2018 ESA Annual Meeting (August 5 -- 10)

SYMP 7-4 - Dynamic indicators of ecosystem resilience

Tuesday, August 7, 2018: 3:10 PM
River Bend 1, New Orleans Downtown Marriott at the Convention Center
Hao Ye1, Erica Christensen1, S.K. Morgan Ernest2, Juniper L. Simonis1,3 and Ethan P. White1, (1)Wildlife Ecology and Conservation, University of Florida, (2)Department of Wildlife Ecology and Conservation, University of Florida, Gainesville, FL, (3)DAPPER Stats, Portland, OR
Background/Question/Methods

How can we quantify resilience? Because ecosystems contain complex, nonlinear, and even chaotic dynamics, with interactions that occur within and across scales, conventional process models struggle with both structural uncertainty and the practical difficulties of parameterizing complex models from observations. An alternative approach is to allow the data to inform our understanding of ecological dynamics. Recently, Ushio et al. presented a method for computing a measure of dynamic stability, based on the abundances of a fish community. Here, we explore an application of that method to a desert rodent community, to test whether dynamic stability can provide advance indication of impending large-scale shifts in community structure.

Results/Conclusions

Our initial results are mixed. Compared to the analysis conducted by Christensen et al. (where the Latent Dirichlet Allocation inferred community groupings, to which a change-point model was applied), the maximal eigenvalue identified somewhat different times at which the ecosystem was unstable. The clear large excursions in dynamic stability occurred in 5 time periods: December 1989-March 1990, December 2002-April 2003, June 2005-August 2007, July 2011-October-2012, and March 2014-April 2015. Although there is some agreement with the transition periods identified in Christensen et al., there are large changes in dynamic stability (e.g. around 2003) that do not appear as community shifts, and vice-versa. Indeed, while dynamic stability is a measure of sensitivity to potential change, it does not necessarily indicate that a change will occur. For example, the realized changes in community structure could be precipitated by environmental events that would not necessarily appear in measures based on biological dynamics.

These results suggest that care needs to be applied to the general usage of such data-driven indicators. In some cases, restricting the empirical measures to focal species can be more informative about management targets for policy interventions, with the caveats of being relatively insensitive to changes in other species' abundances or the potential impacts of stochastic external events, such as weather, or invasive species.