Spatial aggregation effects on the simulation of landscape pattern and ecological processes in southern California plant communities

Abstract

Accurate representation of the processes and components of natural systems is necessary for reliable ecological models, yet data generalization is often needed to reduce unneeded detail and to increase model efficiency. A spatially explicit, raster-based simulation model of disturbance and succession (LANDIS) was used to examine the effects of spatial aggregation on modeled pattern (species composition) and process (fire disturbance). At systematically increased levels of data aggregation, the model was tested on two landscapes, one based on species patterns that were initially random and one based on more realistic distributions, over 500-year time periods for a southern California (Mediterranean-climate) landscape. In both landscapes, spatial aggregation resulted in less frequent, more unpredictable, yet higher-severity fires, and plant species cover became more variable over time in response to infrequent, high-severity fire. The systematic effects of aggregation on pattern, process, and species response suggest that modelers can detect ranges of resolutions for which parameters hold, helping to identify appropriate levels of spatial generalization for their research.