|
|
Type of Document Dissertation Author Sable, Shaye Ellen Author's Email Address ssable1@lsu.edu URN etd-07062007-155333 Title A Comparison of Individual-Based and Matrix Projection Models Applied to Fish Population and Community Dynamics Degree Doctor of Philosophy (Ph.D.) Department Oceanography & Coastal Sciences Advisory Committee
Advisor Name Title Kenneth A. Rose Committee Chair Gregory Stone Committee Member James H. Cowan, Jr. Committee Member James P. Geaghan Committee Member Megan K. Lapeyre Dean's Representative Keywords
- community
- simulation
- fish population
- individual-based model
- matrix projection model
Date of Defense 2007-06-12 Availability unrestricted Abstract Matrix projection models and individual-based models (IBM) are commonly used for theanalysis and management of fish populations. Matrix models break down the population into
age or stage classes, while IBMs track individuals. I perform a series of quantitative
comparisons between the predictions of the two modeling approaches using the IBM as the
standard of comparison to demonstrate when individual variation, species interactions, and
spatial heterogeneity adversely affect matrix model performance. I first evaluate the matrix
approach for predicting yellow perch population responses when perch are involved in size-specific
predator-prey interactions with walleye. I created density-dependent and stochastic age-structured and stage-within-age matrix models from an Oneida Lake walleye-yellow perch IBM,
and then changed perch survival rates within the matrix models and IBM and compared their
predicted responses. The matrix models simulated yellow perch responses reasonably well when
density-dependent YOY survival was correctly defined. At least 20 years of data (IBM output)
were needed to correctly estimate the density-dependent relationships in the matrix models.
Second, I developed a 2-species matrix model by linking the elements between perch and
walleye matrix models. The 2-species model simulated yellow perch prey responses reasonably
well, but was unable to correctly predict walleye predator responses. Third, I developed a new
IBM that simulated a 6-species tidal marsh community on a fine-scale spatial grid of habitat
cells. The IBM was used to scale individual-level effects of lowered dissolved oxygen and
habitat degradation to population-level responses, and used to estimate relatively simple stage-based matrix models for grass shrimp and gulf killifish populations. Equilibrium analysis of the simple matrix models was insufficient for predicting population responses. This study showed
that stochastic, density-dependent matrix projection models were able to mimic density-
dependent survival processes and species interactions relatively well, while equilibrium analysis of simple matrix models was inadequate. The matrix approach consistently had trouble
estimating density-dependent and inter-specific growth relationships that were important for
accurate model predictions. I recommend the use of IBMs and relatively complicated matrix
models (stage-within-age, stochastic, density-dependent, multispecies) for simulation of fish
population and community dynamics.
Files
Filename Size Approximate Download Time (Hours:Minutes:Seconds)
28.8 Modem 56K Modem ISDN (64 Kb) ISDN (128 Kb) Higher-speed Access Dissertation_ss07final.pdf 4.75 Mb 00:21:58 00:11:18 00:09:53 00:04:56 00:00:25