Fire and forest landscapes in the Georgia Piedmont: an assessment of spatial modeling assumptions

Abstract

Landscape simulation models are widely used to study the behavior of ecological systems. As computing power has increased, these models have become more complex and incorporated more realistic spatial representations of landscape patterns and ecological processes. The goal of this research was to examine the sensitivity of simulated landscape patterns to fundamental spatial modeling assumptions. The LANDIS simulator was parameterized for forests of the Georgia Piedmont and used to model landscape-scale community dynamics at fire return intervals from 20 to 100 years. A base scenario incorporating localized seed dispersal along with landform-related variation in species establishment rates and disturbance regimes was contrasted with three alternative scenarios. The uniform habitat scenario applied the same set of species establishment coefficients across all landforms. The uniform dispersal scenario removed the effects of seed source abundance and pattern on species establishment. The uniform disturbance scenario assumed identical disturbance regimes on all landforms. At the shortest fire return intervals, fire severities were low and the stand age distribution was dominated by older forests. At longer fire return intervals, fire severities were high and the stand age distribution was skewed toward younger forests. Species composition generally followed a gradient from fire-resistant species at short fire return intervals to fire-sensitive species at longer fire return intervals. However, some species exhibited bimodal distributions with high abundances at both short and long fire return intervals. Landscape responses to fire were similar in the uniform habitat scenario and the base scenario. Communities were less sensitive to fire return interval and had more fire-sensitive species in the uniform dispersal scenario than in the base scenario. Species composition in the uniform disturbance scenario was similar to the base scenario for the longest fire-intervals, but was more sensitive to changes in the fire regime at shorter fire return intervals. In models of Piedmont forest landscapes, accurate spatial representations of dispersal and fire regime heterogeneity are essential for predicting landscape-scale species composition under changing fire regimes. In contrast, the precise spatial representation of species–habitat relationships may be considerably less important.