Cooperative Ecosystem Studies Unit, Rocky Mountain CESU

archived

The purpose of the Northern Rocky Mountain Science Center (NOROCK) funding opportunity is to explore two methods of modeling complex multivariate wetland vegetation system: Bayesian Belief Networks (BBN) and Bayesian Graphical Models (BGM). Both approaches represent the complex multivariate system by a graph composed of edges and nodes. Variables are represented by nodes and edges can be directed representing a “causal” relationship or undirected representing a correlation (spatial or temporal). The research objectives are as follows: 1) Build a predictive model of wetland vegetation species abundance as a function of abiotic factors within Red Rock Lakes NWR using BBN and/or BGM that incorporates time lags at different temporal scales. 2) Investigate tools for BBN and BGM model selection to establish if the typical information theoretic criteria are appropriate to use to select among different hypothetical causal structures and if the criteria are appropriate in a system with spatial and/or temporal correlation. And 3) Determine if a BBN or a BGM could be a useful tool for practitioners to address the effects of climate change on natural systems.

High