SYMP 24-3 - Translating conceptual to parameterized models for multiple taxonomic groups

Friday, August 12, 2011: 8:30 AM
Ballroom G, Austin Convention Center
Len Thomas1, James S. Clark2 and John Harwood1, (1)University of St Andrews, (2)Duke University, Durham, NC
Background/Question/Methods

For a given scenario, the PCAD framework developed by the National Research Council Committee on Potential Effects of Ambient Noise in the Ocean on Marine Mammals can be translated into a set of nested, stochastic spatio-temporal models, linking levels of disturbance to changes in individual animal behavior, then to consequent changes in individual survival and reproductive success, and finally to population-level changes in vital rates and hence population growth or decline.  Diverse data are often available to help parameterize each level of the hierarchy, although for some important model parameters there is little more than expert opinion.  One potential approach to estimate model parameters is to construct a detailed, mechanistic, hierarchical statistical model and attempt to use all available data to simultaneously fit all model parameters: an “integrated” modeling approach.  However, because different levels operate at quite different temporal and spatial scales, this is currently infeasible.  Further, there is little lost in treating each level separately (or a few together, if convenient), and using the output of one as input to the next. 

Results/Conclusions

The approach we have taken is to construct mechanistic models of individual animal behavior, and fit these using likelihood-based or Bayesian methods.  Outputs of these models, such as the estimated distribution of calf survival as a function of different levels of disturbance, are then used as inputs for simulation-based studies of the effects of changes in vital rates on population outcomes, such as growth rate.

Copyright © . All rights reserved.
Banner photo by Flickr user greg westfall.