United States
Environmental Protection
Agency
Policy, Planning,
And Evaluation
(2122)
EPA 220-R-95-004
April 1995
Ecological Impacts From
Climate Change:
An Economic Analysis Of
Freshwater Recreational Fishing
"JET. v-i;.j--~---••.«*.«•<
^^iK^^iCjM^S-'"^.
v>^**St«, ,:
-------
Recycled/Recyclable • Printed with Vegetable Based Inks on Recycled Paper (20% Postconsumer)
-------
Ecological Impacts From Climate Change:
An Economic Analysis of Freshwater Recreational Fishing
A Report Prepared for
The United States Environmental Protection Agency
Office of Policy, Planning and Evaluation
Climate Change Division
Adaptation Branch
EPA-230-R-95-004
Project Manager
Susan S. Herrod
Chief, Adaptation Branch
Dr. Joel D. Scheraga
-------
-------
Ecological Impacts From Climate Change:
An Economic Analysis of Freshwater Recreational Fishing
Authors
Greg Michaels
Kirk O'Neal
Joanna Humphrey
Kathleen Bell
Rodolfo Camacho
Rob Funk
Abt Associates, Inc.
EPA-230-R-95-004
-------
-------
Photo Credits - Cover Page
Elizabeth Joy: NFS (Fly fisherman)
William S. Keller: NFS (man fishing)
Steve Delaney: EPA (girl fishing)
-------
-------
TABLE OF CONTENTS
Table of Contents i
List of Exhibits iii
Acknowledgements vii
Preface viii
Executive Summary . ix
1. INTRODUCTION 1-1
2. HABITABILITY ASSESSMENT FOR RECREATIONAL FISH TN RIVERS
AND STREAMS
2.1 SAMPLE SITES 2-2
2.1.1 Assigning Fishable Acres to Representative Locations 2-2
2.1.2 Estimating Water Temperatures 2-7
2.2 THERMAL TOLERANCES AND GUILD ASSIGNMENTS FOR FISH
SPECIES 2-14
2.3 FISH PRESENCE AT SAMPLE LOCATIONS 2-21
2.3.1 Checking Model Predictions for Baseline Presence 2-21
2.3.2 Other Determinants of Habitat: Development of "Screens" .... 2-33
2.3.3 Effects of Climate Change on Maximum Temperatures ...... 2-34
Using GCMs to Predict Changes to Fish Presence ......... 2-34
2.4 RESULTS 2-43
3. AN ECONOMIC ASSESSMENT OF RECREATIONAL FISHING IMPACTS
3.1 INTRODUCTION 3-1
3.2 ECONOMIC MODEL 3-2
3.3 ALTERNATIVES TO THE SELECTED MODELLING APPROACH . . 3-4
3.4 IMPLEMENTATION OF THE ECONOMIC MODEL 3-5
3.5 ANNUAL ECONOMIC WELFARE EFFECTS, SELECTED YEARS
AND MODELS . 3-19
3.6 PRESENT DISCOUNTED ECONOMIC WELFARE EFFECTS .... 3-28
4. EVALUATION OF THE ANALYTICAL FRAMEWORK
4.1 INTRODUCTION 4-1
4.2 FISHING-DAY VALUES 4-1
4.3 CLIMATE AND EMISSIONS SCENARIO SENSITIVITY 4-4
4.3.1 Equilibrium models 4-5
-------
4.3.2 Transient models .............................. 4-9
4.4 FISH THERMAL TOLERANCE DESIGNATIONS ............ 4-12
4.4.1 Equilibrium Models 4-12
4.4.2 Transient Models 4-15
4.5 FISH HABITAT DESIGNATIONS . 4-17
4.5.1 Equilibrium models 4-18
4.5.2 Transient Models ............................. 4-21
4.6 WARM-WATER FISHING BEHAVIOR ................... 4-25
4.7 COLD-WATER SUBSTITUTABILITY .................... 4-28
4.8 POTENTIAL IMPACTS ON RECREATIONAL FISHING FROM
CHANGES IN RUNOFF 4-29
4.8.1 Climate Change Effects on Runoff 4-29
4.8.2 Implications of Runoff Changes in Fish Recreational Values . . . 4-33
4.8.3 Economic Implications of Runoff Changes .............. 4-39
4.9 SUMMARY AND CONCLUSION ...................... 4-40
Appendix A Descriptions of Species Included in Study
Appendix B GISS Equilibrium Scenario
U.S. Maps Showing Impacts on Recreational Fish
Appendix C OSU Equilibrium Scenario
U.S. Maps Showing Impacts on Recreational Fish
Appendix D UKM Equilibrium Scenario
U.S. Maps Showing Impacts on Recreational Fish
Appendix E Transient GFDL Scenario - 2050
U.S. Maps Showing Impacts on Recreational Fish
Appendix F Transient GFDL Scenario - 2100
U.S. Maps Showing Impacts on Recreational Fish
Appendix G Summary of Selected Economic Studies of the Value
of Freshwater Fishing Days by Region
-------
LIST OF EXHIBITS
Exhibit 1-1
Exhibit 1-2
Exhibit 1-3
Exhibit 2-1
Exhibit 2-2
Exhibit 2-2
Exhibit 2-3
Exhibit 2-4
Exhibit 2-5
Exhibit 2-6
Exhibit 2-7
Exhibit 2-8
Exhibit 2-9
Exhibit 2-10
Exhibit 2-11
Exhibit 2-12
Exhibit 2-13
Exhibit 2-14
Exhibit 2-15
Exhibit 2-16
Exhibit 2-17
Exhibit 2-18
Exhibit 2-19
Exhibit 2-20
Exhibit 2-21
Exhibit 2-22
Exhibit 2-23
Exhibit 2-24
Exhibit 2-25
Exhibit 2-26
Exhibit 2-27
Exhibit 2-28
Exhibit 2-29
Exhibit 2-30
Organization of the Analysis: General Overview ............... 1-2
Organization of the Analysis: Details on Thermal Component .' 1-4
Organization of the Analysis: Details on Economic Component ...... 1-5
Geographic Distribution of Stations ........... ...........
Fishable Acres of Rivers and Streams by State ...............
(continued) Fishable Acres of Rivers and Streams by State ........
Maximum Weekly Average Air Temperature .......... ......
Maximum Weekly Average Water Temperature (Method 2) .......
Maximum Weekly Average Water Temperature (Method 3) .......
Comparison of Laboratory and Field-Derived Thermal Tolerance
Values .................... . ...... .............
Laboratory and Field-Derived Thermal Tolerance Values for 32 Fish
Thermal Tolerance Values Used in Model ............... ...
Additional Species in FTDMS Database ... ....... ..... .....
Natural Ranges for Cold-Water Guild .....................
Natural Ranges for Cool- Water Guild ..... ................
Natural Ranges for Warm- Water Guild .................. . .
Natural Ranges for Rough Guild . . ................. .....
Comparison of Habitat Definitions (Method 1) ...............
Comparison of Habitat Definitions (Method 2) ...............
Comparison of Habitat Definitions (Method 3) ......... . .....
Baseline Habitability for Cold-Water Fish ..................
Baseline Habitability for Cool-Water Fish ..................
Baseline Habitability for Warm-Water Fish ... ...............
Baseline Habitability for Rough Fish ........... ..... . .....
Baseline Habitability by Guild .................. ...... . .
Global Annual Mean Temperature Increases Used to Scale Equilibrium
GCM Results .... ...... ..........................
Global Annual Temperature Increases Used to Scale Transient GCM
Results . ....... ... .............................
Average Increment to Maximum Weekly Temperature from CO2
Doubling ......................................
Maximum Weekly Average Temperature After CO2 Doubling (GS1) . .
Loss of Habitat by Guild (GS1) .........................
Loss of Habitat for Cold-Water Fish (GS1) .................
Loss of Habitat for Cool-Water Fish (GS1) .................
Loss of Habitat for Warm-Water Fish (GF1) .......... ...... .
Loss of Habitat for Rough Fish (GF1) .......... ...........
2-4
2-5
2-6
2-9
2-10
2-13
2-15
2-20
2-23
2-24
2-25
2-26
2-27
2-28
2-29
2-30
2-31
2-36
2-37
2-38
2-39
2-40
2-42
2-43
2-44
2-45
2-46
2-47
2-48
2-49
2-50
-------
Exhibit 3-1
Exhibit 3-2
Exhibit 3-3
Exhibit 3-4
Exhibit 3-5
Exhibit 3-6
Exhibit 3-7
Exhibit 3-8
Exhibit 3-9
Exhibit 3-10
Exhibit 3-11
Exhibit 3-12
Exhibit 3-13
Exhibit 3-14
Exhibit 3-15
Exhibit 3-16
Exhibit 4-1
Exhibit 4-2
Exhibit 4-3
Exhibit 4-4
Exhibit 4-5
Exhibit 4-6
Exhibit 4-7
Exhibit 4-8
Distribution of Acreage by Fishing Type, Baseline and GCM
Simulations
First Stage: Predicting the Probability of General Fishing Participation1
Second Stage: Predicting the Probability of Participation by Fishing
Category1
Predicting the Number of Person Days Per Year Devoted to Fishing by
Guild1
Net Economic Values per Recreation Day Reported by TCM and CVM
Demand Studies from 1968 to 1988 (1993 Dollars)1
Fishing-Day Value Specifications 1993 Dollars
Estimated Annual Economic Welfare Effects Primary Specification
(Damages) and Benefits in Millions of Dollars
Changes in Fishing Days by Type—Different Equilibrium GCMs ....
Estimated Annual Economic Welfare Effects Specification: Equates Cool-
and Warm-water Fishing Day Values1
Estimated Annual Economic Welfare Effects Specification: Equates Cool-
and Warm-water Fishing Day Values
Estimated Annual Economic Welfare Effects Specification Equates Cool-
and Warm-Water Fishing Day Values
Changes in Fishing Days by Type—Different Transient Benchmarks . .
Present Discounted Value: Equilibrium GCM— Assumed Distribution of
Damages
Present Discounted Value: Transient GCM Assumed Distribution of
Damages
Present Discounted Welfare Effects: Net (Damages) or Benefits Primary
Specification
Present Discounted Welfare Effects: Net (Damages) or Benefits
Specification: Equal Values of Cool- and Warm-water Fishing Days . .
Assessing Sensitivity to Fishing-Day Values
Global Annual Mean Temperature Increases Used to Scale Equilibrium
GCM Results
Alternate Climate Sensitivity Assumptions for Primary Fishing-day Values
Annual (Damages) and Benefits in Millions of Dollars
Global Annual Temperature Increases Used to Scale Transient GCM
Results
Alternate Climate Sensitivity Assumptions for Primary Fishing Day
Values .
High, Primary, and Low Thermal Tolerance Results for Primary Fishing-
Day Values
High, Primary, and Low Thermal Tolerance Results for Primary Fishing-
Day Values (Damages) and Benefits hi Millions of Dollars
Narrow and Wide Screen Results for Primary Fishing-Day Values . . .
3-10
3-12
3-13
3-15
3-17
3-19
3-22
3-23
3-24
3-25
3-26
3-27
3-30
3-31
3-32
3-33
4-3
. 4-6
4-7
4-10
4-11
4-13
4-16
4-19
IV
-------
Exhibit 4-9 Narrow and Wide Screen Results for Low Fishing-Day Values1 .....
Exhibit 4-10 Narrow and Wide Screen Results for High Fishing-Day Values .....
Exhibit 4-11 Sensitivity Analysis of the Model's Treatment of Warm-Water Fishing
Days .
Exhibit 4-12 Sensitivity Analysis of the Model's Treatment of Warm-Water Fishing
Days
Exhibit 4-13 Sensitivity Analysis of the Model's Treatment of Warm-Water Fishing
Days
Exhibit 4-14 Sensitivity Analysis of the Model's Treatment of Cold-Water Acreage
Transitions
Exhibit 4-15 Sensitivity Analysis of the Model's Treatment of Cold-Water Acreage
Transitions .....................................
Exhibit 4-16 Sensitivity Analysis of the Model's Treatment of Cold-water Acreage
Transitions
Exhibit 4-17 Effects of Global Climate Change on Runoff
Exhibit 4-18 Marginal Values Per Acre-Foot in 1980 dollars
Exhibit 4-19 Sensitivity Analysis Comparison: Relative Changes in the Dollar Value
and the Absolute Total Dollar Value By Global Climate Change
Scenario .......................................
Exhibit B-l Maximum Weekly Average Temperature at Doubled CO2 (GISS) ....
Exhibit B-2 Loss of Habitability by Guild (GISS) .....................
Exhibit B-4 Loss of Habitability for Cold Water Species (GISS)
Exhibit B-5 Loss of Habitability for Cool Water Species (GISS)
Exhibit B-5 Loss of Habitability for Warm Water Species (GISS)
Exhibit B-6 Loss of Habitability for Rough Water Species (GISS)
Exhibit C-l Maximum Weekly Average Temperature at Doubled CO2 (OSU) ....
Exhibit C-2 Loss of Habitability by Guild (OSU)
Exhibit C-4 Loss of Habitability for Cold Water Species (OSU)
Exhibit C-5 Loss of Habitability for Cool Water Species (OSU)
Exhibit C-5 Loss of Habitability for Warm Water Species (OSU)
Exhibit C-6 Loss of Habitability for Rough Water Species (OSU)
Exhibit D-l Maximum Weekly Average Temperature at Doubled CO2 (UKMO) . . .
Exhibit D-2 Loss of Habitability by Guild (UKMO)
Exhibit D-4 Loss of Habitability for Cold Water Species (UKMO)
Exhibit D-5 Loss of Habitability for Cool Water Species (UKMO)
Exhibit D-5 Loss of Habitability for Warm Water Species (UKMO) ..........
Exhibit D-6 Loss of Habitability for Rough Water Species (UKMO) ..........
Exhibit E-l Maximum Weekly Average Temperature at Doubled CO2 (GFDL 2050)
Exhibit E-2 Loss of Habitability by Guild (GFDL 2050)
Exhibit E-4 Loss of Habitability for Cold Water Species (GFDL 2050)
Exhibit E-5 Loss of Habitability for Cool Water Species (GFDL 2050)
Exhibit E-5 Loss of Habitability for Warm Water Species (GFDL 2050)
4-22
4-23
4-26
4-27
4-27
4-30
4-31
4-32
4-34
4-39
4-42
B-l
B-2
B-3
B-4
B-5
B-6
C-l
C-2
C-3
C-4
C-5
C-6
D-l
D-2
D-3
D-4
D-5
D-6
E-l
, E-2
E-3
E-4
E-5
-------
Exhibit E-6 Loss of Habitability for Rough Water Species (GFDL 2050) ........ E-6
Exhibit F-l Maximum Weekly Average Temperature at Doubled CO2 (GFDL 2100) . F-l
Exhibit F-2 Loss of Habitability by Guild (GFDL 2100) F-2
Exhibit F-4 Loss of Habitability for Cold Water Species (GFDL 2100) .......... F-3
Exhibit F-5 Loss of Habitability for Cool Water Species (GFDL 2100) . F-4
Exhibit F-5 Loss of Habitability for Warm Water Species (GFDL 2100) . F-5
Exhibit F-6 Loss of Habitability for Rough Water Species (GFDL 2100) F-6
Exhibit G-l Summary of Selected Economic Studies of the Value of Freshwater
Fishing Days by Region .............................. G-3
VI
-------
ACKNOWLEDGEMENTS
The contributions of several people made many things possible in this report that
otherwise would have been -well beyond our budget. The authors would like to single out a few
individuals whose help -was particularly valuable.
Dr. Joel Scheraga, Chief of EPA's Adaptation Branch, initiated the project that led to
this report and has continually kept this study focused on issues that are important to current
policymaking. Ms. Susan Herrod, also of EPA's Adaptation Branch, served as the project
manager. She kept the project on track at critical junctures and, more than any other reviewer,
closely scrutinized our calculations for consistency and accuracy.
A number of reviewers earned our appreciation for identifying improvements and
corrections that had escaped us. First among them is Dr. Richard Park who served as technical
reviewer for Abt Associates. His attention to detail helped assure continued improvement in this
report through a succession of drafts. External reviewers included Dr. John Everett, Dr.
Warren Fisher, Mr. Drew Laughland, and Dr. Clifford Russell. These reviewers' suggestions
as well as their challenging questions helped us anticipate the reactions of a larger audience of
readers.
Finally, -we would like to acknowledge the key technical input of three individuals. Special
thanks go to Dr. John Eaton, of EPA's Environmental Research Laboratory in Duluth,
Minnesota, for sharing the results of on-going research on thermal tolerances of freshwater fish
and also for serving as an external reviewer, and to Mr. Jim Wallis, of IBM Research, for
loaning us precipitation and temperature data in a readily accessible form. Their contributions
were central to our analytical framework. We also relied on Mr. Vijay Narayanan, of Technical
Resources International, who provided sound technical advice on the use of climate change
estimates from general circulation models.
Vll
-------
-------
PREFACE
Ecosystem functions and the downstream services they provide may be significantly affected by
climate change. The rate and magnitude of potential losses of ecosystem-derived services is a concern to
the global community since such losses would negatively affect our welfare and the welfare of future
generations. In response to these concerns, the Climate Change Division of the U.S. Environmental
Protection Agency (EPA) initiated a scoping study in order to understand the feasibility of assessing
these types of impacts, and to gain a better understanding of the extent and geographic distribution of the
impacts. The scoping study identified recreational fishing as a feasible service for which damages could
be estimated, and as a service with significant economic value that may be at risk from climate change.
Based on the findings of the scoping study, EPA conducted further analyses to estimate the potential
economic losses to recreational fishing for the entire United States. This report provides the results of
these analyses.
The study makes use of extensive work on thermal habitats for various fish species and data
from almost one thousand gauging stations and nearby meteorological stations. This information was
combined with projections of climate change from four equilibrium general circulation models (GCMs)
and one transient GCM to predict present-day and future thermal conditions. A national model of
recreational fishing provided estimates of changes in the characteristics of recreational fishing due to
climate-induced changes in fish habitat. The results of this analysis support the possibility that
substantial damages could result from climate-induced losses of cold and cool water fishing
opportunities.
The work described in this report represents one of the first efforts to establish a framework that
integrates thermal and economic modeling in order to estimate the impacts to freshwater fish and the
socio-economic implications of these physical impacts. This work also represents one of the first
attempts to estimate national economic damages to recreational fishing from climate change. Because
EPA's goal was to produce a national assessment, a simplified approach is taken for linking climate
induced ecosystem changes and changes in recreational fishing behavior. Complex relationships, such as
reduced productivity and the translation offish survival into changes in recreational fishing behavior, or
characteristics of an array of alternative fishing opportunities available around the country to entire
populations of potential anglers, are not considered in this study. Instead of a highly site specific
approach, this study focuses on a characterization of fishing opportunities at a higher level of
aggregation, and on the characteristics of the anglers themselves.
This study was conducted for the Environmental Protection Agency by Abt Associates, under
subcontract to Technical Resources International Incorporated. Before producing the final report,
reviews were solicited from EPA staff and experts outside the agency. Their comments are gratefully
acknowledged and have been addressed in preparing this final report.
Susan Herrod
Project Manager
U.S. Environmental Protection Agency
-------
-------
EXECUTIVE SUMMARY
As greenhouse gases accumulate in the atmosphere, the climate will be altered. These
changes will consist of rising temperatures, changes in precipitation, and changes in other
weather patterns. Ecological processes and ecosystem services are likely to be affected, although
determining the subsequent effects on human welfare is often difficult. A dearth of information
exists on the linkages from changes in the climate to changes in ecosystems to changes in
services valued by humans. One area in which there are likely to be risks from climate change,
and where there is enough information to develop estimates of physical impacts and effects on
human welfare is fishing. Recreational fishing is both a popular activity in the United States, as
exhibited by the millions of Americans who participate annually, and has been the subject of
extensive studies, including research on fish tolerances and the potential impacts of climate
change on species survival. This study focuses on freshwater fishing in rivers and streams as a
starting point for estimating potential impacts to recreational fishing from climate change.
The findings of this report fall into two categories: projected physical impacts resulting
from climate change, and the economic value of those impacts. There are fairly substantial
physical losses predicted under all of the selected climate scenarios:
• Losses of cold and cool water .habitat range from almost 1.7 million acres to 2.3 million acres
by about 2060. These acres translate into a complete loss of available cold or cool water
fishing in eight to ten states, and a fifty percent loss in eleven to sixteen additional states.
• By the year 2100, ten states loose all cold and cool water opportunities, and another 17 loose
over 50 percent.
• Cold water guild losses occur throughout their entire range, whereas cool water guild losses
are concentrated in the southern sections of their habitable range.
• Species losses tend to be greatest for Brook Trout and somewhat less for Brown Trout,
Rainbow Trout and Cutthroat Trout. Of the cool water species, Walleye suffer some loss,
primarily in the southern states.
Economic losses are partially offset by the opportunity for anglers to fish in lakes and
impoundments, or to turn to other types of fishing (cool, warm, and rough):
• Annual damages of $95 million and $85 million (1991$) are projected using two equilibrium
GCM models, with two other models projecting annual benefits of about $80 million each.
The results depend on whether gains in cool and warm water fishing will more than offset
losses in cold water fishing.
• The GFDL transient 2050 scenario produces the largest welfare losses of the scenarios
considered. Annual losses are projected to be $320 million.
• The transient 2100 scenario predicts cold water acreage losses twice as large as those losses
predicted by 2050, but the losses are offset slightly by gains in cool, warm and rough acres,
resulting in an annual loss of $266 million.
• Analyses conducted to test assumptions and modeling methods reveal that results are most
sensitive to the model's treatment of cold water acreage substitution, and to the designation
offish habitat. When most of the alternative modeling methods are employed, or assumptions
IX
-------
relaxed, estimated damages increase substantially, providing evidence of the conservative
approach taken in estimating potential damages to recreational fishing from climate change.
The first step in this study was to conduct thermal modeling to estimate the effects of
temperature changes on habitat conditions within different geographic areas, and the subsequent
effect on the ranges offish species. This was done by using data for air temperatures at different
climatological stations to estimate baseline water temperatures, and then simulating the effect of
climate change on baseline air and water temperatures to determine changes to the ranges offish
habitat. The effect of climate change was simulated using temperature increases provided by
transient and equilibrium General Circulation Models (GCMs). The most thermally tolerant
species in each guild was used as the indicator of whether that guild could survive at each
location.
For the second step, the Vaughan and Russell model was used to project changes in
recreational fishing behavior based on measures of habitat changes estimated by the thermal
model. The model predicted changes in total days spent fishing for recreation, by classes offish
(cold, warm/cool, rough), expressed as a function of changes in fishable acreage for the different
classes. To calculate the lost fishing opportunities, the assumption was made that every species
in the entire guild must disappear in order for an opportunity to be lost. This assumption is likely
to have produced conservative estimates of the losses. Using values per day spent fishing for
each class offish, the annual damages for recreational fishing were calculated.
The thermal modeling predicted significant losses of cold water fish. Losses were
generally greatest in the southern border of a species' natural range, where baseline temperatures
were closest to thermal tolerances. Cold water species were most affected, but significant losses
were also predicted for the cool water guild and for individual members of warm water and
rough guilds (e.g., crappie, rock bass, smallmouth bass and white sucker).
Losses of cold and cool water acres ranged from almost 1.7 million to 2.3 million (GFDL
and UKMO, respectively). The distribution of these acres geographically represent a loss of
available cold or cool water fishing in eight to ten states, and a fifty percent loss in eleven to
sixteen additional states, depending on the GCM equilbrium scenario. For most of the scenarios,
cold water guild losses occurred throughout their entire range, whereas cool water guild losses
were concentrated in the southern sections of their habitable range. Species losses tended to be
greatest for Brook Trout and somewhat less for Brown Trout, Rainbow Trout and Cutthroat
Trout. Of the cool water species, Walleye suffered some loss, primarily in the southern states.
The transient GFDL scenario produced similar results to equilibrium scenarios for the year 2050.
However, by the year 2100, losses increased. Ten states lost all cold and cool water
opportunities, and another 17 lost over 50 percent. Sixteen states lost all brook trout. Brown trout
was lost in 2 states and reduced by over 50 percent in another 15 states. Rainbow trout was lost
in four states, and Chum and Pink Salmon were lost in three states. The cool water species most
affected was Walleye, with four states loosing the species completely.
Although projected losses of cold water fishing opportunities resulting from climate
change are fairly significant when considered in isolation, the economic losses may be partially
x
-------
offset by the opportunity for anglers to fish in lakes and impoundments. These sources of fishing
opportunities will be less affected by climate change. The losses may also be offset by anglers
turning to other types of fishing (cool, warm, and rough). Since the thermal modeling has not
adequately addressed changes hi habitability conditions for lakes and impoundments, this study
assumes that no change occurs in the fishing opportunities there. Thus, the economic results
should be viewed as primarily illustrative of how the physical impacts predicted by the thermal
modeling might be translated into welfare effects, given modeling constraints and assumptions,
and the fact that results are not adjusted to a common year.
The changes in fishing opportunity due to climate change produced mixed results for the
equilibrium scenarios. Two scenarios resulted in estimated annual damages of $95 million and
$85 million (1991$). However, in the remaining two scenarios, gams in cool and warm water
fishing offset losses in cold water fishing, resulting hi annual benefits of about $80 million each.
When specification of fishing day values were altered to make cool and warm water fishing day
values equal, (i.e., the gains in cool water fishing days estimated by these models are worth less),
three of the four scenarios produced net welfare losses, with only one scenario producing a gain.
The GFDL transient 2050 scenario produced the largest welfare losses of the scenarios
considered ($320 million annually). Unlike the equilibrium GCM results, this is primarily due to
the loss of cool water acreage. The transient 2100 scenario produced similar results: the cold
water acreage losses were twice as large, but were offset by gains in cool, warm and rough acres,
resulting in an annual loss of $266 million. Using the alternative specification of equal values
for cool and warm water fishing days reversed this trend. The net damages in the 2050 scenario
dropped by one-fourth, but the net damages estimated for the 2100 scenario increased slightly to
$286 million annually.
Because these results represent a series of conservative choices and assumptions,
evaluations were conducted using valid alternative climate, ecological and economic
assumptions and modeling methods. When alternative modeling methods were employed, or
assumptions relaxed, estimated damages increased substantially. More specifically, for all of the
equilibrium model scenarios, the majority of the largest relative changes were associated with
modeling assumptions that increase estimated damages.
The areas examined were fishing-day values, climate sensitivity and emissions scenarios,
fish thermal tolerances, fish habitat designations, warm-water fishing behavior, cold-water
substitutability, and runoff. Results were most sensitive to the model's treatment of cold water
acreage substitution, and to the designation offish habitat. Following these were the climate and
emissions scenario and the fish thermal tolerance specifications.
For cold water acreage substitution, instead of assuming that the loss of cold water habitat
could be offset by increases in opportunities for other types of fishing, the analysis assumed that
the loss of cold water habitat could not be offset - the loss was complete. This modeling change
increased damages, which indicates that the degree to which losses in cold water fishing can be
offset by other types of fishing is a critical determinant of the magnitude of damages estimated
for recreational fishing.
XI
-------
The designation offish habitat was altered to employ a wider screen to designate fish
habitat by state. This change caused an increase in the number of best-use cool- and warm-water
acres in the baseline model. "Best-use" is defined as the preferred type of fishing a given water
body will support. For this analysis, the assumption was made that anglers prefer cold-water
fishing to warm-water fishing and both of these to rough fishing. The result was greater damages
in the wide-screen scenario for all equilibrium climate scenarios. This is because the shifts in
best-use cool-water acres to warm-water acres were larger than the shifts resulting from the
narrow scenario, and the value associated with the loss of cool-water fishing days outweighed
that of the gain in warm-water fishing days. Damages increased by at least a factor of five for the
OSU and UKMO climate scenarios. The GFDL and GISS model scenarios reversed from
projected benefits of $80 and $91 million respectively under the primary specification to
damages of $443 million and $451 million under the wide screen designation. For both transient
scenarios, the wide screen specification yielded higher losses in best-use cold acreage relative to
the narrow screen specification, but the effect on value was a net increase (benefit) because the
wide screen had more initial warm-water acreage, making warm-water acreage a better substitute
for cool and cold water acreage than it was under the narrow screen. This lead to more benefits as
the temperature increased.
Analysis of climate and emission assumptions showed that for all scenarios with high
temperature increases, total damages were projected. The annual damages ranged from $131
million hi GISS to $751 million in UKMO. For scenarios with low temperature increases, total
benefits were predicted in all scenarios. The estimated benefits ranged from $36 million (GISS)
to $395 million (GFDL). The high temperature transient scenarios resulted in projected damages.
However, the low temperature increase scenarios produced mixed results. The GFDL transient
(2050) scenario produced damages and the GFDL transient (2100) scenario produced benefits.
Finally, the results of the analysis examining the potential effect of including changes in
runoff illustrate that runoff could have a significant effect on economic losses in recreational
fishing and should be considered further. The analysis employed simplifying assumptions about
volume of annual runoff, potential runoff changes, seasonal variations in runoff changes,
marginal values per acre-foot lost, and runoff depth. The results were an estimated loss of
between $4 million and $1 billion annually, illustrating that climate-induced changes in runoff
alone could lead to significant economic losses in recreational fishing.
Despite mixed results, the analysis in this report supports the possibility that substantial
damages could be induced by climate change. The results provide useful guidance for further
research and site-specific analyses. Where damages are predicted, they range from $85 million to
$320 million annually for the primary modeling specification. These damages only represent
losses to recreational fishing in freshwater rivers and streams. Potential gains (benefits) are also
not ruled out in this analysis. However, in order to produce benefits, one has to believe that warm
water fishing opportunities will expand when cold water habitat is reduced.
The sensitivity analyses conducted on the thermal and economic assumptions made in the
recreational fishing analysis further support the conclusion that substantial economic damages
-------
could occur. For all of the equilibrium models, the largest relative changes are generally
associated with modeling assumptions that increase the estimated damages. In other words, if the
primary specification had included selected alternative assumptions tested in the sensitivity
analyses, estimated damages would be much larger. Additionally, consideration of the effects of
temperature alone may drastically underestimate total damages to recreational fishing in
freshwater rivers and streams. Some of the factors excluded from this analysis could contribute
to much larger projected losses, as illustrated by the runoff analysis.
These conclusions should be considered in light of the fairly substantial uncertainties
involved and the limitations described in the report. However, the potential for substantial losses
point to the need for conducting further research to address uncertainties, and to learn more
about the magnitude of actual damages that could occur from climate change.
xni
-------
-------
CHAPTER 1
1.
INTRODUCTION
Ecological systems are likely to be vulnerable to effects that could be posed by climate
change hi the next one hundred years. Rapid and large changes in temperature and precipitation
could alter the fundamental circumstances that determine the viability of aquatic and terrestrial
ecosystems and their constituent organisms. Article 2 of the Framework Convention on Climate
Change highlights concerns about the magnitude and speed of these changes, citing as its
ultimate objective the stabilization of greenhouse gas concentrations at a level to prevent
"dangerous anthropogenic interference with the climate system" and stipulating this level "should
be achieved within a time frame sufficient to allow ecosystems to adapt naturally" (United
Nations General Assembly, 1992).
The way that plant and animal species and their associated communities respond to these
changes can have significant implications. Three ways generally describe the array of potential
responses. Organisms or communities can adapt where they are, they can die and possibly
become extinct, or they can migrate to more hospitable environs (Buddemeier, 1991). Of
greatest concern would be the extinction of a species. However, even the disappearance of a
species from selected individual locations, despite moving or surviving elsewhere, could be a
matter of some urgency. Furthermore, others have argued that the transformation of ecosystems
induced by climate change will result in a general loss of biodiversity (Markham, et al., 1993).
An earlier report by Abt Associates identified freshwater fish species as one particular
element of aquatic ecosystems that could be at risk in the United States (Michaels et al., 1992).
The social significance of such a risk is not small. Society has very much at stake when
freshwater fish are threatened. In particular, part of what is at stake manifests itself in the
recreational fishing activities of millions of Americans each year. Because of this importance,
a growing body of empirical studies focused on recreational fishing has been developing over
the last twenty-five years. The previous Abt Associates' report concluded that recreational
fishing should be a primary focal point in EPA's efforts to build an economic assessment of the
impact of climate change on ecological systems, and that the means exist for conducting a
preliminary economic assessment of recreational fishing. This report represents the culmination
of this preliminary effort and provides its findings.
Two analytical tools were designed and implemented by Abt Associates to conduct the
assessment - a thermal model and an economic model. For convenience, each tool is referred
to as a "model" but, more importantly, together they constitute one integrated framework. The
overall framework is depicted in general terms in Exhibit 1-1. The objective hi developing this
framework was to generate policy-relevant information from analyses combining physical
assessment of climate impacts on ecological systems with economic assessments of the
1-1
-------
CD
-a
w
o
'§ -,
CD O
45 ^!
H ^
-------
CHAPTER 1
significance of these impacts. Physical measures of any substantial risks to freshwater fish could
by themselves provide sufficient bases for concern about potential impacts from climate change.
Trying to understand the social implications of these physical effects provides a further basis for
examining the significance of climate change. Here, social significance is measured in economic
terms.
The development of the framework applied in this study is in part attributable to an
intentional effort to push the boundaries of existing knowledge in the natural sciences and
economics as much as possible. The current body of empirical work in these areas was not
developed with climate change hi mind and is limited in its capacity to support the types of
analyses developed here. Given the limitations of existing knowledge, the results from the
aggressive approach taken hi this study can be viewed as a possible source of insights on the
implications of climate change for recreational freshwater fish but not as a definitive basis for
establishing that there are indeed significant threats to these fish and their habitats. A number
of uncertainties are acknowledged and explored in this report. Many of them should be the
focus of a longer-term research agenda. Still, the results of the current study are presented here
with the expectation that certain insights can stand even when major analytical uncertainties are
acknowledged.
The remainder of this report has the following organization. The thermal model,
introduced and evaluated in Chapter 2, provides a link between the current understanding of
temperature changes anticipated hi different parts of the U.S. and the resulting reductions in the
habitat of different recreational fish species as their thermal tolerances are exceeded. Additional
details on the organization of the thermal component are given in Exhibit 1-2. Inputs to the
thermal model include temperature changes projected by different general circulation models
(GCMs), estimated thermal thresholds for recreational fish species, data characterizing rivers
and streams, and definitions of baseline habitat for different guilds of fish (cold-water, cool-
water, warm-water, and rough). The thermal model provides estimates of the extent of each
guild's habitat in rivers and streams, in the baseline and after climate change. The economic
model, adapted from a national model of recreational fishing, uses measures of habitat changes
estimated by the thermal model to project changes in the frequency and type of recreational
fishing. This report presents estimates of the economic value of these changes. Additional
details are provided in Exhibit 1-3. Inputs to the economic model include estimates from the
thermal model of guild-specific habitat in rivers and streams, habitat designations for the
remainder of freshwater bodies (lakes and impoundments), pertinent data on recreational fishing,
and estimates of the value of recreational fishing. The economic model produces estimates of
the characteristics of recreational fishing, such as number of days spent by all anglers fishing
for cold-water species, which provide the basis for estimating damages from climate change.
The economic model and these estimates are presented in Chapter 3. The report closes with an
extensive set of model evaluations hi Chapter 4. These evaluations consider alternative
assumptions in all of three major components of this assessment - the general circulation models
that estimate temperature changes, the thermal model, and the economic model. Several
appendices provide useful, supplemental information.
1-3
-------
C/5
0>
bO
s> o
o t-1
si
5 §.
1 I
W H
CD o
•a
w
-------
-a
-------
CHAPTER 1
REFERENCES
Buddemeier, R. W. 1990. Climate Change and Biology: A Proposal for Scientific Impact
Assessment and Response. In: The Unity of Evolutionary Biology: Proceedings of the
Fourth International Congress of Systematic and Evolutionary Biology, Volume I.
Discorides Press, Portland, Oregon.
Markham, A., N. Dudley, and S. Stolton. 1993. Some Like It Hot: Climate Change,
Biological Diversity and the Survival of Species. WWF International, Gland,
Switzerland.
Michaels, G., K. Sappington, L. Akeson, M. Wojcik, T. Aagaard, and D. DeWitt. 1992.
Ecological Impacts from Climate Change: Scoping Study for an Economic Assessment.
Prepared for the Adaptation Branch, Climate Change Division, Office of Policy,
Planning, and Evaluation, U.S. Environmental Protection Agency. Abt Associates Inc.,
Bethesda, Maryland. November 11.
United Nations General Assembly. 1992. Framework Convention for Climate Change. United
Nations, New York.
1-6
-------
CHAPTER 2
2. HABITABILITY ASSESSMENT FOR RECREATIONAL FISH IN RIVERS AND
STREAMS
Temperature is a very important characteristic of the habitat in which fish live, and the
temperature of streams often is closely related to air temperature. Many fish that are adapted
to cold- and cool-water conditions are living close to the limits of their thermal tolerances at the
present time. These fish could be at risk from global warming in the next century.
Relationships between air temperatures, water temperatures, and ecological effects
(including viability of fish) are complex. The relationship between air temperature and water
temperatures, for example, depends in part on the physical dimensions of each water body,
hydrology, and the extent to which particular water bodies are shaded by riparian vegetation.
Vegetation is in turn likely to be influenced by climate. Changes in precipitation would also be
likely to affect the relationship between air and water temperatures, and might affect the
management of impoundments for water storage. Changes in releases from these impoundments
might affect temperature and flow in downstream surface waters. Finally, response by any
single species of fish will be influenced by responses of other species to which it is ecologically
linked. A rigorous analysis of expected impacts of climate change on fish and recreational
fishing would need to consider these and other dimensions of these complex relationships.
Several researchers have used heat balance models and other analytic techniques to
predict impacts of climate change for specific regions or water bodies. For example, researchers
at the St. Anthony Falls Hydraulic Laboratory of the University of Minnesota (in cooperation
with the U.S. Environmental Research Laboratory in Duluth, Minnesota), have used heat and
oxygen transport models to predict impacts of climate change on 32 fish species in 5 streams and
27 classes of lakes in Minnesota (Stefan, et al., 1992). However, because the goal of this study
is to estimate impacts on a national scale with limited time and resources, simpler analytic
methods are used to derive rough estimates of impacts at high levels of geographic aggregation.
In this study, estimates of changes in thermal habitability in streams have been obtained
for each of the forty-eight contiguous states for each of thirty-two species of fish belonging to
cold-, cool-, and warm-water, and rough-fish guilds. The natural and well-established ranges
of these species were used to verify the approach; the ranges also were used to prevent the
prediction of the occurrence of fish outside their ranges. The long-term viability and the
potential for exclusion of species is considered, rather than changes in biomass or yield, because
the results are used in determining shifts that might occur from one type of recreational fishery
to another. Given the need to estimate impacts at the national level and the fact that this is the
first phase of a continuing effort, the study focuses on undisturbed rivers and streams for which
good data exist and that are amenable to analysis. Almost a thousand gauging stations and
nearby meteorological stations and the results from four equilibrium global circulation models
(GCMs) and one transient GCM, run for four scenarios of emissions of greenhouse gases, were
used to predict present-day and future thermal conditions.
2-1
-------
CHAPTER!
2.1 SAMPLE SITES
The basic approach for the analysis is patterned after Vaughan and Russell (1982), in
which the effects of changed conditions within a representative sample of surface water locations
are scaled upwards to derive estimates for effects on fish or recreational fishing over a broad
geographical area. Ideally, a large base of high-quality data would be available to describe for
each sample location:
• baseline populations of key fish species,
• baseline water temperatures (sampled daily over an extended period of time), and
• baseline air temperatures (sampled daily over a matching time period).
Unfortunately, a national base of such linked data has not been assembled to date. For
this analysis, a database assembled by Wallis et al., (1990) for the purpose of studying
climatological change and verifying output from global circulation models was used instead. For
each of 996 USGS gauging stations, these data contain reports of daily and monthly stream flow,
state, latitude and longitude, elevation, size of watershed, and a description of the station's
location. To facilitate analysis of climate-related trends in the hydrologic regime, the Wallis team
chose stations as free as possible from regulation and with data as free as possible from missing
observations. The stream flow stations included in their set are stations with at least 40 years
of daily record from streams categorized as Class I (no upstream diversions or regulation) or
Class n (minimal upstream diversions and regulation). The 48 contiguous states are all
represented in the set, though the East and Northwest have higher density of stations than the
Southwest. Exhibit 2-1 shows how these stations are distributed across the U.S. To derive
estimates of water temperatures, a matching set of data for air temperatures from about 1000
NOAA climatological stations was used (Wallis et al., 1990). CD-ROM versions of both data
sets were obtained for this study.
2.1.1 Assigning Fishable Acres to Representative Locations
As will be discussed in the next chapter, the chosen methods for relating the availability
of freshwater fish to fishing behavior and value require estimates for the number of fishable
acres of surface water in which each guild offish (i.e., cold-water, cool-water, warm-water, or
"rough") is present in each state. To derive such estimates for each sample station requires an
estimate for the number of fishable acres represented by that station, and an appropriate weight
for each station to scale the results to the state (and ultimately to the national) level. First, the
number of fishable acres of rivers and streams available in each state are estimated. These
estimates are derived from values reported in Vaughan and Russell (1982). As shown in
Exhibit 2-2 Vaughan and Russell's totals represent total fishable acres of surface water in each
state, and must be adjusted to reflect only those fishable acres available in the undisturbed rivers
or streams studied by Wallis, et al. From data compiled by the U.S. Fish and Wildlife Service
for the 1985 National Survey of Fishing, Hunting, and Wildlife-Associated Recreation, the
2-2
-------
CHAPTER 2
fraction of reported fishing days in each state attributable to rivers and streams was determined.
This fraction for each state was derived as the number of reported days spent fishing in rivers
and streams divided by the total number of reported fishing days and was used to approximate
the fraction of fishable acres available in this category of surface waters. Applying this ratio
to the total results in the estimates listed hi the third column of Exhibit 2-2.
The second step is to apportion these estimates of each state's total fishable acres on
rivers and streams among the individual sample locations used in the model. This is done
assuming that the total fishable acres available within each state are represented by individual
stations in proportion to the area of monitored watershed in the state represented by each station.
For example, if the area of the contributing watershed for a particular station represents
5 percent of the total of such areas for all stations in the sample for a given state, that station
is used to represent 5 percent of the fishable acres available in rivers and streams in that state.
The median catchment area represented by a station in the set is 294 square miles,
varying by region from an average of 103 to 575 square miles. Because of the selection criteria
used to assemble the database, these watersheds may be more remote and at higher elevation
than average. The effect of this selection might therefore be a downward bias in water
temperatures for the sample locations used hi our modelling. Such a bias would also reduce
predicted temperatures after climate change, and might result hi conservative estimates of the
effects of climate change. Alternately, extrapolating from thermal damages of undisturbed rivers
and streams to managed rivers and streams may contribute an upward bias to the overall
estimated damages. Managed rivers and streams encompass impoundments which permit some
control of water temperatures through appropriately timed releases. Furthermore, impoundments
can provide cold-water species more refuges with suitable temperatures. Although impacts of
temperature and runoff-changes on impoundments are currently being investigated in a separate
Abt Associates study, the net effect of these biases is unknown.
2-3
-------
Exhibit 2-1
Geographic Distribution of Stations
50-
40-
30-
20-
130
120
110
100
90
80
70
-------
CHAPTER 2
Exhibit 2-2
Fishable Acres of Rivers and Streams by State
State
Alabama
Arkansas
Arizona
California
Colorado
Connecticut
Delaware
Florida
Georgia
Iowa
Idaho
Illinois
Indiana
Kansas
Kentucky
Louisiana
Massachusetts
Maryland
Maine
Michigan
Minnesota
Mississippi
Montana
North Carolina
Total Fishable Acres
(thousands)1
574
811
200
863
176
48
5
1,968
625
335
570
371
129
301
526
2,215
133
151
1,638
940
2,594
551
1,015
822
Rivers and Streams as
Percent of Total2
40.2%
34.3%
19.8%
33.7%
24.4%
31.9%
23.7%
20.8%
21.7%
38.5%
42.0%
24.0%
19.0%
19.5%
25.0%
30.5%
26.4%
32.2%
33.5%
19.8%
11.2%
21.6%
42.3%
28.9%
Fishable Acres of
Rives and Streams
(thousands)
231
278
40
291
43
15
1
410
136
129
239
89
24
59
131
675
35
49
548
186
292
119
429
238
'Vaughan, WJ. and C.S. Russell. 1982. Freshwater Recreational Fishing: The National Benefits of Water
Pollution Control. Resources for the Future, Washington, D.C.
2Data from the 1985 Survey of Fishing, Hunting, and Wildlife-Associated Recreation (U.S. Department of the
Interior, U.S. Fish and Wildlife Service, 1988).
2-5
-------
CHAPTER!
Exhibit 2-2 (continued)
Fishable Acres of Rivers and Streams by State
State
North Dakota
Nebraska
New Hampshire
New Jersey
New Mexico
Nevada
New York
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Virginia
Vermont
Washington
Wisconsin
West Virginia
Wyoming
Total
Total Fishable Acres
(thousands)
541
176
197
119
135
351
1,735
241
904
592
122
NA
689
723
663
1,469
372
452
139
1,158
1,138
141
414
30,737
Rivers and Streams
as Percent of Total2
32.4%
23.6%
34.0%
31.4%
22.8%
27.0%
36.6%
21.8%
13.1%
58.9%
47.8%
22.5%
26.0%
27.1%
33.4%
17.2%
26.1%
33.6%
38.0%
33.4%
24.6%
64.2%
33.2%
Fishable Acres of
Rivers and Streams
(thousands)
176
42
67
37
31
95
635
53
118
349
58
NA
179
196
221
253
97
152
53
387
280
91
137
8,394
'Vaughan, W.J. and C.S. Russell. 1982. Freshwater Recreational Fishing; the National Benefits of Water Pollution
Control. Resources for the Future, Washington, D.C.
2Data from 1985 Survey of Fishing, Hunting, and Wildlife-Associated Recreation (U.S. Department of the Interior,
U.S. Fish and Wildlife Service, 1988).
2-6
-------
CHAPTER 2
2.1.2 Estimating Water Temperatures
The database of 996 stations used in this study does not contain information for water
temperatures (or the presence of fish species) under baseline conditions. Temperatures are
therefore predicted for each location as a function of historical records of air temperature at the
nearest climate station. Maximum and minimum daily air temperatures are averaged to estimate
daily average temperature, with years of data excluded if they contain values flagged as suspect
within the data base. Daily average temperatures are then averaged across the 28 years of
available data for each climate station to calculate a 28-year average temperature for each
calendar date. Finally, since this study focuses on the annual maximum weekly average
temperature, as will be explained in the next section, the daily averages are averaged across
consecutive 7-day periods to determine weekly average temperatures, and (for reasons to be
discussed in Sections 2.2 and 2.3) the maximum of these weekly values is identified for each
station. Exhibit 2-3 summarizes results for each state in the U.S., with the lower map
displaying maximum weekly average air temperature for the station with the lowest value in each
state, and the upper map displaying the maximum weekly average temperature for the station
with the highest value. It shows, for example, that maximum weekly average air temperatures
for stations in Texas ranged from 27.6-29.4 ° C. The narrowest range of maximum temperatures
is observed in Connecticut (21.5-21.9°) and in Maine, where a single climate station was closest
to all flow stations in the sample. The greatest range is observed in California, where maxima
range from 19.0-33.6°C.
The best method for relating air to water temperatures would be to use site-specific
modelling of heat balance for each water body of interest. Because such analyses are not feasible
within the limitations of this project, however, this analysis must rely on less sophisticated
methods. Four alternative methods were investigated for this study. Each is examined in turn
below.
The simplest analytic option (Method 1), and the one ultimately chosen for this study,
is to use unadjusted maximum weekly average air temperatures to represent maximum weekly
average water temperature. The magnitude of error introduced by this simplifying step will
naturally depend on the relationship between air and water temperatures for the stations modeled
in this study. To the extent that extremes of water temperature are moderated by groundwater
inflows, for example, maximum weekly average water temperatures are likely to be closer to
annual average values (i.e., lower) than maximum weekly average air temperatures (Stefan and
Preud'homme, 1993). If maximum water temperatures do in fact tend to be generally lower
than corresponding maximum air temperatures for most stations, the method of using unadjusted
air temperatures is likely to yield overly high estimates of maximum water temperatures under
baseline conditions and therefore to underestimate the natural ranges of cold-water fish species.
Because the estimates of impacts from climate change are driven by the warming of surface
waters beyond fish species' tolerance limits, an overestimation of baseline temperatures might
also lead to overestimation of the impacts of global warming on fish habitat, as true baseline
temperatures might not be as close to tolerance limits as those modeled. Moreover, if the
2-7
-------
CHAPTER 2
amplitude of seasonal swings in water temperature is smaller than the corresponding amplitude
for air temperatures, "forcing" of temperatures from global climate change should likewise be
damped when modelling effects on surface waters. Conversely., in locations where maximum
weekly average water temperatures exceed maximum weekly air temperatures at the nearest
climate station, both the impact of climate change will be under-predicted if unadjusted air
temperatures are used as a surrogate for water.
Stefan and Preud'homme (1993) have used linear regression to investigate the relationship
between air and surface water temperatures for 11 streams in the central U.S. (Mississippi River
basin). They found that weekly average water temperatures could be related to weekly average
air temperatures for their stations (in aggregate) with the relationship:
Tw=A+BTa (2-1)
where:
Ta
A
B
weekly average water temperature (°C),
weekly average air temperature (°C),
intercept, and
regression coefficient.
This regression equation provides predictions of weekly average water temperature (Tw)
as a function of weekly average air temperature (Ta). Values of B derived from their analysis
ranged from 0.669 to 1.026, with seven of the eleven estimated coefficients falling between 0.75
and 1.0. Values of A ranged from 1.40°C to 5.41°C. Standard deviations of measured Tw
compared to Tw from the regressions for individual rivers ranged from 0.65°C to 3.17°C, and
averaged 1.50 °C. The average value of B for the 11 rivers studied was 0.864 with a standard
deviation of 0.11, and the average for A was 2.91 ° C with a standard deviation of 1.43 ° C. The
authors tested water temperatures predicted by these averaged coefficients against actual values
to calculate a standard deviation of 2.16° C. They note the improvement achieved by using
regression equations established for individual rivers as compared to the averaged result.
One possible method for relating air to water temperatures (Method 2) would be to use
Equation 2-1 directly to predict water temperatures for each sample location used in this
analysis, but there are two problems with using the equation for that purpose. First, data used
in Stefan and Preud'homme's analysis were restricted to the central U.S.; their generalization
to other regions and watersheds is questionable. Second, Equation 2-1 adjusts air temperature
downwards when calculating water temperatures from air temperatures greater than about 21 °C,
and upwards for lower air temperatures. For the approximately 25 percent of climate stations
used in this analysis for which mean annual air temperatures are lower than 21 °C, this
relationship would suggest that mean annual water temperatures should exceed mean annual air
temperatures. Similarly, approximately 5 percent of the stations report maximum weekly
2-8
-------
Exhibit 2-3
Maximum Weekly Average Air Temperature
Highest Maximum per State
Lowest Maximum per State
-------
Exhibit 2-4
Maximum Weekly Average Water Temperature
(Method 2)
Highest Maximum per State
Lowest Maximum per State
-------
CHAPTER 2
average air temperatures below 21° C; for those stations, Equation 2-1 would predict maximum
water temperatures exceeding air temperatures by up to about 1 °C. These systematic upward
adjustments of lower temperatures may not be physically justifiable.
Exhibit 2-4 shows water temperatures predicted with Equation 2-1 (where A=2.91 ° C and
B=0.864). As is evident from a comparison of Exhibits 2-3 and 2-4, differences between
(unadjusted) air temperatures and water temperature predicted by Equation 2-1 range from a low
of about -1 °C for the coldest station in New Mexico (13.5°C compared to 14.5°C) to a high of
about +1.8°C for the hottest station hi California (33.6°C compared to 31.8°C). Further
inspection of these exhibits suggests that this difference is equivalent to the temperature change
associated with perhaps 200-400 km of latitude.
A third alternative (Method 3) begins by assuming that mean annual water temperatures
should generally approximate mean annual air temperatures for natural, undisturbed streams
without heat sources or sinks (as suggested by Song et al., 1973). If so, then:
and:
imply:
T = A + BT
T « T
w a
(l-B)
or:
Ta(l-B) + BTa
(2-2)
where:
A
B
annual average water temperature (°C),
annual average air temperature (°C),
constant (°C), and
regression coefficient (dimensionless).
Based on Equation 2-2, this method for predicting maximum water temperatures might
use the 5=0.864 result from Stephan and Preud'homme (1993) but adjust the constant A in the
equation to match mean annual air temperature for each location modelled. The technique would
adjust air temperatures downward for all locations, increasing the predicted presence of cold-
and cool-water fish guilds in southern states of the U.S. With both Equations 2-1 and 2-2,
2-11
-------
CHAPTER 2
expected increments to maximum weekly average water temperature (as a consequence of
climate change) could be related to increments in maximum weekly average air temperature as:
AT; = 0.864 AT;
Results are summarized in Exhibit 2-5. As expected, water temperatures predicted by this
method are consistently lower than those predicted with Method (1).
As a fourth alternative method (Method 4) for estimating Tw, the STORET data base
maintained by the U.S. EPA was accessed to obtain all records of air and water temperatures
available for the USGS gauging stations used this study. This source provided 56,840 records
of water temperatures for 785 of the 996 stations used for this analysis (an average of 72 records
per site). STORET also contained 19,484 records of measurements of air temperatures at these
same locations, for those observations containing both air and water data, the ratio of water to
air temperature (in °C) was approximately 0.88. A serious limitation of these data, however,
is their irregular sampling (generally no more than 1-2 samples per month per station per year,
taken at irregularly scheduled times of day and year). For example, more than 94 percent of
the samples were taken between the hours of 8 AM and 5 PM; the effect of this bias appears
to vary from site to site, depending on the lag between air and water temperatures. Because
water temperatures tend to vary significantly by tune of day (see for example, Stefan et al.,
1993), estimating maximum weekly average temperatures from such constrained data is difficult.
Nevertheless, non-linear regression techniques were used in an attempt to adjust each station's
data for site-specific time-of-day bias, and a sine function describing seasonal variation in
temperature was then fit to the adjusted data for each station. These results were used to
estimate maximum weekly average water temperature for those gauging stations with sufficient
data. Initial results from these attempts, however, did not produce regression equations with
sufficient statistical significance, and this effort was discontinued because of constraints in time
and resources.
In summary, four methods were tested for predicting water temperatures based on air
temperatures. Method (1), the simple assignment of air temperatures to water, is likely to over-
predict water temperatures in areas where groundwater inflows are significant. Methods (2) and
(3) require questionable application of regression coefficients outside their intended contexts.
Method (4) is theoretically preferable but failed for limitations imposed by available data and
resources. As will be discussed in Section 2.3.1, Method (1) has been selected for providing
"best estimates" of the impacts of climate change, for two reasons. First, its predictions for
baseline ranges of fish habitat match reported ranges for the fish species considered about as
well as Method (2), and better than Method (3). Second, Methods (2) and (3) require that
regression results from Stefan and Preud'homme (1993) be used outside their appropriate
context.
2-12
-------
Exhibit 2-5
Maximum Weekly Average Water Temperature
(Method 3)
Highest Maximum per State
Lowest Maximum per State
-------
CHAPTER 2
2.2 THERMAL TOLERANCES AND GUILD ASSIGNMENTS FOR FISH SPECIES
To predict impacts of climate change on the species of freshwater fishes present at
representative stations, quantitative links must be established between climate and the likely
presence of individual species or guilds of freshwater fish. Two general approaches are
available for this step. First, coupled data for surface water temperatures and the presence of
fish species can be used to determine thermal ranges. With this approach, it is assumed that fish
of a particular species cannot survive outside the range of surface water temperatures the species
currently inhabits. A second approach uses results from laboratory analyses of thermal
tolerances. As the result of several decades of scientific interest in the thermal tolerances of
fish, both approaches have been well-developed through previous and ongoing research. Effects
such as changes in ecosystems inside and around thermal plumes from cooling water discharges
have attracted much scientific attention, and the effects of different temperatures on freshwater
fishes have been studied extensively in the laboratory since the 1940s. For many important
fishes, researchers have determined acute lethal temperatures at short time intervals, preferred
temperatures with respect to acclimation, and the effects of temperature on fish bioenergetics.
Others have continued the study of behavioral responses of fish to thermal discharges and the
long-term effects of elevated temperature on growth and mortality of certain fish.
Several different indices are commonly used to describe thermal tolerances of fish in
laboratory experiments. Values for the following measures are summarized for 32 important
species in Exhibit 2-6:
• UILT = Upper Incipient Lethal Temperature. This value is the point of 50 percent mortality
for an experimental group that has been acclimated to a certain temperature before being
exposed to the experimental temperature for a given exposure period (Hokanson and Beisinger
1989). The lethal temperature generally increases with increasing acclimation temperature,
until a maximum is reached, beyond which no increase in UILT is seen. This maximum
UILT is listed without an acclimation temperature and called the Ultimate Upper Incipient
Lethal Temperature, or UUILT (Fry et al., 1946, in Eaton 1993). Eaton et al., list available
UILTs and UUILTs together as Upper Thermal Tolerance Limits (UTTL).
• Short-Term Maximum. This is a calculated criterion based on the upper incipient lethal
temperature and is designed to provide a measure of safety for all the organisms. It is
calculated by fitting experimental data on a straight line on a semi-logarithmic scale, with
exposure time on the logarithmic scale and temperature on the linear scale. The calculated
short-term maximum temperature is:
short
-2°C
2-14
-------
1
I
I
®
>e td
i4
•** ^L
& 3
II
•§
«
f
3
'S
§
at
!
u
&
1
to
OH
H
«
h.J
S3>
j^ ij
s •
b
/^
^N
S3
53 • •
I11
U
^*
,3
B
-d
3
«
i
.s
1
VO
O
en
ON
CM
©
m
CM
en
CM
O
en
a
CM
en
^
i
%
Black crappie*
^
en
c?
ON
CM
en
©
en
c — •
en
CM
en
V.Q
en
en
•a
"*
en
en
©
en
o'
en
1
1
*
3
CM
t-H
CM
*>
CM
ON
i— i
?— i
CM
^
CM
m
en
n
ir-
en
•0
en
«\
•^""N.
*g
in
en
§
i
1 Brown bullhead*
oo
en
CM
-— V
e
en
CM
,__!
CM
•o
!>
en
CM
©
en
CM
1
1
i
ON
O
en
CM
©
l>
en
CM
w
T}"
en
en
^
I
5!
U
*
1-
U
T— 1
en
a
s
&
>n
en
\2/
oo
en
CM
en
in
en
en
•o
>n
^^«
en
©
oo
en
^
'bi
1
O
1 Channel catfish*
o
s
"O
oo
_
' ,^
r— (
in
CM
I
U
CO
5
1 Chinook sahnon
CM
ON
ON
•o
en
r*
/"~V
^
00
en
CM
1
o
1
§
Ji
U
en
en
CM
&
o
CM
CM
OO
V— <
en
CM
en
>D
*v
CS
O
U
0
1
O
oo
CM
CM
i
i
1
|
| Cutthroat trout
CM
CM
en
i
n
1
O
1 Flathead catfish*
^
en
CN
^
C*")
»— fc
CM
OO
CM
en
r<
1
1
*
1 Freshwater drum
^
en
m
en
©
>n
en
•o
7— 1
y^*v
^^
S
vd
en
1
CO
3
I Gizzard shad*
-------
VO
^
1
&
s
H
^
l
^
19
b
^
u>
Ix^
"6 § I
CO
/•^.
u
2
C^
H
^
T3
^
O
*
1
I
CO
1— 1
o
en
0
en
/CA
>S3)
O
en
T— 1
o
3
N*-X
r-~
en
i
^
CO
O
0
| Golden shiner*
CN
o
en
*O
T^
?*~*
^-^
en
T3
CN
3
N"*-p'
^*
en
^
(50
2
PH
CO
O
Green sunfish*
vo
o
en
^.
c^
v*"^/
"1
en
CN
en
m
i/^
en
en
•o
en
^^
•^f
en
8
^
1
*
on
Largemouth bas
en
CN
CN
!
S
o
i
,£3
to
sfi
I Mountain white]
00
00
CN
^
^•"S
O
en
©
en
en
en
!
*o
o
o
1
I Muskellunge
00
00
CN
^
v"*"^
O
en
<§)
en
en
en
00
CN
OO
CN
O
en
T3
ocT
(8)
^— '
•<*
00
CN
"o
s
Pk
O
1 Northern pike
oo
00
T— (
en
#\
•*J
-lEB'
ON
en*
CN
3
3
CO
PH
1
ON
00
n
en
ON
CN
•O
en
in
en
(^)
en
1
&
CO
| Smallmouth bas
ON
oo
CN
0
en
oo"
CN
^*J
»-<
en
•g
8
i
1
*3-
i— i
en
>n
en
g
en
CN
©
en
en
i
^
i
*
09
1
i
•*
en
oo
ra
c^
ON
n
©
en
en
1
*
U
*
I
VO
-------
1
as
cs
en
oo
cs
en
cs
CO
CO
ON
CM
ON
O
CN
CM
en
3
A
en
en
-
8
I
.
-------
CHAPTER!
where a and b are calculated from data in Appendix n-C of the National Academy of
Sciences Water Quality Criteria (U.S. EPA 1986.)
Upper Zero Net Growth (UZNG). Under a defined set of experimental conditions (such
as unlimited food, good quality water), there is a thermal zone within which the growth
rate of a population exceeds its mortality rate, so that the population experiences net
growth. The upper limit of this zone is called the Upper Zero Net Growth temperature
(Hokanson and Biesinger, 1989).
Maximum Temperature for Growth (MWAT-Gro). Because growth is such an important
temperature-sensitive biological function (second only to reproduction), the growth-
limiting temperature has received much scientific attention. One approach to estimating
the limiting temperature was to average the optimum temperature for growth (T^ and the
UZNG. Because of the limited availability of UZNG's, the approach adopted by
U.S.EPA in Quality Criteria for Water 1986 was to calculate a maximum temperature for
growth in terms of the UUILT and the optimum temperature (T^,):
MWAT-Gro = T
opt
(UUILT - Top)
3
(U.S. EPA 1986).
Lethal Temperatures (T4). In developing population rate coefficients to use in modelling
reservoir ecosystems, a list of optimum or preferred temperatures, lower lethal
temperatures, and upper lethal temperatures was assembled for the Chief of Engineers of
the U.S. Army. The upper lethal temperatures (T4) are listed along with the acclimation
temperature and lifestage of experimentation, and ultimate lethal temperatures are noted
(Leidy and Jenkins 1977).
These types of tolerances were often derived in response to specific information
needs. In the 1970s, one such priority was to establish scientifically defensible basis for
enforceable regulation of thermal discharges. In response, the U.S. EPA Environmental
Research Laboratory at Duluth, MN (ERL-D) began a project to verify the laboratory-
derived tolerances with field data, which led to the development of a Fish Temperature Data
Matching System or FTDMS (Hokanson, et al., 1989). This database matches the observed
presence of a fish species in streams and rivers with weekly mean temperatures from
geographically linked sites in the continental U.S., creating a "fish/temperature (F/T)" data
set for each observation. With continued interest in field-data based tolerances of fish
species, especially motivated by concern over climate change, the database has grown. In
1993, it contained 141,208 weekly mean "F/T" observations for 29 species (Eaton et al.,
2-18
-------
CHAPTER 2
1993). Additional species contribute F/T data to the set, and results from analyses of data
for these species have become available since 1993. Based on work completed to date, the
Duluth researchers have concluded that 95th percentile maximum weekly average
temperatures (i.e., the 95th percentile of the maximum weekly average water temperatures
determined for all locations where a particular species has been reported) provide the best
indicator of the thermal limits of a species' natural range. These values generally reflect the
highest water temperatures at the southern extent of the species' range in the U.S. The F/T
data sets are further separated into those north and south of the 40° latitude line to reveal
geographic areas where a species may have adapted to higher thermal niche. Wherever 20
or more observations exist for a species on one side of the line, a northern- or southern-
specific FTDMS value has been calculated.
When the naturalized range of a species is clearly contained within the southern
border of the U.S., the FTDMS value generally is lower than laboratory-derived estimates of
lethal temperatures such as the UTTL but higher than temperature criteria previously
calculated from them, such as the Maximum Weekly Average Temperature for Growth or
Short-term Maximum Temperature. For species with naturalized ranges that cross the
southern border of the U.S., however, the FTDMS value does not approximate the highest
temperature where that fish is to be found, because the 95th percentile temperature does not
reflect the high end of the temperatures the fish tolerates, but the high end of temperatures
found in the U.S.
Freshwater fish, by convention, are classified as belonging to a guild based on
thermal habitat, either cold or warm. Studies based on thermal tolerances require greater
resolution, however. The Duluth research team has used a classification system of three
thermal guilds: cold, cool, and warm. However, as their database has become more
extensive and as their statistical techniques improve, they have discovered that the only clear
distinction is between cold-water fish and the others; the line between cool and warm is
indistinct. In fact, using established conventions of recreational anglers as a basis for the
defining thermal guilds leads to guild memberships with overlapping tolerances. For
example, the estimated tolerances for rock bass (28.5) which is conventionally considered a
"warm-water" fish is lower than the estimated tolerance for walleye (29.5), a cool-water fish.
For consistency with economic literature, however, these conventions are retained for the
present study.
Because of the geographical limitations to FTDMS data, Hokanson et al. (1989)
recommend that for all warm-water fish, the Upper Zero Net Growth (UZNG) literature
value be used instead of the FTDMS value. Further, they recommend that when this value is
not available, an average of the available UZNG values for warm-water fish should be used.
Because "warm-water fish" is a rather arbitrary grouping, however, the present study uses
modified tolerances only for those fish whose naturalized range extends beyond the southern
U.S. border. Exhibit 2-7 lists the species included in this study by guild. For most of these
species, the most current FTDMS value is used as the thermal tolerance value.
2-19
-------
CHAPTER 2
Exhibit 2-7
Laboratory and Field-Derived
Thermal Tolerance Values for 32 Fish Species
Thermal Guild
Species
Abbrev.
Thermal Tolerances1
(°C)
Basis
Cold-water Species
Pink salmon
Chum salmon
Brook trout
Mountain whitefish
Cutthroat trout
Coho salmon
Rainbow trout
Brown trout
Chinook salmon
PKS
CMS
BKT
MWH
CUT
COS
RET
BNT
CHS
18.8
19.2
21.2
22.3
22.8
23.3
23.7
23.8
24.0
FTDMS
FTDMS
FTDMS
FTDMS
FTDMS
FTDMS
FTDMS
FTDMS
FTDMS
Cool-water Species
Northern pike
Muskellunge
Walleye
Pumpkinseed
Yellow perch
NOP
MUE
WAE
PMK
YEP
28.8
28.8
28.9
28.9
29.4
FTDMS
FTDMS
FTDMS
FTDMS
FTDMS
Warm-water Species
Smallmouth bass
Rock bass
Golden Shiner*
Gizzard Shad*
Sauger
Black crappie*
White crappie*
White bass*
Largemouth bass*
Bluegill*
SMB
RKB
GOS
CIS
SAR
BLC
WHC
WHB
LMB
BLG
28.4
28.9
33.0
33.0
30.3
30.6
31.4
35.0
35.5
36.0
FTDMS
FTDMS
Average UZNG
Average UZNG
FTDMS
FTDMS
FTDMS
UZNG
UZNG
UZNG
"Rough" Fish
Brown Bullhead*
Carp*
Flathead catfish*
Freshwater drum*
Green sunfish*
Small-mouth buffalo*
White sucker
Channel catfish*
BRB
CAP
FCF
FWD
GSF
SAB
WTS
CCF
33.0
33.0
33.0
33.0
33.0
33.0
30.2
35.0
Average UZNG
Average UZNG
Average UZNG
Average UZNG
Average UZNG
Average UZNG
UZNG
UZNG
Notes:
'These values are primarily the FTDMS values from the Environmental Research Laboratory at Duluth, MN (personal
communication John Eaton, July 7, 1994).
* These species' naturalized ranges extend south beyond the southern border of the U.S. Thermal tolerances for these
species were calculated from the FTDMS data and the available UZNG values as follows: If both values were
available, the higher of the two was chosen as the thermal tolerance. If no UZNG value was available, the species
was assigned the average UZNG value for the species considered with ranges that did extend past the southern border.
2-20
-------
CHAPTER 2
Species with ranges that extend beyond the southern border of the U.S. are marked with an
asterisk. For these species, we use the UZNG value where it is available (see Exhibit 2-6 ),
or the average of values for "southern-border-crossing" fish (33°C). The resulting values in
Exhibit 2-7 are also displayed on Exhibit 2-8.
The original version of the thermal model derived for this study used a set of fish
species for which FTDMS data were available. This set has grown to include the fish
species listed on Exhibit 2-9, and some species in the original set have dropped out because
the data supporting their values could not meet the same level of statistical precision as the
rest. The cutthroat trout and pumpkinseed (a sunfish), two recognized recreational fish
species from the newer list, were substituted in the original set for fish that were dropped.
The muskellunge was retained because of its recreational importance. It was assigned the
thermal tolerance of the northern pike, a fish that has a similar naturalized range within the
U.S. and similar laboratory values.
These values are the basis for the primary model runs presented here. Separately, a
sensitivity analysis was performed to evaluate the effect of increasing and decreasing the
tolerance for each fish. The Duluth research group has calculated standard errors for all the
FTDMS values for all the species in their original set. These standard errors for FTDMS
values used in the present study and the standard deviation of the UZNG for the species with
extended ranges provided the basis for the sensitivity analysis, discussed in a later chapter.
2.3 PISH PRESENCE AT SAMPLE LOCATIONS
2.3.1 Checking Model Predictions for Baseline Presence
Perhaps the most meaningful test of this model's ability to predict natural ranges for
fish species is how well it predicts ranges of fish habitat under baseline conditions. To a
large extent, these "predictions" are circular: data for reported fish presence have been used
to generate 95th percentile limits to maximum weekly average temperature for FTDMS, and
these limits are the principal source of data used to determine the locations in which
individual fish species are likely to occur. For some species of fish, however, the thermal
tolerances used in this study are derived from laboratory data or generalized from similar
fish species. Moreover (as discussed in Section 2.1.2) water temperatures used for this
analysis have been derived from air temperatures reported for nearby weather stations, and
do not necessarily equal true values. For these reasons, baseline ranges of fish habitat
predicted by the model described in this report cannot necessarily be expected to correspond
exactly to the ranges represented by the data used to derive FTDMS, and a comparison is
useful.
2-21
-------
CHAPTER 2
As a test of the accuracy with which the model predicts natural ranges for fish
species, its estimated ranges have been compared to those reported in the Audubon Society
Field Guide to North American Fishes. Whales and Dolphins (Chanticleer Press, Inc., 1983).
This guide provides maps and verbal description of natural habitat for most of the fish
species included in this analysis. According to the Audubon Guide's introduction, its maps
"show natural ranges; areas where the species is introduced are only included if the species is
well established there." 'Because the maps are very small (about 2x2 cm), their interpretation
as precise indicators of range is subjective. Nevertheless, the maps and verbal descriptions
of the Audubon Guide have been used in this study to determine whether each species of fish
should be expected to occur naturally within each of the 48 contiguous states of the U.S.
Descriptions of the species and their ranges are contained in Appendix A. Fish presence
within a particular state has been considered positive if the reported natural range appears to
cover more than about 10 percent of the state's land area. Exhibits 2-10 through 2-13 show
this interpretation of natural ranges as described in the Audubon Guide. Because the ranges
mapped in these exhibits are discrete with respect to states, they are naturally "lumpier" and
more expanded than those provided in the Audubon guide, for which the boundaries of
shaded areas do not correspond to state boundaries.
Exhibits 2-14 through 2-15 (for Methods 1 through 3, respectively) compare model
predictions of baseline habitat for cold-water fish to natural ranges provided by the Audubon
Guide. Exhibit 2-14, compares predictions for Method (1), hi which maximum weekly
average water temperatures are assumed to equal unadjusted maximum weekly average air
temperatures for the nearest climate station. Exhibit 2-15 compares predictions from Method
(2), where Equation 2-1 (4=2.91, 5=0.864) has been used to estimate maximum water
temperatures from maximum air temperatures. Finally, Exhibit 2-16 provides the same
comparison for Method (3), in which water temperatures are predicted with Equation 2-2
(B=0.864). For all three exhibits, states are unshaded (white) if the thermal model does not
predict the presence of a fish in at least one station within the state, and the maps and
discussions in the Audubon Guide do not suggest that the range for that fish covers at least
10 percent of the state's area. States are marked with cross-hatching if the Audubon Guide
suggest the species is present in a state, and the thermal model predicts fish presence for at
least one of the sample locations modelled in that state. States marked with horizontal lines
indicate that the Audubon Guide reports the presence of a fish species but the thermal model
estimates that water is too warm to support the fish (at all sampled locations within the state).
Finally, states are marked with vertical lines if the model estimates that waters within the
state should be cool enough to support the species, but its natural range (as described by the
Audubon Guide) does not cover at least 10 percent of the state.
The abundance of vertical lines ("false positives") in these maps highlights a
fundamental limitation to modelling based on upper thermal limits alone: a simple thermal
model can indicate areas in which waters are too warm for a particular fish species (i.e.,
where the fish species cannot live), but will not necessarily indicate where the species can
2-22
-------
•a
o
CO
CD
(O
CO
eg
i
O
I
CD
cc
E
CD
H
O)
O)
-C
E 2>
O
0 O
5 ]D
2 8
-•08
+ 8
8
8
2 8
+ TJ 8
LO
CO
o
CO
LO
CM
O
CM
LO
-------
CHAPTER 2
Exhibit 2-9
Additional Species in FTDMS Database
Species Name
Black Bullhead
Bluntnose Minnow
Black Nose Dace
Chain Pickerel
Creek Chub
Common Shiner
Cutthroat Trout
Emerald Shiner
Fathead Minnow
Golden Redhorse
Johnny Darter
Longear Sunfish
Longnose Gar
Mosquito Fish
Mottled Sculpin
Northern Hog Suck
Pumpkinseed Sunfish
Red Shiner
Silver Redhorse
Spotted Bass
Spottail Shiner
Spotfin Shiner
Warmouth
White catfish
FTDMS
Value
31.0
29.3
26.0
29.2
27.1
26.5
22.8
31.0
31.1
29.6
26.2
31.0
30.6
31.3
26.2
29.6
28.9
31.8
29.6
30.9
30.2
29.6
30.6
30.9
These species have enough F/T data sets (>20) to calculate a meaningful 95th percentile Maximum Weekly
Mean Temperature.
2-24
-------
Exhibit 2-10
Natural Ranges for Cold Wa t e r Guild
Brook Trout
Brown Trout
Rainbow Trout
Chinook Salmon Chum Salmon
Coho Salmon
Cutthroat Trout
Pink Salmon
Mountain Whitefish
-------
Exhibit 2-11
Natural Ranges for Cool Water Guild
Muskellunge
Northern Pike
Pumpkinseed
Walleye
Yellow Perch
-------
Exhibit 2-12
Natural Ranges for Warm Water Guild
Black Grapple
Bluegill
Gizzard Shad
Golden Shiner
Largemouth Bass
Rock Bass
Sauger
Smallmouth Bass
White Bass
White Crappie
-------
Exhibit 2-13
Natural Ranges for Rough Guild
Brown Bullhead
Carp
Channel Catfish
Rathead Catfish Freshwater Drum Green Sunfish
Small Mouth Buffalo
White Sucker
-------
Exhibit 2-14
Comparison of Habitat Definitions
(Method 1)
Brook Trout
Brown Trout
Rainbow Trout
Chinook Salmon Chum Salmon Coho Salmon
Cutthroat Trout
Pink Salmon Mountain Whitefish
NONE
AUDUB. MODEL BOTH
-------
Exhibit 2-15
Comparison of Habitat Definitions
(Method 2)
Brook Trout
Chinook Salmon
Cutthroat Trout
NONE
Brown Trout
Chum Salmon
Pink Salmon
AUDUB.
Rainbow Trout
Coho Salmon
Mountain Whitefish
MODEL
BOTH
-------
Exhibit 2-16
Comparison of Habitat Definitions
(Method 3)
Brook Trout
Cutthroat Trout
NONE
Brown Trout
Rainbow Trout
Chinook Salmon Chum Salmon Coho Salmon
Pink Salmon Mountain Whitefish
AUDUB. MODEL BOTH
-------
CHAPTER 2
live. Habitat ranges for individual species are not determined by water temperatures alone;
they involve complex interactions of ecological, hydrological, and physical factors. To the
extent thermal tolerances are separable from other limitations to habitat range, one can draw
meaningful conclusions about the likely absence of individual fish species in waters above
specified temperatures. Such conclusions cannot necessarily be drawn, however, about the
expected presence of individual species in waters with lower temperatures. Based on thermal
limitations alone, for example, coho salmon would be expected to thrive hi South Dakota,
but this anadromous species is hi fact limited to the Pacific Coast. That the thermal model
over-predicts natural ranges for all nine species of cold-water fish (especially the anadromous
salmon of the West Coast) is therefore not surprising. Still, there seems to be a tendency for
all three methods for estimating temperatures to over-predict natural ranges in the direction
of warmer waters: there appears to be an approximate 1-2 °C discrepancy between southern
boundaries of range predicted by the model and those described by the Audubon Guide. In
fact, for cool-water, warm-water, and rough guilds of fish species, these models predict that
all 24 species should be present in all 48 states. This prediction is inconsistent with Exhibits
2-11 through 2-13, however, which suggest the existence of southern, thermal boundaries for
muskellunge, northern pike, walleye, rock bass and small mouth bass.
Of special interest are the relatively few cases ("false negatives") where the simple
thermal model predicts water too warm for a fish to be present, but the Audubon Guide
reports presence of the fish. Exhibit 2-14 shows that Method (1) results hi fourteen such
false negative results: brook trout unexpected in Delaware, New Jersey, Connecticut and
Maine, brown trout unexpected in Arizona, Missouri and Arkansas, rainbow trout
unexpected in Arizona, Missouri and Oklahoma, Cutthroat Trout unexpected in Arizona, and
Pink Salmon unexpected hi California, Minnesota and Michigan. As can be seen from
Exhibit 2-15 Method (2) reduces false negative results to thirteen cases, by allowing brown
and rainbow trout in Arizona, but falsely predicting the absence of chum salmon from
California. Those locations hi California where chum salmon were formerly predicted (based
on unadjusted air temperatures) were estimated to have maximum weekly average air
temperatures lower than the tolerance for the salmon, 19.2°C. Because Equation 2-1
(/4=2.91, 5=0.864) predicts water temperatures higher than air temperatures for air colder
than 21 'C, its use results hi the predicted absence of those salmon. Although Method (2)
results hi a net reduction of one false negative result in predicted habitat ranges, it slightly
increases the count of false positive results. Exhibit 2-16 shows that Method (3) eliminates
all but four of the false negative results observed with the other two methods (brook trout hi
Delaware and New Jersey, brown trout hi Arkansas, and rainbow trout hi Oklahoma), but at
the cost of a marked increase in false positive results (209 compared to 158 for Method 1).
With none of the three methods does a thermal model consistently under-predict the southern
boundary of a fish species (i.e., generate false negatives along the southern boundary of a
species' natural range).
Based on a comparison of results displayed in Exhibits 2-14 through 2-16, it appears
that the simplest method for estimating temperatures (Method 1) predicts baseline ranges of
2-32
-------
CHAPTER 2
habitat as well as the other two methods examined. Because of this finding, and because of
concerns that regression results from Stefan and Preud'homme (1993) should not be
generalized outside their appropriate context, this first analytic option has been selected as
the preferred method for this analysis. Sensitivity of results to this selection will be
discussed later.
2.3.2 Other Determinants of Habitat: Development of "Screens"
To overcome our model's tendency to predict fish presence in areas which, for
reasons other than water temperature, are beyond the natural range of habitat for particular
species, we use a "screen" to simulate limits imposed by other, non-thermal, constraints on
natural ranges. For this purpose, we use the maps shown in Exhibits 2-10 through 2-13 to
represent the natural ranges of habitat for each species of fish included in our analysis. A
fish species is assumed present in waters near a particular station in our sample only if two
conditions are met:
1) maximum weekly average water temperatures estimated for that station fall
beneath the limit assigned to that species, and
2) the station falls within the ranges shown in 2-10 through 2-13.
By using such a screen we avoid projecting impacts from climate change for particular fish
species in areas the species do not currently inhabit. Exhibit 2-17 maps screened estimates
of baseline ranges for the nine species of cold-water fish examined in this analysis. For
example, estimated baseline ranges for chinook, chum, coho and kokanee salmon are limited
to the West Coast, even though estimated water temperatures are sufficiently low in several
other states to support their presence. Conversely, rainbow trout are not assumed present in
Missouri and Oklahoma (even though these states are included in the range mapped by the
Audubon Guide) because estimated water temperatures for all sample stations are higher than
the 24 °C tolerance limit established for that species. In some states, estimated maximum
water temperatures span a range that extends both above and below the tolerance of a
particular species. In California, for example, maximum water temperatures at fewer than
50 percent of the stations in the sample are low enough to support chinook salmon (with a
tolerance limit of 24 °C), whereas more than 50 percent are cool enough to support coho
salmon (with a tolerance of 23.3°C). In general, estimated water temperatures suggest that
all nine cold-water species should be present at only selected stations along the southern and
eastern boundaries of their ranges. Exhibits 2-18 through 2-19 display analogous screened
ranges for cool-water, warm-water, and rough fish guilds. As mentioned earlier, the
combinations of estimated water temperatures and tolerance limits used for this analysis
appear to overestimate the southern and eastern extent of ranges from several of these
species, such that expected fish presence along range boundaries (which are determined
entirely by the Audubon Guide screen) is unconstrained.
2-33
-------
CHAPTER!
From these species-specific estimates of habitability under baseline conditions, we
aggregate our results to determine the presence of each thermal guild at each sample station.
If, for example, at least one of the nine species of fish in the cold-water guild is expected
present at a particular sample location, the cold-water guild is assumed represented at that
location. As shown by Exhibit 2-21, members of the cold-water guild are expected in at
least some sample locations in most states outside the Southeast. Members of the cool-water
guild are expected in all states east of the Rocky Mountains excluding Georgia, which falls
just outside the natural ranges of yellow perch and walleye. The aggregated range of the
warm-water guild is similar to that for cool-water, except that Georgia is no longer excluded.
Finally, members of at least one species from the rough guild (usually carp at a minimum)
are expected in every station included in our sample, except where maximum weekly average
temperatures exceed 33.0°C (the tolerance for carp) in California.
2.3.3 Effects of Climate Change on Maximum Temperatures
Using GCMs to Predict Changes to Fish Presence
Once a representation of "baseline" conditions has been constructed, the next analytic
step is to simulate expected changes in fish presence as a function of increasing temperatures.
For this step, location- and month-specific results from several general circulation models
(GCMs) and global projections from several emission scenarios (EPCC -- IS92a, IS92c, and
IS92e) developed by the Intergovernmental Panel on Climate Change (LPCC, 1992) provide
the basis for predicting incremental changes in the maximum weekly average air temperature
expected for each of our 996 sample locations. In the current model, the GCM output
provides a range of possible geographic and temporal patterns of temperature increase, while
the IPCC scenarios set the global mean average temperature increment to which the GCM
output is normalized.
The GCMs have been developed to simulate the physical processes of the atmosphere
and oceans and to calculate climatic parameters under baseline and increased CO2 conditions.
Models originating from four different research groups were used in this report, and both
equilibrium models and transient models are included. Equilibrium models predict the
temperature of a system that was allowed to reach equilibrium with a doubled CO2
concentration, and they show no time-dependence. References for the equilibrium models
include:
• GFDL (Geophysical Fluid Dynamics Laboratory: Manabe and Wetherald, 1987);
• GISS (Goddard Institute for Space Studies: Hansen et al., 1983);
• OSU (Oregon State University: Schlesinger and Zhao, 1988); and
• UKMO (United Kingdom British Meteorological Office: Wilson and Mitchell, 1987).
The National Center for Atmospheric Research (NCAR) made this data available in
electronic form. Transient modelling is intended to give a more realistic representation of
2-34
-------
CHAPTER 2
how quickly the climate will change over time with dynamically increasing greenhouse gas
concentrations. Currently, NCAR has available data from transient runs from four different
models, including the GFDL. The GFDL transient model (Stouffer et al., 1989) uses a
coupled ocean-atmosphere module, which introduces some lag tune due to the damping effect
of the oceans. Transient data from the GFDL were also included hi this study.
While the GCMs are the best available estimates of geographic variation in climate
change, their spatial resolution is still rather coarse: they treat the globe as a set of grids,
varying in size from 4 by 5 degrees of latitude and longitude for OSU to 8 by 10 degrees for
GISS. Obviously there is significant climatological variation within grid cells of this size,
and the boundaries of the grid cells are artificial. However, the performance of the four
equilibrium models in selected control runs under a 1 x CO2 equilibrium scenario was
evaluated and compared to a real-time climatic data set by climatologists in 1991 (U.S. EPA,
1991). They found that the GCM results and the empirical data were especially compatible
over North America, where the temperatures projected by the models matched observed
climate data relatively well, especially in summer, within 2°C in most cases. The patterns
produced by the transient and equilibrium versions of the models are generally similar over
North America.
Greenhouse warming is caused by anthropogenically increased levels of greenhouse
gases. The GCMs are designed to calculate the level of wanning that would be caused by a
certain concentration of these gases in the atmosphere, but they cannot predict what those
concentrations will be. The actual concentrations of greenhouse gases depend on emissions
of CO2 and other greenhouse gases. IPCC (1992) focused intentionally on emissions
scenarios, predicting changes in the global emissions based on current population forecasts,
energy and industry forecasts, political events and changing economic circumstances
worldwide, current studies on tropical deforestation and forest biomass, and the best
estimates of uncertainty. The authors developed a range of scenarios, with IS92a
representing the median estimate, assuming internationally agreed controls on SOx, NOx,
and non-methane VOCs. Scenario IS92c reflects the results of a low-estimate level of
emissions due to low estimates of population growth, and scenario IS92e represents a high-
end level of emissions, resulting from a higher level of economic growth. In addition to this
range of emissions, the IPCC also presented a range of sensitivity to the level of emissions in
scenario IS92a, representing best-, high-, and low-estimate predictions of global mean
temperature increments resulting from the mid-range estimate of emissions. While there are
a number of outstanding issues in the modelling of climate change, such as possible
differentials in day- and night-time temperature increases, the IPCC's view was judged to be
the most reliable interpretation of the current state of science.
This study uses temperature increments from the GCMs, normalized to correspond to
these three emission scenarios. This approach allows the use of temperature increments
specific to the individual months and locations, adjusted to be consistent with IPCC
projections. A total of 22 combined scenarios were compiled for this study. For each
2-35
-------
Exhibit 2-17
Baseline Habitability for Cold-Water Fish
Brook Trout
Brown Trout
Rainbow Trout
Chinook Salmon Chum Salmon Coho Salmon
Cutthroat Trout Pink Salmon Mountain Whitefish
1 - 4 9 % 5 0 - 9 9 % 1 0 0 %
-------
Exhibit 2-18
Baseline Habitability for Cool-Water Fish
Muskellunge
Northern Pike
Pumpkinseed
Walleye
Yellow Perch
0%
11-49% Illlllllllllillllllill 50-99% 100%
-------
Exhibit 2-19
Baseline Habitability for Warm-Water Fish
Black Crappie
Bluegill
Gizzard Shad
Golden Shiner
Largemouth Bass
Sauger
Smallmouth Bass
White Crappie
Rock Bass
White Bass
0%
1 -49%
50-99%
1 00%
-------
Exhibit 2-20
Baseline Habitability for Rough Fish
Brown Bullhead
Carp
Channel Catfish
Flathead Catfish Freshwater Drum Green Sunfish
Small Mouth Buffalo
White Sucker
0%
1-49% 5 0 - 9 9 % 1 0 0 %
-------
Exhibit 2-21
Baseline Habitability by Guild
Cold Water Guild
Warm Water Guild
Cool Water Guild
Rough Guild
1-4
50-99%
1 00%
-------
CHAPTER 2
scenario, the cell-month temperature increments from a GCM (ATL, L for local) are
multiplied by a normalization factor that is the ratio of the global annual mean temperature
increment indicated by the IPCC emissions scenario (GAMTjpCC), for a given climate
sensitivity assumption divided by the global annual mean temperature increment obtained by
averaging all the GCM cells over the entire grid over the 12 months (GAMTocM). The result
is a normalized cell-month temperature increment, ATL'.
ATL' = (GAMT,pCC/GAMTGCM) ATL
The scenarios in this study include both equilibrium and transient runs of the models. The
equilibrium scenarios are constructed from doubled-CO2 equilibrium runs from the four listed
models scaled to each of the three climate sensitivity estimates used by IPCC. The transient
runs of the GFDL model were scaled to time-dependent projections from scenarios IS92a,
IS92c, and IS92e.
This approach produced twelve equilibrium scenarios, for the four GCMs normalized
to each of the three proposed sensitivity estimates as they were applied to emission scenario
IS92a. The resulting scenarios are summarized in Exhibit 2-22. The equilibrium data are not
tied to a specific calendar year, but to a time of doubled CO2. According to 1992 IPCC
Supplement Figure Ax. 4, the year when the global annual mean temperature increment
reaches the level of predicted warming under emission scenario IS92a varies significantly for
different sensitivity estimates. The resulting increase from low estimates of doubled-CO2
sensitivity, 1.5°C, may occur by 2080, while the "best estimate" 2.5° warming may occur at
2090, and the year when the high-sensitivity estimate of 4.5°C would be reached is 2105.
Temperature increments from the transient GFDL runs were obtained for two
decades, referenced to a baseline representing the climate in 1961-90. The "1.16°C decade"
run of the transient GFDL was presented to the IPCC 1992 working groups because its
global average temperature increase most closely matches the 1.16°C global mean
temperature increase projected to occur by 2050 according to the IS92a "best estimate"
projections. The "eighth decade" was also provided for studies designed to consider a
stronger degree of forcing. For the current study, the local temperatures from these decades
were normalized to a range of IS92 scenarios: IS92a with "best," low, and high sensitivity
assumptions; IS92c with "best estimate" sensitivity; and IS92e with "best estimate"
sensitivity. The "1.16°C" decade was normalized to the IS92 projections for 2050, and the
"eighth decade" was normalized to the 2100 projections to coincide with the years generally
considered in climate change analysis. The global annual mean temperatures used in these
normalizations are presented in Exhibit 2-23.
For the calculations reported here, "forcing" values for air temperature (i.e., expected
increments in temperature due to increased CO2) are obtained automatically from electronic
files of scaled GCM model output corresponding to one of the constructed scenarios listed on
2-41
-------
CHAPTER 2
Exhibit 2-22
Global Annual Mean Temperature Increases Used
to Scale Equilibrium GCM Results
GCM Run
GFDL/
Equilibrium
GISS/
Equilibrium
OSU/
Equilibrium
UKMO/
Equilibrium
GCM Climate
Sensitivity
4.0°C
4.2°C
2.8°C
5.2°C
IPCC Climate Sensitivity
Low
1.50C' at doubled
CO2GF2
1.5°C at doubled
CO2 GI2
1.5°C at doubled
CO2 OS2
1.5°C at doubled
CO2UK2
Best estimate
2.5°C1 at doubled
CO2 GF1
2.5°C at doubled
CO2 Gil
2.5°C at doubled
CO2 OS1
2.5°C at doubled
C02 UK1
High
4.50C' at doubled
CO2 GF3
4.5°C at doubled
C02 Gil
4.5°C at doubled
CO2 OS3
4.5°C at doubled
C02UK3
Notes:
1 IPCC 1992 (p. 10)
Exhibits 2-22 or 2-23, based on the latitude, longitude, and calendar dates appropriate for the
weekly average temperatures of each sample location. These values are added to baseline
temperatures, and the maximum weekly average is re-computed. Exhibit 2-24 summarizes
forcing values taken from each of the four equilibrium "best estimate" scenarios for each of
the 48 contiguous states. This exhibit reports averaged values of forcing for each state;
actual values can vary within a state according to each station's location and the month in
which the new maximum temperature occurs. Exhibit 2-25 shows ranges of temperatures
expected in each state after climate change. Similar maps are provided for the Gil, OS1,
UK1, TR1, and TR2 scenarios in Appendices B-F.
To determine how climate change might affect the natural ranges of individual fish
species, these altered estimates of maximum weekly average water temperature are compared
to the thermal tolerances previously established for each fish species. As with the same step
for baseline conditions, the "screen" constrains the areas considered to those within the
naturalized range of each species. Next, the new estimates for fish presence after climate
change (expressed in fishable acres per state) are subtracted from the corresponding estimates
for baseline conditions to derive estimates for the loss of fish presence expected as a result of
increases in temperature.
2-42
-------
CHAPTER 2
Exhibit 2-23
Global Annual Temperature Increases Used to Scale Transient GCM Results
GCM
Run
GFDL/
Transient
"1.16°C
decade"
GFDL/
Transient
"Eighth
decade"
Climate
Sensitivity
1.32°C
2.6°C
Emission Scenario
IS92C
(low)
IS92A
(mid)
IS92e
(high)
IPCC Climate Sensitivity
Best
estimate
l.l°C2in
2050
TR7
1.5°C2in
2100
TR8
Low
0.15°/decade
= 0.9°C3 in
2050
TR3
0.15°/decade
= 1.65°C3 in
2100
TR4
Best
estimate
0.25°/decade
= 1.5°C3 in
2050
TR1
0.25°/decade
= 2.75°C3
in 2100
TR2
High
0.4°/decade
= 2.4°C3 in
2050
TR5
0.4°/decade
= 4.4°C3 in
2100
TR6
Best
estimate
1.8°C2
in 2050
TR9
3.5°C2in
2100
TRIO
Notes:
1 IPCC 1992. (p. 10)
2 IPCC 1992 Annex Figure Ax.3 (p. 174)
3 IPCC 1992 as interpreted by Neil Leary, EPA, in memo to ECF dated April 13, 1994.
2.4 RESULTS
Exhibits 2-26 provides summary maps of results for all four guilds for the GFDL
equilibrium and Exhibits 2-27 through 2-30 provide summary maps of results for each
species, based on forcing values taken from the GFDL model. In each of the maps included
in this exhibit, a state is shaded with cross-hatching if 100 percent of its original fishable
acres of a particular species or guild are lost as a result of CO2 doubling (i.e., the waters at
all sample locations are expected to exceed thermal tolerances for that species or guild). Of
course, if the species or guild was not present under baseline conditions, it cannot disappear
because of climate change: states where no loss occurs (either because the fish was originally
absent or because warming is insufficient to eliminate it) are not shaded. Those states
experiencing partial losses of habitat for a species or guild are marked with horizontal or
vertical lines, where the percent loss is calculated as the number of fishable acres lost divided
by the original number of fishable acres (before climate change). Appendices B through F
provide analogous summary maps based on the other three GCMs.
2-43
-------
Exhibit 2-24
Average increment to Maximum Weekly
Temperature from CO2 Doubling
GF1
GI1
OS1
UK1
-------
Exhibit 2-25
Maximum Weekly Average Temperature
After CO2 Doubling (GF1)
Highest Maximum per State
Lowest Maximum per State
-------
Exhibit 2-26
Loss of Habitat by Guild (GF1)
Cold Water Guild
Cool Water Guild
Warm Water Guild
Rough Guild
0%
1-49% 50-99% ••Mil 00%
-------
Exhibit 2-27
Loss of Habitat for Cold-Water Fish (GF1)
Brook Trout
Brown Trout
Rainbow Trout
Chinook Salmon Chum Salmon Coho Salmon
Mountain Wnitefish
1-49% 5 0-9 9 % 10 0 %
-------
Exhibit 2-28
Loss of Habitat for Cool-Water Fish (GF1)
Muskellunge
Pumpkinseed
Yellow Perch
Northern Pike
Walleye
to%
1-49%
50-99%
1 00%
-------
Exhibit 2-29
Loss of Habitat for Warm-Water Fish (GF1)
Black Crappie
Bluegill
Gizzard Shad
Golden Shiner
Largemouth Bass
Sauger
Smallmouth Bass
White Crappie
Rock Bass
White Bass
0%
1-49%
150-99%
1 00%
-------
Exhibit 2-30
Loss of Habitat for Rough Fish (GF1)
Brown Bullhead
Carp
Channel Catfish
Flathead Catfish
Freshwater Drum
Green Sunfish
Small Mouth Buffalo
White Sucker
1-49%
[50-99%
1 00%
-------
CHAPTER 2
As can be seen from these exhibits, expected losses of cold-water fish are significant
with all four GCMs. Losses (expressed as a percent of original fishable acres) are generally
greatest along the southern border of a species' natural range, where baseline temperatures
are closest to thermal tolerances. Cold-water species are most affected, but significant losses
are also predicted for the cool-water guild, and for individual members (e.g., crappie, rock
bass, smaUmouth bass and white sucker) of warm-water and rough guilds.
Exhibit 2-27 shows the effects of warming for each of the fish guilds as projected by
the GFDL. Cold- and cool-water fishing are hurt the most as eight states lose all of the
cold- or cool-water fishing available to them. In addition, over fifty percent of cold- and
cool-water fishing are lost in fifteen states. Sixteen other states show a cold- or cool-water
loss to a lesser degree. Significant cold-water guild losses occur throughout their range.
Conversely, cool-water guild losses are concentrated in the southern sections of their
habitable range. Specific losses in the warm and rough water guilds were light except for the
elimination of: black crappie in Oklahoma and Texas; smallmouth bass in Oklahoma and
Arkansas; and white sucker in Arkansas. Warming projected by the UKM model
(Appendix D) is similar to that in the GFDL model. Cold-water guild losses are in the same
range while cool-water guild losses are slightly more evident in the UKM. Specifically, cold-
and cool-water guilds are eliminated in nine states while seventeen others lose over
50 percent of their cold- and cool-water fishing capacity. Under the UKM model cold-water
fish are lost across their natural ranges and cool-water fish are often lost in the southern to
middle sections of their ranges. Marginal to significant losses in Wisconsin, Illinois, and
Iowa of cool-water fish (all but the yellow perch in Wisconsin) sets the UKM apart from the
GFDL. Warm water species are lightly effected in most cases but black crappie is reduced
50 percent or more in five states (white crappie in 4) and smaUmouth bass have a 50 percent
or higher loss in four states.
The OSU model predicts warming that has a slightly greater effect on cold-water fish
than the GFDL and UKMO models but has less of an effect on the cool guild (Appendix C).
Cold- and cool-water guilds are completely eliminated in ten states. Over fifty percent of the
cold- and cool-water guilds are eradicated in fourteen states. Within natural ranges losses
happen throughout the cold-water fish species. Cool-water fish losses are less than cold
throughout species' natural ranges except walleye which is eliminated in Kansas, Oklahoma,
Arkansas, and Mississippi. Specific losses in the warm-water guild are masked at the guild
level by available substitutes within the individual states.
Warming predicted by the GISS model has the lowest effect of all of the GCMs,
eliminating cold- and cool-water guilds in five states (Appendix B). Cold and cool-water
guilds are reduced over 50 percent hi a total of eleven states. Warm and rough water guild
losses are insignificant and mask most specific fish species losses. Specific cold-water
species losses occur throughout their ranges, aside from a handful of unaffected states spread
throughout. A lack of substitutability causes the specific walleye losses to almost mirror the
overall cool guild losses.
2-51
-------
CHAPTER 2
REFERENCES
Ayers, M., D. Wolock, G. McCabe, L. Hay, G. Tasker. 1993. Sensitivity of Water
Resources in the Delaware River Basin to Climate Variability and Change. U.S.
Geological Survey. Open-File Report 92-52.
Carpenter, S.R., S.G. Fisher, N.B. Grimm, J.F. Kitchell. 1992. Global Change and
Freshwater Ecosystem. Annual Rev. Ecol. Syst. 23:119-39.
Cohen, SJ. 1986. Impacts of CO2-Induced Climatic Change on Water Resources in the
Great Lakes Region, dim. Change 8:135-53.
Eaton, J.G., J.H. McCormick, B.E. Goodno, D.G. O'Brien, H.G. Stefan, M. Hondzo, and
R.M. Scheller. In press. A Field Information-Based System for Estimating Fish
Temperature Tolerances. Fisheries, 29(4), April, 1995.
Eaton, J.G., J.H. McCormack, B.E. Goodna, D.G. O'Brien, K.E.F. Hokanson, H.G.
Stefan, and M. Hondzo, 1993. A Field Information Database for Estimating Fish
Temperature Requirements. Draft. September 29.
Flaschka, I.M., C.W. Stockton, and W.R. Boggess. 1987. Climatic variation and surface
water resources in the Great Basin Region. Water Res. Bull. 23:45-57.
Frederick, K.D., P.H. 1988. Greenhouse Warming: Abatement and Adaptation.
Proceedings of a Workshop held in Washington, DC. June 14-15. Edited by
Rosenberg NJ, Easterling HI WE, Crosson PR, Darmstadter J.
Gleick, P.H. 1987. Regional hydrologic consequences of increases in atmospheric CO2 and
other trace gases. Climatic Change 10:137-161.
Hansen, J.G, G. Rusell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L Travis.
1983. "Efficient Three-Dimensional Global Models for Climate Studies: Models I
and n,". April Monthly Weather Review, Vol m, No. 4, pp. 609-662
Hokanson, K.E.F., B.E. Goodno, and J.G. Eaton. Undated. Evaluation of Field and
Laboratory Derived Fish Thermal Requirements for Global Climate Warming Impact
Assessments. Landscape Ecology Branch, U.S. Environmental Protection Agency,
Office of Research and Developments, Duluth, MN 55804. Draft Report.
Manabe, S. and R.T. Wetherald. 1987. "Large-scale Changes in Soil Wetness Induced by
an Increase in Carbon Dioxide," April. J. Atmos. Sci., Vol 44, pp. 1211-1235.
2-52
-------
CHAPTER 2
Nash, L.L. and P.H. Gleick. 1993. The Colorado River Basin and Climatic Change.
Sensitivity ofStreamflow and Water Supply To Variations in Temperature and
Precipitation. EPA 230-R-93-009.
The
Poff, N. LeRoy. Regional Hydrologic Response to Climate Change: An Ecological
Perspective. In Global Climate Change and Freshwater Ecosystems. Eds. Firth, P.L.
and Fisher, S.G. Springer-Verlag, NY. 1992.
Revelle, R.R. and P.E. Waggoner. 1983. Effects of a Carbon Dioxide-Induced Climatic
Change on Water Supplies in the Western United States. In Changing Climate.
National Academy of Sciences, National Academy Press, Washington, DC.
Schlesinger, M.E. and Z.-c. Zhao. 1988. "Seasonal Climate Changes Induced by Doubled
CO2 as Simulated by the OSU Atmospheric GCM/Mixed-layer Ocean Model," CRI
Report and J. Climate, Vol 2, pp. 459-495.
Stouffer, R.J., S. Manabe, and K. Bryan. 1989. Interhemispheric asymmetry in climate
response to a gradual increase of atmospheric carbon dioxide. Nature, Vol 342, pp.
660-662.
U.S. Department of the Interior, Fish and Wildlife Service. 1988. 1985 National Survey of
Fishing, Hunting, and Wildlife-Associated Recreation. November.
United States Environmental Protection Agency, Office of Water. STORET Data System.
United States Environmental Protection Agency, Office of Water. 1986. Quality Criteria
for Water 1986, PB87-226759, EPA 440/5-86-001. May 1.
United States Environmental Protection Agency, Office of Policy, Planning and Evaluation.
1991. Global Comparisons of Selected GCM Control Runs and Observed Climate
Data. 21P-2002. April.
Wallis, J.R., D.P. Lettenmaier, and E.F. Wood. 1990. Research Report: A Daily Hydro-
Climatological Data Set for the Continental U.S. IBM Research Report RC 16607
(#70451) 6/11/90.
Wilson, C.A. and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment
with a Global Climate Model Including a Simple Ocean. JGR, Vol 92, No. Dll
(Nov. 20), pp. 13,315 - 13,343.
2-53
-------
CHAPTER 2
This Page Left Blank Intentionally.
2-54
-------
CHAPTERS
3. AN ECONOMIC ASSESSMENT OF RECREATIONAL FISHING IMPACTS
3.1 INTRODUCTION
Climate change could have a tremendous effect on the availability of different fish species
in the rivers and streams of the continental U.S., as the previous chapter demonstrated. These
physical impacts could result hi substantial disruptions in an important American pastime,
recreational fishing. As an earlier study showed, the U.S. has a lot at stake in recreational
fishing (Michaels et al., 1992). Nearly a billion person-days were devoted to recreational
fishing in 1985. Nearly 30 percent of these were associated with cold-water and anadromous
fishing and more than 40 percent to warm-water and rough fishing. This activity could be worth
more than $30 billion annually hi terms of the consumer surplus it generates. While recreational
fishing does not represent fully the economic significance of climate change impacts on
freshwater fish, an economic assessment focused on recreational fishing does capture an
important link between ecosystems vulnerable to climate change and human well-being.
Establishing the link between the availability of different recreational fish species and its
impact on economic welfare requires an understanding of the determinants of recreational fishing
behavior. The kind of fish that can be caught, the chances of success, the quality of the fishing
experience itself, and proximity are a few facets of angling that have been commonly noted as
important. These and other such factors determine which people will fish, what species they
target, where they go, how often they go, and what the experience is worth to them. On a
national scale, some of this behavior has been quantified by the Surveys of Fishing, Hunting,
and Wildlife-Associated Recreation conducted about every five years by the U.S. Fish and
Wildlife Service. These surveys provide national estimates of anglers' socioeconomic
characteristics, species targeted, frequency of activity, and the value of different types of fishing
(trout and bass). It is a greater challenge however to characterize where anglers go, not in terms
of location but in terms of what each fishing designation offers with regard to the chances for
success and the quality of the fishing experience. Since these factors can vary dramatically from
one location to the next, a potential angler has a host of different sites from which to choose.
The likelihood of fishing at a given site and the value attached to that experience will vary from
site to site and both of these will vary from one potential angler to the next. Conducting an
economic study of recreational fishing behavior is an information-intensive undertaking if all of
these factors are to be addressed for a number of sites.
Compounding the difficulty in the current study is the necessity of saying how these
factors will change as the result of changes in the ecosystem supporting the fish. This particular
challenge has been addressed in the current study by assessing the habitability of approximately
900 locations around the country for different recreational fish species before and after climate
change. This approach simplifies the link between the climate-change induced ecosystem
changes and changes in recreational fishing behavior. If a certain species is no longer available
hi a given location, it is assumed that the value of fishing for that species reduces to zero. Other
3-1
-------
CHAPTER 3
losses affecting recreational fishing, such as reduced productivity, could also be anticipated from
climate change but these impacts present considerable difficulties for both a risk and an
economic assessment. In this regard, the current effort provides conservative estimates of the
potential damages to recreational fishing,,
This difficulty resolved, the challenge of translating changes hi fish species survivability
at virtually every single fishing location in the continental U.S. into estimates of changes in
recreational fishing behavior is enormous. It is beyond the scope of the current effort to
implement a study that characterizes the array of alternative fishing opportunities available
around the country to the entire population of potential anglers. As a matter of fact, such an
effort has been beyond the scope of every other recreational fishing study to date.1 Instead of
a highly site-specific approach, this study focuses on a characterization of fishing opportunities
at a higher level of aggregation, that of each state, and on the characteristics of the anglers
themselves.
3.2 ECONOMIC MODEL
The economic model adapted for this study is based upon a framework developed by
Vaughan and Russell (1982). Their framework was constructed to estimate how fishing
behavior, ultimately described hi terms of days spent fishing, would change as a function of
expanding very specific types of fishing (cold-water, warm-water, and rough fishing).2 By
assigning a value to a day of each type of fishing and by estimating how the available acreage
of each type would change with water quality improvements, Vaughan and Russell (1982) were
able to estimate the national benefits of water quality improvements to different types of
recreational fishing. By differentiating fish species primarily in terms of thermal conditions,
their model lends itself immediately to adaptation to an economic assessment of climate change.3
Vaughan and Russell emphasize three issues hi their evaluation of national recreational
fishing benefits: (1) the mechanisms by which control of water pollution discharges create
1 EPA and other federal agencies have initiated such an effort, which is not expected to
bear fruit for a year or so.
2 Selected examples of each fishing type include trout and salmon (cold-water); bass,
muskellunge, sauger, and northern pike (warm-water game fish and panfish), and carp and
catfish (rough).
3 Vaughan and Russell define their three fish types both by water quality (dissolved
oxygen) and temperature requirements, but each fish type also has a unique temperature range.
Cold-water fish are assumed to exist hi waters less than or equal to 18°C, warm-water fish in
waters above 18°C and below 32°C, and rough fish from 32°C to 34°C.
3-2
-------
CHAPTERS
benefits for recreational fishermen; (2) the tools and data necessary to assess these mechanisms;
and (3) the special difficulties of obtaining national benefits estimates. Many of these same
issues are germane to the analysis of global climate change effects in national recreational
benefits. Global climate change will affect the temperature of surface waters throughout the
United States, and the sensitivity of fish populations to anticipated temperature shifts will vary
considerably, as described in Chapter 2. Dramatic shifts in temperature can result in species
composition shifts as well as changes in the number of fishable acres or sites within the United
States. Temperature changes will have two effects: changing fish species composition and
changing the distribution of acreage supportable for different types of fishing activities.
Vaughan and Russell model the effects of pollution control on individuals, each facing
a cross section of available sites of varying quality. The authors couch the analysis in terms of
the household production framework. It is envisioned that consumers produce service flows
using technology, time, and purchased inputs. Changes in water quality are linked with changes
in the supply curve such that assessments of patterns in the production of recreational fishing
days might be estimated using reduced form equations.
The reduced form model treats changes in site quality using a three-stage estimation
process. The first stage predicts the probability of being a fisherman. The second stage predicts
the probability of doing some cold-water, warm-water, and rough fishing. The third stage
predicts the average days per angler devoted to cold-water, warm-water, and rough fishing.
Each of these stages use information on the availability of each fish. Availability is quantified
in terms of acres.4 Each acre of fishable freshwater is uniquely assigned to cold-water, warm-
water, and rough fishing. This assignment is made according to the "best use" that a given
water body will allow. It is assumed that anglers prefer cold-water fishing to warm-water
fishing and both of these to rough fishing. Therefore, if water quality (dissolved oxygen) and
temperature conditions allow, an acre is counted as available for cold-water fishing. Otherwise,
if not suitable for cold-water but still suitable for warm-water fish, the acre is assigned to warm-
water fishing. By a similar process of elimination, the remaining acres are assigned to rough
fishing or designated as unfishable. This approach allows for the far-reaching impacts of
changes in water quality to be addressed in terms of changes in quantity rather than as complex
adjustments for each individual in relation to distances and travel costs.
This approach estimates changes in participation days and then values these changes using
a unit value approach. In short, this reduces to multiplying the estimated changes in warm-
water, cold-water, and rough fishing by relevant activity-day values. Although theoretically the
appropriate value for changes in the number of recreational days are the marginal values, such
4One reviewer of this study suggested that stream-fishing behavior may be more of a
function of stream length than acreage. For narrow streams, this distinction may not be
meaningful. For larger streams, the size of the measurement error involved is unknown.
Consequently, this study adhered to Vaughan and Russell's original specification using acres.
3-3
-------
CHAPTERS
values are rarely observable. It is almost impossible to infer where on the range of equilibria
individuals fall before and after the proposed policy scenarios. An average measure of consumer
surplus is often adopted as an alternative.
3.3 ALTERNATIVES TO THE SELECTED MODELLING APPROACH
In the past decade, there have been no new published studies focused on all four types
of recreational fishing from a national perspective. The economic modelling of recreational
fishing behavior has taken a very site-specific frame of reference. The overview of recreational
fishing studies provided in Appendix G shows how site-specific and, at times, fish-specific
studies have been. Even when multiple sites are modelled, their geographic scope is often
limited. This orientation has probably been influenced by a demand for studies at the sub-
national level. Many of the studies have certainly supported policy analysis for important issues
on a site-specific, regional, and state level. The tendency to construct such models is also
heavily influenced by the prevailing view in the environmental economics profession that better
definition of alternative fishing sites, especially in terms of environmental quality, is an
important element in describing recreational fishing behavior.
Understandably, given the large demand for economic analysis to support policy analysis
at the national level, environmental economists have been groping for substitutes that can be
built upon the existing body of site-specific literature. The primary candidate has been the
general process of "benefits transfer" through which the estimates of one or more studies can
be applied to one or more new locations that have not been analyzed to the same degree but
which are similar "enough" that making the inference of economic valuation this way seems
reasonable. For several reasons, this approach is not suitable for the existing body of
recreational fishing studies.
When dealing with changes in site quality such as those imposed by global climate change
dynamics, several important connections must be established. First, the characteristics that
matter to people must be designated and mechanisms of how to measure these characteristics
must be assigned. Second, changes in the quality of many sites must be tied with demand
functions so that changes in site quality will result in demand shifts. On this second point, travel
cost models, which have been commonly used to value recreational fishing experiences, seem
to be inadequate. While versions of the travel cost model can capture the influence of site
quality, they fail when changes occur at multiple sites because these models do not characterize
substitutabUity among sites (Freeman, 1993a).
Another modelling alternative—discrete choice, or random utility, models—is more
promising. They can be constructed to address differences in quality and substitutability among
sites. The difficulty in drawing upon this body of work is that the current studies are too few
and provide only limited geographic coverage. Furthermore, no widely accepted benefits
3-4
-------
CHAPTER 3
transfer protocol has been established to construct a national model of fishing behavior that is
based upon these studies and that is linked to measures of environmental quality.
One final alternative to adapting the framework of Vaughan and Russell is a national
approach based on contingent valuation. This approach would entail using contingent valuation
information on the incremental benefits of water quality improvements necessary to attain
fishability, such as those compiled by Carson and Mitchell (1993). It, like the approach of
Vaughan and Russell, has already been used to estimate the national benefits of controlling water
pollution under the Clean Water Act. Both approaches attempt to capture the recreational fishing
value of clean water but they differ substantially in what they measure. Vaughan and Russell
attempt to do so by making inferences from observed recreational fishing behavior and the
availability of different types of fishing opportunities hi the U.S. In contrast, the work of
Carson and Mitchell does not consider specific attributes of fishing. They estimate, instead, the
value of converting a large amount of water, on a national basis, to fishable; by this they meant
that the water was clean enough that "game fish like bass can live in it" (Carson and Mitchell,
1993, p. 2447). This kind of measurement is potentially useful in a study of climate change and
recreational fishing but it has limited applicability. For example, the thermal model in the
current study calculates the amount of national freshwater that is rendered unfishable because
temperatures become too high for any recreational fish species. The estimated amounts are
seldom greater than 0.5 percent in any of the climate-change scenarios. Even a definition of
fishability based on habitability for game fish, such as trout or bass, results in only slightly
larger impacts. Estimated increases in non-game fish habitat range from 1 percent to 3 percent.
Because it does not relate well to the redistribution of fishing likely to occur with climate
change, a national approach based on the existing contingent valuation of fishability does not
appear to be fruitful.
3.4 IMPLEMENTATION OF THE ECONOMIC MODEL
Following the design employed by Vaughan and Russell (1982), the economic model is
a reduced form model that characterizes the repercussions of changes in water temperature on
recreational fishing behavior using a three stage-estimation process. The first stage describes
the probability of general fishing participation. The second stage predicts the conditional
probability of doing one or some combination of cold-water, cool-water/warm-water, and rough
fishing activities. The third stage characterizes the average number of person days devoted to
cold-water, cool-water/warm-water, and rough fishing per year.
The thermal and economic models are constructed to represent recreational fishing in the
48 contiguous states. Accordingly, the implementation of the economic model requires
tremendous amounts of data. While consistent and comprehensive data on the general population
are available, similar data are not readily available for categorical angler and recreation
populations. Because of such significant information constraints, the economic model does not
calculate coefficients for each of these stages. Rather, the economic model uses coefficients
3-5
-------
CHAPTER 3
derived by Vaughan and Russell (1982) in their national analysis of water pollution controls.
In the implementation process, the economic model relies on information from several sources,
including input data provided by the thermal model that characterizes the recreational fishing
opportunities, input data acquired from empirical sources describing the relevant populations,
and estimated coefficients communicating the likelihood of different behavioral adjustments.
Input data are maintained at a state level; and, when possible, inputs to the model have been
updated to reflect current national and angler population characteristics (i.e., 1990-1991).
Combining input values with the Vaughan and Russell coefficients, the economic model produces
average estimates of the probability of fishing by category, the number of anglers by category,
and the number of fishing days by category. These outputs are estimated for each of the
scenarios and ultimately account for the variation in the proposed benefits/damages estimates.
The results of the previous chapter illustrated how temperature changes that occur
because of climate change can significantly alter the availability of different fish species. Fish
species that have lower thermal tolerances can lose their habitats. From an ecological
perspective, these losses may or may not be offset by gains in the abundance of fish that are
more tolerant of higher temperatures. Such an outcome was not specifically modelled. Instead,
the thermal modelling was organized to coincide with the perspective taken in the economic
model presented here. This model assumes that each fishable acre is uniquely assigned to the
best use allowed by thermal conditions. The ordering of fish guilds and thermal tolerances
approximately coincide. Cold-water fish are preferred to cool-water fish, which are preferred
to warm-water ones. The fourth category, rough fish, are the least preferred and tend to have
high thermal tolerances. A fifth category, "none," is added as a residual to reflect unfishable
acres in the baseline and the additional acres that become unfishable because temperature
increases render them uninhabitable for any recreational fish.
The mutually exclusive assignment of freshwater acreage to the four fish guilds (cold,
cool, warm, rough) implies a narrow view of the fishing opportunities available to U.S. anglers.
Because each fishable acre has a unique assignment (cold, cool, warm, or rough), this
assumption implies, for example, that a cold-water acre will be visited by anglers pursuing cold-
water fish only, and not by anglers seeking fish in other guilds. This assumption could decrease
the size of the net damages estimated in the economic model. If it is true that, say, cold- and
warm-water fishing are not done at a given water body and if this water body converts from a
cold-water "best use" to a warm-water best use as the result of climate change, then the
estimated net damages are not biased. There are however two cases where the estimated net
damages could be understated. In the first case, if both cold- and warm-water fishing are taking
place before and only warm-water fishing afterwards, the net damages could be underestimated.
In the second case, if only cold-water fishing is taking place before climate change and it is
infeasible or unlikely that warm-water fishing will take place after climate change, net damages
3-6
-------
CHAPTER 3
could again be underestimated. The extent of the downward bias is the same for both these
cases assuming that the values of cold- and warm-water fishing hold constant.5
The original developers of the economic model adapted in this study support this
assumption by pointing to the results of studies on angler preferences (Vaughan and Russell,
1982, p. 38), which was the basis for the preference ordering used here. Vaughan and Russell
restate the assumption in terms of supply rather than demand. "[T]he state fishery manager will
ensure that [the water body] is supporting whichever type is higher on a scale of preference
broadly accepted by both fisherman and fishery management professionals." The view taken in
the current study is that the validity of the assumption is better verified through empirical
information on the likelihood of more than one type of angling on specific streams and rivers
before and after climate change. Information from the 1985 Survey of Fishing, Hunting, and
Wildlife-Associated Recreation supports the "best-use" assumption at least for parts of the U.S.
In the region including the states of Colorado, Montana, North Dakota, South Dakota, and
Wyoming, the number of person-days spent cold-water fishing is almost four tunes greater than
the number spent warm-water or rough fishing.6 This outcome occurs despite the fact that these
states are considered habitable by warm-water and rough fish.
It is more difficult to determine the likelihood of joint freshwater angling hi other states,
especially if only a part of a state's waters have been deemed habitable by cold- and cool-water
fish. In such cases, it is possible that cold- and warm-water fishing take place in different parts
of a given state. This pattern of fishing is consistent with the mutually exclusive classification
of waters used to implement the "best-use" assumption. In the current effort, it has not been
5 An example may help clarify this discussion. Assume that C and W are the values of
cold-water and warm-water fishing respectively on a given stream. If both cold- and warm-
water fishing are feasible but only cold-water fishing takes place, then the value of fishing before
climate change is C and afterwards is at best W. The difference is C-W. In this case, there is
no downward bias attributable to the "best use" assumption. If however both cold- and warm-
water fishing are possible before and only warm-water is possible after climate change, the net
damages are (C+W)-W = C rather than C-W. The damages are also C if only cold-water
fishing is possible before and no fishing is possible afterwards. Finally, it is possible that C will
rise as the supply of cold-water fishing opportunities falls (to C') and that W will fall (to W')
if the supply of warm-water fishing opportunities rise. In this case, (C'-W) will exceed (C-W).
In all three cases, the estimated net damages (C-W) would be biased downwards.
6 This comparison is based upon calculations made from state-specific data from the 1985
Survey of Fishing, Hunting, and Wildlife-Associated Recreation that was categorized into cold-
water and warm-water categories in Michaels et al., 1992 (Appendix Al). Approximately 20
million person-days are spent on cold-water fishing hi these states, exclusive of salmon fishing,
versus approximately 5.5 million person-days spent warm-water fishing.
3-7
-------
CHAPTER 3
possible to make a state-by-state determination of the appropriateness of the "best-use"
assumption. Instead, the issue is investigated further through a sensitivity analysis.
Changes in the distribution of the "best-use" acreages drive the simulation of changes in
recreational fishing behavior. The economic model is specified with the total acreage of the
freshwater lakes, impoundments, rivers, and streams as assigned mutually exclusively to cold-,
cool-, warm-water, and rough fishing. Total acreage in the baseline was adopted from Vaughan
and Russell (1982), which provides fishable cold-, cool-/warm-water, and rough acreage for each
state. Because Abt Associates' thermal model is constructed to evaluate rivers and streams only.,
it was necessary to apportion the total acreage into two parts—rivers and streams and lakes and
impoundments. Information from the 1985 Survey of Fishing, Hunting, and Wildlife-Associated
Recreation on fishing days by location was used to apportion a state's total fishable acreage into
lake and impoundment acreage and river and stream acreage. River and stream acreage in each
state was further subdivided into cold-, cool-, warm-water, and rough portions based on
estimates from the thermal model. Since no changes in the allocation of lakes and impoundment
acreage was estimated, determining the baseline allocation of lake and impoundment acreage is
sufficient to specify the economic model. However, because the state acreage estimates from
Vaughan and Russell aggregate cool- and warm-water fishing., whereas the Abt Associates*
thermal model separates the two guilds, total warm- and cool-water acreage in each state had
to be split into warm- and cool-water portions.
This split was based on one of two rules. First, the split between cool- and warm-water
acreage estimated for rivers and streams in each state in the thermal model became the basis for
apportioning total warm-/cool-water acreage in that state, if the thermal model assigns any
fishable acreage in that state to warm- or cool-water fishing. If not, the second rule was to
assign warm-/cool-water acres in the state to the two fishing types according to the number of
cool- and warm-water fishing days estimated for the state. These estimates were derived from
the 1985 Survey of Fishing, Hunting, and Wildlife-Associated Recreation data in Michaels et
al., (1992). Exhibit 3-1 presents the assignment of total freshwater acres to the four fishing
types or to none (for unfishable acres) hi the baseline and for main specifications of the six
GCM models evaluated in this study.
With the exception of changes in river and stream acreages, inputs to the economic model
approximately represent the state of the world in the late 1980's or early 1990's. In particular,
socioeconomic data, such as population and average income, used to specify the model were
drawn from the 1990 U.S. Census. Rather than introduce simulations in population and income
growth, these conditions were assumed to be the status quo with and without climate change.
Growth in population and income could by themselves have a tremendous effect on the amount
and type of fishing occurring in the baseline year of analysis, which is at least 56 years hence
(2050) and in some cases as much as 106 years away (2100).
Although it is known that increases in population and income result in more recreational
fishing, it is unknown what bias is introduced by the exclusion of population and income growth
3-8
-------
CHAPTERS
from the modelling at this stage of development. If such growth were to be incorporated, it
would be necessary to adjust the relative values placed on cold-, cool-, warm-water, and rough
fishing. The current analysis holds these constant under the scenarios, with and without climate
change, because little is known about how these relative values can change with the reallocation
of freshwater acreage as the result of climate change. Clearly this reallocation will change the
relative values, but it is anticipated that population and income growth could have an even larger
impact. Their impact is magnified because they are exponential and could easily double or triple
the estimated amount of fishing in the 2050 and 2100 baseline years.
Simultaneous with this growth there could be a significant restructuring of observed
fishing preferences and the relative values of cold-, cool-, warm-water, and rough fishing. The
relative value of cold-water fishing, the most vulnerable to climate change, is probably also the
most sensitive to income growth, thereby compounding any effects on estimated impacts that are
linked to the growth in income. By contrast, the estimated changes in the allocation of fish
acreages attributable to climate change are, while substantial, still equivalent to less than a
100 percent change in the number of fishing days. At this initial stage of analysis, income and
population growth are held constant in order to reduce the number of compounding uncertainties
while a feasible set of model facets is tested. In this respect, the current analysis should be
interpreted as an initial experiment to gauge the general size of the recreational fishing impacts
and the sensitivity of the economic model to a finite set of alternative specifications.7
The link between behavioral responses of recreational fishing and thermal changes is
established in the structure of the three stages of the economic model. The thermal responses
of the climate change scenarios shift the availability of types of fishing opportunities. These
transitions are expressed in movements in acreage from one best-use category to another. As
acreage shifts occur, the input values to each of the three stages change; and it is these
adaptations that explain the different numbers and compositions of fishing days associated with
each of the climate scenarios. The implications of the changes in best-use acreage are made
clearer by understanding the different stages of the estimation process. Brief descriptions of
each of the modelling stages follow. Throughout the discussion, emphasis is awarded to those
variables that are assumed to be changing across the global climate change scenarios.
7This analysis, like any other analysis of the effects of climate change, is bound by current
understanding of tastes and preferences. One reviewer of this study has suggested that certain
fishing activities, such as Pacific salmon and steelhead fishing, may experience dramatically
increasing relative values in coming decades. If true for cold-water fishing in general, then the
actual damages would be much higher than those estimated in this study. The decision to adhere
to constant consumer surplus estimates in this study underscores how much more could have
been done to increase the estimated damages. Instead, for the sake of better credibility, the
authors chose to adhere to the status quo of fishing values which results in a conservative range
of damage estimates.
3-9
-------
CO
5j
1
f
T3
e
U
0)
1
5
OS
n
u
cc
O wa
00 ®^
o E;
- r-
5 CM
CM
co
co
^H CO
8" 3
CO CM
rt-
CO
OO
vo o
co
co
£
co
vo
f=i
ON
co
CO
oo
CO
CO
o
ng
oo
CO
t> CO
« 00
g
co
co
co
oo
CO
VI
o
CO
2 a
oo"
VO
°
ON
ON
CO
co °
_r%. ^^
^^ *
vo o
1—I
co
OQ
CO
- ^
s
ON
t-. oo
L"°°
t-- . •
oo
ON
^o
VO
^
W-i
CM CO
co '-'
ON
ON
co
°°~ o
vo o
CO
CO
OO
O
00
Tf
O
CO
oo cs
o
m
o\
co
ON
CM
ON
co
**
co
o"
vo^
co
00
g
co
OO
CO
-------
CHAPTERS
The first stage calculates the probability of general fishing participation(POF). Exhibit 3-2
presents the variables used in the estimation of the first stage and notes the source of each
variable. The variables reflect socioeconomic characteristics as well as broad recreational
fishing opportunities. With the exception of fishable acreage per capita, the input values for this
calculation are estimated using the 1992 Statistical Abstract of the United States (U.S.
Department of Commerce 1992). Estimates of the fishable acreage per capita per state are
produced by the thermal and economic models using U.S. population estimates and the modified
Vaughan and Russell (1982) baseline acreage data. The socioeconomic characteristics do not
change across scenarios. Conversely, the fishable acreage variable changes considerably across
scenarios with the changes in temperature. The input values are combined with the Vaughan
and Russell coefficients (1982, Table 3-6, Reduced Model I) to derive the general fishing
participation probability (PGF). Because the coefficients were estimated using a logit
specification, the final general fishing participation probability is calculated as follows: PGF =
1/(1+e"E^*x> where the /3s are the estimated coefficients and the Xs are the estimated input
values.
The second stage calculates the conditional probability of participation by fishing category
(Ppc I POP)- For the purposes of this analysis, the economic model distinguishes fifteen
mutually exclusive fishing categories. Each of these fifteen categories are some combination of
the following types of fishing activity: T (cold-water); BP (cool-water/warm-water); R (rough);
and S (salt or Great Lakes). The fifteen categories are as follows: (1) T; (2) BP; (3) R; (4) S;
(5) TBP; (6) TR; (7) TS; (8) BPR; (9) EPS; (10) RS; (11) TBPR; (12) TBPS; (13) TRS; (14)
BPRS; and (15) TBPRS. The economic model calculates the conditional probability for each
category and then aggregates these probabilities according to designations for cold, cool/warm,
and rough fishing activities. For example, any category including T is counted as cold-water
fishing, and similarly, all categories with BP and R are treated as warm/cool and rough fishing
respectively.
Exhibit 3-3 presents the variables used in each of the estimations by category and notes
the source of each variable. Similar to those used in the first stage, the input variables reflect
socioeconomic characteristics and recreational fishing opportunities. In this stage, however, the
data becomes more activity specific. The socioeconomic data represent the angler population
and the catch rates and fishable acreage information are organized by thermal acreage category.
The input variables used in the second stage include the average age and annual income of U.S.
anglers, the percentage of female, metropolitan-residing, and coastal-residing anglers, the
average numbers of cold-water fish, cool-water/warm-water fish, and rough fish caught per day,
and the ratios of warm-water fishing acreage and rough fishing acreage to total fishing acreage.
The input values for the average annual income, age, gender, and percentages in metropolitan
areas and along marine coastlines are estimated using the 1991 National Survey of Fishing,
Hunting, and Wildlife-Associated Recreation (US DOI, 1993). The average catch rates for cold-
water, cool-water/warm-water, and rough fishing days were derived using Vaughan and Russell
regional data (1982, Table 4-4). A weighted average is calculated based on the regional
3-11
-------
CHAPTER 3
Exhibit 3-2
First Stage: Predicting the Probability of General Fishing Participation1
Variable
Intercept
Age
Age-squared
Gender (female)
Metropolitan area
Western region
Central region
Southern region
Fishable acreage per capita by state
Average income per household
Head of household
Coastal state
Source
Not applicable
US Department of Commerce (1992)
US Department of Commerce (1992)
US Department of Commerce (1992)
US Department of Commerce (1992)
US Department of Commerce (1992)
US Department of Commerce (1992)
US Department of Commerce (1992)
Abt economic and thermal models
US Department of Commerce (1992)
US Department of Commerce (1992)
US Department of Commerce (1992)
1 The data employed in the first stage represent the national population in 1990
and recreational fishing opportunities.
averages with the regional percentages of U.S. anglers serving as the weights. The fishable
acreage ratios are gleaned by the economic and thermal models. In practice, the socioeconomic
characteristics are held constant and the acreage ratios are permitted to vary across the climate
scenarios. The input values are combined with the coefficient estimates provided by Vaughan
and Russell (1982, Table 4-8) for each fishing category. Since the coefficients were estimated
using a weighted least squares approach, the conditional categorical probabilities are calculated
as follows: PFC | PGF = S /3*X where the j8s are the estimated coefficients and the Xs are the
estimated input values. The conditional probabilities of participation are estimated for all fifteen
categories. These probabilities are then aggregated to predict participation probabilities for the
categories of cold-water (PCDF)> cool-water/warm-water (PCWF)> and rough fishing (PRF).
The third stage calculates the average number of days per year devoted to cold-water
ODCDF)J cool-water/warm-water (DCWF), and rough fishing (DR). Separate calculations are
completed for each of these three fishing activities. Exhibit 3-4 presents the variables used to
predict the number of fishing days. The variables include socioeconomic characteristics and
3-12
-------
CHAPTERS
Exhibit 3-3
Second Stage: Predicting the Probability of Participation by Fishing Category1
Variable
Intercept
Average income per household
Age
Age-squared
Gender (female)
Metropolitan area
Coastal area
Average number of cold-water game fish
caught per fishing day
Average number of cool-water and warm-
water gamefish/panfish caught per fishing
day
Average number of roughfish caught per
fishing day
Ratio of cool-water and warm-water fishing
acreage to total fishing acreage
Ratio of rough fishing acreage to total
fishing acreage
Source
Not Applicable
US DOI (1993)
US DOI (1993)
US DOI (1993)
US DOI (1993)
US DOI (1993)
US DOI (1993)
Vaughan and Russell (1982)
Vaughan and Russell (1982)
Vaughan and Russell (1982)
Abt economic and thermal models
Abt economic and thermal models
1 The input variables employed in the second stage represent U.S. anglers and
thermal categories of recreational fishing. Outputs from the model are derived
for 15 mutually exclusive categories using unique sets of category coefficients.
These category probabilities are then aggregated to reflect cold-water, cool-water
and warm-water, and rough fishing participation estimates.
descriptions of recreational fishing opportunities. Socioeconomic variables include average age
and income by angler type and the percentages of female, metropolitan-residing, and coastal-
residing individuals by angler type. The recreational fishing opportunity variables include the
average number of cold-water fish caught per day, cold-water fishing acreage per capita, average
number of cool-water/warm-water fish caught per day, warm-water fishing acreage per capita,
average number of rough fish caught per day, and rough fishing acreage per capita. The
3-13
-------
CHAPTERS
majority of the input values used in this stage of the estimation are taken from Vaughan and
Russell (1982, Table 4-12) because of the limited socioeconomic information available on a
national basis for anglers by fishing type or category. Implementing the model, most of the
socioeconomic and recreational fishing characteristics are held constant across the scenarios.
The fishable acreage estimates per capita are exceptions to the rule, as these vary with each
climate scenario. The input values are combined with the Vaughan and Russell (1982, Table
4-12, Unrestricted Model) coefficients to derive three estimates of average person-days by
fishing activity. Because the coefficients are estimated using a weighted least squares approach,
the average number of fishing days per person are calculated as follows: DCDF = £ /3*X where
the /3s are the estimated coefficients and the Xs are the estimated input values. The average days
are estimated for the categories of cold-water (DCDF), cool-water/warm-water (DCwp)> and rough
fishing
The output of the economic model combines information from all three stages of the
estimation process. To estimate the number of fishing days for one activity such as cold-water
fishing, the probability of general fishing participation (i.e., the output of Stage 1) is first
multiplied by the conditional probability of fishing for the category of interest (i.e. , the output
of Stage 2). This probability is then combined with the estimate of the average number of days
devoted to the fishing activity per year (i.e. , the output of Stage 3) and an appropriate population
estimate (i.e. , Bureau of Census 1992) to derive the total number of fishing days. For example,
the total number of cold-water fishing days predicted under one scenario would be calculated as
follows: TDcDp = PGF * PCDF * DCDF * POPULATION. For each run of the economic model,
this procedure is adapted to estimate the total number of cold-water, cool/warm-water, and rough
fishing days (TDCDF, TDcLF, TD^MP, and TD^). In contrast to Vaughan and Russell (1982), the
economic model distinguishes cool- water and warm- water fishing activities. In doing so, it is
assumed that the average number of fishing days per person is the same for cool-water and
warm-water fishing and that the ratio of the total number of cool-water fishing days and warm-
water fishing days is equal to the ratio of the best-use acreage estimates for cool-water and
warm-water fishing.
After designating the behavioral responses with the results of the three stages, the
economic model values the predicted behavioral responses. The welfare or valuation analysis
of the economic model is couched in relative terms, with values being placed on the changes in
the number of fishing days relative to baseline estimates (CTDCDF, CTDCLF, CTD^^p, and
A discussion of the definition and valuation of recreational fishing days follows.
A recreational fishing day is commonly known as one person on-site for any part of a
calendar day (Walsh et al., 1992). Economists have been deriving fishing day values for over
3-14
-------
CHAPTERS
Exhibit 3-4
Predicting the Number of Person Days Per Year Devoted to Fishing by Guild1
Variable
Intercept
Age
Preference intensity for fishing dummy
Age-squared
Gender (female)
Metropolitan area
Average income per household
Coastal area
Average number of cold-water gamefish
caught per day
Cold-water fishing acreage per capita
Average number of warm-water
gamefish/panfish caught per day
Warm-water fishing acreage per capita
Average number of rough fish caught per
day
Rough fishing acreage per capita
Source
Not applicable
Vaughan and Russell (1982)
Vaughan and Russell (1982)
Vaughan and Russell (1982)
Vaughan and Russell (1982)
Vaughan and Russell (1982)
Vaughan and Russell (1982)
Vaughan and Russell (1982)
Vaughan and Russell (1982)
Abt economic and thermal models
Vaughan and Russell (1982)
Abt economic and thermal models
Vaughan and Russell (1982)
Abt economic and thermal models
1 The input variables employed in the third stage represent specific types of
anglers and specific types of recreational fishing opportunities. Separate
predictions are derived for cold-water, cool-water/warm-water, and rough fishing.
twenty-five years. A number of recent reviews of empirical research efforts have been published
(American Fisheries Society, 1993; Freeman 1993b, Walsh et al., 1992; Walsh et al., 1990; and
Sorg et al., 1984). Several of these review studies have prompted related lines of research such
as the comparison of values from different studies (i.e., Smith and Kaoru 1990; Kling 1988;
Walsh et al., 1990; 1992) and the transferring of values across geographical areas, populations,
and time (i.e., Atkinson et al., 1992, and Deck et al., 1992).
3-15
-------
CHAPTERS
Recreational fishing shares two common features with other natural resource recreational
service flows. First, the value of a recreational fishing day varies with the quality of the natural
resource, including the water quality and type and abundance of fish stock. Secondly, access
and rules governing recreational fishing behavior are not typically determined by market forces.
For example, observed fishing behavior in the United States suggests individuals on average
prefer saltwater fishing to warm-water fishing, trout fishing to catfish fishing, and stream fishing
to lake fishing. Observed behavioral patterns also imply that few recreationists actually pay for
the opportunity to fish. Together, these two features allude to the necessity of relying on
inferred estimates of the value of a recreational fishing day. In the context of the proposed
global climate change scenarios, values must be derived that are associated with a large variety
of waters, fishing activities, and fish species and that reflect the preferences of a wide array of
individuals.
The economic model estimates changes in fishing days for cold-water, cool-water, warm-
water, and rough fishing (VCDF, VCLF, V^Q,, and VRF) and then values these changes using a unit
value approach. In short, this reduces to multiplying the estimated changes in cold-water, cool-
water, warm-water, and rough fishing days by the designated fishing day value for each activity
(i.e., dollar value = VCDF * CTDCDF). Although theoretically the appropriate value for changes
in the number of recreational days are the marginal values, such values are rarely observable.
As a result, an average measure of consumer surplus is adopted as an alternative measure of
value. Exhibit 3-5 presents summary statistics of average consumer surplus values derived by
Walsh et al., 1992 for cold-water fishing, anadromous fishing, warm-water fishing, and
saltwater fishing. These values were calculated as part of an empirical review of recreational
fishing day values. The values represent the dollar amount individuals are willing to pay over
and above their current expenditures to ensure the continued availability of the opportunity to
use recreational fishing resources (Walsh et al., 1992). The values displayed in Exhibit 3-5 are
in 1993 dollars. The figures suggest that salt water fishing days are the highest valued type of
activity day, followed thereafter by anadromous, cold-water, and warm-water fishing days. In
all four categories of fishing activity, the estimates exhibit broad ranges: salt water fishing
($23.08-$271.27), anadromous fishing ($20.81-$157.17), cold-water fishing ($12.44-$145.88),
and warm-water fishing ($10.04-$73.38).
The descriptive statistics presented in Exhibit 3-5 reveal the extent of possible fishing day
values and intimate the difficulty of deriving a specification of fishing day values that is
appropriate for the sample and geographic scale of the economic model. Since Vaughan and
Russell modeled the benefits of water pollution control, much time and effort has been devoted
to the analysis of recreational values. With the advent of benefits transfer and comparison of
valuations research, there may be room for additional refinements of the approach discussed
below and adopted by the economic model. Natural limitations to the refinement process are
defined by the scale of existing data and modelling sources. Data such as the National Survey
of Fishing, Hunting, and Wildlife-Associated Recreation Survey is often organized at the state
or national level, and the modelling precision of national waters is often limited to some state
or regional level. These aggregate levels do not necessarily coincide with the appropriate scale
3-16
-------
CHAPTERS
for discussing recreational fishing opportunities and/or values. The range of fishing-day value
estimates across methods, geographical areas, and fish species is made even clearer in
Exhibit G-l in Appendix G which summarizes the results from a select group of recreational
valuation studies.
Exhibit 3-5
Net Economic Values per Recreation Day
Reported by TCM and CVM Demand Studies from 1968 to 1988
(1993 Dollars)1
Activity
Cold-water
fishing
Anadromous
fishing
Warm-water
fishing
Saltwater
fishing
Number of
Estimates
39
9
23
17
Mean
37.82
66.07
29.08
89.53
Median
35.19
57.11
27.79
65.89
Standard
Error of
Mean
4.94
16.79
3.75
21.43
95% CI
29.97-45.66
40.05-93.35
23.13-35.65
55.51-123.54
Range
12.44-145.88
20.81-157.17
10.04-73.38
23.08-271.27
Source: Walsh et al., (1992).
1 Values have been converted to 1993 dollars using the Consumer Price Index.
Because the economic model follows Vaughan and Russell (1982) quite closely, it is
valuable to first describe what estimates these authors adopted for the average values of fishing
days. Vaughan and Russell independently derived estimates from a two-stage travel cost method
for trout and catfish fishing day values. They then calculated a value for a bass fishing day
using information from Charbonneau and Hay (1978) about the differences in willingness to pay
for catfish and trout days ($6) and bass and trout days ($2). The values for trout ($11), bass
($10), and catfish ($7) fishing days were used to represent the value of cold-water, warm-water,
and rough fishing days respectively.
The economic model employs an approach quite close in spirit to this technique. The
determination of the primary set of fishing day values involves combining two sources of
information. First, the mean values of cold-water, anadromous, and warm-water fishing days
reported in Walsh et al., 1992, were selected to serve as the basis for the values (see
Exhibit 3-5. These values were derived from 71 empirical studies conducted between 1968 and
1988, and the value estimates from these studies were adjusted by the authors to establish some
level of consistency across research methods. Adjusted to 1993 dollars using the Consumer
Price Index, the mean fishing day values are $37.82 for cold-water fishing, $66.70 for
anadromous fishing, and $29.08 for warm-water fishing. Because the thermal model includes
3-17
-------
CHAPTERS
anadromous fish species in its cold-water guild specification, a weighted average of these two
mean values is used to reflect the value of a cold-water fishing day in the economic model. The
weights were based on the number of cold-water and anadromous fishing days reported in the
1991 U.S. Department of Interior Fish and Wildlife Service National Survey of Fishing,
Hunting, and Wildlife-Associated Recreation (U.S. DOI, 1993). This first step established the
fishing day values for cold-water and warm-water fishing days.
Next, it was necessary to take the mean values and combine them with another source
of information to glean the relevant set of cold-water, cool-water, warm-water, and rough fishing
day values. Similar to Vaughan and Russell (1982), the economic model relies on research
completed by Charbonneau and Hay (1978). However, instead of using the differences in levels
of willingness to pay, the economic model utilizes the relative sizes of the reported willingness
to pay values for rough and warm-water fishing (0.79) as well as rough and cold-water fishing
(0.60). Catfish values are used to represent rough fishing; bass values are used to represent
warm-water fishing; and a weighted average between trout and salmon values are used to
represent cold-water fishing. It is important to note that ideally all of the unit values would.
come from one source, but as will become apparent in the subsequent discussion of empirical
works, few studies attempt to capture values for more than one type of fishing activity or more
than one state. The Charbonneau and Hay (1978) study is a rare exception. This study was
conducted in 1975 and employed travel cost and contingent valuation methods to derive
nationally based fishing-day values for a variety of freshwater and saltwater fishing activities.
Three specifications of fishing-day values were adopted to assess the sensitivity of the
model to the relative sizes of these values. Exhibit 3-6 presents the three specifications of
fishing-day values. The primary specification of fishing-day values takes the modified Walsh
et al., (1992) mean values for cold-water and warm-water fishing-day values; cool-water fishing
days are assumed to be valued in the same way that cold-water fishing-day values are. Finally,
the rough fishing-day value represents the product of the warm-water fishing-day value with the
ratio of rough to warm value estimated by Charbonneau and Hay (1978). The high and low
specifications of fishing-day values are alterations of the primary specification. In this context,
high refers to widening the range of fishing-day values; while low suggests a narrowing of the
range of values.8
The implementation of the economic model concludes with the derivation of twelve basic-
outputs for each climate scenario. These outputs include the changes in acreage by thermal
8No distinction is made in this study between net willingness-to-pay (WTP) for stream- and
for lake-fishing. The measurement error involved is unknown. Even the direction of bias is
unknown but it is speculated that stream cold-water fishing has higher net WTP. If so, the
damages from lost stream cold-water fishing could be underestimated. In part, this possibility
is captured by the high specification described here even though that did not explicitly motivate
the specification.
3-18
-------
CHAPTERS
category, the changes in fishing days by thermal category, and the associated changes in value
(benefits or damages) by thermal category. The results section of this chapter is devoted to the
comparison and contrast of these outputs for the different climate scenarios. Sensitivity analyses
are also presented that reveal the importance of various assumptions and specifications employed
by the thermal and economic models.
Exhibit 3-6
Fishing-Day Value Specifications
1993 Dollars
Fishing Activity
Cold-water
Cool-water
Warm-water
Rough
Primary
41.75
41.75
29.08
22.96
High
45.93
41.75
29.08
20.66
Low
41.75
29.08
29.08
24.96
3.5 ANNUAL ECONOMIC WELFARE EFFECTS, SELECTED YEARS AND MODELS
The entire framework for the thermal and economic modelling is constructed to estimate
economic welfare effects for a single year in the future. Which year becomes the target is in
part a function of the GCM used to drive the thermal modelling. The equilibrium GCMs
(GFDL, GISS, OSU, UKMO) are anchored on the year or years during which an equilibrium
state associated with the doubling of CO2 is reached. Under the IPCC's central emission
scenario (IS92a), radiative forcing equivalent to that associated with CO2 doubling is expected
sometime between 2060 and 2070 (IPCC, 1992, Figure Ax.3, p. 174). Different emissions
assumptions could move the date forward or later. For the purposes of this analysis, the year
2060 is taken as the initial date.
For the transient GCMs, it is possible in theory to estimate the thermal profiles for
different years. Exploiting this capability to an extreme would generate an enormous amount
of data to manage, especially if, as is done in this study, regional- and month-specific data are
needed. Consequently, only two time periods are characterized using the estimates from the
transient GCM selected for this study. Those tune periods are linked to two years commonly
studied in climate analyses, 2050 and 2100. The results from these two specifications are
labeled "TR 2050" and "TR 2100," where the TR designates the GFDL transient model.
These six models (GFDL, GISS, OSU, UKMO, TR 2050, TR 2100) constitute the main
set of GCM results applied in the thermal-economic modelling. No attempt has been made to
3-19
-------
CHAPTERS
select one model as the best representative. Instead, estimates of economic impacts are provided
for all six. The four equilibrium models, though normalized to a uniform, annual global climate
sensitivity of 2.5°C, lead to different economic impacts because of spatial and temporal
differences in the distribution of their thermal effects. Consequently, it is important to present
all their results here. Including the transient model results corrects a shortcoming in the
equilibrium GCMs. Because the latter are pegged to a very special and possibly fleeting
circumstance (CO2 doubling), they shed no light on the rate of climate change and present a
significant analytical gap if the realized temperature increase in 2100 is very different from that
in 2050. Unfortunately, the transient model is also not sufficient by itself, given the lack of
consensus in GCM modelling.
This section presents the estimated annual economic welfare effects of climate change on
recreational fishing for these selected GCMs and their associated time periods. Because the
results could not all be adjusted to a common year, they are best viewed as a means for
illustrating how the output of the thermal model (acreages by fish type) translates into different
estimates of welfare effects. The next section will convert these results into comparable terms,
estimating the present discounted value of the impacts in the model year and in subsequent years.
Exhibit 3-7 presents the estimated annual welfare impacts for the four equilibrium GCMs.
For each fish type, both the thermal results, in terms of changes in "best-use" acreage, and the
economic results are provided for each fish type (cold, cool, warm, rough). Damages occur
when there is a loss in the value of a particular fishing type between the baseline and the time
after climate change has occurred. These are depicted in parentheses, indicating negative values.
For example, a loss of 2.5 million cold-water acres under the GFDL model is estimated to cause
$2.2 billion in cold-water fishing losses. The opposite case results in benefits. For example,
a gain of 850 thousand cool-water acres leads to estimated benefits of $305 million under the
GFDL.
The net damages or benefits estimated for a particular model are the sum of the damages
or benefits to cold, cool, warm, and rough fishing. For the primary specification of recreational
fishing-day values, net benefits are estimated for two of the models ($81 and $80 million for the
GFDL and GISS respectively) and net damages for the other two equilibrium models (-$95 and
-$85 million for the OSU and UKMO respectively). Although there are losses in cold-water
acreage estimated by all four models, these losses are offset by gains in cool, warm, and rough
acres. Whether the loss of a cold-water acre is offset by a cool-water of other type of acreage
is critical because of the equivalence assumed for the values of cold-water and cool-water fishing
days in the primary specification. The two GCMs which produce net benefits under the primary
specification (GFDL and GISS) have the largest increases in cool-water acreage. As a result,
larger increases in the number of cool-water fishing days and smaller increases in the lower-
valued warm-water fishing days are estimated in the economic model. Exhibit 3-8 illustrates
these differences in estimated fishing days across the four models. The GFDL and GISS models
are distinguished by their larger increases in cool-water fishing days. These circumstances
provide a telling example of the importance of the specified fishing day values. If cool-water
3-20
-------
CHAPTERS
and warm-water fishing day values were combined, as Vaughan and Russell did, the gains in
cool-water fishing days estimated for the GFDL, GISS, and, to a lesser extent, the UKMO
would be worth a lot less. In contrast, the loss in cool-water fishing days calculated for the
OSU model would have led to smaller economic damages, if cool- and warm-water fishing days
were equally valued. Under this assumption, the net welfare effects for three of the models
change signs. Losses are estimated for the GFDL and GISS models (-$11 and -$64 million
respectively) and a gain is estimated for the OSU model ($14 million). The damages associated
with the UKMO model increase slightly, to -$103 million. These results are shown in
Exhibit 3-9.
The two transient specifications, for the years 2050 and 2100, provide further insight into
the sensitivity of the estimated welfare effects to substitutions among the four fishing types. As
Exhibit 3-10 shows, the largest damages observed for all six GCMs (-$320 million) are estimated
for the TR 2050 model. This phenomenon is influenced less by the loss of cold-water acreage
(which contrary to what one might expect, is the smallest estimated for the six GCM models)
than by the loss of cool-water acreage. This conclusion is reinforced by the results for the TR
2100 model, where the cold-water acreage losses are twice as large but are offset by gains in
cool, warm, and rough acres. As a result, although this model represents the year 2100, when
the average annual temperatures are estimated to be 1.25° C higher than in 2050 (2.75°C vs.
1.5° C), the economic damages are smaller though still substantial (-$266 million per year).
This trend is reversed when the values of cool- and warm-water fishing days are equated.
Exhibit 3-11 shows that the net damages in the TR 2050 model (-$76 million) are one-fourth the
damages observed in the primary specification. Now, however, damages increase over time.
The net damages estimated for the TR 2100 model increase slightly, to -$286 million per year.
Exhibit 3-12 illustrates again how important the relative fishing day values are. For the
TR 2050 model, the number of cold- and cool-water fishing days lost (50 million) are more than
offset by the number of warm-water and rough fishing days gained (64 million) but there is an
overall net economic loss because cold- and cool-water days have been assigned a higher value.
For the TR 2100 model, the number of days gained on net is even greater (24 million) but again,
welfare losses rather than gains are still estimated. These results illustrate another feature of the
economic model. The redistribution of freshwater acreage to warm-water "best uses" leads to
an overall increase in the number of fishing days estimated for U.S. anglers. This tendency in
the economic model underscores the importance of how trade-offs in the different acreage types
are modelled. As the economic model is currently implemented, the loss in cold-water acreage
translates directly into an expansion of cool, warm, and rough fishing opportunities. This
assumption will be investigated further in the sensitivity analyses.
3-21
-------
3
cu
o
O
S 42
fuS
,1
Ll
1
ts
I
S
i
s-
S
?
I
1
0?
g
8
S
is
O
CO
£3
o
-------
Exhibit 3-8
Changes in Fishing Days By Type
Different Equilibrium GCMs
100
co
o
Q
D)
_C
CO
JJL
-100
Cold
Cool Warm
Types of Fish
Rough
BGFDL ^GISS 1OSU ^UKMO
-------
a <-i<
la
^
s
&
S
$
§
o
1
.S
O)
u
.5
O>
sl
S
M
.S
U
^
S1
§
S
«_
Vl"
8
1
1
O
cn
S3
0
S
S
a"
8
Ci
00
00
Q.
£
,0
I
JS
•S
1
.2
|
•B
g"
X
M
2
I
•a
a
«•
8
£
•I
S
I!
i|
^.&
> ^s
tl
i) °
•S "«
en
01
3
-------
I
1
•O
"3
U
& *
Is!
8P
M
1
e
c-
§
8
to
§
o
o
-------
$
1
>>
a* S
5 fl" s
«*-< «9 o
*»-< _ei S
S "S O
1*M ^Q ^<
gS «g
ll i
*5 1
TH w A §
TH *a g «
^ i| .s
« g ^ K>
tS S T3 {g
-S W a s
Si? 1
3 "3 w
So -a
Ju S
^l< C5
"8 JS -g
1 § |
tTj H"i S
*| i
•PM
•^t
2
CO
8,
^
J3
3
o
H
I
fc
I
04
C
i
1
2
u
!bi a3
1
c
||
x "^
£ V9
SP ffl
W SM o
-T
«*1
J «
e
o
-------
Exhibit 3-12
Changes in Fishing Days By Type
Different Transient Benchmarks
100
-100
Cold
Coo) Warm
Fishing Types
Rough
I Transient 2050 ^ Transient 2100
-------
CHAPTER 3
3.6 PRESENT DISCOUNTED ECONOMIC WELFARE EFFECTS
The current analysis is confined to estimating climate change effects for selected years
(2050, 2060, and 2100). To encompass the long time-frame considered relevant for these
effects, it is necessary to extrapolate the results from the selected years to other years in the
future. The results of the previous section provided ambiguous guidance on how any
extrapolation should be made. Even though the temperature increases associated with climate
change are expected to grow continuously from the present until at least the year 2100, the
estimated damages for the transient GCM rose from 2050 to 2100 under one specification of the
fishing day values and fell under another. Consequently, it is difficult to identify the possible
bias of even the simplest extrapolation, such as holding the welfare effects constant into the
future. This approach does, however, have the advantage of reducing the expected bias if
upward and downward tendencies hi the damage estimates are equally likely. On this basis,
holding the estimated damages constant was chosen as the procedure for extrapolating from the
selected years to subsequent years.
Greater care appears to be warranted for the years prior to the selected study years.
Although the estimated annual damages associated with them are likely to be smaller than those
estimated for 2050 and 2060, they are discounted less and therefore cany greater weight in the
present value of welfare effects calculated for 1994. For the current study, a conservative
approach was taken. In the absence of more specific evaluation, zero effects were assumed for
the time period between the present and the years 2050 and 2060, the relevant start years for the
transient and equilibrium scenarios, respectively. This assumption has not yet been considered
in a sensitivity analysis but it should be eventually.
Exhibits 3-13 and 3-14 summarize in graphical terms the assumptions applied to the
estimated annual effects. Though couched in terms of damages, they apply analogously to those
cases where net benefits have been estimated. The approach taken to the equilibrium results is
straightforward. The estimated effects are applied to the approximate year when CO2 doubling
is expected and to all subsequent years until 2193, approximately two hundred years from the
present. Damages in the years after 2100 can have either a little or a tremendous influence on
the present discounted welfare effect, depending on whether a low (1 %) or a high (7%) discount
rate is applied.
The approach taken to the transient results departs slightly from the equilibrium
approach. Damages start sooner (2050), are based on the 2050 transient model, are held
constant until 2100, and change then to the damages estimated for 2100 in the transient model.
The damages are held constant at this level from 2100 to 2193.
Extremely long time-frames magnify the difficulty in selecting an appropriate discount
rate. Present values can differ by two orders of magnitude when rates of 1 percent and
7 percent are applied to this study's results. Evidence reviewed by Freeman suggests that the
after-tax real rate of interest, a basis for estimating individuals' rates of time preference, falls
3-28
-------
CHAPTERS
in the range of 1 percent to 4 percent (Freeman, 1993a), which lowers the difference to an order
of magnitude. Scheraga points to historical experience with real after-tax returns on Treasury
bills and stocks as justification for using 3 percent as a consumption rate of interest (Scheraga,
1989). Several critiques of discounting, including ones from economists, have suggested that
discounting is not suitable for circumstances where multiple generations are involved. Freeman,
for example, argues that the trade-off across tune that is implicit in the derivation of rates of
time preference are not feasible for multiple generations. Taken together, those views support
using an even lower discount rate. The discussion presented below focuses on the results for
a 1 percent discount rate. For completeness, results based on rates from 2 percent to 7 percent
are also provided.
Present discounted welfare effects for the six models, based on the primary specification
of fishing day values, are presented in Exhibit 3-15. As was observed in the previous section,
the equilibrium results are mixed for this specification. Two equilibrium models exhibit benefits
and the other two, damages. For a 1 percent discount rate, the present values range from $4
billion in damages to $3 billion in benefits.
The maximum damages do not change substantially when equal values for cool- and
warm-water fishing days are used, as shown in Exhibit 3-16 but signs of the other three models'
results change. As a result, under this specification, three models have estimated damages
ranging from $424 million to $4.0 billion and one model (OSU) has estimated benefits of $540
million.
Exhibits 3-15 and 3-16 also present the estimated present value of damages for the
transient model. While this model is also affected by alternate assumptions regarding the value
of various types of fishing days, the results for the two specifications are more robust than they
were for the equilibrium models. Damages of -$12.9 billion and -$7.8 billion are estimated for
the primary and alternate specifications of relative fishing day values. These estimated damages
are two to three times larger than any estimated damages based upon the equilibrium models.
3-29
-------
CO
Q
/§
/i
CO
o
CO
CD
O
MD
O
cs
0%
CO
o
-------
m
CO
Q
o
§
"3
CO
s
w r i '"r
^ -a
X c
0>
o
o
CO
s
e
o
••c
C3
&
CO
03
vo
vo
(N
O
CO
11
O
o
CN
o
o
g
S3
XSI^N
en
ON
ON
^^n
bJD
.fl
')— I
^
E
E
1
oo
CN
03 S
"o S
O \o
O r--
0*
W
o
TJ I
C5J g
CJ DJD
S^M R
^^^ C
• ^M i es
w e
a i
c
p
o
O
>*
t-i
-------
>o\
.a>
ve
oo
VO
ON
s
i
o
CO
en
52-
r*
c*
8
r
»
*?
^
s
-------
ve
•0
2060-2193
U
po
0-2
,
en
00
t, 2050-2
-------
CHAPTERS
REFERENCES
American Fisheries Society. 1993. Sourcebook for Investigation and Valuation of Fish Kills.
Prepared by Southwick Associates, Arlington Virginia, Supplement to Special Publication
24.
Atkinson et al., 1992. Bayesian Exchangeability, Benefit Transfer, and Research Efficiency.
Water Resources Research 28(3): 715-722.
Carson, R.T. and R.C. Mitchell. 1993. The Value of Clean Water: The Public's Willingness
to Pay for Boatable, Fishable, and Swimmable Quality Water. Water Resources
Research, Vol. 29, No. 7, pp. 2445-2454, July.
Charbonneau, J.J., and MJ. Hay. 1978. Determinants and Economic Values of Hunting and
Fishing. Transactions of the North American Wildlife and Natural Resources Conference
43: 391-403.
Deck, L. et al., 1992. Benefits Transfer: How Good is Good Enough. Presented at the 1992
AERE Workshop, Benefits Transfer: Procedures, Problems, and Research Needs,
Snowbird, Utah, June 3-5.
Freeman, A. M., UJ. 1993a. The Measurement of Environmental and Resource Values:
Theory and Methods. Resources for the Future, Washington, DC.
Freeman, A.M., HI. 1993b. The Economics of Valuing Marine Recreation: A Review of
Empirical Evidence. Economics Working Paper 93-102, Department of Economics,
Bowdoin College, Brunswick, Maine.
Intergovernmental Panel on Climate Change (IPCC). 1992. 1992 Supplement: Scientific
Assessment of Climate Change. Submission from working Group I.
Kting, C.L. 1988. Comparing Welfare estimates of Environmental Quality Changes from
Recreation Demand Models. Journal of Environmental Economics and Management
15:331-340.
Michaels, G., K. Sappington, L. Akeson, M. Wojcik, T. Aagaard, and D. DeWitt. 1992.
Ecological Impacts from Climate Change: Scoping Study for an Economic Assessment.
Prepared for the Adaptation Branch, Climate Change Division, Office of Policy,
Planning, and Evaluation, U.S. Environmental Protection Agency. Abt Associates Inc.,
Bethesda, MD. November 11.
3-34
-------
CHAPTERS
Scheraga, J. 1989. Supplemental Guidelines on Discounting in the Preparation of Regulatory
Impact Analyses. Mimeo. Economic Studies Branch, U.S. Environmental Protection
Agency, Washington, DC. March.
Smith, V.K., and Y. Kaoru. 1990. Signals of Noise? Explaining the Variation in Recreation
Benefits Estimates. American Journal of Agricultural Economics 72(2): 419-433.
Sorg, C.F. et al., 1985. Net Economic Value of Cold and Warm Water Fishing in Idaho. U.S.
Forest Service, Rocky Mountain Forest and Range Experiment Station, Resource Bulletin
RM-11, Fort Collins, Colorado.
Sorg, C.F., and J.B. Loomis. 1984. Empirical Estimates of Amenity Forest Values: A
Comparative Review. General Technical Report RM-107, Prepared for the Rocky
Mountain Forest and Range Experimental Station, Forest Service, Department of
Agriculture, Fort Collins, Colorado.
U.S. Department of Interior/Fish and Wildlife Service. 1993. 1991 National Survey of
Fishing, Hunting, and Wildlife Associated Recreation. Washington, DC.
Vaughan, W.J. and C.S. Russell. 1982. Freshwater Recreational Fishing: The National
Benefits of Water Pollution Control. Washington, DC: Resources for the Future.
Walsh, Richard G., Johnson, D.M., and J.R. McKean. 1992. "Benefit Transfer of Outdoor
Recreation Demand Studies. 1968-1988." Water Resources Research 28(3): 707-713.
Walsh, Richard G., Johnson, D.M., and J.R. McKean. 1990. "Nonmarket Values from Two
Decades of Research in Recreation Demand." in Advances in Applied Micro-Economics
Volume 5, 167-193.
3-35
-------
CHAPTERS
This Page Left Blank Intentionally.
3-36
-------
CHAPTER 4
4.
EVALUATION OF THE ANALYTICAL FRAMEWORK
4.1 INTRODUCTION
The results and methodology discussed in Chapters 2 and 3 are those employed for the
primary specification of ecologic and economic input variables and modelling assumptions.
Chapter 4 presents several different analyses that incorporate alternative specifications for these
input variables and modelling assumptions. In keeping with the objective of this study to
identify potential damages while attempting to minimize any biases that might overstate or
understate the estimated damages, the primary specification was constructed using conservative
assumptions. It is important to note, however, that several of the alternative specifications may
be equally valid as the primary specification as the appropriate basis for analysis. The current
state of knowledge about recreational fishing behavior does not allow a definitive choice among
alternative specifications. In this sense, the primary specification may be a reasonable choice
but is still highly debatable. What typically distinguishes the primary specification from the
alternatives is that the latter tend to result in much higher or small estimated damaged. The
value of conducting sensitivity analysis is manifested in the following sections, as the importance
of certain variables and modelling assumptions are made clear. These types of discoveries
provide a better understanding of the context in which to view the primary specification results
and often point to areas for further research.
Sections 2 through 8 of Chapter 4 summarize seven evaluations of the models that were
conducted as part of the modelling of ecological impacts from global climate change. The
sections address revisions to ecologic as well as economic modelling components. The seven
analyses include modifications of the treatment of or assumptions associated with (1) fishing-day
values; (2) climate and emission scenarios sensitivity; (3) fish thermal tolerances; (4) fish
habitats; (5) warm-water fishing behavior; (6) cold-water acreage substitutability; and (7) runoff.
The discussions in each of the sections emphasize the extent to which the results change relative
to the primary specification. This provides some indication of the 'importance' of the
assumptions. Additional attention is devoted to the changes in acreage and valuation patterns
within and across global climate change scenarios.
4.2 FISHING-DAY VALUES
As discussed in Chapter 3, the economic model is run using three different recreational
fishing-day value specifications. These specifications are referred to as low, primary, and high.
The specifications employ the following sets of values for cold-, cool-, warm-water, and rough
fishing-day values: low ($41.75, $29.08, $29.08, and $24.96); primary ($41.75, $41.75,
$29.08, and $22.96); and high ($45.93, $41.75, $29.08, and $20.66). The sets of values differ
according to their specification of the relative worth of types of recreational fishing activities.
4-1
-------
CHAPTER 4
The variations are primarily associated with the handling of cool-water fishing days and rough
fishing days.
As discussed earlier, the primary specification of fishing-day values is based on the
Walsh et al. (1992) values for cold-water and warm-water fishing days. The primary
specification rates cold-water fishing days and cool-water fishing days equivalently and assigns
the rough fishing-day value as the product of the warm-water fishing-day value with the rough
to warm-water fishing-day value ratio derived from Charbonneau and Hay(1978) estimates. In
contrast, the low specification of recreational fishing-day values assigns equal values to warm-
water and cool-water fishing days and derives the rough fishing-day value using the Charbonneau
and Hay (1978) rough to cold-water fishing-day value ratio. Relative to the primary
specification, the low specification awards less value to the loss in cool-water fishing days and
more value to the increase in rough fishing days. The high specification of recreational fishing-
day values is defined by extending the range of fishing-day values. To form the high
specification, the primary cold-water fishing-day value is increased by 10 percent and the
primary rough fishing-day value is lessened by 10 percent. These adaptations cause the high
specification to allocate more weight to the loss in cold-water fishing days and less value to the
increase in rough fishing-day values relative to the primary specification.
The economic model produces estimates of the changes in acres and fishing days by best-
use activity and calculates the associated value (benefits or damages) of these transitions.
Central to this calculation are the tradeoffs between the different types of recreational fishing
activities. In short, the economic model characterizes the welfare implications of shifts iii
recreational fishing opportunities. In the context of global climate change, these shifts involve
movements from cool-water fishing activities to warm-water fishing activities. By design, the
final results (benefits or damages) are quite sensitive to the relative magnitudes of the values
across fishing activities.
Exhibit 4-1 displays the output of the economic model for four equilibrium scenarios
(GF1, Gil, OS1, and UK1) and two transient scenarios (TR1 and TR2). For each scenario, the
estimated changes in fishing days and changes in value are presented under the low, primary,
and high fishing-day value specifications. To emphasize the differences across specifications,
total and activity-specific (i.e., cold, cool, warm, and rough) changes are presented. Comparing
the low, primary, and high results, several interesting patterns appear. In the UK1, TR1, and
TR2 scenarios, the size of the estimated damages from climate change Increase successively
moving from the low to primary to high specifications. This same trend holds for the OS1
scenario, but interestingly enough, there is a sign change moving from the low
specification(benefits of 57 million dollars) to the primary specification(damages of 95 million
dollars). The Gil scenario stands out from the other scenarios as the low and high specifications
predict damages of approximately 26 and 136 million dollars and the primary specification
estimates benefits of approximately 80 million dollars.
4-2
-------
CHAPTER 4
Exhibit 4-1
Assessing Sensitivity to Fishing-Day Values
Scenario
GF1
Cold
Cool
Warm
Rough
GII
Cold
Cool
Warm
Rough
OS1
Cold
Cool
Warm
Rough
UK1
Cold
Cool
Warm
Rough
Changes in
Fishing Days
(1,000s)
27,367
(51,876)
7,307
47,488
24,448
19,733
(41,292)
11,301
31,064
18,660
24,918
(45,593)
(8,572)
57,337
21,746
27,793
(59,230)
1,428
59,338
25,257
Changes in Dollar
Value, Low1
(Millions)
38
(2,166)
213
1,381
610
(26)
(1,724)
329
903
466
57
(1,904)
(249)
1,668
543
(50)
(2,473)
42
1,726
655
Changes in Dollar
Value,
Primary2 (Millions)
81
(2,166)
305
1,381
561
80
(1,724)
472
903
428
(95)
(1,904)
(358)
1,668
499
(85)
(2,473)
60
1,726
603
Changes in Dollar
Value, High3
(Millions)
(191)
(2,383)
305
1,381
505
(136)
(1,897)
472
903
386
(335)
(2,094)
(358)
1,668
449
(393)
(2,720)
60
1,726
542
4-3
-------
CHAPTER 4
Scenario
TR1
Cold
Cool
Warm
Rough
TR2
Cold
Cool
Warm
Rough
Changes in
Fishing Days
(1,000s)
13,690
(30,906)
(19,272)
50,441
13,427
24,439
(65,652)
1,576
61,621
26,894
Changes in Dollar
Value, Low1
(Millions)
(49)
(1,290)
(561)
1,467
335
(232)
(2,741)
46
1,792
671
Changes in Dollar
Value,
Primary2 (Millions)
(320)
(1,290)
(805)
1,467
308
(266)
(2,741)
66
1,792
617
Changes in Dollar
Value, High3
(Millions)
(480)
(1,420)
(805)
1,467
277
(602)
(3,015)
66
1,792
556
Notes:
The specifications for fishing-day values include values for cold, cool, warm, and rough water
fishing days.
1 The low specification is 41.7548; 29.0843; 29.0843; and 24.9610.
2 The primary specification is 41.7548; 41.7548; 29.0843; and 22.9562.
3 The high specification is 45.9303; 41.7548; 29.0843; and 20.6606.
4.3 CLIMATE AND EMISSIONS SCENARIO SENSITIVITY
The thermal model simulates changes in fish presence as a function of increasing
temperatures using estimates produced from different general circulation models(GFDL, GISS,
OSU, and UKMO) and emissions scenarios. The general circulation models and emission
scenarios work in tandem, with the general circulation model predicting changes in the level of
warming for the concentration of the gases in the atmosphere designated by the emission
scenarios. The emission scenarios data reflect projected estimates of various greenhouse gas
emissions which in turn are based on various assumptions concerning population growth, energy
use, industrial technologies, economic circumstances, and tropical deforestation and forest
biomass. Because of the complexity of projecting emissions over tune, there is considerable
uncertainty associated with the projected emissions scenarios. The primary specification
discussed in Chapter 2 focuses on the best or median emission estimates (IS92a). In this
sensitivity analysis, the results for the primary (1), low (2), and high (3) emission scenarios are
presented for four equilibrium scenarios (GFDL; GISS; OSU; and UKMO) and two transient
4-4
-------
CHAPTER 4
scenarios (TR1 and TR2). These three specifications represent different levels of climate
sensitivity to greenhouse gas emissions. The high and low specifications respectively result in
greater and lesser global mean temperature increments than the primary specification.
The climate sensitivity assumptions affect the output of the economic model by modifying
the way in which best-use fishing acreage is shifted from one thermal category to another.
These shifts in acreage change the predicted estimates of general fishing participation, fishing
participation by type of activity, and the number of days devoted to the different fishing
activities. In general, it is expected that higher (lower) temperature increments will result in
greater (smaller) decreases in the higher valued cold-water and cool-water fishing days and
greater (smaller) increases in the lesser valued warm-water and rough fishing days relative to
the primary increment. The relative sizes of these counteracting changes in value determine the
estimated sign of the dollar value associated with the predicted recreational fishing behavior
response. The equilibrium and transient models are discussed separately, for the models employ
different sets of temperature increments.
4.3.1 Equilibrium models
The climate sensitivity assumptions adopted for the twelve equilibrium models are based
on three levels of projected increases in temperature: 1.5° C, 2.5° C, and 4.5° C (refer to
Exhibit 4-2). These three increments correspond to the low (2), primary (1), and high (3)
climate sensitivity specifications. Exhibit 4-2 presents the results from the twelve equilibrium
scenarios. The results are organized by general circulation model. The numbering reflects the
different assumptions concerning temperature increments. The lower temperature increment
applies to models GF2, GI2, OS2, and UK2. The higher temperature increase is used for GF3,
GI3, OS3, and UK3. The middle or median level temperature increase applies to GF1, Gil,
OS1, and UK1. Exhibit 4-3 displays the estimated changes in acres and changes (benefits or
damages) in dollar value associated with each scenario.
Best-use cold acreage falls relative to the primary specification for all of the high
temperature increment scenarios. The UK3 scenario shows the highest loss (4.2 million) in best-
use cold-water acreage, while the GI3 and OS3 scenarios reveal the lower decreases (3.6
million) in cold-water acreage. Best-use cool-water acreage decreases in all of the high and low
specifications relative to the primary specification in all but one of the scenarios (OS2). The
OS2 scenario shows higher levels of best-use cool-water acreage in the low specification relative
to the primary specification. The increases in best-use warm-water and rough acres are greater
(smaller) for the high (low) specifications than the primary specification.
4-5
-------
CHAPTER 4
Exhibit 4-2
Global Annual Mean Temperature Increases
Used to Scale Equilibrium GCM Results
GCM Run
GFDL/
Equilibrium
GISS/
Equilibrium
OSU/
Equilibrium
UKMO/
Equilibrium
IPCC Climate Sensitivity1
Low
1.5°C
GF2
1.5°C
GI2
1.5°C
OS2
1.5°C
UK2
Primary
2.5°C
GF1
2.5°C
Gil
2.5°C
OS1
2.5°C
UK1
High
4.5°C
GF3
4.5°C
GD
4.5°C
OS3
4.5°C
UK3
Notes:
1 IPCC 1992. (p. 10.) All GCMs reflect
equilibrium effects from the doubling of
C02.
Several interesting patterns appear in the results of the sensitivity analyses presented in
Exhibit 4-3. The models with high temperature increases return total damages in every scenario.
The estimated damages under the high specification range from 131 million dollars in the GI3
scenario to 751 million dollars in the UK3 scenario. Conversely, the models with low
temperature increases yield total benefits across all scenarios. The estimated benefits range from
36 million dollars in the GI2 scenario to 395 million dollars in the GF2 scenario. The dollar
values associated with the primary specification models all fall between the two extremes except
in the case of the Gil scenario. The Gil primary scenario has higher estimated benefits (80
million dollars) than the low temperature increment scenario GI2 (36 million dollars). This
departure from the expected relationship (higher damages with higher temperatures) is attributed
to the low ratio (1 to 8) of benefits from cool-water fishing for GI2 (56 million dollars)
compared to Gil(472 million dollars). The relative sizes of the cold, warm, and rough benefits
or damages of the GI2 and Gil scenarios are 1 to 2, 2 to 3, and 1 to 2 respectively.
The greatest changes in dollar value relative to the primary climate sensitivity
specification appear for the UK1 and GF1 high specification runs. In these runs, the estimated
dollar values fall relative to the primary specification by 666 and 642 million dollars
respectively. For the low climate sensitivity runs, the changes in dollar value relative to the
4-6
-------
g
.a
0)
u
«
.9
§
o
.a
S
a
.a
I
§
s
8
s
c£
2 "§>
o 3
1
CO
g
oT
e
o
»—c
t-"
I
IS
a
a;
VO^
SS
gf
3
g!
s
s
o
8"
ts
•—
a
§i
I
S!
VI
a.
-------
«
9
«
I
.a
.9
O
o
U
.9
o
U
e »
3
£
o
irT
o\
S
a
9
1
vo
"V
VO
s
§
-------
CHAPTER 4
primary specification range from a 44 million reduction in dollar value under the Gil scenario
to a 314 million increase in dollar value under the GF1 scenario relative to the primary climate
sensitivity specification.
4.3.2 Transient models
The climate sensitivity assumptions adopted for the transient models rest on two sets of
estimated temperature changes: 0.9°C, 1.5°C, and2.4°CforTR3, TR1, andTRS; and 1.65°C,
2.75°C, and 4.4°C for TR4, TR2, and TR6 (refer to Exhibit 4-4). These temperature
increments represent the low, primary, and high specifications. Exhibit 4-5 presents the results
from the six transient model scenarios. The lower temperature increment applies to models TR3
and TR4. The higher temperature increase is used for TR5 and TR6. The middle or median
level temperature increase applies to TR1 and TR2. Exhibit 4-5 displays the estimated changes
in acres and changes(benefits or damages) in dollar value associated with each scenario.
The estimated changes in cold-water, warm-water, and unfishable acres are consistent
with expectations. Higher increases in temperature reduce cold-water acres more and increase
warm-water and unfishable acres more than lower temperature changes. Specifically, the high
specifications (TR5 and TR6) exhibit larger cold-water acreage decreases and larger warm-water
and unfishable average increases than their counterparts in the primary specifications (TR1 and
TR2). Because of threshold effects in the conversion of cold-water acres to other categories,
among other causes, changes in cool-water and rough acreage do not coincide with expectations
based solely on the direction of temperature changes. Higher temperatures increases result in
fewer cool-water and rough acres (TR2 vs. TR6) since warm-water and unfishable acres increase
so dramatically.
The trends in estimated shifts in best-use acreage are reflected in the changes in dollar
value associated with the recreational fishing behavior responses. The transient models with
high temperature increases (TR5 and TR6) both yield total damages. The models with low
temperature increases (TR3 and TR4) yield damages and benefits respectively. The TR4
scenario yields benefits of $256 million, which contrasts with the predicted damages of $266
million of the TR2 primary scenario. The TR3 low scenario results in damages ($70 million)
smaller than those predicted by the primary TR1 scenario ($320 million). The expected
relationship (higher temperatures, greater damages) between climate settings holds true for the
TR6, TR2, and TR4 grouping but not for the TR1, TR3, and TR5 grouping. The damages
estimated for the TR5 scenario ($265 million) are lower than the damages estimated for the TR1
scenario ($320 million) (Exhibit 4-5). The discrepancy is explained by the relative sizes of the
damages from reductions in cold and cool best-use acreage and the benefits of increases in warm
and rough best-use acreage.
The greatest change in dollar value relative to the primary specification results appears
for the TR2 low climate sensitivity specification where the estimated dollar value increases
relative to the primary climate sensitivity specification by $522 million. In the case of the high
4-9
-------
CHAPTER 4
specification for the TR2 model, the change in dollar value falls by $296 million relative to the
primary climate sensitivity result. The TR1 model shows increases in dollar value relative to
the primary climate sensitivity run for both the high ($55 million) and low ($250 million) climate
sensitivity specifications.
Exhibit 4-4
Global Annual Temperature Increases
Used to Scale Transient GCM Results
GCM
Run
GFDL/
Transient
"1.16°C
decade"
GFDL/
Transient
"Eighth
decade"
IPCC Climate Sensitivity1
Low
0.9°C
in 2050
TR3
1.65°C in
2100
TR4
Primary
1.5°C
in 2050
TR1
2.75°C in
2100
TR2
High
2.4°C
in 2050
TR5
4.4°C in
2100
TR6
Notes:
1 The GCMs were normalized to each of the
climate sensitivity assumptions using the
rates of temperature increase observed in
IPCC (1972) estimates as interpreted by
EPA (Leary, 1994).
4-10
-------
£3
£
£
2
"o
u
9
I
fe
ee
s a
1*1
i ° s
§ M
S? ffl
©
o
t-^
I
g
S3
"I
•o"
s
8"
8
n-
oC
°i.
oo
c-
S
S
s
e
0_
oo
r-
&
G.
M
a
o
»-H
•o"
1
s
e
-------
CHAPTER 4
4.4 FISH THERMAL TOLERANCE DESIGNATIONS
The thermal tolerance assumptions determine the manner in which species and therefore
guilds of fish change with the predicted thermal dynamics and shifts in best-use fishing acreage.
There is considerable uncertainty associated with the specification of thermal tolerances as well
as with the methodology for deriving guild information based on the well-being of selected
species within a guild. The purpose of this sensitivity analysis is to examine the ways in which
the tolerance assumptions affect the outputs of the economic model and to compare these effects
across several different global climate models.
The model was run using three specifications of thermal tolerances. The results
discussed in Chapters 2 and 3 focus on the primary specification. The high and low tolerance
specifications are modifications of the primary specification, as the primary, high, and low
tolerances are the Fish-Temperature Data Matching System 95th percentile (FTDMS) numbers,
plus and minus their standard error. For a discussion of the FTDMS, please refer to Chapter 2.
The sets of specifications were adopted for four equilibrium models (GF1, Gil, OS1, and UK1)
and two transient models(TRl and TR2).
The high and low tolerance assumptions should provide smaller and greater changes in
damages respectively, relative to the primary tolerance specification. When thermal tolerances
are raised, the number of fish in each guild that can survive or adapt to increased temperature
dynamics rises. Within the structure of the thermal model, this increase results in fewer shifts
in best-use fishing acreage. Benefits are generated when fewer high valued cold and cool-water
acres (and fishing days) are lost because of the higher thermal tolerances. Lowering thermal
tolerances decreases the number of fish in every guild that can survive or adapt to a temperature
increase. In this situation, more damages are generated as more highly valued cold and cool-
water acres (and fishing days) are shifted to a lesser valued best-use acreage designation.
4.4.1 Equilibrium Models
Exhibit 4-6 presents the results from the thermal tolerance sensitivity runs for the four
equilibrium models (GF1, Gil, OS1, and UK1). Changes in acres and changes hi the dollar
value of fishing days are presented for the different thermal categories. The best-use cold
acreage reductions parallel the thermal tolerance specifications (i.e., low tolerance, higher cold
acreage loss) for all of the models except for UK1. The UK1 model has a higher loss of best-
use cold acreage under the primary specification (3.0 million acres) than under the low
specification (2.2 million acres). This same pattern of best-use acreage shifts holds for cool-
water and warm-water acreage; while the shifts in best-use rough acreage under the different
thermal tolerance specifications vary tremendously across the different models.
Changes in dollar value associated with the changes hi recreational fishing behavior are
consistent across all the equilibrium models. Moving from the high to low tolerance
4-12
-------
1
H
t
f
"I
<-> .9
SB 42
i1.-!
i
a.
M
t-
c-
s
3
1
f
o
1
I
VO
%
.
CJ.
S M
SB
"V
ON
ss-
VI
ON"
a
P
i
p
oC
IS
s
^H UD
3S
tf-
S3
S
•a
g
«.
i
r-
t^
o
g
0?
p
E
cn
^H
4
SI
-------
£
f
q
2
"o
U
£
8P
&
LI
S o S
e
t—
E
I
sr
i
m
P
-------
CHAPTER 4
assumptions, the dollar values successively decrease. In each case, the high tolerance
specification returns estimated benefits. Using the high specification, the estimated benefits
range from $82 million for the UK1 model to $584 million for the GF1 model. The primary
specifications for GF1 and Gil result in estimated benefits; whereas the primary specifications
for OS1 and UK1 result in estimated damages. In all cases, the low tolerance specification
results in estimated damages. The OS1 model shows the greatest damages of all the models
($354 million), followed thereafter by the UK1 model ($342 million), the Gil model ($243
million), and the GF1 model ($195 million).
The greatest changes in dollar value relative to the primary thermal tolerance
specification appear for the high thermal tolerance specifications. Relative to the primary
thermal tolerance specification, the dollar values rise considerably across all of the equilibrium
models. The increases relative to the primary specification range from $111 million for the Gil
model to $503 million for the GF1 model. For the low thermal tolerance specifications, dollar
values consistently fall relative to the primary thermal tolerance specification. The Gil model
($323 million) shows the largest reduction in dollar value relative to the primary specification
followed thereafter by the GF1 model ($276 million), the OS1 model ($259 million), and the
UK1 model ($257 million).
4.4.2 Transient Models
Exhibit 4-7 displays the changes in acres and dollar values for the two transient models
(TR1 and TR2). Results from the high, primary, and low thermal tolerance specifications are
presented. The reductions in best-use cold-water acreage become successively smaller, moving
from the low- to high-thermal tolerance specifications, as do the increases in best-use warm-
water acreage. The increases in best-use cool-water acreage follow this same trend for the TR2
model, but the TR1 model shows increases under the low (0.6 million) and high (1.6 million)
thermal tolerance specification and reveals a decrease (0.1 million) under the primary thermal
tolerance specification. The changes in best-use rough acreage vary across the two models.
Similar to the results exhibited by the equilibrium models, benefits are estimated for the
transient models under the high thermal tolerance specification. The TR1 model has benefits
of $467 million and the TR2 model has benefits of 238 million under the high thermal tolerance
specification. Damages are estimated for both the transient models using the primary and low
thermal tolerance specifications. Damages under the low thermal tolerance specification range
from $339 million for the TR1 model to $402 million for the TR2 model. Relative to the
primary thermal tolerance specification, the changes in dollar value are greater under the high
specification than the low specification. Increases in dollar value relative to the primary thermal
tolerance specification range from $504 million for the TR2 model to $787 million for the TR1
model; while reductions hi dollar value relative to the primary thermal tolerance specification
range from $19 million for the TR1 model to $136 million for the TR2 model.
4-15
-------
.2
o
H
£
If
•1
42
*
O.S
I 1
i s 1
1
42
«-<§
° S
oa
£
VI
s
8
1
S
S
ss-
I
i
g
£
8
8
§
VO
s
a
G-
VO
4
-------
CHAPTER 4
4.5 FISH HABITAT DESIGNATIONS
Fish screening matrices were used for the fish species in this study across all the 48
contiguous United States. The thermal model uses the screens to designate fish habitat by state.
A value of one in a matrix position indicates that a particular fish species is present in a specific
state and a zero indicates the absence of the fish type. The presence of a species was decided
by consulting The Audubon Society: Field Guide to North American Fishes, Whales & Dolphins
(Boschung, et al., 1983) and the 1985 Survey of Fishing, Hunting, and Wildlife Associated
Recreation (U.S. DOI/FWS, 1988). The purpose of this sensitivity analysis is to explore the
effects of the screen designs on the results of the economic model. The results discussed in
Chapters 2 and 3 are for the primary specification. The primary specification uses a narrow
screen to assign fish presence and absence. In this sensitivity analysis, results using a modified
wide screen are presented and then compared with the results using the narrow screen.
The narrow screen is based solely on the Audubon Guide's delineation of natural fish
habitat. The descriptions and maps within the guide were examined to determine whether a state
contained a given fish. If an illustration indicated a ten percent or greater coverage in a state
a value of one was assigned to the matrix position. The term narrow refers to the Audubon
Guide's relatively conservative estimate of fish habitat.
The wide screen is based both on the narrow screen and data from the 1985 National
Survey of Fishing, Hunting, and Wildlife Associated Recreation (U.S. DOI/FWS). The 1985
National Survey groups fish into sets based on identified similarities, with each set containing
one to four fish species. Preliminary screens were developed for each of the sets based on
information concerning catch information by state. The screens developed using this method
typically encompass larger areas for fish species than those of the narrow screen. There are two
primary explanations for the wider coverage. First, the survey data include areas where fish
were introduced, while the Audubon Guide only clearly defines natural fish habitat or long-
established presence of a species. Even if only one of the fish species in a set was actually
caught hi the state, the state would be designated a habitat for all species in the set. For this
reason, the survey-based data are used as the basis for designated a state as habitat only if two
or more species in the fish set are present. This approach is adopted to minimize the chance of
overestimating actual fish coverage. Under this criterion all of the cool-water fish and most of
the warm-water data were used.
The wide screens were incorporated into the thermal model structure and runs were
produced for four equilibrium (GF1, Gil, OS 1, and UK1) and two transient (TR1 and TR2)
climate change models. The results from these sensitivity analyses are discussed in the following
sections. The discussion emphasizes the way in which these modifications affect the output of
the economic model.
4-17
-------
CHAPTER 4
4.5.1 Equilibrium models
Exhibit 4-8 presents the results from the fish screen sensitivity analysis using the primary
fishing-day value specification for the four equilibrium models (GF1, Gil, OS1, and UK1).
Changes hi acres and changes hi dollar value are presented by thermal category for both the
narrow and wide screens. Several interesting changes occur when the wider screens are
adopted. In particular, the mixed results in terms of signs (damages and benefits) that are
observed when the equilibrium models use the narrow screen disappear when the wide screen
is used. The climate scenarios consistently predict damages across all models.
The wide screen increases the number of best-use cool- and warm-water acres in the
baseline model. Greater damages in the wide scenario are the result of larger shifts in best-use
cool-water acres to warm-water acres relative to the narrow scenario. The value associated with
the loss of cool-water fishing days outweighs that of the gain hi warm-water fishing days. In
all of the model runs, the switch from the narrow to wide screen has little impact on the changes
in best-use cold-water acreage. Changes in best-use cool-water acreage are positive for all of
the model runs hi the primary narrow configuration. When the wide screen is adopted, the
changes hi best-use cool-water acreage fall for all models relative to the primary narrow
configuration. In the case of the Gil and OS1 models, these decreases actually result hi
reductions hi cool-water acreage relative to the baseline. Best-use warm-water acreage increases
are larger for the wide screen specifications than those for the narrow specification across all
of the models. It is important to note that the wide screen, as constructed, has the drawback of
underestimating initial best-use cold-water acres since the ranges for salmon and trout were not
extended beyond the "natural" ranges indicated hi the Audubon guide. Consequently, the
criterion of requiring two or more species to be observed hi the 1985 survey data hi order to
designate additional habitat may be too strict. The inclusion of additional best-use cold acres
could result hi higher damages than presented hi this section. Higher damages would occur if
temperature increases hi the model shifted the additional best-use cold-water acres to a lower-
valued best use.
Using the wide screen, all four models generate damages. As was the case using the
narrow screen, the OS1 and UK1 models predict damages under the wide screen but they have
increased by at least a factor of five (from $95 million and $85 million respectively to $574
million and $500 million). While these models predict the largest damages, results from the
GF1 and Gil models experience the largest changes hi the switch from the narrow to the wide
screen. Benefits of $80 million and $81 million are predicted for the GF1 and Gil models,
respectively, under the narrow screen, but damages of $443 million and $451 million are
estimated under the wide screen.
Relative to the primary narrow screen specification, the Gil and GF1 models yield the
higher changes ($531 million, $524 million) as the dollar values decrease from benefits of
approximately $80 and $81 million to damages of $451 and $443 million respectively. The OS1
4-18
-------
S
•
S
£
0)
Is
b
O
•<
.S
S
tr?
ON
*—4
4
-------
«
I
S-
s1
I
LI
s
o
c3
s
2
"o
o ,g
•S
o
I
P
8"
CO
S3
§
1
Sj
vo"
8
t-
l-H
VO
c.
9
g
&
.
-------
CHAPTER 4
($479 million) and UK1 ($415 million) models also show large reductions relative to the dollar
value estimated for the primary narrow specification.
4.5.2 Transient Models
Exhibit 4-8 presents the results from the fish screen sensitivity analysis using the primary
fishing-day value specification for the two transient models (TR1 and TR2). Changes in acres
and changes in dollar value are presented by thermal category for both the narrow and wide
screens. For both the transient models, the changes hi dollar values switch from damages
estimates using the narrow screen to benefits estimates using the wide screen.
For both the transient models, the wide screen specification yields higher losses in best-
use cold acreage relative to the narrow screen specification. Comparing best-use cool and warm
acreage changes for the narrow and wide screens, best-use cool acres increase for the TR1
model and decrease for the TR2 model using the wide screen and best-use warm-water acreage
increases for both the TR1 model and the TR2 model. Changes in best-use rough water acres
switched from increases of 360,728 and 810,271 acres with the narrow screen to no change for
TR1 and TR2 models with the wide screen.
The TR1 model is the less volatile model of the two transient models because it predicts
smaller expected impacts in the near future. In contrast, the TR2 amplifies the additional
acreage included in the wide screen predicting larger effects over time. The wide screen has
more initial warm-water acreage, making warm-water acreage a better substitute for cool and
cold-water acreage than it is under the narrow screen. This leads to more benefits as the
temperature increases in the TR2 model. The effect is noticeable throughout most of the
categories shown in Exhibit 4-8 aside from the cool category where the ratio of the changes is
close for both models.
With the wide screen, the TR1 model shows significantly lower benefits ($73 million)
than the TR2 model ($1,003 million). Relative to the primary narrow specification results, the
TR1 model reveals an increase in value of $393 million and the TR2 model exhibits an increase
in value of $1,269 million using the wide screen.
The results using the high and low fishing-day value specifications for the narrow and
wide screens are presented for the equilibrium and transient models hi Exhibit 4-9 and Exhibit
4-10. These exhibits are presented for comparison purposes only.
4-21
-------
~s
1
1
iwo £
€ Q
^b 5fr*
" o
£ £
3|
5*5
:*||
Id 8
d ^
t« IB"
1 1
II
T3
d
a
|
1
^
OJ
a
f
I
^
1
U
•73
"3
U
i
(Damages)
or
Benefits
(Damages)
or
Benefits
I!
(Damages)
or
Benefits
a> jg
(Damages)
or
Benefits
« %
II
(Damages)
or
Benefits
So ^
n a
goS
«* ^"^
f«u g
4s
W .S
1 v
•8
P
I
oc
1
1
o
*— (
VO
en
00
p
1
en
S
oo
1
VO
£>
«
n
>n
(S,
E.
U)
en
oc
0
NO
£
o
o
8
$
OO
en^
f-4
O
2
oo"
m
S
en
&
^
3
a.
§
I
i
m
vo"
I
I
I
ON
OO
§
90
£
s
«»
^
i
^
•H
3
1
eo
vo"
i
o
i
a
g
1
|
1
§
<*
o\
00
n
^2,
8
a
i
1
I
$
"i
oo"
o
00
VO
g
f— (
1
»
o*
S
i-H
n
e
8
o\
o\
8.
vo"
S
s
o
8
2
1
rJ
I
vo"
'f,
o
U
§
1
e
o
£
I
1
I
VO
vo"
S
I
Ov
3
i
i
Tf
a
oo
00
S
ss
E
en
en
P
vo"
S
o
«
8
VO
5?
1
es"
o
1
3"
b
a
VO"
en
a
en,
2
-i
T3
1
I
vo"
en
8
S
en
I
o\
S
r-4
a-
i
g
s»
^
i
ci
«
•M
g
«
PI
P
VO
en
O
Tl'
OO
OO*
vt
&
1
1
i
i
<£
S'
3
*-H
g
1
1
en_
1
1
*— «
00
l-H
Ov
O
1
OO
i— t
OO
g
&
m
*— <
n
|
q
1
en
p.
vo"
1
O
8"
en
of
§
8
M
S
is
VO
n.
g
^
!
i
0
§
Q
•3
, cool ($29.08), «
1
•o
8
£1
1
3
8-
o
i
1
S
$
%
-------
3
S
o>
I
.a
o
o
O .S
2
"o
I s
a
o
w
R
o
8
.
I
vb"
»H -
fa s
s
K
v>
S"
s
f '
a
d
5
s?
St
VI
vo"
1
00
s
c5
e
1
SI
2
00
t-^
s
P
vo"
0\
s
a
l
aj
*^
en
•4-
-------
I
1
fc
f
^
i
*
1
0
•a
"o
U
(Damages)
or
Benefits
H a
SP ynj
S ® fi
& »
»fi
0 a
1,1
S ® c
5 55
a M
1 |£
CA
•3
o
^
*-»
S
b
1
1
i_1
SO
\o
§
$
S
CO
S
S
S
^p
§
§.
S1
co^
1
^G'
S
*«}•
^
*
81
VO
1
vo"
OO
0
OO
*5
£
o
•i-
r~-
"1,
I
S
m_
oo"
OO
OO
°x
•^
S
o
Ov
tn
^<,
O
I
1
•>
to^
VO
S
o
00
g
— '
___
5>
?— (
,_,
IN
en
52-
&
O
II
S
!
CO
CO
1-i
o
g
IS
CO
*o
^
I
CO
Vi
ob"
^
s-x
S3
VD
^H
^
S2-
II
E
-------
CHAPTER 4
4.6 WARM-WATER FISHING BEHAVIOR
The economic model uniformly treats and values additions (reductions) in recreational
fishing days within each of the acreage categories as new or expanded (decreased) recreational
opportunities. The economic model takes the baseline division of best-use acreage and then
modifies this distribution according to the predicted thermal dynamics. The modifications are
also shaped by the assumption that cold-water fishing is preferred to cool-water fishing which
is preferred to warm-water fishing and so on. This chain of logic drives the designation of best-
use acreage, and it is important to acknowledge that all best-use acreage designations are
mutually exclusive. Thus, any changes from one acreage category to another are treated as
transfers in recreational opportunities. In doing so, it is possible that the impacts on recreational
fishing behavior are overstated by the economic model. This sensitivity analysis addresses the
potential for the model to overestimate the changes in warm-water fishing days established under
the various climate change scenarios.
The potential for overcounting follows from the best-use designation of the model where
it is always assumed that cold-water and cool-water fishing are preferred to warm-water fishing.
It is evident that in many states these two types of fishing coincide and that waters of states
jointly provide opportunities for both activities. In short, the types of recreational fishing
service flows supplied by acreage may not always be mutually exclusive. In states where waters
jointly provide opportunities, the economic model will tend to give more weight to the
recreational service flows provided by the cold-water acreage than those provided by the warm-
water acreage by virtue of the best-use designation process. When the model shifts acreage from
the cold-water and cool-water best-use acreage categories to the warm-water category, increases
in the number of warm-water fishing days result, and in turn there are valued by the model as
changes in opportunities or new fishing days. In cases where warm-water fishing is already
common place, such changes may not truly reflect the extent of new recreational opportunities
but rather may overestimate the availability of such opportunities.
Using activity day data from the U.S. Department of Interior (1988) 1985 Survey of
Fishing, Hunting, and Wildlife Associated Recreation Survey, the South Atlantic and Gulf Coast
Region was identified as an area where overcounting might potentially occur. The U.S.
Department of Interior (1993) has published a 1991 Survey of Fishing, Hunting, and Wildlife
Associated Recreation. However, this more recent survey does not break out fishing days by
activity type by state. The 1985 Survey provides estimates of the number of person-days
associated with cold-water, cool-water, warm-water, anadromous, warm-water, and saltwater
fishing. The states comprising the region include: Alabama, Arkansas, Florida, Georgia,
Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, and Texas. In each of these
states warm-water fishing days were the dominant type of fishing activity reported. For
example, the percentage of freshwater fishing days devoted to warm-water fishing in this region
range from a low of 56 percent in Tennessee to a high of 86 percent in Mississippi.
4-25
-------
CHAPTER 4
To test the sensitivity of the economic modelling results with regards to the treatment of
warm-water fishing days, the function that calculates the change in warm-water fishing days was
modified. This adaptation of the model was based on the assumption that no increases in warm-
water fishing days will occur hi the South Atlantic and Gulf Coast Regions due to global climate
change. It is important to note that this is an extreme assumption, for it is likely that some new
warm-water fishing days will result. New warm-water days are likely to result as some anglers
substitute for lost cold-water and cool-water fishing opportunities. The adapted model was run
for four equilibrium scenarios (GF1, Gil, OS1, and UK1) and two transient scenarios (TR1 and
TR2). The output from these modified runs using the primary, high, and low fishing day value
specifications appears in Exhibits 4-11, 4-12, and 4-13, respectively.
The effects of these changes are illustrated in the results presented in these exhibits. By
limiting the extent of changes in the South Atlantic and Gulf Coast region, predicted changes in
warm-water fishing days fall across all scenarios. In turn, the predicted benefits (damages) fall
(increase) or stay the same in size. For the primary specification, the changes in total dollar
value of recreational fishing days range from no change in the Gil and TR2 scenarios to
reductions of approximately $15 million in the GF1, OS1, and UK1 scenarios. Similar results
are exhibited in the high and low fishing-day value specification tables where no changes in total
value result in the Gil and TR2 scenarios and the higher decreases in total dollar value occur
in the UK1 (17,16), GF1 (15,15), and OS1 (15,14) scenarios.
Exhibit 4-11
Sensitivity Analysis of the Model's Treatment of Warm-Water Fishing Days
Scenario
Warm Days
(1,000s)
Total
Dollar
Value
(Millions)
Primary Specification Changes
GF1
Gil
OS1
UK1
47,488
31,064
57,337
59,338
81
80
(95)
(85)
Warm Days
(1,000s)
Total
Dollar
Value
(Millions)
Change in
Total
Dollar
Value
(Millions)
Modified Specification Changes
46,985
31,064
56,827
58,784
67
80
(110)
(101)
14
0
15
16
• . •• • ' .•'"":" •'•."''".'•• '.'••'•!'::;"-;i'i-';'-''j;: ••••••«••••••• ;••••: : -f ::~ :•,•;,•[;.:.:>:» :::AV'-:*VX^:f:.:,l::: V ;t!
TR1
TR2
50,441
61,621
(320)
(266)
50,189
61,621
(327)
(266)
7
0
4-26
-------
CHAPTER 4
Exhibit 4-12
Sensitivity Analysis of the Model's Treatment of Warm-Water Fishing Days
Scenario
Warm Days
(1,000s)
Total
Dollar
Value
(Millions)
High Specification Changes
GF1
Gil
OS1
UK1
47,488
31,064
57,337
59,338
(191)
(136)
(335)
(393)
Warm Days
(1,000s)
Total
Dollar
Value
(Millions)
Change in
Total
Dollar
Value
(Millions)
Modified Specification Changes
46,985
31,064
56,827
58,784
(206)
(136)
(350)
(409)
15
0
15
16
TR1
TR2
50,441
61,621
(480)
(602)
50,189
61,621
(487)
(602)
7
0
Exhibit 4-13
Sensitivity Analysis of the Model's Treatment of Warm-Water Fishing Days
Scenario
Warm Days
(1,000s)
Total
Dollar
Value
(Millions)
Low Specification Changes
GF1
Gil
OS1
UK1
47,488
31,064
57,337
59,338
38
(26)
57
(50)
Warm Days
(1,000s)
Total
Dollar
Value
(Millions)
Change in
Total
Dollar
Value
(Millions)
Modified Specification Changes
46,985
31,064
56,827
58,784
23
(26)
43
(67)
15
0
14
17
, -
TR1
TR2
50,441
61,621
(49)
(232)
50,189
61,621
(56)
(232)
7
0
4-27
-------
CHAPTER 4
4.7 COLD-WATER SUBSTITUTABILrrY
The economic model currently assumes costless transitions across all categories of best-
use fishing acreage. In truth, natural conditions may not permit such transitions to occur without
costs such as time delays or species shifts. The purpose of this sensitivity analysis is to assess
the effects of altering the costless transition assumption. Specifically, this sensitivity analysis
addresses the transfer of cold-water acreage to cool, warm, and rough best-use fishing acreage.
In practice, the evaluation of cold-water acreage substitutability takes quite an extreme posture.
It is assumed that best-use cold-water acreage losses due to thermal changes are effectively lost
recreational service flows. In other words, these acreage losses are not transferable in any way.
The model is adapted to effectively remove these acres from the analysis rather than to nKwe
these acres to the best-use cool-water, warm-water, or rough acreage categories.
To test the sensitivity of the economic modelling results with regards to the treatment of
cold-water acreage, the model was revised so that any changes from baseline best-use cold-water
acreage that resulted from the global climate change scenarios are shifted to an acreage category
specified as the none category. This category is termed the none category, for it is assumed that
no recreational fishing service flows are provided by this acreage. For states with no baseline
best-use cold fishing acreage, the modelling functions were not altered and the scenarios were
run as usual. For states with baseline best-use cold acreage, it was assumed that no changes
from baseline occurred in the cool-water, warm-water, and rough acreage categories. This
assumption was necessary because of the difficulty of disentangling the determinants of flows
of acreage from one category to another. These adaptations of the model were based on the
assumption that cold-water acreage would not readily move to the other categories of fishing
acreage. This is an extreme assumption, for it is not likely that all these a,cres would be entirely
lost. In addition, it is also likely that shifts from other types of acreage categories would also
be limited by natural and other forces. The adapted model was run for four equilibrium
scenarios (GF1, Gil, OS1, and UK1) and two transient scenarios (TR1 and TR2). The output
from these modified runs using the primary, high, and low fishing-day value specification appear
in Exhibits 4-14, 4-15, and 4-16, respectively.
The effects of these changes to the model are made clear in the results presented in these
exhibits. By eliminating the transfer of best-use cold acreage to other best-use acreage
designations, the damages of global climate change are heightened. Losses in best-use cold-
water acreage and their associated recreational service flows are not transferable to other acreage
designations thereby eroding the possibility for substitution across fishing activities by anglers.
In turn, the predicted benefits (damages) fall (increase) across all of the scenarios. In the
primary specification, the decreases in total dollar value of recreational fishing days range from
approximately $508 million in the TR1 scenario to $1,044 million in the UK1 and TR2
scenarios. Comparing the magnitude of the results of the primary and modified scenarios, the
influence of the reductions in the number of cool-water fishing days is evident. It appears that
in many of the scenarios cold-water fishing acreage is transferred to the cool-water acreage
category rather than to the other categories of fishing acreage. These cool-water fishing days
4-28
-------
CHAPTER 4
are highly valued by the model and this is reflected in the changes in dollar value of the primary
specification scenarios. By eliminating the occurrence of such shifts, the substitution possibilities
across fishing activities are markedly limited and the resulting damages measured in fishing-day
values are accentuated. Similar patterns surface in the output from the high and low
specifications. For the high specification, the decreases in the total dollar value of recreational
fishing days ranges from approximately $468 million in the TR1 scenario to $970 million in the
UK1 scenario. Whereas for the low specification, decreases in the total dollar value of
recreational fishing days range from approximately $167 million in the TR1 scenario to $346
million in the UK1 scenario.
4.8 POTENTIAL IMPACTS ON RECREATIONAL FISHING FROM CHANGES IN
RUNOFF
Climate change can have a significant impact on runoff due to changes in precipitation patterns
and watershed characteristics. In this section, the effects of climate change on runoff are
characterized. Then, potential economic losses on recreational fishing due to changes in runoff
are evaluated. This section is included because it addresses the uncertainties associated with
modelling recreational fishing responses under climate change scenarios. In contrast to the other
analyses presented in this chapter, no modifications were made to the model structure nor were
results generated for scenarios using alternative specifications. Rather, the discussion presented
here is for illustrative purposes and seeks to emphasize the significance of runoff assumptions
when modelling the effects of climate change scenarios.
4.8.1 Climate Change Effects on Runoff
Quantification of the effects of climate change on stream runoff for the entire United States is
difficult. We performed a literature review on this subject to quantify this effect for the
different physiographic regions (see Exhibit 4-17). However, as shown in Exhibit 4-17, wide
ranges of increases and decreases in runoff due to climate change are documented. This result
is hi part due to the use of different modelling assumptions in the climate change models, as well
as the geographical location and site characteristics of each study. Nevertheless, based on the
current state-of-the-art in runoff modelling of climate change effects, the following can be
concluded from these studies:
4-29
-------
es
1
&
1
i
oo
ON
p-
oo
•>
oo
r-
oo
CO
oo
00
5*
oo
s
PH
o
a\
G\
\0
ON
en
00
w
00
-
o
o
en
3
en
00
en
CM
Cf\
en
«n
oo
CM
CM
S5
en
ON
oo
o
en
en
VO
es
CM
ON
en
en
ON
|
oo
oo
oo
oo
en
en
r-
o\
«n
CM
CM
OO
o
CM
00
00^
oo"
oo
CM
•>
>T5
vo
«s
<3\
en
^
<3\
P*
••
en
en
oo
'ON
en
F^
•\
oo
vo
«n
'S
VO
ON
OO
CM
-------
•S -s
a «
rt ^^ 5 ® •*•
lilil
ON
i— 1
ON
3
3
00
1
oo
CO
«n
ON
p—i CO
0 >>
o n! •>
UPC
Si
£
n
00
en
00
oo
cs
vo
oo
cs~
CO
»n
»— <
oo"
ON
o
en
CM
S5
co
CM
ON
CO
8
co
CO
CO
cs
ON
o
cs
>n
•\
oo
ON
wo
ON
es
oo
oo
1— (
cs
ON
OO
ON
co
T—<
OO
»—I
§
o
CO
co
00
oo
ON
•0
00
oo
5?
vo
O
CO
co
oo
CO
CO
«v
ON
its
u
1
JS
oo
00
vo
CN
CO
ON
oo
vcT
•I
o
-------
ill
*;
«
ts
Q d-
i—I t» O
a c?^
U Q jf
|
1
cc
U
I
#
i
U
sa
c/a
I
en
CO
ON
oo
oo"
OO
cn
en
rs
t>
OO
OO
00
cs
o
8
CM
-
en
oo
00
«n
CN
>n
00
es"
o
0
vo
o
en
vq^
oo"
00
o
CO
R
OO
CM
CM
S5
en
oo
o
en
en
CN
VI
en
ON
en
en
CM~
ON
O
es
m
«\
00
in
00
oo
en
en^
oC
v~>
t>
r|
CM
en
vo
CM
oo
•s
oo
oo
!S
cs
•\
ON
CN
oC
oo
en
OO
»n
5
en
so
00
-------
CHAPTER 4
• Runoff modelling of climate change scenarios, considering changes in temperature,
precipitation, and basin characteristics, show significant effects on the seasonal amounts of
runoff. For instance, seasonal variations in runoff are more important than annual changes
because of the different fish responses to runoff conditions under different seasons.
• Changes in temperature and rainfall have a significant effect on the seasonal distribution of
runoff. Winter runoff may increase while spring and summer runoff may decrease due to
a decrease in snow accumulations. For instance, Lettenmaier and Gan (1990) found that in
watersheds dominated by spring runoff from snowmelt, such as the Sacramento and San
Joaquin basins, winter runoff increased twice as much, while spring and summer runoff
decreased. This shift in runoff can be more important than the overall annual change in
runoff due to the impact on the available runoff in the summer, which is a critical period for
recreational fishing (Johnson and Adams, 1988). In addition, increases in the frequency of
high flows during spring can have detrimental effects because they can scour spawning areas.
• Most of the studies summarized in Exhibit 4-17 only consider changes in temperature and
precipitation. Most recently, Kite (1993) also considered changes in watershed
characteristics such as vegetation, biomass production, soil processes, erosion, and slope
stability among others. For instance, increases in CO2 can alter the photosynthesis and
transpiration of vegetation, decreasing stomatal conductance and increasing the efficiency of
the plant, making more water available for runoff. Kite (1993) found that under a 2 X CO2
scenario the frequency of high flows increased and the snowpack decreased significantly.
4.8.2 Implications of Runoff Changes in Fish Recreational Values
There are some studies in the literature addressing the recreational value of streamflow for
different regions of the country. Among those, Hansen and Hallam (1991) provided marginal
values per acre-foot of water for recreational fishing. Marginal values were estimated for 99
river subbasins for trout and bass. Exhibit 4-18 shows the interquantile ranges for the marginal
values per acre-foot of water from the Hansen and Hallam (1991) study. Johnson and Adams
(1988) estimated benefits for the recreational fishing of steelhead in the John Day river in
Oregon. A value of $2.36 (in 1987 dollars) for an additional acre-foot of water hi the summer
was estimated. Values of -$0.32 and $0.18 were estimated for the spring and winter
respectively.
4-33
-------
1
a
o
0)
I
&e
4
-------
•
1
0
.9
o\ '53
^TS
a*
*1
JJ 8
O £
« II
Revelle
Revelle
oo
M
i
0)
\r>
OO CO V)
•*•*•*•*•*
1
60
CO
CO
.2
+
foo
T
ON
«S
S
I
^ .s
o S
•c
-------
.a
s>g8
fs
ll
1
<2
00
£
t.
S o
+ +
o
^s
il
B&
m
3
'So
£
•-
O
M
<-*
O G-O
o
VO
-------
in
McCabe & Ay
etal., 1992)
Is
Note:
y in
of b
change, especi
northern porti
TO
- ON
da f
8 S
1 a
Z U
o?
1
*
00
en
cs
o
VO
I
o o o
<» r
S3 E
S S
CCCGC
2*CO2(N
Stomatal
n
8
§
1
i
.a
I
8
1
Q
u
.a
I
a
-------
-------
CHAPTER 4
Exhibit 4-18
Marginal Values Per Acre-Foot in 1980 dollars
Quantile
25%
50% (median)
75%
Trout
1.54
3.45
10.50
Bass
1.35
3.13
7.14
4.8.3 Economic Implications of Runoff Changes
Accurate estimation of the economic loss due to changes in runoff would require
estimates of seasonal changes in runoff for different climate change scenarios by subbasin for
the entire United States. In addition, baseline runoff volumes by basin would be needed for
each basin. However, as pointed out before, magnitudes of the runoff changes are not
available for the entire nation. In addition, available basin-specific studies, as summarized in
Exhibit 4-17, showed a wide range of effects which are mainly due to the climate change
model used and the underlying assumptions. Nevertheless, to provide a crude estimate of
potential economic losses due to changes in runoff, some simplified assumptions can be
made. Rough estimates of volume of annual runoff can be obtained for the entire nation
(USGS, 1989). Then, potential runoff changes can be assumed along with marginal values
per acre-foot lost to obtain potential economic losses due to changes in runoff.
To estimate economic losses, the following assumptions were adopted:
• a weighted average runoff depth (i.e. the volume of water covering a drainage area to a
depth expressed hi inches) of 15 niches per year for the entire U.S.,
• a range of 20 percent to 50 percent reduction in runoff during the summer with one
quarter of the annual runoff assigned to this period, and
• an economic loss of $3 per acre-foot of water.
The resulting estimated economic loss for recreational fishing could range from $0.4 to $1.0
billion per year. This crude estimate includes many simplifying assumptions and is given
only for illustrative purposes. Nevertheless, it is important to point out that the effect of
runoff changes due to climate change is an important issue for further consideration. The
general conclusion at this point is that climate change could affect seasonal runoff and that
decreases hi runoff during the summer could lead to significant economic losses in
recreational fishing.
4-39
-------
CHAPTER 4
4.9 SUMMARY AND CONCLUSION
Chapter 4 presents discussions of seven analyses of the model. These discussions
emphasize the role and/or significance of different assumptions and methods in modelling the
ecological impacts from climate change. The topics examined include fishing-day values
(4.2), climate sensitivity and emissions scenarios (4.3), fish thermal tolerances (4.4), fish
habitat designations (4.5), warm-water fishing behavior (4.6), cold-water substitutability
(4.7), and runoff (4.8). Within this set of topics, assumptions regarding both ecologic and
economic issues are addressed and various pathways of effects on the output of the economic
model are explored.
It is difficult to compare and contrast the findings of the different sections, for the
proposed modifications to the modelling assumptions vary in nature as do the uncertainties
associated with the modified specifications. In particular, the effects of altering ecologic and
economic assumptions are likely to have markedly different influences on the output of the
model. The ecologic changes are often reflected in shifts in best-use acreage, whereas the
economic changes can directly affect participation probabilities or fishing-day values. For
some analyses, the relevant sensitivity indicators might be acres, and for others, it might be
fishing days or changes in fishing-day values. In some instances, it might be useful to
compare the relative variation in size of the different model estimates. In other cases, it
might be of more value to compare the absolute variation in size or the variation in signs of
the estimates.
For simplicity, Exhibit 4-19 presents the results from six of the seven different analyses
in terms of the estimated relative changes in recreational fishing value. These six use the
thermal and economic framework based upon Vaughan and Russell. In this context, relative
change refers to the difference between the sensitivity analysis estimate and the primary
specification estimate of the total dollar value of the ecological change. The relative changes
provide some measure of the sensitivity of the primary specification results to the modified
assumption of interest. It is important to note that these numbers do not reveal the
percentage of the change in value. The total changes in dollar estimates are also presented
for review. These numbers should provide some understanding of the magnitude of the shifts
in value. By comparing the total changes in dollars for the different scenarios with the
primary specification dollar estimates shown in the top portion of the table, it is possible to
check the calculation of the relative change and to discover the nature of the transition.
Exhibit 4-19 displays the results of the various analyses by global climate modelling scenario.
The different specifications examined within the analyses are noted.
The single largest relative change is estimated for the fish habitat designation. The
adoption of the wide screen for the TR2 climate scenario results in an increase in value of
$1,269 million. Other large relative changes appear in the cold-water substitution analysis
with two climate scenarios (TR2 and UK1) showing changes of approximately $1,044
million. Several of the specifications modeled in the cold-water substitution and fish habitat
4-40
-------
CHAPTER 4
designation analyses reveal relative changes upwards of $500 million. The climate and
emissions scenario and the fish thermal tolerance sensitivity analyses show the next greatest
levels of sensitivity in terms of relative value changes with most changes falling above $200
million. The fishing-day value sensitivity analysis follows with most changes above $100
million. The warm-water fishing behavior analysis results in the smallest relative changes,
with changes ranging from $0 to $17 million. The range of total dollar values across all
specifications is very large when compared to the range given on the primary specifications.
The later ranged from -$320 million to $81 million but these alternative specifications range
from -$1,6 billion to $1 billion in annual impacts.
The runoff impacts presented in Section 4.8 should be considered separately since they
were developed outside of the thermal and economic framework used in most of this study.
While the estimated runoff impacts are speculative, their large magnitude ($0.4 to 1.0 billion
per year) suggests that runoff changes could be a significant, additional element in the impact
of climate change on recreational fishing.
In closing, it is important to note that the interpretation of the results from the primary
and alternative specifications is critical. The results may indeed be ambiguous since both
large damages ($1.6 billion per year) and large benefits ($1 billion per year) can be derived
from alternative specifications but it would be a mistake to conclude that "on average" these
results cancel and there is no impact. On the contrary, this study suggests that both are
possible. Nonetheless, taking the thermal and runoff changes together may tilt the judgment
to be made from this study more toward substantial damages. In sum, while the study could
appropriately be interpreted as inconclusive about the size and direction of the impacts of
climate change on recreational fishing, the study does indicate the possibility of substantial
economic damages. In this sense, substantial economic damages to recreational fishing are a
contingency. Determining what weight to attach to this contingency given limited
information and large uncertainties is a difficult choice for public policymakers. At a
minimum, the body of evidence and analysis compiled in this report indicate that this
contingency should not be dismissed.
4-41
-------
o> -<
U
O
e.
a
1
£
1
3
o
H
B
a
c-
a
c-
in1
en
ss
fa
a
g
Ci
1
8
0
§
1
I
2
(M
•s
I
A >
I
CQ i3
II
6?
Q.
a
-
I
low
to
J "
P
s
a
I
l
ll
_ rt ctf
-------
a
R
O
O
8
g
g
1
a
a
a
•^~
G.
I
E?
•a
>
•a
s?
a
s
a-
a.
a
hing
havio
W
Fi
Cold-water
Acreage
Substitution3
« g "3 §
•S i 4 a
?«.•««>*=;
o .B
Illf
•g
2 r
-------
CHAPTER 4
REFERENCES
Ayers M., D. Wolock, G. McCabe, L. Hay, G. Tasker. 1993. Sensitivity of Water
Resources in the Delaware River Basin to Climate Variability and Change. U.S.
Geological Survey. Open-File Report 92-52.
Boschung, H.T., J.D. Williams, D.W. Gotshall, D.K. Caldwell, and M.C.Caldwell. 1983.
The Audubon Society: Field Guide to North American Fishes, Whales, and Dolphins.
Alfred A. Knopf, New York.
Carpenter S.R., S.G. Fisher, N.B. Grimm, J.F. Kitchell. 1992. Global Change and
Freshwater Ecosystem. Annual Rev. Ecol. Syst. 23:119-39.
Charbonneau, J.J., and M.J. Hay. 1978. Determinants and Economic Values of Hunting and
Fishing. Transactions of the North American Wildlife and Natural Resources Conference
43: 391-403.
Cohen, SJ. 1986. Impacts of CO2-induced climatic change on water resources in the Great
Lakes Region. Clim. Change 8:135-53.
Flaschka I.M., C.W. Stockton, and W.R. Boggess. 1987. Climatic variation and surface
water resources in the Great Basin Region. Water Res. Bull. 23:45-57.
Frederick K.D., P.H. 1988. Greenhouse Warming: Abatement and Adaptation.
Proceedings of a Workshop held in Washington, DC. June 14-15. Edited by Rosenberg
NJ, Easterling m WE, Crosson PR, Darmstadter J.
Freeman, A.M. 1993b. The Economics of Valuing Marine Recreation: A Review of
Empirical Evidence. Economics Working Paper 93-102, Department of Economics,
Bowdoin College, Brunswick, Maine.
Gleick, P.H. 1987. Regional hydrologic consequences of increases in atmospheric CO2 and
other trace gases. Climatic Change 10:137-161.
Hansen, L.T. and A. Hallam, 1991. National Estimates of the Recreational Value of
Streamflow. Water Resources Research, Vol 27: No 2. pp 167-175.
Johnson, N.S. and R.M Adams, 1988. Benefits of Increased Streamflow: The Case of the
John Day River Steelhead Fishery. Water Resources Research, Vol 24: No 11. pp 1839-
1846.
4-44
-------
CHAPTER 4
Kite, G.W. 1993. Application of a Land Class Hydrological Model to Climatic Change.
Water Resources Research. Vol.29. No.7. pp 2377-2384.
Klings C.L. 1988. Comparing Welfare estimates of Environmental Quality Changes from
Recreation Demand Models. Journal of Environmental Economics and Management
15:331-340.
Leary, N. 1994. Memorandum to ICF, Inc. April 13.
Lettenmaier, D.P. and T. Y. Gan. 1990. Hydrologic sensitivities of the Sacramento- San
Joaquin River Basin, California, to global warming. Water Resources Res. 26:69-86.
Loomis, J., D.M. Donnelly, C. Sorg, and L. Nelson. 1985. Net Economic Value of
Recreational Steelhead Fishing in Idaho. U.S. Department of Agriculture Forest Service,
Rocky Mountain Forest and Range Experimental Station, Fort Collins, CO.
Nash L.L. and P.H. Gleick . 1993. The Colorado River Basin and Climatic Change. The
Sensitivity of Streamflow and Water Supply To Variations in Temperature and
Precipitation. EPA230-R-93-009.
Poff, N. LeRoy. Regional Hydrologic Response to Climate Change: An Ecological
Perspective. In Global Climate Change and Freshwater Ecosystems. Eds. Firth, P.L. and
Fisher, S.G. Springer-Verlag, NY. 1992.
Revelle, R.R. and P.E.Waggoner. 1983. Effects of a Carbon Dioxide-Induced Climatic
Change on Water Supplies in the Western United States. In Changing Climate. National
Academy of Sciences, National Academy Press, Washington, DC..
Smith, V.K., and Y. Kaoru. 1990. Signals of Noise? Explaining the Variation in
Recreation Benefits Estimates. American Journal of Agricultural Economics 72(2): 419-
433.
U.S. Department of the Interior, Fish and Wildlife Service. 1993. 1991 National Survey of
Fishing, Hunting, and Wildlife-Associated Recreation. March.
U.S. Department of the Interior, Fish and Wildlife Service. 1988. 1985 National Survey of
Fishing, Hunting, and Wildlife-Associated Recreation. November.
USGS, 1989. Average Annual Runoff in the United States (1951-1980). Hydrologic
Investigations Atlas HA-710.
Walsh, Richard G., Johnson, D.M., and J.R. McKean. 1992. Benefit Transfer of Outdoor
Recreation Demand Studies. 1968-1988. Water Resources Research 28(3): 707-713.
4-45
-------
CHAPTER 4
Walsh, Richard G., Johnson, D.M., and J.R. McKean. 1990. Nonmarket Values from Two
Decades of Research in Recreation Demand, in Advances in Applied Micro-Economics
Volume 5, 167-193.
4-46
-------
APPENDICES A THROUGH F
-------
-------
APPENDIX A
FISH DESCRIPTIONS
-------
-------
Coldwater Fish
Pink Salmon (Oncorhynchus gorbuscha)
Pink Salmon are up to 3'3" in length, blue-green in color above, and silvery-white below. Has
large dark ovals spots on back. Pink Salmon are found in waters inshore usually at mid-depth
or near the surface. They spawn upstream sometimes far inland and are valued as game fish.
Inhabits California, Idaho, Michigan, Minnesota, Oregon, and Washington. Related to the Coho
Salmon. (Boschung et al., 389)
Chum Salmon (Oncorhynchus keta)
The Marine variety of the Chum are up to 33" in length, blue-green above, and silvery below.
The freshwater variety are bright red with a pale green head. Females are sometimes
characterized by green and yellow blotches. The Chum occurs in freshwater streams, rivers,
and lakes that contain tributary systems for spawning. Chum inhabit California, Washington,
and Oregon. (Boschung et al., 390)
Cutthroat Trout (Salmo clarki)
Cutthroat Trout are up to 30" in length and 41 Ibs in weight. They are characterized by a dark
olive back, variable color sides from silvery to yellow-orange, and a lighter belly. Cutthroat
Trout are valued as a game fish by fisherman. The Cutthroat is common to inshore marine
waters; lakes; and coastal, inland, and alpine streams. It can be found in Washington,
California, Oregon, Montana, Idaho, Wyoming, Utah, Colorado, New Mexico, Arizona, and
Nevada. (Boschung et al., 393)
Mountain Whitefish (Prosopium williamsoni)
The Mountain Whitefish weighs up to four and one half pounds and can grow to twenty-two
inches in length. Greenish to blue-gray on back and silvery on sides. The Mountain Whitefish
inhabits lakes and streams in Washington, Oregon, California, and Idaho. (Boschung et al.,
392)
Coho Salmon (Oncorhynchus kisutch)
The Coho can grow to a length of three feet and three inches, is blue-green with irregular dark
spots on the back and silvery-white below. They are a highly prized game fish. Coho inhabit
inshore waters and spawn in coastal streams. Coho are found hi Washington, Oregon,
California, and Idaho. (Boschung et al., 389)
Rainbow Trout (Salmo gairdneri)
Rainbow Trout can be 3'9" in length and just over forty-two pounds. They are characterized
A-l
-------
by a metallic-blue coloring above, silvery-white coloring below, and small, black spots on the
back and sides. The freshwater variety have a distinctive red band on the side and more
prominent spots. Rainbow Trout are a valued game fish. Rainbow Trout inhabit lakes and
rivers in Arizona, California, Colorado, Connecticut, Idaho, Indiana, Maine, Massachusetts,
Michigan, Missouri, Nevada, New Hampshire, New York, North Dakota, Ohio, Oklahoma,
Oregon, Pennsylvania, Rhode Island, Utah, Vermont, Washington, Wisconsin, and Wyoming.
(Boschung et al., 394)
Chinook Salmon (Oncorhynchus tshawytscha)
The Chinook Salmon grows to a maximum of 4'10" in length and 126 pounds. The Chinook
has a greenish-blue to black above with oblong, black spots, and is silvery white below. The
freshwater variety is very dark overall. Chinook Salmon are very highly prized game fish in
northern California. The Chinook inhabits freshwater streams in California, Washington,
Oregon, and Idaho (Boschung et al., 391)
Brook Trout (Salvelinus fontinalis)
Brook Trout are a maximum of twenty-one inches and fourteen and a half pounds. The marine
coloration is a bluish-green back, becoming silvery on the side with a white belly. The
freshwater variety has red or yellowish tint on back and sides and red spots in blue halos on
sides. Brook Trout inhabit cool freshwater streams and are found in California, Connecticut,
Delaware, Idaho, Maine, Massachusetts, Michigan, Montana, Nevada, New Hampshire, New
Jersey, New York, Oregon, Pennsylvania, Rhode Island, Utah, Vermont, Virginia, Washington,
West Virginia, Wisconsin, and Wyoming. (Boschung et al., 396)
Brown Trout (Salmo trutta)
Brown Trout may grow to forty inches in length and just over thirty-nine pounds in weight.
Brown Trout have an olive back and sides with red or orange spots, often with a halo, and a
silvery belly. Brown Trout are moderately desired game fish. Brown Trout inhabit high
gradient freshwater streams and are found in Arizona, Arkansas, California, Colorado,
Connecticut, Massachusetts, Minnesota, Missouri, Montana, Nebraska, Nevada, New
Hampshire, New Mexico, New York, Ohio, Oregon, Pennsylvania, Rhode Island, Utah,
Vermont, Virginia, Washington, West Virginia, Wisconsin, and Wyoming. (Boschung et al.,
396)
Coolwater Fish
Pumpkinseed (Lepomis gibbosus)
Pumpkinseeds are dark greenish gold mottled with reddish orange on their backs, have greenish
yellow, mottled orange, and blue-green sides, and a yellow-orange belly. They can grow to 10"
in length and 1 Ib in weight. Pumpkinseeds are found in cool shallow streams, ponds, marshes,
A-2
-------
and lakes with heavy vegetation. Pumpkinseeds are sought by beginner anglers. They are
native to Connecticut, Delaware, Illinois, Indiana, Iowa, Maine, Maryland, Massachusetts,
Michigan, Minnesota, New Hampshire, New Jersey, New York, North Carolina, Ohio,
Pennsylvania, Rhode Island, South Carolina, Vermont, Virginia, West Virginia, and Wisconsin.
(Boschung et al., 554)
Muskellunge (Esox masquinongy)
The Muskellunge can grow to a length and weight of six feet and one hundred pounds. The
back of a Muskellunge is greenish to light brown on the back and sides greenish-gray to silvery
with dark spots or diagonal bars and a creamy-white belly. The "Musky" is sought by anglers.
Inhabits lakes and reservoirs with heavy vegetation and slow, meandering streams and rivers
with heavy plant cover. Muskellunge are native to Illinois, Indiana, Iowa, Kentucky, Michigan,
Minnesota, New Hampshire, New York, Ohio, Pennsylvania, Tennessee, Vermont, Virginia,
West Virginia, and Wisconsin. (Boschung et al., 403)
Northern Pike (Esox lucius)
The Northern Pike has a maximum length and weight of 4'4" and 46 1/8 Ibs. They are
characterized by a dark olive-green to greenish-brown back, lighter sides, and irregular rows of
yellow spots and a small gold spot on the exposed edge of it's scales. The Northern Pike is a
valued game fish and inhabits lakes, reservoirs, and large streams with low current and heavy
vegetation. They are common in Connecticut, Delaware, Illinois, Indiana, Iowa, Massachusetts,
Michigan, Minnesota, Missouri, Montana, Nebraska, New Hampshire, New Jersey, New York,
North Dakota, Ohio, Pennsylvania, Rhode Island, South Dakota, Vermont, and Wisconsin.
(Boschung et al.s 402)
Yellow Perch (Perca flavescens)
The largest Yellow Perch grow to 15" and 4 1/4 Ibs. The Yellow Perch is brassy-green to
golden yellow above with 5-8 dusky bars across it's back almost to the belly. The Yellow Perch
is a sport fish and inhabits open streams, lakes, ponds, and reservoirs with clear water and
aquatic vegetation. The Yellow Perch is found in Alabama, Connecticut, Delaware, Florida,
Illinois, Indiana, Iowa, Maine, Maryland, Massachusetts, Michigan, Minnesota, Missouri,
Montana, Nebraska, New Hampshire, New Jersey, New York, North Carolina, North Dakota,
Ohio, Pennsylvania, Rhode Island, South Carolina, South Dakota, Vermont, Virginia, West
Virginia, and Wisconsin. (Boschung et al., 578)
Walleye (Stizostedion vitreum)
Walleye are up to three feet and five inches hi length and twenty-five pounds. Walleye are
olive-brown to brassy greenish-yellow above with dusky to black mottling. The belly is whitish
with a yellow-green tinge. They are highly sought after game fish. Walleye live in deep waters
of large streams, lakes, and reservoirs. Walleye are found in Alabama, Arkansas, Colorado,
A-3
-------
Illinois, Indiana, Iowa, Kansas, Kentucky, Michigan, Minnesota, Mississippi, Missouri,
Montana, Nebraska, New York, North Dakota, Ohio, Oklahoma, Pennsylvania, South Dakota,
Tennessee, Vermont, West Virginia, Wisconsin, and Wyoming. (Boschung et al., 585)
Warmwater Fish
Rock Bass (Ambloplites rupestris)
Rock Bass can grow to a length and weight of 13" and 3 5/8 Ibs. Rock Bass are
characteristically have an olive mottled back with dark saddles and bronze blotches and are
lighter below with rows of dusky spots. Despite their small size Rock Bass are a popular game
fish. Their habitat is cool, clear, rocky streams and shallow lakes with vegetation or other
cover. Illinois, Indiana, Iowa, Kentucky, Massachusetts, Michigan, Minnesota, Missouri, New
York, North Dakota, Ohio, Pennsylvania, South Dakota, Tennessee, Vermont, Virginia, West
Virginia, and Wisconsin. (Boschung et al., 549)
Black/White Crappie (Pomoxis nigromaculatus/annularis)
Black Crappie (and White) are up to 16" in length and 5 Ibs in weight. Their coloration is
greenish back and silvery green sides with dark green to black scattered mottling. Black and
White Crappie live in quiet warm, clear streams, ponds, lakes, and reservoirs. They are popular
game fish. Black and White Crappie are found in Alabama, Arkansas, Florida, Georgia,
Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maryland, Michigan, Minnesota,
Mississippi, Missouri, Nebraska, New York, North Carolina, North Dakota, Ohio, Oklahoma,
Pennsylvania, South Carolina, South Dakota, Tennessee, Texas, Vermont, Virginia, West
Virginia, and Wisconsin. (Boschung et al., 560)
Smallmouth Bass (Micropterus dolomieui)
Smallmouth Bass can grow to a maximum length and weight of twenty- four inches and twelve
pounds. Smallmouth Bass have a dark olive to brown back, greenish yellow sides, and dark
mottling in the form of midlateral bars. Smallmouth Bass are very popular game fish. They
live in cool, clear streams with moderate flow and in lakes and reservoirs. Smallmouth are
found in Arkansas, Illinois, Indiana, Iowa, Kentucky, Michigan, Minnesota, Missouri, New
York, Ohio, Oklahoma, Pennsylvania, Tennessee, Vermont, West Virginia, and Wisconsin.
(Boschung etal., 557)
Sauger (Stizostedion canadense)
Sauger have a maximum length of 28" and weight of 8 3/4 Ibs. Sauger are gray to dull brown
and brassy to orange on the sides with dark markings and a whitish belly. The Sauger is an
important game fish. Their habitat is dingy waters of large creeks with moderate to swift
currents, lakes, and reservoirs. Alabama, Arkansas, Colorado, Illinois, Indiana, Iowa, Kansas,
Kentucky, Maryland, Michigan, Minnesota, Mississippi, Missouri, Montana, Nebraska, New
A-4
-------
York, North Dakota, Ohio, Oklahoma, Pennsylvania, South Dakota, Tennessee, Vermont,
Virginia, West Virginia, Wisconsin, and Wyoming. (Boschung et al., 584)
Golden Shiner (Notemigonus crysoleucas)
Golden shiners are up to 12" in length. Golden shiners have a golden to olive back, light olive
sides with silvery reflections, and a silvery-yellow belly. Goldens inhabit clear, quiet streams,
lakes, ponds, and swamps over mud, sand, or rocks. They are native from the East United
States west to Montana, Wyoming, Colorado, Nebraska, and Texas . (Boschung et al., 431)
Gizzard Shad (Dorosoma cepedianum)
The Gizzard Shad can grow to sixteen inches in length. They are dark blue or gray, have
silvery sides, and a white belly. Gizzard Shad inhabit freshwater hi large rivers, reservoirs,
lakes, and estuaries. Gizzard Shad can be found in Alabama, Arkansas, Colorado, Delaware,
Florida, Georgia, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maryland, Michigan,
Minnesota, Mississippi, Montana, Nebraska, New Jersey, New Mexico, New York, North
Carolina, Ohio, Oklahoma, Pennsylvania, South Carolina, South Dakota, Tennessee, Texas,
Vermont, Virginia, West Virginia, and Wisconsin. (Boschung et al., 383)
White Bass (Morone chrysops)
White Bass grow to a maximum of 18" in length and 5 1/4 Ibs in weight. They are black-olive
to silvery-gray hi color and have silvery to white sides with 6-9 dark narrow strips. White Bass
are important game fish. The White Bass are found in large streams, lakes, and reservoirs in
moderately clear water. White Bass can be found in Arkansas, Illinois, Indiana, Iowa, Kansas,
Kentucky, Louisiana, Michigan, Minnesota, Mississippi, Missouri, Nebraska, New Mexico,
New York, North Dakota, Ohio, Oklahoma, Pennsylvania, South Dakota, Tennessee, Texas,
Vermont, West Virginia, and Wisconsin. (Boschung et al., 534)
Largemouth Bass (Micropterus salmoides)
Largemouth Bass are up to 3'2" in length and 22 1/4 Ibs in weight at maximum growth.
Largemouth are characterized by an olive to dark green mottled back and greenish yellow sides
with a midlateral stripe. They are highly valued as a sport fish. Largemouth Bass inhabit quiet,
clear streams, ponds, lakes, and reservoirs. They can be found in Alabama, Arkansas, Florida,
Georgia, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Michigan, Minnesota,
Mississippi, Missouri, New York, North Carolina, Ohio, Oklahoma, Pennsylvania, South
Carolina, Tennessee, Texas, West Virginia, and Wisconsin. (Boschung et al., 559)
Bluegill (Lepomis macrochirus)
Bluegill reach a maximum length and weight of twelve niches and four and three quarters
pounds. They are dark olive-green hi color, have lighter sides with brassy reflections or dusky
A-5
-------
bars and have white bellies. They are the most popular game fish in the United States. Bluegill
inhabit clear, warm pools or streams, lakes, ponds, sloughs, and reservoirs, usually in shallow
water with vegetation. Bluegill can be found in Alabama, Arkansas, Florida, Georgia, Illinois,
Indiana, Iowa, Kansas, Kentucky, Louisiana, Michigan, Minnesota, Mississippi, Missouri,
Nebraska, New Mexico, New York, North Carolina, North Dakota, Ohio, Oklahoma,
Pennsylvania, South Carolina, South Dakota, Tennessee, Texas, Vermont, Virginia, West
Virginia, and Wisconsin. (Boschung et al., 555)
Rough Fish
White Sucker (Catostomus commersoni)
White Suckers can grow to 24" in length. They are dusky-olive and have greenish-yellow sides
with a brassy luster. The White Sucker lives in cool, clear streams and lakes. The White
Sucker is native to Arkansas, Colorado, Connecticut, Delaware, Illinois, Indiana, Iowa, Kansas,
Kentucky, Maine, Massachusetts, Michigan, Minnesota, Missouri, Montana, Nebraska, New
Hampshire, New Jersey, New Mexico, New York, North Dakota, Ohio, Oklahoma,
Pennsylvania, Rhode Island, South Dakota, Tennessee, Vermont, Virginia, West Virginia,
Wisconsin, and Wyoming. (Boschung et al., 459)
Green Sunfish (Lepomis cyanellus)
Green Sunfish are up to 10" long and weigh as much as 2 1/4 Ibs. Green Sunfish are yellowish
olive their sides sometimes have dusky bars and their belly is pale olive. They inhabit clear to
turbid waters with low current including small streams, swamps, and ponds. Alabama,
Arkansas, Colorado, Connecticut, Delaware, Illinois, Indiana, Iowa, Kansas, Kentucky,
Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nebraska, New Mexico,
New York, North Carolina, North Dakota, Ohio, Oklahoma, Pennsylvania, South Dakota.,
Tennessee, Texas, Virginia, West Virginia, Wisconsin, and Wyoming. (Boschung et al., 553)
Smalhnouth Buffalo (Ictiohus bubalus)
The Smalhnouth can grow to a length of 3' and a weight of 51 Ibs. They are dark-olive to gray
with grayish to bronze sides. The Smallmouth Buffalo is found in clear to slightly turbid waters
with moderate current and in lakes and reservoirs. It is native to Alabama, Arkansas, Colorado,
Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Minnesota, Mississippi, Missouri,
Montana, Nebraska, New Mexico, North Dakota, Ohio, Oklahoma, South Dakota, Tennessee,
Texas, and West Virginia. (Boschung et al., 462)
Flathead Catfish (Pylodictis olivaris)
The Flathead Catfish reach a maximum length of 4'5" and weight of 91 1/4 Ibs. They are olive-
yellow to light brown with dark mottling. The Flathead is a good sport fish and inhabits large
creeks, rivers, and reservoirs usually near natural debris. The Flathead is native to Alabama,
A-6
-------
Arkansas, Colorado, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Minnesota,
Mississippi, Missouri, Nebraska, New Mexico, New York, North Dakota, Ohio, Oklahoma,
Pennsylvania, South Dakota, Tennessee, Texas, West Virginia, and Wisconsin (Boschung et al.,
475)
Freshwater Drum (Aplodinotus grunniens)
The Freshwater Drum has a maximum length of thirty-five inches and weight of just over fifty-
four pounds. They are silvery-bluish above, silvery on the sides, and whitish on the bottom.
The Freshwater Drum is widely fished for sport and lives in small to large rivers with slow to
moderate current and lakes and reservoirs, usually in deeper water. They are located in
Alabama, Arkansas, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maryland, Michigan,
Minnesota, Mississippi, Missouri, Montana, Nebraska, New York, North Carolina, North
Dakota, Ohio, Oklahoma, Pennsylvania, South Dakota, Tennessee, Texas, Virginia, West
Virginia, and Wisconsin. (Boschung et al., 616)
Carp (Cyprinus carpio)
The maximum length and weight of a Carp are 30" and 60 Ibs. They are dark olive on the
back, have lighter sides, and are yellowish below. Carp are a mildly popular game fish and
inhabit clear to turbid waters in sloughs, streams, lakes, ponds, and reservoirs. They are found
mainly in bodies of water with aquatic cover and are most common in warm water. Carp range
throughout the United States except for Maine. (Boschung et al., 412)
Brown Bullhead (Ictalurus nebulosus)
Brown Bullheads reach a length of 19" and a weight of 5 1/2 Ibs. They have an olive to black
back, lighter sides mottled with brownish blotches, and a whitish belly. Brown Bullheads are
sought by anglers and found in clear water in deep pools with heavy vegetation. They are native
from the Dakotas, Minnesota, Missouri, Arkansas, and Louisiana throughout the East United
States. (Boschung et al., 470)
Channel Catfish (Ictalurus punctatus)
Channel Catfish can grow to a maximum of 3'11" hi length and 58 Ibs in weight. They are
blue-gray on the back, have light blue to silvery sides with scattered dark olive to black spots,
and a white belly. The Channel Catfish is a very popular sport fish. It lives in rivers and large
creeks with slow to moderate current and in ponds, lakes, and reservoirs. Channel Catfish are
found in Alabama, Arkansas, Colorado, Florida, Georgia, Illinois, Indiana, Iowa, Kansas,
Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nebraska, New
Mexico, New York, North Carolina, North Dakota, Ohio, Oklahoma, Pennsylvania, South
Dakota, Tennessee, Texas, West Virginia, Wisconsin, and Wyoming. (Boschung et al., 470)
A-7
-------
REFERENCES
Boschung, Herbert T., Jr., James D. Williams, Daniel W. Gotshall, David K. CaldweU, and
Melba C. CaldweU. 1983. The Audubon Society: Field Guide to North American
Fishes. Whales & Dolphins. New York: Knopf.
A-8
-------
APPENDIX B
GISS EQUILIBRIUM
U.S. Maps Showing Increased Temperature
and Habitat Changes for Recreational Fish
-------
-------
EXHIBIT B-l
Maximum Weekly Average Temperature at Doubled CO2
GISS Equilibrium Results
(Scaled to IPCC 92 "Best Estimate" Climate Sensitivity)
Highest Maximum per State
Lowest Maximum per State
B-l
-------
EXHIBIT B-2
Loss of Habitability by Guild
GISS Equilibrium - Doubled CO2
Cold Water Guild
Cool Water Guild
Warm Water Guild
Rough Guild
1-49%
150-99%
1 00%
B-2
-------
EXHIBIT B-3
Loss of Habitability for Cold Water Species
GISS Equilibrium - Doubled CO2
Brook Trout
Brown Trout
Chinook Salmon
Chum Salmon
Rainbow Trout
Coho Salmon
Cutthroat Trout
Pink Salmon
Mountain Whitefish
1 -49%
I 50-99%
1 00%
B-3
-------
EXHIBIT B-4
Loss of Habitability for Cool Water Species
GISS Equilibrium - Doubled CO2
Muskellunge
Pumpkinseed
Northern Pike
Yellow Perch
0%
1-49%
Walleye
150-99%
1 00%
B-4
-------
EXHIBIT B-5
Loss of Habitability for Warm Water Species
GISS Equilibrium - Doubled CO2
Black Crappie
Bluegill
Gizzard Shad
Golden Shiner
Sauger
Largemouth Bass
Smallmouth Bass
White Crappie
Rock Bass
White Bass
1 -49%
50-99%
1 00%
B-5
-------
EXHIBIT B-6
Loss of Habitability for Rough Water Species
GISS Equilibrium - Doubled CO2
Brown Bullhead
Rathead Catfish
Carp
Freshwater Drum
Channel Catfish
Green Sunfish
Small Mouth Buffalo
1 -49%
White Sucker
50-99%
1 00%
B-6
-------
APPENDIX C
OSU EQUILIBRIUM
U.S. Maps Showing Increased Temperature
and Habitat Changes for Recreational Fish
-------
-------
EXHIBIT C-l
Maximum Weekly Average Temperature at Doubled COZ
OSU Equilibrium Results
(Scaled to IPCC "Best Estimate" Climate Sensitivity)
Highest Maximum per State
Lowest Maximum per State
C-l
-------
EXHIBIT C-2
Loss of Habitability by Guild
OSU Equilibrium - Doubled CO2
Cold Water Guild
Cool Water Guild
Warm Water Guild
0%
1 -49%
Rough Guild
50-99%
1 00%
C-2
-------
EXHIBIT C-3
Loss of Habitability for Cold Water Species
OSU Equilibrium - Doubled CO2
Brook Trout
0%
Brown Trout
1 -49%
Rainbow Trout
Chinook Salmon Chum Salmon Coho Salmon
r\
Cutthroat Trout Pink Salmon Mountain Whitefish
OO%
C-3
-------
EXHIBIT C-4
Loss of Habitability for Cool Water Species
OSU Equilibrium - Doubled CO2
Muskellunge
Northern Pike
Pumpkinseed
Walleye
Yellow Perch
0%
1-49%
50-99%
1 00%
C-4
-------
EXHIBIT C-5
Loss of Habitability for Warm Water Species
OSU Equilibrium - Doubled CO2
Black Crappie
Golden Shiner
Sauger
White Crappie
Bluegill
Gizzard Shad
Largemouth Bass
Smallmoutfi Bass
Rock Bass
White Bass
C-5
-------
EXHIBIT C-6
Loss of Habitability for Rough Water Species
OSU Equilibrium - Doubled CO2
Brown Bullhead
Carp
Channel Catfish
Flathead Catfish Freshwater Drum
Green Sunfish
Small Mouth Buffalo
White Sucker
1-49% 5 0-9 9 % 100%
C-6
-------
APPENDIX D
UKMO EQUILIBRIUM
U.S. Maps Showing Increased Temperature
and Habitat Changes for Recreational Fish
-------
-------
EXHIBIT D-l
Maximum Weekly Average Temperature at Doubled CO2
UKM Equilibrium Results
(Scaled to IPCC "Best Estimate" Climate Sensitivity)
Highest Maximum per State
Lowest Maximum per State
D-l
-------
EXHIBIT D-2
Loss of Habitability by Guild
UKM Equilibrium - Doubled CO2
Cold Water Guild
Cool Water Guild
Warm Water Guild
0%
Rough Guild
1-49% 0111150-99% |i|||||||||||||||lOO%
D-2
-------
EXHIBIT D-3
Loss of Habitability for Cold Water Species
UKM Equilibrium - Doubled CO2
Brook Trout
Chinook Salmon
Cutthroat Trout
0%
Brown Trout
Chum Salmon
Pink Salmon
Rainbow Trout
Coho Salmon
Mountain Whitefish
1 -49%
I 50-99%
1 00%
D-3
-------
EXHIBIT D-4
Loss of Habitability for Cool Water Species
UKM Equilibrium - Doubled CO2
Muskellunge
Pumpkinseed
Northern Pike
Yellow Perch
1-49%
Walleye
50-99%
1 00%
D-4
-------
EXHIBIT D-5
Loss of Habitability for Warm Water Species
UKM Equilibrium - Doubled CO2
Black Crappie
Golden Shiner
Sauger
White Crappie
Bluegill
Gizzard Shad
Largemouth Bass
Smalimouth Bass
Rock Bass
White Bass
to%
.1- «
1-49%
50-99%
1 00%
D-5
-------
EXHIBIT D-6
Loss of Habitability for Rough Water Species
UKM Equilibrium - Doubled CO2
Brown Bullhead
Carp
Channel Catfish
Flathead Catfish
Freshwater Drum
Green Sunfish
Small Mouth Buffalo
White Sucker
o%
1 -49%
I 50-99%
1 00%
D-6
-------
APPENDIX E
TRANSIENT GFDL - 2050
U.S. Maps Showing Increased Temperature
and Habitat Changes for Recreational Fish
-------
-------
EXHIBIT E-l
Maximum Weekly Average Temperature at Doubled CO2
GFDL Transient Results - 2050
(Scaled from "1.16°C Decade")
Highest Maximum per State
Lowest Maximum per State
E-l
-------
EXHIBIT E-2
Loss of Habitability by Guild
GFDL Transient Results - 2050
Cold Water Guild
Cool Water Guild
Warm Water Guild
Rough Guild
o%
1-49%
50-99%
1 00%
E-2
-------
EXHIBIT E-3
Loss of Habitability for Cold Water Species
GFDL Transient Results - 2050
Brook Trout
Brown Trout
Rainbow Trout
Chinook Salmon Chum Salmon Coho Salmon
Kokanee Salmon* Pink Salmon Mountain Whrtefish
1-49% llilllllllllllllllllll 50-99% 100%
E-3
-------
EXHIBIT E-4
Loss of Habitability for Cool Water Species
GFDL Transient Results - 2050
Muskellunge
Shovel-Nose Sturgeon*
0%
Yellow Perch
1-49%
Northern Pike
Walleye
50-99%
1 00%
E-4
-------
EXHIBIT E-5
Loss of Habitability for Warm Water Species
GFDL Transient Results - 2050
Black Grapple
Bluegill
Gizzard Shad
Golden Shiner
Sauger
White Crappie
Largemouth Bass
Smallmouih Bass
Rock Bass
White Bass
0%
1 -49%
50-99%
1 00%
E-5
-------
EXHIBIT E-6
Loss of Habitability for Rough Water Species
GFDL Transient Results - 2050
Brown Bullhead
Carp
Channel Catfish
Rathead Catfish Freshwater Drum Green Sunfish
Small Mouth Buffalo
White Sucker
1 -49%
| C Q _ Q Q Q£ 100 %
E-6
-------
APPENDIX F
TRANSIENT GFDL - 2100
U.S. Maps Showing Increased Temperature
and Habitat Changes for Recreational Fish
-------
-------
EXHIBIT F-l
Maximum Weekly Average Temperature at Doubled CO2
GFDL Transient Results - 2100
(Scaled from "Eighth Decade")
Highest Maximum per State
Lowest Maximum per State
F-l
-------
EXHIBIT F-2
Loss of Habitability by Guild
GFDL Transient Results - 2100
Cold Water Guild
Cool Water Guild
Warm Water Guild
Rough Guild
1-49% iilil50-99%
F-2
-------
EXHIBIT F-3
Loss of Habitability for Cold Water Species
GFDL Transient Results - 2100
Brook Trout
Kokanee Salmon
0%
Brown Trout
Chinook Salmon Chum Salmon
Rainbow Trout
Coho Salmon
Pink Salmon Mountain Whitefish
1 -4
50-99% Mill
F-3
-------
EXHIBIT F-4
Loss of Habitability for Cool Water Species
GFDL Transient Results - 2100
Muskellunge
Northern Pike
Shovel-Nose Sturgeon
Yellow Perch
1-49%
Walleye
50-99%
1 00%
F-4
-------
EXHIBIT F-5
Loss of Habitability for Warm Water Species
GFDL Transient Results - 2100
Black Crappie
Bluegill
Gizzard Shad
Golden Shiner
Sauger
Largemouth Bass
Smallmouth Bass
White Crappie
Rock Bass
White Bass
F-5
-------
EXHIBIT F-6
Loss of Habitability for Rough Water Species
GFDL Transient Results - 2100
Brown Bullhead
Carp
Channel Catfish
Rathead Catfish
Freshwater Drum
Green Sunfish
Small Mouth Buffalo
White Sucker
0%
1-49%
50-99%
1 00%
F-6
-------
APPENDIX G
SUMMARY OF SELECTED ECONOMIC STUDIES OF THE
VALUE OF FRESHWATER FISHING DAYS BY REGION
-------
-------
G. Summary of Selected Economic Studies of the Value of Freshwater Fishing Days
Appendix G provides a brief assessment of current empirical methods employed to
value recreational fishing days and then presents a summary of selected economic studies that
value freshwater fishing days within the United States. The research discussion and presentation
of study results are meant to serve as supplements to the Chapter 3 discussion of the
implementation of the economic model. The designation of values for coldwater, coolwater,
warmwater, and rough fishing days was complicated by the limited number of nationally
representative studies, the limited number of studies that derive values for different types of fish
species, and the variation hi estimated values across studies for single fish species.
G.I Empirical Methods
Tbere are two primary mechanisms for deriving fishing day values: (1) travel cost
methods (TCM) and contingent valuation methods (CVM). Travel cost studies are based on the
implicit trading of travel and time costs for access to fishing sites and rely on the existence of
widespread variation hi travel costs across individuals and fishing sites to assess recreational
values. Within this methodological category, there are two different fundamental approaches
to modeling travel cost behavior: continuous neoclassical models and discrete choice continuous
models. Numerous styles of both of these models have surfaced hi the environmental economics
literature. Distinctions in style are often connected with unique interpretations of travel or time
costs, functional form specifications, and the treatment of substitute or alternative recreational
sites (Freeman, 1993a, Bockstael et al., 1988). Contingent valuation studies are based on
individual responses to hypothetical situations. These studies directly ask individuals what they
would be willing to pay for some hypothetical change hi the natural resource and hi turn the
recreational service flows.
Travel cost methods and contingent valuation methods are respectively referred to as
imputed and indirect market valuation methods. While travel cost methods focus on the
explanation of observed trips to sites based on variations in travel costs and site quality,
contingent valuation methods rely on hypothetical surveys and intended payments (Walsh et al.,
1992). When comparing these two types of studies, it might be expected that travel cost values
are higher than contingent values because the travel cost estimates reflect the value of the entire
trip while the contingent values capture only the value of the recreational activity.
Questions have arised regarding the appropriateness of different methods employed to
determine recreational fishing site and day values. Travel cost methods ulimately often use some
form of consumer surplus measure (i.e., average consumer surplus) to arrive at fishing day
values. Several studies address the limitations of these continuous neoclassical approaches (i.e.,
Morey 1994). Alternative methods to derive site values have been developed. These methods
involve gleaning indirect utility functions and calculating exact welfare measurements using
discrete choice frameworks (i.e, Bockstael et al., 1986). The continuous neoclassical models
appear more suited for analyses where individuals visit many of the sites sampled and visits are
G-l
-------
common across individuals. In turn, the discrete choice models place greater emphasis on
corner solutions (i.e., zero visits) and substitutability across sites (Bockstael et al., 1988).
Several attempts (Walsh et al., 1992, Smith and Kaoru 1990) have been made to address
concerns with the valuation of recreational days such as explaining differences in estimates
across techniques, determining the appropriateness of different techniques, and assessing what
types of factors influence the appropriateness decision. Variations in estimates are often
associated with differences in the characteristics of the anglers, the quality of sites, or the
research methods used for estimation.
G.2 Summary of Selected Empirical Studies
The discussion presented here relies heavily on American Fisheries Society 1993 and
Walsh et al., 1992. Both of these works review several empirical studies and make revisions
to the valuation estimates to allow for some form of meaningful comparison across studies.
Using travel cost values, the recreational fishing day value is calculated by dividing the trip
value by the reported number of days per trip. Alternatively, annual values are divided by rates
of participation or household values are divided by number of persons and days of participation
per person. Adjustments are made to reported study values to account for the omission of travel
time in travel cost studies, the use of individual observations in travel cost methods, the arbitrary
restriction of sample to in-state residents, the omission of a protest mechanism in contingent
valuation studies, and the dollar basis for value estimates (American Fisheries Society, 1993).
The range of fishing day value estimates across methods, geographical areas, and fish
species is made clear by Exhibit[SFXFVAL] which summarizes the results from a select group
of recreational valuation studies. The studies are organized by region and for each study the
reference, value, method, fishing type, and modeling scale are presented. Travel cost and
contingent valuation studies are represented in the table. It is important to that note that in the
context of this analysis the global climate scenarios are linked with changes in the numbers of
four different types of recreational fishing days: cold water, cool water, warm water, and rough.
This economic structure necessitates the derivation of four unit values.
Reviews of the empirical literature suggest that certain species, types of fishing, and areas
of the country have been better studied than others. For example, trout fishing has been broadly
studied as has salmon fishing. In contrast, warm, cool, and rough species have been less well
studied. These research tendencies complicate the derivation of fishing day values on a national
basis by broad activity area and limit the way in which the economic model could specify fishing
day values. There are virtually no studies that assess values for the four fishing categories of
interest to the economic model. In addition, there is wide variation in estimates for single
species across and within regions and estimation method types and little theoretical basis for
deeming one value better than other. These and other issues shaped the designation of the
primary, high, and low fishing day value specifications outlined in Chapter 3.
G-2
-------
CO
3
CO
SL'i
£*
CO
-a
1
0)
1
nce
3
O
H
u
•a 1
1
-
a) 3 tff
<- O C!
CJ
1
Freshws
General
1
09
Louisia
South-c
0
•
=3
CO
Coldw
Genera
cj
CO
(1969)2
6
-------
MD
05
1
1
I
W
3
H
S S
oo ^ « ^ \p
• \o g p ^
l
I
water
al
Wa
Ge
S
3
Coldwat
General
water
N
III
Coldwater
General
08.32
on 60.
Oregon
Washin
II
8
-------
1
00
-a
0
1
£
a
I
•o
?
9
gi
O
l §1
1
00
5 2
88. a
1 at
«§ II
2
a
s g
H 0
1
Califo
u o
a
s?
1
o
-------
a
o
•a
1
ef
Q
M
-g
r£
H
•a
<8
*S w
iH § 'w'
•••* ^^
ra a
»C3 t5
a
Q
fO
w 'H
'-3
B
CO
i
^
W
T>
o>
I
"S
CO
o
1"
«
E
CO
1
CO
en
|
^
^
^> A
H co
-a
Q
"S
S
1
Reference
1
1
a
|| Mountain (CO;
S
1
L
w 3
a) p
^
u
t^ m
CO CO
ON ON (*.
Io ON o • o i—t
•§ c Q *2 Q .f
•§ f f 1 H g.
OO
CJ\
S
1
d
M
1
a
3
2
Freshwater
Steelhead T
g
01
»g
H
go S
S?
H U
OO
Donnelly et al.,
Montana
Coldwater
General
*g
H
tN
00
en en
oo en
Lakes 90.
Streams 1
i
Duffield et al.,
1
a
Coldwater
General
S
1
S
Gordon (1970)2
«
.2 '§
(f J1
.S .
S 2" ^
I -1
ill
Coldwater
General
^
u
1
cs
Hansen (1977)2
Colorado
Coldwater
General
*g
U
m
o\
OO
O
¥
Johnson and Ws
i§
2
"§
£
Colorado
Cache la
Coldwater
Trout
*5
u
^
fe
OO
o\
¥
Johnson and Ws
o
3
« 2
Coldwater
Anadromou
Steelhead T:
*3
U
o
CO
8?
g
W)
Loomis and Sor
o
Coldwater
General
^
H
1
CO
ON
t— )
|| Miller and Hay
-------
las
s-
Colorado
Coldwat
General
at
al
Co
Ge
C
Coldwater
General
ethod
u
J
vo CO
to -o
8 "8 8
12 § ^
O\
a
I
M
%
I
8?
«
•S
•3
I
3
o\
-------
a>
3
CO
5
a>
-a
I
1
Ohio
Lake Eri
Freshwater
Lake
Ohio
Lake Erie
Freshwater
Private Boat
Walleye, Yel
Eri
Ohio
Lake
Yel
War
Wall
lleye 5. 6
ch 4.61
1
York
York
§
.» a
fe H
00
6
S
I
M
-------
i
&
f^
W>
.S
•i
i
'S
TH
« ^ o
i§ jg °
"S ^ o\
=S
1
1
o
g
w
p^
Qt
"oS
CA
*s
e?
«
s
1
1
(9JD
fl
"i
S
*o g
>> ft
C"' CA
-d
e
A
1
1
I
S
fid
•a
«
ij
New York
St. Lawrence Riv
Chautauqua Lake
«
Coolwater
Muskellunj
H
1
H
sT
§•
I
I
: S^
I
1
1
USA by State
J
u
o\ o
^ g 2 8
• c^ -a E
m «j 3 °
His
^K
i-H
a
I
ca
Missouri
Lake Taneycomo
Coldwater
1
VO
04
CO
ON
J
HH
1
J
s
S
|
I
1
Arizona
Coldwater
General
S
o
o\
Tf
&
00
o\
•a
1
USA by State
||
S
^.
OO f^
rt O
2 •§ 2
ill
Sj^i
OO
I
•a
sS
1
'S
&
a
New Mexico
Elephant Butte Ri
«
_§"
•S
Coldwater
Bass, Catfi
1
vo
t^
en
,^v
i— t
1
03
1
O
-------
§
'&
C/l
E>
03
0
Jl
?
^
*
^
ijjH
^
o>
1
Reference
I
«f »
ifll
*g»
O
H
VI
S ^
t ~
II
1
Martin et al., (
c
.2
si
I ^
5 &a
5 .&
11
1 "s
•§ g
O fl)
0 O
v»
S
K>
u
o\
o
a
I
1
§
2 3
2jS
^j
H
00*
oo
o\
v— t
t_l
King and Walke
of
1 1
II
II
« I
.5
o
1
W
5
£j
H
S
5?
OO
o\
CM
C
1
g,
a
Arizona
ll
o 5
U O
5
CJ
H
•*t
oo
oo
o\
^H
Miller and Hay
1
1
. 1
"3 «
sf|l
% " J -s"
•g »" ^S "§
fe po J2 CU
8
H
r^
5-
"
ON
00
S
*-< i
O \O* *O
c^j v^ o ^
S; o -4 d
g ,-< m „ ' *""
§ «- of i if ^ 1
^ . « 5? g .g « ^
U (S h! •< (2 U ffi
1
CO
i
O «
||
I
S5
*5
ii
o £
2
^
u
s;
8
ll
IU.S. Fish and V
and Bureau of C
(1982)2
o
vH
6
-------
a
.S
8P
*
cS
Q
I
.a
fe
*M
S
eS
5
CM
O ^
sil
.•B ^ o
.-2 ^ °
H 0 Oj
j3
0
a
Selected E
CM
O
1
1
ce
-2
OS
cc
60
1
S
CM
o gg
o> "8
« 5
^co
1
^
1
i
2
CM
*
<
O3
S
S
£ g
g
H
Cold 44.81
Warm 39.79
Rough 29.76
=3
I
1
4S«n
3 »
<;
&
S
| "g
•S g
to O
a
H
Trout 58.24
Catfish 38.75
,
i
i
C tj^
"Mg
§ o\
f
S S «- «w
« S -a o o
o S * *
*
M M
o- cx,
-sll
l-i-ijsf-i-i
1^2-a Iff
-------
r
REFERENCES
Agnello, R.J. 1989. The Economic Value of Fishing Success: An Application of
Socioeconomic Survey Data. Fishery Bulletin 87 (1): 223-232.
American Fisheries Society. 1993. Sourcebook for Investigation and Valuation of Fish Kills.
Prepared by Southwick Associates, Arlington Virginia, Supplement to Special Publication
24.
Atkinson et al., 1992. Bayesian Exchangeability, Benefit Transfer, and Research Efficiency.
Water Resources Research 28 (3): 715-722.
Bell, F.W. 1979. Recreational vs. Commercial Fishing in Florida: An Economic Impact
Analysis, Policy Sciences Program, Florida State University Sciences Program,
Tallahassee, Florida.
Bell, F.W., et al., 1982. The Economic Impact and Valuation of Saltwater Recreational
Fisheries in Florida. Florida Sea Grant College. Report No. 47, Gainesville, Florida.
Bockstael et al., 1991. Recreation hi J. Braden and C.Kolstad (editors). Measuring the
Demand for Environmental Quality. North Holland: Elsevier Science Publishers 227-
270.
Bockstael et al., 1989. Measuring the Benefits of Improvements in Water Quality: The
Chesapeake Bay. Marine Resource Economics 6:1-18.
Bockstael et al., 1988. Benefits From Improvements in Chespeake Bay Water Quality. Volume
n of Benefit Analysis Using Indirect or Imputed Market Methods. Prepared for Office
of Policy Analysis, Office of Policy and Resource Management, US EPA. College Park,
Maryland: Department of Agricultural and Natural Resource Economics.
Bockstael et al., 1987. Time and the Recreational Demand Model.
Agricultural Economics 69 (2): 293-302.
American Journal of
Boyle et al., 1988. Analyzing the Effects of Glen Canyon Dam Releases on Colorado River
Recreation Using Scenarios and Unexpected Flow Conditions, in J. Loomis, ed., W-133
Benefits and Costs in Natural Resources Planning. University of California, Davis,
Division of Environmental Studies.
Brown, G.M. and M.J. Hay. 1987. Net Economic Recreation Values for Deer and Waterfowl
Hunting and Trout Fishing, 1980. U.S. Fish and Wildlife Service, Division of Policy
and Directives Management. Working Paper No. 23, Washington, D.C.
G-12
_
-------
Brown, G. et al., 1979. Recreation Valuation: An Economic Analysis of Nontimber Uses of
Forestland in the Pacific Northwest. Washington State University, Forest Policy Project.
January.
Brown, W.G. and P.M. ShaUoof. 1986. Recommended Values for Columbia River Salmon and
Steelhead for Current Fishery Management Decisions. Oregon State University,
Department of Agricultural and Resource Economics, Corvalis.
Brown, W.G. et al., 1983a. Recommended Values for Columbia River Salmon and Steelhead
for Current Fishery Management Decisions. Oregon State University, Department of
Agricultural and Resource Economics, Corvalis.
Brown, W.G. et al., 1983b. Using Individual Observations to Estimate Recreation Demand
Functions: A Caution. American Journal of Agricultural Economics 65 (1): 154-157.
Brown, W.G. et al., 1980. Estimated Expenditures by Sport Anglers and Net Economic Values
of Salmon and Steelhead for Specified Fisheries hi the Pacific Northwest. Oregon State
University, Department of Agricultural and Resource Economics, Corvalis.
Brown, W.G. et al., 1965. Net Economic Value of the Oregon Salmon-Steelhead Sport
Fishery. Journal of Wildlife Management 29: 266-279.
Buerger, R. and J.R. Kahn. 1986. New York Value of Chesapeake Striped Bass.
Resource Economics 6: 19-25.
Marine
Charbonneau, J.J., and M.J. Hay. 1978. Determinants and Economic Values of Hunting and
Fishing. Transactions of the North American Wildlife and Natural Resources Conference
43: 391-403.
Cropper, M. et al., 1991. Environmental Economics: A Survey. Washington, DC: Resources
for the Future. Discussion Paper QE90-12-REV.
Deck, L. et al., 1992. Benefits Transfer: How Good is Good Enough. Presented at the 1992
AERE Workshop, Benefits Transfer: Procedures, Problems, and Research Needs,
Snowbird, Utah, June 3-5.
Donnelly, D.M. et al., 1985. Net Economic Value of Recreational Steelhead Fishing in Idaho.
U.S. Forest Service, Rocky Mountain Forest and Range Experiment Station, Resource
Bulletin RM-9, Fort Collins, Colorado.
G-13
-------
DuffieldetaL, 1987. The Net Economic Value of Fishing in Montana. Montana Department
of Fish, Wildlife, and Parks, Helena. Dutta, N. 1984. The Value of Recreational
Boating and Fishing in the Central Basin of Ohio's Portion of Lake Fjie. Ohio State
University, Ohio State University Sea Grant Program. OSHU-TB-18.
Fiore, J., and F.A. Ward. 1987. Managing Recreational Water Resources to Increase
Economic Benefits to Anglers in the Arid Southwest. New Mexico State University,
Agricultural Experiment Station, Research Report 609, Las Graces.
Freeman, A.M. 1993a. The Measurement of Environmental and Resource Values.
Washington, DC: Resources for the Future.
Freeman, A.M. 1993b. The Economics of Valuing Marine Recreation: A Review of Empirical
Evidence. Economics Working Paper 93-102, Department of Economics, Bowdoin
College, Brunswick, Maine.
Gibbs, K. 1974. Evaluation of Outdoor Resources: A Note. Land Economics 50 (3): 309-
311.
Gordon, D. 1970. An Economic Analysis of Project Number F18R15 Idaho Sport Fisheries.
Report of Idaho Cooperative Fishery Unit to Idaho Department of Fish and Game, Boise.
Green, T.G. 1984. Compensating and Equivalent Variation of the Florida Saltwater Tourist
Fishery. Florida State University. Doctoral Dissertation, Florida State University,
Tallahassee.
Gum, R.L. and W.E. Martin. 1975. Problems and Solutions in Estimating the Demand for the
Value of Rural Outdoor Recreation. American Journal of Agricultural Economics 57:
559-566.
Hansen, C. 1977. A Report on the Value of Wildlife. U.S. Forest Service, Intermountain
Region, Ogden, Utah.
Huppert, D.D. 1989. Measuring the Value of Fish to Anglers: Application to Central
California Anadromous Species. Marine Resource Economics 6: 89-108.
Huppert, D.D. and C.L. Thomson. 1984. Demand Analysis of Partyboat Angling in California
Using the Travel Cost Method. National Marine Fisheries Service, SouthWest Fisheries
Center, February.
G-14
-------
Hushak, L.J., and H.Y. Jeng. 1990. Recreation Demand Model With Endogenous Trip
Duration and Trip Costs. Ohio State University, Department of Agricultural and
Resource Economics and Rural Sociology and Ohio Sea Grant College Program,
Columbus.
Hushak, L.J., J.M. Winslow, and N. Dutta. 1988. Economic Value of Great Lakes
Sportfishing: The Case of Private-Boat Fishing in Ohio's Lake Erie. Transactions of
the American Fisheries Society 117: 363-373.
Hyatt. W. A. 1984. An Angler Survey and Economic Study of the Farmington River Fishery
Resource. Connecticut Department of Environmental Protection, Hartford.
Johnson, D.M., and R.G. Walsh. 1988. Effects of Participant Skill on the Value of Alternative
Fishing Management Practices. Colorado State University, Department of Agricultural
and Resource Economics, Fort Collins.
Johnson, D.M., and R.G. Walsh. 1980. Economic Benefits and Costs of Fish Stocking
Program at Blue Mesa Reservoir, Colorado. Colorado Water Resources Research
Institute, Colorado State University, Technical Report 49, Fort Collins.
Johnson, D.M., and Adams. 1989. On the Marginal Value of a Fish: Some Evidence from a
Steelhead Fishery. Marine Resource Economics 6: 43-55.
Kalter, R. and L. Gosse. 1969. Outdoor Recreation in New York State: Projection of Demand,
Economic Value, and Pricing Effects for the Period 1970-1985. Cornell University,
Special Series 5, Ithaca, New York.
Kaoru, Y. 1991. Valuing Marine Recreation by the Nested Random Utility Model: Functional
Structure, Party Composition and Heterogeneity. Woods Hole Oceanographic Institution,
Marine Policy Center. Working paper.
Kay, D.L., T.L. Brown, and D.J. Allen. 1987. The Economic Benefits of the Restoration of
Atlantic Salmon to New England Rivers. New York State College of Agriculture and
Life Sciences. Working paper, November.
King, D. and J.G. Hof. 1985. Experimental Commodity Definition in Recreation Travel Cost
Models. Forest Science 31: 519-529.
King, D. and A. Walker. 1980. A Market Analysis of Trout Fishing on Fort Apache Indian
Reservation. University of Arizona, School of Renewable Natural Resources, Tucson.
G-15
------- |