Risk Analysis, Vol. 24, No. 4, 2004

Risk Analysis for Invasive Species: General Framework and Research Needs Mark C. Andersen,1 ∗ Heather Adams,1 Bruce Hope,2 and Mark Powell3

A joint workshop was convened by the Society for Risk Analysis Ecological Risk Assessment Specialty Group and the Ecological Society of America Theoretical Ecology Section to provide independent scientific input into the formulation of methods and processes for risk assessment of invasive species. In breakout sessions on (1) the effects of invasive species on human health, (2) effects on plants and animals, (3) risk analysis issues and research needs related to entry and establishment of invasive species, and (4) risk analysis issues and research needs related to the spread and impacts of invasive species, workshop participants discussed an overall approach to risk assessment for invasive species. Workshop participants agreed on the need for empirical research on areas in which data are lacking, including potential invasive species, native species and habitats that may be impacted by invasive species, important biological processes and phenomena such as dispersal, and pathways of entry and spread for invasive species. Participants agreed that theoretical ecology can inform the process of risk assessment for invasive species by providing guidelines and conceptual models, and can contribute to improved decision making by providing a firm biological basis for risk assessments. KEY WORDS:

Invasive species; risk assessment; theoretical ecology

1. INTRODUCTION: WORKSHOP BREAKOUT SESSIONS

risk analysis for invasive species, as discussed in the introductory article of this special section.(3) Workshop participants were given the opportunity to discuss an overall approach to risk assessment for invasive species, as well as research needs and unique features of invasive species problems relative to other types of risk analysis, in four breakout sessions. These sessions focused on (1) the effects of invasive species on human health, (2) effects on plants and animals, (3) risk analysis issues and research needs related to entry and establishment of invasive species, and (4) risk analysis issues and research needs related to the spread and impacts of invasive species. Here we present the results of each breakout session. We conclude by briefly placing the findings of the workshop in the broader context of both risk analysis and invasive species ecology, discussing the limitations of the risk analysis approach, and presenting the implications of the workshop’s findings for invasive species policy implementation.

United States Executive Order 13112 of 1999 mandates the formulation of a National Invasive Species Management Plan, which must include a risk assessment program.(1) Implementation of the plan will require careful application of the principles of risk assessment (for a comparable situation, see Lonsdale et al.).(2) Theoretical ecology has much to contribute to a deeper understanding of the unique features of

Department of Fishery and Wildlife Sciences, New Mexico State University, Las Cruces, NM 88003-003, USA. 2 Oregon Dept of Environmental Quality, Portland, OR 972152654, USA. 3 USDA Office of Risk Assessment and Cost Benefit Analysis, Washington, DC 20250, USA. ∗ Address correspondence to Mark C. Andersen, Department of Fishery and Wildlife Sciences, New Mexico State University, Las Cruces, NM 88003-003, USA; [email protected]. 1

893

C 2004 Society for Risk Analysis 0272-4332/04/0100-0893$22.00/1 

894 2. BREAKOUT SESSION 1: EFFECTS OF INVASIVE SPECIES ON HUMAN HEALTH Workshop participants observed that epidemiologic theory and methods provide a well-developed foundation for assessing the risks to public health associated with nonindigenous species. Different disease etiologies, however, suggest a continuum of analytic approaches. In terms of ecological complexity, the simplest end of the continuum is marked by obligate human pathogens that are transmitted exclusively person-to-person, either directly or indirectly via formites. The most basic epidemic model is the deterministic Susceptible-Infective-Removed (SIR) model, in which the population is divided into three classes—susceptibles, infectives, and removed.(4) Population-level epidemic models with additional complexity may incorporate demographic and/or environmental stochasticity and additional population classes (e.g., exposed individuals who are not yet infective). Epidemiologic models that incorporate community-level dynamics are required for vector-borne diseases and zoonoses.(5–7) In terms of ecological complexity, vector-borne diseases and zoonoses with few intermediate hosts represent intermediate points on the continuum. The far end of the spectrum is marked by pathogens characterized by intricate life cycles and complex ecological interactions among multiple animal hosts and environmental reservoirs. Spatially explicit epidemiologic models(8) can be developed over a range of spatial scales, depending upon the availability of spatially referenced data. The uncertainty of population projections for any organism also complicates longterm predictions of epidemic trajectories. In assessing health risks from nonindigenous organisms, it is also useful to distinguish between emergent pathogens(9) and pathogens that are endemic outside the borders of a region or country. Emergent pathogens include previously benign organisms that have acquired virulence factors through evolution; pathogens that have been eradicated from a region but are potentially reemergent due to changing environmental conditions; and pathogens previously affecting nonhuman populations to which humans recently have become exposed due to changing activity patterns and/or environmental conditions, or due to the introduction of a new vector that permits transmission to humans. Note also that introductions may be intentional or unintentional. Regarding research and data acquisition needs for public health endpoints, workshop participants

Andersen et al. underscored the need for data that can be used to identify and characterize the flows of commercial trade and travel representing significant pathways for the introduction of nonindigenous pathogens. Useful additional epidemiologic data can be acquired in a variety of modes, including reactive (for surprising diagnoses), passive, and active public health surveillance, as well as surveillance of domestic animals and wildlife populations that may harbor human pathogens. Although there may be overlapping authorities and responsibilities among various public agencies (e.g., public health, animal health, plant protection, and resource management), the workshop participants recommended that decisions regarding sanitary and phytosanitary measures designed to protect animal or plant health should consider the potential for unintended public health consequences.

3. BREAKOUT SESSION 2: EFFECTS OF INVASIVE SPECIES ON PLANTS AND ANIMALS Workshop participants emphasized that several types of systems may be impacted by invasive species, including production agricultural systems, urban and other human-dominated nonagricultural systems, and natural systems. For the first two, there are wellunderstood predefined currencies with which to evaluate the impacts of invasive species, such as crop or livestock losses, control costs, real estate values, etc. For natural systems, however, there are no commonly agreed-upon currencies with which to evaluate impacts on biodiversity (e.g., species richness), ecosystems (through impacts on ecosystem functioning and ecosystem services), and potential resources (rangelands, timber, etc.).(10) While workshop participants understand and agree that it is the responsibility of stakeholders to provide valuation systems for potential impacts, all agreed that there is a need for more well-developed systems of societal valuation of natural resources, and that associated currencies and utilities need to be operationalized and clarified.(11,12) The desired output of any risk assessment for an invasive species may be considered to be a map of the risks of various kinds of impacts over an operationally defined region, with separate maps for each resource impacted or other potential impact. The initial steps in attaining this goal should include definition of assessment endpoints through enumeration of the finite set of resources that may be affected (with each operationally defined in an explicit way), and the

Risk Analysis for Invasive Species development of a rigorous conceptual model for the suite of processes leading to potential impacts. This conceptual model must satisfy two criteria, namely, a structural criterion (the model can be neither too complex to be tractable nor too simple to be realistic) and a utility criterion (the model has to be useful for assessing potential impacts of the invasive species). Construction of this conceptual model is crucial to the success of any risk assessment.(13) The model must include any a priori constraints on the model (e.g., factor X must be included, while factor Y cannot be). These constraints can be either imposed by the stakeholders and decision makers (e.g., spatiotemporal boundaries) or by the biology of the system being considered. The process of model construction needs to recognize that mechanistic models on the one hand and ad hoc phenomenological models on the other are ends of a continuum, and that informative models can be constructed that fall practically anywhere along this continuum. Once the assessments endpoints have been identified and a conceptual model has been constructed, the risk assessment process should continue with gathering available data on the processes included in the conceptual model and sensitivity analyses of a stochastic simulation model proceeding in parallel. Construction of the model should proceed through a process of iterative model refinement subject to time and budgetary constraints. The model structure and analysis may indicate the need for additional data. The model ideally should reflect a multidisciplinary perspective, incorporating organismal biology, theoretical/mathematical ecology, ecosystem ecology, and economics. The model should include explicit recognition of the various phases of the invasion process and the different potential impacts and risks associated with each phase. In addition, the ideal model is applicable and effective at multiple spatial scales and explicitly recognizes jurisdictional boundaries and other features of the societal landscape. Theoretical ecology can inform the process of risk assessment for invasive species in a number of ways. These include at least the following (note that all of these may be subject to other inputs as well): 1. Guidelines based on sound ecological principles for operationalizing the region to be considered and the processes likely to be operating. 2. Guidelines for data collection (not just how to collect data, but what variables are likely to be important).

895 3. A conceptual basis for identifying assessment endpoints. 4. A conceptual basis for assessing aptness of the model as well as the appropriate level of complexity. 5. Guidelines for inference, preferably through empirical Bayesian(14) or resampling(15) schemes. 6. A conceptual basis for predicting the range of options, outputs, and dynamical behaviors possible for a given model. Workshop participants identified the following research needs: 1. Spatially explicit, multiscale decision support systems will contribute to better decision making through enhanced credibility, an explicit and direct relationship with managing for sustainability, and explicit illustration of tradeoffs and the cost of inaction. 2. Better understanding of landscape ecology and landscape structure in relation to invasive species biology will contribute to better decision making by providing an explicit, habitatoriented spatial context for decisions, and will perhaps also allow identification of expanded management options. 3. Investigation of linkages between actual or potential invasive species and other stressors, including other species (such as other invasive species, or threatened and endangered species), would contribute to better decision making by providing a broader context for the assessment of potential risks. 4. Studies of the impacts of invasive species in natural systems (especially in terms of the impacts on biodiversity, and on ecosystem function/services), particularly if such studies could lead to better prediction of impacts, would contribute to better decision making by providing a more complete understanding of the possible range and likelihood of impacts. 5. A better conceptual framework for the economic valuation of biodiversity and ecosystem services would contribute to better decision making by providing for better risk communication among analysts, decision makers, and stakeholders, especially in terms of clarifying options with respect to different impacts. 6. Expanded pest lists would provide reliable baseline information of uniformly high

896

Andersen et al. quality and availability; such lists would be particularly useful for managers of natural areas. Pest lists should be expanded both taxonomically and in terms of the biological information they provide, and would be produced according to a uniform set of standards (possibly involving the use of risk assessments to produce the lists in the first place). An example of such a pest list (for plant pathogens) can be found at: http://www.apsnet.org/online/feature/exotic/.

4. BREAKOUT SESSION 3: ENTRY AND ESTABLISHMENT OF INVASIVE SPECIES Fig. 1 shows the basic risk assessment framework applied to entry and establishment of invasive species. Workshop participants agreed that the entry phase of biological invasion lends itself to a number of questions that could be addressed by theoretical ecologists. Thorough exploration of these avenues of research would have several benefits for the process of risk assessment for invasive species. 1. Gap analysis is a method for evaluating the status of biodiversity protection at large spatial scales by examining data on the spatial distribution of habitat types, the geographic ranges of native biota, and land use.(16–18) This approach and other coarse-filter biogeographic GIS-based approaches should be extended to assessment and planning relevant to invasive species. In general, there should be a thorough investigation of the relative contributions that could be made to risk assessment Source

exposure

Habitat

effect

Impact

(location)

(process)

(location)

(process)

(location)

Native range

transport

Point of entry

demography

Invading population

Fig. 1. The basic risk assessment framework applied to the entry and establishment of invasive species. The upper portion of the figure represents the standard ecological risk assessment framework; the lower portion shows the framework’s application to invasive species. The vertical line divides processes and locations pertaining to entry (on the left) from those pertaining to establishment (on the right). Note how different processes at different locations (and perhaps at different scales) interact in the overall processes of entry and establishment, and how these processes and locations are explicitly accounted for by the risk assessment paradigm.

for invasive species by species-based versus habitat-based approaches. 2. Pathway analyses(19) may be applied to the entry of invasives to yield estimates of the probability of entry. The approach may be refined to account for the timing, frequency, level, and location of entry. Applicable research approaches for theoretical ecologists to use in studying invasive species establishment are more mature than for entry, since there are many well-developed theoretical tools for studying the dynamics of small populations. It seems, in fact, that one of the major questions for the study of establishment is whether establishment is truly the “flip side” of extinction. If this is the case, then a number of well-studied theoretical tools can readily be brought to bear on questions of establishment. Indeed, population viability analysis(20) may prove to be more successfully applied to analyzing the establishment of invasive species than to predicting the extinction of threatened and endangered species because of the shorter time horizons involved in risk assessments for invasive species. We recommend a thorough exploration of the extent to which the methodologies of population viability analysis could be adapted for use on invasive species rather than endangered species. Associated with this must be an exploration of the relative merits of considering intrinsic biological thresholds (imposed by, for example, Allee effects(21) or other forms of density dependence) versus thresholds of acceptable risk (which may actually represent a viable established population). In addition, we recommend that theoretical ecologists examine the following questions with respect to the establishment of invasive species: 1. What is the importance of constraints imposed by the novel environment in which establishing invasives find themselves? In particular, how does the novel environment affect the importance of the Allee effects and carrying capacity? To what extent are the life histories of establishing invasives nonoptimal in the novel environment, and what role is played by phenotypic or behavioral plasticity and possibly rapid evolution? 2. What is the importance of hybridization and introgression with related native species? In particular, what are the relative impacts of these processes on slowing the rate of invasion versus the loss of native genotypes?

Risk Analysis for Invasive Species Theoretical ecologists should also strive to find ways to minimize or reduce the uncertainty associated with predictions of establishment probability, perhaps through applications of sensitivity and elasticity analyses, and assessments of the sensitivity of establishment probabilities to the various uncertainties.(22,23) This strategy would fit in well with an empirical Bayes approach to parameter estimation, using uninformative priors for the most highly uncertain parameters.(14) There is a strong need for “informed intuition” in scientific judgment via theory in the absence of situation-specific information about crucial system components, but in the presence of urgent need for action. We must recognize the spectrum from ad hoc empirical models to mechanistic models derived from first principles, and the potential usefulness of the entire spectrum for informing our scientific judgment. By addressing these research needs and recommendations, we will enable decision makers to make better decisions in a number of ways. In particular, we will be able to provide improved decision-making tools with a firm foundation in the well-developed theory of population biology. Better decision-making tools will provide the means to ensure that scientific arguments are thoroughly considered in the policymaking process, along with other legitimate societal values. A sound risk management policy is based on a process that is both analytic and deliberate, rather than one that is purely science-driven or negotiated. 5. BREAKOUT SESSION 4: SPREAD AND IMPACTS OF INVASIVE SPECIES Workshop participants observed that a number of models are currently available for predicting the spread of nonindigenous species.(23,24) The simple diffusive transport model(25) represents a continual random motion of individuals that results in a rate of spread proportional to the population concentration gradient and a normal distribution of abundance extending in two dimensions from the initial site of establishment. While the growth of the nonindigenous population can be incorporated into the diffusion modeling framework through local reproduction or repeated introductions, diffusion remains essentially a small-scale phenomenon. Spread models can be developed on the basis of population-based data (e.g., average rates or distances), but additional biological realism is introduced by the acquisition and incorporation of data that represent interindividual variability. Spread models reflecting the invading species’ potential for jump dispersal are required to incorpo-

897 rate the establishment of population centers at distances beyond the leading edge of the contiguous population. The ability of the new centers to behave as breeding colonies in their own right can significantly alter the spread process. Mechanistic spread models are required to account for directed movements (e.g., weather events or vehicular traffic). Various models currently used to estimate demand for transportation routes and forecast traffic flow and congestion patterns have potential applications to the spread of nonindigenous species assisted by travel and trade pathways.(26) One of the most widely used vehicular trip distribution models is the “gravity model,” which assumes that the number of trips between zones depends on the number produced at and attracted to each zone, and on the travel cost between zones. Realistically accounting for environmental heterogeneity and stochasticity entails additional modeling complexity and requires more detailed, geographically explicit data, as well as longitudinal data. Spread models that incorporate metapopulation dynamics require information regarding habitat patchiness and movement of individuals among patches. Spread models that incorporate demographic stochasticity may be needed to genuinely account for the low population density dynamics at the leading edge of the process and metapopulation processes. Despite its theoretical qualities, computer simulation of invasion scenarios through cellular automata(27) can also provide valuable insights into preventing or managing the spread of nonindigenous species, primarily when the environment can be considered to consist of discrete units (individual hosts or habitat patches) rather than as a spatial continuum. While increased realism and empirical detail in modeling the spread of invasive species are salutary scientific developments, the risk management question, the available data, and the magnitude of the decision should guide the choice of model in a risk assessment context. Under some circumstances, simple empirical models requiring minimal input data are adequate to inform decision making despite their lack of realism and richness of detail. In other cases, extensive ecosystemspecific studies may be warranted. Workshop participants judged that methods development is not a priority for better understanding the spread phase of biological invasions. Instead, a combination of individual- and population-level empirical studies should have precedence. Participants noted the distinction between the objectives of sanitary and phytosanitary inspection measures, which are primarily intended to ensure regulatory compliance, and those of monitoring and surveillance

898

Andersen et al.

activities, which seek to estimate targeted population parameters. Participants also emphasized the need to evaluate the efficacy of sanitary and phytosanitary measures designed to limit the spread of harmful nonindigenous species and stressed that efficient monitoring and surveillance efforts require closely linking measures to management objectives and careful attention to design issues. Workshop participants observed that Parker et al.(28) provide a useful framework for considering the ecological impacts of invaders. Conceptually, three factors determine the overall impact (I) of an invasive nonindigenous species: the total area occupied (R), abundance (A), and some measure of the impact per individual (E): I = R × A× E. While this simple equation provides a useful conceptual model, workshop participants underscored the importance of the time dimension, since the rates of establishment, proliferation, and spread may significantly impact the discounted present value of damages or the ability to adjust management practices to limit impacts. Furthermore, an empirical assessment of the extent and severity of impacts of an introduced pest would require, among other things, detailed information on the geographic distribution, timing, frequency, and level of introductions. Depending upon the management objectives and the nonindigenous species and recipient communities being considered, impacts may be assessed at one or more levels: individual, population, community, or ecosystem. However, there is an unavoidable tradeoff in that individual morbidity and mortality are generally the simplest, most precise, and least relevant ecological impact assessment endpoints. Analytical complexity, data demands, and scientific uncertainty increase with the level of ecological organizational at which impacts are assessed. Furthermore, predicting the economic and/or biological consequences of invasive species inevitably entails some degree of extrapolation beyond the available scientific evidence—at a minimum from impacts observed in regions where the pest is indigenous to predicting the consequences of its spread to areas where it presently does not occur. In order to be responsive to risk management questions and legitimate stakeholder concerns, a series of scientifically plausible assumptions may be required to link proxies that can be reliably measured or predicted to impact assessment endpoints that represent the relevant economic and environmental values whose protection is of ultimate interest.

Workshop participants judged the development of stressor-response relationships relating the level (i.e., abundance or prevalence) of specific nonindigenous species of concern to the resultant ecological effects as a high priority for invasive species impact assessment. Depending upon the formulation of the problem at hand, the ecological response resulting from exposure to the biological stressor (i.e., the invasive species) may be expressed in terms of the likelihood, magnitude, timing, and/or severity of impacts. Finally, because the consequences of biological invasion are highly dependent on the recipient communities, the participants underscored the importance of acquiring data on the identity, location, and volume of pathways by which potentially harmful nonindigenous species are most likely introduced.

6. DISCUSSION There were a number of common threads running through the various breakout session discussions. Workshop participants agreed on the need for extensive empirical research on specific areas in which good baseline data are lacking, including potential invasive species, native species and habitats that may be impacted by invasive species, important biological processes and phenomena, such as dispersal, and pathways of entry and spread for invasive species. In addition to the implications for the application of risk analysis in the implementation of national invasive species policy, the workshop’s findings have implications for national and academic research priorities. Just as threatened and endangered species became an important focus for conservation biology research because of concerns and national priorities arising from the implementation of the Endangered Species Act, invasive species should now become a major priority for research in response to national priorities identified in EO 13112. There are implications for other fields as well. Ecological economists should consider problems associated with the valuation of biodiversity and ecosystem services in light of potential invasive species problems; restoration ecologists should investigate ways to mitigate damage caused by invasives; integrated pest management researchers should continue their vigorous search for means to control invasive species; and risk analysts need to formulate a practical, scientific approach to inform invasive species risk management decisions that strikes a reasoned balance among competing objectives and societal values.

Risk Analysis for Invasive Species APPENDIX: Workshop Participants Alexei Sharov James Andreasen Brenda Rashleigh Brian Leung Bruce Hope Chris Dionigi Diana Kimberling Fermin George Chavez, Jr. Hallie Dozier Harald Scherm Heather Adams Jay Bancroft John M. Drake Lynn Maguire Mark Andersen Mark Hovorka Mark Powell Michael Neubert Orrin Myers Richard Fite Scott Ferson Steve Bartell Steve Yaninek T. Mal Wayne Landis Stacy Scott Kathryn Thomas Tsegegaye Habtemariam Berhanu Tameru Joel Brown Phillippe Rossignol Christopher Brand Ron Hiebert Shyam Nair Jon Lane Kent Prior Ann Nichols Carol Stepien Ron Sequeira Kimberly With Michelle Marvier David Chen Craig Chionio Susan Cohen Michael Firko Richard Orr David Oryang Robin Powell Michael Smith Robert Wiedenmann Robert Zajicek

Virginia Polytechnic Institute and State University United States Environmental Protection Agency United States Environmental Protection Agency Notre Dame University Oregon Department of Environmental Quality National Invasive Species Council (USA) United States Environmental Protection Agency Natural Resource Conservation Service (USA) Louisiana State University University of Georgia New Mexico State University University of Delaware Notre Dame University Duke University New Mexico State University Canadian Wildlife Service United States Department of Agriculture Woods Hole Oceanographic Institution University of New Mexico United States Department of Agriculture Applied Biomathematics, Inc. Cadmus Group, Inc. Purdue University Cleveland State University Western Washington University United States Department of Agriculture United States Geological Survey Tuskegee University Tuskegee University United States Department of Agriculture Oregon State University United States Geological Survey National Park Service (USA) Cadmus Group, Inc. United States Army Corps of Engineers Canadian Wildlife Service Cleveland State University Cleveland State University United States Department of Agriculture Kansas State University Santa Clara University United States Environmental Protection Agency United States Department of Agriculture United States Department of Agriculture United States Department of Agriculture United States Department of Agriculture United States Department of Agriculture Walker River Paiute Tribe United States Department of Agriculture Illinois Natural History Survey Florida Department of Agriculture and Consumer Services

899 REFERENCES 1. NISC. (2001). Management Plan: Meeting the Invasive Species Challenge. Washington, DC: National Invasive Species Council. 2. Lonsdale, W. M., Briese, D. T., & Cullen, J. M. (2001). In E. Wajnberg, J. K. Scott, & P. C. Quimby (Eds.), Evaluating Indirect Ecological Effects of Biological Control. New York: CABI Publishing. 3. Andersen, M. C., Adams, H., Hope, B., & Powell, M. (2004). Introduction: Risk analysis for invasive species. Risk Analysis, 24, 787–793. 4. Anderson, R. M. (Ed). (1982). Population Dynamics of Infectious Diseases: Theory and Applications. New York: Chapman and Hall. 5. Shang, X. S., Holt, J., & Colvin, J. (2000). A general model of plant-virus disease infection incorporating vector aggregation. Plant Pathology, 49(4), 435–444. 6. Porco, T. C. (1999). A mathematical model of the ecology of Lyme disease. IMA Journal of Mathematics Applied in Medicine and Biology, 16(3), 261–296. 7. Hethcote, H. W. (2000). The mathematics of infectious disease. SIAM Review, 47(4), 599–653. 8. Fuentes, M. A., & Kuperman, M. N. (1999). Cellular automata and epidemiological models with spatial dependence. Physica A, 267, 471–486. 9. Dobson, A., & Foufopoulos, J. (2001). Emerging infectious pathogens of wildlife. Philosophical Transactions of the Royal Society of London Series B—Biological Sciences, 356(1411), 1001–1012. 10. Suter, G. (1995). Adapting ecological risk assessment for ecosystem valuation. Ecological Economics, 14, 137– 141. 11. Kiessling, K. L., & Landberg, H. (Eds.). (1994). Population, Economic Development, and the Environment (p. 284). Oxford: Oxford University Press. 12. Constanza, R., Cumberland, H., Daly, H., Goodland, R., & Norgaard, R. B. (Eds.). (1997). An Introduction to Ecological Economics (p. 275). St. Lucie: CRC Press. 13. Landis, W. G., & Wiegers, J. A. (1997). Design considerations and a suggested approach for regional and comparative ecological risk assessment. Human and Ecological Risk Assessment, 3, 287–297. 14. Hilborn, R., & Mangel, M. (1997). The Ecological Detective: Confronting Models with Data. Princeton: Princeton University Press. 15. Efron, B., & Tibshirani, R. J. (1993). An Introduction to the Bootstrap. New York: Chapman and Hall. 16. Scott, J. M., Csuti, B., Estes, J. E., & Anderson, H. (1989). Status assessment of biodiversity protection. Conservation Biology, 3, 85–87. 17. Scott, J. M., Davis, F., Csuti, B., et al. (1993). Gap analysis: A geographical approach to protection of biological diversity. Wildlife Monographs 123. 18. Scott, J. M., Heglund, P. J., Morrison, M. L., et al. (Eds.). (2002). Predicting Species Occurrences: Issues of Accuracy and Scale (p. 868). Washington, DC: Island Press. 19. Bedford, T., & Cooke, R. (2001). Probabilistic Risk Analysis: Foundations and Methods (p. 393). Cambridge: Cambridge University Press. 20. Burgman, M. A., Ferson, S., & Akcakaya, H. R. (1993). Risk Assessment in Conservation Biology. London: Chapman and Hall. 21. Dennis, B. (1989). Allee effects: Population growth, critical density, and the chance of extinction. Natural Resource Modeling, 3, 481–538. 22. Caswell, H. (2000). Prospective and retrospective perturbation analyses: Their roles in conservation biology. Ecology, 81, 619–627.

900 23. Neubert, M. G., & Caswell, H. (2000). Demography and dispersal: Calculation and sensitivity analysis of invasion speed for structured populations. Ecology, 81, 1613–1628. 24. Neubert, M. G., Kot, M., & Lewis, M. A. (2000). Invasion speeds in fluctuating environments. Proceedings of the Royal Society of London B, 267, 1603–1610. 25. Skellam, J. G. (1951). Random dispersal in theoretical populations. Biometrika, 38, 196–218. 26. Bossenbroek, J. M., Kraft, C. E., & Nekola, J. C. (2001). Prediction of long-distance dispersal using gravity models: Zebra

Andersen et al. mussel invasion of inland lakes. Ecological Applications, 11, 1778–1788. 27. Keymer, J. E., Marquet, P. A., & Johnson, A. R. (1998). Pattern formation in a patch occupancy metapopulation model: A cellular automata approach. Journal of Theoretical Biology, 194, 79–90. 28. Parker, I. M., Simberloff, D., Lonsdale, W. M., et al. (1999). Impact: Toward a framework for understanding the ecological effects of invaders. Biological Invasions, 1, 3– 19.