1. How well do environment- based models predict species abundances at a coarse scale? Volker Bahn and Brian McGill McGill University CSEE, Toronto, May, 2007 www.volkerbahn.com

2. Distribution Map Rose-breasted Grosbeak http://www.natureserve.org/

3. Distribution Map Rose-breasted Grosbeak http://www.mbr-pwrc.usgs.gov/bbs/bbs.html

4. Species Distributions Central to ecology –Krebs, C. J. 1972. Ecology: The experimental analysis of distribution and abundance. –Andrewartha, H. G., and L. C. Birch. 1984. The ecological web: More on the distribution and abundance of animals. Conservation of species

5. How does distribution modelling work? –Occurrence or abundance data at some locations –Record environmental conditions –Build statistical model relating sample data to environmental predictors –Predict occurrence for non-surveyed areas Distribution Modelling

8. Research Questions How well does niche-based distribution modelling work? How can one assess the predictive ability of distribution models? Which influence does the evaluation scheme have on the assessment of the models?

9. Methods Breeding Bird Survey 1996-2000 1293 locations 79 - 190 species Environmental data Regression trees/ Random forests

10. Contagion

11. Results DependentIndependentR 2 * Bird abundanceEnvironment0.32 Bird abundanceContagion0.43 Sim. RangesEnvironment0.24 *Averaged over 190 species Bahn, V and McGill, B.J. (2007) Can niche-based distribution models outperform spatial interpolation? Global Ecology and Biogeography: online early.

12. Resubstitution – no split

13. Random split

14. Strips

15. Halves

17. Conclusion When training and test data are interspersed, interpolation does the job just as well as niche-based models Niche-based models predict poorly into new areas Evaluations are dependent on information content and testing scheme

18. Outlook If environmental conditions are not a good predictor then what are we missing? We don’t get the right information from remotely sensed data Processes are not stationary Spatial processes: dispersal and population dynamics

19. Acknowledgements Thousands of volunteers, CWS & USGS for BBS data Grad students, friends and collaborators in the lab and beyond Family Funding from NSERC

20. Discussion?

21. Species Peak at Optimum? Typically not Mueller-Dombois, D. & Ellenberg, H. (1974) Aims and methods of vegetation ecology. Wiley, New York. Rehfeldt, G.E., Ying, C.C., Spittlehouse, D.L. & Hamilton, D.A., Jr. (1999) Genetic responses to climate in Pinus contorta: Niche breadth, climate change, and reforestation. Ecological Monographs, 69(3), 375-407.

22. Species Peak at Optimum? Wang et al. 2006. Use of response functions in selecting lodgepole pine populations for future climates. J Global Change Biology 12(12):2404-2416 Frazier, M., R.B. Huey, and D. Berrigan. 2005. Thermodynamics constrains the evolution of insect population growth rates: "warmer is better." American Naturalist 168:512-520.

23. Mueller-Dombois and Ellenberg (1974)

25. Dispersal Bahn, V., W.B. Krohn, and R.J. O'Connor. Under review. Dispersal leads to autocorrelation in animal distributions: a simulation model. Submitted to Journal of Applied Ecology.

26. Before/after Dispersal

28. Conditional Autoregressive Y = Xβ + ρC(Y – Xβ) + ε

29. Temp max ≥ 28.6 Yearly var precip ≥ 0.2 Seasonal var precip ≥ 0.3 Precip ≥ 66.1 Temp min ≥ 3.2 0.7 n = 115 1.2 n = 145 1.0 n = 66 1.7 n = 273 1.6 n = 24 2.8 n = 54