PARENT SESSION
Oral Session #34: Forests: Models of function and structure. Presiding: M. Lechowicz.
Tuesday, August 7, 2001. 1:00 PM to 3:45 PM. Hall of Ideas F.
Modeling and mapping the distribution of ground flora in the Missouri Ozarks.
Hooten, Mevin1, Larsen, David 1, Wikle, Christopher2, 1 2
ABSTRACT- Forested systems of Southeast Missouri are home to some 500+ species of plants. Research has proven that the occurrence of certain plants is correlated with several site-defining variables. The Missouri Ozark Forest Ecosystem Project, designed to study long-term management effects on forests, collects landscape level floristic data at many spatial scales. The purpose of our research is to create a robust methodology for modeling organisms on a landscape using MOFEP ground flora data. Analysis of the spatial structure for several understory plants has shown that in addition to environmental effects, the distribution of species is influenced by uncharacterized spatial random effects. Using a hierarchical Bayesian framework we have accounted for these effects as well as uncertainty related to the change of spatial support between observation and prediction scales. This methodology has successfully allowed us to map the patterns of several species in the genus Desmodium by utilizing posterior distribution information gained from the model. Similar but separate data collected in the MOFEP area has been useful in performing preliminary evaluation of model accuracy and extent of application. Potential applications of this type of model include but are not limited to: Spatio-temporal mapping of wildlife forage availability, analysis of interspecies spatial interaction, landscape level identification of areas susceptible to exotic invasion, analysis of spatial and temporal patterns of biodiversity.
KEY WORDS: Bayesian modeling, vegetation prediction, spatial statistics