Introduction

Strong advances in understanding forest ecology and forest-ecosystem responses to disturbance and environmental change have come through the application of systems analysis and simulation in studies of small forested ecosystems. For example, the whole series of gap models of forest succession (e.g. JABOWA (Botkin, Janak & Wallis 1972), FORET (Shugart & West 1977), FORTNITE (Aber & Melillo 1982), LINKAGES (Pastor & Post 1985), FORENA (Solomon 1986) and FORSKA (Leemans 1989); see Shugart (1984) for an overview of several of these models) examines forest ecosystems at the spatial scale of one or several large trees, i.e. about 0.1 ha. These models represent areas of sufficient spatial extent to represent adequately important processes such as inter- and intra-specific competition, and soil–vegetation–atmosphere interactions. Results from such models might reasonably be scaled up spatially to the level of the stand, where stand is defined as an ecosystem with a relatively homogenous community of trees and relatively homogenous site conditions compared with neighboring ecosystems. However, generalization of results from stand-level models (as the gap and other microcosm models will now be called) to forest ecosystems containing many stands, without the use of specially formulated forest simulation models, can be dangerous, if not absolutely misleading.

Ecologists generally recognize that ecosystem boundaries are more or less arbitrary. Some such boundaries are easy to assign and are ecologically very meaningful, such as the perimeter of an island in a lake, or the perimeter of a farm woodlot surrounded by field crops.