SYMP 8-1
Overview: Pretty darn good control of ecological systems

Tuesday, August 12, 2014: 1:30 PM
Gardenia, Sheraton Hotel
Paul R. Armsworth, Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN
Michael Bode, ARC Centre of Excellence for Environmental Decisions, University of Melbourne, St. Lucia, Australia
Carl Boettiger, Environmental Science, Policy, & Management, UC Berkeley, Berkeley, CA
Iadine Chades, EcoSciences Precinct - Dutton Park, CSIRO, Dutton Park, Australia
Megan J. Donahue, Hawaii Institute of Marine Biology, Kaneohe, HI
Alan Hastings, Department of Environmental Science and Policy, University of California, Davis, Davis, CA
Mandy A. Karnauskas, Southeast Fisheries Science Center, Miami, FL
Jacob LaRiviere, Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN
Claire Paris, The Rosenthiel School of Marine and Atmospheric Science, University of Miami, Miami, FL
Daniel Ryan, National Institute for Mathematical and Biological Synthesis, Knoxville, TN
James N. Sanchirico, Dept. Environmental Science and Policy, University of California-Davis, Davis, CA
Carl Toews, Mathematics, University of Puget Sound, Tacoma, WA
Background/Question/Methods

Much ecological theory underpinning studies in ecosystem management and life history evolution draws on a body of mathematics known as optimal control. In one set of applications, ecosystems managers are assumed to be seeking ‘optimal’ decisions over how to allocate and manage natural resources; in the other, an ‘optimal’ behavior or phenotype is predicted given the particular ecological conditions that an organism will face. In both situations, the underlying ecological systems involved can be noisy, high dimensional and nonlinear; they also vary over multiple space and time scales and are only ever observed imperfectly. However, much of this richness is assumed away in theoretical ecology in a quest to identify optimal solutions. Instead, what is needed is a theoretical approach that embraces the messy complexity of ecological questions without giving up on the optimizers’ drive to find effective management or life history strategies.

Results/Conclusions

We will present a series of such approaches that together offer an alternative to traditional optimal control. When taken together, we refer to these alternatives as ‘pretty darn good’ control. We will illustrate pretty darn good control through a series of applications in marine ecology and marine ecosystem management.

Some recurring lessons from our efforts to apply pretty darn good control to ecological questions include that

i) behaviors that can seem suboptimal when compared to simplified problem formulations might instead be rational and effective once a fuller representation of the relevant optimization problem is considered.

ii) the detailing of how you represent constraints and limitations on the choice sets that individuals face is very important for obtaining sensible predictions from your model.

iii) only some sources of ecological variation need to be reflected in a pretty darn good life history or management strategy and only some entry-points of uncertainty need to be resolved to deliver effective performance.

iv) and finally, when developing a strategy that can respond to ecological variation, a small amount of responsiveness earns big pay-offs, but finessing further offers comparatively little gain.