Longleaf pine ( Pinus palustris) and hardwood dynamics in a fire-maintained ecosystem: A simulation approach

Abstract

Longleaf pine ( Pinus palustris) savannas of the southeastern U.S. represent an archetype of a fire dependent ecosystem. They are known to have very short fire return intervals (∼1–3 years) that perpetuate understory plant diversity (up to 50 species m −2), support pine recruitment, and suppress fire sensitive hardwoods. Understanding the relationships that regulate longleaf and southern hardwoods is especially critical. With decreased fire frequency, insufficient intensity, or lack of underground competition, a woody mid-story rapidly develops, dominated by fire sensitive trees and shrubs that in-turn suppress more fire dependent species (including pine seedlings). This may occur in forest gaps, where pine-needle abundance is diminished, reducing fire spread potential. The interactions between longleaf pine, hardwoods, forest fuels, and fire frequency are complex and difficult to understand spatially. The objective of this study was to develop a spatially explicit longleaf pine–hardwood stochastic simulation model (LLM), incorporating tree demography, plant competition, and fuel and fire characteristics. Data from two longleaf pine study sites were used to develop and evaluate the model with the goal to incorporate simple site-specific calibration parameters for model versatility. Specific model components included pine seed masting, hardwood clonal sprouting, response to fire (re-sprouting, mortality), and tree density driven competition effects. LLM spatial outputs were consistent with observed forest gap dynamics associated with pine seedling establishment and hardwood encroachment. Changes in fire frequency (i.e., fire probability = 0.35–0.05) illustrated a shift in community structure from longleaf pine dominated to a hardwood dominated community. This approach to assessing model response may be useful in characterizing longleaf ecosystem resilience, especially at intermediate fire frequencies (e.g., 0.15) where the community may be sensitive to small changes in the fire regime. Height distributions and population densities were similar to in situ findings (field and LIDAR data) for both study sites. Height distributions output by the LLM illustrated fluctuations in population structure. The LLM was especially useful in determining knowledge gaps associated with fuel and fire heterogeneity, plant–plant interactions, population structure and its temporal fluctuations, and hardwood demography. This is the first known modeling work to simulate interactions between longleaf pine and hardwoods and provides a foundation for further studies on fire and forest management, especially in relation to ecological forestry practices, restoration, and site-specific applications.