Download File

HUMAN-MEDIATED DISPERSAL OF AQUATIC NONINDIGENOUS SPECIES:

IMPACTS AND INTERVENTIONS

A Dissertation

Submitted to the Graduate School

of the University of Notre Dame

in Partial Fulfillment of the Requirements

for the Degree of

Doctor of Philosophy

John D. Rothlisberger

David M. Lodge, Director

Graduate Program in Biological Sciences

Notre Dame, Indiana

August 2009

HUMAN-MEDIATED DISPERSAL OF AQUATIC NONINDIGENOUS SPECIES:

IMPACTS AND INTERVENTIONS

Abstract

John D. Rothlisberger

The introduction and establishment of species beyond the boundaries of their native ranges is an environmental issue of increasing scope and seriousness. This dissertation examines the consequences of the establishment of aquatic nonindigenous species (NIS) in the Laurentian Great Lakes (GL) region and also investigates alternatives for reducing anthropogenic spread of nuisance aquatic NIS.

I first investigate the pathways by which aquatic NIS are introduced to the GL to learn if introduction pathway is related to where species originate and how likely they are to have spread beyond the GL basin. My analysis shows that ballast water release is highly likely to introduce new aquatic NIS to North America, whereas unauthorized release of organisms in trade tends to introduce to the GL aquatic NIS already established in North America. Moreover, it appears that it is primarily a matter of time before novel

NIS that become established in the GL appear in other North American waterways. I also consider the relationship between introduction pathway and species impacts, finding that John D. Rothlisberger there is an apparent relationship, but that further study of species-specific impacts is needed to verify this finding.

Given the importance of ballast water release in bringing novel species to the GL,

I use a novel technique to estimate the economic impacts in the region of ecological changes caused by populations of aquatic NIS introduced by this pathway. This study concludes that the economic impacts of ballast water species are large, but are also uncertain. Nevertheless, policies that aim to reduce the likelihood of additional invasions via this pathway appear to be economically justifiable.

As nuisance aquatic NIS in the GL region spread to other waterways, they bring with them ecological and economic impacts. The detrimental nature of these impacts motivates efforts to reduce the rate of spread. To inform such efforts, I test the efficacy of multiple methods for removing aquatic NIS from recreational boats and trailers. I found that visual inspection and hand removal is highly effective in removing the nuisance macrophyte Myriophyllum spicatum, but that high-pressure washing is needed to effectively remove small-bodied organisms, including the exotic predatory zooplankter

Bythotrephes longimanus.

Beyond the tactics of how to clean boats, I evaluate efforts to strategically place boat cleaning stations on the landscape. My results show that a common predictive model is limited in its ability to predict which uninvaded lakes cleaning stations should protect. Instead, it appears that placing cleaning stations at invaded lakes to block the transport of invasive propagules is generally more likely to reduce landscape-level spread than protecting uninvaded lakes.

John D. Rothlisberger

Aquatic NIS are only one of many environmental and cultural factors that affect ecosystems and societal interactions with the natural environment. To put the importance of aquatic NIS in context with other potential drivers of change in GL fisheries over the next two decades, I interviewed experts on GL fisheries, asking them to predict changes and to identify the most likely drivers of the changes they predicted. This study revealed that changing cultural interests are the main reason for expected declines in GL fisheries, but that NIS are the predominant environmental driver of change.

The ecological and social issues surrounding NIS are complex and multi-faceted.

As human populations grow, causing global environmental changes and taxing the supply of natural resources, the line separating ecological concerns from social ones is increasingly blurred. In this dissertation, I have included humans as a key component of the ecosystems of the GL region and considered the ecological effects of human actions with respect to their introduction of and intervention against aquatic NIS. In so doing, this dissertation presents several case studies of aquatic NIS in the GL with the aim of providing insights regarding opportunities and pitfalls for efforts to improve NIS policy and management.

To my parents, Dana and Ann. Thank you for your love and support.

CONTENTS

Figures...... vii

Tables ...... xiii

Acknowledgements ...... xv

Chapter 1: Dissertation introduction ...... 1 1.1 A role for ecology in natural resource policy and management ...... 1 1.2 Nonindigenous species...... 6 1.3 Human influence on the ecology of the Laurentian Great Lakes region ...... 9 1.4 Matching scientific inquiries on ecological topics to pertinent policy and management questions ...... 14 1.5 Dissertation outline ...... 15

Chapter 2: The Laurentian Great Lakes are a freshwater invasion beachhead: pathways of nonindigenous species introduction predict prior distribution, subsequent spread, and potential impacts ...... 20 2.1 Abstract ...... 20 2.2 Introduction ...... 22 2.3 Methods...... 25 2.3.1 Pathway and distribution prior to discovery in Great Lakes ...... 25 2.3.2 Pathway and spread beyond the Great Lakes ...... 27 2.3.3 Pathway and impacts...... 28 2.4 Results ...... 34 2.4.1 Pathway and distribution prior to discovery in Great Lakes ...... 34 2.4.2 Pathway and spread beyond the Great Lakes ...... 37 2.4.3 Pathway and impacts...... 41 2.5 Discussion ...... 42 2.5.1 The Great Lakes as a beachhead ...... 42 2.5.2 The Great Lakes as a melting pot ...... 45 2.5.3 Pathways and impacts ...... 47 2.5.4 Conclusion ...... 49 2.6 Acknowledgements ...... 50

iii

Chapter 3: Ship-borne nonindigenous species diminish Great Lakes ecosystem services ...... 51 3.1 Abstract ...... 51 3.2 Introduction ...... 51 3.3 Results ...... 54 3.4 Discussion ...... 64 3.5 Acknowledgements ...... 68

Chapter 4: Aquatic invasive species transport via trailered boats: what is being moved, who is moving it, and what can be done ...... 69 4.1 Abstract ...... 69 4.2 Introduction ...... 70 4.3 Methods...... 75 4.3.1 Observational study ...... 75 4.3.2 Mail survey ...... 77 4.3.3 In-person Northwoods survey ...... 78 4.3.4 Experiment ...... 78 4.4 Results ...... 81 4.4.1 Observational study ...... 81 4.4.2 Mail survey ...... 88 4.4.3 In-person Northwoods survey ...... 89 4.4.4 Experiment ...... 89 4.5 Discussion ...... 90 4.6 Acknowledgements ...... 100

Chapter 5: Limitations of gravity models in predicting the spread of Eurasian watermilfoil (Myriophyllum spicatum) ...... 101 5.1 Abstract ...... 101 5.2 Introduction ...... 102 5.3 Methods...... 107 5.3.1 Relationship between propagule pressure and probability of establishment ...... 108 5.3.2 Model validation ...... 111 5.3.3 Cost-effectiveness of alternative interventions ...... 113 5.4 Results ...... 115 5.4.1 Relationship between propagule pressure and probability of establishment ...... 115 5.4.2 Model validation ...... 115 5.4.3 Efficacy of alternative intervention strategies ...... 117 5.5 Discussion ...... 117 5.6 Management recommendations ...... 129 5.7 Acknowledgements ...... 129

Chapter 6: Future declines of the binational Laurentian Great Lakes fisheries: recognizing the importance of environmental and cultural change ...... 130 6.1 Abstract ...... 130 6.2 Introduction ...... 131 6.3 Methods...... 133 6.3.1 Expert selection and interviews ...... 133 6.3.2 Analysis, combination, and reporting of results ...... 135 6.4 Results ...... 141 6.4.1 US commercial fishery ...... 143 6.4.2 Canadian commercial fishery ...... 143 6.4.3 US sport fishery ...... 145 6.4.4 Canadian sport fishery ...... 146 6.5 Discussion ...... 146 6.5.1 Drivers of change in Great Lakes fisheries ...... 146 6.5.2 Conclusions ...... 149 6.6 Acknowledgements ...... 150

Chapter 7: Dissertation conclusion ...... 151 7.1 Ecology and society ...... 151 7.2 Dissertation overview ...... 152 7.3 Possibilities and pitfalls ...... 156 7.4 Conclusion ...... 159

Appendix A: List of nonindigenous aquatic species established in the Great Lakes and data on these species compiled for analyses in Chapter Two ...... 161

Appendix B: Supporting text for Chapter Three: ship-borne nonindigenous species diminish Great Lakes ecosystem services...... 170 B.1 Methods ...... 170 B.1.1 Selection of experts ...... 170 B.1.2 Briefing book...... 171 B.1.3 Individual interviews ...... 171 B.1.4 Performance measures and combination of expert judgments: the classical model ...... 174 B.1.5 Percent impacts on ecosystem services ...... 180 B.1.6 Economic valuation of impacts ...... 181 B.2 Results ...... 189 B.2.1 Economic valuation of impacts ...... 194 B.2.2 Expert rationales ...... 199

Appendix C: Expert elicitation protocol for ecological and economic impacts of ship- borne nonindigenous species on the Great Lakes ...... 208 C.1 Purpose ...... 208 C.2 Scope ...... 209 C.3 Method ...... 209 v

C.3.1 Format ...... 209 C.3.2 What is a good probability assessor? ...... 211 C.3.3 Expert names ...... 212 C.3.4 Assumptions ...... 212 C.4 Questions ...... 213 C.4.1 Commercial fishing ...... 213 C.4.2 Commercial fishing effort ...... 216 C.4.3 Sport fishing ...... 218 C.4.4 Fouling water intake for power plants & industry ...... 222 C.4.5 Wildlife watching ...... 223 C.5 Answers to training questions ...... 224

Literature cited ...... 225

FIGURES

Figure 1.1. The multi-stage process of biological invasions (Panel A), general policy and management options for dealing with each stage of an invasions (Panel B), and selected research questions for which scientific inquiry can improve the policy and management of invasive species (Panel C). Each of these questions is addressed for specific cases in this dissertation. See the text in the Dissertation Outline section of this chapter for how each of these questions is addressed in the chapters specified above. Panels A and B are redrawn with permission from Lodge et al. (2006)...... 7

Figure 2.1. Proportion of nonindigenous freshwater species that came to the Great Lakes via major introduction pathways that either were first discovered in North America in the Great Lakes basin or, alternatively, outside of the Great Lakes basin. The total number of species introductions attributed to each pathway is shown above each bar. (GL = Great Lakes) ...... 35

Figure 2.2. Nonindigenous freshwater species whose first North American discovery was in the Great Lakes (n = 63) that remain confined to the Great Lakes basin as well as those that have spread beyond the basin. Sub-plots group species according to their pathway of introduction and bars in each sub-plot group species according to their taxonomic category...... 36

Figure 2.3. Misclassification rates as a function of the number of explanatory variables, from leave-one-out cross-validation for binary decision trees intended to predict the current distribution of nonindigenous freshwater species whose first discovery in North America was in the Great Lakes...... 38

Figure 2.4. The binary decision tree with the lowest combined misclassification rates for species that are confined to and which have spread beyond the Great Lakes basin. This tree classifies observations according to a single explanatory variable: years since a species was discovered in the Great Lakes. If it has been less than 74 years since a species was discovered in the Great Lakes, then this model predicts that it is still confined to the basin. The numbers of species correctly and incorrectly classified are shown at the two terminal nodes of this tree (i.e., 55 species were classified as being confined to the Great Lakes basin, 48 correctly so and 7 incorrectly; 8 species were classified as having spread beyond the basin, 7 correctly and 1 incorrectly)...... 39 vii

Figure 2.5. Current distribution (within or outside the Great Lakes) of nonindigenous freshwater species that were first discovered in North America in the Great Lakes versus years since initial discovery. Symbols depict the pathway that introduced each species to the Great Lakes (see Legend). The vertical position of species discovered in the same year has been adjusted to avoid hidden data...... 40

Figure 3.1. Distributions of ship-borne species percent impacts on US commercial fish landings (for each lake), sport fishing effort (for each lake) and expenditures (aggregated across all five lakes), and wildlife viewing effort (aggregated across all five lakes) in 2006. Distributions are performance-based combinations of expert assessments. Solid black lines designate medians, indicating the most likely percentage by which each quantity would have been greater if ship-borne species were not present...... 56

Figure 3.2. Distributions of economic impacts as lost consumer surplus (fishing and wildlife viewing) or additional costs (raw water users), aggregated across lakes, of ship-borne nonindigenous species on ecosystem services in the Great Lakes in the US: A, commercial fishing; B, sport fishing; C, wildlife viewing; D, raw water use). Solid black lines indicate the median and dotted lines the 90% uncertainty range of each distribution. Note differences in scale of horizontal and vertical axes of plots...... 58

Figure 3.3. Ninety percent uncertainty ranges for economic impacts in the United States of ship-borne NIS on multiple ecosystem services in the Great Lakes...... 60

Figure 3.4. Scenarios of future cumulative ship-borne invasive species damage relative to cumulative transportation savings from ocean-going shipping into the GL...... 62

Figure 4.1. Aquatic vegetation found attached to boats and trailers during field survey. Panel A is a histogram of the total mass of fragments on individual boats (bin width = 1g). Panel B shows a histogram of the mass of individual vegetation fragments (bin width = 0.5g)...... 83

Figure 4.2. Average number and type of small-bodied organisms washed from recreational boats and trailers arriving at (n= 36) or departing from (n= 49) lakes in the northern Wisconsin and the Upper Peninsula of Michigan. See Table 4.3 for further detail on taxa included in each taxonomic category...... 84

Figure 4.3. Results of experimental removal of biological materials from boat and trailer via boat washing or visual inspection. Panel A shows removal of Myriophyllum spicatum with different wash pressures and durations, and with visual inspection and hand-removal. Panel B shows data from the same treatments for the removal of small-bodied organisms...... 91

viii

Figure 5.1. Number of Eurasian watermilfoil invasions in Wisconsin lakes larger than 25 ha between 1990 and 2006, inclusive. Bars show the number of new invasions in each year and the line graph shows the cumulative number of invasions in Wisconsin...... 110

Figure 5.2. Evaluation of the gravity model’s ability to accurately predict lake-specific probability of invasion. Each bin holds 100 lakes and bins are arranged from highest predicted probability of invasion on the left to lowest on the right. Bars indicate the number of lakes in each bin that were actually invaded. The average per lake predicted probability of invasion in the top 200 at-risk sites ( pˆ ) is shown for each year. Also shown for each year is the probability or p-value (p) of observing the actual number of newly invaded sites given the value of pˆ for that year. The vertical dashed line shows the cut-off to the left of which are the bins for the 200 lakes predicted most likely to be invaded in each year...... 114

Figure 5.3. Analysis of alternative intervention strategies for slowing the spread of aquatic invasive species. Contour lines show the percent reduction in average per site probability of invasion relative to no intervention versus number (x-axis) and cleaning efficacy (y-axis) of intervention sites deployed. Dashed lines show the protection strategy, where intervention is deployed to prevent introductions at the specified number of uninvaded sites identified by the gravity model as having the highest probability of invasion (i.e., greatest propagule pressure from invaded locations). Solid lines show the containment strategy, where intervention is deployed to keep propagules from leaving the specified number of invaded sites with the greatest probability of initiating new invasions (i.e., highest propagule pressure to uninvaded locations)...... 118

Figure 6.1. Historical and projected commercial and recreational fisheries in the US and Canadian waters of the GL. Angler effort in recreational fisheries is shown as insets in the upper right of each panel. Vertical range bars are performance-based combinations of expert assessments where lower and upper limits show, respectively, 5th and 95th percentiles of the combined expert subjective probability distributions. Hollow circles depict the 50th percentile of each distribution. Note different vertical scales across countries, lakes, and fishery types. Canadian commercial catch (Panel i) is for all Canadian waters of the GL. Historical recreational fisheries data were taken from the USFWS National Survey of Fishing, Hunting, and Wildlife-associated Recreation. Commercial catch data dating back to 1971 were obtained for the US from the USGS Great Lakes Science Center and for Canada from the Department of Fisheries and Oceans. 136

change. Note that, even though most distributions extend to the left of -100%, the value of these variables in 2025 actually cannot be more than 100% less than they were in 2006...... 137

Figure 6.3. Individual and combined expert assessments for US commercial fish landings (left column) and sport fishing effort (right column) for 2006 and 2025 for each of the GL. The lakes are shown in the following order: Superior, Michigan, Huron, Erie, Ontario. Two rows of panels represent each lake, the first showing 2006 assessments and the second 2025. Dashed lines divide 2006 and 2025 estimates for each lake. Each panel shows the 5th to 95th percentile range graphs for individual and combined expert assessments, with filled circles showing 50th percentile estimates. Assessments are shown in the same order in each panel: top to bottom, Expert 1, 2, 3, 4, 5, 6, 7, 8, 9, equally-weighted combination, performance-based combination. For calibration variables, light gray vertical bars show the actual value of the variable being estimated, which became known after the elicitation was finished. Note differences in scale for each lake and fishery type...... 138

Figure 6.4. Individual and combined expert assessments for Canadian commercial fish landings, aggregated across lakes (a, b), and sport fishing effort, divided by lake (c – j), for 2006 and 2025 for each of the GL. For sport fishing, the lakes are shown in the following order: Superior, Huron, Erie, Ontario. Two rows of panels represent each lake, the first showing 2006 assessments and the second 2025. Dashed lines divide 2006 and 2025 estimates for each lake. Each panel shows the 5th to 95th percentile range graphs for individual and combined expert assessments, with filled circles showing 50th percentile estimates. Assessments are shown in the same order in each panel: top to bottom, Expert 1, 2, 3, 4, 5, 6, 7, 8, 9, equally-weighted combination, performance-based combination. For calibration variables, light gray vertical bars show the actual value of the variable being estimated, which became known after the elicitation was finished. Note differences in scale for each lake and fishery type...... 139

Figure 6.5. Individual and combined expert assessments for US GL recreational fishing expenditures for 2006 (a) and 2025 (b). Each panel shows the 5th to 95th percentile range graphs for individual and combined expert assessments, with filled circles showing 50th percentile estimates. Assessments are shown in the same order in each panel: top to bottom, Expert 1, 2, 3, 4, 5, 6, 7, 8, 9, equally- weighted combination, performance-based combination. The vertical light gray bar shows actual 2006 expenditures (USFWS 2007), which became known after the elicitation was over...... 140

Figure 6.6. Number of experts that mentioned various potential drivers of change in explaining their expectations for declines in US and Canadian commercial and recreational fisheries between 2006 and 2025...... 144

Figure 7.1. Scope and content of research projects presented in this dissertation. Chapter numbers accompany a graphical depiction of the topic of each chapter. Biological invasions are a multi-stage process that occur at multiple spatial and temporal scales. For example, the global transport of ballast water in shipping vessels introduces species to the Great Lakes region (Chapter 2) which have ecological and economic impacts in the Great Lakes (Chapter 3). Some of these species and others introduced to the region by other pathways (e.g., commerce in living organisms) spread to other waterways in and beyond the region via anthropogenic mechanisms (Chapters 4 and 5). The impacts of nonindigenous freshwater species in the Great Lakes region take place within a broader context of other environmental and cultural factors that also drive environmental change (Chapter 6). See the text in the Dissertation Overview section of this chapter for explanation of the subject and conclusions of each chapter...... 154

Figure B.1. Schematic of welfare changes related to commercial fishing, illustrating the market model approach taken to estimate economic impacts of ship-borne species...... 184

Figure B.2. Schematic of welfare changes related to outdoor recreation, illustrating the inferred market model approach used to estimate economic impacts of ship-borne species...... 188

Figure B.3. Individual and combined expert assessments showing the impact of ship- borne species on US sport fishing effort (left column) and US commercial fish landings (right column) in 2006. There are two rows of panels for each lake with the first row showing expert assessments for the variable (i.e., angler-days or pounds of commercially landed fish) with ship-borne species (i.e., actual condition) and the second row showing assessments for the variable if ship-borne species had never been introduced (i.e., hypothetical condition). Dashed lines divide assessments with and without ship-borne species. The order of the lakes is, from top to bottom, Superior (a-d), Michigan (e-h), Huron (i-l), Erie (m-p), and Ontario (q-t). Within each panel expert assessments are arranged in order, from top to bottom, Expert 1, 2, 3, 4, 5, 6, 7, 8, 9, equally-weighted combination, and performance-based combination. Vertical light gray bars in panels a, b, e, f, i, j, m, n, q, and r show the realization of the variable in question, which was unknown at the time of the elicitation. Note that the scale of the horizontal axes varies...... 191

Figure B.4. Individual and combined expert assessments of the impact of ship-borne species on US wildlife viewing effort in 2006. The first row shows expert assessments given the presence of ship-borne species (i.e., actual condition) and the second row shows assessments for the variable if ship-borne species had never been introduced (i.e., hypothetical condition). Dashed lines divide assessments with and without ship-borne species. Within each panel, expert assessments are arranged in order, from top to bottom, Expert 1, 2, 3, 4, 5, 6, 7, 8, 9, equally- xi

weighted combination, and performance-based combination. The vertical light gray bar in panel a shows the actual number of wildlife viewing participant-days in 2006, which was unknown at the time of the elicitation...... 192

Figure B.5. Individual and combined expert assessments of the annual per facility impacts of ship-borne species on US raw water users in 2006. The order of the raw water users, from top to bottom, is nuclear power plants (a), water treatment plants (c), fossil fuel power plants (e), and industrial facilities (g). Within each panel expert assessments are arranged in order, from top to bottom, Expert 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, equally-weighted combination, and performance-based combination. Note that the scale of the horizontal axis is different for each user type...... 193

xii

TABLES

Table 2.1. Summary of scientific literature documenting ecological and economic impacts attributable to ballast water species established in the Great Lakes ...... 30

Table 2.2. Summary of scientific literature documenting ecological and economic impacts attributable to nonindigenous freshwater species introduced by commerce in living organisms to the Great Lakes ...... 32

Table 3.1 Summarized distributions of percent impacts of ship-borne nonindigenous species on ecosystem services on ecosystem services in the Great Lakes in the United States in 2006 ...... 55

Table 4.1. Questions and responses from mail and in-person surveys...... 79

Table 4.2. Aquatic plant species and the respective number of fragments of each found on boats and trailers during observational field survey in Northern Wisconsin in Summer 2006 ...... 85

Table 4.3. Taxa collected from boats and trailers during field survey in northern Wisconsin in 2006...... 86

Table 4.4. Nonindigenous species established in the Great Lakes that are morphologically similar to species collected in boat washing samples...... 95

Table 6.1. Experts interviewed and the professional title, affiliation, and qualifications of each (listed alphabetically)...... 134

Table 6.2. Calibration, informativeness, and weights of the nine experts, their equal- weight combination (EQUAL), and their performance-based combination (PBC) for changes in Great Lakes fisheries between 2006 and 2025 ...... 142

Table A.1. Nonindigenous aquatic species established in the Great Lakes as of 2008 .. 162

Table B.1. Experts interviewed and the professional title, affiliation, and qualifications of each (listed alphabetically)...... 172 xiii

Table B.2 Summary statistics on values drawn from the literature on own-price elasticity of demand of commercial fish, the value of sport fishing in the Great Lakes, and the value of wildlife viewing in the Great Lakes region ...... 186

Table B.3. Calibration, informativeness, and weights of the nine experts, their equal- weight combination (EQUAL), and their performance-based combination (PBC) for the impacts of ship-borne species on the Great Lakes in 2006 ...... 190

Table B.4. Summaries of commercial fishery consumer surplus prediction distributions in the Great Lakes in the United States in 2006 ...... 195

Table B.5. Summaries of outdoor recreation consumer surplus prediction distributions in the Great Lakes in the United States in 2006 ...... 196

Table B.6. Additional annual operating costs to raw water users attributable to ship- borne species in the Great Lakes region in the United States in 2006 ...... 200

xiv

ACKNOWLEDGEMENTS

I offer my sincere gratitude to my PhD advisor, Dr. David M. Lodge, for his mentorship during my studies at the University of Notre Dame. His example and intellect have benefited me profoundly, both professionally and personally. My completion of this degree would not have been possible without his guidance and encouragement. I hope to represent him well as I move forward in my career. I also thank my other committee members---Gary Belovsky, Jessica Hellmann, and Gary

Lamberti---for their assistance and input. My associations with these professors have consistently motivated me to improve my work and myself. I have the utmost respect for each of them.

I have also had the opportunity to work closely with senior researchers at other academic institutions. Roger Cooke of Resources for the Future and the Technical

University of Delft and David Finnoff of the University of Wyoming have generously shared their time and expertise with me. Their help has made parts of this dissertation possible that would not otherwise have been so.

I am grateful for the friendly culture and spirit of cooperation that exist in the

Lodge lab and for the opportunity that I have had to interact with many excellent lab members during my graduate career. I thank Joanna McNulty for all she has done to help the projects and grants I have worked on flow smoothly. The positive experiences I have had as a doctoral student have largely been shaped by my interactions with the xv

postdoctoral researchers in the lab---Jon Bossenbroek, Kevin Drury, Darren Yeo, Chris

Jerde, and Andy Mahon---and with my fellow graduate students---Sadie Rosenthal,

Reuben Keller, Jody Peters, James Larson, Konrad Kulacki, Brett Peters, Matthew

Barnes, Andrew Deines, and Ashley Baldridge. As an honorary lab member, Lindsay

Chadderton also belongs on this list. I thank each of these individuals for their constructive influence on my life, as colleagues and as friends.

Many technicians have helped to bring about this research. Mark Drew, Penny

Nichols, Sarah Sutton, Tim Campbell, Mike McCann, and Rebecca Hale have all made significant contributions to this work. I also thank Sheila Kennedy, Jeff Delfeld, Brandon

Feasel and Neil Wallace for their data collection efforts.

Finally and always, I am most grateful to my wife, Emily, for her love and support during our pursuit of this degree. She is everything to me. I also express my deep love and gratitude for my sons---Matthew, Jacob, and Daniel.

xvi

CHAPTER 1:

DISSERTATION INTRODUCTION

1.1 A Role for Ecology in Natural Resource Policy and Management

Among the most significant current challenges for the science of ecology are issues involving the interactions of humans with the natural environment (May 1999,

NRC 2001, MEA 2005). Given the significant effects of humans on biodiversity and ecological processes, there is a great need for ecological guidance in addressing current and future environmental crises (Ehrlich and Ehrlich 2004, Cabrera et al. 2008). The most serious ways in which human activities are modifying the natural world include habitat fragmentation, greenhouse gas release, nitrogen deposition, and the spread of invasive nonindigenous species.

In seeking to provide guidance on such pressing concerns, ecology must sometimes reach beyond its disciplinary boundaries, especially when social issues are involved, working together with researchers in economics, geography, sociology, and other disciplines. Understanding ecological change requires interdisciplinary work because the indirect drivers of ecological change are often economic and cultural.

Furthermore, interdisciplinary work is intellectually challenging because it requires researchers to extend beyond traditional disciplinary boundaries and approaches. For ecological research to be relevant to natural resource policy and management, it must investigate and increase our understanding of the important drivers of ecological change. 1

In this dissertation, I have applied scientific principles to address important practical questions pertaining to the policy and management of invasive nonindigenous species.

This work is grounded in ecology, but at times also connects to other disciplines, in an effort to enhance the potential relevance of the findings of this research.

Humans interact with the natural world in multiple ways. Some human interactions with the environment involve deriving benefits (i.e., goods and services) from ecosystems (Costanza et al. 1997, Daily 1997). Other human-environment interactions are the byproducts of economic activities and can sometimes result in environmental degradation. Natural resource policy and management actions are generally intended either to maintain or enhance ecosystem services (i.e., the benefits of nature to society) or to limit environmental damage from other anthropogenic sources

(Daily and Matson 2008). Scientific research in ecology is a tool to understand how human actions, including natural resource policy and management decisions, may affect the ecosystems with which humans interact (Naidoo et al. 2008).

Nevertheless, understanding anthropogenic effects on ecosystems is difficult because ecological systems are complex (Levin 1999). With this complexity comes a high degree of scientific uncertainty about the functioning, and even the structure, of these systems. Some of this scientific uncertainty arises from the difficulty of connecting cause to effect in ecological dynamics. The vast complexity of ecosystems and the challenge of understanding how they operate have led to various approaches for gaining knowledge about them. In recent decades, one of the most common strategies in ecological research has been a reductionist approach, whereby ecological interactions are stripped down to their simplest form. This simplification allows controlled, replicated

experiments to be performed. Using experiments, researchers can investigate how one or a few variables of interest respond to the manipulation of one or a few potentially important explanatory variables (Pickett et al. 1994, Underwood 1997). In conjunction with inferential statistics, this is a powerful approach in learning how organisms interact with one another and with their physical environment (Platt 1964).

Despite its usefulness in discovering the mechanisms of ecological interactions, not all important questions in ecology and environmental science can be adequately addressed using reductionist science. Logistical considerations generally constrain controlled, manipulated experiments to short time frames, small spatial scales, and questionable realism (Diamond 1986). Recent efforts have been made, however, to increase the spatial scale and realism of controlled ecological experiments. Notable among such efforts are the Free-Air Carbon dioxide Enrichment (FACE) experiments that are investigating the effects of elevated atmospheric carbon dioxide on forest ecosystems (DeLucia et al. 1999). These experiments are being conducted at the acre- plus scale and will perhaps provide a more integrated understanding of ecosystem responses to elevated carbon dioxide than will data from smaller-scale experiments conducted in environmental chambers. A classic historical example of manipulating an entire ecosystem and documenting the effects of the manipulation versus an unmanipulated control are whole-lake phosphorus amendments in the northern temperate region (Schindler et al. 1978, Elser et al. 1986, Carpenter and Kitchell 1993).

Though the value of these large-scale experiments is clear, it is also clear that as system complexity increases, so does the difficulty of tracking the causal linkages from manipulated features of an ecosystem to the components hypothesized to respond to

manipulation. Moreover, the more complex (i.e., realistic) a system is, the greater the uncertainty that the same system or a highly similar one would respond in the same way to nearly the same manipulation, if repeated. Even the comparatively simple dynamics of insect populations, for example, can exhibit extreme sensitivity to initial conditions, or chaos, making the outcome of manipulations uncertain in the absence of complete knowledge of the initial conditions (May 1974). Nevertheless, most ecological studies are conducted on plots that are less than 3 square meters and for periods of time less than

5 years, and are not repeated by independent investigators (Levin 1992, Lodge et al.

1998).

If one of the goals of ecological research is to provide knowledge to manage ecosystems according to scientific principles, there are many instances where short, small-scale experiments may have limited utility in providing such guidance. One reason for such limitations is that the data and inferences from such investigations may not match well with critical policy and management questions in the real world.

Furthermore, other interactions in an ecosystem may be more important drivers of the system’s dynamics than the interactions that have been studied. Thus, well-studied mechanisms may be swamped in importance by relatively unknown interactions that operate on a higher organizational level. Such higher-level interactions may remain unstudied because of their limited tractability under traditional reductionist methods.

One approach that has been used to address higher-order ecological phenomena that operate at large spatial scales has been to take advantage of naturally occurring manipulations to ecosystems (i.e., perturbations), observing how variables in the ecosystem respond to natural events or conditions. For example, hurricanes have been

used to study the role of disturbance and the process of secondary succession (Turner et al. 1998); bird communities on islands of different sizes have been used to study colonization, niche breadth, and community assembly rules (Diamond 1970, Diamond

1973); other cases of the biogeography of islands have been used to learn more about dispersal, colonization, extinction, and community dynamics (MacArthur and Wilson

1967), and volcanic eruptions allow for the study of primary succession (Wood and

Delmoral 1987). These types of opportunities have been called natural experiments

(Diamond 1986). In natural experiments, differences in conditions among locations because of varying degrees of perturbation (i.e., across space) are employed to make inferences about how ecosystems develop through time with respect to their structure and functioning. The space-for-time swap is a common approach in ecology that allows conclusions to be drawn about temporal processes in a much shorter timeframe, months or years versus centuries or millennia, than would be possible by actually observing the processes over time. In other words, because we can observe what happens given varying ecological conditions across space, we can make predictions about what will occur as ecological conditions vary through time.

Biological invasions provide an opportunity to reverse the typical space-for-time swap in ecological research to a time-for-space swap. By retrospectively assessing the process and consequences of previous biological invasions, many of which have occurred in the recent past, and for which some degree of historical data are available, we can make predictions about patterns and processes of biological invasions of the same or similar species that are likely to occur in other locations (i.e., elsewhere in space). This

kind of analysis may prove to be highly informative for natural resource policy and management of biological invasions.

1.2 Nonindigenous Species

Harmful nonindigenous species are an environmental problem of global concern.

Invasive species have been implicated in declines of native biodiversity (Nalepa et al.

1996, Wilcove 1998, Gurevitch and Padilla 2004), with many of their greatest impacts expected to occur in freshwater ecosystems (Sala et al. 2000). Economic losses from invasive species are also substantial, over $100 billion per year in the United States

(Pimentel et al. 2005). Because biological invasions are linked with trade, invasions are expected to increase in the coming decades as trade also increases (Levine and

D’Antonio 2003). Thus, there is a critical need for science to help improve policy and management related to biological invasions.

Biological invasions result from a multi-stage process (Figure 1.1A) that begins with species being transported beyond their native range via some transportation pathway or vector. There are two major categories of anthropogenic transportation of live organisms. The first is the intentional movement of organisms for commercial purposes

(e.g., live food trade, pet trade) and the second is the inadvertent transport of live organisms that is incidental to commerce and travel (e.g., planktonic species in ballast water in ships, wood-boring beetles in packaging materials). Via a wide array of transportation vectors in these two categories, species are introduced to areas outside their native range. A percentage of introduced species (typically 10-50%) become established in a new area when they form one or more self-sustaining 6

General Policy and Selected Research A) Invasion Process B) Management Options C) Questions Species in pathway Prevention

Transported and released alive Early detection, rapid response, and Is there a relationship between introduction pathway and eradication origin of nonindigenous species established in a particular Population region? (Chapter 2) established What species does a particular pathway spread? (Chapter 4) What strategies and tactics most effectively limit the spread Control and of species by particular transportation pathways? (Chapters 4 and 5) Spread slow the spread How accurately can models predict the spread of invasions? (Chapter 5) Are species’ impacts related to introduction pathway? Human adaptation (Chapter 2) Ecological, (change behavior Have NIS introduced by a particular pathway to a particular human health, or and bear the costs) economic impact region modified ecosystem goods and services, and what have been the economic impacts of changes? (Chapter 3) What is the importance of nonindigenous species’ impacts relative to other drivers of change in determining how humans interact with the environment? (Chapter 6)

populations there (Williamson 1996, Jeschke and Strayer 2005). When this occurs, species become part of the nonindigenous flora or fauna of the region. A nonindigenous species is classified as invasive if it spreads throughout the new range and has net negative impacts on ecosystems, human health, or economic interests (Lodge et al. 2006).

Of course, assessments of whether the impacts from a nonindigenous species are on the net positive or negative depend on society’s perceptions. It is, therefore, not the role of ecologists alone to specify whether a species is invasive or not.

In the chapters that follow, I focus on nonindigenous aquatic species that are currently established in the Laurentian Great Lakes (GL) and the surrounding region.

Some of these species are widely acknowledged as being invasive, including, for example, Eurasian watermilfoil Myriophyllum spicatum and zebra mussel Dreissena polymorpha. Others, however, are not, including many of the species on the list of 95 aquatic species established in the GL that I consider in Chapter Two (see Appendix A).

Thus, depending on the chapter and its focus and on the species involved, I refer to some species as aquatic invasive species (AIS) and to others as freshwater nonindigenous species (NIS).

Biological invasions have important ecological and evolutionary implications

(Sax et al. 2005). When a species joins a biological community beyond its native range, it interacts with species that may be very different from those with which it has interacted through evolutionary time. These interactions may include predation, competition, and parasitism, but may also occur through more indirect means, for example by changing abiotic environmental conditions. These novel interactions can affect the evolutionary trajectory of the nonindigenous and native species in a community (Wares et al. 2005).

Potentially novel abiotic conditions can also place new selective pressures on NIS, at times resulting in rapid evolution (Huey et al. 2005). On shorter time scales, the establishment of NIS in a community can have substantial effects at the population, community, and ecosystem levels (Sakai et al. 2001, D'Antonio and Hobbie 2005,

Blackburn and Gaston 2005). Human societies deem some of these ecological effects to be undesirable. Negative impacts include population declines of native species, nuisance-level abundance of NIS, increased disease of humans and organisms important to humans, and alterations to ecosystem functioning (e.g., nutrient cycling; Hall et al.

2003). As previously mentioned, aquatic NIS with these types of ecological effects are often considered to be AIS.

Although the invasion process is relatively well understood (Figure 1.1A) and general recommendations for dealing with invasions are available (Figure 1.1B), many biological invasions have not been studied in sufficient detail to use this understanding to support informed natural resource policy and management decisions (Figure 1.1C). More in-depth scientific study regarding specific stages of specific invasions, such as presented in this dissertation, will help to identify aspects of NIS policy and management that could be improved through scientific research. Similar efforts could also reveal areas where additional scientific developments are most needed before valid scientific advice can be offered.

1.3 Human Influence on the Ecology of the Laurentian Great Lakes Region

The geographic focus of this dissertation is on the GL and the states that surround the lakes. The history of GL is an instructive study in the interactions between humans 9

and natural resources. Beginning in the early 1800s, humans have extracted fishery resources from the GL at a sometimes unsustainable pace (Bogue 2000). These resources contributed to rapid regional population growth and prosperity. Larger populations demanded more resource extraction, more intensive agriculture, and the development of industries. Erosion and water pollution, byproducts of agriculture and industry, further modified the GL environment, often to the detriment of populations of native species, especially fisheries. In an effort to conserve these valuable natural resources, a variety of management actions were taken. Early management actions generally aimed to maintain or enhance utilitarian production of fishery resources, but later actions sought to rehabilitate and restore native ecological structure and function (Brown et al. 1999).

Anthropogenic changes to the surrounding landscape also affected the ecology of the GL. Dams placed on GL tributaries for irrigation and power generation limited reproductive opportunities for potadromous fish species. For example, Atlantic salmon

(Salmo salar), once abundant in L. Ontario, had nearly disappeared by 1850, a casualty of overfishing and spawning habitat loss (Coon 1999). Shipping canals provided new hydrological connections among the lakes and to the lakes from other waterways, facilitating species introductions, which in some cases significantly altered GL foodwebs.

The Welland Canal, first completed in 1829, circumvented the barrier of Niagara Falls, which until then had prevented species (and ship) movement upstream from L. Ontario to

L. Erie and the upper GL. Via the Welland Canal, sea lamprey (Petromyzon marinus) invaded L. Erie and the upper GL early in the 20th century, exerting massive predation pressure on lake trout (Salvelinus namaycush).

Victim to overfishing, spawning habitat loss, and sea lamprey predation, lake trout populations crashed across the GL from 1940 to 1950 (Eshenroder and Amatangelo

2002). Lake trout had historically been an abundant top piscivore in the lakes and its dramatic decline prompted mitigation efforts: commercial fisheries were heavily regulated, some spawning habitat was restored, and lamprey control was initiated.

Despite these efforts, lake trout had already been essentially extirpated from all GL except for L. Superior. The loss of the lakes’ top predator led to substantial foodweb alterations. The most striking of these changes was the invasion of alewife (Alosa pseudoharengus), a North Atlantic planktivore, via the Erie Canal, which connects the

Hudson-Mohawk Rivers to L. Erie.

In the absence of large piscivores and with abundant plankton, a byproduct of cultural eutrophication, to consume alewife populations exploded (O’Gorman and

Stewart 1999). Alewife reached nuisance levels in the GL, especially L. Michigan, in the early to mid-1960s. Through egg and larval predation and competition with juveniles and adults, alewife further changed native fish assemblages, causing declines in yellow perch

(Perca flavescens) and deepwater ciscoes (Coregonus spp.) (Crowder 1980).

Humans also experienced the negative effects of the alewife outbreak directly when particularly cold overwinter temperatures led to a massive alewife die-off in the spring of 1967. Tons of dead, rotting alewife fouled beaches near major population centers, prompting public outcry. In response, fishery managers in the US began stocking piscivorous Pacific salmonines, including chinook (Oncorhynchus tshawytscha), coho (Oncorhynchus kisutch), and steelhead salmon (Oncorhynchus mykiss), as biocontrol against alewife and in hopes of enhancing sport fishing opportunities. The

voracious and fast-growing exotic salmonines consumed vast quantities of alewife

(Stewart and Ibarra 1991) and supported an extremely economically valuable put-grow- take sport fishery (Gale 1987, Talhelm 1988), shifting the emphasis from commercial to sport fisheries in the US waters of the GL. The Pacific salmonine-alewife predator-prey system, however, proved unstable, experiencing fluctuations due to stressors such as climatic factors (e.g., cold water temperatures) and disease (e.g., bacterial kidney disease in salmonines).

In the late 1980s, around the same time sport fisheries were booming, new non- native species began appearing in the GL. These species, including zebra and quagga mussels (Dreissena polymorpha and D. bugensis, respectively), spiny and fishhook waterfleas (Bythotrephes longimanus and Cercopagis pengoi, respectively), Eurasian ruffe (Gymnocephalus cernuus), and round goby (Neogobius melanostomus), became predominantly integrated near the base of GL food webs, sometimes making their ecological impacts on fisheries difficult to isolate. These species and over 50 others, many of them from the Ponto-Caspian and Baltic regions of Europe (Grigorovich et al.

2003), were unintentionally introduced to the GL via the release of ballast water and sediments by interoceanic shipping vessels (Ricciardi 2006). A consequence of the increasing rate and scale of human commerce associated with globalization, access to the

GL by interoceanic shipping vessels, and hence these NIS, was made possible by the opening of the St. Lawrence Seaway in 1959. Since its opening, concern for the effects of unintentionally introduced NIS on the ecology of the GL has greatly increased (Mills et al. 1993).

Since the earliest use of the GL fishery by European colonizers, society’s emphasis has shifted from investment in natural resource extraction to investment in natural resource conservation in the highly modified GL ecosystem. The GL have been and continue to be affected by multiple anthropogenic stressors. Current threats include

Asian carp moving north up the Mississippi River toward Lake Michigan, emerging fish pathogens such as viral hemorrhagic septicemia virus (VHSv), the establishment and range expansion of additional ballast water NIS (particularly those native to regions of new trading partners, e.g., Asia), changes in lake levels, hypoxic conditions in portions of

Lake Erie, and regional climate change.

On the horizon for human influence on the ecology of the GL is a sizeable restoration effort that will seek to reverse some of the anthropogenic damage previously done to the lakes. This impending effort, known as the Great Lakes Restoration

Initiative, will involve numerous state and federal natural resource management agencies, all of whom will require scientific guidance to achieve their goals and mandates effectively (http://www.glrc.us/). It is expected that this GL restoration endeavor will be similar in scale to the massive Everglades restoration project (DeAngelis et al. 1998, http://www.evergladesplan.org/).

The ecology of the GL region has been and will continue to be heavily influenced by society’s choices about natural resource policy and management. In particular, the GL region has become a laboratory for the science and policy of NIS (Kelly et al. 2009). As described above, numerous NIS are established in the GL, some of which have had large ecological and economic impacts (Mills et al. 1993). This has led to substantial foment

and energy for research and policy work on NIS in the region. The research that follows in this dissertation seeks to build upon and contribute to this body of work.

1.4 Matching Scientific Inquiries on Ecological Topics to Pertinent Policy and

Management Questions

In 2006, the Ecological Society of America issued a document titled “Biological

Invasions: Recommendations for US Policy and Management” that stated, “The

Ecological Society of America is committed to assist all levels of government and provide scientific advice to improve all aspects of invasive-species management” (Lodge et al. 2006, italics added). This statement recognizes the essential role for ecologists in guiding natural resource policy and management pertaining to biological invasions.

There remain numerous open questions regarding biological invasions. Additional, and focused, scientific inquiry is required to find solutions and discover options that are available for dealing with invasions (Figure 1.1C). Thus, to keep this commitment, some ecologists must engage in research suited to providing scientific guidance on specific policy and management issues. This dissertation is an effort to apply ecological understanding, as well as to develop the necessary intellectual tools for connecting ecological and economic analysis, to conduct this sort of novel and highly relevant research.

This type of effort is needed because there is often a mismatch between critical questions regarding biological invasions and the scientific data that are available to address these questions. If oriented properly and conducted at the appropriate level of detail, ecological research can provide relevant information for addressing policy and 14

management questions. One of the intellectual challenges of this type of research is to seek out and apply the scientific tools appropriate for addressing relevant questions.

1.5 Dissertation Outline

One of the principal goals of scientific inquiries regarding invasive species is to predict the identity of species that are likely to become invasive if introduced (Kolar and

Lodge 2002, Lodge et al. 2006, Keller et al. 2007). Retrospective historical analysis of past invasions is an information-rich way to make predictions about the identity and impacts of future invasive species (Kolar and Lodge 2001, Kolar and Lodge 2002, Leung et al. 2002). In Chapter Two, I perform a retrospective analysis on the introduction pathways of freshwater NIS in the GL. My analysis shows that ballast water release is highly likely to introduce new freshwater NIS to North America, whereas a separate pathway---commerce in living organisms---tends to introduce NIS already established elsewhere in North America. Moreover, it is primarily a matter of time before novel NIS established in the GL appear in other North American waterways. I also investigate whether the ecological impacts of these species are related to their pathway of introduction. I find that there is an apparent relationship between pathways and impacts, but that further study of species-specific impacts is needed to verify this preliminary conclusion. This research provides a better understanding of which pathways bring which species to the GL and the impacts that often result from NIS introduced by particular pathways. This understanding could inform policies to limit future introductions.

Given the importance of ballast water release in bringing novel species to the GL, in Chapter Three I use an established technique, structured expert judgment (SEJ), in a novel application to estimate the economic impacts of ecological changes caused by ballast water NIS. Well-known direct impacts of these species include invasive zebra and quagga mussels clogging the pipes of power plant cooling systems in the GL region, necessitating costly maintenance and retrofitting (O’neill 1996). These additional operating costs are passed on to energy consumers, affecting nearly every business and household in the GL region. The zebra and quagga mussel invasion has also had indirect impacts on the GL as the filter-feeding bivalves consume large quantities of photoplankton. This has significantly increased the proportion of primary production in the GL that is drawn to the benthos (Vanderploeg et al. 2001). This change has in turn altered the energy flow in the pelagic food web of the lakes, reducing sport and commercial fish production in some instances (Mills et al. 2003). Such food web changes that alter sport and commercial fishing can have repercussions for the entire regional economy as revenues from these sectors decline (Lupi et al. 2003, Perrings et al.

2002).

When an issue affects a large proportion of society and when policy and management decisions on the issue must be made, addressing the issue in a way that reckons the implications of alternative policies or management activities in common units that are widely understood is often an effective strategy (Daly and Farley 2004). In most cases, the common units of reckoning are monetary (e.g., dollars). The conversion of the ecological impacts of invasive species or of the environmental effects of alternative management plans into dollars is a pragmatic way to account for the value that society

ascribes to a wide variety of goods and services, including those affected by biological invasions, in terms that are widely understood and which can be benchmarked against other societal issues and concerns. Chapter Three takes this type of bio-economic approach, employing SEJ to estimate the current annual economic impacts of ballast water NIS in the GL. This study concludes that the economic impacts of ballast water species are large, but also uncertain. Nevertheless, policies that aim to reduce the likelihood of additional invasions via this pathway appear to be economically justifiable.

Knowledge of species dispersal is crucial to understanding the distribution and abundance of biota in the environment. Species may establish populations in patches of suitable habitat, but individuals must first disperse to such suitable patches. Therefore, understanding species dispersal is necessary to understand population spread and range expansion (Kot et al. 1996, Clark et al. 2003). Thus, the study of species dispersal has long been an important topic in ecology (Skellam 1951, Huffaker 1958, Howe and

Smallwood 1982). Chapters Four and Five of my dissertation contribute to this branch of research by considering how human movements of recreational boats and trailers move

AIS throughout the landscape. It is necessary to study boats and trailers instead of, for example, bird droppings or other natural dispersal mechanisms because it has become clear that the most important vectors for the spread of AIS are anthropogenic (Johnson et al. 2001, Hastings et al. 2005).

As AIS in the GL region spread to other waterways, they bring with them ecological and economic impacts, motivating efforts to reduce their rate of spread.

Chapter Four is one of the first scientific studies to investigate what species are actually being transported via the overland movement of recreational boats on trailers. This

chapter describes my research intended to inform efforts to slow AIS spread. I collected data on the type and volume of aquatic species being transported by the overland movement of recreational boats and trailers. I also tested the efficacy of multiple methods for removing aquatic NIS from recreational boats and trailers. I found that visual inspection and hand removal is highly effective in removing the nuisance macrophyte Myriophyllum spicatum, but that high-pressure washing is needed to effectively remove small-bodied organisms, including the exotic predatory zooplankter

Bythotrephes longimanus.

In Chapter Five I go beyond the tactics of how to clean boats and consider the strategic placement of inspection and boat-cleaning stations on the landscape. My results show that a common predictive modeling approach (gravity modeling) is limited in its ability to predict which uninvaded lakes are likely to be invaded next. Therefore, the lakes that cleaning stations should protect cannot be identified. Instead, it appears that placing cleaning stations to block the transport of invasive propagules away from invaded lakes is generally more likely to reduce landscape-level spread than protecting uninvaded lakes.

Aquatic NIS are only one of many environmental and cultural factors that affect ecosystems and, in turn, affect societal interactions with the natural environment. To put the importance of aquatic NIS in context with other potential drivers of change in GL fisheries over the next two decades, I report the results of interviews with experts on GL fisheries in Chapter Six. In these interviews, I asked experts to predict changes in GL fisheries and to identify the most likely drivers of the changes they predicted. This study

reveals that changing cultural interests are the main reason for expected declines in GL fisheries, but that NIS are the predominant environmental driver of change.

Thus, in this dissertation, I provide case studies of real world problems pertaining to biological invasions in the GL region. I selected cases for which scientific inquiry had strong potential to improve the policy and management of invasive species. I report on the findings of my research and its implications for these real world problems.

Furthermore, I contend that the pursuit of such knowledge, in a form that is relevant to management and policy, contributes to the science of ecology by pushing its boundaries in ways that may benefit the science more broadly. For example, in seeking to assess the damage from invasive species across an entire region, my research introduces an approach that may be used to assess the effects of a variety of large-scale ecological perturbations. Similarly, by trying to predict which uninvaded lakes will be invaded in the near future, my work demonstrates the value of retrospective analyses in ecology to test our ability to predict future environmental change.

CHAPTER 2:

THE LAURENTIAN GREAT LAKES ARE A FRESHWATER INVASION

BEACHHEAD: PATHWAYS OF NONINDIGENOUS SPECIES INTRODUCTION

PREDICT PRIOR DISTRIBUTION, SUBSEQUENT SPREAD, AND POTENTIAL

IMPACTS1

2.1 Abstract

Biological invasions alter ecosystems and reduce societal welfare. Resources to manage invasions are limited, and efforts to prevent the arrival and establishment of new invaders are often the most cost-effective management approach to prevent future damage. Effective prevention, however, requires knowledge that is rare: how introduction pathways affect the process and consequences of invasions. Using the

Laurentian Great Lakes (GL) as a case study, we investigated the relationships between different pathways of introduction and (a) the prior global distribution of freshwater species introduced; (b) the likelihood of freshwater species to spread beyond the GL basin once they are established first there; and (c) the ecological and economic impact of nonindigenous freshwater species. We focused on two categories of pathways: shipping and commerce in living organisms (e.g., horticulture and pet industries). Among other

1 The publication status of this chapter is: Rothlisberger, J.R. and D.M. Lodge. The Laurentian Great Lakes are a freshwater invasion beachhead: pathways of nonindigenous species introduction predict prior distribution, subsequent spread, and potential impacts. Diversity and Distributions (in review). 20

ideas, we tested the hypothesis that the shipping pathway makes the GL a beachhead of invasions of freshwater organisms to North America: we predicted that ship-related introductions in the GL are often first-time introductions to North America, and that from the GL, the same species often colonize many additional North American freshwater ecosystems. Results of our analyses of data on the global distribution of species pre- and post-introduction to the GL were consistent with the beachhead hypothesis. We found that the distribution of species prior to their discovery in the GL was related to introduction pathway, with ballast water releases more likely than other pathways to introduce new freshwater species to North America (85% of ballast water introductions).

In contrast, commerce in living organisms was most likely to introduce to the GL freshwater species already established in North America (90% of species introduced by commerce in living organisms). Pathway, however, was a poor predictor of current distribution. Instead, time since discovery in the GL was the best predictor of current distribution: 88% of species discovered more than 74 years ago are now dispersed beyond the GL basin. To examine the relationship between pathway and impacts, we reviewed the scientific literature on species introduced to the GL. Results suggest that species introduced via ballast water are more likely to prey on native species, while species introduced via commerce in living organisms are more likely to impair recreational opportunities. Our results indicate that once a nonindigenous freshwater species is established in the GL it is only a matter of time before it appears in North

American waterways beyond the GL basin.

2.2 Introduction

The establishment of species beyond their native range is a large and growing environmental problem (Vitousek et al. 1997, Mack et al. 2000). Some nonindigenous species become invasive, threatening native species biodiversity and ecosystem services

(Wilcove et al. 1998, Sala et al. 2000), and causing large economic losses (Perrings et al.

2005, Pimentel et al. 2005). Increasingly globalization of commerce is the main driver of nonindigenous species introductions (Levine and D’Antonio 2003, Ruiz and Carlton

2003). The two principal categories of pathways by which humans transport species are

(1) unintentional transport of organisms while conducting other activities and (2) commerce in live organisms, where the purpose of the activity is to move specific organisms (Lodge et al. 2006).

Pathways have been proposed as the appropriate target for policy and management aimed at reducing future invasions (Lodge et al. 2006, Hulme et al. 2008).

Management and policy that focus on pathways can simultaneously reduce the probability of establishment of the multiple species that may be transported in a pathway currently and in the future. With pathways as the fundamental unit for policy and management intended to reduce the introduction of alien species, practical actions to implement such policies are relatively straightforward: reduce the volume of traffic or the number of viable organisms in the pathway, at least of species that are likely to be invasive. Despite recent emphasis on pathway-based policy and management and their apparent benefits, there has been little study of the possible relationships between introduction pathways and how invasions proceed, including the eventual ecological impacts of the species they introduce (Hulme 2008). In other words, do pathways, as a

function of the species they are responsible for introducing, differ with respect to the magnitude or type of threat they pose to ecoystems and human welfare (e.g., ecosystem goods and services, infrastructure)? Knowing how the origin, spread, and harm caused by invasive species are related to pathways of introduction may help to inform policy response to particular pathways.

Here, we use biological invasions of the Laurentian Great Lakes (GL) by freshwater species that are not native to North America as a case study to investigate how the impacts and subsequent spread of nonindigenous species are related to pathways of introduction. The GL offer an excellent opportunity for such an investigation because they contain many nonindigenous freshwater species that have a range of ecological impact types and severities (Mills et al. 1993b, Kelly et al. 2009, Chapter 3). Moreover, these species have been introduced via numerous pathways, including authorized release

(e.g., fish stocking), commerce in living organisms (e.g., aquarium dumping), ballast water release, and solid ballast dumping (Mills et al. 1993b, Ricciardi 2006). In recent years, ballast water release has been the most important pathway for new introductions

(Ricciardi 2006). Nonindigenous freshwater species in the GL region are relatively well- studied and pertinent policy and management efforts to limit additional damage are on- going (Mills et al. 1993b, NRC 2008). Thus, the GL have been and will likely continue to be an important nexus for science, policy, and management of nonindigenous freshwater species.

The GL may also be a beachhead for the invasion of freshwater species into the rest of North America. In other words, as the largest freshwater ecosystem on the North

American continent, the lakes may the site of initial introduction for nonindigenous

freshwater species, from which they can spread readily to other freshwaters on the continent. The recent spread of zebra and quagga mussels, whose original colonization site in North America was the GL, to Lake Mead and other waterways as far west as Utah and California, provides anecdotal evidence of this possible phenomenon (Stokstad

2007), which we refer to here as the beachhead hypothesis. The possibility that the GL are an initial harbor for nonindigenous freshwater species that may eventually become widespread in North America bears further investigation given that 63 of the 391 nonindigenous freshwater species currently established in North America were first discovered in the GL (http://nas.er.usgs.gov/). A fraction of these 63 species have already spread beyond the GL basin and the rest could perhaps follow.

We address three specific questions and associated hypotheses. First, is there a relationship between a species’ pathway of introduction to the GL and its distribution prior to being discovered in the GL? That is, are some pathways more likely than others to bring to the GL species that are novel to North America? We hypothesized that, of the four pathways we considered, ballast water release would be more likely than the others to introduce species to the GL that were not previously found in North America than would be expected by random chance.

Second, for freshwater species that were first discovered in North America in the

GL, we tested whether their pathway of introduction to the GL was related to their current North American distribution (i.e., contained within the GL or spread beyond the

GL). We also quantified which combination of explanatory variables (i.e., years since discovery in the GL, taxonomic identity, endemic region, first GL invaded), along with introduction pathway, could most accurately classify the current distribution species.

Identifying a relationship between pathway of introduction to the GL and spread beyond the GL would allow for targeted efforts to prevent the spread of species most likely to expand their range beyond the GL. We hypothesized that introduction pathway alone would be insufficient to predict current North American distribution accurately, but thought that by combining introduction pathway with other variables, we would be able to reliably predict current distribution.

Third, we asked whether a species’ pathway of introduction to the GL predicted the type and magnitude of impacts caused by the species in the GL or other waterways.

To do this, we reviewed the scientific literature on ecological and economic impacts of species introduced to the GL via ballast water release and commerce in living organisms.

Knowing how pathways are related to impacts would allow managers and policy-makers to direct their efforts at the pathways most likely to introduce damaging invaders in the future. Because many of the species introduced by ballast water release are planktonic or small benthic organisms, we hypothesized that the ecological impacts of species from ballast water would be diffuse food web alterations, in contrast to the more taxonomically diverse set of species introduced by commerce in living organisms, whose impacts we expected to be more varied.

2.3 Methods

2.3.1 Pathway and distribution prior to discovery in Great Lakes

We extracted a list of 95 established nonindigenous freshwater species in the GL from the database of all 184 established nonindigenous species known from the GL

(http://www.glerl.noaa.gov/res/Programs/ncrais/glansis.html). From the same database, for each species, we recorded the taxonomic category, year of discovery in the GL, endemic region, and pathway of introduction to the GL. The 95 species on which we focus are entirely freshwater (i.e., wetland plants are not included), and are nonindigenous to North America (Appendix A). Species in the database are grouped into the broad taxonomic categories of plant, benthic algae, phytoplankton, zooplankton, benthic crustacean, mollusk, annelid, other invertebrate, fish, and virus. Pathways of introduction in the database are categorized as ballast water release, solid ballast dumping, deliberate release (i.e., authorized fish stocking), aquarium dumping, bait release, unintentional release, canals, and unknown. For the purposes of our study, we refer to deliberate release as authorized release and contrast it with ballast water release, solid ballast dumping, and commerce in living organisms, a composite category containing species in the aquarium dumping, bait release, and unintentional release categories. The canal pathway is irrelevant to our study because no freshwater species that are not native to North America have been introduced to the GL via canals. We added to this database whether, prior to their discovery in the GL, each species was not yet known in North America or already established as a nonindigenous species elsewhere in North America. Information on North American distribution prior to GL discovery came mainly from species-specific sources cited in Mills et al. (1993), but some information also came from sources cited in the species accounts available through the

United States Geological Survey’s Nonindigenous Aquatic Species database

(http://nas.er.usgs.gov/).

2.3.2 Pathway and spread beyond the Great Lakes

To investigate relationships between invasion pathway and whether an invader has spread beyond the GL, we analyzed data for the subset of NIS whose first North

American occurrence was in the GL (63 of the 95 species considered above). These species were categorized based on their current distribution (i.e., confined to the GL basin versus beyond the GL basin). For this analysis, we added information to our database on the current North American distribution of each of these 63 species from the USGS NAS database (http://nas.er.usgs.gov/) (Appendix A).

We used recursive partitioning (a nonparametric statistical pattern-finding technique; De’ath and Fabricius 2000) to test whether introduction pathway was consistently related to current distribution. The response categories are the two options for current North American distribution relative to the GL basin (within or outside the GL basin). Other potential explanatory variables that we included in this analysis are year of discovery in the GL, lake where originally discovered, endemic region, and taxonomic category. The decision tree created by a recursive partitioning analysis is a result of balancing accurate classification of the training dataset (i.e., a sufficiently complex tree could correctly classify 100% of the observations in the training dataset) against parsimony sufficient to make the tree robust in accurately classifying new observations

(i.e., the decision tree is not overfit to the training data). Testing the predictions of a decision tree against the known response values of an independent dataset is the ideal method for assessing whether or not this balance has been achieved, but this approach is not advised when the dataset is <100 species, as is the case here (Venables and Ripley

2002). Instead, we used leave-one-out cross-validation (LOOCV) to assess the accuracy

and robustness of our decision trees. This technique entails running the algorithm to fit the decision tree with n-1 of the observations in the dataset.

For this analysis, we considered all possible combinations of our five explanatory variables, producing a total of 31 candidate models containing from one to five explanatory variables, to find the model(s) with the lowest combined misclassification rate. We calculated misclassification rate as the number of species wrongly predicted to be in a category divided by the total number of species predicted to be in that category.

For example, if a model predicted 55 species to be confined to the basin, but seven of those 55 had already spread beyond the basin the ‘in basin’ misclassification rate would be 13% (i.e., 7/55).

2.3.3 Pathway and impacts

To assess the relationship between pathway and impacts of species introduced to the GL, we conducted a review of the scientific literature on the environmental and economic impacts of species transported by two important pathways: ballast water release and commerce in living organisms. We searched databases of peer-reviewed publications (e.g., Web of Science), gray literature (e.g., http://graylit.osti.gov/), and books (e.g., WorldCat). In our searches we used scientific (including synonyms) and common names of the species as keywords. Our literature search included the years 1940 through 2008.

Since the opening of the St. Lawrence Seaway in 1959, 54 nonindigenous freshwater species attributed to ballast water release have been discovered in the GL. For these 54 species, we found published information on only 23. For these 23 species, we

found fewer than five published studies on all but eight species, and many of those studies provided very little evidence for the impacts reported. Because of the paucity of information for most species, we report results only for eight species, which we have lumped into seven taxonomic headings (i.e., zebra (Dreissena polymorpha) and quagga

(D. bugensis) mussels are combined). For these eight species, we discovered 138 sources, from which we recorded any mention of impacts. We grouped the reported ecological and economic impacts into ten categories (Table 2.1).

The commerce in living organisms pathway has resulted in the introduction of 21 species into the Great Lakes. For these 21 species we found published information on all but three. However, for the 18 species with information available, we found five or more studies reporting ecological, economic, or human health impacts for only seven species.

Because of the lack of information on most species, we report results for only seven species. These seven species were predominantly brought to North America by the aquarium or watergarden trades. We summarize the environmental impacts reported in the 133 publications regarding these species using the same ten impact categories we used for ballast water species (Table 2.2). We tested for a statistically significant difference in the distribution of impacts across categories between the two pathways we studied (e.g., ballast water and commerce in living organisms) using a chi-square contingency table analysis.

TABLE 2.1.

SUMMARY OF SCIENTIFIC LITERATURE DOCUMENTING ECOLOGICAL AND

ECONOMIC IMPACTS ATTRIBUTABLE TO BALLAST WATER SPECIES

ESTABLISHED IN THE GREAT LAKES

Species Competition with natives Predation/parasitism on natives Other food web disruption (e.g.,poor food resource for natives) Habitat modification Hybridization/losses in native biodiversity Risks to human health Taste and odor problems in water supply Recreational/aesthetic impairment Fisheries impairment Infrastructure fouling Dreissenid mussels (Dreissena polymorpha 20a 6b 5c 32d 4e 3f 2g 4h 2i 8j and Dreissena rostriformis bugensis) Round Goby (Apollonia 4k 7l 6m 4n

melanostomus) Eurasian Ruffe (Gymnocephalus 8o 2p 1q

cernuus) New Zealand mud snail (Potamopyrgus 2r

antipodarum) Spiny waterflea (Bythotrephes 3s 11t 7u 1v

longimanus) Fish-hook waterflea 4w 2x 2y 1z 1aa (Cercopagis pengoi) Non-native amphipod (Echinogammarus 2ab

ischnus)

Note: Numbers in cells indicate the number of unique studies in the primary literature that document a particular impact of a particular species. This number is not necessarily a metric for the severity or importance of the impact, but represents the range of impacts attributable to ship-borne species and the level of scientific attention to each

. 30

References to Table 2.1: a Kryger and Riisgard 1988, Graham et al. 1992, MacIssac et al. 1992, Dermott and Munawar 1993, Mills et al. 1993a, Dahl et al. 1995, Schloesser et al. 1996, Dermott and Kerec 1997, Nalepa et al. 1998, Trometer and Busch 1999, Johannsson et al. 2000, Landrum et al. 2000, Nalepa et al. 2000, O’Gorman et al. 2000, Lozano et al. 2001, Pothoven et al. 2001, Vanderploeg et al. 2001, Beeton and Hageman 2002, Thorp and Casper 2002, Raikow 2004 b Holland 1993, Nicholls and Hopkins 1993, Fahnenstiel et al. 1995, Lavrentyev et al. 1995, MacIsaac et al. 1995, Munawar et al. 1999 c Hamilton et al. 1994, French and Bur 1996, Hoyle et al. 1999, Pothoven et al. 2001, Pothoven and Madenjian 2008 d Reeders and Bij de Vaate 1990, Effler and Siegfried 1994, Stewart and Haynes 1994, Johengen et al. 1995, Lowe and Pillsbury 1995, Nalepa and Fahnenstiel 1995, Skubinna et al. 1995, Arnott and Vanni 1996, Krieger et al. 1996, Jones et al. 1997, Ricciardi et al. 1997, Madenijan et al. 1998, Stewart et al. 1998, Bailey et al. 1999, Gonzalez and Downing 1999, Knapton and Petrie 1999, Millard et al. 1999, Nicholls et al. 1999, Petrie and Knapton 1999, Bially and MacIsaac 2000, Johannsson et al. 2000, Makarewicz et al. 2000, MacLennan et al. 2000, Marsden and Chotkowski 2001, Cobb and Watzin 2002, Kolar et al. 2002, Vanderploeg et al. 2002, Ratti and Barton 2003, Beekey 2004, Marsden 2004, Reed et al. 2004, Haynes et al. 2005 e Schloesser and Nalepa 1994, Nalepa et al. 1996, Schloesser et al. 1996, Strayer 1999 f Bruner et al. 1994, Mazak et al. 1997, Hogan et al. 2007 g Lange and Wittmyer 1996, Vanderploeg et al. 2001 h Fahnenstiel et al. 1995, MacIsaac 1996, Vanderploeg et al. 2001, Higgins et al. 2005 i Marsden and Robilliard 2004, Higgins et al. 2005 j Dermott and Munawar 1993, LePage 1993, OTA 1993, Hushak 1995, O'Neill 1996, Lewis et al. 1997, Pimentel et al. 2000, Connelly et al. 2007 k Dubs and Corkum 1996, French and Jude 2001, Janssen and Jude 2001, Lauer et al. 2004 l Chotkowski and Marsden 1999, Kuhns and Berg 1999, Weimer and Sowinski 1999, French and Jude 2001, Nichols et al. 2003, Steinhart et al. 2004a, Janssen et al. 2007 m Jude et al. 1995, Jude 1997, Somers et al. 2003, Steinhart et al. 2004b, Truemper et al. 2006, Yule et al. 2006b n Petrie and Knapton 1999, Morrison et al. 2000, Yule et al. 2006a, Hogan et al. 2007 o Ogle et al. 1996, Savino and Kolar 1996, Sierszen et al. 1996, Fullerton et al. 1998, Gunderson et al. 1998, Ogle 1998, Kolar et al. 2002, Bronte et al. 2003 p DeSorcie and Edsall 1995, Selgeby 1998 q Leigh 1998 r Zaranko et al. 1997, Strayer 1999 s Branstrator 1995, Francis et al. 1996, Johannsson et al. 1999 t Lehman 1991, Lehman and Caceres 1993, Vanderploeg et al. 1993, Burkhardt and Lehman 1994, Makarewicz et al. 1995, Yurista and Schulz 1995, Johannsson et al. 1999, Madenjian et al. 2002, Yan et al. 2002, Barbiero and Tuchman 2004, Pangle et al. 2007 u Barnhisel 1991, Bur and Klarer 1991, Barnhisel and Harvey 1995, Coulas et al. 1998, Parker et al. 2001, Pothoven and Vanderploeg 2004, Stetter et al. 2005 v Ojaveer et al. 2001 w Francis et al. 1996, Benoit et al. 2002, Bushnoe et al. 2003, Laxson et al. 2003 x MacIsaac et al. 1999, Ojaveer et al. 2001 y Barnhisel 1991, Barnhisel and Harvey 1995 z Ojaveer et al. 2001 aa MacIsaac et al. 1999 ab Dermott et al. 1998, Ratti and Barton 2003

TABLE 2.2.

SUMMARY OF SCIENTIFIC LITERATURE DOCUMENTING ECOLOGICAL AND

ECONOMIC IMPACTS ATTRIBUTABLE TO NONINDIGENOUS FRESHWATER

SPECIES INTRODUCED BY COMMERCE IN LIVING ORGANISMS TO THE

GREAT LAKES

Competition with natives Other food web disruption (e.g.,poor food resource for natives) Habitat modification Risks to human health Taste and odor problems in water supply Recreational/aesthetic impairment Fisheries impairment Infrastructure fouling Species Predation/parasitism on natives Hybridization/losses in native biodiversity Eurasian watermilfoil (Myriophyllum 16a 3b 19c 2d 3e 13f 1g 3h

spicatum) Water chestnut 6i 7j 10k 2l (Trapa natans) Asiatic clam 4m 2n 7o 11p 2q (Corbicula fluminea) Non-native waterflea 4r 4s (Daphnia lumholtzi) Non-native oligochaete 3t 3u (Branchiura sowerbyi) Largemouth bass virus (LMBV) (Ranavirus 6v

sp.) European frogbit (Hydrocharis morsus- 4w 1x

ranae)

Note: Numbers in cells indicate the number of unique studies in the primary literature that document a particular impact by a particular species. This number is not necessarily a metric for the severity or importance of the impact, but represents the range of impacts attributable to each species and the level of scientific attention to each. Many of the documented impacts were studied in the Great Lakes, but some were studied and recorded elsewhere, outside of the Great Lakes basin.

References to Table 2.2: a Lind and Cottam 1969, Nichols and Mori 1971, Oglesby and Vogel 1976, Bayley et al. 1978, Aiken et al. 1979, Carpenter 1980, Godmaire and Planas 1986, Smith and Barko 1990, Madsen et al. 1991, Bowman and Mantai 1993, Valley and Newman 1998, Boylen et al. 1999, Whyte and Francko 2001, Chase and Knight 2006, Trebitz and Taylor 2007, Riis et al. 2009 b Keast 1984, Toetz 1997, Olson et al. 1998 c Nichols and Keeney 1976, Aiken et al. 1979, Prentki et al. 1979, Barko and Smart 1980, Dvorak and Best 1982, Landers 1982, Keast 1984, Newroth 1985, Pardue and Webb 1985, Godmaire and Planas 1986, Smith and Adams 1986, Kilgore et al. 1989, Engel 1995, Dibble and Harrel 1997, Unmuth et al. 2000, Cheruvelil et al. 2001, Valley and Bremigan 2002a, Valley and Bremigan 2002b, Linden and Lehtiniemi 2005 d Moody and Les 2002, Moody and Les 2007 e Smith et al. 1967, Aiken et al. 1979, Bates et al. 1985 f Smith et al. 1967, Bayley et al. 1978, Aiken et al. 1979, Wile et al. 1979, Davis and Brinson 1983, Dearden 1983, Mikol 1985, Miller and Trout 1985, Newroth 1985, Andrews 1986, Lillie 1986, Madsen et al. 1989, Eiswerth et al. 2000 g Smith et al. 1967 h Smith et al. 1967, Aiken et al. 1979, Dearden 1983 i Winne 1950, Kiviat 1987, Kiviat 1993, Groth et al. 1996, Hummel and Kiviat 2004, Hummel and Findlay 2006 j Bickley and Cory 1955, Connor 1978, Tsuchiya and Iwakuma 1993, Caraco and Cole 2002, Nieder et al. 2004, Hummel and Findlay 2006, Goodwin et al. 2008 k Gwathmey 1945, Winne 1950, Beaven 1955, Beaven 1959, Countryman 1978, Besha and Countryman 1980, Boguki et al. 1980, Gangstad and Cardarelli 1990, Bonopartis 2001, Giddy 2003 l Hummel and Kiviat 2004, Smyth et al. 2009 m Hakenkamp and Palmer 1999, Vaughn and Hakenkamp 2001, Sousa et al. 2005, Yeager et al. 1994 n McMahon 1991, Strayer 1999 o Narbonne et al. 1999, Tran et al. 2001, Cataldo et al. 2001a, Cataldo et al. 2001b, Achard et al. 2004, Cantanhede et al. 2008, Sousa et al. 2008b p Asmus and Asmus 1991, Phelps 1994, Johnson and McMahon 1998, Strayer 1999, Crooks 2002, McMahon 2002, Gutierrez et al. 2003, Cherry et al. 2005, Cooper et al. 2005, Werner and Rothhaupt 2007, Sousa et al. 2008b q Darrigran 2002, Sousa et al. 2008a r Kolar et al. 1997, Johnson and Havel 2001, Dzialowski et al. 2003, Acharya et al. 2006 s Havel et al. 1995, Work and Gophen 1995, Kolar and Wahl 1998, Havel and Graham 2006 t Yokoyama et al. 1991, Yokoyama et al. 1993, Yokoyama et al. 1995 u Kikuchi and Kurihara 1982, Wang and Matisoff 1997, Matisoff et al. 1999 v Grizzle et al. 2002, Woodland et al. 2002, Grizzle and Brunner 2003, Grant et al. 2005, Inendino et al. 2005, Schramm and Davis 2006 w Lumsden and McLachin 1988, Catling and Porebski 1995, Catling et al. 2003, Houlahan and Findlay 2004 x Catling et al. 2003

2.4 Results

2.4.1 Pathway and distribution prior to discovery in Great Lakes

Pathways of introduction differ with respect to the number of freshwater species they have introduced. The ballast water release category has introduced the most nonindigenous freshwater species of any pathway to the GL (54 spp.), followed by commerce in living organisms (21 spp.). Authorized release and solid ballast are tied for the introduction of the fewest species (6 spp. each), with a catch-all category of unknown containing the next smallest number of species (8 spp.) (Figure 2.1).

A significant relationship existed between introduction pathway and whether a species was first discovered in North America in the GL basin or outside the GL basin (χ2

= 46.7, d.f. = 4, p < 0.001), with species introduced via ballast water, solid ballast, and

unknown means more likely to have been first discovered in the GL. Species introduced

by authorized release and commerce in living organisms were more likely to have first

occurred elsewhere in North America (Figure 2.1).

Ballast water release introduced the most diverse range of taxa directly into the

GL (i.e., initial introductions in North America) of any pathway we considered, drawing

species from 9 of 10 possible taxonomic categories of nonindigenous freshwater species

in the GL (Figure 2.2). Vascular plants are the one type of taxa that ballast water release

is not known to have introduced to the GL.

21 6 6 54 8 1.0 1st in GL 1st outside GL

0.8

0.6

0.4

Proportion of species 0.2

0.0

Unknown

SolidBallast BallastWater

AuthorizedRelease

Commerce LivingOrg.

Pathway of Introduction to GL

In GL Basin Outside GL Basin

Authorized Release 14

CommerceUnauthorized in Living Release Organisms 14

7 0

Ballast Water 14 7

Solid Ballast 14 Number Number of species 7

Unknown 14 7 0 Fish Plant Virus Annelid Mollusk Zooplankton Benthic Algae Benthic Phytoplankton OtherInvertebrate Benthic Crustacean Benthic

Taxonomic Category

2.4.2 Pathway and spread beyond the Great Lakes

Decision tree models predicted well which species were confined to the GL basin, and predicted less well which species have spread beyond the basin. The average misclassification rate in predicting which species are confined to the GL basin was 17% for all 31 possible candidate models, and as low as 14% for the best single model. On the other hand, the average misclassification rate in predicting which species have spread beyond the basin was 41% for all 31 possible candidate models, but was as low as 14% for the best single model.

The models with the lowest average misclassification rates were those with a single explanatory variable: time since discovery (Figure 2.3). Adding additional variables increased misclassification rates. The decision tree using only time since discovery had the lowest misclassification rate both for species remaining confined to the

GL basin (14%) and species that had dispersed beyond the GL basin (14%). The prediction of this best model was that if a species was discovered in the GL 74 or fewer years ago, it would still be confined to the GL basin (Figure 2.4). Otherwise, the model predicted that the species had spread beyond the GL basin. Of the 55 freshwater nonindigenous freshwater species first discovered in North America in the GL 74 or fewer years ago, 48 are still confined to the basin. Of the 8 nonindigenous freshwater species first discovered in North America in the GL more than 74 years ago, all but one have established populations beyond the basin.

Because years since discovery in the GL was the only explanatory variable in the best model, because nearly all the candidate models that performed relatively well also

60 Predicted Out of GL Basin Predicted In GL Basin 50

Misclassification Rate . 10

SE Misclassification Rate ± 0 0 1 2 3 4 5

Number of Variables

included this variable, and because models without this variable did not perform well, we further investigated this variable. We plotted species’ North American distribution against time since discovery (Figure 2.5). We were particularly interested in learning more about the seven species classified as being confined to the GL basin that have actually spread beyond the GL basin. The difference between ‘in basin’ versus ‘out of basin’ misclassification rates suggests that knowing which species will escape the basin in less than the 74 years is more challenging than predicting which species will remain confined to the basin longer than expected. Six of the seven species that have escaped

Years Since Discovery ≤ 74

In GL Basin Outside GL Basin 48/7 1/7

Outside GL Basin

Ballast water Solid ballast Authorized release Commerce Unknown

GL Basin

North Distribution Current American

74 0 20 40 60 80 100 120 140

Years Since Discovery in Great Lakes

the GL basin less than 74 years ago were introduced via ballast water release (Figure

2.5): Apollonia (Neogobius) melanostomus, Dreissena bugensis, Dreissena polymorpha,

Bythotrephes longimanus, Ripistes parasita, Eubosmina coregoni), with the introduction vector of the seventh unknown, but is also likely to be ballast water as the species is a planktonic copepod (Neoergasilus japonicus). The ballast water release pathway

therefore seems to be positively associated with a propensity to disperse beyond the GL

basin faster than predicted by our best model.

As only one species that was discovered in the GL more than 74 years ago has not spread beyond the basin, it appears that range expansion beyond the GL basin is only a matter of time for species that first appear in North America in the GL.

2.4.3 Pathway and impacts

Although the freshwater nonindigenous species of the GL are among the highest profile and best-studied suites of nonindigenous species in the world, there is a lack of research on the impacts they have on native ecosystems and the human economy. For the species whose impacts have been studied, more than half (6 of 8, 75%) of those introduced via ballast water release are benthic animals (fish, gastropods, or arthropods)

(Table 2.1), whereas nearly half (3 of 7, 43%) of those introduced by commerce in living organisms are vascular plants (Table 2.2). Probably because of these taxonomic differences, the impacts associated with each set of species differ in their mechanisms, but were fairly similar as to the broad categories of impacts covered and the relative frequency with which impacts were reported from some impact categories. Studies on species from both types of pathways frequently reported competition and habitat modification (Tables 2.1, 2.2). One exception to these similarities was that species introduced by commerce in living organisms were more often reported to interfere with recreational opportunities and aesthetics, primarily because of the dense stands of freshwater vegetation that interfere with swimming and boating and aesthetics. For ballast water species, studies frequently reported predation on natives, in contrast to few cases of such predation reported for species introduced via commerce in living organisms. Moreover, there were no reports of species from commerce in living

organisms causing taste and odor problems in the water supply, whereas two studies found this to be the case with ballast water species (Tables 2.1, 2.2). Because of these differences in impacts between the pathways, we found a statistically significant difference, with respect to their overall distribution of impacts, between pathways (χ2 =

28.5, d.f. = 9, p < 0.001).

2.5 Discussion

2.5.1 The Great Lakes as a beachhead

Ballast water release in the GL introduces novel freshwater species to the North

American continent. No other pathway brings species new to North America into the GL with similarly high frequency. Nearly one-sixth (63 of 391) of the nonindigenous freshwater species established in North America first appeared in the GL, a higher proportion than any other single freshwater ecosystem in North America (Ricciardi

2007). At least 14 of these 63 species introduced first to the GL have already spread to

North American freshwaters beyond the GL, and our results suggest that the GL will continue to serve as a beachhead for freshwater invasions throughout North America.

It appears that once a species is established in the GL, it is only a matter of time before it spreads beyond the GL basin. Our predictive models had low misclassification rates for identifying which species remain confined to the GL basin, indicating that it is highly unusual for a species that first appears in North America in the GL to spend more than 74 years in the lakes without escaping the basin and invading other freshwaters on the continent. International shipping introduced many of the novel nonindigenous

freshwater species in the GL after the opening of the St. Lawrence Seaway in 1959 and therefore most of these species were discovered in the GL less than 50 years ago. Given our prediction that species novel to North America escape the GL basin in 74 years or less, some portion of the 49 novel freshwater species still confined to the basin are likely to spread to other freshwater ecosystems in North America in the near future. Thus, ballast water release in the GL has created a situation where the GL are poised to be an even more important beachhead than they have been up until now for freshwater invasions throughout North America.

One assumption in this expectation, however, is that species in our database

(Appendix A) that were first discovered in North America in the GL and which are now found outside the GL basin were not introduced outside the basin by other introduction events, independent of the established populations of the species in the GL.

Nevertheless, whether the discovery of a species novel to North America in the GL indicates that the GL will serve as the beachhead for dispersal to other freshwaters in

North America or whether it is an early warning that some other independent introduction event will soon lead to the establishment of the species elsewhere in North America, our research suggests that the North American distribution of new freshwater species in the

GL does not remain restricted to the GL for long.

Some species have spread beyond the GL basin in substantially fewer years than the 74-year threshold of their discovery in the GL indicated by our best recursive partitioning model. To date, nearly all of the species that have spread beyond the GL basin in less than 74 years since their discovery there were introduced via ballast water release. It is possible that the species traits that enhance quick intra-continental dispersal

include some of the same traits that are related to inter-continental transport by ships. For example, the planktonic veliger larvae of zebra mussels (Dreissena polymorpha) are easily pumped into ballast tanks, and were also easily dispersed in moving water that connects the GL to other watersheds (e.g., the Chicago Sanitary and Ship Canal that connects the GL to the Mississippi River basin). Further study of how species traits relate to how much time elapses before a species escapes the GL basin could be fruitful.

We suggest that policies to reduce the likelihood of the introduction and establishment of additional nonindigenous freshwater species via ballast water release in the GL are important for the conservation of freshwater resources across the entire continent (Drake and Bossenbroek 2004, Drake and Lodge 2006). Some such policies that are in development or in the early stages of implementation include more stringent ballast water release regulations and improved on-board ballast water treatment technologies (e.g., ultraviolet irradiation) (NRC 2008). While promising, these initiatives will require close monitoring to quantify their efficacy (Costello et al. 2007).

Even as the risks of ballast water introductions of novel species are reduced, the risk of species already established in the GL spreading beyond the basin remains a threat.

Learning more about the anthropogenic pathways that aid the spread of these species throughout North America is an important area for future research that could help to limit spread (Chapter 4). Similarly, learning if there is a relationship between species traits and propensity to spread beyond the GL basin rapidly could help to target specific species currently confined to the GL for vigorous spread prevention efforts (Vander Zanden and

Olden 2008).

2.5.2 The Great Lakes as a melting pot

Pathways other than ballast water release, particularly commerce in living organisms has introduced to the GL, through accidental or intentional release, many nonindigenous freshwater species that were already established elsewhere in North

America. The types of nonindigenous freshwater species transported for trade in the GL region have received recent attention in the literature (Keller and Lodge 2007, Duggan et al. 2006), as has the propagule pressure to a portion of the GL ecosystem arising from one facet of commerce in living organisms, aquarium dumping (Cohen et al. 2007).

These studies have highlighted that numerous organisms are moved into and around the

GL region, in some cases with little attention paid to what exactly is being transported and sold (Keller and Lodge 2007). Furthermore, some of the species documented in trade, like the aquarium plant Egeria densa, a species known to be invasive in other parts of North America, are likely to have serious negative ecological impacts if they become established in the GL, an almost certain prospect given that hundreds of these plants are released into the GL ecosystem annually (Cohen et al. 2007). Thus, the GL are not just a beachhead, but also a melting pot for nonindigenous freshwater species already established in North America.

As there are 391 nonindigenous freshwater species documented in North America

(http://nas.er.usgs.gov/) and only 95 of these are currently found in the GL

(http://www.glerl.noaa.gov/res/Programs/ncrais/glansis.html), there exists the strong potential for new biological invasions in the GL to occur as species already found in

North America are introduced to the lakes. Our findings suggest that learning which of the nearly 200 freshwater species that are in North America, but not yet in the GL, may

pose the greatest threats to the GL could help to better protect the GL. Such forecasts would enable the development of appropriate early detection and rapid response programs tailored to the species posing the highest risk of invasion.

Canals are numerous in the GL region, connecting the GL watershed to other drainages (Ashworth 1986, Annin 2006), and therefore may also be an important vector of spread of species established in other North American freshwaters to the GL. Vice- versa, canals may also facilitate spread of species away from the GL to other parts of

North America, as was the case with zebra mussel (Rodriguez et al. 2005). In both these respects the connection of the GL to the Mississippi River drainage via the Chicago Ship and Sanitary Canal (CSSC) is of major importance. For instance, round goby (Apollonia

(Neogobius) melanostomus) have spread from the GL to colonize the Mississippi River drainage via the CSSC (Charlebois et al. 2001). We did not include canals in our analyses because, as of yet, no freshwater species that are not native to North America have invaded the GL through canals (Ricciardi 2006). A particularly worrisome set of nonindigenous freshwater species---Asian carp (e.g., bighead and silver carp,

Hypothalmichthys nobilis and H. molitrix, respectively), however, is on the cusp of doing so via the CSSC (Whitledge 2009, http://www.epa.gov/glnpo/invasive/asiancarp/).

Moreover, multiple species native to North America, but not to the GL basin have established populations in the GL after entering through canals (Mills et al. 1993b,

Ricciardi 2006). Thus, it appears that canals are likely to be an increasingly important vector for species to and from the GL in the future.

2.5.3 Pathways and impacts

Extensive information on the impacts of most nonindigenous freshwater species in the GL is not available in the scientific literature. For the species on which we did locate information, we assumed that the impacts that have been reported are representative of the kinds of impacts a species actually has in an ecosystem. We think this assumption valid because most often the impacts that are first studied scientifically are those that are most easily noticed and readily observed. From the information that is available, it is apparent that the types of impacts and their mechanisms are widely varied.

It also appears that, at least for the species that have been studied, there is a relationship between the pathway of introduction and the type of taxa that are introduced by that pathway. Thus, because taxonomic identity is related to the type of ecological impacts species have, prevailing pathways of introduction play a role in determining the ecological impacts that arise from biological invasions.

For the two pathways we evaluated in our literature review we found some similarities, but more differences in the types of impacts attributable to the species each had introduced. We hypothesized that food web effects would be reported for ballast water species and that the ecological impacts of species from commerce in living organisms would be more diverse. We found that both sets of species were implicated in food web effects with similar frequency, including competition and habitat modification.

Ballast water species, however, were reported to engage in direct predation on native species more often than species from commerce in living organisms. Many of the impacts reported for ballast water species arise from introduced benthic species

(dreissenid mussels, round goby, and non-native amphipod) and a few predatory

zooplankton (e.g., spiny and fish-hook waterflea). The position of these nonindigenous species in the food web, and their lack of native ecological analogues in the GL food webs, have led to impacts which largely result from alterations in energy flow through these food webs, either into the benthos and away from pelagic food web or through additional linkages in the pelagic food web (Madenjian et al. 2002, Mills et al. 2003).

In contrast to the complete absence of vascular plants in the set of species introduced by ballast water release, some of the most important species introduced by commerce in living organisms are vascular plants. The primary impact mechanisms of these plants, which include Eurasian watermilfoil and water chestnut, tend to be competition for space and light with native plants (Smith and Barko 1990) and physical modifications that change habitat structure for fish (Valley and Bremigan 2002a) and result in recreational and aesthetic impairment for human use (Eiswerth et al. 2000).

It seems that the specifics of particular anthropogenic transport mechanisms have an effect on which species are transported. In other words, the characteristics of species

‘selected’ for transport in a particular pathway depend on the details of the pathway.

Broadly speaking, these differences in species characteristics may explain the differences in the types of impacts that we found reported in the literature for the pathways we studied. Specifically, ballast water intake appears to select for planktonic species or benthic species with planktonic larvae (Vanderploeg et al. 2002, Grigorovich et al. 2003).

On the other hand, live organisms in trade are often plants that are chosen for their growth form and hardiness (Dehnen-Schmutz et al. 2007, Keller and Lodge 2007). Other organisms, such as tropical aquarium fish, that are chosen for trade are selected for features other than their ability to survive in the natural environment of the GL region

and therefore pose less of a threat for invading the GL (Rixon et al. 2005). Trait-based risk assessments (Keller et al. 2007) could help to prioritize which species in trade pose the most risk to the GL and could potentially support more effective voluntary or regulatory restrictions on the sale in the region of the riskiest species (Keller and Lodge

2007).

2.5.4 Conclusion

The GL are both a beachhead for nonindigenous freshwater species becoming established for the first time in North America, and a melting pot including nonindigenous species dispersing into the GL from other parts of North America. Ship- related introductions are primarily responsible for the beachhead phenomenon, while commerce in living organisms contributes to the melting pot phenomenon. What invades the GL does not stay in the GL. Our analyses suggest that without additional management, most of the nonindigenous freshwater species currently still confined in the

GL will colonize other freshwater ecosystems in North America in the next few decades.

Species that were first introduced by ships are likely to be predatory and/or benthic animals and thus are likely to have food web impacts as they disperse. In contrast, species introduced via commerce in living organisms are likely to be plants and those have impacts related to excessive biomass and canopy development. Therefore, the GL are likely to continue to be a focal point for science, policy, and management regarding freshwater invasions. Without improved voluntary or regulatory management, invasive species will continue to accumulate in the GL and hence in other regions of North

America. In particular, given the beachhead and melting pot phenomena, more effective

initiatives to limit the intra-continental spread of invasive species from and to the GL, respectively are warranted (Bossenbroek et al. 2007).

2.6 Acknowledgements

We thank Penny Nichols for her assistance with the literature review. The NOAA

National Sea Grant Program (Award No. NA16RG2283) through the Illinois-Indiana Sea

Grant College Program (Subaward No. 2003-06727-10) funded this research. A Schmitt

Graduate Research Fellowship from U Notre Dame and a research fellowship from the

Center for Aquatic Conservation at the U Notre Dame also supported JDR. Thanks to the

Lodge lab, particularly Reuben Keller, Chris Jerde, and Jody Peters, for comments that improved this manuscript.

CHAPTER 3:

SHIP-BORNE NONINDIGENOUS SPECIES DIMINISH GREAT LAKES

ECOSYSTEM SERVICES2

3.1 Abstract

For ecosystem services in the Great Lakes, North America, we used structured expert judgment and economic analysis to quantify impacts for 2006 of aquatic invasive species introduced by ships. For US waters, median damages aggregated across multiple ecosystem services were $150 million, while a 5% chance existed that for sport fishing alone the losses exceeded $800 million. Plausible scenarios of future damages were of similar magnitude to the benefits of ocean-going shipping in the Great Lakes, suggesting more serious consideration is warranted for policy options to reduce the risk of future invasions via the St. Lawrence Seaway.

3.2 Introduction

Environmental problems often go unaddressed because the value of lost ecosystem services is not expressed in units commensurate with financial investments needed to solve the problem. Invasive species are a leading environmental problem

2 The publication status for this chapter is: Rothlisberger, J.R., D.M. Lodge, R.M. Cooke, and D.C. Finnoff. Ship-borne nonindigenous species diminish Great Lakes ecosystem services. Science (in preparation). 51

globally (Sala et al. 2000), reducing ecological integrity (Carlsson et al. 2004), leading to the occasional extinction of native species (Mills et al. 1994, Nalepa et al. 1996), altering ecosystem functioning (Mills et al. 1994), and thereby reducing human welfare via losses of ecosystem goods and services (Pimentel et al. 2005). Despite the urgent need to quantify lost ecosystem services, biological and economic researchers using traditional methods struggle to quantify invasive species impacts in units that allow comparisons with the costs of possible private or public remedies. External costs cannot be internalized or otherwise remedied if they are not quantified (Ehrlich and Pringle 2008,

NRC 2008).

Here we use Structured Expert Judgment (SEJ) to estimate distributions of the biological and economic impacts of invasive species introduced to the Laurentian Great

Lakes (GL) via ships since the 1959 opening of the St. Lawrence Seaway (Cooke 1991).

We focus on current ecosystem services and services two decades into the future. SEJ is an established technique for probabilistic risk assessment (Apostolakis 1990) and consequence analysis (Cooke and Goossens 2000), but has been applied rarely to environmental problems and never to invasive species.

In SEJ, consideration of relevant scientific research and evaluation of possible future influences are used to generate estimates for important variables. Invasive species impacts, for example, could in theory be empirically measured with very large scale, long-term experiments; in practice, however, logistical, technical, and ethical constraints prevent such experiments (Cooke 1991). In SEJ, experts estimate the distributions of response variables in such hypothetical experiments. The structured process explicitly quantifies uncertainty and treats a subset of expert estimates as hypotheses that are tested

against real data to assess expert accuracy. Thus, the values of unknown variables and uncertainty ranges are estimated. Furthermore, SEJ allows the combination of judgments from multiple experts into a single distribution (Cooke 1991).

Here we focus on invasive species delivered via a single vector (i.e., shipping) because management efforts, especially those designed to prevent unwanted introductions, are most efficiently focused on vectors (Lodge et al. 2006). Globally, shipping is the major vector for aquatic invasive species, and at least 57 alien species introduced by ships have become established in the GL, including zebra and quagga mussel (Dreissena polymorpha and D. bugensis), round goby (Apollonia melanostomus), and spiny waterflea (Bythotrephes longimanus) (Ricciardi 2006). With 35 million people living in the GL basin, ecosystem services from the GL benefit a large number of households and communities, affecting a substantial regional economy (Austin et al.

2007). For the US GL region, we focus on four ecosystem services that are important to the regional economy and for which reliable historical data are available. These are commercial fish landings, sport fishing participation, wildlife viewing, and raw water usage. In 2006, for example, the market revenues of commercial fishing from the GL were $14.5 million (USGS 2008), while expenditures on GL sport fishing were more than

$1.5 billion (USFWS 2007). Nearly 1000 municipal water supplies, industrial facilities, and power generation plants in the US draw raw water from the GL (Deng 1996). We compare each of these ecosystem services in their current invaded condition to a hypothetical benchmark of an ecosystem state without ship-borne species. In making this comparison, we assume that all other factors (i.e., environmental and economic conditions) would have remained exactly the same with and without ship-borne species.

To translate 2006 ecosystem impacts of ship-borne species into dollar values, we used simple economic methods to estimate consumer surplus, a large improvement over most previous studies of the economic impact of invasive species, which have employed replacement cost methods (see Appendix B). By converting distributions of ecosystem service impacts into dollar units, we provide benchmarks to inform policy-makers about the predicted consequences of future invasions based on invasion-associated impacts up until 2006. These benchmarks could be used in evaluating the benefits of policy and management choices to reduce the probability of future invasions (e.g., more stringent requirements for ballast water treatment on ships, closure of the St. Lawrence Seaway).

Our approach to assessing ecosystem-scale effects of invasive species also provides a possible template for similar efforts in different ecosystems and for other environmental stressors. Such assessments could be valuable for evaluating policy and management alternatives to prevent or mitigate many kinds of environmental damage.

3.3 Results

Distributions predicted by experts indicated that without ship-borne invasive species the GL would be providing larger commercial fishery harvests and more participation in sport fishing, with median damage estimates for 2006 ranging among lakes from 13-33% in commercial fisheries and 11-35% in sport fisheries (Table 3.1,

Figure 3.1). Because of large discrepancies in expert assessments for impacts on sport fishing in L. Superior (see Appendix B), the 35% impact predicted there is not taken as credible (because it reflects a calibration variable that was deemed to be unreliable post hoc) and, therefore, is not included in our calculation of economic impacts below. The 54

uncertainties surrounding participation in wildlife viewing, which encompasses various ecotourism-related activities, are very large and participation levels are predicted to be almost as likely to increase as to decrease without ship-borne invasive species (Table 3.1,

Figure 3.1).

TABLE 3.1

SUMMARIZED DISTRIBUTIONS OF PERCENT IMPACTS OF SHIP-BORNE

NONINDIGENOUS SPECIES ON ECOSYSTEM SERVICES IN THE GREAT LAKES

IN THE UNITED STATES IN 2006

Ecosystem Lake Median % Impact % of distribution Service Estimate 0%above: 100% impact impact Commercial Fishing Superior 13 59 9 Michigan 21 62 16 Huron 23 62 16 Erie 18 68 1 Ontario 33 57 39 Sport Fishing Superior 35 66 35 Michigan 11 59 2 Huron 30 62 31 Erie 15 65 1 Ontario 14 62 2 Wildlife Viewing All 1 51 2

A) Commercial B) Sport Fishing Effort 0.008 Superior

0.004

0.000 0.010

Michigan

0.005

0.000 0.006 Huron

0.003

0.000 0.012 Erie

0.006

0.000

0.008 Ontario

0.004 ProbabilityDensity

0.000 C) Sport Expenditures D) Wildlife Viewing

0.012 Expenditures Wildlife Viewing

0.006

0.000 -100 -50 0 50 100 150 200 -100 -50 0 50 100 150 200 Percent Impact

Experts provided mechanistic explanations for their assessments of the impacts that differed according to lake and ecosystem service (see Appendix B). While some experts described some aspects of the impacts of selected nonindigenous species as positive, in only a few instances did an expert (almost always the same individual) indicate that the median value of a variable would be greater with ship-borne species than without. In this study, the dependence between impacts with and without invasives was not modeled, and the default assumption of independence was applied. If factors causing low impact (e.g., demographic factors) tend to affect impacts regardless of invasives, then the independence assumption will overestimate uncertainty (see Appendix B). This may explain the portions of the percent impact distributions to the left of zero (Figure 3.1), which suggest ship-borne species are beneficial. Although techniques for dependence elicitation are well-established (Cooke and Goossens 2000), it would considerably augment the elicitation burden, and given the novelty of this approach for the experts involved, we chose to leave the subject of dependence for a later study.

Assessing the economic value of ship-borne species impacts (as changes in consumer surplus, see Appendix B) introduces an additional layer of uncertainty and makes the range of each economic impact distribution (Figure 3.2) wider than the range of the percent impact distributions (Figure 3.1). For the commercial fishery, predicted economic impacts are likely greater than zero, with a median value of $5.3 million

(Figure 3.2A). The estimated economic impact on sport fishing is greater than for commercial fishing (median impact of $106 million), but sport fishing impact distributions also indicate more uncertainty (Figure 3.2B). Economic impacts on wildlife viewing were extremely uncertain, at least 10 times more uncertain than for sport fishing

0.050 A)

0.025

0.000 -20 -10 0 10 20 30 40

0.002 B)

0.001

0.000 -500 0 500 1000

2e-04 C)

1e-04 Probability Density Density Probability

0e+00 -10000 -5000 0 5000 10000

0.0350 D)

0.0175

0.0000

0 10 20 30 40 50 Difference in Consumer Surplus or Additional Costs from Biofouling (Millions of 2007 USD)

and 200 times more uncertain than for commercial fishing, and were centered almost on zero, with a median impact estimate of only $12 million (Figure 3.2C). For biofouling impacts on raw water use, median additional operating costs aggregated over all GL facilities is $27 million (Figure 3.2D). The municipal water treatment sector experiences the greatest losses (as calculated by the number of facilities in the region multiplied by additional costs per facility), while nuclear power generation experiences the least (see

Table B.3 in Appendix B).

Summing median impact estimates for these four ecosystem services produces an overall median estimate of economic losses in 2006 of $150 million (Figure 3.3).

However, considering only sport fishing—a large economic sector for which expert distributions were skewed strongly in the direction of negative impacts—a 5% chance exists that impacts are as high as $800 million (Figure 3.3). So much uncertainty characterizes the wildlife viewing sector that we do not emphasize those potentially very large damages (Figure 3.3).

Halting the entry of ocean-going ships into the GL would not reduce the impacts we report here because the set of ship-borne species we considered would remain in the lakes. Of course, additional invasions causing additional damages are likely to occur if ballast water releases continue. However using the impact distributions reported here to compare costs and benefits of alternative future ballast water policies is not straightforward because the assessments for 2006 are only a snapshot of dynamic and stochastic invasion processes that occurred since the opening of the St. Lawrence Seaway in 1959. Furthermore, the ecological and economic systems of the GL have changed in

Commercial Fishing Biofouling Sport Fishing Wildlife Viewing

-4775 4950

-500 0 500 1000 90% Uncertainty Range for Economic Impacts (Millions of 2007 USD)

Figure 3.3. Ninety percent uncertainty ranges for economic impacts in the United States of ship-borne NIS on multiple ecosystem services in the Great Lakes.

response to the gradually increasing presence of ship-borne species, further complicating predictions of the future consequences of new introductions.

In the absence of direct knowledge about how the interacting ecological and economic systems of the GL would transition into the future with and without additional invasions, we considered four plausible scenarios for how economic impacts might accumulate if shipping patterns and ballast water releases continue apace (Figure 3.4).

We compared these costs of invasions to the benefits of shipping, as estimated by a previous study that found the St. Lawrence Seaway provides current annual transportation savings of $55 million over using other transport modes (e.g., truck or rail) to move all the goods and materials that are currently carried into the GL region on ocean-going ships (Taylor and Roach 2009). Carrying these annual transportation savings 50 years into the future with a 2% discount rate yields $1.7 billion in cumulative savings (Figure 3.4).

One plausible scenario (‘Constant Increase’) is that impacts from new invasive species will grow at the same constant average annual rate over the next fifty years as they did in the past (assuming linearly increasing impacts during the previous 5 decades

(i.e., $150 million in 2006 divided by 48 years of accumulating impacts = $3.1 million growth in impacts per year). Applying a 2% discount rate (which we applied also to other scenarios below), we find that preventing future ship-borne invasions would avoid the cumulative loss of more than $2.1 billion in additional ecosystem services over the next half century (Figure 3.4). Thus, under this scenario, over the next 50 years potential losses from ship-borne invasions are nearly $400 million more than the losses in transportation savings. A moratorium on the passage of ocean-going ships in the St.

4000 Transport Savings Invasive Damage (Growing Increase) Invasive Damage (Constant Increase) 3000 Invasive Damage (Decreasing Increase until Plateau) Invasive Damage (Exponential Increase from 0 to 150)

2000

1000 (Millions of 2007 USD) 2007 USD) of (Millions

orLosses Savings Future Cumulative 0 0 5 10 15 20 25 30 35 40 45 50 Years into Future

Figure 3.4. Scenarios of future cumulative ship-borne invasive species damage relative to cumulative transportation savings from ocean-going

shipping into the GL.

Lawrence Seaway, the most draconian measure proposed to stop future ship-borne invasions in the GL, would produce net benefits over this time horizon. The point at which cumulative invasive species damages become greater than cumulative transportation savings occurs 38 years in the future (Figure 3.4). Other scenarios yield different results.

Another plausible scenario (‘Growing Increase’) is for annual impacts to grow at an accelerating rate. For example, according to the formula xt = xt-1+b+c(t-1), where xt is the annual impact in year t, b is the base rate of impact growth, and c is amount by which the added impact grows from one year to the next (Figure 3.4). If we set the base rate of impact growth (b) to be the same as the linear model of impact growth (i.e., $3.1M) and c

to be $0.1M, the additional cumulative losses from ship-borne invasions over the next 50 years ($3.1B) would be $1.4 billion more than transportation savings from shipping, with cumulative damages becoming greater than cumulative savings 29 years into the future

(Figure 3.4). If, however, additional annual impacts of invasions accrue at a decreasing rate, eventually reaching a plateau at which annual impacts remain the same from one year to the next, cumulative invasive damages may never be greater than shipping savings (‘Decreasing Increase until Plateau,’ Figure 3.4). Under this scenario, whether the net benefit of preventing invasions is positive or negative depends on the rate of annual decrease ($100,000 in our example) and the level at which impacts plateau (here

$50M above the $150M/year level in 2006). In a fourth scenario (‘Exponential Increase from $0 to $150M), we assume that additional annual impacts will grow exponentially from $0 to $150M/year over the next 50 years, at which time cumulative losses to invasive species will not yet have surpassed cumulative transportation savings (Figure

3.4)

Of course, in reality annual impacts of invasions must eventually level off when there is either no value left to lose at a state of utter ecosystem degradation or when the impacts of any future invaders are redundant with any existing impacts. The $200M plateau in impacts in the ‘Decreasing Increase until Plateau’ scenario above is likely at the low end of the range for possible plateaus, given that much of the distribution of damages estimated for 2006 is above $200M (Figure 3.3).

3.4 Discussion

We provide the first ecosystem-scale estimated distributions of the bioeconomic impacts of invasive species, and the first such estimates that are pathway-specific.

Analyses like the one we report here are a first step to understanding the consequences of biological invasions in units that can inform more rigorous benefit-cost analyses of alternative policies to prevent future invasions. By explicitly quantifying uncertainty inherent in both the biological and economic systems, we have enabled policy makers to make choices about prevention policies with fuller than usual knowledge about risks of future damages.

Because the value of commerce, including the ship-driven commerce considered here, is obvious and often well quantified, decisions in the absence of information on ecosystem services tend to strongly discount the negative environmental side effects of commerce. While the range of our estimates of the collective impact of invasive species are large, our median estimates (i.e., the impact levels experts thought most likely) and scenarios for the accumulation of future economic damage suggest that substantial new investments in reducing ship-borne invasions in the Great Lakes and globally are warranted.

Previous estimates of the impacts of invasive species in the Great Lakes have concentrated on raw water users (O'Neill 1996, NRC 2008) and experts in this study indicate that these impacts persist but are small relative to impacts on other ecosystem services. Specifically, the economic consequences of ship-borne invasions for sport fishing are substantial. Because sport fishing is a large economic sector and is important to the region, it provides the bulk of the predicted impacts. Wildlife viewing is a much

larger economic sector (USFWS 2007), but experts estimated minimal impacts. Our findings thus provide a fuller understanding of the relative importance to the regional economy of recreational ecosystem services, especially sport fishing, that have been affected by ship-borne invasive species.

Many previous efforts to quantify the economic impacts of invasive species have often poorly documented, sometimes reporting worst-case scenarios as actual impacts

(Hoagland and Jin 2006). Misleading estimates of the economic impacts of invasive species can promote policies that are economically wasteful (Hoagland and Jin 2006), highlighting the value of transparent methodologies like those we employed here. A few other studies have estimated the economic damage of invasive species in aquatic ecosystems using traditional non-market valuation techniques that are also transparent

(e.g., travel cost method, contingent valuation; Nunes and van den Bergh 2004, Nunes and Markandya 2008). These methods apply economic theory to produce scientifically valid measurements of economic damage, but are limited in their ability to measure the impacts of invasive species (Nunes and Markandya 2008). For example, when the general public is surveyed for a contingent valuation study it is often not possible to communicate the complex food web effects of invasive species. Furthermore, such studies are generally limited in scope, focusing on one aspect of the economic value of an ecosystem in one particular location (Nunes and Markandya 2008). Trying to extend beyond these limitations, we sought comprehensive impact estimates that included potentially ecologically complex effects of ship-borne species.

Ideally, estimates of biological and economic damage attributable to alien species would be based on the measurement and comparison of key response variables before

and after the invasion, while controlling for all other simultaneously changing factors and conditions that could affect the response variables (Hoagland and Jin 2006). Obtaining such data for the GL region is not possible, making it necessary to seek an alternative approach to quantifying damage to ecosystem services. In SEJ, we found a novel approach to estimate invasive species impacts, representing an important advance because it is highly structured, clearly documented, and explicitly quantifies uncertainty.

On the other hand, several significant omissions from our work make it likely that our median impact estimates are lower than actual damages. First, we were not able to include analogous damages in the Canadian portion of the GL basin. Second, we were not able to include several large US economic sectors (e.g., recreational boating, beach use) that are probably affected by ship-borne invasions. Third, we do not consider losses to ecosystem services that are in the US but outside the GL region. Among the species considered in this study are some, including zebra and quagga mussels, which were introduced originally into the GL by ships, but have already caused substantial damage in other parts of the country, increasingly including western waterways. Unlike other forms of pollution, these living species continue to increase in abundance, spread, and further reduce ecosystem goods and services throughout the continent (Drake and Bossenbroek

2004, Bossenbroek et al. 2007). The current dreissenid mussel invasion of Lake Mead and various California waterways is one such example that is ultimately attributable to shipping in the GL (Stokstad 2007).

Finally, to better assess the net value of efforts to reduce the impact of current or future invaders from specific pathways, the findings of this study need to be augmented with a better understanding of the pattern of invasions that led to these consequences,

how this pattern is expected to continue on into the future, and how it might change under alternative policies. Simply put, to assess what damages would be avoided if additional management efforts were implemented requires more information in addition to that reported herein. For example, the efficacy of ballast water exchange remains debatable, and requires further evaluation (Costello et al. 2007). In spite of these omissions, we hope the results reported here will prompt fuller analyses of various policy options to better protect ecosystem services and human welfare from future biological invasions.

Completely stopping the introduction of invasive species to the GL via ocean- going vessels is unlikely (NRC 2008). Nevertheless, our study provides a useful estimate of the value, in terms of likely damage to ecosystem services avoided, of efforts to prevent future invasions by ship-borne species. We narrowed the focus of our study to estimate the economic impacts arising from ecological perturbations caused by invasive species in a particular system, the GL, associated with a particular introduction pathway, shipping. Estimates of the impacts of species delivered via a certain pathway can support decision making regarding pathway-based policy and management. Our estimates of economic impact provide a figure for comparison against the costs of implementing management activities to prevent invasions via the shipping pathway. Comparison of our results with the results of an earlier study on the transportation savings from shipping

(Taylor and Roach 2009, Fig. 4) illustrates how our results might be used to evaluate alternative policies like those considered by the recent NRC report (e.g., on-board ballast water treatment; NRC 1996, NRC 2008). Whether or not net savings will result from the prevention of future invasions will depend on the cost of modifying transportation systems and on how the magnitude of invasive species impacts change in the future.

Learning the rate at which annual impacts may change in the future and where these impacts are likely to plateau will be a critical next step for evaluating ballast water policy and management.

3.5 Acknowledgements

We thank the experts for their participation and cooperation. The NOAA

National Sea Grant Program (Award No. NA16RG2283) through the Illinois-Indiana Sea

Grant College Program (Subaward No. 2003-06727-10) partially funded this research.

The US EPA's National Center for Environmental Economics also provided support

(Contract No. EP-W-05-022). A Schmitt Graduate Research Fellowship from U Notre

Dame supported JDR. Special thanks to Scott Nelson of the USGS Great Lakes Science

Center for help with the 2006 US commercial fish landing data. Ashley Baldridge, Matt

Barnes, Chris Jerde, Reuben Keller, Brett and Jody Peters, and Darren Yeo provided helpful comments. Thanks to Joanna McNulty for invaluable administrative support.

Although the research described in this article has been funded in part by the US EPA, the opinions expressed here are those of the authors, and do not necessarily express the views of the United States Environmental Protection Agency.

CHAPTER 4:

AQUATIC INVASIVE SPECIES TRANSPORT VIA TRAILERED BOATS: WHAT IS

BEING MOVED, WHO IS MOVING IT, AND WHAT CAN BE DONE3

4.1 Abstract

The overland movement of trailered boats has been linked to the spread of a wide range of aquatic invasive species (AIS), including Eurasian watermilfoil (Myriophyllum spicatum), spiny waterflea (Bythotrephes longimanus), and dreissenid mussels (Dreissena polymorpha and Dreissena bugensis). Most efforts to stem the spread of all varieties of

AIS via trailered boats have focused on pre-launch boat inspections at uninvaded waterways and campaigns to educate the general public on actions individuals can take to reduce the likelihood of transporting AIS. There has been, however, little empirical research on the type and quantity of AIS being transported, nor on the efficacy of management interventions (e.g., inspection crews, boat washing). In an observational study designed to investigate the prevalence and types of organisms attached to small- craft boats and trailers, we collected aquatic organisms representing 17 orders and terrestrial organisms representing 12 orders, including some species that are morphologically similar to known AIS. Additionally, returns from a mail survey of

3 The publication status of this chapter is: Rothlisberger, J.D., W.L. Chadderton, J. McNulty, and D.M. Lodge. Aquatic invasive species transport via trailered boats: what is being moved, who is moving it, and what can be done. Fisheries (in review). 69

registered boaters (n=944) indicated that 68% do not always take steps to clean their boats prior to moving among waterways. Regarding intervention efficacy, we found via a controlled experiment that visual inspection and hand removal can reduce the amount of macrophytes attached to each boat by 88%±5% (mean±SE), with high-pressure washing equally as effective (83%±4%) and low-pressure washing less so (62%±3% removal rate). For removing small-bodied organisms like zooplankton (including resting stages), and plant seeds, high-pressure washing was most effective with a 91%±2% removal rate; low-pressure washing and hand removal were less effective (74%±6% and

65%±4% removal rates, respectively). This research supports the widespread belief that trailered boats are an important vector in the spread of AIS, and suggests that many boaters have not yet adopted consistent and effective boat cleaning habits. Therefore, additional management efforts may be appropriate. In such efforts, pressure washing is highly effective when the aim is to remove small-bodied organisms like spiny waterflea, but visual inspection and hand cleaning appears to be sufficient for the removal of macrophytes.

4.2 Introduction

Much of the on-going spread of aquatic invasive species (AIS) to inland waters throughout North America can be attributed to the overland movement of small-craft boats (Bossenbroek et al. 2001, 2007; Johnson et al. 2001; Puth and Post 2005; Leung et al. 2006). Small-craft boats are vessels less than 40 feet in length, including powerboats, small commercial and recreational fishing boats, sailboats, personal watercraft, canoes and kayaks, and pontoon boats, that can be towed overland on trailers. Translocation of 70

organisms by boaters can be intentional (e.g., as bait; Keller et al. 2007), but is often unintentional (Johnson et al. 2001, Puth and Post 2005), with organisms inadvertently carried in bilge water, live wells, and bait buckets. Organisms can also be entrained on boat exteriors, e.g., entangled on propellers and trailers, attached to other entangled organisms (Johnson et al. 2001). Thus, every time a boat is transported overland after use in an invaded waterway, there is the possibility that it will transfer AIS to uninvaded waterways.

Overland transport of small-craft boats is thought to be responsible for the spread of spiny waterflea Bythotrephes longimanus (MacIsaac et al. 2004, Muirhead and

MacIsaac 2005), Eurasian watermilfoil Myriophyllum spicatum (Buchan and Padilla

2000), and zebra and quagga mussels Dreissena spp. (Johnson et al. 2001; Leung et al.

2004, 2006; Schneider et al. 1998; Stokstad 2007). These organisms are known to have considerable negative effects on the aquatic ecosystems they invade, with impacts including damages to fisheries (Vanderploeg et al. 2002, Mills et al. 2003, Marsden and

Robillard 2004), interference with raw water usage (O’Neill 1996, Leung et al. 2002), decreased property values (Halstead et al. 2003), extirpation of native species (Nalepa et al. 1996, Strayer 1999), and threats to human health (Vanderploeg et al. 2001, Yule et al.

2006, Hogan et al. 2007). The recent invasion of the Great Lakes by Viral Hemorrhagic

Septicemia (VHS, a fish virus) (Lovell and Drake 2009) taken together with the spread of the virus to inland lakes, most likely by trailered boats, further emphasizes the potentially serious consequences of moving biological materials among waterways (Elsayed et al.

2006).

The Great Lakes region provides an opportunity to study how to better manage the risks of aquatic invasive species spread by small-craft boaters. There are numerous aquatic resources in the region, including the Great Lakes as well as abundant inland waterways. Moreover, recreational boating is an important driver of the regional economy (RMRC 2006). In the eight U.S. states bordering the Great Lakes, there are 4.2 million small-craft boats, nearly a third of all those currently in use in the U.S. (Thorp and Stone 2000). Likewise, in the Canadian provinces of Ontario and Quebec, there are over 2 million recreational boats (Thorp and Stone 2000).

The quality of the region’s aquatic resources is threatened by aquatic invasive species. For example, over 300 lakes in the region and multiple rivers have been invaded by zebra mussel, fouling water intakes of industrial facilities and reducing native biodiversity (Johnson et al. 2006). Eurasian watermilfoil, an invasive macrophyte that impairs navigation and recreation and displaces native macrophytes, is present in nearly

1000 lakes in Michigan, Wisconsin, Illinois, and Indiana. The impacts of these and other species, combined with the importance of the resources they harm, have resulted in the region becoming a test bed for science and policy pertaining to the ecology and impacts of AIS. Thus, the stakeholders in the region tend to be generally aware of AIS issues and are concerned about reducing AIS impacts. In some cases, however, stakeholders lack empirical data about the spread of AIS by small-craft boaters and about the effectiveness of various techniques proposed to restrict spread. This lack of knowledge can limit the confidence of managers and the public that management interventions intended to limit future spread of AIS are worthwhile.

To date, in the Great Lakes region, as in most others regions, efforts to stem the spread of AIS via trailered boats have focused on pre-launch boat inspections at uninvaded waterways and on campaigns to educate the general public on actions that individuals can take to reduce the likelihood of transporting AIS. For example, the Clean

Boats/Clean Waters programs of Wisconsin and Michigan train employees and volunteers to inspect watercraft and trailers as a viable means for preventing AIS introductions (http://www.uwsp.edu/cnr/uwexlakes/cbcw/). Furthermore, such regional campaigns and national programs such as Protect Your Waters

(http://protectyourwaters.net) seek to educate the public, particularly those individuals that own and use recreational boats, about five steps to take to reduce the likelihood of transporting AIS: “(1) inspect and remove aquatic plants, animals, and mud from boat, trailer, and equipment before leaving the landing, (2) drain all water from boat, motor, live wells, bilge, bait buckets and other containers before leaving the landing, (3) ice your catch; don’t leave landing with any live fish, bait, or fish eggs, (4) dispose of unused bait in trash, not in the water or on land, and (5) rinse boat and equipment with hot or high pressure water or dry boat for at least five days”

(http://www.uwsp.edu/cnr/uwexlakes/cbcw/Pubs/AISprevention_steps.pdf). Pertaining to this fifth recommendation, some natural resource managers and private citizens advocate boat-washing stations on the public landings of waterways, contending that high-pressure washing is necessary to remove biological materials effectively.

Surprisingly, however, no rigorous scientific research is available on the efficacy of the main techniques advocated for removing organisms from trailered boats.

Furthermore, few empirical efforts have quantified the types and numbers of organisms

in transport. Moreover, data on boater compliance with the above-listed recommendations for preventing the spread of AIS are also lacking and it is unknown if different sub-groups of boaters (e.g., recreational boaters, professional fishing guides) differ in their boat hygiene behaviors and, therefore, in their likelihood to transport organisms. A better understanding of each of these aspects of the trailered boat pathway is critical to improve policy and management intended to reduce the threat of additional invasions.

Drawing on data from an observational study, two surveys, and an experiment, this paper aims to fill some of the important gaps in our understanding of the risks of AIS spread posed by the trailered boat pathway and to address the question of how well various cleaning techniques remove organisms from the pathway. Thus, to quantify the number of organisms being transported, we present data on the type and quantities of organisms collected from the external surfaces of boats and trailers. To learn about the steps boaters take to prevent AIS transport and how these behaviors may differ across sub-groups of boaters, we surveyed registered boaters by mail and in person. Finally, we experimentally tested the efficacy of the three most common boat-cleaning methods (i.e., visual inspection and hand removal, low-pressure washing, and high-pressure washing) in removing organisms (i.e., macrophytes, zooplankton, and plant seeds) from the exterior surfaces of boats and trailers.

4.3 Methods

4.3.1 Observational Study

We washed 85 boats arriving to (36 boats washed) and departing from (49 boats washed) two popular boat landings in the Northern Highlands Lake District of northern

Wisconsin and the Upper Peninsula of Michigan (Big St. Germain Lake, Vilas Co., WI

[Latitude: 45.9344, Longitude: -89.5163] and Lake Gogebic, Gogebic Co., MI [Latitude:

46.4999, Longitude: -89.5835]), between August 26 and September 5, 2006 to gather data on the types and quantities of aquatic organisms inadvertently transported by recreational boaters. We selected these landings because of their popularity and because the design of the boat launch allowed for convenient set up of our boat washing equipment. Invasive spiny waterflea Bythotrephes longimanus are present in Lake

Gogebic, but no aquatic invasive species likely to be inadvertently transported by recreational boaters are known to exist in Big St. Germain Lake.

All arriving and departing boats were washed using a portable high-pressure wash and reclaim system, which was a modified version of a portable noxious weed removal system (WB500, Spika Manufacturing, Mocassin, MT). This system, originally developed by the US Forest Service to clean weed seeds and plant pathogens off vehicles and equipment used to fight wildfires (Trent et al. 2002), supplied the high-pressure wash and the water filtration capabilities needed for this study. The wash water was captured on a waterproof mat and then pumped through a filtration and reclamation system, passing through a food-grade polyethylene filter (nominal pore size: 100 m) that trapped materials removed from washed boats.

Although we washed a total of 85 boats, for logistical reasons, each filter collected the materials washed from four to seven boats. The main reason for this pooling of samples was that boats tended to arrive at our washing station clustered together in time. Changing the filter in our washing unit took approximately 10 minutes.

We estimate that at least one-half of the boats we washed would have bypassed our washing station because of their unwillingness to wait for filter changing. Because one of the main objectives of this aspect of our study was to obtain organisms from as many boats as possible, we chose to pool samples from multiple boats on to each filter. Thus, for the statistical analysis of this component of our research, the filter is the replicated unit of study. As filters are the replicated unit for this study, this gave us a sample size of six (filters) for arriving boats and eleven (filters) for departing boats.

We used separate filters for departing versus arriving boats so that organisms originating from a lake could be distinguished from those arriving from elsewhere. In the lab we removed and weighed all material collected in the filters. We then subsampled the material from each sample (i.e., filter) by spreading it evenly over a flat-bottomed sorting tray divided into 12 equally-sized sectors. We used a random numbers table to select four sectors from which to collect material for detailed sorting and identification and enumeration of organisms and other biological materials. When drawing off material from a subsampled sector, we used an enclosed sectioning device with a foam bottom to form a watertight seal with the bottom of the tray to separate the sector from those adjacent to it and to prevent the inclusion in the subsample of any materials not in the chosen sector. We used information on the total wet mass of material collected in a filter

(i.e., all collected material weighed---not subsampled), the number of boats washed onto

that filter, and the mean number of aquatic organisms in the four subsamples from each sample to calculate estimates of the quantity of biological materials moved over land on the exterior of recreational boats.

Before the washing described above, each boat and trailer were inspected visually for vegetation fragments, all of which were removed, identified, and weighed.

4.3.2 Mail Survey

A mail survey was conducted in August 2005 to obtain data from a broad sample of small-craft boaters about their boat cleaning habits, particularly about their movements from boats from one waterway to another. We mailed a total of 10000 surveys to a random sample of registered boaters in Wisconsin and Michigan (i.e., 5000 to each state), with the number of surveys sent to each county proportional to the number of registered boaters in each. We used the boater registration databases for the two states to select survey recipients. Some addresses in the databases were outdated, resulting in 1382 surveys returned as undeliverable. A total of 515 boaters from Michigan and 429 from

Wisconsin returned usable surveys, a response rate of 11% of delivered surveys. For the analysis, we combined the responses from both states. In the survey we posed a number of questions about boaters’ movement habits and other boating-related activities. Our main interests were how frequently boaters noticed and removed aquatic weeds attached to their boat and trailer, how regularly they cleaned their boat, what methods they used for boat cleaning, and how frequently they launched their boat in different lakes (Table

4.1).

4.3.3 In-person Northwoods Survey

We also interviewed small-craft boaters in person to gather additional data on travel patterns and boat cleaning practices of boaters in the same region where we conducted our observational boat washing study. These interviews, conducted between

May 28 and August 15, 2007, occurred at sites (e.g., lake association meetings, bait shops, campgrounds, and boat ramps) in several counties in and near the Northern

Highlands Lake District of northern Wisconsin and the Upper Peninsula of Michigan, including Vilas and Oneida in Wisconsin and Iron, Gogebic, and Marquette in Michigan.

We asked the same questions as those asked in the mail survey for these interviews

(Table 4.1).

For the in-person survey, we interviewed two categories of boaters: general recreationalists (n = 424) and professional fishing guides (n = 35) to learn if these two categories of boaters had different movement patterns and boat hygiene practices that might affect their risk of spreading AIS. Only 49 of the individuals (46 recreationalists and 3 guides), we approached for interviews declined participation, giving a 90.2% response rate.

4.3.4 Experiment

We performed two experiments to test the effects of cleaning method and duration on the removal of aquatic macrophytes (first experiment) and small-bodied animals and plant seeds (second experiment) from the exterior of recreational boats and trailers. In the macrophyte removal experiment, we used the invasive aquatic plant Eurasian watermilfoil (Myriophyllum spicatum) as the test organism. In the small-bodied

TABLE 4.1.

QUESTIONS AND RESPONSES FROM MAIL AND IN-PERSON SURVEYS.

Questions Responses Before going from one lake or river to Not Always Sometimes Never another, how often do you: applicable

Clean your boat by rinsing, pressure washing, or drying (a) Mail (n = 396) 27% 34% 4% 5% (b) In-person (i) Guides (n = 35) 11% 75% 0% 4% (ii) Recreational boaters (n = 135) 24% 42% 3% 3%

Notice weeds attached to your boat or trailer (a) Mail 9% 43% 0% 8% (b) In-person (i) Guides 11% 86% 0% 3% (ii) Recreational boaters 42% 45% 9% 4%

Remove any aquatic weeds attached to your boat or trailer (a) Mail 57% 14% 13% 16% (b) In-person (i) Guides 96% 0% 0% 4% (ii) Recreational boaters 87% 10% 1% 2%

If you trailer your boat among waterways, in how many different waterways have you launched your boat in the past two weeks? (mean±SE) (a) Mail 2.66±0.14 (b) In-person (i) Guides 5.41±0.80 (ii) Recreational boaters 2.72±0.42

Note: Sample sizes are for the number of transient boaters (i.e., boaters that launch in more than one waterway during the season) that responded to the surveys.

organism experiment, our test organisms were the invasive cladoceran Bythotrephes longimanus and the seeds of three species of wetland plants (Alisma subcordatum,

Verbena hastata, and Carex frankii). The six cleaning treatments were identical in both experiments, and resulted from the factorial crossing of three levels of cleaning method with two levels of cleaning duration (90 seconds and 180 seconds). The three levels of cleaning method were: 40 psi wash water pressure (“low pressure” hereafter); 1800 psi wash water pressure (“high pressure” hereafter); and visual inspection of the boat and trailer accompanied by hand removal of organisms. We repeated both cleaning experiments seven times for each of the six treatments.

During each experiment, one person---the same individual for all replicates--- placed a known quantity of biological materials (52-153g of milfoil for the macrophyte experiment; 100 each of three species of seeds and Bythotrephes) on a boat and its trailer, recording the placement locations of all materials. Milfoil was placed on and around the propeller, on the trailer bunks, and on other protruding parts of the boat and trailer where it could plausibly become attached. Small-bodied organisms were adhered to the boat hull and trailer using water-based gel (L.A. Looks Mega-Hold hair styling gel, Henkel

Consumer Goods, Inc., Irvine, CA), mimicking mud or foam with organisms embedded in it that could stick to a boat or trailer. The same boat, a general-purpose 16-foot aluminum V-hull motorboat (1993 Fisher 1675 Plus, Springfield, MO), and single-axle steel trailer were used in all replicates.

To quantify how effectively visual inspection would detect organisms, a second person (the same person for all replicates) inspected the boat for 120 seconds while

recording the locations of any biological materials discovered. A third person then washed the boat according to the specified pressure/time treatment. We captured and filtered all water used in each washing replicate using the same portable wash and reclaim system used in the observational study described above. Finally, the person who had initially placed the materials on the boat and trailer recovered any items still attached.

To calculate percent removal for the macrophyte experiment, we divided the initial minus the final mass of M. spicatum on the boat by its initial mass. To measure removal rates for seeds and Bythotrephes, we enumerated the seeds and zooplankton captured in the filtration system for each replicate and divided this by the number of small-bodied organisms originally on the boat (i.e., 300). To determine statistical significance of differences in percent removal among treatments, we used a two-way ANOVA on the data from each experiment (i.e., macrophyte and small-bodied organisms), followed by a post-hoc Tukey HSD test for multiple comparisons.

4.4 Results

4.4.1 Observational Study

Of the 85 boats we inspected and washed during the observational study, 38

(45%) carried one or more plant fragments, but, of these, 30 had little material attached

(i.e., <5g, Figure 4.1A). Boats and trailers leaving the lakes were three times more likely to be carrying vegetation than those arriving: seven of 36 boats (19%) arriving at a lake had vegetation attached, whereas, 31 out of 49 boats (63%) leaving a lake had vegetation attached (Figure 4.2). The average biomass of macrophytes attached to a single boat and

trailer was 6.4±2.9g (mean±SE), with no statistically significant difference between boats leaving a lake and those arriving (Welch two-sample t-test: t = -0.17, df = 20.96, p =

0.87).

Of the 13 species of macrophytes collected from boats, none were invasive species (Table 4.2). We collected seven fragments of Myriophyllum heterophyllum, a native milfoil species, that is morphologically similar to the invasive M. spicatum, a widespread nuisance species in North America. Most of the individual vegetation fragments we collected were very small, but some were quite large (Figure 4.1B).

We also collected 51 taxa of small-bodied organisms from the filter samples (Table 4.3), including 28 aquatic animals, among them amphipods, gastropods, and cladocerans. No

AIS were collected. Among the aquatic organisms, 8 of the 18 orders we collected were crustaceans, including zooplankton species (Table 4.3). Numerically, however, crustaceans, particularly zooplankton were rarely encountered, with the exception of amphipods, which were abundant (Table 4.3). Aquatic insect larvae had lower taxonomic richness than crustaceans in our samples (4 of 18 orders encountered), but were numerically more common than the crustaceans. Midge larvae (Family:

Chironomidae) were by far the most common aquatic organisms in our samples (Table

4.3). All three of the orders of mollusks we found in our samples were also relatively common numerically (Table 4.3). Most of the terrestrial organisms collected were either flying insects or tree seeds, primarily birch and elm (Table 4.3, Figure 4.2).

The average number of aquatic organisms transported on the boats and trailers we washed was 37.2. We cannot calculate the variability around this mean (e.g., standard error of the mean) because of the lost replicate identity that resulted from our pooling of

Number boatsof

0 2 4 6 8 10 12 14

0 20 40 60 80 Mass of attached vegetation (g) B)

Number fragmentsof

0 5 10 15 20 25 30

0 10 20 30 40 50 60 70 Mass of vegetation fragment (g)

200 nn == 491125 Terrestrial Organisms TerreTerrestrialstrial Seedsand Aquatic Seeds Aquatic Organisms (Miscellaneous) Zooplankton 150 Aquatic Mollusks

….. Aquatic Insect Larvae

nn === 36606 100

ofNumberOrganisms 50

0 DepartingArriving DepartingArriving Direction of Travel

Figure 4.2. Average number and type of small-bodied organisms washed from recreational boats and trailers arriving at (n= 6 filters; 36 boats washed) or departing from (n = 11 filters; 49 boats washed) lakes in the northern Wisconsin and the Upper Peninsula of Michigan. See Table 4.3 for further detail on taxa included in each taxonomic category.

TABLE 4.2.

AQUATIC PLANT SPECIES AND

THE RESPECTIVE NUMBER OF FRAGMENTS OF EACH

FOUND ON BOATS AND TRAILERS

DURING OBSERVATIONAL FIELD SURVEY

IN NORTHERN WISCONSIN

IN SUMMER 2006

Plant Species # Fragments Vallisneria americana 18 Potamogeton gramineus 9 Ceratophyllum demersum 8 Myriophyllum heterophyllum 7 Potamogeton pusillus 5 Potamogeton zosteriformis 5 Elodea canadensis 4 Najas sp. 4 Potamogeton richardsonii 2 Potamogeton robinsii 2 Zosterella dubia 2 Chara sp. 1 Potamogeton amplifolius 1

TABLE 4.3.

TAXA COLLECTED FROM BOATS AND TRAILERS DURING FIELD SURVEY IN NORTHERN WISCONSIN IN 2006

Category Order Suborder Family Genus Instar Common Name Total number collected from 85 boats washed (estimated from sub-samples) Aquatic (Miscellaneous) Amphipoda Adult Amphipod 209 Isopoda Adult Isopod 3 Oligochaete Adult Freshwater 3 segmented worm Ostracoda Adult Ostracod 3 Prostigmata Adult Water mite 206

86 Aquatic insect larvae Diptera Tipulidae Larval Cranefly 56 Diptera Chironomidae Larval Midge 740 Diptera Cuculidae Larval Mosquito 1 Ephemeroptera Baetidae Larval Baetid mayfly 18 Ephemeroptera Larval Other mayfly 65 Odonata Zygoptera Larval Damselfly 18 Odonata Anisoptera Larval Dragonfly 89 Trichoptera Hydropsychidae Larval Caseless caddisfly 24 Trichoptera Leptoceridae Larval Leptocerid caddisfly 18 Trichoptera Larval Other caddisfly 50 Aquatic mollusks Mesogastropoda Viviparidae Campeloma Adult Campelomid snail 191 Pulmonata Planorbidae Adult Planorbid snail 18 Pulmonata Physidae Physa Adult Physid snail 228 Sorbeoconcha Hydrobiidae Amnicola Adult Amnicola snail 314 Zooplankton Calanoida Adult Calanoid copepod 6 Cladocera Bosminidae Bosmina Adult Waterflea 27

TABLE 4.3 (Continued) Category Order Suborder Family Genus Instar Common Name Total number collected from 85 boats washed (estimated from sub-samples) Cladocera Daphniidae Daphnia Adult Waterflea 12 Cladocera Sididae Diaphanasoma Adult Waterflea 27 Cladocera Adult Waterflea 1 Cyclopoida Adult Cyclopoid copepod 6 Phylum: Rotifera Adult Rotifer 6 Subclass: Copepoda Larval Copepod nauplius 3 Subclass: Copepoda Adult Copepod 6 Terrestrial (Miscellaneous) Araneae Adult Spider 205 Coleoptera Adult Beetle 53 Coleoptera Larval Beetle 62 Collembola Adult Springtail 51 Diptera Adult Other dipteran 153 Diptera Drosophilidae Drosophila Adult Fruit fly 6 Diptera Ceratopogonidae Adult Gnat 285 87 Diptera Muscidae Adult Housefly 123 Diptera Chironomidae Adult Midge 695 Diptera Cuculidae Adult Mosquito 458 Diptera Ichneumondiae Adult Ichneumonid wasp 200 Ephemeroptera Adult Mayfly 3 Homoptera Aphididae Adult Aphid 6 Homoptera Cicadelliae Adult Leafhopper 17 Homoptera Adult True Bug 14 Hymenoptera Formicidae Adult Flying ant 117 Hymenoptera Formicidae Adult Ant 342 Hymenoptera Halictidae Adult Sweat bee 6 Ixodida Adult Tick 294 Trichoptera Adult Caddisfly 9 Terrestrial seeds Fagales Betulaceae Betula Seed Birch tree seed 2931 Rosales Ulmaceae Ulmus Seed Elm tree seed 3596

multiple boats on to each filter and because of the uneven pooling of these samples (i.e., not every filter had the same number of boats washed on to it).

4.4.2 Mail Survey

More than half (58%) of the registered boaters responding to our survey reported that they keep their boat in the same waterway and therefore do not pose any risk of transporting AIS overland. The other 42% of respondents were transient boaters who launched their boat in multiple waterways during the boating season. For these boaters, the average number of different waterways in which they launched their boat in a two- week period was 2.66±0.14 (mean±SE). Of transient boaters, 27% said they always wash and/or dry their boat before launching it in a different waterway, 34% did this sometimes, and 34% never cleaned their boat (Table 4.1). For reasons that are unknown to us, the remaining 5% said that boat cleaning was not applicable to them.

The majority (57%) of transient boaters reported always removing aquatic weeds when noticed from their boats and trailers, but 14% said they did so only sometimes and 13% said they never removed aquatic weeds when they saw them (Table 4.1). The remaining

16% indicated that weed removal was not applicable to them, presumably because they never saw aquatic weeds attached to their boat or trailer. Thus, 68% of transient boaters did not always wash or dry their boat when moving it overland among waterways and

27% did not always remove from their boat and trailer the aquatic weeds they see attached.

4.4.3 In-person Northwoods Survey

Of the recreational boaters interviewed in person, most (68%) reported keeping their boat on a single lake for the entire season (e.g., spend summer camping by the only lake on which they launch their boats), and thus pose a low risk of spreading AIS. In our survey, a total of 135 recreational boaters (32%) reported using their boats at multiple lakes during the summer of 2007. When asked about AIS hygiene practices, 87% percent of recreational boaters reported always removing aquatic plants that they noticed attached to their boat or trailer, but 33% never pressure wash their boat or trailer (Table 4.1). In contrast to recreational boaters, professional fishing guides (n = 35 surveys) reported visiting nearly two times as many unique lakes in a two-week period (5.41±0.80 vs.

2.72±0.42 lakes, mean±SE) (Table 4.1). Furthermore, fishing guides were less likely than others to always clean their boats with washing and/or drying when moving between waterways (11% vs. 24%) (Table 4.1). Fishing guides were also less likely to always notice aquatic weeds attached to their boats or trailers than recreational boaters (11% vs.

42%), but guides were more likely than others to always remove the weeds that they saw

(96% vs. 87%) (Table 4.1).

4.4.4 Experiment

High-pressure washing, and visual inspection combined with hand removal, removed a significantly greater percentage of macrophyte vegetation than low-pressure washing (F2, 36=21.1, p<0.001; >80% vs. ~63%; Figure 4.3A). High-pressure washing removed a significantly higher percentage of small-bodied organisms (i.e., wetland plant seeds and Bythotrephes longimanus) than did low-pressure washing or visual inspection

plus hand removal (F2, 36=15.4, p<0.001; 90% vs ~75%; Figure 4.3B). Duration of cleaning effort (90 vs. 180 s) did not significantly affect the percent removal of biological materials in either experiment (macrophytes: F1, 36=0.81, p=0.37; small-bodied organisms: F1, 36=1.68, p=0.20; Figure 4.3). There was also no significant interaction between cleaning method and duration of effort in either experiment (macrophytes: F2,

36=0.30, p=0.74; small-bodied organisms: F2, 36=0.26, p=0.77; Figure 4.3).

4.5 Discussion

Widespread recognition that overland movements of boats are often responsible for spreading invasive plants (Buchan and Padilla 2000, Puth and Post 2005) and animals

(Johnson et al. 2001, Muirhead and MacIsaac 2005,Keller and Lodge 2007) has prompted increased management concern. To date, however, management actions have largely focused on mitigating the impacts of these AIS through control and eradication efforts once they are already established in a body of water and inflicting harm there (Simberloff et al. 2005, Lovell et al. 2006). Prevention efforts have been rarer, and most that have been implemented concentrate on attempting to educate boaters about how individuals can reduce their likelihood of being a vector

(http://www.uwsp.edu/cnr/uwexlakes/CBCW/).

There are, however, no published studies that rigorously quantify the effectiveness of such education efforts in slowing spread. Management actions specifically aimed at removing AIS from transportation pathways, such as recreational boats and trailers, may be a complementary and efficient way to reduce the spread of AIS

(Lodge et al. 2006, Drury and Rothlisberger 2008). 90

100 A) b b b b 80 a a

100 B) b b a a 80 a a 60

Percent Removed Percent(Mean±SE) Removed 40

0 Short Long Short Long Short Long Low High Visual

Cleaning Method and Duration

Effectively managing the risk of AIS spread by small-craft boaters requires increased knowledge about what organisms are being transported, who is transporting organisms (i.e., how various sub-groups of boaters differ relative to their risk of transporting organisms), how often organisms are being transported, and how effectively various boat cleaning alternatives remove potentially harmful organisms from the pathway. Recent research efforts with implications for such decision-making have focused on predicting spread based on network models of boater traffic among lakes

(Leung et al. 2004, 2006; Drury and Rothlisberger 2008). For example, Drury and

Rothlisberger (2008) demonstrated that for a wide range of hypothetical cleaning efficiencies (i.e., percentage of organisms removed through cleaning of boats and trailers) placing a given number of inspection and cleaning stations at invaded lakes slows landscape-level spread of AIS more effectively than placing the same number of stations at uninvaded lakes. Implicit in this and similar modeling efforts, however, are assumptions about the types and quantities of organisms being transported and about the ability of cleaning efforts to remove them from boats and trailers. This study provides some of the empirical data that was previously lacking, including the types of organisms boaters in the Upper Midwest transport, the quantity of organisms boaters transport, and the effectiveness of various boat cleaning techniques. Our hope is that these data will inform improved risk management of AIS spread.

We found that organisms that are evolutionarily and morphologically similar to

AIS in the Great Lakes region (e.g., Eurasian watermilfoil, spiny waterflea, and

Echinogammarus ischnus) are being transported on small-craft boats and trailers (Table

4.4). Because we did not specifically target lakes known to have multiple invaders (only

one of the two study lakes was known to harbor one invasive species---spiny waterflea), it was not surprising that we did not sample any animal or plant AIS. We did however sample several taxa similar to invaders known to be spreading in the region, e.g., spiny water flea and New Zealand mud snail, including the cladoceran Diaphanosoma spp., and several types of aquatic gastropods (e.g., hydrobids and physids). We also collected amphipods in our filter samples, suggesting that the non-native amphiphod

Echinogammarus ischnus that is currently in the GL could be spread to inland lakes by boaters.

Regarding our estimate of the number of organisms on each boat, it is important to note that because we pooled samples unevenly, we could not calculate the margin of error for this estimate. Our results nevertheless provide the first estimate of the number of small-bodied organism transported on the exterior of recreational boats and trailers.

Future studies that do not pool samples would be worthwhile to better understand the variability in the number of organisms attached to any given boat and its trailer.

As with small-bodied organisms, no invasive macrophyte species were collected from boats during our field survey, but the species we collected are representative of common aquatic vegetation communities in Northwoods lakes (e.g., Vallisneria americana, Potamogeton gramineus, and Ceratophyllum demersum; Wagner et al. 2007).

As with small-bodied AIS, we would have been surprised to collect any invasive macrophytes, such as M. spicatum, in our samples, because the lakes where we washed boats are not known to contain invasive macrophytes. This expectation applied also to arriving boaters, because none of the nearby lakes (i.e., within 15 mile radius) had invasive macrophyte populations that were not under chemical control. If we had been

working on a lake with a population of M. spicatum, it is highly likely that we would have found milfoil on boats, perhaps in even greater quantities than the native vegetation we found, given the tendency of M. spicatum to form dense mats of vegetation on the water’s surface, enabling entanglement on boats (Smith and Barko 1990). Nevertheless, the native vegetation we found on boats is a useful surrogate for demonstrating the propensity of small craft to transport aquatic vegetation over land.

Despite many years of campaigns to educate boaters on how to avoid transporting organisms, our results demonstrate that overland transport of aquatic organisms by boaters still occurs frequently. If relatively diffuse educational campaigns stimulated boaters to take responsibility for their own boat hygiene, it would be a relatively inexpensive way to save the public the expense of equipment and employees required to clean boats. However, from our data on self-reported cleaning rates and our observations of organisms attached to boats and trailers, it does not appear that existing and previous education campaigns have yet resulted in consistently high cleaning rates by boaters or in the use of highly effective cleaning practices in the region we studied.

In Michigan and Wisconsin, states where educational efforts have been among the most vigorous in the US, two-thirds of the boaters who responded to our surveys do not always clean their boat when moving to another waterway, and more than a quarter do not always remove aquatic weeds when they see them attached to their boat or trailer.

TABLE 4.4.

NONINDIGENOUS SPECIES ESTABLISHED IN THE GREAT LAKES THAT ARE

MORPHOLOGICALLY SIMILAR TO SPECIES COLLECTED IN BOAT WASHING

SAMPLES.

Morphological Selected nonindigenous taxa Representative taxa Category/ in Great Lakes collected in boat washing Description samples Plankton Bythotrephes longimanus (spiny waterflea), Bosmina spp., Daphnia Cercopagis pengoi (fish-hook waterflea), Daphnia spp., Diaphanasoma spp., galeata galeata, Daphnia lumholtzi, Eubosmina Rotifers, Copepods coregoni, Eubosmina maritime, Copepods (5 spp.), Diatoms (17 spp.), Green alga (4 spp.)

Small benthic Echinogammarus ischnus, Hemimysis anomala Amphipoda, Isopoda, crustaceans and (bloody-red shrimp), Gammarus tigrinus macroinvertebrates

Small benthic Potamopyrgus antipodarum (New Zealand mud snail), Campeloma spp., Physa mollusks Dreissena polymorpha (zebra mussel), Dreissena spp., Amnicola spp. rostriformis bugensis (quagga mussel), Corbicula fluminea (Asiatic clam), Viviparus georgianus, Valvata piscinalis, Bithynia tentaculata, Sphaerium corneum, Pisidium henslowanum, Pisidium supinum, Cipangopaludina chinensis malleata, Pisidium amnicum, Pisidium moitessierianum, Elimia virginica, Gillia altilis

Other benthic Oligochaetes (6 spp.: Branchiura sowerbyi, Gianius Oligochaetes organisms aquaedulcis, Potamothrix bedoti, Potamothrix moldaviensis, Potamothrix vejdovskyi, Ripistes parasita)

Macrophytes Potamogeton crispus (curlyleaf pondweed), Vallisneria americana, Myriophyllum spicatum (Eurasian waterfoil), Potamogeton spp., Hydrocharis morsus-ranae (European frogbit), Ceratophyllum Cabomba caroliniana (fanwort) demersum, Myriophyllum heterophyllum

This is not highly surprising in that social marketing research indicates that rates of behavioral change are relatively low in cases where compliance benefits society, but where the individual who is being asked to take action receives little or no immediate benefit or gratification, particularly when the desired action is inconvenient to the individual (McKenzie-Mohr 2000). As this is the situation with boat cleaning, it is likely that to achieve high compliance rates educational efforts will need to be augmented with staffed cleaning stations placed at strategic locations and, possibly, enforcement and disincentives for non-compliance (i.e., fines). Two US states in the Great Lakes region have already put laws in place making it illegal to launch a boat if there are potentially invasive aquatic species attached to the boat, trailer, or other equipment (Wisconsin Act

16, Section 30.715; Minnesota Statute 84D). Enforcement of these laws, however, remains a challenge and the strategic deployment of boat cleaning and inspection stations could be an efficient way to help increase compliance substantially.

Our experimental results can help guide choices about the kind of inspections and boat cleaning that may be most appropriate to a given situation. Understanding how species’ characteristics affect the removal rates of those species from boat and trailers is an important factor in selecting effective cleaning techniques. We found that transport of high-risk macrophytes can be prevented with a high probability through visual inspection and hand removal. However, visual inspection failed to detect small-bodied organisms, seeds, and resting stages of other invasive species. Examples of small-bodied organisms in the GL region include the spiny and fish-hook waterfleas, Bythotrephes longimanus and Cercopagis pengoi, respectively, or even smaller, the deadly fish pathogen VHSv. If the spread of such small biological materials and organisms is a concern, visual

inspection will not provide detection and removal with high probability. Overcoming this limitation, high-pressure washing can remove over 90% of small-bodied organisms, making it the most effective option for preventing the transport of small organisms. The failure of visual inspection to detect a high percentage of small-bodied organisms is not surprising, but it is troubling because visual inspection of incoming boats and trailers is by far the most common type of government-sponsored or volunteer-organized intervention employed at boat ramps in the GL region. During our field inspections of boats, we noticed that clear visual clues (e.g., mud or foam deposits), aside from the discovery of macrophytes, that small-bodied organisms might be attached to a boat or trailer were rare. Thus, it is unlikely that visual inspections in practice under field conditions will discover and prompt removal of small-bodied organisms at a rate any higher than the ~63% rate in our experimental trials, particularly because the gel we used to adhere organisms to our boat and trailer was not as visible as mud would have been.

A limitation of our study, which may mean that our experiment overestimated the effectiveness of all boat hygiene methods, is that we were focused on techniques to clean only boat hulls and trailers. We did not sample the interior surfaces or standing water in boats. These surfaces and water reservoirs include carpets, live wells, bait buckets, and bilge water, all of which could harbor AIS, especially small-bodied organisms. In fact, spiny waterflea have been found in bilge water samples (J. Muirhead, U. Alberta, personal communication), and the release of bilge and live well water from lakes where

VHSv is present into uninfected lakes may be a key vector in the spread of this deadly fish pathogen (Wisconsin Natural Resources Rule FH-40-07(E)). The prevalence of transport of VHSv and other pathogens in water held in boats merits further investigation,

as does the effectiveness of washing in removing pathogens from the exterior and interior of boats.

For efficient risk management of the spread of AIS by small-craft boaters, it is also important to determine if any sub-groups of boaters pose a disproportionately greater risk of transporting organisms among waterways. Our surveys indicated the existence of three different categories of boaters, for which management attention might appropriately differ. First, the majority of boaters (mail survey: 58%, in-person survey: 68%) keep their boat on the same body of water during the entire boating season and, therefore, pose a minimal threat for the overland spread of AIS. Second, in both the mail and in-person surveys, transient boaters reported visiting approximately three different waterways during a two-week period of the boating season, indicating a higher probability of AIS spread. Third, the professional fishing guides we surveyed reported visiting an average of more than five different waterways every two weeks. These data suggest that fishing guides pose the greatest risk of AIS spread, especially because they did not employ effective boater hygiene practices at a higher rate than other boaters. Focused efforts to ensure the inspection and cleaning of these most frequently moving boats-- which may be analogous to superspreader individuals, i.e., individuals with many topological connections on the transmission network, in the human disease context (Riley 2007)--- would likely pay high dividends in slowing AIS spread.

Our findings lead naturally to two major recommendations for management actions to slow the spread of AIS from the GL to inland waterways and among inland waterways in the GL region. First, it is clear that some boaters pose a greater risk of spreading organisms than others. This suggests that increased management attention to

identify and communicate with high-risk boaters could be an important way to reduce the spread of AIS. Such outreach would require more targeted efforts than the broad educational efforts that have been employed previously. Our survey data suggest that professional fishing guides are one sub-group of small-craft boaters that move among waterways with extraordinary frequency and who currently employ less than ideal boat- cleaning practices.

Second, knowledge of the geographic location of invasive species within a region should inform efforts to manage the risk of future spread. Indeed, landscape-level approaches are increasingly recognized as highly important for effective management of natural resources, particularly aquatic ones (Post et al. 2008, Drury and Rothlisberger

2008, Vander Zanden et al. 2008). Given our experimental results, for example, knowing which lakes contain small-bodied AIS versus which contain only invasive macrophytes will have obvious implications for the type of boat cleaning strategy employed to keep organisms from being transported away from already invaded lakes. In the GL region there are particular locations where high-pressure washing would be useful. Such sites include high-traffic boat landings on the GL (e.g., landings near major cities such as

Green Bay, Cleveland, Chicago, and Toronto) in which numerous small-bodied AIS (e.g.,

Bythotrephes, Hemimysis) are present, inland waterways currently invaded with spiny waterflea Bythotrephes longimanus (e.g., L. Gogebic, Gile Flowage, WI), and waterways where VHSv is known to occur (e.g., L. Winnebago, WI). For inland waterways that harbor only invasive macrophytes, visual inspection and hand removal at the landings of these sites would effectively contain the spread of such plants.

4.6 Acknowledgements

We thank Mark Fedora, Nick Schmal, and Mike Ielmini of the US Forest Service for their help in organizing this project. Sheilah Kennedy and Jeff and Richard Delfeld of

SK-Environmental (Okanogan, WA) furnished the portable wash-reclaim system and gave essential technical support and field assistance. Ted Ritter, AIS Coordinator for

Vilas County, WI, provided invaluable logistical and field support. Special thanks are due to Jody Peters for help identifying macrophytes and to Kevin Pangle for supplying

Bythotrephes for our experiments. Elizabeth D. Tucker helped to design the mail survey.

Reuben Keller’s help was crucial to the successful implementation of the in-person surveys. We are grateful to Brandon Feasel and Neil Wallace for personally interviewing scores of boaters and to Matt Barnes, Mike McCann, Sarah Sutton, and Tim Campbell for their tireless lab work. Funding from the National Science Foundation for the ISIS project (DEB 02-13698 to DML), the Great Lakes Protection Fund (Grant #797 to DML), and a University of Notre Dame Schmitt Fellowship (to JDR) made this work possible.

100

CHAPTER 5:

LIMITATIONS OF GRAVITY MODELS IN PREDICTING THE SPREAD OF

EURASIAN WATERMILFOIL (MYRIOPHYLLUM SPICATUM)4

5.1 Abstract

The impacts of invasive species are costly and environmentally damaging, but resources to slow their spread and reduce their impacts are scarce. Accurately forecasting the locations most likely to experience new invasions could improve management actions aimed at slowing spread. Beyond identifying vulnerable sites, evaluating the efficacy of various strategies and approaches to removing invasive species from the transportation pathways that spread them could guide the allocation of resources. Here, we assess the predictive accuracy of models of the overland movement of aquatic invasive species on trailered boats in identifying where new invasions will occur. We performed this assessment for the specific case of Eurasian watermilfoil invasions (Myriophyllum spicatum) in Wisconsin. The predictive model is based on milfoil occurrence data in

Wisconsin lakes larger than 25 hectares (n=1803) between 1990 and 1999. We derived propagule pressure estimates from gravity models of boater movements. We investigated the predictive accuracy of our fitted model by forecasting the annual probability of

4 The publication status of this chapter is: Rothlisberger, J.R. and D.M. Lodge. Limitations of gravity models in predicting the spread of Eurasian watermilfoil (Myriophyllum spicatum). Ecological Applications (in review). 101

invasion at lakes for each year between 2000 and 2006, and comparing the modeled probabilities to the actual occurrence of invasions. We tested the hypothesis that our model accurately predicted how many of the 200 lakes predicted as the most likely to be invaded would actually be invaded. Our analysis rejected this hypothesis, concluding that a lake’s predicted probability of invasion was not correlated with whether or not a lake actually became invaded. Given the low predictability of lake-specific invasions, we then compared the efficacy of a prevention strategy that does not require predictions (i.e., containing invaded sites) with one that does (i.e., protecting uninvaded sites). We show that containing invaded sites reduces the likelihood of spread more than protecting uninvaded sites at all effort levels. We suggest that although a model commonly used to predict invasions is not a universally reliable tool for identifying which sites should be protected, efforts to slow the spread of invasive species should not be discouraged because preventing the transport of organisms away from invaded lakes is highly effective and does not require such predictions.

5.2 Introduction

Invasive species are a serious environmental problem, causing ecological and economic harm and posing difficult challenges to natural resource managers and policy makers. Invasive species are important drivers of environmental change, including reductions in global biodiversity (Sala et al. 2000) and alterations of ecosystem structure and functioning (Crowl et al. 2008). The most comprehensive estimates value the damage from invasive species at over $100 billion per year in the US alone (Pimentel et al. 2000, 2005). While humans suffer some of the harm invasive species cause, humans 102

are also responsible for promoting their spread. Indeed, human transportation networks provide ready pathways for the translocation of invasive species, regionally and globally

(Lodge et al. 2006). With the number of invasions growing and financial resources for invasion management becoming increasingly limited, greater attention to the strategic allocation of resources is required (Hoagland and Jin 2006, Lodge et al. 2006, Finnoff et al. 2007).

For management efforts to achieve their desired goals, it is necessary to determine how best to reduce the probability of invasion and slow the rate of spread (Muirhead et al. 2009). One of the most effective ways to do this is to focus on pathways of spread

(Lodge et al. 2006, Hulme et al. 2008). The development of models of species dispersal via various pathways is beneficial in supporting this focus. Pathway-based models of species dispersal can provide estimates of propagule pressure to uninvaded locations.

Accurately quantifying propagule pressure is important to predicting when and where invasions will occur (Lockwood et al. 2005, Cohen et al. 2007). At local scales, propagule pressure is the single most important predictor of establishment of invading species and if high enough can overwhelm environmental resistance to species invasions

(VonHolle and Simberloff 2005).

Here, we assess the predictive accuracy of models used to describe the volume of propagules of a particular class of aquatic invasive species---invasive macrophytes--- delivered to susceptible sites---inland lakes---via a particular pathway---overland movements of recreational boats. Recreational boaters are an important dispersal mechanism for multiple aquatic invasive species (AIS) (Puth and Post 2005) including zebra mussel Dreissena polymorpha (Johnson et al. 2001), spiny waterflea Bythotrephes

103

longimanus (MacIsaac et al. 2004), and Eurasian watermilfoil Myriophyllum spicatum

(Johnstone et al. 1985, Buchan and Padilla 1999). In this paper, we focus on invasive aquatic plants because they are geographically widespread, affecting natural waterways in every state in the US. Additionally, the impacts of invasive aquatic plants are often highly visible and damaging, and therefore garner substantial attention from the public and from managers (Skogerboe et al. 2003). Among their numerous impacts, invasive macrophytes often reach nuisance levels of abundance (Unmuth et al. 2000): displacing native vegetation (Eiswerth et al. 2000), altering habitat structure for fish and invertebrates (Warfe and Barmuta 2004, Cheruvelil et al. 2002), impairing navigation

(Skogerboe et al. 2003), and lowering property values (Halstead et al. 2003). Tens of millions of dollars are spent annually by states, lake associations, and private citizens in the US to control invasive macrophytes. Among the species of invasive macrophytes in the US are hydrilla (Hydrilla verticillata), water hyacinth (Eichhornia crassipes), and

Eurasian watermilfoil (Myriophyllum spicatum). The particular species we study here is

M. spicatum. Myriophyllum species can reproduce vegetatively and, even under dry conditions, M. spicatum can remain viable out of water for seven to nine hours, and even longer given favorable temperature and humidity conditions, making it a good candidate for overland dispersal by recreational boaters (Dove and Wallis 1981, Buchan and Padilla

1999).

Investigations of the human-mediated spread of AIS have predominantly modeled overland movements of recreational boaters using gravity models (Schneider et al. 1998;

Bossenbroek et al. 2001, 2007; MacIsaac et al. 2004; Muirhead and MacIsaac 2005;

Leung et al. 2006). Gravity models are a type of spatial interaction model, analogous to

104

Newton’s laws of attraction, which describe how separation distance and site attractiveness affect the geographic movement of people (Thomas and Hugget 1980).

Gravity models are attractive for studying the spread of AIS because they provide a mechanistic representation of a primary vector of AIS transport (i.e., inter-waterway movements of recreational boaters) and therefore may provide a reliable index of AIS propagule pressure. Moreover, the data needed to create a gravity model are relatively easy to obtain, requiring as little as lake locations (latitude and longitude), lake characteristics (e.g., surface area), and boater registration information (Leung et al.

2006).

In some cases, previous studies have used gravity models to forecast risk of invasion by various AIS (e.g., Eurasian watermilfoil, zebra mussel) at the scale of counties (Buchan and Padilla 1999, Bossenbroek et al. 2001). To provide practical guidance for management efforts, however, predictions of probability of invasion for individual lakes are desirable because decisions about how to manage the risk of invasion are often made at the scale of individual lakes. As such, several investigations using gravity models have reported lake-specific risk forecasts (Leung et al. 2004, 2006;

Muirhead and MacIsaac 2005; Bossenbroek et al. 2007). These forecasts may provide cost-effective management guidance by identifying which waterways have the highest probability of future invasions, flagging these locations as high-priority sites for protective screening to guard against AIS introductions. It is, however, possible that gravity models could misdirect management efforts if the sites identified as having the highest probability of invasion do not closely correspond with the sites that actually would have been invaded in the absence of protective management.

105

Previous tests of the accuracy of gravity model predictions have indicated that gravity models perform generally well in predicting where invasions will occur

(Schneider et al. 1998, Bossenbroek et al. 2001, Leung et al. 2004). The spatial and temporal resolution of the assessments used to test these models, however, were not entirely analogous to the ways in which the models would most likely be expected to perform in management applications. Bossenbroek et al. (2001) made and tested predictions at the spatial resolution of counties. Schneider et al. (1998) and Leung et al.

(2004) generated predictions for individual lakes, but assessed model predictions aggregated over multiple years. For management applications (e.g., the stationing of inspection crews or boat cleaning stations), gravity models would ideally be able to provide reliable predictions of which lakes are most likely to be invaded in the upcoming year. This study is the first to perform an assessment of the predictive accuracy of gravity models at the scale of individual lakes in individual years.

Bossenbroek et al. (2007) recently used a gravity model to correctly predict that the Lake Mead watershed was among the most likely of any western watershed to be invaded by dreissenid mussels. The situation that Bossenbroek et al. (2007) modeled, however, was different from the one we address here: for the western spread of dreissenid mussels, there were many invaded eastern watersheds sending propagules to relatively few western destinations. In contrast, in the situation we model here, there are relatively few invaded sources of propagules and many uninvaded sites that could become invaded.

This is the more common situation, especially early in any invasion when management efforts can be very cost effective (Lodge et al. 2006).

106

Thus, we seek to assess the predictive accuracy of gravity models in forecasting the probability of invasion in a case where there are many potential destinations for colonization and relatively few sources of propagules. Our primary aim here is to evaluate, at an appropriate spatial and temporal scale, the reliability of gravity models in guiding management decisions. Our results suggest that an alternative approach— preventing the egress of propagules from invaded sites—is likely to be more cost effective than trying to prevent the arrival of propagules at sites identified by gravity models as having a high risk of invasion.

5.3 Methods

Using data on the road network, lake locations, boater registration addresses, and the sequence of M. spicatum invasions in Wisconsin’s waterways, along with parameters from the literature, we created a gravity model of recreational boater movements to provide an index of propagule pressure from invaded to uninvaded sites. To relate predicted boater traffic to the likelihood of invasion, we used survival analysis to fit a statistical model of this relationship. This allowed us to evaluate the average probability of invasion and the range of probabilities of invasion in our system of lakes, given our modeled predictions of boater traffic.

Next, to assess the predictive accuracy of our models, we categorized uninvaded lakes according to their modeled probability of invasion, retrospectively comparing these categorizations to where invasions actually occurred. We tested the null hypothesis that of the 200 lakes identified by our gravity model as being the most likely to be invaded, the number that actually became invaded was not different than the number expected to 107

be invaded given the average predicted probability of invasion for these 200 lakes. We chose to evaluate the 200 sites with the highest probability of invasion because 200 is a realistic number of lakes at which protective management could be implemented across a state. We also assessed our models using previously published diagnostic tests (Leung et al. 2004), comparing these with the findings of our hypothesis-testing diagnostic approach.

5.3.1 Relationship between propagule pressure and probability of establishment

To estimate the probability of invasion at lakes in Wisconsin currently uninvaded by M. spicatum, we used data on the number of boaters in each county and values from the literature to parameterize a gravity model of recreational boater overland movements from invaded to uninvaded lakes. The structure of the gravity model we used was identical to that used in previous investigations of the spread of AIS by recreational boaters (Leung et al. 2004, 2006). It is a production-constrained gravity model that relates the number of visits (Tj) by recreational boaters to a particular destination waterway (j) to the number of boaters (Oi) in a source county (i), the surface area of the destination waterway (Wj), the distance between each source and destination (Dij), and the availability of other destinations (Ai). These relationships are expressed mathematically as

K = −d UAOWDj ∑ i i j ij (5.1) i=1 and

= 1 Ai L (5.2) −d ∑WDj ij j=1 108

where Uj is an index of boater traffic to a specified destination waterway, K is the number of source counties, L is the number of waterway destinations, and d is a distance coefficient that describes the relationship between separation distance and travel among locations. We followed the equations and approach of Leung et al. (2004, 2006) to generate an index of propagule pressure from invaded waterways to uninvaded ones through time, with propagule pressure to uninvaded lakes increasing over time as more lakes were invaded.

Data on the number of boaters in each county came from the statewide boater registration database obtained from the Wisconsin Department of State

(http://www.sos.state.wi.us/). We obtained data on lake surface areas from the National

Hydrographic Dataset (http://nhd.usgs.gov/). For modeling traffic to boat ramps on the

Great Lakes, we did not use the surface area of the Great Lakes as Wj. Instead, we used a fitted value obtained from the literature that equated a single Great Lake boat ramp to a body of water with a surface area equal to g = 16000 ha (Leung et al. 2004, Leung et al.

2006). We also obtained from the literature an estimate of the distance coefficient for the

Great Lakes, d = 2.15 (Leung et al. 2004).

The Wisconsin Department of Natural Resources supplied detailed data on the location and date of discovery of M. spicatum invasions at all lakes larger than 25 ha in the state from 1990 to 2006 (Figure 5.1). The number of lakes larger than 25 ha in

Wisconsin is 1803, of which 335 were known to be invaded by M. spicatum as of 2006.

We used the first 8 years of these data to approximate the functional relationship between propagule pressure, indexed by boater traffic as predicted using the gravity model, and the probability of M. spicatum establishment. To fit this relationship of probability of

109

350

300

250

200

150

100

Invaded (Number) lakes 0

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Year

establishment versus propagule pressure, we used a metric derived from survival analysis

(Collett 2003): the negative log-likelihood of the joint probabilities of each site remaining uninvaded as long as it does given its history of propagule pressure is minimized using the simplex algorithm (Press 1992). Following the literature, we assumed the standard asymptotic curve as the basic functional relationship between probability of establishment (P(Est)) and propagule pressure (N):

α P(Est)=1− e-Nl,t , (5.3) where N is the number of propagules arriving at location l at time t and α is a coefficient describing the shape of the curve and is optimally fit via the simplex algorithm (Leung et

110

al. 2004, Jerde and Lewis 2007). This approach produces the modeled fit with the strongest statistical support given the observed data (Collett 2003).

5.3.2 Model validation

To validate our model, we tested its ability to predict known invasions that occurred from 2000 to 2006 using the shape parameter relating propagule pressure to probability of establishment (i.e., α) found using invasion data from 1990 to 1999. For each year in the validation dataset, we calculated the probability of invasion in the next year for each uninvaded site, given the number and location of currently invaded sites and boater movements among sites. This gave predictions for each waterway’s probability of invasion for each year of the validation dataset. Using three different approaches, we compared these predictions with data on actual invasions for each year to assess the model’s predictive accuracy.

First, to objectively compare the ability of our model to accurately assess risk of

M. spicatum invasion with that of a similar modeling effort that assessed zebra mussel invasion risk (Leung et al. 2004), we used the same model assessment method as the previous study. Under this method, we developed a null (random) model of risk that assumed that all lakes are equally likely to be invaded during the period under consideration (i.e., as if lake-specific information were unavailable). Under the null model, the per lake probability of invasion between 2000 and 2006 was calculated as the number of lakes that actually became invaded in a given year divided by the initial number of uninvaded lakes available for invasion in that year. We used our model and

111

the null model to predict the top 10, 100, and 200 lakes at risk of being invaded and compared these predictions with actual invasions occurring from 2000 to 2006, inclusive.

Second, to further assess our model’s ability to discriminate between high- and low-risk sites, we used the Area Under the receiver-operator characteristic Curve (AUC) metric (Jerde and Bossenbroek 2009). The AUC is a plot of the sensitivity of a model’s predictions versus the complement of its specificity at a series of thresholds for a positive outcome. A positive outcome is defined as the model predicting that a site will be invaded given a particular probability of invasion threshold. True positive outcomes occur when these predictions actually come to pass. The sensitivity of a model is the ratio of true positives (i.e., actual invasions) to total positives (i.e., predicted invasions).

Model specificity is the ratio of true negatives (i.e., sites predicted to remain uninvaded that actually do remain uninvaded) to total negatives (i.e., true and false negatives)

(Hosmer and Lemeshow 2000). Therefore, an AUC of 0.5 would mean the discriminatory ability of a model is only as good as flipping a coin; if less, it is worse than random guessing.

Third, we took a novel approach to test the ability of the model to identify lakes most likely to be invaded, we categorized sites according to their modeled probability of invasion on an annual basis---the timescale at which intervention decisions are likely to be made, from high to low probability of invasion, and tabulated how many lakes in each category were actually invaded in each year. For each year, we tested the hypothesis that the number of newly invaded sites in the top 200 sites predicted to have the highest probability of invasion in a given year was no different than the number actually invaded in that year. To test this hypothesis, we calculated the probability or p-value (shown as p

112

on Figure 5.2), according to the binomial distribution, of observing the actual number of newly invaded sites given the average probability of invasion in the top 200 high- probability sites as predicted by our model. In symbols, we calculated

n! k n− k P() K= k = pˆ(1− p ˆ ) , (5.4) k!!() n− k where pˆ is the average per lake probability of invasion in the top 200 at-risk sites as predicted by our model, K is the expected number of new invasions given p, k is the number of new invasions actually observed, and n is the number of lakes being considered (n = 200). In Results, we present these three approaches in reverse order to their description here.

5.3.3 Cost-effectiveness of alternative interventions

We also modeled the outcomes of alternative intervention strategies for slowing the spread of M. spicatum by recreational boaters. The alternative interventions we modeled were the protection of lakes that have the highest probability of invasion versus the containment of lakes (i.e., prevention of propagules from leaving) most likely to supply propagules for new invasions. Our response variable was the percent reduction in the mean per site probability of invasion. We calculated the value of this response variable for all combinations of intervention site efficacy, specified as the proportional reduction in propagule load, ranging from 0.1 to 1, on an individual boat and trailer at any given intervention site and the total number of intervention sites, either at invaded or uninvaded lakes on the landscape. The number of lakes where intervention was placed in our simulations ranged from a single lake up to half of all uninvaded lakes in the system

(~700) for protection and up to the entire set of invaded lakes for containment (~335).

113

114

Thus, we were able to compare differences in average probability of invasion across the entire system for each strategy (protection v. containment) given the same effort level

(i.e., number of intervention sites and site-wise propagule removal efficiency).

5.4 Results

5.4.1 Relationship between propagule pressure and probability of establishment

As determined by the simplex optimization routine applied to our eight years of training data, the best-fitting shape coefficient for the standard asymptotic curve we used to model the relationship between propagule pressure and probability of establishment was α = 4.19E-5 (Equation 5.1).

5.4.2 Model validation

Propagule pressure, as indexed by boater traffic from invaded to uninvaded lakes, was not a good predictor of actual probability of establishment by M. spicatum in the validation portion of our dataset (Figure 5.2). If the model were doing a good job of predicting lake-specific probability of invasion, most invaded lakes would be clustered in the bins containing the lakes predicted to have the highest probability of invasion (at the left end of Figure 5.2). Instead, we found that in all years considered for model validation (2001-2006), the majority of new invasions occurred in waterways that were not in the highest probability of invasion bins (Figure 5.2). In other words, we overestimated the probability of invasion at the top 200 high-predicted probability waterways (the average of which is pˆ , Figure 5.2). Furthermore, across all sites the

115

probability of invasion was not correlated with the number of sites in each bin that were actually invaded in the next year (Figure 5.2).

We also assessed our predictive model using two other previously published diagnostic techniques. We compared the predictions of the model using probability of establishment versus propagule pressure, aggregated over all years in our validation data set, to a null model that assumed equal probability of invasion among lakes (Leung et al.

2004). Out of 1547 previously uninvaded lakes, seventy-nine new invasions occurred in

2000-2006. The null model prediction was that each lake had the same 4.4% (i.e.,

79/1547) chance of being invaded. Under the null model, if we selected 10, 100, and 200 lakes from the set of initially uninvaded lakes at random, we would expect to correctly predict the invasion of 0.44, 4, and 9 of those lakes, respectively. Using our model to rank uninvaded lakes according to probability of invasion, we correctly identified 0, 5, and 22 lakes that became invaded, respectively, among the top 10, 100, and 200 highest probability of invasion lakes. Thus, according to this diagnostic, our model provided no improvement over the null model in predicting the top ten sites most likely to be invaded, marginal improvement in predicting the top 100, but was nearly two and a half times better than the null model at identifying sites in the top 200.

Finally, the AUC diagnostic indicated that our modeled predictions of probability of invasion (with probability of invasion aggregated over all years) were substantially better than random guessing (AUC = 0.71), suggesting relatively strong discriminative power of the model (Hosmer and Lemeshow 2001, Jerde and Bossenbroek 2009).

116

5.4.3 Efficacy of alternative intervention strategies

In comparing average per site reduction in probability of invasion under alternative strategies (protect v. contain), we found that for any combination of per site intervention efficacy and number of intervention sites on the landscape, containing source populations was more effective in reducing system-wide probability of new invasions than trying to protect uninvaded sites identified by our gravity model as being at the highest risk of invasion (Figure 5.3). For example, containing 100 invaded sites with

85% efficacy reduced average per site probability of invasion 17.7% more than the same level of effort in protecting the uninvaded sites with the highest probability of invasion

(47.5% reduction in system-wide average probability of invasion vs. 29.8%). For intervention at 200 sites at 90% efficacy, containment is 27.1% better than protection

(70.0% vs. 42.9%). Put another way, to achieve a 50% reduction in average per site probability of invasion, one would either have to contain, for example, 101 invaded lakes with 85% efficacy, or protect 255 uninvaded lakes with 95% efficacy.

5.5 Discussion

Accurate a priori identification of which sites are most likely to be invaded could improve the preventative management of biological invasions (Vander Zanden and Olden

2008). Propagule pressure, a combination of the frequency with which viable organisms or resting stages are introduced into a novel environment and the number of individuals introduced with each event, strongly determines successful population establishment

(Lockwood et al. 2005, Von Holle and Simberloff 2005, Drake and Lodge 2006).

Gravity models of human transportation networks have been recommended as an 117

118

effective way to rapidly assess the propagule pressure experienced by and, hence, the vulnerability of numerous potential sites across landscapes (Bossenbroek et al. 2007,

Vander Zanden and Olden 2008). We found, however, that even if one uses a widely accepted approach to modeling propagule pressure, the ability to predict the sites of new invasions is more limited than previously recognized

Specifically, when we used a gravity model of recreational boater movements in

Wisconsin to forecast new invasions of M. spicatum one year into the future, we predicted significantly more invasions would occur in the 200 lakes predicted to have the highest probability of invasion than actually did. Furthermore, it also appears that whether a site actually became invaded or not was not correlated with its predicted probability of invasion (i.e., predicted high-probability sites did not become invaded more often than predicted low-probability sites). One important reason that it is difficult to accurately predict specific sites at which new invasions will occur is because population establishment is highly stochastic and there is a very low probability that any given introduction event will result in establishment (Lewis and Pacala 2001). Thus, while gravity models may provide an accurate index of propagule pressure, they may nevertheless fail to reliably predict where new invasions will occur in cases like the one we consider here where progagules from relatively few sources are distributed to many potential colonization sites that all receive fairly similar levels of propagule pressure (i.e., the range of difference in predicted probability of invasion at high-probability sites and low-probability sites is relatively narrow). In this type of situation, the stochastic ‘noise’ inherent in population establishment appears to overwhelm the probability-of- establishment versus propagule pressure (i.e., boater traffic) ‘signal’ generated by the

119

gravity model, leading to low predictability of the location and timing of population establishment. The consequences of high stochasticity and the low probability of establishment on the predictive accuracy of spread models have received limited attention

(but see Jerde and Lewis 2007). Our study extends previous efforts by comparing modeled predictions against actual invasions and suggests that population establishment is so stochastic that the range of probability of establishment predictions we modeled here is too small to distinguish accurately between sites that have high versus low probability of actually being invaded.

The usefulness of gravity models in predicting invasions may therefore depend strongly on the circumstances of the invasion being modeled. For example, we contrast our case with that of Bossenbroek et al. (2007), who used a gravity model parameterized for the US to accurately predict the invasion of Lake Mead by dreissenid mussels

(Stokstad et al. 2007). Our study and Bossenbroek et al. (2007) lie at nearly opposite ends of the spectrum of efforts that seek to use gravity models to forecast the location of new invasions. In the present study, the range of predicted probabilities of invasion is narrow because traffic from a few invaded sites is divided among numerous uninvaded sites. In Bossenbroek et al. (2007), the predicted probability of invasion gradient is broad because, in working at the watershed scale as they do, their gravity model concentrates recreational boater traffic originating from numerous invaded watersheds east of the 100th meridian into a relatively small number of western watersheds with enough surface water to make them attractive destinations for recreational boaters. The Lake Mead watershed, with its tremendous surface water area relative to all watersheds in the surrounding region, had a much higher probability of invasion than any other watershed near it

120

(Bossenbroek et al. 2007). Thus, it was highly likely to be invaded before any other western waterway and, in retrospect, the signal produced by the gravity model was indeed stronger than the noise associated with the uncertainty inherent in predicting where and when invasions will occur. The opposite appears to be true for predicting M.

spicatum invasions in the upper Midwest.

If we had used only the relatively general diagnostics of model performance such as the null model comparison and the AUC analysis, both of which have been presented as support for the predictive accuracy of gravity models in previous investigations (Leung et al. 2004, Jerde and Bossenbroek 2009), we would have concluded that our model assessed site-specific probability of invasion fairly accurately and, therefore, could identify waterways to be protected in a management context. However, when we assessed gravity model performance with the additional technique discussed above, one that had not previously been applied to gravity models of human-mediated AIS dispersal, we found that our gravity model’s accuracy in predicting where new invasions will occur in a given year is seriously limited. The novel diagnostic test we used summarizes the full range of data provided by the gravity model and compares it against the data on actual invasions in individual years, allowing for a hypothesis test of the correspondence between the data and the predictions. Neither the null model comparison nor the AUC analysis, tests a hypothesis about the accuracy of the model predictions. Furthermore, the null model comparison aggregates data on invasions across all years of the validation data set, as opposed to our diagnostic, which compares model predictions against validation data on a year-to-year basis. Also, the null model comparison considers only the lakes predicted most likely to be invaded, and provides no insights regarding the more

121

numerous lakes predicted less likely to be invaded. In contrast with our diagnostic, which does so graphically, the null model comparison does not detail where in the ranked probability of invasion the other lakes that actually become invaded are found.

The AUC analysis is also not an ideal diagnostic for the type of model we consider here because it glosses over details about where misclassifications occur in the probability of invasion rankings. This is because the AUC is a composite score of all misclassifications, false positives and false negatives, over the full range of probability threshold values (0, 1). As such, the same AUC may be produced by many different patterns of misclassifications (Hosmer and Lemeshow 2003) and, therefore, does not provide a clear indication as to which kind of sites are predominantly being misclassified.

For example, it is not clear from the AUC analysis if sites with a high-predicted probability of invasion are not being invaded (i.e., false positives) or if those with low predicted probabilities of invasion are being invaded (i.e., false negatives) or if there is some intermediate mixture of misclassifications. The model predictions whose accuracy are most important for guiding management decisions are that those lakes predicted to have a high probability of invasion are actually those which become invaded. In other words, false positives are the most subversive to management actions because it is at these lakes where management actions will be taken and, thus, where inaccurate predictions are likely to lead to wasted effort. Thus, while the AUC analysis is a useful metric for the overall performance of predictive models that involve classification

(Hosmer and Lemeshow 2003), it does not provide specific detail regarding misclassifications. It is this kind of specific detail that is most relevant in evaluating the

122

usefulness of gravity models for making decisions about intervention actions for slowing the spread of Eurasian watermilfoil and many other AIS.

Many decisions regarding intervention efforts are made on an annual basis. Thus, our diagnostic analysis assesses model performance at the same temporal scale at which such a model would be used to guide management decisions. Our results suggest that the uncertainty about which particular lakes will be invaded in the next year is too great to identify reliably which ones should be protected.

On the other hand, efforts to quarantine lakes where populations of invasive species are present are certain to reduce propagule transport throughout the system of lakes connected by anthropogenic pathways. This means that efforts that prevent the export of propagules from invaded sites (i.e., containment) are generally preferable in two ways to protecting uninvaded sites. First, containment offers greater reductions in the probability of new invasions at the landscape-level. Second, protecting uninvaded lakes may result in wasted effort because the probability is high that an uninvaded lake, even one predicted to have a high probability of invasion in a given year, will not be invaded, even if left unprotected.

The consistent message that emerges from our study is that preventing egress of propagules from newly invaded lakes is critical to slowing AIS spread across landscapes, thereby avoiding widespread impacts, which are known in some cases to be exceedingly economically damaging (Leung et al. 2002, Halstead et al. 2003). Our findings also emphasize the value of Early Detection/Rapid Response programs in slowing the spread of invasive macrophytes, because the sooner invaded sites are detected and managed, the less opportunity there will be for propagule egression. Consistent with the

123

recommendations emerging from our study, a theoretical study demonstrated that when less than half the lakes on a landscape are invaded, containing propagules in invaded lakes is the most effective management strategy to prevent landscape-level spread of invasive species (Drury and Rothlisberger 2008). In striking contrast to these findings, the most common intervention strategy actually employed---often by riparian owners associations---is the attempted protection of uninvaded lakes. This is the case in spite of the fact that in most regions less than half the waterways are invaded.

An analogy with human and livestock disease management is apt. Quarantine of infected individuals is recognized as a highly effective means for reducing the transmission of infectious diseases throughout populations (Anderson and May 1991).

Preventing the movement of invasive species away from locations that they have already colonized (i.e., containment) is conceptually similar to the quarantine of sick people or animals, where the invasive species that colonize ecosystems are analogous to the bacteria and viruses that infect organisms. Drawing on this analogy, natural resource managers have already used containment in some cases to restrict the spread of AIS.

Containment of invaded sites, sometimes even to the extreme of quarantine that forbids the movement of any vessels to or away from the invaded site, has typically been used when a novel invader is discovered in one or a few sites in a region. One example is the discovery in 2006 of the highly invasive macrophyte Hydrilla verticillata in Lake

Manitou in north central Indiana, which at the time was hundreds of miles further north and west than any other known Hydrilla population in the United States. After this small population of Hydrilla was found, managers quickly quarantined the lake and initiated herbicide applications to eradicate the invader

124

(http://www.in.gov/dnr/files/LAKE_MANITOU_HYDRILLA_FAQ.pdf). Thus, when there is a single focal site that requires containment to prevent further spread, putting containment measures in place is intuitive. Our study’s conclusions resonate with this intuitive understanding: the probability of invasion at any given uninvaded lake is so low and so contingent upon many unpredictable factors that we cannot accurately predict which particular lakes will be invaded next. Furthermore, as would be expected with a single invaded lake, so it is with multiple invaded lakes: the most cost-effective approach to reducing the probability of invasions in the surrounding area is to keep propagules from leaving invaded lakes.

On the other hand, a case can also be made for protection of some individual uninvaded waterways because some sites may be identified as sufficiently valuable to warrant such shielding, regardless of their predicted vulnerability to invasion (or the limited ability of models to accurately predict invasions). Intervention to prevent the invasion of a particular lake will most likely reduce that lake’s probability of invasion more than containment of any single invaded lake, near or far. On this point, Vander

Zanden and Olden (2008) suggest that managers can apply three layers of geographic and environmental filters to identify where protective intervention should occur. These filters include (1) assessing propagule pressure to uninvaded sites (using models similar to those we assess here) to understand where introductions are occurring, (2) evaluating how well attributes of sites receiving substantial propagule pressure match the AIS requirements to predict where population establishments are possible, and (3) assessing the likelihood of adverse impacts resulting from successful invasion of each candidate site. Examples of this triage approach show that the number of candidate sites for management action, be it

125

protective intervention or surveillance to detect invasions early on, can be reduced up to

95% (e.g., from 800 candidate sites to 30 and from >5000 to 269) (Vander Zanden et al.

2004, Mercado-Silva et al. 2006). While this triage-based approach, largely inspired by conservation planning procedures, theoretically shows promise for making AIS spread manageable (Vander Zanden and Olden 2008), our present analysis reveals limitations in the value of quantifying the first and, arguably, most critical filter (i.e., propagule pressure).

Although gravity models are the state-of-the-art approach to predicting propagule pressure arising from inadvertent human-mediated spread of invasive species among waterways and though previous studies have demonstrated the often overriding importance of propagule pressure in determining population establishment (Williamson

1996, Lockwood et al. 2005, Von Holle and Simberloff 2005, Colautti et al. 2006), our predictions of probability of invasion based on propagule pressure estimates were not strongly correlated with actual invasion events. In response to the potential argument that perhaps the second filter (i.e., suitable environmental conditions) needs to be applied to improve on predictions from gravity models, we point out that mismatches between M.

spicatum requirements and environmental conditions in Wisconsin lakes do not likely explain this lack of predictive ability because nearly all lakes in our dataset provide at least some suitable M. spicatum habitat (i.e., our analysis implicitly includes the second filter) (Buchan and Padilla 2000). We suggest instead that the low probability of a successful invasion occurring at any particular lake, owing to environmental and demographic stochasticity, is the primary consideration limiting the ability of gravity models to predict invasions in situations of few sources and many potential destinations.

126

We believe a more likely source of error in our modeling approach is that sampling effort to detect invasions may not have been uniform across the lakes in our data set (i.e., some of the occurrence data come from incidental discoveries from non- standardized sampling), perhaps resulting in the lack of invasion detection at some sites that are actually invaded. The M. spicatum distribution dataset we used is the compilation of findings from agency field sampling and confirmed reports from the public of M. spicatum presence (J. Bode, WDNR, personal communication). We could only assume M. spicatum absence from the lack of public reports and lack of detection during DNR surveys (i.e., no absence can be 100% verified). This could lead to the underestimation of the number of invaded lakes on the landscape, but it is most likely that any unreported invaded lakes would fall in the low predicted probability of invasion bins of our models because these sites receive fewer visits from recreational boaters and less attention from natural resource managers, owing to their small size and relatively remote locations (i.e., these are the reasons these lakes had the lowest traffic scores in the gravity model). In contrast, the lakes in the high-probability bins are likely to receive the most attention, therefore it is most likely that invasions at these lakes will be quickly detected and reported (Johnson and Padilla 1996). In other words, we think the lag time between

M. spicatum population establishment and its discovery is shorter at high-traffic (i.e., high-probability) lakes than at low-traffic lakes. Therefore, because we found that fewer high-traffic lakes and more low-traffic lakes became invaded than predicted by our gravity model and because invasion detection is more likely at high-traffic lakes than low-traffic ones, perfect knowledge of the invasion status of all lakes would likely reveal a disproportionate number of low-traffic lakes currently identified as uninvaded as

127

actually being invaded, strengthening our finding that the gravity model did not reliably forecast invasions.

Another possibly confounding factor could be that some predicted high invasion probability sites were not invaded because of effective protective intervention. This is not likely to be the case, however, because relatively few comprehensive protective interventions were in place during the period we studied

(http://www.uwsp.edu/cnr/uwexlakes/cbcw/). Moreover, the protective intervention efforts that may have existed were certainly not strategically directed at the sites with the highest predicted invasion probabilities, but instead were situated where local citizen groups or managers happened to be concerned and active. Also, effective protective intervention would not explain away the invasion of sites predicted to have very low invasion probabilities.

Future work that might help to better understand the pattern of M. spicatum

invasions on the landscape would be an analysis of stream connections between invaded

and uninvaded lakes. In highlighting the shortcomings of a dispersal model of M.

spicatum focused only on human-mediated spread (i.e., gravity model), our present analysis points to the potential importance of such natural dispersal corridors for invasion spread. In other words, according to our gravity model, boaters are not likely to have brought milfoil to all of the low-traffic lakes that were invaded. Dispersal via streams is an additional potentially important vector Other studies have found hydrological connectivity important in the spread of other AIS such as zebra mussels (Bobeldyk et al.

2005) and rusty crayfish (Puth and Allen 2005).

128

5.6 Management Recommendations

This work contributes to the management of AIS and other invasive organisms that are spread anthropogenically. We assessed the predictive ability of a commonly used model of invasive species spread and found that it did not provide highly accurate predictions as to which lakes became invaded on an annual basis. Management interventions based on such modeling efforts are unlikely to be cost effective. On the other hand, we found that intervention efforts aimed at containing invaded sites reduce landscape-level risk more than similar levels of effort employed to protect uninvaded sites. Thus, we recommend that to prevent the spread of invasive macrophytes to uninvaded lakes, natural resource managers should strategically place interventions (e.g., inspection crews, boat cleaning resources) at boat landings of invaded lakes. The application of these findings to inform intervention efforts for slowing the spread of AIS could improve the effectiveness of such programs.

5.7 Acknowledgements

We thank Mark Drew for GIS support and for helpful conversations that advanced this research. Chris Jerde, Lindsay Chadderton, Reuben Keller, and Kevin Drury also contributed valuable ideas and feedback that improved this manuscript. Jeff Bode of

Wisconsin DNR provided Eurasian watermilfoil distribution data. Financial support for this work came from the National Science Foundation ISIS project (IRCEB DEB 02-

13698, PI: DML) and the Great Lakes Protection Fund (Grant #797, PI: DML) and a

University of Notre Dame Schmitt Fellowship (to JDR).

129

CHAPTER 6:

FUTURE DECLINES OF THE BINATIONAL LAURENTIAN GREAT LAKES

FISHERIES: RECOGNIZING THE IMPORTANCE OF ENVIRONMENTAL AND

CULTURAL CHANGE5

6.1 Abstract

It is increasingly clear that future long-term environmental challenges (e.g., climate) are driven strongly by economic and cultural choices, as well as by physical and biological mechanisms. We tested to what extent this applies to potential future changes in fisheries in the Laurentian Great Lakes (GL). These fisheries rank among the most valuable in the world, but have changed markedly in recent decades. To investigate how the fisheries might develop in the future, we used structured expert judgment to elicit projections from experts in fisheries and related fields. Experts provided assessments on variables relating to US and Canadian commercial (pounds landed) and sport

(participation and expenditures) fisheries for the years 2006 and 2025. We measured each expert’s ability to probabilistically characterize uncertainty, producing performance- weighted combinations of expert estimates. All experts expected commercial fisheries to decline from 2006 to 2025, with greater declines in the US (25%) than Canada (9%).

5 The publication status of this chapter is: Rothlisberger, J.R., D.M. Lodge, R.M. Cooke, and D.C. Finnoff. Future declines of the binational Laurentian Great Lakes fisheries: recognizing the importance of environmental and cultural change. Frontiers in Ecology and the Environment (in press). 130

Expectations for sport fishing differed more among lakes and less between countries, with median expected declines ranging from 1% to 13%. Experts attributed expected declines primarily to changes in economic market demands and shifts in societal interests. Increased attention to social and economic trends could aid GL fishery policy and management.

6.2 Introduction

Recent IPCC reports indicate that future climate will be a function, in part, of human choices about cultural and economic priorities (Parry et al. 2007). Recent experience with marine and freshwater fisheries in many countries also demonstrates the importance of the interaction of cultural practices and economic policies with fish population dynamics in determining the current status of fish stocks (Hilborn et al. 2003).

Here we apply this perspective of co-determination---the two-way interactions between human and natural systems (Crocker and Tschirhart 1992, Knowler and Barbier 2001)--- by using structured expert judgment to estimate the future magnitude of fisheries-derived ecosystem services in the Laurentian Great Lakes (GL).

The commercial and sport fisheries of the GL are valuable natural resources that have been used extensively since European colonization, but which in recent decades have changed substantially because of various factors including water quality regulation, increased societal interest in recreational relative to commercial fishing, and invasions by nonindigenous species. Beginning in the early 1800s, humans have extracted fishery resources from the GL at a fast and sometimes unsustainable pace (Bogue 2000).

Growing human populations and rising prosperity brought more resource extraction, 131

more intensive agriculture, and the development of industries, including a wide variety of

GL-based recreation. However, erosion and water pollution, byproducts of agriculture and industry, modified the GL environment, often to the detriment of fish populations that had been instrumental in promoting economic growth (Beeton 1969).

Other environmental changes affecting GL fish that accompanied human population growth and economic activity included the construction of dams on GL tributaries, the opening of shipping canals, the intentional stocking of non-native fish species, and the inadvertent introduction of numerous other non-native taxa via multiple vectors (Mills et al. 1993, Coon 1999, Ricciardi 2006).

We used structured expert judgment (SEJ) (Cooke 1991) to incorporate multiple interacting social, economic, political, and environmental factors in ecosystem-scale assessments of potential future conditions of GL fisheries. For decades, fisheries councils have used informal expert opinion approaches to set allowable catch quotas

(e.g., Sazonova et al. 1999, Boydstun 2001), but our study is the first outside this realm to use expert opinion to forecast the commercial or recreational harvest of fish stocks.

Specifically, we used SEJ to estimate the magnitude of changes to GL fisheries between 2006 and 2025, and to quantify the uncertainty surrounding each prediction. To identify likely drivers of change, including continuing natural and anthropogenic stresses such as future biological invasions, regional climate change, economic market changes, and changing trends in pastime activities, experts provided rationales, explaining the reasoning behind their predictions. Understanding these drivers could guide research priorities and assess the potential efficacy of natural resource policy and management.

132

6.3 Methods

In SEJ, experts are seen as independent scientific instruments that provide assessments of future conditions based on current trends and on past and possible future drivers of variability. Expert estimates are obtained via a structured survey. Answers to some of the questions asked are unknown at the time of questioning, but are available by the time of analysis. These questions (12 of 31 questions in this study) allow investigators to treat expert estimates and uncertainties as hypotheses that are then compared against new data to assess each expert’s statistical accuracy and precision, thereby determining how well calibrated each expert is. The judgments of multiple experts are combined into a single estimate that accounts for the calibration of each expert and is bounded by an uncertainty range (Cooke 1991).

6.3.1 Expert selection and interviews

Based on publication record, individual area of expertise (e.g., food web ecology, environmental economics, fishery management), and recommendations from senior GL researchers, we invited nine individuals with excellent general knowledge and wide credibility on the social and biological dynamics of the GL region to participate (Table

6.1). We randomly assigned each expert a number, making it impossible to link specific assessments to individual participants.

Before their interview, each expert received the elicitation survey and a briefing book with historical data on GL fisheries and training materials on uncertainty and probabilistic assessment. Both the briefing book and the survey are available at http://aquacon.nd.edu/research/ and the survey is also included as Appendix C. We

133

TABLE 6.1.

EXPERTS INTERVIEWED AND THE PROFESSIONAL TITLE, AFFILIATION,

AND QUALIFICATIONS OF EACH (LISTED ALPHABETICALLY).

Name Title, affiliation, and qualifications Richard Natural resource economist with the US Fish and Wildlife Service. His office Aiken administers and analyzes the USFWS National Survey of Fishing, Hunting, and Wildlife-Associated Recreation (USFWS 2007).

Mark P. Fisheries assessment biologist with the Chippewa-Ottawa Resource Authority. Ebener Ebener has chaired the Lake Superior and Lake Huron Technical Committees and served on the Lake Michigan Technical Committee of the Great Lakes Fishery Commission (GLFC).

Leroy J. Professor Emeritus of Agricultural, Environmental and Development Economics at Hushak The Ohio State University. Hushak has conducted research on the value of recreation in the Great Lakes and the effects of dreissenid mussels on Great Lakes basin water treatment facilities, electric power plants and industrial water users.

Roger L. Lake Erie Fisheries Program Administrator for the Ohio Department of Natural Knight Resources, Division of Wildlife. Knight serves on the Lake Erie Committee and the Council of Lake Committees of the GLFC.

Frank Lupi Associate Professor of Environmental and Natural Resource Economics at Michigan State University. Lupi studies fish and wildlife demand and valuation and the economics of ecosystem services in the Great Lakes region.

Lloyd C. Fisheries Assessment Team Leader for the Upper Great Lakes Management Unit of Mohr the Ontario Ministry of Natural Resources. Mohr has been active in and chaired the GLFC’s Lake Huron Technical Committee.

Charles R. Senior Extension Associate with New York Sea Grant and the director of Sea O’Neill, Jr. Grant’s National Aquatic Nuisance Species Clearinghouse. O’Neill has led research initiatives regarding the fouling effects of dreissenid mussels on raw water users in the Great Lakes region and has served for the past four years as a member of the Federal Invasive Species Advisory Committee.

Donald Professor of Natural Resources at the University of Michigan and Director of Scavia Michigan Sea Grant. Scavia oversees several large-scale research projects on drivers and conditions of Great Lakes ecosystems.

Roy A. Stein Professor of Evolution, Ecology and Organismal Biology and Director of the Aquatic Ecology Laboratory at The Ohio State University. Stein served as a US Commissioner on the GLFC during 1998-2004.

134

encouraged experts to use the book and any other materials of their choosing as much as desired prior to and during their interview.

We interviewed each expert individually during October 2007. Each expert provided the 5th, 50th, and 95th percentiles of his subjective cumulative probability density function for 31 variables relevant to GL fisheries.

A typical pair of questions took the following form.

How many total pounds of commercial fish were landed from the US waters of Lake Erie

in 2006? 5%______50%______95%______

How many total pounds of commercial fish WILL BE landed from the US waters of Lake

Erie in 2025? 5%______50%______95%______

Following each such pair of questions, we asked the expert to provide their rationale for the predicted changes in the variable between 2006 and 2025, identifying, if possible, the most important expected driver of change. We made notes and audio recordings of these responses.

6.3.2 Analysis, combination, and reporting of results

We report the experts’ answers as combined assessments for each variable

(Figures 6.1, 6.2) and as individual assessments (Figures 6.3, 6.4, 6.5). We combined expert assessments in two ways: (1) equal weight given to each expert and (2) weighting based on performance on calibration variables (Cooke 1991).

For each variable, we calculated expected percent change between 2006 and

2025, and the associated uncertainty range. To make this calculation, we assumed independence of all variables and took the convolution of the joint probabilities of the 135

Figure 6.1. Historical and projected commercial and recreational fisheries in the US and Canadian waters of the GL. Angler effort in recreational fisheries is shown as insets in the upper right of each panel. Vertical range bars are performance-based combinations of expert assessments where lower and upper limits show, respectively, 5th and 95th percentiles of the combined expert th subjective probability distributions. Hollow circles depict the 50 percentile of each distribution. Note different vertical scales across countries, lakes, and fishery types. Canadian commercial catch (Panel i) is for all Canadian waters of the GL. Historical recreational fisheries data were taken from the USFWS National Survey of Fishing, Hunting, and Wildlife-associated Recreation. Commercial catch data dating back to 1971 were obtained for the US from the USGS Great Lakes Science Center and for Canada from the Department of Fisheries and Oceans.

136

Commercial Sport

0.006 US Superior US Superior CAN Superior 0.003 0.000 0.008 US Michigan US Michigan

0.004

na 0.000 0.006 US Huron US Huron CAN Huron

0.003

0.000 0.008 US Ontario US Ontario CAN Ontario 0.004 Probability Density Probability 0.000

0.010 US Erie US Erie CAN Erie 0.005

0.000

0.008 CAN All US Spending

0.004 na 0.000 -200 -100 0 100 200 -200 -100 0 100 200 -200 -100 0 100 200 Projected Percent Change

Figure 6.2. Probability density functions of PBC-projected percent change between 2006 and 2025 in US and Canadian commercial fish landings (lbs landed) and sport fishing effort (angler-days) and expenditures (2007 US$). Black lines show median of each distribution. Dotted lines provide a reference to zero percent change. Note that, even though most distributions extend to the left of -100%, the value of these variables in 2025 actually cannot be more than 100% less than they were in 2006.

137

a b

c d

0 1 2 3 4 5 6 7 0.0 0.5 1.0 1.5 2.0 2.5 e f

g h

0 5 10 15 0 2 4 6 8 i j

k l

0 2 4 6 8 0 1 2 3 4 m n

o p

0 2 4 6 8 0 2 4 6 8 10 q r

s t

0.00 0.05 0.10 0.15 0.20 0 1 2 3 4 5

Commercially Landed Fish (Millions of Pounds) Sport Fishing Effort (Millions of Angler-days)

138

0 10 20 30 40 50 c

0.0 0.2 0.4 0.6 0.8 1.0 1.2 e

0 1 2 3 4 5 g

0.0 0.5 1.0 1.5 2.0 2.5 3.0 i

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Commercially Landed Fish (10^6 lbs, a-b) or Sport Fishing Effort (10^6 Angler-days, c-j)

Figure 6.4. Individual and combined expert assessments for Canadian commercial fish landings, aggregated across lakes (a, b), and sport fishing effort, divided by lake (c – j), for 2006 and 2025 for each of the GL. For sport fishing, the lakes are shown in the following order: Superior, Huron, Erie, Ontario. Two rows of panels represent each lake, the first showing 2006 assessments and the second 2025. Dashed lines divide 2006 and 2025 estimates for each lake. Each panel shows th th the 5 to 95 percentile range graphs for individual and combined expert assessments, with filled circles showing 50th percentile estimates. Assessments are shown in the same order in each panel: top to bottom, Expert 1, 2, 3, 4, 5, 6, 7, 8, 9, equally-weighted combination, performance-based combination. For calibration variables, light gray vertical bars show the actual value of the variable being estimated, which became known after the elicitation was finished. Note differences in scale for each lake and fishery type.

139

0 2 4 6 Sport Fishing Expenditures (Billions 2007 US $)

Figure 6.5. Individual and combined expert assessments for US GL recreational fishing expenditures for 2006 (a) and 2025 (b). Each panel shows the 5th to 95th percentile range graphs for individual and combined expert assessments, with filled circles showing 50th percentile estimates. Assessments are shown in the same order in each panel: top to bottom, Expert 1, 2, 3, 4, 5, 6, 7, 8, 9, equally-weighted combination, performance-based combination. The vertical light gray bar shows actual 2006 expenditures (USFWS 2007), which became known after the elicitation was over.

140

2006 and 2025 performance-based combination distributions. We then divided the 5th,

50th, and 95th percentiles of this distribution of differences by the appropriate 2006 median assessment and multiplied by 100 (i.e., [(2025 – 2006)5th, 50th, or 95th percentile)/200650th]*100) to find expert-expected percent change. We extracted from the experts’ rationales the driver of change each one identified as being most important to the future condition of GL fisheries. We also tabulated how many experts mentioned each driver of change.

6.4 Results

Equal weighting and performance-based combination (PBC) of expert assessments produced median results in the same ballpark. The equal weight combination, however, was statistically inaccurate, whereas the PBC variables were statistically accurate (Table 6.2). Furthermore, the PBC allowed us to weight experts according to the precision and accuracy of their assessments (Table 6.2; Figures 6.3, 6.4).

Therefore, we report here the results only of the PBC. In our figures, we provide PBC uncertainty ranges (Figure 6.1) and probability distributions (Figure 6.2). In the text, we focus on median values because they reflect the outcomes experts saw as most likely, and experts’ rationales for their predictions correspond to these medians. Moreover, because we assumed independence between 2006 and 2025 when calculating projected percent changes, the ranges of these projections (Figures 6.1, 6.2) are overly broad relative to the experts’ narrative descriptions of how GL fisheries might change between 2006 and

2025.

141

TABLE 6.2.

CALIBRATION, INFORMATIVENESS, AND WEIGHTS OF THE NINE EXPERTS,

THEIR EQUAL-WEIGHT COMBINATION (EQUAL), AND THEIR PERFORMANCE-BASED COMBINATION

(PBC) FOR CHANGES IN GREAT LAKES FISHERIES BETWEEN 2006 AND 2025

Expert or Calibration Weight un- combination (p-value) Mean relative information normalized Weight normalized All variables Calibration variables Without combo With combo 1 2.66E-4 8.50E-1 9.01E-1 (2.39E-4) 0 6.70E-4 142 2 3.11E-1 8.66E-1 7.09E-1 2.20E-1 7.10E-1 6.17E-1 3 6.04E-4 1.46 1.19 7.15E-4 2.30E-3 2.00E-3 4 6.04E-4 1.15 1.03 6.19E-4 1.99E-3 1.73E-3 5 4.77E-3 8.31E-1 1.21 5.77E-3 1.86E-2 1.62E-2 6 3.11E-1 3.10E-1 2.65E-1 8.24E-2 2.65E-1 2.31E-1 7 6.04E-4 1.22 1.01 6.10E-4 1.96E-3 1.71E-3 8 2.55E-5 1.60 1.28 (3.25E-5) 0 9.11E-5 9 1.88E-8 2.54 2.92 (5.50E-8) 0 1.54E-7 EQUAL 1.49E-1 2.89E-1 3.11E-1 4.63E-2 – 1.30E-1 PBC 8.24E-1 3.23E-1 2.46E-1 2.03E-1 – 3.95E-1 PBC no 8.24E-1 3.31E-1 2.46E-1 2.03E-1 – 3.95E-1 optimization

6.4.1 US Commercial Fishery

The actual 2006 commercial catch data are enclosed in the PBC 90% uncertainty estimates in all cases (Figure 6.1, Table 6.2). The PBC projected 19-31% decreases in commercial catches for Lakes Superior, Michigan, Huron, and Erie in 2025 (Figure 6.2).

Six of nine experts (67%) indicated that the most important reason for predicted declines in commercial fishing was the decreasing economic viability of the fishery, resulting from a combination of cheaper supplies of fish entering the marketplace (e.g., aquaculture) and sport fishing continuing to receive political preference (e.g., higher allocation of total allowable catch quotas). The other experts gave invasive species impacts (one expert), climate change (one expert), and limitations in ecosystem productivity (one expert) as the primary cause of declines. The most commonly mentioned reasons for decline were invasive species, economic markets, and fishery policy (Figure 6.6).

6.4.2 Canadian Commercial Fishery

Historical data on individual lakes were not available for the Canadian commercial fishery; therefore, experts were asked for estimates for landings aggregated from all Canadian waters of the GL. The PBC assessments accurately predicted the 2006

Canadian commercial catch (Figure 6.1) and forecast that Canadian commercial landings will decline 9% by 2025 (Figure 6.2). Five of nine experts (56%) cited failing economic viability of the fishery as the main reason for the declines they predicted. One expert attributed expected declines primarily to food web disruptions by invasive species. Three

143

Commercial-US Commercial-CANADA

Sport-US Sport-CANADA

Cultural Factors 9 Environmental 8 7 6 5 4 3 2 1 Number of mentions 0

Drivers of change

144

experts predicted stability or even slight increase in the Canadian commercial catch by

2025, owing largely to strong political support for commercial fishing in Canada.

Invasive species and changing economic markets were the most commonly mentioned factors driving declines or limiting growth (Figure 6.6).

6.4.3 US Sport Fishery

The PBC estimates were not as accurate for 2006 sport effort as for commercial landings, but the actual data for 2006 fell within the 90% uncertainty range for four of the five lakes (Figure 6.1). For four lakes, the PBC projected 9-13% declines in fishing effort in 2025 relative to 2006. In contrast, the PBC predicted that effort will increase 1% in L. Huron (Figure 2), which several experts attributed to 2006 being a particularly bad year for L. Huron because of food web disruptions involving alewife population declines.

Six out of nine experts (67%) gave as their primary reason for expected declines in angling effort reduced cultural interest in consumptive outdoor recreation (e.g., fishing) and, for the population segment still interested in angling, a preference for inland waterways over the GL. These experts attributed the shifting preference away from the

GL to the rising costs of participation in GL fishing as nearshore fisheries decline and as costs climb for equipment and fuel to participate in offshore fishing. One expert said invasive species will be the primary reason for declines by 2025, one expected stability in angler participation, and one did not identify a primary cause for declines. Invasive species were the most commonly mentioned secondary driver of declines (Figure 6.6).

145

The PBC performed well in predicting 2006 angler expenditures aggregated over all lakes (Figure 6.5), which are expected to fall 24% (or about $330 million in 2007 US dollars) by 2025 (Figure 6.2, USFWS 2007).

6.4.4 Canadian Sport Fishery

The PBC forecast that effort in Canadian sport fisheries will decline 1-12% by

2025 (Figures 6.1, 6.2). Most experts (six out of nine) expressed greater uncertainty in their predictions for Canadian sport fisheries than for their US counterparts, partially because of their currently small size, possibly causing volatility. Five experts indicated changing trends in pastimes would be the primary driver of declines, but two said invasive species impacts would be the main driver of declines. On the other hand, one expert expected stability in the fishery and one expected a slight increase in angler effort.

Shifts away from fishing as a pastime and harm caused by invasive species were mentioned most often as drivers of declines for the Canadian sport fishery (Figure 6.6).

6.5 Discussion

6.5.1 Drivers of change in GL fisheries

According to experts, the projected declines in fisheries in the US and Canadian waters of the GL (Figure 6.2) will largely be a function of social trends, economic conditions, and political decisions (Figures 6.6). Thus, when considering the future of fisheries in the GL, a comprehensive view of societal dynamics and bio-economic

146

feedbacks must be taken into account. This is especially the case for the economically valuable sport fisheries (GLFC 1988), on which we focus our discussion here.

The importance of biological and cultural factors on GL fisheries is clear when considering the rehabilitation of self-sustaining populations of native species in the lakes.

Lake trout rehabilitation is a major emphasis in GL fishery management, particularly for

US federal agencies (Holey et al. 1995, Knuth et al. 1995). Using L. Superior as a case study, Kitchell et al. (2000) point out that the ecological turn-over rate of the historical fish community in the upper Great Lakes, consisting of lake trout and lake herring as primary predator and prey, was much slower than the currently predominant Pacific salmon-alewife and rainbow smelt predator-prey system. They conclude that the further restoration (i.e., augmentation) of lean lake trout stocks, which is currently specified as one of the major management goals in L. Superior (GLFC 2003), will be extremely challenging. In fact, L. Superior may be as close to a restored state in this regard as possible. Hence, any continued efforts to enhance lean lake trout populations in L.

Superior may unintentionally force declines in Pacific salmon, a valuable component of the sport fishery, possibly reducing angler participation and, hence, economic benefits

(Kitchell et al. 2000).

This apparent conflict in objectives in Lake Superior, however, may be a false dilemma, as may be the case with other similar situations across the GL, because cultural trends may now be stronger drivers of fishery participation levels than stock sizes. Even as managers actively try to balance recreational fishery (Pacific salmon) enhancement with native fish community (lake trout) restoration, experts expressed substantial uncertainty regarding the value of these efforts for improving fisheries, mainly because of

147

doubts as to whether or not angler effort will increase in response to these management efforts. Of course, management decisions regarding the restoration of native communities and the enhancement of particular fish stocks are complex issues, the goals of which certainly extend beyond creating successful fisheries (e.g., existence value of native species). Nevertheless, our focus here is on expectations for GL fisheries and their expected drivers of change.

Therefore, increased consideration of the response of anglers to changes in fish stocks relative to cultural change could help to inform fishery policy and management.

For example, if the decision were made to try to increase participation levels in GL sport fisheries in the coming years, it would be valuable to know the trade-offs between investing in programs to improve sport fish stocks versus efforts to recruit more anglers to the fishery through advertising and educational programs. According to our experts, the latter would probably be a more effective tool in growing or maintaining the fishery because, as long as the target species are not extremely scarce, most of the projected declines in sport fishing on the GL will occur because of weakening societal interest in the pastime, driven by less discretionary free time and the pursuit of other pastimes, not by declining abundance of target species. This is counter to some reports that fishing pressure has previously responded to shifts in abundance of popular species in the GL

(e.g., spike in Ohio fishing license sales in the 1980s in response to walleye recovery in

L. Erie (Hatch et al. 1987), drop in L. Huron angling in the 2000s when Chinook salmon fishery collapsed (Dobiesz et al. 2005)) and bears further investigation.

148

6.5.2 Conclusions

The highly complex biological and human interactions in the GL could never be tested directly with standard research methods, even if an enormous budget were available to support such research. SEJ provides a tool to overcome this obstacle and to predict future states of the GL fisheries with explicit attention to the substantial uncertainties involved. Results that suggest the importance of social and cultural factors in the future of GL ecosystem services do not in any way diminish the well known importance of environmental degradation and unsustainable harvest practices that have led to past stock collapses and declines in productivity in many of the world’s most important marine (Pauly et al. 2002, Hilborn et al. 2003) and freshwater (Allan et al.

2005) fisheries. Indeed, the importance of cultural factors relative to environmental ones

(Figure 6.6) may be of greater magnitude for the GL than for other major fisheries because of the growing primacy of sport fishing in the lakes. Our study underscores the continued and increasing relevance of cultural factors, including economic considerations, when designing plans and setting objectives for fishery management, particularly in the North American Great Lakes. Specifically, further research on the drivers of change in GL fisheries, especially sport fisheries, is needed to support scientifically sound and socially beneficial fishery policy and management. The primary drivers of change that experts identified point the way for future inquiries, particularly the causes and consequences of people spending less time fishing recreationally on the Great

Lakes.

149

6.6 Acknowledgements

We thank the experts for their participation. The NOAA National Sea Grant

Program (Award No. NA16RG2283) through the Illinois-Indiana Sea Grant College

Program (Subaward No. 2003-06727-10) partially funded this research. The US EPA's

National Center for Environmental Economics also provided support (Contract No. EP-

W-05-022). A Schmitt Graduate Research Fellowship from U Notre Dame supported

JDR. Special thanks to Scott Nelson of the USGS Great Lakes Science Center for his help in obtaining the 2006 US commercial fish landing data. Feedback from Reuben

Keller, Darren Yeo, and two anonymous reviewers improved this manuscript. Although the research described in this article has been funded in part by the US EPA, the opinions expressed here are those of the authors, and do not necessarily express the views of the

United States Environmental Protection Agency.

150

CHAPTER 7:

DISSERTATION CONCLUSION

7.1 Ecology and Society

Ecology is the study of the workings and interconnectedness of the physical and biological components of the natural world, with each other and with themselves. Given this definition, it is, to say the least, a broad field of study. Biological invasions exemplify well the ecological principle of interconnectedness, demonstrating that the natural world extends beyond the traditional view of nature as an unsullied patch of forest deep in a reserve or a pristine stream untouched by human influence. Human influence upon the natural world is vast (Vitousek et al. 1997) and often detrimental (Parry et al.

2007), but humans are not separate from nature. We cannot abdicate responsibility for how our choices affect the natural world or deny that the natural world affects the choices available to us (Lodge and Hamlin 2006).

In light of potentially serious environmental crises on the horizon (MEA 2005,

Parry et al. 2007), it is more important than ever for ecologists to engage in research that is socially relevant, enabling ecology to contribute science-based recommendations for natural resource policy and management. For a variety of natural resource issues, including invasive species, it seems that the unwillingness of ecologists to provide practical and objective guidance could lead to an information vacuum. Such information vacuums are often filled with the voices of partisan interest groups, regardless of how 151

accurate the information they present is. This kind of situation could be avoided because ecologists work at many levels of biological organization and many spatial and temporal scales. Thus, within the discipline, there should be room for ecologists to work at the spatial and temporal scales and the levels of biological organization most relevant to natural resource policy and management. Working at such scales, ecologists can bring to bear the research of other ecologists who work at other scales on critical natural resource issues. If ecologists are unwilling to allow their work to be interpreted by other ecologists in ways that make it relevant to natural resource policy and management, or if they are unwilling to interpret their work for these purposes themselves, then ecology concedes to allow those with training in other disciplines to interpret ecological research according to their own disciplinary paradigms and purposes.

The work presented in the foregoing chapters is connected to the environmental issue of biological invasions. Biological invasions often occur when human movements transport species beyond their native ranges, followed by these nonindigenous species harming native species, agricultural concerns, or human infrastructure. Because invasive species interact with other species and with the physical environment to cause impacts, it is necessary to study the interactions and the interplay amongst components of the natural world (i.e., ecology) to understand the impacts of biological invasions on humans and to discover the options available to humans in seeking to mitigate these impacts.

7.2 Dissertation Overview

In this dissertation I focused on biological invasions that have already occurred in a particular geographic region of the world---the Laurentian Great Lakes (GL) and the 152

states that surround them (Figure 7.1). Knowing more about current invasions can help society better respond to the threat of future invasions, and to deal more effectively with

NIS that are already established. Furthermore, lessons learned from biological invasions in one region can provide insights regarding the ecology of biological invasions, and, possibly, appropriate policy and management responses in other parts of the world. With this in mind, one of the ultimate objectives of my research was to provide information and analysis on biological invasions to support the development of policies and management strategies to reduce the damaging effects of future invasions. I did not, however, try to predict the identity of species that may invade in the future. Instead, my research aimed to improve the understanding and quantification of the impacts and spread of NIS that have already established populations in the GL region.

I studied how these species arrived in the region and what clues their pathway of introduction may give about their future spread and impacts (Chapter Two). In this chapter, I retrospectively analyzed pathways of invasion to the GL. I found that there is a relationship between pathway of introduction and the likely impacts of a species. I also found that the GL appear to serve as an important beachhead of invasion for freshwater

NIS that eventually establish populations elsewhere in North America.

I also investigated the effects aquatic NIS introduced to the GL by ballast water release on GL ecosystems and thus on ecosystem services that are economically valuable to people living in the region (Chapter Three). In this chapter I used a novel method of expert elicitation to quantify economic losses from ship-borne NIS. I found that the economic damage from NIS introduced by this one introduction pathway are large,

153

CH. 2

CH. 6

CH. 5 CH. 4 154 CH. 3

Figure 7.1. Geographical scope and content of research projects presented in this dissertation. Biological invasions are a multi-stage process that occur at multiple spatial and temporal scales. Chapter numbers accompany a graphical depiction of the topic of each chapter. Each chapter investigates the ecology of biological invasions and offers recommendations for natural resource policy and management. The global transport of ballast water in shipping vessels introduces species to the Great Lakes region (Chapter 2) which have ecological and economic impacts in the Great Lakes (Chapter 3). Some of these and other species introduced to the region by other pathways (e.g., commerce in living organisms) spread to other waterways via anthropogenic mechanisms (Chapters 4 and 5). The impacts of nonindigenous freshwater species in the Great Lakes region take place within a broader context of other environmental and cultural factors that also drive environmental change (Chapter 6).

possibly justifying additional management of the pathway to reduce the risk of future invasions.

I also gathered empirical data on the basics of an important pathway for spread among inland lakes: the overland towing of recreational boats. This pathway has been much discussed in the scientific literature (MacIsaac et al. 2004, Puth and Post 2005,

Leung et al. 2006), but my chapter (Chapter Four) is one of the first scientific studies of what species are actually being transported and how effective various cleaning methods are for avoiding the transport of potentially harmful NIS. This study showed that, despite public campaigns to educate boaters on how to avoid the transport of NIS, many boaters do not always take steps to remove organisms from their boats and trailers. My research also showed that visual inspection and hand removal is as effective as power washing for removing macrophytes from boats and trailers, and that high-pressure washing is the most effective option for removing small-bodied organisms such as zooplankton.

I also investigated how accurately the human-mediated spread of aquatic NIS can be predicted (Chapter Five). To evaluate the accuracy of predictions of future invasions by an invasive aquatic macrophyte, I took data from the past and used it to predict a future that has already occurred (i.e., the more recent past and present). This allowed me to assess the predictive ability of a modeling approach that is widely touted for identifying where future invasions are most likely. This study showed that this modeling approach provides low predictive accuracy in pinpointing which uninvaded lakes were the most likely to be invaded next. This finding highlighted the value of considering approaches other than trying to protect uninvaded lakes for slowing the landscape-level spread of invasive species. In this regard, I found that management actions that keep

155

invasive species from leaving invaded lakes reduce the average probability of invasion at uninvaded sites across the landscape more than trying to protect uninvaded lakes.

Finally, I considered the importance of NIS relative to other environmental changes and cultural influences in driving society’s use of a particular set of ecosystem services (Chapter Six). In this study, I investigated the sport and commercial fisheries of the GL, comparing their current size to what experts expect their size will be in 2025. I learned that all GL fisheries, sport and commercial, in the US and Canada, are expected to decline by 2025. I also found that cultural changes, rather than environmental factors, are expected to be the primary cause for the fisheries’ predicted decline. Environmental factors were also important for these predictions, with the effects of NIS being the most- often mentioned environmental factor leading to fishery declines. To achieve the desired ends of environmental policies and management actions, it is necessary to identify what factors most affect the desired endpoints of the natural system under consideration.

Chapter Six does this, highlighting the combined importance of environmental and cultural drivers of change.

7.3 Possibilities and Pitfalls

In this dissertation, for a handful of specific instances (Figure 7.1), I have tried to look beyond the seeming simplicity of the invasion process to better understand the challenging complexities of the spread and impacts of aquatic NIS. The research presented here is readily applicable to current concerns regarding the policy and management of biological invasions. My research points out some of the possibilities as

156

well as some of the pitfalls involved in dealing with aquatic invasive species (AIS) in the

GL region.

One possibility that my research uncovers is that managing particular introduction pathways to reduce the likelihood of new invasions via these pathways can lower the chances of incurring particular types of impacts. This is because species introduced to the GL by different pathways appear to have different kinds of typical impacts. It also appears that the invasion of the GL by freshwater NIS often precedes their appearance in other North American watersheds. Thus, measures to reduce the likelihood of additional invasions in the GL will likely guard waterways throughout North America against invasions.

The spread of aquatic NIS from the GL to other waterways in the region and other watersheds throughout North America has been tied to the overland transport of recreational boats. Part of my research provided empirical data on how effective various boat cleaning methods are for removing AIS from this pathway, showing that the efficacy of various cleaning methods depends on the species to be removed. Beyond the tactics of boat cleaning, my research also considered how the strategic placement of boat-cleaning stations affects the probability of invasion for waterways across a landscape. In most practical applications, boat-cleaning stations that prevent the transport of invasive propagules away from invaded lakes reduces the chances of invasions at the landscape- level more than trying to protect uninvaded lakes.

The management of introduction pathways and efforts to slow the spread of AIS are costly, and invasive species are only one of multiple environmental issues that demand attention. Better understanding the economic impacts arising from the ecological

157

changes linked to NIS establishments is an important way to benchmark how much should be invested in dealing with invasions. In Chapter Three, I used a novel technique to estimate the economic impacts of NIS introduced to the GL via ballast water release.

This study concluded that the economic impacts of these species are large and that it may be economically beneficial to implement policy and management changes that reduce the chances of future invasions.

Just as it is important to evaluate the economic impacts of NIS, it is also helpful to consider the importance of NIS relative to other environmental and cultural factors in driving ecological change. In Chapter Six, I investigated the relative importance of cultural and environmental factors, including NIS, in determining the future state of GL fisheries. I concluded that cultural factors will be more important drivers of fishery condition that environmental factors. Among environmental factors, however, NIS are expected to be among the leading drivers of change.

One pitfall associated with NIS policy and management highlighted here is the possibility for overzealousness regarding NIS. That is, while NIS are currently a prominent environmental concern, there are other important drivers of change, particularly cultural ones, that also affect ecosystem services. Furthermore, the mechanistic details of the impacts of NIS currently established in the GL are still little known. Continued research effort is needed to better grasp how NIS in the GL region are changing aquatic ecosystems. Failure to gain better understanding of the biology and ecology of these species will inevitably lead to pitfalls in attempts to manage AIS.

Finally, this research identified the serious pitfall in predicting the spread of invasions.

Namely, because new invasions of inland waterways cannot be accurately predicted using

158

gravity models, protecting pristine lakes that are predicted likely to be invaded may result in wasted effort and resources.

7.4 Conclusion

Biological invasions are a multi-stage process. Broadly speaking, the invasion process is conceptually logical and empirically verified (Lodge et al. 2006, Jerde and

Lewis 2007). At each stage in the unfolding of any given biological invasion, society’s options for responding to the invasion are different. As an invasion progresses from species introduction, to population establishment, to spread and impacts, alternatives for avoiding environmental harm and economic damage typically become for limited and more costly (Lodge et al. 2006). For example, if the introduction of Eurasian watermilfoil Myriophyllum spicatum to North America as an ornamental watergarden plant in the 19th-century had been prevented then states like Wisconsin would not now need to spend millions of dollars annually to control nuisance populations of the species

(Carpenter et al. 2007).

There are at this time many NIS, distributed across all steps in the invasion

process, in the GL region. While there are notions regarding what management options

are available for dealing with a generic invasion, many of the specifics are still murky.

For a particular variety of NIS---freshwater species---in a particular region---the GL and

surrounding states---this dissertation has sought to shed light on the specifics of the

efficacy of various practices in limiting the spread of established invasive species and

also elucidate the damage to ecosystem services that NIS cause (Figure 7.1). In other

words, the goal of this dissertation has been to go beyond the clear-cut linear progression 159

of the invasion process as it is often depicted and to provide a better understanding of what society can do about biological invasions, taking into account how much investment might be justifiable to counter biological invasions. Much of the research here is analogous to the epidemiological research that informs the management of flu viruses and other infectious pathogens (Colizza et al. 2006, Crowl et al. 2008).

The introduction and establishment of additional NIS in the GL region, and in many other regions of the world, is inevitable, as global trade and travel continue apace

(Levine and D'Antonio 2003, Taylor and Irwin 2004). Some future NIS will have environmentally undesirable and economically damaging consequences for the areas they invade. While more research on a wider variety of cases is needed, the work presented here demonstrates the value of scientific inquiry to better understand the aspects of current invasions upon which natural resource policy and management decisions hinge.

These highly influential aspects of invasions include their economic impacts, their importance relative to other drivers of change, their predictability, and the options available to reduce the damage they cause. Each of these is addressed in this dissertation.

It is clear that there is much to be done in improving NIS policy and management in the GL region and beyond. Success in this work will require the cooperative efforts of environmental researchers along with practitioners of natural resource policy and management.

160

APPENDIX A:

LIST OF NONINDIGENOUS AQUATIC SPECIES ESTABLISHED IN THE GREAT

LAKES AND DATA ON THESE SPECIES COMPILED FOR ANALYSES IN

CHAPTER TWO

161

TABLE A.1.

NONINDIGENOUS AQUATIC SPECIES ESTABLISHED IN THE GREAT LAKES AS OF 2008

Scientific Name Common Name First Current Endemic Region Year Pathway Taxonomic Lake Discovered NA Distri- Discovered Category Where First in NA in bution Discovered the Great Lakes? Cyprinus carpio common carp No Out of GL Ponto-Caspian 1879 Authorized Fish Wide- Basin release spread

162 Potamogeton curlyleaf No Out of GL Eurasia 1879 Authorized Plant Lake crispus pondweed Basin release Ontario Salmo trutta brown trout Yes Out of GL Eurasia 1883 Authorized Fish Lake Basin release Michigan Marsilea European water No Out of GL Eurasia 1925 Authorized Plant Lake quadrifolia clover Basin release Ontario Najas minor minor naiad Yes Out of GL Eurasia 1932 Authorized Plant Lake Erie Basin release Cipangopaludina Oriental No Out of GL Asia 1940 Authorized Mollusk Lake Erie japonica mystery snail Basin release Sphaerium corneum fingernail clam Yes Out of GL Eurasia 1924 Ballast water Mollusk Lake Basin Ontario Stephanodiscus diatom Yes In GL Eurasia 1938 Ballast water Phyto- Lake binderanus Basin plankton Michigan Actinocyclus diatom Yes In GL Eurasia 1938 Ballast water Phyto- Lake normanii fo. Basin plankton Ontario subsalsa

TABLE A.1. (Continued) Scientific Name Common Name First Current Endemic Region Year Pathway Taxonomic Lake Discovered NA Distri- Discovered Category Where First in NA in bution Discovered the Great Lakes? Diatoma diatom No Out of GL Widespread 1938 Ballast water Phyto- Lake ehrenbergii Basin plankton Michigan Stephanodiscus diatom Yes In GL Eurasia 1946 Ballast water Phyto- Lake subtilis Basin plankton Michigan Cyclotella diatom No Out of GL Widespread 1946 Ballast water Phyto- Lake pseudostelligera Basin plankton Michigan Pisidium supinum humpback pea Yes In GL Europe 1959 Ballast water Mollusk Lake clam Basin Ontario Thalassiosira diatom Yes In GL Widespread 1962 Ballast water Phyto- Lake St. weissflogii Basin plankton Clair 163 Chroodactylon red alga Yes In GL Atlantic 1964 Ballast water Benthic Lake Erie ramosum Basin Alga Bangia red alga No Out of GL Atlantic NA 1964 Ballast water Benthic alga Lake Erie

atropurpurea Basin Cyclotella cryptica diatom No Out of GL Widespread 1964 Ballast water Phyto- Lake Basin plankton Michigan Cyclotella woltereki diatom No Out of GL Widespread 1964 Ballast water Phyto- Lake Basin plankton Michigan Potamothrix oligochaete Yes In GL PontoCaspian 1965 Ballast water Annelid Lake Erie vejdovskyi Basin Eubosmina waterflea Yes Out of GL Eurasia 1966 Ballast water Zooplankton Lake coregoni Basin Michigan Dugesia polychroa flatworm Yes In GL Europe 1968 Ballast water Other Lake Basin Invertebrate Ontario Nitokra hibernica harpacticoid Yes In GL Eurasia 1973 Ballast water Benthic Lake copepod Basin Crustacean Ontario

TABLE A.1. (Continued) Scientific Name Common Name First Current Endemic Region Year Pathway Taxonomic Lake Discovered NA Distri- Discovered Category Where First in NA in bution Discovered the Great Lakes? Skeletonema diatom Yes In GL Eurasia 1973 Ballast water Phyto- Lake Erie subsalsum Basin plankton Thalassiosira diatom Yes In GL Widespread 1973 Ballast water Phyto- Lake Erie guillardii Basin plankton Thalassiosira diatom Yes In GL Widespread 1973 Ballast water Phyto- Lake Erie pseudonana Basin plankton Sphacelaria brown alga Yes In GL Unknown 1975 Ballast water Benthic Lake lacustris Basin Alga Michigan Hymenomonas cocco- No Out of GL Eurasia 1975 Ballast water Phyto- Lake roseola lithophorid alga Basin plankton Huron 164 Thalassiosira diatom Yes In GL Eurasia 1978 Ballast water Phyto- Lake Erie lacustris Basin plankton Chaetoceros diatom Yes In GL Unknown 1978 Ballast water Phyto- Lake

muelleri Basin plankton Huron Ripistes parasita oligochaete Yes Out of GL Eurasia 1980 Ballast water Annelid Lake Basin Huron Daphnia galeata waterflea Yes In GL Eurasia 1980 Ballast water Zooplankton Lake Erie galeata Basin Bythotrephes spiny waterflea Yes Out of GL Eurasia 1982 Ballast water Zooplankton Lake longimanus Basin Ontario Gianius oligochaete Yes In GL Europe 1983 Ballast water Annelid Lake Erie (Phallodrilus) Basin aquaedulcis Nitellopsis obtusa green alga Yes In GL Eurasia 1983 Ballast water Phyto- Lake St. Basin plankton Clair Gymnocephalus Eurasian ruffe Yes In GL PontoCaspian 1986 Ballast water Fish Lake cernuus Basin Superior

TABLE A.1. (Continued) Scientific Name Common Name First Current Endemic Region Year Pathway Taxonomic Lake Discovered NA Distri- Discovered Category Where First in NA in bution Discovered the Great Lakes? Dreissena zebra mussel Yes Out of GL PontoCaspian 1988 Ballast water Mollusk Lake St. polymorpha Basin Clair Thalassiosira diatom Yes In GL Europe 1988 Ballast water Phyto- Lake baltica Basin plankton Ontario Eubosmina waterflea Yes In GL Eurasia 1988 Ballast water Zooplankton Lake Erie maritima Basin Dreissena quagga mussel Yes Out of GL PontoCaspian 1989 Ballast water Mollusk Lake rostriformis Basin Ontario bugensis Proterorhinus tubenose goby Yes In GL PontoCaspian 1990 Ballast water Fish Lake St. 165 semilunaris Basin Clair Apollonia round goby Yes Out of GL PontoCaspian 1990 Ballast water Fish Lake St. (Neogobius) Basin Clair melanostomus Potamopyrgus New Zealand No Out of GL Australasia 1991 Ballast water Mollusk Lake antipodarum mud snail Basin Ontario Ichthyocotylurus digenean fluke Yes In GL PontoCaspian 1992 Ballast water Other Lake St. pileatus Basin Invertebrate Clair Dactylogyrus monogenetic Yes In GL Eurasia 1992 Ballast water Other Lake amphibothrium fluke Basin Invertebrate Superior Dactylogyrus monogenetic Yes In GL Eurasia 1992 Ballast water Other Lake hemiamphibothrium fluke Basin Invertebrate Superior

Neascus digenean fluke Yes In GL Eurasia 1992 Ballast water Other Lake brevicaudatus Basin Invertebrate Superior Timoniella sp. digenean fluke Yes In GL Eurasia 1992 Ballast water Other Lake Basin Invertebrate Superior

TABLE A.1. (Continued) Scientific Name Common Name First Current Endemic Region Year Pathway Taxonomic Lake Discovered NA Distri- Discovered Category Where First in NA in bution Discovered the Great Lakes? Trypanosoma flagellate Yes In GL PontoCaspian 1992 Ballast water Other Lake acerinae Basin Invertebrate Superior Echinogammarus amphipod Yes In GL PontoCaspian 1994 Ballast water Benthic Lake Erie ischnus Basin Crustacean Scolex pleuronectis cestode Yes In GL PontoCaspian 1994 Ballast water Other Lake St. Basin Invertebrate Clair Sphaeromyxa mixosporidian Yes In GL PontoCaspian 1994 Ballast water Other Lake St. sevastopoli Basin Invertebrate Clair Heteropsyllus nr. harpacticoid Yes In GL Atlantic 1996 Ballast water Benthic Lake nunni copepod Basin Crustacean Michigan 166 Acineta nitocrae suctorian Yes In GL Eurasia 1997 Ballast water Other Lake Erie Basin Invertebrate Schizopera harpacticoid Yes In GL PontoCaspian 1998 Ballast water Benthic Lake borutzkyi copepod Basin Crustacean Michigan Cercopagis pengoi fish-hook Yes In GL PontoCaspian 1998 Ballast water Zooplankton Lake waterflea Basin Ontario Psammonobiotus testate amoeba Yes In GL PontoCaspian 2001 Ballast water Other Lake communis Basin Invertebrate Ontario Psammonobiotus testate amoeba Yes In GL PontoCaspian 2002 Ballast water Other Lake linearis Basin Invertebrate Ontario Psammonobiotus testate amoeba Yes In GL PontoCaspian 2002 Ballast water Other Lake dziwnowi Basin Invertebrate Ontario Enteromorpha green alga No Out of GL Widespread 2003 Ballast water Benthic alga Lake flexuosa Basin Michigan Hemimysis mysid shrimp Yes In GL PontoCaspian 2006 Ballast water Benthic Lake anomala Basin Crustacean Michigan Najas marina spiny naiad No Out of GL Eurasia 1864 Solid ballast Plant Lake

TABLE A.1. (Continued) Scientific Name Common Name First Current Endemic Region Year Pathway Taxonomic Lake Discovered NA Distri- Discovered Category Where First in NA in bution Discovered the Great Lakes? Basin Ontario Bithynia faucet snail Yes Out of GL Eurasia 1871 Solid Ballast Mollusk Lake tentaculata Basin Michigan Pisidium pea clam Yes Out of GL Eurasia 1895 Solid Ballast Mollusk Lake moitessierianum Basin Superior Pisidium amnicum pea clam Yes Out of GL Eurasia 1897 Solid Ballast Mollusk Lake Basin Ontario Valvata piscinalis European valve Yes Out of GL Eurasia 1897 Solid Ballast Mollusk Lake snail Basin Ontario Pisidium henslow's pea Yes In GL Eurasia 1916 Solid Ballast Mollusk Lake 167 henslowanum clam Basin Ontario Carassius auratus goldfish No Out of GL Asia 1878 Unauthorized Fish Wide- Basin release spread Radix auricularia European ear No Out of GL Eurasia 1901 Unauthorized Mollusk Lake snail Basin release Michigan Aeromonas furunculosis No Out of GL Unknown 1902 Unauthorized Other Unknown salmonicida Basin release invertebrate Acentropus niveus aquatic moth No Out of GL Eurasia 1927 Unauthorized Other Lake Erie Basin release invertebrate Nymphoides peltata yellow floating No Out of GL Eurasia 1930 Unauthorized Plant Lake Erie heart Basin release Cipangopaludina Oriental No Out of GL Asia 1931 Unauthorized Mollusk Niagara chinensis malleata mystery snail Basin release River

Craspedacusta freshwater No Out of GL Asia 1933 Unauthorized Other Lake Erie sowerbyi jellyfish Basin release invertebrate

TABLE A.1. (Continued) Scientific Name Common Name First Current Endemic Region Year Pathway Taxonomic Lake Discovered NA Distri- Discovered Category Where First in NA in bution Discovered the Great Lakes? Misgurnus Oriental Yes In GL Asia 1939 Unauthorized Fish Lake anguillicaudatus weatherfish Basin release Huron Branchiura oligochaete No Out of GL Asia 1951 Unauthorized Annelid Lake sowerbyi Basin release Michigan Myriophyllum Eurasian No Out of GL Eurasia 1952 Unauthorized Plant Lake Erie spicatum watermilfoil Basin release Cordylophora hydroid No Out of GL Ponto-Caspian 1956 Unauthorized Other Lake Erie caspia Basin release invertebrate Trapa natans water chestnut No Out of GL Eurasia 1959 Unauthorized Plant Lake Basin release Ontario 168 Glugea hertwigi protozoan No Out of GL Eurasia 1960 Unauthorized Other Lake Erie Basin release invertebrate Myxobolus salmonid No Out of GL Unknown 1968 Unauthorized Other Lake Erie

(Myxosoma) whirling Basin release invertebrate cerebralis disease Hydrocharis European No Out of GL Eurasia 1972 Unauthorized Plant Lake morsus-ranae frogbit Basin release Ontario Sphacelaria brown alga Yes In GL Asia 1975 Unauthorized Benthic Lake fluviatilis Basin release Alga Michigan Corbicula fluminea Asiatic clam No Out of GL Eastern Asia 1980 Unauthorized Mollusk Lake Erie Basin release Argulus japonicus parasitic No Out of GL Asia 1988 Unauthorized Zooplankton Lake copepod Basin release Michigan Scardinius rudd No Out of GL Eurasia 1989 Unauthorized Fish Lake erythrophthalmus Basin release Ontario Daphnia lumholtzi waterflea No Out of GL Africa 1999 Unauthorized Zooplankton Lake Erie Basin release

TABLE A.1. (Continued) Scientific Name Common Name First Current Endemic Region Year Pathway Taxonomic Lake Discovered NA Distri- Discovered Category Where First in NA in bution Discovered the Great Lakes? Ranavirus sp. largemouth No Out of GL Unknown 2002 Unauthorized Virus Lake bass virus Basin release Michigan (LMBV) Tanysphyrus aquatic weevil Yes In GL Eurasia 1943 Unknown Other Unknown lemnae Basin Invertebrate Potamothrix bedoti oligochaete Yes In GL PontoCaspian 1950 Unknown Annelid Lake Basin Ontario Potamothrix oligochaete Yes In GL PontoCaspian 1952 Unknown Annelid Lake moldaviensis Basin Ontario Neoergasilus copepod Yes Out of GL Asia 1994 Unknown Zooplankton Lake 169 japonicus Basin Huron Megacyclops viridis cyclopoid Yes In GL Europe 1994 Unknown Zooplankton Lake copepod Basin Huron Heterosporis sp. microsporidian Yes In GL Unknown 2000 Unknown Other Lake Basin Invertebrate Ontario Rhabdovirus carpio SVC spring Yes In GL Eurasia 2001 Unknown Virus Lake viraemia of Basin Michigan carp Cylindrospermopsis cyanobacterium Yes In GL South America 2002 Unknown Phyto- Lake raciborskii Basin plankton Michigan

APPENDIX B:

SUPPORTING TEXT FOR CHAPTER THREE: SHIP-BORNE NONINDIGENOUS

SPECIES DIMINISH GREAT LAKES ECOSYSTEM SERVICES6

B.1 Methods

In structured expert judgment (SEJ), experts are used as scientific instruments, the precision and accuracy of which can be calibrated, to distill the available knowledge on a topic of interest (Cooke 1991). Structured expert judgment has previously been used in various environmental applications to assess the likelihood of natural disasters (e.g., volcanic eruption, dam failure; Aspinall et al. 2003), the consequences of nuclear accidents (Cooke and Goossens 2000), the drivers of climate change (Morgan et al. 2006,

Lenton et al. 2008), and increases in mortality attributable to air pollution (Cooke et al.

2007). The National Research Council and the US EPA are both considering increased use of SEJ for environmental risk assessment and consequence analysis (NRC 2002, IEI

2006).

B.1.1 Selection of Experts

6 This appendix will be included as supporting on-line materials when the journal article based on Chapter Three is submitted for publication to Science. 170

Based on a review of the relevant scientific literature and conversations with senior Great Lakes (GL) researchers, we identified experts on the GL whose professional opinions are generally well respected and who could evaluate the effects of ship-borne species in the context of multiple interacting factors (e.g., social trends, economic issues, land use change, management activities). We invited eleven experts to participate. Ten accepted and are identified in alphabetical order in Table B.1. One expert provided assessments only in regard to biofouling of raw water intakes. For confidentiality, the experts were randomly assigned the titles “Expert 1” through “Expert 10”.

B.1.2 Briefing Book

We prepared a 136-page briefing book that contained a list identifying ship-borne and other alien species in the GL, historical data on the commercial and recreational fisheries of the GL, and training materials on uncertainty and probabilistic assessment.

We encouraged experts to review the booklet prior to their interview and to refer to it as desired during the interview. Along with the booklet, we also provided the experts a complete copy of the elicitation questionnaire well in advance of their interview. A full copy of the briefing materials and questionnaire are available at http://aquacon.nd.edu/research/. The questionnaire is also provided as Appendix C.

B.1.3 Individual Interviews

One to three members of our research team, including Rothlisberger, Cooke, and

Lodge, conducted expert interviews. We started each interview by briefly presenting information on our project, the general approach of SEJ, and the quantification of uncertainty. Each expert also responded to several practice questions, similar to those on

171

TABLE B.1.

EXPERTS INTERVIEWED AND THE PROFESSIONAL TITLE, AFFILIATION,

AND QUALIFICATIONS OF EACH (LISTED ALPHABETICALLY).

Name Title, affiliation, and qualifications Richard Aiken Natural resource economist with the US Fish and Wildlife Service. His office administers and analyzes the USFWS National Survey of Fishing, Hunting, and Wildlife-Associated Recreation (USFWS 2007). Renata Claudi Former employee of Ontario Power Generation whose main duties included dealing with biofouling problems, active organizer of the annual International Conference on Aquatic Invasive Species, and owner of a biofouling consulting firm. Mark P. Ebener Fisheries assessment biologist with the Chippewa-Ottawa Resource Authority. Ebener has chaired the Lake Superior and Lake Huron Technical Committees and served on the Lake Michigan Technical Committee of the Great Lakes Fishery Commission (GLFC). Leroy J. Hushak Professor Emeritus of Agricultural, Environmental and Development Economics at The Ohio State University. Hushak has conducted research on the value of recreation in the Great Lakes and the effects of dreissenid mussels on Great Lakes basin water treatment facilities, electric power plants and industrial water users. Roger L. Knight Lake Erie Fisheries Program Administrator for the Ohio Department of Natural Resources, Division of Wildlife. Knight serves on the Lake Erie Committee and the Council of Lake Committees of the GLFC. Frank Lupi Associate Professor of Environmental and Natural Resource Economics at Michigan State University. Lupi studies fish and wildlife demand and valuation and the economics of ecosystem services in the Great Lakes region. Lloyd C. Mohr Fisheries Assessment Team Leader for the Upper Great Lakes Management Unit of the Ontario Ministry of Natural Resources. Mohr has been active in and chaired the GLFC’s Lake Huron Technical Committee. Charles R. O’Neill, Jr. Senior Extension Associate with New York Sea Grant and the director of Sea Grant’s National Aquatic Nuisance Species Clearinghouse. O’Neill has led research initiatives regarding the fouling effects of dreissenid mussels on raw water users in the Great Lakes region and has served for the past four years as a member of the Federal Invasive Species Advisory Committee. Donald Scavia Professor of Natural Resources at the University of Michigan and Director of Michigan Sea Grant. Scavia oversees several large-scale research projects on drivers and conditions of Great Lakes ecosystems. Roy A. Stein Professor of Evolution, Ecology and Organismal Biology and Director of the Aquatic Ecology Laboratory at The Ohio State University. Stein served as a US Commissioner on the GLFC during 1998-2004.

172

the questionnaire, for which they received immediate feedback as to the true value of the variable they were assessing. During the elicitation, we asked experts to provide the 5th,

50th, and 95th percentiles of their subjective cumulative probability distribution function for 41 variables pertaining to the impacts of ship-borne species on four ecosystem services in 2006. The variables we asked experts about were pounds of commercially landed fish from the US waters of the GL, angler-days of sport fishing effort on the US waters of the GL, overall expenditures for sport fishing in the US waters of the GL, participant-days of wildlife viewing across all 8 US states bordering the GL, and additional costs to raw water users in the GL region of the US.

A typical pair of questions took the following form, which asks first for the actual value of the variable in 2006 (given that ship-borne species are present) and then asks for what the value of the variable would have been if no ship-borne species were present.

How many total pounds of commercial fish were landed from the US waters of

Lake Erie in 2006?

5%______50%______95%______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological

and commercial factors unchanged. How many total lbs of commercial fish WOULD

HAVE BEEN landed from the US waters of Lake Erie in 2006?

5%______50%______95%______

Following each such pair of questions, the interviewer said to the expert, “Please explain your thinking as to the mechanisms of ship-borne NIS impacts on this variable.”

We made notes and audio recordings of these responses, transcribing them after the interviews.

173

B.1.4 Performance measures and combination of expert judgments: the classical model

We report below the assessments of each individual expert for each variable, as well as combined assessments for each variable. Following Cooke’s classical model

(Cooke 1991), we combined expert assessments in two ways: (a) each expert’s assessment was given equal weight or (b) individual assessments were weighted according to the expert’s performance (both accuracy and informativeness) on the calibration questions. We refer to the result of this second combination method as the performance-based combination (PBC) and use it in the results reported above. We provide more technical details on both approaches below.

There are two generic, quantitative measures of performance, calibration and

information. Calibration measures the statistical likelihood that a set of empirical observations correspond, in a statistical sense, with the expert’s assessments. Information measures the degree to which a distribution is concentrated.

For each variable, each expert divides the range into four inter-quantile intervals for which his/her probabilities are known, namely p1 = 0.05: less than or equal to the 5% value, p2 = 0.45: greater than the 5% value and less than or equal to the 50% value, etc.

If N variables are assessed, each expert may be regarded as a statistical hypothesis, namely that each realization falls in one of the four inter-quantile intervals with probability vector

p= (0.05, 0.45, 0.45, 0.05).

Suppose we have realizations x1,…xN of these quantities (i.e., calibration variables). We may then form the sample distribution of the expert's inter-quantile intervals as:

174

s1(e) = #{ i | xi ≤ 5% quantile}/N

s2(e) = #{ i | 5% quantile < xi ≤ 50% quantile}/N

s3(e) = #{ i | 50% quantile < xi ≤ 95% quantile}/N

s4(e) = #{ i | 95% quantile < xi }/N

s(e) = (s1,…s4).

Note that the sample distribution depends on the expert e. If the realizations are indeed drawn independently from a distribution with quantiles as stated by the expert then the quantity

2NI(s(e) | p) = 2N ∑i=1..4 si ln(si / pi) ...... (1)

is asymptotically distributed as a chi-square variable with 3 degrees of freedom. This is

the so-called likelihood ratio statistic, and I(s | p) is the relative information of

distribution s with respect to p. If we extract the leading term of the logarithm we obtain the familiar chi-square test statistic for goodness of fit.

There are advantages in using the form in Eq. 1 (Cooke 1991). For example, if after a few realizations the expert were to see that all realizations fell outside his 90% central uncertainty intervals, he might conclude that these intervals were too narrow and might broaden them on subsequent assessments. This means that for this expert the uncertainty distributions are not independent, and he learns from the realizations. Expert learning is not a goal of an expert judgment study and his joint distribution is not elicited.

Rather, it is preferable that experts do not need to learn from the elicitation. Hence, the combination method scores Expert e as the statistical likelihood of the hypothesis

He: “the inter-quantile interval containing the true value for each variable is

drawn independently from probability vector p.”

175

A simple test for this hypothesis uses the test statistic (Eq. 1), and the likelihood, or p-value, or calibration score of this hypothesis, is:

Cal(e) = p-value = Prob{ 2NI(s(e) | p)≥ r | He},

where r is the value of Eq. 1 based on the observed values x1,…xN. The resulting p-value is the probability under hypothesis He that a deviation at least as great as r should be observed on N realizations if He were true.

Although the calibration score uses the language of simple hypothesis testing, it must be emphasized that we are not rejecting expert hypotheses; rather we are using this language to measure the degree to which the data supports the hypothesis that the expert's probabilities are accurate. Low scores, near zero, mean that it is unlikely that the expert’s probabilities are correct.

The second scoring variable is information. Loosely, the information in a distribution is the degree to which the distribution is concentrated. Information cannot be measured absolutely, but only with respect to a background measure. Being concentrated or “spread out” is measured relative to some other distribution.

Measuring information requires associating a density with each quantile assessment of each expert. To do this, we use the unique density that complies with the experts' quantiles and is minimally informative with respect to the background measure.

This density can easily be found with the method of Lagrange multipliers. For a uniform background measure, the density is constant between the assessed quantiles, and is such that the total mass between the quantiles agrees with p. The background measure is not elicited from experts as indeed it must be the same for all experts; instead it is chosen by the analyst.

176

The uniform and log-uniform background measures require an intrinsic range on which these measures are concentrated. The classical model implements the so-called k% overshoot rule: for each item we consider the smallest interval I = [L, U] containing all the assessed quantiles of all experts and the realization, if known. This interval is extended to

I* = [L*, U*]; L* = L – k(U-L)/100; U* = U + k(U-L)/100.

The value of k is chosen by the analyst. A large value of k tends to make all experts look quite informative, and tends to suppress the relative differences in information scores. In this study, we used a uniform background measure and selected k to be 0.1.

The information score of Expert e on assessments for uncertain quantities 1…N is Inf (e) =Average Relative information w.r.t. Background = (1/N) ∑i = 1..N I(fe,i | gi),

where gi is the background density for variable i and fe,i is expert e's density for item i.

This is proportional to the relative information of the expert's joint distribution given the background, under the assumption that the variables are independent. As with calibration, the assumption of independence here reflects a desideratum of the combination method and not an elicited feature of the expert's joint distribution. The information score does not depend on the realizations. An expert can give himself a high information score by choosing his quantiles very close together. The information score of e depends on the intrinsic range and on the assessments of the other experts. Hence, information scores cannot be compared across studies.

Of course, other measures of concentrated-ness could be contemplated. The above information score is chosen because it is familiar, tail insensitive, scale invariant, and

177

slow. The latter property means that relative information is a slow function; large changes in the expert assessments produce only modest changes in the information score.

This contrasts with the likelihood function in the calibration score, which is a very fast function. This causes the product of calibration and information to be driven by the calibration score.

The combined score of Expert e will serve as an (unnormalized) weight for e:

wα(e) = Cal (e) × Inf (e) × 1α(Cal(e) ≥ α), ...... (2)

where 1α(Cal(e)α) = 1 if Cal(e) ≥ α, and is zero otherwise. The combined score thus depends on α. If Cal(e) falls below cut-off level α Expert e is unweighted. The presence of a cut-off level is imposed by the requirement that the combined score be an asymptotically strictly proper scoring rule. That is, an expert maximizes his/her long run expected score by and only by ensuring that his probabilities p= (0.05, 0.45, 0.45, 0.05)

correspond to his/her true beliefs. α is similar to a significance level in simple hypothesis

testing, but its origin is indeed different. The goal of scoring is not to “reject” hypotheses,

but to measure “goodness” with a strictly proper scoring rule.

A combination of expert assessments is called a “decision maker” (DM). All decision makers discussed here are examples of linear pooling. The classical model is essentially a method for deriving weights in a linear pool. “Good expertise” corresponds to good calibration (i.e., high statistical likelihood, high p-value) and high information.

We want weights which reward good expertise and which pass these virtues on to the decision maker.

The reward aspect of weights is very important. We could simply solve the following optimization problem: find a set of weights such that the linear pool under

178

these weights maximizes the product of calibration and information. Solving this problem on real data, one finds that the weights do not generally reflect the performance of the individual experts. As we do not want an expert's influence on the decision maker to appear haphazard, and we do not want to encourage experts to game the system by tilting their assessments to achieve a desired outcome, we must impose a strictly proper scoring rule constraint on the weighing scheme.

The scoring rule constraint requires the term 1α(calibration score), but does not say what value of α we should choose. Therefore, we choose α so as to maximize the combined score of the resulting decision maker. Let DMα(i) be the result of linear pooling for item i with weights proportional to (2):

DMα(i) = ∑e=1,..E wα(e) fe,i / ∑e=1,..E wα(e) ...... (3)

The optimized global weight DM is DMα* where α* maximizes

calibration score(DMa) × information score(DMα)...... (4)

This weight is termed global because the information score is based on all the assessed seed items.

A variation on this scheme, which we employ here, allows a different set of weights to be used for each item. This is accomplished by using information scores for each item rather than the average information score:

wα (e,i) = 1α(calibration score)×calibration score(e) × I(fe,i | gi)...... (5)

For each α we define the Item weight DMα for item i as

IDMα(i) = ∑e=1,..E wα(e,i) fe,i / ∑e=1,..E wα(e,i)...... (6)

The optimized item weight DM is IDMα* where α* maximizes

calibration score(IDMa) × information score(IDMα)...... (7)

179

The non-optimized versions of the global and item weight DMs are obtained simply by setting α = 0.

We used item weights for generating PBC “decision makers” because they allow an expert to up- or down- weight him/herself for individual items according to how much

(s)he feels (s)he knows about that item. “Knowing less” means choosing quantiles further apart and lowering the information score for that item. Of course, good performance of item weights requires that experts can perform this up/down-weighting successfully. Both item and global weights can be described as optimal weights under a strictly proper scoring rule constraint. In both global and item weights, calibration dominates over information, information serves to modulate between more or less equally well calibrated experts. Further details on assessing expert performance and combining expert judgments can be found elsewhere (Cooke 1991).

We used 12 of the 41 variables elicited as calibration variables (i.e., realizations).

The true values of these variables were not known at the time of the study, but became available soon after. These calibration variables included commercial landings and sport fishing participation and expenditures and wildlife viewing participation for 2006. It is worth noting that the experts' p-values are quite uneven; only Experts 2 and 6 exhibit good statistical accuracy. The equal-weight combination's statistical accuracy is marginal, whereas that of the PBC is quite good.

B.1.5 Percent impacts on ecosystem services

We calculated estimates of median percent impacts and the associated 90% uncertainty range using the convolution of the joint probabilities of the distributions of

180

the ‘without ship-borne species’ PBC minus the ‘with ship-borne species’ PBC, assuming independence of all variables. This produced a single distribution of differences between the ‘without ship-borne species’ and ‘with ship-borne species’ assessment for each variable (e.g., sport fishing participation in 2006). We then divided the 5th, 50th, and 95th percentiles of this distribution of differences by the associated median ‘with ship-borne species’ PBC assessment and multiplied the quotient by 100 to generate percent impact of ship-borne species.

The sport fishery of L. Superior is small relative to the sport fisheries of the other

GL and because of this is more volatile than the fisheries of the other lakes in terms of percent impacts. There were large discrepancies among the experts as to median impacts here, with several indicating zero impact from ship-borne species, but with others estimating impacts at 70 to 150% (Figure B.3). These discrepancies reduced our confidence in PBC median impact assessment for this variable (35%, Table 3.1), especially in light of studies showing that L. Superior has been relatively minimally impacted by ship-borne invasive species (Phaneuf and Smith 2005). Therefore, we viewed the PBC’s percent impact (35%) of ship-borne species on participation in the

Lake Superior sport fishery with skepticism and we chose not to include this likely overestimate in our economic impact calculations, so as to avoid erroneous inflation of these estimates.

B.1.6 Economic Valuation of Impacts

We present a methodological advance as we apply an economic approach to capturing the economic value of the consequences of ship-borne species on the Great

Lakes region. From an economic viewpoint, if nonindigenous species affect the 181

provisioning of ecosystem services, they can result in lost consumer surplus (i.e., opportunity costs to consumers). Consumer surplus is the benefit to consumers of a market outcome and accrues whenever consumers pay less than their maximum willingness to pay for that unit of a good. An example might be of a consumer going to their local fish market and paying $5 for a pound of walleye, knowing that the maximum they would be willing to pay for that pound is $10. As what they actually pay is less than their maximum willingness to pay, the $5 difference is a measure of the benefits to the consumer from the exchange.

Market prices capture the relative rate at which the market is willing to exchange one good for another. The method often employed in estimating the value of lost economic activity is the product of market price and a change in quantity, or engineering

(replacement) cost estimates. These tend to be rejected as valid estimates of economic impact because they have no relationship to surplus measures (Phaneuf and Smith 2005).

Estimating economic surplus, however, can be a challenge because it requires more information than market prices and quantities.

To calculate the surplus measures we used two standard methods under the following assumptions: each estimate was calculated in isolation of the other (i.e., neglecting any interaction effects) and under the presumption that everything else (e.g., environmental conditions, economic conditions) would have remained exactly the same with and without the invaders. Given these assumptions, for commercial fishing we used a simple market model of demand. For the recreation-based values of sport fishing and wildlife viewing, we employed a simple benefits transfer method.

182

The market model employed for commercial fishing is illustrated in Figure B.1.

The SEJ provides predicted quantities of commercial landings with (Qin) and without

(Qwo) invaders for each lake. (The figure illustrates a situation when invaders lead to

lower landings.) For a given price per unit (Pin), the engineering cost method would calculate the welfare loss as lost revenues due to invaders (Figure B.1a, Area [QinabQwo]) as a measure of the welfare loss imposed. However, as noted, a more accurate measure of the welfare loss to consumers is lost consumer surplus, which requires a specification of the market and how the changes in predicted landings are reflected in the market. We assume that the predicted changes in landings are a result of enhanced populations of species harvested and these enhancements only serve to increase the quantity landed (i.e., they do not result in any change in the consumer demand function). In this restricted view, all that is required additionally is an estimate of the relevant demand curve (Figure

B.1b).

What is critical about the market method is that the price consumers are willing to pay depends in an inverse fashion upon how much they are able to buy (Figure B.1b,

Demand curve). Thus, if there were more landings without invaders, the price consumers would be willing to pay per unit would decline as they bought more fish. The change in consumer surplus is then given by area [PwoPinac] which may or may not correspond to the engineering cost estimate of area [QinabQwo].

The demand curve (Figure B.1b) for commercial fishing in each lake provides a means of estimating how the quantity change predicted by the SEJ influences market prices given consumer tastes, which are reflected by their willingness to pay. While estimating the demand curve for each lake is beyond the scope of this research, an 183

(a) Predicted changes and replacement costs

Price per unit

a b

Qin Qwo Quantity

(b) Predicted changes and consumer surplus

Price per unit

200 a b Pin 6

Demand curve

Q Q in wo Quantity

Figure B.1. Schematic of welfare changes related to commercial fishing, illustrating the market model approach taken to estimate economic impacts of ship- borne species.

184

approximation can be made by assuming demand to be linear (as sketched). Within this specification the only additional data necessary are estimates of own-price elasticities of demand. The own-price elasticity of demand is a parameter that measures the responsiveness of consumers to changes in price. Unfortunately, specific estimates were not available for the aggregated Great Lakes fisheries, so we constructed a distribution of fish demand own-price elasticity estimates by accessing all relevant estimates from the

USDA’s database (http://www.ers.usda.gov/Data/Elasticities/Query.aspx ) and augmenting these with estimates from the seminal work in the literature (Cheng and

Capps 1988). Summary statistics for elasticities are presented in Table B.2.

While the market-based methods for commercial fishing are straightforward, how to measure the value of the changes on sport fishing and wildlife watching is more complicated. The problem is that when considering these outdoor recreation activities, the goods (e.g., sport fishing and wildlife watching) are not traded in well-defined markets (as are fish caught commercially) making the employment of a market model as developed above difficult. Furthermore, with recreation, how to interpret the drivers of the expert predictions is also not readily apparent. For example, even if the combined expert assessments predict an increase in sport fishing without invaders, it is still not clear what characteristic of sport fishing causes the increase.

The usual method employed to deal with the first problem (i.e., missing markets) is to focus on related goods. There are complimentary goods that consumers purchase when recreating (i.e., expenditures on time and travel) and these goods are traded in markets. The likely answer to the second question (i.e., what drives the change) is that an improvement in quality of the resource (i.e., improvements in environmental quality) 185

TABLE B.2

SUMMARY STATISTICS ON VALUES DRAWN FROM THE LITERATURE ON

OWN-PRICE ELASTICITY OF DEMAND OF COMMERCIAL FISH, THE VALUE

OF SPORT FISHING IN THE GREAT LAKES, AND THE VALUE OF WILDLIFE

VIEWING IN THE GREAT LAKES REGION

Median Mean Standard Min Max Sample size deviation (# estimates) Category Fish Elasticity -0.48 -0.52 0.25 -1.13 -0.10 27 Sport Fishing 19.35 34.94 42.58 0.26 194.66 53 Value Wildlife 25.23 33.52 18.81 2.57 103.13 65 Viewing Value

SOURCE: Cheng and Capps 1988, Boyle et al. 1998, http://www.ers.usda.gov/Data/Elasticities/Query.aspx

NOTE: Fish own-price elasticity of demand is unitless. Outdoor recreation values are given as consumer surplus per day in 2007 US dollars.

186

leads to increased demand for outdoor recreation (and complementary goods) and vice versa.

The method assumes that the effects of changes in the consumption of complimentary goods (arising from a change in environmental quality) provide an indirect indication of the value of recreation. This assumption of ‘weak complementarity’ provides the basis of a large amount of research in environmental economics. Various sources provide excellent descriptions of this problem and its resolutions (Cheng and Capps 1988, Spash and Vatn 2008).

The case of recreational demand, for either sport fishing or wildlife watching, is shown in Figure B.2. The combined expert assessments provide predicted quantities of days spent recreating in both activities. The change in recreation is assumed to be due to a change in the quality of the resources, with (qin) and without (qwo) invaders, which leads to a change in consumption of related goods with (Xin) and without (Xwo) invaders. The figure illustrates a situation when invaders lead to a lower level of environmental quality and lower related consumption of related goods. For a given price per unit of the related good (Px) the engineering cost method would calculate the welfare loss as lost revenues due to invaders (Area [XinabXwo]) as a measure of the welfare loss imposed. Changes in consumer surplus arise because of the change in environmental quality, which shift the demand curve for the related good from D(qin) to D(qwo). The change in consumer surplus is given by the area [abcd], which again may or may not correspond to the engineering cost estimate.

However, estimating the demand for related goods that is a function of environmental quality is not trivial. It is here that we employ benefits transfer methods,

187

Price per unit of related c good

a b Px D(qin) D(qwo)

Xin Xwo Quantity of related good, X

Figure B.2. Schematic of welfare changes related to outdoor recreation, illustrating the inferred market model approach used to estimate economic impacts of ship-borne species.

where benefits estimates from a specific study are “transferred” to another similar case or situation (Spash and Vatn 2008). While there are many assumptions and caveats that come with the technique, it has been the focus of a great deal of research and allows one way to address some of these problems. We follow the intent of the method in a very simple fashion and use distributions of previously estimated consumer surplus for Great

Lakes sport fishing and wildlife watching in conjunction with the SEJ prediction. Table

B.2 summarizes the per day sportfishing estimates derived from a query of all Great

Lakes fishing from the “Sportfishing Values Database” (Boyle et al. 1998, http://www.indecon.com/fish/). Estimates of the value of wildlife watching are not as regionally precise. The best available distribution is one of wildlife watching consumer

188

surplus estimates for the Northeast region of the United States (Kaval and Loomis 2003,

Table B.2).

For raw water usage, we did not perform any economic modeling because the

values we elicited from experts were per facility costs resulting directly from biofouling

by ship-borne species for four different facility types (i.e., nuclear power generation

plants, fossil fuel power generation plants, industrial facilities, and municipal water

plants). We scaled these additional costs from biofouling up to the regional level by

multiplying per facility costs by the number of facilities of each type that draw water

from the GL in the US (Deng 1996).

B.2 Results

Equal weighting and performance-based combination (PBC) of expert assessments produced similar results with respect to median percent impacts, but the equal-weighted combination was not statistically accurate (p<0.05), whereas the null hypothesis of statistical accuracy of the PBC could not be rejected. Furthermore, the

PBC allowed us to weight experts according to the precision and accuracy of their assessments (Table B.3; Figures B.3, B.4). For this reason and for brevity, we report here the results of the PBC, focusing on median values because they reflect the outcomes experts saw as the most probable.

In all categories, for any given variable, uncertainty ranges varied substantially across individual experts (Figures B.3, B.4, B.5). Relative uncertainty ranges seemed to depend more on expert identity than on the variable being assessed. Uncertainty was

189

TABLE B.3.

CALIBRATION, INFORMATIVENESS, AND WEIGHTS OF THE NINE EXPERTS, THEIR EQUAL-WEIGHT COMBINATION

(EQUAL), AND THEIR PERFORMANCE-BASED COMBINATION (PBC) FOR THE IMPACTS OF SHIP-BORNE SPECIES ON

THE GREAT LAKES IN 2006

Expert or Calibration Weight Mean relative information Weight normalized combination (p-value) un-normalized All Calibration 190 Without combo With combo variables variables 1 4.029E-5 0.801 0.9401 3.788E-5 0 0.000183 2 0.09646 1.007 0.7019 0.06771 0.3524 0.3272 3 0.0001173 1.522 1.259 0.0001476 0 0.0007134 4 0.0001173 1.508 1.003 0.0001177 0 0.0005686 5 0.0007465 0.5782 1.131 0.0008441 0 0.004079 6 0.4539 0.4212 0.2741 0.1244 0.6476 0.6013 7 0.0001173 1.17 0.9895 0.0001161 0 0.0005608 8 4.856E-6 1.372 1.367 6.637E-6 0 3.208E-5 9 1.912E-9 2.336 2.86 5.469E-9 0 2.643E-8 EQUAL 0.04411 0.2671 0.3066 0.01353 – 0.06536 PBC 0.9281 0.389 0.2086 0.1936 – 0.5019 PBC no 0.9281 0.3827 0.2061 0.1913 – 0.4972 optimization

Note: Values in parentheses in Column 5 are un-normalized weights of experts below the significance cut-off. The relative contribution of each expert to the PBC, which is used as the primary metric of interest throughout this paper, is shown in Column 6. The sum of the last column is 1 when the PBC weights, which were derived from other normalizations, are excluded. For details on the meaning of columns not fully explained here, see Tuomisto et al. 2008.

a b c d

0.0 0.5 1.0 1.5 2.0 2.5 3.0 0 2 4 6 8 e f g h

0 2 4 6 8 10 12 14 0 10 20 30 40 i j

k l

0 2 4 6 8 0 5 10 15 20 m n o p

0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 q r

s t

0 2 4 6 8 10 0.0 0.1 0.2 0.3 0.4 0.5 Sport FishingSport Effort Fishing Effort (Millions Commercialof Angler Commercially Landings Landed Fish (Millions of (Millions of Angler-days) (Millions of Pounds)

Figure B.3. Individual and combined expert assessments showing the impact of ship-borne species on US sport fishing effort (left column) and US commercial fish landings (right column) in 2006. There are two rows of panels for each lake with the first row showing expert assessments for the variable (i.e., angler-days or pounds of commercially landed fish) with ship-borne species (i.e., actual condition) and the second row showing assessments for the variable if ship-borne species had never been introduced (i.e., hypothetical condition). Dashed lines divide assessments with and without ship- borne species. The order of the lakes is, from top to bottom, Superior (a-d), Michigan (e-h), Huron (i-l), Erie (m-p), and Ontario (q-t). Within each panel expert assessments are arranged in order, from top to bottom, Expert 1, 2, 3, 4, 5, 6, 7, 8, 9, equally-weighted combination, and performance-based combination. Vertical light gray bars in panels a, b, e, f, i, j, m, n, q, and r show the realization of the variable in question, which was unknown at the time of the elicitation. Note that the scale of the horizontal axes varies.

191

b ae f

bg dh

0 50 100 150 200 250 300 0 100 200 300 400 Wildlife ViewingWildlife Effort Watching Effort (Millions of Participant-days) (Millions of Participant-days)

192

0 100 200 300 400 500

0 50 100 150 c

0 50 100 150 200 250

0 20 40 60 80 100 120 Expense (Thousands of 2007 US $)

Figure B.5. Individual and combined expert assessments of the annual per facility impacts of ship-borne species on US raw water users in 2006. The order of the raw water users, from top to bottom, is nuclear power plants (a), water treatment plants (c), fossil fuel power plants (e), and industrial facilities (g). Within each panel expert assessments are arranged in order, from top to bottom, Expert 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, equally-weighted combination, and performance- based combination. Note that the scale of the horizontal axis is different for each user type.

193

almost universally greater for ‘without ship-borne species’ assessments than for ‘with ship-borne species’ assessments. Combined assessments for a given variable ‘with’ and

‘without ship-borne species’ differed more from one another than did the assessments of any single expert for the same ‘without-with’ pair (Figures B.3, B.4, B.5).

B.2.1 Economic Valuation of Impacts

Given the lack of precision in the impact distributions (Figure 3.1), employing a point estimate to describe them is likely to lead to misleading inferences. However, what is useful is whether or not each pair-wise comparison of a variable with and without ship- borne species reveals a clear shift in the distribution one way or another. To capture this,

Table 3.1 includes (for the ‘without-with’ distributions) percentages of each of these distributions where the net is above zero and above 100. This metric demonstrates the general direction of changes in the distributions.

To evaluate the prediction distributions for commercial fishing and outdoor recreation from a policy or economic perspective, given uncertainty in economic parameters and SEJ predictions, joint distributions of the impacts by economic activity were generated by combining distributions of the SEJ predictions with the distributions of economic parameters, where each distribution was assumed to be independent of the other. For each ecosystem service, 50,000 randomly drawn SEJ predictions were combined with ecosystem service-specific 50,000 randomly drawn economic parameters to calculate distributions of the changes in consumer surplus for each ecosystem service.

Tables B.4 and B.5 summarize these distributions of economic impacts.

Without direct knowledge of the true distribution of the economic parameters, for convenience and in accord with the elicited distributions, we assumed all economic 194

TABLE B.4.

SUMMARIES OF COMMERCIAL FISHERY

CONSUMER SURPLUS PREDICTION DISTRIBUTIONS

IN THE GREAT LAKES IN THE UNITED STATES IN 2006

Note: The state of the lakes without any ship-borne nonindigenous species is indicated by W/O, as opposed to the actual invaded state of the lakes (INV).

195

TABLE B.5.

SUMMARIES OF OUTDOOR RECREATION CONSUMER SURPLUS

PREDICTION DISTRIBUTIONS IN THE GREAT LAKES

IN THE UNITED STATES IN 2006

Wildlife Watching Total* | State US Total | INV $6,061 $604 $15,111 US Total | W/O $6,098 $598 $15,348 US Total | W/O - INV $12 5,818 $6,030 50%

Note: The state of the lakes without any ship-borne nonindigenous species is indicated by W/O, as opposed to the actual invaded state of the lakes (INV).

196

parameters were distributed according to uniform and triangle distributions. These forms mesh well with the limited data available and either place the same weight upon each point of the distribution (i.e., uniform) or place more weight on those values in the middle of the distribution (i.e., triangle). The lake-by-lake results shown are based on the uniform distribution. Results based on the triangle distribution are similar to, but with a slightly more narrow range, than those based on the uniform distribution.

As would be expected, given additional uncertainty in the economic parameters, the economic distributions become wider than the elicited distributions and as such have their mass spread more thinly, reducing confidence further in any single point estimates.

An informative description can perhaps best be provided by the 90% interval of the distributions. In addition, given other policy relevant costs of comparison, a useful description can be provided by the proportion of each distribution above certain values

(this is in addition to a focus only on the median). This method is most useful when describing changes in consumer surplus between the ‘without’ and ‘with ship-borne NIS’ states. Figure 3.2 provides the proportion of each difference in consumer surplus distributions above certain dollar values chosen arbitrarily for policy inference.

For estimated declines in consumer surplus for commercial fishing, only Lakes

Erie, Huron, and Michigan have over 50% of their predicted distributions greater than

$0.5 million. While these results suggest that impacts on commercial fishing are likely greater than zero, they are small.

In contrast, for sport fishing, all lakes have more than 50% of their distributions being greater than $5 million. In addition, Lakes Erie, Huron, Michigan, and Superior all have more than 50% of their distributions above $10 million. Lake Erie has more than

197

50% above $50 million. The implications are that, within the assumptions outlined above, the experts believe there are likely impacts in the millions for sport fishing.

However the only lake that looks likely to have tens of millions in impacts is Lake Erie, which appears to have several times the magnitude of impact than any other lake.

For each lake, the results for commercial and sport fishing are characterized by a high degree of uncertainty. In the face of this uncertainty, aggregating across lakes helps to make inferences about some overall trends in the distributions. At an aggregate level the distributions have obvious regions of highest relative frequency and indicate that economic losses from ship-borne species are likely greater than zero (Figure 3.2).

In the commercial fishery, 90% of the distribution with ship-borne species lies between $5-29 million in consumer surplus. Without ship-borne species, this 90% interval increases to $6-55 million. Seventy-four percent of the difference in the two distributions (with minus without ship-borne species) lies above zero, 70% exceeds $1 million, while the proportion above $5 million falls to 50%, with higher values increasingly unlikely. This distribution provides a clear indication that the experts expect an ecosystem without ship-borne species would provide higher overall fishery landings

(Figure 3.2A).

The uncertainty in the sport fishery is also significant, with 90% of the distribution with ship-borne species within the range of $205-2,434 million in consumer surplus. Without ship-borne species, the distribution shifts and becomes wider, with 90% of its mass lying within a range of $232-2,833 million. Seventy-two percent of the difference distribution (i.e., without minus with ship-borne species) is positive, more than

65% of is above $50 million and 50% lies above $100 million (Figure 3.2B). This

198

distribution therefore provides a strong indication that a system without ship-borne species is expected to provide a substantially greater amount of sport fishing (Figure

3.2B).

The difference in consumer surplus for wildlife watching has a median of $12.2 million, with 50% of the distribution being greater than this number. The range of the distribution is so great that it is reasonable to infer little more than that wildlife watching has a 50% chance of experiencing impacts greater than zero.

The estimated distributions for raw water users were elicited directly. That is, experts gave estimated costs of biofouling for each type of facility or, in other words, the costs of the invaded state. Table B.6 presents summary statistics of the elicited distributions. The 90% intervals provide a good description of the distributions (e.g., fossil fuel facilities are predicted to have biofouling costs between $1.7-13.9 million).

B.2.2 Expert Rationales

The explanations experts gave for their assessments provide a sense of the mechanisms through which ship-borne species affect the GL. We briefly summarize these rationales here, organized by ecosystem service, lake (where possible), and year. In these summaries, we seek to be inclusive, mentioning the mechanisms of impact given by each expert. Therefore, the statements about mechanisms of impact in these summaries do not necessarily represent the consensus view of the experts. Indeed, according to the

SEJ protocol we used, no effort is made to build consensus among the experts, each of whom responded to the survey independently, having no information on the responses of the other expert participants. However, to give a sense of whether or not there was broad

199

TABLE B.6.

ADDITIONAL ANNUAL OPERATING COSTS TO

RAW WATER USERS ATTRIBUTABLE TO

SHIP-BORNE SPECIES IN THE GREAT LAKES REGION

IN THE UNITED STATES IN 2006

Median Per Facility Cost # of Regional Cost Facility Type (Thousands of 2007 US$) Facilities (Millions of 2007 US$) Nuclear 118.1 (43.5, 211.3) 13a 1.54 (0.57, 2.75) Power Plant Fossil Fuel 28.1 (6.6, 53.5) 260a 7.31 (1.72, 13.91) Power Plant Municipal 32.5 (4.9, 61.3) 436b 14.17 (2.14, 26.73) Water Plant Industrial 30.4 (4.6, 56.7) 117b 3.56 (0.54, 6.63) Facility Total – – – 26.57 (4.96, 50.02)

Note: Regional costs are median per facility costs from combined expert assessments multiplied by the number of each facility type in the GL region. Ninety percent uncertainty ranges appear in parentheses.

a Power Plants in the Great Lakes Basin, Northeast Midwest Institute Report (http://www.nemw.org/Power%20plants%20in%20the%20Great%20Lakes%20Basin.pdf, accessed: 5/6/2008)

b Approximated from Deng (1996) and O’neill (1996).

200

acknowledgement of an impact mechanism, we also indicate how many experts referred to the same mechanisms for each variable. For some variables, not all experts provided a mechanistic rationale for their assessments, often providing an overarching rationale for each broad category of variables (e.g., commercial fishing), without providing specific, individual rationales for each separate lake with respect to that variable. While our focus here is on the effects of ship-borne species, experts sometimes identified as important the interactions ship-borne species have with alien species delivered by vectors other than shipping and with various other factors, including eutrophication, nutrient abatement, pollution, land use change, and cultural change.

Commercial Fishing

All experts thought ship-borne species are an important factor in the continuing decline of commercial fishing in the GL. Two experts indicated that the decline of commercial fishing is also a result of sport fishing being more economically valuable and having stronger political support in the US. Two experts said that commercial fishing is shrinking because of the relative instability in production of the GL system, causing some global markets to lose interest and shift to other sources of fish, including aquaculture.

These experts thought that some of this instability in the GL commercial fishery is because of the presence of ship-borne species.

L. Superior

Seven experts said that the L. Superior commercial fishery was functioning well as of 2006, with impacts from ship-borne species being relatively small. Lake trout are reproducing naturally and whitefish are abundant, which bodes well for the commercial fishery. Although fish stocks are doing well in L. Superior, one expert said the market

201

for commercial fish is currently weaker than it has been historically, leading to declining harvest levels.

L. Michigan

Ship-borne species have had a variety of direct impacts on the food web of L.

Michigan. Four experts pointed out that zebra and quagga mussels have caused major changes. One important effect of these mussels, specifically mentioned by two experts, is that they have reduced the food supply of whitefish, which are currently 50% lighter at age than prior to the dreissenid invasion. The decline of the benthic crustacean Diporeia sp., an energy-rich staple of whitefish diet, is linked to the presence of dreissenids, said one expert. According to one expert, the salmon catch is also lower than it would have been without ship-borne species; this is because dreissenids consume phytoplankton, making it unavailable to alewives; with less phytoplankton to eat, alewives grow less, resulting in less food for salmon. One expert said the predatory spiny waterflea

Bythotrephes longimanus has caused substantial reductions in energy flow to fishes by adding a link in the pelagic food web between phytoplankton and planktivorous fish.

Experts also mentioned indirect impacts of ship-borne species on food webs. For instance, two experts said that in L. Michigan (and L. Huron), it is almost impossible to fish with nets at certain times of year because of green algae blooms (Cladophora spp.).

These experts said these blooms occur because dreissenids concentrate phosphorus in the benthos, fertilizing the algae. The same two experts indicated that, as filter feeders, dreissenids also increase water clarity, allowing increased light penetration and promoting algal photosynthesis. This smothering ‘wall of green’ occasionally eliminates the gillnet fishery, occasionally reducing fishing effort by 20-30%.

202

L. Huron

Three experts specifically said L. Huron fisheries are currently failing, indicating that the effects of ship-borne species on L. Huron’s commercial fishery were similar to those on that of L. Michigan (see above). In contrast, one expert characterized the fishery as stable, but currently at a lower equilibrium point than if dreissenids had not driven down Diporeia populations. The alewife, whose numbers are currently declining, has been an important forage fish in the food web in the past 50 years, but is itself not native to the GL. One expert pointed out that it is difficult to say how the absence of ship-borne species would have altered the role of alewife (which did not arrive by ship) in the food web.

L. Erie

According to two experts, since the early 1980s, there has been less and less commercial fishing on L. Erie, caused by a ban on gillnetting yellow perch and legislative buyouts of many commercial licenses for conversion to sport licenses. Compared to changes in fishery allocation, ship-borne species have not affected commercial catch, in part because high biological productivity. According to these two experts, there is ample food for commercial fish species, despite dreissenid-induced food web alterations. One expert said that damages from round gobies have been limited and may have even helped to improve the fishery because dreissenid mussels are food for round gobies, which in turn are eaten by piscivorous fish such as walleye (Sander vitreus).

L. Ontario

Three experts said that the impacts of ship-borne species, especially dreissenid mussels and spiny waterflea, have been immense in L. Ontario, making the commercial

203

fishery much smaller than it would have been if these species were not present. One expert stated that continuing declines of alewife and other forage species are probably linked to food web changes attributable to dreissenids. According to one expert, L.

Ontario has been particularly sensitive to impacts caused by ship-borne species because of the sharp contrast between nearshore and offshore habitats and their associated biological communities. The other Great Lakes have more transitional habitat between nearshore habitats, which are now dominated by dreissenids, and offshore zones. Such transition zones have given native species some latitude in adjusting to the presence of ship-borne species and have thus provided a buffer against impacts of ship-borne species in all the GL except L. Ontario. The same expert said that commercial catch could have been orders of magnitude higher without ship-borne species, but other factors, including contaminants, lack of prey base for piscivores, and habitat destruction (e.g., loss of gravel beds for spawning, nursery wetlands) have also impaired commercial catch. Three experts pointed out that recent declines in commercial catch are driven by allocation decisions made by New York state, which no longer issues commercial licenses, favoring instead sport fishing licenses. In fact, according to one expert, the current management emphasis in L. Ontario is to promote recreational fishing for Chinook, coho, and Atlantic salmon.

Sport Fishing

All experts agreed that overall, if not for ship-borne species, fishing opportunities would have been better and more people would have participated in sport fishing in 2006.

However, three experts indicated that societal changes that are not clearly linked to ship- borne species are also causing declines in participation. For example, sport fishing

204

participation in 2006 is probably largely driven by fuel prices, which in 2006 were nearly

50% higher than in 2001, the date of the most recent historical sport fishing participation estimates. Moreover, two experts said that target species availability and the breadth of species available are more important drivers of sport fishing participation than fish abundance. According to these experts, ship-borne species probably have had little effect on species availability. Nevertheless, in lakes where the benthic community has been affected by ship-borne species (i.e., all but L. Superior), four experts asserted that there would have been better angling opportunities without ship-borne species, leading to more angler participation.

L. Superior

The five experts who provided rationales indicated that ship-borne species have had little effect on the L. Superior sport fishery. However, the four experts who did not provide rationales indicated high percent impacts on this fishery. Partly because of this lack of rationale from the experts, we did not include Lake Superior in our economic calculations.

L. Michigan

One expert said that salmon fishing in L. Michigan has been tremendous recently, but that there might have been more participants if not for angler attrition attributable to perceived damage from ship-borne species. A different expert said that round gobies might be having an effect on the sport fishery because they may be reducing yellow perch populations.

205

L. Huron

Three experts said that ship-borne species have probably played a major role in the collapse of L. Huron’s sport fishery. The mechanism they provided in explanation is that alewife populations, the prey base for salmon, have collapsed and now salmon are following suit, with total salmon landings declining to half what they recently were.

According to these experts, alewife are collapsing because zooplankton, a key component of alewife diet, have declined dramatically recently, probably because of dreissenids’ consumption of phytoplankton.

L. Erie

Four experts indicated that the effects of alien species in L. Erie have been minor, but also said that the whole system would have been somewhat better off without ship- borne species. Two of these experts pointed out that the ship-borne species in L. Erie have been there for many years now and the lake still supports a solid sport fishery.

According to one expert, the phosphate ban that resulted in lower ecosystem productivity in L. Erie is a more important limitation on sport fishing opportuinities there (because of nutrient limitation) than the effects of ship-borne species.

L. Ontario

Three experts stated that without ship-borne species, participation might have been slightly higher in L. Ontario, even though fishing opportunities there have been generally good recently. These experts said ship-borne species have destabilized the food web of L. Ontario. Were it not for ship-borne species in the system, the food web would have been more stable, leading to better fishing opportunities and resulting in greater angler participation.

206

Raw water use (biofouling)

Five experts expected biofouling costs to remain similar to historical costs reported in the literature. Three experts said that technological advances have lessened present-day costs and pointed out that historical costs often included retrofitting expenses, whereas now that anti-biofouling devices are installed, the primary costs are only for their maintenance.

Wildlife viewing

Seven experts stated that ship-borne species have had little or no impact on wildlife viewing in the eight states bordering the GL. However, one expert said a possible mechanism for negative impacts of ship-borne species on wildlife viewing is the concentration of botulism toxins in gobies and dreissenids that may increase the mortality rates of waterfowl that consume these benthic species. A different expert said that bird kills from botulism may not necessarily reduce the number of bird watchers. Three experts thought that the consumption of dreissenid mussels by scaups and other waterfowl could, on the contrary, increase bird populations, leading to more bird watching opportunities.

207

APPENDIX C:

EXPERT ELICITATION PROTOCOL FOR ECOLOGICAL AND ECONOMIC

IMPACTS OF SHIP-BORNE NONINDIGENOUS SPECIES ON THE GREAT LAKES

C.1 Purpose

Two waves of nonindigenous species (NIS) are responsible for ecological perturbations in the Great Lakes.

The first wave (e.g., rainbow smelt, alewife, sea lamprey) occurred prior to 1959,

and was not associated with transoceanic shipping. The supplied background materials

include a complete list of nonindigenous species in the Great Lakes and their probable

introduction vector.

The second wave (e.g., spiny and fishhook waterfleas B. longimanus and

Cercopagis pengoi; dreissenid mussels D. polymorpha, D. bugensis; round goby

Neogobius melanostomus, viral hemorrhagic septicemia) began in 1959 with the opening

of the St. Lawrence Seaway. A current list of ship-borne nonindigenous species is

included in the supplied background materials.

The full consequences of both invasion waves may not have occurred yet.

The purpose of this elicitation is to quantify the economic impacts – with

uncertainty – of nonindigenous species introduced through ballast water of transoceanic

shipping since 1959. We therefore focus on the second wave of NIS.

208

C.2 Scope

We are interested in quantifying the impacts up to the present and out to 2025,

taking account of currently established species. The dimensions of economic damages are

1. Commercial fishing, direct impact from changes in harvest plus associated changes in sales and services 2. Sport fishing, also direct impact plus effects on sales and services 3. Zebra mussel fouling, costs incurred by o Nuclear power plants o Fossil fuel o Water treatment plants o Industrial facilities 4. Wildlife watching, direct impact plus sales and services

C.3 Method

To the extent possible, the assessments will be based on available data. However, gaps and shortfalls in data necessitate an appeal to structured expert judgment. Structured expert judgment has been widely applied in risk analysis for many years but - understandably - still meets skepticism among researchers, stakeholders and general public. Use of structured expert judgment also involves greater uncertainty. For these reasons, it is imperative to document fully all steps in the process, and to validate the uncertainty assessments to the extent possible. Validation requires eliciting uncertainty on variables whose true values will be known within the time frame of the study.

C.3.1 Format

All the questions will have the same format. You will be given the description of

an uncertain quantity taking values in a continuous range. You are asked to quantify your

uncertainty by giving 5, 50, and 95 percentiles of your uncertainty distribution. For

example: 209

How many total pounds of yellow perch were caught in 2005 in Lake Michigan by commercial fishing? ...... 5% ______50% ______95% ______

Presumably, this number is uncertain. If you fill in

How many total pounds of yellow perch were caught in 2005 in Lake Michigan by commercial fishing? ...... 5% ___17,000 50% __25,000_ 95% _40,000_

this means that you believe there is a 5% chance that the actual number is below 17,000; there is a 50-50 chance that it is below 25,000, and a 95% chance that it is below 40,000.

The true value is 23,609lbs. This is not a surprising value relative to this assessment. If the value were 15,000 this would be surprising, as would 45,000. In each case the realization would be outside the 90% confidence band.

An expert’s probabilistic assessments are statistically accurate if 10% of the realizations fall outside the 90% confidence band; 50% of the realizations fall on either side of the median (50% value).

If your assessments had been

5% __2,000___ 50% ___50,000__ 95% __150,000__

You would have been equally un-surprised, but your assessments would be less informative.

To get a feeling for this format, please complete the following assessments:

Data from the US Geological Survey on the total lbs of selected fish species caught by

US commercial fishing operations in Lake Michigan and Lake Erie are given below.

210

Lake Michigan Lake Erie Year Rainbow Smelt Walleye Walleye 1971 1,309,217 10,008 55,525 1980 974,202 2,265 80,505 1990 3,120,668 806 10,190 1995 1,422,539 245 41,145 2000 387,819 12,718 186 2004 415,828 21,969 300

How many total pounds of walleye were caught in 2005 in the US waters of Lake Erie by commercial fishing? ...... 5% ______50% ______95% ______

How many total pounds of rainbow smelt were caught in 2005 in the US waters of Lake

Michigan by commercial fishing? 5% ______50% ______95% ______

How many total pounds of walleye were caught in 2005 in the US waters of Lake

Michigan by commercial fishing? 5% ______50% ______95% ______

C.3.2 What is a good probability assessor?

A good probability assessor is one whose assessments, taken together, show good statistical accuracy, and which are informative. Of these two, statistical accuracy is more important, informativeness is important to discriminate between statistically accurate assessments.

It is essential for the credibility of the results that the combined expert judgments display good statistical accuracy and high informativeness. For this reason, we will ask you in the sequel to assess items whose true values will become know within the time frame of the study.

211

C.3.3 Expert Names

The policy regarding the use of expert names reflects the desire to shield experts from intrusive “expert shopping” by interested stakeholders, while at the same time satisfying the demands of scientific reproducibility and transparency. Expert names and affiliations are part of the published documentation, as are the individual assessments.

The association of names and assessments are preserved only in the unpublished records of the research group. However, the association of names with individual assessments is never included in the open publications.

C.3.4 Assumptions

For many questions below you are asked what various quantities would have been if ship-borne NIS were not present in the Great Lakes. In forming your answer please consider that the following are among possible causes of change.

• First wave of nonindigenous species • Loss and/or restoration of habitat • Overfishing and/or stock restoration • Environmental degradation and/or rehabilitation • Economic factors such as changes in trade competition and market demands • Climate change • Eutrophication and/or pollution control • Changes in social and cultural trends • Changes in fishing regulations

212

C.4 Questions

C.4.1 Commercial fishing

Data from the US Geological Survey on the total lbs of fish caught by US commercial fishing operations is given below. The supplied reference materials contain more extensive data and graphs.

(Source: http://www.glsc.usgs.gov/main.php?content=products_data_fishingreports&title=Data0&menu=products)

Total lbs caught in commercial fishing (USGS) Year Total Superior Michigan Huron Erie Ontario 1971 65,022,239 6,147,423 46,971,316 2,769,782 8,826,329 307,389 1980 48,790,513 3,330,179 23,702,050 2,357,060 19,191,273 209,951 1990 40,242,125 3,566,605 25,630,568 4,876,925 6,025,295 142,732 1995 26,532,062 1,820,615 14,624,952 4,930,205 5,092,092 64,198 2002 17,474,630 2,043,044 6,900,790 4,031,475 4,458,238 41,083 2005 18,402,001 2,175,415 7,362,035 4,034,824 4,822,333 7,394

How many total lbs of commercial fish were landed in the US waters of the GL in 2006?(units: pounds) (1) Lake Superior ...... 5% ______50% ______95% ______(2) Lake Michigan ...... 5% ______50% ______95% ______(3) Lake Huron ...... 5% ______50% ______95% ______(4) Lake Erie ...... 5% ______50% ______95% ______(5) Lake Ontario ...... 5% ______50% ______95% ______Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. How many total lbs of commercial fish WOULD HAVE BEEN landed from the US waters of the GL in 2006? (units: pounds) (6) Lake Superior 5% ______50% ______95% ______(7) Lake Michigan ...... 5% ______50% ______95% ______

(8) Lake Huron ...... 5% ______50% ______95% ______

(9) Lake Erie ...... 5% ______50% ______95% ______(10) Lake Ontario 5% ______50% ______95% ______

213

How many total lbs of commercial fish WILL BE landed from the US waters of the GL in 2025? (units: pounds) (11) Lake Superior ...... 5% ______50% ______95% ______(12) Lake Michigan ...... 5% ______50% ______95% ______(13) Lake Huron ...... 5% ______50% ______95% ______(14) Lake Erie ...... 5% ______50% ______95% ______(15) Lake Ontario ...... 5% ______50% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. How many total lbs of commercial fish WOULD HAVE BEEN landed from the US waters of the Great Lakes in 2025? (units: pounds) (16) Lake Superior ...... 5% ______50% ______95% ______(17) Lake Michigan ...... 5% ______50% ______95% ______(18) Lake Huron ...... 5% ______50% ______95% ______(19) Lake Erie ...... 5% ______50% ______95% ______(20) Lake Ontario ...... 5% ______50% ______95% ______

214

Data from Fisheries and Oceans Canada on the total lbs of fish landings by

Canadian commercial fishing operations in the province of Ontario (primarily in the

Great Lakes bordering Ontario) is given below. The supplied reference materials contain more extensive data and graphs.

(Source: http://www.dfo-mpo.gc.ca/communic/statistics/commercial/landings/index_e.htm)

Total lbs caught by commercial fishing in the province of Ontario 1971 42,793,930 1980 58,866,629 1990 55,029,585 2002 31,739,952 2004 32,846,672 2005 29,506,669

How many total lbs of commercial fish were landed in the province of Ontario in 2006? (units: pounds) (21) 5% ______50% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. How many total lbs of commercial fish WOULD HAVE BEEN landed in the province of Ontario in 2006? (units: pounds) (22) 5% ______50% ______95% ______

How many total lbs of commercial fish WILL BE landed in the province of Ontario in 2025? (units: pounds) (23) 5% ______50% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. How many total lbs of commercial fish WOULD HAVE BEEN landed in the province of Ontario in 2025? (units: pounds) (24) 5% ______50% ______95% ______

215

C.4.2 Commercial Fishing Effort

Data from several sources on commercial fishing effort with a variety of gears is given below.

Commercial Fishing Effort Superior Huron Large-mesh Small-mesh Trap Net Large-mesh Small-mesh Trap Net Gillnet Gillnet (1000s net Gillnet Gillnet Year (1000s net lifts) (1000s km) (1000s km) lifts) (1000s km) (1000s km) 1981 17638 6076 38.47 12584 9213 38.40 1985 19056 4605 31.44 14382 12809 37.20 1990 33775 2610 29.57 12584 10112 44.80 1995 12224 2093 31.81 16854 10337 35.60 1999 10681 1464 41.83 14157 6180 29.73

What was the commercial fishing effort on the Great Lakes listed below in 2006? (units: thousands of net kilometers or lifts) Large-mesh Gillnet (km) Small-mesh Gillnet (km) Trap Net (lifts) Lake Superior (25) (26) (27) 5% ______5% ______5% ______50% ______50% ______50% ______95% ______95% ______95% ______Lake Huron (28) (29) (30) 5% ______5% ______5% ______50% ______50% ______50% ______95% ______95% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. What WOULD HAVE BEEN the commercial fishing effort on the Great Lakes listed below in 2006? (units: thousands of net kilometers or lifts) Large-mesh Gillnet (km) Small-mesh Gillnet (km) Trap Net (lifts) Lake Superior (31) (32) (33) 5% ______5% ______5% ______50% ______50% ______50% ______95% ______95% ______95% ______Lake Huron (34) (35) (36) 5% ______5% ______5% ______50% ______50% ______50% ______95% ______95% ______95% ______

216

What WILL BE the commercial fishing effort on the Great Lakes listed below in 2025? (units: thousands of net kilometers or lifts) Large-mesh Gillnet (km) Small-mesh Gillnet (km) Trap Net (lifts) Lake Superior (37) (38) (39) 5% ______5% ______5% ______50% ______50% ______50% ______95% ______95% ______95% ______Lake Huron (40) (41) (42) 5% ______5% ______5% ______50% ______50% ______50% ______95% ______95% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. What WOULD HAVE BEEN the commercial fishing effort on the Great Lakes listed below in 2025? (units: thousands of net kilometers or lifts) Large-mesh Gillnet (km) Small-mesh Gillnet (km) Trap Net (lifts) Lake Superior (43) (44) (45) 5% ______5% ______5% ______50% ______50% ______50% ______95% ______95% ______95% ______Lake Huron (46) (47) (48) 5% ______5% ______5% ______50% ______50% ______50% ______95% ______95% ______95% ______

217

C.4.3 Sport Fishing

Data on the total number of recreational angler days on the US waters of the Great

Lakes from the US Fish and Wildlife Service are presented below.

(Numbers in thousands; Source: http://federalaid.fws.gov/surveys/surveys.html, 1991: Table 32, 1996: Table 26, 2001: Table 27, Preliminary 2006: Table 1)

Thousands of Angler days on US waters TOTAL Superior Michigan Huron Erie Ontario 1991 17562 883 5090 2113 7082 2394 1996 17277 1301 4338 2059 6421 3158 2001 17916 601 4836 1171 7748 3560 2006 14447 Data not yet available

What was the total number of angler days on the US waters of the GL in 2006? (units: thousands of angler days) (49) Lake Superior ...... 5% ______50% ______95% ______(50) Lake Michigan ...... 5% ______50% ______95% ______(51) Lake Huron ...... 5% ______50% ______95% ______(52) Lake Erie ...... 5% ______50% ______95% ______(53) Lake Ontario ...... 5% ______50% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. What WOULD HAVE BEEN the total number of angler days on the US waters of the Great Lakes in 2006? (units: thousands of angler days) (54) Lake Superior ...... 5% ______50% ______95% ______(55) Lake Michigan ...... 5% ______50% ______95% ______(56) Lake Huron ...... 5% ______50% ______95% ______(57) Lake Erie ...... 5% ______50% ______95% ______(58) Lake Ontario ...... 5% ______50% ______95% ______

What WILL BE the total number of angler days on the US waters of the GL in 2025? (units: thousands of angler days) (54) Lake Superior ...... 5% ______50% ______95% ______(55) Lake Michigan ...... 5% ______50% ______95% ______(56) Lake Huron ...... 5% ______50% ______95% ______(57) Lake Erie ...... 5% ______50% ______95% ______(58) Lake Ontario ...... 5% ______50% ______95% ______

218

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. What WOULD HAVE BEEN the total number of angler days on the US waters of the Great Lakes in 2025? (units: thousands of angler days) (64) Lake Superior ...... 5% ______50% ______95% ______(65) Lake Michigan ...... 5% ______50% ______95% ______(66) Lake Huron ...... 5% ______50% ______95% ______(67) Lake Erie ...... 5% ______50% ______95% ______(68) Lake Ontario ...... 5% ______50% ______95% ______

Data from Fisheries and Oceans Canada on the total number of recreational angler days on the Canadian waters of the Great Lakes are presented below.

(Numbers in thousands; Source: http://www.dfo-mpo.gc.ca/communic/Statistics/recreational/great_lakes/index_e.htm)

Thousands of Angler days on Canadian waters TOTAL Superior Huron Erie Ontario 1980 11,727 Data disaggregated by lake not available 1985 10,679 1990 11,364 444 4,887 2,377 3,656 1995 6,576 375 2,559 1,536 2,106 2000 5,142 202 2,497 1,134 1,309

What was the total number of angler days on the Canadian waters of the GL in 2006? (units: thousands of angler days) (69) Lake Superior ...... 5% ______50% ______95% ______(70) Lake Huron ...... 5% ______50% ______95% ______(71) Lake Erie ...... 5% ______50% ______95% ______(72) Lake Ontario ...... 5% ______50% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. What WOULD HAVE BEEN the total number of angler days on the Canadian waters of the GL in 2006? (units: thousands of angler days) (73) Lake Superior ...... 5% ______50% ______95% ______(74) Lake Huron ...... 5% ______50% ______95% ______(75) Lake Erie ...... 5% ______50% ______95% ______(76) Lake Ontario ...... 5% ______50% ______95% ______

219

What WILL BE the total number of angler days on the Canadian waters of the GL in 2025? (units: thousands of angler days) (77) Lake Superior ...... 5% ______50% ______95% ______(78) Lake Huron ...... 5% ______50% ______95% ______(79) Lake Erie ...... 5% ______50% ______95% ______(80) Lake Ontario ...... 5% ______50% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. What WOULD HAVE BEEN the total number of angler days on the Canadian waters of the GL in 2025? (units: thousands of angler days) (81) Lake Superior ...... 5% ______50% ______95% ______(82) Lake Huron ...... 5% ______50% ______95% ______(83) Lake Erie ...... 5% ______50% ______95% ______(84) Lake Ontario ...... 5% ______50% ______95% ______

220

Data on the aggregated expenditures of recreational anglers (expressed in 2007

dollars) on the US waters of the Great Lakes from the US Fish and Wildlife Service are

presented below.

(Numbers in thousands; Source: http://federalaid.fws.gov/surveys/surveys.html, 1991: Table 19, 1996: Table 15, 2001: Table 16)

Thousands of Dollars Spent by Recreational Anglers fishing on US waters TOTAL (X 1000) (expressed in 2007 dollars) 1991 2,019,162 ($0.791/angler) 1996 1,843,686 ($0.904/angler) 2001 1,482,002 ($0.802/angler)

Give estimates as sum total expenditures, not per angler amounts.

How many total dollars did anglers spend in connection with fishing excursions on the US waters of the GL in 2006? (units: 2007 dollars) (85) All Great Lakes ...... 5% ______50% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. How many dollars WOULD anglers HAVE SPENT in connection with fishing excursions on the US waters of the Great Lakes in 2006? (units: 2007 dollars) (86) All Great Lakes ...... 5% ______50% ______95% ______

How many dollars WILL anglers spend in connection with fishing excursions on the US waters of the GL in 2025? (units: 2007 dollars) (87) All Great Lakes ...... 5% ______50% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. How many dollars WOULD anglers HAVE SPENT in connection with fishing excursions on the US waters of the Great Lakes in 2025? (units: 2007 dollars) (88) All Great Lakes ...... 5% ______50% ______95% ______

221

C.4.4 Fouling water intake for power plants & industry

From O’Neill (1996), a survey of 766 infrastructure owners/operators in ZM

infested region with 56.92% response rate. Total sector-wide expenses (in 2007 dollars)

increased from $315,851 in 1989 to $23,945,787 in 1995.

Facility Type Sector mean annual per facility costs of responding facilities (1989-1995), expressed in 2007 dollars Nuclear Power plants (NPP) $151,600 Water treatment plants (WTP) $41,310 Fossil fuel generating plants (FFPP) $28,063 Industrial facilities (IF) $32,189 The above figures report biological fouling costs associated with zebra mussels

alone. Our questions below invite you to also include costs associated with other nonindigenous fouling organisms, like quagga mussels, and other organisms, like

Microcystis, that may cause taste and odor problems in drinking water. Though

Microcystis is not a nonindigenous species its abundance appears to be strongly linked to

the presence of zebra mussels.

In your estimates, please also consider the possible effects of future invaders on

raw water users.

What was the mean per facility cost in each sector, including non- respondents, of biological fouling in 2006? (units: 2007 dollars) NPP WTP FFPP IF All (89) (90) (91) (92) Great 5% ______5% ______5% ______5% ______50% ______50% ______50% ______50% ______Lakes 95% ______95% ______95% ______95% ______

What WILL BE the mean per facility cost in each sector of biological fouling in 2025? (units: 2007 dollars) NPP WTP FFPP IF All (93) (94) (95) (96) Great 5% ______5% ______5% ______5% ______50% ______50% ______50% ______50% ______Lakes 95% ______95% ______95% ______95% ______

222

C.4.5 Wildlife watching

The US Fish and Wildlife service reports the following participant days of wildlife watching in the US states bordering the

Great Lakes.

(Numbers in thousands; Source: http://federalaid.fws.gov/surveys/surveys.html, 1991: Table 72, 1996: Table 63, 2001: Table 68)

Thousands of participant days in Wildlife Watching in Great Lakes US States TOTAL Minnesota Wisconsin Michigan Illinois Indiana Ohio Pennsylvania New York 1991 98,610 10,378 12,914 14,159 8,464 7,135 12,769 20,062 12,729 1996 84,449 6,807 12,154 16,162 9,416 5,912 11,418 13,123 9,457 2001 123,783 13,243 16,499 13,999 7,656 11,999 19,814 18,990 21,583

223 In total, how many participant-days of wildlife watching occurred in the US states bordering the Great Lakes in 2006? (units: thousands of participant days)

(97) All Great Lakes...... 5% ______50% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. How many participant-days of wildlife watching WOULD HAVE OCCURRED in the US states bordering the Great Lakes in 2006? (units: thousands of participant days) (98) All Great Lakes...... 5% ______50% ______95% ______

In total, how many participant-days of wildlife watching WILL OCCUR in the US states bordering the Great Lakes in 2025? (units: thousands of participant days)

(99) All Great Lakes...... 5% ______50% ______95% ______

Suppose ship-borne NIS were NOT present, with all other unrelated ecological and commercial factors unchanged. How many participant-days of wildlife watching WOULD HAVE OCCURRED in the US states bordering the Great Lakes in 2025? (units: thousands of participant days) (100) All Great Lakes ...... 5% ______50% ______95% ______

C.5 Answers to Training Questions 224

How many total pounds of walleye were caught in 2005 in the US waters of Lake Erie by commercial fishing?

830

How many total pounds of rainbow smelt were caught in 2005 in the US waters of Lake Michigan by commercial fishing?

675,880

How many total pounds of walleye were caught in 2005 in the US waters of Lake Michigan by commercial fishing?

19,176

LITERATURE CITED

Achard, M., M. Baudrimont, A. Boudou, and J. P. Bourdineaud. 2004. Induction of a multixenobiotic resistance protein (MXR) in the Asiatic clam Corbicula fluminea after heavy metal exposure. Aquatic Toxicology 67:347-357.

Acharya, K., J. D. Jack, and A. S. Smith. 2006. Stoichiometry of Daphnia lumholtzi and their invasion success: are they linked? Archiv fur Hydrobiologie 165:433-453.

Aiken, S. G., P. R. Newroth, and I. Wile. 1979. Biology of Canadian Weeds .34. Myriophyllum-Spicatum L. Canadian Journal of Plant Science 59:201-215.

Allan, J. D., R. Abell, Z. Hogan, C. Revenga, B. W. Taylor, R. L. Welcomme, and K. Winemiller. 2005. Overfishing of inland waters. Bioscience 55:1041-1051.

Anderson, R. M., and R. M. May. 1991. Infectious diseases of humans: dynamics and control. Oxford University Press, Oxford ; New York.

Andrews, J. H., editor. 1986. Nuisance vegetation in the Madison lakes: current status and options for control. University of Wisconsin, Madison, WI.

Annin, P. 2006. The Great Lakes water wars. Island Press, Washington.

Apostolakis, G. 1990. The concept of probability in safety assessments of technological systems. Science 250:1359-1364.

Arnott, D. L., and M. J. Vanni. 1996. Nitrogen and phosphorus recycling by the zebra mussel (Dreissena polymorpha) in the western basin of Lake Erie. Canadian Journal of Fisheries and Aquatic Sciences 53:646-659.

Ashworth, W. 1986. The late, Great Lakes : an environmental history. 1st edition. Knopf, New York.

Asmus, R. M., and H. Asmus. 1991. Mussel beds: limiting or promoting phytoplankton? Journal of Experimental Marine Biology and Ecology 148:215-232.

Aspinall, W. P., G. Woo, B. Voight, and P. J. Baxter. 2003. Evidence-based volcanology: application to eruption crises. Journal of Volcanology and Geothermal Research 128:273-285.

225

Austin, J. C., S. Anderson, P. N. Courant, and R. E. Litan. 2007. Healthy waters, strong economy: the benefits of restoring the Great Lakes ecosystem. The Brookings Institution, Washington, DC.

Bailey, R. C., L. Grapentine, T. J. Stewart, T. Schaner, M. E. Chase, J. S. Mitchell, and R. A. Coulas. 1999. Dreissenidae in Lake Ontario: impact assessment at the whole lake and Bay of Quinte spatial scales. Journal of Great Lakes Research 25:482- 491.

Barbiero, R. P., and M. L. Tuchman. 2004. Changes in the crustacean communities of Lakes Michigan, Huron, and Erie following the invasion of the predatory cladoceran Bythotrephes longimanus. Canadian Journal of Fisheries and Aquatic Sciences 61:2111-2125.

Barko, J. W., and R. M. Smart. 1980. Mobilization of sediment phosphorus by submersed freshwater macrophytes. Freshwater Biology 10:229-238.

Barnhisel, D. R. 1991. Zooplankton spine induces aversion in small fish predators. Oecologia 88:444-450.

Barnhisel, D. R., and H. A. Harvey. 1995. Size-specific fish avoidance of the spined crustacean Bythotrephes: field support for laboratory predictions. Canadian Journal of Fisheries and Aquatic Sciences 52:768-775.

Bates, A. L., E. R. Burns, and D. H. Webb. 1985. Eurasian watermilfoil (Myriophyllum spicatum L.) in the Tennessee Valley: an update on biology and control. Pages 104-115 in Proceedings of the First International Symposium on the watermilfoil (Myriophyllum spicatum) and related Haloragaceae species. Aquatic Plant Management Society, Washington, D. C.

Bayley, S., V. D. Stotts, P. F. Springer, and J. Steenis. 1978. Changes in the submerged aquatic macrophyte populations at the head of the Chesapeake Bay, 1958-1975. Estuaries 1:73-84.

Beaven, G. F. 1955. Water chestnut threatens disaster to Maryland water areas. Maryland Tidewater News 12:1-2.

Beaven, G. F. 1959. The water chestnut down, but not out! Maryland Tidewater News 15:8.

Beekey, M. A. 2004. Zebra mussels affect benthic predator foraging success and habitat choice on soft sediments. Oecologia 141:164-170.

Beeton, A. M. 1969. Changes in the environment and biota of the Great Lakes. Pages 150-187 in N. A. o. S. (U.S.), editor. Eutrophication: causes, consequences, correctives. National Academy of Sciences, Washington, D. C.

226

Beeton, A. M., and J. Hageman. 2002. Impact of Dreissena polymorpha on the zooplankton community of western Lake Erie. Verh. Internat. Verein. Limnol. 27:3798-3804.

Benoit, H. P., O. E. Johannsson, D. M. Warner, W. G. Sprules, and L. G. Rudstam. 2002. Assessing the impact of a recent predatory invader: the population dynamics, vertical distribution, and potential prey of Cercopagis pengoi in Lake Ontario. Limnology and Oceanography 47:626-635.

Besha, J. A., and W. D. Countryman. 1980. Feasibility assessment of anaerobic digestion of European water chestnuts (Trapa natans L.). 80-13, New York State Energy Research and Development Authority, Albany, NY.

Bially, A., and H. J. MacIsaac. 2000. Fouling mussels (Dreissena spp.) colonize soft sediments in Lake Erie and facilitate benthic invertebrates. Freshwater Biology 43:85-97.

Bickley, W. E., and E. N. Cory. 1955. Water caltrop in the Chesapeake Bay. Association of Southeastern Biologists Bulletin 2:27-28.

Blackburn, T. M., and K. J. Gaston. 2005. Biological invasions and the loss of birds on islands: insights into the idiosynchrasies of extinction. Pages 85-110 in D. F. Sax, J. J. Stachowicz, and S. D. Gaines, editors. Species invasions : insights into ecology, evolution, and biogeography. Sinauer Associates, Sunderland, Mass.

Bobeldyk, A. M., J. M. Bossenbroek, M. A. Evans-White, D. M. Lodge, and G. A. Lamberti. 2005. Secondary spread of zebra mussels (Dreissena polymorpha) in coupled lake-stream systems. Ecoscience 12:339-346.

Bogucki, D. J., G. K. Gruendling, and M. Madden. 1980. Remote sensing to monitor water chestnut growth in Lake Champlain. Journal of Soil and Water Conservation 35:79-81.

Bogue, M. B. 2000. Fishing the Great Lakes : an environmental history, 1783-1933. University of Wisconsin Press, Madison, Wisconsin.

Bonopartis, N. 2001. Details few in Esopus drownings. Page 1B Poughkeepsie Journal, Poughkeepsie, NY.

Bossenbroek, J. M., L. E. Johnson, B. Peters, and D. M. Lodge. 2007. Forecasting the expansion of zebra mussels in the United States. Conservation Biology 21:800- 810.

Bossenbroek, J. M., C. E. Kraft, and J. C. Nekola. 2001. Prediction of long-distance dispersal using gravity models: zebra mussel invasion of inland lakes. Ecological Applications 11:1778-1788.

227

Bowman, J. A., and K. E. Mantai. 1993. Submersed aquatic plant communities in western New York: 50 years of change. Journal of Aquatic Plant Management 31:81-84.

Boydstun, L. B. 2001. Ocean salmon fishery management. Pages 183-195 in R. L. Brown, editor. Fish Bulletin 179: Contributions to the biology of Central Valley salmonids, Volume 2. Scripps Institution of Oceanography Library, San Diego, CA.

Boyle, K., R. Bishop, J. Caudill, J. Charbonneau, D. Larson, M. A. Markowski, R. E. Unsworth, and R. W. Paterson. 1998. A database of sportfishing values. Industrial Economics, Incorporated, Cambridge, MA.

Boylen, C. W., L. W. Eichler, and J. D. Madsen. 1999. Loss of native aquatic plant species in a community dominated by Eurasian watermilfoil. Hydrobiologia 415:207-211.

Branstrator, D. K. 1995. Ecological interactions between Bythotrephes cederstroemi and Leptodora kindtii and the implications for species replacement in Lake Michigan. Journal of Great Lakes Research 21:670-679.

Bronte, C. R., M. P. Ebener, D. R. Schreiner, D. S. DeVault, M. M. Petzold, D. A. Jensen, C. Richards, and S. J. Lozano. 2003. Fish community change in Lake Superior, 1970-2000. Canadian Journal of Fisheries and Aquatic Sciences 60:1552.

Brown, R. W., M. Ebener, and T. Gorenflo. 1999. Great Lakes commercial fisheries: historical overview and prognosis for the future. Pages 307-354 in W. W. Taylor and C. P. Ferreri, editors. Great Lakes fisheries policy and management : a binational perspective. Michigan State University Press, East Lansing, Michigan.

Bruner, K. A., S. W. Fisher, and P. F. Landrum. 1994. The role of the zebra mussel, Dreissena polymorpha, in contaminant cycling: II. Zebra mussel contaminant accumulation from algae and suspended particles, and transfer to the benthic invertebrate, Gammarus fasciatus. Journal of Great Lakes Research 20:735-750.

Buchan, L. A. J., and D. K. Padilla. 1999. Estimating the probability of long-distance overland dispersal of invading aquatic species. Ecological Applications 9:254- 265.

Buchan, L. A. J., and D. K. Padilla. 2000. Predicting the likelihood of Eurasian watermilfoil presence in lakes, a macrophyte monitoring tool. Ecological Applications 10:1442-1455.

Bur, M. T., and D. M. Klarer. 1991. Prey selection for the exotic cladoceran Bythotrephes cederstroemi by selected Lake Erie fishes. Journal of Great Lakes Research 17:85-93.

228

Burkhardt, S., and J. T. Lehman. 1994. Prey consumption and predatory effects of an invertebrate predator (Bythotrephes, Cladocera, Cercopagidae) based on phosphorus budgets. Limnology and Oceanography 39:1007-1019.

Bushnoe, T. M., D. M. Warner, L. G. Rudstam, and E. L. Mills. 2003. Cercopagis pengoi as a new prey item for alewife (Alosa pseudoharengus) and rainbow smelt (Osmerus mordax) in Lake Ontario. Journal of Great Lakes Research 29:205-212.

Cabrera, D., J. T. Mandel, J. P. Andras, and M. L. Nydam. 2008. What is the crisis? Defining and prioritizing the world's most pressing problems. Frontiers in Ecology and the Environment 6:469-475.

Cantanhede, G., N. S. Hahn, E. A. Gubiani, and R. Fugi. 2008. Invasive molluscs in the diet of Pterodoras granulosus (Valenciennes, 1821) (Pisces, Doradidae) in the Upper Parana River floodplain, Brazil. Ecology of Freshwater Fish 17:47-53.

Caraco, N. F., and J. J. Cole. 2002. Contrasting impacts of a native and alien macrophyte on dissolved oxygen in a large river. Ecological Applications 12:1496-1509.

Carlsson, N. O. L., C. Bronmark, and L. A. Hansson. 2004. Invading herbivory: the golden apple snail alters ecosystem functioning in Asian wetlands. Ecology 85:1575-1580.

Carpenter, S. R. 1980. The decline of Myriophyllum spicatum in a eutrophic Wisconsin Lake. Canadian Journal of Botany 58:527-535.

Carpenter, S. R., B. J. Benson, R. Biggs, J. W. Chipman, J. A. Foley, S. A. Golding, R. B. Hammer, P. C. Hanson, P. T. J. Johnson, A. M. Kamarainen, T. K. Kratz, R. C. Lathrop, K. D. McMahon, B. Provencher, J. A. Rusak, C. T. Solomon, E. H. Stanley, M. G. Turner, M. J. Vander Zanden, C. H. Wu, and H. L. Yuan. 2007. Understanding regional change: A comparison of two lake districts. Bioscience 57:323-335.

Carpenter, S. R., and J. F. Kitchell. 1993. The Trophic Cascade in Lakes. Cambridge University Press, Cambridge ; New York.

Cataldo, D., D. Boltovskoy, J. Stripeikis, and M. Pose. 2001a. Condition index and growth rates of field caged Corbicula fluminea (Bivalvia) as biomarkers of pollution gradients in the Parana River delta (Argentina). Aquatic Ecosystem Health and Management 4:187-201.

Cataldo, D., J. C. Colombo, D. Boltovskoy, C. Bilos, and P. Landoni. 2001b. Environmental toxicity assessment in the Parana River delta (Argentina): simultaneous evaluation of selected pollutants and mortality rates of Corbicula fluminea (Bivalvia) early juveniles. Environmental Pollution 112:379-389.

229

Catling, P. M., G. Mitrow, E. Haber, U. Posluszny, and W. A. Charlton. 2003. The biology of Canadian weeds. 124. Hydrocharis morsus-ranae L. Canadian Journal of Plant Science 83:1001-1016.

Catling, P. M., and Z. S. Porebski. 1995. The spread and current distribution of European frogbit, Hydrocharis morsus-ranae L, in North America. Canadian Field Naturalist 109:236-241.

Charlebois, P. M., L. D. Corkum, D. J. Jude, and C. Knight. 2001. The round goby (Neogobius melanostomus) invasion: current research and future needs. Journal of Great Lakes Research 27:263-266.

Chase, J. M., and T. M. Knight. 2006. Effects of eutrophication and snails on Eurasian watermilfoil (Myriophyllum spicatum) invasion. Biological Invasions 8:1643- 1649.

Cheng, H., and O. Capps. 1988. Demand analysis of fresh and frozen finfish and shellfish in the United States. American Journal of Agricultural Economics 70:533-542.

Cherry, D. S., J. L. Scheller, N. L. Cooper, and J. R. Bidwell. 2005. Potential effects of Asian clam (Corbicula fluminea) die-offs on native freshwater mussels (Unionidae) I: water-column ammonia levels and ammonia toxicity. Journal of the North American Benthological Society 24:369-380.

Cheruvelil, K. S., P. A. Soranno, and J. D. Madsen. 2001. Epiphytic macroinvertebrates along a gradient of Eurasian watermilfoil cover. Journal of Aquatic Plant Management 39:67-72.

Cheruvelil, K. S., P. A. Soranno, J. D. Madsen, and M. J. Roberson. 2002. Plant architecture and epiphytic macroinvertebrate communities: the role of an exotic dissected macrophyte. Journal of the North American Benthological Society 21:261-277.

Chotkowski, M. A., and J. E. Marsden. 1999. Round goby and mottled sculpin predation on lake trout eggs and fry: field predictions from laboratory experiments. Journal of Great Lakes Research 25:26-35.

Cobb, S. E., and M. C. Watzin. 2002. Zebra mussel colonies and yellow perch foraging: spatial complexity, refuges, and resource enhancement. Journal of Great Lakes Research 28:256-263.

Cohen, J., N. Mirotchnick, and B. Leung. 2007. Thousands introduced annually: the aquarium pathway for non-indigenous plants to the St Lawrence Seaway. Frontiers in Ecology and the Environment 5:528-532.

Colautti, R. I., I. A. Grigorovich, and H. J. MacIsaac. 2006. Propagule pressure: a null model for biological invasions. Biological Invasions 8:1023-1037. 230

Colizza, V., A. Barrat, M. Barthelemy, and A. Vespignani. 2006. The role of the airline transportation network in the prediction and predictability of global epidemics. Proceedings of the National Academy of Sciences of the United States of America 103:2015-2020.

Collett, D. 2003. Modelling survival data in medical research. 2nd edition. Chapman & Hall/CRC, Boca Raton, Fla.

Connelly, N. A., C. R. O'Neill, B. A. Knuth, and T. L. Brown. 2007. Economic impacts of zebra mussels on drinking water treatment and electric power generation facilities. Environmental Management 40:105-112.

Connor, J. 1978. Osprey trapped by waterchestnut. Auk 95:610-611.

Cooke, R. M. 1991. Experts in uncertainty: opinion and subjective probability in science. Oxford University Press, New York, New York.

Cooke, R. M., and L. J. H. Goossens. 2000. Procedures guide for structured expert judgment. Nuclear science and technology series EUR 18820, Office for Official Publications of the European Communities, Luxembourg.

Cooke, R. M., A. M. Wilson, J. T. Tuomisto, O. Morales, M. Tainio, and J. S. Evans. 2007. A probabilistic characterization of the relationship between fine particulate matter and mortality: elicitation of European experts. Environmental Science & Technology 41:6598-6605.

Coon, T. G. 1999. Ichthyofauna of the Great Lakes basin. Pages 55-72 in W. W. Taylor and C. P. Ferreri, editors. Great Lakes fisheries policy and management : a binational perspective. Michigan State University Press, East Lansing, Michigan.

Cooper, N. L., J. R. Bidwell, and D. S. Cherry. 2005. Potential effects of Asian clam (Corbicula fluminea) die-offs on native freshwater mussels (Unionidae) II: pore- water ammonia. Journal of the North American Benthological Society 24:381- 394.

Costanza, R., R. dArge, R. deGroot, S. Farber, M. Grasso, B. Hannon, K. Limburg, S. Naeem, R. V. Oneill, J. Paruelo, R. G. Raskin, P. Sutton, and M. vandenBelt. 1997. The value of the world's ecosystem services and natural capital. Nature 387:253-260.

Costello, C., J. M. Drake, and D. M. Lodge. 2007. Evaluating an invasive species policy: ballast water exchange in the Great Lakes. Ecological Applications 17:655-662.

Coulas, R. A., H. J. Macisaac, and W. Dunlop. 1998. Selective predation on an introduced zooplankter (Bythotrephes cederstroemi) by lake herring (Coregonus artedii) in Harp Lake, Ontario. Freshwater Biology 40:343-355.

231

Countryman, W. D. 1978. Nuisance aquatic plants in Lake Champlain: Lake Champlain Basin Study. PB-293-439, U.S. Department of Commerce, National Technical Information Service, Burlington, VT.

Crocker, T. D., and J. Tschirhart. 1992. Ecosystems, externalities, and economies. Environmental and Resource Economics 2:551-567.

Crooks, J. A. 2002. Characterizing ecosystem-level consequences of biological invasions: the role of ecosystem engineers. Oikos 97:153-166.

Crowder, L. B. 1980. Alewife, rainbow smelt and native fishes in Lake Michigan: competition or predation. Environmental Biology of Fishes 5:225-233.

Crowl, T. A., T. O. Crist, R. R. Parmenter, G. Belovsky, and A. E. Lugo. 2008. The spread of invasive species and infectious disease as drivers of ecosystem change. Frontiers in Ecology and the Environment 6:238-246.

Dahl, J. A., D. M. Graham, R. Dermott, O. E. Johannsson, E. S. Millard, and D. D. Myles. 1995. Lake Erie, 1993, western, west-central, and eastern basins: changes in trophic status, and assessment of the abundance, biomass and production of the lower trophic levels. Technical Report 2070, Canadian Fisheries and Aquatic Sciences, Ottawa, Canada.

Daily, G. C. 1997. Nature's services: societal dependence on natural ecosystems. Island Press, Washington, DC.

Daily, G. C., and P. A. Matson. 2008. Ecosystem services: From theory to implementation. Proceedings of the National Academy of Sciences of the United States of America 105:9455-9456.

Daly, H. E., and J. C. Farley. 2004. Ecological economics : principles and applications. Island Press, Washington.

D'Antonio, C. M., and S. E. Hobbie. 2005. Plant species effects on ecosystem processes: insights from invasive species. Pages 65-84 in D. F. Sax, J. J. Stachowicz, and S. D. Gaines, editors. Species invasions : insights into ecology, evolution, and biogeography. Sinauer Associates, Sunderland, Mass.

Darrigran, G. 2002. Potential impact of filter-feeding invaders on temperate inland freshwater environments. Biological Invasions 4:145-156.

Davis, G. J., and M. M. Brinson. 1983. Trends in submersed communities of the Currituck Sound: 1909-1979. Journal of Aquatic Plant Management 21:83-87.

DeAngelis, D. L., L. J. Gross, M. A. Huston, W. F. Wolff, D. M. Fleming, E. J. Comiskey, and S. M. Sylvester. 1998. Landscape modeling for Everglades ecosystem restoration. Ecosystems 1:64-75.

232

Dearden, P. 1983. Anatomy of a biological hazard: Myriophyllum spicatum L. in the Okanagan Basin, British Columbia. Journal of Environmental Management 17:47-61.

De'ath, G., and K. E. Fabricius. 2000. Classification and regression trees: a powerful yet simple technique for ecological data analysis. Ecology 81:3178-3192.

Dehnen-Schmutz, K., J. Touza, C. Perrings, and M. Williamson. 2007. The horticultural trade and ornamental plant invasions in Britain. Conservation Biology 21:224- 231.

DeLucia, E. H., J. G. Hamilton, S. L. Naidu, R. B. Thomas, J. A. Andrews, A. Finzi, M. Lavine, R. Matamala, J. E. Mohan, G. R. Hendrey, and W. H. Schlesinger. 1999. Net primary production of a forest ecosystem with experimental CO2 enrichment. Science 284:1177-1179.

Deng, Y. 1996. Present and expected costs of zebra mussel damages to water users with Great Lakes water intakes. Ph.D. Dissertation. The Ohio State University, Columbus, OH.

Dermott, R., and D. Kerec. 1997. Changes to the deepwater benthos of eastern Lake Erie since the invasion of Dreissena: 1979-1993. Canadian Journal of Fisheries and Aquatic Sciences 54:922-930.

Dermott, R., and M. Munawar. 1993. Invasion of Lake Erie offshore sediments by Dreissena, and its ecological implications. Canadian Journal of Fisheries and Aquatic Sciences 50:2298-2304.

Dermott, R., J. Witt, Y. M. Um, and M. Gonzalez. 1998. Distribution of the Ponto- Caspian amphipod Echinogammarus ischnus in the Great Lakes and replacement of native Gammarus fasciatus. Journal of Great Lakes Research 24:442-452.

Desorcie, T. J., and T. A. Edsall. 1995. Feeding rate of young-of-the-year ruffe on eggs of lake whitefish. Journal of Freshwater Ecology 10:225-229.

Diamond, J. 1986. Overview: laboratory experiments, field experiments, and natural experiments. Pages 3-22 in J. Diamond and M. Cody, editors. Community Ecology. Harper and Row, New York.

Diamond, J. M. 1970. Ecological consequences of island colonization by Southwest Pacific birds: 1. Types of niche shifts. Proceedings of the National Academy of Sciences of the United States of America 67:529-535.

Diamond, J. M. 1973. Distributional ecology of New Guinea birds. Science 179:759-769.

Dibble, E. D., and S. L. Harrel. 1997. Largemouth bass diets in two aquatic plant communities. Journal of Aquatic Plant Management 35:74-78.

233

Dobiesz, N. E., D. A. McLeish, R. L. Eshenroder, J. R. Bence, L. C. Mohr, M. P. Ebener, T. F. Nalepa, A. P. Woldt, J. E. Johnson, R. L. Argyle, and J. C. Makarewicz. 2005. Ecology of the Lake Huron fish community, 1970-1999. Canadian Journal of Fisheries and Aquatic Sciences 62:1432-1451.

Dove, R., and M. Wallis. 1981. Studies on aquatic macrophytes part XXXVI. The 1981 aquatic plant quarantine project. APD Bulletin 17, Province of British Columbia Ministry of Environment, Victoria, British Columbia, Canada.

Drake, J. M., and J. M. Bossenbroek. 2004. The potential distribution of zebra mussels in the United States. Bioscience 54:931-941.

Drake, J. M., and D. M. Lodge. 2006a. Forecasting potential distributions of nonindigenous species with a genetic algorithm. Fisheries 31:9-16.

Drake, J. M., and D. M. Lodge. 2006b. Allee effects, propagule pressure and the probability of establishment: risk analysis for biological invasions. Biological Invasions 8:365-375.

Drury, K. L. S., and J. D. Rothlisberger. 2008. Offense and defense in landscape-level invasion control. Oikos 117:182-190.

Dubs, D. O. L., and L. D. Corkum. 1996. Behavioral interactions between round gobies (Neogobius melanostomus) and mottled sculpins (Cottus bairdi). Journal of Great Lakes Research 22:838-844.

Duggan, I. C., C. A. M. Rixon, and H. J. MacIsaac. 2006. Popularity and propagule pressure: Determinants of introduction and establishment of aquarium fish. Biological Invasions 8:377-382.

Dvorak, J., and E. P. Best. 1982. Macroinvertebrate communities associated with the macrophytes of Lake Vechten: structural and functional relationships. Hydrobiologia 95:115-126.

Dzialowski, A. R., J. T. Lennon, W. J. O'Brien, and V. H. Smith. 2003. Predator-induced phenotypic plasticity in the exotic cladoceran Daphnia lumholtzi. Freshwater Biology 48:1593-1602.

Effler, S. E., and C. Seigfried. 1994. Zebra mussel (Dreissena polymorpha) populations in the Seneca River, New York: impact on oxygen resources. Environmental Science & Technology 28:2216-2221.

Ehrlich, P. R., and A. H. Ehrlich. 2004. One with Nineveh : politics, consumption, and the human future. Island Press : Shearwater Books, Washington.

Ehrlich, P. R., and R. M. Pringle. 2008. Where does biodiversity go from here? A grim business-as-usual forecast and a hopeful portfolio of partial solutions. Proceedings

234

of the National Academy of Sciences of the United States of America 105:11579- 11586.

Eiswerth, M. E., S. G. Donaldson, and W. S. Johnson. 2000. Potential environmental impacts and economic damages of Eurasian watermilfoil (Myriophyllum spicatum) in western Nevada and northeastern California. Weed Technology 14:511-518.

Elsayed, E., M. Faisal, M. Thomas, G. Whelan, W. Batts, and J. Winton. 2006. Isolation of viral haemorrhagic septicaemia virus from muskellunge, Esox masquinongy (Mitchill), in Lake St Clair, Michigan, USA reveals a new sublineage of the North American genotype. Journal of Fish Diseases 29:611-619.

Elser, M. M., J. J. Elser, and S. R. Carpenter. 1986. Paul and Peter lakes: a liming experiment revisited. American Midland Naturalist 116:282-295.

Engel, S. 1995. Eurasian Watermilfoil as a fishery management tool. Fisheries 20:20-27.

Eshenroder, R. L., and K. L. Amatangelo. 2002. Reassessment of the lake trout population collapse in Lake Michigan during the 1940s. Technical Report 65, Great Lakes Fishery Commission, Ann Arbor, Michigan.

Fahnenstiel, G. L., T. B. Bridgeman, G. A. Lang, M. J. McCormick, and T. F. Nalepa. 1995. Phytoplankton productivity in Saginaw Bay, Lake Huron: effects of zebra mussel (Dreissena polymorpha) colonization. Journal of Great Lakes Research 21:465-475.

Finnoff, D., J. F. Shogren, B. Leung, and D. Lodge. 2007. Take a risk: preferring prevention over control of biological invaders. Ecological Economics 62:216-222.

Francis, J. T., S. R. Robillard, and J. E. Marsden. 1996. Yellow perch management in Lake Michigan: a multi-jurisdictional challenge. Fisheries 21:18-20.

French, and Bur. 1996. The effect of zebra mussel consumption on growth of freshwater drum in Lake Erie. Journal of Freshwater Ecology 11.

French, J. R. P., and D. J. Jude. 2001. Diets and diet overlap of nonindigenous gobies and small benthic native fishes co-inhabiting the St. Clair River, Michigan. Journal of Great Lakes Research 27:300-311.

Fullerton, A. H., G. A. Lamberti, D. M. Lodge, and M. B. Berg. 1998. Prey preferences of Eurasian ruffe and yellow perch: comparison of laboratory results with composition of Great Lakes benthos. Journal of Great Lakes Research 24:319- 328.

Gale, R. P. 1987. Resource miracles and rising expectations: a challenge to fishery managers. Fisheries 12:8-13.

235

Gangstad, E. O., and N. F. Cardarelli. 1990. The relation between aquatic weeds and public health. Pages 85-90 in A. H. Pieterse and K. J. Murphy, editors. Aquatic weeds: the ecology and management of nuisance aquatic vegetation. Oxford University Press, New York, NY.

Giddy, I. 2003. The Hudson River Water Trail guide. Hudson River Watertrail Association, New York, NY.

(GLFC) Great Lakes Fishery Commission. 1988. Economics of Great Lakes fisheries: a 1985 assessment. Technical Report 54, Great Lakes Fishery Commission, Ann Arbor, Michigan.

(GLFC) Great Lakes Fishery Commission. 2003. Fish community objectives for Lake Superior. Special Publication 03-01, Great Lakes Fishery Commission, Ann Arbor, MI.

Gonzalez, M. J., and A. Downing. 1999. Mechanisms underlying amphipod responses to zebra mussel (Dreissena polymorpha) invasion and implications for fish- amphipod interactions. Canadian Journal of Fisheries and Aquatic Sciences 56:679-685.

Goodwin, K., N. Caraco, and J. Cole. 2008. Temporal dynamics of dissolved oxygen in a floating-leaved macrophyte bed. Freshwater Biology 53:1632-1641.

Graham, D. M., W. G. Sprules, and S. J. Nepzy. 1992. Effects of zebra mussels on the diets and growth of juvenile yellow (Perca flavescens) and white perch (Morone americana) in Lake Erie. Journal of Shellfish Research 11:226-227.

Grant, E. C., K. R. Inendino, W. J. Love, D. P. Philipp, and T. L. Goldberg. 2005. Effects of practices related to catch-and-release angling on mortality and viral transmission in juvenile largemouth bass infected with largemouth bass virus. Journal of Aquatic Animal Health 17:315-322.

Grigorovich, I. A., R. I. Colautti, E. L. Mills, K. Holeck, A. G. Ballert, and H. J. Maclsaac. 2003. Ballast-mediated animal introductions in the Laurentian Great Lakes: retrospective and prospective analyses. Canadian Journal of Fisheries and Aquatic Sciences 60:740-756.

Grizzle, J. M., I. Altinok, W. A. Fraser, and R. Francis-Floyd. 2002. First isolation of largemouth bass virus. Diseases of Aquatic Organisms 50:233-235.

Grizzle, J. M., and C. J. Brunner. 2003. Review of largemouth bass virus. Fisheries 28:10-14.

Groth, A. T., L. Lovett-Doust, and J. Lovett-Doust. 1996. Population density and module demography in Trapa natans (Trapaceae), an annual clonal aquatic macrophyte. American Journal of Botany 83:1406-1415. 236

Gunderson, J. L., M. R. Klepinger, C. R. Bronte, and J. E. Marsden. 1998. Overview of the International Symposium on Eurasian ruffe (Gymnocephalus cernuus): biology, impacts, and control. Journal of Great Lakes Research 24:165-169.

Gurevitch, J., and D. K. Padilla. 2004. Are invasive species a major cause of extinctions? Trends in Ecology & Evolution 19:470-474.

Gutierrez, J. L., C. G. Jones, D. L. Strayer, and O. O. Iribarne. 2003. Mollusks as ecosystem engineers: the role of shell production in aquatic habitats. Oikos 101:79-90.

Gwathmey, J. H. 1945. Potomac River cleared of 'floating island'. Maryland Conservation 21:1-3.

Hakenkamp, C. C., and M. A. Palmer. 1999. Introduced bivalves in freshwater ecosystems: the impact of Corbicula on organic matter dynamics in a sandy stream. Oecologia 119:445-451.

Hall, R., J. L. Tank, and M. F. Dybdahl. 2003. Exotic snails dominate nitrogen and carbon cycling in a highly productive stream. Frontiers in Ecology and the Environment 1:407-411.

Halstead, J. M., J. Michaud, S. Hallas-Burt, and J. P. Gibbs. 2003. Hedonic analysis of effects of a nonnative invader (Myriophyllum heterophyllum) on New Hampshire (USA) lakefront properties. Environmental Management 32:391-398.

Hamilton, D. J., C. D. Ankney, and R. C. Bailey. 1994. Predation of zebra mussels by diving ducks: an exclosure study. Ecology 75:521-531.

Hastings, A., K. Cuddington, K. F. Davies, C. J. Dugaw, S. Elmendorf, A. Freestone, S. Harrison, M. Holland, J. Lambrinos, U. Malvadkar, B. A. Melbourne, K. Moore, C. Taylor, and D. Thomson. 2005. The spatial spread of invasions: new developments in theory and evidence. Ecology Letters 8:91-101.

Hatch, R. W., S. J. Nepszy, K. M. Muth, and C. T. Baker. 1987. Dynamics of the recovery of the western Lake Erie walleye (Stizostedion vitreum vitreum) stock. Canadian Journal of Fisheries and Aquatic Sciences 44:15-22.

Havel, J. E., and J. L. Graham. 2006. Complementary population dynamics of exotic and native Daphnia in North American reservoir communities. Archiv fur Hydrobiologie 167:245-264.

Havel, J. E., W. R. Mabee, and J. R. Jones. 1995. Invasion of the exotic cladoceran Daphnia lumholtzi into North American reservoirs. Canadian Journal of Fisheries and Aquatic Sciences 52:151-160.

237

Haynes, J. M., N. A. Tisch, C. M. Mayer, and R. S. Rhyne. 2005. Benthic macroinvertebrate communities in southwestern Lake Ontario following invasion of Dreissena and Echinogammarus: 1983 to 2000. Journal of the North American Benthological Society 24:148-167.

Higgins, S. N., E. T. Howell, R. E. Hecky, S. J. Guildford, and R. E. Smith. 2005. The wall of green: the status of Cladophora glomerata on the northern shores of Lake Erie's eastern basin, 1995-2002. Journal of Great Lakes Research 31:547-563.

Hilborn, R., T. A. Branch, B. Ernst, A. Magnussson, C. V. Minte-Vera, M. D. Scheuerell, and J. L. Valero. 2003. State of the world's fisheries. Annual Review of Environment and Resources 28:359-399.

Hoagland, P., and D. Jin. 2006. Science and economics in the management of an invasive species. Bioscience 56:931-935.

Hogan, L. S., E. Marschall, C. Folt, and R. A. Stein. 2007. How non-native species in Lake Erie influence trophic transfer of mercury and lead to top predators. Journal of Great Lakes Research 33:46-61.

Holey, M. E., R. W. Rybicki, G. W. Eck, E. H. Brown, J. E. Marsden, D. S. Lavis, M. L. Toneys, T. N. Trudeau, and R. M. Horrall. 1995. Progress toward lake trout restoration in Lake Michigan. Journal of Great Lakes Research 21:128-151.

Holland, R. E. 1993. Changes in planktonic diatoms and water transparency in Hatchery Bay, Bass Island Area, western Lake Erie since the establishment of the zebra mussel. Journal of Great Lakes Research 19:617-624.

Hosmer, D. W., and S. Lemeshow. 2000. Applied logistic regression. 2nd edition. Wiley, New York.

Houlahan, J. E., and C. S. Findlay. 2004. Effect of invasive plant species on temperate wetland plant diversity. Conservation Biology 18:1132-1138.

Howe, H. F., and J. Smallwood. 1982. Ecology of seed dispersal. Annual Review of Ecology and Systematics 13:201-228.

Hoyle, J. A., T. Schaner, J. M. Casselman, and R. Dermott. 1999. Changes in lake whitefish (Coregonus clupeaformis) stocks in eastern Lake Ontario following Dreissena mussel invasion. Great Lakes Research Review 4:5-10.

Huey, R. B., G. W. Gilchrist, and A. P. Hendry. 2005. Using invasive species to study evolution: case studies with Drosophila and salmon. Pages 139-164 in D. F. Sax, J. J. Stachowicz, and S. D. Gaines, editors. Species invasions : insights into ecology, evolution, and biogeography. Sinauer Associates, Sunderland, Mass.

238

Huffaker, C. B. 1958. Experimental studies on predation: dispersion factors and predator- prey oscillations. Hilgardia 27:343-383.

Hulme, P. E., S. Bacher, M. Kenis, S. Klotz, I. Kuhn, D. Minchin, W. Nentwig, S. Olenin, V. Panov, J. Pergl, P. Pysek, A. Roques, D. Sol, W. Solarz, and M. Vila. 2008. Grasping at the routes of biological invasions: a framework for integrating pathways into policy. Journal of Applied Ecology 45:403-414.

Hummel, M., and S. Findlay. 2006. Effects of water chestnut (Trapa natans) beds on water chemistry in the tidal freshwater Hudson River. Hydrobiologia 559:169- 181.

Hummel, M., and E. Kiviat. 2004. Review of world literature on water chestnut with implications for management in North America. Journal of Aquatic Plant Management 42:17-28.

Hushak, L. 1995. The cost of zebra mussel monitoring and control. Aquatic Nuisance Species Digest 1:5.

Industrial Economics, I. 2006. Expanded expert judgment assessment of the concentration-response relationship between PM 2.5 exposure and mortality. Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina.

Inendino, K. R., E. C. Grant, D. P. Philipp, and T. L. Goldberg. 2005. Effects of factors related to water quality and population density on the sensitivity of juvenile largemouth bass to mortality induced by viral infection. Journal of Aquatic Animal Health 17:304-314.

Janssen, J., and D. J. Jude. 2001. Recruitment Failure of Mottled Sculpin Cottus bairdi in the Calumet Harbor, Southern Lake Michigan, Induced by the Newly Introduced Round Goby Neogobius melanostomus. Journal of Great Lakes Research 27:319- 328.

Janssen, J., J. E. Marsden, C. R. Bronte, D. J. Jude, S. P. Sitar, and F. W. Goetz. 2007. Challenges to deep-water reproduction by lake trout: pertinence to restoration in Lake Michigan. Journal of Great Lakes Research 33:59-74.

Jerde, C. L., and J. M. Bossenbroek. 2009. Uncertain invasions: a biological perspective. Pages 126-150 in R. P. Keller, D. M. Lodge, M. A. Lewis, and J. F. Shogren, editors. Bioeconomics of Invasive Species: Integrating Ecology, Economics, Policy, and Management. Oxford University Press, New York.

Jerde, C. L., and M. A. Lewis. 2007. Waiting for invasions: a framework for the arrival of nonindigenous species. American Naturalist 170:1-9.

239

Jeschke, J. M., and D. L. Strayer. 2005. Invasion success of vertebrates in Europe and North America. Proceedings of the National Academy of Science USA:0501271102.

Johannsson, O. E., R. Dermott, D. M. Graham, J. A. Dahl, E. S. Millard, D. D. Myles, and J. LeBlanc. 2000. Benthic and pelagic secondary production in Lake Erie after the invasion of Dreissena spp. with implications for fish production. Journal of Great Lakes Research 26:31-54.

Johannsson, O. E., D. M. Graham, D. W. E. Einhouse, and E. L. Mills. 1999. Historical and recent changes in the Lake Erie zooplankton community and their relationship to ecosystem function. Pages 169-196 in M. Munawar, T. Edsall, and I. F. Munawar, editors. Biological Invasions in Lake Erie, Past, Present and Future in State of Lake Erie : past, present and future. Backhuys, Leiden.

Johengen, T. H., T. F. Nalepa, G. L. Fahnenstiel, and G. Goudy. 1995. Nutrient changes in Saginaw Bay, Lake Huron, after the establishment of the zebra mussel (Dreissena polymorpha). Journal of Great Lakes Research 21:449-464.

Johnson, J. L., and J. E. Havel. 2001. Competition between native and exotic Daphnia: in situ experiments. Journal of plankton research 23:373-387.

Johnson, L. E., J. M. Bossenbroek, and C. E. Kraft. 2006. Patterns and pathways in the post-establishment spread of non-indigenous aquatic species: the slowing invasion of North American inland lakes by the zebra mussel. Biological Invasions 8:475- 489.

Johnson, L. E., and D. K. Padilla. 1996. Geographic spread of exotic species: ecological lessons and opportunities from the invasion of the zebra mussel Dreissena polymorpha. Biological Conservation 78:23-33.

Johnson, L. E., A. Ricciardi, and J. T. Carlton. 2001. Overland dispersal of aquatic invasive species: a risk assessment of transient recreational boating. Ecological Applications 11:1789-1799.

Johnson, P. D., and R. F. McMahon. 1998. Effects of temperature and chronic hypoxia on survivorship of the zebra mussel (Dreissena polymorpha) and Asian clam (Corbicula fluminea). Canadian Journal of Fisheries and Aquatic Sciences 55:1564-1572.

Johnstone, I. M., B. T. Coffey, and C. Howardwilliams. 1985. The role of recreational boat traffic in interlake dispersal of macrophytes: a New Zealand case study. Journal of Environmental Management 20:263-279.

Jones, C. G., J. H. Lawton, and M. Shachak. 1997. Positive and negative effects of organisms as physical ecosystem engineers. Ecology 78:1946-1957.

240

Jude, D. J. 1997. Round gobies: cyberfish of the third millennium. Great Lakes Research Review 3:27-34.

Jude, D. J., J. Janssen, and G. Crawford. 1995. Ecology, distribution, and impact of the newly introduced round and tubenose gobies on the biota of the St. Clair and Detroit Rivers. Pages 447-460 in M. Munawar, T. Edsall, and J. H. Leach, editors. The Lake Huron ecosystem: ecology, fisheries and management. SPB Academic Publishing, Amsterdam.

Kaval, P., and J. Loomis. 2003. Updated outdoor recreation use values with emphasis on National Park recreation. U.S. Department of the Interior, National Park Service, Washington, D.C.

Keast, A. 1984. The introduced aquatic macrophyte, Myriophyllum spicatum, as habitat for fish and their invertebrate prey. Canadian Journal of Zoology 62:1289-1303.

Keller, R. P., A. N. Cox, C. Van Loon, D. M. Lodge, L.-M. Herborg, and J. Rothlisberger. 2007. From bait shops to the forest floor: earthworm use and disposal by anglers. American Midland Naturalist 158:321-328.

Keller, R. P., J. M. Drake, and D. M. Lodge. 2007. Fecundity as a basis for risk assessment of nonindigenous freshwater molluscs. Conservation Biology 21:191- 200.

Keller, R. P., and D. M. Lodge. 2007. Species invasions from commerce in live aquatic organisms: problems and possible solutions. Bioscience 57:428-436.

Keller, R. P., D. M. Lodge, and D. C. Finnoff. 2007. Risk assessment for invasive species produces net bioeconomic benefits. Proceedings of the National Academy of Sciences of the United States of America 104:203-207.

Kelly, D. W., G. A. Lamberti, and H. J. MacIsaac. 2009. The Laurentian Great Lakes as a case study of biological invasion. Pages 205-225 in R. P. Keller, D. M. Lodge, M. A. Lewis, and J. F. Shogren, editors. Bioeconomics of Invasive Species: Integrating Ecology, Economics, Policy, and Management. Oxford University Press, New York.

Kikuchi, E., and Y. Kurihara. 1982. The effects of the oligochaete Branchiura sowerbyi Beddard (Tubificidae) on the biological and chemical characteristics of overlying water and soil in a submerged ricefield soil system. Hydrobiologia 97:203-208.

Kilgore, K. J., R. P. Morgan II, and N. B. Rybicki. 1989. Seasonal and temporal distribution and abundance of fishes association with submersed aquatic plants. North American Journal of Fisheries Management 9:101-111.

Kitchell, J. F., S. P. Cox, C. J. Harvey, T. B. Johnson, D. M. Mason, K. K. Schoen, K. Aydin, C. Bronte, M. Ebener, M. Hansen, M. Hoff, S. Schram, D. Schreiner, and 241

C. J. Walters. 2000. Sustainability of the Lake Superior fish community: Interactions in a food web context. Ecosystems 3:545-560.

Kiviat, E. 1987. Water chestnut (Trapa natans). Pages 31-38 in D. J. Decker and J. W. Enck, editors. Exotic plants with identified detrimental impacts on wildlife habitats in New York State. Cornell Cooperative Extension, Cornell University, Ithaca, NY.

Kiviat, E. 1993. Under the spreading water chestnut. News from Hudsonia 9:1-6.

Knapton, R. W., and S. A. Petrie. 1999. Changes in distribution and abundance of submerged macrophytes in the Inner Bay at Long Point, Lake Erie: implications for foraging waterfowl. Journal of Great Lakes Research 25:783-798.

Knowler, D., and E. Barbier. 2005. Importing exotic plants and the risk of invasion: are market-based instruments adequate? Ecological Economics 52:341-354.

Knuth, B. A., S. Lerner, N. A. Connelly, and L. Gigliotti. 1995. Fishery and environmental managers' attitudes about and support for lake trout rehabilitation in the Great Lakes. Journal of Great Lakes Research 21:185-197.

Kolar, C. S., J. C. Boase, D. F. Clapp, and D. H. Wahl. 1997. Potential effect of invasion by an exotic zooplankter, Daphnia lumholtzi. Journal of Freshwater Ecology 12:521-530.

Kolar, C. S., A. H. Fullerton, K. M. Martin, and G. A. Lamberti. 2002. Interactions among zebra mussel shells, invertebrate prey, and Eurasian ruffe or yellow perch. Journal of Great Lakes Research 28:664-673.

Kolar, C. S., and D. M. Lodge. 2001. Progress in invasion biology: predicting invaders. Trends in Ecology & Evolution 16:199-204.

Kolar, C. S., and D. M. Lodge. 2002. Ecological predictions and risk assessment for alien fishes in North America. 298:1233-1236.

Kolar, C. S., and D. H. Wahl. 1998. Daphnid morphology deters fish predators. Oecologia 116:556-564.

Krieger, K. A., D. W. Schloesser, B. A. Manny, C. E. Trisler, S. E. Heady, J. J. H. Ciborowski, and K. M. Muth. 1996. Recovery of burrowing mayflies (Ephemeroptera: Ephemeridae: Hexagenia) in western Lake Erie. Journal of Great Lakes Research 22:254-263.

Kryger, J., and H. U. Riisgard. 1988. Filtration rate capacities in 6 species of European freshwater bivalves. Oecologia 77:34-38.

242

Kuhns, L. A., and M. B. Berg. 1999. Benthic invertebrate community responses to round goby (Neogobius melanostomus) and zebra mussel (Dreissena polymorpha) invasion in southern Lake Michigan. Journal of Great Lakes Research 25:910- 917.

Landers, D. H. 1982. Effects of naturally senescing aquatic macrophytes on nutrient chemistry and chlorophyll a of surrounding waters. Limnology and Oceanography 27:428-439.

Landrum, P. F., D. C. Gossiaux, T. F. Nalepa, and D. L. Fanslow. 2000. Evaluation of Lake Michigan sediment for causes of the disappearance of Diporeia spp. in southern Lake Michigan. Journal of Great Lakes Research 26:402-407.

Lange, L., and S. A. Wittmeyer. 1996. The contribution of zebra mussels (Dreissena spp.) feces and pseudofeces production to taste and odor episodes in the Niagara River and Lake Erie. Pages 225-243 in F. M. D'Itri, editor. Zebra Mussels and Aquatic Nuisance Species. Ann Arbor Press, Chelsea.

Lauer, T. E., P. J. Allen, and T. S. McComish. 2004. Changes in mottled sculpin and johnny darter trawl catches after the appearance of round gobies in the Indiana waters of Lake Michigan. Transactions of the American Fisheries Society 133:185-189.

Lavrentyev, P. J., W. S. Gardner, J. F. Cavaletto, and J. R. Beaver. 1995. Effects of the zebra mussel (Dreissena polymorpha Pallas) on protozoa and phytoplankton from Saginaw Bay, Lake Huron. Journal of Great Lakes Research 21:545-557.

Laxson, C. L., K. N. McPhedran, J. C. Makarewicz, I. V. Telesh, and H. J. Macisaac. 2003. Effects of the non-indigenous cladoceran Cercopagis pengoi on the lower food web of Lake Ontario. Freshwater Biology 48:2094-2106.

Lehman, J. T. 1991. Causes and consequences of cladoceran dynamics in Lake Michigan: implications of species invasion by Bythotrephes. Journal of Great Lakes Research 17:437-445.

Lehman, J. T., and C. E. Caceres. 1993. Food web responses to species invasion by a predatory invertebrate: Bythotrephes in Lake Michigan. Limnology and Oceanography 38:879-891.

Leigh, P. 1998. Benefits and costs of the Ruffe control program for the Great Lakes fishery. Journal of Great Lakes Research 24:351-360.

Lenton, T. M., H. Held, E. Kriegler, J. W. Hall, W. Lucht, S. Rahmstorf, and H. J. Schellnhuber. 2008. Tipping elements in the Earth's climate system. Proceedings of the National Academy of Sciences of the United States of America 105:1786- 1793.

243

LePage, W. L. 1993. The impact of Dreissena polymorpha on waterworks operations at Monroe, Michigan: a case history. Pages 333-358 in T. F. Nalepa and D. W. Schloesser, editors. Zebra mussels: biology, impacts, and control. Lewis Publishers, Boca Raton, Fla.

Leung, B., J. M. Bossenbroek, and D. M. Lodge. 2006. Boats, pathways, and aquatic biological invasions: estimating dispersal potential with gravity models. Biological Invasions 8:241-254.

Leung, B., J. M. Drake, and D. M. Lodge. 2004. Predicting invasions: propagule pressure and the gravity of allee effects. Ecology 85:1651-1660.

Leung, B., D. M. Lodge, D. Finnoff, J. F. Shogren, M. A. Lewis, and G. Lamberti. 2002. An ounce of prevention or a pound of cure: bioeconomic risk analysis of invasive species. Proceedings of the Royal Society of London Series B-Biological Sciences 269:2407-2413.

Levin, S. A. 1992. The problem of pattern and scale in ecology. Ecology 73:1943-1967.

Levin, S. A. 1999. Fragile dominion: complexity and the commons. Perseus Publishing, Cambridge, Massachusetts.

Levine, J. M., and C. M. D'Antonio. 2003. Forecasting biological invasions with increasing international trade. Conservation Biology 17:322-326.

Lewis, D. P., J. M. Piontkowski, R. W. Straney, J. J. Knowlton, and E. F. Neuhauser. 1997. A method for assessing the risk of zebra mussel Dreissena polymorpha infestation in industrial fire protection systems. Fire Technology 33:214-229.

Lewis, M. A., and S. Pacala. 2000. Modeling and analysis of stochastic invasion processes. Journal of Mathematical Biology 41:387-429.

Lillie, R. A. 1986. The spread of Eurasian watermilfoil Myriophyllum spicatum in Davis Lake, Sauk County, Wisconsin. Lake and Reservoir Management 2:64-68.

Lind, C. T., and G. Cottam. 1969. The submerged aquatics of University Bay: a study in eutrophication. American Midland Naturalist 81:353-369.

Linden, E., and M. Lehtiniemi. 2005. The lethal and sublethal effects of the aquatic macrophyte Myriophyllum spicatum on Baltic littoral planktivores. Limnology and Oceanography 50:405-411.

Lockwood, J. L., P. Cassey, and T. Blackburn. 2005. The role of propagule pressure in explaining species invasions. Trends in Ecology & Evolution 20:223-228.

Lodge, D. M., S. C. Blumenshine, and Y. Vadeboncoeur. 1998. Insights and application of large-scale, long-term ecological observations and experiments. Pages 202-235

244

in W. J. Resetarits and J. Bernardo, editors. Experimental ecology : issues and perspectives. Oxford University Press, New York.

Lodge, D. M., and C. Hamlin. 2006. Religion and the new ecology : environmental responsibility in a world in flux. University of Notre Dame Press, Notre Dame.

Lodge, D. M., S. Williams, H. J. MacIsaac, K. R. Hayes, B. Leung, S. H. Reichard, R. N. Mack, P. B. Moyle, M. Smith, D. A. Andow, J. T. Carlton, and A. McMichael. 2006. Biological invasions: recommendations for U.S. policy and management. Ecological Applications 16:2035-2054.

Lovell, S. J., and L. A. Drake. 2009. Tiny stowaways: analyzing the economic benefits of a US Environmental Protection Agency permit regulating ballast water discharges. Environmental Management 43:546-555.

Lovell, S. J., S. F. Stone, and L. Fernandez. 2006. The economic impacts of aquatic invasive species: a review of the literature. Agricultural and Resource Economics Review 35:195-208.

Lowe, R. L., and R. W. Pillsbury. 1995. Shifts in benthic algal community structure and function following the appearance of zebra mussels (Dreissena polymorpha) in Saginaw Bay, Lake Huron. Journal of Great Lakes Research 21:558-566.

Lozano, S. J., J. V. Scharold, and T. F. Nalepa. 2001. Recent declines in benthic macroinvertebrate densities in Lake Ontario. Canadian Journal of Fisheries and Aquatic Sciences 58:518-529.

Lumsden, H. G., and D. J. McLachlin. 1988. European frogbit, Hydrocharis morsus- ranae, in Lake Ontario marshes. Canadian Field Naturalist 102:261-263.

Lupi, F., J. P. Hoehn, and G. C. Christie. 2003. Using an economic model of recreational fishing to evaluate the benefits of sea lamprey (Petromyzon marinus) control on the St. Marys River. Journal of Great Lakes Research 29:742-754.

MacIsaac, H. J. 1996. Potential abiotic and biotic impacts of zebra mussels on the inland waters of North America. American Zoologist 36:287-299.

MacIsaac, H. J., J. V. M. Borbely, J. R. Muirhead, and P. A. Graniero. 2004. Backcasting and forecasting biological invasions of inland lakes. Ecological Applications 14:773-783.

MacIsaac, H. J., I. A. Grigorovich, J. A. Hoyle, N. D. Yan, and V. E. Panov. 1999. Invasion of Lake Ontario by the Ponto-Caspian predatory cladoceran Cercopagis pengoi. Canadian Journal of Fisheries and Aquatic Sciences 56:1-5.

Macisaac, H. J., C. J. Lonnee, and J. H. Leach. 1995. Suppression of microzooplankton by zebra mussels: importance of mussel size. Freshwater Biology 34:379-387.

245

Macisaac, H. J., W. G. Sprules, O. E. Johannsson, and J. H. Leach. 1992. Filtering impacts of larval and sessile zebra mussels (Dreissena polymorpha) in western Lake Erie. Oecologia 92:30-39.

Mack, R. N., D. Simberloff, W. M. Lonsdale, H. Evans, M. Clout, and F. A. Bazzaz. 2000. Biotic invasions: causes, epidemiology, global consequences, and control. Ecological Applications 10:689-710.

MacLennan, D. S., R. C. Haas, and M. V. Thomas. 2000. Lake St. Clair: a case history of change, 1970-1999.in Salmonid communities in oligotrophic lakes symposium. Great Lakes Fishery Commission.

Madenjian, C. P., G. L. Fahnenstiel, T. H. Johengen, T. F. Nalepa, H. A. Vanderploeg, G. W. Fleischer, P. J. Schneeberger, D. M. Benjamin, E. B. Smith, J. R. Bence, E. S. Rutherford, D. S. Lavis, D. M. Robertson, D. J. Jude, and M. P. Ebener. 2002. Dynamics of the Lake Michigan food web, 1970-2000. Canadian Journal of Fisheries and Aquatic Sciences 59:736-753.

Madenjian, C. P., D. W. Schloesser, and K. A. Krieger. 1998. Population models of burrowing mayfly recolonization in western Lake Erie. Ecological Applications 8:1206-1212.

Madsen, J. D., J. W. Sutherland, J. A. Bloomfield, L. W. Eichler, and C. W. Boylen. 1991. The decline of native vegetation under dense Eurasian watermilfoil canopies. Journal of Aquatic Plant Management 29:94-99.

Madsen, J. D., J. W. Sutherland, J. A. Bloomfield, and K. Roy. 1989. The Lake George aquatic plant survey. Lake George Park Commission Final Report., New York State Department of Environmental Conservation, Albany, New York.

Makarewicz, J. C., P. Bertram, T. Lewis, and E. H. Brown. 1995. A decade of predatory control of zooplankton species composition of Lake Michigan. Journal of Great Lakes Research 21:620-640.

Makarewicz, J. C., P. Bertram, and T. W. Lewis. 2000. Chemistry of the offshore surface waters of Lake Erie: pre- and post-Dreissena introduction (1983-1993). Journal of Great Lakes Research 26:82-93.

Marsden, J. E., and M. A. Chotkowski. 2001. Lake trout spawning on artificial reefs and the effect of zebra mussels: fatal attraction? Journal of Great Lakes Research 27:33-43.

Marsden, J. E., and S. R. Robillard. 2004. Decline of yellow perch in southwestern Lake Michigan, 1987-1997. North American Journal of Fisheries Management 24:952- 966.

246

Matisoff, G., X. S. Wang, and P. L. McCall. 1999. Biological redistribution of lake sediments by tubificid oligochaetes: Branchiura sowerbyi and Limnodrilus hoffmeisteri/Tubifex tubifex. Journal of Great Lakes Research 25:205-219.

May, R. 1999. Unanswered questions in ecology. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 354:1951-1959.

May, R. M. 1974. Biological populations with nonoverlapping generations: stable points, stable cycles, and chaos. Science 186:645-647.

Mazak, E. J., H. J. MacIsaac, M. R. Servos, and R. Hesslein. 1997. Influence of feeding habits on organochlorine contaminant accumulation in waterfowl on the Great Lakes. Ecological Applications 7:1133-1143.

McKenzie-Mohr, D. 2000. Promoting sustainable behavior: an introduction to community-based social marketing. Journal of Social Issues 56:543-554.

McMahon, R. F. 1991. Mollusca: Bivalvia. Pages 315-399 in J. H. Thorp and A. P. Covich, editors. Ecology and Classification of North American Freshwater Invertebrates. Academic Press, Inc., New York.

McMahon, R. F. 2002. Evolutionary and physiological adaptations of aquatic invasive animals: r selection versus resistance. Canadian Journal of Fisheries and Aquatic Sciences 59:1235-1244.

(MEA) Millennium Ecosystem Assessment. 2005. Ecosystems and human well-being: Synthesis. Island Press, Washington, DC.

Meng, C. 2005. Performance-based experts aggregation. M.S. thesis. Technical University of Delft, Delft, Netherlands.

Mercado-Silva, N., J. D. Olden, J. T. Maxted, T. R. Hrabik, and M. J. Vander Zanden. 2006. Forecasting the spread of invasive rainbow smelt in the Laurentian Great Lakes region of North America. Conservation Biology 20:1740-1749.

Mikol, G. F. 1985. Effects of harvesting on aquatic vegetation and juvenile fish populations in Saratoga Lake, New York. Journal of Aquatic Plant Management 23:59-63.

Millard, E. S., E. J. Fee, D. D. Myles, and J. A. Dahl. 1999. Comparison of phytoplankton photosynthesis methodology in Lakes Erie, Ontario, the Bay of Quinte and the Northwest Ontario Lake Size Series. Pages 441-468 in M. Munawar, T. Edsall, and I. F. Munawar, editors. Biological Invasions in Lake Erie, Past, Present and Future in State of Lake Erie : past, present and future. Backhuys, Leiden.

247

Miller, G. L., and M. A. Trout. 1985. Changes in the aquatic plant community following treatment with the herbicide 2,4-D in Cayuga Lake, New York. Pages 126-138 in Proceedings of the First International Symposium on the watermilfoil (Myriophyllum spicatum) and related Haloragaceae species. Aquatic Plant Management Society, Washington, D. C.

Mills, E., J. H. Leach, J. T. Carlton, and C. L. Secor. 1994. Exotic species and the integrity of the Great Lakes: lessons from the past. Bioscience 44:666-676.

Mills, E. L., J. M. Casselman, R. Dermott, J. D. Fitzsimons, G. Gal, K. T. Holeck, J. A. Hoyle, O. E. Johannsson, B. F. Lantry, J. C. Makarewicz, E. S. Millard, I. F. Munawar, M. Munawar, R. O'Gorman, R. W. Owens, L. G. Rudstam, T. Schaner, and T. J. Stewart. 2003. Lake Ontario: food web dynamics in a changing ecosystem (1970-2000). Canadian Journal of Fisheries and Aquatic Sciences 60:471-490.

Mills, E. L., R. M. Dermott, E. F. Roseman, D. Dustin, E. Mellina, D. B. Conn, and A. P. Spidle. 1993b. Colonization, ecology, and population structure of the quagga mussel (Bivalvia, Dreissenidae) in the lower Great Lakes. Canadian Journal of Fisheries and Aquatic Sciences 50:2305-2314.

Mills, E. L., J. H. Leach, J. T. Carlton, and C. L. Secor. 1993. Exotic species in the Great Lakes: a history of biotic crises and anthropogenic introductions. Journal of Great Lakes Research 19:1-54.

Moody, M. L., and D. H. Les. 2002. Evidence of hybridity in invasive watermilfoil (Myriophyllum) populations. Proceedings of the National Academy of Sciences of the United States of America 99:14867-14871.

Moody, M. L., and D. H. Les. 2007. Geographic distribution and genotypic composition of invasive hybrid watermilfoil (Myriophyllum spicatum x M-sibiricum) populations in North America. Biological Invasions 9:559-570.

Morgan, M. G., P. J. Adams, and D. W. Keith. 2006. Elicitation of expert judgments of aerosol forcing. Climatic Change 75:195-214.

Morrison, H. A., D. M. Whittle, and G. D. Haffner. 2000. The relative importance of species invasions and sediment disturbance in regulating chemical dynamics in western Lake Erie. Ecological Modelling 125:279-294.

Muirhead, J. R., A. M. Bobeldyk, J. M. Bossenbroek, K. J. Egan, and C. L. Jerde. 2009. Estimating dispersal and predicting spread of nonindigenous species. Pages 103- 125 in R. P. Keller, D. M. Lodge, M. A. Lewis, and J. F. Shogren, editors. Bioeconomics of Invasive Species: Integrating Ecology, Economics, Policy, and Management. Oxford University Press, New York.

248

Muirhead, J. R., and H. J. MacIsaac. 2005. Development of inland lakes as hubs in an invasion network. Journal of Applied Ecology 42:80-90.

Munawar, M., and I. F. Munawar. 1999. The changing phytoplankton biomass and its composition in Lake Erie: a lake wide comparative analysis. Pages 125-154 in M. Munawar, T. Edsall, and I. F. Munawar, editors. Biological Invasions in Lake Erie, Past, Present and Future in State of Lake Erie : past, present and future. Backhuys, Leiden.

Naidoo, R., A. Balmford, R. Costanza, B. Fisher, R. E. Green, B. Lehner, T. R. Malcolm, and T. H. Ricketts. 2008. Global mapping of ecosystem services and conservation priorities. Proceedings of the National Academy of Sciences of the United States of America 105:9495-9500.

Nalepa, T. F., and G. L. Fahnenstiel. 1995. Dreissena polymorpha in Saginaw Bay, Lake Huron ecosystem: overview and perspective. Journal of Great Lakes Research 21:411-416.

Nalepa, T. F., D. J. Hartson, J. Buchanan, J. F. Cavaletto, G. A. Lang, and S. J. Lozano. 2000. Spatial variation in density, mean size and physiological condition of the holarctic amphipod Diporeia spp. in Lake Michigan. Freshwater Biology 43:107- 119.

Nalepa, T. F., D. J. Hartson, D. L. Fanslow, G. A. Lang, and S. J. Lozano. 1998. Declines in benthic macroinvertebrate populations in southern Lake Michigan, 1980-1993. Canadian Journal of Fisheries and Aquatic Sciences 55:2402-2413.

Nalepa, T. F., D. J. Hartson, G. W. Gostenik, D. L. Fanslow, and G. A. Lang. 1996. Changes in the freshwater mussel community of Lake St Clair: from Unionidae to Dreissena polymorpha in eight years. Journal of Great Lakes Research 22:354- 369.

Narbonne, J. F., J. E. Djomo, D. Ribera, V. Ferrier, and P. Garrigues. 1999. Accumulation kinetics of polycyclic aromatic hydrocarbon adsorbed to sediment by the mollusc Corbicula fluminea. Ecotoxicology and Environmental Safety 42:1-8.

(NRC) National Research Council (U.S.). Committee on Estimating the Health-Risk- Reduction Benefits of Proposed Air Pollution Regulations., and National Research Council (U.S.). Board on Environmental Studies and Toxicology. 2002. Estimating the public health benefits of proposed air pollution regulations. National Academies Press, Washington, D.C.

(NRC) National Research Council (U.S.). Committee on Grand Challenges in Environmental Sciences. 2001. Grand challenges in environmental sciences. National Academey Press, Washington, D.C.

249

(NRC) National Research Council (U.S.). Committee on Ships' Ballast Operations. 1996. Stemming the tide: controlling introductions of nonindigenous species by ships' ballast water. National Academy Press, Washington, D.C.

(NRC) National Research Council (U.S.). Committee on the St. Lawrence Seaway:Options to Eliminate Introduction of Nonindigenous Species into the Great Lakes Phase 2. 2008. Great Lakes shipping, trade, and aquatic invasive species. Transportation Research Board, Washington, D.C.

Newroth, P. R. 1985. A review of Eurasian water milfoil impacts and management in British Columbia. Pages 139-153 in Proceedings of the First International Symposium on the watermilfoil (Myriophyllum spicatum) and related Haloragaceae species. Aquatic Plant Management Society, Washington, D. C.

Nicholls, K. H., and G. J. Hopkins. 1993. Recent changes in Lake Erie (North Shore) phytoplankton: cumulative impacts of phosphorus loading reductions and the zebra mussel introduction. Journal of Great Lakes Research 19:637-647.

Nicholls, K. H., S. J. Standke, and G. J. Hopkins. 1999. Effects of dreissenid mussels on nitrogen and phosphorus in north shore waters of Lake Erie. Pages 323-336 in M. Munawar, T. Edsall, and I. F. Munawar, editors. Biological Invasions in Lake Erie, Past, Present and Future in State of Lake Erie : past, present and future. Backhuys, Leiden.

Nichols, S. A., and D. R. Keeney. 1976. Nitrogen nutrition of Myriophyllum spicatum uptake and translocation of 15N by shoots and roots. Freshwater Biology 6:145- 154.

Nichols, S. A., and S. Mori. 1971. The littoral macrophyte vegetation of Lake Wingra. Transactions of the Wisconsin Academy of Sciences, Arts, and Letters 59:107- 119.

Nichols, S. J., G. Kennedy, E. Crawford, J. Allen, J. French, G. Black, M. Blouin, J. Hickey, S. Chernyak, R. Haas, and M. Thomas. 2003. Assessment of lake sturgeon (Acipenser fulvescens) spawning efforts in the lower St. Clair River, Michigan. Journal of Great Lakes Research 29:383-391.

Nieder, W. C., E. Barnaba, S. E. G. Findlay, S. Hoskins, N. Holochuck, and E. A. Blair. 2004. Distribution and abundance of submerged aquatic vegetation and Trapa natans in the Hudson River estuary. Journal of Coastal Research:150-161.

Nunes, P. A. L. D., and A. Markandya. 2008. Economic value of damage caused by marine bio-invasions: lessons from two European case studies. Ices Journal of Marine Science 65:775-780.

250

Nunes, P. A. L. D., and J. C. J. M. van den Bergh. 2004. Can people value protection against invasive marine species?: Evidence from a joint TC-CV survey in the Netherlands. Environmental & Resource Economics 28:517-532.

Ogle, D. H. 1998. A synopsis of the biology and life history of ruffe. Journal of Great Lakes Research 24:170-185.

Ogle, D. H., J. H. Selgeby, J. F. Savino, R. M. Newman, and M. G. Henry. 1996. Predation on ruffe by native fishes on the St. Louis estuary, Lake Superior, 1989- 1991. North American Journal of Fisheries Management 16:115-123.

Oglesby, R. T., and V. A. 1976. Changes in submerged plants at the south end of Cayuga Lake following Tropical Storm Agnes. Hydrobiologia 48:251-255.

O'Gorman, R., J. H. Elrod, R. W. Owens, C. P. Schneider, T. H. Eckert, and B. F. Lantry. 2000. Shifts in depth distributions of alewives, rainbow smelt, and age-2 lake trout in southern Lake Ontario following establishment of dreissenids. Transactions of the American Fisheries Society 129:1096-1106.

O'Gorman, R., and T. J. Stewart. 1999. Ascent, dominance, and decline of the alewife in the Great Lakes. Pages 489-514 in W. W. Taylor and C. P. Ferreri, editors. Great Lakes fisheries policy and management : a binational perspective. Michigan State University Press, East Lansing, Michigan.

Ojaveer, N., L. A. Kuhns, R. P. Barbiero, and M. L. Tuchman. 2001. Distribution and population characteristics of Cercopagis pengoi in Lake Ontario. Journal of Great Lakes Research 27:10-18.

O'Neill, C. R., Jr. 1996. National zebra mussel information clearinghouse infrastructure economic impact survey: 1995. Dreissena! 7:1-5, 1-12.

(OTA) U.S. Office of Technology Assessment. 1993. Harmful nonindigenous species in the United States. OTA-F-565:63-68. U.S. Government Printing Office, Washington, D.C., USA.

Palmquist, R. B. 2005. Weak complementarity, path independence, and the intuition of the Willig condition. Journal of Environmental Economics and Management 49:103-115.

Pangle, K. L., S. D. Peacor, and O. E. Johannsson. 2007. Large nonlethal effects of an invasive invertebrate predator on zooplankton population growth rate. Ecology 88:402-412.

Pardue, W. J., and D. H. Webb. 1985. A comparison of aquatic macroinvertebrates occurring in association with Eurasian watermilfoil (Myriophyllum spicatum L.) with those found in the open littoral zone. Journal of Freshwater Ecology 3:69-79.

251

Parker, S. L., L. G. Rudstam, E. L. Mills, and D. W. Einhouse. 2001. Retention of Bythotrephes spines in the stomachs of eastern Lake Erie rainbow smelt. Transactions of the American Fisheries Society 130:988-994.

Parry, M. L., O. F. Canziani, J. P. Palutikof, P. J. van der Linden, and C. E. Hanson, editors. 2007. Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK.

Pauly, D., V. Christensen, S. Guenette, T. J. Pitcher, U. R. Sumaila, C. J. Walters, R. Watson, and D. Zeller. 2002. Towards sustainability in world fisheries. Nature 418:689-695.

Perrings, C., S. Dalmazzone, and M. Williamson. 2005. The economics of biological invasions. Pages 16-35 in H. A. Mooney, R. N. Mack, J. A. McNeely, L. E. Neville, P. J. Schei, and J. K. Waage, editors. Invasive Alien Species. Island Press, Washington.

Perrings, C., M. Williamson, E. B. Barbier, D. Delfino, S. Dalmazzone, J. Shogren, P. Simmons, and A. Watkinson. 2002. Biological invasion risks and the public good: an economic perspective. Conservation Ecology 6:-.

Petrie, S. A., and R. W. Knapton. 1999. Rapid increase and subsequent decline of zebra and quagga mussels in Long Point Bay, Lake Erie: possible influence of waterfowl predation. Journal of Great Lakes Research 25:772-782.

Phaneuf, D. J., and V. K. Smith. 2005. Recreation demand models. Pages 671-751 in K.- G. Maler and J. R. Vincent, editors. Handbook of Environmental Economics. Elsevier, New York.

Phelps, H. L. 1994. The Asiatic clam (Corbicula flumineaI) invasion and system-level ecological change in the Potomac River Estuary near Washington, D.C. Estuaries 17:614-621.

Pickett, S. T., J. Kolasa, and C. G. Jones. 1994. Ecological understanding. Academic Press, San Diego.

Pimentel, D., L. Lach, R. Zuniga, and D. Morrison. 2000. Environmental and economic costs of nonindigenous species in the United States. Bioscience 50:53-65.

Pimentel, D., R. Zuniga, and D. Morrison. 2005. Update on the environmental and economic costs associated with alien-invasive species in the United States. Ecological Economics 52:273-288.

Platt, J. R. 1964. Strong inference. Science 146:347-353.

252

Post, J. R., L. Persson, E. A. Parkinson, and T. van Kooten. 2008. Angler numerical response across landscapes and the collapse of freshwater fisheries. Ecological Applications 18:1038-1049.

Pothoven, S. A., and C. P. Madenjian. 2008. Changes in consumption by alewives and lake whitefish after dreissenid mussel invasions in Lakes Michigan and Huron. North American Journal of Fisheries Management 28:308-320.

Pothoven, S. A., T. F. Nalepa, P. J. Schneeberger, and S. B. Brandt. 2001. Changes in diet and body condition of lake whitefish in southern Lake Michigan associated with changes in benthos. North American Journal of Fisheries Management 21:876-883.

Pothoven, S. A., and H. A. Vanderploeg. 2004. Diet and prey selection of alewives in Lake Michigan: Seasonal, depth, and interannual patterns. Transactions of the American Fisheries Society 133:1068-1077.

Prentki, R. T., M. S. Adams, S. R. Carpenter, A. Gasith, C. S. Smith, and P. R. Weiler. 1979. The role of submersed weedbeds in internal loading and interception of allochthonous materials in Lake Wingra, Wisconsin, USA. Archiv Hydrobiologia Supplement 57:221-250.

Press, W. H. 1992. Numerical recipes in C : the art of scientific computing. 2nd edition. Cambridge University Press, Cambridge ; New York.

Puth, L. M., and T. F. H. Allen. 2005. Potential corridors for the rusty crayfish, Orconectes rusticus, in northern Wisconsin (USA) lakes: Lessons for exotic invasions. Landscape Ecology 20:567-577.

Puth, L. M., and D. M. Post. 2005. Studying invasion: have we missed the boat? Ecology Letters 8:715-721.

Raikow, D. F. 2004. Food web interactions between larval bluegill (Lepomis macrochirus) and exotic zebra mussels (Dreissena polymorpha). Canadian Journal of Fisheries and Aquatic Sciences 61:497-504.

Ratti, C., and D. R. Barton. 2003. Decline in the diversity of benthic invertebrates in the wave-zone of eastern Lake Erie, 1974-2001. Journal of Great Lakes Research 29:608-615.

(RMRC) Recreational Marine Research Center. 2006. Recreational Boating Statistical Abstract 2005. National Marine Manufacturers Association, Chicago, IL.

Reed, T., S. J. Wielgus, A. K. Barnes, J. J. Schiefelbein, and A. L. Fettes. 2004. Refugia and local controls: benthic invertebrate dynamics in lower Green Bay, Lake Michigan following zebra mussel invasion. Journal of Great Lakes Research 30:390-396. 253

Reeders, H. H., and A. Bij de Devaate. 1990. Zebra mussels (Dreissena polymorpha): a new perspective for water quality management. Hydrobiologia 200:437-450.

Ricciardi, A. 2006. Patterns of invasion in the Laurentian Great Lakes in relation to changes in vector activity. Diversity and Distributions 12:425-433.

Ricciardi, A. 2007. Are modern biological invasions an unprecedented form of global change? Conservation Biology 21:329-336.

Ricciardi, A., F. G. Whoriskey, and J. B. Rasmussen. 1997. The role of the zebra mussel (Dreissena polymorpha) in structuring macroinvertebrate communities on hard substrata. Canadian Journal of Fisheries and Aquatic Sciences 54:2596-2608.

Riis, T., T. V. Madsen, and R. S. H. Sennels. 2009. Regeneration, colonisation and growth rates of allofragments in four common stream plants. Aquatic Botany 90:209-212.

Riley, S. 2007. Large-scale spatial-transmission models of infectious disease. Science 316:1298-1301.

Rixon, C. A. M., I. C. Duggan, N. M. N. Bergeron, A. Ricciardi, and H. J. Macisaac. 2005. Invasion risks posed by the aquarium trade and live fish markets on the Laurentian Great Lakes. Biodiversity and Conservation 14:1365-1381.

Rodriguez, J. P., T. D. Beard, Jr., J. Agard, E. M. Bennett, S. Cork, G. Cumming, D. Deane, A. P. Dobson, D. M. Lodge, M. Mutale, G. C. Nelson, G. D. Peterson, and T. Ribeiro. 2005. Interactions among ecosystem services. Pages 431-448 in S. R. Carpenter, P. L. Pingali, E. M. Bennett, and M. B. Zurek, editors. Ecosystems and human well-being: Scenarios. Island Press., Washington, D. C.

Ruiz, G. M., and J. T. Carlton. 2003. Invasive species : vectors and management strategies. Island Press, Washington, DC.

Sakai, A. K., F. W. Allendorf, J. S. Holt, D. M. Lodge, J. Molofsky, K. A. With, S. Baughman, R. J. Cabin, J. E. Cohen, N. C. Ellstrand, D. E. McCauley, P. O'Neil, I. M. Parker, J. N. Thompson, and S. G. Weller. 2001. The population biology of invasive species. Annual Review of Ecology and Systematics 32:305-332.

Sala, O. E., F. S. Chapin, J. J. Armesto, E. Berlow, J. Bloomfield, R. Dirzo, E. Huber- Sanwald, L. F. Huenneke, R. B. Jackson, A. Kinzig, R. Leemans, D. M. Lodge, H. A. Mooney, M. Oesterheld, N. L. Poff, M. T. Sykes, B. H. Walker, M. Walker, and D. H. Wall. 2000. Global biodiversity scenarios for the year 2100. Science 287:1770-1774.

Savino, J. F., and C. S. Kolar. 1996. Competition between nonindigenous ruffe and native yellow perch in laboratory studies. Transactions of the American Fisheries Society 125:562-571. 254

Sax, D. F., J. J. Stachowicz, and S. D. Gaines, editors. 2005. Species invasions : insights into ecology, evolution, and biogeography. Sinauer Associates, Sunderland, Mass.

Sazonova, L., G. Osipov, and M. Godovnikov. 1999. Intelligent system for fish stock prediction and allowable catch evaluation. Environmental Modelling & Software 14:391-399.

Schindler, D. W., E. J. Fee, and T. Ruszczynski. 1978. Phosphorus input and its consequences for phytoplankton standing crop and production in Experimental Lakes area and in similar lakes. Journal of the Fisheries Research Board of Canada 35:190-196.

Schloesser, D. W., and T. F. Nalepa. 1994. Dramatic decline of Unionid bivalves in offshore waters of western Lake Erie after infestation by the zebra mussel, Dreissena polymorpha. Canadian Journal of Fisheries and Aquatic Sciences 51:2234-2242.

Schloesser, D. W., T. F. Nalepa, and G. L. Mackie. 1996. Zebra mussel infestation of unionid bivalves (Unionidae) in North America. American Zoologist 36:300-310.

Schneider, D. W., C. D. Ellis, and K. S. Cummings. 1998. A transportation model assessment of the risk to native mussel communities from zebra mussel spread. Conservation Biology 12:788-800.

Schramm, H. L., and J. G. Davis. 2006. Survival of largemouth bass from populations infected with largemouth bass virus and subjected to simulated tournament conditions. North American Journal of Fisheries Management 26:826-832.

Selgeby, J. 1998. Predation by ruffe (Gymnocephalus cernuus) on fish eggs in Lake Superior. Journal of Great Lakes Research 24:304-308.

Sidman, C. F., and T. J. Fik. 2005. Modeling spatial patterns of recreational boaters: Vessel, behavioral, and geographic considerations. Leisure Sciences 27:175-189.

Sierszen, M. E., J. R. Keough, and C. A. Hagley. 1996. Trophic analysis of ruffe (Gymnocephalus cernuus) and white perch (Morone americana) in a Lake Superior coastal food web, using stable isotope techniques. Journal of Great Lakes Research 22:436-443.

Simberloff, D. 2003. How much information on population biology is needed to manage introduced species? Conservation Biology 17:83-92.

Simberloff, D., I. M. Parker, and P. N. Windle. 2005. Introduced species policy, management, and future research needs. Frontiers in Ecology and the Environment 3:12-20.

255

Skellam, J. G. 1951. Random dispersal in theoretical populations. Biometrika 38:196- 218.

Skogerboe, J. G., A. G. Poovey, K. D. Getsinger, and G. Kudray. 2003. Invasion of Eurasian watermilfoil in lakes of the western Upper Peninsula, Michigan. ERDC/EL TR-03-10, U.S. Army Corps of Engineers, Washington, D.C.

Skubinna, J. P., T. G. Coon, and T. R. Batterson. 1995. Increased abundance and depth of submersed macrophytes in response to decreased turbidity in Saginaw Bay, Michigan. Journal of Great Lakes Research 21:476-488.

Smith, C. S., and M. S. Adams. 1986. Phosphorus transfer from sediments by Myriophyllum spicatum. Limnology and Oceanography 31:1312-1321.

Smith, C. S., and J. W. Barko. 1990. Ecology of Eurasian watermilfoil. Journal of Aquatic Plant Management 28:55-64.

Smith, G. E., T. F. Hall, and R. A. Stanley. 1967. Eurasian Watermilfoil in Tennessee Valley. Weeds 15:95-&.

Smyth, R. L., M. C. Watzin, and R. E. Manning. 2009. Investigating public preferences for managing Lake Champlain using a choice experiment. Journal of Environmental Management 90:615-623.

Somers, C. M., M. N. Lozer, V. A. Kjoss, and J. S. Quinn. 2003. The invasive round goby (Neogobius melanostomus) in the diet of nestling double-crested cormorants (Phalacrocorax auritus) in Hamilton Harbour, Lake Ontario. Journal of Great Lakes Research 29:392-399.

Sousa, R., C. Antunes, and L. Guilhermino. 2008a. Ecology of the invasive Asian clam Corbicula fluminea (Muller, 1774) in aquatic ecosystems: an overview. Annales De Limnologie-International Journal of Limnology 44:85-94.

Sousa, R., L. Guilhermino, and C. Antones. 2005. Molluscan fauna in the freshwater tidal area of the River Minho estuary, NW of Iberian Peninsula. Annales De Limnologie-International Journal of Limnology 41:141-147.

Sousa, R., A. J. A. Nogueira, M. B. Gaspar, C. Antunes, and L. Guilhermino. 2008b. Growth and extremely high production of the non-indigenous invasive species Corbicula fluminea (Muller, 1774): Possible implications for ecosystem functioning. Estuarine Coastal and Shelf Science 80:289-295.

Spash, C. L., and A. Vatn. 2006. Transferring environmental value estimates: issues and alternatives. Ecological Economics 60:379-388.

256

Steinhart, G. B., E. A. Marschall, and R. A. Stein. 2004a. Round goby predation on smallmouth bass offspring in nests during simulated catch-and-release angling. Transactions of the American Fisheries Society 133:121-131.

Steinhart, G. B., R. A. Stein, and E. A. Marschall. 2004b. High growth rate of young-of- the-year smallmouth bass in Lake Erie: a result of the round goby invasion? Journal of Great Lakes Research 30:381-389.

Stetter, S. L. P., L. D. Witzel, L. G. Rudstam, D. W. Einhouse, and E. L. Mills. 2005. Energetic consequences of diet shifts in Lake Erie rainbow smelt (Osmerus mordax). Canadian Journal of Fisheries and Aquatic Sciences 62:145-152.

Stewart, D. J., and M. Ibarra. 1991. Predation and production by salmonine fishes in Lake Michigan, 1978-88. Canadian Journal of Fisheries and Aquatic Sciences 48:909- 922.

Stewart, T. W., and J. M. Haynes. 1994. Benthic macroinvertebrate communities of southwestern Lake Ontario following invasion of Dreissena. Journal of Great Lakes Research 20:479-493.

Stewart, T. W., J. G. Miner, and R. L. Lowe. 1998. Macroinvertebrate communities on hard substrates in western Lake Erie: structuring effects of Dreissena. Journal of Great Lakes Research 24:868-879.

Stokstad, E. 2007. Feared quagga mussel turns up in western United States. Science 315:453-453.

Strayer, D. L. 1999. Effects of alien species on freshwater mollusks in North America. Journal of the North American Benthological Society 18:74-98.

Taylor, B. W., and R. E. Irwin. 2004. Linking economic activities to the distribution of exotic plants. Proceedings of the National Academy of Sciences of the United States of America 101:17725-17730.

Taylor, J. C., and J. L. Roach. 2009. Ocean shipping in the Great Lakes: an analysis of industry transportation cost savings. Transportation Journal 48:53-67.

Thomas, R. W., and R. J. Huggett. 1980. Modelling in geography : a mathematical approach. Barnes & Noble Books, Totowa, N.J.

Thornton, K., and C. Laurin. 2005. Soft sciences and the hard reality of lake management. Lake and Reservoir Management 21:203-208.

Thorp, J. H., and A. F. Casper. 2002. Potential effects on zooplankton from species shifts in planktivorous mussels: a field experiment in the St. Lawrence River. Freshwater Biology 47:107-119.

257

Thorp, S., and J. Stone. 2000. Recreational boating and the Great Lakes–St. Lawrence Region. Great Lakes Commission, Ann Arbor, Michigan.

Toetz, D. 1997. Does Eurasian watermilfoil, Myriophyllum spicatum, contribute to the diet of animals in a turbid reservoir? Journal of Freshwater Ecology 12:545-551.

Tran, D., A. Boudou, and J. C. Masabuau. 2001. How oxygenation level influences cadmium accumulation pattern in the Asiatic clam Corbicula fluminea: a laboratory and a field study. Environmental Toxicology and Chemistry 20:2073- 2080.

Trebitz, A. S., and D. L. Taylor. 2007. Exotic and invasive aquatic plants in Great Lakes coastal wetlands: distribution and relation to watershed land use and plant richness and cover. Journal of Great Lakes Research 33:705-721.

Trent, A., D. Karsky, and S. Gilmour. 2002. MTDC portable vehicle washer: interim report. 0234-2836-MTDC, USDA Forest Service Technology and Development Program, Missoula, MT.

Trometer, E. S., and W. D. N. Busch. 1999. Changes in age-0 fish growth and abundance following the introduction of zebra mussels Dreissena polymorpha in the western basin of Lake Erie. North American Journal of Fisheries Management 19:604- 609.

Truemper, H. A., T. E. Lauer, T. S. McComish, and R. A. Edgell. 2006. Response of yellow perch diet to a changing forage base in southern Lake Michigan, 1984- 2002. Journal of Great Lakes Research 32:806-816.

Tsuchiya, T., and T. Iwakuma. 1993. Growth and leaf life-span of a floating leaved plant, Trapa natans L., as influenced by nitrogen influx. Aquatic Botany 46:317-324.

Tuomisto, J. T., A. Wilson, J. S. Evans, and M. Tainio. 2008. Uncertainty in mortality response to airborne fine particulate matter: combining European air pollution experts. Reliability Engineering & System Safety 93:732-744.

Turner, M. G., W. L. Baker, C. J. Peterson, and R. K. Peet. 1998. Factors influencing succession: lessons from large, infrequent natural disturbances. Ecosystems 1:511-523.

Underwood, A. J. 1997. Experiments in ecology : their logical design and interpretation using analysis of variance. Cambridge University Press, Cambridge [England] ; New York, NY, USA.

Unmuth, J. M. L., R. A. Lillie, D. S. Dreikosen, and D. W. Marshall. 2000. Influence of dense growth of Eurasian watermilfoil on lake water temperature and dissolved oxygen. Journal of Freshwater Ecology 15:497-503.

258

(USFWS) United States Fish and Wildlife Service. 2007. 2006 National survey of fishing, hunting, and wildlife-associated recreation. U.S. Department of the Interior, Fish and Wildlife Service, and U.S. Department of Commerce, U.S. Census Bureau, Washington, D.C.

(USGS) United States Geological Survey, Great Lakes Science Center. 2008. 2006 Commercial fishing report: total pounds and dollar value of commercial catch in U.S. waters of the Great Lakes by year, state, lake, and species. U. S. Department of the Interior, Washinton, D.C.

Valley, R. D., and M. T. Bremigan. 2002. Effects of selective removal of Eurasian watermilfoil on age-0 largemouth bass piscivory and growth in southern Michigan lakes. Journal of Aquatic Plant Management 40:79-87.

Valley, R. D., and M. T. Bremigan. 2002a. Effects of selective removal of Eurasian watermilfoil on age-0 largemouth bass piscivory and growth in southern Michigan lakes. Journal of Aquatic Plant Management 40:79-87.

Valley, R. D., and M. T. Bremigan. 2002b. Effects of macrophyte bed architecture on largemouth bass foraging: Implications of exotic macrophyte invasions. Transactions of the American Fisheries Society 131:234-244.

Valley, R. D., and R. M. Newman. 1998. Competitive interactions between Eurasian watermilfoil and northern watermilfoil in experimental tanks. Journal of Aquatic Plant Management 36:121-126.

Vander Zanden, M. J., and J. D. Olden. 2008. A management framework for preventing the secondary spread of aquatic invasive species. Canadian Journal of Fisheries and Aquatic Sciences 65:1512-1522.

Vander Zanden, M. J., J. D. Olden, J. H. Thorne, and N. E. Mandrak. 2004. Predicting occurrences and impacts of smallmouth bass introductions in north temperate lakes. 14:132-148.

Vanderploeg, H. A., J. R. Liebig, W. W. Carmichael, M. A. Agy, T. H. Johengen, G. F. Fahnenstiel, and T. F. Nalepa. 2001. Zebra mussel (Dreissena polymorpha) selective filtration promoted toxic Microcystis blooms in Saginaw Bay (Lake Huron) and Lake Erie. Canadian Journal of Fisheries and Aquatic Science 58:1208-1221.

Vanderploeg, H. A., J. R. Liebig, and M. Omair. 1993. Bythotrephes predation on Great Lakes zooplankton Measured by an in situ method: implications for zooplankton community structure. Archiv fur Hydrobiologie 127:1-8. 259

Vanderploeg, H. A., T. F. Nalepa, D. J. Jude, E. L. Mills, K. T. Holeck, J. R. Liebig, I. A. Grigorovich, and H. Ojaveer. 2002. Dispersal and emerging ecological impacts of Ponto-Caspian species in the Laurentian Great Lakes. Canadian Journal of Fisheries and Aquatic Sciences 59:1209-1228.

Vaughn, C. C., and C. C. Hakenkamp. 2001. The functional role of burrowing bivalves in freshwater ecosystems. Freshwater Biology 46:1431-1446.

Venables, W. N., and B. D. Ripley. 2002. Modern applied statistics with S. 4th edition. Springer, New York.

Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M. Melillo. 1997. Human domination of Earth's ecosystems. Science 277:494-499.

Von Holle, B., and D. Simberloff. 2005. Ecological resistance to biological invasion overwhelmed by propagule pressure. Ecology 86:3212-3218.

Wagner, K. I., J. Hauxwell, P. W. Rasmussen, F. Koshere, P. Toshner, K. Aron, D. R. Helsel, S. Toshner, S. Provost, M. Gansberg, J. Masterson, and S. Warwick. 2007. Whole-lake herbicide treatments for Eurasian watermilfoil in four Wisconsin lakes: effects on vegetation and water clarity. Lake and Reservoir Management 23:83-94.

Wang, X. S., and G. Matisoff. 1997. Solute transport in sediments by a large freshwater oligochaete, Branchiura sowerbyi. Environmental Science & Technology 31:1926-1933.

Wares, J. P., A. R. Hughes, and R. K. Grosberg. 2005. Mechanisms that drive evolutionary change: insights from species introductions and invasions. Pages 229-257 in D. F. Sax, J. J. Stachowicz, and S. D. Gaines, editors. Species invasions : insights into ecology, evolution, and biogeography. Sinauer Associates, Sunderland, Mass.

Warfe, D. M., and L. A. Barmuta. 2004. Habitat structural complexity mediates the foraging success of multiple predator species. Oecologia 141:171-178.

Weimer, M., and M. Sowinski. 1999. Diet of the round goby (Neogobius melanostomus) in Lake Erie. Dreissena 10:7-12.

Werner, S., and K. O. Rothhaupt. 2007. Effects of the invasive bivalve Corbicula fluminea on settling juveniles and other benthic taxa. Journal of the North American Benthological Society 26:673-680.

Whitledge, G. W. 2009. Otolith microchemistry and isotopic composition as potential indicators of fish movement between the Illinois River drainage and Lake Michigan. Journal of Great Lakes Research 35:101-106.

260

Whyte, R. S., and D. A. Francko. 2001. Dynamics of a pioneer population of Eurasian watermilfoil (Myriophyllum spicatum L.) in a shallow Lake Erie wetland. Journal of Aquatic Plant Management 39:136-139.

Wilcove, D. S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying threats to imperiled species in the United States. Bioscience 48:607-615.

Wile, I., G. Hitchin, and G. Beggs. 1979. Impact of mechanical harvesting on Chemung Lake. Pages 145-159 in Aquatic Plants, Lake Management, and Ecosystem Consequences of Lake Harvesting. University of Wisconsin-Madison, Madison, WI.

Williamson, M. H. 1996. Biological Invasions. 1st edition. Chapman & Hall, London ; New York.

Winne, W. T. 1950. Water chestnut: a foreign menace. Bulletin to the Schools 36:230- 234.

Wood, D. M., and R. Delmoral. 1987. Mechanisms of early primary succession in sub- alpine habitats on Mount St. Helens. Ecology 68:780-790.

Woodland, J. E., A. D. Noyes, and J. M. Grizzle. 2002. A survey to detect largemouth bass virus among fish from hatcheries in the southeastern USA. Transactions of the American Fisheries Society 131:308-311.

Work, K., and M. Gophen. 1995. The invasion of Daphnia lumholtzi (Sars) into Lake Texoma (USA). Archiv fur Hydrobiologie 133:287-303.

Yan, N. D., R. Girard, and S. Boudreau. 2002. An introduced invertebrate predator (Bythotrephes) reduces zooplankton species richness. Ecology Letters 5:481-485.

Yeager, M. M., D. S. Cherry, and R. J. Neves. 1994. Feeding and burrowing behaviours of juvenile rainbow mussels Villoa iris (Bivalvia: Unionidae). Journal of the North American Benthological Society 13:217-222.

Yokoyama, H., K. Ogawa, and H. Wakabayashi. 1991. A new collection method of actinosporeans, a probable infective stage of myxosporeans to fishes, from tubificids and experimental infection of goldfish with the actinosporean, Raabeia sp. Fish Pathology 26:133-138.

Yokoyama, H., K. Ogawa, and H. Wakabayashi. 1993. Some biological characteristics of actinosporeans from the oligochaete Branchiura sowerbyi. Diseases of Aquatic Organisms 17:223-228.

Yokoyama, H., K. Ogawa, and H. Wakabayashi. 1995. Myxobolos cultus N-Sp (Myxosporea, Myxobolidae) in the goldfish Carassius auratus transformed from

261

the actinosporean stage in the oligochaete Branchiura sowerbyi. Journal of Parasitology 81:446-451.

Yule, A. M., J. W. Austin, I. K. Barker, B. Cadieux, and R. D. Moccia. 2006a. Persistence of Clostridium botulinum neurotoxin type E in tissues from selected freshwater fish species: implications to public health. Journal of Food Protection 69:1164-1167.

Yule, A. M., I. K. Barker, J. W. Austin, and R. D. Moccia. 2006b. Toxicity of Clostridium botulinum type-E neurotoxin to Great Lakes fish: implications for avian botulism. Journal of Wildlife Diseases 42:479-493.

Yurista, P. M., and K. L. Shultz. 1995. Bioenergetic analysis of prey consumption by Bythotrephes cederstroemi in Lake Michigan. Can. J. Fish. Aquat. Sci. 52:141- 150.

Zaranko, D. T., D. G. Farara, and F. G. Thompson. 1997. Another exotic mollusc in the Laurentian Great Lakes: the New Zealand native Potamopyrgus antipodarum (Gray 1843) (Gastropoda, Hydrobiidae). Canadian Journal of Fisheries and Aquatic Sciences 54:809-814.

262