cation of mechanistic or simulation models, and the interpretation of results
(Van Winkle and Dale 1998). Using models can be a part of the scientific
process even when initial information about a system is sparse. Models can
be used to organize existing information, indicate the sensitivities of the
system, and identify gaps in knowledge. For example, Aber and Driscoll
(1997) point out that “models are often more interesting when they fail than
when they succeed.” These failures often point to problems with original
hypotheses, data collection and/or storage, development and interlinking of
algorithms, or the data interpretation. Interim conclusions from modeling
often cause modifications in the original hypothesis and, possibly, the model
itself, thus setting the stage for the next iteration of asking questions via the
scientific process.
The model-building process itself is an iterative process. Many models
begin with simple assumptions or are based on general assumptions. Refinement of the model comes with more experience (e.g., data) and improved
means to express the experience (e.g., more powerful formulae with statistical analysis to better define the boundaries).
Pressing needs for decisions to be made or policy actions to be taken in
the face of uncertainty often force the use of incomplete or untested
models. Yet it is in those instances in which information is deficient that the
modeling process may be most useful. Many cases arise in which qualitative information is valuable. In fact, models typically use both qualitative
and quantitative information and do not always result in quantitative
projections. Modeling the effects of climate change is an example. No one
knows how much a change in temperature or precipitation will alter biota
in a given region, but it is still valuable to use models to explore the possible implications of various scenarios of climate change. Such scenario exploration informs policymakers about which aspects of the ecological systems
they should be most concerned.
Furthermore, a clear distinction between qualitative and quantitative
information used in models is neither realistic nor appropriate because
information forms a continuum (Van Winkle and Dale 1998). Frequently,
the lack of confidence about information is expressed by using inequalities
or upper and lower bounds. At other times, a rough mean tendency is used
to represent a general understanding about some unmeasured quantity
[such as the assumption that past windstorms removed 20% of the biomass
of impacted forests in New England (Aber and Driscoll 1997)]. This type
of semiquantitative or categorical knowledge is frequently the basis of
equations and parameter values that are used in models and can be important to increasing the understanding of the ecological system.
16.3.4 Exploring Alternative Futures
The purpose of many model analyses is to help predict future conditions
built on a basis of “what if.” Some models have short-term outlooks, while
16. New Directions in Ecological Modeling for Resource Management
317
(Van Winkle and Dale 1998). Using models can be a part of the scientific
process even when initial information about a system is sparse. Models can
be used to organize existing information, indicate the sensitivities of the
system, and identify gaps in knowledge. For example, Aber and Driscoll
(1997) point out that “models are often more interesting when they fail than
when they succeed.” These failures often point to problems with original
hypotheses, data collection and/or storage, development and interlinking of
algorithms, or the data interpretation. Interim conclusions from modeling
often cause modifications in the original hypothesis and, possibly, the model
itself, thus setting the stage for the next iteration of asking questions via the
scientific process.
The model-building process itself is an iterative process. Many models
begin with simple assumptions or are based on general assumptions. Refinement of the model comes with more experience (e.g., data) and improved
means to express the experience (e.g., more powerful formulae with statistical analysis to better define the boundaries).
Pressing needs for decisions to be made or policy actions to be taken in
the face of uncertainty often force the use of incomplete or untested
models. Yet it is in those instances in which information is deficient that the
modeling process may be most useful. Many cases arise in which qualitative information is valuable. In fact, models typically use both qualitative
and quantitative information and do not always result in quantitative
projections. Modeling the effects of climate change is an example. No one
knows how much a change in temperature or precipitation will alter biota
in a given region, but it is still valuable to use models to explore the possible implications of various scenarios of climate change. Such scenario exploration informs policymakers about which aspects of the ecological systems
they should be most concerned.
Furthermore, a clear distinction between qualitative and quantitative
information used in models is neither realistic nor appropriate because
information forms a continuum (Van Winkle and Dale 1998). Frequently,
the lack of confidence about information is expressed by using inequalities
or upper and lower bounds. At other times, a rough mean tendency is used
to represent a general understanding about some unmeasured quantity
[such as the assumption that past windstorms removed 20% of the biomass
of impacted forests in New England (Aber and Driscoll 1997)]. This type
of semiquantitative or categorical knowledge is frequently the basis of
equations and parameter values that are used in models and can be important to increasing the understanding of the ecological system.
16.3.4 Exploring Alternative Futures
The purpose of many model analyses is to help predict future conditions
built on a basis of “what if.” Some models have short-term outlooks, while
16. New Directions in Ecological Modeling for Resource Management
317