UW Aquatic & Fishery Sciences Quantitative Seminar

Jim Cronan

School of Environmental and Forest Sciences

Managing disturbance in the longleaf pine ecosystem: effects of managed fire regime characteristics on fire hazard and community ecology at multiple spatial scales

Abstract

When asked to name areas of the US prone to wildfires, most people would probably not say Florida. After all, the bulk of the fires that make the news every summer occur west of the Mississippi River, in places with steep terrain, hot, dry summers, and parched vegetation. Qualities that are pretty much the antithesis of Florida. While it is true that Florida does not have many headline grabbing wildfires, in many respects nowhere else in the country are ecosystems quite so primed to burn. After a brief explanation of why this is so, this seminar will focus on the quantitative methods I have used in my dissertation research to address questions surrounding fire management in the uniquely flammable ecosystems of the southeastern US. My dissertation has two parts. The first was a field study designed to test how prescribed fires conducted in the winter versus the summer can influence re-vegetation and potential fire behavior in the longleaf pine forests of the southeastern US. The second is a simulation exercise using a type of computer program called a landscape-fire succession model to predict how forest structure and fire hazard change on an air force base in northwestern Florida after decades of different fire regimes. While both look at the relationships between fuels and fire behavior, the data and methods are quite different and reflect a gradual shift from relatively simple statistical tests involving few explanatory variables and small datasets to more complex tests with multiple explanatory variables and large datasets. The fire seasonality study represents the more traditional approach. Winter or summer burns were assigned equally to 20 sites in two regions in longleaf pine forests in northwestern Florida. The difference in fuel categories from pre-fire conditions was measured for two consecutive years after the treatment burns. Repeated measures ANOVA was conducted to test for differences in fuels and modeled fire behavior. The landscape-fire-succession model study is ongoing and represents an increasingly common approach that uses computer models that act on spatially explicit datasets to forecast changes in the distribution of some variable of interest. Existing datasets were used to build a 30-meter resolution raster map of fuel types across that air force base that was then ground-truthed with two independent datasets. A landscape-fire-succession model was developed to simulate disturbance regimes and vegetation response at annual time steps. Ensemble modeling will be conducted on the fuel type maps for fifty-year simulations under four prescribed fire regimes. Output maps will be examined to test how changes in prescribed burning could affect forest structure and fire hazard over multi-decadal management periods.