72.
Czekanowski, V. 1909. Zur differential diagnose der
Neandertalgruppe. Korrespbl. dt. Ges. Antwop. 40:44-
47
Digby, P.G.N., and R.A. Kempton. 1987. Multivariate Analy-
sis of Ecological Communities. Chapman and Hall,
London.
Gauch, H. 1982. Multivariate Analysis in Community Ecol-
ogy. Cambridge Univ. Press, London.
Gower, R. 1971. A general coefficient of similarity and some
of its properties. Biometrics 27:857-72.
Green, R.H. 1979. Sampling Design and Statistical Methods
for Environmental Biologists. John Wiley & Sons, New
York.
Hilsenhoff, W. 1982. Using a Biotic Index to Evaluate Water
Quality in Streams. Tech. Bull. No. 132. Wise. Dep. Nat.
Resour., Madison.
Hurlbert, S.H. 1971. The nonconcept of species diversity: a
critique and alternative parameters. Ecology 52:577-86.
Jaccard, P. 1901. Distribution de la flore alpine dans le Bassin
des Dranses and dans quelques r)gions voisines. Bull.,
Soc. vahd. Sci. Nat. 37:241-72.
Karr, J. K.D. Fausch, P.L. Angermier, P.R. Yant, and I.
Schlosser. 1986. Assessing biological integrity in run-
102
-------
Biological Criteria: Research and Regulation, 1991
Species removed
A
C
C
D
F
F
N
N
N
N
N
P
S
S
S
S
sp2
sp2
sp3
sp1
sp1
sp2
sp1
sp3
sp4
sp5
sp6
sp1
sp1
sp2
sp3
sp6
Influence
-5.915
-.674
- .746
-.592
-.387
-.337.
-5,915,
-.699,
- .640
- .696
-.489
- .669
-.659
-.667
-.686
-7.015
B
Trt. Trt. Critical value
1
1
1
1
1
2
2
2
2
3
3
3
4
4
5
2
3
4
5
6
3
4
5
6
4
5
6
5
6
6
.78949
.79348
.77944
.79077
.82022
.81480
.79738
.79219
.78752
.79658
.79791
.82326
.79434
.80127
.78662
Between distance
.30638
.46888
.63923
.90158 *
.98559 *
.37119
.52340
.80488 *
.90666 *
.43593
.66137
.79031
.54372
.69137
.53696
Note: The critical values below are for the multiple
comparisons of the mean between distance.
Reject if the mean between distance is larger than
the critical value. Asterisk (*) denotes detected
difference.
Figure 8.—(A) Species influence scores based oh Euclidian
distance determined from rank abundance. (B) Palrwlse
tests of treatment differences based on between-treatment
similarity. Differences are detected at p = 0.1.
ning waters: a method and its rationale. Spec. Publ. 5.111.
Nat. History Surv., Champaign.
Kolkwitz, R. and M. Marsson. 1908. kologie der pflanzlicheri
Saprobien. Berichte Deuts. Botan. Gesell. 26A:505-19.
Legendre, L. and P. Legendre. 1983. Numerical Ecology.
Elsevier, Amsterdam.
Levine, S.N. 1989. Theoretical and methodological reasons
for variability in the responses of aquatic ecosystem
processes to chemical stresses. Pages 145-80 in S.A.
Levin, M.A. Harwell, J.R. Kelly, and K.D. Kimball, eds.
Ecotoxicology: Problems and Approaches. Springer-
Verlag, New York.
Lowe, R.L. 1974. Environmental Requirements and Pollution
Tolerance of Freshwater Diatoms. EPA-670/4-74-005.
U.S. Environ. Prot. Agency, Washington, DC.
Margalef, R. 1958. Information theory in ecology. Gen. Svs
3:36-71. *
Metcalfe, J.L. 1989. Biological water quality assessment of
running waters based on macroinvertebrate communi-
ties: history and present status in Europe. Environ.
Pollut. Serv. A 60:101-39.
Odum, E.P. 1985. Trends Expected in Stressed Ecosystems.
BioScience 35(7):419-22.
-.1990. Field experimental tests of ecosystem-level
hypotheses. Trends Ecol. Evol. 5:203-36.
Ohio Environmental Protection Agency. 1987. Biological Cri-
teria for the Protection of Aquatic Life: Vol. II. Users
Manual for Biological Field Assessment of Ohio Surface
Waters. Ohio Div. Water Qual. Monit. Assessment, Co-
lumbus.
Omernik, J.M. 1987 Ecoregions of the conterminous United
States. Ann. Ass. Am. Geogr. 77:118-25.
Pielou, E.G. 1984. The Interpretation of Ecological Data
Wiley, New York.
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/444/4-89-001. Off. Water, U.S. Environ. Prot.
Agency, Washington, DC.
Pratt, J.R.'1990 Aquatic community response to stress: pre- •
diction and detection of adverse effects. Pages 16-26 in
W.G. Landis and W.H. van der Schalie, eds. Aquatic
Toxicology and Risk Assessment, 13th Vol. Am. Soc.
Test. Mater,, Philadelphia, PA.
Pratt, J.R., J. Cairns, and P.M. Stewart. 1985. Development of
microbial communities in mined lands. Pages 239-59 in
R.P. Brook, D.E. Samuel, and J.B. Hill, eds. Wetlands and
Water Management of Mined Lands. Pa. State Univ.,
University Park.
Roback, S.S. 1974. Insects (Arthropoda: Insecta). Pages 275- ,
376 in C.W. Hart and S.L.H. Fuller, eds. Pollution Ecol-
ogy of Freshwater Invertebrates. Academic Press, New :
York.
Schaeffer, D.J., E.E. Herricks, and H.W. Kerster. 1988. Ecosys-
tem Health I. Measuring ecosystem health. Environ.
Manage. 12:445-55.
Schindler, D.W. 1987. Detecting ecosystem responses to an-
.thropogenic stress. Can. J. Fish. Aquat. Sci. 44(SuppL):5-
Shannon, C.E. and W. Weaver. 1949. The Mathematical The-'
ory of Communication, Univ. of 111. Press, Urbana.
Smith, E.P., K.W. Pontasch, and J. Cairns. ,1990. Community
similarity and the analysis of multispecies environmen-
tal data: a unified statistical approach. Water Res.
24:507-14.
Sorensen, T. 1948. A method of establishing groups of equal
amplitude in plant sociology based on similarity in spe-
cies content. Biol. Skr. K. danske Vidensk. Selsk. 5:1-34.
Whittier, T.R., R.M. Hughes, and D.P. Larsen. 1988. Corre-
spondence between ecoregions and spatial patterns in
stream ecosystems in Oregon. Can. J. Fish; Aquat. Sci
45:1264-78.
103
-------
Errors in Errors in Hypothesis Testing
Brock B. Bernstein
Robert W. Smith
EcoAnalysis, Inc.
Ojai, California
ABSTRACT
Clear hypotheses are the.basis for evaluating whether biological criteria are met in achieving
environmental goals. Their logical structure reflects fundamental management concerns and
scientific questions. Unfortunately, hypotheses are often tested or evaluated improperly in en-
vironmental studies. This paper demonstrates the importance of selecting the proper error
term for statistical tests of impact or compliance. A straightforward, nontechnical discussion
of the role of the error term in hypothesis tests shows how the choice of error term controls the
question that is actually being asked. Simple simulation results demonstrate how the choice of
the wrong error term can completely invalidate significance tests. This is an important issue
for two reasons. First, the use of inappropriate error terms in environmental literature sug-
gests that environmental scientists are not yet familiar enough with these concepts. Second,
many packaged statistical programs automatically default to error terms that are not always
appropriate, thus increasing the chance of producing erroneous hypothesis tests.
Introduction
Hypothesis testing is often considered one of the
central activities of science, separating it from other
technical and intellectual disciplines (Platt, 1964).
The absence of clear hypotheses and unambiguous
hypothesis tests forms the basis for much of the
criticism of environmental studies (e.g., Beanlands
and Duinker, 1983; Natl. Res. Counc. 1986, 1990;
Fritz et al. 1980). Such critiques have resulted, in an.
emphasis on planning environmental studies
around null hypotheses that can provide the basis
for conclusions about impacts and other changes.
This emphasis has, on the whole, been benefi-
cial. However, too often null hypotheses are
thought of as merely simplistic statements that ex-
press the opposite of an expected event. Thus, "the
discharge will have no effect on fish populations" is
typical of hypotheses of this type. At a somewhat
more sophisticated level of study design, the impor-
tance of beta as well as alpha errors in hypothesis
tests is taken into account (e.g., Henkel, 1976; Sokal
and Rohlf, 1981). This has led to a recent focus on
the utility of power tests and optimization analyses
(e.g., Bernstein and Zalinski, 1983) that systematize
the statistically efficient allocation of limited sam-
pling resources.
Both these improvements have enhanced the
rigor arid utility of environmental studies. How-
ever, another fundamental issue of overriding im-
portance affects study design and the ability to
make logical inferences. This is the selection of the
proper error term in hypothesis tests. The choice of
error term controls the questions that are asked and
affects, or should affect, the sampling or measure-
ment design. In addition, inadvertently using the
wrong error term can completely invalidate signifi-
cance tests.
These ideas are not new and in fact are well
known among statisticians. In spite of this, there is a
consistent lack of attention to this issue and/or use
of inappropriate error terms in environmental stud-
ies. This is particularly true in the gray literature of
project reports and policy documents, suggesting
that these ideas are perhaps inaccessible as pre-
sented in the statistical literature. They are therefore
worth restating, in simpler terms and with straight-
forward examples that make them more under-
standable to practicing environmental scientists.
1 Thus, as much as possible, this report avoids formal
statistical terminology and definitions. Similarly,
only the main ideas are presented, without the de-
tail that a more developed treatment would include.
104,
-------
Biological Criteria: Research and Regulation, 1991
Error Terms Plain and Simple
The structure of the routine F test in the analysis of
variance (ANOVA) illustrates some basic princi-
ples. The F test, which is based on the ratio of two
variances, is analogous to a signal to noise ratio.
The numerator of the test contains the signal plus
background noise and the denominator of the test
only the background noise (Equation 1).
signal + noise
F = — — (Equation 1)
noise
If the numerator is larger than the denominator
by a predetermined amount, it can be concluded
that there is a change, impact, or effect (some kind
of signal) that is not due to chance. Since this is a
ratio test, the choice of the denominator term is crit-
ical.
The background noise in the F test is the vari-
ability in the system that stems from sources other
than those that create the signal, but that could ob-
scure, confound, or bias our perception of the sig-
nal. As one example, the success of a study of how
some part of the environment responds to improve-
ments in effluent treatment over a period of years
could be affected by the b^cicgrbund noise of naru- '
ral temporal changes in the environment. Many im-
portant technical elements are involved in a formal
statistical definition of the signal and noise terms.
However, for these purposes, only the three general
points are stressed.
1. The noise term can be made up of several
kinds or sources of variability (see Fig. 1).
Rarely does only one kind of variability con-
tribute to the background noise in a system!
For instance, in the effluent treatment exam-
ple given, natural temporal variability cafi
include diel, seasonal, yearly, and longer-
term patterns, as well as less predictable dis-
turbances.
2. The noise terms in the numerator and de-
nominator of the F test should be identical,
including the same sources of variability.
The only difference between the variances in
the numerator and denominator should be
the presence of the signal term in the numer-
ator. If the two noise terms are not identical,
then the denominator will be artificially in-
flated or deflated and the significance test
i . will be biased.
3. The sampling or measurement plan should
be carefully designed to capture the impor-
tant sources of background noise that can-
hot be controlled for. These cannot be
1 "• '• included in the hypothesis test if they are
not measured properly.
What is the Real Question?
Hypothesis tests can be thought of as evaluating
questions about events or changes in the environ-
ment. Because they do this by comparing the po-
tential signal to the background noise or error term
(see Equation 1), different error terms result in dif-
ferent questions, even if the signal is identical in all
cases. Lack of awareness of this fact leads to two re-
lated kinds of mistakes in environmental studies.
First> researchers may in fact.be asking one ques-
^^^^
STUDY DESIGN
Figure 1
QUESTIONS
SIGNAL
Condition X
Location
interaction
, ERROR TERM
time . . . -
time X location
random sampling error
EQUIVALENT
QUESTION
Question 1A
Question 1B
fr
the
105
-------
S. B. BERNSTEIN and R. W. SMITH
Elements of Sampling
Plan
• Sample impact and
control locations
• Sample several
stations within each
location
• Sample several
instantaneous
replicates at each
station
• Sample all stations
several times in the
before impact
condition and also in
the after impact
condition
• Average data from
the impact stations at
each sampling time.
Do the same for the
control stations.
• Calculate the
difference between
the impact and
control averages to
derive a difference
score for each
sampling time.
Schematic
Impact Control A
Before:
tn
A
A
After:
A
A
Sources of Variation
Condition: difference
between the before
and after conditions
Location: difference
between control, and
impact locations
Condition X Location:
change from the
before to the after
condition, in the
difference between the
control and impact
locations
*Time: changes over
time common to all
stations
*Time X Location:
changes over time in
the differences
between locations
*Random Sampling
Error: irreducible
differences between
samples collected at
the same time and
place
Flaura 1 —The example study design used to Illustrate the importance of selecting the proper error term. This study design is
termed the BACI model (Before, After, Control, Impact) (Bernstein and Zalinski, 1983; Stewart-Oaten et al. 1986). The sche-
matic shows the logical structure of the design, where the t's represent successive sampling times in each period and the a
the difference between Impact and control locations at each time. Important sources of variation are shown. Those marked by
an asterisk are components of the background noise.
tion when they think they are asking another. Sec-
ond, the question that in fact is being asked may be
ecologically meaningless or irrelevant.
Figure 1 shows an example study design that is
the basis for discussion in this section. Table 1 illus-
trates how specifying different error terms in this
design leads to the evaluation of quite different
questions. Typically, questions or null hypotheses
specify only the signal term, equivalent to the first
part of the questions in Table 1. In contrast to this,
the complete questions in Table 1 include descrip-
tions of both the expected signal and the back-
ground noise. Framing questions more thoroughly
in this way makes it easier to identify and avoid the
two kinds of mistakes described in the preceding
paragraph.
106
-------
Biological (Maria: Research and Regulation, 1991
In the example shown in Figure 1 and Table 1,
the Condition x Location interaction is the numera-
tor (or signal plus noise term) in the F test. This is
because an impact has occurred if the difference be-
tween the impact and control locations, averaged
over several times in the after condition, changes
compared to the difference between the impact and
control locations, averaged over several times in the
before condition (Green, 1979; Bernstein and
Zalinski, 1983; Stewart-Oaten et al. 1986). Figure 2
shows how this interaction, a change in the differ-
ence between control and impact, can be visualized.
Even with this correct signal term, however, com-
paring it to the different error terms in Table 1 leads
to quite different conclusions about whether an im-
pact has occurred and about the nature of the im-
pact.
For example, comparing this signal to the ran-
dom sampling error indicates an impact has oc-
curred if the Condition x Location interaction is
large compared to the differences among samples
taken at a single point in time and space. In contrast,
using the Time x Location interaction (see Figs.,1
and 2) as the error term means that an impact has
occurred only if the signal is large compared to the
natural temporal variability in the differences be-
tween locations. This is more ecologically appropri-
ate because it includes the natural temporal
variability in the spatial comparison (between im-
pact and control locations) that is the basis of the
impact hypothesis (see Fig. 2).
Far from being an obscure statistical detail,
choosing alternative error terms establishes quite
different ecological criteria for deciding if an impact
has occurred. By doing so, these different terms also
establish quite different definitions of just what an
impact is. In the first case, an impact will be any
change larger than those that occur on a very small
spatial scale at a single point in time. In the second
case, an impact will be any change larger than those
that occur between more widely spaced locations
over short to moderate time scales. Since these alter-
native error terms will most likely differ in magni-
tude, a signal that would be a significant impact
when compared to one would not be statistically
significant when compared to the other.
The other potential error term shown in Table 1
raises similar issues. Using as the background noise
the changes over time that are common to all sta-
tions (as suggested by Green, 1984, 1987) includes
temporal variability but ignores the necessary spa-
tial component of a study that compares two loca-
tions. It is easy to imagine a situation in which a
large temporal change affects the entire study area
(e.g., storm, El Nino, regional anoxia, population
-------
B. B. BERNSTEIN and R. W. SMITH
pact is. The examples also show that some of these
criteria are not ecologically meaningful and that ap-
propriate error terms must be selected carefully.
Practitioners must also exercise care in this regard
when using statistical software packages. In most
packages, the default error term is the random sam-
pling error and not all packages give users the op-
tion of defining error terms as needed.
When Significance Tests Lie
The preceding discussion has shown how the
choice of error term influences the question that
hypotheses tests are actually asking. This choice
also affects the validity of statistical tests of signifi-
cance. Tests performed with inappropriate error
terms are essentially meaningless since the numer-
ator of Equation 1 is being compared to an irrele-
vant denominator. Such faulty significance tests
can be insidious since they appear outwardly well-
founded. The following example presents a simple
simulation study that illustrates the serious conse-
quences of basing significance tests on incorrect
error terms. It shows that using the incorrect error
term can produce significant results a high percent-
age of the time even when the simulated data con-
tain no impact.
The results of significance tests were simulated
with the study design described (Figs. 1 and 2)
using two different error terms. In each simulation,
the data contained no impact, only random noise. In
the first case, the correct error term was used, in-
cluding both the Time x Location interaction and
the random sampling error. In the second case, the
interaction was omitted and the random sampling
error was used alone. In both cases, the ratio be-
tween the random sampling error arid the interac-
tion was systematically varied. In the second case,
this allowed investigation- of what happens to the
significance test as the error term reflects a greater
or lesser proportion of the actual background noise
in the data.
Line B in Figure 3 shows that the significance
test using the correct error term finds significant re-
sults (i.ev a false impact) approximately 5 percent of
the time, thus reflecting the true alpha level in the
simulation of .05. Since the error term contains both
components of the background noise, the test is not
sensitive to their relative magnitudes. In contrast,
Line A in Figure 3 shows that significance tests
using only the random sampling error as the back-
ground noise, or error term, produced incorrect and
highly variable results, even though the simulated
data contained no impact.
$ 0.8
OC
I
&
S)°-6
OT
s
0.4
o
0.2
0 OS 1.0 1.5 2.0 2.5 3.0 3.6 4.0 4.5 5.0 5.5 6.0 M 7.0 7.5 8.0
SD(REP)/SD(TXL)
Figure 3.—Sensitivity of a significance test to two different
•rror terms. Uns A shows results of simulated significance
tests using only the random sampling error (REP) as the
error term. Line B shows results of simulated tests using
the correct error term, including the standard deviation (SD)
of both random sampling error (HEP) and the Time x Loca-
tion variability (TxL). The simulated data contain no Impact,
only random noise. Line A shows that the correct error term
produces falsely significant results only 5 percent of the
time, equal to the expected alpha level of false Impacts. Line
B shows that the rate at which the incorrect error term pro-
duces falsely significant results Is extremely dependent on
the ratio of the two background variabilities. (See text for
further detail.)
The percentage of simulations that indicated a
false impact depended on how much of the total
background noise was made up by the random
sampling error. Thus, when the random sampling
error was roughly equal to the Time x Location in-
teraction, the test produced falsely significant re-
sults a high percentage of the time (83 percent). It
was not until the random sampling error became
much larger than the Time x Location interaction (to
the right on Fig. 3) that the percentage of falsely sig-
nificant results fell and the test became more accu-
rate. In this second case, the accuracy of the
significance test depended, not on the size of any
impact (there was none), but only on how much of
the total background noise happened to be captured
by the random sampling error.
This example demonstrates that a flawed signif-
icance test can produce a wide variety of results, de-
pending on how inaccurate the error term is in
108
-------
relation to the true background error. It should go
without saying that statistical tests that produce
such wildly variable results are worse than useless.
They are dangerous because they can indicate an
impact is present when in fact none has occurred.
Conclusions
Hypothesis testing is a critical step in determining
whether environmental conditions have changed
or whether compliance criteria have been met.
Framing clear and specific null hypotheses helps in
ensuring that hypothesis testing will produce valid
and useful information. In this context, several
points will help environmental scientists construct
and carry out valid and powerful hypothesis tests.
1. Hypothesis tests are ratio tests, analogous to
signal to noise ratios, in which the variance
of an expected signal term is compared to
the variance of a background noise term.
2. Null hypotheses will be more informative
and accurate if they are framed in terms of
both the expected signal and the back-
ground noise against which the signal will
be compared.
3. The background noise, or error term, must
, be carefully structured to cbntain the correct
components. If this is not done, the error
term will be artificially inflated or deflated,
leading to biased significance tests. '
4. The choice of error term controls the ques-
tion actually being asked by the significance
test. Questions resulting from inappropriate
error terms can be ecologically meaningless.
5. Using the wrong error term can lead to erro-
neous significance tests and to conclusions
that an impact has occurred when in fact
none exists.
Biological Criteria: Research and Regulation, 1991
While these concepts are familiar to statisti-
cians, we have presented them, along with related
examples, in straightforward and relatively non-
technical language. It is our hope that this will make
them more accessible and understandable to envi-;
ronmental scientists who may not have the time or
the background to interpret the statistical literature.
References
Beanlands, G. E. and P. N. Duinker. 1983. An Ecological
Framewprk for Environmental Impact Assessment in
Canada. Inst. Resour. Environ. Stud., Dalhpusie Univ.,
Halifax, Nova Scotia. ' •'
Bernstein, B. B. and J. Zalinski. 1983. An pptimum sampling
design and power tests for environmental biologists. J.
Environ. Manage. 16:35-43.
Fritz, E. S., P. J. Rago, and 1.D. Murarka. 1980. Strategy for as-
sessing impacts of power plants on fish and shellfish
populations. FWS/OBS-80/34. Natl. Power Plant Team,
Off. Biolog. Serv. Fish Wildl. Serv., U.S. Dep. Interior,
Ann Arbor, ML
Green, R. H. 1979. Sampling Design and Statistical Methods
. for Environmental Biologists. Wiley-Interscience, John
Wiley and Sons, New York.
. 1984. Statistical and nonstatistical considerations for
environmental monitoring studies. Environ. Monitor.
Assess. 4:293-301.
—-. 1987- Statistical and mathematical aspects: distinc-
tion between natural and induced variation. Pages 335-
54 in V.B. Vouk, G. C. Butler, A. C. Upton, D. V. Parke
and S. C. Asher, eds.. Methods for Assessing the Effects
of Mixtures of Chemicals. Wiley Publisher, Chichester,
England.
Henkel, R. E. 1976. Tests of Significance. Sage Publications,
Beverly Hills, CA.
National Research Council. 1986. Ecological Knowledge and
Environmental Problem Solving. Natl. Acad. Press,
Washington, DC. .
-. 1990. Managing Troubled Waters: The Role of Marine
Environmental Monitoring. National Academy, Press,
Washington, DC.
Platt, J. R. 1964. Strong inference. Science 146:347-52.
Sokal, R. R. and F.J. Rohlf. 1981. Biometry. W. H. Freeman
and Co., New York.
Stewart-Oaten, A.,W. W. Murdoch, and K. R. Parker. 1986.
Environmental impact assessment: "pseudoreplication"
in time? Ecology 67: 929-40.
109
-------
The Integrated Biosurvey as a Tool for
Evaluation of Aquatic Life Use Attainment
and Impairment in Ohio Surface Waters
Chris O.Yoder
Division of Water Quality Planning & Assessment
Ecological Assessment Section
Ohio Environmental Protection Agency
Columbus, Ohio
ABSTRACT
The Ohio Environmental Protection Agency recently incorporated biological criteria ("biocriteria") into its
water quality standards regulations. Numerical biological criteria were derived by utilizing the results of
sampling conducted at "least impacted" regional reference sites. Fish and macroinvertebrate data from
more than 300 Ohio reference sites were used to establish attainable, baseline expectations within the
framework of an existing system of tiered aquatic life use designations. Attainment status is determined as
beine "full" (all biocriteria are met), "partial" (one organism group reflects attainment, but the other does
not), or non (none of the biocriteria are met or one organism group reflects a poor or very poor condition).
An attainment status table is constructed using these guidelines. The diagnosis of observed aquatic life use
impairment relies on an integrated assessment of available biological, chemical, physical habitat, bioassay,
pollution source, and general watershed information. This approach is employed extensively in the Clean
Water Act section 305b reporting process and in support of regulatory program efforts. While all available
biological and chemical criteria are utilized, considerable reliance is placed on the integrated interpretation
of these data by the scientists who actually conduct the field sampling and evaluate the results. Detailed,
site-specific knowledge of complex study areas in combination with these varied types of monitoring data
is necessary to accomplish an environmentally accurate assessment. No single tool alone can accomplish
this level and power of assessment. A common criticism of biosurvey information is that it lacks the ability
to distinguish between different types, causes, and sources of impairment. The emergence of mulhmetric
biological evaluation tools and a rigorous, standardized approach to field assessment has provided the de-
tail necessary to establish biological response patterns and distinguish between general impact types: The
Ohio Environmental Protection Agency is currently working to develop biological' response signatures
that consist of key response components of the biological data that consistently indicate one type of impact
over another. Further refinement of this tool should have a profound influence on both site specific and
statewide assessments and should be an important consideration in some of the biocriteria policy issues
that are currently being debated.
Introduction
The monitoring of surface waters and evaluation of
the biological integrity goal of the Clean Water Act
have historically been dominated by nonbiological
measures such as chemical/physical water quality
(Karr et al. 1986). While this approach may have
fostered an impression of empirical validity and
legal defensibility, it did not sufficiently measure
the ecological health and well-being of the aquatic
resource. This point was demonstrated in a com-
parison of the abilities of chemical water quality
criteria and biological criteria to detect aquatic life
impairment based on ambient monitoring in Ohio.
Of the 645 waterbody segments analyzed, biologi-
cal impairment was evident in 49.8 percent of the
cases where no violations of chemical water quality
criteria were observed (Ohio Environ. Prot.
Agency, 1990a).
110
-------
While this "discrepancy" may be remarkable on
the surface, the reasons are many and complex. Bio-
logical communities respond to and integrate a
wide variety of chemical, physical, and biological
variables in the environment, whether of natural or
anthropogenic origin. These include several factors
that chemical water quality criteria alone cannot ad-
equately discriminate or detect; two examples are
the habitat and siltation. Often it is the cumulative
combination of chemical and physical factors that
result in impaired biological community structure
and function.
The Ohio Environmental Protection Agency re-
cently adopted biological criteria in its water quality
standards regulations. These biocriteria are based
on a system of tiered aquatic life uses from which
numerical criteria were derived using a regional ref-
erence site approach (Ohio Environ. Prot. Agency,
1987, 1990b). The numerical expressions of biologi-
cal goal-attainment criteria are essentially the end
product of an ecologically complex derivation and
assessment system. While numerical biological indi-
ces have been criticized for oversimplifying com-
plex ecological processes, the need to distill such
information to commonly comprehended expres-
sions is both practical and necessary. The advent of
"new" generation evaluation mechanisms such as
the Index of Biotic Integrity (IBI) (Karr, 1981; Fausch
et al. 1984; Karr et al. 1986), the Index of Well-Being
(Iwb) (Gammon, 1976; Gammon et al. 1981), the In-
vertebrate Community Index (ICI) (Ohio Environ.
Prot. Agency, 1987) have filled important theoretical
gaps left by previous indices.
Such multimetric evaluations extract ecologi-
cally relevant information from biological commu-
nity data while preserving the opportunity to
analyze such data on a multivariate basis. The prob-
lem of biological data variability is also addressed
within this system. Variability is controlled by spec-
ifying standardized methods and procedures, com-
pressed through the application of multimetric
evaluation mechanisms, and stratified by account-
ing for regional and physical variability and poten-
tial. This has yielded evaluation mechanisms such
as the IBI and ICI that have acceptably low, replicate
variability (Rankin and Voder, 1990).
Ecoregionai Biocriteria and
Determination of Use
Attainment
Biological criteria in Ohio are based on two princi-
pal organism groups, fish and macroinvertebrates.
Numerical biological criteria for rivers and streams
Biological Criteria: Research and Regulation. 1991
were derived by utilizing the results of sampling
conducted at more than 300 "least impacted" refer-
ence sites. This information was used within the
existing framework of tiered aquatic life uses to es-
tablish attainable, baseline expectations on a re-
gional basis. Resultant criteria for two of the
"fishable, swimmable" uses, Warmwater Habitat
(WWH) and Exceptional Warmwater Habitat
(EWH), are shown in Figure 1.
Procedures for determining the use attainment
status of Ohio's lotic surface waters were also devel-
oped. Using the numerical biocriteria as defined by
the Ohio water quality standards, use attainment
; status is determined as follows:
• FULL — use attainment is considered full if all
of the applicable numeric indices exhibit at-
tainment of the respective biological criteria.
• PARTIAL — at least one organism group ex-
hibits nonattainment of the numeric biocrite-
ria, but no lower than a "Fair" narrative
rating, and the other group exhibits attain-
ment.
• NON — none of the applicable indices exhibit
attainment of the regional biocriteria; or, one
organism group reflects a "Poor" or "Very
Poor" narrative rating, even if the other group
exhibits attainment.
A use-attainment table based on these rules is
constructed on a longitudinal mainstem or water-
shed basis. Data included in the table are sampling
location (river mile index), biological index scores,
the Qualitative Habitat Evaluation Index (QHEI)
score, attainment status, and comments about im-
portant site-specific factors such as proximity to
pollution sources. The following examples demon-
strate the use of the biological criteria as an assess-
ment tool and the overall biosurvey design as an
integrated diagnostic approach.
Blacklick Creek
Table 1 shows a completed attainment table for
Blacklick Creek located in the East Corn Belt Plains
ecoregion of central Ohio. The lower section of this
stream is impacted by a privately owned and
poorly operated wastewater treatment plant. The
results are typical of a small stream (50 sq. mi.
drainage area) impacted by municipal sewage—
full attainment upstream, nonattainment down-
stream, with eventual recovery to partial and full
attainment. Field observations included sewage
111
-------
C.O. YODER
Fish — Boat Sites
Fish —Wading Sites
Fish — Headwater Sites
EWH
EWH
Huron Erie Lake Plain - HELP
1 Interior Plateau - IP
EWH
^ Eastern-Ontario Lake Plain-'EOLP
3 Western Allegheny Plateau - WAP
Eastern Corn Belt Plains - ECBP
Flfluro 1.—Ohio biological criteria for the Warmwater Habitat (WWH) and Exceptional Warmwater Habitat (EWH) use designa-
tions arranged by biological Index, site type for fish, and ecoreglon. Index values on each map are the WWH blocrlterla that
vary by ecoreglon as follows: IBI/Mlwb for Boat Sites (upper left), IBI/Mlwb for Wading Sites (upper right), IBI for Headwater
Sites (lower left), and the ICI (lower right). The EWH criteria for each Index and site type appear In the boxes located outside of
•achmap.
sludge deposits, elevated ammonia-nitrogen, and
continuous dissolved oxygen concentrations that
were depressed below applicable water quality cri-
teria. Extensive experience with this type of im-
pact, the good correlation of the biological
impairment with the dissolved oxygen profile, and
the proximity of the source to the observed impair-
ment made diagnosis relatively easy. Localized
habitat alterations were not a predominant factor
in the results.
Wills Creek
Table 2 shows results from Wills Creekdn the east-
ern coal-bearing region of Ohio (W. Allegheny Pla-
teau ecoregion). The upper watershed is impacted
by runoff from surface mining, resulting in the ex-
tensive siltation of the substrates. The mains'tem is
additionally impacted by two municipal wastewa--
ter treatment plants1 and several small industries.
The extensive siltation from the ndnacidic mine
runoff combined with the relatively low gradient
results in an overlying physical impact that masks
most of the influence of the point sources. While
some localized upstream/downstream patterns are
evident (e.g., ICI downstream Byesville), the over-
all pattern of nonattainment of the Warm Water
Habitat (WWH) use is affected by the predomi-
nance of diffuse squrces. Particularly important in
diagnosing this situation was the Qualitative Habi-
112
-------
Biological Criteria: Research and Regulation, 1991
for the Warmwater Habitat (WWH) use designation in macmck Creek
RIVER MILE
FISH/INVERT.
Blacklick Creek
5.6/4.8
4.7/4.7
3.3/3.6
2.1d/2.1
Big Walnut Creek
15.8d/15.9
14.9
-------
C.O. YODER
Table 2.—Aquatic life use attainment status for the Warmwater Habitat (WWH) use designation in the Wills Creek
RIVER MILE
FISaiNVERT.
75.9/75.8
74.0/71.0
68.1/68.1
66.5/66.7
65.3/65.1
62.4/62.7
61 81
O 1 »O/ ^^
60.7/60.1
58.4/58.6
56.4/56.5
53.5/53.5
46.6/46.6
37.7/ —
27.0/—
IB!
33*
24*
22*
29*
28*
27*
22*
25*
24*
26*
29*
26*
28*
26*
MODIFIED
Iwb
7.7*
5.8*
5.3*
7.0*
6.4*
6.9*
5.7*
7.7*
6.3*
6.6*
7.8*
6.2*
6.5*
5.8*
ICI
30*
34ns
14*
16*
18*
22*
:.
28*
20*
20*
34hs
22*
—
—
QHEI"
52 '•;
34
41 .,
33"
38
48,
54
52 .. '
• 3?:/
42.
55 '-.
42
39
37
WWH
ATTAINMENT
STATUS
Non
Non
Non
Non
Non
Non
(Non)b
Non
Non
Non
Partial
Non
(Non)b
(Non)b
COMMENT
Upstream all point sources
Upstream Byesville WWTP
Downstream Byesville WWTP
Downstream Natl. Cash Register
Downstream sewer line break
Downstream Cambridge WWTP
Downstream Crooked Creek
Downstream Salt Fork
Downstream numerous mines
•All UUamaUVB nauliai evaluation inuex vwncij vaiuoa aio uaocu wiima mw^i .w^/m »^.«.v.
bUse attainment status based on one organism group Is parenthetically expressed.
'Significant departure from ecoregion biocriteria; poor and very poor results are underlined.
""Nonsignificant departure from ecoregion biocriteria (4 IBI or ICI units; 0.5 Iwb units).
1989).
Ecoregion Biocriteria: Western Allegheny Plateau (WAP)
Index— Site Type WWH EWH
IBI-Boat 40 48
Mod. Iwb-Boat 8.6 9.6
ICI 36 46
MWH1
24
5:5
30
'Modified Warmwater Habitat for mine-affected areas.
Sourco: Ohio Environ. Prot. Agency (1990b).
site-specific problems, it does provide the concep-
tual support necessary to operate a "case-specific"
regulatory program, such as the issuance of Na-
tional Point Discharge Elimination System (NPDES)
permits. .
The strength of the biological data is the infor-
mation it provides about whether or not an impair-
ment exists and its severity. Often, this is assumed
to be the limit of the usefulness of biological data.
Prior to the development of many recent concepts
and tools (i.e., regional reference sites, multimetric
indices) this may have been true. However, much
more of the biological information can now be uti-
. lized in multivariate approaches that begin to "sort
out" and communicate important patterns of bio-
logical community response termed "biological re-
sponse signatures."
Biological Response
"Signatures"
The availability of a comprehensive, standardized
ambient biological database from a variety of envi-
ronmental settings has permitted certain patterns
and characteristics of biological community re-
sponse to be identified. A common criticism of am-
bient biological survey data has been its inability to
determine the cause or source of an impaired con^
dition. While this is probably valid for some of the
traditional diversity indices (e.g., Shannon indices),
number of species, biomass, and other single-di-
mension indices, it does not apply .equally to the
"new" generation multimetric indices such as the
IBI and Invertebrate Community Index. When the
response patterns of the various metrics and com-
ponents of these indices were examined from areas
where the predominant impairment causes and
sources are well known, some consistent patterns
emerged. Unique combinations of biological com-
munity characteristics that identify one impact
type over others are referred to as "biological re-
sponse signatures." These proved valuable in as-
signing causes and sources to the aquatic life use
impairments analyzed in the 1990 305(b) report
(Ohio Environ. Prot. Agency, 1990a).
A database including 25 similarly sized streams
and rivers (drainage area range 90-450 square miles)
from the Eastern Corn Belt Plains (ECBP) and
Huron/Erie Lake Plain (HELP) ecoregions was ar-
ranged. Sampling generally took place between
1982 and 1989 and followed Ohio Environmental
Protection Agency procedures
-------
Biological Criteria: Research and Regulation, 1991
Maumee R. and tribsl
Raccoon
Little Cuyahoga River
Tuscarawas RJNimishillen Cr
1985-1989
I BlanchardR. \ \ Maumee River \
, \ Cuyahoga River \
, | Little Cuyahoga River\
•^ I Mahoning River \
Tuscarawas RJNimishillen Cr
Muskingum River \
-f Arsenic
^. Cadmium
O Chromium
O Copper
© Lead
Zinc
Fijjure 2.—Distribution of'biological sampling sites with at least one biological index value reflecting "Poor" or "Very Poor"
performance (upper) and sediment chemistry sites with highly elevated or extremely elevated heavy metal concentration*.
115
-------
C. O. YODER
fraction of the summer base flow of the re-
ceiving stream and where one or more of the
following have occurred: serious instream
chemical water quality impairments involv-
ing toxics; recurrent whole effluent toxicity;
fish kills; and severe sediment contamina-
tion involving toxics. This may include
areas that have combined sewer overflows
and/or urban areas located upstream from
the point sources.
2. Conventional Municipal/Industrial: This
includes impacts from municipal wastewa-
ter treatment plants that discharge conven-
tional substances and where no serious or
recurrent whole effluent toxicity is evident
(these may or may not dominate stream
flows). It may also include impacts from
small industrial discharges that may be
toxic, but that do not comprise a significant
fraction of the summer base flows; other in-
fluences such as combined sewer overflows
and urban runoff may be present upstream
from the point sources.
3. Combined Sewer Overflows/Urban: In-
cluded are impacts from combined sewer
overflows and urban runoff within cities
and metropolitan areas that are in direct
proximity to sampling sites. This includes
both free-flowing and impounded areas up-
stream from the major wastewater treat-
ment plant discharges. Minor point sources
may also be present in some areas.
4. Channelization: Areas impacted by exten-
sive, large-scale channel modification pro-
jects and where little or no habitat recovery
has taken place comprise this impact type.
Some minor point source influences may be
present.
5. Agricultural Nonpoinfc This includes areas
that are principally impacted by runoff from
the row crop agriculture that is the predomi-
nant land use in the ECBP and HELP
ecoregions. Some minor point source and lo-
calized habitat influences may be present.
6. Other: Includes impacts hot mentioned
above — i.e., quarries, sand and gravel exca-
vation, sanitary landfills, and flow alter-
ations (immediate tailwater areas below
dams).
One of these impact types was assigned to each
of 225 sites sampled for fish and 111 sites sampled
for macroinvertebrates. Assignments were based on
the predominant impact that was directly influenc-
ing the site at the time the sampling took place. The
assignments were based on the site-specific knowl-
edge of the study area gained by the Ohio Environ-
mental Protection Agency while conducting
biological surveys of the 25 streams and rivers. The
extent of spatial overlap between different impact
types throughout this database is somewhat vari-
able. However, the key objective of this analysis was
to determine whether or not the feedback from the
biological community can communicate .about and
characterize these differences. t
Some preliminary results of this ongoing proj-
ect are provided in Figures 3 and 4 and Table 3. The
work thus far has concentrated on two- and three-
dimensional analyses of IBI, Mlwb, and ICI metrics
and sub-components. Analysis of "smaller" com-
munity components (e.g., species level) is also being
attempted. An example of one analysis is portrayed
in Figure 3. Three components of the fish commu-
nity data are included in a three-dimensional plot
that illustrates the concept behind examining the
combinant biological response characteristic? of
each impact type. The IBI, Mlwb, and the frequency
(percentage) of deformities, eroded fins, lesions,
and tumors (DELT) on individual fish are different
expressions of the relative health of the fish commu-
nity at a given location. The Complex Toxic impact
type (1) was compared on a three-dimensional basis
to each of the five other impact types (2 through 6).
In each comparison the Complex Toxic impact
type exhibited a fairly distinct pattern as compared
to the other impact types. The amount of overlap
was least with the Agricultural Nonpoint Source
type (5) and greatest with the CSO/Urban impact
type (3). The response characteristics of the Com-
plex Toxic impact type generally include an IBI of
less than 20-25, Mlwb of less than 5.0-6.0, and DELT
anomalies greater than 5-10 percent.
While other impact types may have one or two
of these characteristics in common, very seldom do
they have all three. One sample in the Agricultural
Nonpoint Source type possessed all three of these
characteristics and clustered with the Complex
Toxic impact type. Upon further investigation it was
learned that this particular site was downstream
from an experimental "no-till" agricultural demon-
stration plot where pesticide usage was atypical.
The resultant biological response confirmed that
this particular impact fit the Complex Toxic impact
type both biologically and culturally.
116
-------
Biological Criteria: Research and Regulation, 1991
b
LU
Q
0\
ao
90
40
30
20
10
0
*
f-
^
: <
/
&
10
ao
b
IU
Q
^
ao
98
40
30
ao
10
0
£
|-
^
1 <
/
^
10
10
O coivip>ukx.-roxic
-------
C.O. VODER
(X 1000)
4
LU
O
-
£M*
M -a
0) J
LU
a
SE 2
CO
»— 1
a:
o
0
i i i i i i i i i i i i i • » i » » • > i * * * * i * * * * i_
_
mm —
X
0 X sx
; O A « Xx X ^
• fVSt ^/s^1^ ^
1 V -
1 . . f . 1 . 1 B . t ' • • • ' 1 — t — 1 — 1 — 1 — t — t — I_J — I — t — 1 — 1 1 1
r— I — 1 — 1 — i 1 I I I t I I i 1 r—i | I— I I 1 | I > 1 « I ' < ' ' [
• . :
- ~
;
l n -
n Q-I
n ^-tP
- o cP r-a ~
; ^ A $ ^ i-p •
' o5^ <"> ^ H_H rfl
"•..., ' .^TT . 1 .... 1 i t i • 1 i t i i 1
10 20 30 40 50 60
I C I
10 20 30 40 50
I C I
60
Figure 4.—TNwo-dlmenslonal comparison of She Invertebrate Community Index (1CI) and organism density between the Com-
plex Toxic (O) and Conventional Municipal (+), CSO/Urban (*), Channelization (X), and Agricultural NPS (D) Impact types at
106 sampling sites.
Some of the CSO/Urban impacted sites over-
lapped into the Complex type cluster in terms of the
IBI and Mlwb, but much less so in terms of percent-
age of fish exhibiting DELT anomalies. Some of
these sites are located in areas with significant in-
dustrial sources that discharge into the municipal
sewer system and that may have "mimicked" the
Complex Toxic impact type. This example high-
lights the need to rely more on the biological re-
sponse signature to characterize an environmental
impact rather than the traditional process of cul-
tural impact characterization based on chemical,
physical, and process characteristics alone.
Table 3 shows the distribution of fish commu-
nity index scores by impact type for the five narra-
tive performance classes (Ohio Environ. Prot.
Agency, 1987,1990c). Sampling sites predominantly
affected by the Complex Toxic impact type were
most frequently in the "Very Poor" (55 percent) per-
formance category, followed by the CSO/Urban im-
pact type (23 percent). The highest narrative
category reached by Complex Toxic impacted sites
was "Fair" (18 percent). It should be noted that the
Fair sites in this impact category were the farthest
downstream from the major sources and represent
the initial recovery along the longitudinal profile.
The Conventional impact type had no sites in the
"Very Poor" range and along with the Agricultural
NPS and Other impact types, was the only type to
have sites that attained the "Exceptional" perfor-
mance level (equivalent to the EWH use). These
three impact types also had the highest number of
118
-------
Biological Criteria: Research and Regulation, 1991
IaabnlL3;7^1Str-ibUtJ0n*uf y°\09lc?1 sa"iP"n9 results from 225 sites between the five narrative biological
mance categories for the Index of Biotic Integrity and Modified Index of Well-Beina for siy mainr imn»*t
IMPACT TYPE
1
Complex
Toxic
2
Conventional
Muni./lnd.
3
CSO/Urban
4
Channelization
Agricultural
NFS
6
Other
VERY
POOR
55%
(n = 12)
0
23%
(n = 9)
7%
(n = 1)
2%
(n = 1)
6%
(n = 1)
POOR
27%
(n = 6)
13%
(n = 10)
10%
(n=4)
33%
(n = 5)
2%
(n = 1)
18%
(n = 3)
FAIR
18%
(n = 4)
44%
(n=35)
67%
(n=27)
60%
(n = 9)
29%
(n = 15)
23%
(n-4)
GOOD ^XCE
perfor-
types.
0
34% 9%
(n = 27). (n = 7)
0
0
52% 15%
(n = 27) (n=8)
47% e
(n=8) in
%
sites in the "Good" performance level (equivalent to
the WWH use). The CSO/Urban and Channeliza-
tion impact types, along with the Complex Toxic
type, had no sites attaining the "Good" or "Excep-
tional" performance levels.
Using the macroinvertebrate community, a
comparison of the relationship between the Inverte-
brate Community Index and organism density
(number/square feet) demonstrates the need to ac-
cess community information beyond the index re-
sult or the metrics that comprise the index. A
comparison of the Complex Toxic impact type with
the Conventional Municipal, CSO/Urban, Channel-
ization, and Agricultural NFS impact types was
made in a two-dimensional framework. In the com-
parison of the Complex Toxic and CSO/Urban im-
pact types the ICI alone yields equally low results
for each type of impact. Thus, this index alone was
not able to discriminate the impacts. However, or-
ganism density, which is not a direct component of
the ICI, yielded an improved separation of the two
impact types (Fig. 4). This was also true in compari-
sons with the Conventional Municipal and Chan-
nelization impact types. The ICI alone separated
most of the Agricultural NFS impacts. Thus, it is im-
portant to experiment with other aggregations of
the community data that are not direct metrics of
the indices used as the biological criteria for each or-
ganism group.
Although the Complex Toxic impact type sepa-
rates well from the other impact types in this analy-
sis, more overlap exists between the other five
types. For example, the statistics in Table 3 are sim-
ilar for the Conventional and Agricultural NFS im-
pact types. Scatter plots also show a great deal of
overlap. This is not surprising since the chemical
and physical manifestations of each are functionally
similar in the aquatic environment. Nevertheless,
some differences may exist and are likely discern-
ible by using more complex and iterative analyses
than those demonstrated in Figures 3 and 4, and
TableS.
The Ohio Environmental Protection Agency is
currently cooperating with Bolt, Beranek, & New-
man, Inc., to evaluate techniques by which some of
these more subtle differences might be defined
using biological response signatures (Anderson et
al. 1990). This involves the use of genetic algorithms
employing artificial intelligence and machine learn-
ing techniques. One initial finding was the utility of
one of the IBI metrics used for the headwaters site-
type sensitive species. This metric combines the in-
tolerant metric of the wading and boat site types
with moderately intolerant species (Ohio Environ.
Frot. Agency, 1987). This aggregation of the commu-
nity data was by itself found to consistently indicate
the Complex Toxic impact type with a reliability of
82 percent in stream and river sizes outside of its de-
signed use in the Headwaters IBI.
Another way to describe the attributes of ambi-
ent biological data for characterizing different types
of environmental impacts is with a conceptual
model. Figure 5 shows a model of the response of a
fish community to increasing stress from a ''least
impacted" to "severely degraded" condition. The
comparison of numbers and/or biomass with the
IBI shows this conceptual relationship. Beneath the
graphic are narrative descriptions of biological com-
119
-------
C.O. YODER
1
NUMBERS/
BIOMASS
X
MODERATE
IMPACT
DEGRADED
/
SEVERELY
DEGRADED
ENRICHMENT
A
LEAST 4 — — UNIMPACTED
IMPACTED
12 20 30 40 50
INDEX OF BIOTIC INTEGRITY
60
%INSECTIVORES
%INTOLERANTS'
NO. SPECIES —
%OMNIVORES
%TOLERANTS
%ANOMALIES
(ARROWS INDICATE INCREASES IN
EACH METRIC RELATIVE TO THE IBI)
Conceptual Model^ of Community Response - Narrative Descriptions
Attributes
1) Community
Condition
Character-
istics
2) Chemical
Conditions
3) Physical
Conditions
4) Examples
of Pertur-
bations
Severely
Degraded
(Very Poor)
No community
organization -
few or no species
very low numbers
only most toler-
ant, high %
anomalies
Acutely toxic
chemical con-
ditions
andlor
Total habitat
loss, extremely
contaminated
sediments
Toxic discharges.
dessication, acid
mine drainage,
severe thermal
conditions
Degraded
(Poor)
Poorly organized
community -
few species, low
numbers, tolerant
species only,
many anomalies
, ' ; w „
Low D.O. with
chronic toxicity
andlor
Severe habitat
degradation,
severe sediment
contamination'
Municipal and
industrial dis-
charges, inter-
mittent acute
impacts
Moderate
Impact
(Fair)
Reorganized
community -
tolerant species,
intolerants in
v. low numbers,
bmnivores pre-
dominate
Low D.O.,
nutrient enriched,
no recurrent ,
toxicity
Modified stream
channel, heavy
siltation, canopy
removal
Municipal sewage,
combined sewers,
heavy agricul-
tural use, non-
acid mine drain-
age, moderate
thermal increases
Enrichment
(Good)
Good community
organization -
good numbers of
sensitive species,
some intolerants
..' ' '
Adequate D.O.,
no acute/chronic
effects, elevated
nutrients
Good habitat,
no significant
channel modifi-
cations
Minor sewage
inputs, most
agricultural
non-point affected
areas
Least
Impacted'
(Exceptional)
Highly organized
community -
insectivores,
top carnivores,
intolerants pre-
dominate, high
diversity
No effects
evident, back-
ground condi-
tions, good D.O.
Excellent habitat.
no modifications
evident
No perturbations
evident
Figure 5.—Conceptual model of the response of the fish community as portrayed by the Index of Blotlc integrity and other
community metrics with narrative descriptions of Impact types and corresponding narrative biological performance expecta-
tions.
munity characteristics, chemical conditions, physi-
cal conditions, and examples of environmental per-
turbations that are typical of biological community
response across the five narrative performance
classes. These are necessarily general and are not in-
variable. However, this model was developed from
the Ohio Environmental Protection Agency's expe-
rience in analyzing biological, chemical, and physi-
cal data over a 12-year period and on a statewide
basis. Thus, the model has a good foundation in the
observation of actual environmental conditions and
associated biological community responses.
120
-------
Biological Criteria: Research and Regulation, 1991
Actual results of the IBI from four similarly
sized streams and rivers were plotted together (Fig.
6) in an attempt to demonstrate the real application
of these model concepts. These results demonstrate
the utility of using the vertical scale of the IBI (or
ICI, Mlwb, etc.) to differentiate between different
types of impacts. The left column lists the "gradi-
ent" of impact types associated with the vertical
scale of the IBI with the actual impacts present in
each of the four streams being listed in the right col-
umn. Riverine biological communities may experi-
ence spatially different impacts on a longitudinal
upstream/downstream basis with the degree of de-
parture and recovery dependent on the severity and
type(s) of impacts being exerted on the biota. Wal-
nut Creek and the Hocking River are examples (Fig.
6). Other streams are relatively unimpacted or have
only moderate departures (Big Darby Creek) while
others may be uniformly devastated (Rush Creek).
These examples correspond to the narrative de-
scriptions of community response and the attributes
of the various impact types in Figure 5.
The point here is that the biota itself integrates
differing types and degrees of environmental im-
pacts on a spatial and temporal basis, providing
feedback that is more accurate than can be achieved
using cultural, surrogate, or process characteriza-
tions alone. Also, insight can be gained on what to
expect as the predominant impacts in a particular
segment change over time as a result of decreasing
or increasing pollution levels. For example, a pre-
dominantly toxic impact should be expected to
change to a conventional impact when the sources
of toxicity are controlled. This may be evident on
both a temporal and spatial scale. Often times im-
pacts are "layered" in rivers, with the less severe im-
pact types being masked by those that presently
result in more severe degradation. As the more se-
vere problems are reduced or eliminated the
"lesser" problems may become evident in the re-
sults. An example of this is being observed in Ohio
where the abatement of municipal and industrial
point source problems is revealing nonpoint source
impacts.
Summary
Definite patterns in biological community data
exist and can be used in determining whether or
not a water body is attaining its designated use
and, if not, identifying the predominant causes of
impairment. The Ohio Environmental Protection
Agency has used this approach in producing the bi-
ennial Clean Water Act 305b report required by
EPA, specifically the assignment of causes and
sources of aquatic life use impairment (Ohio Envi-
I
B
I
60
50
40
30
20 -
10
IMPACT TYPE I
— "O Q A PlICNITll 1
LariAUIbNr *
Least impacted,
"Reference"
Conditions
Minor sewage and
NFS impacts
Moderate enrich-
ment,siltation,low
DO, habitat impacts
f\flf-*f u
chronic toxicity
Complex toxic
(acute), acid mine,
toxic sediments
i - • i
BIOLOGICAL RESPONSE
EXCEPTIONAL
/"\ I ' . ^^
V ~7r\
i GOOD / y *
GL f "•"
r^&
__ r ' , \^f
(7> (/} V7
FAIR i^y y-" \.
' fS' ' •
• H / /
H^. fo ;
Q^r / FLOW
POOR ©-, „ . / t
VERY POOR H' ^L_rn *?
1 : 1 1
STREAM/
It an X OTO
IMrAoTo
BIG DARBY CR.
' (Municipal, Agr.
NPS)
WALNUT CR.
s (Industrial/
Conventional,
Municipal) ,
HOCKING R.
» (Municipal w/Pre-
treatment,CSO)
. • ... . ... , .,
RUSHCR. ; ^
Drainage)
J • . i . :
RIVER MILE
Figure 6.—Biological community response as portrayed by the Index of Blotlc Integrity (IBI) In four similarly sized Ohio rivers
with different types of point and nonpoint source Impacts. ;..-.. ''.
121
-------
C.O. YODER
ton. Prot. Agency, 1990a). Other uses include sup-
porting enforcement and litigation proceedings.
For example, this type of assessment has been used
to refute positions taken by the National Point Dis-
charge Elimination System permit holders that the
degradation measured was due to poor habitat or
factors unrelated to their discharge. The biological
response signatures can be particularly useful in
demonstrating that the degradation is related to
specific discharges, especially those involving the
Complex Toxic impact type. While the legal re-
quirements of the Clean Water Act may be viewed
as sufficient to require entities to reduce pollutant.
loadings, the system of challenging these mandates
requires the regulatory agency to defend the rea-
sonableness of its regulatory actions. This type of
ambient response data is particularly valuable in
meeting that need.
Minimum requirements for using this type of
data include having sufficient information to em-
ploy the use of multimetric evaluation mechanisms,
a standardized approach to data collection, and
consistent and responsive management of the
database. Other factors that further increase the an-
alytical power of the biological data include the use
of multiple organism groups, an integrated ap-
proach to conducting ambient surface water assess-
ments, and the inclusion of ancillary data such as
biomass for fish.
It is important to make these and other data col-
lection decisions early in the process. An example of
the importance of these early decisions was the
Ohio Environmental Protection Agency's decision
to record external anomalies on fish, which 10 years
later allowed the development of the %DELT metric
of the IBI. This metric proved key in identifying the
Complex Toxic impact type. At the time it was not
known that this use would be possible; however,
failure to include it as a quantitative measurement
early in the process would have resulted in an un-
fortunate and irreplaceable loss of data. Thus, the
ability to utilize biological data for diagnosis in
Ohio was partly the result of decisions made more
than 10 years ago, not only regarding which organ-
ism groups to sample, but about the types of infor-
mation that should be recorded from each sampling
effort. Frequently, biological monitoring programs
are pressured to sacrifice data quantity and quality
to meet regulatory and financial constraints. As
seen here, such decisions can have far-reaching con-
sequences over the long term.
References
Anderson, K., A. Boulanger, H. Gish, J. Kelly, and J. Morrill.
1990. Using machine learning techniques to visualize
and refine criteria for biological integrity. Unpubl. mss.
Bolt, Beranek, & Newman, Cambridge, MA.
Fausch, D.O., J.R. Karr, and P.R. Yant. 1984. Regional applica-
tion of an index of biotic integrity based on stream fish
communities. Trans. Am. Fish. Soc. 113:39-55.
Gammon, J.R. 1976. The fish populations of the middle 340
km of the Wabash River. Tech. Rep. 86. Water Resourc.
Res. Center, Purdue Univ. Lafayette, IN.
Gammon, J.R., A. Spacie, J.L. Hamelink, and R.L. Kaesler.
1981. Role of electrofishing in assessing environmental
quality of the Wabash River. Pages 307-24 in J.M. Bates
and C.I. Weber, eds. Ecological Assessments of Effluent
Impacts on Communities of Indigenous Aquatic Organ-
isms. Am. Soc. Test. Mater. STP 703. Philadelphia, PA.
Karr, J.R. 1981. Assessment of biotic integrity using fish com-
munities. Fisheries 6(6): 21-27.
Karr, J.R., K.D. Fausch, P.L. Angermier, P.R. Yant, and I.J.
Schlosser. 1986. Assessing biological integrity in run-
ning waters: a method and its rationale. Spec. Publ. 5.
111. Nat. Hist., Urbana.
Ohio Environmental Protection Agency. 1987. Biological Cri-
teria for the Protection of Aquatic Life: Volume II. Users
Manual for Biological Field Assessment of Ohio Surface
Waters. Div. Water Qual. Plann. Assess., Surface Water
Sect, Columbus.
. 1989a. Addendum to Biological Criteria for the Pro-
tection of Aquatic Life: Users Manual for Biological
Field Assessment of Ohio Surface Waters. Div. Water
Qual. Plann. Assess., Surface Water Sect., Columbus.
' . 1989b. Biological Criteria for the Protection of
Aquatic Life: Volume III. Standardized Biological Field
Sampling and Laboratory Methods for Assessing Fish
and Macroinvertebrate Communities. Div. Water Qual.
Plann. Assess., Ecol. Assess. Sec., Columbus.
. 1990a. Ohio Water Resource Inventory. Rankin, E.T.,
C.O. Yoder, D. Mishne, eds. Executive Summary and
Vol. I. Div. Water Qual. Plann. Assess., Ecol. AssessmenV
Section, Columbus.
—. 1990b. Uses of Biocriteria in the Ohio Environmental
Protection Agency Surface Water Monitoring and As-
sessment Program. Div. Water Qual. Plann. Assess.,
Ecol. Assessment Section, Columbus.
. 1990c. Compendium of Biological Results from Ohio
Rivers, Streams, and Lakes. 1989 ed. Div. Water Qual.
Plann. Assess., Ecol. Assessment Section, Columbus.
Rankin, E.T. 1989. The Qualitative Habitat Evaluation Index
(QHEI): Rationale, Methods, and Application. Div.
Water Qual. Plann. Assess., Columbus.
Rankin, E.T. and C.O. Yoder. 1990. The nature of sampling
variability in the index of biotic integrity (IBI) in Ohio
streams. Pages 9-18 in Proc. 3rd Midw. Pollut. Biologists
Conf., U.S. Environ. Prot. Agency, Region V, Chicago,
IL.
122
-------
POSTERS
-------
-------
Biological Criteria: Research and Regufati'on, f 99t
Using Machine Learning Techniques
to Visualize and Refine Criteria for
Biological Integrity
Kenneth Anderson
Albert Boulanger
Herbert Gish
James Kelly
Jeffrey Morrill
Bolt Bemnek & Newman Inc.
Systems & Technologies Division
Cambridge, Massachusetts
Lawrence Davis
TICA Associates
Cambridge, Massachusetts
ABSTRACT
Several techniques from the fields of artificial intelligence/machine learning and statistics
were used to examine and analyze biological criteria (biocriteria) data. The goals of this re-
search were: (1) to analyze biocriteria data, discovering patterns and relationships to aid water
quality scientists in assessing environmental integrity of surface water sites; and (2) to de-
velop, compare, and contrast computer techniques for aiding in the visualization and analysis
of large databases. Specific classification techniques used were CART, IDS, and a Genetic Algo-
rithm. An interactive data exploration tool was built based on IDS. The research produced sev-
eral results: (1) Complex industrial sites can be identified with an 82 percent accuracy; (2) other
sites are strongly corrupted by secondary impact, yielding an average accuracy of between 60
and 65 percent; (3) biological criteria have comparable accuracy to chemical, while combining
both biological and chemical indicators produces slightly better results; (4) raw fish scores
classify as well as composite indices; (5) IDS is a valuable tool for identifying important attri-
butes in a high dimensional attribute space; and (6) an attribute, "all intolerant species," pres-
ently used only for headwaters by the Ohio EPA, was identified for potential broad use.
Data
The biocriteria data used in this research are equiv-
alent to 461 records and consist of measurements
taken at water sites in Ohio. Each record holds the
values of 495 attributes, including site information,
biological and chemical measurements, and com-
pound variables. Records include the primary and
secondary pollution impact on that site (e.g., agri-
cultural runoff) as determined by Ohio EPA water
quality scientists.
The data were merged from several database
sources into one file consisting of physical, biologi-
cal (individual fish and insect species data), and
chemical attributes, totaling 495 in all. These data
were drawn primarily from one ecoregion of Ohio
containing medium and large streams, and include
reference sites. Each data record was classified by its
123
-------
K. ANDERSON, A. BOULANGER, H. GISH, J. KELLY,
J, MORFIILL, and L DAVIS
primary and secondary impact class as determined
by Ohio EPA water quality scientists into eight cate-
gories. These are (1) Complex Municipal/Indus-
trial; (2) Conventional Municipal/Industrial; (3)
Combined Sewer Overflows; (4) Channelization; (5)
Agricultural Nonpoint; (6) Impoundments; (7)
Combined Sewer Overflows with Toxics; and (8)
Other.
The data also include combined or derived at-
tributes according to a tiered methodology. Fish and
insect species are grouped together into subclasses
by taxonomic and other means such as habitat.
These subclasses are then used to build a scoring
and normalizing system similar to the indicators
used to measure the health of the economy. The
Ohio EPA uses three: IBI (Index of Biological Integ-
rity), IWB (Index of Well Being), and ICI (Inverte-
brate Community Index). IWB is based on
structural attributes of the fish community, whereas
the IBI also incorporates functional characteristics.
ICI is based on the insect community. All of the de-
rived attributes used to build these indices were
also part of the data.
The data are very typical of real data sets. Char-
acteristics include or imply:
• Noise, in terms of both measurement and
impact type identification error;
• Unknown, missing, or by-case not-relevant
data fields;
• A rich attribute structure (species taxonomy,
tiered attributes);
• Distribution in several interlinked database
files; and
• Multiple questions requiring more research.
Several questions were investigated with these
biocriteria data:
• Can methods be developed for automating
the determination of impact type(s) of a new
site based on chemical, biological, and
combined attributes?
• What attributes are more and less relevant
when determining the impact type of a new
site?
• Does the use of biological (or chemical)
measurements alone provide superior
classification results to that obtained using
just chemical (or biological) data?
• How can chemical, biological, and combined
attributes help discriminate the primary
impact types and both the primary and.
secondary impact types taken in
combination?
Can single-species counts be used as a
discriminator of impact type?
Can the Index of Biological Integrity be used
to generate a measure of impact severity? If
so, can this measure of impact severity
discriminate as part of the class label?
1 Can the utility of some of the combined
attributes for discriminating impact type be
improved?
1 Does an initial partitioning of the data, such
as the separation of medium from large
streams, help the discrimination task?
Technical Approach
Classification
The initial goal of this research was to correlate the
human determination of impact type with the mea-
sured attributes so that future determinations
would be automated. The data were analyzed with
four classification algorithms.
For each technique the general methodology
was to withhold a certain percentage of the data
and train each specific algorithm with the remain-
der. By then classifying the withheld cases using
each algorithm arid matching the automated deter-
mination against the prior human determination,
the effectiveness of each approach can be measured.
In addition, running the systems on the biological
and chemical attributes separately made it possible
to compare the utility of each approach.
Tree Classifers: CART and ID3
ID3 is an algorithm for inductively synthesizing a
binary decision tree for classification given a set of
labeled training examples in the form of feature
vectors (Quinlan, 1983; Pao, 1989). As in the game
of "20 questions," the object is to find as few ques-
tions as possible that will correctly classify the
data. Thus, IDS determines the binary question that
provides the most information about the identity of
the data. Each question divides the dataset, S, into
two groups, St and Sf, depending on whether the
answer to the question is true or false for a particu-
lar datum. IDS is then applied recursively to each
group until the data cannot be partitioned further.
124
-------
Biological Criteria: Research and Regulation, T9Sf
At each stage, IDS asks the question that maxi-
mizes the .information (most reduces the uncer-
tainty) about the class membership of the data. The
entropy, or uncertainty, existing before the question
is asked is:
H(S)—
where Pi is the fraction of the elements in S belong-
ing to class Q.
After the binary question Q is applied, the data
are divided into two groups, St and Sf, and the re-
maining entropy is:
H(S,Q) = P(St)H(St) + P(Sf)H(Sf)
where PCSt)is the fraction of the elements of S for
which the question, Q, is true. Similarly for PCSf).
The information gained by asking Q is then
H(S)-H(S,Q)
and the best question to ask is the one that maxi-
mizes I(Q).
When the training data have binary valued fea-
tures, the feature, F, for which H(S,F) is maximum,
is chosen. When the values of the data features are
continuous, as they are here, a feature, F, and a
threshold, T, for which H(S,F>T) is maximized,
must be determined. For a given feature, the best
threshold can be found by a linear search.
IDS is problematic in that it overfits the data.
C4, and CART (Breiman et al. 1984; Crawford, 1989)
try to solve this problem by trimming tree limbs
when they don't improve some heuristic measure of
the goodness of the tree. MDL was used to prune
the initial IDS-generated trees. The utility of CART
with some of the initial data obtained from the Ohio
EPA was also explored.
Genetic Algorithm/K Nearest
Neighbors
Genetic algorithms are a learning paradigm based
loosely on the model of biological evolution and
first described in Holland (1975). Briefly, a popula-
tion set of solutions to the problem is initialized
and a reproduction /evaluation cycle is initiated.
Reproduction includes operations such as cross-
over and random mutation of solutions. Reproduc-
tion rates for each member are determined by the
evaluation function, which provides a measure of
the ability to solve the problem at hand. Genetic al-
gorithms have been applied successfully toward
many tasks, including network layout and semi-
conductor design.
The nearest neighbors clustering algorithm is a
well-known technique for categorization based on
the distances between data points. In this study the
Euclidean distance metric was used. A new classifi-
cation algorithm was developed in which the ge-
netic algorithm learns real valued weights for each
attribute. By assuming higher weights to the attri-
butes more attributes relevant for classification
measures are adjusted so that like data points are
determined to be in close proximity to each other.
Assuming the weights have two significant digits,
an exhaustive search would entail the examination
of 100number of attributes wejght combinations/ neces.
sitating enormous computing capabilities. The ge-
netic algorithm approach allows all areas in the
space of possible solutions to be explored, albeit in-
completely.
Neural Networks
Several machine-learning techniques can be used
in conjunction with one another. For example,
while neural networks are quite powerful, they can
require considerable computer time to develop.
Tree classifiers such as IDS and CART are often
used initially because they produce classifiers
quickly. By extending the IDS approach to use lin-
ear combinations of features, a neural net-like clas-
sifier can be quickly produced. Cluster analysis can
also be used to develop gaussian features that pro-
vide a hidden layer of a neural network, with or
without supervised training data. Neural networks
(backpropagation and cascade correlation training)
were applied initially to the EPA data, but the num-
ber of attributes and training set size led to training
times that were much too long for initial explora-
tion of the data.
Data Visualization
Any machine learning technique can be misapplied
when treated as a black box. Displaying and ex-
ploring the data are important at each point in the
analysis. Even classifiers, such as a decision tree,
are quite useful as a data browsing technique in
their own right. The classifier can be thought of as
an apprentice to the domain expert; together they
iteratively refine their understanding of the situa-
tion under investigation. Thus, the rules produced
by a classifier are a compact description of the data;
125
-------
K. ANDERSON, A. BOULANGER, H. GISH, J. KELLY,
J. MORRILL, and L DAVIS
in the process of understanding these rules, the ex-
pert may be led to refine the classification problem.
The TD3 tree classifier was used for interactive
data exploration (see Fig. 1). A system was built that
graphically presents the results of IDS, using the
hypertext paradigm to probe nodes in the tree for
information not currently displayed (e.g., number
of examples covered by this node, ID numbers of
the examples, attribute value for each example,
etc.).
A graphic display for presenting the Genetic Al-
gorithm/K Nearest Neighbors results was also con-
structed. Figure 2 displays the sets of weights
learned during five training runs of the algorithm to
discriminate agricultural nonpoint impacts (impact
class 5) from all others. This display allows a person
to quickly view results and discover patterns. The
results in Figure 2 suggest that the number of darter
species (row 1) is more important for the identifica-
tion of agricultural nonpoint pollution than the
number of intolerant species (row 4), as this species
received a larger weight during each training run.
Results
Classification
These biological techniques showed that biological
indicators were more sensitive than chemical mea-
sures in determining impact type of the data
withheld for test. However; as a much greater per-
centage of chemical data had missing values, fur-
ther experiments would be required to prove that
.biological analysis is in fact better than chemical.
Second, no technique had an accuracy rate greater
than 75 percent in the classification of withheld
sites. The Complex Municipal/Industrial category
had much better classification performance than
the other categories.
cgtnnc
) Oiminuerj
X: OniotcAS AUJNT
1 OtacccaS iOi
> loom. 'CD uses tr.ese **
er Comrana, Ptrcition Subt
: DATUM 'ALUtJ") S,0»
i*f Commano. *A>ndow Operj
isr Commana-,
on fee Oruoepas..npac.-*ree '"
c.5iS , All Oatasets !2 Subsets] i
OhioepaS CiS1 Datums, 1 .Subset]
AUUNT <,S.S :^6I Datums]
li/iiion Plot .^PARTITION '
HOOttUKS < 36.41 /
0.44942U4 ./
l^
r *" ' *•
02147S4S4- ^^
030600£91 0
SDSOCKPC < 2.7£
« 33 B 1 !5 9 • 9 ^
reeVHHHHHHIHHHHB
1 £3455791 \ 1 Z 3 4 5 « 7 8 < |
CUff 0
Q.94SS531
~1 I34S6
12345
4498368
1 2 3 45
0 34 15 4 1:
CllJf !
0.1 MIS',
12 345
d 18 Q 0 M
onnwoR < t
'12 34-5
3 32 6 e 12
0^32537
HMO <2^
ALLIIM
' S
. ' 8
Decision Class: IMPACT pe'
AH Attributes
Classifiers
impact-tree
;., '•' c:ass2-tree
c'3issi*no-ibi
c.i$,si-rtree . • „• ,'L -._ ;.• ' "
Class -abels C1S More]
Oertved Attributes C2 More] r v
Baw's«cW+ C4 MoreJ '
Raw Soecies C127 More]
Rankin's Selections C1S3 More] MJ
Chemical Features CSS More] ^
T < S.3 *LLINT vs. I8I
'. *
' & O
• <> O
«i""§ »
•'• *<>sTcl|
Q • (^g 3. £&«
A
* •> *
' 4 A A . Q , "' "
'. °' A tiDiti,AA A" '...
*=*$ ^i^i '" A- . '•••''' b :v •
™ ' • ' • • :•'-..;.••; - '••* '.- ;., .:
A . _ . ..,.;- ,,„,;..' 'ff .
.,....,. ,;,;?;-
Allint
Moiisu"! 5 JldOCt thiii winclow; MOUK«-K: Mcnit ot opoi-atu
To aoo other commands, press Shift, Coiitr«pl. Mcta, Meta-Ghift, or Super.
LToi lo iec 5:21:32. Keyooara CL^ySER: . 'Jsar input
Flflur* 1—A »cro«n Image of the Data Exploration Interface, Beginner. Note: Beginner incorporates an ID3 tree classifier at •
way of doing exploratory data analysis.
126
-------
Biological Criteria: Research and Regulation, 1391
Training Mode: Single Class KM; Class: 5
darter species
sunfish species
sucker species
intolerant species
, top carnivores
carnivores
round bodied suckers
deformities
species minus exotics
, sensitive species
sculpin £ darter species
minnow species
insectivores
headwater species
pioneering species
% lith. spawners
' f lith. spawners
Kl
K2
K3
Training Brror Rate
Testing Error Rate
(
0.84
0.71
6.40
0.37
0.85
0.23
0.54
0.16
0.34
1.00
0.31
0.73
0.35
0.03
0.47
0.19
6.56
0.56
0.94
0.95
1234) (
— 0. 66
— 0.78
— 0.18
— — 0.53
0.41
— 0.56
0.58
— 0.22
' 0.95
- 0.99
— . 0.80
— 0.46
0.19
— 0.49
0.27
— 0.03
•• 0.28
— - 0 . 63
_ 0.68
123 5) (
— 0. 71
— — 0.33
• 0.34
— 0.73
— 1.00
— 0.15
— — 0.94
• 0.92
. 0.98
— ~ 0.71
— ,. 0.70
— 0.63
0.94
— 0.28
- 0 . 13
0.05
• 0.76
— 0.64
— 0.36
1245) (
— • 0.38
— 0.28
— 0.46
— — 1.00
— 0.29
0.22
' 0.16
—— 0 .32
— 0.71
— 0.65
— 0.68
— 0.29
— — 0.78
• 0.32
• 0.04
0;77
— 0.84
•— 0.46
— 0.78
1345) (
— 0..81
— 0.85
-. 0.21
— i 0.37
. 0.82
- 0.55
0.43
0.00
— 0.28
— — 0.62
— 0.84
— ° • 17
0.53
— — 0.20
- , 0.57
0.40
— 0.06
— 0.52
— 0.82
— • 0.69
2 3
MBMM
mi^^m
m '.
-.
•MH
_
_
_
__.
^••DB
•
_
•
_,
_
_
maam
mmmm
0.199 0.191 0.214 0.176 0.219
0.304 0.348 0.344 0.269 0.355
Figure 2—Genetic Algorithm/K nearest neighbor results.
In the process of running this analysis, it be-
came obvious that it was necessary to first under-
stand the data more thoroughly, especially where it
presented noisy, missing, or invalid values. It was
also necessary to identify subproblems of predicting
impact using biocriteria and chemical data from the
Ohio EPA dataset. To accomplish this, the interac-
tive visualization capabilities of our machine learn-
ing and visualization environment. Beginner, was
used, enabling interactive exploration of the prob-
lem space and identification of tractable sub-
problems.
Visualization
By using the visualization tool, the domain expert
was able to discover previously unknown relation-
ships among the data. For example, preliminary re-
sults indicate that the Ohio EPA measure of
intolerant species, INTOLS, would be more dis-
criminating if it were broadened to include moder-
ately tolerant species — as was done for another at-
tribute, ALLINT. ALLINT, which was used in head-
water cases by the Ohio EPA, was more useful than
INTOLS as a discriminator of impact type. Com-
plex Municipal/Industrial sites can be easily iden-
tified. Combined Sewer Overflow sites are difficult
to separate from Agricultural Nonpoirit sites using
biological criteria alone, but relative degree of im-
pact can be identified.
Summary
In summary, this project generated several results
important in exploring complex databases:
• Differences in setup, run time, and results of
. machine learning algorithms were compared
and contrasted.
• Many issues associated with large databases
were explored. For example, the handling of
127
-------
K. ANDERSON, A. BOULANGER, H.GISH, J.KELLY,
J. MORRILL, and L DAVIS
missing values and data records that can be
classified into multiple categories was
investigated.
• A data exploration/visualization tool that
successfully developed, confirmed, and
disproved hypotheses in a time- effective
manner was constructed.
This project also generated important results in
the use of criteria for biological integrity:
• an attribute currently being used in
evaluating headwaters, ALLINT, is a useful
discriminating attribute for all rivers;
• individual fish species that are ubiquitous in
Ohio rivers can serve as indicators of
biological integrity; and
• the presence of some fish species helps in
determining a river's change in biological
integrity. For example, low values of the
tiered attribute, HEADWTR (headwater
species), was used in discriminating large
moderately impacted municipal/industrial-
class streams, possibly indicating recovery.
ACKNOWLEDGMENTS: Charles T. Walbridge of the
U.S. EPA directed the authors to the biocriteria domain,
and suggested the use of genetic algorithms. Chris Yoder
and Ed Rankin of the Ohio EPA not only provided the
data but also many insightful observations during the
course of this work. Discussions involving members of
the Machine-Learning and Optimization Group at Bolt
Beranek & Newman Inc. (BBN) provided feedback
throughout this project. Finally, the authors would like to
thank BBN Systems and Technologies Division for sup-
porting this research.
References
Breiman, L., J. H. Friedman, R. A. Olsen, and C. J. Stone. 1984.
Classification and Regression Trees. Wadsworth, Bel-
mont, CA.
Crawford, S. L. 1989. Extensions to the CART algorithm. Int.
J. Man-Mach. Stud. 31(2):197-217.
Holland, J. 1975. Adaptation in Natural and Artificial Sys-
tems. Univ. Michigan Press, Ann Arbor.
Kelly, J. and D. Davis. A Hybrid Genetic Algorithm for Clas-
sification. Subm. 12th Int. Joint Conf. Artif. IntelL
Pao, Y-H. 1989. Adaptive Pattern Recognition and Neural
Networks. Addison-Wesley, New York.
Quinlan J. R. 1983. Learning efficient classification proce-
dures and their application to chess end games/Pages
463-82 in R.S. Michalski, J.G. Carbbriell, and T.M. Mitch-
ell, eds. Artificial Intelligence Approach. Tioga Publ.
Co., Palo Alto, CA.
Quinlan J. R. and R.L. Rivest. 1989. Inferring decision trees
using the computation. Pages 80,227-48 in Information
and Computation. Academic Press, Reading, MA.
128
-------
Biological Criteria: Research and Regulation, 1991
Poeciliopsis: A Fish Model for Evaluating
Genetically Variable Responses to
Environmental Hazards
Lawrence E. Hightower
R. JackSchultz
Marine/Freshwater Biomedical Sciences Center
The University of Connecticut
Starrs, Connecticut
Recent studies of responses to cellular stress
(such as the heat shock response) com-
bined with earlier studies of inducible de-
toxification systems (such as the cytochromes P450)
have shown that organisms in general have sensi-
tive genetic systems for monitoring environmental
stressors. These methods for measuring stress use
an organism's own stress response induction path-
ways as onboard biosensors and induction prod-
ucts in the form of stress mRNA and proteins as
indicators of environmental stress.
This new approach proposes a new indicator,
cellular stress responses, as early warning systems
to signal the need for remedial action before severe
ecosystem disturbances result in loss of species. The
practice of relying on massive fishkills as indicators
of environmental problems does not take into ac-
count the numbers of weakened survivors that later
quietly die of secondary infectious diseases or can-
cer. As environmentalists become more prevention-
oriented, the monitoring of stress in feral animals,
as well as establishing species-specific risk levels,
may become the preferred goals. Several applica-
tions of this approach to environmental problems
are presented.
One example relates to the recent concern that
over the next decade global warming of roughly
2°C will be experienced. It is thought that this
warming could be progressive if remedial action is
not taken over the next decade. Climatic shifts of
this sort contain the ingredients of mass extinctions.
The effects depend on the degree of genetic varia-
tion available in populations. This experiment used
six species of the Sonoran topminnow, Poeciliopsis
and eight of its all-female hybrid clones as models
to evaluate genetic deployment of resistance to heat
stress in natural populations. It was established that
thermal resistance can be conferred by preheating
the fish to just below killing temperatures (37-38°C)
for one hour, after which survival at the normal kill-
ing temperature (39-41°C, depending on the bio-
type) is considerably enhanced. When liver tissues
from preheated fish are examined, using two-di-
mensional polyacrylamide gel electrophoresis, they
contain stress proteins that are not present in fish at
normal temperatures.
In addition, there is extensive biochemical di-
versity in the isoforms of two major families of heat
shock proteins (hsp70 and hspSO families), suggest-
ing that genetic variation in these proteins may con-
tribute to differences in thermal resistance among
the Poeciliopsis biotypes. The hsp70 and hspSO pro-
teins are thought to repair thermal damage and to
protect cells from lethal damage, respectively. Re-
cent studies have confirmed earlier hypotheses that
the heat stock response is keyed to protein damage
and appears directed toward restoring protein ho-
meostasis in cells subjected to a variety of stressors
in addition to heat, including heavy metal ions, ar-
senicals, amino acid analogues, and tissue trauma.
Differences in susceptibility to chemically in-
duced hepatocarrinogenesis, both in tumor inci-
dence and in tumor type, are found within species
as well as between species of Poeciliopsis. This diver-
sity of response among genotypes maintained in the
laboratory aquarium facility allows examination of
physical, chemical, and genetic factors that may
contribute to differences in tumor induction among
genotypes, The assemblage of biotypes in the col-
ony includes inbred and outcrossed stocks, as well
as unique all-female species that reproduce clonally,
thus allowing multiple replicates of wild genotypes
to be held constant while the environment is manip-
ulated. These fish are being used to examine
129
-------
L E HIGHTOWER and R. J. SCHULTZ
whether carcinogens become effective at lower-
than-threshold doses when hepatocyte proliferation
is initiated with independent stimuli such as heat
stress and chemical toxicants. Using dimethyl-
benz[a]anthracene (DMBA) as a toxicant in cell pro-
liferation studies, it has been established that the
number of days after treatment to maximum levels
of mitosis (2-12 days) is highly predictable and is in-
fluenced by the time of exposure (10-22 hours) to a
toxic concentration (5ppm) of DMBA.
Exposing Poetiliopsis to sublethal temperatures
for 30 to 60 minutes results in the death of embryos
in pregnant females and in liver cell damage to
adults. Hepatocyte proliferation is thus stimulated,
which peaks 2 to 3 days after the imposition of heat
stress. Subsequent studies will determine if prior
initiation of cell proliferation will enable tumors to
be induced at lower concentrations than in fish that
have not been exposed to heat. Since fish seeking
food enter water that is hot enough to risk their
lives, presumably it is hot enough to cause cell dam-
age and initiate unscheduled cell proliferation.
Many compounds do not become toxicants or
carcinogens until they are metabolically activated
by an oxidative enzyme, a cytochrome P450. Using
the Poetiliopsis hepatoma cell line, it was demon-
strated that cytochrome P450 activity can be in-
duced in cell culture. It has thus been possible to
carry out dose-response studies of hepatotoxicity
and modulation of P-450 inducers and inhibitors
(benzoflavone) on the effects of DMBA and
benzo[a]pyrene (BaP). Preliminary studies suggest
that the toxic levels of DMBA and BaP for cells in
culture are generally comparable to those for live
fish.
Working with nitrosodiethylamine (NDEA), it
has been determined that among Poeciliopsis geno-
types NDEA deethylase activity (liver microsomal
cytochrome P450pj) varies both in maximal activity
and in optimal temperature. Metabolic responses
after exposure to different concentrations of NDEA
will now be compared over a range of tempera-
tures, examining production of phase 1 metabolites
and the formation of DNA adducts.
This system enhances the understanding of fish
as monitors of domestic water supplies; it also pro-
vides a means to assess the variation in response to
chemical and thermal stress that is stored in the
gene pools of wild populations. Such variation may
play a major role in how manmade forces of selec-
tion will shape future populations.
ACKNOWLEDGMENTS: Members of the Ma-
rine/Freshwater Biomedical Sciences Center who have
contributed to this research are: J.F. Crivello, L.A.E.
Kaplan, P.J. dilorio, M.E. Schultz, J.J. Stegeman, and C.N.
White.
130
-------
Biological Criteria: Research and Regulation, 1991
Assessing Biological Integrity Using
EPA Rapid Bioassessment Protocol II
The Maryland Experience
Niles L. Primrose
Walter L. Butler
Ellen S. Friedman
Maryland Department of the Environment
Water Management Administration :
Annapolis, Maryland
The Maryland Department of the Environ-
ment, Water Quality Monitoring Division,
has begun using the EPA Rapid Bioassess-
ment Protocol II as part of a statewide water quality
monitoring network, and in selected special stud-
ies. The Protocol II, with 100+ organism subsamples
identified to family level, was considered an effi-
cient method for obtaining quality data from a large
number of streams.
Over 200 rapid assessment samples were com-
pleted during the 1990 field season. The streams
sampled were located in a number of different
ecoregions. The results from two ecoregions, the Po-
tomac drainage from the Allegany Plateau and the
coastal watersheds of the Choptank and Chester
Rivers, were chosen to illustrate our experience (the
Maryland experience).
Stations were chosen from 1/62,500-scale
county maps. The most downstream third-order
reach with a road crossing was the first choice. If
this location was inaccessible or lacked proper habi-
tat, the next closest road crossing was used. All but
minor impacts to the upper watershed of a stream
were assured to be reflected at the third-order sam-
pling location.
The sampling methods were based on the
Rapid Bioassessment Protocol II described by U.S.
EPA (Plafkin et al. 1989). Aim2 kick seine sample
was collected from the best available habitat, and a
random 100+ organism subsample obtained. The
subsample was identified to the family level in the
field, and the sample was archived for future refer-
ence. No Course Particulate Organic Matter sam-
ples were collected because of the varied
availability of this substrate. The habitat was as-
sessed at each station with habitat characteristics
customized for each ecoregion.
Reference streams were chosen from each ecore-
gion for both biological and habitat indices. The
choice of a reference stream was an intuitive deci-
sion taking into consideration all biotic, habitat, and
water quality factors. The -biological reference
stream was not necessarily the habitat reference
stream. ; > ,
The information from the family-level identifi-
cation was processed through the various metrics
described for Protocol II and a biological score-was
obtained. The biological and habitat scores were
then taken as a percentage of their respective refer-
ences. The results are plotted in Figures 1 and 2.
The divisions of unimpaired, moderately im-
paired, supporting, etc., are based on the character-
izations of poor, fair/good, and excellent that are
applied to the raw biological and habitat scores.
Streams that fell in the lower right-hand portion of
the graph tended to have water quality problems
such as acid mine drainage, STP effluents, and agri-
cultural runoff. The macroinvertebrate community
was severely impacted.
For purposes of initial characterization and
monitoring, those streams that fell in the severely
impaired classification were candidates for a more
intensive study to better define the source of the im-
pact. This study would use the Protocol III or other
quantitative methods.
The indices used in the assessment calculations
were described by Plafkin et al. (1989). Some of
these, particularly the ratios, did not correlate well
with the others except in cases of impacts. In an ef-
fort to overcome this shortcoming, a number of al-
ternate indices are being considered as substitutes,
such as Chironomidae/Total Diptera, Inverte-
brates/Total Trichoptera, and Non-insect Inverte-
brates/Total Sample. If these or other alternates
131
-------
N. L PRIMROSE, W. L BUTLER, and E S. FRIEDMAN
100-
90-
0)
o 80-
I 70-
0)
0= 60-
"5
^ 50-
co 40-
§> 30-
ffi 20-
10-
Unimpaired
Moderately
Impaired
Severely
Impaired
Nonsupporting
State Forest -
N,
Agricultural
"^" ,
Industrial
Moderately
0 10 20 30 40 50
Supporting
^^
• i
^^
i
/Acid Mine
• Supporting
60 70 80 90 100
Habitat - % of Reference
Figure 1 .—Potomac trlbutarles-AIIegany Plateau.
100-
90-
8 o«
8 80-
0)
a) 70-
Io5
cc 60-
5 50-
1
•§) 30-
o
.0 20-
CQ
10-
Unimpaired
Moderately
Impaired
Severely
Impaired
Nonsupporting
v 6 10 20
•
Ditched
^*^m
Heavy Woodland Buffer ^"JT™
Landfill ^ •
^
?
STP Effl.
m
Moderately Supporting
30 40 50 60
• • i
""w"
i
Agricultural
Supporting
70 80 90 10C
Habitat - % of Reference
Figure 2.—Choptank and Chester Rivers tributaries.
provide better correlation and information about
the sample, they will be included in all past and fu-
ture assessment calculations.
We feel that the U.S. EPA Rapid Bioassessment
Protocol II is an effective tool for initial characteriza-
tion and monitoring of streams in Maryland. It has
allowed broader coverage of state waters with mini-
mal additional time and expense, and has contrib-
uted toward the U.S. EPA goal of "fishable/swim-
mable."
References
Plafkin, J.L., M.T. Barbour, K.D. Porter, S.K., and R.M.
Hughes. 1989. Rapid Bioassessment Protocols For Use
in Streams and Rivers: Benthic Macroinvertebrates and
Fish. Assess. Watershed Prot. Div., U.S. Environ. Prot.
Agency, Washington, DC.
132
-------
Biological Criteria: Research and Regulation, 1991
The Use of the Qualitative Habitat
Evaluation Index for Use Attainability
Studies in Streams and Rivers in Ohio
Edward T. Rankin
Ohio Environmental Protection Agency
Columbus, Ohio
The Qualitative Habitat Evaluation Index (QHEI) has been developed to help distinguish the influ-
ence of habitat effects on fish communities in Ohio streams. The index is a composite of six habitat
variables: substrate, instream cover, riparian characteristics, channel characteristics, pool and riffle
quality, and gradient and drainage area. The index relies on visual estimates of several characteristics of each
habitat variable and can be completed in less than an hour for a 200-500 meter stream segment. Components
of each variable have been assigned scores based on observed or predicted relationships with fish species di-
versity and/or measures of community integrity. The QHEI was significantly correlated with the Index of
Biotic Integrity in Ohio streams and rivers, however the nature of the relationship varied by ecoregion
Ecoregion-level, reach-level, and subbasin-level habitat quality factors appear to act a "covariates" that
likely limit the site-specific predictability of any habitat indices that fail to consider them. The use of the
QHEI for use attainability analyses includes the compilation QHEI subcomponents by the aquatic life use
they are most strongly associated with (modified warmwater, warmwater, and exceptional warmwater hab-
itat) and the ratios of these subcomponents to one another in a given stream reach.
If you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
133
-------
The Use of the Amphipod Leptocheirus
Plumulosusto Determine Sediment Toxicity
in Chesapeake Bay: Development and Field
Applications
i
C.E. Schlekat
B.M. McGee
E. Reinharz
Maryland Department of the Environment
Baltimore, Maryland
Ideal species for testing the toxicity of estuarine sediments exhibit sensitivity to sediment contaminants
and are physiologically adapted to extreme variability in salinity and sediment type. We propose the
use of the amphipod Leptocheirus plumulosus to test the toxicity of sediments in Chesapeake Bay and
other east coast estuaries. This species is an ecologically important inhabitant of both oligqhaline and
mesohaline sections of Chesapeake Bay, and is found infaunaUy in sediments ranging from fine sand to very
fine mud. No significant differences were observed in amphipod survival among four salinity treatments (5,
15,20, and 32 ppt). Additionally, no significant differences were observed in amphipod survival among four
salinity treatments varying in particle size,and organic, content. Acute 96-hour LC-50 values for aqueous
cadmium at a salinity of 6 ppt were 0.26 mg cd/L and 0.19 mg cd/L for L. plumulosus and Hyalella aztesa, a
common freshwater sediment test organism, respectively. A field survey was conducted in which
Leptodieirus plumulosus were exposed to sediments from a variety of sites within Chesapeake Bay. The .sites
ranged from highly industrialized harbors to embayments containing commercial and community marinas.
Ambiguity between qualitative benthic analysis and amphipod survivorship at a portion of the test'sites
highlight the need to implement toxicity tests utilizing sublethal endpoints. Laboratory experiments con-
ducted to this end indicate that significant growth of juvenile L. plumulosus occurs under laboratory condi-
tions, and that morphological features allowing for the differentiation of male and female amphipods
appear after 20 days.
If you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
134
-------
Biological Criteria: Research and Regulation, 1991
Compliance Monitoring of the Aquatic
Biota in Vermont
Doug Burnham
Steve Fiske
Rich Langdon
Vermont Department of Environmental Conservation
Waterbury, Vermont
ABSTRACT
In 1986, Vermont Water Quality Statutes were amended to require all land-based waste-
water disposal systems with a capacity of greater than 6,500 gallons per day, including
spray irrigation and community sub-surface systems, to obtain an Indirect Discharge Permit
(IDR) from the Department of Environmental Conservation (DEC). This statutory amend-
ment established a narrative compliance criterion of "no significant alteration of the
aquatic biota" in surface waters adjacent to such systems (Vt. Dep; Environ. Conserv. 1986).
The DEC was charged with developing the rules and regulations that would implement this
criterion (Vt. Dep. Environ. Conserv. 1990). The legislative intent of the regulation was pri-
marily to protect fragile, high-elevation streams that were at risk from impacts that altered
the basic biological and aesthetic character of surface waters adjacent to high-volume waste
disposal systems installed to service recreational development. Under this interpretation,
discharges that caused "benign" alterations to biological integrity (e.g., benign enrichment),
but impaired other values and uses (e.g., proliferation of algal growth), would be consid-
ered significant under this criterion. The DEC was therefore required to develop numeric bi-
ological criteria that would evaluate alterations of the aquatic biota in terms of both
biological integrity and impact to nonbiological values and uses.
The Vermont Program
A protocol document for making determinations of
significant alteration in the context described in the
Abstract was prepared (Vt. Dep. Environ. Conserv.
1987). The document addressed the following
major considerations: (1) target monitoring com-
munity; (2) sampling strategy and site selection; (3)
sampling and processing methods; and (4) data
analysis and criteria development. For each of
these factors, a series of goals and objectives was
specified and methodologies that best met those
goals and objectives were determined. -
Overall, the Department of Environmental Con-
servation felt that aquatic macroinvertebrate sam-
pling was the most cost-effective means of
establishing a database from which compliance de-
cisions could be made. Macroinvertebrates were
chosen as the target community for a variety of rea-
sons, including sensitivity to perturbation, ability to
integrate impacts over time and across trophic lev-
els, and the high informational content of
macroinvertebrate samples. The existence of well-
established sampling and analytical methods, and
in-house data base and expertise were also impor-
tant considerations.
A paired site (control/impact) sampling proto-
col with on-site controls was selected as the primary
sampling strategy. The objective of site selection
was to isolate the stream reach potentially impacted
by the discharge to 'minimize the effect of non-
discharge-related perturbations.
The sampling methods were intended to reduce
variability within and between control and impact
sites caused by sampling error and habitat heteroge-
neity. A data quality objective (DQO) was estab-
135
-------
D. BURNHAM, S. FISKE, and R. LANGDON
lished consisting of population abundance estimate
with a percent standard error of less than 20 per-
cent. Available data indicated that these goals could
be consistently met or with the use of rock-filled
basket artificial substrates at a replication level of 5.
A minimum density of 300 animals per replicate
would be required to meet the DQO of 20 percent.
Sample processing following standard preservation
and in-lab sorting under magnification procedures
were specified, with special protocols for subsampl-
ing if necessary. The protocol specified taxonomy to
the lowest possible level using standard reference
keys for individual orders of animals. Sample ar-
chiving and reference collections were also re-
quired.
Because compliance monitoring was to be con-
ducted by the permittee or its agent, a relatively
high level of DEC oversight would be necessary to
ensure high quality data. Prior to initiating monitor-
ing, the permittee would be required to submit a de-
tailed QA/QC (quality assurance/quality control)
plan to the Department of Environmental Conser-
vation for approval. In addition, the Department
personnel were to conduct joint site visits and eval-
uations, process split samples, examine reference
collections, review all data submitted, and generally
maintain close contact with all agents generating
data.
A great variety of metrics can be used to de-
scribe the functional and structural characteristics
of macroinvertebrate communities. For the pur-
poses of determining significant alterations, the De-
partment of Environmental Conservation chose
four metrics for comparing alterations between con-
trol and impact sites.
1. The Pinkham-Pearson Coefficient of Simi-
larity was selected as a screening metric to •
make an initial evaluation of the degree of
community structure similarity between
control and impact sites.
2. A modification of the Hilsenhoff Biotic
Index was selected as an evaluation metric
primarily because of its sensitivity to alter-
ations in nutrient dynamics, the major antic-
ipated impact.
3. Ephemeroptera (mayfly), Plecoptera
(stonefly), Trichoptera (caddisfly) taxa
richness (EPT) was selected as an evalua-
tion metric because of the anticipated preva-
lence of these orders in target streams and
their sensitivity as an indicator of high qual-
ity diversity.
4. Finally, relative abundance was selected as
an evaluation metric because of the general
sensitivity of abundance to toxics, trophic
alterations, and overall habitat alterations.
The relatively extensive macroinvertebrate
database maintained by the Department of Environ-
mental Conservation was evaluated by Department
staff to determine the amount of change in the se-
lected metrics that would be indicative of signifi-
cant alteration. With the exception of relative
abundance, these change criteria were determined
independently of statistics and were selected to rep-
resent alterations of biological significance. In the
case of relative abundance, the biological and statis-
tical significances (p < .05, Mann-Whitney U-Test)
were thought to be equivalent. Exceeding the allow-
able change criterion in any one metric would result
in a determination of significant alteration. Because
of excessive data variability (failure to meet DQOs),
the change criteria could be exceeded without pro-
ducing a statistically significant change. Therefore,
a confirmation of statistical significance would be
required (p < .05, Mann-Whitney U-Test) prior to a
significant alteration determination.
Results
To date, more than 25 Indirect Discharge Permits
requiring biological monitoring have been issued
by the Department of Environmental Conserva-
tion. Monitoring frequency is determined primar-
ily by the size of the system and ranges from twice
per year (winter and late summer) to once during
the five-year life of the permit. The regulated com-
munity has been very responsive and cooperative
in implementing compliance monitoring. The De-
partment of Environmental Conservation and
monitoring personnel have maintained close com-
muni- cations and have worked together to resolve
problems as they arose. Communication is facili-
tated by the small size of the state of Vermont and
the relatively small number of consultants in-
volved in monitoring activities.
In general, performance expectations have been
met. Macroinvertebrates have proven to be an excel-
lent monitoring community providing data ade-
quate for making informed decisions. Consulting
biologists have demonstrated considerable exper-
tise in conducting monitoring activities and pro-
cessing and analyzing samples.
At the same time, sampling strategy and sam-
pling methods have not always attained the stan-
dards set by the regulations. In some cases,
appropriate paired sites on the same stream are not
136
-------
Biological Criteria: Research and Regulation, 1991
available because of physical limitations or poten-
tial degradation of control sites by nonregulated
perturbations. Attempts to locate control sites in ad-
jacent watersheds have met with mixed success. By
far the greatest problem has been the physical dis-
placement of rock baskets during exposure because
of extreme hydrologic events. This problem has
been minimized by close observation of weather
conditions and substrates during exposure. Another
persistent problem has been low productivity of re-
ceiving waters, resulting in less than the minimum
number of organisms per replicate required to meet
DQOs. As a consequence, the data have not been
adequate for making regulatory decisions in some
cases. Selection of criteria indicating significant al-
teration of the aquatic biota has proven to be appro-
priate in most cases, although a review to evaluate
the need for modifications is underway.
The majority of sites being monitored have
demonstrated compliance with the criteria. In cases
where alteration of biota has been found, enforce-
ment response has ranged from increased monitor-
ing intensity to major modifications to treatment
systems.
The use of numeric biological criteria for regula-
tory purposes is a potentially contentious process. It
is critical that final decisions regarding noncompli-
ance and subsequent enforcement response not be
executed in a technical vacuum, but rather with full
consideration for interpretations of the monitoring,
data using the best professional judgment of both
regulating and consulting biologists.
In addition to providing compliance informa-
tion, biological data collected through the Vermont
program have provided additional benefits. General
knowledge of stream ecology and the response of
stream biota to low levels of pollution has been
greatly expanded. Chemical monitoring permit re-
quirements provide data with which to evaluate
dose/response observations in the receiving water.
The general awareness of aquatic biota on the part
of regulators and the regulated community has im-
proved, to the ultimate benefit of pragmatic water
quality management. In some cases, data generated
through this program have detected water quality
degradation in receiving streams caused by sources
unrelated to the discharge being monitored, such as
failed erosion control systems.
, In summary, biological compliance monitoring
in Vermont has resulted in the production of high-
quality da,ta describing actual in-stream impacts
from indirect discharges. These data have been suc-
cessfully used to make compliance decisions that
are acceptable to both regulators and the regulated
community. This program has demonstrated that
biological monitoring, when applied in a program
with clear goals and objectives, a high degree of
QA/QC, numerical standards, and cooperation be-
tween regulators and the regulated community can
be a valuable and extremely pragmatic water qual-
ity management tool.
References
Vermont Department of Environmental Conservation. 1986.
Compliance Monitoring of the Aquatic Biota—Pursuant
to 1986 Legislative Amendments to 10 V.S. A. Chapter 47
(Actl99,S.B.S-42).
—'• .-1987. Biological Compliance Monitoring Methods
Manual. Vt. Dep. Environ. Conserv., Waterbury.
. 1990. Indirect Discharge Rules. Chapter 14 in Envi-
ronmental Protection Rules. Vt. Dep. Environ. Conserv.,
Waterbury.
137
-------
A Method for Rapid Bioassessment of
Streams in New Jersey Using Benthic
Macroinvertebrates .
James Kurtenbach
U.S. Environmental Protection Agency, Region II
Edison, New Jersey
A family level rapid bioassessment method (RBM) using benthic macroinvertebrate communities
was developed and field tested for water quality evaluations in New Jersey streams. The method is
a regional modification of EPA's family level rapid bioassessment protocol developed to screen
and prioritize sites having impaired water quality. Depending on geographic location, macroinvertebrates
are sampled from riffle areas or multiple instream habitats using kick net procedures. The RBM was applied
to a set of approximately 200 sites over a two year period. The community analysis used to determine bio-
logical condition consist of five biometrics: 1) total taxa richness, 2) Ephemeroptera, Plecoptera and
Trichoptera (EPT) richness, 3) percent dominance, 4) percent EFT and 5) modified biotic index. Biological
criteria were established for three categories of water quality (non-impacted, moderately impacted and se-
verely impacted), and the natural variability associated with individual biometrics was examined. Replicate
sample comparisons made in several unimpacted reference streams did not result in assignment of differing
water quality categories, suggesting that variability associated with individual biometrics was not sufficient
enough to cause inaccurate water quality assessment. The rapid bioassessment appears to provide an ac-
ceptable approach to accurately screen for water quality impairment.
If you would like further details on this subject matter, pkase feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
138
-------
Biological Criteria: Research and Regulation, 1991
Development of Biological Impairment
Criteria for Streams in New York State
Robert W. Bode
New York State Department of Environmental Conservation
Albany, New York
; ABSTRACT
Biological criteria were recently developed for measuring Significant water quality impairment in flow-
ing waters of New York state (Bode et al. 1990). The criteria established are based on sampling benthic
macromvertebrate communities; they measure impairment as a quantitative change from conditions
upstream of a given discharge. :The sampling methods used are the traveling kick method for stream
segments with wadeable riffles, and the multiple-plate artificial substrate sampler for stream segments
; Without wadeable, riffles. Replication in samplingis necessary tp insure reliability of data The parame-
, ters on which thecriteria are based are bioticindex, EPT value, species richness; species dominance, and
percent model affinity. Because the criteria are directed, toward enforcement rather than detection they
are numerical rather than narrative, and site-specific, rather than regional. Site-specific criteria have ad-
, vantages over regional criteria in accounting for naturalyariability by comparing results to an upstream
, control si^e; this approach is also able to target the causeof impairment to specific discharges. Habitat
comparability criteria were established to ensure high habitat similarity between the upstream arid
downstream sites. The parameters measured are current speed, substrate particle size, substrate
embeddedness, and canopy. The proposed criteria were drawn from data sets collected from flowing
waters in New York state over a 17-year period (1972-89). Preliminary criteria were based on changes
between levels of an existing four-tiered classification of water quality used in New York state The cri-
teria were then tested over a 2-year period and modified as necessary. Issues tested include sensitivity
and accuracy of the criteria, adequacy of the 2-minute/5-m kick sample, replicate variability, adequacy
of the habitat criteria, and seasonal variability. Data from sites designated as having significant biologi-
cal impairment were corroborated with available chemical data to confirm possible impairment
Specifications of Biological
Impairment Criteria
Sampling Methods
Two sampling methods are used, dependent on the
availability of wadeable riffles. For streams with
wadeable riffles in the desired reach, the traveling
kick method is used, taking three 2-minute/5 m
samples. One hundred organisms are subsampled
from each sample. For streams without available
wadeable riffles, multiple-plate artificial substrate
samplers are used, with three 5-week exposure pe-
riods. Multiple-plate samplers have been used in
New York state since 1972, with the modifications
of 'using ;three hardboard plates^ each 6 inches
square, suspended 1 m below the water surface.
Index Levels
Significant biological impairment is indicated for
kick samples when one or more of the levels in in-
dices a-e is exceeded and the change is also shown
to be statistically significant at the level of P = .05.
Significant biological impairment is indicated for
multiplate samples when one or more of the levels
in indices a-d is exceeded and the change is also
139
-------
R. W. BODE
shown to be statistically significant at the level of
P = .05.
a. Biotic index. The biotic index is calculated
by multiplying the number of individuals of
each species by its assigned tolerance value,
summing these products, and dividing by
the total number of individuals. Tolerance
values have been assigned on a scale of 0 to
10. The criterion for this parameter is + 1.5;
an increase of 1. or more exceeds the allow-
able amount of change.
b. EFT value. The total number of species in
the orders Ephemeroptera (mayflies),
Plecoptera (stoneflies), and Trichoptera
(caddisflies) found in the sample or subsam-
ple. The criterion for this parameter is -4; a
decrease of 4 or more exceeds the allowable
amount of change.
c. Species richness. The total number of spe-
cies found in the sample or subsample. The
criterion for this parameter is -8; a decrease
of 8 or more species exceeds the allowable
amount of change.
d. Species dominance. This is the percent con-
tribution of individuals of the most numer-
ous species or taxon in the sample. The
criterion for this parameter is + 15; an in-
crease of 15 or more exceeds the allowable
amount of change.
e. Percent model affinity. This hew index is a
measure of similarity to a model non-
impacted community based on percent
abundance in seven major groups (Novak
and Bode, in prep.). Percentage similarity is
used to measure similarity to a community
of 40 percent Ephemeroptera, 5 percent
Plecoptera, 10 percent Trichoptera, 10 per-
cent Coleoptera, 20 percent Chironomidae, 5
percent Oligochaeta, and 10 percent Other.
The criterion for this parameter is -20; a de-
crease of 20 or more Exceeds the allowable
amount of change.
Procedures for Application of
Biological Impairment Criteria
1. Choose appropriate sampling method (kick
sampling or multiplate sampling) by deter-
mining availability of wadeable riffles.
2. Select an upstream site and downstream site
that meet the habitat criteria for site compa-
rability for current speed, substrate particle
size, substrate embeddedness, and canopy.
3. Conduct sampling at the upstream and
downstream site using kick sampling in
streams with wadeable riffles and multipl-
ate sampling in all other streams. For kick
sampling, four replicates are collected at
each site; for multiplate sampling, three 5-
week exposures are conducted.
4. Conduct laboratory sorting and identifica-
tion of samples, using the level of taxonomy
required for each group.
5. For kick samples, use percentage similarity
to calculate similarity between three of the
replicates at each site. If similarity if less
than 50 for any replicate pairing, resubsam-
ple 100 organisms from the replicate with
the lowest average similarity. If similarity is
: still less than 50 for the replicate pairing,
subsample a fourth replicate from the site. If
50 percent similarity cannot be achieved
with these replicates or subsamples, resam-
pling is necessary.
6. Calculate parameters a-e for kick samples
and parameters a-d for multiplate samples.
Compute the average for the three samples
from each site.
• Biotic index
• EPT value
• Species richness
• Species dominance
• Percent model affinity
7. Compare values from the downstream site
to those from upstream site (Fig. 1). For kick
samples, violation of one or. more criteria for
parameters a-e indicates provisional im-
pairment. For multiplate samples, violation
of one or more criteria for parameters a-d
indicates provisional impairment.
• Biotic index: + 1.5 (0-10 scale)
• EPT value:-4
• Species richness: -8
• Species dominance: + 15
• Percent model affinity: -20
140
-------
Biological Criteria: Research and Regulation, 1991
E. PERCENT
MODEL AFFINITY*
D. SPECIES
DOMINANCE
C. SPECIES
RICHNESS
B. EPT
VALUE
A. BIOTJC
INDEX
IMPAIRMENT IS BASED ON
EXCEEDING ANY 1 CRITERION
Figure 1.—Biological Impairment criteria for flowing waters In New York state. *Percent model affinity Is not used with multi-
plate samples.
8. For sites with provisional impairment, per-
form the Student's T-test to determine if re-
sults are statistically significant at the level P
= .05. If results are significant, biological im-
pairment is indicated.
References
Bode, R.W., M.A. Novak, and L.E. Abele. 1990. Biological im-
pairment criteria for flowing waters in New York State.
Tech. Rep. New York State Dep. Environ. Conserv., Al-
bany.
Novak, M.A. and R.W. Bode. In prep. Percent model affinity,
a new measure of macroinvertebrate community com-
position.
141
-------
Development of Sediment Criteria for
the Protection and Propagation of
Salmonid Fishes
Timothy A. Burton
William H. Clark
Geoffrey W. Harvey
Terry R. Maret
Idaho Department of Health and Welfare
Division of Environmental Quality
Water Quality Bureau
Boise, Idaho
ABSTRACT
Salmonid spawning and rearing are protected beneficial uses of waters in most western states.
Nonpoint source activities causing accelerated sedimentation to streams can adversely affect sal-
monid growth and survival. Water quality criteria, proposed for inclusion in Idaho's water qual-
ity standards, have therefore been focusing on protection of developing embryos an4 young fish
from the detrimental effects of sediment. The approach is supported in the literature and has
been verified by field testing in Idaho. . , ,
Introduction
The feedback loop concept of nonpoint source pol-
lution control has been incorporated into Idaho
water quality standards (Ida. Dep. Health and
Welfare, 1990). This concept requires development
of in-stream criteria to protect the beneficialuses of
the State's waters. Feedback from in-stream moni-
toring is compared to the in-stream criteria to de-
termine whether or not best management practices
(BMPs) applied to nonpoint source activities are ef-
fectively protecting the beneficial uses.
Fine sediment pollution that impairs habitat for
rearing and reproducing salmonid fishes has been
reported in 90 percent of all impacted stream seg-
ments in Idaho (Ida. Dep. Health and Welfare,
1988). This condition prompted water quality ex-
perts to initiate an extensive review of the literature
covering sediment effects to fish. Based on this re-
view (Chapman and McLeod, 1987) and the advice
of 50 local and regional technical experts, sediment
criteria were proposed for inclusion in the State
water, quality standards (Harvey, 1989).
One criterion is designed to protect incubating
salmonid eggs from the detrimental effects, of fine
sediment on critical dissolved oxygen delivery
through the substrate. Another prevents increases
in sediment accumulation in cobble rearing spaces
critical to over-winter survival of young salrnonids.
Salmonid Embryo Survival in
the Spawning Redd
Methods
Chapman and McLeod (1987) concluded that for
incubating salmonid embryos, survival to emer-
gence is inversely related to the proportion of fine
sediment increases in the incubation environment.
142
-------
Biologioal Criteria: Research and Regulation,
P
P
m
12
I
G 10
D a
O .
5 10 15 20 25 30 35 40 45 50 55 60
PERCENT FINES (< 6.3 mm)
Figure 1 .—Effect of fine sediment on dissolved oxygen in the intragravel incubation environment
measured on Rock Creek, Idaho.
Also, survival of salmonid embryos is positively
correlated with apparent velocity and permeability.
Dissolved oxygen affects both emergence success
and timing.
A methodology for monitoring sediment im-
pact on incubation of salmonid embryos in-situ has
been developed. The technique measures in-
tragravel dissolved oxygen, percent intragravel fine
sediment, and percent survival of embryos to emer-
gence in artificial egg pockets. Monitoring in natu-
ral egg pockets has proven ineffective and
destructive to the beneficial use. The artificial redd
technique permits measurement of the fine sedi-
ment infiltrating egg pockets and the dissolved oxy-
gen concentration surrounding the incubating
embryos. These values are compared with egg sur-
vival and alevin escapement from the artificial egg
pockets.
Results
Testing in Idaho has shown the technique to be use-
ful in varying seasons and stream conditions. The
validation work also verified that fine sediment im-
pairs permeability within egg pockets resulting in
dissolved oxygen depression sufficient to suffocate
incubating eggs (Maret et al. in prep.). Figure 1
shows that as fine sediment approaches 40 percent,
dissolved oxygen within the intragravel environ-
ment decreases to levels impairing growth and sur-
vival of incubating embryos. Tests conducted in
substrates with coarser sediments showed little or
no dissolved oxygen depression, but mortalities by
entrapment were observed among developed al-
evins trying to escape heavily sedimented egg
pockets (Burton et al. 1990). Excessive fine sedi-
ment may also affect growth and condition of sur-
viving embryos as indicated by conclusions from
the Rock Creek Rural Clean Water Program study
(Maret et al. in prep.).
The proposed salmonid spawning criterion is
based on dissolved oxygen concentrations within
the incubation environment. Attempts to establish a
permeability criterion indicated that it would be
technically unfeasible. In practice, intergravel dis-
solved oxygen should be a good surrogate for
gravel permeability. No standard methodology cur-
rently exists to quantify escapement success. An in-
terim standard is being developed for use until
functional relationships between percent fine sedi-
ment and alevin survival to emergence have been
established.
The proposed criterion is:
• Nonpoint source activities shall not cause in-
tragravel dissolved oxygen in spawning grav-
els to decline, below a weekly average of 6
milligrams per liter.
Salmonid Survival in the
Intercobble Environment
Methods
The interstitial space found in streambed cobble
habitats is important to survival of juvenile salmo-
nids. These fish use the interstitial space primarily
for feeding and refuge cover, especially in winter.
When this habitat has been replaced by the intru-
143
-------
T. A BURTON, W. H. CLARK, G. W. HARVEY, and
T.R. MARET
sion of fine sediment, salmonids must find other
suitable habitat either by migrating from the
stream reach or within the same stream. Chapman
and McLeod (1987) found in their literature review
that real and detectable relationships exist between
land-disturbing activities and increased fine sedi-
ment in the aquatic environment. The weight of ev-
idence indicates that areas ;with high
embeddedness tend to have lower densities of sal-
monids and additional sediment that reduces liv-
ing space, increases mortality.
A protocol for measuring embeddedness has
been developed by Burns and Edwards (1985). The
method was further refined for sampling design
and statistical treatments by Skille and King (1989).
Using this approach, cobbles within a specified size
range are drawn from a 60 cm diameter sampling"
plot on the bed of the stream. Each cobble is mea-
sured for depth embedded in fine sediment. Areas
within the plot completely covered by fine sediment
are weighted as fully embedded. The mean of all
measurements on the plot is counted as one sample.
A number of samples (or plot measurements) are
collected from random locations in the stream over
a stream reach equal to 20 times the channel width.
The number of samples (plots) ranges from 10 to 50,
depending on the sample variability of the stream.
As variability increases, more samples are required.
The standard sample size is equal to the number
needed to predict the mean cobble embeddedness
within a 95 percent confidence interval on the t sta-
tistic.
Results
Using this technique for measuring cobble
embeddedness has allowed quantification of inter-
cobble habitat degradation resulting from exces-
sive sedimentation to the stream. Embeddedness
measurements have always demonstrated a high
inverse correlation to living space used by the fish.
In addition, streams impacted by nonpoint source
activities show significantly higher embeddedness
as compared with minimally impacted control
sites.
The proposed criterion is:
• No statistically demonstrable increase, at the
95 percent confidence interval, in natural base-
line percent embeddedness as the result of
nonpoint source activities shall be permissible
in salmonid rearing habitats. Impacts of sedi-
mentation on interstitial space habitats impor-
tant to salmonid rearing will be assessed by
measurement of cobble and rubble percent
embeddedness. Baseline percent embedded-
ness will be determined by a quantitative tech-
nique in-stream reaches with similar
geomorphology and stream power which are
unaffected by nonpoint source sedimentation.
A percent embeddedness value will consist of
a mean at the 95 percent precision level of the t
statistic.
Summary
Excessive fine sediment impairs salmonid growth
and recruitment. The effect within the incubation
environment is to reduce intragravel flow velocity
and therefore the delivery of oxygen to developing
embryos. Fine sediment intrusion in the top layers
of the egg pocket may also restrict emergence after
development of the embryos.
Intercobble space is a critical habitat for juvenile
salmonids. Replacement by fine sediment severely
degrades this environmental requisite.
Quantitative methods for estimating and moni-
toring sediment effects on salmonids have been
specified. As a result, biocriteria have been devel-
oped and proposed for inclusion in Idaho's water
quality standards.
References
Burns, D.G. and R.E. Edwards. 1985. Embeddedness of sal-
monid habitat of selected streams on the Payette Na-
tional Forest. U.S. Dep. Agric. Forest Serv., Payette Natl.
Forest. McCall, ID.
Burton, T.A., G.W. Harvey, and M.L. McHenry. 1990. Proto-
cols for assessment of dissolved oxygen, fine sediment
and salmonid embryo survival in an artificial redd. Rep.
1. Idaho Dep. Health Welfare, Div. Environ. Qual.,
Boise.
Chapman, D.W. and KP. McLeod. 1987. Development of Cri-
teria for Fine Sediment in the Northern Rockies Ecpre-
gion. Final Rep. EPA 910/9-87-162. U.S. Environ. Prot.
Agency, Washington, DC. ' '•
Harvey, G.W. 1989. Technical review of sediment criteria,
Idaho Dep. Health Welfare, Div. Environ. Qual., Boise. ,
Idaho Department of Health and Welfare. 1988. Idaho Water
Quality Status Report and Nonpoint Source Assess-
ment. Div. Environ. Qual., Boise.
. 1990. Rules and Regulations, Title 1, Chapter 2, Rules
governing water quality standards and wastewater
treatment requirements. Div. Environ. Qual., Boise.
Maret, T.R., T.A. Burton, G.W. Harvey, and W.H. Clark. In
prep. Evaluating agricultural impacts on brown trout
spawning success in Rock Creek, Twin Falls, Idaho.
Idaho Dep. Health Welfare, Div. Environ. Qual., Poise.
Skille, J. and J. King. 1989. Proposed cobble embeddedness
sampling procedure. U.S. Dep. Agric. Forest Serv., Inter-
mount. Res. Sta. Coeur d'Alene, ID.
144
-------
Biological Criteria: Research and Regulation, 1991
Biomonitoring Methods in the
Tennessee Valley
Anne E.Keller
St. John's River Water Management District
Palatka, Florida
Neil E. Carriker
Water Resources
Tennessee Valley Authority
Chattanooga, Tennessee ,
ABSTRACT
The Tennessee Valley Authority is expanding its water monitoring program to include
biomonitoring methods. Methods being compared include EPA's Rapid Bioassessment
Protocols for both fish and macroinvertebrates, Karr's IBI as modified by Saylor, extensive
quantitative and qualitative macroinvertebrate sampling,the EFT evaluation of the insect
community used by North Carolina Department of Environmental Management, peri-
phyton sampling, and both laboratory and ambient toxicity tests where appropriate. This
effort will require the adaptation of techniques and analyses to reflect conditions in the
chemistry, and stream habitat quality will also be^ assessed. These comparisons will be
performed on a set of rivers that have been impacted by agricultural use, heavy metals, or
xenobiotics. Results will be used to determine the efficiency of each method and its utility
in evaluating specific impacts. From these data an organized, cost-effective, and adaptable
approach to water quality assessment will be developed for the Tennessee Valley.
Introduction
The Tennessee Valley Authority (TVA) performs bi-
ological surveys on a wide variety of streams and
rivers in the Tennessee River basin. The sampling
strategy often is designed to complement chemical
and physical monitoring, and typically requires
collecting both fish and macroinvertebrates. Recent
efforts have included analyses of community-level
changes relative to pollution (e.g., IBI), toxicity
tests, and measurements of various contaminants
in fish tissue. Even with intensive field work from
March to September, water quality can be assessed
on only a fraction of the watershed.
To improve and expand this monitoring effort
with a fixed or decreasing budget and limited man-
power, efficiency must be increased. The TVA is
conducting a comparison of biomonitoring meth-
ods in Spring 1991.
TVA has employed a variety of techniques to as-
sess the water quality of streams and rivers, includ-
ing chemical, physical, and biological surveys (fish,
macroinvertebrates), and toxicity tests. While new
methods appear in the literature every few years
and others are modified, the appropriateness of the
methods TVA is currently using or the utility Of
newer methods, such as the rapid procedures being
promoted by the U.S. Environmental Protection
Agency (1989) have never been evaluated.
Currently, Aquatic Biology staff evaluate water
quality from analysis of the fish community using
several standard procedures: the Index of Biotic In-
tegrity (IBI) developed by Karr (1981); an assess-
ment of the macroinvertebrate community with
EPT taxa (Ephemeroptera, Plecbptera, Trichoptera),
total taxa, and percent composition; and chemical
analyses of water samples. These methods are labor
intensive, expensive, and require months to pro-
duce results. Such a delay is often unacceptable.
With the expansion of TVA biomonitoring ef-
forts, it is more important than ever to improve field
and analytical efficiency. The recently published
145
-------
A E KELLER and N. £ CARRIKER
EPA manual on rapid bioassessment procedures in-
cludes fester, more field-oriented methods to deter-
mine the health of the fish and macroinvertebrate
communities (U.S. Environ. Prot. Agency, 1989).
These methods reduce sampling and processing
time, and permit fester determination of water qual-
ity. However, little information is available on the
validity of these rapid methods in comparison with
traditional methods or their utility under a wide va-
riety of circumstances. In addition, the project will
determine the feasibility and validity of in situ toxic-
ity tests, which may be less expensive than labora-
tory toxicity tests. Finally, the project will assess the
effectiveness of algal indices of water quality.
The comparison study will be conducted in
Spring 1991, on four rivers (Middle Fork Holston,
Pigeon, Big Sandy, and Oostaunala Creek). These
rivers vary in size and have different water quality
problems. The traditional IBI and macroinvertebr-
ate assessments will be coupled with rapid methods
proposed by EPA, in situ toxicity tests with fathead
minnows and Ceriodaphnia dubia, algal identifica-
tion, and analyses of water chemistry. The project
hopes to determine which methods are fastest, least
expensive, and most accurate for characterizing
water quality in various sized streams with differ-
ent water quality problems.
Summary
Materials and Methods
• Biological and chemical samples taken in four
rivers (see Fig. 1).
• Analytical methods
•IBI
• Complete macroinvertebrate sampling
• Rapid bioassessment methods
• In situ toxicity tests
• Water chemistry analyses
Results
• Comparability of various biomonitoring
methods in identifying water quality problems
• Ability of various methods to discriminate
between point and nonpoint source pollution
• Sensitivity of methods in streams and
watersheds of different sizes
• Cost/benefit analysis of methods
• List of suggested methods, based on budget,
known problems, stream conditions.
References
Karr, J.R. 1981. Assessment of biotic integrity using fish com-
munities. Fisheries 6:21-27.
U.S. Environmental Protection Agency. 1989. Rapid
Bioassessment Protocols for Use in Streams and Rivers.
Benthic macroinvertebrates and fish. EPA 444/4-89-001.
Assess. Watershed Prot. Div., Washington, DC.
Big Sandy R.
VA
Middle Fork Holston R.
NC
Pigeon R.
Oostanaula Cr.
FIgurt 1.—Tannessoa River watershed (40,910 sq. mi.).
146
-------
Biological Criteria: Research and Regulation, 1991
Development of Biological Criteria
for Use in Maine's Water Quality
Classification Program
Susan P. Davies
Leonidas Tsomides
David L. Courtemanch
Maine Department of Environmental Protection
Augusta, Maine
Francis Drummond
Department of Entomology
University of Maine
Orono, Maine
ABSTRACT
The Maine Department of Environmental Protection has begun the process of incorporating the
use of biocriteria into its water assessment program. The Department established narrative
standards that described uses and characteristics of each class of water within its classification
system. Macroinvertebrate data were used as the basis of a three-stage protocol to assign water
classifications to specific waterbodies and target those that could benefit from remediation. The
first stage classifies samples into one of four water quality groups according to macroinvertebr-
ate characteristics, based on a linear discriminant model. This model produces a 74 to 88 per-
cent probability of correctly placing a biological community sample into its appropriate
classification. The second stage refines the prediction by applying class-specific criteria to the
samples to assess attainment of the unique standards of a given classification. The third stage
utilizes the expertise of biologists to adjust the scores obtained in stages one and two. The result
is a rapid and accurate decisionmaking tool that utilizes both statistical probability and human
judgment to assess aquatic life in waterbodies in the state of Maine.
Introduction
The State of Maine began its development of
biocriteria by establishing narrative standards for
each of the classes within its water classification
system. The purpose of these narrative standards
was to identify specific conditions of the biological
community that supported the uses and character-
istics of each class of water, rather than to merely
establish a ranking from "good" to "poor." These
standards and accompanying statutory definitions
identify specific attributes of the biological commu-
nity for evaluation (Courtemanch and Davies
1987).
The benthic macrpinvertebrate community was
chosen as a practical community component for
sampling and evaluation. Each class is distinct from
the others and thus requires a different set of met-
rics or different criteria values for the metrics.
Maine is now at the stage of proposing numeric cri-
teria to interpret and evaluate the narrative stan-
dards. •
The original proposal of a hierarchical test de-
sign (Courtemanch and Davies, 1987) has been ex-
panded to include a three-stage test. The first stage
includes general tests of the macroinvertebrate data,
followed by the application of the data to a linear
147
-------
S. P. DAV1ES, L TSOMIDES, D. L COURTEMANCH, and
F. DRUMMOND
discriminant model that determines the probability
for placement in any class. The second stage tests
specific attributes of the assigned class, using both
data compared to a reference site when available,
and unreferenced data. The final stage provides for
the use of professional judgement to adjust the
scores of the first two stages, taking into account
both ambiguous findings resulting from data collec-
tion and processing, and unique habitat conditions
that influence community development.
First-Stage Criteria
To develop the first-stage model, an initial classifi-
cation assignment reflecting the narrative stan-
dards had to be constructed. Three agency
biologists, familiar with macroinvertebrate data in-
terpretation, independently and blindly (site iden-
tity unknown) evaluated data from 145 samples
and assigned one of four classes (A, B, C, or non-
attainment of the lowest class.) Next, results of this
evaluation were compared. There was unanimous
agreement in the assignment for 114 (79 percent) of
the samples. The remaining samples were then re-
evaluated collectively by the biologists with site in-
formation revealed and a consensus classification
was assigned. A subset of the data was also pro-
vided blindly to two nonagency biologists from a
technical review committee overseeing the criteria
development process. Concurrence was found for
86 percent of the samples between the agency biol-
ogist assignments and nonagency biologists. It was
concluded that there was substantial agreement in
the biological interpretation of the statutory lan-
guage.
Following classification of the test data set, it
was possible to develop a model that best simulated
the biologists' decisions. To do this, all the criteria
that any of the agency biologists had used in mak-
ing their determinations were quantified. The out-
come was 31 variables (Table 1). No reference
comparative variables (e.g., percent change, similar-
ity) were included at this stage.
Factor analysis and stepwise discriminant anal-
ysis reduced this number to nine quantitative vari-
ables (Table 1). These nine discriminating variables
were used to build a linear discriminant model
(Green and Vascotto, 1978). A jackknife procedure
(Mosteller and Tukey, 1977) was used to assure the
stability of the model using four runs, each with 25
percent of the samples removed. The model assigns
a classification based on the highest probability for
membership in one of the four classes.
Table 1.—Metrics used in initial biologist classifica-
tion. Underlined metrics are those selected for the
discriminant model.
1. Total abundance (log transformed)
2. Generic richness
3. Ephemeroptera abundance (log transformed)
4. Plecoptera abundance (log transformed)
5. Ephemeroptera abundance / Total abundance
6. Plecoptera abundance / Total abundance
7. Ephemeroptera richness
8. Ephemeroptera richness / Generic richness
9. Plecoptera richness
10. Plecoptera richness / Generic richness
"11. Ephemeroptera + Plecoptera + Trichoptera (EPT) richness
12. EP richness / Generic richness
13. EPT richness / Diptera richness
14. Non-EPT richness / Generic richness
15. Oligochaete abundance / Total abundance
16. Hirudinea abundance / Total abundance
17. Gastropoda abundance / Total abundance
18. Chironomidae abundance / Total abundance (log transformed)
19. Diptera richness / Generic richness
20. Tanypodinae abundance / Total abundance
21. Tribelos abundance / Total abundance
22. Chironomus abundance / Total abundance
23. Hvdropsyche abundance
24. Hydropsyche abundance / Total abundance
25. Glossosoma abundance / Total abundance
26. Brachycentrus abundance / Total abundance
27. Percentage of predator abundance
28. Ratio of collector-filterers + collector-gatherers to
predators + shredders
29. Number of functional feeding groups represented
30. Hilsenhoff Biotic Index
31. Shannon-Wiener diversity index
Table 2 compares the classification assignment
to the model-predicted classification. In making a
decision on attainment, the model must correctly
predict class assignment; however, errors where the
model predicts a higher class are tolerable since at-
tainment decisions are based on minimum condi-
tions. Conversely, the model can err by predicting
that a community is a lower class than assigned
(right of the matrix line). The frequency of this type
of error was judged to be acceptable because the po-
tential remains for the error to be corrected in the
second and/or third stages of the protocol. The clas-
sification probabilities are used as the raw score for
the first-stage test decision.
Table 2.—Percentage of concurrence between as-
signed and model predicted classification.
Model predicted class
Assigned class
• A
B
C
NA
A
74
13
0
0
B
26
76
25
0
C
0
11
75
12
NA
0
0
0
88
148
-------
Biological Criteria: Research and Regulation, 1991
Second-Stage Criteria
Second-stage analysis is separated into that
for samples with high water quality reference
sites (compatible with Class A standards), and
those without reference sites. While the first-
stage analysis provides a prediction and prob-
ability based on variables or discriminators
common to all classes, the second stage pro-
vides further testing of the data using unique
tests within each class. These tests are chosen
to address aspects of the standards that are
specific for each class, and which have not
been used to build the linear discriminant
model.
Second-stage criteria first assess the pres-
ence and abundance of indicator taxa. Indica-
tor taxa were selected based on: (1) significant
occurrence in the entire data set (occurring at
&10 percent of all sites), thus being sufficiently
common to be a good indicator; (2) dominant
abundance in the specific class (a60 percent of
total abundance in the dataset occurring in the
specified class); and (3) significant occurrence
within the specified class (occurring at &25 per-
cent of the sites in the specified class). Some of
the strongest indicator taxa include:
s
• Class A: Leucrocuta, Paragnetina
(Plecoptera); Serratella, Eurylophella
(Ephemeroptera); Bmchycentrus,
Psilotreta (Trichoptera)
• Class B: Baetis and Tricorytkodes
(Ephemeroptera); Chimarra, Neuredipsis
and Lepidostoma (Trichoptera); Simulium
(Diptera)
• Class C: Dicrotendipes and Conchapelopia
(Diptera).
An additional approach used in the second
stage of analysis draws on information available
from comparisons of the test site to a clean water
"upstream" reference site. Reference site criteria
were set to ensure closely matched habitat condi-
tions (reference sites usually located upstream on
the same waterbody), and very high water quality
(reference sites must attain Class A standards).
Comparative indices are computed for these sites
and the results are used to strengthen confidence in
the likelihood of a site belonging to a given classifi-
cation (Fig. 1).
Jaccard Index
C
B
Class A
0
1 ^
i< i i
~ \ i i j —
°-l 0.2 0.3 0.4
1
0.5
Dominant Taxa Similarity
Class A
1 1
1
1 1 1 1
0 0.1 0.2 0.3
0.4 0.5
Percent Similarity
Class
B
A
0
1 1 1 \ 1 1 f
0.1 0.2 0.3 0.4 0.5 ,0.6 0.7
Coefficient of Community Loss
c
B
Class A .
1
1 j
' 1
0.2
0.6
1.0
Figure 1.—Confidence Intervals (95 percent) for selected compara-
tive metrics used In second stage.
Third-Stage Criteria
The third-stage analysis does not set criteria, but
rather provides a mechanism for adjusting the
scores in stages one and two. This process relies on
professional biological judgement, as well as docu-
mented evidence of conditions that can result in
atypical findings. Examples of conditions that
could trigger adjustment mechanisms are unusual
habitats, natural or human-induced disturbance of
the sample site, or known or suspected problems
with sample collection or analysis. Following are
some examples of unusual habitats:
• Lake outlets/regulated flows: Influence vari-
ables including total abundance, variables
based on relative abundance, Diptera abun-
dance, hydropsyche abundance, percent col-
lector-filterers.
149
-------
S. P. DAVIES, L TSOMIDES, D. L COURTEMANCH, and
F.DRUMMOND
• Substrate character (especially soft-bottom
habitats): Influences the biological pool avail-
able to colonize the artificial substrates, thus
affecting richness-related measures.
• Low velocity: Tend to suppress total abun-
dance and filtering activity.
• Tidal movement: Alters community composi-
tion compared to unidirectional flow, loss of
filtering activity.
• Anomalous samples: Unusual samples, par-
ticularly where disturbance is suspected (e.g.,
spates/ flow control, vandalism), may be dis-
carded.
Summary
The State of Maine has combined statistically de-
rived predictions of classification attainment with
criteria based on class-specific ecological attributes
and professional biological judgement to develop a
water quality classification attainment decision
protocol based on samples of benthic
macroinvertebrates. The protocol can be used for
sites with or without an associated high quality ref-
erence site. The result is a statistically defensible
and reproducible decisionmaking tool that also al-
lows for the exercise of professional biological
judgement.
References
Courtemanch, D.L. and S.P. Davies. 1987. Implementation of
biological standards and criteria in Maine's water clas-
sification law. EPA-905/9-89/003. Pages 4-9 in Proc.
First Natl. Workshop on Biological Criteria. U.S. Envi-
ron. Prot. Agency, Washington, DG.
Green, R.H. and G.L. Vascott. 1978. A method for the analysis
of environmental factors controlling patterns of species
composition in aquatic communities. Water Res. 12:583-
90.
Mosteller, R and J.W. Tukey. 1977. Data Analysis and Regres-
sion. Addison-Wesley, Reading, MA.
150
-------
Biological Criteria: Research and Regulation, 1991
The Relationship of Chironomidae (Diptera)
Community Structure to Chlordane Levels
in Sediments of Streams Near St. Louis,
Missouri
Christopher A. Wright
Kansas Biological Survey and Department of Entomology
University of Kansas
Lawrence, Kansas
Norman H. Crisp
U.S. Environmental Protection Agency, Region VII
Kansas City, Kansas
ABSTRACT
Chironomidae pupal exuviae were collected along tributaries of the Meramec River near St. Louis, Mis-
souri, at sites with varying levels of technical chlordane in the sediments to test the efficiency of this col-
lection technique as a method for detecting stream sites impacted with moderate to high levels of
chlordane. Exuviae were identified to lowest possible taxonomic level and analyzed relative to chlor-
dane levels found at the collection sites. Sites were placed into high and low chlordane categories and
comparisons were made between them using the Kruskal-Wallis test. Of the 18 total sites sampled, 13
were used in the analysis. Five of the sites were excluded because of extraneous factors that may have
altered the faunal composition (e.g., sewage treatment plants, other pollutants, spring influence, disrup-
tion of the site). Comparisons of mean community index values between high and low chlordane cate-
gories and mean percentage abundance of taxonomic groups, individual taxon, and functional feeding
groups between high and low chlordane categories revealed many significant differences. Overall,
numbers of taxa and pollution-sensitive taxa were reduced in the high chlordane category. The high
chlordane sites were dominated by detritus-feeding taxa (mostly Chironomini), pollution-tolerant al-
gavores (mostly Cricotopus species), and pollution-tolerant predators within the genus Procladius. It was
concluded from these results that stream sites containing sediments with high chlordane levels affect
the chironomid community structure in a predictable pattern and that the collection of pupal exuviae
was an effective technique for sampling the chironomid community. This sampling protocol could be
iised for preliminary surveys of sites suspected of chlordane contamination or long-term monitoring of
impacted sites, thus decreasing time required in the field and the lab.
Background
When adult Chironomidae (nonbiting midges)
emerge from their aquatic environment, they leave
behind a pupal exuvia that floats for a period of up
to two days and tends to accumulate with other
flotsam at catch points along the course of a stream
or river (Wiederholm, 1986). Collecting and identi-
fying chironomid pupal exuviae has shown to be
an effective method for surveying lotic systems and
for studying the impacts of organic sewage enrich-
ment and heavy metals on those systems
(Ferrington, 1987; Wilson, 1988). This project was
undertaken to examine the applicability of this
sampling methodology for detecting stream sites
contaminated with relatively high levels of chlor-
dane as compared to other local streams. If distinct
differences or predictable patterns could be found
151
-------
C. A. WRIGHT and N. H. CRISP
in the chironomid communities based on the rela-
tive chlordane levels in the sediments at the sites,
this method of biological monitoring could prove
to be useful as a preliminary survey technique for
detecting high levels of chlordane at sites or as a
long-term monitoring technique to observe recov-
ery at sites.
Chironomids are extremely valuable as a study
organism for this project because of theft high spe-
cies diversity, common abundance in most freshwa-
ter systems, and generally low mobility as mature
larvae (Ashe et al. 1987). Presently, many chlordane
monitoring programs are based on tissue analysis of
fish (carp), which are highly mobile and not contin-
ually exposed to the sediments where chlordane
tends to accumulate. Therefore, studying only carp
or other fish increases the probability of wrongly as-
sessing a site as nonimpacted. Many species of
Chironomidae spend most of their larval life bur-
rowing and feeding on sediments. Other species
may live on top of the sediments or feed on sus-
pended particles of organics (Oliver, 1971).
Chironomid larvae occupy numerous micro- hab-
itats, thus relaying increased amounts of informa-
tion about the dynamics of a freshwater system to
the researcher.
Methodology
On May 10 and 11,1988, chironomid pupal exuviae
were collected from tributaries along the Meramec
River, approximately 40 miles southwest of St.
Louis, Missouri. The collections coincided with a
sediment sampling survey undertaken by the Mis-
souri Department of Conservation to isolate possi-
ble sources of chlordane in the tributaries.
The pupal exuviae samples were collected
using a standard protocol in which each site is sam-
pled by dipping a white pan into catch point areas
along the stream edge where flotsam has accumu-
lated (Ferrington, 1987). "Dips" from all possible
catch points at a site are poured through a 125 mi-
cron sieve for 10 minutes. The debris retained in the
sieve is then washed out into the pan using 80 per-
cent ethanol and poured into a labelled jar. The
samples within the jars are sorted in the lab by
hand-picking the exuviae under a dissecting micro-
scope. Common and conspicuous exuviae may be
identified unmounted; however, rare and minute
taxa must be mounted on slides and identified
using a compound microscope. At least two repre-
sentatives of each taxon are permanently mounted
for positive identification and voucher material. All
identifications and abundances are recorded on
data sheets for future analysis.
Discussion
Correlations, cluster analysis, and Kruskal-Wallis
tests were initially applied to the data. Only the re-
sults of the Kruskal-Wallis tests are presented, as
they reflect the overall patterns found in all the
other analyses (Fig. 1).
The chironomid fauna at sites with high techni-
cal chlordane in the sediments were dominated by
pollution-tolerant taxa such as Chironomus species,
Dicrotendipes spp., Glyptotendipes sp. gp. A,
Cricotopus species, and Procladius spp. Most of the
Chironomini species are detritivores that burrow
and feed on sediments in streams. These taxa
should have been exposed to the highest levels of
chlordane. Species within the genus Procladius are
predators that usually occur on top of the sediments
in slower-moving water (Beck, 1978).
High chlordane sites also showed less omni-
vores and filterers. These two functional feeding
groups, although not. found burrowing in sedi-
ments, are still exposed to them. Omnivore is a gen-
eral category that describes taxa feeding on a
variety of foodstuffs (algae, sediments, animals)
and that tend to move between microhabitats.
Filterers may build tubes on the surface of sedi-
ments, rocks, or plants and feed on suspended or-
ganic material. Their exposure to contaminated
sediments while possibly not as high as detritivores,
may have been enough to decrease their abundance
in highly contaminated sites. Taxa within the omni-
vore and filterer groups tend to be less tolerant of
stressful conditions than taxa within the detritivore
group (Beck, 1978). • • -
Analysis of algavores revealed a large shift from
high chlordane sites to low chlordane sites. High
chlordane sites contained mostly Cricotopus species,
whereas low chlordane sites had the largest abun-
dance being among Orthocladius species. Cricotopus
species were present at low chlordane sites, but in
fewer numbers than at high chlordane sites. The
most obvious difference between high and low
chlordane sites was the presence of more algae-
feeding taxa besides Cricotapus and Orthocladius
species at low chlordane sites. These taxa, such as
.^Tvetenia cf. calvescens, Thienemanniella sp., Eu-
kiefferiella spp., Pam-metriocnemus spp., and
Rheocricotopus sp., are usually listed as occurring in
relatively "good" water quality. Algavores usually
restrict feeding to filamentous mats of algae or sin-
gle-celled algae on the surfaces of rocks, plants, and
the benthic substrate. This functional feeding group
is probably least exposed, to chlordane in the sedi-
ments; however, they may be affected more by
chlordane in the water column than other groups.
Because chlordane usually enters the stream system
152
-------
Biological Criteria: Research and Regulation, 1991
100
80
Heb
I
13
20-
Mean percentage abundance of
dominant taxa for high and low
chlordane categories
I I low chlordane
B=S3 high chlordene
R
Crlcotopuc Orthocladlus . Crlcotoput cf. Chlronomu* Orthool«dlut cf.
»PP- app-M aylvtitrla" ' ap. gp. 1 "" mailochl "*
Mean percentage abundance of
dominant taxa for high and low
chlordane categories
- 12'
10-
o
1 8"
c
E e-
decreased presence of al-
gavores may reflect recent "slugs"' of chlordane in
the water column.
Summary
Collections of Chironomidae pupal exuviae have
been shown to be useful in assessing impacts of or-
ganic sewage enrichment and heavy metals on
stream systems by past researchers. The technique
has also been advantageous over larval collections
for several reasons: (1) larval collections usually
miss very small larvae; (2) larval collections tend to
miss some of the microhabitats in which larvae
occur; (3) processing in the field and lab isi usually
much faster using the exuviae method; and (4) exu-
viae represent only one age class, whereas larvae
153
-------
C. A WRIGHT and N. H. CRISP
may be collected from many age classes, thus com-
pounding difficulties in their identification.
In this study, collections of pupal exuviae were
successful in revealing differences between stream
sites impacted with high and low levels of technical
chlordane. High chlordane sites contained less taxa
than low chlordane sites, and the taxa present at
high chlordane sites tended to be pollution-tolerant.
These results suggest that the collection of
chironomid pupal exuviae may be useful as a pre-
liminary survey tool or long-term monitoring tech-
nique in lotic systems impacted with chlordane.
Factors that could influence faunal distribution in
lotic systems besides a target pollutant, such as
chlordane, must be taken into consideration. For ex-
ample, it may be impossible to distinguish the ef-
fects of a sewage treatment plant and chlordane on
the same stream system.
At present, no indicator species can be identi-
fied to detect the presence of chlordane. As this is
the first project exploring the relationship of
Chironomidae communities to chlordane in the sed-
iments in which they occur, the results should be
analyzed for general patterns, perhaps identifying
indicator communities instead of individual taxon.
Analysis of ecological factors, such as feeding habits
of larvae, may prove to be most useful for applying
this technique to other ecoregional areas because
taxa in similar order streams may shift between
ecoregions, while the functional feeding groups
should remain stable between ecoregions.
ACKNOWLEDGMENTS: This project was funded in
part by the U.S. Environmental Protection Agency
through the National Network for Environmental Man-
agement Studies fellowship program (Project Control
Number U-913134-01-0). Work space and technical assis-
tance were provided by the Water Quality and Freshwa-
ter Ecology section of the Kansas Biological Survey at the
University of Kansas.
References
Ashe, P., D.A. Murray, and F. Reiss. 1987. The zoological dis-
tribution of Chironomidae (Insecta:Diptera). Ann.
Limnol. 23(1): 27-60.
Beck, W.M., Jr. 1977. Environmental Requirements and Pol-
lution Tolerance of Common Freshwater Chironomidae.
EPA-600/4-77-024. U.S. Environ. Protect. Agency. Cin-
cinnati, OH. , -..';
Ferrington, L.C., Jr. 1987. Collection and identification of sur-
face-floating pupal exuviae of Chironomidae for use in
studies of surface water quality. Stand. Operat. Proc.
No. FW130A. U.S. Environ. Prot. Agency, Region VII,
Kansas City, KS. ';.:
Oliver, D.R. 1971. Life history of the Chironomidae. Ann.
Rev. Ent. 16:211-30.
Wiederholm, T., scientific ed. 1986. Chironomidae of the
Holarctic region. Part 2: Pupae. Ent. Scand. Suppl. 28.
Wilson, R.S. 1988. A survey of the zinc-polluted River Nent
(Cumbria) and the East and West Allen (Northumber-
land), England, using chironomid pupal exuviae. Spixi-
ana. Suppl. 14:167-74.
154
-------
Biological Criteria: Research and Regulation, 1991
Regional Standardization of Taxonomy
Lawrence L. Lovell
Taxonomic Consultant ,
Vista, California
Ronald G. Velarde
Marine Biology Laboratory, City of San Diego
San Diego, California
ABSTRACT
The Clean Water Act requires the implementation of aquatic habitat monitoring, and numerous surveys
have been conducted to characterize those biological communities. A fundamental component of these sur-
veys is taxonomic data that list the taxa and their abundance. Typically, several surveys are conducted by
both public and private organizations within a given geographic region. A problem in realizing the full po-
tential of these data is the lack of taxonomic consistency among surveys, The Southern California Associa-
tion of Marine Invertebrate Taxqnomists (SC AMIT) was formed in 1982 to provide regional standardization
among benthic marine surveys in the southern California bight SCAMIT schedules a yearly agehda of tax-
onomic topics, including regular exchange of specimens that have been noted as inconsistently identified or
are new to science. National and regional taxonomic experts lead workshops presenting innovative identifi-
cation techniques, new taxonomiokeys, and review of voucher collections. A central voucher collection is
maintained, consisting of specimens exchanged and reviewed at meetings. The results of these meetings
and workshops are distributed in a monthly newsletter. Several aspects of biological criteria development
can benefit from regional standardization of taxonomy. The biological survey design should include a com-'
ponent for regular calibration of taxonomic data. Selection and assessment of regional reference sites
should utilize regionally standardized data. Selection of aquatic community components for detailed anal-
ysis, whether for statistical manipulation or toxicity testing, should be supported by the survey's taxonomic
data. Biological indices, commonly used as regulatory tools to manage complex environmental impact is-
sues, are dependent on quality and comparability of the underlying taxonomic data. SCAMIT's activities
have greatly enhanced taxonomic quality control and standardization within and among benthic marine
data bases in southern California. Implementation of taxonomic standardization in other regions should
serve to improve national biological criteria for surface water programs.
"What's the use of their having names," the
Gnat said, "if they won't answer to them ? "
"No use to them," said Alice: "but it's useful
to the people who name them, I suppose."
— Lewis Carroll, Through the Looking Glass
The Clean Water Act passed in 1972 requires
that treated wastewater and industrial
flows into aquatic habitats must be moni-
tored for impacts on biological communities. These
monitoring surveys collect biological data on a reg-
ular basis and typically generate large data sets
representing hundreds of species or taxa. The basic
component of a biological survey is taxonomic
data, a listing of names and abundance levels for
organisms collected. Taxonomic data are analyzed
utilizing various statistical methods. The results are
interpreted, reported, and used in regulatory
decisionmaking. Taxonomic consistency, subse-
quent analyses, interpretations, and regulatory de-
cisions cannot be made with confidence. Imple-
mentation of taxonomic standardization both
within and between regional monitoring programs
is a means of achieving and maintaining taxonomic
consistency.
155
-------
L L LOWELL and P. G. VELARDE
In heavily populated and industrialized coastal
regions such as southern California, several pro-
grams monitoring marine biological communities
are conducted by both public and private organiza-
tions, and considerable amounts of money and ef-
fort are expended. A largely unrecognized obstacle
to realizing the full potential of these data is the lack
of taxonomic consistency. Taxonomic consistency is
difficult to maintain for a variety of reasons: (1)
published or unpublished regional taxonomic liter-
ature is rarely comprehensive (if it exists at all); (2)
undescribed species are common in many habitats;
(3) standard taxonomic usage changes over time as
new research becomes available and new species
are recognized; (4) taxonomic staffs vary in skill, ex-
perience, and capabilities; (5) personnel changes
within a program lead to change—drift over time;
(6) sibling species or sexual dimorphism within a
species are common.
Statistical analyses are affected when taxonomic
inconsistencies exist in the supporting data. Com-
munity indices can be inaccurate due to overestima-
tion or underestimation of species numbers by
different taxonomists within a survey. Changes over
time in the taxonomy can lead to inaccurate com-
munity classification analyses. Tests of statistical
significance may be compromised by inconsistency
between taxonomists working on the same survey.
Comparison of data between programs becomes
difficult or impossible when taxonomists either
within or among monitoring programs are inconsis-
tent.
Other types of data can be affected by inconsis-
tent taxonomy. Chemical results from bioaccumula-
tion samples can be compromised if the animals
used are not consistently identified during collec-
tion. Inconsistently or incorrectly identified species
used in toxicity testing could lead to misinterpreta-
tion of test results. Easily confused sibling species
may respond differently to tested contaminants.
There are a number of consequences for regula-
tory decisionmaking when the analysis of the data
is affected. Test sensitivity may be reduced by taxo-
nomic inconsistency. Failure to identify organisms
consistently may lead to spurious variability in a
database, and decrease the sensitivity of a statistical
test by increasing the variance at reference (control)
stations. Evidence may be contradictory because of
taxonomic confusion. Contradictory evidence can
be introduced if the same organism is identified dif-
ferently in different samples within a single survey,
or over time. Thus an "indicator" species could be
present (or absent) depending, not on the environ-
mental conditions, but on inconsistent taxonomy.
FAlse violations may result from invalid toxicity
test results. Toxicity tests based on mixed lots of test
animals from similar, but different, species may lead
to spurious test results. These caveats might results
in the appearance of a discharge permit violation,
when none had actually occurred. Approval/denial
decisions may rest on questionable evidence. Re-
quests for environmentally safe dischargers could
be denied; or worse, requests for environmentally
safe dischargers approved, because decisions are
based on evidence from flawed taxonomic data. All
of the above types of compromised data can lead to
inappropriate action at the regulatory level, since
such actions must be based on the "best available
data and analyses."
SCAMIT was formed in 1982 by a group of ma-
rine biologists who recognized the value of region-
ally standardized data. The nonprofit organization
is supported by contributions from regulated agen-
cies, grants from industry, and annual dues.
SCAMIT's goals are "to promote the study of ma-
rine invertebrate taxonomy and develop a region-
ally standardized taxonomy." A variety of activities
help achieve these goals.
An annual agenda of monthly meetings cover-
ing taxonomic problem areas is scheduled (see
Table 1). Members regularly exchange specimens
that have been inconsistently identified or are new
to science. National and regional experts lead taxo-
nomic workshops presenting innovative identifica-
tion techniques, new keys, and review of voucher
collections. Information from meetings and work-
shops, new literature citations, species voucher
sheets, and announcements of interest are distrib-
uted to members through a monthly newsletter.
Centralized literature and specimen voucher collec-
tions are maintained and updated by SCAMIT.
Membership in SCAMIT is open to anyone, and
currently over 100 individuals and more than 25 or-
ganizations participate.
SCAMIT's activities have greatly enhanced tax-
onomic quality control and regional standardiza-
tion within and among benthic marine invertebrate
data bases in southern California. The organization
recommends the following guidelines for imple-
menting taxonomic standardization: (1) recognize
the nature and extent of the problem; (2) improve
communication between taxonomists; (3) generate
new taxonomic information; (4) establish and main-
tain centralized literature and specimen voucher
collections; and (5) support long-term maintenance
of ecological survey samples.
Several aspects of biological criteria develop-
ment can benefit from regional standardization of
taxonomy. Biological survey design should include
a component for regular calibration of taxonomic
156
-------
Biological Criteria: Research and Regulation, 1991
Table 1.—SCAMIT Agenda 1990-91.
June 11 Nassarius by Don Cadien at Cabrillo
Marine Museum, San Pedro, California
July 9 Hydrozoa by John Ljubenkov at MEG
Analytical Systems, Carlsbad, California
August 13 • Polychaete Scale Worms by Ross Duggan
at Allan Hancock Foundation, University of
Southern California
September 10 Latin Grammar for Taxonomy by John
Ljubenkov at Cabrillo Marine Museum, San
Pedro, California
October 15 • Epitoniidae by Helen DuShane at Los
Angeles County Museum of Natural History
November 19 Hesionidae by Ron Velarde at Allan
Hancock Foundation, University of
Southern California
December 10, 11 Barnard Amphipod Workshop at Los
Angeles County Museum of Natural History
January 14 Polyclad Flatworms by John Ljubenkov,
Carol Paquette, Tony Phillips at Cabrillo
Marine Museum, San Pedro, California
February 11 Lovell Taxonomic Consulting, Vista,
California
March 11 , Nuculanidae by Paul Scott at Santa
Barbara Museum of Natural History
April 8 Tharyx by Tony Phillips at Allan Hancock
Foundation, University of Southern
California
May 13, 14 ' Bryozoan Workshop with Dr. William Banta
at Cabrillo Marine Museum, San Pedro,
California •
data through quality assurance/quality control pro-
grams. Selection and assessment of regional or pro-
gram reference sites should' utilize regionally
standardized data. In some areas, reference sites
might be shared by more than one discharger. Selec-
tion of aquatic community components for statisti-
cal manipulation and detailed analysis should be
discussed with taxonomists to avoid selection of
problem species. Taxonomic data should support
the selection of endemic species for toxicity testing.
Biological indices, commonly used as regulatory
tools to manage complex environmental impact is-
sues, are dependent upon the quality and compara-
bility of the underlying taxonomic data.
Implementation of taxonomic standardization in
other regions should serve to improve national bio-
logical criteria for surface waters programs.
"If one does not know the names, one's
knoivledge of things is useless."
— Isidorus
ACKNOWLEDGMENTS: The authors would like to
express their appreciation to Donald B. Cadien, Thomas
Parker, the County Sanitation Districts of Los Angeles
County, the City of San Diego, and MEC Analytical Sys-
tems for their help and support in producing this paper
and associated poster presentation. This paper is publica-
tion number 5 as contributed by the Southern California
Association of Marine Invertebrate Taxonomists.
157
-------
Integrated Chemical and Biological
Monitoring of Sun Creek, McPhearson
County, Kansas, U.S.A.
N.H. Crisp
L.C. Ferrington
L. Cowles
U.S. Environmental Protection Agency, Region VII
Kansas City, Kansas and • , ' '
University of Kansas
Lawrence, Kansas "''""',
Water quality in a portion of the Sun Creek catchment, a stream system which receives the efflu-
ents from a municipal wastewater treatment facility and an oil refinery, was evaluated based on
collection of surface floating Chirohomid pupal exuviae and water chemistry. Utilizing bluster
analyses as the initial basis for evaluation/both pupal exuviae and chemical data provided supportive infor-
mation. Cluster analysis of the Chironomid data identified tolerant taxa, facilitative taxa and intolerant taxa.
For the chemical parameters the analyses identified a group of parameters associated with organic enrich-
ment and a group of parameters which define ambient water quality. Both chemical data and Chironomid
data classified sites similarly, however, Chironomid data provided somewhat better resolution of differences
between sites. Correlation analyses between the chemical data and the Chironomid data revealed that the
taxa identified as tolerant were positively correlated with the enrichment parameter while those identified
as intolerant were negatively correlated with the enrichment parameter. The simultaneous collection of both
chemical and biological data provided complementary information which water quality managers could use
to make decisions and plan options/Each element provided a slightly different perspective which, when
taken together, clearly defines water quality problems and processes.
If you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of'this document,
158
-------
Biological Criteria: Research and Regulation, 1991
The Use of Biocriteria in the Ohk) EPA
Biological Monitoring and Assessment
Program • . •-"-. ?'**•? ^::-<^;,,:- .v^^i-
Chris O.Yoder
Ohio Environmental Protection Agency "•'•••<''
Columbus, Ohio .
Ohio EPA has operated a program of biological surveys since the late 1970s. Their initial.purpose
was to provide an integrated set of biological and chemical data for use in monitoring and report-
ing activities and the water quality standards (WQS) program. An outgrowth of this effort has
been the development of biological criteria ('biocriteria") as an ambient aquatic life use goal assessment tool.
Biocriteria were recently adopted (February 1990), as a part of .the Ohio WQS regulations. Concepts impor-
tant to this approach include, a practical definition of biological integrity, the role of ecoregions, the regional
reference site approach, and recognizing the characteristics inherent to chemical assessment ("bottom up"
approach) and biocriteria ("top down" orientation). These are important .concepts-, in the development and
application of biocriteria. Initiating and implementing a biocriteria progranx requires that several initial deci-
sions be made. These include how to incorporate biocriteria into the existing structure of the iWQ.S regula-
tions, selection of appropriate prganism groups, selection of evaluation topis, selection of reference sites,, and
regionalization considerations. All of these affect how well biocriteria work in a state water quality manage-
ment program.
Biological field sampling,procedures are also summarized with cost,and resource requirements. The
Index of Biotic Integrity (IBI), modified for application in Ohio, and the Invertebrate Community Index (ICI)
are two of the principal evaluation tools used by Ohio EPA. The derivation, calibration, and variability of
each is described. Current program uses of biocriteria include water quality standards (use designations, use
attainability analysis), NPDES permitting, State Revolving Loan 'Fund, basic monitoring/reporting (e.g
305b report), nonpoint source assessment, enforcement/litigation, 404/401 dredge and fill issues, and
CSO/stormwater management. An emerging area of use is with Natural Resource Damage Assessments. Ex-
amples of biocriteria application are illustrated and include stream specific assessment, trend reporting and
assessment, and providing information about rare and endangered species.
If you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of'this document. '"•••- "'•
159
-------
Water Quality Indicators for Rivers and
Streams: Selection, Stratification and
Aggregation for Decisionmaking
J. Harrison
U.S. Environmental Protection Agency, Region IV ,
Atlanta, Georgia
Current state and national assessments of water quality status and trends (the state and national re-
ports pursuant to Section 305(b) of the Clean Water Act are perceived by many to be less than ade-
quate for important management and information purposes. This stems from use of inappropriate
measures of status and trends that are inconsistently reported and summarized. Consistent stratification of
meaningful indicators of water quality status coupled with standardized methods to aggregate assessment
information can greatly enhance the utility of our reporting mechanisms. An aggregation method is devel-
oped using four important indicators pertinent to water quality problems in rivers and streams: (1) severity,
(2) extent, (3) trend, and (4) recovery potential. .
A consistent scheme for reporting problem severity is presented allowing use of ecological (in-stream bi-
ological survey) data, chemical water column data, toxicity testing information, or risk estimates for con-
sumption of contaminated fish or water supplies. Problem extent for rivers and streams incorporates both
flow and segment length. Each indicator is stratified with values corresponding to each level. The logic for
the stratification scheme and values is described. Options for formal development of values are discussed.
The aggregation method (with initial straw values developed by the author) is applied to the Deep River
system (North Carolina) demonstrating its utility for objective quantification of water quality status, for
measuring water quality improvement over time and for illustrating control program effectiveness. Conclu-
sions and potential applications are discussed. These include: (1) coherent summarization of status and
trends for a wide range of geographic levels of resolution, (2) better targeting of priority problems, and
(3) easier evaluation of effectiveness of control programs.
If you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
160
-------
Biological Criteria: Research and Regulation,, 1991
Aquatic Macroinvertebrates as Biological
Indicators of Water Pollution in Arizona
J.F. Boggs, C. Olson
M. Lowry, M. Longsworth
R. Williams, J. Clayton
E. Swanson, F. Woodwick
Arizona Department of Environmental Quality
Tucson, Arizona
The Arizona Department of Environmental Quality, in cooperation with the University of Arizona, is
reviewing a variety of biological assessment methods to enhance Arizona's surface water quality
monitoring program. Very little data regarding water quality and its relationship to aquatic
macroinyertebrate communities in arid regions are available. This information will be necessary for future
management of Arizona's surface water resources. A semi-quantitative approach of macroinvertebrate sam-
pling (U.S. EPA Rapid Bioassessment Protocols) is currently being tested at 6 sites on the Verde and Santa
Cruz Rivers. Physical, chemical and biological data have been collected during spring and summer 1990 to
characterize conditions in streams that are either minimally or moderately impacted by anthropogenic activ-
ities. Species of macroinvertebrates collected during spring represented 8 classes (Insecta, Crustacea, Gas-
tropoda, Bivalvia, Oligochaeta, Hirudinea, Turbellaria) including 17 orders of aquatic macroinvertebrates.
Aquatic insects (immature and mature forms) dominate these communities with representatives of 29 fami-
lies. Aquatic communities from Verde River collection sites appeared similar whereas those sampled from
the Santa Cruz River demonstrated lower diversities (family level) in the effluent dominated sites than the
control site. Species typically tolerant of low dissolved oxygen and mediocre habitat (Chironorninae) domi-
nated the fauna at effluent dominated sites. The data, while providing valuable aquatic community ecology
information, will be useful for the implementation of programs consistent with U.S. EPA guidelines for bio-
logical standards and monitoring mandated by the 1987 Clean Water Act.
// you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
161
-------
Development of Diagnostic Procedures
to Evaluate Aquatic Resources In
Regional Watersheds
John W. Arthur
U.S. Environmental Protection Agency
Environmental Research Laboratory
Ditluth, Minnesota
The U.S. EPA laboratory at Duluth, Minn, has been participating in regional water quality studies
within the states of Illinois, Minnesota, and Michigan. Locations are in the Upper Illinois, Minnesota
and Saginaw river basins. The objectives are to develop diagnostic procedures to identify impacts
and assist the sponsoring agency in the integration of this information into a regional goal setting process.
Our overall project goal with these regional studies is to determine biological and hydrological linkages that
transcend geographical boundaries and serve as guidelines for defining watershed health. The example pre-
sented is the Minnesota River study being coordinated by the Minnesota Pollution Control Agency and con-
ducted by a multiagency task force composed of state, university, and federal participants. The general
approach is to evaluate a variety of procedures to define watershed health. The physical procedures measure
habitat quality, and the chemical and laboratory toxicity tests measure the quality of the ambient surface and
sediments. The biosurveys (with fish and macroinvertebrates) define the present status of instream biota
and reveal the severity of degradation. All procedures assist in defining stressors and. aid in the categoriza-
tion of priority reaches needed for further control measures by regulatory authorities.
// you would like further details on this subject matter, please feel
free to contact the participant; addresses can be found in the Atten-
dees List starting on page 163 of this document.
162
-------
Biological Criteria: Research and Regulation, 1991
Biological Criteria: Research and Regtilatioh
Sponsored by the
Office of Water
U.S. Environmental Protection Agency
December 12r-13,1990 • Hyatt Regency Crystal City • Arlington, Virginia
ATTENDEES LIST
P. David Allen, II
U.S. Environmental Protection Agency,
Region 5
5WQS-TUB8, 230 S. Dearborn Street
Chicago, IL 60604
Howard Alexander
Dow Chemical Company
2030 Dow Center
Midland, Ml 48674
Lisa Almodouar
U.S. Environmental Protection Agency
401 M Street, S.W. (OW/CSD)
Washington, D.C. 20460
Dennis Anderson
Colorado Department of Health
4210 E. 11th Avenue
Denver, CO 80220
Terry Anderson
Supervisor, Standards & Specification
Section
Kentucky Division of Water
18Reilly Road
Frankfort, KY 40601
Paul Angermeier
Virginia Cooperative Fish and Wildlife
Research Unit
Virginia Polytechnic Institute and State
University
Blacksburg, VA 24061
John Arthur
Environmental Research
Laboratory/ORD/Duluth
6201 Congdon Boulevard
Duluth, MN 55804
David Bailey
Potomac Electric Power Company
(PEPCO)
1900 Pennsylvania Avenue, N.W.
Suite 4100
Washington, D.C. 20068
Michael Barbour
Sr. Scientist
EA Engineering Science & Technology,
Inc. '
15 Loveton Circle
Sparks, MD 21152 _
Carole Ann Earth
Alliance for Chesapeake Bay - CRIS
6110 Executive Blvd.
Rockville, MD 20852
John Bender
Nebraska Department of
Environmental Control
P.O. Box 98922
Lincoln, NE 68509-8922
Brock Bernstein
ECO Analysis
221E Matilija Street, Suite A
Ojai, CA 93023
Robert Berger
East Bay Municipal Utilities District
P.O. Box 24055
Oakland, CA 94623
Albert Boulanger
Computer Scientist
BBN ,
, 'ip Moulton Street
Cambridge, MA 02138
Tim Brush
RMC Environmental Services Inc.
1921 River Road
P.O. Box 10
Drewmore, PA 17518
Daniel Boward
Maryland Department of the
Environment
2500 Broening Highway
Baltimore, MD21224
163
Stephanie Braden
Louisiana Department of ,
Environmental Quality ,.,
Environmental Quality Specialist
625 N. Fourth Street -- '
P.O. Box 44091
Baton, Rouge, LA 70804,
Bob Brody ^
St. Johns River Water Management
District .,.--.
P.O. Box 1429 ,
Palatka, FL 32178
Robert P. Brooks
Forest Resources Laboratory
Pennsylvania State University
University Park, PA 16802
Melvin S. Brown
Senior Environmental Scientist
Law Environmental, Inc.
112 Town Park Drive
Kennesaw, GA30144
Steven Brown
Manager, Biological Services
SMC Environmental Services Group
P.O. Box 859
Valley Forge, PA 19001
Bob Burm
U.S. Environmental Protection Agency,
Region 8
999 18th Street, Suite 500
Denver, CO 80202
Robert Byren
Wildlife Management Institute
1101 14th Street, N.W.
Suite 725
Washington, D.C. 20005
Clalr Buchanan
Aquatic Ecologist
Interstate Commission on the Potomac
River Basin
6110 Executive Blvd. Suite 300
Rockville, MD 20852
-------
Con ferenca Attendee List
Doug Burnham
Aquatic Biologist
Vermont Department of Environmental
Conservation
103 S. Main Street-1 ON
Waterbury, VT 05676
Tim Burton
Hydrologlst/Fishery Biologist
Idaho Division of Environmental Quality
1410 N. Hilton
Boise, ID 83706-1253
Robert Campalgne
Director Engineering and
Environmental Affairs
The Upjohn Company
410 Sackett Point Road
North Haven, CT 06473
Anthony Carlson
U.S. Environmental Research
Laboratory/Duluth
6201 Congdon Boulevard
Duluth, MN 55804
Catherine CekoIIn
Office of Marine and Estuarine
Protection
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460
Cynthia Chritton
Environmental Quality Specialist -
Water Quality Standards
Louisiana Department of
Environmental Quality
625 N. Fourth Street
P.O. Box44091
Baton Rouge, LA 70804
David L. Clough
Director, Water Quality Division
Vermont Department of Environmental
Conservation
103 South Main Street
Waterbury, VT 05676
Steve Constable
Dupont Engineering
P.O. Box 6090
Louviers 33E34
Newark, DE19711-6090
Robert W. Cooner
Chief
Special Studies Section
1751 Federal Drive
Montgomery, AL36130
David Courtemanch
Maine Department of Environmental
Protection
State House #17
Augusta, ME 04333
Clayton Creager
Principal/Senior Scientist
Western Aquatics, Inc.
Executive Park - Suite 220
1920 Highway 540
Durham, NC 27713
William Creal
Michigan DNR
P.O. Box 30028
Lansing, Ml 48909
Norman Crisp
Environmental Services Division
U.S. Environmental Protection Agency,
Region 7
25 Funston Road
Kansas City, KA 66115
Stephen Crowley
Program Director, Wetlands and Water
Resources
Vermont Natural Resources Council
9 Bailey Avenue
Montpelier, VT 05602
Kenneth Cummins
University of Pittsburgh
Biological Science Department
Pittsburgh, PA 15260
James Cummins
Associate Director of Living Resources
Interstate Commission on the Potomac
River Basin
Suite 300
6110 Executive Blvd.
Rockville, MD 20852
Elleanore Daub
Virginia Water Control Board
2111 N. Hamilton Street
P.O. Box11143
Richmond, VA 23230
Wayne Davis
U.S. Environmental Protection Agency,
Region 5
536 S. Clark Street, 10th Floor
Chicago, Illinois 60605
Michele Dionne
Virginia Polytechnic Institute and State
University
Post-Doctoral Associate
Biology Department
Blacksburg, VA 24061-0406
George E. Dissmeyer
S&PF Water Program Manager
U.S.D.A. Forest Service
1720 Peachtree Road, N.W.
Atlanta, Georgia 30367
Randy Dodd
Research Triangle Institute
P.O. Box12194
Research Triangle Park, NC 27709
Daniel Drozdowski
U.S. Testing Company, Inc.
1415 Park Avenue
Hoboken, NJ 07030
Dan Dudley
Ohio EPA
1030 King Avenue
Columbus, OH 43212
Bill Ettinger
RMC Environmental Services Inc.
1921 River Road
P.O. Box 10
Drewmore, PA 17518
Brigitte Farren
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20406
Chris Faulkner
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460
Robbin Finch
Boise City Public Works Department
150 N. Capitol Blvd., P.O. Box 500
Boise, ID 83701-0500
William Fowler
Forest Hydrologist
U.S. Forest Service
P.O. Box1008
Russellville, AR 72801
Toby Frevert
Illinois Environmental Protection
Agency
Manager Planning Section
2200 Churchill Road
Springfield, IL 62794
Russell Frydenborg
Biologist
Florida Department of Environmental
Regulation
2600 Blairstone Road
Tallahassee, FL 32399-2400
Rod Fujilta
Environmental Defense Fund
257 Park Avenue South
New York, NY 10010
John Gannon
US Rsh and Wildlife Service
National Fisheries Wildlife Research
Center
1451 Green Road
Ann Arbor, Ml 58105
Robin Gariby
The Advent Group
P.O. Box 1147
Brentwood, TN 37024-1147
164
-------
Biological Criteria: Research and Regulation, 1991
Allan Gaulke
American Electric Power
1 Riverside Plaza
Columbus, Ohio 43215
Reid Garrett
Carolina Power & Light
The Harris Energy & Env. Center
Rt. 1,Box327
New Hill, N.C. 27562
George Gibson
U.S. Environmental Protection Agency
Health and Ecological Criteria Division
401 M Street, S.W.
Washington, D.C. 20460
Steve Glomb
US EPA, Office of Marine and
Estuarine Protection, WH 556F
401 M Street, SW
Washington, D.C. 20460
Tom Gieason
U.S. Environmental Protection Agency
Office of Research and Development
RD-689
401 M Street, S.W.
Washington, D.C. 20460
James Green
U.S. Environmental Protection Agency,
Region 3
303 Methodist Building
Wheeling, WV 26003
William Gregonis
MCMD, Merck & Co., Inc.
P.O. Box 600
Danville, PA 17821
Tom Grovhoug
Larry Walker Assoc.
509 A 4th Street
Davis, CA 95616
Sharon Gross
Tetra Tech, Inc.
10306 Eaton Place, Suite 340
Fairfax, VA 22030
George Guillen
Texas Water Commission
5144 E. Sam Houston Parkway North
Houston, TX 77015
Richard Hafele
Supervisor, Biomonitoring Section
Oregon Department of Environmental
Quality
1712 S.W. 11th
Portland, OR 97201
Cindy Hagley
ASCI
6201 Congdon Boulevard
Duluth, MN 55804
LeAnne Hamilton
Project Engineer
County Sanitation District of Los
Angeles County
P.O. Box 4998
Whittier, CA 90607
Jim Harrison
U.S. Environmental Protection Agency,
Region 4
345 Courtland Street
Atlanta, GA 30365
David Hart
Academy of Natural Sciences
19th and the Parkway
Philadelphia, PA 19103
Margarete Heber
Biologist
U.S. Environmental Protection Agency
(EC-338)
401 M Street, SW
Washington, D.C. 20460
Mark Hicks
Water Quality Standards Coordinator
Washington State Department of
Ecology
Water Quality Program, Mail Stop
PV-11
Olympia, WA 98504-8711
Gary Hickman
Biologist
Tennessee Valley Authority
Water Research Division
Torestry Building
Norris, TN 37828-2017
Lance Himmelberger
PA Department of Environmental
Resources
P.O. Box 2063
Harrisburg, PA17105
Lawrence Hightower
Professor
The University of Connecticut
Dept. of Molecular & Cell Biology
Storrs, CT 06269-3044
Linda Hoist
Regional Water Quality Standards
Coordinator
U.S. Environmental Protection Agency,
Region 3
841 Chestnut Building
Philadelphia, PA 19107
A. Frederick Holland
Vice President
Versar, Inc.
EMS Operations
9200 Rumsey Road
Columbia, MD 21045-1934
Evan Hornig
Biologist
U.S. Environmental Protection Agency,
Region 6
1445 Ross Avenue (6E-SA)
Dallas, TX 75202
Frank Horvath
Michigan Dept. Natural Resources
Land & Water Management Division
P.O. Box 30028
Lansing, Ml 48909
John Houlihan
U.S. Environmental Protection Agency
726 Minnesota Avenue
Kansas City, KS 66101
Hoke Howard
Environmental Services Division
Aquatic Biologist
U.S. Environmental Protection Agency
College Station Road
Athens, GA 30613
Joseph Hudek
Ambient Monitoring Section
Acting Chief
U.S. Environmental Protection Agency,
Region 2
2890 Woodbridge Avenue
Building 209, MS-220
Edison, NJ 08837
Bob Hughes
NSI Technical Services
200 SW 25th Street
Corvallis, OR 97333
William Hunley
Hampton Roads Sanitation District
P.O. Box 5000
Virginia Beach, VA 23455
John Jackson
Unified Sewage Agency of Washington
County
155 North 1st
Hillsboro, Oregon 97124
Roman Jesien
Horn Point Environmental Lab
University of Maryland
P.O. Box 775
Cambridge, Md 21613
Peter Johnson
Graduate Student
University of Maryland at Horn Point
P.O. Box 775
Cambridge, MD 21613
David Jordahl
Maryland Department of Natural
Resources
580 Taylor Avenue
Annapolis, Maryland 21401
165
-------
Conference Attendee List
Stephen Jordan
Maryland Department of Natural
Resources
Tidewater Administration
B-3 Tawes State Office Building
580 Taylor Avenue
Annapolis, MD 21401
Michael Kadlec
Water Quality Technician
St. Regis Mohawk Tribe
Community Building
Hogansburg, NY 13655
James Karr
Harold H. Bailey Professor of Biology
Virginia Polytechnic Institute and State
University
Department of Biology
Blacksburg, VA 24061 -0406
Anne Keller
Environmental Scientist
Tennessee Valley Authority
HB 25 270C-C
311 Broad Street
Chattanooga, TN 37402-2801
James Kennedy
University of North Texas
Biology Department
Denton, Texas 76203
BHIIe Kerans
Post-Doctoral Associate
Virginia Polytechnic Institute and State
University
Department of Biology - Derring Hall
Blacksburg, VA 24061 -0406
Steve Kllpatrlck
Manager Water Issues
Dow Chemical Company
2030 Dow Center
Midland, MI 48674
Warren Klmball
Environmental Engineer
Massachusetts Division of Water
Pollution Control
Lyman School
Route 9
Westboro, MA01581
Lionel! Klikoff
Arizona Department of Environmental
Quality
20005 North Central
Phoenix, AZ 85004
(602)257-2270
Edward Krueger
Potomac Electric Power Company
(PEPCO)
1900 Pennsylvania Avenue, N.W.
Suite 4100
Washington, D.C. 20068
Ernest C. Ladd
Environmental Resources
Management, Inc. •
2125 Pier Drive
Ruskin, Florida 33570
Phil Larsen
U.S. Environmental Protection Agency
200 S.W. 35th Street
Corvallis, Oregon 97333
Willie Lane
U.S. Environmental Protection Agency,
Region 6 (6E-5A)
1445 Ross Avenue
Dallas, Texas 75202
James Lazorchak
U.S. Environmental Protection Agency
Environmental Monitoring Systems Lab
3411 Church Street
Cincinnati, Ohio 45244
Brian Lee
Air Water Pollution Report
951 Pershing Drive
Silver Spring, Maryland 20910-4464
Stewart Lehman
Maryland Department of Natural
Resources
580 Taylor Avenue
Annapolis, MD 21401., .-
Dave Lenat
NC DEHNR
P.O. Box 27687
Raleigh, NC 27611
Paul Leonard
EA Engineering, Science, and
Technology Inc.
15 Loveton Circle
Sparks, MD 21152
J. van Leawen
Ministry of Housing Physical Planning
and Environment
P.O. Box 450 : •
2260 MB Leidschendam
The Netherlands .
Gordon Linam
Aquatic Biologist . ,.
Texas Parks and Wildlife Department,
P. O. Box 947
San Marcos, TX 78667
James Loar
Environmental Sciences Division
Oak Ridge National Lab
P.O. Box 2008
Oak Ridge, TN 37831 -6036
Catherine Long
U.S. Environmental Protection Agency
401 M Street, S.W. (PM-221)
Washington, D.C. 20460
166
Lawrence Lovell
The Southern California Association of
Marine Invertebrate Taxonomists
1036 Buena vista Drive
Vista, CA92083
Kenneth Lubinski
Environmental Management Technical
Center
U.S. Fish and Wildlife Service
575 Lester Avenue
Onalaska, Wisconsin 55650
Anthony Maclorowski
Battelle
505 King Avenue
Columbus, Ohio 43201
John Magnuson
Academic Program Director
Center for Limnology
680 North Park Street
University of Wisconsin
Madison, Wl 53706
Sally Marquis
U.S. Environmental Protection Agency
1200 Sixth Avenue
Seattle, WA 98101
Suzanne Marcy
U.S. Environmental Protection Agency
Criteria and Standards Division
401 M Street, S.W.
Washington, D.C. 20460
Scott D. Matchett
Hunton & Willjams
P.O. Box1535
Richmond, VA 23212
John Maxted
Delaware Dept. of Natural Res. and
Env. Control
89 Kings Highway
P.O. Box1401
Dover, Delaware 19903
Roland McDaniel
FTN Associates
Number 3 Innwood Circle
Suite 220
Little Rock, Arkansas 72211
Beth McGee
Aquatic Biologist
Maryland Dept. of the Environment
2500 Broening Highway
Baltimore, MD 21224
James E. Mclndoe
Alabama Dept. of Env. Management
Planning/Projects Branch
1751 Federal Drive
Montgomery, AL36130
Ossi Meyn
U.S. Environmental Protection Agency
Office of Toxic Substances
P.O. 60X16090
Arlington, Virginia 22215
-------
Biological Criteria: Research and Regulation, 1991
David Vana-Miller
Water Quality Monitoring Coordinator
U.S. Environmental Protection Agency,
Region 8
P.O. Box 25366
Lakewood, CO 80225
John Miller
U.S. Environmental Protection Agency,
Region 5
536 S. Clark Street
Chicago, IL 60605
Michael Mills
Environmental Biologist Chief
Kentucky Division of Water
18ReillyRoad
Frankfort, KY 40601
Reid Miner
Program Director
NCASI
260 Madison Avenue
New York, NY 10016
William Morrow
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460
Doyle Mosier
Lower Colorado River Authority
P.O. Box 220
Austin, TX 78767
Regina Mulcahy
Environmental Scientist
U.S. Environmental Protection Agency,
Region 2
2890 Woodbridge Avenue, Building 209
Edison, NJ 0883-3649
Mark Munn
E.V.S. Consultants
2517 Eastlake Avenue, East
Seattle, WA 98102
Deirdre Murphy
Head, Water Quality Toxics Section
Maryland Department of Environment
2500 Broening HGWY
Baltimore, MD 21224
Arleen Navarret
City & County of San Francisco
Department of Public Works
750 Phelps Street
San Francisco, CA94124
Bruce Newton
Assessment and Watershed Protection
Division .
U.S. Environmental Protection Agency
401 M. Street, S.W. (WH 553)
Washington, D.C. 20460
Steven Newhouse
Indiana Department of Environmental
Management
5500 W. Bradbury Avenue
Indianapolis, IN 46241
Tim Oakes
RMC Environmental Services Inc.
1921 River Road
P.O. Box 10
Drewmore, PA 17518
Grace Ordaz
Public Health Engineer
Maryland Department of the
Environment
2500 Broening Highway
Baltimore, MD21224
Cheryl Overstreet
Water Quality Standards Coordinator
U.S. Environmental Protection Agency,
Regipn 6
1445 Ross Avenue
Dallas, TX 75202
Jimmie Overton
NC DEHNR
PO Box 27687
Raleigh, NC 27611
Jack Paar
U.S. Environmental Protection Agency,
Region 1
60 Westview Street
Lexington, MA 02173
Loys Parrish
U.S. Environmental Protection Agency,
Region 8
P.O. Box25366
Denver Federal Center
Denver, CO 80225
Ron Pasch
Projects Manager
Tennessee Valley Authority
Water Research Division
HB 25 270C-C
311 Broad Street
Chattanooga, TN 37402-2801
Dean Pasko
City of San Diego, Ocean Monitoring
Program
4077 North Narbor Drive, MS-45A
San Diego, CA 92101
Steve Pawlowski
Rule Development Specialist
Arizona Department of Environmental
Quality
2005 N. Central Ave.
Phoenix, AZ 85004
Jim Pendergast
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460
Clayton Penniman
Narraganset Bay Project
291 Promenade Street
Providence, Rl 02908
167
David Penrose
North Carolina DEHNR
P.O. Box 27687
Raleigh, NC 27611
Patrick Pergola
U.S. Environmental Protection Agency,
Region 2
26 Federal Plaza
Room: 813 - 2WMD-WSP
New York, New York 10278
Mark Pifher
Anderson, Johnson & Gianunzio
104 South Cascade, Suite 204
Colorado Springs, CO 80903
JayPitkin
Manager, Engineering & Water Quality
Management
Utah Bureau of Water Pollution Control
P.O. Box 16690
Salt Lake City, UT 84116
Ernest Pizzuto, Jr.
Senior Environmental Analyst
Connecticut Department of
Environmental Protection
Water Management Bureau
122 Washington Street
Hartford, CT 06106
Donald Porteous
60 West View Street
Lexington, MA 02173 '
Ken Potts
U.S. Environmental Protection Agency
401 M Street, S.W. (WH-585)
Washington, D.C. 20460
James Pratt
Assistant Professor, Aquatic Ecology
The Pennsylvania State University
213 Ferguson Building
University Park, PA 16802
Ronald Preston
U.S. Environmental Protection Agency,
Regions
303 Methodist Bldg, 11th and Chapline
Streets .
Wheeling, WV 26003
Niles Primrose
Biologist Natural Resources IV
Maryland Department of the
Environment
416 Chinquapin Round Road
Annapolis, MD 21401
Edward Rankin
Water Quality Scientist .
Ohio.EPA, Division Water Quality
Planning & Assess.,
Ecological Assessment Section
1030 King Avenue
Columbus, OH 43212
-------
Conference Attendee List
Frederic Reid
Ducks Unlimited
9823 Old Winery Place, Suite 16
Sacramento, CA 95827
Chris Reiter
SOCMA
1330 Connecticut Avenue, N.W.
Suite 300
Washington, D.C. 20036
Doreen Robb
U.S. Environmental Protection Agency
Office of Wetlands A-104F
401 M Street, S.W.
Washington, D.C. 20460
Glenn Rodriguez
U.S. Environmental Protection Agency,
Region 8
P.O. BOX25366
Denver Federal Center
Denver, CO 80225
Peter Ruffier
Director of Technical Services
Association of Metropolitan Sewerage
Agencies
1000 Connecticut Avenue, NW, Suite
1006
Washington, D.C. 20036
Keith Sappington
Environmental Specialist
Maryland Dept. of the Environment
2500 Broering HWY
Baltimore, MD 21224
Louis Scarano
ENVIRON Corp.
4350 N. Fairfax Drive
Arlington, VA 22203
Chris Schlekat
Aquatic Biologist
Maryland Dept. of the Environment
2500 Broening Highway
Baltimore, MD 21224
Jack Schultz
Professor
The University of Connecticut
75 North Eagleville Road
Ecology & Evolutionary Biology
Storrs.CT 06269-3042
Sonja Schuyler
Johnson Company
5 State Street
Montpelier, VT 05602
Walter Schoepf
U.S. Environmental Protection Agency
ERRD
Pre-Remedial Technical Support
Section
Room: 13-100
26 Federal Plaza
New York, NY 10278
Duane Schuettpelz
Planning Analyst
Wisconsin Dept. of Natural Resources
P.O. Box 7921
Madison, Wl 53707
Dick Schwer
El Dupont Co.
P.O. Box 6090
Newarek, Delaware 19794
Lawrence Shepard
Life Scientist
U.S. Environmental Protection Agency,
Region 5
230 S. Dearborn
Chicago, IL 60604
Russell Sherer
Director, Div. Water Quality
Assessment and Enforcement
S.C. Dept. Health and Environmental
Control
2600 Bull Street
Columbia, SC 29201
Debbie Smith
California Water Quality Control Board
Los Angeles Region
101 Centre Plaza Drive
Monterey Park, CA 91754
Jerry Smrchek
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460
Mark Southerland
Dynamac Corporation
11140RockvillePike
Rockville. MD 20852
Richard Spear
U.S. Environmental Protection Agency
2890 Woodbridge Avenue, Bldg. 209
Edison, NJ 08837
Mark Sprenger
U.S. Environmental Protection Agency
2890 Woodbridge Avenue, Bldg. 18
Edison, NJ 08837-3679
Charles S. Steiner
U.S. Environmental Protection Agency,
Region 5
536 S. Clark Street
Chicago, IL 60605
Gordon Stuart
USDA Forest Service - CF '
P.O. Box 96090
Washington, D.C. 20090-6090
Susan Swenson
U.S. Environmental Protection Agency,
Superfund Region 6
Mail 6H-SR
Dallas, Texas 75202
168
Diane Switzer
U.S. Environmental Protection Agency,
Region 1
60 Westview Street
Lexington, MA 02173 .
Mary Teves
Env. Planning Office
Hawaii Department of Health
5 Waterfront Plaza
Suite 250
Honolulu, Hawaii 96813
Ronald Thorn
Battelle
Marine Research Laboratory
439 W. Sequim Bay Road
Sequim, Washington 98382-9099
Roger Thoma
Ohio Environmental Protection Agency
1030 King Avenue
Columbus, Ohio 43212 ,
William Turetsky
GAP Chemical Corp.
1361 Alps Road
Wayne, NJ 07470
Stephen Twidwell
Environmental Quality Specialist
Texas Water Commission
1700 North Congress Avenue
P.O. Box13087
Capitol Station
Austin, TX 78711
Fritz Wagener
U.S. Environmental Protection Agency,
Region 4
345 Courtland Street
Atlanta, GA 30365
Robert Ware
Manager, Water Quality Branch
Kentucky Division of Water
ISReillyRoad
Frankfort, KY 40601
Diane Wehner
NOAA
U.S. Environmental Protection Agency,
Region 3 (3HW02)
841 Chestnut Building
Philadelphia, PA 19107
David Wefring
Senior Environmental Technical
Specialist
International Paper
6400 Poplar Avenue
Memphis, TN 38197
Molly Whitworth
Senior Staff Ecologist
U.S. Environmental Protection Agency
Office of Policy Analysis/OPPE
401 M Street, S.W.
Washington, D.C. 20460
-------
Biological Criteria: Research and Regulation, 1991
Wendy Whiltse
U.S. Environmental Protection Agency,
Region 9
75 Hawthorne Street (W-3-1)
San Francisco, California 94105
Melissa Wieland
Baltimore Gas & Electric
1000 Brandon Shores Road
Baltimore, MD 21226
Joe Winfield
Applied Marine Research Laboratory
Old Dominion University
Norfolk, VA 23529-0456
Susan Woods
Environmental Analyst
NEIWPCC
85 Merrimac Street
Boston, MA 01879
Forest Woodwick
Manager Water Quality Standards Unit
AZ Dept. of Env. Qual.
2655 E. Magnolia
Phoenix, AZ 85034
Jack Word
Battelle
Marine Research Laboratories
439 W. Sequim Bay Road
Sequim, WA 98382-9099
ChiehWu
U.S. Environmental Protection Agency
ORD/RD682 "
401 M Street, S.W. •
Washington, D.C. 20460
Bill Wuerthele
U.S. Environmental Protection Agency,
Regions •'•.'''
999 18th Street, 8WM-SP
Denver, CO 80202
Chris Yoder
Environmental Manager
Ohio EPA - Division of Water Quality
Planning & Assessment
1800 Water Mark Drive
Columbus, OH 43266-0149
Carl Young
U.S. Environmental Protection Agency,
Region 6
1445 Ross Avenue
Dallas, TX 75214
Edward Younglner
Manager, Water Quality Monitoring
Section
S.C. Dept. Health and Environmental
Control
2600 Bull Street .
Columbia, SC 29201
Deborah Zmarzly
Marine Biologist
City of San Diego, Ocean Monitoring
Program
4077 North Narbor Drive, MS-45A
San Diego, CA 92101
169
-------
-------
INDEX
Anderson, Kenneth 123
Arnold, Ray 61
Arthur, John W 162
Barbour, Michael T. 25
Bernstein, Brock B 104
Bode, Robert W 139
Boggs, J.F. 161
Boulanger, Albert 123
Brooks, Robert P. 81
Burnham, Doug 135
Burton, Timothy A 142
Butler, Waiter L 131
Clark, William H 142
Clayton, J 161
Courtemanch, David L 147
Cowles, L 158
Crisp, N.H. 158
Croonquist, Mary Jo 81
Cummins, Kenneth W. 3
D'Silva, Elizabeth T. 81
Davies, Susan P. 147
Dickson, Kenneth L 61
Dionne, Michele .62
Dudley, Daniel R 15
Ferrington, L.C. 158
Fiske, Steve . 135
Friedman, Ellen S .131
Gallagher, Joseph E 81
Gannon, John E. 46
Gish, Herbert . . 123
Harrison, J 160
Harvey, Geoffrey W. 142
Hightower, Lawrence E 129
Hoist, Linda 13
Jordan, Stephen J 73
Karr, James R 62
Keller, Anne E. 145
Kelly, James ....
Kennedy, James H.
Kurtenbach, James
Langdon, Rich . . .
Lazorchak, James M.
Longsworth, M. . .
Lovell, Lawrence L.
Lowry, M ,
Lubinski, Kenneth S.
Magnuson, John . . ,
Maret, Terry R
Maxted, John R. . . .
McGee, B.M
Morrill, Jeffrey . . . .
Newton, Bruce J. . .
Olson, C
Overton, Jimmie . . .
Pendergast, James F.
Pratt, James R. . . .
Preston, Ronald . . .
Primrose, Miles L. . .
Prothro, Martha . . .
Rankin, Edward T. . .
Reid, Frederic A. ...
Reinharz, E
Schlekat.C.E
Schultz, R. Jack . . .
Stribling, James B. .
Swanson, E
Thorn, Ronald M. . .
Tsomides, Leonidas .
Waller, William T. . .
Williams, R
Woodwick, F. .....
Word, Jack Q
Wright, Christopher A.
Voder, Chris O. ...
. . . .123
.... 61
. . . .138
. . . .135
.... 90
. . . .161
. . , .155
. . . .161
.... 45
. ... 39
. . . .142
. ... 47
. . . .134
. . . .123
12
. . . .161
. ... 19
. .... 9
. ... 91
. ... 13
. . . .131
1
. . . .133
. ... 41
.'. . .134
. . .134
. . .129
... 25
. . .161
... 55
. . .147
... 61
. . .161
. . .161
... 40
. . .151
110, 159
•frU.S. GOVERNMENT PRINTING OFFICE: 1993-717-814/61004
171
-------
-------
-------
TO
CO
------- |