COS 22-1 - Simulating the transition from even-aged longleaf plantations to old growth savannas using a coupled multiple model framework

Tuesday, August 9, 2016: 8:00 AM
Palm A, Ft Lauderdale Convention Center
Wendell P. Cropper Jr.1, Carlos A. Gonzalez-Benecke2, Daniel J. Leduc3, Davut Atar1, Salvador Gezan1, Timothy A. Martin1 and Lisa J. Samuelson4, (1)School of Forest Resources and Conservation, University of Florida, Gainesville, FL, (2)Department of Forest Engineering, Resources and Management, Oregon State University, Corvallis, OR, (3)USDA Forest Service, (4)School of Forestry and Wildlife Sciences, Auburn University, Auburn, AL

We developed an integrated longleaf pine (Pinus palustris) model framework consisting of an even-aged plantation model, an artificial neural network (ANN), and an individual tree savanna model. This model system was designed to simulate the transition from even-aged management to longleaf sand hill savanna  ecosystems. The plantation model (EAM) integrates a growth and yield model with longleaf specific allometric equations to simulate variable planting density, thinning, prescribed fire, and biomass dynamics. The savanna model (LLM) is an individual tree based model that simulates competition, growth, recruitment, and mortality for longleaf and hardwood trees. In order to provide for a transition between the even-aged stand model and the LLM, we used an ANN to generate a list of individual trees consistent with a 60 yr. old EAM stand, each tree assigned the age, a height, and a diameter. The LLM is spatially explicit, which required random placement of trees in this list into the simulation grid. We used multiple runs of the LLM model with the same initial tree list assigned to different locations to evaluate the sensitivity of the LLM to tree placement. Additional sensitivity analyses included mean fire return times ranging between 2 and 10 years.


The neural network model simulated tree size distributions with fit indices  for the thinned and unthinned  stand models of 0.84 and 0.93 respectively. Longleaf stands simulated over 250 yrs. had characteristics generally similar to published values for uneven-aged old growth longleaf pine savannas (e.g. mean DBH = 20.2 cm, max DBH = 44.1 cm; mean height = 14.2 m, max = 31.6 m; mean age = 65.6 yr., max = 310 yr.). The LLM showed only modest sensitivity to the initial placement of trees within the grid, coupled with stochastic event variability. After 250 yrs. of LLM simulation, the mean longleaf pine above-ground carbon stock was 17.1 ton/ha with a range of 14.8 to 19.4 over 15 runs (CV= 7.2%). Hardwood mean above-ground carbon was only 0.65 ton/ha (CV= 42.4%). Burning return times between 2 and 4 years showed modest sensitivity, but increasing the mean return time to ten years resulted in increased hardwood dominance and decreased longleaf pine regeneration. Between the 4 yr. burn interval and the 10 yr. interval  mean longleaf pine above-ground carbon decreased from 16.2 to 6.8 tons/ha whereas the mean hardwood above-ground carbon increased from 0.7 to 5.4 tons/ha.