EPA/600/R-08/066F | November 2008 | www.epa.gov/ncea
United States
Environmental Protection
Agency
Predicting Future Introductions
of Nonindigenous Species to the Great Lakes
National Center for Environmental Assessment
Office of Research and Development, Washington, DC 20460
-------
EPA/600/R-08/066F
November 2008
Predicting Future Introductions of
Nonindigenous Species to the Great Lakes
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20460
-------
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
ABSTRACT
The Great Lakes of the United States have been subjected to adverse ecological and
economic impacts from nonindigenous species (NTS). Ballast water from commercial shipping
is the major means by which NTS have entered the Great Lakes. To help resource managers
assess the future arrival and spread of invasive species, 58 species were initially identified as
having a moderate or high potential to spread and cause ecological impacts to the Great Lakes.
Using a species distribution model (the Genetic Algorithm for Rule-Set Production or GARP),
areas within the Great Lakes where 14 of these 58 potential invasive species could find suitable
habitat, were identified. Based on the model and species depth tolerances, all of Lake Erie and
the shallow water areas of the other four Great Lakes are most vulnerable to invasion by the
14 modeled species. Analysis of ballast water discharge data of vessels entering the Great Lakes
via the St. Lawrence Seaway revealed that the original source of most ballast water discharges
came from Canada and Western Europe. The Great Lakes ports at greatest risk for invasion by
the 14 modeled species from ballast water discharges are Toledo, Ashtabula and Sandusky, OH;
Gary, IN; Duluth, MN; Milwaukee and Superior, WI; and Chicago, IL. Since early detection is
critical in managing for NTS, these results should help focus monitoring activities on particular
species at the most vulnerable Great Lakes ports. This assessment demonstrates that successful
invasions are best predicted by knowing the propagule pressure (i.e., the number of
larvae/individuals entering a new area) and habitat matching (i.e., how similar is the invaded area
to the native range of the species).
Preferred citation:
U.S. EPA (Environmental Protection Agency). (2008) Predicting future introductions of nonindigenous species to
the Great Lakes. National Center for Environmental Assessment, Washington, DC; EPA/600/R-08/066F. Available
from the National Technical Information Service, Springfield, VA, and http://www.epa.gov/ncea.
11
-------
CONTENTS
DISCLAIMER ii
ABSTRACT ii
LIST OF TABLES v
LIST OF FIGURES vi
LIST OF ABBREVIATIONS AND ACRONYMS viii
AUTHORS, CONTRIBUTORS, AND REVIEWERS ix
1. EXECUTIVE SUMMARY 1
2. INTRODUCTION—NONINDIGENOUS SPECIES POSE A THREAT TO LAKE
ECOSYSTEMS 4
2.1. NONINDIGENOUS SPECIES AND THE GREAT LAKES 5
2.1.1. Origin and Patterns of Species Invasions 6
2.1.2. Ballast Water and NIS 7
2.1.3. Measures to Control the Release of Ballast Water Containing NIS 7
2.1.4. NOBOB Vessels and Species Invasions 9
2.1.5. Other Options for Controlling Species Invasions From Ballast Water 10
2.2. CONCEPTUAL FRAMEWORK 11
2.2.1. Propagule Pressure 11
2.2.2. Habitat Suitability 12
2.2.2.1. Species Distribution Modeling 12
2.2.2.2. Genetic Algorithms for Rule-Set Production (GARP) 13
2.2.2.3. Modification of GARP for Aquatic Systems 14
3. METHODS 15
3.1. HABITAT SUITABILITY USING THE GARP MODEL 15
3.1.1. Selection of Modeled Species 15
3.1.2. Model Inputs and Environmental Data Layers 16
3.1.2.1. Environmental Data Layers 16
3.1.2.2. Environmental Data Sources 18
3.1.3. Use of GARP model 19
3.1.4. Assumptions and Limitations 20
3.1.4.1. Data Errors 20
3.1.4.2. Biological Errors 22
3.1.5. Testing the GARP Model Performance 22
3.1.6. Determining GARP's Power to Predict 23
3.2. DETERMINING PROPAGULE PRESSURE USING BALLAST WATER
DISCHARGE DAT A AND VESSEL TRAFFIC PATTERNS 24
3.2.1. Analysis of Ballast Water Discharge Data 24
3.2.2. Assumptions and Uncertainty 26
in
-------
CONTENTS (continued)
4. RESULTS 27
4.1. COMPARISON OF THE GREAT LAKES TO THE PONTO-CASPIAN SEA 27
4.1.1. Temperature 27
4.1.2. Chlorophyll a Concentrations 27
4.1.3. Diffuse Attenuation Coefficient (K490) 30
4.1.4. Normalized Water-Leaving Radiance 30
4.2. HABITAT SUITABILITY FOR MODELED SPECIES 30
4.2.1. Blueback Herring 33
4.2.2. Quagga Mussel 35
4.2.3. Round Goby 35
4.2.4. Fishhook Waterflea 37
4.2.5. Zebra Mussel 37
4.2.6. Ruffe 39
4.2.7. Monkey Goby 40
4.2.8. New Zealand Mud Snail 40
4.2.9. Tubenose Goby 42
4.2.10. Rudd 42
4.2.11. Corophium Curvispinum (an Amphipod) 44
4.2.12. Sand Goby 45
4.2.13. Roach 46
4.2.14. Tench 46
4.3. VESSEL TRAFFIC AND GREAT LAKES PORTS 47
4.3.1. Analysis of Vessels With Ballast on Board (BOB) 47
4.3.2. Analysis of Vessels with No Ballast on Board (NOBOB) 51
5. DISCUSSION 54
5.1. PREDICTING THE SPREAD OF SPECIES 54
5.2. POTENTIAL MONITORING SITES BASED ON VESSEL TRAFFIC 55
6. REFERENCES 59
APPENDIX A. LIST OF NONINDIGENOUS SPECIES THAT HAVE BEEN
REPORTED AS OCCURRING IN THE GREAT LAKES A-l
APPENDIX B. NONINDIGENOUS SPECIES THAT MAY SPREAD TO THE
GREAT LAKES B-l
APPENDIX C. GARP MODEL VALIDATION C-l
APPENDIX D. GARP POWER OF PREDICTION ANALYSIS D-l
APPENDIX E. TABLES DEPICTING SOURCES AND DESTINATION OF
BALLAST WATER DISCHARGES IN U.S. GREAT LAKES E-l
IV
-------
LIST OF TABLES
1. Fourteen species modeled using GARP and the source of occurrence data 17
2. Environmental variables used to predict locations that would provide suitable
habitat for the 14 modeled species in the Great Lakes 18
3. Composite results for 14 species modeled using GARP 56
-------
LIST OF FIGURES
1. The five Great Lakes, some of the Great Lakes ports, and surrounding region.
2. Sources of Great Lakes species invasions from 1960-2006 (based on data provided
in Appendix A, which are derived fromNOAA, 2007a)
3. Conceptual framework for predicting future introductions of nonindigenous species
into the Great Lakes of the United States 12
4. Maximum monthly mean temperature (MMT) (°C) as determined by AVHRR sensor
(1985-2001) 28
5. Average chlorophyll a concentration (mg/m3) as determined by MODIS
(2001-2005) 29
6. Average diffuse attenuation coefficient (m"1) at 490 nm as determined by MODIS
(2001-2005) 31
7. Average normalized water leaving radiance mW/(cm2 |im sr) as determined by MODIS
(2001-2005) 32
8. GARP-predicted habitat suitability of blueback herring (Alosa aestivalis) in the
Great Lakes 34
9. GARP-predicted habitat suitability of quagga mussel (Dreissena bugemis) in the
Great Lakes 36
10. GARP-predicted habitat suitability of round goby (Neogobius melanstromus) in the
Great Lakes 36
11. GARP-predicted habitat suitability of fishhook waterflea (Cercopagis pengof) in the
Great Lakes 38
12. GARP-predicted habitat suitability of zebra mussel (Dreissena polymorpha) in the
Great Lakes 38
13. GARP-predicted habitat suitability of ruffe (Gymnocephalus cernuus) in the Great
Lakes 40
14. GARP-predicted habitat suitability of monkey goby (Neogobius fluviatilis) in the
Great Lakes 41
15. GARP predicted habitat suitability of New Zealand mud snail (Potamopyrgus
antipodarum) in the Great Lakes 41
16. GARP-predicted habitat suitability of tubenose goby (Proterorhinus marmoratus)
in the Great Lakes 43
VI
-------
LIST OF FIGURES (continued)
17. GARP-predicted habitat suitability of rudd (Scardinius erythrophthalmus) in the
Great Lakes 43
18. GARP-predicted habitat suitability ofCorophium curvispinum (no common name
reported) in the Great Lakes 44
19. GARP-predicted habitat suitability of sand goby (Potamoschistus minutus) in the
Great Lakes 45
20. GARP-predicted habitat suitability of roach (Rutilus rutilus) in the Great Lakes 46
21. GARP-predicted habitat suitability of tench (Tinea tinea) in the Great Lakes 47
22. Location of the original source of ballast water taken-on prior to ballast water
exchange and discharge in the U.S. Great Lakes 49
23. Location of the source of ballast water taken-on from Canadian ports in or near
the Gulf of St. Lawrence prior to ballast water exchange and discharges in the
U.S. Great Lakes 49
24. Location of the source of ballast water taken-on from European ports prior to
ballast water exchange and discharges in the U.S. Great Lakes 50
25. Frequency, volume, and original source of ballast water (prior to ballast water
exchange) discharged into U.S. Great Lakes ports, from sources outside the
Great Lakes 50
26. Frequency and volume of ballast water discharges (after ballast water exchange
at sea) from ballast on board vessels, when the original source of ballast water
came from outside the Great Lakes 51
27. Possible source locations of residual materials discharged from vessels that
entered the St. Lawrence Seaway with no ballast on board, based on last five
ports of call in 2006 52
28. Frequency and volume of discharges in 2006 from NOBOB-RM vessels 53
vn
-------
LIST OF ABBREVIATIONS AND ACRONYMS
AVHRR Advanced Very High Resolution Radiometer
BOB ballast on board
BWD ballast water discharge(s)
BWE ballast water exchange
CFR Code of Federal Regulations
EPA Environmental Protection Agency
GARP Genetic Algorithm for Rule-Set Production
GBIF Global Biodiversity Information Facility
GIS geographic information systems
K490 diffuse attenuation coefficient at 490 nm
MMT mean monthly water surface temperature
MODIS Moderate Resolution Imaging Spectroradiometer
NBIC National Ballast Information Clearinghouse
NIS nonindigenous species
NOAA National Oceanic and Atmospheric Administration
NOBOB no ballast on board
NOBOB-RM no ballast on board but the vessel contains residual material
NVMC National Vessel Movement Center
nLW normalized water-leaving radiance
ppt parts per thousand
sr steradian (units of a solid angle and can also be called a squared radian)
USCG United States Coast Guard
USGLP U.S. Great Lakes ports (or ports of call)
Vlll
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
The National Center for Environmental Assessment within the Office of Research and
Development prepared this document. TN & Associates, Oak Ridge, TN, under contract
numbers 68-C-04-004 and EP-D-06-072, performed the modeling and analysis, most of the
literature review, and helped prepare this report.
Contract Officer Representative:
Vic Serveiss, U.S. EPA, National Center for Environmental Assessment, Washington, DC
Authors:
Vic Serveiss, U.S. EPA, National Center for Environmental Assessment, Washington, DC
Don Catanzaro, TN & Associates, Inc., Oak Ridge, TN
Matt Fitzpatrick, TN & Associates, Inc., Oak Ridge, TN
William Hargrove, U.S. Forest Service, Asheville, NC
Art Stewart, TN & Associates, Inc., Oak Ridge, TN
David Eskew, TN & Associates, Inc., Oak Ridge, TN
Contributors of Data and Technical Input:
Jean Mayo, Jennifer Hendricks, Robert V. O'Neill, TN & Associates, Inc., Oak Ridge, TN
David Reid, National Oceanic and Atmospheric Administration, Ann Arbor, MI
Paul Bunje, American Association for the Advancement of Science, Washington, DC
Betsy vonHolle, University of Central Florida, Orlando, FL
Sarah Bailey, Fisheries and Oceans Canada, Burlington, Ontario, Canada
Barry Burns, Michigan Department of Environmental Quality, Lansing MI,
Formatting and Editing Contributors:
Danelle Haake, TN & Associates, Inc., Oak Ridge, TN
Cris Broyles, Intellitech Systems, Inc., Cincinnati, OH
U.S. EPA Reviewers:
Henry Lee, National Health and Environmental Effects Research Laboratory, Newport, OR
Britta Bierwagen, National Center for Environmental Assessment, Washington, DC
Marilyn Katz, Office of Water, Washington, DC
Jeffrey B. Frithsen, National Center for Environmental Assessment, Washington, DC
Michael Slimak, National Center for Environmental Assessment, Washington, DC
Elizabeth Murphy, Great Lakes National Program Office, Chicago, IL
Marc Tuchmann, Great Lakes National Program Office, Chicago, IL
External Reviewers:
Becky Cudmore, Fisheries and Oceans Canada, Burlington, Ontario, Canada
John M. Drake, Odum School of Ecology, Athens, GA
Edward L. Mills, Cornell University Biological Field Station, Bridgeport, NY
IX
-------
Acknowledgements:
The authors thank James Andreasen and Dan Kluza for the idea of applying the Genetic
Algorithm for Rule-Set Production (GARP) model, originally developed for predicting
movements of endangered species in terrestrial environments, to aquatic systems; Michael
Slimak for the suggestion to apply GARP to the Great Lakes; and Anne Sergeant who served as
Project Officer from 2005-2006 and provided technical direction for the literature review and
initial species distribution models. The authors also thank Richard Everett, Stuart White, Bivan
Patnaik, and Keith Donohue of the U. S. Coast Guard (USCG) for explaining USCG regulations,
helping us interpret the National Ballast Information Clearinghouse database, and enabling their
contractor, Whitman Miller of Smithsonian Environmental Research Center, to download and
further explain the database.
-------
1. EXECUTIVE SUMMARY
Nonindigenous species (NTS) are organisms that enter an ecosystem beyond their native
spatial range. The Sea Lamprey (Petromyzon marinus) was the first to enter the Great Lakes
during the 1830s facilitated by the Erie and Welland canals. Since then, at least 185 other
species have invaded the Great Lakes. Thirteen of these species have been labeled as invasive
by causing ecological or economic harm. The zebra mussel (Dreissenapolymorpha)., for
example, has impacted many Great Lakes native species and has imposed large expenses on the
utility industry by clogging water intake pipes.
The objective of this report is to develop data and tools that U.S. Great Lakes resource
managers can use to more effectively prevent the establishment of aquatic NTS. This study maps
the habitats of the Great Lakes most vulnerable to the entry of aquatic NTS and identifies
particular NTS with the potential to enter U.S. Great Lakes ports (USGLP).
Since the St. Lawrence Seaway opened in 1959, ballast water released from transoceanic
vessels during commercial shipping operations has been identified as the predominant pathway
for NTS to enter the Great Lakes. Transport of NTS occurs when a vessel takes-on ballast water
containing NTS from outside the Seaway, the species survives in a ballast tank during transit, and
is released when the ballast water is discharged into the Great Lakes. To become established in
the new environment, the organisms must be able to survive, reproduce, and spread. To predict
future invasions of NTS in the Great Lakes, the two most important determinants of successful
invasions were evaluated: whether there is suitable habitat in the Great Lakes for nonnative
species and whether there are a sufficient number of these organisms and their larvae arriving in
the Great Lakes. First, a species distribution model was used to identify the areas of the Great
Lakes which could provide suitable habitat for NTS of concern. Second, commercial shipping
and ballast water discharge data were used to evaluate if there are a sufficient number of these
organisms entering the Great Lakes to become established.
Based on a literature review of NTS life-history characteristics and invasion histories,
58 species that pose high or medium risk for becoming established in the Great Lakes and for
causing ecological harm were identified. To predict the possible distributions of each of these
species within the Great Lakes, spatial data sets that characterize aquatic conditions on a global
scale were analyzed. These data sets were derived from remotely sensed space-based platforms,
operated and made available by the National Aeronautics and Space Administration and the
National Oceanic and Atmospheric Administration. Six of these data sets, each at a 4.6 km
9
(21 km ) spatial resolution, were found to be useful for NTS modeling. Three of the
environmental variables are direct measures of water temperature and the other three indirectly
relate to primary productivity and water clarity, indicators of habitat suitability.
1
-------
The Genetic Algorithm for Rule-Set Production (GARP) model was used to determine
habitat suitability. GARP predicts the potential distribution of species by comparing the
environmental conditions of locations currently inhabited by the species (the reference area) with
the environmental conditions in the region of concern. Adequate spatial distribution data were
available to model only 9 of the 58 potential invasive species because the GARP model requires
at least 30 spatially unique latitude and longitude points that describe the distribution of a
particular species. In addition to these nine species, GARP was also used to predict regions
within the Great Lakes that would provide suitable habitat for five species of concern that were
selected by the U.S. Environmental Protection Agency's Great Lakes National Program Office.
Since the existing location of these five species was already known, the model was validated by
comparing the reported locations of three of these species with the predicted locations. Results
from GARP modeling were used to produce 14 range maps, one for each of the modeled species,
predicting their locations of suitable habitat within the Great Lakes. The overall results varied
with each modeled species, but generally showed that all of Lake Erie and the shallower portions
of the other Great Lakes appear to be most vulnerable for invasion by the 14 modeled species.
Water depth appears to be the predominant factor limiting the potential spread of many of the
modeled species. Yet, at least one species, the quagga mussel (Dreissena bugensis\ is surviving
at greater depths in the Great Lakes than in its native habitat.
Releases of ballast water into USGLP were analyzed using 2006-2007 data obtained
from the National Ballast Information Clearinghouse. The ports that received the most ballast
water discharges from vessels entering the Seaway with ballast on board (BOB), after ballast
water exchange outside the Seaway, are Duluth, MN; Toledo, OH; Superior, Green Bay, and
Milwaukee, WI; and Gary, IN. The most frequent original sources of ballast water came from
Antwerpen, Belgium; Puerto Cabello, Venezuela; Haraholmen, Sweden; and Bremen, Germany.
It is important to note that there were no clear relationships between foreign and USGLP relative
to ballast water uptake and releases. For instance, 13 vessels that discharged ballast water in
Toledo obtained ballast water from 12 different foreign ports.
Some vessels enter the St. Lawrence Seaway without ballast water, but may still contain
residual water or sediment containing NIS in their ballast tanks, and are referred to as no ballast
on board vessels containing residual material (NOBOB-RM). After entering the Seaway, these
vessels can off-load cargo and pick up ballast water which would mix with the residual material
and be subsequently released into Great Lakes ports. There were considerably more discharges
into USGLP from NOBOB-RM vessels than from those vessels with ballast on-board. Those
ports receiving the most ballast water from NOBOB-RM vessels are Toledo, Ashtabula, and
Sandusky, OH; Superior, WI; and Duluth, MN. Assuming the observed results for 2006 and
2007 are representative of discharge and shipping patterns over the past several years, the port of
-------
greatest concern for receiving sufficient propagules and providing the most suitable habitat is
Toledo, OH. Toledo is located on Lake Erie, a region that the GARP model predicted would
have a high chance of providing suitable habitat for the modeled species. Other ports of concern
for receiving sufficient propagules and offering suitable habitat are Gary, IN; Ashtabula and
Sandusky, OH; Milwaukee, WI; and Chicago, IL. Duluth, MN and Superior, WI, with high
transport potential but low habitat suitability, could be a source of interlake transport of NTS.
This study involved numerous assumptions resulting in uncertain findings. A major
source of uncertainty for the GARP modeling is the lack of complete occurrence data for many
of the modeled species, for many parts of the globe. Another source of uncertainty is due to the
lack of an ideal suite of data for characterizing aquatic environments. Data on abiotic factors
such as salinity, bathymetry, substrate, pH, and nutrient levels were not available globally at the
21 km2 scale. The lack of species-specific data on significant biotic factors, such as competition
and predation, also lead to uncertainty in the modeling results. Despite these limitations, a model
validation exercise confirmed that GARP and the environmental variables used could produce
useful predictions of potential NTS distributions. These predictions were validated using
occurrence data from other regions to develop models that predicted known occurrences of three
NTS already widespread in the Great Lakes.
There were also limitations with the vessel traffic and ballast water discharge analyses,
used as a surrogate for measuring propagule pressure. First, the analysis was only based on 2
years of data, 2006-2007. A second source of uncertainty is due to the self-reporting nature of
data entered into the National Ballast Information Clearinghouse. Self-reporting by vessels is
not guaranteed to be accurate or complete records of actual vessel practices and should be used
with caution. The analysis of discharges from NOBOB-RM vessels is also uncertain because the
source of the residual material cannot be known for certain and could even be from ports within
the Great Lakes. Further, this data set only includes information on the last five ports of call and
species could remain in ballast tanks from visits to previous ports.
Both Canada and the United States implemented ballast water exchange procedures in
1989 and 1993, respectively. Although new NTS continue to be detected in the Great Lakes, it is
possible the NTS were transported prior to 1993 and took several years to detect. Despite these
procedures and subsequent regulations, it is likely that nonindigenous species will continue to
arrive in the Great Lakes.
These findings support the need for detection and monitoring efforts at those ports
believed to be at greatest risk. This study also demonstrates the importance of understanding
invasion biology by evaluating the two most important predictors of invasion: propagule
pressure and suitable habitat. Further, this may be the first time that remote sensing data were
used in conjunction with GARP to predict the spread of aquatic invasive species.
-------
2. INTRODUCTION—NONINDIGENOUS SPECIES POSE A THREAT TO
LAKE ECOSYSTEMS
The U.S. Great Lakes have suffered ecological damage and economic costs from a
number of aquatic nonindigenous species (NTS) that have successfully invaded this region (Mills
et al., 1994; NOAA, 2007a). NTS that enter an ecosystem beyond their native spatial range are
expected to continue to enter the Great Lakes (Ricciardi, 2006). Preventing the transport of NTS
to the region is the best way to avoid their potential adverse impacts, but if this is not possible,
the next best alternative is to monitor for their arrival and control their spread. Resource
managers are most concerned with NTS that may become invasive. Invasive species are
nonindigenous species that are likely to cause economic or environmental harm or harm to
animal or human health losses, ecological impacts, or adversely affect human health (National
Invasive Species Council, 2007). Our primary goal is to help scientists and managers to better
focus aquatic NIS monitoring activities and resources by identifying new invasive species, their
potential to spread, and the U.S. Great Lakes ports (USGLP) most susceptible to invasion.
Another goal is to demonstrate the use of a habitat suitability model and ballast water discharge
data to predict invasion potential. Clients for this report include the U.S. Environmental
Protection Agency's (EPA's) Great Lakes National Program Office, Great Lakes port officials,
the U.S. Coast Guard (USCG), environmental organizations, agencies in the United States and
Canada concerned about invasive species, and invasion biologists.
Our findings are intended to improve detection and monitoring programs by providing
managers with an approach for (1) identifying newly established populations of invasive species,
(2) tracking or detecting spatial-range expansions, and (3) estimating potential impacts of
introductions or spread by gathering baseline data on pre-existing populations and habitat.
Nonindigenous species are one of the greatest threats to the world's ecosystems (Elton,
1958), and represent the greatest threat to biodiversity in lakes worldwide (Sala et al., 2000).
Nonindigenous invasive species are the second most important threat to threatened and
endangered species in the United States, after habitat loss or alteration (Wilcove et al., 1998). To
date, about 50,000 species have been introduced into the United States (Pimentel et al., 2000).
While many beneficial food crops, such as corn, wheat, and rice are included in this number,
about 4,500 introduced species are free-ranging and up to one-fifth of these are invasive (U.S.
Congress, 1993) and cause economic losses, ecological impacts, or adversely affect human
health. The economic cost of invasive species to the United States has been estimated at $97
billion (U.S. Congress, 1993) and $137 billion (Pimentel et al., 2000) annually. Crop weeds and
crop plant pathogens are the most costly ($26 and $21 billion, respectively) followed by rats and
-------
cats ($19 and $17 billion, respectively). Pimentel (2005) estimated the total environmental and
economic impact (damage and control costs) of biological invaders to the Great Lakes Basin at
$5.7 billion per year.
Biological invasion occurs when an organisms arrives somewhere beyond its previous
range. Currently, most invasions are a result of human actions, deliberate or accidental.
Fortunately, most invaders do not become pests, or reach invasive levels, but predicting those
that do is difficult, at best. Invasions and introductions have long fascinated biologists from a
theoretical perspective. As the economic consequences of invaders has increased, however, this
fascination must now be used to develop tools that will allow one to predict future invaders,
especially those that may affect whole ecosystems, such as the Great Lakes.
2.1. NONINDIGENOUS SPECIES AND THE GREAT LAKES
The Great Lakes have been subjected to biological invasions since the 1830s, when the
sea lamprey (Petromyzon marinus) became the first recorded species to enter the Great Lakes
from the Atlantic Ocean (Mills et al. 1993). Ricciardi (2006) reports that 182 NTS are now
established in the Great Lakes and the National Oceanic and Atmospheric Administration
(NOAA) reports a similar number of 185 (NOAA, 2007a; Appendix A). While any NTS may
cause alterations to ecosystem structure or function, 13 of the reported NTS have become
invasive (Mills et al., 1994). The zebra mussel (Dreissenapolymorpha), illustrates the impact of
an invasive NTS. The zebra mussel is out-competing Diporeia, a deep-water macroinvertebrate,
for food (IAGLR, 2002). Diporeia is a key source of food for many Great Lakes fish and has
been a dominant benthic organism since the Great Lakes were formed (IAGRLR, 2002). The
loss of Diporeia from the Great Lakes system affects the structure and function of the food web
and commercially important fish such as the lake whitefish (IAGLR, 2002). Zebra mussels also
appear to be responsible for more frequent occurrences of toxic algal blooms (Microcystis) by
selectively rejecting blue-green algae as food and removing competing algae (Vanderploeg et al.,
2001). From an economic standpoint, dense populations of zebra mussels have clogged water
intake pipes, imposing large costs on utilities.
The St. Lawrence Seaway, which opened in 1959, is a system of canals and locks that
permit ocean-going vessels (as large as 225.6 m long, 23.8 m wide, and 7.9 m deep), to travel
from the Atlantic Ocean to the Great Lakes. While shipping pathways to the Great Lakes existed
prior to 1959, the opening of the Seaway and technological changes in commercial shipping
drastically increased international trade. The opening of the Seaway resulted in an increase in
the number of ships entering the Great Lakes (Sala et al., 2000; Maclsaac et al., 2001; Duggan et
al., 2003), larger ships conveying larger volumes of ballast water, and ships that have plied the
-------
waters in many geographic locations distant from the Great Lakes (Grigorovich et al., 2003a;
Holeck et al., 2004; Drake et al., 2005; Duggan et al., 2005). While the transportation of goods
has been economically beneficial, the unintended side effect of increased international trade has
resulted in the long-range transport of NTS to the Great Lakes.
2.1.1. Origin and Patterns of Species Invasions
Most NTS that have become established in the Great Lakes since 1985 are native to the
Ponto-Caspian region or the Black, Azov, and Caspian Seas (Ricciardi and Maclsaac, 2000;
Appendix A). The Baltic Sea has also served as the source of many invaders in part because it
has a climate very similar to the Great Lakes (Leppakoski et al., 2002) The NTS from these seas
include a diverse array of taxa including fish: the round goby (Neogobius melanostomus), the
tubenose goby (Proterorhinus marmoratus)., the rudd (Scardinius erythrophthalmus), the zebra
mussel, the quagga mussel (Dreissena bugensis), and several cladocerans (e.g., the fishhook
water flea, Cercopagispengoi), amphipods (e.g., Echinogammarus ischmis), and harpacticoid
copepods (e.g., Nitocra incerta; Maclssaac et al., 2001; NOAA, 2007a). The success of
Ponto-Caspian species may be related to their ability to survive ballast water exchange due to a
broader salinity tolerance developed through a geological history that includes fluctuating water
levels and salinities (Dumont, 1998).
There is no direct (i.e., nonstop) shipping traffic between the Great Lakes and the
Ponto-Caspian Sea (Colautti et al., 2003), implying that the dominance of the Ponto-Caspian
region as a source of invaders might be due to indirect linkages. To account for invasions where
no direct shipping connections exist between the occupied spatial range and the range that may
be invaded, the potential natural and human transport patterns need to be considered. NTS can be
transported from the Ponto-Caspian region to the Great Lakes via an intermediate step in
Western Europe. In addition to direct invasion pathways from the Ponto-Caspian region to
Western Europe, Maclsaac et al. (2001) proposed four indirect pathways along the major rivers:
the Danube and Rhine River pathway; (2) the Dnieper, Pripiat, Nemuna, and Vistula River
pathway; (3) the Volga River system pathway; and (4) the Don and Volga River pathway. Many
of these connections are completed through man-made canals and waterways which have
allowed considerable exchange of species between water bodies (Reid and Orlova, 2002). To
fully understand past indirect linkages (which, in turn, might help predict future indirect
linkages) it would be necessary to have complete shipping data both from the Great Lakes to the
intermediate port and from the intermediate port to the Ponto-Caspian region.
The natural construction of the Great Lakes, whereby water flows and boat traffic moves
from one lake into another, facilitates natural and human-induced dispersal within and between
-------
the lakes (Duggan et al., 2003). These dispersal patterns are likely to hasten the spread of a NTS
once it has entered the Great Lakes but are unlikely to add new species.
2.1.2. Ballast Water and NIS
Cargo vessels frequently take on ballast water to maintain stability when traveling from
port to port and especially when crossing an open sea. Some or all of the ballast water is later
released when cargo is loaded at various ports and, with regards to this study, those Great Lakes
ports shown in Figure 1.
Ballast water is the largest source of NIS to the Great Lakes as shown in Figure 2.
Additional sources of NIS to the Great Lakes include fish stocking programs, private
aquaculture, the bait industry, the aquarium and ornamental pond industry, live fish food
markets, recreational boating, and canals and diversions (Kerr et al., 2005).
While ballast water discharge (BWD) is the most prevalent pathway, an increase in BWD
does not directly translate to more species invasions. Most discharges of ballast water in the
Great Lakes occur in Lake Superior (Colautti et al., 2003), yet Lake Superior has less invasive
species than any of the other Great Lakes (Grigorovich et al., 2003b). The low NIS colonization
rate in Lake Superior may be due to any of several factors including cooler temperatures, a high
ratio of deeper waters, low food availability due to low productivity, and low calcium
concentrations (Grigorovich et al., 2003b).
2.1.3. Measures to Control the Release of Ballast Water Containing NIS
In response to NIS invasions stemming from ballast water releases in the Great Lakes,
voluntary ballast water exchange (BWE) guidelines were implemented by Canada in 1989 and
made mandatory in 2006. Mandatory BWE regulations were instituted by the USCG in 1993.
These regulations require vessels carrying ballast water and entering the U.S. Great Lakes from
outside the U.S. Exclusive Economic Zone (usually 200 miles away from the United States) to
comply with one of the following three options:
1) Vessels may exchange ballast water in open-ocean waters more than 200 nautical miles
from any shore, and in waters more than 2,000 m deep, before entering the Snell Lock, at
Massena, New York, provided that salinity of the ballast water is at least 30 parts per
thousand (ppt).
2) Vessels may retain their ballast water on board (vessels in this status are referred to as
Ballast-on-Board, or BOB vessels).
-------
) /Michiga
y
Wisconsin /
Green Bay
Figure 1. The five Great Lakes, some of the Great Lakes ports, and
surrounding region.
Shipping/Ballast
Water
65%
Unintentional
Release
11%
Figure 2. Sources of Great Lakes species invasions from 1960-2006 (based
on data provided in Appendix A, which are derived from NOAA, 2007a).
8
-------
3) Vessels may use an alternative environmentally sound method of ballast water
management that has been submitted to, and approved by, the Commandant of the USCG
or an authorized representative before the vessel's voyage (33 CFR 151.1510).
Compliance with these ballast regulations has been high. From July 1999 to June 2001,
93% of regulated ships reported performing the necessary level of active BWE before arriving in
Massena. The remaining 7% of ships were forced by the USCG to perform some sort of
alternative action, such as decontamination, prior to being allowed to enter the Great Lakes
(USCG, 2001). USCG reported high rates of BWE compliance (89 +/- 10%) for the period
1992-2004 (Ruiz and Reid, 2007).
Ballast water exchange at sea works first by the dilution effect. Assuming a homogenous
distribution of flora and fauna in the ballast tank, 95-99% of the fresh water (and organisms)
would be replaced by seawater (NRC, 2008). Second, BWE can be effective since most
remaining freshwater organisms in the ballast tank are killed by the resulting high salinity levels.
Despite these ballast water regulations, at least 13 new NTS are believed to have entered
the Great Lakes from ballast water since 1993 (Appendix A; JAGLR, 2002; Holeck et al., 2004;
NOAA, 2007a). It is possible that BWE has been effective and that all the species found after
1993 were introduced before 1993; it just took many years to detect and report them
(Costello et al., 2007a). Others have noted that organisms can survive BWE, and that BWE
practices have not been completely effective in terminating the flow of NTS into the Great Lakes
(e.g., Grigorovich et al., 2003a, b; Holeck et al., 2004; Drake et al., 2005; Ricciardi, 2006).
Recently, more stringent regulations have been implemented (e.g., 73 FR 37, p.9950), which
should reduce the flow of NTS into the Great Lakes from commercial shipping.
2.1.4. NOBOB Vessels and Species Invasions
Vessels fully loaded with cargo generally carry no ballast water on board. Vessels with
no ballast-on board, commonly called NOBOB vessels, entering USGLP were not required to
flush their ballast tanks or use an alternative treatment method until 2006. It is possible that
invasions may have occurred from NOBOB vessels arriving in the Great Lakes (Maclsaac et al.,
2002; Johengen et al., 2005). The almost completely empty ballast water tanks in NOBOB
vessels often still contain residual sediment and water from previous ballasting operations.
These residuals cannot be pumped from the ballast water tanks since the pump-out ports cannot
be closer than several inches from the bottom of the tank. Residual material in ballast water
tanks of NOBOB vessels can contain thousands of live organisms, their resting eggs and cysts,
and microorganisms, including human pathogens, all of which may be discharged into Great
Lakes waters (Johengen et al., 2005). When a NOBOB vessel off-loads cargo at a Great Lakes
-------
port of call it often takes on Great Lakes water into its ballast tanks to reestablish ballast. This
pumped in freshwater mixes with the residual material in the ballast tanks, thereby increasing the
viability of organisms. When such a NOBOB vessel then moves to another port to take on
cargo, it may then discharge some or all of the recently acquired ballast water, along with the
NTS from the earlier residual material.
NOBOB vessels currently account for about 90% of all inbound traffic to the Great Lakes
(Maclsaac et al., 2002). Due to the number of potential invasive species in the residual material
in NOBOB vessels and due to the large relative proportion of NOBOB vessels entering the Great
Lakes, NOBOB vessels could pose a significant invasion risk. Maclsaac et al. (2002) found that
for bacteria, copepods, cladocerans, and rotifers, NOBOB vessels may be exerting 10 to
100 times as much propagule pressure as vessels with ballast on-board complying with the
regulations.
2.1.5. Other Options for Controlling Species Invasions From Ballast Water
As a result of the threat from NOBOB vessels, Canada developed mandatory regulations
in 2006, requiring that transoceanic NOBOB vessels arriving in Canada undergo ballast flushing
to eliminate fresh or brackish water residuals in their ballast tanks. Coastal vessels entering
Canadian ports must comply with fairly similar requirements, only the BWE or ballast flushing
must occur in an area only 50 nautical miles from shore (GLBWWG, 2008). Since August 2005,
NOBOB vessels entering U.S. waters have been strongly encouraged, but not required, to
conduct saltwater flushing before entering the Great Lakes (71 FR 18, pages 4,605-4,606). The
St. Lawrence Seaway Development Corporation published regulations, which became effective
at the start of the 2008 navigation season, requiring all NOBOB vessels that have operated
outside the exclusive economic zone (usually 200 miles from the United States) to conduct
saltwater flushing of their ballast tanks before transiting the St. Lawrence Seaway, regardless of
whether their destination is a U.S. or Canadian port (73 FR 37, p. 9,950).
It is not yet possible to measure the effectiveness of recent regulations or guidelines
because there is a time lag between when a species is transported, colonizes, and reproduces to a
large enough population, to be detected and reported. However, the National Research Council
recommends that a binational science-based surveillance program be established to monitor for
aquatic invasive NTS (NRC, 2008). The recommended program should involve dedicated lake
teams, as well as academic researchers, resource managers, and local citizens groups, and it
should leverage existing monitoring activities whenever possible.
NOAA is testing the effectiveness of BWE along with other various methods to treat
ballast water using mechanical (e.g., filtration and separation), physical (e.g., sterilization by
10
-------
ultraviolet light, ozone, heat, electric current, or ultrasound), and chemical (e.g., chlorine
dioxide) methods (NOAA, 2007b). The State of Michigan has established its own ballast water
legislation, and other Great Lakes states are considering similar regulations (NRC, 2008). In
2007, the Michigan Department of Environmental Quality started prohibiting ballast water
releases from oceangoing vessels into Michigan waters until a permit was issued by the state.
Permits require one of four approved treatments, either sodium hypochlorite, chlorine dioxide,
ultraviolet light radiation treatment preceded by suspended solids removal, or de-oxygenation
(MDEQ, 2008). Because Michigan is currently an import state, there have been no permit
applications to discharge ballast water into Michigan ports since Michigan's law was
implemented in 2007 (telephone conversation on August 8, 2008 between Barry Burns,
Michigan Department of Environmental Quality, and Vic Serveiss, U.S. EPA, NCEA).
Therefore, oceangoing vessels visiting Michigan ports have not needed to install the Michigan
approved ballast water treatment methods.
2.2. CONCEPTUAL FRAMEWORK
Owing to the fact that invasive species are a major threat to ecosystems, there is a need to
develop predictive tools and to demonstrate their use to natural resource managers as they
consider ways to manage the problem. The approach used in this assessment, as shown in
Figure 3, is based on Williamson's (1996) conceptual framework regarding biological invasions.
Specifically, Williamson's thesis states that successful invasions are best predicted by knowing
the propagule pressure (i.e., the number of larvae/individuals entering a new area) and matching
the invaded habitat with the habitat in the invader's historical range.
2.2.1. Propagule Pressure
Propagule pressure is a composite measure of the number of individuals of a species
released into a region to which they are not native. It incorporates estimates of the absolute
number of individuals involved in any one release event (propagule size) and the number of
discrete release events (propagule number). The probability of establishment of an introduced
species increases as propagule pressure increases (Menges, 1998, 2000; Simberlof and von Holle
1999; Kolar and Lodge, 2001). In considering the sources of propagules to the Great Lakes,
ballast water becomes the primary concern as shown in Figure 3.
11
-------
Translocation of species
from native range
Propagule pressure
determined by ballast
water discharges
Habitat suitability
determined by
GARP
Sufficient number
of propagules
1
Suitable habitat
Successful invasion
Figure 3. Conceptual framework for predicting future introductions of
nonindigenous species into the Great Lakes of the United States.
The condition and life stage (resilient resting stages compared with sensitive juvenile
stages) of propagules will also strongly affect the probability of establishment
(Smith et al., 1999; Hayes and Hewitt, 2000; Wonham et al., 2001). Thus, management actions
that reduce the number of released individuals, the number of introduction events, and the health
of individuals released are likely to reduce the risk of invasion. Unfortunately, detailed
quantification of these factors is limited and thus surrogate measures become necessary to
estimate propagule pressure. As shown in Figure 3, the conceptual approach for this study uses
ballast water discharge data as a surrogate measure for propagule pressure. Current scientific
understanding of invasion biology suggests strongly that consideration of propagule pressure
should be a major component of an assessment.
2.2.2. Habitat Suitability
2.2.2.1. Species Distribution Modeling
Assessing the degree to which a new environment is similar to the donor environment is a
reasonable starting point to try to answer the question "Is a species likely to survive in this
12
-------
environment if it were introduced here?" Good computer-based tools are available that provide
a first-cut broad geographical answer to the question. Standard methods for modeling suitable
habitat include traditional multivariate statistical methods (e.g., discriminate analysis, multiple
regression, logistic regression), often coupled with geographic information systems (GIS) (e.g.,
Ramcharan et al., 1992; Buchan and Padilla, 2000). More recent methods that are tailor-made
for identifying potential ranges include CLIMEX (Sutherst et al., 1999), Genetic Algorithm for
Rule-Set Production (GARP) (Peterson and Vieglais, 2001; Drake and Bossenbroek, 2004) all of
which are embodied in user-friendly and readily available software. Predicting suitable habitat is
also possible for aquatic environments (Drake and Bossenbroek, 2004; Marchetti et al., 2004a),
but currently less tractable than for terrestrial habitats because: (1) fewer aquatic
physico-chemical data are available in appropriate electronic formats, and fewer distribution data
have been collected for aquatic species; (2) terrestrial climatic data are often poor predictors of
the aquatic environment; and (3) strongly predictive environmental variables for establishment
are unknown for many aquatic species (Carlton et al., 1995). All of these predictive models have
at least two intrinsic limitations. First, environment matching assumes that no evolution will
occur in the nonindigenous species with respect to habitat requirements (Cox, 2004;
Sakai et al., 2001). Second, biotic interactions in a new environment may limit or facilitate
establishment independent of any climatic match (Torchin and Mitchell, 2004).
2.2.2.2. Genetic Algorithms for Rule-Set Production (GARP)
GARP develops predictions of the potential geographic extent of an invasion by first
modeling relationships between known occurrences of a species and the corresponding abiotic
environmental variables, and then projecting the modeled species-environment relationships to a
region of interest. GARP modeling requires two types of inputs: (1) spatial data describing the
location of species based on occurrence data and (2) digital data layers describing environmental
conditions at locations coinciding with the species occurrence data. GARP develops outcomes
consisting of a set of conditional rules in the form of 'if-then' statements that describe the
ecological conditions of the species in its studied habitat (Stockwell and Peters, 1999). Habitats
are matched by searching iteratively for nonrandom correlations between a species' known
location and a variety of environmental parameters.
The GARP method is considered to be based on models of genetic evolution (Holland,
1975) because GARP models are built by an iterative process of rule selection, evaluation,
testing, and incorporation or rejection of the rules produced (Peterson et al., 1999). With each
iteration, rules are modified by selection, crossover, and mutation—resembling the genetic
process. In the first phase, GARP selects a random population, based on a combination of initial
13
-------
prediction rules, which might represent suitable solutions for the problem. The fitness to the
characteristics of the population is then evaluated for each pixel in the search space. If the
performance of the rule is adequate as determined by the rule's significance measure, the rule is
retained for further runs of the algorithm, until an end condition—consisting of a convergence
limit and maximum number iterations—is satisfied (Stockwell and Peters, 1999). One of the
main advantages of GARP is its ability to apply different types of rules at once to explain
complex nonlinear relationships between the species occurrence and predictive variables. This
implies that the algorithm can 'learn' through each iteration and apply the type of rule that
describes best the relationship among the variables for any particular portion of the search space
(i.e., all possible combinations of variables) (Stockwell and Peters, 1999).
2.2.2.3. Modification of GARP for Aquatic Systems
GARP has been used to predict a variety of species distributions including birds in
Mexico (Feria and Peterson, 2002; Stockwell and Peterson, 2002; Anderson et al., 2003) and
North America (Peterson and Cohoon, 1999); rodents in South America (Anderson et al., 2002
and Anderson et al., 2003); and invasive vector disease insects in South America (Peterson et al.,
2002). This may be the first time that remote sensing data were used in conjunction with GARP
to predict invasive freshwater aquatic species.
14
-------
3. METHODS
For a nonindigenous species (NTS) to become established in the Great Lakes, the species
must (1) move or be transported from its existing spatial range to the Great Lakes and (2) be able
to colonize, become established, and spread in the new environment (Williamson, 1996;
Theoharides and Dukes, 2007). Others have also combined these two analyses to predict NTS
spread, though different names were used to characterize their respective efforts. Leung and
Mandrak (2007) combined invasability and propagule pressure, to make predictions about zebra
mussel spread. Herborg et al. (2007) combined introduction effort and environmental niche
models to predict the potential spread of the Chinese mitten crab (Eriocheir sinemis) in North
America. To address both requirements for successful invasion, we used information on ballast
water discharges as a surrogate for propagule pressure and the Genetic Algorithm for Rule-Set
Production (GARP) model to determine the suitability of habitat by matching the invaded habitat
in the Great Lakes with the species native habitat.
3.1. HABITAT SUITABILITY USING THE GARP MODEL
Habitat suitability was modeled using a species distribution model to compare the
environmental conditions associated with the distribution of invasive species in their home range
with the conditions found in the Great Lakes. GARP was selected because it is a well
established model, is one of the few models that accepts presence-only distribution data (e.g.,
locations where the species has been observed without corresponding information on where the
species has not been observed), and incorporates multiple statistical approaches into a single
framework.
3.1.1. Selection of Modeled Species
The first step to using GARP is to select the species to be evaluated. Species of interest
(i.e., those thought to be potential invaders of the Great Lakes system) were identified based
upon a review of the literature and best professional judgment. We searched for species'
scientific names and the keywords "invasive" and "Great Lakes" in publications after 1990 using
Web of Science and international databases, such as Fishbase (Froese and Pauly, 2007) and
Global Invasive Species (IUCN, 2006). We augmented this general search strategy to include
authors who have studied Great Lakes invasive species to find potentially relevant papers that
did not specifically include the terms "invasive" or "Great Lakes" in the article's title, abstract,
or keywords. Other sources include the U.S. Geological Survey, the States' Department of
Natural Resources (for states adjacent to the Great Lakes), the Canadian Wildlife Federation, and
15
-------
the Great Lakes Panel on Aquatic Species. We initially identified 156 species of concern based
on a review of the literature (see Appendix B).
Of the 156 species identified, using best professional judgment it was determined that 58
of these species pose the most risk for their potential to invade the Great Lakes and reach
population levels that could cause ecological impact (see shaded entries in Appendix B).
Twenty-eight of the 58 species identified are already in the Great Lakes. The remaining
30 species, not yet reported in the Great Lakes, were evaluated to see if sufficient data was
available to run the GARP model. GARP requires at least 30 spatially unique occurrence points
(i.e., latitude-longitude coordinates of locations where the species has been reported) to develop
robust predictions (Stockwell and Peterson, 2002). For a variety of reasons, only 9 of the
30 species had sufficient data to be modeled. Of these nine, five species have not yet been
detected in the Great Lakes, and the other four have been reported only infrequently. At the
request of EPA's Great Lakes National Program Office, we modeled five additional species
already found in the Great Lakes. Two of the five species, the zebra mussel and round goby, are
currently widespread throughout the Great Lakes. The three other species, ruffe
(Gymnocephalus cernuus), quagga mussel, and New Zealand mud snail (Potamopyrgus
antipodarum\ have been reported as established in the Great Lakes but are not yet widespread
(USGS, 2007). Thus, a total of 14 NTS species were evaluated using the GARP model for the
availability of suitable habitat in the Great Lakes (Table 1).
3.1.2. Model Inputs and Environmental Data Layers
3.1.2.1. Environmental Data Layers
Six specific parameters were used to define environmental variables suitable to develop
data layers for GARP: mean, maximum, and minimum monthly surface water temperature;
chlorophyll a concentration; the diffuse attenuation coefficient; and normalized water-leaving
radiance (Table 2). These six parameters were chosen because they represent important
environmental variables that tend to control the distribution of species. Three of the parameters
are measures of temperature that affects species distribution worldwide. The other three are
related to the productivity of aquatic systems. Chlorophyll a is an indicator of biological
productivity. Water clarity, as measured by diffuse attenuation coefficient and the water-leaving
radiance, is an indicator of the trophic state of the system. Water clarity also influences the
depth of the photic zone and the ability of primary producers to acquire sunlight and flourish.
Although some of these six data layers may be covariant, GARP is considered to be relatively
robust to collinearity (Kluza et al., 2007). For the species modeled in this report, no literature
16
-------
Table 1. Fourteen species modeled using GARP and the source of
occurrence data
Species common name and year reported
Description
Useful
occurrence
data records
Data source
Species already widespread in the Great Lakes
Gymnocephalus cernuus, ruffe — 1986
Dreissena polymorpha, zebra mussel — 1988
Dreissena bugensis, quagga mussel — 1989
Neogobius melanostromus, round goby — 1990
Potamopyrgus antipodarum, New Zealand mud
snail— 1991
fish
mollusk
mollusk
fish
mollusk
229
268
83
145
867
GBIFa USGSb
GBIF, USGS
USGS
GBIF, USGS
GBIF, USGS
Species reported in the Great Lakes but either not extensive or lacking spatial data
Cercopagis pengoi, fishhook waterflea — 1998
Scardinius erythrophthalmus, rudd — 1989
Proterorhinus marmoratus, tubenose goby — 1990
Alosa aestivalis, blueback herring — 1995
crustacean
fish
fish
fish
152
57
171
408
GISC
GBIF, CIMSd
CIMS, BSRDB6
GBIF
Species not yet reported in the Great Lakes
Corophium curvispinum, N/A
Neogobius fluviatilis, monkey goby
Pomatoschistus minutus, sand goby
Rutilus rutilus, roach
Tinea tinea, tench
amphipod
fish
fish
fish
fish
65
50
102
117
50
GBIF, CIMS
CIMS
GBIF, BSRDB
GBIF, CIMS
CIMS
aGlobal Biodiversity Information Facility, 2007 (http://www.gbif.org/').
bUSGS Nonindigenous Aquatic Species Database, 2007 (http://www.usgs.gov/pubprod/maps.htmlX
'Regional Biological Invasions Center. INVADER, 2007
(http ://www.zin. ru/proj ects/invasions/gaas/invader/invader. htm).
dCaspian Interactive Map Service, 2007 (http://ipieca.unep-wcmc.org/imaps/ipieca/caspian/viewer.htmX
eBlack Sea Environment Programme Red Data Book, 2007 (http://www.grid.unep.ch/bsein/redbook/index.htmX
17
-------
Table 2. Environmental variables used to predict locations that would
provide suitable habitat for the 14 modeled species in the Great Lakes. The
spatial resolution of each of these six data layers is -21 km2
Variable
Mean monthly temperature
Maximum mean monthly temperature
Minimum mean monthly temperature
Chlorophyll a concentration
Diffuse attenuation coefficient (K490)
Normalized water-leaving radiance
(nLW551)
Units
°C
°C
°c
mg/m3
m-1
mW/(cm2 |im sr)
Source
AVHRRa
AVHRR
AVHRR
MODISb
MODIS
MODIS
Collection
period
1985-2002
1985-2002
1985-2002
2001-2005
2001-2005
2001-2005
"Advanced Very High Resolution Radiometer.
bModerate Resolution Imaging Spectroradiometer.
reliably supports the a priori weighting of any of the selected environmental variables as more
important than any other variable.
3.1.2.2. Environmental Data Sources
Water temperature was derived from the satellite-based Advanced Very High Resolution
Radiometer (AVHRR) Oceans Pathfinder Sea Surface Temperature Data set and is accurate to
within 0.5°C (http ://podaac-www.jpi.nasa.gov/sst/). We used temperature data from 1985
through 2001. We used the MMT data to calculate three data layers for use by the GARP
models: Maximum MMT, Mean MMT, and Minimum MMT. Maximum MMT represents the
highest value of each of the 12 sets of monthly averages of data. To calculate the mean MMT,
we assigned each pixel the average of the 12 monthly averages. The minimum MMT represents
the lowest value of each of the 12 monthly averages. Chlorophyll a concentrations were
obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor. The
Diffuse Attenuation Coefficient (K490) relates to the presence of light-scattering organic and
inorganic particles in the water column and is inversely related to water clarity
(http://oceancolor.gsfc.nasa.gov/PRODUCTS/k490.html). The normalized water-leaving
radiance is the radiance of reflected light at 551 nm. Since water absorbs very little light at 551
18
-------
nm, increases are due to light reflection out of the water, which are usually caused by
nonabsorbing particles such as suspended sediments.
3.1.3. Use of GARP model
A stand-alone version of Desktop GARP (version 1.1.6) (Scachetti-Pereira, 2002) was
used to model the distributions of the first nine NTS; Desktop GARP (version 1.1.6) within Open
Modeler (version 1.0.5) was used for five of the species. There is no functional difference
between the two desktop versions used. GARP relies on species occurrence or presence data and
synthetic species absence data, termed pseudo-absence data. The use of pseudo-absence data is
an intrinsic and accepted part of GARP modeling. To develop pseudo-absence data,
investigators must identify a region surrounding the occupied range of the target species to which
the species could easily spread. It is assumed that the reason the species is absent from the
surrounding region is because environmental conditions are different and outside the species
tolerance limits. The established range includes the species' occupied range and a surrounding
unoccupied range. The pseudo-absence data are selected from a subset of locations within the
investigator-defined study area that are not currently occupied by the species. The pseudo-
absence points represent the locations presumed to be unsuitable for the target species and
provide a contrast against which occurrence models can be developed. A new, random selection
of pseudo-absence data was made for each model iteration.
GARP divides the occurrence data into training and test data sets. Test data sets are
reserved to test predictive performance of models that are developed using the training data sets.
GARP then uses the training data and one of the individual algorithms to develop a model, and
the model is tested and improved until the best solution is found. For each of the 14 modeled
species, GARP randomly assigned the data into 50/50 splits of training and test data sets. We
produced 1,000 model runs from the training data sets (i.e., 14,000 total GARP runs) that are all
slightly different and vary in predictive ability. Each individual GARP model run produces a
map of 'Os and Is,' representing predicted absence and presence, respectively. The area of
predicted presence for each run is simply the proportion of pixels that have a predicted value of
1. We used a procedure described by Anderson et al. (2003) to select the best subset of runs
generated for each species to develop a final composite range map. The model is run for each
species and compared to see which runs best predicted the known occurrence locations and
omitted the fewest of the known occurrence locations of the modeled species. For each species,
we retained the models with the lowest omission error rates when compared to the results of the
test data sets.
19
-------
Next, the median area of predicted presence for each species (simply the number of
pixels that have a predicted value of 1) was determined from the models with the lowest error
rates. The subset of models within 50% of this median value was then selected to form the final
prediction of habitat suitability. Every pixel on the model output represents either a predicted
presence or a predicted absence. The final composite prediction maps reflect the sum of the
results from the models; that is, each pixel was assigned a value ranging from zero (i.e., no
models predict presence) to 100 (i.e., all models predict presence). This value represents the
relative environmental suitability of each location.
The final habitat suitability maps reveal the frequency with which a pixel is predicted to
provide suitable habitable and depict the repeatability of that prediction with different models
developed from different, randomly divided, training data sets. The higher the value of a pixel,
the more likely the modeled species is expected to find suitable habitat at that location. Another
interpretation is that pixels that have higher values represent higher quality habitat for the
modeled species because these locations are predicted to be suitable by more models. Those
pixels that are predicted to provide highly suitable habitat within the Great Lakes are
characteristic of environments similar to those known to be occupied by the modeled species in
their natural range.
3.1.4. Assumptions and Limitations
The use of species distribution models to predict the spread of NTS required three
assumptions: (1) the available distribution data describe the full range of environmental
conditions that the modeled species can tolerate; (2) the environmental variables selected to
model potential spread govern the current and future geographic ranges of the NTS under study;
and (3) biotic factors do not influence species distributions, unless such biotic factors can be
included explicitly as environmental data layers. Failure to meet these three assumptions can
limit the ability of GARP to predict invasive spread and can result in two types of errors, which
arise from two broad sources: (1) limitations inherent in the data or the model themselves (these
are often termed data errors), and (2) ecological processes relevant to the distribution of the
species that are not included in the model (often called biological errors).
3.1.4.1. Data Errors
A major source of error in GARP modeling is the lack of complete occurrence data for
the modeled species. GARP requires occurrence points that are both representative of the full
range of environmental conditions associated with presence of the species and of the area
inhabited by the species. In reality, the actual occurrence points reflect bias in both sampling and
20
-------
reporting efforts, which is influenced by resources, accessibility, transportation corridors, and
visitation frequency. Further, the occurrence data extracted from online databases were collected
ad hoc, and not for the purposes of constructing distribution models. If occurrence data do not
describe the full environmental tolerances of the species, predictions will underestimate areas
where a NTS could survive and establish itself in a particular region of the Great Lakes. For
example, occurrence data for the monkey goby and the tench were available only for their
distributions within the Iranian portion of the Caspian Sea. This may be part of the reason these
two species have the smallest predicted distributions of any modeled species within the Great
Lakes, perhaps suggesting under-prediction. Therefore, the predicted habitat suitability might
not include all environments which the NTS could invade.
Some occurrence data were discarded because they did not fall in waters defined by the
21 km2 spatial resolution available for the environmental data layers. These discarded
r\
occurrence points are more than likely in lakes and rivers smaller than 21 km that could not be
resolved by the satellite sensors used in this study. GARP requires that species absence data be
developed by accurately selecting the region for which the species is absent because
environmental conditions are different and outside the species tolerance limits. Determining the
extent of the GARP prediction region assumes that these pseudo-absence points really are
uninhabitable, and not, for example, simply suitable environments to which the target species has
not yet dispersed.
Model errors can also result from modeling habitat suitability with a limited set of
environmental variables. While we know that each of the six selected variables has a strong
influence on species distributions, other abiotic factors known to influence species distribution
are not captured by the variables that we used. For example, salinity impacts the survival of
many aquatic organisms (Bailey et al., 2005), but salinity is not included in the GARP analysis.
We were unable to locate a global database with spatial salinity data at the same scale of
resolution as the six variables included in this study. Including salinity at a coarser resolution
would have introduced coarse range boundaries where salinity was the limiting factor.
Bathymetry data were available, but not used, because species occurrence data did not include
the depth at which the species was found. As many aquatic species may only survive in waters
to a certain depth, the model would show some deeper waters as suitable habitat and may
contradict what is known about the depth limitations of a particular species. Nutrient levels may
also be biologically important to some NIS. Calcium concentrations, for example, are likely a
limiting factor for zebra and quagga mussel distributions (Cohen, 2007), but global spatial
databases of calcium concentrations are not available. Failure to include such key factors can
lead to over-predictions (predicting that the species can survive in an area where the habitat is
actually unsuitable due to the environmental variable not included in the model). Detailed
21
-------
knowledge of the target species is required to reliably determine if one of more key factors have
been excluded from the models. Because the consequences of under-prediction (failing to
identify a place where NTS can establish) are much greater than those resulting from
over-prediction, our approach is conservative and errs on the side of over-prediction.
3.1.4.2. Biological Errors
Even if the environmental variables could accurately reflect the abiotic factors controlling
species distributions, the predictions are developed without considering biotic factors such as
competition, predation, and parasitism. Biotic factors also are important determinants of the
distributions of species, but it is not ordinarily possible to obtain data on biotic factors for
incorporation into GARP, and failure to consider such factors can lead to poor predictions (e.g.,
Fitzpatrick et al., 2007). In a new environment, a species may be freed of restrictions (e.g., a
predator may not exist), encounter new challenges in a new environment (e.g., competition from
a species with a similar niche), or evolve and adapt. Thus, it is difficult to predict whether the
impacts of excluding biotic factors would inhibit establishment or expand the colonization range
of an introduced species. The fire ant in the southeastern United States is an example of an
introduced NTS that established itself beyond the predicted range of a species distribution model,
perhaps due to biotic factors that encouraged the species successfulness (Fitzpatrick et al., 2007).
3.1.5. Testing the GARP Model Performance
Despite the limitations described above, species distribution models are currently one of
the few techniques readily available to predict the potential for an invasive NTS to become
established in an area of interest (Peterson, 2003). Therefore, species distribution models should
be considered a key component of a multi-faceted NTS prevention and management plans (Mack,
1996; Peterson and Vieglais, 2001).
The GARP model outputs were validated by testing the ability to correctly predict
independent data that were not used to develop the model. Specifically, we evaluated how well
GARP performed by assessing how well the model predicted the known distributions of three
NTS that are already widespread throughout the Great Lakes using distribution data collected
outside of the Great Lakes. Thus, occurrence data for the zebra mussel, ruffe, and New Zealand
mud snail within the Great Lakes were withheld from the GARP model runs and the model
tested for its ability to correctly predict suitable habitat. The performance of the GARP model
was assessed using area under the curve of the Receiver Operating Characteristic curve (Sing et
al., 2005). Area under the curve is a threshold-independent evaluation of model performance
that measures the ability of the model to differentiate between sites where a species is present
22
-------
from sites where it is considered absent. Area under the curve represents the probability that,
when a predicted-present site and a predicted-absent site are drawn at random, the predicted-
present site will have a higher predicted value. The effectiveness of the GARP modeling is
based on the scale for determining model performance devised by Swets (1988). More details on
the model validation approach we used are provided in Appendix C.
3.1.6. Determining GARP's Power to Predict
GARP and other species distribution models make predictions about the suitability of
habitat for a particular species within a region of concern. These models are developed by
comparing the environmental conditions in the region containing the species to those found in
the region of concern, in this case the Great Lakes. As noted previously, predictions from GARP
and other species distribution models are valid only for the range of environmental conditions on
which the model was developed. Reliable predictions cannot be made for any environment
within the Great Lakes that are not similar to those found within the region containing the
distribution of the study species. GARP does not have a method for determining when a reliable
prediction cannot be made, and, instead, may report such environments as a predicted absence
when they may indeed be habitable by the NTS. Reporting such areas as unsuitable habitat may
be erroneous and could misdirect management attention away from these potentially susceptible
areas.
We used a technique called "power-of-prediction analysis," devised expressly for this
project, to distinguish between areas of predicted absence from areas for which a reliable
prediction cannot be made (null prediction). Like GARP models for individual species, power of
prediction analysis uses GARP to develop predictions. However, instead of developing a model
of environments represented by the distribution of the study species, power of prediction analysis
attempts to (1) model all environments within the region containing the distribution of the
species and (2) compare these environments with those characterizing the Great Lakes.
To perform a power of prediction analysis, we identified a region encompassing the full
range of environmental conditions to which the species is known to occur. For example,
consider a hypothetical species in the Caspian Sea reported in regions with water temperatures
between 15 and 20°C but not reported in regions with a water temperature from 10 to 15°C.
GARP would then predict that places in the Great Lakes with water temperatures between 15 and
20°C provide suitable habitat and that all regions with temperatures less than 15°C are
unsuitable. However, temperatures in the Great Lakes range from 6 to 20°C. This presents a
problem regarding areas within the Great Lakes that range from 6 to 10°C, which is below any
temperature found in the Caspian Sea. When considering temperature, in an isolated, univariate
23
-------
way, it is likely the species would not tolerate temperatures from 6 to 10°C if it could not tolerate
10 to 15°C. Yet, it is not correct to assume that all locations beyond a particular extreme, in this
case less than 10°C, are unsuitable. The species experiences the environment in a multivariate
manner and that could produce surprising and counterintuitive results. For example, a terrestrial
species might be able to survive and reproduce in locations that were hotter, if they were also
wetter.
Even though GARP cannot make a reliable prediction for such areas, GARP and many
other species distribution models will report areas with temperatures outside of the range from 10
to 20°C as unsuitable for the species, when in reality the GARP model has no information to
draw such a conclusion (Heikkinen et al., 2006). We used power of prediction analysis to denote
the geographic extent of predictive power. We performed power of prediction analyses for 11 of
the 14 modeled species. Power of prediction analysis was not performed for two of the invasive
species already established in the Great Lakes—quagga mussel and round goby—due to the lack
of occurrence data outside the Great Lakes. Also, no power of prediction analysis was needed
for the blueback herring because GARP model runs predicted that the blueback herring can
encompass essentially the entire area of the Great Lakes. Appendix D provides more details on
how we applied the power of prediction analysis to this study.
3.2. DETERMINING PROPAGULE PRESSURE USING BALLAST WATER
DISCHARGE DATA AND VESSEL TRAFFIC PATTERNS
The probability that a NTS can become established increases with increased propagule
pressure (Simberloff and Von Holle, 1999; Kolar and Lodge, 2001; Lockwood et al., 2005).
Propagule pressure, as explained in the introduction, is the number of individuals (including
larvae, seeds, and spores) released in a nonnative region over a specified period of time
(Simberloff and Von Holle, 1999). We used two sources of data as a surrogate for propagule
pressure: Data from the U.S. Coast Guard's (USCG's) National Vessel Movement Center
(NVMC) and the National Ballast Information Clearinghouse (NBIC). Ultimately, the NBIC
data proved to be the most useful in predicting propagule pressure. The NBIC collects, analyzes,
and interprets data on ballast water management practices of commercial ships that operate in the
United States. NBIC was created by the USCG and the Smithsonian Environmental Research
Center (NBIC, 2008) pursuant to the National Invasive Species Act of 1996 (16 USC 67 § 4712).
NBIC's data are electronic and are accessible on the Internet (http://invasions.si.edu/nbic/).
3.2.1. Analysis of Ballast Water Discharge Data
The principal aim of the NBIC database is to quantify the amounts and origins of ballast
water discharges in U.S. coastal systems and to determine the degree to which such water has
24
-------
undergone open-ocean exchange or alternative treatments designed to reduce the likelihood of
ballast-mediated invasions by exotic species (NBIC, 2008). NBIC data come from national
ballast water management reporting forms submitted to the USCG by vessels arriving to ports
and places in the United States. The data includes port of arrival, date of arrival, and last port of
call, along with the source of ballast water (either a specific port or a latitude/longitude
coordinate at sea), date of ballast water intake, type of ballast water management, date
discharged, and the volume discharged.
This database allowed us to locate the source of ballast water and to determine those
Great Lakes ports receiving the most ballast water discharges with the most potential to transport
NTS. NBIC data from for 2004-2007 for Illinois, Indiana, Michigan, Minnesota, New York,
Ohio, Pennsylvania, and Wisconsin were downloaded and analyzed using a relational database
(Microsoft® Office Access). The original data set contained records of 44,461 vessel arrivals and
121,031 ballast water discharges. By excluding records of vessels arriving in ports outside the
Great Lakes system, the NBIC data set was reduced to 63,574 ballast water discharges.
Since NIS that were in the ballast tank before ballast water exchange at sea may survive
the exchange and can later be released in the Great Lakes, we needed to determine the original
source of ballast water. Discharges of ballast water that originated within the Great Lakes
(which we defined as west of Quebec City, see Figure 1) was excluded along with discharges of
ballast water that was derived from outside 200 nautical miles of any shore and deeper than
2,000 m. We analyzed the remaining 618 ballast water discharges because these waters have the
most risk of transporting NIS. We identified the most common original source of ballast water
and the U.S. Great Lakes ports (USGLP) receiving the most discharges.
As discussed previously, some vessels enter the St. Lawrence Seaway declaring to have
no ballast on board (NOBOB) but as they traverse the Great Lakes, they take on ballast water
which can mix with residual water or sediment in the ballast tanks. We combined NBIC with
NVMC data since the later includes information on the last five ports of call. Starting with
NBIC records, we matched the NBIC arrival port and arrival date to the corresponding data in
the NVMC data set. For each of the vessel records with matching sets of data, we obtained the
last five ports of call records from the NVMC database. If one of more of the last five ports of
call were not in the Great Lakes, we considered the vessel to have entered the Great Lakes with
no ballast on board but containing residual material (NOBOB-RM). We then calculated the
number of ballast water discharges and the volume of ballast water discharged from each of
these vessels.
25
-------
3.2.2. Assumptions and Uncertainty
Although NBIC employs a rigorous quality assurance and quality control protocol, the
accuracy and completeness of the self-reported data cannot be guaranteed (NBIC, 2008). The
NVMC data set also has limitations. Although a vessel may have stopped at a foreign port
during one of its last five ports of call it does not necessarily mean that ballast water was taken
on at that port. It is possible that any residual material in the ballast tank may be from within the
Great Lakes and not the foreign port of call included in the last five ports of call records. This
would lead to over-predicting the potential for NIS release. Similarly, it is also possible that we
missed a source of residual material because it may have been picked-up sometime earlier than
the last five ports of call. This would lead to an under-estimation of NIS releases. Finally, we
assumed that the data from 2006-2007 used in this study are representative of discharge and
shipping patterns over the past several years.
26
-------
4. RESULTS
We first present results comparing the Great Lakes with the rest of the world, especially
the Ponto-Caspian region with respect to six environmental parameters used to model species
distribution. We next identify those locations within the Great Lakes that would provide suitable
habitat for each of the 14 modeled nonindigenous species (NTS). And third, we identify the ports
within the U.S. Great Lakes that received the most ballast water discharges from the vessel
traffic we analyzed, including the identification of ports around the globe that provided the
original source of ballast water discharged at a U.S. Great Lakes port.
4.1. COMPARISON OF THE GREAT LAKES TO THE PONTO-CASPIAN SEA
The Ponto-Caspian region has been identified as a significant source of nonindigenous
species entering the Great Lakes. The comparison of the Great Lakes to the Ponto-Caspian
region, based on the six environmental parameters used in the Genetic Algorithm for Rule-Set
Production (GARP) modeling, reveals that the regions are indeed quite similar. Figures 4-7
illustrate the environmental conditions for those parameters used in the habitat suitability
modeling as shown in Table 2. Latitudinal differences and discernable patterns in deeper open
ocean waters are clearly evident.
4.1.1. Temperature
Overall, the maximum monthly temperature (MMT) shows a strong latitudinal gradient
(Figure 4). The Great Lakes, shown mostly in yellow, have a spatial mean of 9.9°C and range
from 5.9°-13.7°C. Lake Superior is colder than the other four lakes and has a spatial mean of
7.9°C with a range of 5.9°-11.1°C. The Caspian and Black Seas are somewhat warmer and less
variable (in terms of maximum MMT) than the Great Lakes, with a spatial mean of 13.7°C and a
range of 12.1°-16.1°C. The impact of climate change will likely cause the Great Lakes to reach
MMT levels even more similar to the Ponto-Caspian Sea region (IPCC, 2007).
4.1.2. Chlorophyll a Concentrations
Due to upwelling, the western edges of continents display high concentrations of
chlorophyll a (dark blue color in Figure 5), a surrogate measure of productivity. Colder arctic
waters are also more productive. The global spatial mean concentration of chlorophyll a is
0.24 mg/m3. The Great Lakes, shown mostly in blue in Figure 5, show a mean spatial
productivity level of 1.7 mg/m3, ranging from 0.19-62.3 mg/m3. The Caspian and Black Seas
are similar to the Great Lakes with a spatial mean of 2.2 mg/m3 and range from 0.2-56.2 mg/m3.
27
-------
Figure 4. Maximum monthly mean temperature (MMT) (°C) as determined
by AVHRR sensor (1985-2001). Warmest temperatures are indicated by red;
cooler temperatures are indicated by shades of blue. Global view (top),
Ponto-Caspian region (left), and Great Lakes (right).
28
-------
Figure 5. Average chlorophyll a concentration (mg/m ) as determined by
MODIS (2001-2005). High chlorophyll a concentrations are represented with
blue and dark green. Brown and yellow indicate low concentrations of
chlorophyll a. Global view (top), Ponto-Caspian region (left), and Great Lakes
(right).
29
-------
4.1.3. Diffuse Attenuation Coefficient (K490)
Overall, open ocean waters are generally clearer than waters with higher biological
productivity. The more light that is scattered from the surface, the Diffuse Attenuation
Coefficient (K490), the greater the amount of suspended solids, a measure of productivity.
Greater K490 values imply more light attenuation and lower water clarity. The global mean
K490 is 0.032/m, which translates to a photic zone depth of-144 m. The K490 of the Great
Lakes, shown mostly in blue in Figure 6, is much higher than the global average and has a spatial
mean of 0.099/m (equivalent to a photic zone depth of-47 m) and ranges from 0.037-0.741/m.
The Caspian and Black Seas also are fairly turbid and similar to the Great Lakes with a spatial
mean of 0.11/m (photic zone depths of-42 m) and range from 0.05-0.72/m.
4.1.4. Normalized Water-Leaving Radiance
As with the Diffuse Attenuation Coefficient, this is a measure of the productivity of
waters. As shown in Figure 7, and consistent with the previous figures, waters near the
continents are generally more productive. The Great Lakes have a spatial mean of 1.5 mW/(cm2
|im sr) and a range of water clarity from 0.0-5.1 mW/(cm2 jim sr). Lake Superior, being much
deeper (averaging 147 m and with a maximum depth of 406 m) than the other Great Lakes, has a
spatial mean of 0.5 mW/( cm2|im sr) indicating that Lake Superior is less productive. The
r\
Caspian and Black Seas have a spatial mean of 2.4 mW/(cm jim sr), similar to the lower Great
Lakes.
4.2. HABITAT SUITABILITY FOR MODELED SPECIES
The results of using the GARP species distribution model reveals that the Great Lakes
offers suitable habitat for all of the 14 modeled species in this study, with Lakes Erie and Ontario
the most likely to be invaded. Five of the species modeled are already established in the Great
Lakes and the remaining nine, selected from an original list of 156 species, are most likely to
invade and become established in the Great Lakes. Figures 8 through 10 illustrate the suitability
of habitat within the Great Lakes for the blueback herring (Alosa aestivalis\ quagga mussel
(Dreissena bugensis) and round goby (Neogobius melanostromus). Unfortunately, due to limits
with occurrence data, a power of prediction analysis could not be performed for the quagga
mussel and round goby. There was no reason to perform a power of prediction analysis for the
blueback herring because it is predicted to find suitable habitat throughout the Great lakes.
Figures 11 through 21 show the suitable habitat for the fishhook waterflea (Cercopagispengoi),
zebra mussel (Dreissenapolymorpha\ ruffe (Gymnocephalus cernus), monkey goby (Neogobius
fluviatilis), New Zealand mud snail (Potamopyrgus antipodarum), tubenose goby (Proterorhinus
30
-------
Figure 6. Average diffuse attenuation coefficient (m"1) at 490 nm as
determined by MODIS (2001-2005). Yellow and green colors indicate less
light absorption, blues indicate greater attenuation of light. Global view (top),
Ponto-Caspian region (left), and Great Lakes (right).
31
-------
Figure 7. Average normalized water leaving radiance (mW/cm um sr) as
determined by MODIS (2001-2005). Blues are higher values (i.e., higher
concentrations of particles in the water which reflect more light), and reds and
yellows indicate lower values (i.e., less light is emitted). Global view (top),
Ponto-Caspian region (left), and Great Lakes (right).
32
-------
marmomtus), rudd (Scardinius erythrophthalmus), an amphipod (Corophorum curvispmum),
sand goby (Potamoschistus minutus), roach (Rutilus rutilus\ and tench {Tinea tinea). These
figures also show areas where no reliable prediction could be made based on a power of
prediction analysis.
The habitat suitability maps are based on the best 100 of 1,000 model runs for each
species. The color scale (from blue to red) in the figures reflects the number of GARP model
runs, from 0 to 100, that predict the modeled species would find suitable environmental
conditions in the location being considered. Specifically, the red colored regions indicate where
nearly all GARP model runs predicted that NTS would find suitable habitat. The blue-colored
areas indicate where few or no GARP models runs predicted that NTS would find suitable
habitat. The power of prediction analysis helps to distinguish between areas with a low
likelihood of providing suitable habitat from areas where a prediction could not reliably be made.
The gray regions denote areas where no reliable prediction can be made about the potential
distribution of an invader. The color scale does not imply any measure of credibility or
precision, but rather expresses commonality among predictions developed via a stochastic
process of model generation.
To further validate the GARP model, results for the three species already reported in the
Great Lakes (zebra mussel, ruffe, and the New Zealand mud snail) were compared with their
current spatial distribution. This analysis indicates that GARP modeling is a good predictor of
habitat suitability according to the scale for evaluating the performance of species distribution
models devised by Swets (1988). The model performance scores are 0.79 for the zebra mussel
and ruffe and 0.74 for the New Zealand mud snail. These scores, representing the area under the
curve of predicted accuracy (Sing et al., 2005), suggest that the six environmental data layers that
were selected as inputs for the GARP modeling are appropriate for predicting the locations that
would provide suitable habitat. Appendix C provides more information on model validation.
4.2.1. Blueback Herring
If the blueback herring, a medium-sized fish, enters the Great Lakes it is very likely to
find suitable habitat throughout the Great Lakes system, according to GARP (Figure 8). Only
the deeper portions of Lake Superior and other isolated spots in other Lakes may not provide
suitable habitat for this species. Without a power of prediction analysis, it is not possible to
know if the blue colored areas reflect unsuitable habitat or areas where no prediction is possible.
The blueback herring and alewife are of similar shape and general appearance, and
distinguishing between them is difficult. Bluebacks tend to have a smaller eye than alewives,
with the eye diameter usually smaller than the snout length. As their name implies, these fish
33
-------
often have dark blue backs. An anadromous fish, the blueback herring spends the greater part of
its life in salt water and returns to fresh water to spawn. It usually spawns later in the spring than
the alewife, when water temperatures are a bit warmer. During spawning, many eggs are
deposited over the stream bottom where they stick to gravel, stones, logs, or other objects. A
few surviving, spent fish move back to the sea after spawning. Young fish usually move to sea
when about 1 month old and 1 1/2 to 2 inches long. Bluebacks feed on plankton, various small
floating animals, small fish fry, and fish eggs. Although the Great Lakes are distant from marine
waters, the blueback herring can spend its whole life and develop reproducing populations
entirely in freshwater (VA Inland Fisheries, 2008). If blueback herring became established in the
Great Lakes, they could impede recovery of depressed populations of indigenous fishes such as
cisco and lake trout (Owens et al., 1998).
Blueback herring (Alosa aestivalis)
Canada
Predicted
Suitability
•^ High (100)
Mid (50)
•• Low (0)
es
Figure 8. GARP-predicted habitat suitability of blueback herring (Alosa
aestivalis) in the Great Lakes.
34
-------
4.2.2. Quagga Mussel
The quagga mussel, a mollusk, already occupies most shoreline areas in Lakes Erie and
Ontario and southern Michigan (Figure 9). According to GARP modeling, the rest of Lake Erie
and the southern shoreline zones of Lakes Michigan and Huron also are likely to provide suitable
habitat for this species. Without a power of prediction analysis, predictions cannot reliably be
made for the other regions. However, the species has already been reported in several of these
locations, including the shorelines of Lake Michigan, Lake Superior, and Lake Huron. Quaggas
are prodigious water filterers, removing substantial amounts of phytoplankton and suspended
particulate from the water. As such, their impacts are similar to those of the zebra mussel. By
removing the phytoplankton, quaggas in turn decrease the food source for zooplankton, therefore
altering the food web. Impacts associated with the filtration of water include increases in water
transparency, decreases in mean chlorophyll a concentrations, and accumulation of pseudofeces
(Claxton et al., 1998). Quagga mussels prefer silty or sandy lake bottoms and can live in warm
or cold water. Maclsaac (1994) correctly speculated that the quagga mussel was still expanding
its nonindigenous range in the Great Lakes. It has spread to depths greater than it occupies in its
native range (Mills et al., 1996) and is abundant to a depth of 150 m (Wisconsin DNR, 2008) and
174 m in Lake Ontario (Watkins et al., 2007). By 1999, the quagga mussel dominated southern
Lake Ontario, where the zebra mussel was once dominant (Mills et al., 1999), and it continues to
spread into regions previously occupied by the zebra mussel (Watkins et al., 2007). The ability
to spread to areas that can be potentially occupied by the zebra mussel further supports the notion
that spread and colonization may occur until the species reaches its depth limitation.
4.2.3. Round Goby
The GARP model predicts the round goby, a medium-sized, bottom-dwelling fish, would
find suitable habitat throughout Lakes Erie and Ontario and along the shorelines of the other
Lakes (Figure 10). In fact, this species became established in all five Great Lakes by 1998
(Rasmussen, 2002). Round gobies perch on rocks and other substrates in shallow areas, yet they
have also been reported to flourish in a variety of habitat types (USGS, 2008a). Gobies also
have a well developed sensory system that enhances their ability to detect water movement. This
allows them to feed in complete darkness, giving them an advantage over other fish in the same
habitat (Wisconsin Sea Grant, 2008). Zebra mussels may have facilitated the invasion of the
round goby and other Eurasian species by providing an abundant food source (Ricciardi and
Maclsaac, 2000). The distribution of the round goby around the inshore areas of the Black and
Caspian seas indicates their potential for widespread occupation of inshore habitats with cover,
especially plants, in the lower Great Lake, yet they can migrate to deeper water (50-60 m) in
35
-------
Quagga mussel (Dreissena bugensis)
Canada .«-,
Predicted
Suitability
™ High (100)
Mid (50)
_ Low (0)
ares
Figure 9. GARP-predicted habitat suitability of quagga mussel (Dreissena
bugensis) in the Great Lakes. Inset map shows the locations where the species
has been reported.
Round goby (Neogobius melanostromus)
Canada .«-
l\ u
Predicted
Suitability
^" High (100)
Mid (50)
^H Low (0)
Figure 10. GARP-predicted habitat suitability of round goby (Neogobius
melanstromus) in the Great Lakes. Inset map shows the locations where the
species has been reported.
36
-------
winter (Jude et al., 1992). The numbers of native fish species have declined in areas where the
round goby has become abundant (Grossman et al., 1992). This species has been found to prey
on darters, other small fish, and lake trout eggs and fry in laboratory experiments. They also
may feed on eggs and fry of sculpins, darters, and logperch (Marsden and Jude, 1995) and have
also been found to have a significant overlap in diet preference with many native fish species.
They compete with rainbow darters (Etheostoma caeruleum), logperch (Percina caprodes), and
northern madtoms (Noturus stigmosus) for small macroinvertebrates (French and Jude, 2001).
4.2.4. Fishhook Waterflea
According to the GARP model, if transported to the Great Lakes, the fishhook waterflea,
a free-swimming macroinvertebrate, would likely find suitable habitat throughout the region,
except for the deeper waters of Lake Superior (Figure 11). The fishhook waterflea has been
reported in Lakes Ontario, Michigan (USGS, 2008b), and Erie (University of Minnesota, 2006).
The species is predicted to spread to the other Great Lakes, and, once established, it becomes
difficult to eradicate (University of Minnesota, 2006). Unlike several of the other modeled
species, population densities of the fishhook waterflea increase with distance from shore (IUCN,
2006), suggesting that this species may be able to occupy, given sufficient time, the entire region
including the deeper waters of Lake Superior.
4.2.5. Zebra Mussel
The zebra mussel, a mollusk, has already invaded the shoreline areas of all five Great
Lakes (Figure 12). The GARP model predicts the zebra model could potentially find suitable
habitats throughout most of the Great Lakes region. Zebra mussels were first discovered in
North America in 1988 in the Great Lakes. The first account of an established population came
from Canadian waters of Lake St. Clair, a water body connecting Lake Huron and Lake Erie. By
1990, zebra mussels had been found in all the Great Lakes. The following year, zebra mussels
escaped the Great Lakes basin and found their way into the Illinois and Hudson rivers. Zebra
mussels are notorious for their biofouling capabilities by colonizing water supply pipes of
hydroelectric and nuclear power plants, public water supply plants, and industrial facilities.
Zebra mussels can have profound effects on the ecosystems they invade. They primarily
consume phytoplankton, but other suspended material is filtered from the water column
including bacteria, protozoans, zebra mussel veligers, other microzooplankton, and silt. Large
populations of zebra mussels in the Great Lakes and Hudson River reduced the biomass of
phytoplankton significantly following invasion. Diatom abundance declined over 80% and
37
-------
Fishhook waterflea (Cercopagis pengoi)
Canada *-.
Predicted
Suitability
^m High (100)
Mid (50)
• Low (0)
United
States
Figure 11. GARP-predicted habitat suitability of fishhook waterflea
(Cercopagis pengoi) in the Great Lakes.
Zebra mussel (Dreissena polymorpha)
Predicted
Suitability
^B High (100)
Mid (50)
^m Low (0)
No Prediction
Possible
United
States
Figure 12. GARP-predicted habitat suitability of zebra mussel (Dreissena
polymorpha) in the Great Lakes. Inset map shows the locations where the
species has been reported.
38
-------
transparency, as measured by Secchi depth, increased by 100% during the first years of the
invasion in Lake Erie (Holland, 1992). Zebra mussels represent one of the most important
biological invasions into North America, having profoundly affected the science of Invasion
Biology, public perception, and policy. Zebra mussels are described as poor 62 regulators,
which may explain their low success rate in colonizing eutrophic lakes and the hypolimnion.
9-1-
Mellina and Rasmussen (1994) noted that calcium (Ca ) levels and water temperatures in the
open waters of Lake Superior are too low for the zebra mussel. Zebra mussels require 10 mg/L
9-1-
of Ca to initiate shell growth and 25 mg/L to maintain shell growth (USGS, 2008c). Zebra
mussels are generally within 2 to 7 m of the water surface (O'Neill, 2004) but, on rare occasions,
have been found at depths exceeding 90 m (Watkins et al., 2007). The depth limitation of the
species should further restrict the maximum potential spread of the species to Lake Erie and to
the shallower waters of the other four Great Lakes (NOAA, 2008). Competition with the quagga
mussel also appears to limit zebra mussel spread. Zebra mussels were outcompeted and almost
completely replaced by quagga mussels in Lake Ontario between 1995 and 2003 and this trend
could occur in other Lakes (Watkins et al., 2007).
4.2.6. Ruffe
The ruffe, a small to medium-sized fish, has already invaded Lake Superior and GARP
modeling predicts it will find suitable habitat almost everywhere in all five Great Lakes
(Figure 13). GARP models are not able to make a prediction about some of the deeper waters of
Lake Superior. Established in the western portion of Lake Superior since about 1988 it has
expanded in an easterly direction. It has now become the dominant species in the St. Louis River
estuary (Leigh, 1998). Based on bottom trawl samples, ruffe make up an estimated 80% offish
abundances in the southwestern regions of Lake Superior (Leigh, 1998). Ruffe exhibit rapid
growth and high reproductive output, and adapt to a wide range of habitat types; therefore, the
species may pose a threat to native North American fish. There is much concern that ruffe may
have a detrimental effect on the more desirable species in Lake Superior, such as yellow perch
and walleye, by feeding on the young of these species or by competing for food (Fuller and
Jacobs, 2008). Ruffe are often associated with bottom waters and can tolerate lacustrine and
lotic systems and depths to 85 m (Sandlund et al., 1985). The species intolerance to deeper
waters may limit its range of potential suitable habitat to Lake Erie, southern Lake Michigan, and
the shallower waters of the other Great Lakes.
39
-------
Ruffe (Gymnocephalus cernuus)
Predicted
Suitability
^m High (100)
Mid (50)
^m Low (0)
United
States
Figure 13. GARP-predicted habitat suitability of ruffe (Gymnocephalus
cernuus) in the Great Lakes. Inset map shows the locations where the species
has been reported.
4.2.7. Monkey Goby
The GARP model predicts that the monkey goby, a member of the goby fish family,
could find suitable habitat in most of Lake Erie and in some portions of Lake Ontario and Lake
Huron (Figure 14). Predictions cannot be made for most of the Great Lakes because of data
limitations. The monkey goby is closely related to the round goby. Currently, the monkey goby
is confined to Eurasia but it has traveled up the Danube, Dnieper, and Volga Rivers from its
native waterways and is becoming an invasive nuisance in these areas. Similar to other
Gobiidaes, the monkey goby prefers shallow water and would likely not survive in deeper
waters.
4.2.8. New Zealand Mud Snail
The New Zealand mud snail, another mollusk, is predicted by GARP modeling to find
suitable habitat in most if not all of Lakes, Erie, Ontario, and Michigan (Figure 15) and
shorelines of Lakes Huron and Superior. It was first established in Lake Ontario in 1991
(Zaranko et al., 1997) and in Lake Erie in 2005 (Levri et al., 2007). It may also be established in
Lake Superior, where some individuals were found in 2001 (Grigorovich et al., 2003b). Mud
40
-------
Monkey goby (Neogobius fluviatilis)
Predicted
Suitability
• High (100)
Mid (50)
• Low (0)
No Prediction
Possible
United
States
Figure 14. GARP-predicted habitat suitability of monkey goby (Neogobius
fluviatitis) in the Great Lakes.
New Zealand mud snail (Potamopyrgus antipodarum
Predicted
Suitability
^ High (100)
Mid (50)
• Low (0)
United
States
Figure 15. GARP predicted habitat suitability of New Zealand mud snail
(Potamopyrgus antipodarum) in the Great Lakes. Inset map shows the
locations where the species has been reported.
41
-------
snail populations consist mostly of asexually reproducing females that are born with developing
embryos in their reproductive system. This species can be found in all types of aquatic habitats
from eutrophic mud bottom ponds to clear rocky streams. It can tolerate a wide range of water
temperatures (except freezing), salinity, and turbidity in clean as well as degraded waters. They
feed on dead and dying plant and animal material, algae, and bacteria. It can tolerate a broad
range of ecological factors thus facilitating its further spread. In moist conditions, this snail can
withstand short periods of desiccation. Since this snail is found at depths from 5 to 45 m (Levri
et al., 2007) it is unlikely the species will survive in deeper waters.
4.2.9. Tubenose Goby
The tubenose goby, another member of the goby fish family, is predicted by the GARP
model to become established in Lake Erie and the shoreline areas of the other Great Lakes
(Figure 16). Predictions could not be made for most of the rest of the region. Their distribution
around the inshore areas of the Black and Caspian Seas indicates their potential for widespread
occupation of inshore habitats where cover, especially plants, occurs in the lower Great Lakes
(Jude et al., 1992). Tubenose gobies have been shown to have a significant overlap in diet
preference with rainbow darters, Etheostoma caeruleum, and may compete with these native fish
for food (French and Jude, 2001). The usual habitat for this species is shallow bays, offshore
banks, or flowing water of streams. However, the tubenose goby also can be found in ponds and
canals overgrown with vegetation. Where current is strong, it hides under boulders. It is often
found under stones or among weeds, to which it retreats rapidly if disturbed. Some individuals
can be found at depths greater than 3 m in the sea. The preferred conditions probably restrict its
probable range of suitable habitat to shallower waters.
4.2.10. Rudd
Already occurring in the Great Lakes with an unknown frequency at this time, significant
portions of Lakes Erie and Ontario as well as portions of Superior and Michigan are prone to
invasion by the rudd, a medium-sized, thick-bodied fish (Figure 17). The rudd's tolerance of a
variety of habitats has likely contributed to its widespread distribution. In streams and rivers,
this fish usually prefers long, slow pools and backwaters. The rudd can be expected to compete
for invertebrate food sources with native fishes. In addition, being omnivorous, the rudd can
shift its diet to plants, unlike most native fishes. Because rudd are fairly hardy, Nico et al. (2008)
indicate that the fish will fare better than many native fishes in waters that are eutrophic or
42
-------
Tubenose goby (Proterorhinus marmoratus)
Can a da
Predicted
Suitability
•I High (100)
Mid (50)
• Low (0)
No Prediction
Possible
Figure 16. GARP-predicted habitat suitability of tubenose goby
(Proterorhinus marmoratus) in the Great Lakes.
Rudd (Scardinius erythrophthalmus)
Canada .-»-,
Predicted
Suitability
• High (100)
Mid (50)
• Low (0)
No Prediction
Possible
United
States
Figure 17. GARP-predicted habitat suitability of rudd (Scardinius
erythrophthalmus) in the Great Lakes.
43
-------
polluted. Predictions cannot be made about the habitat suitability for rudd in parts of Lake
Superior, but, given the species preference for littoral waters, it is unlikely the rudd would find
suitable habitat in the deeper regions of all the lakes.
4.2.11. Corophium Curvispinum (an Amphipod)
According to the GARP model, almost all of Lake Erie and the southern shores of Lakes
Ontario, Huron, and Michigan are prone to invasion by the amphipod Corophium curvispinum.
Predictions for the other locations in the Great Lakes were not possible due to limited data
(Figure 18). This amphipod builds tubes on firm surfaces such as rocks, wood, submerged
vegetation, or bivalve shells on otherwise sandy or muddy substrata in shallow waters
(Frammandearter, 2008). C. curvispinum prefers rivers, estuaries, and other areas with brackish
water, but it can also tolerate freshwater environments.
Corophium curvispinum
Canada
Predicted
Suitability
~ High (100)
Mid (50)
Low (0)
United
States
Figure 18. GARP-predicted habitat suitability of Corophium curvispinum
(no common name reported) in the Great Lakes.
44
-------
4.2.12. Sand Goby
The sand goby is predicted by GARP modeling to find suitable habitat almost
everywhere in all five Great Lakes (Figure 19). This occasionally schooling species occurs
primarily in inshore sandy and muddy areas (Froese and Pauly, 2008). The sand goby is a
coastal goby of European waters from the Baltic to the Mediterranean Sea and can grow up to
10 cm in length. Some variation from the GARP modeling prediction is expected because
similar to other gobys, the sand goby is unlikely to find suitable habitat in deeper waters.
Sand goby (Potamoschistus minutus)
Canada
Predicted
Suitability
^^ High (100)
Mid (50)
^m Low (0)
No Prediction
Possible
Figure 19. GARP-predicted habitat suitability of sand goby (Potamoschistus
minutus) in the Great Lakes.
45
-------
4.2.13. Roach
GARP predicts that Lakes Erie and Ontario would provide suitable habitat for the roach,
a medium-sized fish in the carp family. Most of the other regions would be unsuitable (Figure
20) although predictions cannot be made about the suitability of habitat of parts of Lake
Superior. Adults inhabit slow-flowing or still muddy waters and are abundant in their native
rivers, lakes, canals, and reservoirs. Brackish water populations in the Baltic and the Black Sea
are anadromous and they are known to thrive in poor quality, even polluted water (Nico and
Fuller, 2008). As omnivores, they feed on insects, crustaceans, mollusks, and plants.
Roach (Rutilus rutilus)
Canada *•**
Predicted
Suitability
^* High (100)
Mid (50)
•• Low (0)
No Prediction
Possible
United
States
Figure 20. GARP-predicted habitat suitability of roach (Rutilus rutilus) in
the Great Lakes.
4.2.14. Tench
The tench, a medium-sized fish already established in many rivers within the United
States, is likely to find suitable habitat in most of Lake Erie and small portions of Lake Ontario
and Lake Huron (Figure 21) according to GARP models. The diet consists mainly of aquatic
insect larvae and mollusks. Nico and Fuller (2008) considered it a potential competitor for food
with sport fishes and native cyprinids and noted that the species is known to stir up bottom
sediments, possibly affecting water quality, similar to the common carp.
46
-------
Tench (Tinea tinea)
Canada
Predicted
Suitability
^m High (100)
Mid (50)
^ Low (0)
No Prediction
Possible
United
States
Figure 21. GARP-predicted habitat suitability of tench (Tinea tinea) in the
Great Lakes.
4.3. VESSEL TRAFFIC AND GREAT LAKES PORTS
The U.S. Great Lakes receive substantial vessel traffic from around the world because of
the commodities shipped in and out of the area. The second phase of this study was to better
understand whether there is sufficient numbers of propagules (e.g., larvae, seeds, spores, adults)
entering the Great Lakes for species to become established. As described in the methods section,
we used vessel traffic and ballast water discharges as a surrogate for propagule pressure since no
data exists that actually measures the number of propagules released from discharges. By
analyzing ballast water discharges, we were able to identify those ports that are at greatest risk.
4.3.1. Analysis of Vessels With Ballast on Board (BOB)
In accordance with U.S. Coast Guard (USCG) regulations, ballast water from
transoceanic vessels with ballast on board is exchanged at sea prior to entering the St. Lawrence
Seaway. Despite the ballast water exchange (BWE), some ballast water and residue may remain
and NTS may survive in the ballast tank and then potentially be released when the ballast water is
discharged. We evaluated ballast water discharges (BWD) into U.S. Great Lakes ports from
vessels entering the Seaway whose original source of ballast water (i.e., the ballast in the tank
47
-------
prior to open ocean exchange) was taken-on from areas outside of the Great Lakes. To interpret
these results one must consider that transoceanic vessels carry multiple ballast tanks and each
tank may have a different history of ballast water source, exchange, and discharge. Therefore,
each ballast tank discharge was counted as a BWD. A transoceanic vessel may carry over
20,000 metric tons of ballast water and have as many as 20 ballast tanks, implying one
vessel-trip could have up to 20 ballast discharges at any one Great Lakes port.
Our study found that in the period of 2006-2007, 618 ballast tanks and 382 thousand
metric tons of ballast water were discharged at Great Lakes ports from 107 different vessels.
From a global perspective, the BWDs that were evaluated and could be linked to a particular port
usually came from the eastern and western areas of the northern Atlantic Ocean (Figure 22). The
Gulf of St. Lawrence region, near the St. Lawrence Seaway was the original source of ballast
water for over 1/3 of the 618 discharges (Figure 23). Western Europe was the second-most
common source, with most of the ballast originating from the southeastern portion of the North
Sea (Figure 24).
Fifty-eight different foreign ports provided the original source of ballast water ultimately
discharged at Great Lakes ports. Figure 25 identifies the most important ports based on the
number of vessels leaving these ports and entering the Great Lakes. However, the most common
source of ballast water was obtained while in transit and not at any particular port of call. The
ports of Antwerpen, Belgium; Puerto Cabello, Venezuela; Haraholmen, Sweden; and Bremen,
Germany are responsible for the greatest number of discharges, not including those ports near
the entrance to the St. Lawrence Seaway (Figure 25). These ports, however, are not necessarily
the source of the greatest volume of BWD. For instance, Haraholmen was ranked fourth among
nonNorth American ports in terms of number of tanks discharged, but was ranked eighth in
terms of metric tons of BWD (Figure 25 and Appendix E, Table E-l). In this study it was rare to
find more than one vessel originating from the 58 foreign ports (Appendix E, Table E-l). Only
six ports from outside North America were the source of two or more vessels included in this
analysis.
Duluth received more than twice the BWDs and twice the volume of ballast water as any
other Great Lake port in 2006-2007 (Figure 26). The ports of Toledo, Superior, Green Bay,
Gary, and Milwaukee, also received over 10,000 metric tons of ballast water (Figure 26 and
Appendix E, Table E-2). Appendix E, Table E-3 provides detailed information for those vessels
with ballast on board discharging to Great Lakes ports in 2006-2007.
48
-------
Number of Ballast Water Discharges
o 10 10°
Figure 22. Location of the original source of ballast water taken-on prior to
ballast water exchange and discharge in the U.S. Great Lakes. The area of
each green circle is proportional to the number of ballast tank discharges.
Number of Ballast
Water Discharges
Atlantic Ocean
Figure 23. Location of the source of ballast water taken-on from Canadian
ports in or near the Gulf of St. Lawrence prior to ballast water exchange
and discharges in the U.S. Great Lakes. The area of each green circle is
proportional to the number of ballast tank discharges.
49
-------
Number of Ballast
Water Discharges
Figure 24. Location of the source of ballast water taken-on from European
ports prior to ballast water exchange and discharges in the U.S. Great
Lakes. The area of each green circle is proportional to the number of ballast
tank discharges.
80
60
40
D Ballast Tanks Discharged
DVolume Discharged
40,000
1M
90,000
80,000
.fc
0>
30,000 g.
10,000 i
m
St. Lawrence Estuary / Gulf of St. Lawrence
Figure 25. Frequency, volume, and original source of ballast water (prior to
ballast water exchange) discharged into U.S. Great Lakes ports, from
sources outside the Great Lakes.
50
-------
450
400
^•—
|100
re
.c
u
w
Q
c
"
re
m 50
1
It
L
~
D Ballast Tanks Discharged
• Volume of Ballast Water Discharged
200000
„ n
x s // V / ^
? «?<• <& & J1 3?
(S" v A N^ V
^J» s? 0<^ ^J9
180,000 =
u
80,000 -iT
S1
re
Ł
i/i
a
Ł
i
tfl
40,000 =
m
3
Volume of
Figure 26. Frequency and volume of ballast water discharges (after ballast
water exchange at sea) from ballast on board vessels, when the original
source of ballast water came from outside the Great Lakes.
4.3.2. Analysis of Vessels with No Ballast on Board (NOBOB)
Some vessels enter the St. Lawrence Seaway with no ballast on board but may have
organisms that remain and survive in the residual material left in the ballast tanks, and are
referred to as NOBOB-RM vessels. These NOBOB-RM vessels can then take-on ballast water
in the Seaway (most likely when cargo is off-loaded) and later discharge the ballast water along
with residual materials at a Great Lakes port. We combined the 2006 NVMC data with the 2006
NBIC data to identify these types of vessels. There were 1,730 discharges of ballast water at
Great Lakes ports from NOBOB-RM vessels in 2006. This is substantially more than the
discharges from vessels with ballast on board, supporting the notion that NOBOB vessels may
pose a much greater risk. The distribution of the potential sources of ballast water from
NOBOB-RM vessels is somewhat similar to the vessels with ballast on board. Over half of the
last five ports of call by these vessels were in southeastern Canada with Western Europe the
second most common source of ballast water (Figure 27 and Appendix E, Table E-4).
Some of the foreign ports of origin are different between BOB and NOBOB vessels. For
example, ten NOBOB-RM vessels included a stop at Europa Point, Gibraltar as one of the last
five ports of call before discharging ballast water into the Great Lakes. Other vessels stopped at
51
-------
Riga, Latvia; and Santos, Brazil (Figure 27 and Appendix E, Table E-4). Yet, we did not find
any vessels with ballast on board stopping at these ports. Some ports are visited by both types of
vessels, especially Sept lies, Canada; and Ijmuiden, Netherlands (Figure 27 and Appendix E,
Table E-4).
Several of the Great Lakes ports, including Duluth, Toledo, and Superior, receive ballast
discharges from both NOBOB-RM vessels and vessels with ballast on board. Most Great Lakes
ports received far more ballast discharges from NOBOB-RM vessels than BOB vessels in 2006.
Several ports receive most of their ballast water from NOBOB-RM vessels, including Sandusky,
Conneaut, Buffalo, and Calumet as shown in Figure 28 and Appendix E, Table E-5. Ashtabula
was the extreme case, receiving 297 discharges from NOBOB-RM vessels (Figure 28) and only
one discharge from a BOB vessel (Figure 26). When both vessel types are considered, the Great
Lakes ports at greatest risk of receiving sufficient propagule pressure to facilitate invasion are
Duluth, MN; Superior and Milwaukee, WI; Toledo, Ashtabula, and Sandusky, OH; Gary, IN;
and Chicago, IL.
s 16°]
o
•5
fi
o
Q.
Ł
ig 120
"tn
_re
ips from specified port withi
Ł §
T
-------
400
300
E?
TO
« 200
n
o
TO
100
1 Ballast Tanks Discharged
I Volume of Ballast Water Discharged
(I
r.
700,000
600,000
In
a
500,000 o
400,000 ,
-------
5. DISCUSSION
The Great Lakes system has been adversely affected by invasive species. Preventing the
transport of these species to the Great Lakes from outside the system is the best way to avert
potential ecological and economic impacts. Our analysis of ballast water discharges using vessel
traffic data, evaluating similar habitats using the Genetic Algorithm for Rule-Set Production
(GARP) niche model, and a literature review indicate that invasions are likely to occur over the
next decade or so. If it is not possible to eliminate the transport of nonindigenous species (NTS)
to the Great Lakes, the next best alternative is to monitor for the arrival of potentially invasive
species and to control their spread as soon as they arrive.
Since we began our investigation, additional ballast-water control measures have been
implemented. Beginning with the 2008 navigation season, all vessels must either undergo ballast
water exchange (BWE) or flushing before entering the St. Lawrence Seaway (73 FR 37,
p. 9,950), even those vessels that heretofore were declared as having no ballast on board.
However, even with the more extensive requirements, additional NTS may still reach the Great
Lakes. Some saltwater tolerant species may survive the BWE or flushing, and other vectors
(e.g., hull fouling organisms) continue to pose a threat. This report provides information that
may help resource managers prioritize monitoring efforts by identifying potential invaders and
ports at risk.
The National Research Council recommends that a binational (United States and Canada)
science-based surveillance program be established to monitor for aquatic NTS and that the
program involve dedicated lake teams, as well as academic researchers, resource managers, and
local citizens groups (NRC, 2008). Since early detection and rapid response is a priority of the
National Invasive Species Council (NISC, 2007), the ports and species we identified could be
used to structure an early warning and detection system to help evaluate the effectiveness of
ballast exchange regulations and practices.
5.1. PREDICTING THE SPREAD OF SPECIES
This study identified 30 potentially invasive species with medium or high risk for
spreading to the Great Lakes and causing ecological impacts and another 28 potentially invasive
species that have already become established in one or more of the Great Lakes (see
Appendix B). Habitat suitability maps are provided along with a summary of invasion potential
for 14 modeled species. All of the modeled species are predicted to have the capability to
colonize Lake Erie and the shallower waters of the other lakes. Several species may be able to
colonize the entire Great Lakes region. Literature regarding the species environmental
54
-------
tolerances reveals that the predominant limitation to the spread of several modeled species is
their tolerance to water depth. However, managers need to recognize that when NTS are
transported to a new environment, species-tolerance "surprises" can occur. For instance, the
quagga mussel has been found at deeper depths in the Great Lakes than in its native range
(Watkins et al., 2007).
Table 3 summarizes the habitat suitability and current status for the 14 modeled species.
The modeled species are categorized into two groups: (1) NTS already established in the Great
Lakes and having the potential to spread to at least parts of all five lakes; and (2) NTS, not yet
established but with the potential to invade the Great Lakes.
5.2. POTENTIAL MONITORING SITES BASED ON VESSEL TRAFFIC
The source of most of the ballast water discharged into the Great Lakes came from
58 different ports located predominantly in Canada and Western Europe, thereby complicating
surveillance programs. If just a few foreign ports were the original source of ballast water (prior
to exchange) then programs could focus on species found at those foreign ports. The six ports
which received the most ballast water from vessels with ballast on board in 2006-2007, in rank
order, were Duluth, MN; Toledo, OH; Superior, WI; Green Bay, WI; Gary, IN; and Milwaukee,
WI (see Figure 26 and Appendix E, Table E-2). The first three ports, Duluth, Toledo, and
Superior, account for 86% of the total volume of ballast water discharged into the Great Lakes.
There was no evidence of a frequent, repeated connection from any specific foreign port to a
specific port within the Great Lakes. For instance, 11 different vessels discharged ballast water
in Toledo in 2006-2007 (see Appendix E, Table E-3). If all 11 vessels obtained ballast water
from a single foreign port than monitoring could be targeted for those species occurring at that
particular port. Unfortunately, the 11 vessels discharging ballast water in Toledo took-on ballast
from 10 different foreign ports.
Invasive species can also be transported to the Great Lakes via vessels with no ballast on
board but with residue left in the tanks (NOBOB-RM vessels). A different set of foreign ports
were found to be the source of ballast water from these vessels (see Figure 27), although ports in
Canada and Western Europe predominated. Consistent with the vessels with ballast on board, it
was rare to find a frequent connection between particular Great Lakes ports and foreign ports.
The ports receiving the greatest volume of ballast water from NOBOB-RM vessels are Duluth,
MN, Toledo, OH, and Superior, WI accounting for 54% of the total volume discharged from
these vessels in 2006-2007. Ashtabula and Sandusky, OH receive 32% of the ballast water
released from NOBOB-RM vessels.
55
-------
Table 3. Composite results for 14 species modeled using GARP
Species/common name
Summary of invasion potential
Species already established in the Great Lakes and potential for spread to all five Great Lakes
Neogobius melanostromus —
round goby
Potamopyrgus
antipodarum — New Zealand
mud snail
Dreissena bugensis — quagga
mussel
Dreissena polymorpha —
zebra mussel
Gymnocephalus cernuus —
ruffe
Already spread to all five Great Lakes, with large populations in Lakes Erie
and Ontario. Likely to find suitable habitat throughout Lake Erie and in all
Great Lakes waters at depths less than 60 m.
Already occurs in isolated areas of Lakes Erie, Ontario, and Superior.
Likely to find all shallower waters (<50 m depth) as suitable habitat. High
spread potential.
Already found in all five Great Lakes, with large populations established in
Lakes Erie and Ontario. The only possible identified limitation for spread
is a species depth limitation which is questionable and currently appears to
be as deep as 200 m.
Already occurs in all five Great Lakes. Likely to find suitable habitat in
most of Lake Erie and portions of other lakes where water depth is less than
60 m. May be outcompeted by the quagga mussel.
Already found in Lakes Superior, Michigan, and Huron. The species is
probably capable of colonizing most areas within the Great Lakes where
water depth is less than 85 m.
Species that may Invade at least parts of all five Great Lakes
Alosa aestivalis — blueback
herring
Pomatoschistus minutus —
sand goby
Rutilus rutilus — roach
Scardinius
erythrophthalmus — rudd
Cercopagis pengoi —
fishhook waterflea
Tinea tinea — tench
Proterorhinus
marmoratus — tubenose goby
Corophium curvispinum —
(an amphipod)
Neogobius fluviatilis —
monkey goby
Models predict it could find the entire region as suitable habitat, except
possibly the deeper waters of Lake Superior.
It is likely this species would find all shallower waters as suitable habitat.
Already reported in Lakes Erie and Ontario. Predicted to find suitable
habitat throughout these lakes, and probably into other shoreline areas.
Predicted to find suitable habitat throughout Lake Erie and into the
shallower waters of the other four Great Lakes.
Established in Lake Ontario and reported in Lakes Erie and Michigan.
Predicted to find suitable habitat throughout the region. Densities increase
in deeper waters.
Found currently in St. Lawrence River. Potential to spread to shallower
waters of most of Lake Erie, and to isolated portions of the other Lakes.
Tench can spread rapidly once established.
Already reported as present in Lake St. Clair and western Lake Erie. May
be able to occupy all shallow waters of all five Great Lakes.
Capable of invading Lake Erie and shallower waters. Not enough data is
available to predict if it can find suitable habitat elsewhere.
May be capable of inhabiting shallower waters of all five Great Lakes.
56
-------
Since it is unknown which type of vessel (ballast on board [BOB] or NOBOB-RM) is
more likely to transport NTS, ports receiving ballast water from either type of vessel are
presumed to be at risk. In order to recommend ports for monitoring we cannot just consider the
transport potential, we also need to consider the potential to find suitable habitat. The results of
GARP modeling and the literature review reveal that Lake Erie and shallower portions of the
other Lakes provide the most favorable habitat for the modeled species, and that the deeper
portions of Lake Superior are less hospitable to species invasions (Grigorovich et al., 2003b).
However, the shallower portions of Lake Superior, especially the Duluth-Superior harbor, are at
greater risk for invasion.
Assuming the observed vessel traffic and ballast-water discharge information for 2006
and 2007 is representative, the port of greatest concern for receiving sufficient propagules and
providing the most suitable habitat is Toledo, OH. Toledo is located on Lake Erie, a region that
the GARP model predicted would have a high chance of providing suitable habitat for the
modeled species. Other ports of elevated concern for receiving sufficient propagules and
offering suitable habitat are Gary, IN; Milwaukee, WI; Chicago, IL; and Ashtabula and
Sandusky, OH. Ports with high transport potential but generally low habitat suitability are
Duluth, MN and Superior, WI. The spread of invasive species from beyond the Duluth-Superior
harbor may be limited by the colder and deeper waters in the main portion of Lake Superior.
Yet, since inter-lake transport can occur to other ports, Duluth and Superior also warrant a
monitoring program. Managers may wish to emphasize detection programs at the ports of
concern that were identified and may wish to focus on the list of 58 potentially invasive species
with a moderate or strong chance to invade and cause ecological or economic impacts. For the
14 modeled species, the focus can be narrowed based on the summary shown in Table 3.
Given the new regulations which require all vessels entering the Seaway to undergo
either ballast water exchange or flushing at sea, additional research on the tolerance of invasive
species to saltwater would enable managers and scientists to better focus monitoring activities on
those species that are likely to survive salt water flushing. Subsequent analyses of the NBIC
database is recommended to determine if the 2006-2007 data are indeed representative.
In summary, we have provided a list of NIS of concern, predicted locations that would
provide suitable habitat for 14 modeled species, identified those U.S. Great Lakes ports receiving
the most ballast water from sources originating from outside the Great Lakes, and predicted the
ports most at risk of invasion. Our findings support the need for detection and monitoring efforts
at those ports believed to be at greatest risk. This study also demonstrates the importance of
understanding invasion biology by evaluating the two most important predictors of invasion, as
suggested by Williamson (1996): propagule pressure and suitable habitat. Further, this may be
57
-------
the first time that remote sensing data were used in conjunction with GARP to predict the spread
of aquatic invasive species.
58
-------
6. REFERENCES
Anderson, RP; Laverde, M; Peterson, AT. (2002) Geographical distributions of spiny pocket
mice in South America: Insights from predictive models. Global Ecology and Biogeography
11:131-141.
Anderson, RP; Lew, D; Peterson, AT. (2003) Evaluating predictive models of species
distributions: criteria for selecting optimal models. Ecol Mod 162(3):211-232.
Bailey, SA; Reid, DF; Colautti, RI; et al. (2005) Management options for control of
nonindigenous species in the Great Lakes. J Great Lakes L Sci and Policy 5:101-112.
Buchan, LJ; Padilla, DK. (2000) Predicting the likelihood of Eurasian watermilfoil presence in
lakes, a macrophyte monitoring tool. Ecol Appl 10:1442-1445.
Carlton, JT; Reid, DM; van Leeuwen, H. (1995) Shipping study- the role of shipping in the
introduction of non-indigenous aquatic organisms to the coastal waters of the United States
(other than the Great Lakes) and an analysis of control options. In: The National Biological
Invasions Shipping Study. United States Coast Guard and the National Sea Grant Program,
Washington, D.C., USA.
Claxton, WT; Wilson, AB; Mackie, GL; et al. (1998) A genetic and morphological comparison
of shallow- and deep-water populations of the introduced dreissenid bivalve Dreissena bugensis.
Can J Zool 76(7): 1269-1276.
Cohen, AN. (2007) Potential distribution of zebra mussels (Dreissenapolymorpha) and quagga
mussels (Dreissena bugensis) in California phase 1 report. California Department of Fish and
Game. Available on-line at http://www.dfg.ca.gov/invasives/quaggamussel/docs
/PotentialDistributionZebraQuaggalnCA.pdf (accessed March 2008).
Colautti, RI; Niimi, AJ; Van Overdijk, CD A; et al. (2003) Spatial and temporal analysis of
transoceanic shipping vectors to the Great Lakes. In: Ruiz, GM; Carlton, JT; eds. Invasive
species: vectors and management strategies. Washington, DC: Island Press; pp. 227-246.
Costello, C; Drake, J; Lodge, D. (2007a) Evaluating an invasive species policy: ballast water
exchange in the Great Lakes. Ecol Appl 17(3):655-662.
Cox, GW. (2004) Alien species and evolution: the evolutionary ecology of exotic plants,
animals, microbes, and interacting native species. Washington, DC: Island Press.
Grossman, EJ; Holm, E; Cholmondeley, R; et al. (1992) First record for Canada of the rudd,
Scardinius erythrophthalmus, and notes on the introduced round goby, Neogobius melanostomus.
Canadian Field-Naturalist 106(2):206-209.
59
-------
Drake, JM; Bossenbroek, JM. (2004) The potential distribution of zebra mussels in the United
States. Biol Sci 54:931-941.
Duggan, 1C; Bailey, SA; Colautti, RI; et al. (2003) Biological invasions in Lake Ontario: past,
present and future. In: Munawar, M.., ed. State of Lake Ontario (SOLO) - past, present and
future. Ecovision World Monogr. AquatEcosyst Health Manage, pp. 1-17.
Duggan, 1C; Van Overdijk, DA; Bailey, SA; et al. (2005) Invertebrates associated with residual
ballast water and sediments of cargo-carrying ships entering the Great Lakes. Can J Fish Aquat
Sci 62(11):2463-2474.
Dumont, HJ. (1998) The Caspian Lake: history, biota, structure, and function. Limnol Oceanogr
43:44-52.
Elton, CS. (1958) The ecology of invasions by animals and plants. London, UK: Metheun and
Co., Ltd., 181 pp.
Feria, TP; Peterson, AT. (2002) Using point occurrence data and inferential algorithms to predict
local communities of birds. Diversity and Distributions 8:49-56.
Fitzpatrick, MC; Weltzin, FF; Sanders, NJ; et al. (2007) The biogeography of prediction error:
Why does the introduced range of the fire ant over-predict its native range? Global Ecol and
Biogeogr 16:24-33.
Frammandearter. (2008) List of introduced marine species (Sweden): Caspian mud shrimp
(Corophium curvispinum). Available online at
http://www.frammandearter.Se/0/2english/pdf/Corophium curvispinum.pdf (accessed April
2008).
French, JRP, III; Jude, DJ. (2001) Diets and diet overlap of nonindigenous gobies and small
benthic native fishes co-inhabiting the St. Clair River, Michigan. J. Great Lakes Res
27(3):300-311.
Froese, R; Pauly, D. Editors. (2007) FishBase version 11/2007: Coitus gobio, Bullhead.
Available online at http://www.fishbase.org/Summarv/SpeciesSummary.php?id=2439 (accessed
December 19, 2007).
Fuller, P; Jacobs, G. (2008) Gymnocephalus cernuus. USGS Nonindigenous Aquatic Species
Database, Gainesville, FL. Available online at
(revision date: 7/2/2007).
GLBWWG (Great Lakes Ballast Water Working Group). (2008). 2007 summary of Great Lakes
ballast water management. Available on-line at
http://www.greatlakes-seaway.com/en/pdf/Summ_Ballast_Exams_Report_2007.pdf (accessed
July 23, 2008).
60
-------
Grigorovich, IA; Colautti, RI; Mills, EL; et al. (2003a) Ballast-mediated animal introductions in
the Laurentian Great Lakes: retrospective and prospective analyses. Can J Fish Aquat Sci
60:740-756.
Grigorovich, IA; Korniushin, AV; Gray, DK; et al. (2003b). Lake Superior: an invasion
coldspot? Hydrobiol 499(1): 191-210.
Hayes, KR; Hewitt, CL. (2000) Risk assessment framework for ballast water introductions -
Volume II. CRIMP Technical Report 21, Centre for Research on Introduced Marine Pests,
CSIRO Marine Research, Hobart, Australia.
Heikkinen, RK; Luoto, M; Araujo, MB; et al. (2006) Methods and uncertainties in bioclimatic
envelope modelling under climate change. Prog Phys Geogr 30:751-777.
Herborg, LM; Jerde, CL; Lodge, DM; et al. (2007) Predicting invasion risk using measures of
introduction effort risk and environmental niche models. Ecol Appl 17(3):663-674.
Holeck, KT; Mills, EL; Maclsaac, HJ; et al. (2004) Bridging troubled waters: biological
invasions, transoceanic shipping, and the Laurentian Great Lakes. Biol Sci 54(10):919-929.
Holeck, KT; Mills, EL; Maclsaac, HJ; et al. (2005) When did the discovery rate for invasive
species in the North American Great Lakes accelerate? Biol Sci 55:4-5.
Holland, J. (1975) Adaptation in natural and artificial systems. Ann Arbor, MI: The University
of Michigan Press.
Holland, J. (1992) Genetic algorithms. Scientific American 267(l):66-72.
IAGLR (International Association for Great Lakes Research). (2002) Research and management
priorities for aquatic invasive species in the Great Lakes. Available online at
http://iaglr.org/scipolicy/ais/index.php (Revision date November 15, 2002).
IPCC (Intergovernmental panel on climate change). (2007) Climate change 2007 - Impacts,
adaptation and vulnerability. Contribution of Working Group II to the Fourth Assessment
Report of the IPCC.
IUCN (International Union for Conservation of Nature). (2006) Global invasive species
database: Cercopagispengoi (crustacean). Available online at
http://www.issg.org/database/species/ecology.asp?fr=l&si=l 18 (accessed February 2008).
Johengen, T; Reid, D; Fahnenstiel, G; et al. (2005) Assessment of transoceanic NOBOB vessels
and low-salinity ballast water as vectors for non-indigenous species introductions to the Great
Lakes NOAA. Great Lakes Environmental Research Laboratory and University of Michigan,
Cooperative Institute for Limnology and Ecosystems Research, Ann Arbor, pp. 287.
61
-------
Jude, DJ; Reider, RH; Smith, GR. (1992) Establishment of Gobiidae in the Great Lakes basin.
Can J Fish Aquat Sci 49:416-421.
Kerr, SJ; Brousseau, CS; Muschett, M. (2005) Invasive aquatic species in Ontario: a review and
analysis of potential pathways. Fish 30:21-30.
Kluza, DA; Vieglais, DA; Andreasen, JK; et al. (2007) Sudden oak death: geographic risk
estimates and predictions of origins. Plant Pathol 56:580-587.
Kolar, CS; Lodge, DM. (2001) Progress in invasion biology: predicting invaders. TRENDS in
EcolEvol 16(4): 199-204.
Leigh, P. (1998) Benefits and costs of the ruffe control program for the Great Lakes fishery. J
Great Lakes Res 24(2):351-360.
Leppakoski, E; Gollasch, S; Gruszka, P; et al. (2002) The Baltic - a sea of invaders. Can J Fish
Aquat Sci 59(7): 1175-1188.
Leung, B; Mandrak, NE. (2007) The risk of establishment of aquatic invasive species: joining
invisibility and propagule pressure. Proc R Soc Bull 274:2603-2609.
Levri, EP; Kelly, AA; Love, E. (2007) The invasive New Zealand Mud Snail
(Potamopyrgus antipodarum) in Lake Erie. JGreat Lakes Res 33:1-6.
Lockwood, JL; Cassey, P; Blackburn, T. (2005) The role of propagule pressure in explaining
species invasions. TRENDS in Ecol Evol 20(5):223-228.
Maclsaac, HJ. (1994) Comparative growth and survival of Dreissenapolymorpha andDreissena
bugensis, exotic mollusks introduced to the Great Lakes. J Great Lakes Res 20(4):783-790.
Maclsaac, HJ; Colautti, RI; Niimi, AJ; et al. (2001) Spatial and temporal analysis of transoceanic
shipping vectors to the Great Lakes. Proceedings of the Second International Conference on
Marine Bioinvasions, April 9-11, New Orleans, LA.
Maclsaac, HJ; Robbins, TC; Lewis, MA. (2002) Modeling ships' ballast water as invasion
threats to the Great Lakes. Can J Fish Aquat Sci 59:1245-1256.
Mack, RN. (1996) Predicting the identity and fate of plant invaders: emergent and emerging
approaches. Biol Conserv 78(1-2):107-121.
Marchetti, MP; Light, T; Moyle, PB; et al. (2004) Fish invasions in California watersheds:
testing hypotheses using landscape patterns. Ecol Appl 14:1507-1525.
Marsden, JE; Jude, DJ. (1995) Round gobies invade North America. Fact sheet produced by Sea
Grant at Ohio State University, Columbus, OH.
62
-------
Melina, E; Rasmussen, JB. (1994) Patterns in the distribution and abundance of zebra mussel
(Dreissenapolymorphd) in rivers and lakes in relation to substrate and other physicochemical
factors. Can J Fish Aquat Sci 51:1024-1036.
Menges, ES. (1998) Evaluating extinction risks in plant populations. In: Fielder P.L; Karieva,
PM; ed. Conservation biology. New York, NY: Chapman and Hall; pp 49-65.
Menges, ES. (2000) Population viability analyses in plants: challenges and opportunities. Trends
in Ecology and Evolution 15:51-56.
MDEQ (Michigan Department of Environmental Quality). (2008) Ballast water control general
permit. Available online at http://www.michigan.gov/deq/0,1607,7-135-3313_3682_3713-
153446-,00.html (accessed August 7, 2008).
Mills, EL; Leach, JH; Carlton, JT; et al. (1993) Exotic species in the Great Lakes: a history of
biotic crises and anthropogenic introductions. J Great Lakes Res 19(1): 1-54.
Mills, EL; Leach, JH; Carlton, JT; et al. (1994) Exotic species and the integrity of the Great
Lakes: lessons from the past. Biol Sci 44(10):666-676.
Mills, EL; Rosenberg, G; Spidle, AP; et al. (1996) A review of the biology and ecology of the
quagga mussel (Dreissena bugensis), a second species of freshwater Dreissenid introduced to
North America. Amer Zool 36:271-286.
Mills, EL; Chrisman, JR; Baldwin, B; et al. (1999) Changes in the Dressenid community in the
lower Great Lakes with emphasis on Southern Lake Ontario. J Great Lakes Res 25(1): 187-197.
National Invasive Species Council. (2007) 2008-2012 National invasive species management
plan, draft for public comment, December 20, 2007. Available online at
http://www.invasivespeciesinfo.gov/council/mpdraft07.pdf (accessed April 2008).
NBIC (National Ballast Information Clearinghouse). (2008) NBIC Online Database. Electronic
publication, Smithsonian Environmental Research Center & United States Coast Guard.
Available online at http://invasions.si.edu/nbic/search.html. (accessed July 2008).
Nico, L; Fuller, P; Jacobs, G. (2008) Scardinius erythrophthalmus. USGS Nonindigenous
Aquatic Species Database, Gainesville, FL. Available online at
http://nas.er.usgs.gov/queries/FactSheet.asp?speciesID=648 (revision date: 8/21/2007).
Nico, L; Fuller, P. (2008) Tinea tinea. USGS Nonindigenous Aquatic Species Database,
Gainesville, FL. Available online at http://nas.er.usgs.gov/queries/FactSheet.asp?speciesID=652
(revision date: 4/24/2006).
NOAA (National Oceanic Atmospheric Administration). (2007a) Great Lakes list. National
Center for Research on Aquatic Invasive Species. Available online at
63
-------
http://www.glerl.noaa.gov/res/Progratns/ncrais/docs/great-lakes-list.xls (accessed March 5,
2008).
NOAA (National Oceanic Atmospheric Administration). (2007b) Sea Grant National: Ballast
water technology projects funded by NOAA and FWS. Available online at
http://www.seagrant.noaa.gov/roe/ballastwatertechprojects.html (accessed August 15, 2007).
NOAA (National Oceanic and Atmospheric Administration). (2008) Great Lakes bathymetry.
National Environmental Satellite, Data, and Information Service. Available online at
http://map.ngdc.noaa.gov/website/mgg/greatlakesbathy/viewer.htm (accessed February 2008).
NRC (National Research Council). (2008) Great Lakes shipping, trade, and aquatic invasive
species, Transportation Research Board, Special Report 291. Available online at
http://onlinepubs.trb.org/Onlinepubs/sr/sr291prepub.pdf (accessed July 23, 2008).
O'Neill, CR, Jr. (2004) Biological pollution of the Great Lakes: the nonindigenous aquatic
species invasion of North America's freshwater seas. New York Sea Grant. Available online at
http://www.sgnis.org/publicat/papers/neil2004.pdf (accessed March 2008).
Owens, RW; O'gorman, R; Mills, EL; et al. (1998) Blueback herring (Alosa aestivalis) in Lake
Ontario: First record, entry route, and colonization potential. J Great Lakes Res 24(3):723-730.
Peterson, AT. (2003) Predicting the geography of species' invasions via ecological niche
modeling. Q Rev Biol 78(4):419-433.
Peterson, AT; Cohoon, KP. (1999) Sensitivity of distributional prediction algorithms to
geographic data completeness. Ecol Model 117:159-164.
Peterson, AT; Vieglais, DA. (2001) Predicting species invasions using ecological niche
modeling: new approaches from bioinformatics attack a pressing problem. Biol Sci 51(5):363-
371.
Peterson, AT; Soberon, J; Sanchez-Cordero, V. (1999) Conservatism of ecological niches in
evolutionary time. Science 285:1265-1267.
Peterson, AT; Stockwell, DRB; Kluza, DA. (2002) Distributional prediction based on ecological
niche modeling of primary occurrence data. In: Scott, JM; Heglund, PJ; Morrison, ML; eds.
Predicting species occurrences: issues of scale and accuracy. Washington, DC: Island Press, pp.
617-623.
Pimentel, D; Lach, L; Zuniga, R; et al. (2000) Environmental and economic costs of
nonindigenous species in the United States. Biol Sci 50:53-65.
Pimentel, D. (2005) Aquatic nuisance species in the New York State Canal and Hudson River
Systems and the Great Lakes Basin: an economic and environmental assessment. Environ
Manage 35(5):692-701.
64
-------
Ramcharan, CW; Padilla, DK; Dodson, SI. (1992) Models to predict potential occurrence and
density of the zebra mussel, Dreissenapolymorpha. Can J Fish Aquat Sci 49:2611-2620.
Rasmussen, JL. (2002) The Cal-Sag and Chicago sanitary and ship canal: a perspective on the
spread and control of selected aquatic nuisance fish species. U.S. Fish and Wildlife Service, La
Crosse, Fisheries Resource Office. Available online at
http://www.fws.gov/midwest/LaCrosseFisheries/reports/Connecting_Channels_Paper_Final.pdf
(accessed March 2008).
Reid, DF; Orlova, MI. (2002) Geological and evolutionary underpinnings for the success of
Ponto-Caspian species invasions in the Baltic Sea and North American Great Lakes. Can J Fish
Aquat Sci 59:1144-1158.
Ricciardi, A. (2006) Patterns of invasion of the Laurentian Great Lakes in relation to changes in
vector activity. Divers Distrib 12:425-433.
Ricciardi, A; Maclsaac, HJ. (2000) Recent mass invasion of the North American Great Lakes by
Ponto-Caspian species. TREE 15(2):62-65.
Ruiz, G; Reid, DF. (2007) Current state of understanding about the effectiveness of ballast water
exchange (BWE) in reducing aquatic nonindigenous species (ANS) introductions to the Great
Lakes Basin and Chesapeake Bay, USA: synthesis and analysis of existing information.
National Oceanic and Atmospheric Administration, Technical Memorandum GLERL-142, Great
Lakes Environmental Research Lab, Ann Arbor, Michigan. 127 pp.
Sakai, AK; Allendorf, FW; Holt, JS; et al. (2001) The population biology of invasive species.
Ann Rev Ecol and Syst 32:305-332.
Sala, OE; Chapin, FS, III; Armesto, J; et al. (2000) Global biodiversity scenarios for the year
2100. Science 287:1770-1774.
Sandlund, OT; Naesje, TF; Klyke, L; et al. (1985) The vertical distribution offish species in
Lake Mjosa, Norway as shown by gillnet catches and echo sounder. Inst Freshwater Res —
Drottningholm. 62:136-149.
Scachetti-Pereira, R. (2002) Desktop GARP version 1.1.6. Available online at
http://www.lifemapper.org/desktopgarp/ (accessed April 3, 2006).
Simberloff, D; Von Holle, B. (1999) Positive interactions of nonindigenous species: invasional
meltdown? Biol Inv 1:21-32.
Sing, T; Sander, O; Beerenwinkel N; et al. (2005) ROCR: visualizing classifier performance in
R. Bioinformatics 21:3940-3941.
65
-------
Smith, LD; Wonham, MJ; McCann, LD; et al. (1999) Invasion pressure to a ballast-flooded
estuary and an assessment of inoculant survival. Biol Inv 1:67-87.
Stockwell, DRB; Peters, D. (1999) The GARP modeling system: problems and solutions to
automated spatial prediction. Int J Geogr Inf Syst 13(2): 143-158.
Stockwell, DRB; Peterson, AT. (2002) Effects of sample size on accuracy of species distribution
models. Ecol Model 148(1): 1-13.
Sutherst, RW; Maywald, GF; Yonow, T; et al. (1999) CLEVIEX user guide—predicting the
effects of climate on plants and animals. Collingwood, Victoria, Australia: CSIRO Publishing.
Swets, JA. (1988) Measuring the accuracy of diagnostic systems. Science 240:1285-1293.
Theoharides, KA; Dukes, JS. (2007) Plant invasion across space and time: factors affecting
nonindigenous species success during four stages of invasion. New Phytol 176:256-273.
Torchin, ME; Mitchell, CE. (2004) Parasites, pathogens, and invasions by plants and animals.
Front Ecol Env 2:183-190.
University of Minnesota. (2006) Cercopagis pengoi invades Lake Ontario fact sheet: October
1999. Ontario Federation of Anglers and Hunters, University of Minnesota. Available online at
http://www.seagrant.umn.edu/exotics/fishhook.html (accessed February 2008).
USCG (United States Coast Guard). (2001) Report to Congress on the voluntary national
guidelines for ballast water management. United States Coast Guard Office of Operating and
Environmental Standards. Available online at http://www.USCG.mil/hq/g-m/mso/bwm.htm
(accessed May 1,2007).
U.S. Congress. (1993) Harmful nonindigenous species in the United States. Office of
Technology Assessment, Superintendent of Documents, P.O. Box 371954, Pittsburgh, PA
15250-7974; OTA-F-565 (phone 202-512-1800).
USGS (United States Geological Survey). (2007) Nonindigenous aquatic species database.
Maps, Imagery, and Publication. Available online at http://www.usgs.gov/pubprod/maps.html
(accessed 12/04/07).
USGS (United States Geological Survey). (2008a) Great Lakes Science Center: Round Goby.
Available online at
http://www.gl sc.usgs.gov/main.php?content=research_invasive_goby&title=Invasive%20FishO&
menu=research invasive fish (revision date: January 31, 2008).
USGS (United States Geological Survey). (2008b) Nonindigenous aquatic species: Cercopagis
pengoi. Available online at http://nas.er.usgs.gov/queries/FactSheet.asp?speciesID=163
(revision date: April 25. 2008).
66
-------
USGS (United States Geological Survey). (2008c)NAS nonindigenous aquatic species:
Dreissena polymorpha. Available online at
http://nas.er.usgs.gov/queries/FactSheet.asp?speciesID=5 (Revision Date: 1/18/2008).
Vanderploeg, HA; Liebig, JR; Carmichael, WW; et al. (2001) Zebra mussel (Dreissena
polymorpha) selective filtration promoted toxic microcysts blooms in Saginaw Bay (Lake
Huron) and Lake Erie. Can J Fish Aquat Sci 58:1208-1221.
VA Inland Fisheries (Virginia Department of Inland Fisheries). (2008) Blueback herring.
Available online at http://www.dgif state.va.us/wildlife/fish/details.asp?fish=010045 (accessed
February 2008).
Watkins, JM; Dermott, R; Lozano, SJ; et al. (2007) Evidence for remote effects of Dreissenid
mussels on the amphipod Diporeia: Analysis of Lake Ontario benthic surveys 1972-2003. J
Great Lakes Res 33:642-657.
Wilcove, DS; Rothstein D; Dubow J; et al. (1998) Quantifying threats to imperiled species in the
United States. Biol Sci 48(8):607-615.
Williamson, M. (1996) Biological invasions. Chapman and Hall: London.
Wisconsin DNR (Wisconsin Department of Natural Resources). (2008) Quagga mussel.
Available on-line at http://www.dnr.state.wi.us/org/caer/ce/eek/critter/invert/quaggamussel.htm
(accessed February 2008).
Wisconsin Sea Grant. (2008) Round Goby, fish of the Great Lakes. University of Wisconsin Sea
Grant Institute. Available online at http://www.seagrant.wisc.edu/greatlakesfish/roundgoby.html
(accessed March 2008).
Wonham, MJ; Walton, WC; Ruiz, GM; et al. (2001) Going to the source: role of the invasion
pathway in determining potential invaders. Marine Ecology Progress Series 215:1-12.
WSR (World Shipping Register). (2007) World shipping register: world sea ports. Available
online at http://www.world-register.com/ports.php (accessed February 2, 2007).
Zaranko, DT; Farara, DG; Thompson, FG. (1997) Another exotic mollusk 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.
67
-------
APPENDIX A. LIST OF NONINDIGENOUS SPECIES
THAT HAVE BEEN REPORTED AS OCCURRING IN THE GREAT LAKES
A-l
-------
>
Year of
Invasion
1840
1843
1843
1843
1843
1843
1843
1843
1847
1860
1862
1864
1865
1866
1867
1869
1871
1873
1873
1874
1876
1878
1879
1879
1880
1882
Species
Rumex obtusifolius
Echinochloa crusgalli
Solarium dulcamara
Mentha piperita
Conium maculatum
Poa trivalis
Mentha spicata
Polygonum persicaria
Rorippa nasturtium-aquaticum
Elimia virginica
Juncus gerardii
A/a/as marina
Sonchus arvensis
Carex disticha
Chenopodium glaucum
Lythrum salicaria
Bithynia tentaculata
Alosa pseudoharengus
Oncorhynchus tshawytscha
Epilobium hirsutum
Oncorhynchus mykiss
Carassius auratus
Cyprinus carpio
Potamogeton crispus
Typha angustifolia
Lysimachia nummularia
Common Name
bitter dock
barnyard grass
bittersweet nightshade
peppermint
poison hemlock
rough-stalked
meadow grass
spearmint
lady's thumb
water cress
snail
black-grass rush
spiny naiad
field sow thistle
sedge
oak leaved goose foot
purple loosestrife
faucet snail
alewife
Chinook salmon
great hairy willow herb
rainbow trout
goldfish
common carp
curlyleaf pondweed
narrow leaved cattail
moneywort
Type
Plant
Plant
Plant
Plant
Plant
Plant
Plant
Plant
Plant
Mollusk
Plant
Plant
Plant
Plant
Plant
Plant
Mollusk
Fish
Fish
Plant
Fish
Fish
Fish
Plant
Plant
Plant
Endemic
Region
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Atlantic NA
Atlantic NA
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Atlantic NA
Pacific NA
Eurasia
Pacific NA
Asia
Ponto-Caspian
Eurasia
Eurasia
Eurasia
Location of
First Sighting
Widespread
Widespread
Widespread
Widespread
Widespread
Widespread
Widespread
Widespread
Lake Ontario
Erie Canal
Chicago
Lake Ontario
drainage
Central NY
Lake Ontario
Lake Ontario
drainage
Lake Ontario
Lake Michigan
Lake Ontario
Widespread
Lake Ontario
Lake Huron
drainage
Widespread
Widespread
Lake Ontario
drainage
Lake Ontario
Lake Ontario
Vector
Unknown
Release (deliberate)
Release (deliberate)
Release (deliberate)
Release (deliberate)
Release (deliberate)
Release (deliberate)
Unknown
Release (deliberate)
Canals
Shipping, Solid
Ballast
Shipping, Solid
Ballast
Release
(unintentional)
Shipping, Solid
Ballast
Railroads and
Highways
Canals
Shipping, Solid
Ballast
Canals
Release (deliberate)
Release
(unintentional)
Release (deliberate)
Release
(unintentional)
Release (deliberate)
Release (deliberate)
Canals
Release (deliberate)
-------
>
Year of
Invasion
1882
1883
1884
1884
1886
1886
1886
1886
1886
1892
1893
1894
1895
1895
1896
1897
1897
1901
1901
1902
1902
1903
1905
1906
1912
Species
Alopecurus geniculatus
Salmo trutta
Agrostis gigantea
Rorippa sylvestris
Salix fragilis
Salix purpurea
Myosotis scorpioides
Salix alba
Iris pseudacorus
Lycopus asper
Puccinellia distans
Stellaria aquatics
Juncus compressus
Pisidium moitessierianum
Carex flacca
Valvata piscinalis
Pisidium amnicum
Rumex longifolius
Radix auricularia
Aeromonas salmonicida
Sonchus arvensis var.
glabrescens
Lycopus europaeus
Butomus umbellatus
Viviparus georgianus
Impatiens glandulifera
Common Name
water foxtail
brown trout
redtop
creeping yellow cress
crack willow
purple willow
true forgot- me- not
white willow
yellow flag
western water
horehound
weeping alkali grass
giant chickweed
flattened rush
pea clam
sedge
European valve snail
pea clam
yard dock
European ear snail
furunculosis
smooth field sow
thistle
European water
horehound
flowering rush
banded mystery snail
Indian balsam
Type
Plant
Fish
Plant
Plant
Plant
Plant
Plant
Plant
Plant
Plant
Plant
Plant
Plant
Mollusk
Plant
Mollusk
Mollusk
Plant
Mollusk
Other
Invertebrate
Plant
Plant
Plant
Mollusk
Plant
Endemic
Region
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Mississippi R.
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Unknown
Eurasia
Eurasia
Eurasia
Mississippi R.
Asia
Location of
First Sighting
Lake Erie
Lake Michigan
drainage
Lake Superior
Lake Ontario
Widespread
Widespread
Lake Ontario
Widespread
Lake Ontario
Lake Erie
Lake Ontario
Lake St. Clair
Lake Ontario
drainage
Lake Superior
Detroit River
Lake Ontario
Lake Ontario
Lake Superior
Lake Michigan
Unknown
Lake Erie
Lake Ontario
Detroit River
Lake Michigan
Lake Huron
Vector
Release (deliberate)
Release (deliberate)
Release (deliberate)
Shipping, Solid
Ballast
Release (deliberate)
Release (deliberate)
Release (deliberate)
Release (deliberate)
Release (deliberate)
Release, (unintention
al)
Shipping, Solid
Ballast
Unknown
Release
(unintentional)
Shipping, Solid
Ballast
Unknown
Shipping, Solid
Ballast
Shipping, Solid
Ballast
Release (deliberate)
Release
(unintentional)
Release
(unintentional)
Release
(unintentional)
Shipping, Solid
Ballast
Shipping, Solid
Ballast
Aquarium release
Release (deliberate)
-------
>
Year of
Invasion
1912
1913
1913
1913
1915
1915
1916
1916
1918
1922
1923
1924
1925
1926
1927
1928
1928
1929
1930
1931
1932
1933
Species
Osmerus mordax
Alnus glutinosa
Lysimachia vulgaris
Rhamnus frangula
Mentha gentilis
Veronica beccabunga
Pluchea odorata var.
purpurescens
Pisidium henslowanum
Gillia altilis
Juncas inflexus
Gambusia affinis
Sphaerium corneum
Marsilea quadrifolia
Enteromorpha intestinalis
Acentropus niveus
Noturus insignis
Lepomis microlophus
Lepomis humilis
Nymphoides peltata
Cipangopaludina chinensis
malleata
A/a/as minor
Oncorhynchus kisutch
Common Name
rainbow smelt
black alder
garden loosetrife
glossy buckthorn
creeping whorled mint
European brookline
salt-marsh fleabane
henslow's pea clam
snail
rush
western mosquitofish
fingernail clam
European water clover
green alga
aquatic moth
margined madtom
redearsunfish
orange spotted
sunfish
yellow floating heart
Oriental mystery snail
minor naiad
coho salmon
Type
Fish
Plant
Plant
Plant
Plant
Plant
Plant
Mollusk
Mollusk
Plant
Fish
Mollusk
Plant
Benthic Alga
Other
Invertebrate
Fish
Fish
Fish
Plant
Mollusk
Plant
Fish
Endemic
Region
Atlantic NA
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Atlantic NA
Eurasia
Atlantic NA
Eurasia
Mississippi R.
Eurasia
Eurasia
Atlantic NA
Eurasia
Atlantic NA
Southern U.S.
Mississippi R.
Eurasia
Asia
Eurasia
Pacific
Location of
First Sighting
Lake Michigan
drainage
Widespread
Lake Ontario
Lower Great
Lakes, Ontario
Lake Ontario
Lake Ontario
Lake Erie
drainage
Lake Ontario
Lake Ontario
drainage
Lake Ontario
drainage
Lake Michigan
drainage
Lake Ontario
drainage
Lake Ontario
drainage
Lake Ontario
St. Lawrence
R., Montreal
Lake Ontario
drainage
Lake Michigan
drainage
Lake Erie
drainage
Lake Erie
drainage
Niagara River
Lake Erie
drainage
Lake Erie
Vector
Release (deliberate)
Release (deliberate)
Release (deliberate)
Release (deliberate)
Release (deliberate)
Shipping, Solid
Ballast
Release
(unintentional)
Shipping, Solid
Ballast
Canals
Unknown
Release (deliberate)
Shipping, Ballast
Water
Release (deliberate)
Release
(unintentional)
Release
(unintentional)
Canals
Release (deliberate)
Canals
Release
(unintentional)
Release (aquarium)
Release (deliberate)
Release (deliberate)
-------
>
Year of
Invasion
1933
1934
1935
1936
1938
1938
1938
1939
1940
1940
1943
1946
1946
1950
1950
1950
1950
1950
1950
1951
1951
1952
1952
Species
Craspedacusta sowerbyi
Lophopodella carter!
Cabomba caroliniana
Sparganium glomeratum
Actinocyclus normanii fo. subsalsa
Diatoma ehrenbergii
Stephanodiscus binderanus
Misgurnus anguillicaudatus
Cipangopaludina japonica
Glyceria maxima
Tanysphyrus lemnae
Cyclotella pseudostelligera
Stephanodiscus subtilis
Phenacobius mirabilis
Morone americana
Oncorhynchus nerka
Cirsium palustre
Potamothrix bedoti
Pluchea odorata var. succulenta
Branchiura sowerbyi
Carex acutiformis
Myriophyllum spicatum
Potamothrix moldaviensis
Common Name
freshwater jellyfish
bryozoan
fanwort
bur reed
diatom
diatom
diatom
Oriental weatherfish
Oriental mystery snail
reed sweet-grass
aquatic weevil
diatom
diatom
suckermouth minnow
white perch
kokanee
marsh thistle
oligochaete
salt-marsh fleabane
oligochaete
swamp sedge
Eurasian watermilfoil
oligochaete
Type
Other
Invertebrate
Other
Invertebrate
Plant
Plant
Phytoplankton
Phytoplankton
Phytoplankton
Fish
Mollusk
Plant
Other
Invertebrate
Phytoplankton
Phytoplankton
Fish
Fish
Fish
Plant
Annelid
Plant
Annelid
Plant
Plant
Annelid
Endemic
Region
Asia
Asia
Southern U.S.
Eurasia
Eurasia
Widespread
Eurasia
Asia
Asia
Eurasia
Eurasia
Widespread
Eurasia
Mississippi R.
Atlantic NA
Pacific NA
Eurasia
Ponto-Caspian
Atlantic NA
Asia
Eurasia
Eurasia
Ponto-Caspian
Location of
First Sighting
Lake Erie
drainage
Lake Erie
Lake Michigan
drainage
Lake Superior
Lake Ontario
Lake Michigan
Lake Michigan
Lake Huron
drainage
Lake Erie
Lake Ontario
Unknown
Lake Michigan
Lake Michigan
Lake Erie
drainage
Lake Ontario
Lake Ontario
drainage
Lake Superior
Lake Ontario
Lake Ontario
drainage
Lake Michigan
drainage
Lake Michigan
drainage
Lake Erie
Lake Ontario
Vector
Release
(unintentional)
Canals
Release (aquarium)
Unknown
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Release
(unintentional)
Release (deliberate)
Release (deliberate)
Unknown
Shipping, Ballast
Water
Shipping, Ballast
Water
Canals
Canals
Release (deliberate)
Unknown
Unknown
Unknown
Release
(unintentional)
Unknown
Release (aquarium)
Unknown
-------
>
Year of
Invasion
1956
1956
1958
1959
1959
1959
1960
1960
1962
1962
1963
1964
1964
1964
1964
1964
1965
1966
1966
1967
Species
Cordylophora caspia
Oncorhynchus gorbuscha
Eurytemora affinis
Trapa natans
Lasmigona subviridis
Pisidium supinum
Glugea hertwigi
Polygonum caespitosum var.
longisetum
Lepisosteus platostomus
Thalassiosira weissflogii
Skeletonema potamos
Bangia atropurpurea
Cyclotella atomus
Cyclotella cryptica
Cyclotella woltereki
Chroodactylon ramosum
Potamothrix vejdovskyi
Eubosmina coregoni
Epilobium parviflorum
Skistodiaptomus pallidus
Common Name
hydro id
pink salmon
calanoid copepod
water chestnut
mussel
humpback pea clam
protozoan
Bristly Lady's Thumb
shortnose gar
diatom
diatom
red alga
diatom
diatom
diatom
red alga
oligochaete
waterflea
small flowered hairy
willow herb
calanoid copepod
Type
Other
Invertebrate
Fish
Zooplankton
(crustacean)
Plant
Mollusk
Mollusk
Other
Invertebrate
Plant
Fish
Phytoplankton
Phytoplankton
Benthic Alga
Phytoplankton
Phytoplankton
Phytoplankton
Benthic Alga
Annelid
Zooplankton
(crustacean)
Plant
Zooplankton
(crustacean)
Endemic
Region
Ponto-Caspian
Pacific NA
Widespread
Eurasia
Atlantic NA
Europe
Eurasia
E. Asia
Mississippi R.
Widespread
Widespread
Atlantic NA
Widespread
Widespread
Widespread
Atlantic Ocean
Ponto-Caspian
Eurasia
Eurasia
Mississippi R.
Location of
First Sighting
Lake Erie
Lake Superior
Lake Ontario
Lake Ontario
drainage
Lake Ontario
drainage
Lake Ontario
Lake Erie
Lake Erie
drainage
Lake Michigan
drainage
Detroit River
Lake Erie
drainage
Lake Erie
Lake Michigan
Lake Michigan
Lake Michigan
Lake Erie
Lake Erie
Lake Michigan
Lake Michigan
drainage
Lake Ontario
Vector
Release
(unintentional)
Release
(unintentional)
Shipping, Ballast
Water
Release (aquarium)
Canals
Shipping, Ballast
Water
Release
(unintentional)
Unknown
Canals
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping? Ballast
Water?
Shipping, Ballast
Water
Unknown
Release
(unintentional)
-------
>
Year of
Invasion
1968
1968
1969
1971
1972
1972
1973
1973
1973
1973
1975
1975
1975
1975
1975
1978
1978
1978
1979
1979
1980
Species
Myxobolus (Myxosoma) cerebralis
Dugesia polychroa
Solidago sempervirens
Enneacanthus gloriosus
Cyclops strenuus
Hydrocharis morsus-ranae
Skeletonema subsalsum
Thalassiosira guillardii
Thalassiosira pseudonana
Nitocra hibernica
Sphacelaria fluviatilis
Lotus corniculatus
Renibacterium (Corynebacterium)
salmoninarum
Sphacelaria lacustris
Hymenomonas roseola
Biddulphia laevis
Chaetoceros hohnii
Thalassiosira lacustris
Notropis buchanani
Enteromorpha prolifera
Corbicula fluminea
Common Name
salmonid whirling
disease
flatworm
seaside goldenrod
bluespotted sunfish
copepod
European frogbit
diatom
diatom
diatom
harpacticoid copepod
brown alga
birdsfoot trefoil
bacterial kidney
disease
brown alga
cocco-lithophorid alga
diatom
diatom
diatom
ghost shiner
green alga
Asiatic clam
Type
Other
Invertebrate
Other
Invertebrate
Plant
Fish
Zooplankton
(crustacean)
Plant
Phytoplankton
Phytoplankton
Phytoplankton
Benthic
Crustacean
Benthic Alga
Plant
Bacteria
Benthic Alga
Phytoplankton
Phytoplankton
Phytoplankton
Phytoplankton
Fish
Benthic Alga
Mollusk
Endemic
Region
Unknown
Europe
Atlantic NA
Atlantic NA
Hudson Bay
Eurasia
Eurasia
Widespread
Widespread
Eurasia
Asia
Eurasia
Unknown
Unknown
Eurasia
Widespread
Unknown
Eurasia
Mississippi R.
Atlantic NA
East Asia
Location of
First Sighting
Lake Erie
drainage
Lake Ontario
Lake Michigan
Lake Ontario
drainage
Lake Superior
Lake Ontario
Lake Erie
Lake Erie
Lake Erie
drainage
Lake Ontario
Lake Michigan
drainage
Lake Superior
Lake Superior
Lake Michigan
Lake Huron
Lake Michigan
Lake Huron
Lake Erie
Lake St. Clair
drainage
Lake St. Clair
drainage
Lake Erie
Vector
Release
(unintentional)
Shipping, Ballast
Water
Release
(unintentional)
Release (aquarium)
Canals (water
diversion)
Release
(unintentional)
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Release (aquarium)
Release (deliberate)
Release
(unintentional)
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Release (bait)
Unknown
Release (aquarium)
-------
>
oo
Year of
Invasion
1980
1980
1982
1982
1983
1983
1985
1986
1986
1988
1988
1988
1988
1989
1989
1990
1990
1991
1992
1992
Species
Ripistes parasita
Daphnia galeata galeata
Lupinus polyphyllus
Bythotrephes longimanus
Nitellopsis obtusa
Gianius (Phallodrilus) aquaedulcis
Salmincola lotae
Gymnocephalus cernuus
Apeltes quadracus
Argulus japonicus
Thalassiosira baltica
Bosmina maritima
Dreissena polymorpha
Scardinius erythrophthalmus
Dreissena bugensis
Neogobius melanostomus
Proterorhinus marmoratus
Potamopyrgus antipodarum
Acanthostomum sp.
Ichthyocotylurus pileatus
Common Name
oligochaete
waterflea
lupine
spiny waterflea
green alga
oligochaete
copepod
Eurasian ruffe
fourspine stickleback
parasitic copepod
diatom
waterflea
zebra mussel
rudd
quagga mussel
round goby
tubenose goby
New Zealand mud
snail
digenean fluke
digenean fluke
Type
Annelid
Zooplankton
(crustacean)
Plant
Zooplankton
(crustacean)
Phytoplankton
Annelid
Zooplankton
(crustacean)
Fish
Fish
Zooplankton
(crustacean)
Phytoplankton
Zooplankton
(crustacean)
Mollusk
Fish
Mollusk
Fish
Fish
Mollusk
Other
Invertebrate
Other
Invertebrate
Endemic
Region
Eurasia
Eurasia
Eurasia
Eurasia
Eurasia
Europe
Eurasia
Ponto-Caspian
Atlantic NA
Asia
Europe
Eurasia
Ponto-Caspian
Eurasia
Ponto-Caspian
Ponto-Caspian
Ponto-Caspian
Australasia
Eurasia
Ponto-Caspian
Location of
First Sighting
Lake Huron
Lake Erie
Lake Superior
Lake Ontario
Lake St. Clair
Niagara River
Lake Superior
Lake Superior
Lake Superior
Lake Michigan
Lake Ontario
Lake Erie
Lake St. Clair
Lake Ontario
Lake Ontario
St. Clair River
St. Clair River
Lake Ontario
Lake Superior
Lake Superior
Vector
Shipping, Ballast
Water
Shipping, Ballast
Water
Release
(unintentional)
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Unknown
Shipping, Ballast
Water
Shipping, Ballast
Water
Release (aquarium)
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Release (Bait)
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
-------
>
Year of
Invasion
1992
1992
1992
1992
1994
1994
1994
1994
1994
1995
1996
1997
1998
1998
1999
1999
2000
2001
2001
2001
Species
Neascus brevicaudatus
Trypanosoma acerinae
Dactylogyrus amphibothrium
Dactylogyrus hemiamphibothrium
Echinogammarus ischnus
Scolex pleuronectis
Sphaeromyxa Sevastopol!
Neoergasilus japonicus
Megacyclops viridis
Alosa aestivalis
Heteropsyllus nr. nunni
Acineta nitocrae
Cercopagis pengoi
Schizopera borutzkyi
Daphnia lumholtzi
Nitocra incerta
Heterosporis sp.
Rhabdovirus carpio
Gammarus tigrinus
Psammonobiotus communis
Common Name
digenean fluke
flagellate
monogenetic fluke
monogenetic fluke
amphipod
cestode
mixosporidian
copepod
cyclopoid copepod
blueback herring
harpacticoid copepod
suctorian
fish-hook waterflea
harpacticoid copepod
waterflea
harpacticoid copepod
microsporidian
spring viraemia of
carp (SVC)
Amphipod
testate amoeba
Type
Other
Invertebrate
Other
Invertebrate
Other
Invertebrate
Other
Invertebrate
Benthic
Crustacean
Other
Invertebrate
Other
Invertebrate
Zooplankton
(crustacean)
Zooplankton
(crustacean)
Fish
Benthic
Crustacean
Other
Invertebrate
Zooplankton
(crustacean)
Benthic
Crustacean
Zooplankton
(crustacean)
Benthic
Crustacean
Other
Invertebrate
Virus
Benthic
Crustacean
Other
Invertebrate
Endemic
Region
Eurasia
Ponto-Caspian
Eurasia
Eurasia
Ponto-Caspian
Ponto-Caspian
Ponto-Caspian
Eastern Asia
Europe
Atlantic NA
Atlantic NA?
Eurasia
Ponto-Caspian
Ponto-Caspian
Africa,
Australasia
Ponto-Caspian
Unknown
Eurasia
Atlantic NA
Ponto-Caspian
Location of
First Sighting
Lake Superior
Lake Superior
Lake Superior
Lake Superior
Detroit River
/Lake Erie
Lake St. Clair
Lake St. Clair
St. Clair River
Lake Huron
Lake Ontario
Lake Michigan
Lake Erie
Lake Ontario
Lake Michigan
Lake Erie
Detroit River
Lake Ontario
Lake Michigan
drainage
Lake Superior
(Lake Huron in
2002)
Lake Ontario
Vector
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Unknown
Unknown
Canals
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Shipping, Ballast
Water
Release
(unintentional)
Shipping, Ballast
Water
Unknown
Release (aquarium)
Shipping, Ballast
Water
Shipping, Ballast
Water
-------
Year of
Invasion
2002
2002
2002
2002
2002
2003
2005
2006
Species
Ranavirus sp.
Psammonobiotus linearis
Psammonobiotus dziwnowi
Cylindrospermopsis raciborskii
Piscirickettsia cf. salmonis
Enteromorpha flexuosa
Novirhabdovirus sp.
Hemimysis anomala
Common Name
largemouth bass virus
testate amoeba
testate amoeba
cyanobacterium
muskie pox
green alga
VMS
bloody-red mysid
Type
Virus
Other
Invertebrate
Other
Invertebrate
Phytoplankton
Other
Invertebrate
Benthic Alga
Virus
Benthic
Crustacean
Endemic
Region
Unknown
Ponto-Caspian
Ponto-
Caspian?
South
America?
Unknown
Widespread
Atlantic NA?
Ponto-Caspian
Location of
First Sighting
Lake Michigan
drainage
Lake Ontario
Lake Ontario
Lake Michigan
drainage
Lake St. Clair
Lake Michigan
drainage
Lake Ontario
Lake Michigan
drainage
Vector
Release
(unintentional)
Shipping, Ballast
Water
Shipping, Ballast
Water
Unknown
Unknown
Shipping
Shipping?, Ballast
Water?
Shipping, Ballast
Water
Source: http://www.glerl.noaa.gov/res/Programs/ncrais/docs/great-lakes-list.xls (accessed March 5, 2008).
-------
APPENDIX B. NONINDIGENOUS SPECIES THAT MAY
SPREAD TO THE GREAT LAKES
B-l
-------
Table B-l. Summary of Literature Review for potentially invasive species. The 58 species shown as shaded represent those with the
greatest risk.
Type of Organism
Common Name
(Scientific Name)
cnidarian
Hydroid (Cordylophora caspia)
crustacean
Amphipod (Pontogammarus crassus)
crustacean
Amphipod (Corophium curvispinum)
crustacean
Amphipod (Dikerogammarus villosus)
crustacean
Amphipod (Pontogammarus
robustoides)
crustacean
Amphipod (Echinogammarus ischnus)
crustacean
Amphipod (Dikerogammarus
haemobaphes)
crustacean
Amphipod (Corophium sowinskyi)
crustacean
Amphipod (Echinogammarus
warpachowskyi)
crustacean
Amphipod (Pontogammarus aralensis)
In
GL?a
Yes
No
No
No
No
Yes
No
No
No
No
Invasion
History"
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Spread
Potential in
GLC
High
NEK
High
High
High
High
High
Med
High
High
Ecol.
Impact in
GLd
Med
NEK
Med
High
NEK
Med
Med
Med
Med
Med
Species
Origin6
Ponto-
Caspian
Sea
Caspian
Sea
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Possible
Source'
aquarium;
ballast
water
ballast
water
ballast
water
ballast
water;
canals
(Europe)
ballast
water;
canals
(Europe)
ballast
water
ballast
water;
canals
(Europe)
canals
ballast
water
ballast
water
Concern, Consequence, Invasion9
established in L. Erie, Baltic Sea basin; invasion
history
established in Baltic Sea; not established yet,
may be introduced
found in L. St. Clair; invasion history; displaced
filter-feeding caddisflies widely distributed in
Europe; high densities in Baltic ports; established
itself in British Isles
not established yet; established in Baltic Sea
basin; eat and replace native amphipods;
invading and decreasing natives in Europe; have
rapid impact on macroinvertebrate survival,
leading to population declines; invading most of
Western Europe's hydrosystems
not established yet; established in Baltic Sea
basin
established in L. Erie, Huron, Ontario; invasion
history; limited dispersal capability; displaced
native amphipod Gmelinoides fasciatus
not established yet; established in Baltic Sea
basin
spreading across Europe; not established yet;
could alter littoral communities and food webs
not established yet; established in Baltic Sea
basin
not established yet; established in Baltic Sea
basin
Lit. Source11
13,19,52
13, 19, 30
14, 19, 23,29,
30
6,7,8, 13,19,
30
13,19,30
4,9, 13,19,25,
26, 29, 32, 52,
53
13, 19, 30
30
13.19
13
-------
td
crustacean
Amphipod (Echinogammarus berilloni)
crustacean
Amphipod (Echinogammarus
trichiatus)
crustacean
Amphipod (Gammarus tigrinus)
crustacean
Amphipod (Gmelinoides fasciatus)
crustacean
Amphipod (Iphigenella shablensis)
crustacean
Amphipod (Pontogammarus
maeoticus)
crustacean
Amphipod (Pontogammarus obesus)
crustacean
Amphipod (Pontogammarus
subnudas)
crustacean
Isopod (Jaera istri)
crustacean
Isopod (Jaera sarsi)
crustacean
Isopod (Proasellus coxalis)
crustacean
Isopod (Proasellus meridianus)
No
No
No
No
No
No
No
No
No
No
No
No
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
North
America
Asia
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
13
13
13
13
13
13
13, 30
13
13
13
13
13
-------
crustacean
Mysid shrimp (Hemimysis anomala)
crustacean
Mysid shrimp (Paramysis ullskyi)
crustacean
Mysid shrimp (Limnomysis benedeni)
crustacean
Mysid shrimp (Paramysis intermedia)
crustacean
Mysid shrimp (Paramysis lacustris)
crustacean
Pseudocumid (Pseudocuma
cercaroides)
crustacean
Pseudocumid (Pterocuma pectinata)
crustacean
Rusty crayfish (Orconectes rusticus)
fish
Alewife (Alosa pseudoharengus)
fish
Arowana (Osteoglossum bicirrhosum)
No
No
No
No
No
No
No
No
Yes
No
Yes
Yes
Yes
Yes
Yes
NEK
NEK
Yes
Yes
NEK
High
Med
NEK
NEK
NEK
NEK
NEK
High
Present
NEK
Med
NEK
NEK
NEK
NEK
NEK
NEK
High
High
NEK
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Southeas
t United
States
Atlantic
coast of
North
America
South
America
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
bait
bait fish;
canals
ornamental
not established yet; established in Baltic Sea
basin; reduction of zooplankton; biomagnification
of contaminants at higher trophic levels; adaption
to shallow, warm waters
not established yet; established in Baltic Sea
basin; reduction of zooplankton; biomagnification
of contaminants at higher trophic levels; adaption
to shallow, warm waters
not established yet; established in Baltic Sea
basin; reduction of zooplankton; biomagnification
of contaminants at higher trophic levels; adaption
to shallow, warm waters
not established yet; established in Baltic Sea
basin; reduction of zooplankton; biomagnification
of contaminants at higher trophic levels; adaption
to shallow, warm waters
not established yet; established in Baltic Sea
basin; reduction of zooplankton; biomagnification
of contaminants at higher trophic levels; adaption
to shallow, warm waters
not established yet, may be introduced
not established yet, may be introduced
found in inland waters of Michigan (not Great
Lakes); invasion history; spreading north into
Ontario, northern Midwest; reduction of
vegetation important to native fish for food and
cover; habitat destruction; affects native crayfish
industry (Ontario)
found in L. Ontario, Erie, Michigan, Superior;
restructure a lake's food web, leaving less food
for native fish; contributed to extinction of some
native species
potential there but have slow growth rate; inability
to survive temperatures below 58 degrees
Fahrenheit
13,19,30
30
13, 19,30
13, 30
13, 19,30
13
13
4, 17,46, 48,
49,56
5,9,11,29,30,
35, 44, 49, 52,
55, 56
11,31
-------
fish
Bighead carp
(Hypophthalmichthys nobilis/
Aristichthys nobilis)
fish
Black carp (Mylopharyngodon piceus)
fish
Black sea silverside (Aphanius boyeri)
fish
Bleak (Alburnus alburnus)
fish
Blue catfish (Ictalurus furcatus)
fish
Blue tilapia (Oreochromis aureus)
fish
Blueback herring (Alosa aestivalis)
fish
Bullhead (Coitus gobio)
fish
Caspian shad (Caspialosa caspia)
fish
Caucasian goby (Knipowitschia
caucasica)
Yes
No
No
No
No
No
Yes
No
No
No
Yes
No
NEK
Yes
Yes
Yes
No
NEK
NEK
Yes
High
NEK
High
High
Med
NEK
Present
High
Med
High
Med
NEK
NEK
Med
Med
NEK
High
NEK
NEK
Med
China
China
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
North
America
Africa/
Eurasia
Atlantic
coast of
North
America
Europe
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
aquaculture;
fish market;
canals
aquaculture
bait;
aquaculture
aquaculture;
bait
aquaculture;
sport fish
aquaculture;
sport fish
canals
bait
fish market
ballast
water
few isolated cases in L. Erie; tolerate low temp;
established in Mississippi River basin; invasion
history; vast mobility; voracious consumption
habits; clog fishing nets and scare away
commercial fish; competes with native fish
not established yet; reports found in conclave of
Ohio, Missouri, and Mississippi Rivers; vast
mobility; voracious consumption habits; clog
fishing nets and scare away commercial fish; risk
to commercial shellfish stocks
invasive in Baltic; feed on crustaceans
spreading outside natural range; will eat any
species offish they can catch, along with
crayfish, freshwater mussels, frogs, and other
readily available aquatic food sources
spreading in U.S.; invasive in FL; local
abundance and high densities in certain areas
have resulted in marked changes in fish
community structure; private culture in Ontario
invasive in L. Ontario; impede recovery of
depressed populations of indigenous fishes such
as Cisco and lake trout; cold water may prevent
establishment
often behave aggressively towards one another,
and competition for shelter and foraging space
can be intense
invasion in Baltic; predators which feed on small
benthic animals
11, 12,14, 17,
31,34,39,43,
44, 46, 49, 56
11, 18,34,39,
43,46
18
11,18
11,18
11,17,18,42,
52
9,11,28,29,
44, 52, 53
11, 18
11, 18
11, 18
-------
fish
Cherry salmon (Oncorhynchus masou)
fish
Chub (Leucaspius cephalus)
fish
Chum salmon (Oncorhynchus keta)
fish
Clown loach (Botia macracanthus)
fish
Common dace (Leuciscus leuciscus)
fish
Common tilapia (Oreochromis
mossambica)
fish
Eurasian minnow (Phoxinus phoxinus)
fish
European perch (Perca fluviatilis)
fish
European ruffe (Gymnocephalus
cernuus)
fish
European whitefish (Vendace)
(Coregonus albula)
fish
Fourspine stickleback (Apeltes
quadracus)
No
No
No
No
No
No
No
No
Yes
No
Yes
NEK
NEK
NEK
NEK
Yes
NEK
Yes
Yes
Yes
Yes
NEK
Med
NEK
High
NEK
High
NEK
High
High
Present
High
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
Med
High
NEK
NEK
North
Pacific;
Japan
Ponto-
Caspian
Sea
Asia/Nort
h Pacific
Indonesia
Ponto-
Caspian
Sea
Africa
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Europe
Europe/
Atlantic
coast of
North
America
sport fish
ornamental;
live bait;
sport fish
sport fish
ornamental
bait
fish market;
aquaculture
bait
sport fish
ballast
water;
canals; bait
sport fish
ballast
water;
ornamental
widespread
widespread
potential there but have slow growth rate
widespread in Europe, showed invasion in one
country
private culture in Ontario
invasive in Baltic
widespread prized for angling (introduced in
many countries); because widespread,
designated as invasive due to impacts on native
species; cannibalism is common
invaded L. Superior, Huron; now in L. Michigan;
invasive history in Europe; displace native
species; predation on native fish eggs;
competition with native fish; L. Superior native
fish declined since introduction of ruffe
some invasion history
confined to coasts; rapid increases of Apeltes in
Thunder Bay (L. Superior) suggests the species
is displacing native sticklebacks at a rapid rate
11,18
11,18
11, 18
11, 31
11, 18
11, 17, 18, 42
11, 18
1 1 , 1 8, 49
11, 16,23, 29,
30, 37,43, 16,
48, 49, 52, 56
11, 18
11, 52
-------
fish
Ghost (or glass) catfish
(Kryptopterus bicirrhis)
fish
Giant or red snakehead
(Channa micropeltes)
fish
Goldfish (Carassius auratus)
fish
Grass carp
(Ctenopharyngodon idella)
fish
Inland silverside (Menidia beryllina)
fish
Koi (common) carp (Cyprinus carpio)
fish
Longtail goby (Ctenogobius sagittula)
fish
Monkey goby (Neogobius fluviatilis)
fish
Mummichog (Fundulus heteroclitus)
fish
Nile tilapia (Oreochromis niloticus)
No
No
Yes
No
No
Yes
No
No
No
No
NEK
NEK
Yes
Yes
Yes
Yes
NEK
Yes
NEK
NEK
NEK
NEK
High
High
NEK
High
NEK
High
NEK
NEK
NEK
NEK
High
High
NEK
High
NEK
High
NEK
NEK
Asia
Asia
China/
Japan
Eastern
Asia
Eastern
U.S.
coast
Eurasia
Eastern
Pacific
Eurasia
Western
Atlantic/
Eastern
U.S.
coast
Africa
ornamental
ornamental
ornamental
aquaculture;
fish market
aquaculture;
sport fish
ornamental;
aquaculture
ballast
water
bait;
aquarium
fish market;
aquaculture
potential there but have slow growth rate
potential there but have slow growth rate
of the large established populations are recorded
from the vicinity of western L. Erie, widespread;
extensive invasion history; concern of impacts on
community, increasing turbidity, predation of
native fish, help facilitate algal blooms; tolerate
low temperatures
found isolated in L. Erie, Ontario, Huron; grazes
on aquatic vegetation reducing plant density or
removing all aquatic vegetation from a body of
water; competes with native fish; tolerate low
temperatures; invasion history
could replace native fish; replaced native fish in
CA, OK; found in Mississippi conclave of Illinois
and Ohio Rivers
found in Great Lakes; extensive invasion history;
globally widespread; uproot and destroy
submerged aquatic vegetation and therefore may
be detrimental to duck and native fish
populations; tolerate low temperatures
invasion history in Europe; mass invasion of
monkey goby is connected with the intensive
consumption of plankton crustaceans; competes
with other small fish for food and space
private culture in Ontario; Mississippi
11,31
11,31
9,11,31,49
11, 14,17,31,
43, 44, 46, 52
11, 18,52
1 1 , 31 , 49
11, 18
11, 14, 18, 30,
51
11, 18
11, 17,18,42
-------
fish
Oriental weatherfish (Misgurnus
anguillicaudatus
fish
Pacu (Colossoma macropomum)
fish
Pike killifish (Belonesox belizanus)
fish
Pontic shad (Alosa pontica)
fish
Racer goby
(Neogobius gymnotrachelus
fish
Rainbow smelt (Osmerus mordax)
fish
Red tail botia (Botia modesta)
fish
Redear sunfish (Lepomis microlophus)
fish
Roach (Rutilus rutilus)
fish
Round goby (Neogobius
melanostromus)
fish
Rudd (Scardinius erythrophthalmus)
Yes
No
No
No
No
Yes
No
Yes
No
Yes
Yes
Yes
NEK
NEK
NEK
Yes
Yes
NEK
NEK
Yes
Yes
Yes
High
NEK
NEK
Med
High
Med
NEK
High
High
Present
Present
Med
NEK
NEK
NEK
Med
High
NEK
NEK
Med
High
High
Asia
South
America
Central
America
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Atlantic
coast of
Central
America
Asia
Atlantic
coast of
North
America
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Eurasia
ornamental;
aquaculture
aquaculture;
ornamental
aquarium
fish market;
aquarium
canals;
ballast
water
sport fish
ornamental
aquaculture;
sport fish
sport fish
ballast
water
bait fish;
ballast
water
established in Shiawassee River and L.
Michigan; reduce populations of aquatic insects
important as food to native fishes; invasion
history; tolerate low temperatures
may compete with larvae of native fish species
for plankton
voracious predator and has been known to
reduce populations of eastern mosquitofish
(Gambusia holbrooki) and other native poeciliid
and cyprinodontid populations
spreading in Europe
Erie Canal; Lake Ontario, Michigan, Superior;
Minnesota; dominant prey form for salmonids;
contributed to extinction of blue pike; affect
imperiled species
potential there but have slow growth rate
found in L. Michigan; introduced redearare
associated with ecological changes in
populations of pumpkinseed L. gibbosus, a native
molluscivore; preference for mollusks
invasive in Baltic; nuisance once established
found in Great Lakes; invasion history; spreads
rapidly; zebra mussels dominate diet; altering
benthic communities; aggressive; lake trout egg,
sturgeon egg predators
introduced into L. Ontario; expanded in L. Erie;
habitat degradation for native fish; rudd
introduced to open waters will hybridize with
golden shiner
1 1 , 1 4, 31 , 44
11,18
11,18
11,18
11,18,51
1 1 , 44, 52, 53,
55, 56
11, 31
11, 18,52
11, 18
3, 9, 11, 29,30,
32, 43, 46, 48,
49, 51 , 52, 53,
56
11, 17,29, 41,
46, 49
-------
fish
Sand goby (pomatoschistus minutus)
fish
Silver carp (Hypophthalmichthys
molitrix)
fish
Snakehead (Channa argus argus)
fish
Starry goby (Benthophilus stellatus)
fish
Striped bass (Morone saxatilis)
fish
Sunbleak (Leucaspius delineatus)
fish
Tench (Tinea tinea)
fish
Toothed carp (Aphanius fasciatus)
fish
Tubenose goby (Proterorhinus
marmoratus)
fish
Tyulka / Caspian kilka
(Clupeonella cultriventris/caspia)
No
No
No
No
No
No
No
No
Yes
No
Yes
NEK
Yes
Yes
Yes
Yes
Yes
NEK
Yes
Yes
High
High
High
High
NEK
High
High
High
Med
High
Med
NEK
Med
NEK
NEK
High
High
NEK
High
NEK
Eastern
Atlantic/
Ponto-
Caspian
Sea
China
Asia
Ponto-
Caspian
Sea
Atlantic
coast of
North
America
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea/
Europe/
Asia
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
aquarium;
aquaculture
aquaculture
fish market
ballast
water?
fish market
canals
ornamental;
canals;
sport fish
canal
ballast
water
ballast
water
abundant; invasion in Baltic; predators that feed
on small benthic animals
documented in Cal-Sag Channel of IL Waterway;
spread along Mississippi River; competition with
commercially significant native fish for food and
habitat; vast mobility; voracious consumption
habits; clog fishing nets; widespread in U.S. Gulf
states
sporadic places; devastation to freshwater
ecosystems of the U.S. because of its predacious
nature, lack of natural predators, high fertility,
and adaptability to a wide range of environmental
conditions
spreading in Europe
invasion history
widespread in Europe; affecting coarse fish
populations in England (limited range); feeds on
phytoplankton and zooplankton
been captured in St. Lawrence River; may head
to L. Ontario; competition with benthic feeders,
native fish; potential competitor for food with
sport fishes and native cyprinids; widespread
competes with native fish; declining in
Mediterranean
found/restricted to L. St. Clair, western L. Erie;
Invasion history but not spread rapidly;
aggressive
expanded in Volga River; suppressed native fish
11,18
11,12, 18,34,
39, 43, 44, 46,
52,56
11,12, 46,49
11, 18,30
11,31
11,18
11,17, 18,37,
49
11,18
1 1 , 29, 32, 43,
51,52
11, 14,18,30
-------
fish
Ukranian or nine-spined stickleback
(Pungitius platygaster)
fish
Weather fish (Misgurnus fossilis)
fish
White cloud mountain minnow
(Tanichthys albonubes)
fish
White perch (Morone americana)
fish
Zander (Sander lucioperca)
flat worm
Flatworm (Dendrocoelum
romanodanubiale)
flat worm
Trematode (Apophallus muehlingi)
flat worm
Trematode (Nicolla skrjabini)
flat worm
Trematode (Rossicotrema donicum)
insect
Apterogote (wingless)
(Campodea staphylinus)
microcrustacean
Baltic water flea (Bosmina coregoni)
microcrustacean
Calanoid copepod (Eurytemora affinis)
Yes
No
No
Yes
No
No
No
No
No
No
Yes
Yes
NEK
Yes
Yes
Yes
Yes
NEK
NEK
NEK
NEK
Yes
Yes
Yes
Present
NEK
High
Present
High
NEK
NEK
NEK
NEK
High
Present
Present
NEK
NEK
Med
High
High
NEK
NEK
NEK
NEK
NEK
High
NEK
Ponto-
Caspian
Sea
Europe
China
Atlantic
coast of
North
America
Ponto-
Caspian
Sea
North
Sea
basin
Black
Sea
Black
Sea
Black
Sea
Eurasia
Eurasia/
Ponto-
Caspian
Ponto-
Caspian
Sea
canal?;
bait?
ornamental
ornamental
fish market;
canals
sport fish;
canals;
aquaculture
ballast
water
ballast
water
ballast
water
ballast
water; fish
host
ballast
water
ballast
water
ballast
water
found in L. Huron, Michigan
tolerate low temperature; established invasion in
Italy, Spain, Croatia
tolerate low temperature; extensive invasion
history; high occurrence frequency
already invaded Great Lakes; competition with
native fish; potential to cause declines of walleye
populations; prey on eggs of walleye
found everywhere in Europe; depleted stocks of
native fish; hunts in packs
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
not established yet, may be introduced
widespread; been recorded in North American
fresh waters; tolerate range of salinity
established in L. Ontario, Superior; invasion
history
widespread; invasion history; established in
Great Lakes; high populations
11, 18,52
11,31
11, 14,31
9,11, 17,31,
44, 46, 48, 52,
53, 56
11, 18
13
13
13
13
10
13
13
-------
microcrustacean
Calanoid copepod (Heterocope
appendiculata)
microcrustacean
Calanoid copepod (Calanipeda aquae-
dulcis)
microcrustacean
Calanoid copepod (Heterocope
caspia)
microcrustacean
Cladoceran (Daphnia cristata)
microcrustacean
Cladoceran (Bosmina obtusirostris)
microcrustacean
Cladoceran (Podonevadne trigona
ovum)
microcrustacean
Cladoceran (water flea) (Daphnia
lumholtzi)
microcrustacean
Cladoceran (Cornigerius maeoticus
maeoticus)
microcrustacean
Cyclopoid copepod (Cyclops strenuus)
microcrustacean
Cyclopoid copepod (Cyclops kolensis)
microcrustacean
Fishhook waterflea (Cercopagis
pengoi)
microcrustacean
Harpactacoid copepod (Ectinosoma
abrau)
No
No
No
No
No
No
No
No
Yes
No
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
NEK
High
High
High
High
High
High
Med
High
Present
High
Present
High
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
NEK
High
NEK
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Australia;
SE Asia
Ponto-
Caspian
Sea
Eurasia
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water;
boating;
waterfowl
ballast
water
ballast
water
ballast
water
ballast
water;
boating
ballast
water
established in Baltic Sea basin
established in Baltic Sea basin
established in Baltic Sea basin
established in Baltic Sea basin
established in Baltic Sea basin
established in Baltic Sea basin
expanding in U.S.; widespread; prefers warm
water; competes with native daphnia for food and
of its ability to avoid predation
established in Baltic Sea basin
established in Great Lakes
established in Baltic Sea basin
established in Great Lakes except L. Huron,
Superior; Baltic Sea basin; compete with small
fish for zooplankton; invasive in Europe; clogs
nets for fisheries; damage cost data available
potential to invade Great Lakes
13
13
13
13
13
13
1 4, 34, 52, 53
13
10
13
9,13, 19,20,
29, 32, 38, 49,
52, 53, 54, 56
13
-------
microcrustacean
Harpactacoid copepod
(Onychocamptus mohammed)
microcrustacean
Harpactacoid copepod
(Paraleptastacus spinicaudata triseta)
microcrustacean
Harpactacoid copepod (Schizopera
borutzkyi)
microcrustacean
Harpactacoid copepod (Nitocra
incerta)
microcrustacean
Spiny waterflea (Bythotrephes
longimanus)
microcrustacean
Spiny waterflea (Bythotrephes
cederstroemi)
microsporidian
Fish parasite (Heterosporis sp.)
mollusc
Asian clam (Corbicula fluminea)
mollusc
Basket (European) shell (Corbula
gibba)
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
No
NEK
Yes
Yes
Yes
Yes
Yes
Yes
Yes
NEK
Present
High
Present
Present
High
Present
NEK
NEK
NEK
NEK
NEK
High
NEK
High
High
NEK
NEK
NEK
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Danube
River
delta of
Black
Sea
Black
Sea
Great
Britain;
Europe
Eurasia
Europe/
Asia?
Asia
Atlantic
coast of
Europe
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water;
boating
ballast
water
fish release;
bait
ornamental;
fish
market/bait
ballast
water
established in L. Ontario
invasion history
established in L. Michigan; altering species
composition of nearshore communities
established in L. Michigan
established in Great Lakes; Great Lakes - has
caused major changes in the zooplankton
community structure; invasion history; reproduce
rapidly; competes directly with small fish and can
have impact on zooplankton community; damage
cost data available
established in Great Lakes; the invasion into the
Laurentian Great Lakes has resulted in
substantial and sustained decreases in the
populations of a number of (mostly cladoceran)
native zooplankton species; coincided with
dramatic declines in abundance of Daphnia
found in L. Ontario, attacks muscle cells in yellow
perch; found in Wisconsin
established in L. Michigan, Superior, Erie; does
not tolerate low temperatures; fouled water
plants; damage cost data available
capacity to achieve very high population
densities, giving it the potential to affect the
growth and recruitment of a wide range of soft
bottom organisms
13, 14
13
1,10, 13,14,
15,28
13
13, 21,33, 38,
48, 49, 52, 53,
54
32, 49, 50, 56
9,46
13, 46,49, 52,
54
13
-------
mollusc
Clam (Hypanis colorata)
mollusc
European fingernail clam (Sphaerium
corneum)
mollusc
Golden mussel (Limnoperna fortune!)
mollusc
Chinese mystery snail [(prosobranch]
(Cipangopaludina chinensis malleata)
mollusc
New Zealand mudsnail [hydrobid]
(Potamopyrgus antipodarum)
mollusc
Quagga mussel (Dreissena bugensis)
mollusc
Snail (hydrobid) (Lithoglyphus
naticoides)
mollusc
Snail (neritid) (Theodoxus fluviatilis)
mollusc
Snail (neritid) (Theodoxus pallasi)
mollusc
Zebra mussel (Dreissena polymorpha)
No
Yes
No
Yes
Yes
Yes
No
No
No
Yes
NEK
Yes
Yes
Yes
Yes
Yes
NEK
NEK
NEK
Yes
NEK
Present
Med
High
Present
Present
NEK
NEK
NEK
Present
NEK
NEK
Med
Med
High
Med
NEK
NEK
NEK
High
Caspian
Sea
Eurasia
China/
Asia
Asia
New
Zealand
Ukraine/
Ponto-
Caspian
Sea
Ponto-
Caspian
Baltic/
Black
Sea
Caspian
Sea
Ponto-
Caspian
Sea
ballast
water
unknown -
canals?
ballast
water
fish market;
ornamental
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
potential to invade Great Lakes
found in L. Huron, Ontario; effect unknown
produces a rapid change in benthic communities
and threatens native biodiversity; produces
macrofouling in the water systems of facilities;
spreading in South America
isolated pop. in L. Erie and upper St. Lawrence
River; established L. Michigan; clog screens of
water intake pipes; vectors for the transmission
of parasites and diseases
established in L. Ontario, Superior; invasion
history; reduce native species, harm trout
populations; suspected that can alter primary
production of streams; spread rapidly
found in L. Erie, Ontario; invasion history;
historically dominated the biomass of
transition(30-50m) and profundal regions(>50m);
negative impact on raw water-using industries,
potable water treatment plants; damage cost
data available
potential to invade Great Lakes
potential to invade Great Lakes
potential to invade Great Lakes
already established in Great Lakes; invasion
history; interferes with native molluscs' ecological
functions; damage cost data available
13,30
52
27, 49
31 , 42, 46, 52
13, 23,28, 44,
46, 49, 52
9, 26, 29, 30,
32, 44, 49, 52,
53, 54, 56
13
13
13
2, 9, 29, 32, 44,
48, 49, 52, 53,
54, 56
-------
plant
Curly-leaf pondweed (Potamogeton
crispus)
plant
Dwarf hygrophila (Hygrophila
polysperma)
plant
Brazilian elodea or waterweed (Egeria
densa)
plant
Eurasian watermilfoil (Myriophyllum
spicatum)
plant
European frogbit (Hydrocharis morus-
ranae)
plant
European water clover (Marsilea
quadrifolia)
plant
Fan wort (Cabomba caroliniana)
plant
Flowering rush (Butomus umbellatus)
plant
Giant salvinia (Salvinia molesta)
Yes
No
No
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Present
High
Med
High
Med
Med
Present
Med
Med
Med
Med
High
High
High
Med
High
NEK
Med
Eurasia
India
Central/
South
America
Eurasia
Eurasia
Eurasia
Southeas
t United
States
Eurasia
South
America
ornamental;
horticulture
ornamental
ornamental
ornamental;
boaters
ornamental;
horticulture;
boaters
ornamental
ornamental
ornamental;
horticulture
ornamental;
horticulture
established in U.S.; invasion history; causes
problems due to excessive growth
tolerate low temp; established in FL, TX, VA;
invasion history; clogs irrigation and flood-control
canals
established in OR, NY, MD, CT; invasion history;
slow dispersal; create dense mats that can
impede water recreation; water movement is
restricted causing fish population imbalances;
cause fluctuations in water quality
invaded L Ontario; interferes with water
recreation; canopy can crowd out important
native plants; decrease oxygen levels when plant
decays; rapid colonization; damage cost data
available; threat factor to fish
invaded L. Ontario (only lake found); creates
dense mats of vegetation and thus prevents light
and nutrients from reaching submerged
vegetation; plants die in the fall so depleted
dissolved oxygen is possible
found in L. Ontario; poses a realistic nuisance
threat to ecosystems; affect local molluscan
communities
Already invaded L. Michigan; clogs drainage
canals and freshwater streams; form dense
stands crowding out previously well-established
plants
spread in limited areas in Great Lakes (Erie,
Ontario, Michigan); actively expanding; competes
with native shoreline vegetation
established in lower U.S.; invasion history;
impede the flow of water to irrigation pipes and
other water intake pipes; rapidly expanding
populations can overgrow and replace native
plants, resulting dense surface cover prevents
light and atmospheric oxygen from entering the
water; decomposing material drops to the
bottom, greatly consuming dissolved oxygen
needed by fish and other aquatic life
22, 41 , 47, 48,
49,56
14, 31
14,31,36,46,
49
9,17,23,47,
48, 49, 54, 55,
56
9,17,22,36,
41 , 46, 49, 56
36
17, 31,36, 49,
22, 46, 48, 49,
56
22, 41 , 46, 49
-------
td
plant
Hydrilla (Hydrilla verticillata)
plant
Minor (slender) naiad (Najas minor)
plant
Parrot's feather (Myriophyllum
aquaticum)
plant
Purple loosestrife (Lythrum salicaria)
plant
Sessile joyweed (Alternanthera
sessilis)
plant
Spiny naiad (Najas marina)
plant
Variable-leaved watermilfoil
(Myriophyllum heterophyllum)
plant
Water primrose (Ludwigia
uruguayensis)
plant
Waterchestnut (Trapa natans)
No
Yes
No
Yes
No
Yes
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
NEK
Yes
High
NEK
High
Present
Med
Med
High
NEK
High
High
NEK
High
High
NEK
NEK
Med
NEK
Med
Central
Africa
Europe
South
America
Eurasia
Asia
Eurasia
Eastern
North
America
South
America
Eurasia
ornamental;
horticulture
ornamental
ornamental
ornamental;
horticulture
ornamental;
horticulture
ornamental
ornamental;
boaters
ornamental
ornamental;
horticulture
established in southern U.S.; high movement;
invasion history; adaptable; dense mats of
hydrilla will alter the waters chemistry by raising
pH, cause wide oxygen fluctuations, and
increase water temperature; eliminate native
plants
limited in L. Erie; can form dense, monospecific
stands in shallow water and hinder swimming,
fishing, boating, and other forms of water contact
recreation
invasion history; very adaptive to variety of
environments
established in Great Lakes; invasion history;
plant can form dense, impenetrable stands which
are unsuitable as cover, food, or nesting sites for
a wide range of native wetland animals; damage
cost data available
invasion history; aggressive
found in L. Ontario; widespread; interferes with
recreational boating; pose a realistic nuisance
threat to ecosystems
spreading in New England states; out compete
native aquatic vegetation, resulting in nearly
monotypic growth with less habitat value;
interferes with recreational boating; pose a
realistic nuisance threat to ecosystems
large colonies can prevent small boat navigation
and recreational use of shoreline areas
released in L. Ontario; reproduce rapidly; habitat
degradation through floating mats, hindering
navigation of waters and inhibiting the growth of
native aquatic plant species; damage cost data
available
22,36,41,46,
49,56
36,
1 4, 31 , 49
22,23,41,47,
48, 53, 54, 56
22
36
31,36,40
36
9, 36, 44, 46,
49, 54, 56
-------
plant
Yellow floating heart (Nymphoides
peltata)
protozoans
Amoebae (testatej (Psammonobiotus
communis)
protozoans
Amoebae (testate) (Psammonobiotus
dziwnowii)
protozoans
Amoebae (testate) (Psammonobiotus
linearis)
virus
Largemouth bass virus (Iridoviridae
family)
worm
Oligochaete (Potamothrix bedoti)
worm
Oligochaete (tubificid) (Potamothrix
heuscheri)
worm
Oligochaete (Potamothrix
moldaviensis)
worm
Oligochaete (Potamothrix vejdovskyi)
worm
Polychaete worm (Hypania invalida)
No
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
High
NEK
NEK
NEK
High
NEK
NEK
NEK
NEK
NEK
Med
NEK
NEK
NEK
Med
NEK
NEK
NEK
NEK
NEK
Eurasia
Baltic
Sea
Baltic
Sea
Baltic
Sea (?)
?
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
Ponto-
Caspian
Sea
ornamental;
horticulture
ballast
water
ballast
water
ballast
water
bait fish
ballast
water
ballast
water
ballast
water
ballast
water
ballast
water
spreading in New England; grows in dense
patches, excluding native species and even
creating stagnant areas with low oxygen levels
underneath the floating mats
found in L. Ontario; newly discovered in Great
Lakes
found in L. Ontario; newly discovered in Great
Lakes
found in L. Ontario and other Great Lakes
extended range in southeast basin; spreading in
L. Michigan; kills fish (commercial, native,
invasive)
established in Great Lakes; invasion history in
Baltic
invasion history in Baltic; continuously dispersing
to the west over Central Europe and to the north-
west towards the Baltic Sea
established in Great Lakes; invasion history in
Baltic
established in Great Lakes; invasion history in
Baltic
invasion history in Baltic; slow spread in Europe
36, 46, 49
9,24
9,24
9
34, 45, 46
13
13
13
13
13,30
-------
worm
Polychaete worm (Hypaniola
kowalevskyi)
No
Yes
NEK
NEK
Ponto-
Caspian
Sea
ballast
water
invasion history in Baltic; slow spread in Europe
13
aPresent in Great Lakes is scored "yes" if the organism has been reported for any of the Lakes.
blnvasion history is scored "yes" if the species is reported to have an invasion hi story/exotic in any other country of the papers cited for that
species.
°Spread potential within the Great Lakes after introduction scored "High" or "Med" based on comments in two or more of the papers cited for
that species, or NEK (not enough known).
Ecological impact if a species becomes established in Great Lakes scored as "High" or "Med" if so reported in at least two of the papers cited
for that species; else, it is scored as NEK (not enough known).
eSpecies origin notes the probable native area or region of occurrence for the species in question, as cited in the papers reviewed for the species.
fThe known, suspected, or probable mechanism of introduction of the species prior invasion history is listed based on the papers cited and prior
invasion history.
gComments related to investigators' concerns about the consequences of invasion for the species in question are abstracted from papers cited for
the species in question.
Articles referenced here were identified on the basis of species names plus reference to "Great Lakes". Thus, the listed references do not
include many papers that consider the biology of the organism in its native range.
-------
Table B-2. References that identify 156 invasive species as described in Table B-l.
Reference
Number
References
1
Horvath, TG; Oberdick, EJ; Last, LL; et al. (2001) Invading species dominate the Harpacticoid fauna in the nearshore sands of southern
Lake Michigan. NABS Annual Meeting, La Crosse Wl (abstract)
Buchan, LAJ; Padilla, DK. (1999) Estimating the probability of long-distance overland dispersal of invading aquatic species. Ecol Appl
9(1):254-265.
Corkum, LD; Sapota, RM; Skora, KE. (2004) The Round Goby, Neogobius malanostomus, a fish invader on both sides of the Atlantic
Ocean. Biol Invasions 6(2):173-181.
Cristescu, MEA; Witt, JDS; Grigorovich, IA; et al. (2004) Dispersal of the Ponto-Caspian amphipod Echinogammarus ischnus: invasion
waves from the Pleistocene to the present. Heredity 92:197-203.
Daniels, RA. (2001) Untested assumptions: the role of canals in the dispersal of Sea Lamprey, Alewife, and other fishes in the eastern
United States. Environ Biol Fishes 60(4):309-329.
Devin, S; Piscart, C; Beisel, JN; et al. (2003) Ecological traits of the amphipod invader Dikerogammarus villosus on a mesohabitat
scale. Arch Hydrobiol 158(1):43-56.
Dick, JTA; Platvoet, D. (2000) Invading predatory crustacean Dikerogammarus villosus eliminates both native and exotic species. Proc
Biol Sci 267(1447):977-983.
td
oo
Dick, JTA; Platvoet,D; Kelly, DW. (2002) Predatory impact of the freshwater invader Dikerogammarus villosus (Crustacea: Amphipoda).
Can J Fish Aquat Sci 59(6):1078-1084.
Duggan, 1C; Bailey, SA; Colautti, Rl; et al. (2003) Biological invasions in Lake Ontario: past, present, and future. In: Munawar, M, ed.
State of Lake Ontario (SOLO) - Past, Present, and Future.
10
Duggan, 1C; van Overdijk, DA; Bailey, SA; et al. (2005) Invertebrates associated with residual ballast water and sediments of cargo-
carrying ships entering the Great Lakes. Can J Fish Aquat Sci 62:2463-2474.
11
Froese, R; Pauly, D, Eds. (2005) FishBase.World Wide Web electronic publication, www.fishbase.org, version (10/2005). [12/07/05]
12
Goehle, M. (2004) Keeping track of aquatic invasive species in the Great Lakes. ANS Update. 9(4):1-2. Available online at
http://www.glc.org/ans/ansupdate/pdf/2004/Update-02-04.pdf
13
Grigorovich, IA; Colautti, Rl; Mills, EL; et al. (2003) Ballast-mediated animal introductions in the Laurentian Great Lakes: retrospective
and prospective analyses. Can J Fish Aquat Sci 60:740-756.
14
Holeck, KT; Mills, EL; Maclsaac, HJ; et al. (2004). Bridging troubled waters: Biological invasions, transoceanic shipping, and the
Laurentian Great Lakes. Bioscience 54(10):919-929.
15
Horvath, TG; Whitman, RL; Last, LL. (2001) Establishment of two invasive crustaceans (Copepoda: Harpacticoida) in the nearshore
sands of Lake Michigan. Can J Fish Aquat Sci 58:1261-1264.
16
Keppner, SM; Busiahn, TR; McClain, J; et al. (1997) Surveillance for Ruffe in the Great Lakes - an overview. G Lakes Res Rev
3(1):17-26.
17
Kerr, SJ; Brousseau, CS; Muschett, M. (2005) Invasive aquatic species in Ontario: a review and analysis of potential pathways for
introduction. Fisheries 30(7):21-30.
-------
Reference
Number
References
18
Kolar, CS; Lodge, DM. (2002) Ecological predictions and risk assessment for alien fishes in North America. Science 298:1233-1236.
19
Maclsaac, H; Grigorovich, IA; Ricciardi, A. (2001) Reassessment of species invasions concepts: the Great Lakes basin as a model.
Biol Invasions 3:405-416.
20
Maclsaac, HJ; Robbins, TC; Lewis, MA. (2002) Modeling ships' ballast water as invasion threats to the Great Lakes. Can J Fish Aquat
Sci 59:1245-1256.
21
Maclsaac, HJ; Borbely, JVM; Muirhead, JR; et al. (2004) Backcasting and forecasting biological invasions of inland lakes. Ecol Appl
14(3)773-783.
22
Maki, K; Galatowitsch, S. (2004) Movement of invasive aquatic plants into Minnesota (USA) through horticultural trade. Biol Conserv
118:389-396.
23
Mills, EL; Leach, JH; Carlton, JT; et al. (1994) Exotic species and the integrity of the Great Lakes. BioScience 44(10):666-676.
24
Nicholls, KH. (2005) Psammonobiotus dziwnowi and Corythionella georgiana, two new freshwater sand-dwelling testate amoebae
(Rhizopoda: filosea). Acta Protozool 44:271-278.
25
Palmer, ME; Ricciardi, A. (2004) Physical factors affecting the relative abundance of native and invasive amphipods in the St. Lawrence
River. Can J Zool 82:1886-1893.
td
26
Palmer, ME; Ricciardi, A. (2005) Community interactions affecting the relative abundances of native and invasive amphipods in the St.
Lawrence River. Can J Fish Aquat Sci 62:1111-1118.
27
Ricciardi, A. (1998) Global range expansion of the Asian mussel Limnoperna fortune! (Mytilidae): Another fouling threat to freshwater
systems. Biofouling 13(2):97-106.
28
Ricciardi, A. (2001) Facilitative interactions among aquatic invaders: is an "invasive meltdown" occurring in the Great Lakes? Can J
Fish Aquat Sci 58:2513-2525.
29
Ricciardi, A; Maclsaac, HJ. (2000) Recent mass invasion of the North American Great Lakes by Ponto-Caspian species. TREE
15(2):62-65.
30
Ricciardi, A; Rasmussen, JB. (1998) Predicting the identity and impact of future biological invaders: a priority for aquatic resource
management. Can J Fish Aquat Sci 55:1759-1765.
31
Rixon, CAM; Duggan, 1C; Bergeron, NMN; et al. (2005) Invasion risks posed by the aquarium trade and live fish markets on the
Laurentian Great Lakes. Biodiversity Conserv 14:1365-1381.
32
Vanderploeg, HA; Nalepa, TF; Jude, DJ; et al. (2002) Dispersal and emerging ecological impacts of Ponto-Caspian species in the
Laurentian Great Lakes. Can J Fish Aquat Sci 59(7):1209-1228.
33
Yan, ND; Girard, R; Boudreau, S. (2002) An introduced invertebrate predator (Bythotrephes) reduces zooplankton species richness.
Ecol Lett 5:481-485.
34
International Association for Great Lakes Research. (2002) Research and management priorities for aquatic invasive species in the
Great Lakes. Available online at http://www.iaglr.org/scipolicy/ais/aisjaglr02.pdf
35
CANOE News. (2005) Invasive fish species found in Lake Champlain, say experts from Quebec, Vermont. Online at
http://cnews.canoe.ca [11/21/05]
-------
Reference
Number
References
36
Madsen, J. (1999) A quantitative approach to predict potential nonindigenous aquatic plant species problems. ANS Update Fall/Winter
1999. Available online at http://www.glc.org/ans/ansupdate/pdf/ansdec99.pdf [11/22/05]
37
Cherwaty, S; Shear, H. (2002) Great Lakes indicators target aquatic invasive species. ANS Update Fall/Winter 2002. Available online
at http://www.glc.org/ans/ansupdate/pdf/ansv8n4.pdf [11/22/05]
38
Sturtevant, R. (2002) The Great Lakes waterflea report. ANS Update Spring/Summer 2002. Available online at
http://www.glc.org/ans/ansupdate/pdf/ansv8n2.pdf [11/22/05]
39
Froese, R; Pauly, D, Eds. (2005) FishBase.World Wide Web electronic publication, www.fishbase.org, version (10/2005). [12/07/05]
40
Rendall, J. (2003) Watercraft inspections: an opportunity to prevent the spread of aquatic nuisance species. ANS Update Winter/Spring
2003. Available online at http://www.glc.org/ans/ansupdate/pdf/Update-05-03.pdf [11/22/05]
41
Maki, K; Galatowitsch, S. (2003) Mail-order sales of aquatic plants: a pathway for ANS introduction. ANS Update Summer 2003.
Available online at http://www.glc.org/ans/ansupdate/pdf/Update08-03.pdf [11/22/05]
42
Higbee, E; Glassner-Shwayder, K. (2004) The live food fish industry: new challenges in preventing the introduction and spread of
aquatic invasive species. ANS Update Fall/Winter 2004. Available online at
http://www.glc.org/ans/ansupdate/pdf/2004/ANSUpdateFW.pdfI11/22/05]
td
to
o
43
Rasmussen, J. (2002) The Cal-Sag and Chicago sanitary and ship canal: a perspective on the spread and control of selected aquatic
nuisance fish species. US Fish and Wildlife Service, East Lansing Ecological Services Field Office, Great Lakes Coastal Program.
Online at http://www.fws.gov/midwest/LaCrosseFisheries/reports/Connecting_Channels_Paper_Final.pdf [11/30/05]
44
Canadian Wildlife Federation. (2003) Invasive Species in Canada. Available online at http://www.cwf-fcf.org/invasive/chooseSC.asp
[12/05/05]
45
Whelan, G. (2004) Largemouth bass virus continues to spread in Michigan waters. Michigan Department of Natural Resources,
Lansing, Ml. Available online at httpV/www.michigan.gov/dnr/0,1607,7-153-10364_10950-93547--,00.html [12/07/05]
46
Indiana Department of Natural Resources. (2005) Invasive species. Available online at http://www.in.gov/dnr/invasivespecies/
[12/09/05]
47
Klein, B. (2004) Making A List: Prevention Strategies for Invasive Plants in the Great Lakes States. Environmental Law Institute,
Washington D.C. Available online at http://www.elistore.org [11/23/05]
48
Minnesota Department of Natural Resources. (2005) Invasive species. Available online at
http://www.dnr.state.mn.us/invasives/index.html [12/09/05]
49
Invasive Species Specialist Group Global Invasive Species Programme. (2005) ISSG Global Invasive Species Database. Available
online at http://www.issg.org/database [01/2006]
50
Berg, D. (1992) The spiny water flea, Bythotrephes cederstroemi Another unwelcome newcomer to the Great Lakes. Ohio Sea Grant
College Program, Fact Sheet 49. Available online at http://sgnis.org/publicat/papers/bergdj92.pdf [01/18/06]
51
Stepien, CA; Tumeo, MA. (2006) Invasion genetics of Ponto-Caspian gobies in the Great Lakes: a 'cryptic' species, absence of founder
effects, and comparative risk analysis. Biol Invasions 8:61-78.
52
USGS (2005) NAS - Nonindigenous Aquatic Species. Available online at http://nas.er.usgs.gov/ [01/17/06]
-------
td
to
Reference
Number
53
54
55
56
References
Bailey, SA; Reid, DF; Colautti, Rl; et al. (2005) Management of nonindigenous species in
Science & Policy 5:101-112.
the Great Lakes. Toledo J Great Lakes' Law,
Pimentel, D. (2005) Aquatic nuisance species in the New York State Canal and Hudson River Systems and the Great Lakes Basin: An
economic and environmental assessment. Environ Manage 35(5):692-701 .
Dextrase, AJ; Mandrak, NE. (2006) Impacts of alien invasive species on freshwater fauna
Wisconsin Department of Natural Resources. (2005) Aquatic Invasive Species Program.
http://www.dnr.wi.gov/invasives/aquatic.htm [12/09/05]
at risk in Canada. Biol Invasions 8:13-24.
Available online at
-------
APPENDIX C. GARP MODEL VALIDATION
C-l
-------
GARP MODEL VALIDATION
The most rigorous evaluation of any model is a test of its ability to correctly predict
independent data that have not been processed by the model. In this report, Genetic Algorithm
for Rule-set Production (GARP) models were evaluated by comparing the known occurrences of
three invasive species already within the Great Lakes with the predictive ability of GARP models
developed for these species using occurrence data from other regions.
Model performance was assessed using area under the curve of the receiver operating
characteristic curve (Sing et al., 2005) using R 2.4.0 (R Development Core Team, 2006). Area
under the curve is a threshold-independent evaluation of model performance that, in this case,
measures the ability of the model to differentiate between sites where a species is considered
present versus where it is considered absent. Area under the curve represents the probability
that, when a predicted-present site and a predicted-absent site are drawn at random, the
predicted-present site will have a higher predicted value than the predicted-absent site. Because
true absence data were not available, randomly generated absence data, termed pseudo-absence
data (i.e., points selected randomly from sites where the species have not been recorded as
present within the Great Lakes), were used to validate the GARP models. This is a standard
approach when true absence data are not available (Graham et al., 2004).
GARP produces predictions of habitat suitability ranging from 0 to 100 that can be
converted to a binary prediction of presence or absence by selecting a threshold. For the purpose
of model validation, values above this threshold (i.e., 50) are considered present (assigned a
value of 1) while values below this threshold are considered absent (assigned a value of 0). The
threshold that is selected is typically the threshold that maximizes model performance which
may bias estimates of model performance. Area under the curve avoids the subjectivity in the
threshold selection process and, therefore, provides an unbiased evaluation of model
performance by plotting the false-positive rate (i.e., over-prediction, the rate at which the model
predicts the species to be present at sites at which it is considered absent) versus the true-positive
rate (i.e., the rate at which the model correctly predicts known presences as present) across all
possible thresholds. For these reasons, area under the curve is considered one of the best
approaches for model validation (Pearce and Ferrier, 2000). Nonetheless, area under the curve
poses three important limitations. Notable to his study are that (1) it weights over-prediction and
C-2
-------
under-prediction errors equally, (2) it does not give information about the spatial distribution of
prediction errors, and (3) the size of the study area to which models are projected influences the
rate of correctly predicted absences and the area under the curve scores.
Explanation of figures
The first figure, Figure C-l, is explained to help interpret the set of three figures. Figure
C-l shows the results from model validation for the zebra mussel, including a plot of the receiver
operating characteristic curve with the area under the curve statistic. The colors along this curve
correspond to the colors in the map of the zebra mussel predicted habitat suitability. Thus, by
moving along the curve, one can stop at any color transition, say between yellow and orange (or
a threshold value of 0.81 as determined by the right-hand y-axis or 81 from the legend in the
map). By moving horizontally from this threshold value to the left-hand y-axis, one can
determine the rate at which known presence correctly are predicted as present, also known as the
true-positive rate (about 0.75 in this example). By moving vertically downward from this
threshold to the x-axis, one can determine the rate at which pseudo-absences were predicted as
present (the true-false rate, which is about 0.3 in this case). In other words, if values greater than
81 in the map are considered as present and values less than 81 as absent, we would get roughly
75% of the known occurrences correctly predicted, but it would also predict roughly 30% of
presumed absences as present. In this manner, the receiver operating characteristic curve
provides a means to assess the rates of false-positive predictions (predicting a species present
where it is considered absent) and false-negative predictions (predicting a species absent where it
is known to be present).
Model Evaluation Results
Swets (1988) suggested the following scale for determining model performance using
area under the curve: 0.90-1.00 = excellent; 0.80-0.90 = good; 0.70-0.80 = fair; 0.60-0.70 =
poor; <0.60 = fail. The area under the curve for all 3 species falls between 0.74 and 0.79, so the
models would fall into the 0.70 to 0.80 category of "fair" (see Table C-l).
C-3
-------
Table C-l. Summary of area under the curve (AUC) values and occurrence
data sets used for model construction (training points) and evaluation points.
Evaluation points within the Great Lakes are shown as hollow points on
Figures C-l to C-3
Species and common name
Dreissena polymorpha
zebra mussel
Gymnocephalus cernuus
ruffe
Potamopyrgus antipodarum
New Zealand mud snail
No. of Great Lakes
evaluation points
238
46
10
No. and location of
training points
24 (Europe)
183 (Europe)
844 (Europe, Australia)
AUC
0.79
0.79
0.74
Model Evaluation
All three model validation Figures (C-l to C-3) show (1) the predicted habitat suitability
for each species within the Great Lakes when using only occurrence data from outside the Great
Lakes, (2) the corresponding receiver operating characteristic curve plot and area under the curve
value and bootstrap statistics for each model,
and (3) the occurrence data within the Great Lakes withheld from GARP and used for evaluation
of predictive performance (hollow points). Note that for reasons discussed under "Selecting
Species to Model and Development of Occurrence Data" in Section 3.1, these occurrence points
may not be inclusive of all known occurrences in the Great Lakes and represent only those
suitable for model evaluation.
Taken together, these area under the curve scores and the predicted distributions suggest
three important conclusions. First, in our tests, GARP models adequately predict the known
distributions of potential invasive species within the Great Lakes and, therefore, may be capable
of accurately identifying areas of the Great Lakes susceptible to aquatic invasive species that
have yet to be introduced. Thus, distribution data from a species' existing range can produce
useful predictions of invasion potential using these GARP methods. Second, the observed
patterns of invasion closely match those predicted for both known and potential invaders,
suggesting that Lakes Erie and Ontario, near-shore areas of all of the Great Lakes in general,
Saginaw Bay in Lake Huron, Lake St. Clair (located between Lakes Erie and Huron), and
Thunder Bay in Lake Superior are particularly prone to future invasion when considering
C-4
-------
environmental tolerances alone. Finally, the universally high area under the curve scores suggest
that the six environmental data layers we selected as inputs for the GARP models provide useful
information for predicting the potential distributions of invasive species within the Great Lakes.
In sum, the model validation exercise suggests that GARP predictions provide a useful
assessment of invasion potential, given the availability of adequate occurrence data outside the
Great Lakes.
REFERENCES FOR APPENDIX C.
Graham, CH; Ferrier, S; Huettman, F; et al (2004) New developments in museum-based
informatics and applications in biodiversity analysis. TRENDS Ecol Evol 19(9): 497-503.
Pearce, J; Ferrier, S. (2000) Evaluating the predictive performance of habitat models developed
using logistic regression. Ecol Model 133(3): 225-245.
R Development Core Team (2006). R: A Language and Environment for Statistical Computing.
R Foundation for Statistical Computing, Vienna, Austria. Available on-line at http://www.R-
project.org.
Sing, T; Sander, O; Beerenwinkel N; et al. (2005) ROCR: visualizing classifier performance in
R. Bioinformatics 21:3940-3941.
Swets, JA. (1988) Measuring the accuracy of diagnostic systems. Science 240:1285-1293.
C-5
-------
O
Zebra mussel (Dreissena polymorpha)
Predicted
Suitability
High (100r
Mid (50)
Low (0)
•n
O
Is
AUC:0.79
Bootstrap Statistics:
Bias: 0.001
Std, Error: 0.0125
Figure C-l. GARP model validation for zebra mussel showing predicted suitability and area under the curve (inset).
-------
O
Ruffe (Gymnocephalus cernuus)
Predicted
Suitability
High (100)%
Mid (50)
Low (0)
S
AUC: 079
Bootstrap Statistics:
Bias: 0 001
Std. Error: 0.038
04 "if,
False posrtire rale
08
Figure C-2. GARP model validation for ruffe showing predicted suitability and area under the curve (inset).
-------
O
oo
New Zealand mud snail (Potamopyrgus antiparuni)
Predicted
Suitability
High (100)'
Mid (50)
Low (0)
AUC: 0.74
Bootstrap Statistics:
Bias: N/A
Std. Error: N/A
ft
(J
04 06 08 10
•Ol
Figure C-3. GARP model validation for New Zealand mud snail showing predicted suitability and area under the curve (inset).
-------
APPENDIX D. GARP POWER OF PREDICTION ANALYSIS
D-l
-------
DETAILS OF HOW GARP POWER OF PREDICTION ANALYSIS WAS USED
If environmental conditions within a particular region of the Great Lakes differ from
those near a species' currently occupied range, then a prediction about the suitability of habitat
cannot be made (null prediction). A prediction cannot be made because no information exists to
predict whether a species under these novel conditions will or will not find suitable habitat.
Unfortunately, the Genetic Algorithm for Rule-set Production (GARP), along with most species
distribution models, simply predicts environments that are non-analogous to the environments
found near the species' currently occupied range to be unsuitable. But, in reality, such "null"
areas may actually be habitable by the species. GARP power of prediction analysis was
conceived to distinguish areas that do not provide suitable habitat (predicted absence) from areas
for which a prediction cannot be made (null prediction). More details about the theoretical basis
for using GARP power of prediction analysis are provided in Section 3.5. The remainder of
Appendix D presents the specific application of GARP power of prediction analysis to this study.
Power of prediction analyses were performed for 11 of the 14 species to help distinguish
the regions for which a prediction could be made from regions where a prediction could not be
made for each species. Power of prediction analysis was not performed for two of the invasive
species already established in the Great Lakes, quagga mussel (Dreissena bugensis) and round
goby (Neogobius melanostromus), because of a lack of occurrence data outside the Great Lakes
(i.e., these species had no distribution data describing their home range, the basis for a power of
prediction analysis). Also, because GARP models predict that the blueback herring (Alosa
aestivalis) may potentially find the entire Great Lakes region as suitable habitat, no power of
prediction analysis is needed, because power of prediction analyses are only concerned with
areas of predicted absence.
Power of prediction analyses for species can be grouped if they share a similar home
range region, eliminating the need to perform separate power of prediction analysis for each
species studied. The 11 species had home-range occurrence data originating from one or more of
four large regions: the northwest Pacific coast of North America, the northeast Atlantic coast of
North America, the southeastern coast of Australia, and a large region comprising the northern
coast of Europe and Northern Africa. Thus, we performed four power of prediction analyses.
D-2
-------
As discussed, power of prediction analyses are only concerned with areas of predicted
absence. Therefore, no power of prediction analyses were required for those species originating
from southeast Australia and the Atlantic U.S. regions because the models developed for species
from these regions predicted the entire Great Lakes region as suitable habitat. Similarly, no
power of prediction analysis was performed for the U.S. Pacific coast region because (1) only the
New Zealand mud snail, (Potamopyrgus antipodarum) had occurrence points within this region
and (2) this species was already predicted to be fairly widespread within the Great Lakes. Thus,
only Europe and North Africa remained as regions on which to focus power of prediction
analyses. This complex region was subdivided into three sub-regions: the United Kingdom and
northern Europe (designated UK) shoreline, the Baltic Sea, and the Ponto-Caspian Sea. GARP
power of prediction analyses were performed on combinations of these regions. The gray shaded
regions in Figures 7-17 and 24-27 reflect the results of the GARP power of prediction analysis.
Reverse power of prediction analysis, which processes in the opposite direction from
invaded range to home range, was also performed, but these results are not included here. When
we looked globally for places with conditions similar to those in the Great Lakes, we found
largely salt-water environments at homologous latitudes. This result is not surprising because
the vast majority of the earth's surface waters are oceanic. Assuming that salinity represents a
dispersal barrier, most species found in these saline, but otherwise similar aquatic environments,
would not be able to survive in the freshwater of the Great Lakes.
GROUPED RESULTS FOR 11 SPECIES
Figures Dl through D3 show the results of the GARP power of prediction analyses.
Multiple species are shown in most figures because the species are grouped according to the
presence of species occurrence points in each sub-region. Each of the 11 species collectively
shown in Dl through D3, can be associated with a corresponding GARP figure (Figures 8-17
and 31-34). Similar to the GARP predictions, the power of prediction analyses produced a
continuous measure of the ability to predict by adding the binary presence/absence outcomes
from the best 100 GARP models out of 1,000 total models constructed. For clarity, we
simplified the figures by converting these continuous predictions into two areas: areas where
prediction is possible (shown in red) and areas where prediction is not possible (shown in black).
Red areas represent location where more than 50 percent of the models have predictive power
D-3
-------
and black areas depict locations in which less than 50 percent of the models could predict
presence or absence.
The power of prediction analyses for the fishhook waterflea, the roach, the rudd, and the
ruffe (see Figure Dl) indicate that GARP can make predictions for these species everywhere
throughout the Great Lakes except for a central deep-water portion of Lake Superior. GARP
originally predicted a rather limited group of areas as susceptible for invasion by roach and rudd.
Coupled with the power of prediction analysis for these two species, we can conclude that
indeed, most of the areas thought to be low-risk areas are in fact predicted by the models as low-
risk areas. Similarly, Lakes Erie and Ontario were predicted to be particularly suitable for
invasion by both species, and the power of prediction analysis showed that the modeling
approach had good predictive capabilities within these two lakes. Susceptibility to invasion by
these two species in the upper Great Lakes was predicted to be limited primarily to near-shore
areas, and the power of prediction analysis shows that many of the areas predicted to be not
susceptible to invasion by these two species fall well within the red areas determined to be
predictable by GARP. Thus, these, too, are valid predictions of low risk of invasion by roach
and rudd. Therefore, the GARP model results, coupled with the power of prediction analysis,
successfully identify large areas of the Great Lakes as definitely unsuitable for invasion by roach
or rudd. These areas represent an opportunity for effectively focusing effort and resources for
monitoring these species.
The power of prediction analysis for the zebra mussel and New Zealand mud snail
indicate a large area within the Great Lakes that is predictable by GARP (see Figure D2 top
panel). Much of the area that is indicated as predictable based on the power of prediction
analysis is predicted to be highly susceptible to invasion by these two already-established NTS.
The GARP models predict that the deep-water areas of Lake Superior should be unsuitable and,
therefore, not at risk of future invasion by these species. However, the power of prediction
analysis for these species indicates that this area is simply not predictable. Therefore, the risk to
central Lake Superior from invasion by these two species is, at this point, undeterminable by
GARP models alone. GARP predictions show parts of Lake Huron to be at low risk from the
New Zealand mud snail and the power of prediction analysis shows that these areas can be well
predicted by GARP.
D-4
-------
The set of power of prediction analyses that were performed for Corophium curvispinum,
monkey goby, tubenose goby, and tench indicate that large areas within the Great Lakes are not
predictable by GARP (see Figure D2 lower panel). GARP predictability zones (red areas) are
restricted to Lake Erie, the southern half of Lake Ontario, and the southernmost tip of Lake
Michigan. All four of these potential invasive NTS were predicted by GARP models to
encounter large areas of the Great Lakes that were unsuitable as habitat, which could be
classified as immune to invasion. Many of these predicted absence areas actually fall within the
black zone of the power of prediction analysis, within which no GARP prediction can be made.
From the GARP predictions alone, these extensive areas would have been interpreted by
managers as no-invasion-risk zones. However, the areas actually are no-prediction possible
zones, meaning they might be susceptible to invasion and they might not. Without the benefit of
the power of prediction analysis, this condition would not have been known from the GARP
predictions alone. The power of prediction analysis prevents this critical misinterpretation.
When viewed in conjunction with the corresponding GARP power of prediction analysis
figures, the results present a clearer prediction of invasion-susceptible habitat areas. Without
power of prediction analysis, only 5 species (the blueback herring, the sand goby, the zebra
mussel, the ruffe, and the New Zealand mud snail) are predicted by GARP models to be able to
find suitable habitat throughout the Great Lakes. With the power of prediction analyses, it is
possible that the entire Great Lakes region could also be susceptible to three additional potential
invasive species: C. curvispinum, the tubenose goby, and the round goby. Managers cannot rule
out much of the Great Lakes as unsuitable for these three species, as was suggested by the GARP
predictions alone.
D-5
-------
POP Analysis: Fishook waterflea, Roach, Rudd
| Prediction Possible
No Prediction Possible
POP Analysis: Ruffe
| Prediction Possible
No Prediction Possible
Figure D-l. Power of prediction analysis for fishhook waterflea, roach, and rudd (top) and
ruffe (bottom)
D-6
-------
POP Analysis: Zebra mussel, New Zealand mud snail
Prediction Possible
No Prediction Possible
POP Analysis: C. curvispinum, monkey goby, tubenose goby, tench
| Prediction Possible
No Prediction Possible
Figure D-2. Power of prediction analysis for zebra mussel and New Zealand mud snail (top)
and C. curvispinum, monkey goby, tubenose goby and tench (bottom).
D-7
-------
POP Analysis: Sand goby
| Prediction Possible
No Prediction Possible
Figure D-3. Power of prediction analysis for sand goby.
D-8
-------
APPENDIX E. TABLES DEPICTING SOURCES AND DESTINATION OF
BALLAST WATER DISCHARGES IN U.S. GREAT LAKES
E-l
-------
Table E-l. Source of ballast water discharges (prior to ballast water exchange) into U.S.
Great Lakes ports (2006-2007). Includes only discharges from vessels whose source of ballast
water (prior to ballast water exchange) came from outside the Great Lakes.
Source of Ballast Water
Not from POC
Port C artier
Antwerpen
Baie Comeau
Port Alfred
Puerto Cabello
Sept lies
Haraholmen
Bremen
Manfredonia
St John
Ghent
Amsterdam
New Haven
Safi
Casablanca
Hafnarfjordur
Santander
Albany
Rotterdam
Hamburg
Houston
Thisvi
Workington
Yokkaichi
Agadir
Country
—
Canada
Belgium
Canada
Canada
Venezuela
Canada
Sweden
Germany
Italy
Canada
Belgium
Netherlands
US
Morocco
Morocco
Iceland
Spain
US
Netherlands
Germany
US
Greece
UK
Japan
Morocco
Ballast
Tanks
Discharged
79
57
40
34
34
29
28
22
20
18
15
14
13
13
13
12
12
12
11
8
7
7
7
7
7
6
Volume
Discharged
(metric tons)
31,323
80,233
14,134
46,353
18,699
14,267
16,014
3,586
5,858
4,364
3,680
7,332
3,613
17,252
7,231
10,552
9,718
10,633
1,427
1,389
4,374
1,593
1,998
1,477
1,512
3,383
Vessels
Discharging
20
8
4
5
2
3
5
1
1
1
1
1
1
1
1
2
1
2
1
2
1
2
1
1
1
1
E-2
-------
Source of Ballast Water
Rouen
South Shields
Bahia Quintero
Mostaganem
Philadelphia
Porsgrunn
Algeciras
Baltimore
Barranquilla
Ciwandan
Dunkerque (east)
Hancock
Troy
Brake
Shuwaikh
Al Burayqah
Dammam
Gros Cacouna
Holyhead
Mina Jabal Ali
Mobile
Oulu
Searsport
Southampton
Chiba Ko
Halifax
Montevideo
Port Of Pasajes
Santa Marta
Country
France
UK
Chile
Algeria
US
Norway
Spain
US
Colombia
Indonesia
France
US
US
Germany
Kuwait
Libya
Saudi Arabia
Canada
UK
UAE
US
Finland
US
UK
Japan
Canada
Uruguay
Spain
Colombia
Ballast
Tanks
Discharged
6
6
5
5
5
5
4
4
4
4
4
4
4
O
O
2
2
2
2
2
2
2
2
2
1
1
1
1
1
Volume Discharged
(metric tons)
2,006
1,926
16,839
1,196
5,299
877
490
8
5,098
491
2,194
9,632
57
970
739
350
360
2,426
1,374
577
612
182
395
548
379
910
237
126
160
Vessels
Discharging
1
1
1
1
1
1
1
1
2
1
1
1
4
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
-------
Source of Ballast Water
Sundsvall
Thessaloniki
Veracruz
Wilmington
Country
Sweden
Greece
Mexico
US
TOTAL
Ballast
Tanks
Discharged
1
1
1
1
618
Volume Discharged
(metric tons)
254
1,000
719
1,500
381,927
Vessels
Discharging
1
1
1
1
107
E-4
-------
Table E-2. Ballast water discharges at U. S. Great Lakes during 2006-2007. Includes only
vessels whose original source of ballast water (prior to ballast water exchange) came from
outside the Great Lakes.
US Great
Lake Port
Duluth
Toledo
Superior
Green Bay
Gary
Milwaukee
Oswego
Chicago
Ludington
Erie
Lorain
Menominee
Ashtabula
TOTAL
Tanks
Discharged
407
85
50
18
17
11
8
7
5
4
3
2
1
618
Volume
Discharged
(metric tons)
184,844
65,335
78,085
5,984
11,154
10,768
1,239
17,916
1,913
490
2,320
380
1,500
381,927
Vessels
Discharging
58
13
10
4
4
2
5
O
1
1
3
2
1
107
E-5
-------
Table E-3. Source of 2006-2007 ballast water discharges from vessels whose original source
of ballast water (prior to ballast water exchange) came from outside the Great Lakes.
Sorted by U.S. Great Lake port of call.
US Great
Lake Port
Ashtabula
Chicago
Chicago
Chicago
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Source of Ballast
Water
Wilmington
Bahia Quintero
Not from POC
Thessaloniki
Agadir
Al Burayqah
Albany
Amsterdam
Antwerpen
Baie Comeau
Casablanca
Chiba Ko
Ciwandan
Dammam
Dunkerque (east)
Ghent
Hamburg
Haraholmen
Houston
Manfredonia
Mina Jabal Ali
Mobile
Montevideo
Mostaganem
New Haven
Not from POC
Oulu
Country
US
Chile
—
Greece
Morocco
Libya
US
Netherlands
Belgium
Canada
Morocco
Japan
Indonesia
Saudi Arabia
France
Belgium
Germany
Sweden
US
Italy
UAE
US
Uruguay
Algeria
US
—
Finland
Tanks
Discharged
1
5
1
1
6
2
11
13
33
O
12
1
4
2
4
14
7
22
7
18
2
2
1
5
13
66
2
Volume
Discharged
(metric
tons)
1,500
16,839
77
1,000
3,383
350
1,427
3,613
11,668
1,922
10,552
379
491
360
2,194
7,332
4,374
3,586
1,593
4,364
577
612
237
1,196
17,252
27,054
182
Vessels
Discharging
1
1
1
1
1
1
1
1
2
1
2
1
1
1
1
1
1
1
2
1
1
1
1
1
1
13
1
E-6
-------
U.S. Great
Lake Port
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Duluth
Erie
Gary
Gary
Gary
Gary
Green Bay
Green Bay
Green Bay
Green Bay
Lorain
Lorain
Lorain
Source of Ballast
Water
Philadelphia
Port Alfred
Port Cartier
Puerto Cabello
Rotterdam
Safi
Santa Marta
Santander
Searsport
Sept lies
Shuwaikh
Southampton
St John
Thisvi
Workington
Yokkaichi
Algeciras
Baie Comeau
Not from POC
Porsgrunn
Rotterdam
Antwerpen
Brake
Rouen
South Shields
Barranquilla
Not from POC
Veracruz
Country
US
Canada
Canada
Venezuela
Netherlands
Morocco
Colombia
Spain
US
Canada
Kuwait
UK
Canada
Greece
UK
Japan
Spain
Canada
—
Norway
Netherlands
Belgium
Germany
France
UK
Colombia
—
Mexico
Tanks
Discharged
5
27
13
26
6
13
1
12
2
11
O
2
15
7
7
7
4
9
1
5
2
3
3
6
6
1
1
1
Volume
Discharged
(metric
tons)
5,299
11,679
15,803
11,719
1,249
7,231
160
10,633
395
6,024
739
548
3,680
1,998
1,477
1,512
490
10,077
60
877
140
1,082
970
2,006
1,926
780
821
719
Vessels
Discharging
1
1
2
2
1
1
1
2
1
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
E-7
-------
U.S. Great
Lake Port
Ludington
Menominee
Menominee
Milwaukee
Milwaukee
Oswego
Oswego
Superior
Superior
Superior
Superior
Superior
Superior
Superior
Toledo
Toledo
Toledo
Toledo
Toledo
Toledo
Toledo
Toledo
Toledo
Toledo
Toledo
Toledo
Source of Ballast
Water
Sept lies
Port OfPasajes
Sundsvall
Baie Comeau
Not from POC
Not from POC
Troy
Baie Comeau
Barranquilla
Hancock
Not from POC
Port Cartier
Puerto Cabello
Sept lies
Antwerpen
Baie Comeau
Baltimore
Bremen
Gros Cacouna
Hafnarfjordur
Halifax
Holyhead
Not from POC
Port Alfred
Port Cartier
Sept lies
Country
Canada
Spain
Sweden
Canada
—
—
US
Canada
Colombia
US
—
Canada
Venezuela
Canada
Belgium
Canada
US
Germany
Canada
Iceland
Canada
UK
—
Canada
Canada
Canada
TOTAL
Tanks
Discharged
5
1
1
9
2
4
4
8
3
4
1
28
O
O
4
5
4
20
2
12
1
2
O
7
16
9
618
Volume
Discharged
(metric
tons)
1,913
126
254
10,077
691
1,182
57
16,988
4,318
9,632
95
42,855
2,548
1,649
1,384
7,290
8
5,858
2,426
9,718
910
1,374
1,343
7,020
21,576
6,428
381,927
Vessels
Discharging
1
1
1
1
1
1
4
1
1
1
1
4
1
1
1
1
1
1
1
1
1
1
1
1
2
1
107
-------
Table E-4. Number of vessel-trips including specified port as a last five
ports of call, for NOBOB-RM vessels (2006). NOBOB-RM vessels are
vessels that entered the Seaway without ballast on board, picked up ballast
water in the Great Lakes, and then discharged the ballast water along with
residual material at a Great Lake port. NOBOB-RM vessels must also have
visited a port outside the Great Lakes during one of the last five ports of call.
Port outside St.
Lawrence Seaway
Sept lies
Port Cartier
Ijmuiden
Baie Comeau
Pointe Noire
Halifax
Europa Point
Belledune
Antwerpen
Casablanca
Riga
Santos
Rotterdam
Stephenville
Amsterdam
Ashdod
Police
Praia Mole
Tynemouth
Dunkerque (east)
Tunis
Ghent
Gdansk
Swinoujscie
Oran
Country
Canada
Canada
Netherlands
Canada
Canada
Canada
Gibraltar
Canada
Belgium
Morocco
Latvia
Brazil
Netherlands
Canada
Netherlands
Israel
Poland
Brazil
UK
France
Tunisia
Belgium
Poland
Poland
Algeria
NOBOB
L5POC
145
77
23
23
22
12
10
10
9
9
8
8
7
7
7
7
6
6
5
5
5
5
5
5
5
E-9
-------
Port outside St.
Lawrence Seaway
Syros Island
Puerto Quetzal
Magdalen
Sankt-peterburg
Kao-hsiung
Vitoria
Charlottetown
Hamburg
La Goulette
Richards Bay
Coatzacoalcos
Balboa
Corinto
Piombino
Barranquilla
Bahia San Nicolas
Arkhangelsk
Puerto Cabello
Rocky Point
Rouen
Tarragona
Puntarenas
Summerside
Porto De Maceio
Paranagua
Santander
Shanghai
Chi -lung
Setubal
Bunbury
Country
Greece
Guatemala
Canada
Russia
Taiwan
Brazil
Canada
Germany
Tunisia
South Africa
Mexico
Panama
Nicaragua
Italy
Colombia
Peru
Russia
Venezuela
Jamaica
France
Spain
Costa Rica
Canada
Brazil
Brazil
Spain
China
Taiwan
Portugal
Australia
NOBOB
L5POC
5
4
4
4
3
3
3
3
3
3
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
1
1
1
1
E-10
-------
Port outside St.
Lawrence Seaway
Singapore
Cartagena
Campana
Ceuta
Brunsbuttel
Sao Francisco
Sluiskil
Fos
Szczecin
Terneuzen
Avonmouth
Trieste
Venezia
Argentia
Aratu
Xingang
Annaba
Bremen
Recife
Port Alfred
Port St Joe
Point Tupper
Yosu-bando
Porto Empedocle
Map Ta Phut
Liverpool
Las Palmas
La Plata
Durban
Kingston Upon Hull
Country
Singapore
Spain
Argentina
Spain
Germany
Brazil
Netherlands
France
Poland
Netherlands
UK
Italy
Italy
Canada
Brazil
China
Algeria
Germany
Brazil
Canada
USA
Canada
Republic of Korea
Italy
Thailand
UK
Spain
Argentina
South Africa
UK
NOBOB
L5POC
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
E-ll
-------
Port outside St.
Lawrence Seaway
Djen-djen
Itea
Reka Luga
Huelva
Hong Kong
Guayaquil
Safi
Geraldton
San Lorenzo
Port Of Spain
Santa Marta
Pusan
Country
Algeria
Greece
Russia
Spain
Hong Kong
Ecuador
Morocco
Australia
Argentina
Trinidad and Tobago
Colombia
Republic of Korea
TOTAL
NOBOB
L5POC
1
1
1
1
1
1
1
1
1
1
1
1
544
E-12
-------
Table E-5. Frequency and volume of ballast tank discharges into U.S.
Great Lakes ports of call from NOBOB-RM vessels (2006). NOBOB-RM
vessels are vessels that entered the Seaway without ballast on board, picked up
ballast water in the Great Lakes, and then discharged the ballast water along
with residual material at a Great Lake port. NOBOB-RM vessels must also
have visited a port outside the Great Lakes during one of the last five ports of
call.
U.S. Great
Lake port
Toledo
Superior
Ashtabula
Duluth
Sandusky
Milwaukee
Gary
Chicago
Conneaut
Buffalo
Calumet
Lorain
Two Harbors
Brevort
TOTAL
Ballast Tanks
Discharged
353
340
297
289
116
89
60
48
43
37
30
13
8
7
1730
Volume Discharged
(metric tons)
511,181
524,446
579,785
379,999
243,137
68,522
52,987
76,607
76,335
73,197
33,974
25,729
13,444
20,914
2,680,255
E-13
------- |