Question: Underlying ecological processes have often been inferred from the analysis of spatial patterns in ecosystems. Using an individual-based model, we evaluate whether basic assumptions of species' life-history, drought-susceptibility, and shade tolerance generate dynamics that replicate patterns between and within forest stands.

Location: Virginia piedmont, USA.

Method: Model verification examines the transition in forest composition and stand structure between mesic, intermediate and xeric sites. At each site, tree location, diameter, and status were recorded in square plots ranging from 0.25 to 1.0 ha. Model validation examines the simulated spatial pattern of individual trees at scales of 1–25 m within each forest site using a univariate Ripley's K function.

Results: 7512 live and dead trees were surveyed across all sites. All sites exhibit a consistent, significant shift in pattern for live trees by size, progressing from a clumped understorey (trees ≥ 0.1 m in diameter) to a uniform overstorey (trees > 0.25 m). Simulation results reflect not only the general shift in pattern of trees at appropriate scales within sites, but also the general transition in species composition and stand structure between sites.

Conclusions: This shift has been observed in other forest ecosystems and interpreted as a result of competition; however, this hypothesis has seldom been evaluated using simulation models. These results support the hypothesis that forest pattern in the Virginia piedmont results from competition involving species' life-history attributes driven by soil moisture availability between sites and light availability within sites.

Nomenclature: Harlow et al. (1996).

Abbreviations: ALB = Above-ground live biomass; BA = Basal area; CSR = Complete spatial randomness; DBH = Diameter at breast height; LAI = Leaf area index; MUSE = Multistrata spatially explicit model; PET = Potential evapo-transpiration.