Numeerisen analyysin ja laskennallisen tieteen seminaari

7.3.2005 klo 14.15 U322

Otso Ovaskainen, Helsingin Yliopisto, Metapopulation Research Group

Estimating habitat-specific movement parameters from mark-recapture data

When a biologist is interesting in measuring the movements of individuals (such as butterflies), she usually conducts a mark-recapture study. A fundamental problem in the analysis of spatial mark-recapture data is that the observed movement pattern depends not only on the properties of the species but also on the structure of the landscape, and on the design of the study. This can be overcome by parameterizing a movement model that accounts for the structure of the landscape. Assuming that the organism performs a (correlated) random walk, a plausible model is given by diffusion in a domain that may consist of several habitat types. The caveat is that the problem becomes computationally demanding, as the likelihood of the data needs to be constructed from a time-dependent solution to the diffusion equation. I illustrate how the estimation can be done using finite-element methods in combination with adaptive Bayesian modeling.