neric Nonindigenous Aquatic Organisms Risk Analysis Review Process recently up- dated in the United States (28), are based on expert opinion or qualitative assessments, and not on rigorously quantitative statistics. In contrast, the approach illustrated here is quantitative, repeatable, and transparent— characteristics a recent National Research Council report urges should apply to the next generation of risk assessments (11). Although the results presented here are specific to the Great Lakes, this approach to constructing predictive models could be ap- plied to a diversity of plant and animal taxa inhabiting a variety of terrestrial and aquatic ecosystems. As highlighted by the U.S. Na- tional Management Plan on invasive species (5), the urgent need to focus attention on prevention requires the development of spe- cies risk-assessment protocols. As alien spe- cies move along the invasion sequence (from transport to introduction, establishment, spread, and impact), management options be- come more limited. Even in the rare cases in which the knowledge and technology exist to control an established species, such efforts are expensive and must be practiced in per- petuity (29). For example, the United States and Canada jointly spend about U.S. $15 million annually to control sea lamprey (Petromyzon marinus) in the Great Lakes (30); these costs have been incurred since 1956 and will continue as long as sea lamprey control remains a management goal. Quanti- tative risk assessments that identify the alien species most likely to establish, spread quick- ly, and become a nuisance could be the foun- dation for efforts to prevent future expensive and environmentally damaging invasions. References and Notes 1. H. A. Mooney, R. J. Hobbs, Invasive Species in a Changing World (Island Press, Washington, DC, 2000). 2. F. J. Rahel, Science 288, 854 (2000). 3. O. E. Sala et al., Science 287, 1770 (2000). 4. D. Pimentel, L. Lach, R. Zuniga, D. Morrison, Bio- science 50, 53 (1999). 5. National Invasive Species Council, National Manage- ment Plan: Meeting the Invasive Species Challenge (National Invasive Species Council, Washington, DC, 2001). 6. D. M. Lodge, Trends Ecol. Evol. 8, 133 (1993). 7. R. H. Groves, F. D. Panetta, J. G. Virtue, Weed Risk Assessment (CSIRO Publishing, Collingwood, Victoria, Australia, 2001). 8. Committee on the Scientic Basis for Predicting the Invasive Potential of Nonindigenous Plants and Plant Pests in the United States, Predicting Invasions of Nonindigenous Plants and Plant Pests (National Acad- emy Press, Washington, DC, 2002). 9. M. Enserink, Science 285, 1834 (1999). 10. C. S. Kolar, D. M. Lodge, Trends Ecol. Evol. 16, 199 (2001). 11. Committee on Environment and Natural Resources of the National Science and Technology Council, Ecological Risk Assessment in the Federal Govern- ment, Report CENR/5-99/001 (May 1999). 12. S. H. Reichard, C. W. Hamilton, Conserv. Biol. 11, 193 (1997). 13. M. Rejmanek, D. M. Richardson, Ecology 77, 1655 (1996). 14. Detailed Materials and Methods are available as sup- porting material on Science Online. 15. Method suggested by A. Grafen, Philos. Trans. R. Soc. London Ser. B 326, 119 (1989). Classification of P. B. Moyle, C. C. Cech Jr., Introduction to Ichthyology (Prentice Hall, Upper Saddle River, NJ, ed. 3, 1996). 16. Function coefficients for the establishment DA (signs indicate association with establishment): constant = 9.8025, closer to mature length by 2 years = 0.0704, wide temperature range = 0.1430, wide salinity tol- erance = 1.5844, history of invasiveness = 1.4602. 17. The reliability of our models (and all other models) is a function of the accuracy of the prediction and the frequency with which the event occurs at all (i.e., the “base rate,” or proportion of introduced species that establish). Low base rate probability inflates the number of “false-positives” identified by the screen- ing tool (species predicted to become established, spread quickly, or be perceived as a nuisance) (31). See online material for a discussion of the influence of base rate on our results. 18. CART was performed with CART software (Salford Systems, San Diego, CA). 19. Discriminant function coefficients for the spread DA (signs indicate association with spreading quickly): constant = 24.68635, survive higher temperatures = –1.52543, wide temperature range = 0.95508, closer to mature length by 2 years = –0.0000076. 20. Discriminant function coefficients for the impact DA (signs indicate association with being a nuisance): constant = –1.29465, egg diameter = –2.41764, minimum temperature threshold = –0.17902, salin- ity range = 0.73598. 21. A. Ricciardi, H. J. MacIsaac, Trends Ecol. Evol. 15, 62 (2000). 22. J. Illies, Limnofauna Europea (Fischer Verlag, Stutt- gart, Germany, 1978). 23. C. J. Krebs, Ecological Methodology (Harper & Row, New York, 1989). 24. A. Ricciardi, J. B. Rasmussen, Can. J. Fish. Aquat. Sci. 55, 1759 (1998). 25. B. Cudmore-Vokey, E. J. Crossman, “Checklists of the fish fauna of the Laurentian Great Lakes and their connecting channels,” Canadian Manuscript Report of Fisheries and Aquatic Sciences 2550 (Canadian De- partment of Fisheries and Oceans, Ottawa, Canada, 2000). 26. Fed. Regist. 67, 49280 (2002). 27. E. Sharp, “Beware big fish: Asian carp are threat to lakes,” Detroit Free Press, 11 April 2002; available at www.freep.com/sports/outdoors/outcol11_200204411. htm. 28. Risk Assessment and Management Committee, Ge- neric Nonindigenous Aquatic Organisms Risk Analysis Review Process (Report to the Aquatic Nuisance Spe- cies Task Force, 21 October 1996); available at www. anstaskforce.gov/gennasrev.htm. 29. U.S. Congress, Harmful Nonindigenous Species in the United States (Office of Technology Assessment, Washington, DC, 1993). 30. C. I. Goddard, Great Lakes Fishery Commission, cir- cular letter (8 July 1997). 31. C. S. Smith, W. M. Lonsdale, J. Fortune, Biol. Invas. 1, 89 (1999). 32. We thank R. Sparks, E. Marsden, and D. Schneider for thoughtful discussion. We also thank J. Drake, J. Feder, K. Filchak, T. Kreps, B. Leung, and anonymous reviewers for comments on earlier versions of the manuscript. Thanks are also due to the fish managers and exotic species specialists who responded to our nuisance survey. This research is part of a larger project (National Sea Grant 643-1532-04). Addition- al support was provided by the Great Lakes Fishery Commission, Environmental Protection Agency STAR fellowship (R-82889901-0), the NSF Biocomplexity Initiative, a Clare Boothe Luce Presidental Fellowship (C.S.K.), and the Equal Opportunities Section of the American Fisheries Society (C.S.K.). Supporting Online Material www.sciencemag.org/cgi/content/full/298/5596/1233/ DC1 Materials and Methods SOM Text Tables S1 to S9 References and Notes 4 July 2002; accepted 19 September 2002 Avian Persistence in Fragmented Rainforest Luc Lens, 1,2 * Stefan Van Dongen, 1 Ken Norris, 3 Mwangi Githiru, 2,4 Erik Matthysen 1 What factors determine the persistence of species in fragmented habitats? To address this question, we studied the relative impacts of forest deterioration and fragmentation on bird species in 12 rainforest fragments in Kenya, combining 6 years of individual capture-recapture data with measurements of current captures and museum specimens. Species mobility, as estimated from species-specific dis- persal rates, and tolerance to habitat deterioration, as estimated from change in fluctuating asymmetry with increasing habitat disturbance, explained 88% of the variation in patch occupancy among eight forest bird species. Occupancy increased with mobility and with tolerance to deterioration, where both variables contributed equally to this relationship. We conclude that individual-level study, such as of dispersal behavior and phenotypic development, can predict patterns of persistence at the species level. More generally, for conservation tactics to stand a high chance of success, they should include action both within sites, to minimize habitat deterioration, and across landscapes, to maximize dispersal. Anthropogenic habitat deterioration is impos- ing new selection pressures on organisms, increasing local extinction rates (1). Simulta- neously, reduced movement among remnant patches lowers colonization rates, which fur- ther negatively affects demographic and ge- netic population parameters (2). From a con- servation perspective, the impacts of habitat deterioration and the impacts of habitat frag- mentation might demand different strategies. Whereas the former often requires manage- ment of populations within local ( protected) R EPORTS 8 NOVEMBER 2002 VOL 298 SCIENCE www.sciencemag.org 1236