Modeling fire-induced tree mortality in longleaf pine using spatial data

Graduation Date

2014

Document Type

Thesis

Program

Other

Program

Thesis (M.S.)--Humboldt State University, Environmental Systems: Mathematical Modeling, 2014

Committee Chair Name

Christopher Dugaw

Committee Chair Affiliation

HSU Faculty or Staff

Keywords

Forestry, Humboldt State University -- Theses -- Mathematical Modeling, Math, Modeling, Spatial, Longleaf, Statistics

Abstract

Restoration of fire-excluded pine ecosystems is a major conservation and management goal. Land managers have used prescribed fires to reduce accumulated forest floor fuels and restore pinelands. While these fires have been successful at reducing fuels, tree mortality in these restoration fires can be as high as 75 to 95 percent. In this study, I model post-fire mortality of longleaf pine (in two stands in the Mountain Longleaf National Wildlife Refuge in northeastern Alabama) using logistic models composed of only predictors that can be measured pre-fire. These methods are extended by the addition of spatial competition functions since such measures are drawn from available data for managers, and have been shown to have effects on growth and mortality. The best models, selected by AIC, predict survival as well or better than earlier models but have high false mortality predictions. The spatial dynamics of the stand are also studied pre-fire and post-fire to determine the effect of prescribed fire and inter-tree competition. Post-fire, spatial distribution of the trees did not significantly change. Earlier studies detected differences in the spatial dynamics between different classes of longleaf pine. These dynamics were confirmed in some results and in others questioned. Small trees (DBH 30 cm) clustered away from large trees (DBH 30 cm). Large trees were more dispersed than a random process (Poisson) would dictate. Juveniles (DBH 5 cm) were shown to cluster away from large trees, but aggregate around other small trees.

https://scholarworks.calstate.edu/concern/theses/47429c72c

Share

 
COinS