Ecosystem management implies a concern over time periods of tens to hu
ndreds of years for sites on a scale of tens to hundreds of hectares.
Decision makers need to be able to model the likely consequences of al
ternative management strategies at these temporal and spatial scales.
They therefore require models that can be constructed quickly and chea
ply, that capture the key components of the ecosystem, that respond pl
ausibly to management actions, and are easy to explain, modify and und
erstand. This paper presents the frame-based modeling paradigm as a re
sponse to these needs. Frame-based modeling is used to examine the eff
ects of soil, weather, fire and deer population density on management
of the white pine ecosystem in northern Minnesota. The management obje
ctive is to maintain white pine forest. The paper describes how curren
t understanding of seedling establishment, tree growth, competition, h
erbivory and the effects of fire and high winds, can be captured at a
consistent level of resolution in a model that can be presented, compl
etely, in a few pages of text. The paper goes on to describe how the m
odel was tested at three sires in northern Minnesota, and was then use
d to explore alternative management strategies. Our results confirm th
at it is comparatively easy to maintain a forest in early successional
stages by burning or clear-cutting, and in late stages by suppression
of fires and control of cutting. Establishment and maintenance of mid
-successional stages, such as red and white pine, is much more difficu
lt and requires a finely-tuned balance between natural disturbance and
management action. For example, the pine forest was only maintained o
n average for 117 years out of 1000 years (average among 1000 simulate
d stands) on poor soil with high fire frequency and high deer density.
When the deer population was low, when all wild fi es were suppressed
, and when prescribed ground fires were used to promote pine establish
ment, years in pine was 804 out of 1000 years. The model provides guid
ance for management decisions to maintain the desired conditions. The
paper draws some conclusions from this particular modeling exercise th
at are likely to be generally applicable. For example, the modeling ex
ercise illustrates limits to what management can hope to achieve, the
importance of maintaining strategies over long time periods, and the d
ifficulties of predicting and measuring success when the time horizon
is hundreds of years. (C) 1997 Published by Elsevier Science Ltd.