COS 121-6
Improving regional abundance dynamics and range-shift forecasts by coupling individual-based dispersal models with matrix-based population projections

Thursday, August 14, 2014: 3:20 PM
314, Sacramento Convention Center
Kevin T. Shoemaker, Ecology & Evolution, Stony Brook University, Stony Brook, NY
Damien A. Fordham, University of Adelaide, Adelaide, Australia
Nathan H. Schumaker, US EPA, Corvallis, OR
H. Resit Ak├žakaya, Ecology & Evolution, Stony Brook University, Stony Brook, NY
Background/Question/Methods

Simulation models for projecting regional abundance and range dynamics are indispensible tools for assessing the biodiversity consequences of global change. Increasingly, many of the mechanisms and interactions that drive species distributions are modeled explicitly, yielding key insights for ecology and conservation in a changing world. However, simple distance-based dispersal kernels continue to form the basis for modeling regional connectivity despite the clear importance of connectivity as a determinant of range dynamics and (via connectivity enhancement) as a proposed solution for biodiversity conservation under climate change. Previous efforts to forecast regional population and range dynamics typically have not considered the interactions between the behavior of dispersing individuals and the structural elements and landscape processes that affect dispersal. Here we link an individual-based dispersal model (IBM) to a regional population model to test the implications of this omission. We apply this novel approach to study local extinctions, regional abundance dynamics and range shifts for a well-studied Australian freshwater turtle whose persistence is threatened by two synergistic drivers of global change: predation by invasive species and overexploitation.

Results/Conclusions

We show that projections of local extinctions and range dynamics in this study system change substantially when the processes driving connectivity are modelled explicitly. In our study system, the coupled IBM/metapopulation model yielded higher estimates of local extinction risk, and lower estimated rates of range contraction, relative to forecasts using standard methods. We also show that the new approach can correct known biases that can arise from distance-based dispersal models, such as excessive movement of individuals from very large populations into adjacent small populations. We conclude that a coupled IBM/metapopulation modeling approach can minimize an important source of bias in predictions of shifts in species distributions and abundances under global change, especially for organisms whose dispersal behaviours are strongly affected by landscape structural elements and processes.