EPA/600/R-04/190
September 2004
POPULATION MODELS FOR STREAM
FISH RESPONSE TO HABITAT AND
HYDROLOGIC ALTERATION: THE CVI
WATERSHED TOOL
by
Brenda Rashleigh, M. Craig Barber, Michael J. Cyterski, John M. Johnston,
Rajbir Parmar and Yusuf Mohamoud
Ecosystems Research Division
Athens, GAS0605
National Exposoure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
Notice
The research described in this document was funded by the U.S. Environmental Protection Agency
through the Office of Research and Development. The research described herein was conducted at the
Ecosystems Research Division of the U.S. Environmental Protection Agency National Exposure
Research Laboratory in Athens, Georgia. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use. Although this work was reviewed by EPA and
approved for publication, it may not necessarily reflect official Agency policy.
Acknowledgments
Numerous people contributed administrative, database, software development, and technical support to
this project. Special thanks are given to Lawrence Burns (USEPA/NERL/ERD), Lourdes Prieto
(USEPA/NERL/ERD), Frank Stancil (USEPA/NERL/ERD), Kurt Wolfe (USEPA/NERL/ERD), Alan
Herlihy (USEPA/NHEERL/WED), Curt Seelinger (USEPA/NHEERL/WED), Marlys Cappaert
(USEPA/NHEERL/WED), Benjamin Daniel (CSC, Athens GA), Ronald Beloin (CSC, Athens GA),
Michael Galvin (CSC, Athens GA), Stephen Alberty (CSC, Athens GA), Tom DeMoss (USEPA/Region
3 and CVI), Paul Kinder(CVI), Jennifer Newland (CVI), and Ron Preston(CVI). We also thank Drs Joan
Baker (USEPA/NHEERL/WED) and Robert Swank for their technical and editorial reviews of this
report.
-------
Foreword
Streams and rivers provide important ecological services, including habitat for fishes and other organisms,
and drinking water supplies, yet these ecosystems are among the most impaired across the country.
Management for these ecosystems involves the assessment of probable causes of impairments and
management alternatives, as well as the forecasting of future condition in a scientifically defensible fashion
to more effectively protect and restore valued ecosystems. Communities, watershed groups and states require
decision support tools for managing the quality of aquatic systems. Community-based environmental
management is a long-term goal of the Agency, and providing the methods/tools and technical transfer
mechanisms to achieve this goal are critical to the role of ORD. Effective client collaborations are the most
efficient means to achieve this.
This report is the result of a collaboration with the Canaan Valley Institute (CVI) in which a decision analysis
toolkit was produced in order to support management of fisheries in the Mid-Atlantic Highlands. Although
there are many ecological endpoints that are important indicators of the condition of aquatic communities
and their associated watersheds, fish health is arguably one of the most important, since fishability is a
principal designated use for surface waters under the Clean Water Act. The approach used here can be
applied to aid CVI and other agencies in the management of aquatic resources in the Mid-Atlantic Highlands,
and may serve as a model for management tools for aquatic systems in other regions.
Rosemarie C. Russo, Ph.D.
Director
Ecosystems Research Division
Athens, Georgia
-------
Abstract
The Canaan Valley Institute (CVI) is dedicated to addressing the environmental problems in the Mid-
Atlantic Highlands (MAH). Their goal is to develop and implement solutions to restore damaged areas
and protect aquatic systems. In most wadeable streams of the Mid-Atlantic Highlands region of the
eastern United States, habitat alteration resulting from agriculture and development is the primary
stressor for fish communities. Sedimentation is the primary source of habitat degradation in Highlands
streams, and productive, sustainable fisheries, i.e., trophy trout streams, are the valued aquatic endpoints.
Planned restoration activities in the region include riparian zone restoration and stream channel design to
mitigate near stream inputs and stabilize streambanks. Natural Stream Channel Design (NSCD) is also
being investigated by CVI for further optimization of instream habitats for fish communities. Models that
predict the responses of fish populations and communities to key habitat characteristics are necessary for
CVI's watershed management goals, both for determining where to restore and how, as well as
evaluating the most probable outcome of various alternatives. The USEPA National Exposure Research
Laboratory (NERL) has developed a suite of modeling tools to be used for this purpose. The CVI
Watershed Health Assessment Tool Investigating Fisheries, WHAT IF, contains four components: 1) a
Hydrology Tool for predicting hydrologic characteristics of new streams of interest; 2) a Clustering Tool
for assigning the most probable fish assemblages to unsampled Mid-Atlantic Highlands streams, 3) a
Habitat Suitability Calculator, which evaluates habitat suitability of streams to support fish species and
families, and 4) the Bioaccumulation and Aquatic System Simulator (BASS) model, a generalized
aquatic ecosystem model that simulates fish community dynamics with time, which permits the
evaluation of comparative risk regarding instream restoration combined with fisheries management. The
USEPA Environmental Monitoring and Assessment Program (EMAP) surface water dataset (available
online, two index periods in the 1990's) is the basis of the habitat associations derived for fish species
and communities. Additionally, a tutorial is provided for the user to examine existing scenarios for fish
stocking, harvest and restoration combined. Stakeholders interact with the software interface to frame the
problem by: selecting valued endpoints of concern and analytical methods, accessing data and models to
establish the causal relationships between stream habitat characteristics and changes in endpoint
status/trend, and performing multiple model executions and visualizations of projected outcomes to span
the range of various management scenarios that might be taken so that associated costs and benefits can
be evaluated.
-------
Table of Contents
1. Introduction
10
2. Hydro Tool: Predicting Mean Depth, Width, and Streamflow for Small Streams 15
2.1. Introduction 15
2.2. What is the average flow depth, streamflow, and water temperature in my stream? 15
2.2.1. A ratio approach for estimating mean monthly streamflow 17
2.2.2. Stream Temperature Predictions 18
3. Clustering Tool: Predicting Fish Assemblages in the Mid-Atlantic Highlands 20
3.1. Background 20
3.2. What fishes might I have in my stream? 22
3.2.1. Methods 22
3.2.2. Discriminatory Power 27
3.2.3. Use of the Cluster/Discriminant Fish Assemblage Tool 28
3.3. Intended Audience 29
4. Habitat Suitability Tool 31
4.1. Introduction to habitat suitability assessment 31
4.2. How suitable is my stream for specific fish species? 32
4.2.1. Methodology 32
4.2.2. Results and Discussion 34
4.2.3. Interactive Tool 38
5. BASS: a Simulation Model for Fisheries Management 43
5.1. Using Simulation Models for Fisheries Management 43
5.2. Model Description 45
5.3. Developing Case Studies to Evaluate Mid-Atlantic Fisheries 46
5.3.1. Developing Initial Conditions 48
5.3.2. Parameterization of BASS Physiological and Ecological Processes 48
-------
5.4. Community Responses to Riparian Alteration and Fisheries Management 50
5.4.1. Responses of Non-game Streams to Riparian Restoration 50
5.4.2. Responses of Non-game Streams to Trout Stocking 51
6. Example Application of the CVI Watershed Toolkit 55
References 60
Appendix A. Clustering Methods and Algorithms 76
A.I. Density-to-Biomass Conversion Algorithm 76
Appendix B. BASS Bioenergetic and Population Dynamics Algorithms 86
B.I. Modeling Temperature Effects on Individual Growth 86
B.2. Modeling Growth of Fish 88
B.3. Modeling Trophic Interactions and Predatory Mortalities 93
B.4. Modeling Dispersal, Non-Predatory Mortalities, and Recruitment 97
B.5. Modeling Habitat Effects 99
B.6. Modeling Non-fish Compartments 100
-------
List of Figures
Figure 1.1 Questions to guide the user through the CVI tool 13
Figure 2.1 Hydro Tool Interface for Hogue Creek, VA 18
Figure 3.1 Primary members of Cluster 10, according to percent of total assemblage biomass 25
Figure 3.2 Software tool for predicting stream fish assemblages based on stream and watershed
characteristics 30
Figure 4.1 Location of sample sites within the Mid-Atlantic Highlands region 35
Figure 4.2 Interface for the Habitat Suitability Tool in the CVI Watershed 40
Figure 4.3 Interface for Best Management Practice (BMP) scenarios in the CVI Habitat Suitability Tool.
42
Figure 6.1 The use of the WHAT-IF tool to address a more complex management question for a
particular stream site 55
Figure 6.2 Predicted biomass dynamics of Hogue Creek, VA before riparian restoration 59
Figure 6.3 Predicted biomass dynamics for Hogue Creek, VA after riparian restoration 59
Figure A.1 Plot of observed versus predicted biomasses. Indicated line represents the identity
relationship of observed biomass equals predicted biomass 85
Figure B.I Maximum daily ingestion of brown trout (Salmo trutta) as a function of temperature. Data
from Elliott (1976b) 87
-------
List of Tables
Table 2.1 Regional regression equations for the three physiographic provinces of the Mid-Atlantic
region. All length units are in feet; flow is cubic feet/sec; flow area is in square feet; drainage
area is in square miles 16
Table 3.1 Fundamental fish assemblages of the Mid-Atlantic Highlands. Species representing more than
10% of the assemblage biomass are listed in the cluster name 24
Table 3.2 Relating the impairment index to fish communities in MAH streams 27
Table 3.3 Relating stream characteristics to fish communities in MAH streams 28
Table 4.1 Instream habitat measures that were selected from the EMAP dataset as possible explanatory
variables for the presence offish species in the Mid-Atlantic Highlands region 33
Table 4.2 Results from logistic regression models of presence of Mid-Atlantic Highlands stream fish
species/groups. Results include the number of occurrences in the data set (N) and goodness-of-fit
statistics for overall logistic regression models: the Wald Chi-square value and its significance
(P), and the percentage of correct classifications determined using cross-validation (a higher
value indicates greater predictive ability) 36
Table 4.3 Parameter estimates and odds ratios from logistic regression models of presence of Mid-
Atlantic Highlands stream fish species/groups 37
Table 4.4 Scenarios of Best Management Practices (BMPs) provided within the CVI Habitat Suitability
tool 41
Table 5.1 EMAP fish sites selected as fisheries and riparian alteration case studies 47
Table 5.2 Summary of default species assignments for parameterizing BASS for MAH genera 49
Table 5.3 Summary of conversion factors for EMAP macroinvertebrate data 49
Table 5.4 Summary of annual mean biomasses and fluxes predicted by BASS for Bell Run PA, Flat
Creek VA, and Tuscarora Creek WV for status quo conditions, with a 25% restoration of riparian
canopy and ground cover, and with trout stocking 54
Table 6.1 Status quo and restoration HSI for Hogue Creek, VA 58
Table A.1 Data sources for supplementing Carlander (1969, 1977, 1997) 81
Table A.2 Summary of growth data used for biomass estimation of MAH fish species 83
-------
List of Acronyms
BASS Bioaccumulation and Aquatic System Simulator
BMP Best Management Practice
CVI Canaan Valley Institute
EMAP Environmental Monitoring and Assessment Program
EPA Estimated Prediction Accuracy
ERD Ecosystems Research Division
GIS Geographic Information Systems
GUI Graphical User Interface
HSI Habitat Suitability Index
IBI Index of Biological Integrity
MAH Mid-Atlantic Highlands
NERL National Exposure Research Laboratory
USEPA U.S. Environmental Protection Agency
USGS U.S. Geological Survey
-------
1. Introduction
Scientists recognize that fish assemblages in developed watersheds are affected primarily by nonpoint
source anthropogenic stressors that result from land use development, in particular alteration of physical
habitat (Williams et al. 1989, Richter et al. 1997, Wilcove et al. 1998). Over half of the streams in the
Mid-Atlantic Highlands (MAH) have fish communities that are in fair or poor condition, and the USEPA
concluded that physical habitat alteration represents the greatest potential stressor across this region
(USEPA 2000). Habitat alteration can occur both in terms of habitat quantity and quality. Loss or
destruction of habitat quantity reduces the total amount of habitat available to aquatic species, and can
isolate patches of suitable habitat within a stream, which reduces species' survival and alters natural fish
movement and migration patterns (Reeves et al. 1995). Loss of habitat quantity is often associated with
significant hydrologic alterations, such as impoundments (Yeager 1993), whereas loss of habitat quality
can be due to factors such as landscape development and alteration of flow patterns on the landscape.
The mission of the Canaan Valley Institute (CVI) is to address the environmental problems in the Mid-
Atlantic Highlands through a program of environmental stewardship that considers and integrates natural,
economic, and human concerns in the management of natural resources (CVI 2002). Their goal is to
develop and implement solutions to restore damaged areas and protect aquatic systems. To achieve this
goal, they require sound science that combines theory, detailed knowledge, monitoring and modeling
(CVI 2002). Toward this goal, CVI developed its own geographic information system (GlS)-based
management tool, Landscape Analyst, based on the proprietary Arc View GIS (ESRI, Redlands, CA), for
estimating land use change impacts on water quantity and quality (http://www.canaanvi.org/). However,
the adoption and widespread application of the tool has been less than anticipated. In order to access
Landscape Analyst, users must posses the required GIS software and utilities and be familiar with GIS
software and its operation. Complexity in the software user interface, as well as the supporting science
modules, is a barrier to widespread adoption of the tools. Refining the aquatic assessment needs further,
CVI posed these questions on behalf of regional stakeholders: What conditions will sustain aquatic
endpoints in the long-term? Can we evaluate restoration techniques like natural stream channel design
(NSCD) and agriculture and forestry stream best management practices (BMPs) for their effectiveness in
improving aquatic endpoints? Can we create "what if scenarios and evaluate management actions based
on the response of aquatic endpoints?
10
-------
In support of CVI, the USEPA National Exposure Laboratory's Ecosystem Research Division (ERD) has
conducted research to develop watershed modeling tools for CVI and their stakeholders in the Mid-
Atlantic Highlands. Specifically, models have been developed that support the information required to
conduct aquatic ecosystem assessments, including:
• the assignment of instream hydrological habitat quantities for ungauged/unmeasured streams
of interest: stream flow, depth, width and temperature (Hydro Tool)
• the assignment of the most probable fish communities for unsampled streams (Clustering
Tool)
• the evaluation of the physical habitat variables that affect suitability for fish species and
families/guilds to further refine the fish community assignment and address restoration and BMP
actions for selected species (Habitat Suitability Calculator)
• the evaluation of expected trophic dynamics of the dominant fish species under various
fisheries management and restoration actions for cumulative/comparative risk assessment 3,5 or
10 years in the future (BASS bioaccumulation and population dynamics model).
The result is a suite of tools for regional application to wadeable Mid-Atlantic Highlands streams, the
Canaan Valley Institute - Watershed Health Assessment Tools Investigating Fisheries, CVI-WHAT IF.
WHAT IF is entirely open source and does not require proprietary software. It is not GIS-based, nor does
it contain models difficult to apply for all but expert users. Models are complementary in the information
provided and are designed to be efficient and problem focused. For example, WHAT IF incorporates
statistical hydrology in the toolset rather than a watershed hydrology model. Models that require a high
degree of input data processing and model setup put undue burden on novice users (Doherty and
Johnston 2003). A fitted parameter model such as the Hydrologic System Program Fortran (HSPF),
though widely used, is complex and calibration intensive and does not match the aquatic assessment
needs for evaluating outcomes of near and instream restorations. Similarly, models with a large degree of
overhead in their use, including sediment hydrodynamic models, are unsuitable for the majority of CVI
stakeholders and are not straightforward in application.
11
-------
WHAT IF consists of models of varying complexity, suited to the problem needs of the client rather than
imposing a 'one model fits all needs' restriction. In this manner, the toolset is question driven as opposed
to model-focused. WHAT IF is stream-based, permitting users to evaluate streams of interest for
outcomes of management approaches and specific restoration actions. Habitat quality and aquatic
ecosystem response models have been linked to a regional hydrologic model that simulates habitat
characteristics (e.g., water depth, current velocity and water temperature) that determine the survival,
reproduction, and recruitment of fish and aquatic invertebrates. To facilitate the use and application of
these models, graphical user interfaces (GUI), supporting databases, and libraries of management
scenarios were developed. The Canaan Valley Institute (CVI) software/toolset is a stream-based decision
support tool that is object-oriented in design and easily maintained. Ultimately, what has been developed
using available data collected by the USEPA Environmental Monitoring and Assessment Program
(EMAP) is a framework based on the biogeography of fish suitable for applying all models for regional
assessments of important fish health issues in the Mid-Atlantic Highlands. It is set in the form of a series
of questions for the user (Figure 1.1).
The EMAP surface water dataset (http://www.epa.gov/nheerl/arm/) is a stratified random sampling
design and contains over 300 sites and 600 samples (multiple site visits) used in the final research
product. Fish count data were pre-processed by conversion to biomass for both clustering and habitat
suitability analyses, using a utility associated with the BASS model. In this manner, actual communities
were derived from the base data, so that conclusions could be made with respect to biomass per unit area
and ultimately the carrying capacity of impaired versus restored streams. The EMAP dataset represents
two index periods (1993-1995 and 1997-1998) and habitat associations are statistically valid within the
region of the Mid-Atlantic Highlands, which includes portions of Pennsylvania, Maryland, West Virginia
and Virginia. Watershed characteristics, in stream and near stream habitat quantitative data, are used in
addition to fish, benthic insect, and periphyton data (i.e., attached algae). Field data were collected from
late Spring to late Summer and span a range of stream sizes and watershed areas throughout the
Highlands.
The overall flow of information between the various models is as follows. The Hydrology Tool supplies
stream conditions for sites in the software archive, as well as a means of calculating conditions (mean
monthly flow, depth and water temperature) for new streams of interest. The Clustering Tool provides a
means of assigning the most probable fish community to a stream based on combinations of remotely
12
-------
sensed and field collection data. This tool also permits a user to perform simple screening level
assessments of habitat changes (water flow, temperature and percent fines) and the probable changes in
the assigned stream community. After identifying a stream of interest and assigning a fish community,
the Habitat Suitability Calculator is used to evaluate that stream's potential to support members of the
potential community, on a fish-by-fish basis. Habitat features of interest are those that relate to key
aspects offish life history requirements: water velocity, water depth, stream bottom composition
(substrate) and amount of refugia (cover, riparian vegetation, etc.). The BMP tool that is part of the
Suitability Calculator provides a means of translating habitat changes (i.e., restoration actions under
consideration) into suitability scores for each fish in the community. BASS is a generalized aquatic
ecosystem simulator capable of treating a variety of managed freshwater ecosystems and accepts the
assigned fish community as an initial condition, as well as any habitat multipliers investigated using the
BMP tool. Stream depth and temperature information is also passed from the Hydrology Tool to BASS.
With these tools, environmental managers are better able to characterize and quantify relationships
between stressors and stream responses for valued ecological resources in a manner that supports
diagnosis of current condition and assists in management activities.
13
-------
Figure 1.1 Questions to guide the user through the CVI tool.
EH cvi r oois i.o
File View Window Help
;i ' mini*!
A
R^fto A|
•••• 'i — •
What is the averaae flow depth, streamflow. and water temperature in mv ft ream?
Clustering
What fishes mioht 1 have in rrw stream?
How mtqht species composition chanqe with changes in watershed arid stream characteristics?
Habitat Suitab%
How suitable is my stream for specific fish loecies?
How midit suitability be irrcraved for these species?
BASS
How averaoe lerrth weiahl and other attributes of fish peculations anc
ehanoe thtouoh time when BMPs aie aDoied?
corrmunities
How does the proportion of trophy-sized trout chanqe with various stocking programs?
Tutorial: Review Expected Community Dynamics
How do trout respond to riparian zone restoration? 1
How do trout respond to stocking?
How do trout respond to a combination ot riparian restoration and stockina?
Advanced
«l
Close I w 1
I >U
14
-------
2. Hydro Tool: Predicting Mean Depth, Width, and Streamflow for
Small Streams
2.1. Introduction
Instream physical habitat characteristics, such as water depth, temperature, and mean flow velocity, are
important to the growth and survival offish species at different life stages. The ecologically-focused
average annual discharge is highly correlated with drainage area. In addition, bankfull discharge and
average annual discharge are also highly correlated with each other. Both are important data for resource
managers and planners who would like to know the maximum flow frequency that fills and forms the
channel shape (bankfull) and the average discharge available to sustain fish and aquatic habitats
throughout the year. In this work, a regional regression method is developed for the determination of
mean water depth and mean flow velocity variables.
2.2. What is the average flow depth, streamflow, and water temperature in my
stream?
The regional regression method is based on the quantification of relationships between drainage area and
stream hydraulic characteristics. To enhance the predictive potential of the regression equations and to
reduce the percentage of the variability not explained by the model, we developed separate regression for
each physiographic province. The Mid-Atlantic region consists of four physiographic provinces:
Appalachian Plateau, Blue Ridge, Ridge and Valley, and Piedmont. We combined the Ridge and Valley
and Blue Ridge Physiographic Provinces as one and developed a total of three sets of regression
equations.
In each physiographic province, about 25 streams with variable drainage areas were selected. The criteria
used to select the study watersheds were based on watershed area, land use type, and availability of over
10 years of observed streamflow data. All selected watersheds had a drainage area ranging between 2 to
about 400 square miles. All selected watersheds had high percent of forest cover and agricultural land
use and a low percent of impervious surface cover. Only watersheds with USGS gaging stations were
15
-------
Table 2.1 Regional regression equations for the three physiographic provinces of the Mid-Atlantic region. All length units are in
feet; flow is cubic feet/sec; flow area is in square feet; drainage area is in square miles.
Regression Model Summary Statistics
Model R2 N Coefficients Value Standard
Error
Lower
95% CB
Upper
95% CB
Appalachian Plateau Physiographic Province
Q = aDAb 0.96 25 a
b
W = aQb 0.81 25 a
b
A = aQb 0.86 25 a
b
D=aQb 0.84 19 a
b
Ridge and
Q = aDAb 0.95 25 a
b
W = aQb 0.86 25 a
b
A = aQb 0.91 25 a
b
D=aQb 0.88 19 a
b
3.41
0.85
6.09
0.47
3.16
0.67
0.33
0.27
Valley Physiographic
2.98
0.82
3.82
0.58
1.12
0.89
0.31
0.3
0.78
0.04
1.72
0.05
1.13
0.06
0.05
0.03
Province
0.87
0.05
1.30
0.06
0.51
0.08
0.05
0.03
1.83
0.76
2.54
0.37
0.81
0.54
0.22
0.21
1.17
0.71
1.13
0.45
0.07
0.72
0.21
0.23
4.97
0.94
9.66
0.58
5.51
0.80
0.44
0.34
4.78
0.94
6.51
0.72
2.20
1.06
0.42
0.36
Piedmont Physiographic Province
Q = aDAb 0.98 25 a
b
W = aQb 0.90 19 a
b
A = aQb 0.91 19 a
b
D=aQb 0.86 19 a
b
1.35
0.99
7.39
0.47
3.67
0.65
0.40
0.23
0.23
0.03
1.32
0.04
1.01
0.05
0.04
0.02
0.87
0.92
4.65
0.39
1.57
0.54
0.30
0.18
1.84
1.06
10.3
0.54
5.76
0.76
0.48
0.28
selected for the development of regional regression equations. For each gaging station, the mean annual
streamflow was determined from the historical streamflow data. Hydraulic channel geometry data such as
width, depth, velocity, and cross-sectional area plus flow velocity that correspond to the mean
streamflow were determined from USGS stream measurement data. Using regression analysis, we
16
-------
developed quantitative relationships between the drainage area of a watershed and its mean stream
hydraulic properties such as mean streamflow rate, mean flow depth, mean flow width, and mean cross-
sectional flow area. Regression equations for the three physiographic province groups are shown in
Table 2.1. These regional regression equations can be used to estimate mean streamflow rate, mean
depth, mean width, and mean cross-section area for small streams located in ungaged watersheds.
Our study is ecologically-oriented, and its focus is to determine micro-habitat variables needed as input
variables into ecological endpoint models such as the Bioaccumulation in Aquatic Systems Simulator
(BASS) Model. For this reason, the Table 2.1 regression equations have been incorporated into the CVI
Watershed Tool to automatically provide input for various components of the tool. The interface for the
Hydro Tool is shown in Figure 2.1. This tool calculates the annual mean watershed hydraulic parameter
values using regional regression equations, but also uses a ratio approach to estimate monthly values
based on the predicted annual mean as well.
2.2.1. A ratio approach for estimating mean monthly streamflow
The ratio approach is based on the assumption that, within a relatively large watershed, normalized mean
streamflow remains nearly constant across all sub-watersheds of the large watershed. Based on this
assumption, normalized streamflow can be transferred from a nearby gaged watershed to an ungaged
subwatershed of interest within the large watershed. One requirement is that the gaged watershed must
have larger drainage area than the ungaged watershed. In general, it is desirable to select a gaged
watershed from downstream of the ungaged watershed. Note that this approach uses both drainage area
and mean streamflow normalization as basis for transfer of streamflow data from a gaged watershed to an
unaged watershed. This approach serves two purposes. First, it addresses the limitations of the regional
regression equations that are based on the assumption that all watersheds within a physiographic
province can be represented by a single regression equation. The use of the ratio approach to transfer a
streamflow data from a nearby gaged station therefore adjusts the prediction of the regional equations to
address the variability in streamflow within each physiographic province. Second, the ratio approach is a
transfer method that does not use precipitation as a predictor variable but rather uses streamflow to
transfer streamflow data from a gaged watershed to an ungaged watershed. The ratio approach can be
written as:
17
-------
Qy=
(2-1)
where Qy is the monthly streamflow from the unaged watershed; Ay is the drainage area of the ungaged
watershed; Qx is the monthly streamflow of the gaged watershed; Qx is the monthly streamflow of the
gaged watershed; Qar is the mean annual streamflow determined by the regional regression method; and
Qmean ^s mean annual streamflow of the gaged watershed determined from the long-term monthly
streamflow data.
Figure 2.1 Hydro Tool Interface for Hogue Creek, VA
Hydrology Tool
HOGUE CREEK, VA, community: Smallmouth, condition: fair.
Instructions: The first table contains the annual averaged hydrology parameters for the selected
site. The second table has the monthly parameter values. Highlighted values correspond to the
period of data collection for fishes and habitat. Statistical relationships for community and habitat
suitability analyses are derived during this period.
Month
> January
February
March
April
May
June
July
August
September
October
November
December
Temperature (C)
3.61
7.06
11
16.06
16.89
21.11
22.22
23.28
19.89
15.39
11.67
3.89
Depth (ft)
0.98
1.06
1.16
1.12
1.03
0.89
0.77
0.76
0.73
0.81
0.84
0.94
Streamflow (cfs) T
50.01
66.38
96.33
83.91
59.37 |;
33.11
18.1
16.75
14.42
22.75
26.65
41.43 i
Previous
Next
Cancel
2.2.2. Stream Temperature Predictions
In this study, relationships between air and water temperature developed by Stefan and Preudhomme
18
-------
(1993) were used to estimate stream water temperature from daily air temperature measured from a
nearby weather station. The results of their study concluded that stream water temperatures follow
closely with air temperature with some lag that varies with time scale, i.e., hours or days and with
increasing stream depth. Small shallow rivers had smaller temperature deviations than large, deep rivers.
Their regression equation is written as:
Tw= 5 +0.75*7; (2-2)
where Tw is stream water temperature and Ta is air temperature in degree Celsius.
Webb et al. (2003) also studied the water-air temperature relationships and its moderation by streamflow.
They found that the relationships are stronger and more sensitive to flows below the median flow and
that water temperature is inversely related to streamflow for all time scales and drainage areas. They
concluded that streamflow had greater impact on stream temperature variations at short time scales and in
larger watersheds.
The procedure followed to determine water temperature was to identify the nearest weather station for
each stream and then use the relationships given in Equation (2-2) to estimate stream water temperature
from air temperature. In watersheds with no nearby weather stations or in some cases where significant
elevation differences exist between the stream location and weather station location, development of
elevation-latitude-time of the year and air temperature relationships may be needed.
19
-------
3. Clustering Tool: Predicting Fish Assemblages in the Mid-Atlantic
Highlands
3.1. Background
To predict fish community response to various proposed environmental restoration actions in the MAH
Region using an empirical approach, one must first have information on fish abundance and diversity.
We used data collected by the USEPA EMAP program to identify the fundamental fish assemblages in
wadeable streams of the Mid-Atlantic Highlands Region. EMAP staff, working with personnel from the
US Fish and Wildlife Service, as well as state and contract personnel, set out to assess the physical,
chemical, and biological condition of MAH streams (USEPA 2000). Given that there are approximately
80,000 total stream miles in the MAH, it was not possible to sample each one. Instead, a spatially-
constrained, randomized statistical design was developed to choose a subset of 1st through 3rd order
wadeable streams where data collection would occur. The EMAP objective was to provide unbiased
estimates of stream condition throughout the MAH, quantify the proportion of stream miles that are
biologically degraded or impaired, and examine the relative importance of various stressors on stream
fish communities in the region.
To gather data on the fish species present in the sampled streams, the EMAP team used backpack
electroshocking as their primary method of collection. At each sampling location, they performed three
passes over a reach length approximately 40 times the mean wetted channel width at the midpoint of the
reach, with a minimum distance of 150 m. This insured sampling was performed across a range of
habitats at each site (runs, riffles, rapids, and pools). The abundance and diversity offish species were
recorded at each site. Habitat measurements were also taken at many of these sites (e.g., stream flow,
dissolved oxygen, temperature, sediment levels, riparian cover) as well as watershed characteristics (e.g.,
total watershed area and percents of agriculture, forest, and disturbed land areas in the watershed). Our
focus in developing this model was on smaller, wadeable streams, so we only analyzed data from those
EMAP sites that were less than 20 meters in width. Fish densities recorded by EMAP personnel at nearly
80% of streams over this size were less than 100 fish per hectare. For streams under 20 meters in width,
this low fish density was seen at 3% of sites. This discrepancy in fish densities between small and large
streams likely stems from bias in the sampling technique (backpack electroshocking); capture efficiency
20
-------
declines appreciably in larger lotic systems (Reynolds 1996). Due to the inherent bias of the sampling
techniques that were used, these EMAP data can be used to estimate relative abundance of fishes within
and between sites, but should not be used to estimate absolute fish abundances at the sampled sites.
Numerous researchers have used multivariate statistical approaches to find patterns in fish assemblage
data using a variety of sorting criteria, amongst them taxonomic, geographic, limnologic, and
physiographic. Angermeier and Winston (1999) used several multivariate techniques to examine fish
communities in Virginia streams at several spatial scales. They also examined relationships between
community composition and landscape variables, such as stream order, channel slope, and elevation.
They found that ecological and taxonomic characterizations of community composition produced similar
results. Madejczyk et al. (1998) used cluster analysis to investigate electrofishing data and related the
presence and absence of various species to artificial and natural habitats along the shoreline of the upper
Mississippi River. Saiki and Martin (2001) used Ward's minimum variance method of cluster analysis to
find two dominant fish communities in Abbotts Lagoon in Point Reyes National Seashore. Their analysis
was based on data from gill nets and minnow traps. Kendrick and Francis (2002) used both Ward's
cluster analysis and Canonical Correspondence Analysis to analyze the species assemblages of the
Hauraki Gulf, New Zealand. They worked with trawl data, and found four dominant fish assemblages,
with an additional four species showing no relation to these groups. They caution that year-to-year
variation in fish abundance can greatly affect any assemblage analysis. Mathews and Robinson (1988)
identified five faunal regions within Arkansas and related these regions to geography, meteorology, and
physical variables. Wilkinson and Edds (2001) performed a space-constrained cluster analysis, along
with a principal coordinates analysis, to conclude that three distinct fish fauna existed in the Spring River
basin, and that these communities were related to differences in geographic and habitat differences
between the Ozark Highlands, Central Plains, and mainstream regions of the basin. The spatial and
temporal patterns of distribution and abundance of a tropical fish community were investigated via
cluster analyses by Ornellas and Coutinho (1998). Their findings indicated the fish community was
greatly influenced by the aquatic macrophyte beds which determined the availability of space and
habitats. Ansari et al. (1995) looked at data from trawl hauls on the structure and seasonal variation of
fish species at Goa on the West Coast of India. Their cluster analysis showed a strong seasonal
component in the determination of species groupings. Fish communities in freshwater lakes from
watersheds near Lake Ontario were identified using cluster analysis by Kelso and Johnson (1991).
21
-------
McCormick et al. (2000) used a subset of the MAH dataset we analyzed. They performed a cluster
analysis on the fish collection data to arrive at primary fish assemblages for the region. Along with this
analysis, they also investigated whether or not fish assemblages could be defined based on spatial data,
such as ecoregions and catchments, as well as stream order. Their analyses were based on count data, and
they concluded that any historic fine-scale structures of Highland fish assemblages have been overridden
by intricate zoogeographic patterns and many years of human disturbance. We have chosen a different
route, opting to analyze relative fish biomass rather than abundance. Because biomass is always
conserved while numbers of individuals are not, the former is seen as a more robust indicator of
ecological importance within fish communities (Diana 1995). In addition, there is the necessity of having
biomass measures for creating the input files that will drive our BASS modeling effort.
3.2. What fishes might I have in my stream?
3.2.1. Methods
A software tool was developed using the EMAP stream sampling data from the MAH and incorporated
into WHAT-IF. This statistical model predicts a stream's fish assemblage using stream and watershed
characteristics. Step one in the tool development was a k-means cluster analysis (Fisher 1958) that
grouped streams with similar dominant fish species. The data matrix input into SAS for analysis
consisted of 562 rows, one per sampled site, and 105 columns (representing the various fish species
sampled across all sites). For each site, the relative biomass of the three most abundant species was
recorded in the appropriate columns; the procedure used to estimate species biomasses from EMAP's
reported count/density data is outlined in Appendix A. Species of lesser abundance at each site were not
considered because our goal was to identify the dominant species/assemblages of the MAH. Other
statistical techniques would need to be applied to this dataset if the focus were on prediction of the
presence of rare species in a given stream. By examination of the results of the SAS clustering output, we
found further justification for using only the three most abundant species to represent each site, i.e., the
presence of subdominant species could warp the clustering results. We observed that observations/
clusters could be grouped with other observations/clusters if the two shared a number of subdominant
species, despite the fact that they did not share the same dominant, or even secondary, species. This
result was dissatisfying, and so we choose to cluster based only on the three most dominant species
present at each site.
22
-------
Clustering techniques typically define each cluster by a multidimensional mean, or centroid. In our case,
this centroid was the vector of biomass values for every species in the cluster. During each iteration of
the process, observations (streams) were added to the cluster with the nearest centroid. However, we
specified that an observation had to be within 48 units of Euclidean distance of a cluster centroid to be
permitted to join that cluster. This distance was based on the EMAP data, which showed on average that
the dominant species at a site accounted for 55% of total fish biomass, while the second most abundant
species on average accounted for 21% offish biomass at the site. If two sites had a reversal of their
dominant and secondary species, but shared the same third-most abundant species, the Euclidean distance
between these sites would be y2x(55 - 21)2 ~ 48. In essence, we designated that a reversal of the
dominant and secondary species was a highly significant change and should be used as a threshold of
difference between sites. After any observation was added to a cluster, that cluster centroid was
recalculated. The process was terminated only after all observations ceased to change their cluster
membership from iteration to iteration. Clusters with less than 11 members (2% of the total sample size
of 562 streams) were deleted at the end of any iteration step, meaning any observations in these clusters
had to be reassigned in the next iteration. These smaller clusters were judged not to be representative of
dominant fish assemblages in the MAH.
The first step in the k-means clustering analysis is selection of sites to serve as "seeds" that define the
initial clusters at the start of iteration 1. This selection of initial seeds influences the end results of the
analysis. Due to the subjectivity of this process, we choose to adopt a random initial seed selection,
within the constraint that any new seed had to be at least 48 units of Euclidean distance from any existing
seed. We set the maximum number of initial seeds to 25, which was roughly equivalent to the total
number of different species that were dominant at more than one site in the EMAP dataset. Because
initial seed selection was random, results of the cluster analysis varied from run to run. We arrived at the
final results by running the clustering algorithm hundreds of times (each time with a different set of
initial seeds), then choosing the run with the "best" combination of three outputs:
1) Cubic Clustering Criterion (CCC Sarle 1983) - a measure of the variation in the dataset explained by
the clusters
2) Nearest Neighbor distances - the distance between each cluster centroid and the nearest cluster's
centroid. Runs with the very highest CCC values tend to have larger numbers of clusters, but the fish
23
-------
assemblages of some of the resulting clusters can be very similar. This result would hinder a subsequent
discriminant analysis. The analyst should thus decide on a minimum tolerable nearest neighbor distance
between the final clusters. After examining many pairs of nearest neighbor clusters, we concluded that a
distance of approximately 30 Euclidean units was appropriate for this data set.
3) Root Mean Square Standard Deviation (RMSSTD) - the variability within each cluster. Taking the
average RMSSTD value across all clusters is an indication of how well-defined the clusters are.
After generating approximately 1000 simulations, we saved the output from the runs with the top 5 CCC
values. We then looked at the nearest neighbor distances and the average RMSSTD values for these five
runs. As our designated "winner," we chose the run with the third largest CCC value. This run had good
separation between the final clusters, with no nearest neighbor distance below 30 Euclidean units, and it
had the second-lowest average RMSSTD value. The run with the highest CCC value also had the lowest
average RMSSTD value, but it had three pairs of clusters with nearest neighbor distances below 30
Euclidean units, one of which was below 20 Euclidean units.
Table 3.1 Fundamental fish assemblages of the Mid-Atlantic Highlands. Species representing more than 10%
of the assemblage biomass are listed in the cluster name.
duster N Dominant Species Biomass RvlSSTD hND stance
1
2
3
A
5
6
7
8
9
10
11
12
13
14
15
16
19
26
36
31
27
28
45
33
27
22
32
70
16
20
36
45
Brook Trout
FaJIfishMhite Sucker
Ftck Bsss/Northern Hoosucker/Smdlmouth Bass
Backnose DaceAjonanose DaceVvli it e Sucker
Creek Chub
Blacknose Dace/Creek Chub
Northern hbqsucker/Smdl mouth Bass
Bluehead ChubCreekChub
Creek Chu bWh it e Su cker
Torrent Sucker
Blecknose Dace
White SuckerBlocknose Dace
Creek Chubsucker
Rock Bass/vVh it e Sucker
Creek Chub/Bleckncse Dace
White Sucker
86
35/29
"10/18/11
40/18/10
91
57/25
54/10
52/10
40/39
71
94
52/10
70
74/10
60/24
79
2.19
2.89
2.88
3.11
1.14
2.56
3.02
2.94
2.07
2.41
1.27
2.65
3.06
2.23
2.21
1.75
83.2
32.5
37.7
36.0
38.4
36.0
37.7
50.3
32.0
68.8
46.6
32.0
64.3
41.0
38.4
34.2
Our chosen run produced 16 clusters (Table 3.1). High cluster variability (as indicated by larger
RMSSTD values for Clusters 4, 7, and 13) is an indication that the assemblage is less well-defined, i.e.,
there are a large number of species associated with the cluster and differences in species-specific biomass
from stream to stream. This also means that one's success in predicting the precise species composition
24
-------
of any one stream within these highly-variable clusters is lessened. It is important to note that the species
listed in Table 3.1 are not the only species found in their respective clusters. For instance, the primary
members of Cluster 16 (according to percent of total assemblage biomass) are shown in Figure 3.1.
We note that the assemblage of fishes defined by one of these clusters is not identical to the fish
community found in any one stream in the set, but instead provides a pool of species that could be found
in streams of this type. For example, Cluster 8 is defined by a set of 20 species, but an actual stream from
this group might only have a subset of six to ten species from the total species pool. The fish community
of Stream A could consist of bluehead chub, creek chub, redbreast sunfish, northern hogsucker, and
green sunfish, while the community of stream B is bluehead chub, green sunfish, rock bass, redbreast
sunfish, and blacknose dace. These two streams belong to the same cluster and have three species in
common, but they do not have identical fish communities.
Figure 3.1 Primary members of Cluster 10, according to percent of total assemblage
biomass.
RoanokE
Golden Redhorse, Hogsucker, 1.4
1.5 i
White Sucker, 1.7
Fallfish,5.4
Bluehead Chub
6.6
Blacknose Dace,
7.4
Other Species ,4.9
Torrent Sucker,
71.1
Discriminant Analysis
Step two in the tool development process was a discriminant analysis that produced a system of equations
to predict a stream's fish assemblage based on characteristics of that stream and its watershed. Some of
the characteristics used in the analysis can be quantified by simply looking at maps or using available
25
-------
databases, while other data must be collected at actual stream sites or calculated via regression equations
in the Hydrology Tool. Below is a list of variables of each type that we included.
Remotely-Sensed
Latitude
Longitude
Watershed Area (ha)
% Agricultural Area in Watershed
% Urban Area in Watershed
% Forested Area in Watershed
Stream Slope
Mean Stream Elevation (m)
Annual Mean Precipitation (m)
Watershed Road Density (m per ha)
Watershed Population Density (people
per km2)
Measured On-Site/Determined via Regression
Mean Stream Depth (cm)
Mean Stream Width (m)
Mean Stream Flow (cfs)
Dissolved Oxygen in Stream (mg per liter)
Stream Temperature (degrees C)
% of Stream Bank with Riparian Cover
% of Stream Bottom Covered in Fine Sediments
To begin, the correlation matrix of all predictors was examined, and for predictor pairs with a correlation
above 0.75, the most easily interpretable and understandable member the pair was retained for analysis.
Then, using the clusters from Table 3.1 as the response variable, all remaining predictors were added to
form the full model (Minitab software was used to perform the analysis). At that point, one by one, the
least useful predictors, in terms of their effect on predictive success probabilities, were dropped from the
model, i.e., a backwards-selection methodology was used. The process of deleting predictors was stopped
when the predictive success probability dropped by more than 5% after removing a predictor from the
model. The final model contained the following predictors: latitude, longitude, stream elevation, stream
slope, % disturbed area in the watershed, stream width, mean thalweg depth, stream flow, temperature,
dissolved oxygen, % of the stream bed covered in fine sediment.
Basic summary statistics were also used to examine the clusters of Table 3.1. Table 3.2 gives the
26
-------
assemblages sorted by an Impairment Index that was calculated as the sum of four parameters: %
Agricultural Landuse, % Urban Landuse, % Fine Sediments, and (100 - % Riparian Zone). Watersheds
with sizeable disturbed areas will typically experience more soil erosion and have higher in-stream
sedimentation than streams in heavily forested watersheds. Species tolerant of sediments, such as white
sucker and creek chubsucker are more prevalent in agricultural watersheds, whereas intolerant species,
such as the brook trout and northern hogsucker, are found in primarily undisturbed, forested watersheds.
Table 32 Relating the impairment index to fish communities in MAH streams.
Dominant Species Impairment Index
Brook Trout 15
N orthem H ogsucker/Smallm outh Bass 38
Torrent Sucker 48
Rock Bass/Northern Hogsucker/Smallmoutri Bass 51
Blacknose Dace/Long nose Dace/White Sucker 54
Creek Chub/Blacknose Dace 54
Blacknose Dace 59
Creek C hub/Wh ite Sucker 60
Bluehead Chub/Creek Chub 60
Creek Chub 63
White Sucker/Blacknose Dace 68
R oc k B as s/W hite S u cker 69
Blacknose Dace/Creek Chub 71
Fallfish/White Sucker 73
Creek Chubsucker 89
White Sucker 97
Table 3.3 summarizes the averages of measured on-site variables for each cluster in Table 3.1. This
information quickly indicates the type of stream each assemblage would likely inhabit. For example, a
brook trout dominated assemblage would most likely be found in a small, cool, well-oxygenated stream
with little sediment and extensive riparian buffer.
3.2.2. Discriminatory Power
Using the 16 functions (one for each cluster) output by the discriminant analysis, we could predict the
correct stream cluster for streams in our dataset with approximately 41% accuracy. Given that there were
16 total clusters included, the random chance of picking the correct stream cluster would only be about
6% (1/16). The power of prediction was thus increased by nearly seven times by using the stream and
27
-------
watershed characteristics. In addition, the correct stream cluster was predicted to be one of the three most
likely outcomes (as opposed to the more rigorous case discussed above) for 70% of the stream sites.
Randomly, given three choices, one would only have a 3 in 16 chance of picking the correct cluster
(19%).
Table 33 Relating stream characteristics to fish communities in MAH streams.
Mean Ainual Stream Bank % Bottom Cohered
Dominant Species Depth (cm) Temp(oC) DO (mg/\) Riparian% by Fine Sediment
Brook Trout
F all fish White Sucker
Rock Bass/NorthernHogsucker/Smallmouth Bass
BlacknoseDace/Longnose DacejWhite Sucker
Creek Chub
Black nose DaceCreek Chub
Northern HogsuckerfSmallmouth Bass
Bluehead Chub/Creek Chub
Creek ChubMMe Sucker
Torrent Sucker
Black nose Dace
White Suckerffllacknose Dace
Creek Chufasucker
Rock BassWhite Sucker
Creek Chubffllacknose Dace
White Sucker
13.8
23.2
31.7
21.5
13.7
16.0
29.0
19.2
18.9
22.2
13.1
25.2
22.3
27.6
17.5
25.1
14.1
18.9
18.3
16.3
18.4
16.5
19.2
19.7
17.3
17.1
15.4
18.2
20.3
16.1
18.1
17.2
8.6
8.1
8.2
8.8
7.8
8.2
8.2
8.0
8.2
9.0
8.3
8.4
6.1
9.1
7.9
8.5
89
88
89
84
71
77
88
88
89
76
81
80
86
68
76
77
2
20
14
9
12
14
9
19
21
8
11
14
47
12
9
20
3.2.3. Use of the Cluster/Discriminant Fish Assemblage Tool
1) At the model execution screen (see Figure 3.2), input the stream and watershed characteristics for the
stream of interest, then click the "Calculate" button.
The "Estimated Prediction Accuracy" (EPA) that appears indicates the overall success rate for predicting
a stream's actual cluster using our EMAP dataset. A successful prediction was defined to be when the
stream's actual cluster was one of the 3 most likely assemblages predicted by the model.
Above the species list of each assemblage is the Relative Score for that assemblage. As opposed to the
EPA, the Relative Score of the assemblages will change as a user changes parameter values. This
numerical score gives an indication of the relative likelihood of each predicted assemblage. The Relative
Scores are calculated by dividing the discriminant score for each assemblage by the largest discriminant
28
-------
score (i.e., the discriminant score for the most likely assemblage). These scores are scaled so that the
most likely assemblage always receives a Relative Score of 1.0 and the least likely assemblage receives a
Relative Score of 0.0. If the second and third most likely assemblages have Relative Scores near 1.0, that
should lead the user to conclude that any of the three could easily be correct. However, if the second
and/or third most probable assemblages have Relative Scores much less than the most probable
assemblage, then the user would have confidence that the most probable assemblage is much more likely
to be correct.
2) For the predicted assemblages, note the actual species and the relative dominance of each within the
assemblage. The % of total fish biomass for each species is given in parentheses. If you see Bluehead
Chub (33.3) in one of the assemblages, it means the biomass of Bluehead Chub on average comprises
33.3% of the total fish biomass in streams from this cluster.
3) If examining the influence of stream restoration/degradation is desired, change the characteristics of
the stream in accordance with the predicted outcomes of best management practices, stream restoration
efforts, or environmental degradation and rerun the analysis. This could include reducing/increasing the
% forested area in the watershed, or the values of stream slope and/or mean depth. After the desired
changes are made, recalculate the most probable fish assemblages and note differences.
If changing the stream and watershed characteristics produced a new set of most probable fish
communities, note that the time scale over which the fish assemblage could be expected to change cannot
be predicted. This process could take months, years, or decades, depending on how different the new
communities are from the originals, whether or not species would be stocked into the stream, and rates of
natural immigration from other locations. If the species of the assemblage were not predicted to change,
but only the relative dominance of species within the assemblage, the change process would likely be
faster. If an assemblage of entirely new species were predicted, one could expect this process to be
relatively much slower.
3.3. Intended Audience
We envision this tool being used by a wide diversity of stakeholders, from private landowners and public
interest groups to municipal planners and developers to environmental management professionals. One
29
-------
goal would be to predict fish communities in streams for which basic watershed and stream
characteristics are known, when actual sampling of the stream is cost prohibitive. Users could also
investigate potential impacts of hypothesized environmental restoration/degradation scenarios by altering
stream and watershed characteristics, then noting subsequent changes in the predicted fish community.
For researchers, this tool's basic fish community information can be passed to more complex,
mechanistic fish community models that can examine the effects of specific stressors on stream fish
communities.
Figure 3.2 Software tool for predicting stream fish assemblages based on stream and watershed characteristics.
Jr
I C'jjjjjjjlijiJiy C!lJj!3j"ijj:-i Turj!
HOGUE CREEK, VA, community: Smallmouth, condition: fair.
j Please provide input data and click the 'Calculate' button to generate a result set. Saving a trial will preserve the inputs for a specific calculation.
, Estimated Predication Accuracy = 0.56 , I
Remotely Sensed Stream Characteristics
Latitude
^J I _2_j Trial 8 1
On-Site Stream Characteristics
Longitude
Elevation (rn)
% Disturbed
[watershed]
Stream Flow (cfs) JO
% Fines [22"
Mean Stream Slope J3.51
Watershed Area (ha) j
Mean Stream Width (m) fTtfT
Temperature (C) J27.1
Thalweg Depth (cm) [595
Dissolved Oxygen (mg/l) U 7
Dominant Species Threshold (%} [g
Results r Compare Trials
Save Trial
Calculate
Note: On-Site Stream Characteristics should
represent means during the period April-July.
Display Scientific Names Cluster for model run:
Actual
Actual Community
smallmouth_bass
rock_bass
mottled_sculpin
central_stoneroller
longnose_dace
redbreast_sunfish
bluntnose_minnow
creek_chub
northern_hogsucker
longear_sunfish
fallfish
creek_chubsucker
goldenjedhorse
Most Probable Assemblage
Relatives core-1
Creel Chub|46|
Blacknose Dace (4.5)
Rock Bass (1.9)
Northern Hogsucker (1.5)
Slimy Sculpin (1.4)
Brown Trout (1.3)
Brook Trout (1.2)
Longnose Dace (1.1)
Second Most Probable Assemblage Third Most Probable Assemblage
Relative Score = .9803
Roc! Bat: 11291
White Sucker (5. 6)
I Northern Hogsucker (5.6)
Stoneroller (3.6)
Walleye (2.6)
Largemouth Bass (2.5)
RedbreastSunfish(2.1)
Silver Redhorse (2.1)
Creek Chub (1.9)
Flathead Catfish (1.9)
Fallfish (1.9)
Sauger(1.5)
Yellow Bullhead (1.5)
BlackJurnprock (1.3)
Torrent Sucker (1.2)
(Bluntnose Minnow (1.2)
Relative Score = .9633
Rod Bar: 1162)
White Sucker I62I
|SmallmouthBass(5.1)
I Stoneroller (5)
I Creek Chub (4.1)
I Bluntnose Minnow (3.7)
I Longnose Dace (2.9)
I River Chub (2.6)
lBlueheadChub(2.5)
I Blacknose Dace (2.2)
I Redbreast Sunfish (2.1)
I Fallfish (1.3)
I Golden Redhorse (1.7)
lBlacUumprock(1.2)
I Spotted Bass (1.1)
i Yellow Bullhead (1)
Previous
Next
Cancel
30
-------
4. Habitat Suitability Tool
4.1. Introduction to habitat suitability assessment
The suitability of instream physical habitat for particular fish species has been assessed using a variety of
quantitative methods. One of the earliest approaches to assess habitat suitability was through the use of
habitat suitability index (HSI) models. These index models are composite scores of the suitability of
multiple habitat variables. For each of the habitat variables, suitability ranges from 0 (unsuitable) to 1
(fully supporting of the species). These models are based on the assumption that there is a positive
relationship between the suitability index and habitat carrying capacity (USFWS 1981). A similar
approach is the development of guild-based habitat suitability criteria, which have been used to represent
species groups, or guilds, that utilize similar habitats in similar manners (Leonard and Orth 1988,
Aadland 1993, Vadas and Orth 2001). In a recent review of predictive habitat distribution models,
Guisan and Zimmermann (2000) reported that multiple regression models are a very popular approach to
predict habitat distribution, and that neural networks, ordination and classification, and Bayesian
methods are also used. All of these techniques have successfully been applied to stream fish (Lefwich et
al. 1997, Mastrorillo et al. 1997, Rieman et al. 2001). Here, multiple logistic regression was used to
develop quantitative relationships between instream habitat variables and the presence or absence of
selected fish species. Although we initially considered an HSI approach, few species known from the
MAH had developed HSIs (McCormick et al. 2001). While other techniques, in particular neural
networks, may give higher correct percentages, logistic regressions perform reasonably well in
comparison (Olden and Jackson 2002) and may be easier to interpret.
In general, studies relating stream fish occurrence and habitat quality have found that such relationships
exist at multiple scales, including reach, watershed, and landscape scale (Angermeier et al. 2002 , and
references therein). In this analysis, only reach-scale habitat variables were used as predictors, because
recent analyses of fish data in this region showed that little differences in fish communities occurred
across regional-scale variables of physiographic provinces, drainage basins, and other geographic
features (McCormick et al. 2000, McCormick et al. 2001). Local-scale habitat variables that are most
often included in habitat suitability studies are instream cover, substrate, and physical characteristics
(e.g., Edwards et al. 1983, Freeman et al. 1997, Barbour et al. 1999). For example, Lefwich et al. (1997)
31
-------
found that the presence of a darter species in Virginia could be predicted by multiple logistic regression
models that included elevation, depth, width, and substrate particle size. Wilson and Belk (2001) found
that the occurrence of the leatherside chub in the Great Basin of the western United States was related to
water velocity, water depth, and substrate composition. Guay et al. (2000) used a logistic model that
included water depth, current speed, and substrate composition to predict the distribution of Atlantic
salmon in a Quebec river. Habitat variables such as these were chosen for this analysis.
4.2. How suitable is my stream for specific fish species?
4.2.1. Methodology
The fish and habitat data used in this analysis were produced by the USEPA EMAP program for MAH
surface waters during 1993-6 (http://www.epa.gov/emap/html/datal/surfwatr/data/ma9396.html). A
description of the methods and quality assessment of the EMAP habitat data is given in Kauffman et al.
(1999). Samples outside of the MAH were excluded from the dataset using a Geographic Information
Systems (GIS) approach: a GIS coverage of the boundary of the MAH region was overlaid on a map of
the sites, and only sites that fell within those boundaries were selected for use in the analysis.
Fish species were either assessed individually or in groups. The decision was made based on McCormick
et al. (2000), who identified dominant species for the MAH. All dominant species mentioned by
McCormick et al. (2000) were assessed individually; in general these were the most common species in
the dataset. Rarer species were grouped taxonomically, generally following the metrics developed by
McCormick et al.(2001). Grouping of species facilitates analysis because many of the species were rare,
and sufficient occurrences are needed to develop a robust model (Harrell 2001).
We used the set of samples from the habitat dataset for which fish collection data were also available. In
order to improve normality, percentage variables were arcsine-squareroot transformed, width and depth
variables were square-root transformed, and slope was log-transformed (Zar 1974). Boxplots were used
to identify extreme outliers (>3 interquartile ranges away from either the sample 25th or 75th percentiles)
(SAS 1989); samples with extreme outliers were deleted. Pearson correlation coefficients were calculated
among the variables, and only those that were correlated at |r|<0.80 were retained for analysis, in order to
reduce multicollinearity (Glantz and Slinker 1990). For each pair, one of the two variables was selected
32
-------
for use in the analysis based on a preliminary examination of their explanatory power in the logistic
regression models (Jongman et al. 1995). Samples with missing data were deleted. This resulted in a final
data set of 337 samples (Figure 4.1) and twenty habitat variables (Table 4.1).
Equations to predict the presence of multiple fish species/groups in the MAH from instream habitat data
were developed using multiple logistic regression analysis with backward stepwise selection in SAS
(SAS 1989). Logistic regression is a statistical technique to predict or model a categorical response
variable from one or more continuous explanatory variables. The categorical response variable of fish
presence-absence was used because it is more robust to sampling biases than measures of densities
(Green 1979). The default of binary logit model with Fisher's scoring for the optimization technique was
used (SAS 1989). A significance value of 0.05 was used for retaining habitat variables into the model.
Table 4.1 Instream habitat measures that were selected from the EMAP dataset as possible explanatory variables for the presence
of fish species in the Mid-Atlantic Highlands region.
Variable
Boulder
Coarse gravel
Cobble
Depth
Fine gravel
Fine sediments
Glide
Overhanging
Pool
Riffle
Rip ground
Rip layers
Sand
SD_depth
Slope
Slow
Temp
Undercut
Width
Woody
EMAP Code
XFC RCK
PCT BIGR
PCT CB
XDEPTH
PCT GF
PCT FN
PCT GL
XFC OHV
PCT POOL
PCT RI
XG
XPCMG
PCT SA
SDDEPTH
XSLOPE
PCT SLOW
TEMP FLD
XFC UCB
XWIDTH
XFC LWD
Habitat measure (units)
Boulder/rock ledge cover (%)
Coarse gravel substrate (>16 mm) (%)
Cobble substrate (64-240 mm) (%)
Mean thalweg depth (cm)
Gravel substrate (2- 16mm) (%)
Fine sediment, i.e., silt/clay/muck (%)
Glide habitat (%)
Overhanging vegetation cover (%)
Pool habitat (%)
Riffle habitat (%)
Riparian vegetation (ground layer) (%)
Riparian vegetation (3 layers) (%)
Sand substrate(0.6-2mm) (%)
Standard deviation of depth (cm)
Water surface gradient over reach (%)
Slow-water habitat (% glide and pool)
Temperature (C)
Undercut bank cover (%)
Mean wetted width (m)
Large woody debris cover (%)
Mean
18.3
62.76
26.5
34.09
9.7
12.8
36.3
11.9
14.5
42.3
56.0
80.3
13.0
15.9
1.58
50.8
17.8
4.9
8.40
3.5
S.D.
19.9
25.38
19.2
19.07
9.9
15.4
24.7
13.1
19.3
23.3
24.4
26.4
16.1
9.3
1.72
25.9
4.4
6.5
8.75
5.2
Min
0
0
0
3.77
0
0
0
0
0
0
20
0
0
2.4
0.055
0
7.8
0
0.43
0
Max
87.5
100
100
98.59
54.6
100
100
79.3
99
100
100
100
87.3
65.7
11.25
100
30.1
41.8
51.75
38.2
The outputs of each logistic regression model included habitat variables retained in the model, maximum
likelihood estimates and significance values for the coefficients, a test of the significance of the residuals,
33
-------
and goodness-of-fit statistics for the overall model. Goodness of fit statistics included the Hosmer and
Lemeshow test, for which a low significance value indicates a poorly fit model (Hosmer and Lemeshow
2000), and the percent of samples correctly classified. This percentage was derived via leave-one-out
cross-validation, that is, each sample was sequentially left out, model parameters were reestimated, and
the predicted probability of presence (the species/group was assumed to be present at a predicted
probability ofp>0.5) was compared to the observation for that sample. The probability of occurrence (p)
of each species at a particular site was calculated by the logit equation
p =
x
1 + exp(- r)
where r denotes the species fitted logistic regression equation, i.e.,
r = 10 + E ltXt (4-2)
1= i
See SAS (1989). Assuming that the probability of a species occurrence at a site is representative of that
site's habitat suitability for that fish species, the aforementioned equations and fitted regressions were
then used to develop an interactive computer tool for analyzing stream habitat suitability for eighteen fish
species and higher taxonomic groups.
4.2.2. Results and Discussion
Wald chi-square results indicated that all of the overall logistic regression models to estimate the
probability of presence of MAH stream fish species/groups were significant (p<0.05) (Table 4.2).
Hosmer and Lemeshow tests and residuals tests were nonsignificant (p>0.05) for all models except for
suckers (p=0.0023). Our results for the percent correctly predicted are in the range of 65-81%, and half of
the models are >75%, which has been suggested as acceptable model performance for managers and
researchers (Hurley 1986). Moderate to high values occurred for both sensitivity (the ability to predict an
occurrence correctly) and specificity (the ability to predict an absence correctly) (Table 4.2). High values
for both sensitivity and specificity are desirable. Low sensitivity, which occurs for more rare species,
implies that it will be more difficult to predict the occurrence of organisms whose conservation may be
most critical (Olden and Jackson 2002). Sensitivity can be increased by lowering the threshold at which
34
-------
presence is assumed (Fielding 2002), which may be desirable for certain applications, such as the
identification of sites for species relocation.
Figure 4.1 Location of sample sites within the Mid-Atlantic Highlands region.
300 Kilometers
These results indicate that instream physical habitat does have significant relationships with the presence
of fish species in MAH streams, although there may be other factors influencing the presence of fishes,
including population dynamics, species interactions, historical factors, barriers to movement, and spatial
autocorrelation. Species interactions, including predation, can affect presence/absence. However, the
majority of the sites were relatively small and predation is of less importance in smaller streams
(Grossman et al. 1998). Barriers to movement can cause species absence even when suitable habitat
exists, and this may be more significant for poor dispersers (Angermeier et al. 2002). Although spatial
autocorrelation, that is, patchiness due to factors other than habitat characteristics, can influence models
(Guisan and Zimmermann 2000), McCormick et al. (2000) concluded that large-scale patchiness was not
a factor in explaining fish species occurrences. Finally, there may be imperfect detection of presence in
the streams due to sampling inefficiencies(MacKenzie etal. 2004). Logistic regression models are
limited in that they assume a linear response of the fishes to habitat variables; this is possibly a limiting
assumption, since nonlinear responses also occur. Because of these considerations, the models should be
used cautiously, and not considered to be absolute predictors of presence/absence. However, they can be
35
-------
a useful way to conduct assessment of stream ecosystems (Oberdorff et al. 2001).
Table 4.2 Results from logistic regression models of presence of Mid-Atlantic Highlands stream fish species/groups. Results
include the number of occurrences in the data set (N) and goodness-of-fit statistics for overall logistic regression models: the
Wald Chi-square value and its significance (P), and the percentage of correct classifications determined using cross-validation (a
higher value indicates greater predictive ability).
Species/
group name
black bass
blacknose dace
brook trout
brown trout
catfish
chub
creek chub
dace
darters
longnose dace
northern
hogsucker
rock bass
sculpin
shiners
stoneroller
suckers
sunfish
white sucker
Species within group
Micropterus sp.
Rhinichthys atratulus
Salvelinus fontinalis
Salmo trutta
Ameiurus sp., Ictalurus sp.,
Noturus sp., Pylodictus sp.
Nocomis sp., Semotilus, sp.,
Exoglossum sp.
Semotilus atromaculatus
Margariscus sp., Phoxinus sp.
Etheostoma sp., Percina sp.
Rhinichthys cataractae
Hypentelium nigricans
Ambloplites rupestris
Cottus sp.
Notropis sp.
Campostoma anomalum
Hypentileum sp., Moxostoma sp.
Lepomis sp.
Catostomus commersoni
N
103
242
76
92
96
272
200
81
227
137
125
110
161
100
160
152
154
186
Wald Chi-
square [df]
73.52 [5]
45.39 [5]
45.70 [4]
32.05 [6]
59.33 [4]
47.79 [4]
56.03 [5]
40.61 [4]
71. 09 [7]
62.80 [10]
77.98[5]
67.95 [5]
36.57 [4]
70.56 [5]
59.15 [7]
76.73 [6]
72.30 [9]
60.48 [6]
P % correct Sensitivity Specificity
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
O.OOOl
81.3
76.3
80.1
71.2
76.3
83.4
70.0
76.3
78.9
70.9
81.4
76.6
65.3
81.0
68.8
81.6
70.9
73.3
59.2
93.4
26.3
14.1
42.7
96.7
80.5
9.9
89.4
57.7
71.2
52.7
59.0
57.0
63.8
78.3
64.9
81.2
91.0
32.6
95.8
93.1
89.6
27.7
54.7
97.3
57.3
80.0
88.2
88.1
71.0
91.1
73.4
84.3
76.0
63.6
Although different fish species/groups responded to different habitat variables, over half the
species/groups analyzed responded to stream temperature, slope, and/or width (Table 4.3). Twelve
species/groups demonstrated significant responses to stream slope. All of these species/groups (i.e., black
bass, brown trout, central stoneroller, chubs, creek chubs, darters, longnose dace, northern hogsuckers,
rock bass, sculpins, suckers, and white suckers) responded negatively to this variable. This finding is
likely due to the fact that this region is mountainous and steep mountain streams provide flashy, unstable
habitats. Eleven species/groups demonstrated significant responses to stream width. Whereas seven
species/groups (i.e., black bass, central stonerollers, darters, northern hogsuckers, rock bass, shiners, and
suckers) responded positively to this variable, four species/groups (i.e., blacknose dace, creek chubs,
36
-------
daces, and white suckers) responded negatively. Eleven species/groups also demonstrated significant
responses to stream temperature. Whereas eight species/groups (i.e., black bass, catfish, central
stonerollers, chubs, creek chubs, northern hogsucker, suckers, and sunfish) responded positively to this
variable, three species/groups (i.e., brook trout, brown trout, and sculpins) responded negatively.
A third half or more of the species/groups analyzed responded to stream depth, large gravel cover, and
riparian cover (Table 4.3). Six species/groups responded significantly to stream depth. Whereas creek
chubs displayed a negative respond to this variable, the remaining species (i.e., catfish, daces, longnose
dace, sunfish, and white suckers) responded positively to increasing stream depth. Vadas and Orth (2001)
found that depth was the most important factor in habitat selection for seven fish guilds in the Roanoake
River drainage in Virginia. Central stonerollers, chubs, longnose dace, and white suckers responded
positively to the presence of large gravel stream bottoms whereas rock bass, shiners, and suckers
responded negatively to this variable. Eight species/groups showed significant positive responses to
increasing riparian cover metrics. In particular, blacknose dace, brown trout, catfish, darters, longnose
dace, and shiners responded to riparian ground cover, and darters, suckers, and sunfish responded to
either overhanging or multilayer riparian vegetation.
Fishes responded to the cover metrics, and the response was generally positive, although some fish
species showed a negative relationship with undercut bank cover. This is surprising, since fishes can use
undercut bank habitat as cover and refuge. However, undercut banks can also occur as a result of landuse
development and physical habitat alteration and these stressors can have a negative effect on fish species
in streams. Large gravel and sand substrate species responded to multiple larger substrates, which
provide for cover and reproduction; a mix of substrates may be most favorable. Many fish species/groups
responded positively to riparian vegetation, which slows water flow, traps sediment and other pollutants,
and stabilizes streambanks.
Fish species/groups responded individually to habitat variables, so management activities that alter these
variables will favor different species in different ways (e.g., slow water habitat). However, most fish
species/groups are predicted to benefit from the maintenance of flow volume, large gravel substrate, and
riparian vegetation.
37
-------
Table 43 Parameter estimates and odds ratios from logistic regression models of presence of Mid-Atlantic Highlands stream fish
species/groups.
Species
Group
black bass
blacknose
dace
brook trout
brown trout
catfish
Variable
Intercept
Cobble
Sand
Slope
Temp
Width
Intercept
Glide
Pool
Rip ground
Slow
Width
Intercept
Boulder
Glide
Temp
Woody
Intercept
Cobble
Fine gravel
Rip ground
SD_depth
Slope
Temp
Intercept
Depth
Rip ground
Slow
Temp
Coefficient
-7.13
2.00
2.50
-3.17
0.13
1.12
1.50
4.97
4.10
1.69
-6.24
-0.47
0.98
1.62
-1.33
-0.15
2.35
-2.74
2.62
2.07
1.10
0.34
-2.49
-0.10
-8.45
0.67
1.40
-1.34
0.19
P-value
O.OOOl
0.0082
0.0007
0.0024
0.0005
O.OOOl
0.0074
0.0016
0.0021
0.0012
0.0002
O.OOOl
0.1565
0.0033
0.0135
O.OOOl
0.0370
0.0097
O.OOOl
0.0163
0.0371
0.0126
0.0030
0.0040
O.OOOl
O.OOOl
0.0083
0.0127
O.OOOl
Species Variable
Group
central Intercept
stoneroller Coarse gravel
Glide
Pool
Slope
Temp
Width
chub Intercept
Coarse gravel
Sand
Slope
Temp
creek chub Intercept
Depth
SD_depth
Slope
Temp
Width
daces Intercept
Depth
SD_depth
Width
Woody
darters Intercept
Boulder
Rip ground
Rip layers
Sand
Slope
Width
Woody
Coefficient
-4.10
1.96
1.13
1.57
-2.79
0.07
0.41
-3.03
2.39
1.73
-4.33
0.21
2.20
-0.50
0.56
-4.02
0.10
-0.57
-0.27
0.74
-0.5
-1.13
-3.01
-3.38
-1.9
2.00
0.99
1.69
-2.23
1.19
-3.55
P-value
O.OOOl
0.0003
0.0472
0.0034
0.0008
0.0393
0.0040
0.0042
0.0007
0.0387
O.OOOl
O.OOOl
0.0121
0.0036
0.0048
O.OOOl
0.0016
0.0017
0.6237
O.OOOl
0.0271
O.OOOl
0.0124
0.0012
0.0052
0.0007
0.0157
0.0157
0.0152
O.OOOl
0.0049
4.2.3. Interactive Tool
The predictive habitat models developed here have been incorporated into an interactive software tool
that outputs changes in the probability of occurrence for these fish species under various habitat
management scenarios (Figure 4.2). The interface provides a list of the fish species from which the user
38
-------
can select one or more by checking the associated checkboxes; habitat variables needed to predict the
suitability for selected fish species are highlighted and variables not needed are grayed-out. The tool
allows the user to vary the instream habitat parameters within their specified ranges. The user can then
recalculate the habitat suitability scores as a separate trial. If the user specifies a value that is out of
range, an error message is generated in the message field that indicates the accepted range.
Table 4.3 Continued.
Species
Group
longnose
dace
northern
hogsucker
rock bass
sculpin
shiners
Variable
Intercept
Coarse gravel
Depth
Fine gravel
Fine sediments
Glide
Pool
Rip ground
Slope
Slow
Intercept
Cobble
Sand
Slope
Temp
Width
Intercept
Slope
Slow
Width
Intercept
Fine sediments
Sand
Slope
Temp
Intercept
Coarse gravel
Cobble
Rip ground
Sand
SD_depth
Width
Coefficient
-8.02
3.14
0.51
2.11
1.67
5.00
2.99
1.57
-2.53
-6.65
-6.43
2.39
3.09
-5.32
0.13
1.09
-3.90
-2.99
1.70
0.98
3.70
-2.11
-2.23
-2.15
-0.10
-4.64
-1.92
1.48
1.89
2.78
0.37
0.57
P-value
O.OOOl
0.0001
O.OOOl
0.0315
0.0457
0.0143
0.0463
0.0036
0.0060
0.0024
O.OOOl
0.0018
O.OOOl
O.OOOl
0.0003
O.OOOl
O.OOOl
0.0023
0.0019
O.OOOl
O.OOOl
0.0001
O.OOOl
0.0011
0.0004
O.OOOl
0.0008
0.0224
0.0003
O.OOOl
0.0327
0.0015
Species Variable
Group
suckers Intercept
Coarse gravel
Cobble
Overhanging
Sand
Slope
Slow
Temp
Width
sunfish Intercept
Coarse gravel
Depth
Fine sediments
Glide
Rip layers
Sand
Slow
Temp
white sucker Intercept
Coarse gravel
Depth
Fine sediments
Sand
Slope
Width
Coefficient
-5.26
-1.78
2.06
1.95
2.99
-3.16
-1.27
0.11
1.29
-10.58
2.50
0.20
3.36
-1.79
0.76
3.04
3.49
0.14
-4.96
3.84
0.47
3.05
2.48
-4.59
-0.42
P-value
O.OOOl
0.0162
0.0058
0.0149
O.OOOl
0.0011
0.0473
0.0042
O.OOOl
O.OOOl
0.0260
0.0447
0.0015
0.0186
0.0352
0.0043
O.OOOl
0.0002
0.0011
0.0006
0.0010
0.0027
0.0154
O.OOOl
0.0185
39
-------
Figure 4.2 Interface for the Habitat Suitability Tool in the CVI Watershed.
J5I;
Habilol SuitaWt; Index Calculator
TUSCARGRA GREEK, PA, community; NorvGame, condrtion, good,
Select one or teas fish, oonptele the fields, and and click 'Cafctilala' to calculate hahiat sutabittj) lot the selected site.
The fish h Ihe list haw letters nerf to Ihnm. These mdicafe retfijred iietkfe.
Fish:
Channel slope [X|
Mean deptfi [cml
Mean width [ml
16ntm). X:
JO, 5
27.1
|63.1
10.9
Riffle en [12
Flow (m3/5se): |: H
Temperature (C) J24.6
Side PoolJ:
Large woo^1 debii* (St):
Bw* and small debii* (X]:
Overhanging vegelafon 1%}:
:l2;r-3
Streamside' vegatation [S]
SaveTiial
J^^i^M^
Ciculso
I
j
K
L
M
N
0
How might suitability be improved for these species?
The Habitat Suitability tool allows the user to examine how habitat suitability for particular species or
groups offish changes in response to changing habitat characteristics, that is, the user can alter any of the
input habitat variables and recalculate the score. The models will be most reliable for smaller levels of
change and with reasonable combinations of variables. Additionally, three scenarios for Best
Management Practices (BMPs), based on best professional judgement and information from the literature
40
-------
(e.g., Rosgen 1996), are provided as options to be applied at a particular stream site (Figure 4.3, Table
4.4). The user can select the level of percent change that occurs in the scenario. The results should be
interpreted in a comparative fashion: higher values for the probability of occurrence indicate a higher
likely suitability for that species.
Table 4.4 Scenarios of Best Management Practices (BMPs) provided within the CVI Habitat Suitability tool.
BMP scenario
Increases
Decreases
Riparian Zone Restoration
Bank stabilization (instream)
Natural stream channel design
Large woody debris
Small woody debris
Overhanging vegetation
Percent coarse gravel
Percent riffle
Width
Temperature
Percent fine sediment
Percent glides and all pool types
Models that predict the presence of stream fish species based on habitat characteristics can be useful in
watershed management: they allow managers and researchers to understand the link between physical
habitat alteration and fish community condition. Such models can be used to assess the effectiveness of
management actions. For example, management actions that may be undertaken in the MAH to improve
physical habitat for stream fish include protecting existing riparian vegetation, restoring stream bank
vegetation, and restoring streams to more natural flow regimes (USEPA 2000). The models can also be
used to aid in species conservation (Angermeier et al. 2002). For example, the models can inform
management about which habitat factors should be considered in conservation and restoration activities
for particular species. Most often endangered species are endangered due to habitat loss. By focusing on
conservation of specific habitats and habitat characteristics favored by species at risk, such species can
be helped before they become endangered. We anticipate that the Habitat Suitability approach will also
be useful as a tool for optimization of restoration efforts in subwatersheds. A next step would be to
couple these results with economic analyses, specifically to relate cost estimates to BMP activities and
associated ecological benefits, in order to assess trade-offs in management and restoration (Holmes et al.
2004). Such analyses can lead to more comprehensive planning and decision-making for watersheds.
41
-------
Figure 4.3 Interface for Best Management Practice (BMP) scenarios in the CVI Habitat Suitability Tool.
BMP Adjustment
Instructions: Choose a BMP option and move the ;lider to Hie right increment
Click OK to applji the BMP.
BMP 0ptiore: I Bank Stabization (in stream]
Decreases:
% fine
Increases:
4 big grave!
25
Caned
42
-------
5. BASS: a Simulation Model for Fisheries Management
5.1. Using Simulation Models for Fisheries Management
Some of the most important assessment questions related to fisheries management in the MAH can be
summarize as follows:
• How will the destruction of riparian buffers impact the expected body sizes and
population abundances of recreationally important fish species such as trout and
smallmouth bass?
• How will the destruction of riparian buffers impact the expected body sizes and
population abundances of non-game fish species such as darters, suckers, and other fish
of special concern?
• How will the restoration of riparian buffers improve the expected body size and
population abundances of recreationally important fish species such as trout and
smallmouth bass?
• How will the restoration of riparian buffers improve the expected body sizes and
population abundances of non-game fish species such as darters, suckers, and other fish
of special concern?
• How can trout stocking programs be managed to provide anglers with abundant catches
of moderately sized fish for consumption, as well as sufficient numbers of trophy size
fish?
• How can trout stocking programs be managed to minimize the almost unavoidable
negative impacts on native non-game fish species due to competition and predation?
Process-based models that describe
43
-------
• the expected growth and trophic dynamics,
• the spawning and recruitment patterns, and the effects that
• the interactions of physical habitat and water quality have on the feeding, metabolism,
reproduction, survival, and dispersal
of the community's dominant fish species are perhaps the most obvious tools that fisheries managers
need to address these and other related issues.
Although several excellent fish bioenergetic models were available that could be used to analyze the
above assessment questions, the Bioaccumulation and Aquatic System Simulator (BASS) was chosen for
this purpose in the CVI Watershed Toolkit. BASS is a general and extremely flexible Fortran 95 model
that simulates not only fish chemical bioaccumulation but also individual and population growth
dynamics of age-structured fish communities using a temporal and spatial resolution of a day and a
hectare, respectively. Because a species' age class can be specified as either a month or a year, users can
readily simulate small, short-lived species, such as daces and minnows, along with larger, long-lived
species such, as suckers, bass, perch, sunfishes, and trout. The community's food web is specified by
defining one or more foraging classes for each fish species based on either body weight, body length, or
age. The dietary composition of each of these foraging classes is then specified as a combination of
benthos, incidental terrestrial insects, periphyton, phytoplankton, zooplankton, and/or other fish species,
including its own.
Although BASS was originally developed to simulate the bioaccumulation of chemical pollutants within
a community or ecosystem context, it is also suited for simulating population and community dynamics
of fish assemblages that are exposed to a wide variety of nonchemical stressors. In particular, BASS can
be easily setup to simulate the population and community dynamics offish assemblages that are
subjected to altered thermal regimes associated with riparian and hydrological alterations. BASS can also
be used to simulate the population and community dynamics offish assemblages that are subjected to
introductions of exotic species or stockings of recreational sport fishes.
Notable capabilities of BASS include:
• there are no restrictions to the number of chemicals that can be simulated;
44
-------
• there are no restrictions to the number of fish species that can be simulated;
• there are no restrictions to the number of cohorts that fish species can have; and
• there are no restrictions to the number of feeding/foraging classes that fish species can
have.
Model output includes:
• summaries of all model input parameters and simulation controls;
• tabulated annual summaries of the bioenergetics of individual fish by species and age
class;
• tabulated annual summaries of the chemical bioaccumulation within individual fish by
species and age class;
• tabulated annual summaries of the community level consumption, production, and
mortality of each fish species by age class; and
• plotted annual dynamics of model variables, as requested by the user, as a function of
fish species age or size classes.
5.2. Model Description
To understand how BASS can be used to address the assessment questions formulated previously, it is
useful to review BASS' basic model structure that solves the following system of differential equations
for each "year" class or cohort offish:
dW,
d- = F- E- R- EX- SDA (5-2)
dt
— =- EM- NM- PM (5_3)
45
-------
In these equations, B and Wd denote the chemical contaminant body burden (fig / fish) and dry body
weight (g[nw] / fish) of the average individual within the cohort, respectively, and TV denotes the cohort's
population density (fish / ha). In Equation (5-1) J and Ji denote the net chemical exchange across a fish's
gills from the water and across its intestine from food, respectively, and/m denotes the chemical's
biotransformation or metabolism. In Equation (5-2) F, E, R, EX, and SDA denote the fish's feeding,
egestion, routine respiration, excretion, and specific dynamic action (i.e., the additional respiratory
expenditure in excess of R required to assimilate food), respectively. Although many physiologically
based models for fish growth are formulated in terms of energy content and fluxes (e.g., kcal / fish and
kcal / day), formulating a physiologically based growth model in terms of dry weight is fundamentally
identical to the former, since the energy densities of fish depend on their dry weight (Kushlan et al. 1986,
Hartman and Brandt 1995). Finally, in Equation (5-3) EM, NMand PM denote the cohort's emigration
(i.e., dispersal), non-predatory, and predatory mortality, respectively. Although immigration can be a
significant process in determining population sizes, this process is not presently modeled in BASS.
Because cohort recruitment and fishery stockings are treated as boundary conditions, the right-hand side
of Equation (5-3) does not require a term to address these processes. Though it may not be immediately
apparent from the symbols and notation used, these equations are tightly coupled to one another. For
example, the cohort's realized feeding depends on the availability (i.e., density and biomass) of suitable
prey. The fish's predatory mortality, in turn, is determined by the individual feeding levels and
population densities of its predators. Finally, the fish's dietary exposure to organic and metallic chemical
pollutants is determined by its rate of feeding and the levels of those contaminants in its prey.
Appendix B summaries how BASS models the mass fluxes in the above system of equations. However,
due to the focus of the assessment questions above, this discussion is restricted to only the cohort's
growth and population Equations (5-2) and (5-3). Readers interested in BASS's bioaccumulation
algorithms should see Barber (2001).
5.3. Developing Case Studies to Evaluate Mid-Atlantic Fisheries
In order to develop initial fishery management and riparian alteration assessment scenarios for the MAH,
18 EMAP surface water stream sites were chosen as case studies. These streams were selected such that
two streams are representative of typical trout, smallmouth, and non-game dominated communities
within each of the states of Pennsylvania, West Virginia, and Virginia. For each of the nine community
46
-------
type x state combinations, one stream was selected that had excellent or minimally disturbed riparian
vegetation, and one stream was selected that had poor or disturbed riparian vegetation. See Table 5.1.
Habitat suitability scores developed in Section 4.2 (i.e., Equations (4-1) and (4-2) and Table 4.3) were
used as habitat suitability multipliers on each species non-predatory mortality and dispersal within these
streams as discussed in Appendix B (see Section B.5 Equation (B-38)). After simulating the nominal
community dynamics of the 18 case study streams, the expected effects of riparian destruction or
restoration were investigated by recalculating each species habitat suitability multiplier assuming a 25%
decrease or increase in the streams' associated riparian ground cover and multilayer vegetation.
Table 5.1 EMAP fish sites selected as fisheries and riparian alteration case studies.
Stream, state
EMAP id
Community
Impactedness
Powdermill Run, PA
Falling Spring, PA
North River, VA
Beaver Creek, VA
Little B lack Fork, WV
South Br Wolf Run, WV
Allegheny Creek, PA
Little Tenmile Creek, PA
Calfpasture River, VA
Hogue Creek, VA
Clifford Hollow, WV
Dillons Run, WV
Kettle Creek, PA
Bell Run, PA
Dunnavant Creek, VA
Flat Creek, VA
Brake Run, WV
Tuscarora Creek, WV
pa531s_1993.1
par01s_1993.1
maia97-132_1997.1
va806s_1994.1
wv774s_1994.1
wv773s_1994.1
maia97-101_1997.1
maia97-028_1997.1
va754s_1994.1
maia97-052_1997.1
wv750s_1997.0
wvr03s_1993.1
maia97-081_1997.1
pa523s_1994.1
maia97-137_1997.1
maia98-115_1998.0
maia97-019_1997.1
wvrOls 1993.1
trout marginal
trout moderate
trout marginal
trout moderate
trout marginal
trout moderate
smallmouth marginal
smallmouth moderate
smallmouth marginal
smallmouth moderate
smallmouth marginal
smallmouth moderate
non-game/northern hogsucker marginal
non-game/northern hogsucker moderate
non-game/bluehead chub marginal
non-game/bluehead chub moderate
non-game/blacknose dace marginal
non-game/blacknose dace moderate
Because trout stocking is one of the most important and widespread fisheries management practices in
the region, a default stocking scenario was developed for application to each of the 18 streams. This
stocking scenario assumes that brook, rainbow, and brown trout are stocked at the rates of 200, 100, and
50 fish/ha, respectively, once a month from February through May. Whereas the stocking length of brook
47
-------
and rainbow trout was assumed to be approximately 7 inches, brown trout were assumed to be 12 inches.
Annual fishing mortality (i.e., harvest) for each trout species was set at 90%.
5.3.1. Developing Initial Conditions
The species compositions of the 18 case study stream communities were assigned to be the most
abundant species in terms of calculated biomass (g[pw]/m2) such that the relative biomasses of the
selected species summed to least 95%. This selection criteria resulted in an average number of species
per stream equal to 5.8.
Species biomasses and initial cohort densities were calculated from EMAP fish data by first converting
the reported species counts into species densities (fish/m2) using the reported stream segment length and
mean width. Total species densities were then converted into a vector of cohort densities using the self-
thinning algorithm described in Appendix A. Initial live body weights of each cohort were than estimated
using the growth rates summarized in Table A.2. Finally, the product of each cohort's estimated body
weight and density were summed to estimate the species total biomass. When growth data were
unavailable for a given species, a default growth rate was assigned based on the species' genus. See
Table 5.2.
Macroinvertebrates biomasses were also estimated from EMAP surface water data. In particular reported
taxa densities were converted into biomasses using the conversion factor summarized in Table 5.3.
5.5.2. Parameterization of BASS Physiological and Ecological Processes
A Fortran 95 fish community program was developed to generate BASS input files automatically while
estimating the species biomasses and cohort densities described in Section 5.3.1. Data required to define
diets and to assign reproductive parameters were taken from Carlander (1969, 1977, 1997), Jenkins and
Burkhead (1994), and Etnier and Starnes (1993). All fish species were assumed to feed on the
generalized prey categories of benthos, periphyton/attached algae, zooplankton/drifting invertebrates, and
fish in direct proportion to the biomasses of these prey as summarized by afore mentioned authorities.
For example, species that generally feed on benthic invertebrates and periphyton were assumed to have
feeding electivities for benthos and periphyton equal to zero. Similarly, species that generally feed on
48
-------
benthic invertebrates and fish were assigned feeding electivities equal to zero for benthos and all fish
species within the stream of interest. See appendix Section B.3 for details. Maximum daily consumption
of all species was modeled using BASS's linear feeding model (see Appendix B Equation (B-18)) and
the growth data and default parameterization assignments summarized in Table A.2 and Table 5.2.
Family-specific respiratory parameters (see Appendix B Equations (B-23) and (B-24)) were estimated for
all species from the OXYREF database (Thurston and Gehrke 1993) that can be downloaded from the
USEPA Center for Exposure Assessment Modeling web site at
http://www.epa.gov/ceampubl/oxyref.htm. All other physiological and ecological parameters were
assigned as interspecies means of data summarized in Barber (2004b).
Table 5.2 Summary of default species assignments for parameterizing BASS for MAH genera.
Default species
Genus
Alosa pseudoharengus
Ameiurus nebulosus
Etheostoma spp
Catostomus commersoni
Cottus cognatus
Etheostoma spp
Erimyzon spp
Hypentelium nigricans
Lampetra spp
Lepisosteus osseus
Lepomis macrochirus
Morone chrysops
Moxostoma spp
Semotilus atromaculatus
Notropis spp
Noturus miurus
Pimephales spp
Percina spp
Rhinichthys spp
Alosa, Dorosoma
Ameiurus, Ictalurus
Ammocrypta, Crystallaria
Catostomus
Cottus
Etheostoma, Psychromaster, Litocara, Allohistrium, Nanostoma,
Doration, Boleosoma, loa, Vallantia, Nothonotus, Fuscatelum,
Belophlox, Ozarka, Oligocephalus, Catonotus
Erimyzon
Hypentelium
Lampetra, Ichthyomyzon
Lepisosteus
Lepomis
Morone
Moxostoma
Exoglossum, Nocomis, Semotilus
Cyprinella,uxilus, Lythrurus, Notropis, Phoxinus, Margariscus,
Clinostomus
Noturus
Pimephales
Percina, Hadropterus, Swania, Alvordius, Ericosma,
Odontopholis, Cottogaster, Imostoma
Rhinichthys, Erimystax, Hybopsis, Phenacobius
49
-------
Table 5.3 Summary of conversion factors for EMAP macroinvertebrate data.
Group Average Weight
(mg dry weight)
Diptera 0.02944
Ephemeroptera 0.34833
Megaloptera 3.94963
Plecoptera 0.12037
Trichoptera 0.57713
Other Insect (primarily Coleoptera and Odonata) 0.14265
Oligochaetes (Annelida andNemotoda) 0.08900
Non-Insect (Amphipods, Cladocerans, Cyclopoids, Isopods, 0.18492
Ostracods, Hydracarina, Harpacticoids)
5.4. Community Responses to Riparian Alteration and Fisheries Management
As an illustration of the use of the BASS CVI Watershed Tool, this section summaries the predicted
community structure during the fifth year of a riparian alteration or fisheries management program for
the following streams:
Bell Run, PA Non-Game/ white sucker
Flat Creek, VA Non-Game/ bluehead chub
Tuscarora Creek, WV Non-Game/ blacknose dace
5.4.1. Responses of Non-game Streams to Riparian Restoration
Table 5.4 summarizes the predicted year 5 mean annual biomasses, community trophic flows, and
average interspecies HSIs for Bell Run, Flat Creek, and Tuscarora Creek with and without 25% riparian
restoration.
The total year 5 fish biomasses predicted for Bell Run, Flat Creek, and Tuscarora Creek without riparian
alteration or trout stocking are 47.5, 86.7, and 2.88 kg[rw]/ha, respectively. These estimated biomasses
are reasonably consistent with results of Randall et al. (1995) who reported average total fish biomasses
for lakes and rivers as 83.8 and 146.1 kg[rw]/ha, respectively. The total year 5 fish densities predicted
50
-------
for Bell Run, Flat Creek, and Tuscarora Creek under status quo conditions are 3130, 10900, and 2002
fish/ha, respectively. The mean interspecies HSIs for Bell Run, Flat Creek, and Tuscarora Creek are
0.757, 0.605, and 0.468, respectively.
Whereas the proportion of stream reach possessing a three-layer riparian cover varies from zero to 0.64
for Bell Run, Flat Creek, and Tuscarora Creek, the proportion of stream reach having riparian ground
cover varies from 0.35 to 0.87. When the proportions of riparian ground cover and three-layer riparian
cover are increased by 25%, the mean interspecies HSIs for Bell Run, Flat Creek, and Tuscarora Creek
increase to 0.797, 0.634, and 0.542, respectively. In other words, mean HSI scores increase by 5 to 15 %.
These restoration HSIs, however, do not include potential changes in fine sediments or water
temperatures that would be expected with increasing riparian vegetation cover (see for example Daniels
and Gilliam 1996, Blann et al. 2002). These restoration HSIs also do not consider the expected increases
to large and small instream woody debris (see Table 4.4). Although the total year 5 fish biomasses
predicted for Bell Run and Flat Creek after riparian restoration are essentially unchanged, the total year 5
fish biomass for Tuscarora Creek increases by 31% to 3.77 kg[rw]/ha. Similar trends are also predicted
for the total year 5 fish densities of these streams. These results are consistent with the fact that the HSIs
for the community dominants of Bell Run and Flat Creek (i.e., white suckers andbluehead chubs) are
independent of riparian cover metrics while the HSI for the community dominant of Tuscarora Creek
(i.e., blacknose dace) responds positively to increasing riparian ground cover (see Table 4.3).
The preceding analysis should only be considered only as the immediate benefits of riparian restoration.
Evaluation of the longer term benefits requires good empirical or process-based models for stream
temperature dynamics, fine sediment transport, and allochthonous inputs of woody debris and terrestrial
invertebrates. These models are planned for development during the next phase of the CVI-WHAT IF
software program.
5.4.2. Responses of Non-game Streams to Trout Stocking
Table 5.4 also summarizes the predicted year 5 mean annual biomasses, community trophic flows, and
average interspecies HSIs for Bell Run, Flat Creek, and Tuscarora Creek with and without trout stocking
as outlined in Section 5.3.
51
-------
The mean interspecies HSIs for Bell Run, Flat Creek, and Tuscarora Creek with trout stocking are 0.652,
0.508, and 0.340, respectively. The decrease in mean interspecies HSIs with respect to these streams'
unstocked condition simply reflects low HSI scores for brook, brown, and rainbow trout in these streams.
The total year 5 fish biomasses predicted for Bell Run, Flat Creek, and Tuscarora Creek with trout
stocking are 37.3, 37.9, and 8.88 kg[pw]/ha, respectively. These values correspond to 22 and 56 percent
reductions for Bell Run and Flat Creek, respectively, and a 308 percent increase for Tuscarora Creek.
The total year 5 fish densities predicted for Bell Run, Flat Creek, and Tuscarora Creek with trout
stocking are 1403, 3301, and 330 fish/ha, respectively. The total year 5 non-game fish biomasses of Bell
Run, Flat Creek, and Tuscarora Creek are predicted to be 22.3, 31.3, and 0.004 kg[rw]/ha, respectively.
These values represent 53.1, 63.9, and 99.9 percent reductions in the native or resident fishes of these
streams. These reductions are the direct consequence of trophic competition with trout for drifting and
benthic macro invertebrates and of predation by trout and are consistent with field studies that have
evaluated the impacts of game fish stocking to native/resident fish populations (see for example Garman
and Nielsen 1982, Fisher Huckins et al. 2000, Vander Zanden et al. 2004, Weidel et al. in review). Negus
(1995) predicted similar results and trends using the Wisconsin Bioenergetics Model (Hewett and
Johnson 1992, Hanson et al. 1997) for salmonid stocking programs in Minnesota waters of Lake
Superior. In a related application, Irwin et al. (2003) used the Wisconsin Bioenergetics Model to evaluate
how largemouth bass might to used to control gizzard shad introductions to small impoundments.
In Bell Run the mean annual biomass of benthic macroinvertebrates for both the stocked and unstocked
trout scenarios is 9.52 kg[ow]/ha since this resource is modeled as a system forcing function. The total
annual consumption of benthic macroinvertebrates by all fish species for the stocked and unstocked
scenarios is 35.1 and 15.1 kg[ow]/ha/yr, respectively. The increased consumption of benthic
macroinvertebrates by stocked trout has the greatest impact on resident non-ominivorous species such as
northern hogsucker, rock bass, and common shiner. Ominivorous species (i.e., white suckers,
stonerollers, river chubs, creek chubs, bluntnose minnows, and blacknose dace) are less impacted since
these fish can augment the lost of benthic macroinvertebrate prey with periphyton. For example, the per
capita total annual consumption of periphyton by non-game fish species for stocked and unstocked
scenarios is 0.0216 and 0.0485 kg[Dw]/fish/yr, respectively.
Piscivory in Bell Run significantly increases with trout stocking. Total annual piscivory for the unstocked
52
-------
Bell Run scenario is predicted to be only 0.0713 kg[ow]/ha/yr. However, trout stocking increases Bell
Run's total annual piscivory by an order of magnitude to 0.977 kg[ow]/ha/yr. Although this increased
piscivory is still small with respect to the total food consumption of Bell Run fish species (i.e.,
approximately 1%), it has significant impacts on the ultimate population numbers of non-game species
since most of the predicted piscivory occurs on small and young-of-year fish. For example, total annual
predatory mortality in Bell Run increases from 301 to 5037 fish/ha/yr with trout stocking. The mean
annual biomass and population density of white sucker, which is the community dominant with and
without stocking, decreases from 31.6 kg[rw]/ha and 726 fish/ha to 17.7 kg[rw]/ha and 436 fish/ha.
The trends predicted for Bell Run are repeated for Flat Creek in which bluehead chub is the community
dominant. In this case, the mean annual biomass and population density of bluehead chub decreases from
76.7 kg[pw]/ha and 10340 fish/ha to 23.4 kg[rw]/ha and 2839 fish/ha. The per capita annual consumption
of periphyton by non-game species increases from 0.00263 kg[Dw]/fish/yr for the status quo scenario to
0.0032 kg[Dw]/fish/yr for the trout stocking scenario. The order of magnitude difference in the per capita
periphyton consumption rates between Bell Run and Flat Creek is directly related their estimated benthic
macroinvertebrate to periphyton biomass ratios that equal 0.133 and 4.0, respectively. See Table 5.4 for
further details.
Unlike Bell Run and Flat Creek trout stocking is predicted to completely displace the resident fishes in
Tuscarora Creek The mean annual biomasses of resident fish and benthic macroinvertebrates in this
stream without trout stocking are only 0.663 and 10.0 kg[ow]/ha, respectively. On the other hand, the
total predicted annual consumption of fish and marcroinvertebrates by brook, brown, and rainbow trout
under the assumed stocking schedule is 19.3 kg[ow]/ha/yr. Thus, while Tuscarora Creek could maintain a
mixed trout fishery, it would do so at the expense of virtually all of its native fish fauna. If a trout fishery
was still desired, it would be advisable to the limit stockings to brook trout, which are less piscivorous
than brown or rainbow trout, and consider stocking rates even lower than those assumed in this analysis.
53
-------
Table 5.4 Summary of annual mean biomasses and fluxes predicted by BASS for Bell Run PA, Flat Creek VA, and
Tuscarora Creek WV for status quo conditions, with a 25% restoration of riparian canopy and ground cover, and with trout
stocking.
Community Variable : Condition
average HSI : status quo
average HSI : restoration
average HSI : stocking
total fish biomass kg(pw)/ha : status quo
total fish biomass kg(pw)/ha : restoration
total fish biomass kg(pw)/ha : stocking
total non-game fish biomass kg(pw)/ha : stocking
total fish density fish/ha : status quo
total fish density fish/ha : restoration
total fish density fish/ha : stocking
total non-game fish density fish/ha : stocking
consumption of macroinvertebrates kg(Dw)/ha/yr : status quo
consumption of macroinvertebrates kg(Dw)/ha/yr : restoration
consumption of macroinvertebrates kg(Dw)/ha/yr : stocking
consumption of periphyton kg(Dw)/ha/yr : status quo
consumption of periphyton kg(Dw)/ha/yr : restoration
consumption of periphyton kg(Dw)/ha/yr : stocking
predatory mortality kg(Dw)/ha/yr : status quo
predatory mortality kg(Dw)/ha/yr : restoration
predatory mortality kg(Dw)/ha/yr : stocking
predatory mortality fish/ha/yr : status quo
predatory mortality fish/ha/yr : restoration
predatory mortality fish/ha/yr : stocking
Bell Run
0.7568
0.7913
0.6524
47.54
47.82
37.31
22.25
3130
3331
1403
626
15.14
15.27
35.09
67.68
67.92
30.38
0.0713
0.0784
0.977
301
332
5037
Flat Creek
0.6049
0.6344
0.5075
86.68
86.50
37.92
31.30
10900
10960
3301
3048
118.8
118.4
54.49
28.72
28.57
9.814
0.286
0.338
1.416
1081
1311
5405
Tuscarora Creek
0.4675
0.5428
0.3400
2.884
3.767
8.881
0.004
2002
2896
330
2.82
0.440
0.655
18.62
5.589
6.579
0.00792
<0.001
<0.001
0.076
<1
<1
1066
54
-------
6. Example Application of the CVI Watershed Toolkit
The tools within the CVI Watershed Toolkit can be used either independently or together to analyze
expected responses of fish populations and communities to proposed stream restoration and/or fisheries
management. Figure 6.1 illustrates how CVI-WHAT IF tools can work together to evaluate a proposed
BMP for riparian stream restoration.
Figure 6.1 The use of the WHAT-IF tool to address a more complex management question for a particular stream site.
How do fish populations and communities respond to
a proposed restoration?
Site selection
I
Hydrology Tool
I
Clustering Tool
Habitat Suitability Tool
Define Restoration Activity
Tool
Recalculate Suitability
Compare
Results
Rerun BASS Tool
55
-------
To illustrate this process, we will now show how a riparian restoration analysis could be conducted for
Hogue Creek, Frederick County, VA which is an impacted smallmouth bass stream. For this example, we
assume that Hogue Creek's riparian ground cover and canopy cover are increased by 25% of their current
values. We also assume that this riparian restoration results in a 25% decrease of the creek's current
percent bottom coverage by fine sediments. The creek's restored stream bottom is then assumed to be
replaced with 50% sands (0.06-2 mm), 25% fine gravel (2-16 mm), 12.5% coarse gravel (16-64 mm), and
12.5% cobble (64-250 mm).
The first step in an integrated analysis of Hogue Creek riparian restoration is the estimation of the creek's
expected mean annual streamflow, depth, and water temperatures. Using the Hydro Tool, these
parameters are estimated to be 44.10 cfs, 0.59 ft, and 14.29 Celsius, respectively. See Figure 2.1.
Hogue Creek's mean annual streamflow, depth, and width can now be ported to the Clustering Tool
where these parameters are augmented with user-specified or CVI-WHAT IF database landscape
variables (i.e., site longitude and latitude, percent agriculture, percent stream slope, etc.) to predict the
stream's expected fish assemblage. See Figure 3.2. If the stream of interest is actually contained in the
CVI-WHAT IF database, as is Hogue Creek, the stream's predicted fish assemblages can be considered
as other fish communities that might exist at stream reaches other than the one contained in the CVI
database. Users can now port the stream's actual or predicted/alterative fish assemblage to either the
Habitat Suitability Tool or the BASS-Clustering Interface Tool.
When the user's ultimate goal is to run a BASS simulation analysis for a predicted/alternative fish
assemblage, the BASS-Clustering Interface Tool is used to translate knowledge of the stream's cluster
membership into a realized stream community and to assign initial conditions to that community. Two
methods are available to accomplish this task. The first of these is the nearest-neighbor option. This
option presents the user a list of all database streams that are members of the stream's predicted cluster.
This list is arranged in ascending order of the Euclidean distance from the stream of concern. The user
then selects any member of this list as a surrogate for the stream of interest. BASS data files for the
selected stream are then loaded into the user's active project. The second method is the ranked relative
biomass option. When choosing this option, the user constructs a fish community for the stream of
interest by selecting fish species from a series of lists. The i-th list in this series identifies those species
that are the i-th most abundant species (based on relative biomass) in one or more streams of the
56
-------
predicted cluster. Generally the number of species lists presented to the user will vary from cluster to
cluster, dependent on the number species needed to account for 95% of the biomass of streams within the
cluster. The BASS-Clustering Interface Tool then constructs initial conditions for the stream of interest
assuming that the stream's total fish biomass equals the mean total fish biomass of the predicted cluster'
streams. Additionally, the rank ordered relative biomasses of the stream of interest are assigned as the
means of the ranked ordered relative biomasses of streams within the predicted cluster. These rank
ordered relative biomasses generally will not be identical to the relative biomasses displayed by the
Clustering Tool itself. See Section 3.2.1 and the discussion concerning Table 3.1 and Figure 3.1. for
details.
The Habitat Suitability Tool is now used to calculate habitat suitability scores for each species within the
selected fish assemblage. Although the resulting habitat suitability scores can be ported directly to the
user's BASS project as habitat multipliers on species non-predatory mortality and dispersal, they can also
be used to corroborate or to redefine the stream's species composition predicted by the Clustering Tool.
Whereas the Clustering Tool essentially predicts a collection of species that is likely to occur in streams
possessing similar landscape features, scores predicted by the Habitat Suitability Tool are actually
probabilities of species occurrences at a stream based instream habitat features. Consequently, although
the Clustering Tool may predict that species A is a resident of the stream of interest (based on landscape
features), the Habitat Suitability Tool could predict (based on instream features) such a low probability
of occurrence for species A that a user might want to exclude species A from any further consideration.
Habitat suitability scores predicted for Hogue Creek before and after the riparian restoration scenario
outlined above are presented in Table 6.1. According to Table 6.1, HSI scores for all of the species
under the status quo scenario are >0.5, which is the generally-accepted criterion for occurrence, so the
results of the HSI are consistent with the observed community at the site. Under the restoration scenario,
the interspecies mean increased from 0.7689 to 0.8240. This analysis demonstrates how a restoration can
be expected to improve overall species habitat and fish community condition.
Having estimated the stream's hydrological features, community structure, and habitat relationships, the
BASS fish community model can now be run to simulate Hogue Creek's responses to the proposed
riparian restoration scenario. Figure 6.2 and Figure 6.3 display the predicted 5 year biomass dynamics
of Hogue Creek fishes with and without riparian restoration. The total year 5 fish biomasses predicted for
Hogue Creek with and without riparian restoration are 14.1 and 17.1 kg[rw]/ha, respectively. Thus, for
57
-------
Hogue Creek a 25% increase of riparian covers is expected to increase total fish biomass by 21.2%. The
stream's average annual fish density, however, is predicted to remain essentially unchanged. In
particular, the total year 5 fish densities predicted for Hogue Creek before and after riparian restoration
are 422 and 437 fish/ha, respectively.
Table 6.1 Status quo and restoration HSI for Hogue Creek, VA
Species
status quo HSI
restoration HSI
white sucker
smallmouth bass
stoneroller
redbreast sunfish
fallfish
bluntnose minnow
creek chub
northern hogsucker
creek chub sucker
golden redhorse
longear sunfish
rockbass
green sunfish
interspecies mean
0.8800
0.7110
0.8342
0.8766
0.9869
0.5150
0.8739
0.8241
0.5406
0.5406
0.8766
0.6597
0.8766
0.7689
0.9022
0.7906
0.8363
0.9232
0.9903
0.6827
0.8739
0.8883
0.6589
0.6589
0.9232
0.6597
0.9232
0.8240
58
-------
Figure 6.2 Predicted biomass dynamics of Hogue Creek, VA before riparian
restoration
Figure 6.3 Predicted biomass dynamics for Hogue Creek, VA after riparian
restoration
S3 BASS Fish
Graph
12000 -
"1" 10000 -
if,
2
•p 6000 -
S
S
2000 -
0 -
Select Variable to
Plot by species
Total Species Biomass (grams)
^Ln
^-,
1
-------
References
Aadland, L.P. 1993. Stream habitat types: their fish assemblages and relationship to flow. North
American Journal of Fisheries Management 13: 790-806.
Allen, M.S., M.V. Hoyer, and D.E. Canfield, Jr. 1998. Factors related to black crappie occurrence,
density, and growth in Florida Lakes. North American Journal of Fisheries Management 18: 864-871.
Angermeier, P.L. and M.R. Winston. 1999. Characterizing fish community diversity across Virginia
landscapes: Prerequisite for conservation. Ecological Applications 9: 335-349.
Angermeier, P.L., K.L. Krueger, and C.A. Dolloff. 2002. Discontinuity in stream-fish distributions:
implications for assessing and predicting species occurrence. In Predicting species occurrences: issues of
accuracy and scale. Edited by J.M. Scott, P.J. Heglund, M.L. Morrison, J.B. Haufler, M.G. Raphael,
W.A. Wall and F.B. Samson. Island Press, Washington, DC. pp. 519-527
Ansari, Z.A., A. Chatterji, B.S. Ingole, R.A. Sreepada, C.U. Rivonkar, and A.H. Parulekar. 1995.
Community structure and seasonal variation of an inshore demersal fish community at Goa, west-coast of
India. Estuarine Coastal and Shelf Science 41: 593-610.
Barber, M.C. 2001. Bioaccumulation and Aquatic System Simulator (BASS) User's Manual Beta Test
Version 2.1. U.S. Environmental Protection Agency, Office of Research and Development, Athens, GA,
USA. EPA/600/R-01/035.
Barber, M.C. 2004a. Bioaccumulation and Aquatic System Simulator (BASS) User's Manual Version
2.2. U.S. Environmental Protection Agency, Office of Research and Development, Athens, GA, USA.
EPA/600/R-04/XXX.
Barber, M.C. 2004b. Bioaccumulation and Aquatic System Simulator (BASS) Version 2.2 Data
Supplement. U.S. Environmental Protection Agency, Office of Research and Development, Athens, GA,
USA. EPA/600/R-04/XXX.
Barbour, M.T., J. Gerritsen, B.D. Snyder, and J.B. Stribling. 1999. Rapid Bioassessment Protocols for
Use in Streams and Wadeable Rivers: Periphyton, Benthic Macroinvertebrates and Fish, Second Edition.
U.S. Environmental Protection Agency, Office of Water, Washington, D.C. EPA 841-B-99-002.
Barrett, J.C. 1992. Turbidity-induced changes in reactive distance of rainbow trout. Transactions of the
American Fisheries Society 121: 437-443.
Beamish, F.W.H. and M. Legrow. 1983. Bioenergetics of the southern brook lamprey, Ichthyomyzon
gagei. Journal of Animal Ecology 52: 575-590.
Beamish, F.W.H. and P.O. MacMahon. 1988. Apparent heat increment and feeding strategy in walleye,
Stizostedion vitreum vitreum. Aquaculture 68: 73-82.
Bergman, E. and L.A. Greenberg. 1994. Competition between a planktivore, a benthivore, and a species
with ontogenetic diet shifts. Ecology 75: 1233-1245.
60
-------
Bettoli, P.W., MJ. Maceina, R.L. Noble, and R.K. Betsill. 1993. Response of a reservoir fish community
to aquatic vegetation removal. North American Journal of Fisheries Management 13: 110-124.
Blann, K., J.F. Nerbonne, and B. Vondracek. 2002. Relationship of riparian buffer type to water
temperature in the Driftless Area ecoregion of Minnesota. North American Journal of Fisheries
Management 22: 441-451.
Bohlin, T., C. Dellefors, U. Faremo, and A. Johlander. 1994. The energetic equivalence hypothesis and
the relation between population density and body size in stream-living salmonids. American Naturalist
143: 478-493.
Bonner, T.H. and G.R. Wilde. 2002. Effects of turbidity on prey consumption of prairie stream fishes.
Transactions of the American Fisheries Society 131: 1203-1208.
Boudreau, P.R. and L.M. Dickie. 1989. Biological model of fisheries production based on physiological
and ecological scalings of body size. Canadian Journal of Fisheries and Aquatic Sciences 46: 614-623.
Bowman, M.L. 1970. Life history of the black redhorse, Moxostoma duquesnei (Lesueur), in Missouri.
Transactions of the American Fisheries Society 99: 546-559.
Bozek, M.A., T.M. Burri, and R.V. Frie. 1999. Diets of muskellunge in northern Wisconsin lakes. North
American Journal of Fisheries Management 19: 258-270.
Breck, J.E. and M.J. Gitter. 1983. Effect offish size on the reactive distance of bluegill (Lepomis
macrochirus) sunfish. Canadian Journal of Fisheries and Aquatic Sciences 40: 162-167.
Brett, J.R., J.E. Shelbourn, and C.T. Shoop. 1969. Growth rate and body composition of fingerling
sockeye salmon, Oncorhynchus nerka, in relation to temperature and ration size. Journal of the Fisheries
Research Board of Canada 26: 2363-2394.
Brodeur, R.D. 1991. Ontogenetic variations in the type and size of prey consumed by juvenile coho,
Oncorhynchus kisutch, and chinook, O. tshawytscha, salmon. Environmental Biology of Fishes 30: 303-
315.
Burr, B.M. and R.L. Mayden. 1982. Life history of the brindled madtom Noturus miurus in Mill Creek,
Illinois (Pisces: Ictaluridae). American Midland Naturalist 107: 2541.
Buynak, G.L., L.E. Kornman, A. Surmont, and B. Mitchell. 1991. Evaluation of a smallmouth bass
stocking program in a Kentucky Stream. North American Journal of Fisheries Management 11: 293-297.
Carlander, K.D. 1969. Handbook of Freshwater Fishery Biology. Volume One: Life History Data on
Freshwater Fishes of the United States and Canada, Exclusive of the Perciformes. Iowa State University
Press, Ames, IA.
Carlander, K.D. 1977. Handbook of Freshwater Fishery Biology. Volume Two: Life History Data on
Centrarchid Fishes of the United States and Canada. Iowa State University Press, Ames, IA.
Carlander, K.D. 1997. Handbook of Freshwater Fishery Biology. Volume Three: Life History Data on
61
-------
Ichthyopercid and Percid Fishes of the United States and Canada. Iowa State University Press, Ames, IA.
Craig, J.F. 1977. The body composition of the adult perch, Perca fluviatilis in Windermere, with
reference to seasonal changes and reproduction. Journal of Animal Ecology 46: 617-632.
CVI. 2002. Mid-Atlantic Highland Action Program. Canaan Valley Institute, Thomas, WV.
Damuth, J. 1981. Population density and body size in mammals. Nature (London) 290: 699-700.
Daniels, R.B. and J.W. Gilliam. 1996. Sediment and chemical load reduction by grass and riparian filters.
Soil Science Society of American Journal 60: 246-251.
Dettmers, J.M., S. Gutreuter, D.H. Wahl, and D.A. Soluk. 2001. Patterns in abundance of fishes in main
channels of the upper Mississippi River system. Canadian Journal of Fisheries and Aquatic Sciences 58:
933-942.
Diana, J.S. 1995. Biology and Ecology of Fishes. Cooper Publishing Group, Carmel, IN.
Dickie, L.M., S.R. Kerr, and P.R. Boudreau. 1987. Size-dependent processes underlying regularities in
ecosystem structure. Ecological Monographs 57: 233-250.
Doherty, J. and J.M. Johnston. 2003. Methodologies for calibration and predictive analysis of a
watershed model. Journal of the American Water Resources Association 39: 251-265.
Dreves, D.P., T.J. Timmons, and J. Henson. 1996. Age, growth, and food of freshwater drum,
Aplodinotus grunniens (Scianidae), in Kentucky Lake, Kentucky/Tennessee. Transactions of the
Kentucky Academy of Science 57: 22-26.
du Preez, H.H. and A.C. Cockroft. 1988a. Nonfaecal and faecal losses ofPomadasys commersonni
(Teleostei: Pomadasyidae) feeding on the surf clam, Donax serra. Comparative Biochemistry and
Physiology A 90: 63-70.
du Preez, H.H. and A.C. Cockroft. 1988b. Nonfaecal and faecal losses of the marine teleost, Lichia amia
(Linnaeus, 1758), feeding on live southern mullet, Liza richardsonii (Smith, 1846). Comparative
Biochemistry and Physiology A 90: 71-77.
Dunbrack, R.L. 1988. Feeding of juvenile coho salmon (Oncorhynchus kisutch): maximum appetite,
sustained feeding rate, appetite return, and body size. Canadian Journal of Fisheries and Aquatic
Sciences 45: 1191-1196.
Dunham, J.B. and G.L. Vinyard. 1997. Relationships between body mass, population density, and the
self-thinning rule in stream-living salmonids. Canadian Journal of Fisheries and Aquatic Sciences 54:
1025-1030.
Dunham, J.B., B.R. Dickerson, E. Beever, R.D. Duncan, and G.L. Vinyard. 2000. Effects of food
limitation and emigration on self-thinning in experimental minnow cohorts. Journal of Animal Ecology
69: 927-934.
62
-------
East, P. and P. Magnan. 1991. Some factors regulating piscivory of brook trout, Salvelinus fontinalis, in
lakes of the Laurentian Shield. Canadian Journal of Fisheries and Aquatic Sciences 48: 1735-1743.
Edwards, E.A., H. Li, and C.B. Schreck. 1983. Habitat suitability index models: Longnose dace. U.S.
Department of the Interior, Fish and Wildlife Service, Washington, DC. FWS/OBS-82/10.33.
Elliott, J.M. 1976a. Body composition of the brown trout (Salmo trutta L.) in relation to temperature and
ration size. Journal of Animal Ecology 45: 273-289.
Elliott, J.M. 1976b. The energetics of feeding, metabolism and growth of brown trout (Salmo trutta L.) in
relation to body weight, water temperature and ration size. Journal of Animal Ecology 45: 923-948.
Elliott, J.M. 1993. The self-thinning rule applied to juvenile sea-trout, Salmo trutta. Journal of Animal
Ecology 62: 371-379.
Elrod, J.H. and R. O'Gorman. 1991. Diet of juvenile lake trout in southern Lake Ontario in relation to
abundance and size of prey fishes, 1979-1987. Transactions of the American Fisheries Society 120: 290-
302.
Ensign, W.E., R.J. Strange, and S.E. Moore. 1990. Summer food limitation reduces brook and rainbow
trout biomass in a Southern Appalachian stream. Transactions of the American Fisheries Society 119:
894-901.
Eschmeyer, P.H. and A.M. Phillips, Jr. 1965. Fat content of the flesh of siscowets and lake trout from
Lake Superior. Transactions of the American Fisheries Society 94: 62-74.
Etnier, D.A. and W.C. Starnes. 1993. The Fishes of Tennessee. University of Tennessee Press,
Knoxville, TN.
Fielding, A.H. 2002. What are the appropriate characteristics of an accuracy measure? In Predicting
species occurrences: issues of accuracy and scale. Edited by J.M. Scott, P.J. Heglund, M.L. Morrison,
J.B. Haufler, M.G. Raphael, W.A. Wall andF.B. Samson. Island Press, Washington, DC. pp. 271-280
Fisher Huckins, C.J., C.W. Osenberg, and G.G. Mittelbach. 2000. Species introductions and their
ecological consequences: an example with congeneric sunfish. Ecological Applications 10: 612-625.
Fisher, W. 1958. On grouping for maximum homogeneity. Journal of the American Statistical
Association 53: 789-798.
Flath, L.E. and J.S. Diana. 1985. Seasonal energy dynamics of the alewife in southeastern Lake
Michigan. Transactions of the American Fisheries Society 114: 328-337.
Flick, W.A. and D.A. Webster. 1992. Standing crop of brook trout in Adirondack waters before and after
removal of non-trout species. North American Journal of Fisheries Management 12: 783-796.
Freeman, M.C., Z.H. Bowen, and J.H. Crance. 1997. Transferability of habitat suitability criteria for
fishes in warmwater streams. North American Journal of Fisheries Management 17: 20-31.
63
-------
Garman, G.C. and L.A. Nielsen. 1982. Piscivory by stocked brown trout (Salmo trutta) and its impact on
the nongame fish community of Bottom Creek, Virginia. Canadian Journal of Fisheries and Aquatic
Sciences 39: 862-869.
Gillen, A.L., R.A. Stein, and R.F. Carline. 1981. Predation by pellet-reared tiger muskelunge on minnows
and bluegills in experimental systems. Transactions of the American Fisheries Society 110: 197-209.
Glantz, S.A. and B.K. Slinker. 1990. Primer of Applied Regression and Analysis of Variance. McGraw-
Hill, Inc., New York, NY.
Gordoa, A. and C.M. Duarte. 1992. Size-dependent density of the demersal fish off Namibia: patterns
within and among species. Canadian Journal of Fisheries and Aquatic Sciences 49: 1990-1993.
Grant, J.W.A. and D.L. Kramer. 1990. Territory size as a predicator of the upper limit to population
density of juvenile salmonids in streams. Canadian Journal of Fisheries and Aquatic Sciences 47: 1724-
1737.
Grant, J.W.A., S.O. Steingrimsson, E.R. Keeley, and R.A. Cunjak. 1998. Implications of territory size for
the measurement and prediction of salmonid abundance in streams. Canadian Journal of Fisheries and
Aquatic Sciences 55 (Suppl. 1): 181-190.
Gray, R.H. and D.D. Dauble. 1979. Biology of the sand roller in the central Columbia River.
Transactions of the American Fisheries Society 108: 646-649.
Green, R.H. 1979. Sampling design and statistical methods for environmental biologists. Wiley, New
York, NY.
Gregory, RS. 1993. Effect of turbidity on the predator avoidance behavior of juvenile chinook salmon
(Oncorhynchus tshawytscha). Canadian Journal of Fisheries and Aquatic Sciences 50: 241-246.
Gregory, RS. and T.G. Northcote. 1993. Surface, planktonic, andbenthic forgaging by juvenile chinook
salmon (Oncorhynchus tshawytscha) in turbid laboratory conditions. Canadian Journal of Fisheries and
Aquatic Sciences 50: 233-240.
Grossman, G.D., RE. Ratajczak, Jr., M. Crawford, andM.C. Freeman. 1998. Assemblage organization in
stream fishes: effect of environmental variation and interspecific interactions. Ecological Monographs
68: 395-420.
Groves, T.D.D. 1970. Body composition changes during growth in young sockeye (Oncorhynchus nerka)
in fresh water. Journal of the Fisheries Research Board of Canada 27: 929-942.
Guay, J.C., D. Boisclair, D. Rioux, M. Leclerc, M. Lapointe, and P. Legendre. 2000. Development and
validation of numerical habitat models for juveniles of Atlantic salmon (Salmo solar). Canadian Journal
of Fisheries and Aquatic Sciences 57: 2065-2075.
Guisan, A. andN.E. Zimmermann. 2000. Predictive habitat distribution models in ecology. Ecological
Modelling 135: 147-186.
64
-------
Habera, J.W., RD. Bivens, B.D. Carter, and C.E. Williams. 2004. Region IV Trout Fisheries Report.
Tennessee Wildlife Resources Agency, Nashville, TN. Fisheries Report No. 04-04.
Hackney, P.A., G.R. Hooper, and J.F. Webb. 1970. Spawning behavior, age and growth, and sport fishery
for the silver redhorse, Moxostoma anisurum (Rafmesque), in the Flint River, Alabama. Proceedings of
the Annual Conference of Southeastern Association of Fish and Wildlife Agencies 24: 569-576.
Hale, RS. 1996. Threadfm shad use as supplemental prey in reservoir white crappie fisheries in
Kentucky. North American Journal of Fisheries Management 16: 619-632.
Hambright, K.D. 1991. Experimental analysis of prey selection by largemouth bass: role of predator
mouth width and prey body depth. Transactions of the American Fisheries Society 120: 500-508.
Hanson, J.M. and W.C. Leggett. 1986. Effect of competition between two freshwater fishes on prey
consumption and abundance. Canadian Journal of Fisheries and Aquatic Sciences 43: 1363-1372.
Hanson, P.C., T.B. Johnson, D.E. Schindler, and J.F. Kitchell. 1997. Fish bioenergetics 3.0 for Windows.
University of Wisconsin, Sea Grant Institute, Madision, WI. U.S. Department of Commerce, National
Technical Information Service PB98-502370.
Harrell, F.R. 2001. Regression modeling strategies with applications to linear models, logistic regression
and survival analysis. Springer-Verlag, New York, NY.
Hartman, K.J. and S.B. Brandt. 1995. Estimating energy density offish. Transactions of the American
Fisheries Society 124: 347-355.
Hewett, S.W. andB.L. Johnson. 1992. Fish bioenergetics model 2. University of Wisconsin, Sea Grant
Institute, Madision, WI. WIS-SG-92-250.
Holling, C.S. 1966. The functional response of invertebrate predators to prey density. Memoirs of the
Entomological Society of Canada 48: 1-86.
Holmes, T.P., J.C. Bergstrom, E. Huszar, S.B. Kask, and F. Orr, El. 2004. Contingent valuation, net
marginal benefits, and the scale of riparian ecosystem restoration. Ecological Economics 49: 19-30.
Hosmer, D.W. and S. Lemeshow. 2000. Applied Logistic Regression, 2nd edition. John Wiley & Sons,
New York, NY.
House, R. and L. Wells. 1973. Age, growth, spawning season, and fecunity of the trout-perch (Percopsis
omiscomaycus) in southeastern Lake Michigan. Journal of the Fisheries Research Board of Canada 30:
1221-1225.
Hurley, J.F. 1986. Summary: development, testing, and application of wildlife-habitat models-the
managers viewpoint. In Wildlife 2000: Modeling habitat relationships of terrestrial vertebrates. Edited by
J.A. Verner, M.L. Morrison and C.J. Ralph. University of Wisconsin Press, Madison, WI. pp. 151-153
Ingram, W. and C.D. Ziebell. 1983. Diet shifts to benthic feeding by threadfin shad. Transactions of the
American Fisheries Society 112: 554-556.
65
-------
Irwin, B J., D.R. DeVries, and R.A. Wright. 2003. Evaluating the potential for predatory control of
gizzard shad by largemouth bass in small impoundments: a bioenergetics approach. Transactions of the
American Fisheries Society 132: 913-924.
Jenkins, R.E. andN.M. Burkhead. 1994. Freshwater Fishes of Virginia. American Fisheries Society,
Bethesda, MD.
Jobling, M. 1981. Mathematical models of gastric emptying and the estimation of daily rates of food
consumption for fish. Journal of Fish Biology 19: 245-257.
Jobling, M. 1986. Mythical models of gastric emptying and implications for food consumption studies.
Environmental Biology of Fishes 16: 35-50.
Jobling, M. 1987. Influence of particle size and dietary energy content on patterns of gastric evacuation
in fishes: test of a physiological model of gastric emptying. Journal of Fish Biology 30: 299-314.
Johnson, B.L., D.L. Smith, and R.F. Carline. 1988. Habitat preferences, survival, growth, foods, and
harvests of walleye and walleye x sauger hybrids. North American Journal of Fisheries Management 8:
292-304.
Johnson, P.C. and G.L. Vinyard. 1987. Filter-feeding behavior and particle retention efficiency of
Sacramento Blackfish. Transactions of the American Fisheries Society 116: 634-640.
Jongman, RH.G., C.J.F. ter Braak, and O.F.R. van Tongeren. 1995. Data analysis in community and
landscape ecology. Cambridge University Press, Cambridge, England.
Juanes, F. 1986. Population density and body size in birds. American Naturalist 128: 921-929.
Juanes, F., R.E. Marks, K.A. McKown, and D.O. Conover. 1993. Predation by age-0 bluefish on age-0
anadromous fishes in the Hudson River estuary. Transactions of the American Fisheries Society 122:
348-356.
Jude, D.J., F.J. Tesar, S.F. Deboe, and T.J. Miller. 1987. Diet and selection of major prey species by
Lake Michigan salmonines, 1973-1982. Transactions of the American Fisheries Society 116: 677-691.
Kaufmann, P.R, P. Levine, E.G. Robison, C. Seeliger, and D.V. Peck. 1999. Quantifying Physical
Habitat in Wadeable Streams. U.S. Environmental Protection Agency, Washington, D.C. EPA/620/R-
99/003.
Keeley, E.R. 2003. An experimental analysis of self-thinning in juvenile steelhead trout. Oikos 102: 543-
550.
Keeley, E.R. and J.W.A. Grant. 1995. Allometric and environmental correlates of territory size in
juvenile Atlantic salmon (Salmo solar). Canadian Journal of Fisheries and Aquatic Sciences 52: 186-196.
Kelso, J.R.M. and M.G. Johnson. 1991. Factors related to the biomass and production offish
communities in small, oligotrophic lakes vulnerable to acidification. Canadian Journal of Fisheries and
Aquatic Sciences 48: 2523-2532.
66
-------
Kendrick, T.H. and M.P. Francis. 2002. Fish assemblages in the Hauraki Gulf, New Zealand. New
Zealand Journal of Marine and Freshwater Research 36: 699-717.
Knight, R.L., FJ. Margraf, and R.F. Carline. 1984. Piscivory by walleyes and yellow perch in western
Lake Erie. Transactions of the American Fisheries Society 113: 677-693.
Knight, W. 1968. Asymptotic growth: An example of nonsense disguised as mathematics. Journal of the
Fisheries Research Board of Canada 25: 1303-1307.
Knouft, J.H. 2002. Regional analysis of body size and population density in stream fish assemblages:
testing predictions of the energetic equivalence rule. Canadian Journal of Fisheries and Aquatic Sciences
59: 1350-1360.
Kuehn, J.H. 1949. A population study of longnose dace (Rhinichthys c. cataractae). Proceeding of the
Minnesota Academy of Science 17: 81-87.
Kunisaki, N., K. Takada, and H. Matsuura. 1986. On the study of lipid content, muscle hardness, and
fatty acid composition of wild and cultured horse mackerel. Nippon Suisan Gakkaishi 52: 333-336.
Kuns, M.M. and W.G. Sprules. 2000. Zooplankton production in Lake Ontario: a multistrata approach.
Canadian Journal of Fisheries and Aquatic Sciences 57: 2240-2247.
Kushlan, J.A., S.A. Voorhees, W.F. Loftus, and P.C. Frohring. 1986. Length, mass, and calorific
relationships of Everglades animals. Florida Scientist 49: 65-79.
Lacasse, S. and P. Magnan. 1992. Biotic and abiotic determinants of the diet of brook trout, Salvelinus
fontinalis, in lakes of the Laurentian Shield. Canadian Journal of Fisheries and Aquatic Sciences 49:
1001-1009.
Lammens, E.H.R.R., H.W. de Nie, and J. Vijverberg. 1985. Resource partitioning and niche shifts of
bream (Abramis brama) and eel (Anguilla anguilla) mediated bypredation of smelt (Osmerus eperlanus)
onDaphnia hyalina. Canadian Journal of Fisheries and Aquatic Sciences 42: 1342-1351.
Lassiter, R.R. 1975. Modeling dynamics of biological and chemical components of aquatic ecosystems.
U.S. Environmental Protection Agency, Office of Research and Development, National Environmental
Research Center, Corvallis, OR. EPA/660/3-75-012.
Lassiter, R.R. and D.K. Kearns. 1974. Phytoplankton population changes and nutrient fluctuations in a
simple ecosystem model. In Modeling the Eutrophication Process. Edited by E.J. Middlebrooks, D.H.
Falkenborg and T.E. Maloney. Ann Arbor Scientific Publications
Ann Arbor, pp. 131-138.
Lefwich, K.N., P.L. Angermeier, and C.A. Dollof. 1997. Factors influencing behavior and transferability
of habitat models for a benthic stream fish. Transactions of the American Fisheries Society 126: 725-734.
Leonard, P.M. and DJ. Orth. 1988. Use of habitat guilds to determine instream flow requirements. North
American Journal of Fisheries Management 8: 399-409.
67
-------
Lewis, W.M., R. Heidinger., W. Kirk, W. Chapman, and D. Johnson. 1974. Food intake of the
largemouth bass. Transactions of the American Fisheries Society 103: 277-280.
Lorenzen, K. 1996. The relationship between body weight and natural mortality in juvenile and adult
fish: a comparison of natural ecosystems and aquaculture. Journal of Fish Biology 49: 627-647.
Lowe, T.P., T.W. May, W.G. Brumbaugh, and D.A. Kane. 1985. National Contaminant Biomonitoring
Program: Concentrations of seven elements in freshwater fish, 1979-1981. Archives of Environmental
Contamination and Toxicology 14: 363-388.
Maceina, M.J., W.B. Wrenn, and D.R. Lowery. 1995. Estimating harvestable largemouth bass abundance
in a reservoir with an electro fishing catch depletion technique. North American Journal of Fisheries
Management 15: 103-109.
MacKenzie, D.I., L.L. Bailey, and J.D. Nichols. 2004. Investigating species co-occurrence patterns when
species are detected imperfectly. Journal of Animal Ecology 73: 546-555.
Madejczyk, J.C., N.D. Mundahl, and R.M. Lehtinen. 1998. Fish assemblages of natural and artificial
habitats within the channel border of the upper Mississippi River. American Midland Naturalist 139:
296-310.
Madenjian, C.P., T.J. DeSorcie, and R.M. Stedman. 1998. Ontogenic and spatial patterns in diet and
growth of lake trout in Lake Michigan. Transactions of the American Fisheries Society 127: 236-252.
Magnan, P. 1988. Interactions between brook charr, Salvelinus fontinalis, and nonsalmonid species:
ecological shift, morphological shift, and their impact on zooplankton communities. Canadian Journal of
Fisheries and Aquatic Sciences 45: 999-1009.
Margenau, T.L., P.W. Rasmussen, and J.M. Kampa. 1998. Factors affecting growth of northern pike in
small Wisconsin lakes. North American Journal of Fisheries Management 18: 625-639.
Mastrorillo, S., S. Lek, F. Dauba, and A. Belaud. 1997. The use of artificial neural networks to predict
the presence of small-bodied fish in a river. Freshwater Biology 38: 237-246.
Matthews, W.J. and H.W. Robison. 1988. The distribution of the fishes of Arkansas: a multivariate
analysis. Copeia 1988: 359-374.
Mattingly, H.T. and M.J. Butler, IV. 1994. Laboratory predation on the Trinidadian guppy: implications
for size-selective predation hypothesis and guppy life history evolution. Oikos 69: 54-64.
Mayden, R.L. and S.J. Walsh. 1984. Life history of the least madtom Noturus hildebrandi (Siluriformes:
Ictaluridae) with comparisons to related species. American Midland Naturalist 112: 349-368.
McCann, J.A. 1959. Life history studies of the spottail shiner of Clear Lake, Iowa, with particular
reference to some sampling problems. Transactions of the American Fisheries Society 88: 336-343.
McCormick, F.H., D.V. Peck, and D.P. Larsen. 2000. Comparison of geographic classification schemes
for Mid-Atlantic stream fish assemblages. Journal of the North American Benthological Society 19: 385-
68
-------
404.
McCormick, F.H., R.M. Hughes, P.R. Kaufmann, D.V. Peck, J.L. Stoddard, and A.T. Herlihy. 2001.
Development of an index of biotic integrity for the Mid-Atlantic Highlands region. Transactions of the
American Fisheries Society 130: 857-877.
McGurk, M.D. 1993. Allometry of herring mortality. Transactions of the American Fisheries Society
122: 1035-1042.
McGurk, M.D. 1999. Size dependence of natural mortality rate of sockeye salmon andkokanee in
freshwater. North American Journal of Fisheries Management 19: 376-396.
Miller, T.J., L.B. Crowder, J.A. Rice, and P.P. Binkowski. 1992. Body size and the ontogeny of the
functional response in fishes. Canadian Journal of Fisheries and Aquatic Sciences 49: 805-812.
Miner, J.G. and R.A. Stein. 1996. Detection of predators and habitat choice by small bluegills: effects of
turbidity and alternative prey. Transactions of the American Fisheries Society 125: 97-103.
Minns, C.K. 1995. Allometry of home range size in lake and river fishes. Canadian Journal of Fisheries
and Aquatic Sciences 52: 1499-1508.
Minor, J.D. and E.J. Grossman. 1978. Home range and seasonal movements of muskellunge as
determined by radiotelemetry. American Fisheries Society Special Publication 11: 146-153.
Mittelbach, G.G. andL. Persson. 1998. The ontogeny of piscivory and its ecological consequences.
Canadian Journal of Fisheries and Aquatic Sciences 55: 1454-1465.
Moore, C.M., R.J. Neves, J.J. Ney, and O.K. Whitehurst. 1985. Utilization of alewives and gizzard shad
by striped bass in Smith Mountain Lake, Virginia. Proceeding of the Annual Conference of the
Southeastern Association of Fish and 39: 108-115.
Morishita, T., U. Kazuaki, N. Imura, and T. Takahashi. 1987. Variation with growth in the proximate
compositions of cultured red sea bream. Nippon Suisan Gakkaishi 53: 1601-1607.
Morris, L.A. 1965. Age and growth of the river carpsucker, Carpiodes carpio, in the Missouri River.
American Midland Naturalist 73: 423-429.
Mueller, G. 1996. Establishment of a fish community in the Hayden-Rhodes and Salt-Gila Aqueducts,
Arizona. North American Journal of Fisheries Management 16: 795-804.
Myers, R.A. 1997. Comment and reanalysis: paradigms for recruitment studies. Canadian Journal of
Fisheries and Aquatic Sciences 54: 978-981.
Myers, R.A. and N.J. Barrowman. 1996. Is fish recruitment related to spawner abundance? Fishery
Bulletin 94: 707-724.
Negus, M.T. 1995. Bioenergetics modeling as a salmonine management tool applied to Minnesota waters
of Lake Superior. North American Journal of Fisheries Management 15: 60-78.
69
-------
Nelson, W.R. 1974. Age, growth, and maturity of thirteen species offish from Lake Oahe during the
early years of impoundment, 1963-68. U.S. Department of Interior, Fish and Wildlife Service,
Washington, DC. Technical Papers of the Bureau of Sport Fisheries and Wildlife No. 77.
Oberdorff, T., D. Pont, B. Hugueny, and D. Chessel. 2001. A probabilistic model characterizing fish
assemblages of French rivers: a framework for environmental assessment. Freshwater Biology 46: 399-
415.
Olden, J.D. and D.A. Jackson. 2002. A comparison of statistical approaches for modeling fish species
distributions. Freshwater Biology 47: 1976-1995.
Ornellas, A.B. and R Coutinho. 1998. Spatial and temporal patterns of distribution and abundance of a
tropical fish assemblage in a seasonal Sargassum bed, Cabo Frio Island, Brazil. Journal of Fish Biology
53: 198-208.
Paloheimo, J.E. and L.M. Dickie. 1965. Food and growth of fishes. I. A growth curve derived from
experimental data. Journal of the Fisheries Research Board of Canada 22: 521-542.
Parker, R.R. and P. A. Larkin. 1959. A concept of growth in fishes. Journal of the Fisheries Research
Board of Canada 16: 721-745.
Parsons, J.W. 1971. Selective food preferences of walleyes of the 1959 year class in Lake Erie.
Transactions of the American Fisheries Society 100: 474-485.
Paulson, L.J. 1980. Models of ammonia excretion for brook trout (Salvelinus fontinalis) and rainbow
trout (Salmo gairdneri). Canadian Journal of Fisheries and Aquatic Sciences 37: 1421-1425.
Pereira, D.L. and G.W. LaBar. 1983. Age and growth of trout-perch in Lake Champlain. New York Fish
and Game Journal 30: 201-209.
Persson, A. and L.-A. Hansson. 1999. Diet shift in fish following competitive release. Canadian Journal
of Fisheries and Aquatic Sciences 56: 70-78.
Peters, R.H. and J.V. Raelson. 1984. Relations between individual size and mammalian population
density. American Naturalist 124: 498-517.
Peterson, I. and J.S. Wroblewski. 1984. Mortality rates of fishes in the pelagic ecosystem. Canadian
Journal of Fisheries and Aquatic Sciences 41: 1117-1120.
Pierce, C.L., M.D. Sexton, M.E. Pelham, and H. Liao. 2001. Dynamics of the littoral fish assemblage in
Spirit Lake, Iowa, and implications for prey availability for piscivores. North American Journal of
Fisheries Management 21: 884-896.
Plante, C. and J.A. Downing. 1989. Production of freshwater invertebrate populations in lakes. Canadian
Journal of Fisheries and Aquatic Sciences 46: 1489-1498.
Priegel, G.R. 1969. Age and rate of growth of freshwater drum in Lake Winnebago, Wisconsin.
Transactions of the American Fisheries Society 98: 116-118.
70
-------
Purkett, C.A., Jr. 1957. Growth of the fishes in the Salt River, Missouri. Transactions of the American
Fisheries Society 87: 116-131.
Quinn, S.P. 1988. Flathead catfish abundance and growth in the Flint River, Georgia. Proceedings of the
Annual Conference of Southeastern Association of Fish and Wildlife Agencies 42: 141-148.
Radwell, A. 2000. Ecological integrity assessment of Ozark rivers to determine suitability for protective
status, pp. 107. University of Arkansas, Fayetteville, AR.
Randall, R.G., J.R.M. Kelso, and C.K. Minns. 1995. Fish production in freshwaters: Are rivers more
productive than lakes? Canadian Journal of Fisheries and Aquatic Sciences 52: 631-643.
Rashevsky, N. 1959. Some remarks on the mathematical theory of nutrition of fishes. Bulletin of
Mathematical Biophysics 21: 161-183.
Reed, M.S. and C.F. Rabeni. 1989. Characteristics of an unexploited smallmouth bass population in a
Missouri Ozark stream. North American Journal of Fisheries Management 9: 420-426.
Reed, R.J. 1959. Age, growth, and food of longnose dace, Rhinichthys cataractae in northwestern
Pennsylvania. Coepia 1959: 160-162.
Reed, R.J. 1971. Biology ofthe fallfish, Semotilus corporalis (Pisces, Cyprinidae). Transactions of the
American Fisheries Society 100: 717-725.
Reed, R.J. and J.C. Moulton. 1973. Age and growth of blacknose dace, Rhinichthys atratulus and
longnose dace, R. cataractae in Massachusetts. American Midland Naturalist 90.
Reeves, G.H., L.E. Benda, K.M. Burnett, P.A. Bisson, and J.R. Sedell. 1995. A disturbance-based
ecosystem approach to maintaining and restoring freshwater habitats of evolutionary significant units of
anadromous salmonids in the Pacific Northwest. American Fisheries Society Symposium 17: 334-349.
Reid, S.M., M.G. Fox, and T.H. Whillans. 1999. Influence of turbidity on piscivory in largemouth bass
(Micropterus salmoides). Canadian Journal of Fisheries and Aquatic Sciences 56: 1362-1369.
Reynolds, J.B. 1996. Electrofishing, Chapter 8. In Fisheries Techniques, 2nd Edition. Edited by B.R.
Murphy and D.W. Willis. American Fisheries Society, Bethesda, MD.
Richards, R.J. 1959. A flexible growth function for empirical use. Journal of Experimental Botany 10:
290-300.
Richter, B.D., D.P. Braun, M.A Mendelson, and L.L. Master. 1997. Threats to imperiled freshwater
fauna. Conservation Biology 11: 1081-1093.
Ricker, W.E. 1979. Growth rates and models. In Fish Physiology, Vol. VIE. Edited by W.S. Hoar, D.J.
Randall and J.R. Brett. Academic Press, Inc., Orlando, FL. pp. 677-743
Rieman, B., J.T. Peterson, J. Clayton, P. Howell, R, Thurow, W. Thompson, and D. Lee. 2001.
Evaluation of potential effects of federal land management alternatives on trends of salmonids and their
71
-------
habitats in the interior Columbia River basin. Forest Ecology and Management 153: 43-62.
Robinson, J.G. and K.H. Redford. 1986. Body size, diet and population density of neotropical forest
mammals. American Naturalist 128: 665-680.
Rosgen, D. 1996. Applied River Morphology. Wildlife Hydrology, Pagosa Springs, CO.
Saiki, M.K. and B.A. Martin. 2001. Survey offish and environmental conditions in Abbotts Lagoon,
Point Reyes National Seashore, California. California Fish and Game 87: 123-138.
Sarle, W.S. 1983. Cubic Clustering Criterion. SAS Institute Inc, Gary, NC. SAS Technical Report A-108.
SAS. 1989. SAS/STAT® users guide, version 6, 4th ed., v. 2. SAS Institute Inc., Gary, NC.
Schnute, J. 1981. A versatile growth model with statistically stable parameters. Canadian Journal of
Fisheries and Aquatic Sciences 38: 1128-1140.
Shepherd, M.E. and M.T. Huish. 1978. Age, growth, and diet of the pirate perch in a coastal plain stream
of North Carolina. Transactions of the American Fisheries Society 107: 457-459.
Shubina, L.I. and T.L. Rychagova. 1981. Dynamics of fat and water metabolism in Caspian shad, Alosa
caspia caspia, in relation to its biological peculiarities (English Translation). Journal of Ichthyology
21(6): 123-127.
Smith, L.L., Jr and R.H. Kramer. 1964. The spottail shiner in the Lower Red Lake, Minnesota.
Transactions of the American Fisheries Society 93: 35-45.
Staples, D.J. and M. Nomura. 1976. Influence of body size and food ration on the energy budget of
rainbow trout Salmo gairdneri Richardson. Journal of Fish Biology 9: 29-43.
Stefan, H.G. andE.B. Preudhomme. 1993. Stream temperature estimation from air temperature. Water
Resources Bulletin 29: 27-45.
Steingrimsson, S.O. and J.W.A. Grant. 1999. Allometry of territory size and metabolic rate as predictors
of self thinning in young-of-year Atlantic salmon. Journal of Animal Ecology 68: 17-86.
Stiefvater, R.J. and S.P. Malvestuto. 1985. Seasonal analysis of predator-prey size relationships in West
Point Lake, Alabama-Georgia. Proceedings of the Annual Conference of the Southeastern Association of
Fish an 39: 19-27.
Stockwell, J.D. and O.E. Johannsson. 1997. Temperature-dependent allometric models to estimate
zooplankton production intemperate freshwater lakes. Canadian Journal of Fisheries and Aquatic
Sciences 54: 2350-2360.
Storck, T.W. 1986. Importance of gizzard shad in the diet of largemouth bass in Lake Shelbyville,
Illinois. Transactions of the American Fisheries Society 115: 21-27.
Swartzman, G.L. and R. Bentley. 1979. A review and comparison of plankton simulation models. ISEM
72
-------
Journal 1: 30-81.
Swedberg, D.V. 1968. Food and growth of the freshwater drum in Lewis and Clark Lake, South Dakota.
Transactions of the American Fisheries Society 97: 442-447.
Sweka, J.A. and KJ. Hartman. 2003. Reduction of reactive distance and foraging success in smallmouth
bass, Micropterus dolomieu, exposed to elevated turbidity levels. Environmental Biology of Fishes 67:
341-347.
Tandler, A. and F.W.H. Beamish. 1981. Apparent specific dynamic action (SDA), fish weight and level
of caloric intake in largemouth bass, Micropterus salmoides Lacepede. Aquaculture 23: 231-242.
Thornton, K.W. and A.S. Lessem. 1978. A temperature algorithm for modifying biological rates.
Transactions of the American Fisheries Society 107: 284-287.
Thurston, R.V. and P.C. Gehrke. 1993. Respiratory oxygen requirements of fishes: description of
OXYREF, a datafile based on test results reported in the published literature. In Proceedings of the
Second International Symposium on Fish Physiology, Fish Toxicology, and Water Quality Management.
Sacramento, CA, Sept. 18-20, 1990. Edited by R.C. Russo and R.V. Thurston. U.S. Environmental
Protection Agency, Office of Research and Development, pp. 95-108
Timmons, T.J., W.L. Shelton, and W.D. Davies. 1980. Differential growth of largemouth bass in West
Point Reservoir, Alabama-Georgia. Transactions ofthe American Fisheries Society 109: 176-186.
USEPA. 2000. Mid-Atlantic Highlands Streams Assessment. U.S. Environmental Protection Agency,
Washington, D.C. EPA-903-R-00-015.
USFWS. 1981. Standards for the development of habitat suitability index models for use in the Habitat
Evaluation Procedures. U.S. Department of Interior, Fish and Wildlife Service, Division of Ecological
Services. ESM 103.
Vadas, R.L., Jr. andD.J. Orth. 2001. Formulation of habitat suitability models for stream fish guilds: do
the standard methods work? Transactions ofthe American Fisheries Society 130: 217-235.
Vandenbyllaardt, L., F.J. Ward, C.R. Braekevelt, andD.B. Mclntyre. 1991. Relationships between
turbidity, piscivory, and development ofthe retina in juvenile walleyes. Transactions ofthe American
Fisheries Society 120: 382-390.
Vander Zanden, M.J., J.D. Olden, J.H. Thorne, andN.E. Mandrak. 2004. Predicting occurrences and
impacts of smallmouth bass introductions in north temperate lakes. Ecological Applications 14: 132-148.
Vogel, J.L. and D.A. Beauchamp. 1999. Effects of light, prey size, and turbidity on reaction distances of
lake trout (Salvelinus namaycush) to salmonid prey. Canadian Journal of Fisheries and Aquatic Sciences
56: 1293-1297.
Wanzenbock, J. and F. Schiemer. 1989. Prey detection in cyprinids during early development. Canadian
Journal of Fisheries and Aquatic Sciences 46: 995-1001.
73
-------
Ware, D.M. 1975. Growth, metabolism, and optimal swimming speed of a pelagic fish. Journal of the
Fisheries Research Board of Canada 32: 3341.
Waters, T.F., J.P. Kaehler, TJ. Polomis, and TJ. Kwak. 1993. Production dynamics of samallmouth bass
in a small Minnesota stream. Transactions of the American Fisheries Society 122: 588-598.
Weatherley, A.H. and H.S. Gill. 1983. Protein, lipid, water, and caloric contents of immature rainbow
trout, Salmo gairdneri Richardson, growing at different rates. Journal of Fish Biology 23: 653-673.
Webb, B.W., P.D. Clack, and D.E. Walling. 2003. Water temperature relationships in a Devon river
system and the role of flow. Hydrological Processes 17: 3069-3084.
Weidel, B.C., D.C. Josephson, and C.E. Kraft, in review. Fish community response to removal of
introduced smallmouth bass in an oligotrophic Adirondack lake.
http://www.dnr.cornell.edu/facstf/kraft/Publications/Weidel_et_al_CJFAS.pdf.
Werner, E.E. 1974. The fish size, prey size, handling time relation in several sunfishes and some
implications. Journal of the Fisheries Research Board of Canada 31: 1531-1536.
Werner, E.E. and J.F. Gilliam. 1984. The ontogenetic niche and species interactions in size-structured
populations. Annual Review of Ecology and Systematics 15: 393425.
Wilcove, D.S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying threats to imperiled
species in the United States. BioScience 48: 607-615.
Wilkinson, C.D. and D.R. Edds. 2001. Spatial pattern and environmental correlates of a Midwestern
stream fish community: including spatial autocorrelation as a factor in community analyses. American
Midland Naturalist 146: 271-289.
Williams, J.E., J.E. Johnson, D.A. Hendrickson, S. Contreras-Balderas, J.D. Williams, M. Navarro-
Mendoza, D.E. McAllister, and J.E. Deacon. 1989. Fishes of North America: endangered, threatened, or
of species concern. Fisheries 14: 2-20.
Wilson, K.W. and M.C. Belk. 2001. Habitat characteristics of leatherside chub (Gila copei) at two spatial
scales. Western North American Naturalist 61: 36-42.
Wolfe, G.W., B.H. Bauer, and B.A. Branson. 1978. Age and growth, length-weight relationships, and
condition factors of the greenside darter from Silver Creek, Kentucky. Transactions of the Kentucky
Academy of Science 39: 131-134.
Woodward, R.L. and T.E. Wissing. 1976. Age, growth, and fecundity of the quillback (Carpiodes
cyprinus) and highfin (C. velifer) carpsuckers in an Ohio stream. Transactions of the American Fisheries
Society 105:411-415.
Wrenn, W.B. 1968. Life history aspects of smallmouth buffalo and freshwater drum in Wheeler
Reservoir, Alabama. Proceedings of the Annual Conference of Southeastern Association of Fish and
Wildlife Agencies 22: 479-495.
74
-------
Wu, L. and D.A. Culver. 1992. Ontogenetic diet shift in Lake Erie age-0 yellow perch (Perca
flavescens): a size-related response to zooplankton density. Canadian Journal of Fisheries and Aquatic
Sciences 49: 1932-1937.
Yang, M.S. and P.A. Livingston. 1988. Food habits and daily ration of Greenland halibut, Reinhardtius
hippoglossoides, in the eastern Bering Sea. Fishery Bulletin 86: 675-690.
Yeager, B.L. 1993. Dams. In Impacts on Warmwater Streams: Guidelines for evaluation. Edited by C.F.
Bryan and DA. Rutherford. American Fisheries Society, Little Rock, AR. pp. 57-111
Zar, J.H. 1974. Biostatistical analysis. Prentice-Hall, Inc., Englewood Cliffs, NJ.
75
-------
Appendix A. Clustering Methods and Algorithms
A.I. Density-to-Biomass Conversion Algorithm
Although EMAP, like most fish surveys or monitoring programs, reported only species counts from
which fish densities (fish/ha) can be straightforwardly estimated, fish standing stocks (kg/ha) are also
important indicators of the condition offish assemblages as a whole. Fortunately, fish biomasses can be
estimated from observed densities if one makes certain simplifying assumptions concerning the
recruitment, mortality, and average body growth dynamics of the species of interest. These assumptions
include:
1) a species' observed density is functionally dependent on its mean body weight, and similarly
the density of each cohort is dependent on its mean body weight;
2) the mean body weights of a species and of its cohorts is determined primarily by the species
physiological growth rather than by size specific predation, environmentally induced mortality,
or dispersal; and
3) the recruitment strength for each cohort or year class within the observed population density
has been relatively constant or has been fluctuating around the long term average for the species.
Using these assumptions, a density-to-biomass conversion algorithm can be developed using the general
empirical observation that population densities of most vertebrates can be adequately characterized by
the self-thinning power function relationship
N= aW~b (A-l)
where TV and W denote the density (inds/area) and the mean body weight, respectively, of a population of
interest. Importantly, the population of interst can be either a collection of species (e.g., a guild or higher
taxonomic grouping), a single species, or the individual cohorts of a species. For fish populations, the
self-thinning exponent b generally varies from 0.75 to 1.5 (Boudreau and Dickie 1989, Grant and Kramer
1990, Gordoa andDuarte 1992, Elliott 1993, Bohlin et al. 1994, Randall et al. 1995, Dunham and
76
-------
Vinyard 1997, Grant et al. 1998, Dunham et al. 2000, Keeley 2003). Larger exponents ranging from 1.5
to 3.0, however, have also been reported (Steingrimsson and Grant 1999).
If Equation (A-l) is differentiated with respect to time, it immediately follows that
baW~b dW
dN_
dt
W dt
= - \iN
(A-2)
where y = W l dWIdt is the specific growth rate for individuals within the population; andp, = b y is
the population's lumped rate of mortality and dispersal. Readers should consult Peterson and Wroblewski
(1984), McGurk (1993, 1999), andLorenzen (1996) for detailed discussions of the theoretical
foundations and implications of Equation (A-2). Also see Section D.4 herein. Equation (A-2), however,
is also equivalent to
dN , dW
= - b
N
W
(A-3)
which can be reintegrated to obtain the following reformulation of Equation (A-l)
N(t) = N(tQ) exp
(A-4)
A species total population density can now be formulated by applying Equation (A-4) to each of its
cohorts, i.e.,
E
N(f) = E Nt(f)
- b ln
Wt(t- a,)
(A-5)
77
-------
where TV,, Wt, and a, denote the density, average body weight, and age, respectively, of the z'-th cohort. If
each cohort is assumed to be recruited into the population with the same initial body weight (W0) and
population density (TV,,), the preceding equation can be simplified to
7V(0 = 7VoEexp - b In
(A-6)
If the growth rate trajectories of each cohort have also remained relatively constant, it also follows that
N(t) = 7VoEexp - b In
Wt(at)
(A-7)
From this equation it should be reasonably clear that given the species current population density (TV) and
a reasonable model for the species body growth, one can straightforwardly calculate the species' apparent
long term year-class strength TV0. Having done so, the species' total biomass can then be estimated
by
B = E Wt(at) Nt(t) = NQ E r.(a.) exp - b In
Wn
(A-8)
Selecting an appropriate growth model to parameterize Equation (A-8), like most model selections, is not
a trivial concern since over the past 50 years at least four different models (i.e., von Bertalanffy,
Richards, Gompertz, and Parker-Larkin models) have become standard tools for characterizing the
growth of fishes. See Ricker (1979) for a detailed discussion of these models and other less commonly
used models.
According to the von Bertalanffy model, a fish's growth rate is the simple mass balance of anabolic
processes that are directly proportional to the fish's surface area and of catabolic processes that are
directly proportional to the fish's body weight. Assuming isometric growth (i.e., W= AL3), the fish's
growth dynamics is therefore governed by the following differential equation
78
-------
= 4, W™ - p W (A-9)
where 4> is the fish's rate of feeding and assimilation; and p is the fish's total metabolic rate. In terms of
body length, this model is also equivalent to
dL p , T T,
~ = -
where L is the fish's body length; and Lmai - 4> / (p A1/3) is the fish's "maximum" body length that is
obtained by setting Equation (A-9) to zero. For further discussion, see Parker and Larkin (1959) and
Paloheimo and Dickie (1965).
The Richard's model (Richards 1959) is a generalization of the von Bertalanffy model that relaxes the
assumption of isometric growth and strict proportionality between a fish's feeding/assimilatory processes
and its absorptive surface areas. In this model, the fish's feeding is simply assumed to be a power
functions of its body weight. The fish's growth is then described by the differential equation
4>, W *2 - p W (A-ll)
at
Although both the von Bertalanffy and the Richards models appear to have a strong physiological
foundation, a more critical inspection of the parameters of these models cast doubts on such assertions.
One particular point of contention is the assumption that a fish's metabolism (i.e., respiration and
excretion) is directly proportional to its body weight. Although this assumption is certainly satisfied or
closely approximated for some fish species, most fish species have metabolic demands that are best
described as power functions of their body weights. Consequently, from a purely physiologically-based
perspective a much better anabolic-catabolic process model for fish growth would be
See Paloheimo and Dickie (1965). Unlike the von Bertalanffy and Richards models, however, this model
79
-------
generally does not have a closed analytical solution. Furthermore, when this model is fit to observed data,
there is no a priori guarantee that the fitted exponents will actually match expected physiological
exponents unless the analysis is suitably constrained.
In light of these criticisms, simpler empirical growth models maybe more than adequate for most
applications. Two such models that have proved useful in this regard are the Gompertz and Parker-Larkin
models. Both of these models are intended to describe the growth of fishes that decreases with the age or
size of the individual. Whereas the Gompertz model describes fish growth by
dW , . „,
— = ej exp( - e2 0 W (A-13)
the Parker-Larkin model (1959) assumes that
dW W7p
~ = '
where the exponent p is generally less than 1.
Although each of the aforementioned models can describe very different growth trajectories, much of the
discussion surrounding their use has focused on whether the models predict asymptotically zero or
indeterminate growth (Parker and Larkin 1959, Paloheimo and Dickie 1965, Knight 1968, Schnute 1981).
Although growth rates of individual fish almost always decrease with increasing age or body size, Knight
(1968) argued that the traditional notion of asymptotically zero growth is seldom, if ever, supported by
studies that have focused on actual growth increments rather than on size at age. Because the Parker-
Larkin model is the only model outlined above that assumes that the growth of fish is fundamentally
indeterminate, this model has conceptual advantages over the von Bertalanffy, Richards, and Gompertz
models. The Parker-Larkin model also does not rely on the a priori assumption that the respiration of
fishes is generally a linear function of their body weight as does the von Bertalanffy and Richards
models.
Estimation of Fish Growth Rates
Wherever possible expected growth rates for MAH fish species were estimated using data summarized
80
-------
by Carlander(1969, 1977, 1997). The Carlander data summaries, however, were also supplemented with
the data sources summarized in Table A.l. Reported body lengths at age were converted to live body
weights using applicable weight-length regressions. Estimated live body weights were then fit to the
analytical solution Parker-Larkin growth model, i.e.,
W(f) = [ W(t0)1 - P + a (1 - P) (t - tQ)} 1/(1 - P) (A-15)
using the NL2SOLV non- linear regression and optimization software. However, because Equation (A-
15) is discontinuous at p=l, the growth parameters a and P for each species were actually obtained by
fitting calculated body weights to the equivalent expression
exp(6)
W(f) = [^0)exP(-*) + a exp(- b) (t - gjexp (A-16)
where exp (-b) - (1 - P). Results of these regressions are summarized in Table A.2.
81
-------
Table A.1 Data sources for supplementing Carlander (1969, 1977, 1997)
Species
Data source
Aphredoderus sayanus
Aplodinotus _grunniens
Carpiodes carpio
Carpiodes cyprinus
Etheostoma blennioides
Moxostoma anisurum
Moxostoma duquesnei
Notropis hudsonius
Noturus spp
Percopsis omiscomaycus
Percopsis transmontana
Rhinichthys atratulus
Rhinichthys cataractae
Semotilus corporalis
Shepherd and Huish (1978)
Dreves et al. (1996), Nelson (1974), Priegel (1969), Purkett (1957),
Swedberg (1968), Wrenn (1968)
Morris (1965), Nelson (1974), Purkett (1957)
Woodward and Wissing (1976)
Wolfe etal. (1978)
Hackney et al. (1970)
Bowman (1970)
McCann (1959), Smith and Kraemer (1964)
Burr and Mayden (1982), Mayden and Walsh (1984)
House and Wells (1973), Pereira and LaBar (1983)
Gray and Dauble (1979)
Reed and Moulton (1973)
Reed and Moulton (1973), Reed (1959), Kuehn (1949)
Reed(1971)
82
-------
Table A.2 Summary of growth data used for biomass estimation of MAH fish species
Species
Alosa pseudoharengus
Alosa sapidissima
Ambloplites rupestris
Ameiurus catus
Ameiurus melas
Ameiurus natalis
Ameiurus nebulosus
Aphredoderus say anus
Aplodinotus _grunniens
Campostoma anomalum
Carpiodes carpio
Carpiodes velifer
Catostomus commersoni
Centrarchus macropterus
Coregonus alpenae
Coregonus artedii
Coregonus clupeaformis
Coregonus hoyi
Coregonus kiyi
Coregonus sardinella
Cycleptus elongatus
Cyprinus carpio
Dorosoma cepedianum
Erimyzon oblongus
Erimyzon spp
Erimyzon sucetta
Esox americanus
Esox lucius
Esox masquinongy
Esox niger
Etheostoma blennioides
Etheostoma spp
Etheostoma zonale
Gila atraria
Hiodon alosoides
Hiodon tergisus
Hypentelium nigricans
Ictalurus furcatus
Ictalurus punctatus
Ictiobus bubalus
Ictiobus cyprinellus
Ictiobus niger
Lampetra spp
Lepisosteus osseus
Lepomis auritus
Lepomis cyanellus
Lepomis gibbosus
Lepomis gulosus
Lepomis humilis
Lepomis macrochirus
max cohorts
5
8
12
14
6
5
5
3
16
4
14
8
7
7
9
10
17
9
8
10
6
10
9
6
6
6
4
24
19
8
5
4
4
10
9
7
9
11
14
14
12
8
6
22
6
6
9
7
3
8
g(fw) range
10
-------
Table A.1 Continued
Species
Lepomis megalotis
Lepomis microlophus
Micropterus coosae
Micropterus dolomieu
Micropterus punctulatus
Micropterus salmoides
Minytrema melanops
Morone americana
Morone chrysops
Morone missis sippiensis
Morone saxatilis
Moxostoma anisurum
Moxostoma carinatum
Moxostoma duquesnei
Moxostoma erthrurum
Moxostoma
macrolepidotum
Moxostoma spp
Nocomis spp
Notemigonus crysoleucas
Notropis cornutus
Notropis hudsonius
Notropis spp
Noturus flavus
Noturus spp
Oncorhynchus aguabonita
Oncorhynchus clarld
Oncorhynchus mykiss
Osmerus mordax
Percaflavescens
Percina spp
Percopsis omiscomaycus
Percopsis transmontana
Pimephales spp
Polyodon spathula
Pomoxis annularis
Pomoxis nigromaculatus
Prosopium cylindraceum
Prosopium williamsoni
Rhinichthys atratulus
Rhinichthys cataractae
Rhinichthys spp
Salmo trutta
Salvelinus fontinalis
Salvelinus namaycush
Semotilus atromaculatus
Semotilus corporalis
Stizostedion canadense
Stizostedion vitreum
Thymallus arcticus
max cohorts
7
7
10
12
7
11
6
10
9
7
14
9
12
10
8
9
12
4
8
5
4
5
9
36
5
7
7
7
11
4
8
6
3
7
8
9
12
9
4
6
5
8
7
16
7
10
11
15
11
g(fw) range
3
-------
To validate the density-to-biomass conversion procedure outlined above, a database of studies that have
reported measured fish densities and associated fish biomasses was compiled from the literature (Quinn
1988, Reed and Rabeni 1989, Ensign et al. 1990, Buynak et al. 1991, Flick and Webster 1992, Bettoli et
al. 1993, Waters et al. 1993, Maceina et al. 1995, Mueller 1996, Allen et al. 1998, Radwell 2000,
Dettmers et al. 2001, Pierce et al. 2001, Habera et al. 2004). Reported fish densities were converted into
estimated biomasses assuming evenly spaced self-thinning exponents b ranging from -0.5 to -1.0 at 0.025
increments. Reduced major axis (RMA) regressions were then calculated for each assumed self-thinning
exponent. The self-thinning exponent that minimized the intercurve area between the calculated RMA
regression line and the identity relationship was 6=-0.800. This regression was
In B , = 0.846 In B - 0.147 (« = 499; r2 = 0.662)
B
obs
0.712 B.
0.846
'est
(A-17)
Figure A.1 displays the data for the regression (A-17) and the identity relationship Bobs = Besf.
Figure A.1 Plot of observed versus predicted biomasses. Indicated line represents the identity
relationship of observed biomass equals predicted biomass.
M 1-03 -
m
a
B
o
-5.31 -2.11 1.09 4.29
ln(biomassest)
density to biomass validation test
85
-------
Appendix B. BASS Bioenergetic and Population Dynamics Algorithms
B.I. Modeling Temperature Effects on Individual Growth
Because one of the principal consequences of riparian destruction/restoration on stream fishes centers on
their bioenergetic responses to altered stream temperatures, it is instructive to outline the temperature
response algorithm that BASS uses to predict fish feeding, metabolism, and growth before actually
describing how these processes are represented in BASS.
Although the temperature dependence of physiological processes are often described using an
exponential response equation, e.g.,
ri=
ro)] (B-l)
where rg and r, are the reaction rates of the process at temperatures T0 and Tl7 respectively, such
descriptions are generally valid only with a range of the organism's thermal tolerance. In most cases, the
process's reaction rate increases exponentially with increasing temperature up to a temperature Tl after
which it decreases. Moreover, in most cases the temperature at which a process's rate is maximal is very
close to the organism's upper thermal tolerance limit. To address this problem, Thornton and Lessem
(1978) developed a logistic multiplier to more realistically describe the temperature dependence of a
wide variety of physiological processes. Although this algorithm has been used successfully in a variety
offish bioenergetic models, BASS uses an exponential-type formulation that is assumed to respond
hyperbolically to increasing temperature.
Let P denote the rate of a physiological process and T\ denote the temperature at which the rate is at its
maximum value. If this process generally exhibits an exponential response to temperature changes well
below ri; then Equation (B-l) can be used to describe this process for rand T0 « T\, i.e.,
- Tn (B-2)
(B-3)
86
-------
where P0 is the process's rate at the low-end reference temperature T0. To incorporate the adverse effects
of high temperatures on this process, the right hand side of Equation (B-3) can be multiplied by a
hyperbolic temperature term that approaches unity as temperature decreases well below Tl7 equals zero at
ri; and becomes increasingly negative as temperatures approach the fish's upper thermal tolerance limit
TL = T2. Modifying Equation (B-3) in this fashion yields
dT
T- r
(B-4)
whose solution is
'o)]
T2- T
T - T
-* -*
(B-5)
If one assumes, without loss of generality, that T0 = 0, the preceding equation can be simplified to
(B-6)
Figure B.I displays the utility of this equation for describing the temperature dependence of the
maximum feeding rate of brown trout (Salmo trutta) as reported by Elliott (1 976b). Although rate
equations of this form apparently have not been used to describe physiological responses offish, results
summarized by Barber (2004a) clearly demonstrate their applicability for doing so. For other applications
of this model see Lassiter and Kearns (1974), Lassiter (1975), and Swartzman and Bentley (1979).
87
-------
Figure B.I Maximum daily ingestion ofbrown trout (Salmo trutta) as a function of temperature. Data from Elliott (1976b).
0.26
CO
Z^
CD
0.16 -
73
\
7.38
10.96 14.54
T[Celsius]
brown trout (Elliott 1976, Table 2)
B.2. Modeling Growth of Fish
As already mentioned, BASS simulates fish growth in terms of a dry weight mass balance of feeding,
egestion, respiration, and excretion. However, because BASS' bioaccumulation algorithms are
necessarily formulated in terms of the fish's live weight, and because many of BASS' basic physiological
parameters are generally determined with respect to the fish's live weight, BASS also calculates a fish's
wet weight (Wj) from its simulated dry weight using the following relationships:
Wf= Wa+
W
(B-7)
(B-8)
Pa=
(B-9)
-------
where Wa, Wd, Wt, and W0 denote the fish's aqueous, dry, lipid, and non-lipid organic weights,
respectively; andPa, Ph andP0 are the corresponding wet weight proportions of these components.
Whereas Equations (B-7) and (B-10) follow directly from mass conservation, Equations (B-8) and (B-9)
are purely statistical in nature. Although Equation (B-8) is assumed because simple power functions of
this form generally describe a wide variety of morphometric relationships for most organisms, Equation
(B-9) is based on the results of numerous field and laboratory studies (Eschmeyer and Phillips 1965,
Brett et al. 1969, Groves 1970, Elliott 1976a, Staples and Nomura 1976, Craig 1977, Shubina and
Rychagova 1981, Beamish and Legrow 1983, Weatherley and Gill 1983, Flath and Diana 1985, Lowe et
al. 1985, Kunisaki et al. 1986, Morishita et al. 1987).
BASS calculates a fish's realized feeding by first estimating its maximum ad libitum consumption and
then adjusting this potential by the availability of appropriate prey as described in the next section.
Because a wide variety of models and methods have been used to describe maximum feeding offish,
BASS is coded to allow a user the option of using any one of four different models to simulate the
feeding of any particular age / size class of fish. The first formulation that can be used is a temperature-
dependent power function
(B-H)
where fi,f2, P, T17 and T2 are empirical constants specific to the fish's feeding.
A commonly used alternative to the preceding allometric model is the Rashevsky-Holling model that is
defined by the equations:
Fmax= (Gmax- G)
89
-------
where 4> is the fish's ad libitum feeding rate (day'); Gmax is the maximum amount of food (g[nw]) that
the fish's stomach / gut can hold; G is the actual amount of food (g[nw]) present in the gut; and A
denotes the rate of food assimilation by the fish (Rashevsky 1959, Rolling 1966). The ad libitum feeding
rate 4> can be estimated using the following equations
dt
In
1 -
F(f)
max
(B-15)
where F(t) denotes the total amount of food consumed during the interval (0, t] (also see Dunbrack
1988).
For planktivores BASS can also estimate a fish's maximum ingestion rate using the clearance volume
model
where Y is the plankton standing stock (g[nw] / L); and Qcl is the planktivore's clearance volume (L /
day) that is assumed to be given by
( r^-7"1'
where q{, q2, P, Tl7 and T2 are empirical constants specific to the fish's filtering rate.
The fourth and final option is based on knowing the fish's projected growth and routine respiratory
demands. In particular, because assimilation, egestion, specific dynamic action, and excretion can be
90
-------
calculated as linear functions of feeding and routine respiration as discussed subsequently, it is then a
straightforward matter to calculate a fish's expected ingestion given its projected growth and respiration.
When a user elects this feeding option, BASS assumes that the fish's specific growth
ratey = Wf ! dWJdt (day"1) is given by
Li
where g1? g2, p, T17 and T2 are empirical constants specific to the fish's growth rate.
When BASS estimates a fish's feeding rate using Equations (B-ll), (B-16), or (B-18) , the fish's
assimilation and egestion are estimated as simple fractions of its realized ingestion F, i.e.,
A = ttfF (B-19)
flF (B-20)
where ay is the fish's net assimilation efficiency that is a weighted average of the fish's assimilation
efficiencies for invertebrate, piscine, and vegetative prey. However, when the Rashevsky-Holling feeding
model is used, BASS calculates these fluxes by substituting F with a function that describes the fish's
pattern of intestinal evacuation. The general form of this function is assumed to be
-r'
EV= eG
where e{, e2, p, Tlf and T2 are empirical constants specific to the fish's gastric evacuation.
The numerical value of this function's exponent, e2, depends both on characteristics of the food item
being consumed and on the mechanisms that presumably control gastro-intestinal motility and digestion
(Jobling 1981, 1986, 1987). For example, when gut clearance is controlled by intestinal peristalsis, e2
should approximately equal % since peristalsis is stimulated by circumferential pressure exerted by the
intestinal contents which, in turn, is proportional to the square root of its mass. On the other hand, when
91
-------
surface area controls the rate of digestion, e2 should be approximately either % or unity. If the fish
consumes a small number of large-sized prey (e.g., a piscivore), e2 = % may be the appropriate surface
area model. On the other hand, if the fish consumes a large number of smaller, relatively uniform-sized
prey (e.g., a planktivore or drift feeder), e2 = I is more appropriate since total surface area and total
volume of prey become almost directly proportional to one another.
A fish's specific dynamic action, i.e., the respiratory expenditure associated with the digestion and
assimilation of food, is modeled as a constant fraction of the fish's assimilation. In particular,
SDA = a A (B-22)
where a is generally on the order of 0.15 to 0.20 (Ware 1975, Tandler and Beamish 1981, Beamish and
MacMahonl988).
BASS assumes that body weight losses via metabolism are due entirely to the respiration of carbon
dioxide and the excretion of ammonia. Respiratory losses R are calculated from a fish's routine oxygen
consumption, R02 (g(O2) / day) using respiratory quotients RQ (L [CO2 ] respired) / L [O2] consumed) as
follows
12 e C m°le
R= 1Z c
mole CO2 22.4 L CO2
22.4 L O7 mole 0, 17 (B"23)
- -• - -'Rn7= —
mole 0. 32 g O. °2 32
BASS calculates a fish's routine oxygen consumption as a constant multiple of its basal or standard
oxygen consumption (Ware 1975) that is specified using the temperature-dependent power function
6, ( 7-W2-rD
R02= Z^/exptpr, 1- — (B-24)
where bt, b2, p, T:, and T2 are empirical constants specific to the fish's oxygen consumption. Although
ammonia excretion could be modeled using an analogous function (Paulson 1980, du Preez and Cockroft
1988a, b), BASS formulates this flux as a constant fraction of the fish's total respiration assuming that
92
-------
fish maintain a constant nitrogen / carbon ratio NC (g[N] / g[C]). Thus, BASS estimates a fish's
excretory loss in body weight as
EX= eNC(R+ SDA) (B-25)
where e = 17/14is the ratio of the molecular weight of ammonia to that of nitrogen.
B.3. Modeling Trophic Interactions and Predatory Mortalities
BASS is designed to simulate aquatic food webs in which each age class of a species can feed upon other
fish species, benthos, incidental terrestrial insects, periphyton / attached algae, phytoplankton, and
zooplankton. The realized feeding of any given age class of fish is determined by the estimated maximum
feeding rate of individuals within the cohort, the cohort's population size, and the biomass of prey that is
available to the cohort; the later quantity being the sum of the current biomass of potential prey minus the
biomass of potential prey that is expected to be consumed by other fish cohorts that are more efficient
foragers / competitors. BASS ranks the competitive abilities of different cohorts using the following
assumptions:
ASSUMPTION 1. The competitive abilities and efficiencies of benthivores and piscivores are positively
correlated with their body sizes (Garman and Nielsen 1982, East and Magnan 1991). Two general
empirical trends support this assumption. The first of these is the trend for the reactive distances,
swimming speeds, and territory sizes of fish to be positively correlated with their body size (Minor and
Grossman 1978, Breck and Gitter 1983, Wanzenbock and Schiemer 1989, Grant and Kramer 1990, Miller
et al. 1992, Keeley and Grant 1995, Minns 1995). Given two differently sized predators of the same
potential prey, these trends would suggest that the larger predator is more likely to encounter that prey
than is the smaller. Having encountered the prey, the other general trend for prey handling times to be
inversely correlated with body size (Werner 1974, Miller et al. 1992) suggests that the larger predator
could dispatch the prey and resume its foraging more quickly than the smaller predator.
ASSUMPTION 2. Unlike benthivores and piscivores, the competitive abilities and efficiencies of
planktivores are inversely related to their body size due to their relative morphologies (Lammens et al.
1985, Johnson and Vinyard 1987, Wu and Culver 1992, Persson and Hansson 1999). Consequently,
"large" planktivores only have access to the leftovers of "small" planktivores.
93
-------
BASS calculates the relative frequencies {...,di,...} of the different prey consumed by each cohort using
dietary electivities, i.e.,
ei = TTT (B-26)
where f ] is the relative availability of the z'-th prey with respect to all other prey consumed by the cohort.
One can easily verify that the range of dietary electivities is - 1 < e. < 1 . One can also verify that if the
fish does not eat a potential food item i that et= - 1 . Similar, if the fish consumes a potential prey item i
in direct proportion to the prey's relative abundance, then et= 0. BASS actually allows users to specify a
fish's diet as either a set of fixed dietary frequencies {...,di ,...}, a set of electivities {...,et,...} , or a
combination of fixed frequencies and electivities {...,di ,...,e ,...} . To calculate a cohort's realized
dietary composition, however, BASS converts all user supplied fixed dietary frequencies into their
equivalent electivities using the current simulated relative abundances {...,ft ,...} of the cohort's prey.
These electivities are then combined with any user specified electivities to form a set of unadjusted
electivities {...,et,...} that is subsequently converted into a consistent set of realized
electivities {...,ef ,...} . Using these realized electivities, BASS then calculates the cohort's realized
dietary frequencies using
4 =
1 +
1 - e
ft
(B-27)
The important step in this computational process is the conversion of the unadjusted electivities into a set
of realized electivities. Readers interested in the details of this process should consult Barber (2001,
2004a).
Because numerous studies have shown that there is generally a strong positive correlation between the
body sizes of piscivorous fish and the forage fish that they consume (Parsons 1971, Lewis et al. 1974,
94
-------
Timmons et al. 1980, Gillen et al. 1981, Knight et al. 1984, Moore et al. 1985, Stiefvater and Malvestuto
1985, Storck 1986, Jude et al. 1987, Johnson et al. 1988, Yang and Livingston 1988, Brodeur 1991, Elrod
and O'Gorman 1991, Hambright 1991, Juanes et al. 1993, Mattingly and Butler 1994, Hale 1996,
Madenjian et al. 1998, Margenau et al. 1998, Mittelbach and Persson 1998, Bozek et al. 1999), when
BASS uses the above procedure to calculate piscivorous interactions, only a specific size range of forage
fish are assumed to be available to a piscivorous cohort. To determine this size range, BASS first
assumes that the mode of the body lengths of forage fish ingested by a piscivore is given by
(B-28)
where A1; A2, and A3 are empirical constants specific to the mode of the predator's ingestable prey length.
Using this relationship, BASS then assumes that the minimum length of prey consumed by the piscivore
is given by
£-,-= 0.5Zmo& (B-29)
mm
As like the mode, the maximum length of prey consumed by the piscivore is assumed to be given by an
equation of the form
where A1; A2, and A3 are again empirical constants specific to the predator's maximum ingestable prey
length. All relative frequencies dt of forage fish in the diet of a piscivorous cohort are then calculated
relative to sum of forage fish biomasses whose body lengths are both greater than Lmin and less than Lmax
minus the biomass of those prey sizes that are predicted to be consumed by more efficient piscivorous
cohorts (see Assumption 1).
When two or more cohorts of a forage species i can be consumed by a piscivore, the relative frequencies
of those cohorts stj in the piscivore's diet are calculated assuming that prey sizes follow a simple
triangular distribution defined by Equations (B-28) - (B-30). For example, letLn andL,2 denote the body
lengths of two age classes of species that are prey for the cohort. If Ptj denotes the triangular distribution
95
-------
function
^ max min/ V mode min/
2 Z "
^ max min-' v max mode'
for LvLmode
(B-31)
the relative frequencies of these two age classes in the cohort's diet are calculated to
/7 = di [PiI/(Pil + Pi2)} and sa = df [Pi2/(Ptl + -P,2)] • If only one a§e class °f a forage species is
vulnerable to the cohort, then sf = dr
If during the calculation of the dietary frequencies of a piscivorous cohort BASS predicts that the
cohort's available prey is insufficient to satisfy its desired level of feeding, BASS reassigns the cohort's
unadjusted electivities {...,et,...} in a manner to simulate prey switching. These reassignments are based
on the following assumption:
ASSUMPTION 3. When forage fish become limiting, piscivores switch to benthic macroinvertebrates or
incidental terrestrial insects as alternative prey. However, piscivores that must switch to benthos or that
routinely consume benthos in addition to fish, are less efficient benthivores than are obligate benthivores
(Hanson and Leggett 1986, Lacasse and Magnan 1992, Bergman and Greenberg 1994). Consequently,
only the leftovers of non-piscivorous benthivores are available to benthic feeding piscivores. If such
resources are still insufficient to satisfy the piscivores' metabolic demands, piscivores are assumed to
then switch to planktivory (Werner and Gilliam 1984, Magnan 1988, Bergman and Greenberg 1994). In
this case, piscivores have access only to the leftovers of non-piscivorous planktivores. Using this
assumption, BASS first assigns the cohort's electivity for benthos to 0 regardless of its previous value.
BASS also reassigns any other electivity which does not equal -1, to 0.
If benthos become limiting for benthivores, or if plankton becomes limiting for planktivores, BASS
assumes that benthivores can shift their diets to include plankton and terrestrial insects and that
96
-------
planktivores can shift their diets to include benthos and terrestrial insects. See, for example, Ingram and
Ziebell(1983)
After BASS has calculated a cohort's dietary composition, it then assigns its realized feeding rate
adjusted for prey availability as
(8.32)
I V1
where Fmax is the cohort's maximum or desired individual ingestion, TV is the cohort's population size, and
ABj is the biomass of prey 7 that is available to that cohort. Using its predicted dietary compositions and
realized feeding rates, BASS then calculates the predatory mortalities for each cohort and non-fish biotic
resource.
B.4. Modeling Dispersal, Non-Predatory Mortalities, and Recruitment
The algorithm that BASS employs to simulate a species' dispersal and non-predatory mortality is based
on the general empirical observation that population densities of most vertebrates can be adequately
characterized by the self-thinning power function relationship
N=aW~b (B-33)
where TV is the species' density (fish / area) and f^is the species' mean body weight (Damuth 1981,
Peters and Raelson 1984, Juanes 1986, Robinson and Redford 1986, Dickie et al. 1987, Boudreau and
Dickie 1989, Gordoa and Duarte 1992, Randall et al. 1995, Dunham and Vinyard 1997, Steingrimsson
and Grant 1999, Dunham et al. 2000). For fish species, the body weight exponent b generally varies from
0.75 to 1.5(Boudreau and Dickie 1989, Grant and Kramer 1990, Gordoa and Duarte 1992, Elliott 1993,
Bohlin et al. 1994, Randall et al. 1995, Dunham and Vinyard 1997, Grant et al. 1998, Dunham et al.
2000, Knouft 2002, Keeley 2003). Larger exponents ranging from 1 .5 to 3.0, however, have also been
reported (Steingrimsson and Grant 1999). If Equation (B-33) is differentiated with respect to time, it
immediately follows that a species' population dynamics can be modeled using the time varying linear
differential equation
97
-------
dN -abW~b dW , ..
(B-34)
where y is the species' specific growth rate. Consequently, by corresponds to the cohort's total mortality
rate. Readers interested in detailed discussions concerning the underlying process-based interpretation
and general applicability of this result should consult Peterson and Wroblewski (1984), McGurk (1993,
1999), and Lorenzen (1996).
Because Equations (B-33) and (B-34) encompass the species predatory mortality, non-predatory
mortality, and dispersal, and because BASS explicitly models the cohort's predatory mortality, BASS
assumes that the cohort's rate of non-predatory mortality and dispersal is simply a fraction 8 of by. In
particular,
EM+ NM= 6byN (B-35)
If community population dynamics are strongly dominated by predation, the fraction 8 will be "small"
(e.g., 8 < 0.5) for forage fishes and "large" (e.g., 8 > 0.5) for predatory species. However, if community
population dynamics are dominated by dispersal mechanisms related to competition for food, space, or
other limiting community resource, the fraction 8 will be large for both forage and predatory species
alike.
BASS estimates a species' recruitment by assuming that each species turns over a fixed percentage of its
potential spawning biomass into new young-of-year (YOY). This percentage is referred to as the species'
reproductive biomass investment (rbi). The species' spawning biomass is defined to be the total biomass
of all cohorts whose body lengths are greater than or equal to a specified minimum value (tl_rO) marking
the species' sexual maturation. When reproduction is simulated, the body weight of each sexually mature
cohort is decreased by its rbi and the total number of YOY that are recruited into the population as a new
cohort is estimated by simply dividing the species' spawned biomass by the species' characteristic YOY
body weight. Although this formulation does not address the myriad of factors known to influence
population recruitment, it is logically consistent with the spawners abundance model for fish recruitment.
See Myers and Barrowman (1996) and Myers (1997).
-------
B.5. Modeling Habitat Effects
Although BASS does not explicitly model physical habitat features of the fish community of concern, it
does allow users to specify habitat suitability multipliers on the feeding, reproduction / recruitment, and
dispersal / non-predatory mortality for any or all species. Because these multipliers are assumed to be
analogous to subcomponents of habitat suitability indices, they are assumed to take values from 0 to 1. If
these multipliers are not specified by the user, BASS assigns them the default value of 1.
When feeding habitat multipliers are specified, BASS uses the specified parameters as simple linear
multipliers on the fish's maximum rate of ingestion, i.e.,
-*max, habitat = "^feeding ''max (B-36)
The resulting adjusted maximum feeding rate (Fmax habita) then replaces Fmax in Equation (B-32). These
multipliers are assumed to modify the fish's ability to perceive or to intercept prey either by effecting the
fish's reactive distance etc. or by providing modified refuges for its potential prey. Habitat interactions
that actually change the abundance of potential prey should not be specified as feeding habitat multipliers
since these interactions are automatically addressed by the algorithms outlined in Section B.3.
Like the aforementioned feeding habitat multipliers, BASS uses any specified recruitment habitat
multipliers as simple linear multipliers on the number of young-of-year that is recruited into the species
population, i.e.,
N0, habitat = ^Recruitment N0 (B-37)
These multipliers can represent either the availability of suitable spawning sites or the ability of the
otherwise successful spawns to result in the expected numbers of young-of-year as discussed in Section
B.3.
Finally, when habitat multipliers are specified for dispersal / non-predatory mortality, the specified
values are assumed to act as hyperbolic multipliers as follows:
(EM+ NM)habUat = { ~^L-\ N (B-38)
I, H^survival)
99
-------
See Equation (B-35). Thus, as habitat suitability decreases dispersal and non-predatory mortality
increases and vicea versa.
Because constructing or estimating the aforementioned habitat suitability multipliers in a general or
standard way is not a trivial issue, BASS relegates their construction to the user. Nevertheless there are
several obvious starting points that users might consider when simulating habitat effects using BASS.
Turbidity, for example, is known the effect the foraging abilities of both prey and predatory fishes, and
one could readily use results of published studies (e.g. Vandenbyllaardt et al. 1991, Barrett 1992,
Gregory 1993, Gregory and Northcote 1993, Miner and Stein 1996, Reid et al. 1999, Vogel and
Beauchamp 1999, Bonner and Wilde 2002, Sweka and Hartman2003) to estimate feeding multipliers for
Equation (B-36) as power functions or polynomials of turbidity. Field-based HSI's are often estimated
by logistic regression of presence-absence data without specifying the underlying mechanisms that
actually determine habitat suitability for a species. Such HSI's could be used as habitat multipliers for a
species' recruitment (Equation (B-37)) or persistence/survival (Equation (B-38)) or depending on the
user's own interpretation of what the indices most likely represent.
B.6. Modeling Non-fish Compartments
BASS assumes that the non-fish components of a community of concern can be treated as 4 lumped
compartments, i.e., benthos, periphyton/attached algae, phytoplankton, and zooplankton. These
compartments can be treated either as community forcing functions or as bona fide state variables. In the
later case, the required compartmental dynamics are simulated using the simple mass balance model
IP-R-F-M (B-39)
where Fis the compartment's biomass (g[nw] / m2) ; and the fluxes IP, R, F, andM, all in units of
(g[ow] -day"1 • m"2), denote the compartment's ingestion or photosynthesis, respiration, predatory
mortality due to fish consumption, and other mortality and dispersal, respectively.
BASS formulates expressions for a compartment's ingestion, photosynthesis and respiration, by first
formulating these processes for the individuals that comprise the compartment. In particular, BASS
100
-------
assumes that the ingestion, photosynthesis, and respiration by individuals within the aforementioned
compartments can be adequately described by temperature dependent power functions of the form
( T\i(T2-Tv
P= ofFjexpiYT-) 1 - -i- (B-40)
where Wd denotes the individual's gram-dry weight. Compartmental expressions for ingestion,
photosynthesis, and respiration are then obtained by simply multiplying these individual-based power
functions by the estimated numbers of individuals within the compartment.
For example, consider the following formulation of benthos consumption. If Wd denotes the average dry
weight of individuals comprising the benthos compartment, it then follows that the expected density of
individuals within the benthos compartment is simply
If the individual ingestion of benthic invertebrates is now given by the simplified power function
r) (B-42)
it also follows that the compartmental consumption of the benthos is given by
IP= CN= rafF
W
d (B-43)
Formulating compartmental ingestion, photosynthesis, and respiration by this method not only leads to an
objective procedure to parameterize BASS, but also produces production estimates that are logically
consistent with results reported by Plante and Downing (1989), Stockwell and Johannsson (1997), and
Kuns and Sprules (2000).
101
-------
Although BASS does not simulate the average individual body sizes of benthos, periphyton/attached
algae, phytoplankton, and zooplankton, users can specify these parameters as functions of time.
102
------- |