Conserving Creatures of The Forest

Conserving Creatures of
the Forest:
A Guide to Decision Making
and Decision Models for
Forest Biodiversity
March 1, 2007
K. Norman Johnson
Dept. of Forest Resources
Oregon State University
Corvallis, Oregon 97331
norm.johnson@oregonstate.edu
Sean Gordon
U.S.D.A. Forest Service
Pacific Northwest Research Station
Sally Duncan
Institute for Natural Resources
Denise Lach
Dept. of Sociology
Brenda McComb
Dept. of Natural Resources Conservation
University of Massachusetts
Keith Reynolds
U.S.D.A. Forest Service
Pacific Northwest Research Station
Table of Contents
Executive Summary
Chapter 1. Introduction Norm Johnson and Sally Duncan
Chapter 2. How Do We Make Decisions? Denise Lach and

Sally Duncan
12
Chapter 3. Conserving Biodiversity: The Challenge

Brenda McComb
21
Chapter 4. DSS: How Might They Help? How Might

They Hurt? Norm Johnson and Sean Gordon
35
Chapter 5. Tips For Successful DSS Use: Its All

About People Sean Gordon
47
Chapter 6. Shopping for the System that Meets

Your NeedsSean Gordon and Keith Reynolds
56
Appendix A. Case Study BriefsSean Gordon
68
Appendix B. DSS Abbreviations and References
88
This research was supported by a grant from the National

Commission on Science for Sustainable Forestry
Pen and ink illustrations by Laura Cummings
Permission to draw illustrations from photographs granted by
Richard VanBuskirk, silverspot butterfly
Tom Greer, Douglas squirrel
Kevin Karlson, purple martin
Chip Krilowicz, winter wren
Bill Leonard, banana slug
Phil Phillips, Greenes mariposa lily
Dilbert cartoons used with permission of United Media
The authors are indebted to Alison Moldenke and Gretchen Bracher for the editing
and layout of this booklet. Alison and Gretchen took a rough draft and turned out
the polished product that follows. The authors deeply appreciate their work.
Executive Summary
ver the last 30 years, we

have seen a revolution
in our laws, policies,
and attitudes toward
the creatures of the
forest. While there has always been
interest in the animals we hunt and
fish, the major change has been that
we now value creatures great and
small simply for their very existence.
The complete collection of living
things in a forest has been called
forest biodiversity. The definition
of biodiversity takes in every living
thing, from the most obscure arthropod to the grizzly bear, from the least
lichen to the most ancient redwood
tree. Biodiversity also takes in all the
processes that nourish and differentiate those living things: water and
nutrient cycles, food webs, energy
flows, insect outbreaks, even fire
intervals.
This booklet is about making
decisions that affect conservation of
biodiversity. Often these decisions
target a select group of species and
the habitats on which they depend
for example, How can we manage
our forests to produce more quail
coveys? Other times, the decisions
address the entire suite of species
How do we provide for all the species
that inhabit old-growth forests?
Forest managers on both public and
private lands make such decisions
regularly, and this publication has the
goal of helping them improve their
own ability, and that of the groups
that advise them, to make these
complex decisions in the context of

their many other objectives. How do
we make those decisions? When can
detailed analysis help? Both of these
questions will be addressed generally,
and then in relation to conservation
of forest biodiversity.
We would not survive if we had
to submit each decision we make in
our lives to a detailed evaluation of
goals, alternatives, and implications.
Thus, people have developed informal rules to help us get through the
day, such as Shoot first and answer
questions later and A bird in the
hand is worth two in the bush. More
generally, we have decision processes
that quickly integrate our goals,
choices, and outcomes in an almost
unconscious way. These processes are
often called on for decisions involving time pressure, high stakes, experienced decision makers, inadequate
information, ill-defined goals, poorly
defined procedures, and dynamic
conditionsexactly the context for
most of the work-related decisions
we face, and all too familiar in the
biodiversity management arena.
Much of the time, the combination of intuitive decision processes
and informal decision rules can
suffice. Sometimes, though, we
may want to take a deeper, more
systematic look. We argue that, as
complexity increases, so might the
usefulness of such moral algebra.
Our approach poses complexity in at
least two dimensions: (1) the social
dimension asks how many players
and how complicated and contentious are their relationships, and (2)
the informational dimension asks
to what degree available data and
relationships must be organized. Our
experience suggests that, as either or
both of these dimensions increase
in complexity, the need and call for
systematic, transparent analysis will
also increase.
In those cases, we may quickly
come face-to-face with sophisticated
decision support systems (DSS).
Fundamentally, DSS are tools that
help evaluate alternative options
or scenarios (decision), deal with
complexity (support), and have a
clear, reproducible protocol (system).
More specifically, DSS is the name
often given to computerized decision
aids to help with complex decisions.
Committing to DSS can have significant ramifications for decision
effectiveness, ranging from exhilarating to disastrous, so whether to use
a decision support system is a major
decision in its own right. In addition, it would be a mistake to think
of DSS merely as models into which
you feed data and the answer to your
biodiversity problems pops out the
other end. Rather, we believe they
should be seen as a possible component in the decision process, a set
of specified tools that can be useful
to enhance deliberation in certain
circumstances.
Thus, our philosophy on using DSS is somewhat cautionary,
in the sense that we suggest a very
clear-eyed approach to whether such

systems should be used at all. In this
booklet, therefore, we first describe
approaches to decision making, from
the intuitive to the deeply analytical. We then discuss the many facets
of complexity that often creep into
biodiversity decisions, look at how
DSS can both help and hinder the
decision process, provide some tips
on successful use of DSS, and evaluate existing systems. In addition, we
provide descriptions of 15 specific
cases where DSS have been used with
various degrees of success.
How do we make
decisions?
Most of us have conducted or
experienced what is known as a
classic decision-making process
composed of the following distinct
stages: characterize the problem
or question, identify a full range of
alternatives and determine criteria
for selecting one, collect information about each option and rate it
on the criteria, then make the final
decision based on the rating. Classic decision models are rational
in that they promise us an orderly
and systematic approach to complex
problems. Whats more problematic
is that this classic understanding
of decision-making has also been
defined as rational in the normative
sense that any other sort of decision
process is by default characterized as
irrational and therefore significantly
flawed, thereby making the resulting decision suspect. This rational
approach to environmental decisions
has provided the framework for federal decision making since passage of
the National Environmental Policy
Act and has led to the development
and use of decision models and deci-
sion support systems.

Recent decision research suggests that relatively few decisions
are actually made by individuals or
organizations following the stages in
the classical rational model. A new
decision-making model has emerged
through research examining people
making real decisions under typical
conditions. This approach is often
referred to as naturalistic decision
making, looking at decisions with
time pressure, high stakes, experienced decision makers, inadequate
most of the decisions we face, and
all too familiar in the biodiversity
management arena.
Expert decision makers do not
typically characterize the range of
possible alternatives as prescribed
by classical decision makingthis
is hard work and takes a lot of time.
Instead, their experience allows them
to recognize familiar elements, even
in unique situations, and identify a
reasonable reaction as the first one
to be considered. Generally, expert
decision makers evaluate each choice
on its own merits, rather than comparing it to other options. Decision
makers may cycle through several
iterations of choices as they determine best fit, but they tend to assess
the choice on its own characteristics.
All of this happens quickly, almost
unconsciously, for expert decision
makers as they search for the right
choice. So, what are decision makers
doing when theyre quickly searching
for and assessing options?
Intuition and mental simulation are two strategies that make for
expert decision makers. Experience
enables experts to see situations as
patterns and relationships that grow
out of past choices and understand

how they can grow into the future.
This intuition includes seeing many
things that are invisible to the rest
of us, such whats critical, what goals
we should be seeking, what to expect
(and not expect), and how to respond
based on our experience with similar
situations. While our ability to mentally simulate events into the future
is rather limited, experts develop the
capacity to work with larger, more
complex chunks of information.
How do we link decision support tools to the way most people
make decisions? Some people have
suggested calling them deliberation
support tools, or tools for informing
debates, dialogues, and deliberations.
They recognize the inherent value
of argumentation to the process of
deliberation, given its ability to open
new avenues of thought, stimulate
learning, further deliberation, and
uncover the multiple aspects of the
problems under consideration. In
this sense, decision support tools can
help engage the intuition and mental simulation skills outlined above.
Also, DSS can be seen as exploratory,
as aids in the process of thinking
problems through, even as games, to
discern the potential outlines of both
problems and solutions. In other
words, engaging a formal DSS ought
not to be the central, definitive step
in the decision-making process but,
instead, part of an ongoing learning process. The quest for a single
objective answer from a DSS is
misguided. Instead, iteratively running a formal DSS with the goal of
considering a variety of alternative
outcomes should in fact provide a
fruitful combination of classic and
naturalistic decision making, each
spiced with a judicious combination
of deliberation and argumentation.
Why are biodiversity

decisions so challenging?
Species are usually considered
the primary currency of biodiversity
conservation. But even conservation of species presents challenges.
Rare, threatened and endangered
species garner much attention politically, and species that are hunted
or are aesthetically appealing (e.g.,
deer, songbirds, wildflowers) are
often used as focal or special interest species in making biodiversity
decisions. But these are tiny drops
in the huge species bucket. Over 1.6
million species have been described
on the earth, and this is known to be
only a fraction of what occurs here.
Furthermore, patterns of species
richness (the number of species in
an area) for one taxonomic group do
not reflect patterns of other groups
very well at all. So there are clear
challenges to conserving biodiversity.
Another approach to understanding the complexity of life and
the functions produced from this
complexity is to understand the
ecosystem processes that support and
maintain life on the planet. Energy
flow, nutrient cycling, population
dynamics, population genetics, and
evolution all interact to maintain this
collage of life, and the many human
activities that disrupt these processes
can lead to cascading effects on different species. Ensuring that both
ecosystem structures and processes
are provided over an area through
management decisions increases the
probability that species and genes
will be conserved.
We have several options in conceptualizing biodiversity in a conservation strategy. A tiered approach to
decision making considers the needs
of some species explicitly, while as-
suming the needs of other species

will be met through a more generalized strategy of habitat protection,
management, or both. Scientists
simplify the problem by taking a
logical step-wise approach, albeit
with significant assumptions. The
coarse-filter/fine-filter approach is
often used as a basis for reducing the
risk of losing a species from an ecosystem. In this approach, the coarse
filter is applied by describing the distribution of biophysical classes (e.g.,
vegetation, slope, or stream classes)
that occur in an area of concern and
documenting the arrangement and
connectivity of these classes across
the landscape. These current conditions may then be projected into
the future under various alternative
management assumptions and compared with reference conditions to
evaluate future habitat suitability.
But the coarse filter doesnt catch
all of the species. Some are likely
to simply be rare enough, have low
enough reproductive rates or large
enough territories, or have been so
adversely affected by chemicals or
habitat loss that their decimated
populations require special attention.
Consequently, a fine filter is constructed that also attempts to maintain the habitat needs of selected,
often imperiled, species.
Multiple factors influence the
degree to which a decision regarding protection of biodiversity will be
effective. The spatial scale(s) at which
the decision is made, its context, and
the level of spatial detail used in the
decision all contribute to effectiveness. How will species ranges be
evaluated? What ownerships will be
considered? How will we deal with
the fact that species often do not recognize ownership boundaries? Similarly, the temporal framework within
which the decision is made is critical.

Will the decision meet the concerns
of constituents now? Ten years from
now? 100 years? And all decisions are
couched within a number of factors
associated with the ever-present uncertainty of ecological and sociological processes. How will we recognize
that we know relatively little about
the habitat needs of many species?
Or the priorities of the winners of the
next election? Or how public values
will shift? How do we keep these uncertainties clearly in mind while still
making effective decisions?
In recent years, many goals for
achieving biodiversity have been
prescribed in federal, state, and
occasionally local policies. Specific
decisions about species listed as
threatened or endangered under the
Endangered Species Act often require
site-specific analysis to address particular issues, such as timber sales.
Court battles may ensue from this
debate, resulting in judicial decisions
that have broader implications. Typically, though, more comprehensive
decisions are made relative to the
Endangered Species Act when recovery plans are drafted or when Habitat
Conservation Plans are developed.
These large plans not only require an
understanding of the effects of management actions on the target species
over large multiownership areas, but
also must consider effects on other
species and people. Also, state-managed forestlands often must conserve
biodiversity in the context of their
Trust duties to produce revenue for
schools and other purposes.
Private landowners, both large
and small, also often face biodiversity decisions. Some want to manage their forests, in part, to provide
quail habitat. Others, particularly
large owners, may want to demon-
strate biodiversity conservation to

their stockholders or to certification
groups. Also, conservation groups
may use biodiversity analysis to identify areas of high conservation value.
DSS have been brought to bear in
many of these decisions. They have
been employed to aid in large-scale
planning and the planning of individual landowners and to quantify
biodiversity hotspots. Thus, many
people and groups involved in biodiversity planning face the decision of
whether to use a DSS.
Very little has been published

about how DSS get used in realworld forest and biodiversity prob-
lem solving. Most of the literature

on relevant models focuses on their
technical specifications, i.e., what
kind of computational procedures
they use. To address this gap, we assembled a diverse team with considerable experience to write this
booklet. However, we also wanted to
ground our advice in a broader base
of experience, so we selected and
researched a variety of recent examples where DSS have been used in
forest biodiversity decision-making
processes. The lessons from these 15
case studies are discussed throughout
the booklet and described in detail in
an appendix.
From our own experience and
the case studies, we identify below a
number of ways in which DSS might
help, and a number of ways in which
they might hurt.
The overall moral of this compar-
DSS: How might they

help? How might they
hurt?
A DSS can help by
enabling consideration of detailed and complex information and relationships

surfacing nonobvious results
providing a mechanism to
help organize and channel
group thinking
providing a common process
for analysis
providing an institutional
memory for decision analyses
improving the information
base
improving transparency and
credibility with stakeholders/
public/policy makers
A DSS can hurt by
committing the organization

to a system beyond its capability
creating a slow, expensive,
cumbersome process
restricting creative thinking
focusing on the wrong problem
deemphasizing unknown or
poorly known aspects of a
system
creating the illusion of certainty and knowledge
creating a modeling priesthood
creating a feeling of powerlessness
ison is that decision support systems

can be very helpful to biodiversity
decisions or very problematic. Thus,
the art of their use is of paramount
importance. You have undoubtedly
noticed some of the similarities in
the help and hinder lists, e.g., DSS
can help focus group work, but they
can also lead groups down the wrong
path. Most of the negative issues
can be overcome with some careful
thought and thorough discussion
before embarking on a decisionmaking effort or conflict-resolution
scenario, just as many of the positive
contributions can fail without such
considerations.
Perhaps our most compelling
general observation is that if users
(that is, people who use the DSS
results, not just the users of the
software) are involved in the original
development of a DSS, they can be
empowered in the decision-making
process. Unfortunately, this is rarely
the case. Analytical approaches,
especially when embedded in computer code, are naturally difficult
to access and understand. It takes
significant effort and communication skills, which are vastly different
from analytical skills, to make DSS
broadly accessible. The roles of communication and analysis tend to get
separated when planning and executing projects. For those responsible
for implementing a DSS, the focus
is usually on the analytical needs
and methods, with communication
regarded as necessary only to transfer results back to the larger project
process. But this is an outdated
one-way modelremember technology transfer?and we all intuitively
know that when something is not
understood, it tends not to be trusted. If the users are empowered in the
development or selection of a DSS,
our points about organizing group

thinking, promoting transparency,
and establishing credibility with the
decision-making community can be
benefits. Conversely, a DSS can hinder decision making if it is imposed
on the participants without taking
the time and effort to facilitate their
involvement.
Tips on successful use of

DSS: Its all about people
Once you have clarified why you
want to use a DSS and which one
might best meet your needs, you
might think that gathering the data,
running the model, and presenting
the results will be fairly straightforward. In our experience, its not.
Undoubtedly, technical problems
will be encountered. Often there are
not enough data of the right sort, the
DSS is difficult to operate, and the
process and results are not easy to
communicate to affected stakeholders. Perhaps less obvious (at least to
us analytical types) are the social and
managerial aspects, which can make
or break the success of the effort.
Who should be involved?

It is a widely shared observation
that the demand for participation
in decisions about forest management is increasing. In this new era
of collaborative decision making,
it is commonly recommended to
(1) identify and include stakeholders, (2) have numerous and varied
opportunities for input throughout
the decision-making process, and
(3) provide opportunities early in
the problem-framing stage, rather
than only after specific alternatives
have been analyzed. While these are
certainly worthy goals, real-world
decision making is bounded by time
and resources. Our experience and

case studies suggest three important
lessons for successful participation
models for biodiversity decision
making:
Think about the mixture of technical expertise and stakeholder
values that you need.
Carefully select the participantsreach out for community leaders who can contribute
positive suggestions for improvement.
Work with relatively small groups
with long-term participation.
Who will do the DSS modeling?

There are two very different types
of needs in using a DSS in a decision-making process: social, relating
to how to involve people and groups
in the decision-making process, and
analytical, addressing how to run the
DSS and analyze the data. In some
cases, it may be possible to fulfill
both needs with one person or team,
but frequently the different skills
required will necessitate integrating
different people or teams. Three basic
choices surfaced in our case studies:
Engage the DSS designer. A plus
is that the work will be done by
an expert. A minus is that the
designer may have some other
objectives, such as trying new,
unproven methods.
Hire a consultant with experience in the DSS of interest.
Engaging such consultants is
probably the quickest and surest way to access DSS expertise.
They tend to be more client-focused than researchers, but also
more expensive. They may not
understand the organizations
decision-making needs as well as
insiders, or, on the other hand,
they might also be able to play

a more neutral role in a multistakeholder process.
Do it yourself. Developing inhouse expertise generally requires more time than the others
and is best selected if there is a
continuing need for the DSS.
How Should the DSS be used?

The process for using the tool can
be as important as the tool itself. Toward that end, the following themes
that emerged from our case studies
might be useful:
Define the subset of the problem
to be addressed by the DSS (realize that it cant address the whole
problem).
Understand that the problem
definition will be dynamic and
plan to communicate changes.
Understand how decision makers
gain confidence in information.
Pilot test on diverse data.
Solicit and incorporate local
information.
Be careful what you presentit is
easy to lose support and credibility.
Look for opportunities for visualization, but dont lull people into
a false sense of accuracy.
Utilize DSS to help structure
group work and accumulate
results.
We also want to emphasize two
additional points, which integrate the
themes from the three questions just
discussed. First, promote learning
about how the DSS works and make
sure that results are viewed criticallydo not take results at face value.
The broader set of users, as well as
the programmer, should be provided
with means to check the results, both

quantitatively and in reality on the
ground. Second, try to ensure that
the users drive the DSS development
and that this role is not ceded to the
scientists, technicians, or, worse yet,
the DSS itself. DSS are tools developed
by people, for people. Maintain the
perspective that the technical experts
work for the stakeholders and decision
makers, not the other way around. It is
important to be sure that the DSS does
not drive the discussion and decision,
but that the stakeholders make those
decisions using information from the
DSS, as well as other sources.
Choosing a DSS
Perhaps counterintuitively, we
address the choice of the particular
DSS to use last. We do this to stress
that serious thinking about whether
and how to use a DSS should occur
first. Our questions about who will
be involved, who will do the work,
and how it should be done should
also influence which DSS is right
for the context. From previous DSS
reviews, journal articles, and interviews with DSS experts, we generated a master list of over 100 DSS.
We then screened this list down to
those systems (~30) most applicable
to forest biodiversity decisions and
surveyed the designers for more in-
depth information.
Our survey covered 20 different
aspects of systems, including descriptions of their purpose and past
applications, specific indicators and
resources covered, costs, and hardware and software needs. Although
quite a few systems are available, we
found they can be grouped into the
following relatively few categories
by their overall focus and the major
functions they perform.
In addition, we present a matrix
that provides a quick reference to
in terms of the coarse and fine filter

approaches to biodiversity assessment described above. Second, we
show which types of major forest
disturbances are modeled by each
program, including harvesting, fire,
pests and pathogens, land use and
climate change. Third, we attempt to
identify the extent to which some of
these systems address the complex
and political nature of many forest
biodiversity decisions, including the
integration of information from different disciplines, the consideration of
Focus Major Function

Biodiversity

Forestry

General
Regional Assessment

Population modeling
Reserve selection
Activity scheduling
Forest growth & management
Evaluation & prioritization
Forest growth & management
Land use simulation & evaluation
Restoration prioritization
three major aspects of DSS functionality. First, we compared the outputs

of the individual DSS to the classes
of biodiversity indicators described
by the Montreal Process Criteria and
Indicators. This comparison will help
users who have identified their needs
DSS Count
2
8
5
9
3
2
2
1
mutiple spatial scales, and the support

provided for social negotiations. We
encourage readers to visit our website
(http://ncseonline.org/ncssf/dss),
where they can review and search the
survey details and find appropriate
contacts for follow up.
Chapter 1. Introduction
Norm Johnson and Sally Duncan
ver the last 30 years, we

have seen a revolution
in our laws, policies,
and attitudes toward
the creatures of the
forest. While there has always been
interest in the animals we hunt and
fish, the major change in that period
has been that we now value creatures
great and small simply for their very
existence. From spotted owls to spotted frogs, from fishers to mushrooms
and bats, we now have national, state,
and local commitments to keep them
thriving in our woods. More people
than ever before are using words like
ecosystem, habitat, viability, and biodiversity, and many of those people
know what theyre talking about!
The complete collection of living
things in a forest has been called
forest biodiversity. The definition
of biodiversity takes in every living
thing, from the most obscure arthropod to the grizzly bear, from the least
lichen to the most ancient redwood
tree. Butand this is crucialbiodiversity also takes in all the processes
that nourish and differentiate those
living things: water and nutrient
cycles, food webs, energy flows,
insect outbreaks, even fire intervals. No matter where you place the
boundaries for your own working
definition, you can be certain of one
thing: biodiversity means complex-
ity, and no decision about conserving

biodiversity, thoughtfully made, can
ever be simple.
This booklet is about making
decisions that affect conservation of
biodiversity. Often these decisions
target a select group of species and
the habitats on which they depend:
for example, how can we manage our
forests to produce more quail coveys?
Other times, the decisions address
the entire suite of species: how do
we provide for all the species that
inhabit old-growth forests? Forest
managers on both public and private
lands make such decisions on a regular basis, and this publication has the
goal of helping them improve their
own ability, and that of the groups
which advise them, to make these
complex decisions in the context of
their many other objectives.
We anchor this guide in a more
general discussion of ways people
make decisions. Few hours go by
without the need for decisions both
large and small: what to wear, where
to eat lunch, whom to marry, which
car or house to buy? How do we
make those decisions? When can
detailed analysis help? Both of these
questions will be addressed generally,
and then in relation to conservation
of forest biodiversity.
We would not survive with our
sanity intact if we had to submit each
decision we make in our lives to a detailed evaluation of goals, alternatives,

and implications, as is often advocated
in books about decision making.
Thus, people have developed informal
rules to help us get through the day,
such as A bird in the hand is worth
two in the bush. The box below offers
a collection of such aphorisms, and,
whether you see them as pithy, wise,
or thoroughly disheartening, youll
recognize their guiding power.
Some Rules for

Decision Making
1. Shoot first and ask questions later.
2. This above all, unto thine
own self be true.
3. Better safe than sorry.
4. A bird in the hand is worth
two in the bush.
5. Dont rock the boat.
6. Its now or never.
7. If your lips are moving, you
arent learning anything.
8. Its easier to get forgiveness
than ask permission.
9. An ounce of prevention is
worth a pound of cure.
10. When you come to a fork in
the road, take it.
are computerized decision aids to

help with complex decisions. Committing to DSS can have significant
ramifications for decision effectiveness, ranging from excellent to disastrous, so whether to use a decision
support system is a major decision in
its own right. In addition, it would be
a mistake to think of DSS merely as
models into which you feed data and
the answer to your biodiversity problems pops out the other end. Rather,
we believe they should be seen as a
possible component in the decision
process, a set of specified tools that
can be useful to enhance deliberation
in certain circumstances.
In certain circumstances. Yes, we
recommend caution and here insist
that the old adage applies: if you only
know how to use a hammer, youll
see everything as a nail. In other
words, DSS come in many different
shapes and sizes and perform many
different kinds and levels of task. If
you dont choose carefully, you may
end up with a Jaguar when you were
seeking a Jeep. Worse, you may end
up with a dump truck when all you
needed was a tape measure.
High
Social Complexity
More generally, we have decision

processes that quickly integrate our
goals, choices, and outcomes in an
intuitive, almost unconscious way.
These processes are often called on
for decisions involving time pressure,
high stakes, experienced decision
makers, inadequate information,
ill-defined goals, poorly defined
procedures, and dynamic conditionsexactly the context for most
of the work-related decisions we face,
and all too familiar in the biodiversity management arena.
Much of the time, the combination of intuitive decision processes
and informal decision rules can
suffice. Sometimes, though, we may
want to take a deeper, more systematic look, seeking to use what Benjamin Franklin so elegantly dubbed
moral algebra: divide a sheet of
paper into two columns, writing Pro
over one and Con over the other; fill
in ideas under the different heads
to clarify thinking; and then cross
out items of comparable weight on
each side until one side or the other
prevails. We argue that, as complexity increases, so might the usefulness
of such moral algebra. Our approach
poses complexity in at least two
dimensions: (1) the social dimension asks how many players and how
complicated and contentious are
their relationships, and (2) the informational dimension asks to what
degree available data and relationships must be organized. Figure 1.1
provides examples in each dimension. Our experience suggests that,
as either or both of these dimensions
increase in complexity, the need and
call for systematic, transparent analysis will also increase.
In those cases, we may quickly
come face-to-face with sophisticated
decision support systems (DSS). DSS
Low
Thus, our philosophy on using

DSS is somewhat cautionary, in the
sense that we suggest a very clear-eyed
approach to whether such systems
should be used at all. In this booklet,
therefore, we first describe approaches
to decision making from the intuitive
to the deeply analytical. We then discuss the many facets of complexity that
often creep into biodiversity decisions
and look at how DSS can both help
and hinder the decision process. We
complete the booklet with descriptions
and critique of existing DSS for those
headed down that path, a little technical guidance on choosing among them,
and descriptions of 15 specific cases
where DSS have been used.
Along the way, we hope to keep
your attention on this potentially
daunting and off-putting subject
through observations on decision
making from our own and others
experiences, as well as sidebars and
cartoons illustrating the wild and
varied approaches people take to
lifes decisions. After all, if we cannot claim to provide you with all the
answers, we want to claim we helped
you enjoy framing the questions.
Decision about
whether to
remove geese
from the city pond
Reintroduction of
grizzly bears into
Montana wilderness
Individual
deciding which
kind of bird seed
to use in a feeder
Firm deciding where to

put harvest units over
the next few decades to
distribute deer forage
across the landscape
on a continuing basis
Low
High
Information Complexity
Figure 1.1 Classification system for informational and social complexity.
10
11
Chapter 2. How Do We Make Decisions?

Denise Lach and Sally Duncan
Traditional
Understandings of
Decision Making
Most of us have conducted or
experienced what is known as a classic
decision making process: characterize
the problem or question, identify a
full range of alternatives and determine criteria for selecting one, collect
information about each option and
rate it on the criteria, then make the
final decision based on the rating.
Various so-called stage models like
this have been proposed from two
stages (idea getting and idea evaluation) to seven or eight (problem sensing, problem definition, alternative
generation, alternative evaluation,
choice, action planning, and implementation). Some of these models
propose that successful decisions require that the stages be completed in
sequence; others propose that effective
decision making may be more flexible
[see Lipshitz and Bar-Ilan (1996) for
a concise review of classical decision
making]. For most stage models,
however, the goal is to optimize, or
come up with the best choice. Most
computer-based decision support
systems are based on the stage model
of decision making, and the computer
allows us to weight and rank multiple
alternatives on many variables very
quickly and easily.
Classic decision models are

rational in the descriptive sense that
they are based in cognitive activities such as decomposition (breaking the problem into small units or
basic elements), decontextualization
(removing the ambiguity inherent in
situations by finding units that are
independent of context), calculation
(applying formal procedures, such as
the rules of logic or statistical analysis), and description (telling others
how we made the decision). Rational decision making promises us an
orderly and systematic approach to
complex problems.
Whats more problematic is that
this classic understanding of decision making has also been defined as
rational in the normative sense.
Any other sort of decision process
is by default characterized as irrational, and therefore significantly
flawed, making the resulting decision
suspect. As natural resource professionals, we want to be perceived as
systematic problem-solvers, working
through transparent processes that
allow others to follow our reasoning
and choices. In some cases, it is mandated by law or normalized through
practice that we use some sort of
decision-support system based on
the classic decision model.
The demand for rational decision making and accountability in the
12
conservation of biodiversity may have

reached a peak with the passage of
the National Environmental Protection Act (NEPA). The Act demanded
analytical accountabilitydeveloping
a range of alternatives and specifying their environmental implications.
In addition, the regulations associated with NEPA call for a Record
of Decision in which the decision
maker defends the decision reached
by weighing the pros and cons of the
different alternatives in a logical manner, much like Benjamin Franklins
moral algebra from long ago.
This rational approach to environmental decisions has provided
the framework for federal decision
making since passage of NEPA and
has led to the development and use
of decision models and DSS. But the
approach sits uneasily in a world
where naturalistic decision making
(described below) tends to dominate.
Irrational Decision
Making?
Recent decision research suggests
that only about 5% of all decisions
are actually made by individuals or
organizations using a classic stage
model (Klein 2001). In many of those
cases, the model was used to justify a
decision made using other strategies
(e.g., Soelberg 1967). Starting from
the premise that effective decisions

are rational decisions, decision research based on classic stage models
has found a lot of incompetent and
nonrational decision makers. Academic careers have been built on
documenting the limitations of human decision makers, and millions
have been spent trying to fix the situation: training in the rational model
of decision making, and development
of decision support systems that
guide us through the rational model.
Yet we make decisionsgood
decisionsall the time. If were not
making rational decisions, then what
are we doing? A new decision-making model has emerged through
research examining people making
real decisions under typical conditions. This approachreferred to as
naturalistic decision makingis in
direct contrast with the laboratory
studies in which most classic decision research is conducted using
inexperienced decision makers (usually undergraduate students) making well-defined decisions. The new
research is looking at decisions with
time pressure, high stakes, experienced decision makers, inadequate
most of the decisions we face, and all
too familiar in the biodiversity management arena (Box, right).
Naturalistic decision-making
researchers have identified several
strategies of effective decision making by studying both novice and
expert decision makers in multiple
settings. Most of these will seem
commonsensical to you, and you
probably utilize these strategies more
or less unconsciously all the time.
The trick to becoming a better decision maker is to become aware of,
Chess, Anyone?
An article on chess in the New Yorker provides an intriguing mathematical perspective on decision making:
With about 10128 possible unique gamesvastly more than there are
atoms in the known universechess is one of mankinds most complex
activities. In an average arrangement on the board, white has thirtyfive moves and black has thirty-five replies, yielding twelve hundred
and twenty-five potential positions after one full turn. With subsequent
moves each of these positions branches out exponentially in further
lines of play1.5 million positions after the second turn, 1.8 billion after
the third---forming a gigantic map of potential games that programmers
call the search tree."
How human beings confront this complexity and seize on a few good
moves remains a mystery. Experienced players rely on subconscious faculties known variously as pattern recognition, visualization and aesthetic
sense. All are forms of educated guessworkaids to making choices
when certainty through exhaustive calculation is impossibleand may
be summed up in a word: intuition.Your Move, Tom Mueller, 12/12/05
Characteristics
of Naturalistic
Tasks and Settings
complex decisions
ambiguous and uncertain
information
large quantity of information to consider
poorly structured problems
shifting, poorly defined, or
competing goals
iterative outcomes due to
ongoing evaluation
high stakes and consequences for decision makers
organizational goals and
norms involved
dynamic decision making
environment
time constrained
adapted from Zsambok and Klein (1997)
13
and expert in, the strategies that we

use to make most decisions.
It turns out that expert decision makers do not typically
identify the range of possible
alternatives, as prescribed by
classical decision makingthis
is hard work and takes a lot of
time. Instead, their experience allows
them to characterize even unique
situations as familiar and identify a
reasonable reaction as the first one to
be considered. This strategy has been
labeled satisficing (Simon 1957) and,
instead of optimizing, expert decision makers evaluate each choice on
its own merits, rather than comparing it to other options. Decision
makers may cycle through several
iterations of choices as they determine best fit, but they always assess
the choice on its own characteristics.
All of this happens quickly for expert
decision makers, almost unconsciously, as they search for the right
choice. So, what are decision makers
doing when theyre quickly searching

for and assessing options?
One of the most powerful decision skills turns out to be what is
commonly called intuition, using our
experience to recognize key patterns
that indicate the dynamics of a situation (Klein 1999: 31). In a recent
New York Times bestseller, Malcolm
Gladwell describes how decisions
happen in a blink (Gladwell 2005).
Expert decision makersas we all often areare able to size up a decision
and almost immediately know how
to respond; our experience affects
the way we see whats happening and,
in many cases, whats not happening that should be. We are able to
ascertain whats critical, what goals
we should be seeking, what to expect,
and how to respond, based on our
experience with similar situations.
And, it all happens in the blink of an
eyeso quickly, its hard to think of it
as conscious consideration! Intuition
has a bad reputation with professionals; we cant usually tell someone how
we used our experience to recognize
patterns in a situation, as it seems
almost accidental and peripheral to
our analytical thinking. If it makes
you feel any better, there appears to
be a biological basis for intuition:
brain-damaged patients are unable
to anticipate consequences of good
and bad decisions. In undamaged
subjects, intuitionlooking for and
discarding patterns and assessing
future consequences of our decisionsappears to be activated long
before we are even aware of making
decisions (Bechara et al. 1997).
Intuition and mental simulation
are two strategies that make for expert
decision makers. Expert decision
makers see situations as patterns and
relationships that grow out of past
choices and understand how they can
Forest Creatures, Old Growth, and Timber Harvest

In the summer of 1989, lawsuits over the adequacy of protection of the northern spotted owl halted old-growth timber harvest
on federal lands in the Pacific Northwest. Those harvests had
been providing almost 15% of the softwood lumber in the United
States, providing employment for tens of thousands of people,
so there was some urgency to create a legally viable plan for the
spotted owl. To cope with this problem, the Chief of the Forest
Service and other agency heads asked Jack Ward Thomas to lead
an effort to develop a scientifically credible conservation strategy
for the northern spotted owl for the 20 million acres of its range.
Toward that end, Thomas gathered around him a cadre of scientific experts, called the Interagency Scientific Committee (ISC).
The ISC had six months to complete its charge and held a series of
public meetings at which professionals and the public could give
their facts, ideas, and proposals. After months of deliberation, the
ISC built a system founded on the fundamentals of conservation
biology: reserves large enough to contain self-sustaining populations of owls close enough together so that owls could disperse
between them. They also suggested new conservation rules for the
intervening forest and concluded that such a system would give
the northern spotted owl a reasonably high probability of survival.
Old-growth forests, though, contain hundreds of species. Congress, therefore, asked four senior scientists (Jack Ward Thomas,
Jerry Franklin, Norm Johnson, and John Gordon) to develop alternatives for conservation of old-growth forests in the Northwest
and estimate the implication of each alternative plan for forest
species and for timber harvest levels. They gave the scientists three
weeks to complete their task.
The four scientists gathered together hundreds of specialists
to help them map old growth and to develop conservation alternatives. They ended up with alternatives that varied in the size and
extent of reserves on federal land and management of the intervening forests and stream systems. They then assessed the risk
to species and ecosystems of each alternative, in terms of five risk
measures, and estimated likely timber harvest for the 20 million
federal acres in their study. Much of the risk assessment for the
different alternatives was done with pencil and paper over a few
days, with the scientists and specialists visualizing the future forest
under each alternativetruly an exercise in mental simulation.
They took six weeks to finish, rather than the three weeks they had
originally been given, and presented a report to Congress that
became the basis for discussion of old-growth forest conservation
for many years.
14
grow into the future. And expert decision makers see many things that are
invisible to the rest of us:
patterns the rest of us miss
events that did not happen but
should have
the big picture
mental models of the way things
work
opportunities and ways to improvise
events that either already happened or are likely to happen
soon
their own limitations
Most of us make most of our
decisions using intuition and mental
simulations. Strategies that take advantage of our experience are generally quite successful at relatively low
cost. We dont want to be developing
extensive decision support systems
for our dailyor perhaps even for
most ofour decisions. Its too timeconsuming and not guaranteed to
produce a better decision.
Group Decision Making

All the discussion to this point
has focused on the individual decision maker. What happens when we
need to make decisions as a group?
After all, decision making in our
professional lives almost always takes
place in a group context. The early
literature on group decision making typically concluded that groups
were prone to faulty decision making
and provided no real advantage over
individual decision making.
Many researchers found a phenomenon they labeled group think,
in which individuals in the group are
pressured into consensus to the point
that tolerance for dissenting view-
points was reduced, sundering the

groups ability to make sound judgments (Janis 1972). They suggested
that groups do not function very well
as cognitive units. In addition, researchers had a very difficult time finding evidence that the popular process
of brainstorming actually produced
better ideas than individuals working
on their own. The value of brainstorming may lie in allowing all group members an opportunity to be heard.
This early research on group decision making was based on the classic decision model described above.
And, even though researchers were
unable to characterize consistently
what it takes to make high quality
group decisions, everyone knew that
good group decisions are being made
all the time. As researchers started
to think about the naturalistic approach, they began asking somewhat
different questions about decision
making in groups.
In particular, cognitive psychologists were looking at processes
that operate beneath our conscious
awareness when were working in
groupslike intuition and pattern
matching. They found that the value
of group decision making is linked
to the complexity of the problem. In
particular, it appears that the more
difficult the problem to be solved,
the more beneficial it may be to
work in a group. In fact, as tasks and
problems become more complex
and multi-layered, it may be critical
to move to a group decision process
(Wilson et al. 2004). In addition, researchers found that groups who are
not dominated by a single individual
usually make better decisions than
those dominated by a strong-willed
or opinionated person whose perspectives override all others. These
latter groups are probably better
15
characterized as individuals looking

for support for their choices.
In a recent book, The Wisdom of
Crowds, Surowiecki (2004) reviews
the research and experiences people
have with group decision making.
Like the researchers described above,
he describes how diversity among the
decision makers is critical to good
group decision making. You cant expect a good decision from a group of
people who know nothing about the
topic at hand, but if you can bring
together a diverse group with varying
degrees of knowledge and insight,
Youre better off entrusting it with
major decisions rather than leaving them in the hands of one or two
people, no matter how smart those
people are (Surowiecki 2004: 31).
The primary reason to bring people
together to make a decision is to
increase the range of options that you
can evaluate and choose from.
It is possible to help people become expert decision makersthey
need to be able to think and learn
like experts. If you want people and
groups to be able to use their intuition to match familiar cases, you
need to expand their experience
base. People need to make many
decisions in many different situations. If few decision opportunities
are available, realistic scenarios can
be used for practice in sizing up numerous situations. You can also help
people and groups to become more
adept at mental simulations through
scenario premortems, or asking
decision makers to imagine what the
future looks like with their decision
in place. Others have found the practice of debriefing decision processes
to be valuable, especially in situations
with few decision opportunities.
Worldviews and Decision

Making
Externally imposed
restrictions on choices
The Fatalist
The Hierarchist
Of course, one of the reasons
we have difficulty making decisions
Fatalistic rationality:
Procedural rationality:
as a group is that we all bring difIt doesnt matter
A place for everything
ferent experiences, knowledge, and
expectations to the group. There is a
Individual
Group
large business (and a lot of money to
be made) in helping dysfunctional
The Individualist
The Egalitarian
groups operate effectively. One powSubstantive rationality:
Critical rationality:
erful tool for thinking about underlyWe
are right; therefore
The bottom line
ing differences that we bring to group
you must be wrong
decision making is a framework
No externally imposed
developed by Douglas, a cultural anrestrictions on choices
thropologist (e.g., Douglas 1982). This
framework has been tested in multiple Figure 2.1. Dimensions of social relationships (adapted from Douglas 1982).
settings in many countries and apthe individualism of the market and
Scientists looking at how we
pears to be a robust characterization
manage natural resources, like
of how people tend to view the world. the inequalities of the hierarchy; they
prefer
the
egalitarian
group
(Schwarz
forests, have found that organizaHer framework is based on two
and
Thompson
1990:
7).
Egalitarians
tions adopt management strategies
central and eternal questions for
stress the values of cooperative and
that reflect their understanding of
humans: Who am I? and How
the human-nature interface (Holling
should I behave? As a social species, volunteer relationships. Finally, there
1979, 1986). These myths of nature
are marginalized members of society
the answer to these questions is decan be graphically represented by a
who
feel
that
they
have
no
capacity
or
termined by how we relate to groups:
ability to influence events, leading to a ball in the landscape. Figure 2.2 is a
How connected are we to strong
fatalist approach in which outcomes, mapping of the worldviews regarding
groups that bind our decisions in
good or bad, are simply to be enjoyed nature onto the dimensions of social
different ways? How constrained are
relationships.
we by those ties to the groups? Figure or endured, but never achieved
Nature benign is a hospitable
2.1 describes how these two dimen(Schwarz and Thompson 1990: 8).
sions generate four basic and stable
Externally imposed
forms of social relationships.
Individualists stress the autonomy
The Fatalist
The Hierarchist
of individuals, especially their freedom to bid and bargain with others.
The bottom line is what they care for,
not relationships with people who
Nature capricious
Nature perverse/tolerant
come together to achieve results.
Individual
Group
Most of our organizations, however,
are made up of orderly and ranked
The Individualist
The Egalitarian
relationships in hierarchies. Organizational attempts to manage these relationships create a sense that it is more
Nature ephemeral
Nature benign
important to regulate who does what
No externally imposed
than to focus on an outcome (if there
even is one). According to scholars,
Figure 2.2: The myths of nature mapped onto the dimensions of social relationships
however, some people reject both
(adapted from Holling 1986).
16
worldview for individualists who

prefer to believe that human behavior and choices are unlikely to have
much impact on a totally resilient
natural world. A belief in an ephemeral nature works well for many
egalitarians who believe that stopping the destruction of the planet is
a moral imperative and can best be
accomplished through reducing our
footprint on the planetincluding
the size of our organizations and
other collective behavior. Nature
perverse/tolerant requires strong
management to ensure that were operating within the boundaries of the
natural world. Knowledge, certainty,
and predictability are needed to keep
everythingand everyonein their
place in this balanced world. Finally,
nature capricious reflects those
without influence in society, where
the natural world is, and remains, a
mysteryunknowable and unmanageable. All you can do is cope as
best you can in situations over which
you have no control.
This is no judgment or argument
that one of these worldviews is a better or a more effective way of seeing
the world. It is an attempt at explaining and understanding the pluralism
we bring to the way we make decisions. The three active perspectives
(hierarchical, individualistic, and
egalitarian) structure the world in
different ways; we have to learn to
use these differences in complementary, rather than conflicting, ways
as we work together in groups to
make decisions. In fact, our strongest
decisions are likely to be those that
blend the perspectives, bringing the
strength of each worldview to the
decision and ensuring that all parties
see how the solution addresses the
problems they see as critical.
Given the complexity of many
Water Rights and Learning to Share

In response to the demands of drought in southern California between 1987 and 1992, several water districts adopted a highly unusual
conservation rate structure. Rather than a cost of service structure, which
is standard for public utilities, the innovative rate structure funds all current operations by a base rate, although each water sector is allocated a
conservation use target. Calculations of the base rate were made with input from many stakeholders, who agreed that the amounts of water they
were expected to use and resulting costs were fair. After the new charges
were installed, water use actually declined.
The genius of this particular solution is its explicit focus on fairness.
Proposals for conservation pricing have foundered elsewhere when
advanced solely on the basis of efficiency. Furthermore, the procedure for
establishing the rate embodies three competing principles of equity, at
least one of which can be seen to appeal to each of the worldviews to be
found among the stakeholders.
First, each household is allocated the same fixed allowance for human
consumption, e.g., for drinking, cooking, bathing, etc., thus meeting the
requirement of parity characteristic of egalitarians, who view water as a
basic need and human right, rather than as a commodity.
Each household then receives an additional variable allowance for
use outside the house, mainly irrigating lots. This allowance is determined
by a formula that includes the area of each lot, the evapotranspiration
rates of typical plantings, and records of seasonal temperatures. Thus the
allowance varies by lot size and by month to allow for efficient irrigation of
gardens. Charts showing usage alongside allocation have proven effective
in correcting householders tendency to overwater their yards late in the
growing season. This second allowance satisfies hierarchical preferences
for proportionality in allocation.
Consumers wishing to consume in excess of these allowances are,
in principle, subject to an escalating scale of charges, the rate rising
more precipitously as consumption increases. This allows for individualist householders to assert their priority in allocation, should they choose
to do so. Individualists are also attracted by the market-like emphasis on
establishing prices and property rights.
Hence, egalitarians see a strong instrument that motivates conservation
as well as protecting everyones access to sufficient water for basic needs.
Hierarchists appreciate the rationality of the strategy and its ability to help
in long term planning. Individualists appreciate how the strategy protects
the customers freedom of choice to use as much water as can be afforded.
Story adopted from Lach et al. (2006).
of the problems we face and the

different worldviews held by our
colleagues, there are some problems that even the smartest among
17
us are unable to resolve. Typically,

decision makers muddle through
problems, with only small policy
changes resulting. We come up
with a satisfying solution and move

on to the next problem. Muddling
through is an idea introduced by
Charles Lindblom (1959) in the late
1950s and still studied today because it seems to describe decision
making so accurately, especially in
the policy realm. Lindblom argued
that this incremental decision
making was probably the smartest
way of doing things because it is
so difficult to evaluate the consequences of more radical decisions.
In addition to muddling through,
Lindblom also argued that redundancyhaving many decision makers working on a problem at the
same timehelps produce better
decisions. He uses The Manhattan Project as an example of how
a team of many scientists was able
to offset all the individual perspectives and errors only through the
inbuilt redundancy and diversity of
a group decision process.
New Roles for Decision

Support Systems
With the two decision-making models in mindrational and
naturalisticit is helpful to reconsider how you might use computerbased decision support systems
in new ways. For example, some
have playfully suggested seeing
DSS instead as DST (Deliberation
Support Tools) or TIDDD (Tools
for Informing Debates, Dialogues
& Deliberations) (OConnor 2004).
Along similar lines, certain scholars are investigating the inherent
value of argumentation to the process of deliberation (Parkins and
Mitchell 2005). Far from merely
being the precursor to entrenched
conflict, argumentation is posed as
a preferred focus for participatory
processes; it can open new avenues

of thought, stimulate learning and
further deliberation, and uncover
the multiple aspects of the problems under consideration. In this
sense, it can help engage the intuition and mental simulation skills
outlined above. However, experienced facilitation is recommended!
Another line of thought suggests
that DSS tools should be seen as
exploratory, as aids in the process of
thinking problems through, even as
games, to discern the potential outlines of both problems and solutions.
In other words, engaging a formal
DSS ought not to be the central, irreversible step in the decision-making
process but, instead, the first step in
an ongoing process.
Each of these different perspectives on the nature and role of DSS
recognizes that the quest for a single
outcome may be misguided. Instead,
iteratively running a formal DSS with
the objective of considering a variety
of alternative outcomes should in
fact provide a fruitful combination
of classic and naturalistic decision
making, each spiced with a judicious
combination of deliberation and
argumentation.
It is important to remember
that perhaps the most crucial step
is among the very first: problem
identification. Hastening through
this part of the process can too easily lead to long hours spent trying
to solve the wrong problem. Several
approaches have been developed
to enhance thorough exploration
of problem frames and thereby test
the clarity of the initial problem
statement. Two are mentioned here
merely as examples of the possible
paths available.
The first is called Soft Systems
Methodology. This well-tested
18
approachoften identified as an
inquiring and learning process
provides a framework for tackling
complex organizational and social
problems that typically resist easy
or quick definition (Checkland and
Scholes 1999, Wilson and Morren
1990) A key to problem definition,
according to Soft Systems practitioners, is in fact to resist the idea of
identifying a problem, for problems
do not exist independently of human beings: they are constructs of
concerned minds, defined by worldviews. Instead, they suggest, seek to
identify an overall situation. With a
situationa somewhat open-ended
conceptidentified and described,
shared perceptions, persuasion, and
debate are more likely to lead to improvements than is grappling with a
closed problem definition.
A second approach to understanding the shape and size of
problems before making decisions
about them is called cooperative inquiry (Heron 1996). This methodology is based on the fundamental
difficulty with empirical scientific
research, which is that, contrary to
the associated worldview, no one
can rightly claim objective knowledge or facts about what other
people want, feel, or believe. The
approach is based on a cycle in
which practical action alternates
with reflection. This cycle, notably,
is based on the premise of adaptive
management, which is not always
implemented in ways that include
or encourage reflection.
But reflection is surely a crucial
component of decision making, and
certainly one in which intuition and
mental simulation come fully into
play. To our cumulative loss, it is too
often passed over as an unproductive
use of time.
References Cited
Floral Growers and New Systems

Floral growers from a struggling cooperative in a semirural area needed help
sorting out their difficulties, so they called on their local agricultural college for assistance (Wilson and Morren 1990). Facilitators decided to use an intensive inquiry
approach called Soft Systems Methodology to fully understand the situation. Initial
inquiries with small and large growers, the state department of agriculture, the
county land use commission, the city council, and related floral commodity associations led to a synthesis of current situations, which revealed six major concerns:
lack of land to expand
insufficient political clout to affect
policy
competition between major floral associations
competition between large and small
growers
lack of relevant research
inadequate information from the
university and associations.
Facilitators then deeply investigated
the five following aspects of the floral
growers situation with them, rather than
trying merely to redefine their problems
for them:
Who will be responsible for carrying
out actions to change the situation?
Who will benefit or be adversely affected by changes?
Who will be the prospective decision
makers in the new situation?
What are the resources and constraints
affecting new activities?
What are the values and assumptions
(worldviews) of all these players?
After much discussion, the growers
and facilitators created three transformation
statements, oriented to the future and based
on developing new human activity systems:
(1) a system to share information and
to facilitate communications among
growers regarding technical, production, and marketing developments;
(2) a system to expand production to
allow for a greater volume, to make a
packing house economically feasible,
and to provide for year-round production and supply;
(3) a system to coordinate government

and private-sector efforts to plan,
promote, and expand the industry,
including making new land available.
The systems thus identified were
fleshed out with details on implementation strategy, performance measures,
expected date of completion, needed
resources, individual and group responsibilities, reporting procedures, and budgets. The floral growers were able to move
ahead in an environment now specifically
designed for change, performance, and
cooperation.
Notable pointers from the experienced facilitators in this exercise:
Do not try to reduce the themes of
concern in the early stages of inquiryif
they are collapsible, this will become apparent during subsequent discussions. If
they are neglected, they may come back
to bite.
In defining a situation, do not allow participants to make lists of
phrasesrequire them to use full
sentences, because observing
grammatical rules fosters orderly and
logical thinking.
Be prepared to go back to groups
individually and collectively many
times to fully comprehend their situations.
Ensure that participants precisely define
the people and groups they believe will
be future actors and owners, as well as
the environmental constraints; vagueness undermines the ability to identify
future workable systems.
(Wilson and Morren 1990)
19
Bechara, A., H. Damasio, D. Tranel, and A.

Damasio. 1997. Deciding advantageously
before knowing the
advantageous strategy.
Science 275:1293
1295.
Checkland, P., and J. Scholes 1999. Soft systems
methodology in action.
2nd ed. John Wiley,
New York.
Douglas, M. 1982. Essays in the sociology of
perception. Rougledge
and Kegan Paul, London.
Gladwell, M. 2005. Blink:
The power of thinking
without thinking. Little
Brown, New York.
Heron, J. 1996. Co-operative inquiry: Research
into the human condition. Sage Publications, London.
Holling, C.S. 1979. Myths
of ecological stability: Resilience and the
problem of failure. In
Studies in crisis management, G. Smart and
W. Stansbury (eds.).
Butterworth, Montreal, Canada.
Holling, C.S. 1989. The
resilience of terrestrial ecosystems: local
surprise and global
change. P. 292317 in
Sustainable development of the biosphere,
Clark, W. and R.
Munn (eds.). Cam-
bridge University Press, Cambridge, UK.

Janis, I. 1972. Victims of groupthink:
A psychological study of foreign
policy decisions and fiascos.
Houghton Mifflin, Boston, MA.
Klein, G. 1999. Sources of power: How
people make decisions. MIT Press,
Cambridge, MA.
Klein, G. 2001. Understanding and
supporting decision making:
An interview with Gary Klein.
Information Knowledge Systems
Management 2(4):291296.
Lach, D., H. Ingram, and S. Rayner.
2006. You never miss the water
till the well runs dry: Crisis and
creativity in California. Chap. 10
in Clumsy solutions for a complex world, M. Verweij and M.
Thompson (eds.). Cambridge
University Press, Cambridge,
United Kingdom.
Lindblom, C. 1959. The science of
muddling through. Public Administration Review 19:7999.
Lipshitz, R., and O. Bar-Ilan. 1996.

How problems are solved: Reconsidering the phase theorem.
Organizational Behavior and Human Decision Processes 65:4860.
OConnor, M. 2004. Environmental
governance through multi-media
multi-stakeholder dialogues?
Proceedings, Georgia Basin
futures project, conference on interactive sustainability, May 2004,
Vancouver, Canada.
Parkins, J.R., and R.E. Mitchell. 2005.
Public participation as public
debate: A deliberative turn in
natural resource management.
Society and Natural Resources
18:529540.
Schwarz, M., and M. Thompson.
1990. Divided we stand: Redefining politics, technology, and social
science. University of Pennsylvania Press, Philadelphia.
Simon, H. 1957. Models of man:
Social and rational. Wiley, New
York.
20
Soelberg, P. 1967. Unprogrammed

decision making. Industrial Management Review 8:1929.
Surowiecki, J. 2004. The wisdom of
crowds. Doubleday, New York.
Wilson, D., J. Timmel, and R. Miller.
2004. Cognitive cooperation:
When the going gets tough,
think as a group. Human Nature
15(3):225250.
Wilson, K., and G.E.B. Morren, Jr.
1990. Systems approaches for
improvement in agriculture and
resource management. Macmillan, New York.
Zsambok, C.E., and G. Klein, eds.
1997. Naturalistic decision making.
Lawrence Erlbaum, Mahwah, NJ.
Chapter 3. Conserving Biodiversity: The Challenge

Brenda McComb
ct
ru
al
Landscape
patterns
si
tio
St
ur
po
Habitat
cientists define biodiversity

as the genes and species of
an area and the ecosystem
processes supporting them.
Most nonscientists, by contrast, view biodiversity as a collage of
species, and many equate biodiversity with those species that are rare
and wild (Figure 3.1). Regional assessments or planning exercises often
apply biodiversity concepts to allow
planners and managers to make informed decisions. For instance, Gap
Analysis can, among other things,
portray areas that are likely to have
high species richness (Scott et al.
1993).
Ecologists have tried to conceptualize this complex term by using a
variety of images, such as that at the
left of Figure 3.1 (Meffe et al. 2002).
Key principles that are often included
in the definition of biodiversity are
those of structure, composition, and
function occurring at various scales
of space and time. But for scientists
and managers to meet the expectations of society for conserving
biodiversity effectively, the collage of
species (right, Figure 3.1) must also
be addressed.
For whole regions to approach
the conservation challenge, techniques such as Gap Analysis are
municate these approaches to the

public.
Species are usually considered
the primary currency of biodiversity
conservation. But even conservation
of species presents challenges. Rare,
threatened, and endangered species
garner much attention politically,
and species that are hunted or are
aesthetically appealing (e.g., deer,
songbirds, wildflowers) are
often used as focal
or special interCo
m
est species in
making bioLandscape
type
diversity
Communities
decisions.
But
Populations
required (Scott et al. 1993). Hence, a

triad of biodiversity perceptions, biodiversity concepts, and biodiversity
assessments must be interconnected
if we are to address biodiversity issues successfully. Clearly, the most
challenging aspects of biodiversity
conservation include deciding how
to understand complexity and uncertainty, protect both known and
unknown species,
and com-
Population
Genetic
Genes
Genetic
processes
Demographic
processes
Ecosystem
processes
Landscape
processes
Fu n c t i o n a l
What is Biodiversity?
Jim Petranka
Figure 3.1. The scientific concept of biodiversity is a nested set of processes and conditions that interact to reflect the breadth of life on the planet (left, adapted from
Meffe et al. 2002). That biological complexity is often perceived as a collage of life
by nonscientists (right, from www.unca.edu/tulula/biodiversity.html; used with permission). Scientific concepts can be used to conserve the collage of life by developing
maps of species richness for various groups of organisms.
21
these are tiny drops in the huge speSpecies and their Ranges
cies bucket. Over 1.6 million species
have been described on earth, and this
The geographic ranges of species and their relationships to the range
is known to be only a fraction of what
vary widely. Knowledge of where in the world a species occurs is the first
occurs. Furthermore, patterns of speconsideration in biodiversity conservation. Some, such as the Siskyou
cies richness (the number of species
mountain salamander (Plethodon stormi), have ranges so small that they
in an area) for one taxonomic group
remain vulnerable to a large disturbance or stresses that may accumulate
do not reflect patterns of other groups
among multiple management actions over time. Others, such as the northvery well at all (Flather et al. 1997). So
ern goshawk (Accipiter gentilis), extend around the globe in boreal and
there are clear challenges to conservnorthern hardwood forests. But because they require large areas to feed
ing biodiversity.
and reproduce, they remain vulnerable to habitat loss as well. Still other
Although species are the primary
species, such as the black-throated green warbler (Dendroica virescens),
currency of conservation, genetic
select eastern hemlock (Tsuga canadensis) forests in the northern part of
representation of species is also
their range, but are more associated with mesic hardwoods in the southimportant. Planners and managern part of their range. Hence one set of assessment criteria applied to the
ers usually assume that genes will
entire species geographic range would not likely give a realistic estimate of
be successfully conserved among
habitat availability (Collins 1983).
individuals within a species if we
can ensure the long-term viability of each species throughout its
3.2). Energy flow, nutrient cycling,
the functions produced from this
geographic range. If we artificially
population dynamics, population
complexity is to understand the
reduce a geographic range through
ecosystem processes that support and genetics, and evolution all interact to
policies or actions, it is assumed that maintain life on the planet (Figure
maintain this collage of life, and the
we reduce the genetic
diversity of the species Disturbance
Forest structure
Seed production
and increase the risk
and dispersal
events
and composition
that the species would
be less able to tolerate
climate change, exotic diseases or
Digestible
Moisture
predators, or further attacks on its
Soils
energy
stress
habitat in the future. This decision
in plants
rule, keep the gene pool intact by
keeping the subpopulations, is an
Litter
Herbivore
Vegetative
important assumption that probably
quality and
biomass
reproduction
is true for many species. We certainly
decomposition
see geographic variation in a species
appearance, diets, habitat selection,
and home range sizes within many
Availablility
Predator
Light
species of vertebrates. A reasonable
of
nutrients/
biomass
availablility
water
assumption is that these differences
reflect some evolutionary advantage
to the species in those places. Very
Competition among
Regeneration
Regeneration
rarely have these assumptions been
trees, shrubs, and herbs
consumption
establishment
tested, so the approach follows the
precautionary principle: err on the
Figure 3.2 Example of a set of ecological processes occurring in forest ecosystems
side of preservation.
that supports and maintains a variety of life. Ensuring that the plant communities,
Another approach to understructural conditions, and processes are all present increases the likelihood that spestanding the complexity of life and
cies and genes will be conserved.
22
many human activities that disrupt these processes

can lead to cascading effects on both described and
undescribed species. Ensuring that both the structure and ecosystem functions are provided over an
area through management decisions increases the Forest plan
probability that species and genes will be conserved.
How Do Scientists Conceptualize

Biodiversity to be Able to
Conserve It?
Species
A
Coarse filter
Legacy retention
Ecosystem state
or process
Mesofilter
If we acknowledge that biodiversity is a resource
that society is increasingly coming to value and understand, and if we recognize that it is mind-bendingly complex, what are we as scientists, managers,
Recovery plans
planners, or decision makers to do to ensure that
biodiversity is conserved for future generations?
Fine filter
How can we hope to understand and consider the
needs for all species in a planning area?
We have several options. A tiered approach
Figure 3.3. Coarse filter goals are met using vegetative types and sucto decision making considers the needs of some
cessional stages that are likely to meet the needs for many species (listspecies explicitly, while assuming the needs of
ed as A through G in this diagram) in a planning area. For those
other species will be met through a more general- not likely to be met using this approach, special features are retained
ized strategy of habitat protection and/or manage- for some species (B D, F, and G) (mesofilter), and a fine filter
ment. So scientists simplify the problem by taking (single-species) plan is developed for one species (D).
a logical step-wise approach, albeit with significant
assumptions. The coarse-filter/fine-filter approach
is often used as a basis for reducing the risk of
losing a species from an ecosystem (Hunter 1999)
Historic range
(Figure 3.3). In this approach, the coarse filter is
applied to the landscape by describing the distribution of biophysical classes (e.g., vegetation, slope,
Reduced variability
and stream classes) that occur in an area of concern,
and documenting the arrangement and connectivity of these biophysical classes across the landscape.
These current conditions may then be projected into
the future under various alternative management
New state
assumptions.
Time
The current and possible future conditions are
Disturbance
then compared to some reference condition(s).
Recently that comparison has quite often been to
Figure 3.4. An ecosystem condition (plant community representathe historical range of variability (HRV) (Landres
tion, successional stage area, wood biomass) fluctuates over time.
et al. 1999) (Figure 3.4). The more that the current
Human activities may reduce the amount of variability over time
and likely future conditions depart from the HRV,
(top line after disturbance) or cause the condition to depart from the
the greater the risk that genes or species may be
historic range (bottom line after disturbance) leading to increased
lost from the system. Scientists assume that species risk to biodiversity (Used with permission of K. McGarigal, pers.
are more likely to persist into the future under the
comm. 2006).
23
conditions that remain most similar

to the conditions that they persisted
in during the past. Hence providing
the structure and functions that fall
within the historical range of variation is likely to catch many species
in this management approach.
But the coarse filter doesnt catch
all of the species. Some are likely
to simply be rare enough, have low
enough reproductive rates or large
enough territories, or have been so
adversely affected by chemicals or
habitat loss that their decimated
populations require special attention. Consequently a fine filter is
constructed that maintains the coarse
filter structure but takes special
management actions to conserve the
set of species identified for fine filter
consideration (Figure 3.3).
What are Typical

Biodiversity Decision
Problems?
Regional biodiversity assessments and the goals that are derived
from them provide the context for
decisions and actions that occur at
smaller spatial scales. In essence,
goals are defined over large areas and
are achieved by implementing actions cumulatively over many small
areas. Hence biodiversity decisions
occur at all levels of management
and policy.
At the broadest spatial scale,
bioregional assessments have been
conducted in many parts of the U.S.
(Johnson et al. 1999). In many past
bioregional assessments and forest
planning efforts, biodiversity goals
were articulated as desired amounts
and distributions of plant communities and other biophysical resources.
Efforts such as the Willamette Valley
Alternative Futures project involved
stakeholders in the process from the

very beginning of the assessment,
and the emerging issues were raised
to the planners and the scientists
by affected constituents (Hulse et
al. 2002; Figure 3.5). These representatives of the public then can
decide, based on estimated gains and
losses in ecological and economic
resources, which land use approach
best meets their objectives. A similar
regional plan was developed by the
federal land management agencies in
the Pacific Northwest (sidebar, p. 25).
Within this context of a bioregional assessment, lets consider several situations as examples. A planner
from a nongovernmental organization (NGO) such as The Nature Conservancy may wish to identify areas
with high potential value for biodiversity conservation for purchase or
a conservation easement. Given the
complexities of land ownership and
multiple layers of information and
NGO goals, areas of high conservation value may best be identified
based on client-driven definitions of
what is most important in this quest
(Figure 3.6). Public involvement is

critical where private landowners will
be involved in determining the success or failure of any planning effort.
The Conservation Assessment and
Prioritization System (CAPS) (McGarigal et al. 2001) is one approach
to identifying high priority parcels in
complex landscapes.
A similar spatial planning problem is faced by a National Forest
planner developing standards and
guidelines for placement and size of
harvest units and seeking to achieve
connectivity goals for rare species.
She must identify the current and
future corridors that might be maintained over time while still allowing
active management in surrounding
stands. Where recreation, access, and
other social concerns also interact
with the corridor design, public
involvement will be just as important
as ecological assessment early in this
planning process.
Finally, consider a nonindustrial private forest manager from
the region who has both timber and
wildlife objectives. She must decide
Conservation
2050
Plan trend
2050
Pre-EuroAmerican
settlement
Circa
1990
Development
2050
Figure 3.5. Past and current conditions and alternative futures for the Willamette
Basin, Oregon. Darker areas represent older forested lands (from Hulse et al. 2002)
Used with permission of Oregon State University Press.
24
The Northwest Forest Plan

The Northwest Forest Plan (NWFP; U.S.D.A. and USDI 1994) was developed to address biodiversity concerns on federal lands and began with a focus on the geographic range of the northern spotted owl. Species
that fell into the high-risk group requiring particular attention to their needs included spotted owls (Strix occidentalis caurina), marbled murrelets (Brachyramphus marmoratous), red tree voles (Arborimus longicaudus),
and many others. Indeed, over 1,000 species were assessed to understand the risks of implementing one of
11 land management options under the NWFP. The ensuing species-by-species assessments constituted a
fine-filter approach designed to ensure protection for those species not captured by the coarse-filter planning strategy (moving more land into later successional stages).
But additional questions arise with a planning process like this. As older forests are allowed to develop,
young forests decline in abundance. This is particularly true of young forests that are structurally or compositionally complex, as would occur after a natural disturbance. Hence, focal species that are known to be associated with particular conditions (snags in young forests, hardwoods, shrubs, etc.) are selected as examples
of how a species might respond to changes in forest conditions. By projecting habitat availability for these
species forward, we can see how their habitat might change under current and alternative policies. In coastal
Oregon, Spies et al. (2007) projected future habitat in coastal Oregon for early successional snag associates
(e.g., western bluebirds) and high-contrast forest edge associates (e.g., olive-sided flycatchers). Current policies either maintain or slightly enlarge the amount of habitat for the species over time, but alternative policies that increase green tree retention at harvest are likely to be better for these species. Hence, inferences
can be made to other species that also might be associated with these conditions. It is important to remember, however, that these are not indicator species. Because each species has its own habitat requirements, the
responses of one species will never accurately reflect the responses of other species.
Figure 3.6. Example of applying

multiple landscape
filters to identify
areas of high and
low conservation
value as a precursor to exploring
conservation easements with private
landowners (from
McGarigal et al.
2001; http://www.
umass.edu/landeco
/research/caps/caps.
html; used with
permission of K.
McGarigal.)
25
Challenges to Making
Effective Biodiversity
Decisions
Multiple factors influence the
degree to which a decision regarding protection of biodiversity will
be effective. The spatial scale(s) at
which the decision is made, its context, and the level of spatial detail
used in the decision all contribute to
effectiveness. Similarly, the temporal framework within which the
decision is made is critical. Will the
decision meet the concerns of constituents now? 10 years from now?
100 years? What is the appropriate
time frame? And all decisions are
couched within a number of factors associated with the ever-present uncertainty of ecological and
sociological processes. How do we
keep these uncertainties clearly in
mind while still making effective
decisions? The following sections
discuss these issues in more detail.
Cumulative Species or Groups
how many trees to leave in a harvest

unit for cavity-dependent species.
She may just need some simple
guidelines regarding tree sizes,
numbers, and distribution and goals
for this set of species that could be
met easily at the stand level. For
instance, Mellen et al. (2002) summarized and organized information
on cavity-nesting birds for various
forest types in the Pacific Northwest
(Figure 3.7). By knowing the range
in snag densities most likely to
meet various goals for cavity-nesting birds, the landowner can decide
what tradeoff between bird habitat
and timber production seems reasonable within the regional standards and guidelines established by
the State Board of Forestry.
6
5
CNB
CND
DTWR
DRCR
CND
DTWR
DTWR
NFSQ
NFSQ
NFSQ
DOSQ
DOSQ
DOSQ
10
20
30
40
50
Tolerance level
30%
50%
80%
60
70
80
90
100
Snag Density (number/ha)

Figure 3.7. This figure shows cumulative species curves for density of snags >25 cm
dbh: Species use of areas for nesting, roosting, and occurrence with documented
snag densities for 30%, 50%, and 80% tolerance levels in the Westside lowland
Conifer-Hardwood Forest Wildlife Habitat Type and Small/medium Trees Structural Condition Class. The needs of cavity-nesting birds in a young managed stand
could be met by retaining or creating from 11 to 95 snags per ha, depending on the
tolerance level goals (degree of security provided) of the landowner. The letter combinations stand for different species of birds.. See DECAID by Mellen et al. (2002;
http://wwwnotes.fs.fed.us:81/pnw/DecAID/DecAID.nsf ) for more information.
Spatial scale
Land ownership implies a certain level of commitment to part of
the earth, and that commitment is
expressed through the accumulation
of individual landowner behaviors
over space and time. It may seem
obvious that one landowner making
a decision to manage for cavity-nesting birds in a stand on her land can
achieve her goal easily. Just leave a
certain number of trees or snags of
certain sizes, and the goal is reached.
Or is it? How will the actions of her
neighbors influence the likelihood
that her biodiversity objectives will
be achieved? And how will her actions influence the achievement of
her neighbors goals to provide a
corridor for migrating elk (Cervus
elaphus)? Can she trust her federal neighbors to follow through on
their plans even as administrations
26
change? Will her private neighbors

sell their land? Subdivide it? Will the
state impose restrictions on private
land management that inhibit her
ability to achieve her goals on her
land and that of her neighbors? Will
an NGO intervene to offer a conservation easement and purchase development rights? All of these questions
are driven by social values, and all
are played out on the patchwork
ownership quilt of the landscape.
Effective biodiversity conservation
decisions cannot proceed without
considering spatial context and the
implications of multiple ownerships.
As noted, biodiversity policies
and bioregional assessments that
support them often are formed at a
broad spatial scale (e.g., Northwest
Forest Plan, state forest practice
rules) and are implemented through
cumulative actions made at fine
spatial scales over time (e.g., forest
plans, habitat conservation plans,

and watershed council restoration
plans). The policy guides the actions
(e.g., how many wildlife trees to leave
in a clearcut, how wide a riparian
buffer strip should be), but decisions
made locally determine where and
how these actions occur.
Many people and agencies do
not appreciate that both this sociopolitical framework and the species
and ecological processes must be
considered in making biodiversity
conservation decisions. Large territory and home range sizes of some
species form one part of the picture,
and so does the need to ensure that
an adequate number of individuals
of each species must be maintained.
How large an area do we need to
consider to make effective decisions
that include habitat for marbled salamanders (Ambystoma maculatum)?
northern goshawks? wolverines
(Gulo gulo)? Where you draw the
line taxonomically in your assessments and decisions will influence
the spatial scale associated with the
decision-making process. Similarly,
dominant ecological processes will
influence the outcome of the decisions. Wildfires, insects, disease,
wind, ice and climate change all have
ranges of frequencies and intensities
associated with various locations,
and the spatial scales associated with
each kind of disturbance must also
be considered.
Planning in a Parcelized Environment

The 750-acre Mt Toby forest in Sunderland, Massachusetts is one of
several university teaching and research forests used by the Department
of Natural Resources Conservation at the University of MassachusettsAmherst (UMass). It was acquired in 1906 by the University and now is
surrounded by homes, roads, and other forest landowners, both public
and private (Figure 3.8). The forest receives thousands of user-days of
recreation use by local citizens and tourists each year. Most frequent uses
are fishing, hiking, biking, bird watching, and horseback riding. UMass
uses the property primarily as a teaching site to support classes and for
research. Active forest management has been limited to thinning of white
pine (Pinus strobus) stands in the past 10 years. Although the primary
goals for management of the property are to support teaching, research,
and demonstration by the Department, many stakeholders wish to have a
voice in management activities. Most vocal among those are nearby landowners. Hence, at the very least, the ability of UMass to meet management goals must consider the likely land use of the mosaic of landowners
adjacent to and near Mt Toby, and the ability of these landowners to meet
their objectives must consider future management actions on UMass
property. GIS and spatial planning tools helped students and faculty at
UMass portray likely future management alternatives to local landowners
in public meetings and to develop alternative management plans currently under consideration. There are two key challenges here: to help people
understand and manage for biodiversity at a landscape scale, and to
emphasize the concept of dynamic landscapes and the constant interaction of processes and
species through
time across
ownerships.
N
Mt. Toby
Forest
Figure 3.8. Map

of landowners in
the Towns of Sunderland and Leverett,
Massachusetts, surrounding the 750-acre University of
Massachusetts-Amherst Mt. Toby
teaching and research forest. Other public
and NGO lands are in white; industry lands are
in black; nonindustrial private forest and agricultural owners are in a multicolored mosaic.
Time
Politicians may view effective
time frames for decisions as days,
weeks, maybe years or, more rarely,
decades. Some of us try to plan for
our financial security by thinking in
multiple decades. And most people
want to leave a legacy of their values
to the next generation. But humans
27
the inherent rates of growth and

disturbance affecting vegetation
the potential and real rate of
population growth for key species
the rate of movement and colonization of habitat for key species.
At least two additional major

factors must be considered when
making biodiversity decisions over
large multiowner areas. Land tenure
can influence achievement of goals,
particularly if the parcels being sold

or inherited change owners having
one set of core values to owners with
another. Timber harvesting intervals
(or rotation lengths) also influence
Historical Range of Variability (3rd)

If we choose to use the historical range of variability (HRV) as a guide
to making decisions at a landscape or watershed scale, then we must
consider the effect of spatial scale on this range of conditions. In general,
as the spatial scale at which a decision is made gets finer, variability in
conditions increases. Take, for instance, the range in proportion of old
forests over the Oregon Coast Range at two different spatial scales (Figure 3.9).
The proportion of the Oregon Coast Range, taken as a whole, that
likely was old forest at any one time is more stable than is the the proportion of old forest likely to be in the national forest within the Range. The
variability is even greater at the scale of an individual reserve within the
national forest. Consequently this concept is not of much value when
making decisions at small spatial scales, but may be quite useful when
guiding decisions at regional scales.
(a) Province scale (5,000,000 acres)

100
80
Percentage Old Growth
have more difficulty thinking in

terms of multiple lifetimes. Therefore, we have trouble considering the
types of decisions required to achieve
biodiversity goals that wont produce
obvious results for many decades
(e.g., the recovery of nesting habitat
for spotted owls) or several centuries
(e.g., returning the Oregon Coast
Range to within the historic range
of variability). Suppose we decide
to designate a part of the landscape
that at present contains many square
miles of young plantations as late
successional reserves. The desired
result could take three human lifetimes (200 years or more) to show
itself, leaving many, or even most,
constituents unable to see how we are
approaching the intended goal.
It is important not to think just
about human lifetimes in the conservation arena. Consider long-lived
species such as box turtles (Terrapene
carolina) and Puerto Rican parrots
(Amazona vittata). These are such
long-lived species (40 years or more)
that options for recovery may be
very limited by the time declines in
populations are detected. Similarly,
recovery, when it is possible, may
take multiple generations, as much as
hundreds of years.
Consequently, the appropriate
time frame for planning biodiversity
conservation does not have a standard time frame, because it needs to
be driven by several key interacting
variables:
human schedules for implementing the plan
60
40
20
0
(b) National forest scale (600,000 acres)

100
80
60
40
20
0
3000
2500
2000
1500
1000
500
Years Before Present

Figure 3.9. The range of variability in old-forest area in one 3000-year simulation for: (a) entire Coast Range and b) national forest. From Davis et al.
2001; used with permission of Waveland Press.
28
the periods during which a forest will

be suitable for a set of species. These
factors are often integrated into
analyses as transition probabilities
to help decision makers consider the
likely direction of landscape change
over the planning period.
Uncertainty
One of the greatest uncertainties facing conservation biologists
and land planners is development
of conservation and management
strategies with incomplete information about the suite of species under
consideration. Past research on many
species allows us to develop reasonable management plans for them. For
some species we know nearly nothing, however, and many species are
yet to be discovered. In the face of
this uncertainty, managers may use
one or several DSS to organize and
synthesize information, hoping to
improve their decisions. Obviously,
DSS rely on what is known, not what
is unknown, and therein lies the biggest challenge. Information-hungry
DSS can do only a minimally useful
job of aiding decisions for poorly understood species, so how helpful are
they likely to be for species that have
not yet been identified?
Thus, biodiversity planners must
recognize that professional judgment is a key part of the decision
process. If biodiversity conservation
is a primary goal, then a reasonable
course of action is to follow the precautionary principle and err on the
side of conservation over resource
extraction. The plan should include
techniques that add to the information required to make decisions, such
as adaptive management and formal
monitoring protocols.
Planning requires identifying the
finest level of information needed,
Estimating Recovery Periods

Consider the potential for documenting responses of organisms to
management actions. The list of species below is ranked by their potential
longevity in yearsthe maximum time needed for a complete turnover in
a generation. The number of generations is then portrayed for a 40-, 100and 200-year rotation. The higher the number of potential generations, the
lower is the risk of displacement (and dispersal) to other home ranges.

Rotation length (years)
Species
40 100
200

Longevity (yr) Generations
Short-tailed shrew (Blarina brevicauda)
3
13
33
67
Winter wren (Troglodytes troglodytes)
4
10
25
50
Spruce grouse (Falcipennis canadensis)
5
8
20
40
Red-cockaded woodpecker (Picoides borealis)
16
3
6
13
27
1
4
8
Great-horned owl (Bubo virginanus)
Box turtle
50
1
2
4
Clearly, natural disturbances have a probability of occurring at a variety

of intervals that also could lead to displacement for these species. As rotation lengths depart from the historic range of natural disturbance frequencies, the risk to long-lived species (which often have low reproductive rates)
increases. In addition, because they are so long lived, changes in populations can be subtle, making it difficult to detect population declines. For the
same reason, documenting recovery of these species can also be difficult,
requiring long periods of population monitoring.
the outer bounds of the problem, and

the surrounding landscape conditions. After making these critical first
decisions, decision makers suddenly
find themselves faced with a number of fundamental uncertainties. A
central one is the degree of continued
social commitment to biodiversity
values. Societal values change and
the decision-making process must
adapt. Values placed on deer, for
example, have evolved from largely
utilitarian, to protection, to recreation, to nuisance, to public health
concerns. How will values change for
spotted owls, Townsends big-eared
bats (Plecotus towsendii), or American burying beetles (Nicrophorus
29
americanus)? We tend to think of

goals and objectives as being relatively stable, but our decisions about biodiversity protection need to reflect
our ability to adapt to new values.
There also are biophysical uncertainties: fires, floods, invasive plants
and animals, disease, and global
climate change, to name a few. For
many of these factors, we have information that can help us understand
probabilities of occurrence over time
at various scales. In such cases, the
uncertainty or likelihood of occurrence can be quantified. This also
allows us to assign risks associated
with these events. Many forest health
issues are framed within this risk as-
sessment paradigm. For instance, an

unhealthy forest is one with a higher
than expected chance of wildfire,
disease, or insect attack, and the
ensuing ecological and social effects
can be predicted.
In some cases, these risks can be
expressed as a departure from the
historic range of variability (HRV),
but in many settings we can no longer squeeze a system back into the
HRV. For instance, in the northeastern U.S. there are no longer passenger pigeons (Ectopistes migratorius),
American chestnut trees (Castanea
dentata), or wolves (Canis lupis).
Atlantic salmon (Salmo salar) are
effectively gone, and society will not
likely pay for restoration to historic
levels. Although some aspects of the
northeastern hardwood forests and
associated streams have recovered,
some aspects will never recover.
Hence, the past may help inform decisions, but it is not invariably a good
template. And even the ability of the
past to inform decisions is weakened
as we see systems develop in ways
they never have before following effects of development, climate change,
and invasive species, for example.
Uncertainties about how aspects of
biodiversity might respond thus tend
to proliferate as historys guidance
becomes increasingly vague.
Finally there are political uncertainties. Although political decisions
are usually an outcome of societal
values, our political system and those
of other countries can result in decisions being made that produce significant constraints on our ability to
achieve biodiversity goals: the Kyoto
agreement, 9/11, economic recovery,
going to war. Once made, these decisions significantly affect the certainty
of achieving our biodiversity conservation objectives. Changes in policy,
Chasing a Mythical Creature: The Ivory-billed Woodpecker

(Campephilus principalis)
For decades, ecologists have recognized that the formerly extensive
bottomland hardwood forests of the lower Mississippi Valley have been
converted to agriculture, and the remaining pieces are deeply fragmented. Clearly a number of species have declined, including Louisiana black
bears (Ursus americanus). Many people had assumed that such habitat
loss and fragmentation also had led to the extinction of the ivory-billed
woodpecker, until the spring of 2004 when a team of biologists found and
filmed one of the elusive birds in the Cache River National Wildlife Refuge
in Arkansas (Figure 3.10). Or did they? Failure to corroborate the sighting
over the past two years has led to questions regarding this potential rediscovery. But if ivory-billed woodpeckers still do exist, then the Cache River
Basin is a likely location.
The Cache River basin, the Pearl River basin, and the Atchafalaya basin
are three of the most extensive remaining tracts of bottomland hardwoods. All three areas have had undocumented reports of this species before 2004. We know virtually nothing about the specific habitat needs of
this species, yet, if confirmed sightings continue, we will now logically be
expected to manage and conserve it. One significant source of uncertainty
lies in the knowledge available versus the knowledge needed to manage
uncommon or rare species using a fine filter approach.
Pre-colonial extent of
forested wetlands
Extent of forested
wetlands in 1982
Cache River
Pearl River
Atchafalaya Basin
Figure 3.10. Historic and recent distribution of bottomland hardwood forest,

with most likely habitat for ivory-billed woodpeckers noted. Map from USGS
http://biology.usgs.gov/s+t/noframe/m1107.htm; photo from http://www.zeiss.de/
C12568CF00206298/GraphikTitelIntern/MI_Tanner_2.gif/$File/MI_Tanner_2.gif .
such as modifications to the Endangered Species Act, Clean Water Act,

the National Forest Management Act,
and others are not only likely, they
30
are inevitable, given the changes in

society and politicians that we can
expect over the next 100 years. As
philosophical beliefs of our elected
Climatic Uncertainty and Puerto Rican Parrots

Climatic uncertainty combined with human impacts (poaching, habitat removal) can place species at risk of extinction by natural disturbances
to which they ordinarily would be
well-adapted. For instance, low
populations of Puerto Rican parrots now face risk of extinction
from hurricanes.
Recent hurricanes have had
significant effects on the remaining populations of this species.
Hurricane Georges recently hit
the population when it consisted
of less than 43 individuals. Approximately the same number of
birds existed in 1989, after Hurricane Hugo had reduced the wild
parrot population by half. Should
the frequency or intensity of hurFigure 3.11. Puerto Rican parrot
ricanes change in the future, as
populations have declined to the point
is predicted under some climate
where they remain vulnerable to a wide
change scenarios, the potential
variety of events, including hurricanes.
for recovery of this species may
Photo from USFWS http://biology.usgs.
be significantly affected.
gov/s+t/lrgimage/b016w01.jpg.
and appointed officials wax and wane,

so will the degree to which policies
provide a workable legal framework
for biodiversity decision making.
Values and Assumptions

Biodiversity goals are influenced
by the values placed on species,
values that are highly diverse. Take
the beaver, for example (Figure 3.12).
We know there are many reasons to
manage habitat for beaver, though
some segments of society would like
to reduce their numbers, and some
would like more of them. To complicate matters further, often people
with differing values are neighbors,
and the beavers dont care where the
property line falls! Nor do spotted
owls, white-tailed deer (Odocoileus
virginianus) deer, or coho salmon
(Oncorhynchus kisutch).
Assessments and plans are always
subject to criticism (and frequently
are criticized). Sometimes these disputes result in a determination that
no options are socially acceptable,
and the process would have to start
It is a keystone Its an exotic

species in
pest in my
riparian
country! What a beautiful animal!
ecosystems.
Will it spread giardia?
It would make a
warm hat!
Will it flood
my road?
Its pond could
improve trout
habitat and
improve fishing!
Will it bite me?
Society
Prism of human values
Will it cut down

my apple trees?
Figure 3.12. Society views natural resources through a prism of values (based on discussion with R. Muth).
31
anew. A similarly likely outcome, especially given the number of assumptions on which future projections
are based, is that the affected parties
reach agreement on a preferred option and implement it, but observed
responses are not what were expected. Should these surprises mean that
important societal values are not being sustained, there may yet again be
a reason to start anew. But even if the
system responds as anticipated and
current social values are sustained,
social values, including biodiversity
goals, are not static. Stuff happens.
Societal expectations evolve. Biodiversity goals change. Evolution of
cultural mores, in addition to unanticipated events (9/11, tsunamis,
wildfires, disease), can drastically
alter the perceived values and importance of biodiversity, and assessments
will need to be revisited as new social
issues emerge.
Setting Biodiversity
Goals
Most biodiversity objectives
reflect the paraphrased text of Aldo
Leopold: the first rule of intelligent tinkering is to save all the
pieces. Indeed, the pieces, genes,
species, and processes are exactly
what are often considered within the
coarse-filter/fine-filter approach to
biodiversity conservation. The key
word in this quote is all, and it demands that we address the question,
How much of each? The answer,
obviously, is Enough. Saving all
the pieces is a noble goal and keeps
future generations in the picture. But
at what price will all the pieces be
saved? It is not a price that society is
necessarily willing to pay in all instances: human self-preservation and
preservation of life styles can trump
noble goals very quickly.

Take for instance the recovery of
wild stocks of salmon in the Pacific
Northwest of the U.S. Years of research indicate that some key factors
probably are working together to
cause wild salmon stocks to be at less
than 10% of historic levels. If society truly wants salmon to recover to
historic levels, then they must make
some choices: remove some or all
dams to improve passage, do not pollute wild genetic stocks with hatchery
fish, reduce or eliminate sport and
commercial fishing, restore freshwater conditions to be acceptable
for spawning, and allow all spawning fish to enter the stream and die
to provide stream nutrients. BUT
Remove a source of hydropower?
Increase electricity bills? Use coal or
nuclear fuels for electricity? Do not
allow salmon harvest? Will society
agree to these options? Not likely.
And this is in a wealthy society. Consider the overgrazing situation in the
dry tropical forests of South America
that has led to desertification. Tell the
campesino to stop grazing for a few
years to allow the rangeland to recover (and it would), and he and his
family will starve. Not likely. So setting biodiversity goals must consider
the genetic resources, the species, the
ecological processes, and the goals
and objectives of the people affected
by a decisionand those objectives
will probably keep changing.
Addressing Biodiversity
Problems with DSS
Many goals for achieving biodiversity are prescribed in federal,
state, and occasionally local policies or are reflected in the goals and
actions of NGOs. Specific decisions
regarding what constitutes take
32
(a legal term for harming a species

directly or through habitat modification) of species under the Endangered Species Act (ESA), for example,
often requires site-specific decisions
to address particular issues such as
timber sales. Court battles may ensue
from this debate, resulting in judicial
decisions that have broader implications. Typically, though, more complex decisions are made relative to
ESA when recovery plans are drafted
or when Habitat Conservation Plans
are developed. These large plans not
only require an understanding of the
effects of management actions on the
target species over large multiownership areas, but also must consider
effects on other species and people.
It is in these planning processes that
DSS are often effectively brought to
bear. Indeed, use of population viability models in species recovery has
become commonplace and expected
when developing recovery plans (e.g.,
spotted owls, marbled murrelets, redcockaded woodpeckers). Similarly,
in cases where water quality issues
are addressed to ensure that requirements of the Clean Water Act are
met, models are used to assess potential effects of clean-up activities on
surface and ground water. The U.S.
Forest Service has long used harvestplanning models, wildlife habitat
relationships models, and other tools
to assist with development of forest
plans required under the National
Forest Management Act.
States also address complex planning problems by using DSS. Plans
for the Elliott State Forest in Oregon and the recently acquired state
lands in northern New Hampshire
prompted the Oregon Department
of Forestry and the New Hampshire
Department of Fish and Game,
respectively, to develop harvest plans
by using a DSS that would evaluate habitat requirements for a set of

focal species while also considering
economic effects on local communities and mills.
Nongovernmental organizations
also use DSS to aid in large-scale
planning efforts. Models of ecoregion
structure and composition have been
used in combination with principles
of landscape ecology to identify
areas of potential high priority for
protection or recovery (Poiani et al.
2000). In so doing, these groups are
ensuring, to the degree possible, that
the needs of those species of most
concern to their constituents are met,
though they also recognize the need
to consider many species and social
values as well.
Large-scale ecoregional assessments often are used as the basis for
making informed decisions regarding biodiversity protection among
other social values. Often these
assessments employ DSS to address
complex processes and value sets.
Efforts such as the Northwest Forest Plan, Connecticut Lakes Planning process, Coastal Landscape
Analysis and Modeling Project
(Spies et al. 2002), Columbia Basin
Ecoregional Assessment (Marcot
et al. 1998), the Willamette Basin
Alternative Futures Analysis (Hulse
et al. 2002), and the Sagebrush
Ecoregional Analysis (Knick et al.
2003) all contributed to a foundation or framework within which
local decisions could be made that
could contribute to broader goals
and objectives. Each of these assessments used DSS to varying degrees.
The success in moving management and policy forward to achieve
broader ecological goals within societal constraints also varied widely
among these projects.
Clearly DSS are not needed at all

levels of decision making, nor have
they been effective in all past assessments. The decision to use a DSS
to assist in making decisions is an
interesting dilemma in itself. Simple
decisions do not require a DSS, but
the reliability of a DSS in addressing
a simple problem is probably high.
Complex problems increase the need
for some sort of DSS to organize and
visualize information, but because of
the complexities and uncertainties, the
reliability of the DSS results is probably less. Furthermore, as problems
become more complex, no single DSS
can address all information needs, so it
becomes easier for critics to challenge
the resulting decisiona challenging
Catch-22 for all concerned, humans
and other species alike.
systems. Cambridge University

Press, Cambridge, UK.
Knick S.T., D.S. Dobkin, J.T. Rotenberry, M.A. Schroeder, W.M.
Vander Haegen, and C. Van
Riper III. 2003. Teetering on the
edge or too late? Conservation
and research issues for avifauna
of sagebrush habitats. The Condor 105:611634.
Johnson, K.N., F. Swanson, M. Herring,
and S. Greene. 1999. Bioregional
assessments: Science at the crossroads of management and policy.
Island Press, Washington, DC.
Landres, P., P. Morgan, and F.J. Swanson. 1999. Overview of the use of
natural variability in managing
ecological systems. Ecological Applications 9:12791288.
Marcot, B., L. Croft, J. Lehmkuhl, R.

Naney, C. Niwa, W. Owen, and
References
R. Sandquist. 1998. MacroecolCollins, S.L., 1983. Geographic variaogy, paleoecology, and ecological
tion in habitat structure of the
integrity of terrestrial species and
black-throated green warbler
communities of the interior Co(Dendroica virens). Auk 100:382
lumbia River Basin and portions
389.
of the Klamath and Great Basins.
U.S.D.A. Forest Service General
Davis, L., K.N. Johnson, P. Bettinger,
Technical Report PNW-410, Paand T. Howard. 2001. Forest Mancific Northwest Research Station,
agement, 4th ed. McGraw-Hill,
Portland, OR.
N.Y. 804 p.
Flather, C.H., K.R. Wilson, D.J. Dean,
and W.C. McComb. 1997. Identifying gaps in conservation networks:
Of indicators and uncertainty in
geographic-based analyses. Ecological Applications 7:531542.
Hulse, D., S. Gregory, and J. Baker.
2002. Willamette River Basin
planning atlas: Trajectories of
environmental and ecological
change. Oregon State University
Press, Corvallis.
Hunter, M.L. Jr. (ed.). 1999. Maintaining biodiversity in forest eco-
33
McGarigal, K. 2001. Biodiversity

Filters. CAPS Summary. Unpublished document. University of
Massachusetts-Amherst, Amherst, MA.
Meffe, G.K., L.A. Nielsen, R.L.
Knight, and D.A. Schenborn.
2002. Ecosystem management:
Adaptive, community-based conservation. Island Press, Washington, DC.
Mellen, K., B.G. Marcot, J.L. Ohmann, K.L. Waddell, E.A. Willhite,
B.B. Hostetler, S.A. Livingston,
and C. Ogden. 2002. DecAID: A

decaying wood advisory model
for Oregon and Washington. P.
527533 in Proceedings of the
symposium on the ecology and
management of dead wood in
western forests. November 24,
1999. Reno, NV. Laudenslayer,
W.F., Jr., P.J. Shea, B.E. Valentine,
C.P. Weatherspoon, and T.E. Lisle
(tech. coords.). U.S.D.A. Forest
Service General Technical Report
PSW-181, Pacific Southwest Research Station, Albany, CA.
Poiani, K.A., B.D. Richter, M.G. Anderson, and H.E. Richter. 2000.
Biodiversity conservation at
multiple scales: functional sites,
landscapes, and networks. BioScience 50:133146
Scott, J. M., F. Davis, B. Csuti, R.
Noss, B. Butterfield, C. Groves,
H. Anderson, S. Caicco, F.
DErchia, T. C. Edwards, Jr., J. Ulliman, and G. Wright. 1993. Gap
analysis: a geographic approach
to protection of biological diversity. Wildlife Monograph 123.
Spies, T.A., K.N. Johnson, P. Bettinger, W.C. McComb, J.L.
Ohmann, and G.H. Reeves. 2002.
Challenges to integrating wood
production, biodiversity, and
other socio-economic values at
broad scales: an example from
Coastal Oregon. P. 1326 in
Congruent management of multiple resources: Proceedings from
the wood compatibility initiative
workshop, Johnson, A.C., R.W.
Haynes, and R.A. Monserud
(eds.). U.S.D.A. Forest Service
General Technical Report PNWGTR 563, Pacific Northwest
Research Station, Portland, OR.
34
Spies, T.A., B.C. McComb, R. Kennedy, M. McGrath, K. Olsen, and

R.J. Pabst. 2007. Habitat patterns and trends for focal species
under current land policies in the
Oregon Coast Range. Ecological
Applications (in press).
U.S.D.A. and U.S.D.I. 1994. Record
of decision for amendments to
Forest Service and Bureau of
Land Management planning
documents within the range
of the northern spotted owl.
U.S.D.A.Forest Service, and
USDI Bureau of Land Management, Washington, D.C.
Wimberly, M.C., T.A. Spies, C.J.
Long, and C. Whitlock. 2000.
Simulating historical variability
in the amount of old forests in
the Oregon Coast Range. Conservation Biology 14:167180.
Chapter 4. DSS: How Might They Help? How Might They Hurt?
Norm Johnson and Sean Gordon
he purpose of the next

three chapters is to provide a framework for considering when and how
to apply decision support systems (DSS). The framework
draws on our experience, the existing
literature, and 15 brief case studies
that we introduce here and describe
in more detail in Appendix A. If you
prefer learning through the experiences of others, you may want to flip
to the appendix after reviewing the
introductory material below and read
a few or all of the cases; they are not
too long. In these chapters, we will
focus more on general principles
but try to illustrate them with short
sidebars and references to the cases.
We have organized our discussion to
answer three broad questions:
Which available DSS might meet

your needs? (Chap. 6)
These questions are presented in
the order we believe they most commonly should be addressed, but they
are also interdependent. Concepts
or choices related to one question
will necessarily reflect back on others, so the given order should not
be interpreted as recommending a
strict stepwise progression. A holistic
perspective is also needed.
booklet. We also wanted to ground

our advice in a broader base of
experience, however, so we selected
and researched a variety of recent
examples where DSS have been used
in forest biodiversity decision-making processes.
Selecting the Cases
We used four major criteria to

identify and organize our cases.
First, we limited our search to cases
occurring in the United States, both
to ensure a somewhat similar policy
environment and to limit the scope
Introduction to the Case
of the task. Second, only cases active
Studies
in the past five years were considVery little has been published
ered because computer technologies
about how DSS get used in realchange rapidly and interviewees
world forest and biodiversity probrecollection of events is likely to
lem-solving processes. Most of the
fade with time. Third, we searched
literature on relevant models focuses
for cases in which DSS were used
Why might you want to use a DSS
on their technical specifications, i.e.,
constructively to solve the problems
(and why might you want to
what kind of computational procebeing addressed. Fourth, we sorted
avoid them)? (Chap. 4)
dures they use. To address this gap,
the cases that met our criteria into
How could you successfully use a DSS we assembled a diverse team with
categories by the type of organization
considerable experience to write this
in problem solving? (Chap. 5)
implementing them.
Organizations from
a variety of sectors and
Key References
levels make decisions
Case studies: Short descriptions of our 15 case studies are found in Appendix A.
affecting biodiversity,
including federal, state,
DSS: Acronyms, full names, and sources of further information can be found in Appendix B.
and local governments,
Project website: Our project website is located at http://ncseonline.org/ncssf/dss.
private industry, and individual landowners. Re-
35
sponsibilities at these levels

are generally split between
landowners and regulators.
A short list of applications
to investigate further was
finalized by selecting one
or two cases for each of the
organizational categories
based on accessibility of
information (documentation and contacts) and
geographic diversity. We
selected 15 cases for review,
trying primarily to get a
diversity of organizational
settings. This short list is
presented in Table 4.1.
Estimating their
Social and Analytical
Complexity
Table 4.1 List of case studies.

Case
Decision-making
number
Sector*
Case Name
Type of Decision
1
2
Mixed

Willamette Basin Futures Analysis

Sandy Basin Anchor Habitats
Regional futures assessment

Aquatic habitat restoration priorities
Federal mgt

NW Forest Plan Watershed Condition

Boise-Payette National Forest Plan
FWS red-cockaded woodpecker
FSP Spatial Analysis Project
Regional assessment
National forest management plan
Federal species recovery planning
Landowner assistance planning
7
8
State mgt

Chesapeake Forest
Oregon Harvest & Habitat Model
State forest management plan

State forest management plan
9
10
State reg

WA state water typing model

TNC WA state ecoregional planning
State forest practice rules

Setting priorities for conservation
11
12
Local government mgt Baltimore watershed plan

Summit County (CO) master plan
City watershed management plan

County land use planning
13
14
Industry mgt

International Paper habitat modeling

forest management certification
Industrial forest management planning

Industry self-regulation
15
Nipf mgt
Consulting foresters
Small landowner management plans
3
4
5
6
mgt = land management process; reg = regulatory process; nipf = nonindustrial private forest owner
To provide a general
context for thinking about
these cases relative to one
another, we have ranked them on the
two dimensions charted in Chapter
Table 4.2. Summary of analytical and social complexity measures.
1: social and analytical complexity.
Description
Score
We gave each case a score from 1
(simplest) to 5 (most complex) on
Analytical complexity
each of these dimensions (Table 4.2,
Small geographic and short temporal extent, simple biodiversity and forest measures
1
with further explanation in Appen$ (gradient of increasing complexity in these factors) $
$
dix A).
Large geographic and long temporal extent, complex biodiversity and forest measures
5
The results of this ranking are
Social complexity
presented in Figure 4.1. At the
Single decision maker
1
simplest end of the spectrum, the
Multiple decision makers within one organization
2
Vermont Consulting Foresters case
Decision shared between two organizations (e.g., regulator and regulated)
3
involved only single decision makOpen stakeholder process with no immediate allocation of resources
4
ers (the landowner) and very simple,
Open stakeholder process involving direct allocation of resources
5
qualitative measures of biodiversity
(Management option A will increase
wild turkey habitat) on relatively
small properties without explicit
50 years, including estimating effects ity was associated with low analytical
complexity or vice versa. One excepanalysis over time. At the other
on as many as 30 species.
tion is the Washington Water Typextreme, the Boise-Payette-Sawtooth
What is most striking about
ing case, which attempted to use a
National Forest Plan involved an
these results is the general correlarelatively simple model to estimate
open stakeholder decision-making
tion found between analytical and
fish habitat for setting riparian harprocess over about 6.6 million acres
social complexity. Few cases were
and projected seven alternatives over found in which high social complex- vest rules. Although some observers
36
DSS: How Might They

Help? How Might They
Hurt?
In the remainder of this chapter we
address the Why? and Why not?
aspects of deciding on whether to use
a DSS. To do so, we draw on a decision-aiding technique described by
Benjamin Franklin as moral algebra:
dividing a sheet of paper into two
columns, writing A DSS can help by
over one and A DSS can hurt by over
the other, and filling in our ideas under
the heads to clarify our thinking.
How DSS Can Help

Problem Solving and
Decision Making
We will start out on a positive
note, with a story about how DSS
helped preserve the desert tortoise,
and then we will discuss the potential
helpful aspects of DSS from our experiences and case studies (remember you can refer to Appendix A for
more specifics on each case). Dont
get too comfortable, though. We will
follow this with the potential problems created by DSS use.
4
11
12
Social Complexity
of these modeling processes have

recommended using simpler models
in complex situations, the relatively
simple Water Typing model was not
able to meet the high level of accuracy demanded by this extremely
socially complex situation (setting
regulations with a large private
impact). Perhaps social complexity
drives analytical complexity? Consistent with this hypothesis, we see that
the large majority of cases fell into
the upper right quadrant, perhaps
because a complex problem is needed
to precipitate the effort required to
implement a DSS.
14
13
8
1
2
7
6
3
10
15
Analytical Complexity
Figure 4.1 Analytical and social complexity of potential case studies (numbered
labels correspond to case numbers in Table 4.1).
A DSS can help by
enabling consideration of detailed and complex information and relationships

surfacing nonobvious results
providing a mechanism to
help organize and channel
group thinking
providing a common process
for analysis
providing an institutional
memory for decision analyses
improving the information
base
improving transparency and
credibility with stakeholders/
public/policy makers
37
A DSS can hurt by
committing the organization

to a system beyond its capability
creating a slow, expensive,
cumbersome process
restricting creative thinking
focusing on the wrong problem
deemphasizing unknown or
poorly known aspects of a
system
creating the illusion of certainty and knowledge
creating a modeling priesthood
creating a feeling of powerlessness
Where DSS Helped: Desert Tortoise Species Recovery

Desert tortoise (Gopherus agassizii) populations are declining in the
Western Mojave Desert. Land managers and scientists studying desert
tortoises are developing responses to this decline by recommending policies
and implementing actions to promote species recovery. These responses
must be grounded in science, but there are gaps, both in knowledge and
data, that pose a challenge to specifying the causes of the species decline
and to identifying the most appropriate policies and actions that would assist species recovery.
A process that helps managers and scientists to identify knowledge and
data gaps, both topically and geographically, benefits the tortoise recovery
effort by identifying linkages between scientific research and the information needs of those who formulate policy and carry out actions designed
to bring about species recovery. The Redlands Institute, University of Redlands Desert Tortoise Project
(http://www.redlands.edu/x12596.xml#x12636) is using the Ecosystem Management Decision Support (EMDS; see Appendix B) system to assist scientists and managers in understanding the interrelationships, uncertainty, and relative influence of scientific knowledge on modeling desert tortoise habitat.
EMDS applications involve three components: a logic model, a landscape assessment, and a decision support system. The EMDS logic model for evaluating desert tortoise habitat is a conceptual map of
desert tortoise knowledge and data. It is constructed through a collaborative process, involving a combination of traditional research and knowledge-modeling workshops with experts in the desert tortoise
community. The EMDS landscape assessment uses Geographic Information Systems (GIS) tools to perform
spatial analysis, based on specifications provided by the logic model.
The project specifically addressed the translocation of tortoises affected by the expansion of the Fort
Irwin National Training Center (some 133,000 acres of prime tortoise habitat). The model combined data
on attributes of tortoise habitat, tortoise threats, tortoise population surveys, and important anthropogenic factors (e.g., land ownership, road status, and projected urban growth) that were available through
a variety of sources. This model was used to construct several scenarios of the suitability of lands in the
west Mojave Desert to serve as receiving areas for the translocation of tortoises. Acceptable translocation sites were identified on the basis of multiple criteria, such as conservation history, land ownership
attributes, habitat suitability, level of habitat disturbance, accessibility by the public, and the presence of
major roads. The scenarios were presented to the Translocation Implementation Committee, so that a final
consensus on the most suitable sites for translocation could be achieved (see map of Good Translocation
Areas).
The DSS process allowed the locations of sites to be determined using the best available data and in
a relatively objective manner. The managers (USFWS, BLM, NPS, DOD, CDFG) were especially pleased that
the model provided them with a transparent process, which allowed them to show how and why sites
were identified. Interestingly, the model surprised managers by identifying one area for translocation
that they had not previously been considered as a viable alternative. Ground-truthing showed that the
alternative area was indeed good habitat if fencing were installed along a specific portion of highway. The
fact that the model identified specific priority locations for fencing was an additional value, as this would
reduce costs by eliminating the need to fence the entire highway.
38
Enable Consideration of
Detailed and Complex
Information and
Relationships
Chapter 2 discussed the strengths
of human decision makers, namely
a very flexible and fast capacity for
pattern matching and mental simulation. But, lets face it, unless you have
savant syndrome (think Rain Man),
your capacity to remember and
manipulate large amounts of numbers or other information is quite
limited. Miller (1956) published a
classic review on cognitive limits
entitled The magical number seven
(plus or minus two). Seven was the
approximate number of unrelated
items a person could reliably hold
in short-term memory, and also the
number of categories a person could
deduce from various types of sensory
input. More recently, Klein (1998)
found that even experts construction
of mental simulations was generally
limited to three factors and six transition states.
As can be seen in the cases we
reviewed, many biodiversity questions are often asked at the level of
thousands or millions of acres and
possibly looking many years in the
futuresimply too much information for mental simulation. Computers are an excellent complement to
people in this regard. Computers
can keep track of and process vast
amounts of information quickly, but
they are not as nimble as the human
brain because they must be carefully
programmed by someone to do so.
Surface Nonobvious Results

Complex models are sometimes
criticized because they incorporate
so many interacting uncertainties
how can one claim that they have
any accuracy? We believe the point

of all this calculation is not so much
to enhance accuracy as it is to bring
forward nonobvious results. What
might happen given the interaction
of all these various elementsdevelopments that our mental simulations alone could not conceive of?
One of the leaders of the Willamette
Basin Alternative Futures Analysis
(WBAFA) considered this surfacing
of nonobvious results as a measure of
the projects success:
One of the critical things about
our analyses is that I think that it
showed we didnt know going into
it what the outcome would be in
terms of the future scenarios.
The example given was that nonradical (politically plausible) behavior changes under the Conservation
scenario led to an unexpectedly large
rebound in biodiversity. The case also
emphasized how the modeling process extended and reinforced some
perhaps better known, but often forgotten, aspects of our management,
such as the general concentration of
conservation in uplands and neglect
of lowlands, and the cumulative impacts of frequent exceptions to land
use regulations.
A significant advantage of a
DSS is that patterns and processes
not immediately apparent to the
users emerge from the results. If
these emergent properties also make
sense (are believable), the users have
learned something new that may be
useful to their collective decisions.
Provide a Mechanism to Help

Organize and Channel Group
Thinking (Build Capacity)
One of the decision researchers
discussed in Chapter 2 described
39
group decision making as often chaotic, plagued by the same problems

as individuals face when making
decisions, such as perceptual filters
and limited memory and attention
(Klein 1998). Especially if deliberation is ongoing or arduous, it can
be difficult to keep a group focused
and to bring their expertise together
in some common framework. The
Northwest Forest Plan Watershed Assessment team, as an example, used
a DSS to help organize the input of
technical specialists in seven regions.
The model first provided a way to assess individual indicators (e.g., water
temperaturewhat are good and bad
values?) and then structured these
indicators into an overall evaluation
of watershed condition. Although the
modeling system limited creativity to
some extent, the common structure
it provided was critical to integrating
the input of multiple experts rapidly
and consistently.
Provide a Common Process

for Analysis (help prevent
technical errors)
International Paper also used
group processes, drawing widely
from throughout the organization,
to develop their Forest Patterns
information system. In this case, the
software did not actually provide a
methodology for the groups. Rather,
it captured the groups knowledge in
such a way that it was now accessible
to all their field personnel. It documents what type of management is
to occur where, integrating a wide
variety of factors such as variable
stream-buffer widths and special forest types. Before foresters plan operations, they check the system to make
sure the operations are compatible
with their overall land use plan. The
system provides the organizational
coordination and control needed to

meet regulatory demands and public
expectations. As one of our interviewees put it,
I think its successful if its being
used; if it is preventing environmental mistakes, hazards, its
certainly successful.
Provide an Institutional
Memory for Decision
Analyses
A major problem with many
watershed (and other habitat) assessments based on expert opinion has
been that the assessment methodology was difficult to document. How
exactly did a group of experts arrive
at a particular assessment, and what
knowledge had they used? One or
two years later, when the assessment
was being considered for another
decision, no one might be able to
recall. Using a DSS leaves an explicit
model of how the assessment was
done, and the model can be updated
as new data or knowledge comes into
play. These were primary reasons the
Sandy Basin Anchor Habitats project
decided to use a model in prioritizing watershed restoration options.
They hoped that they could continue
to refine their results by maintaining
and updating their data bases as new
information became available. In
general, this type of dynamic institutional memory resembles adaptive
management in that it allows decisions to be viewed as hypotheses,
which then can be supported, rejected, or refined as new information
becomes available.
Improve the Information

Base
DSS tend to be data hungry,
meaning they require considerable
data about the problem at hand,

such as forest inventories, habitat
classifications, or species-habitat
relationships. DSS can help organize
available information in a decisionrelevant framework and define missing information needs. In many of
the cases we reviewed, the DSS effort
helped prompt and focus the collection of new data. For example, defining which streams were fish habitat
(and so subject to more stringent
forest harvesting regulations) was the
objective of the Washington Water
Typing case. Despite more than five
years of development, the model is
still not being used in the regulatory
process. Building the model prompted the state agencies to improve
their stream information database,
however, and these data are already
in regulatory use and contributing
to an online information system for
landowners. Improvement of the
data base was also seen as a principal
goal in the Willamette Basin Futures
case. The scientific teams spent the
first two years of the project building up the information store, and the
project has made these data accessible through a web portal.
Improve Transparency and

Credibility with Stakeholders/
Public/Policy Makers
Science and technical analysis
tend to have privileged roles in decision making. Especially with public
or business decisions, such objective thinking is expected
in our culture. DSS
and computer models
are examples of this type of
thinking and so may enhance the
credibility of the decision processes
in which they are used. This credibility may be warranted or not (as
we discuss further below), but the
40
focus here is the potential positive

links between DSS, transparency, and
credibility. To many people, the idea
of computer models increasing the
transparency of decision processes
may be hard to believe. And indeed,
we have seen many an eye glaze over
as we have tried to explain our models as they grow ever more complex
in pursuit of accuracy. However, not
all our experiences, or those in the
cases we reviewed, were negative
in this respect. In fact, a surprising
number of the people we interviewed
pointed to the explicitness of modeling as a major contribution to the
consensus-building process. In some
cases, the consensus was restricted to
the scientists and technical specialists
involved, but in others it extended to
more diverse stakeholders.
For example, International
Papers Forest Patterns system was
designed to make their management
intentions and practices explicit in
order to win greater public trust in
their operations. One of the principal goals of the Willamette Futures
Analysis was to build consensus on
the effects of various development
options. Although their modeling
effort was quite complex, a diverse
group of stakeholders became comfortable with it because the project
ceded them significant control and
invested the time and effort needed
for learning.
How DSS can Hinder

Problem Solving/
Decision Making
Our 15 cases in Appendix A
are better for illustrating the positive points; we looked for cases that
had achieved at least some measure
of success. Nevertheless, any choice
is likely to have both positive and
negative consequences, so we are
able to tease a few suggestions from
these largely positive examples. To
help fill out the negative side of the
score sheet, we present you with the
personal experience of one of us in
raising the 900-pound gorilla of forestry DSS (sidebar, right).
Commit the Organization to

a System Beyond its Capacity
If you take the DSS plunge, its
likely you will need at least one person to devote significant time to running the model. Documentation can
be poor and learning curves steep for
much of this software (we give some
relative level of difficulty ratings in
Chapter 6). Learning by doing will
be long and painful. The Sandy Basin
Anchor Habitats was one case that
began without an experienced DSS
user. One of the leaders had tried to
use the EMDS software on an earlier
project but found he didnt have the
time to master it. They succeeded in
the second attempt because they then
had someone with the desire to make
it work, the time, and, perhaps most
importantly, the patience to tinker
and learn the system by trial and
error.
Another aspect of capacity is
technologicalhaving the computer
hardware and software necessary to
implement a DSS. With the widespread diffusion of inexpensive and
powerful personal computers, tech-
Where DSS Hurt: FORPLAN and the National Forests

In early 1977, the U.S.D.A. Forest Service announced that FORPLAN, a
timber harvest scheduling model, would be used as the primary analysis tool (decision support tool) for forest planning as the Forest Service
developed new plans under the National Forest Management Act. Forest
Service leaders wished to have a standard approach to considering alternative management strategies for managing the national forestsan
understandable goal. With this announcement, the Forest Service committed itself to a single decision support system for all 120 administrative
units covering 190 million acres.
Thus began a futile 15-year odyssey of the Forest Service that consumed hundreds of millions of dollars and thousands of person-years of
effort to develop the optimal amount of timber harvest given the many
goals of the National Forests. In the end, very few of the plans developed
by FORPLAN were ever implemented and many, many people became
disenchanted with decision support models for forest planning.
Use of FORPLAN was unsuccessful for at least five reasons, instructive
for any organization that might choose to use a decision support system:
1. FORPLAN was not a finished or widely understood system before the
National Forest staffs attempted to use it. The developers continued
its testing and repair as it was being implementeda fairly reliable
recipe for disaster.
2. Forest planning on all national forests had to use FORPLAN. This
created enormous resentment because the planning teams did not
have the freedom to pick the decision support methods that best fit
their problems.
3. FORPLAN was difficult to explain. Thus, optimal harvest levels were
produced without planners being able to explain why these solutions were best. This created a sense of powerless and frustration
among both Forest Service planning teams and the public.
4. Use of FORPLAN assumed that the central resource-planning problem on the National Forests was to find a sustainable harvest level
within environmental constraints. But, as history shows, it turned out
that the central resource-planning problem was to find scientifically
credible conservation strategies for at-risk species.
5. FORPLAN was imbedded in a technical, rather than a social, planning
process. The heart of planning was viewed as a technical analysis,
rather than a social negotiation.
Timber harvest scheduling models like FORPLAN can play an important supporting role in decisions that affect forest biodiversity. In fact, we
will see some examples in Chapter 6. But, without a doubt, these five problems combined to bring down forest planning dependent on FORPLAN.
We would do well to heed the valuable lessons they offer about using DSS.
41
nological capacity has become much

less limiting than when FORPLAN
was initiated in the early 1980s. A
hidden cost likely to be more important is that a number of the DSS
reviewed also require other software.
The most common requirement is
for a separate geographic information system (GIS), with the ArcView
program (Environmental Systems
Research Institute, Redlands, CA)
being the most popular. ArcView is
moderately expensive, has fairly high
hardware requirements, and also
requires time to masterall of which
must be figured into technical and
technological capacity needs.
Last, but certainly not least,
data represent a very important and
sometimes overlooked aspect of
capacity. In many of the case studies, we found that data preparation
was the most resource-intensive
activity (even when appropriate data
appeared to be available). On the
biophysical side, common data needs
are land-cover maps, forest inventories, and species-habitat relationships. On the social side, you may
need stakeholder values, management options, and well-structured
goals. The question then is Do you
know what data format will best suit
the system youre using? In considering DSS, carefully think through
what data you will need and where
you will get it. Expect problems with
incomplete, biased, or simply incompatible data sets, especially between,
but also within, organizations.
Create a Slow, Expensive,

Cumbersome Process
Even if you have the capacity
to do modeling, it is likely to take
significant time and effort. Time is
needed to implement a DSS, and this
must be reconciled with the decision
environment. In the literature exploring the difficulties of integrating science into policy, time is often cited as
a major disconnect (Dale et al. 2003).
New research typically takes longer
than decision imperatives allow, and
even science assessments chartered
for specific decision needs can face
this problem (Johnson et al. 1999).
The estimated time needed to gather
data and implement a DSS must be
compared against expected decision
deadlines and the general volatility of
the decision-making process. If the
time needed exceeds the deadline, a
DSS approach is unlikely to succeed.
If the process is volatile, time taken
may either have a calming influence
or cause the entire process to derail.
The sidebar below discusses three

examples in which decision needs
changed so radically while the DSS
was under development that the systems were no longer useful.
Restrict Creative Thinking

Although DSS can help bring
forward nonobvious results, they
also may restrict creative thinking.
Each type of modeling comes with
its own assumptions and limitations.
Further, once time and effort has
been sunk into a DSS, it becomes
the default conceptual framework
for understanding how the world
works and how one can use information to make decisions. If the phase
of problem solving you are in would
Rapidly Changing Decision Environments

Complex planning problems have employed DSS to various degrees,
and the long-term effectiveness of the resulting plan is not necessarily
related to the use of DSS. Consider the following projects in which the
authors of this report have been involved: the Northwest Forest Plan
(federal lands in Washington, Oregon, and California), the Connecticut
Lakes project (state and private lands in New Hampshire), and sustainable forest plan development for Trillium Corporation (private land in
Washington state). All used GIS technology to help organize spatial
information. The Northwest Forest Plan (NWFP) relied on relatively quick
and simple decision analyses based largely on expert opinion, rather
than more data and computationally intensive methods. The plan has
been implemented and has persisted for 10 years (a recent assessment
did use a variety of DSSsee Haynes et al. 2006). In contrast, the Connecticut Lakes and Trillium projects involved heavy use of forest growth
models, landscape planning models, and habitat-quality models, but
the DSS results of these analyses were not used to any significant degree
because of their changing decision environments. The landowners lost
interest in pursuing collaborative land management in the Connecticut
Lakes region, effectively killing the effort, while Trillium decided to sell
their lands before their plan could be implemented. The important message is that DSS do not ensure nor increase the likelihood of success of a
project. They can help people make decisions or they can be of no help
at all, depending on the changing priorities of the decision makers.
42
benefit most from brainstorming of

diverse solutions, there are several
group interaction methods you may
want to look into (for example, see
http://en.wikipedia.org/wiki/
Brainstorming).
Focus Effort on the Wrong

Problem
As mentioned in the sidebar (p.
42), FORPLAN addressed the wrong
central resource-planning problem:
it wasnt sustainable harvest levels
that were important to policy-makers and the public; it was credible
conservation strategies to protect
species at risk. The use of FORPLAN
delayed this realization in at least
two ways. First, it was designed by
drawing from earlier models, which
were even more focused on timber
supply. Naturally it inherited their
approaches and worldview. Second,
running the model required considerable effort, which inevitably took
time that otherwise might have been
devoted to further scoping out what
the important issues really were.
Many decision-making processes
are fast-moving and fragmented. The
primary criteria for decision making can literally change overnight. As
mentioned above, make sure your
problem is somewhat stable before
starting a DSS. The Chesapeake Forest Project is one example from our
cases in which the center of attention
changed radically. The DSS analysis
focused on trade-offs between timber production and habitat for the
endangered Delmarva fox squirrel,
but when the process was opened
for public comment, no one was
interested in this. Public access to the
forestlands for hunting game species
quickly came to dominate the discussion, something that the DSS did not
address at all.
Deemphasize Unknown or
Poorly Known Aspects of a
System
With biodiversity conservation decisions, we are constantly
struggling with balancing outcomes
among species or systems about
which we have highly variable information, e.g., we know a lot about
owls and almost nothing about
sharptail snakes. Keep this firmly in
mind: DSS will tend to focus efforts on
the aspects of the problem for which
we have the most data, but these areas
may not actually be the most important for our overall objectives. Expert
opinion and local qualitative knowledge can become much more important than the typically quantitative
data used in a DSS, but there must
be a process for incorporating this
nonquantitative knowledge. Given
the highly disparate knowledge bases
among the species, how do we take
what we know and project likely outcomes of actions on multiple species
in a way that reflects these disparities
in knowledge? Combining coarseand fine-filter approaches that draw
on different types of information
can help, but realize that these filter
approaches are also almost always
fraught with assumptions.
Create the Illusion of

Certainty and Knowledge
Computers are often associated
with precision, and outputs from DSS
typically look certain. Tables, graphs,
and maps all exhibit vast amounts of
orderly information, one value per
cell. But as we commented under the
last point, much of the information
going into a DSS is in fact likely to be
imprecise. Further, these imprecise
data are combined using our inexact
knowledge about their relationships.
43
The classic example from timber

modeling is how optimal harvest
levels predicted by a model get adopted as the expectations. Such levels
are rarely achievable on the ground,
however, due to myriad factors
that were not incorporated into the
models (see the Oregon State Forests
case for an example of this difficulty
and approaches to overcoming it).
Levels of uncertainty often are not
adequately portrayed in DSS results.
For one thing, they are difficult to
knowhow do you calculate them
given all the interactions involved?
Secondly, most DSS are simply not
designed to provide methods to
calculate or output uncertainty, so it
must be done outside the model. The
illusion of certainty can be addressed
by quantifying uncertainty, but this
is too rarely done. Even if done, such
meta-analyses may be difficult to understand and consequently ignored.
A corollary danger is that the
credibility in the whole process
drops like a rock when stakeholders spot what are (or at least seem to
be) obvious errors (e.g., their family farm shows up as a clearcut or
a shopping mall). The Sandy Basin
case faced this problem when a few
stream segments were rated as good
habitat by the model, but some of
the participants knew them to be
dry. The lessons they drew were
to be careful about what you present and to prepare the audience by
explaining model limitations and
uncertainties. In tabular and graphical output, it should not be that
difficult to provide either qualitative
or quantitative uncertainty estimates. It is less clear how to portray
uncertainty in mapped outputs. The
Willamette Futures Analysis faced
this problem. Users of their mapped
futures might naturally assume that
they are intended to be spatially accuratethink of this as map tyrannywhen in fact another model
run under the same assumptions (or
the same model run with different
assumptions) is likely to produce different results. In a follow-up project,
they are addressing uncertainty by
looking at ways to document and
communicate the variability produced by multiple model runs. Likewise, stakeholders growing familiar
with a modeled approach will reliably
start to ask about the assumptions for
any given model-based projection.
Be carefulpeople are getting wise
to models!
Create a Modeling Priesthood

Because of the illusion of certainty and the ability to handle calculations beyond our unaided abilities,
models often can become seen as
the authoritative source of information in a decision-making situation
[Meadows and Robinson (1985)
wrote a book on this subject.] Developing and running analytical models
usually requires specialized expertise
and therefore is restricted to a few
specialists with this knowledge. The
rest of the participants in the problem-solving process may have little
idea of how the model works. Even
the officially responsible decision
makers have to go through these analysts to get their ideas vetted, as was
the case with use of the FORPLAN
model on many National Forests.
This distanced forest supervisors
from the decision-making process.
They had to rely on their modeling
specialists for formulating options
to fit within the model, and supervisors often would have difficulty in
explaining why certain actions were
ranked as preferable. The FORPLAN
priesthood lasted a particularly long
time, probably because of the complexity of the model and the bureaucratic control the agency was able to
exert. We wish to emphasize that the
evolution of such a priesthood is not
always due to the evil plans of analysts, but rather seems to be a natural
tendency in organizational decisionmaking behavior. As George Bernard
Shaw once said, Every profession
is a conspiracy against the layman.
Methods used to counter this tendency include deliberately keeping
models simple and providing training to many in their use (as in the
International Paper case) or putting
representative laymen in charge of
setting goals and regularly reviewing
modeling progress (as did the Willamette Basin Futures Analysis).
Create a Feeling of
Powerlessness and
Discourage Participation
Many of the potential disadvantages described above can generate a feeling of powerlessness and
discourage participation from the
decision stakeholders. An organization might take on a modeling
project beyond their capacity and
end up feeling that they simply cannot solve the problem (instead of
realizing that the approach needs to
change). Participants may feel that
the wrong problem is being addressed, but that they cannot change
the focus because a (DSS) process
has been locked in. Or, even if the
focus is right, they may feel that
their input is not valued because
it does not fit well into the DSS, or
at least it must be filtered through
an unsympathetic priesthood of
analysts. Then, when any process
gets drawn out by time-consuming
analyses, participation is naturally
likely to wane. These changes may
44
not necessarily be seen as detrimental. In the Washington Water Typing

case, the stakeholder policy group
deliberately assigned the task of
defining fish habitat via a model to a
scientific group. It seemed a logical
move and allowed a contentious negotiation process to finish. However,
the modeling took many years and
ran into a variety of value-related
questions. The scientists have had
difficulty in reengaging the policy
group on the issue, and at the same
time the policy group has become
more accepting of the electrofishing alternative, taking away one
of the fundamental reasons for a
model-based solution.
Summary
We have intended this chapter
as a quick synthesis of our experiences with DSS, both good and
bad. You have undoubtedly noticed
some of the similarities in the Help
and Hinder lists, e.g., DSS can help
focus group work, but they can
also lead groups down the wrong
path. Most of the negative issues
can be overcome with some careful
thought and thorough discussion
before embarking on a decisionmaking effort or conflict-resolution
scenario, just as many of the positive contributions can fail without
such considerations.
Perhaps our most compelling
general observation is that if users (that is, people who use the
DSS results, not just the users of
the software) are involved in the
original development of a DSS,
then they can be empowered in the
decision making. Unfortunately,
this is rarely the case. Analytical
approaches, especially when embedded in computer code, are naturally
Pitfalls of Decision Support Systems

Examples from Classic Literature
Hamlet
Shakespeares plays are full of decision makers, and of persons offering good and bad advice. Probably the most indecisive character is
Hamlet, Prince of Denmark. His problem was the mysterious death of his
father, quickly followed by the marriage of his mother to his uncle and the
coronation of the latter. What should young Hamlet do? He used a decision-support system with a supernatural mechanismstate of the art in
mediaeval Scandinaviahe consulted a ghost.
The ghost of Hamlets father gave him an analysis of the situation and
advised him to take his revenge. In implementing this advice, things did
not go well: Suffice it to say that six people die instead of one, yielding an
efficiency ratio of only 17%.
Several reasons can be offered for this unsatisfactory result. Firstly,
Hamlet doesnt fully trust the ghost. The reliability of the analysis and
advice depends on authenticating the ghosts identityis it really his
father, or is it the devil playing tricks? Secondly, maybe Hamlet doesnt ask
the ghost the right question, thereby failing to characterize the problem
correctly. Freudians, for example, argue that he should have asked about
his mother, since that was his primary anxiety.
The Delphic Oracle
The Delphic oracle can be criticized on the grounds of ambiguity and
obscurity. The oracle was sometimes used to provide advice to individual
decision makers, sometimes to arbitrate in disputes. The priestesses
would receive a percentage of a successful armys spoils. Because both
sides would consult the same oracle, accusations of bribery and corruption abounded. King Pleistoanax of Sparta was alleged to have bribed the
Pythian priestess to give specific advice to the Spartan delegation.
Some lessons from these stories:
A good decision-support system justifies its analysis and its advice, so
that the decision maker is not distracted from the primary problem
onto secondary oneshow reliable is the advice? How authentic is
the source of the advice?
A good decision-support system is open and transparentnot surrounded in mystery.
A good decision-support system prompts the decision maker to
consider all aspects of the problemthe quality of the advice does
not depend solely on the decision makers ability to think of the right
questions.
On these criteria, both Hamlets father and the Delphic Oracle can be
regarded as failures.
Used with permission of Richard Veryard, business technology evangelist
http://www.users.globalnet.co.uk/~rxv/kmoi/decisions.htm#hamlet
45
difficult to access and understand.

It takes significant effort and communication skills, which are vastly
different from analytical skills, to
make DSS broadly accessible. The
roles of communication and analysis
tend to get separated when planning
and executing projects. For those
responsible for implementing a DSS,
the focus is usually on the analytical
needs and methods, with communication regarded as necessary only
to transfer results back to the larger
project process. But this is an outdated one-way modelremember
technology transfer?and we all intuitively know that when something
is not understood, it tends not to be
trusted. If the users are empowered
in the development or selection of
a DSS, our points about organizing
group thinking, promoting transparency, and establishing credibility
with the decision-making community can be benefits. Conversely,
a DSS can hinder decision making
if it is imposed on the participants
without taking the time and effort to
facilitate their involvement.
In the next chapter we will talk
about some more specific strategies
that we think can help you use a DSS.
We recommend you also read these
next two chapters as part of your
decision-making process on whether
to use a DSS at all. The choice should
depend on the benefits and burdens (this chapter), the capacity and
methods needed (Chapter 5), and the
systems available (Chapter 6).
References
Dale, V.H., C. Rewerts, W. van Winkle, M.A.Harwell, M. Vasievich,
and S. Hodapp. 2003. Barriers to
the use of ecological models in
decision making. P. 109122 in
Ecological modeling for resource
management, Dale, V.H. (ed.).

Springer, New York.
Haynes, R.W., B.T. Bormann, D.C.
Lee, and J.R. Martin (tech. eds.).
2006. Northwest Forest Planthe
first 10 years (19942003): synthesis of monitoring and research
results. U.S.D.A. For. Serv. General Technical Report PNW-GTR651, Pacific Northwest Research
Station, Portland, OR.
and policy. Island Press, Washington, DC.

Klein, G.A. 1998. Sources of power:
how people make decisions. MIT
Press, Cambridge, MA.
Meadows, D.H., and J.M. Robinson.
1985. The electronic oracle: com-
Johnson, K.N., F. Swanson,

M. Herring, and S. Greene. 1999.
Bioregional assessments: science
at the crossroads of management
46
puter models and social decisions.

Wiley, New York.
Miller, G.A. 1956. The magical
number seven, plus or minus
two: Some limits on our capacity
for processing information. The
Psychological Review 63:8197.
Chapter 5. Tips For Successful DSS Use: Its All About People
Sean Gordon
nce you have clarified why you want

to use a DSS and
which one might best
meet your needs, you
might think it will be fairly straightforward to gather the data, run the
model, and present the results. In
our experience, its not. Undoubtedly
technical problems will be encountered. Often there are barely enough
data of the right sort, DSS can be
difficult to operate, and the results
may not be easy to communicate to
affected stakeholders. Perhaps less
obvious (at least to those of us who
are DSS geeks) are the social aspects,
which can make or break the success of the effort. In this chapter, we
will discuss typical decisions and
difficulties encountered with DSS
implementation and strategies that
we and others have devised to address them. Although this chapter on
DSS use might logically come after
the next chapter on DSS choice, we
have placed it before to emphasize

the value of thinking through use
before (or concurrently) DSS choice.
We have divided our discussion
into three basic concepts: (1) how to
involve people, (2) who will do the
work, and (3) how to do the work.
How to Involve People?

It is a widely shared observation
that the demand for participation in
decisions about forest management
is increasing (Carr et al. 1998). In
this new era of collaborative decision making, it is commonly recommended to (1) identify and include
stakeholders, (2) have numerous
and varied opportunities for input
throughout the decision-making
process, and (3) provide opportunities early in the problem-framing
stage, rather than only after specific
alternatives have been analyzed.
While these are certainly worthy
goals, real-world decision making
is bounded by time and resources,
so we will discuss strategies to get

the most out of participation given
limited resources.
Evaluate the Type of

Participation Needed
Figure 5.1 provides a framework
for thinking about the type of participation needed based on the state of
knowledge and agreement on values
involved in the problem. These four
possibilities are discussed below with
examples from our case studies.
Occasional Review
Should you be so lucky as to be

making a decision for which knowledge is sufficient and there is little
disagreement in values, little participation beyond the decision maker
(or responsible organization) may
be needed. In such cases, Chess et al.
(1998) simply recommend that the
routine decision-making structure
be reviewed periodically by experts.
In the group of cases we reviewed,
the Sustainable Forestry
Initiative (SFI) certification program was the
Key References
closest match to this
situation. Values and necessary knowledge have
been agreed on to form
the evaluation criteria;
it is simply a matter of
47
State of Knowledge
High
Stakeholder deliberation
Occasional review
Low
Integrated deliberation
Scientist deliberation
Low
High
Agreement on Values
Figure 5.1 Recommended types of deliberation (adapted from Chess et al. (1998)).
the certification team applying these

criteria. The criteria are reviewed and
revised periodically by the Sustainable Forestry Board, a diverse group
of 15 conservationists, scientists,
foresters, public officials, and landowners (SFI 2004).
Scientist Deliberation
When agreement on values is
high but knowledge is low, deliberation by scientists is recommended.
The Northwest Forest Plan Watershed Assessment case fits this
model. Watershed condition could be
defined in many ways, but from the
agencies view their legal mandate to
protect endangered fish species was
clearly dominant. Because knowledge was scarce on how to evaluate
fish habitat over this large area, they
gathered experts on fish biology and
hydrology to deliberate and construct the models used.
Stakeholder Deliberation
Stakeholder deliberation turned
out to be the dominant force in the
Chesapeake Forest Plan. A DSS had
been used to analyze timber harvest versus options for endangered
Delmarva fox squirrel habitat for the

initial plan, but when the process was
opened to the public, access to hunting for other species quickly became
the dominant issue. What ensued
was a process that focused on the
various values involved in the hunting debate.
Integrated Deliberation
Integrated deliberation refers
to the need for an ongoing dialog
involving both scientists and stakeholders. A number of the cases we
reviewed used such an integrated
strategy. The Willamette Basin
Futures Analysis had a stakeholder
group meet regularly with scientists
to review and guide the evolving
analysis. The project leader for the
Summit County biodiversity analysis recruited representatives from
several sectors, such as a county
commissioner, a planner, a developer,
a land owner, a wildlife manager,
and an environmental advocate. In
both these cases, the project leaders
selected who they believed to be the
right mix of representative participants, rather than just opening the
process to anyone interested.
48
On the other hand, the Washington Water Typing case emphasized

the separation of stakeholder and
scientist deliberation because of
concern over policymakers tainting
the science. Early on, the stakeholder
policy group handed the process to a
scientific group. This separation also
proved to be an impediment to the
process, however, because the scientists encountered policy-oriented
questions that needed to be resolved
for their work but had difficulty in
engaging the stakeholders to address
them. At the same time, the policy
discussion continued to evolve, but
the implications of these changes for
model development were not discussed in an integrated forum.
DSS Work Best with Small

Groups and Long-term
Participation
Just as there are many DSS available, so are there many possible participation structures. A whole literature on participation exists, including
various handbooks for practitioners
[(see Webler (1997) for a review of
three)]. Approaches are summarized
in Table 5.1. Because DSS typically
take some time to implement and
understand, the typical public hearing is unlikely to be a good forum
for DSS use, where it can appear as
an untrustworthy black-box solution
to participants who have not had
the chance to learn about its inner
workings. This is not to say that DSS
results cannot be presented in public
forum. In the Boise-Payette-Sawtooth National Forest planning process, focus remained on the DSS results and progress was made towards
a solution without getting bogged
down in the modeling details. In
contrast, the FORPLAN system used
in earlier forest plans became itself
the focus of considerable controversy.

There may be an important threshold
to observe in terms of how dominant
the system appears in the process.
Table 5.1 Participation model alternatives.

Public hearings
Public inquiries
Social surveys
Arbitration
Scientific advisory groups
Citizen advisory committees
Citizen panels (planning cells)
Citizen juries
Citizen initiatives
Negotiated rule making
Mediation
Compensation and benefit sharing
Dutch study groups
Sources: Renn et al. (1995); NRC (1996)
Most of the cases we reviewed

employed small groups of participants who met repeatedly over long
periods. International Paper created an internal group with a mix of
expertise; the NW Watershed Assessment worked with expert groups
organized by the different regions involved. The Willamette Futures effort
had the most unique strategy, comprised of multiple levels of participation. Core technical and stakeholder
groups were chosen to represent the
technical expertise and stakeholder
viewpoints. They met monthly. The
effort also tapped into
two larger regional
groups, to which
it gave quarterly
updates and from
which it solicited feedback.
They also engaged the public, as a third
level, through a newspaper insert, a

public forum, and other publications.
Who is Going to Do the

Work?
Addressing this question second
may seem counterintuitive, but we
argue that determining who needs
to be involved should influence who
does the work, rather than vice-versa.
There are two very different types
of needs in an organization: one social, relating to how to involve people
and groups in the decision-making
process, and one analytical, addressing how to run the DSS and analyze
the data. In some cases, it may be
possible to fulfill both needs with one
person or team, but frequently the
different skills required will necessitate integrating different people or
teams. Some of the cases we reviewed
combined these responsibilities, while
others separated them (Table 5.2).
Factors influencing this choice include
the complexity of participation, the
skills and interests of those involved,
and the constraints from the institutional setting. In a number of cases
in which DSS appeared particularly
well integrated with stakeholders (e.g.
Willamette Futures, Summit County,
Baltimore Watershed), a strong decision-support facilitator was involved.
These facilitators knew the inter-organizational dynamics of the situation,
were respected by all parties, and
understood the uses and limitations
of the DSS. Apart from this type of
leadership, there are generally three
options, presented below, for finding
the capacity needed, especially on the
analytical side.
or federal research labs. Engaging

these system designers directly is one
possibility for securing the capacity to
apply a DSS. Such arrangements may
or may not require compensation for
the researchers time, depending on
the fit of the application to the research organizations mission and the
availability of other research funds.
Formal organizational arrangements
may be necessary, which require time
to set up. It should also be recognized that researchers, especially
self-funded ones, will often have their
own objectives, such as trying new,
unproven methods and publicizing
results, which can conflict with those
of the decision makers.
Hire a Consultant
Expertise in building or running a DSS often can be purchased
from commercial businesses. In
some cases, such as the RAMAS and
Woodstock DSS, one firm sells both
the software and consulting services.
In others, the software may be free,
but consultants make a living helping people apply it to their problems.
Engaging such consultants is probably the quickest and surest way to
access DSS expertise. They tend to be
more client-focused than researchers,
but also more expensive. The Oregon
Department of Forestry hired a consulting firm to do the growth modeling and a university professor to design their harvest scheduling model.
At the same time, the department
largely handled the public outreach
component themselves, working with
the timber and environmental interest groups to design and run their
preferred model scenarios.
Engage the DSS Designer
Do It Yourself
Many of the DSS we reviewed

were developed at either universities
A third way to access DSS capacity is to develop it in-house, either
49
through hiring people

with the existing skills or
training current employees. This option requires
more time than the others and is best selected
if there is a continuing
need for the DSS (as
opposed to a one-time
use). Many of the DSS
reviewed have not been
applied widely enough
for a pool of skilled individuals to develop, but a
few have. International
Papers Forest Patterns
DSS is an example in
which the organization
decided to do the work
itself. At the time of inception, they did not see
any specialized systems
that would meet their
needs, and they had the
resources to build their
own on top of a generic
GIS platform. Further, as
a core business function
of a private company, it
merited the effort and
expenditure needed for
internal development.
Our website lists the
estimated number of applications for each system,
which can help gauge the
potential for finding an
experienced operator. For
in-house training, Table
5.2 provides a general
assessment of the level of
effort needed to learn the
different systems, and our
project website provides
more information on the
documentation and training opportunities available for each system.
Table 5.2. Case studies: Who did the work?

1. Willamette Basin Futures Analysis
One of the academic leaders of the project facilitated the
stakeholder group. On the technical side, they engaged over
30 scientists to create and run different models.
2. Sandy Basin Anchor Habitats
One of the representatives from a participating NGO took
it upon himself to learn and apply the software, but he
decided not to use the GIS portion of the program because
it would require additional learning and expense for the GIS
software. GIS work was contracted out to another NGO with
experience in this area.
3. NW Forest Plan Watershed Condition
An aquatic ecologist and a DSS specialist facilitated the expert
meetings. They then worked with a GIS analyst and database
manager to prepare the data, which were fed back to the DSS
specialist, who constructed and ran the models.
4. Boise-Payette National Forest Plan
National forest planning specialists handled the public outreach aspects and helped translate decision makers needs
to a consultant, who ran the forest scheduling model.
5. FWS Red-Cockaded Woodpecker
A major GIS software company (ESRI) contributed staff time
from one of their programmers to get the project started, and
then Fort Bragg (which manages considerable RCW habitat)
paid for finishing the tool using one of their contractors.
6. FSP Spatial Analysis Project
A federal team leader is working with state representatives
to design the system. State forestry staffs are responsible
for implementing it, including information technology and
forestry extension specialists.
7. Chesapeake Forest
The Conservation Fund hired a private consulting firm to
develop a management plan. Modeling was done by a
university professor affiliated with the firm, whose modeling experience came from working for industry. The state
planning department managed the public forums.
8. Oregon Harvest & Habitat Model
The state contracted with a private consulting firm to do the
inventory estimates and run the forest growth model, and
with a university professor to design and run the scheduling
model. State managers and analysts worked with decision
makers and stakeholders to model their preferred scenarios.
9. Washington State Water Typing Model
The Forest Practices Board and the stakeholder policy group
created a science committee composed of scientists from
the various groups (government, industry, environmental).
The science committee in turn formed a water typing sub-
50
group, which did the data analysis and created the model.
Some private contractors were used to help with data
compilation and analysis.
10. The Nature ConservancyWashington State Ecoregional Planning
The Nature Conservancy (TNC) and the Washington Department of Fish and Wildlife provided most of the core staff,
but hundreds of people from a wide array of organizations
participated. Five specialist teams (plants, animals, freshwater, marine, ecosystems) were assembled for each of the
nine ecoregions. A core team composed primarily of the
specialist team leaders oversaw the entire effort. Much of
the technical modeling work was done by TNC staff, who
have developed and applied the methods in other states.
11. Baltimore Watershed Plan
The city contracted with the state forestry department to
create the plan. The lead forester both implemented the
public outreach and worked with the DSS designer and a
GIS specialist to do the modeling.
12. Summit County (Colorado) Master Plan
Research staff from Colorado State University conceived
of and executed the effort. They refined the initial general
ideas using a collaborative design group, consisting of a
variety of people representing the types of professionals
expected to use the planning tool.
13. International Papers Forest Patterns
The system was designed by a small team chosen to
represent the organization both vertically, from managers to operations foresters, and horizontally, across the
four northeastern states. A Forest Service researcher also
participated on the team, and its draft products were
reviewed by a number of academics, agency personnel,
and water and wildlife consultants. Field foresters use the
system directly for their day-to-day management planning; they are assisted by specialized GIS foresters for the
more complex tasks.
14. SFI Forest Management Certification
A number of consulting firms specialize in forest management
certification, and they send a small team of experts to do each
audit. Their teams do not do any modeling, but they review
any modeling work done by the company and may suggest
methods for improving biodiversity-related analyses.
15. Consulting Foresters
The consulting foresters we spoke with ran the DSS themselves
and presented results to the landowners. They did not use the
wildlife module, however, and landowners did not appear to
use the DSS themselves.
How to Do the Work?

In a previous survey of 50 decision support tools, Johnson and
Lachman (2001) wrote, The process
for using the tool is as important as
the tool. We concur. Here we discuss
several themes that emerged from
our case studies.
Define the Problem to be

Addressed by the DSS
What we wish to reemphasize
here is that a DSS will be able to
address only a subset of your problem, and do so only imperfectly. The
issue must necessarily be reduced
and at least somewhat changed if
it is fit into a DSS (even a custombuilt one). Ralph Strauch, formerly
a senior mathematician with Rand
Corporation, argued that any problem definition can only be represented imperfectly in an analytical
framework (Strauch 1975). A danger
is that the model is taken to be a
true surrogate for the problem, with
modeling results adopted without
further thought. Rather, he said,
models are best thought of as perspectives on a problem to be complemented by decision makers more
qualitative understanding of the issues. The Washington Water Typing
participants, for example, originally
intended to use their model as a surrogate for the problem (i.e., identifying fish-bearing streams), and
the associated regulation would be
directly determined by the model.
The model has not proven accurate
enough across diverse geographies,
however, and participants prefer
electrofishing data where available,
so the role of the model is now being revised to assist in establishing
regulations.
Understand That the Problem

Definition Will be Dynamic
and Plan to Communicate
Changes
Take the time to define the
problem well is standard advice in
modeling texts. Again, we concur,
but would add that even carefully
thought-out problem definitions almost always change over time. Events
external to the modeling will drive
much of this change. Some changes
in problem definition also may be
precipitated by the modeling itself.
Ideally, models are useful as learning tools, and as more is learned, the
problem focus is likely to change. Sociologists and policy scholars will tell
you that social preferences are rarely
preexisting conditions, but rather are
socially constructed as people interact during the decision-making process. While preferences and problems
may evolve in many ways, what we
have seen in our 15 cases is that communicating these changes within the
decision-making community is vital.
Again, the Washington Water Typing
case is instructive here: policymakers
originally defined the problem and
handed it off to a scientist group, but
when this group encountered decisions requiring policy input, they
had a difficult time reengaging the
policy group. Further, concern about
the use of electrofishing in the policy
group (a prime reason to develop the
model) declined, but this change and
its implications for the modeling effort were never clearly articulated.
Understand the Different

Demands for Credibility of
Information by Decision
Makers
Scientists and modelers tend
to look for quantitative accuracy
51
thresholds, but decision makers may

not be able to articulate their needs
in this regard. No specific levels were
discussed in the NW Forest Plan
Watershed Assessment or Willamette
Futures cases; it was just assumed
that the best available information
within the bounds of the projects
budgets would be generated. In the
International Paper case, the information had to be current or the users
would lose faith in the system. The
accuracy threshold in all these cases
was not an absolute scientific measure; rather, it was the level of confidence in the system developed by the
participants. The confidence of these
groups was in turn the primary evaluation metric seen by decision makers. Decision makers in the Washington Water Typing case actually set a
quantitative accuracy threshold, but
the primary lessons from the case
show two ways in which credibility
of information strongly relates to
context. First, when the stakes are
high and immediately related to DSS
results, accuracy demands are likely
to be very high as well. Second, the
case showed the relative nature of
quality judgments; stakeholders had
what they felt was a more accurate
measure by electrofishing and so
were unwilling to accept less accuracy, even though it was much more
costly to landowners and potentially
harmful to fish.
Pilot Test on Diverse Data

Data quality is a major determinant of information quality (garbage
in, garbage out). Natural resource
data is by its nature likely to vary
considerably over different geographies and ownerships. Furthermore,
our knowledge of the life histories
and habitat requirements of various
animal and plant species varies wide-
ly from one species to the next. The

NW Forest Plan Watershed Assessment team struggled with the comparability of public and private roads
and streams data, while in the Washington Water Typing case, the data
available were not precise enough to
model the flatter regions. Pilot testing a DSS on samples reflecting the
diversity of the data is recommended.
Our study identified data preparation
as one of the most resource-intensive
aspects of DSS implementation, so
potential users should plan for these
costs upfront (and perhaps double
this estimate, then double it again, as
we have heard recommended!).
Solicit and Incorporate Local

Information
Generating data accurate enough
to satisfy stakeholders usually appears to require their local input.
Participation of local experts/stakeholders was a success factor in three
of the cases. In the International
Paper case, this was accomplished by
having the field foresters help maintain the data directly. Local experts
were engaged in building the NW
Forest Plan Watershed Assessment
and Willamette Futures models. The
exception, the Washington Water
Typing case, had the participation
of stakeholder representatives, but it
lacked local representation in a geographic sense. Stakeholders are now
demanding that local corrections for
stream location and electrofishing
results be integrated into the WWT
fish presence maps.
Be Careful What You

PresentIt is Easy to Lose
Support and Credibility
DSS tend to make information
explicit. Outcomes are not just qualitative estimates of high/medium/low
but specific numbers (e.g., how many

salamanders are predicted for each
scenario). As discussed in the previous chapter, explicitness makes
errors and omissions very visible
(e.g., one dry stream segment got
a high fish habitat score). A few of
our case study interviewees said that
such obvious errors can quickly lead
people to doubt the whole model. All
models are limited and these limitations need to be carefully explained
to users, but one cannot expect that
such explanations will be effective
for all audiences and situations. The
bottom line is to be very careful of
what is shown. Another variant of
this phenomenon occurred in our
Oregon Harvest and Habitat case.
Although initial harvest volume
results were characterized as relative
between the different scenarios (not
absolute in the sense of being achievable on the ground), various interests
have latched onto these numbers and
continue to use them as the expected
benchmarks for policies.
Look for Opportunities for

Visualization, but Dont Lull
People into a False Sense of
Accuracy
Maps were a principal translation
tool in most of the cases. Scientists
and lay persons may start with quite
different conceptions and interpretations of maps, so arranging time for
joint exploration has been recommended (Duncan and Lach 2006a,b).
Maps may be very limited in the
forest information they convey because of their abstracted nature. The
Willamette Futures Analysis experimented with the use of three-dimensional landscape visualizations, but
did not comment on their effectiveness. Research in this area exists to
help gauge its usefulness and guide
52
its application (Meitner et al. 2005).

From our experience we can say
that the level of realism provided by
computer visualizations can enhance
stakeholder understanding, but it can
also increase the false perception of
model accuracy. Try to think through
and explain the limitations of the
visualization software and the data
underlying the simulation.
Utilize DSS to Help Structure

Group Work and Accumulate
Results
Research has shown how erratic
group decision-making processes can
be (Klein 1998).The cases reviewed
in this study showed how DSS can be
used to provide a structure to the decision-making process. At the same
time, the system needs to be flexible
enough to be adapted to participants needs. In the more successful
cases, participants had a feeling of
control over the DSS. The cases also
demonstrated how DSS are used to
keep track of and integrate diverse
sources of information. They serve
as repositories for group decisions.
These repositories of data and models were directly accessible to users
in a few cases (such as field foresters
accessing International Papers Forest
Patterns), but most provided access
through a facilitator or translation
products such as hardcopy maps, or
tables. Even so, the repositories appeared effective if their structure was
transparent enough for the group to
understand in concept.
Summary
Deciding how to use a DSS
involves a host of interrelated questions. We have focused on three: (1)
Who needs to be involved? (2) Who
is going to do the work? and (3) How
The Importance of PeopleForest Planning on a University Forest

In the early 1990s, the dean of Oregon State Universitys College of Forestry recognized that the College
needed a new management plan for its research forest, located just outside the city of Corvallis. Three pressing issues had brought this decision to the fore: (1) neighbors had wrongly assumed they lived next to a
permanently forested park and objected strongly to the Colleges use of patch clearcuts next to their homes,
(2) the closure of federal forests to manipulative experiments in other parts of the state had placed a higher
demand on the College forest for such research opportunities, and (3) numerous highly critical articles were
published in the local paper. It had become essential to create a long-term plan with broad-based input.
Two interested faculty members elected to use soft systems methods (SSM, discussed in Chapter 2) in an
experimental setting to investigate and define the various situations affecting the forest and its future. Their
research included interviews and meetings with College teachers and administrators, forest researchers and
managers, and interested members of the public, in particular recreationists and neighbors of the forest.
After several months of intensive work, this approach yielded common themes of concern across previously contentious groups of affected people. These themes included the lack of a long-term management
plan; underutilization of the forest for teaching, research, and demonstration; lack of faculty support for
management of the forest; and a deteriorating relationship between the College and neighbors and recreational users regarding management of the forest. In addition, the effort developed transformation statements that indicate the central activities that could occur in an improved future state. These statements
focused on the concern over the lack of a long-term plan and reflected the different views of the forests mission: Set A contained improvements in management to accommodate all uses, set B was organized around
research related to urban fringe forests, and set C emphasized the ecological basis of forestry practices. Most
importantly for the work that followed, the soft systems effort developed a comprehensive set of goals that
could guide the planning effort.
The first goal was the provision of teaching, research, and demonstration opportunities. Next, accommodation of recreation use and sensitivity to the interests of neighbors were of significant concern. Finally, the
revenue that would be produced over time had some importance: the forest was entirely self-funding, and
some of the revenue was used to augment teaching and research funds in the College.
One problem that emerged at the conclusion of the SSM process was a disconnect between some
peoples expectations of it and what it actually produced. It was never intended to produce an actual plan
for the forest. It was intended to map the current situation, so that planners could then take into account the
input from divergent and often conflicting groups, which included teachers, researchers, equestrians, hikers,
neighbors, and many others. In this respect, SSM became a decision-support system in its own right, in that
it was a step in the process, not an ending solution.
A faculty committee was then formed by the Dean to construct the forest plan itselfa plan that would
describe the long-term forest condition that would be sought and the activities (especially harvests) that
would occur in the next 10 years, reflective of the goals suggested in the soft systems work and those developed during further conversations between the Dean and the faculty.
The faculty team met for four hours every week for five months as they developed a vision of a forest
that would best meet all the different goals. As part of that effort, they wished to have a quantitative estimate of the types of harvest that would occur and the harvest volume and revenue that would result. To do
that, they utilized a quantitative decision support system and worked back and forth with the modelers and
the Dean to solidify a plan that would best meet all goals.
Once they settled on a plan that seemed reasonable, they then explained it to a broader group of faculty
and staff and took it to leaders in the community for advice and consultation, generating several further
53
adjustments in the plan and its explanation. Upon its public announcement and description, the local press
that had been so hostile became enormously supportive and the plan was implemented and largely followed for the next decade.
As an interesting side note to this story, the quantitative decisions support system used was FORPLAN
the system discussed in the sidebar on p. 41 as a major contributor to the failure of national forest planning.
What was different here? First, FORPLAN was a mature, well-tested system by this time. Second, FORPLAN
was not the focus of planning and the final answer, but rather a mechanism to try out ideas in a back-andforth discussion about the implications of different proposed plansthe ideal use of
any decision support system. Third, it was deeply embedded in a
social process of plan revision.
The university forest planning approach made use of several
different types of decision support systems and, by using them as
components of a longer process, achieved widely acceptable and usable results.
will you go about applying the DSS?

We would like to leave you with two
additional points, which integrate the
themes from these three questions.
First, promote learning about how
the DSS works and make sure that
results are viewed criticallydo not
take results at face value. The broader
set of users, as well as the programmer, should be provided with means
to check the results, both in a quantitative sense and in reality on the
ground. Second, try to ensure that
the users drive the DSS development
and that this role is not ceded to the
scientists, technicians, or, worse yet,
the DSS itself. DSS are tools developed by people, for people. Maintain
the perspective that the technical
experts work for the stakeholders
and decisions makers, not the other
way around. It is important to be
sure that the DSS does not drive the

discussion and decision, but that the
stakeholders make those decisions
using information from the DSS as
well as other sources.
References
Carr, D.S., S.W. Selin, and M.A.
Schuett. 1998. Managing public
forests: Understanding the role of
collaborative planning. Environmental Management 22:767776
Chess, C., T. Dietz, and M. Shannon,
1998. Who should deliberate
when? Human Ecology Review
5(1):4548.
Duncan, S.L., and D.H. Lach. 2006.
GIS technology in natural resource management: Process as
a tool of change. Cartographica
41(3):201215.
54
Duncan, S.L., and D.H. Lach. 2006.

Privileged knowledge and social
change: Effects on different participants of using GIS technology
in natural resource management.
Environmental Management
38:267285.
Johnson, P., and B. Lachman. 2001.
Rapid scan of decision support
system tools for land-use related
decision making. NatureServe,
Arlington, VA.
Klein, G.A. 1998. Sources of power:
how people make decisions. MIT
Press, Cambridge, MA.
Meitner, M.J., R. Gandy, and
S.R.J. Sheppard. 2005. Reviewing the role of visualization
in communicating and understanding forest complexity. Paper Presented at Ninth
International Conference on
Information Visualisation
(IV05), London, England.
NRC (National Research Council).
1996. Understanding risk: informing decisions in a democratic
society. National Academy Press,
Washington, D.C.
Renn, O., T. Webler, and P. Wiedemann. 1995. Fairness and
competence in public participation: evaluating models for

environmental discourse. Kluwer
Academic Press, Dordrecht, The
Netherlands.
SFI 2004. Sustainable forestry
initiative: 20052009 standard. Arlington, VA: Sustainable Forestry Board. Online:
55
http://www.sfiprogram.org/
Last accessed March 1, 2006.
Strauch, R. 1975. Squishy problems
and quantitative methods. Policy
Sciences 6:174184.
Webler, T. 1997. Organizing public
participation: a critical review of
three handbooks. Human Ecology
Review 3(2):245254.
Chapter 6. Shopping for the System that Meets Your Needs

Sean Gordon and Keith Reynolds
n the previous chapters, we

discussed approaches to decision making for the conservation of biodiversity and reasons why you might want to
consider a DSS. If you have decided
that a DSS may be helpful in your
situation, the next obvious question
is, Which one? That is the subject of this chapter. First, we should
clarify what we mean by decision
support system.
What is a DSS?
The phrase decision support
system has multiple meanings.
Specialists in the field of information
systems use it to describe a type of
computer software, but in more general usage it could refer to any system
for supporting decisions (whether
involving computers or not). Our
review in this chapter uses the term
in the former, more technical, sense.
As discussed in the previous chapters
and in the appendix of cases, how-
Table 6.1 Approaches to defining DSS.

The technology
Data management
Analytical models
User interface
The use
Helps evaluate alternative options or scenarios (decision)
Helps deal with complexity (support)
Has a clear, reproducible protocol (system)
ever, a computer-based DSS always

sits within (and is sometimes even
defined by) a larger decision-making
framework.
Within the large field of computer software, DSS have been distinguished as programs that combine
three elements: (1) data management, (2) analytical models, and (3)
a user interface. DSS have also been
distinguished from other types of
software simply by looking at how
they are used, i.e., Is the program
designed to help people make decisions? For this review, we tried to
keep both approaches in mind (as
summarized in Table 6.1), but the
upshot of all this is that there is no

bright line dividing DSS from nonDSS software. Keep an open mind
and be aware that people will define
them in different ways.
Previous DSS Reviews
DSS have proven helpful in a

wide range of fields, including business planning, medical diagnosis,
and transportation (e.g., our air
traffic control system). In forestry,
they have been used extensively for
timber-harvest scheduling, selection of silvicultural treatments, and
insect and disease management,
but their application to
the more diverse and
Key References
complex goals of biodiversity and ecosystem
management has been
somewhat more limited, at least until quite
recently. Nevertheless,
numerous researchers
56
and practitioners have recognized

the considerable potential of DSS to
assist (remember that verb: to assist)
in sustainable management of natural
resources, given the abilities of DSS
to model complex processes and integrate knowledge across disciplines
(IGDSNRE 1998, Rauscher 1999,
Reynolds et al. 2000).
Previous reviews of DSS relevant
to forests and biodiversity have
focused on DSS capabilities to assist
with National Forest plans (Schuster
et al. 1993), ecosystem management
(Mowrer 1997, Rauscher 1999), and
biodiversity in county-level planning
(Johnson and Lachman 2001).
How We Selected and

Reviewed the DSS
We developed an inventory of
available systems from these previous
DSS reviews, our own review of journal articles, and interviews with a
few DSS experts. We found over 100
systems (a full list is on our website)
and screened this list down to the 32
systems that fit in one of these four
categories: (1) systems focusing on
wildlife and biodiversity, (2) systems
focusing on forestry, (3) general-purpose DSS with known applications
to forest biodiversity issues, and (4)
regional assessments that included
forest biodiversity as a significant
component and relied heavily on
computer models.
We constructed a review template consisting of twenty types of
information that we thought would
be useful in choosing a DSS (Table
6.2; the full template is on our website). For the DSS passing our initial
screening, we filled out the template
for each system as completely as possible, based on available information
(primarily system documentation,
past reviews, journal articles, and

websites). As an additional check, we
e-mailed our reviews to the system
designers for their comments, receiving a response from 24 of the 32).
Table 6.2 Review criteria.
1. System name
2. Brief overview
3. Major DSS components
4. System focus
5. General DSS approaches used
6. Types of information used
7. Typical spatial extent of application
8. Typical unit of data input
9. Spatial analysis & display capabilities
10. Abilities to address interdisciplinary, multiscaled, and political issues
11. Support for specific biodiversity issues
12. Support for general classes of biodiversity
indicators
13. Types of decisions supported
14. Types of outputs produced
15. Examples of use
16. Transferability
17. Extent of use
18. Future development plans
19. Developer/distributor contact
20. Other sources of information
57
Making Your Choice

Many of the 20 aspects covered
in the review might be relevant to
your choice of a DSS. Here we will
highlight a few of the categories
that we believe are most generally
applicable. Our website provides a
number of ways to browse and search
these reviews, and you may end up
looking at the full details of a few of
your top prospects.
Finding the Right DSS

Approach for Your Problem
The 32 systems most applicable
to forest biodiversity issues could
quickly be classed in two ways: by
overall focus and by functional
objective (Table 6.3). Some focused
primarily on biodiversity questions;
others focused on forestry issues;
a few were regional assessments
looking at land use patterns; and
the remainder were general purpose
DSS. The systems also could be categorized into a relatively few major
functional areas (Table 6.3).
Two of the biodiversity DSS
focus on population modelingthat
Forest-focused DSS also fall into two

camps.
The first is concerned with models for
Table 6.3 DSS approaches
scheduling activities. These DSS are similar to
Focus
Major Function
DSS Name
Simulation Optimization Evaluation
models for reserve selection, in that they use
Biodiversity Population modeling
PATCH
X
X
optimization techniques to try to find efficient

RAMAS
X
X
patterns. The most common use is for exam
Reserve selection
BMAS
X
ining possible patterns for timber harvest
CAPS
X
ing over time. Efficiency can be judged in

C-Plan
X
multiple ways, including both economic and

MARXAN/SPEXAN
X

Refuge GAP
X
ecological concerns. The second focuses on

ResNet & Surrogacy
X
simulation of tree growth and management.

Sites
X
X
They can simulate tree growth over time

Vista
X
X
(either at the individual tree or aggregated
Forestry
Activity scheduling
Habplan
X

MAGIS
X
stand level) and apply silvicultural treatments

RELM
X
or natural disturbances (such as fire). Because

Spectrum
X
the models for scheduling activities generally

Woodstock
X
X
are used to look into the future, they often rely

Forest growth & FVS
X

management
on growth and treatment options generated

Harvest
X
X
from a simulation model.

LANDIS
X
We also found three more general pur
LANDSUM
X
pose DSS that had been applied to forest bio
LMS
X
X

NED
X
X
diversity issues. They were designed to help

RMLANDS
X
people with evaluation and prioritization.

SIMPPLLE
X
Technically, you might hear this genre de
VDDT / TELSA
X
scribed as multicriteria decision models. They
General
Evaluation &
DEFINITE
X
X

prioritization
provide a general framework for users to rate

EMDS
X
X
different aspects of the problem and combine

Netica
X
these ratings into an overall assessment, such
Regional Forest growth
CLAMS
X
X
X
as the relative condition of a number of waterassessment & management

MRLAM
X
X
X
sheds, the likelihood of species persistence, or

Land use simulation
LUCAS
X
priorities for habitat restoration.

& evaluation
The final category of DSS types we identi
WBAFA/PNW-ERC
X
X
fied concerns regional assessments. The

Restoration prioritization Restore
X
X
X
primary consideration, from a potential users
point of view, is that these systems have not been designed for distribution, but
is, estimating the size of populations
rather are applications customized for use in one region. Several of these efforts
over time, given different assump(CLAMS, LUCAS, WBAFA) involve multiple models linked into an overall
tions about breeding and available
habitat. The majority of the biodiver- conceptual framework. Nevertheless, we believe their methods and tools are informative and could be transferred, with some investment, to other situations.
sity-focused systems address what
is often referred to as the reserveselection problem. Reserve selection
involves finding the most efficient set
of land parcels for conserving biodiversity, based on a landscape in
which each parcel is assigned one or
more biodiversity values (and potentially costs as well, such as cost of
purchasing the land).
Evaluation, Simulation, and Optimization

A more abstract way of considering what type of DSS could answer your
question(s) is to look at their general problem-solving approaches. Textbooks
on DSS describe a variety of approaches, such as expert systems, group-support
systems, executive information systems, agent-based models, and so on (Marakas 1999, Turban and Aronson 2001). Instead of focusing on the modeling
style, our survey classified DSS approaches into three purpose-based categories:
evaluation, simulation, and optimization. Questions such as What is the best
58
parcel to conserve/silvicultural practice/etc? lead naturally to an optimization approach. Simulation systems

are best at addressing What if?
questions (e.g., What if we change
our streamside buffer widths?).
Tools with evaluation functions are
best for answering questions such as
What is the biodiversity impact of
X change in forest cover? The more
general systems (e.g., EMDS) provide
a framework for setting up an evaluation, while the forest and biodiversity-centric tools tend to provide
specific templates (e.g., LMS includes
a number of metrics to evaluate forest
structural stages).
Table 6.3 shows which systems
include which approaches. Many systems include more than one approach,
usually pairing either optimization or
simulation with evaluation. By definition, optimization systems must include an evaluation function to compare alternatives, but systems received
an X in the evaluation column only if
they provide specific functionality to
help users frame an evaluation (e.g.,
LMS provides a number of structural
metrics for describing stand diversity
from the literature). Often, more than
one type of system will be needed
to answer a question. For example,
models for harvest scheduling (e.g.,
Spectrum) require some estimates of
tree growth and management, which
are usually provided by growth models (e.g., FVS).
Looking at our case studies, we
found that their general goals could
be summarized as one of three types:
1. evaluating habitat (current, past,
or future)
2. simulating future habitat
3. finding efficient, integrated harvest and habitat solutions
Table 6.4 groups the studies by
these categories and briefly describes
Table 6.4 Biodiversity conservation goals and DSS used in the case studies
Categories and cases
1. Evaluating current (and past) habitat
FSP Spatial Analysis Project
Use several GIS layers to evaluate the conservation value of properties that
are or might be enrolled in the stewardship program
FWS Red-Cockaded Woodpecker
Help landowners evaluate woodpecker habitat value and consider the
complex habitat needs of overlapping population clusters
International Papers Forest Patterns
Use a land classification system to evaluate amounts of different habitats
at the landscape scale and guide day-to-day management planning
NW Forest Plan Watershed Condition
Evaluate watershed condition for fish habitat at two points in time (past and
present) to gauge the impacts of the NW Forest Plan
Sandy Basin Anchor Habitats
Score the habitat potential of different stream segments, in order to prioritize
restoration activities
SFI Forest Management Certification
Meet the requirements for planning to conserve native biological diversity (ecolog-
ical communities and individual species) in general, and to locate and protect known
sites associated with viable occurrences of imperiled species and communities
Washington State Water Typing Model
Delineate parts of the stream network that are useful fish habitat in order to set
timber harvest regulations
Baltimore Watershed Plan
Analyze risks to the long-term sustainability of reservoir lands, including
maintaining water quality and enhancing forest habitat as a contribution
towards regional biodiversity
2. Simulating future habitat
Summit County (CO) Master Plan
Understand the impact of development regulations on biodiversity by projecting
the landscape to maximum allowed development density
Willamette Basin Futures Analysis
Simulate the effects of three different land use scenarios 50 years into the future
on a wide variety of terrestrial and aquatic species
3. Finding efficient harvest and habitat solutions
Boise-Payette-Sawtooth National Forest Plan
Help select a management alternative by simulating the effects of different
alternatives on harvest levels and habitat for a variety of species over 50 years
Chesapeake Forest
Analyze trade-offs between timber production and Delmarva fox squirrel
habitat over 50 years
Oregon Harvest & Habitat Model
Model the effects of different harvesting alternatives on timber production and
habitat over 150 years in order to set harvest goals and tactical strategies
TNC-WA State Ecoregional Planning
Prioritize all lands in the state in terms of their biodiversity conservation values
59
DSS used
Custom GIS
application
Custom GIS
application
Custom GIS
application
EMDS
EMDS
Custom GIS
application
Custom GIS
application
NED + custom
GIS application
Custom GIS
application
Variety of custom
models + PATCH
FVS, Spectrum +
custom HSIs
Growth simulator
+ Habplan
FVS + custom
activity scheduler
Sites/Marxan
their major objectives. Note that

custom-built DSS (usually based in
a general-purpose GIS program)
were developed in many of the cases.
It would be nice if we could claim
this pattern was largely due to the
lack of an easily accessible guide to
the available DSS. Alas, we think the
problem is more fundamental. As
reflected even in our small sample,
the objectives of biodiversity analyses
are simply too diverse to be amenable
to a relatively small set of standardized tools. Even the decision on how
to characterize biodiversity is highly
variable, given the different possible
human values and highly localized
nature of species-habitat relationships,
as we will discuss in the next section.
This diversity is especially likely if you
follow our earlier recommendations
about taking time to characterize
problems fully and pursuing full and
open social engagement.
How to Characterize
Biodiversity?
As Chapter 3 discussed, you will
face fundamental choices in deciding how to measure biodiversity. You
may already have decided this point,
in which case you will want a system
that can handle your method, or you
may be more open to seeing what the
different systems have to offer. Our
review covered this point by generalizing an internationally adopted set
of nine forest biodiversity indicators
(from the Montreal Process, MPCI
2003) into eight indicator classes (Table 6.5). Indicators 14 tend toward
the category of coarse-filter measures
of forest structure and management,
while 58 are more fine-filter indicators generally tied to specific species.
We found a split between the forest modeling systems, which tend to
focus on indicator classes 13 (forest
Key for Table 6.5

Biodiversity Indicators Supported
User defined: user can define the indicators
Forest type: DSS tracks areas of different forest types
Forest age: tracks forest age classes and/or successional stages
Mgmt class: tracks land by different management classes (e.g., protected areas)
Fragmentation: calculates the degree of fragmentation of habitat type and/or successional
stages
Species diversity: calculates species diversity measures
Species viability: calculates species viability measures
Species distribution: calculates species distribution measures
Species abundance: calculates species abundance measures
Forest Influences Supported
Silviculture: handles basic human forest management activities, such as harvesting and planting.
Land use change: addresses conversions of land use between forest and other nonforest types
Climate change: predicts effects of changing climatic conditions on forests
Biological threats: models impacts of tree pests and pathogens, such as insects and fungi
Fire: models fire behavior and effects
Complexity
Information integration: evaluates interactions between different basic information types
(biophysical, economic, social) beyond the common management-biophysical interactions. For
example, NED helps to set value-based objectives, simulate growth and evaluate habitat, and
calculate financial returns.
Multiple scales: produces coordinated results for decision makers
operating at different spatial scalesCan
the system produce and integrate analyses
occurring simultaneously at several scales?
(For example, the Restore model provides
site-specific restoration recommendations
and a landscape-level analysis of their cumulative impacts).
Social negotiation: facilitates social negotiation by
design (e.g., Vista includes a filters function, which
allows the quick comparison of different assump-
structure and management), and the

wildlife/biodiversity systems, which
focus more on classes 58 (speciesbased measures, Table 6.5). One of
the forest structure indicators (4.
Fragmentation) is rare in the forest
systems, but common in the wildlife
DSS. If your interest is just in coarsefilter measures of how much of different types of land cover are available,
the forest modeling systems are likely
60
to be enough. In order to use both

coarse- and fine-filter indicators, as
recommended in Chapter 3, you will
likely need a combination of these two
types of systems or a forest modeling
system plus habitat suitability indices
(HSIs) for species of high interest.
Unfortunately, HSIs and other wildlife
models are species specific, and there
seems to be no easily accessible repository for their exchange. They tend to
Table 6.5 Comparison of available DSS to forest biodiversity decision-making considerations (see key, p. 60)
Complexity
Inf
or
ma
ti
Mu on in
teg
lti
pl e
rat
sc a
ion
So
l
e
ci a
s
ln
eg
ot
iat
ion
ag
e
Mg
mt
cl a
ss
Fra
gm
en
tat
Sp
ion
ec
ies
div
ers
Sp
ec
it y
ies
via
bil
Sp
it y
ec
ies
dis
tri
Sp
bu
ec
tio
ies
n
ab
un
S il
da
vic
nc
ul t
e
ur
e
La
nd
us
ec
ha
Cli
ng
ma
e
te
ch
a
ng
B io
e
log
ic a
l th
Fir
rea
e
ts
Forest Disturbances
Fo
res
t
typ
Biodiversity Indicators Supported
Fo
res
t
Category System Name

Biodiversity BMAS
CAPS
C-Plan
MARXAN / SPEXAN
PATCH
RAMAS
Refuge GAP
ResNet & Surrogacy
Sites
Vista
Forestry
FVS
Habplan
Harvest
LANDIS
LANDSUM
LMS
MAGIS
NED
RELM
RMLANDS
SIMPPLLE
Spectrum
VDDT / TELSA
Woodstock
General
DEFINITE
EMDS
Netica
RegAssess CLAMS
LUCAS
MRLAM
Restore
WBAFA
Us
er
d
efi
ne
d
Decision-making Needs
Gaps indentfied in
DSS capabilities
L
System includes
specific support
for this function
L Links to another
system with this
function
L
a
a
a
a
circulate in the scientific literature, so

the best bet is usually to find a species
expert to help you find or define such
measures. Roloff et al. (1999) provide
an excellent (but somewhat technical) reference to the area of wildlife
models. They include flow charts and
tables that can help you select a modeling approach.
Some of the regional assessment systems covered a fuller range
of indicator classes, but only one
addressed species distribution and
abundance measures. This system,
the Willamette Basin Futures Analysis, as well as the LANDIS forest
modeling system, stood out in that
they have established explicit links
between modeling systems for forest growth and wildlife population
(PATCH and RAMAS, respectively).
a
a
a
a
a
a
a
a
a No specific
support, but has
been applied
a
a
a
a
a
a
a
a
a
Forest Disturbances
Most questions involve how
changes in the forest affect biodiversity. The Forest Disturbances
category documents the DSS that
address specified change agents. The
biodiversity systems do not simulate such disturbances; instead, they
depend on the user to provide one
or more existing landscapes, which
they then evaluate. From the documentation we were able to review,
it does not appear as if any of these
systems contains built-in relationships for particular disturbances. In
other words, they do not distinguish
between patches of forest lost to fire
versus timber harvesting. In general,
these systems are quite flexible, so
a user could set up such distinc-
61
a
a
a
a
a
a
tions, but they are not ready-made.

CAPS may be an exception, in that it
contains a built-in filter that evaluates the biodiversity effects of varying
levels of development.
Many of the forest modeling
systems specialize in predicting the
impacts of silviculture, fire, and biological threats, but they generally do
not include mechanisms to address
the impacts of these disturbances on
nontree organisms. A significant gap
in the process of assessing biodiversity impacts from these models is
that they tend to focus on the types
and aspects of trees important for
timber production, whereas ideally many other features would be
included in assessing forest habitat.
Some are now integrating some of
these features, such as downed wood
and snags. LANDIS appears to be

the only system with some designed
capacity to model effects of climate
change. Several modeling efforts
are in progress that link climate and
biodiversity (Lawler et al. 2006), but
we have not found any designed to
transfer such capacity to managers.
While many of the systems for regional assessment model the effects of
silviculture and land-use change, none
of these systems included capabilities
to address the influences of wildfire,
biological threats (pest, pathogens,
invasive species), or climate change on
biodiversity. In the future, we expect
that systems for regional assessment
will include climate change and wildfire influences to a significant degree.
Inclusion of invasive species, though,
remains more problematic because
of the difficulty of predicting what
species might strike and their possible
effects.
Complex Decision Processes

All regional assessment systems perform some integration of
biophysical, social, and economic
information, most often predicting the impact of human activities
(social/economic) on forests and
wildlife (biophysical). Forest modeling systems also generally predict
the impacts of management actions
on the forest resource, and a few of
the more generic DSS enable integration of socioeconomic goals or
constraints. The most frequent integration of information performed
by the wildlife/biodiversity systems
is combining land-use costs with
biophysical information in designing
reserve networks.
Virtually all the DSS manage data
at at least two scales: a minimum
modeling unit (stand, polygon, cell)
and the aggregate of such units at
the level of the full analysis (forest,

landscape, etc.). We therefore chose
to distinguish only systems that also
provide disaggregation (landscape
goals to stand level results), or that
try to address the needs of managers at two or more scales simultaneously. Two of the regional assessment
systems clearly produce results for
decision makers at different scales.
MRLAM is designed to produce
results relevant to individual landowners, as well as for the region as
a whole; CLAMS provides results
relevant to both watershed councils
and regional policymakers; and the
Restore model assists with site-level
allocation of restoration alternatives
and watershed-level evaluation of
cumulative restoration impacts. Two
of the forestry systems (RELM and
Woodstock) have specific design
features to help bring strategic management goals (e.g., harvest levels)
down to the operational level by
spatially placing them according to
constraints with GIS. In the generic
DSS category, EMDS and Netica
systems have been frequently used to
nest resource evaluations over two or
more spatial scales.
We found the category of social
negotiation to be the most difficult
to evaluate. Most of the systems have
been used in situations involving
social negotiation (as most decisions
do), but, as with the other categories, we looked for the inclusion of
specific features designed to support
this aspect. Only one effort stood out
in this regard (and that was more
for its overall project design than
for the software features): the Willamette Basin Futures Analysis made
extensive use of stakeholder groups
in setting up model assumptions and
a variety of visualization techniques
in presenting model results. Two
62
other systems, Vista and EMDS, were

marked in this category because they
provide a scenarios feature which
enables users to quickly change
model assumptions and revisualize results. Although not included
in our matrix, since it is no longer
maintained and distributed, the EZIMPACT system (Bonnicksen 1996)
also deserves mention here because
it was specifically designed to integrate individual values in group decision processes.
Decision Context
In Chapter 5 we presented a
framework for thinking about your
decision context based on levels of
knowledge and agreement on values
(Figure 5.1). Because optimization
approaches require more structure,
they tend to be more appropriate in
situations with relatively high levels
of agreement on the problem definition and relevant information. The
What if questions, answerable by
simulation systems, generally need
less agreement on the problem and
its components. They often are used
to project different means for addressing a problem and do not have
to incorporate an evaluation mechanism for explicitly comparing results.
Evaluation systems focus attention
on explaining differences in means
and ends. As mentioned in the
previous chapter, a DSS is not recommended when the problem is very
unstructured; conceptual modeling
and problem structuring methods
are likely to be more useful. Chapter
2 discussed some approaches more
suitable for problem structuring, and
Rosenhead (1989) has edited a volume that introduces six such problem-structuring methodologies.
Another aspect of decision context to consider is the different needs
in different decision stages. Costanza

and Ruth (2001) discuss modeling
trade-offs between the aspects of
realism (degree to which the model
structure reflects real world processes), precision (in predicting outcomes), and generality (the range of
questions it can address). For the initial scoping stage, they recommend
starting with an emphasis on generality and then moving towards realism in specifying problem structure.
Flow charts and systems dynamics
modeling software are recommended. Two such programs are STELLA
and VenSim; they are not included
in our review, but you should be able
to find further information using an
internet search. The second research
stage typically requires sacrificing
generality to move towards realism
and precision. The final management
stage is most concerned with predicting outcomes and so should move
further towards precision.
Capacity and Time

Capacity issues also figure
into the choice of DSS, including
how much time is available for the
analysis compared with the available technical capacity to implement
various approaches. The level of
capacity needs (Table 6.6), as well
as the case studies, provides some
information on the time needed
to implement various approaches.
Special attention should be given to
data needs, because they often are
the most time consuming to meet. In
a survey of water-quality modeling
efforts, Smith-Korfmacher (1998)
found a tendency to choose the most
complex system available because
such systems appear more credible and state-of-the-art, as well as
more difficult to challenge and more
career-enhancing. We agree with
Table 6.6 Relative needs for DSS technical capacity.

Category
DSS
Capacity needs*
Biodiversity

Regional assessment

Forestry

General

BMAS
CAPS
C-Plan
MARXAN
PATCH
RAMAS
RefugeGAP
ResNet
Sites
Vista
CLAMS
LUCAS
MRLAM
Restore
WBAFA
FVS
Harvest
LANDIS
LANDSUM
LMS
MAGIS
NED
RELM
RMLANDS
SIMPPLLE
Spectrum
TELSA
VDDT
Woodstock
DEFINITE
EMDS
Netica
High
(High)
Medium
Medium
(Medium)
(Medium)
Medium
Low
(High)
(High)
(High)
(Medium)
(High)
(Medium)
Low
High
(Medium)
(High)
(Medium)
(High)
(High)
(Medium)
(Medium)
Low
(High)
Low
(Low)
(Low)
*Items not in parentheses are based on answers to

the specific capacity survey question;items in parentheses indicate the authors estimate based on other
system information available.
her assessment that the costs of this

preference tend to be borne later in
more complex data needs, difficulty
in updating, and inaccessibility to
decision makers.
Case Study Examples

To help put all the above information together, let us look at a few
63
of our case studies to see how their

participants chose what type of DSS to
use, as well as the benefits and drawbacks of their approaches. International Papers principal question of
interest was How do we communicate our forest practices to the outside
(regulators, certifiers, and the general
public)? The question did not involve
simulation or optimization; rather, it
focused primarily on documentation
of their practices. Land allocation is
at the root of these practices and is a
spatial issue, so a GIS-based system
was a natural choice. Their own foresters often used maps, and maps also
appeared to be a good tool for public
communication. After meeting regulations for stream buffers and other
special habitats, their major challenge
was to characterize biodiversity in
general at the landscape scale, which
led them to a coarse-filter approach
of tracking habitat types. They did
not see any existing appropriate tools
available at the time, and, as a large
company, they had the capacity to
develop their own.
The Northwest Forest Plan
Watershed Assessment also asked
an evaluation question: What is the
condition of watersheds? Previous
structuring of the problem within
the Northwest Forest Plan gave them
a decision context with agreed-on
ends/values, but without explicit
agreement on a means for evaluation. In this situation, the Chess et al.
(1998) decision framework (Figure
5.1) would recommend scientist
deliberation. EMDS was chosen
because it was a system designed to
support expert judgment that had
proven fairly easy to use in past watershed evaluations.
The objective in the Washington
Water Typing case was also related to
fish habitat, but in a more absolute,
predictive sense: Are fish expected

to be found in stream segment X?
None of the DSS available on our list
are obviously applicable to this type
of question. A statistical evaluation
approach was taken because an external source of information (electrofishing) was available to calibrate and
check model results.
In contrast to the other cases,
the leaders of the Willamette Futures
Analysis posed a What if? type
question: What would be the results
under different policy scenarios?
Similar to International Paper, they
were interested in the intersection of
many factors at the landscape scale,
and they also saw maps as an integrating and easily accessible format.
For these reasons, they also chose a
GIS database as their core system.
They wished to estimate biodiversity effects at both the coarse and
fine scale, so they used a specieshabitat matrix for the former and
the PATCH model for the latter. For
aquatic diversity, they used statistical
models when relationships were sufficiently well understood and expert
judgment models when definitive
data were lacking. Relative to many
efforts, they had a wealth of time and
capacity and so were able to build
several custom models particularly
suited to their needs.
Possible Scenarios and

Recommendations
In the remainder of this chapter,
we present a range of situations and
discuss systems that may be appropriate for each. Web links to systems
mentioned in this section can be
found in Appendix B. Although there
is no simple, natural order to the sequence of situations presented, they
generally progress from the simple to
the complex.
My commercial organization is
considering a variety of alternatives related to improving or
protecting biodiversity on its forest
land, and we are looking for an
easy-to-use software package to
help structure the decision problem, and evaluate a set of alternatives that have been identified.
Numerous commercial, off-theshelf DSS are available for this type

of classic decision problem. They are
generally referred to as multicriteria
decision models (MCDM). The basic
objective of any MCDM is to rank a
set of alternatives by evaluating the
attributes of each alternative against
a set of decision criteria. Typical
applications of an MCDM in the
context of biodiversity might include
ranking species for protection measures based on such factors as their
rarity, or ranking habitat areas for
protection based on such factors
as their exposure to development
pressures.
Although many different
MCDM techniques have evolved
over the past 25 years, one of the
most popular continues to be the
analytic hierarchy process (AHP)
developed by Saaty (1980). Decision models based on the AHP
have proven popular because they
are relatively easy to build, they
64
provide a rigorous basis for decision

making based on well-established
theory and proven technology, and
their results are easy to explain,
making them effective tools for communication. Two MCDM systems in
relatively wide use, and that implement the AHP, are Criterium DecisionPlus (http://www.infoharvest.
com) and ExpertChoice (http://www.
expertchoice.com). For a general
review of MCDM systems, see Mendoza and Martins (2006).
I am a wildlife biologist, and I
need to assess the population viability of a threatened or endangered species. There are concerns
that the existing or proposed
pattern of management is not conducive to the continued long-term
viability of the species.
RAMAS appears to be one of the

most widely used wildlife population
modeling programs. It is specifically
for modeling metapopulations, that
is, multiple interacting groups of
one species. It incorporates spatial
dynamics, such as the configuration of populations, dispersal, and
recolonization among patches, and
similarity of environmental patterns
experienced by the populations.
The program can be used to predict
extinction risks and explore management options such as reserve design,
translocations, and reintroductions
and to assess human impact on fragmented populations.
Another alternative, PATCH,
models wildlife at the level of individuals. It is a spatially explicit
life-history simulator designed to
project populations of territorial
terrestrial vertebrate species through
time. PATCH generates estimates
of wildlife population size, distribution, and trends. The model permits
landscapes to change with time.

Model inputs include species habitat
and area requirements, vital rates,
and movement ability. Model outputs
include tabular measures of population size and maps of population distribution. Future versions of PATCH
are planned to address a wider range
of species life history traits, stressor
scenarios, and species interaction.
I am a consulting forester to small
private forest landowners who
typically manage 1001000 acres.
My clients are often interested in
making some income from sale
of timber, but are also concerned
about good land stewardship and
want to provide suitable habitat
for locally important wildlife species or protect special land cover
types, such as wetlands.
NED is a public domain DSS

developed by U.S.D.A. Forest Service Research. NED is a collection
of software tools created to help
resource managers develop goals,
assess current and future conditions,
and produce management plans for
forests. NED uses an original system
for prescription design to incorporate management goals for multiple
objectives, analyze current forest
conditions, recommend management alternatives, and predict future
conditions under different alternatives. Core outputs include an
evaluation of goals for any or all of
five resources: visual quality, wildlife,
water, wood production, and general
ecological objectives.
The Landscape Management
System (LMS) provides a common
interface to a set of computer applications that support the analysis and
communication of landscape-scale
decisions on forest management.
They use standard inventory information to integrate many analyses
and predict complex changes in

stands and landscapes over time. The
system coordinates the flow of information among existing growth models, computer visualization software,
and analysis tools to allow the user
to simulate the growth of stands and
landscapes and to view the outcomes
using a point-and-click system.
Preferred management scenarios
are developed in LMS by evaluating
multiple projections that can be done
either at the stand or landscape level.
Companion tools allow the user to
develop and compare many alternatives very rapidly.
Both NED and LMS use the Forest Vegetation Simulator (FVS) for
projections of stand growth because
both systems are fundamentally
stand-based. The Forest Vegetation
Simulator (FVS) is an individualtree, distance-independent growthand-yield model, which means you
need a detailed inventory of your
forest to use it. FVS simulates growth
and yield for most major forest tree
species, forest types, and stand conditions in the United States. FVS can
simulate a wide range of silvicultural
treatments. Variants of FVS provide
growth-and-yield models for many
geographic areas of the United States.
Extensions are available to address
the effects of insects, forest pathogens, and fire and fuels issues. FVS
results can also be fed into other DSS
for biodiversity analyses. California
Spotted Owl Wildlife Habitat Relationships and Multistory Elk Hiding
Cover are two biodiversity-related
post-processors available.
I am a member of a watershed
council, and we are interested in
examining the effects of different
levels of harvesting and development on water quality and forest
habitats. We do not have a techni-
65
cal staff, but may or may not be

able to partner with the local university for some assistance. In any
case, we need to understand the
assumptions behind any system
we use.
If you need to simulate forest

growth and harvesting yourselves,
NED or LMS also may be appropriate for you. Both are relatively userfriendly systems, but require some
dedicated time (a couple of full days
at least) and patience to get to the
point of running your own scenarios.
In addition, both require detailed
forest inventories, although estimates
can always be used. Both include
some measures of forest wildlife
habitat (NED for eastern forests and
LMS for western), but such measures
depend highly on your geographic
location, and you may need to devise
your own.
Very few systems in our survey
evaluate water quality, and again
such an evaluation may be quite geographically dependent. Restore was
designed specifically for evaluating
watershed conditions and generating
optimal restoration options. It has
no facilities to predict tree growth
or development patterns, so you
would need to generate GIS maps
of any alternatives you would like to
investigate. You would likely want to
modify the sample evaluation criteria
included. You also may be able to
extract water criteria and ideas from
some of the more complex models,
such as Willamette Futures, CLAMS,
or MRLAM.
If you are not interested in
simulation or optimization and just
want a very basic comparison of
alternatives, you might want to look
at some of the generic MCDM tools
discussed in the first situation above.
EMDS in particular has been used
for a number of watershed assessments.

I represent a conservation organization whose primary mission
is the conservation of species and
their habitats. We are concerned
with the gradual loss of wildlife
habitat in our region and would
like to propose an overall conservation framework, including
which areas are the highest priorities for preservation.
A number of systems have been

developed to address what is often called the reserve selection
problem (Table 6.3), which involves
prioritizing possible conservation
lands based on a number of criteria,
such as costs and biodiversity values.
SITES has probably been the most
widely applied, since it has been
adopted by the Nature Conservancy
for use in their ecoregional planning
efforts. SITES integrates the SPEXAN/MARXAN program (which does
the prioritization/optimization) into
the ArcView geographic information
system, enabling users to specify spatial criteria in the process and easily
display results in maps. Vista is a successor to SITES, and it operates as an
extension to the newer ArcMap GIS
software. The other reserve selection
programs have somewhat different
ways of selecting and prioritizing;
you can research these features on
our website and those of the program
developers (see Appendix B).
I am a manager for a state (or
federal) natural-resource agency.
The agency is responsible for a
large forestland base and has to
accommodate multiple uses and
values, although conservation of
biodiversity is often an important
consideration in our land management decisions. The agency has
significant technical resources and
computing capabilities. As a public

agency, effective communication
with the public about our decision processes is a very important
consideration.
State and federal forest management agencies are generally expected

to manage their forests for maximum
value (greatest permanent value is
the legal phrasing in Oregon, for example). Traditionally, this value has
been interpreted as maximizing timber revenues within some constraints
for environmental, recreational, and
cultural resources. The state and federal forest planning examples in our
15 cases all used forest growth simulators with some form of optimization software for this purpose (Table
6.4, category 3). FVS is probably
the most widely used forest growth
simulator; it has variants for many
different regions of the U.S., and is
available free from the Forest Service.
It doesnt cover all areas, however,
and many other growth models are
available. [Richie (1999) has done a
review for the Pacific Coast.] Forest growth is highly variable, so any
simulator can benefit greatly from inputs of local growth data. The growth
and management software is used to
provide growth and timber yields for
a number of different management
options. For maximizing values, you
need an activity-scheduling program
(Table 6.3).
The options vary quite a bit,
from free (Spectrum) to relatively
expensive (Woodstock). From our
anecdotal information, Woodstock
appears to be the most popular in
the forest industry, and Spectrum
(a successor to FORPLAN) is the
most widely used by federal managers. You may need a combination of
tools; for example, the Boise-PayetteSawtooth National Forest Plan used
66
Spectrum for their strategic plan (not

considering the geographic location
of resources) and RELM to test the
feasibility of meeting these targets in
test areas, given more on-the-ground
spatial constraints. Woodstock
includes similar strategic and tactical planning tools in its family. The
Oregon and Maryland examples used
planning heuristics, which attempt
both a strategic and tactical plan at
the same time. Taking into account
spatial considerations boosts the
number of possible solutions greatly,
so these heuristics are not able to test
all options and guarantee an optimal
solution. If designed well, however,
they typically come quite close.
I am part of a stakeholder group
tasked with considering an ecosystem management approach to
the forests (public and private) in
our region. Of special concern are
fire and the effects of alternative
mitigation strategies.
Fire can be quite difficult to

model, given the somewhat random
nature of fire starts and the spatial
nature of its spread. Several forestry
DSS include capabilities for modeling the effects of fire on forests (Table
6.5). FVS has a fire and fuels module,
which can simulate effects of fire, as
well as potential fire risks for all trees
in individual stands; however, it does
not simulate the spatial spread of fire.
VDDT is often used to simulate fire
in a similar but even simpler way: the
user inputs transition probabilities,
in this case the probability of fire, for
different vegetation types. Within
each period this percentage will burn
and make the transition to a new
vegetation class. The companion program TELSA can locate these burns
spatially on a landscape. We would
recommend your technician(s) read
an article by Barrett (2001), which

compares four of these landscape
change models in some detail.
Woodstock and RMLANDS also
have capabilities for simulating fire,
but are more complex programs requiring more technical resources.
References
Barrett, T.M. 2001. Models of vegetation change for landscape planning: a comparison of FETM,
LANDSUM, SIMPPLLE, and
VDDT. U.S.D.A. For. Serv. General Technical Report RMRS-76WWW, Rocky Mountain Forest
and Range Experiment Station,
Ogden, UT. Online: http://www.
fs.fed.us/rm/pubs/rmrs_gtr76.pdf
Bonnicksen, T.M. 1996. Reaching
consensus on environmental
issues: the use of throwaway
computer models. Politics and the
Life Sciences 15(1):2334.
Chess, C., T. Dietz, and M. Shannon,
1998. Who should deliberate
when? Human Ecology Review
5(1):4548.
Costanza, R., and M. Ruth. 2001.
Modeling for scoping, research
and management. P. 169178 in
Institutions, ecosystems, and sustainability, Costanza, R., B.S. Low,
E. Ostrom, and J. Wilson (eds.).
CRC Press, Boca Raton, FL.
IGDSNRE 1998. A strategy for the
development and application of
decision support systems for natural resources and the environment.
Interagency Group on Decision
Support for Natural Resources
and the Environment. Online:
http://www.i4sd.org/strat.html.
Last accessed August 4, 2003.
Johnson, P., and B. Lachman. 2001.
Rapid scan of decision support
system tools for land-use related

decision making. NatureServe,
Arlington, VA. Online:
http://ncseonline.org/NCSSF/
DSS/Documents/docs/Johnson2001.pdf
Lawler, J.J., D. White, R.P. Neilson,
and A.R. Blaustein. 2006. Predicting climate-induced rangeshifts: model differences and
model reliability. Global Change
Biology 12:15681584.
Marakas, G.M. 1999. Decision support systems in the 21st century.
Prentice Hall, Upper Saddle
River, NJ. 611 p.
Mendoza, G.A., and H. Martins.
2006. Multi-criteria decision
analysis in natural resource
management: a critical review
of methods and new modelling
paradigms. Forest Ecology and
Management 230:122.
MPCI. 2003) Montreal Process
Criteria and Indicators. Online:
http://www.mpci.org/criteria_
e.html. Last accessed November
12, 2006.
Mowrer, H.T. 1997. Decision support
systems for ecosystem management: an evaluation of existing systems. U.S.D.A. For. Serv.
General Technical Report RM296, Rocky Mountain Forest and
Range Experiment Station. Fort
Collins, CO.
Rauscher, H.M. 1999. Ecosystem
management decision support
for federal forests in the United
States: a review. Forest Ecology
and Management 114:173197.
On-line: www.srs.fs.fed.us/pubs/
viewpub.jsp?index=755.
Reynolds, K., J. Bjork, R.R. Hershey,
D. Schmoldt, J. Payne, S. King, L.
DeCola, M. Twery, and P. Cunningham. 2000. Decision Support for Ecosystem Management.
67
P. 687721 in Ecological Stewardship: A Common Reference for

Ecosystem Management, Johnson,
N.C., A.J. Malk, W.T. Sexton, and
R.C. Szaro. Elsevier Science Ltd.,
Oxford, U.K.
Ritchie, Martin W. 1999. A compendium of forest growth and yield
simulators for the Pacific Coast
states. U.S.D.A. For. Serv. General
Technical Report PSW-174, Pacific Southwest Research Station,
Albany, CA.
Roloff, G.J., G.F.Wilhere, T. Quinn,
and S. Kohlmann. 2001. An overview of models and their role in
wildlife management. P 512536
in Habitat relationships in Oregon
and Washington, Johnson, D.H.,
and T.A. ONeil (eds.). Oregon
State University Press, Corvallis.
Rosenhead, J. 1989. Rational analysis
for a problematic world: problem
structuring methods for complexity, uncertainty, and conflict.
Wiley, New York, Chichester, U.K.
Saaty, T.L. 1980. The analytic hierarchy process : planning, priority
setting, resource allocation. McGraw-Hill, New York, NY.
Schuster, E.G., L.A. Leefers, and J.E.
Thompson. 1993. A guide to
computer-based analytical tools
for implementing national forest
plans. U.S.D.A. For. Serv. General Technical Report INT-296,
Intermountain Research Station,
Ogden, UT.
Smith-Korfmacher, K. 1998. Water
quality modeling for environmental management: lessons
from the policy sciences. Policy
Sciences 31(1):3554.
Turban, E., and J.E. Aronson. 2001.
Decision support systems and
intelligent systems. Prentice Hall,
Upper Saddle River, NJ.
Appendix A. Case Study Briefs

Sean Gordon
Multisectoral applications

Willamette Basin Alternative Futures Analysis

Sandy River Basin Anchor Habitats Project
Forestry Support ProgramSpatial Analysis Project
Washington StateTNC Ecoregional Planning Model
Federal

Northwest Forest Plan Aquatic & Riparian Monitoring Program

Red-cockaded Woodpecker Foraging Matrix Application
State

Chesapeake Forest Project

Oregons State Forest Management Planning
Local

Summit County Lower Blue Subbasin Master Plan

Baltimore Reservoirs Forest Conservation Plan
Private

International Papers Forest Patterns System

SFI Certification
Vermont Consulting Foresters
68
Multisectoral Applications
Time frame: 19962001
DSS used: PATCH + variety of custom models
W
Description
BAFA was designed to help diverse stakeholders understand

the ecological consequences of possible societal decisions
related to changes in human populations and ecosystems in
the Pacific Northwest and to develop transferable tools to
support management of ecosystems at multiple spatial scales.
It has been used to simulate the effects of three possible development scenarios
on eight regional measures of biodiversity over the next 50 years. The process
included a four-tiered stakeholder involvement and outreach plan and multiple
biodiversity modeling efforts. Elements of ecosystem management it addressed
included cutting across ownerships and integrating biophysical and socio-economic information.
Analytical Complexity: High (5)

The WBAFA case involved high analytical complexity in terms of space, time,
forest influences, and biodiversity measures. The overall geographic scope was large
(~30,000 km2) and was modeled in considerable detail (30-m2 cells reflecting 65
land use/cover classes). Natural vegetation change and four other influences were
modeled (urbanization, rural residential, agriculture, forestry) for 10-year intervals
over 60 years. Impacts of these land cover changes on the four core resources were
estimated for each period, using a wide variety of models; for example, seven models
were used for impacts on aquatic life. The principal strategies to limit the analytical
complexity of the cases forest biodiversity were as follows: using explicit, bounded
scenarios, as opposed to an open-ended analysis of trends; evaluating only three
scenarios; and not modeling selective timber harvesting (thinning & uneven age
management).
Social Complexity: Medium-High (4)
Key Points
The project engaged stakeholders in constructing
the analysis, which helped
create buy-in/trust of the
model results.
A facilitated working group
was used as a forum for
translating between experts
and stakeholders.
The project did not engage
stakeholders in defining
overall framework, so it
missed key concerns such
as economic outcome measures.
Project leaders developed a
multi-level communication
strategy to reach different
audiences.
Project leaders took advantage of other deliberative
forums at the same scale.
The project did not directly
involve a political allocation
of resources, so it was less
controversial than some
other cases included here (it
was aimed at deliberation
support rather than decision support).
Social complexity was assessed as medium-high because the citizen group

directing the modeling involved stakeholders representing diverse interests.
Additionally, two other publicly appointed groups regularly reviewed the
modeling process. Its social complexity, however, was a step below a typical
public resource decision process. WBAFAs participation was not as wide-open, and
the project was not making any direct decisions about resource use or regulation. Two
strategies that limited social complexity were the selection of citizen/modeling participants on their willingness to work towards consensus and the separation of scenario
development by citizens from estimating impacts by scientists.
More Information
http://www.orst.edu/Dept/pnw-erc/ (Last accessed June 15, 2006)
http://willametteexplorer.info/ (Last accessed June 15, 2006)
69
Sandy River Basin Anchor Habitats Project

DSS used: EMDS
Description
The convening purpose of this project was to bring all the entities (federal, state,
local governments, watershed council, NGOs) together and develop a basin-wide
watershed restoration strategy for the Sandy River Basin in northwest Oregon. The
process was structured to focus on aquatic habitat and produce a collaborative stakeholder vision across all ownerships. This first phase of the project identified anchor
habitats. These are distinct stream and river reaches that harbor specific life-history
stages of four species of salmon and steelhead to a greater extent than the river system at large, are critical for the creation and maintenance of high quality habitat, or
both. Three data sources were used: empirical data from
existing stream surveys, habitat modeling data generated
Key Points
by the Ecosystem Diagnosis and Treatment model, and
professional judgment from three local experts. Anchor
The DSS helped to combine different sources of
habitat stream segments were identified for the four
information.
species, and these priority areas can now be used to help
The DSS helped to structure an explicit approach
guide habitat restoration planning activities.
to evaluation.
Analytical Complexity: Medium (3)
The model enables future efforts to rerun analysis

with new data or understandings.
The analysis looked at a medium-sized area (325,000

acres) at only one time (the present). They simplified
the task by excluding fish harvest, fish hatcheries, and
hydroelectric influences. Looking at four species using
three measures of habitat quality made the treatment of
biodiversity measures moderately complex.
The DSS required some time and effort to learn,

and so depended on interest and dedication of
one of the individuals involved. (Without this
interest, they probably would have used a simpler
spreadsheet-based scoring procedure.)
Making evaluation procedures and results explicit
also made errors and omissions very visible (e.g.,
one dry reach got a high score); each error made
people doubt the whole model, so careful consideration needs to be given to what is shown.
Social Complexity: Medium-High (4)
Since all parties with an interest in the subject

were invited to participate, the project resembled an
open-stakeholder process. Setting restoration priorities
did not, however, directly allocate resources or drive
regulations. Excluding fish harvest, fish hatcheries, and
hydroelectric influences also simplified the problem socially. Otherwise stakeholders at all levels were invited to participate, and no particular social constraints were
noted.
More information
http://www.oregontrout.org/images/8success/Sandy%20Habitat%20Report.pdf
(Last accessed July 3, 2006)
70
Forestry Support ProgramSpatial Analysis Project

Time frame: 2001present
DSS used: Standard GIS systems plus a new web-based database system
Description
The U.S.D.A. Forest Service Forestry Support Program (FSP) provides technical assistance, through State forestry agency partners, to nonindustrial private forest owners to
encourage and enable active long-term forest management. A primary focus of the program
is the development of comprehensive, multi-resource management plans that provide landowners with the information they need to manage their forests for a variety
of products and services. Under pressure from the Office of Management
Key Points
and Budget to better demonstrate program effectiveness, the FSP has been
Involvement of local (i.e., state) repredeveloping the Spatial Analysis Program to track and summarize informasentatives seems desirable; in the pilot
tion about properties enrolled in the program. It is providing an online
phase it resulted in a broad buy-in
interface that helps create stewardship plans that qualify for the program
from the states.
and stores the information in a central database. It provides a basic set of
Flexibility given to state analytical
GIS data which can be used to evaluate impacts of (and possibly prioritize)
methods has created consistency
stewardship activities. States can add their own data layers and weighting
problems and made quality control/
systems.
oversight more difficult.
Information privacy is a big issue
because the system deals with private
Although the forest and biodiversity measures used are likely to be relalands data; arranging appropriate levtively simple, the overall analytical complexity was rated medium because
els of access for different hierarchical
of the very large geographic scope and likely state-to-state variability of the
levels is a major task. (Records must
analyses. The spatial extent of the DSS is very large, incorporating the 354
be anonymous at higher levels.)
million acres of the U.S. estimated to be under nonindustrial private owner- Although not originally intended to
ship, and spatial resolution needs to be quite fine to pick out properties
prioritize assistance (a somewhat sensidown to 10 acres (or less in some states). The actual analysis area is reduced,
tive issueFSP support has been on a
however, because individual states are responsible for determining their
first-come first-served basis), there is
priority areas and only a small portion of private lands are actually involved
concern that state and federal pressures
for efficiency may drive it that way.
in the program. Twelve basic data layers are proposed nationally (including threatened and endangered species) and more can be added by states, if
they wish.
Social Complexity: Medium-low (2.5)

Social complexity is rated medium-low because the interactions are primarily decisions shared between two organizations: between the state and the Forest Service for
the priority analyses and between the state and the landowner for stewardship plans.
Stewardship plans involve incentives, rather than regulations, so they are not as controversial. The state prioritization processes may become more controversial, as they
will influence the allocation of resources locally (and possibly, eventually, nationally).
More Information
http://www.fs.fed.us/na/sap/
(last accessed June 20, 2006)
71
Washington StateThe Nature Conservancy Ecoregional Planning Model

DSS used: SITES
Description
To guide biodiversity conservation and land use planning across Washington
State, the Washington Departments of Fish and Wildlife (WDFW) and Natural Resources (WDNR) joined with The Nature Conservancy (TNC) in a partnership to do
an ecoregional assessment for each of Washingtons nine ecoregions. Each assessment
attempts to identify and prioritize places for the conservation of all biodiversity in an
ecoregion. The relative priorities are based on such factors as species rarity, species
richness, species representation, site suitability, and overall efficiency. Statistical models for suitability are typically not available, so therefore much of the index is based on
expert opinion. Expert opinion was incorporated by using an abbreviated version of
the analytic hierarchy process. The analysis utilizes an optimization program known
as Sites to find the most efficient set of conservation units.
Key Points
Analytical Complexity: Medium (3.5)
Coarse filter (habitat types) and fine filter

(individual species occurrences) approaches were used to best characterize
biodiversity.
The characterization of biodiversity was quite complex,

with over 800 kinds of biodiversity elements considered, and
the spatial complexity was also medium-high, comprising over
8000 polygons. On the other hand, the analysis only focused
on the current situation (no temporal projection), so dynamic
forest influences were not considered. These high and low complexity aspects led to an overall rating of medium.
There is little statistically rigorous data at

the ecoregional scale, so assessments tend
to rely on expert opinion.
Explicit modeling forced the partners to
come to consensus about evaluating conservation priorities.
The NGO partner was able to set a specific
conservation goal, but the state agencies
did not feel they had the policy authority
to do so.
Social Complexity: Medium (3)

Although the assessments are a public, open stakeholder
process, they do not have an immediate effect on regulations or
the allocation of resources. Participants described the process
as a primarily technical one, although involving multiple stakeholders.
More Information
http://www.ecotrust.org/placematters/assessment.html (Last
accessed July 5, 2006)
http://conserveonline.org/workspaces/cbdgateway/era (Last accessed July 5, 2006)
72
Federal
Northwest Forest Plan Aquatic & Riparian Monitoring Program
DSS used: EMDS
Description
he Aquatic and Riparian Effectiveness Monitoring Program (AREMP) is
charged with monitoring and assessing the condition of aquatic ecosystems
and their riparian and upslope influences in the Northwest Forest Plan
area. The NW Forest Plan is a common management strategy that was
implemented in 1994 across all federal lands in the range of the northern
spotted owl, approximately 24 million acres in western Washington, Oregon, and
northern California. AREMP uses a randomized sampling strategy to sample instream, riparian, and upslope indicators on a total of 250
watersheds (about 10% of the total) every five years.
Key Points
Assessment of watershed condition is a complex task
involving considerable social and scientific uncertainty,
Regulatory agencies were not initially invited to
such as what attributes of the watershed are important
participate (to avoid typically contentious ESA
and what are the influences of the myriad of watershed
consultation procedures).
processes on one another? Many of the past watershed
Although the Northwest Forest Plan has been
assessments had been done using expert teams, who were
highly contentious, this analysis was not, most
able to use their best judgment to make assumptions
likely because it was not directly tied to any alabout these complexities and uncertainties. A group of
location of resources.
experts would be confined to a room with maps and data
and not allowed out until they had colored all the water A relatively simple evaluation model (roads &
sheds good, fair, or poor. The main problems with this
vegetation) was used for a complex concept.
approach were that such decisions were difficult to understand or repeat. A year (or week) down the line, the forest
Neither scientists nor stakeholders complained
supervisor might not be able to explain why a particular
about simplicity, and both supported the attempt
area was rated poor, or a different group of experts
at making the evaluation explicit and quantitamight well come up with a different set of ratings.
tive.
The science team that designed AREMP understood
Results were presented to the Regional Interthese problems and recommended the use of a decision
agency Executive Committee, but there are no
support system called Ecosystem Management Decision
clear mechanisms for how results are expected to
Support (or EMDS for short). EMDS could be used to
influence decision making.
capture the experts assessment criteria, so they would be
documented and consistently applied. Other reasons for
the DSS choice were that it was developed by the Forest
Service and had been tested for watershed evaluations before.
The AREMP team drafted an initial model internally to become familiar with the
process. This initial model did not distinguish the biophysical differences between
different areas of the plan (e.g., water temperature might naturally be higher in some
areas than in others), nor did it capture the range of expertise available. To address
these shortcomings, the AREMP team divided the Plan area into seven biophysical
provinces and held a series of workshops to develop a model for each. A total of 36
experts from the Forest Service and BLM attended the 2-day workshops, and an ad-
73
ditional 41 provided some input but did not attend. For the most part, attendees did
not appear to have a difficult time understanding the modeling process concepts, after
a short Powerpoint presentation on them.
Construction and testing of the different models were relatively rapid (a few days
each), but assembling the data to run through the models took the team most of the
next year. Final formatting of the data to fit the models also took as much effort as
building the models themselves. These runs were presented in a second round of
workshops intended to verify whether the models actually worked as intended. In
most cases, at least a few changes were needed to bring the models into line with the
experts knowledge.
The AREMP team published the model results as part of the 10-year evaluation of
the Northwest Forest Plan. Next they will need to decide when and how the models
should next be updated.

The analytical complexity of the AREMP modeling process was judged to be
medium. The overall geographic scope was large (~24 million acres), but the analysis
was restricted to a sampling of 250 watersheds within the broader region. Only two
past periods were considered, so the model did not require any simulation dynamics.
Many different vegetation, roads, and in-stream indicators were used, and the standreplacing habitat influences of harvesting and fire were integrated. Assembling these
data across three states and two Forest Service regions was a moderately complex task.
Finally, different models were developed to represent the seven provinces.
Analytical complexity was limited in a number of ways: the broader concept of
watershed health was limited to a fish habitat analysis; the in-stream indicators were
limited to those already collected by the monitoring program; landscape (GIS) indicators were limited to those available for roads and derivable from the Forest Service
vegetation inventory; and indicators were scored on their general contribution to
habitat, rather than an absolute influence on fish habitat.
Social Complexity: Medium-low (2.5)

The complexity of the immediate social context of the modeling process was medium-low. The modeling occurred within the highly contentious and political context
of the Northwest Forest Plan, and its Aquatic Conservation Strategy was recently
rewritten to address what the authors perceived as misinterpretations of their intent
by the courts. This new strategy is being challenged in court. In contrast, the monitoring program itself has not been particularly contentious. All sides seem to agree that
monitoring is essential, and the monitoring results are not directly tied to any reallocation of resources. Social complexity was limited by narrowing participation mostly
to staff from land management agencies to avoid potentially contentious arguments
with regulatory agencies and the public.
More Information
http://www.reo.gov/monitoring/watershed/
(Last accessed August 4, 2006)
74

DSS used: Spectrum, RELM, VDDT, web-based mapping
Description
National forests are required to update their management plans every 1015
years. The adjacent Boise, Payette, and Sawtooth National Forests in southern Idaho
and northern Utah decided to update their plans together in order to better understand larger landscape issues and to address their many common concerns
more efficiently. National forest plans do not make specific decisions about
Key Points
timber harvesting or other activities, but rather have been described as more
The scope of the modeling
akin to land use zoning in determining overall rules and activities appropriate
project can change signififor certain areas. As part of planning, forests are required to calculate an Alcantly during the project;
lowable Sale Quantity (ASQ) of timber, which led the forest to use Spectrum,
initial calls for back of the
the standard DSS used for this purpose on national forests. The forests soon
envelope analyses for ASQ
realized that the basic forest growth and harvesting model could be expanded
eventually evolved into a
to help evaluate other effects of the different possible management alternamodel with 120 vegetation
tives. The model was expanded to include 120 vegetation classes (combinaclasses.
tions of vegetation types, successional stages, and canopy closures) that were
The kinds of DSS traditiondistributed across seven land allocation zones over 50 years for each of seven
broad management alternatives. To get a more detailed view of the feasibility
ally used to calculate timof these alternatives, the RELM DSS was used to take these Spectrum outputs
ber harvest levels are now
th
and distribute them further down to 6 field watersheds (about 200 per forbeing used to model more
est). Because fire is an important influence in the region that was not explicitly
complex vegetation dynammodeled by Spectrum and because there was some suspicion of inherent biases
ics over time for a variety of
in optimization modeling, a parallel modeling exercise using the VDDT DSS
resource outputs.
was also undertaken near the end of the planning process. (VDDT had also
Multiple DSS are often
been used to model the unforested parts of the planning area.)
needed to meet complex
needs: separate models
Analytical Complexity: High (5)
were needed to handle the
The forests cover a relatively large area (6.6 million acres), and analyses of
strategic (Spectrum) and
management options were disaggregated to the subwatershed level (approximately
tactical (RELM) aspects
200 per forest). Habitat trends were analyzed for seven alternatives over 50 years.
of planning; a simulation
The effects on approximately 2030 species were analyzed, ~10 quantitatively and
approach (VDDT) was also
the rest qualitatively. The analytical complexity was considerable, but we could say
done to provide an alternait was limited by choosing no more than seven alternatives, modeling them for no
tive view.
more than 50 years (some other national forest plans have gone up to 150 years),
looking at the subwatershed (rather than stand-level), and using groups of indicator species, rather than trying to model effects on all species.
Social Complexity: High (5)

Social complexity was high because national forest planning is an open stakeholder
process that directly affects on the allocation of resources. The social complexity of the
modeling process was somewhat limited by the fact that the public was not directly involved at all points and only saw the end results of the modeling in formal presentations.
More Information
http://www.fs.fed.us/r4/sawtooth/arevision/revision.htm (Last accessed December 21, 2005)
75
Red-cockaded Woodpecker Foraging Matrix Application

DSS used: Custom ArcGIS 9.x extension
Description
The red-cockaded woodpecker (RCW, Picoides borealis) is one of the longest
recognized federally endangered species. It lives only in open, mature, and old-growth
pine ecosystems in the southeastern United States, a habitat that has declined rapidly
due to fire suppression and short-rotation forestry. Its current abundance is estimated
at less than 3% of its abundance at the time of European settlement.
In 2003, the U.S. Fish and Wildlife Service published a major new revision of the
RCW recovery plan that includes updated management guidelines for both federal and nonfederal lands. Applying these guidelines on the ground can be complex
because breeding groups often occupy a cluster of nesting trees, and multiple groups
may be found adjacent to one another. To encourage compliance with the
new regulations, the FWS has developed an extension to the popular ArcGIS
Key Points
software that can assist managers in meeting the new guidelines. The software
As the sophistication of our
is referred to as the RCW Foraging Habitat Matrix Application and can be
understanding of habitat
downloaded free from the internet (http://www.fws.gov/rcwrecovery/matrix.
needs increases, computer
html). It builds on previous work by Fort Bragg on automating habitat evalutools may be able to help
ations based on digital forest inventories. The GIS software company (ESRI)
managers keep up.
and U.S. Army Environmental Center also contributed significant resources
Building flexibility into the
to the effort. One important difference from the past effort is that the new
software allows users to
guidelines require habitat details not normally present in forest inventories,
modify model parameters to
including ground cover and midstory hardwoods.
reflect their local conditions.
The software was released in April 2006, so it is too early to gauge its
impact. The authors expect considerable feedback and refinement of the tool,
Data not commonly coland a central design goal was to build in as much flexibility as possible.
lected for forest inventories
A few RCW modeling efforts simulate how populations of the bird
are often needed for habitat
will fluctuate over time given environmental influences. These models are
evaluations.
considerably more complex in that they simulate individual birds over
time in a spatially explicit manner. Designers of one of these models run it
on their mainframe computer for clients on a contractual basis. They have
also recently (2006) received a contract from the Department of Defense to create a
desktop computer version for managers, expected to be completed in 2009.
Analytical Complexity: Medium-low (2)

The analytical complexity of the tool is rated at medium-low. The geographic
resolution is high (stands with individual tree lists), but the scope considered is very
local (RCW group home ranges, ~80500 acres each). The application enables the
user to enter changes in habitat values (such as a planned clearcut) and see the resulting changes in the suitability scores. Several coarse filter habitat indicators are used,
including forest types, basal area of different forest structural stages, canopy cover,
and fire history. Analytical complexity was limited by considering only group needs in
one location at one point in time. As mentioned, a more complex tool was developed
separately for looking at multiple groups over time.
76
Social Complexity: Medium (3)

The social complexity is rated at medium because the software development and
use process involves mainly shared decisions between two organizations (a regulator
and the regulated). Social complexity has been limited, at least in the development
stages of the software, because the discussion appears to have been dominated by a
well-established community of experts, rather than by a more publicly oriented process, like a national forest plan.
More Information
http://www.fws.gov/rcwrecovery/matrix.html
(Last accessed June 27, 2006)
http://www.serdp.org/Research/upload/SI_FS_1472.pdf
(Last accessed June 27, 2006)
77
State
Chesapeake Forest Project
DSS used: Habplan and GIS
Description
he Chesapeake Forest consists of 58,000 acres of forest land scattered over
the eastern shore of Maryland, and it makes up about 12% of productive
timberland in the region. In 1999, the state of Maryland and the Conservation Fund cooperated to purchase the lands from an exiting industrial
owner. The Conservation Fund transferred title to the state the next year,
but also included a sustainable forestry management plan and ongoing contract with
consultants for management. As part of the management plan, the Habplan DSS was
used to model a possibility curve for endangered Delmarva fox squirrel habitat versus
timber volume extraction. The Habplan model has not been rerun since the transfer;
the project instead relies on relatively simple GIS system for its ongoing planning.
However, a recent forest certification audit recommended more attention to future
habitat modeling.
Key Points
The Habplan model helped frame

broad management options and is still
referred to.
The model was run by a consultant, and

the state agency overseeing the plan
never developed the in-house capacity
to run it (although certification may
now give them an impetus to do so).
The state agency is currently using

simple GIS overlays to screen sites for
management, but would like to project
timber yield.
Analytical complexity was rated medium because of relatively

high spatial and temporal complexity but relatively low complexity of
the forest and biodiversity measures. Spatial complexity was medium, due to moderate size (58,000 acres) but a fairly high resolution
(2080 individual stands), and the area was modeled over 50 years.
Analytical complexity was limited by modeling only two resources
(wood products (saw timber and pulp) and fox squirrel habitat) and
by modeling softwood (pine), not hardwood, stands.
Broad biodiversity issues and analysis

did not generate much interest in the
public planning process; the stakeholder group became dominated by public
input and discussion on opening lands
for hunting.
Social Complexity: Medium (3.5)

Although the project became an open stakeholder process, which
involved the allocation of public resources, this process occurred after the modeling was complete. The model was not a focus of controversy in the public planning process; rather, it was mostly irrelevant
as the process became dominated by the issue of hunting rights. The
model was more relevant in the prepublic phase, where the endangered species was being considered from a regulatory and species
recovery standpoint by the governmental agencies involved.
More Information
http://www.dnr.state.md.us/forests/chesapeakeforestlands.asp
(Last accessed November 11, 2005)
78
Oregons State Forest Management Planning

DSS used: FVS plus a custom-developed scheduling model
Description
To provide decision support to the Board of Forestry for its 2001 revision of the
strategic long-term plans for state forest management, the Oregon Department of
Forestry (ODF) modeled different alternatives on timber production and complex
stand structure development. A public planning process, begun in the mid-1990s,
had identified a range of management options, from industrial forestry to conservation-focused approaches. A compromise active-management approach, referred to as
structure-based management, was identified as the preferred alternative. The first
analytical effort in 1999 compared outputs from structure-based management to five
other scenarios from the identified range of alternatives. Originally scheduled to be
completed in five months, the effort took nearly a year longer because of the extensive
work needed to prepare all the necessary data. A variety of growth and thinning options were generated with the ORGANON growth-and-yield program. These were fed
into a custom-programmed, spatially explicit harvest scheduling model created by a
professor at Oregon State University. The model projected the alternatives for a 200-year planning horizon in 10-year intervals. The primary Key Points
indicators used to describe the results of the alternatives were harvest
Existing data were not adequate (in
volume, net present value and area of land in the oldest two (of five)
accuracy or format) for the modelstructure classes. The Board of Forestry approved the structure-based
ing goals; a new (and expensive) data
management plan in early 2001.
generation initiative was needed.
As with most modeling efforts, both available time and data were major constraints. Stand-level inventories, road access data, and information Although initial results were charon the growth-retarding effects of an emerging disease problem (Swiss
acterized as relative, they created
needle cast) were all weak or not available in formats that could be easily
concrete expectations that have been
incorporated into the model. Further, there was little time to involve the
hard to break.
various district and field foresters in refining the results. Because of these Engaging field personnel in an iterashortcomings, the results from these initial modeled alternatives were
tive model refinement process was
portrayed as relative, not absolute. Nevertheless, when operational estinecessary to incorporate on-themates the districts produced (after the plan had been adopted) came in
ground constraints and generate
at only about half the model-predicted harvest, it became a major politirealistic model outputs.
cal issue with the counties and forest industry that depended heavily on
Involving stakeholders is a widely
revenues from these forests.
recommended strategy, but scenarios
In 2003, the counties and the state agreed to a formal three-year,
they generate may be outside what
$2 million dollar project to enhance the modeling process in order to
decision makers see as their viable
provide decision support for a potential revision to the management
decision space.
plan, to support a decision on whether to pursue a habitat conservation
plan, and to help set harvest levels. This Harvest and Habitat Model
Project (H&H) utilized a new stand-level inventory, improved growth
projections (including updated impacts from Swiss needle cast and the use of the
Forest Vegetation Simulator program), and incorporated the costs and constraints of
silvicultural options and operational harvest units, including associated transportation
systems. District foresters were involved at every stage in the development of model
inputs and in a feedback loop with the modelers to help check and refine the feasibil-
79
ity of model operations. A separate GIS-based tool was developed to help facilitate
this checking. Four alternatives were modeled: the current management plan, using
both a proposed habitat conservation plan and an endangered species take avoidance strategy; the current management plan with only take avoidance; and timber
and conservation-oriented alternatives (elaborated in conjunction with these separate
stakeholder groups). The final results of this second phase were presented to the Board
in early 2006. Although Board members seemed to understand the model results,
they were not clear on their decision space, i.e., how much legal latitude they have
to adjust the plan and what are the specific features they can adjust. Timber interests
questioned the validity of the plan (not the model) because the new model estimates
are considerably below earlier estimates and what they consider sustainable. Although
the H&H project is now officially completed, the tools created will continue to be
refined and used in the states forest planning processes. In August 2006, a peer review
of the model was conducted, providing considerable information on the strengths and
weaknesses of this effort (available at the website cited below).
Analytical Complexity: Medium-high (4)

Analytical complexity of the forest component was high and the biodiversity component medium, leading to an overall rating of medium-high. The forest component
modeled a high number of spatial (~400,000) and temporal units (150 years by 5-year
intervals), requiring many decision variables (~3.6 million binary) plus additional
decision variables for the embedded transportation network and control of the distribution of complex forest patches across the landscape. Stands were modeled by using
individual tree lists. On the biodiversity side, results were grouped into five general
stand structure classes. Thirty-nine wildlife species were placed in three groups (generalists, simple structure specialists, complex structure specialists). Four high-level
management options were modeled, most with various suboptions.
Additional Points from Project Staff

(not elaborated in our description)
Project management should be
initiated at the start of a project.
Quality control and quality assurance procedures should be developed early, and sufficient time
should also be allocated for model
verification and testing.
Modeling projects can provide a
company or agency the opportunity
to examine program functions and
operational considerations for more
efficient operations.
Social Complexity: High (4.5)

Social complexity was quite high because this is a public openstakeholder process that directly allocates resources (to harvest or
conservation). It was ranked slightly below the Washington Water
Typing case (which involved regulation of private lands) and the
Boise-Payette National Forest planning (a federal process involving
more public involvement rules and a wider stakeholder base).
More Information
http://www.oregon.gov/ODF/STATE_FORESTS/Harvest_and_
Habitat_Model_Project.shtml (Last accessed December 5, 2006)
Overhulser, P., J. Sessions, R. Holloway, M. Rasmussen, and D.L.
Johnson. 2006. Oregon State ForestsAn approach to policy analysis
under complex demands. Oregon Department of Forestry, Salem, OR.
Sessions, J., P. Overhulser, P. Bettinger, and D. Johnson. 2006. Linking

Multiple Tools: An American Case. P 223238 in Computer applications in sustainable
forest management, including perspectives on collaboration and integration. Series:
Managing Forest Ecosystems, Vol 11. Shao, G., and K. Reynolds (eds.). Springer-Verlag, New York.
80

DSS used: GIS and custom built logistic regression model
Description
The Washington Forest Practice rules require different riparian buffer widths
to fishbearing and nonfishbearing streams (making this distinction is referred to as
water typing). The regulatory maps in force in the mid-1990s were found to significantly underestimate fish habitat, so the multi-stakeholder group negotiating the new
regulations agreed to develop a new scientific, model-based approach. The state Board
of Forestry adopted a regulation supporting the model-based approach, with the stipulations that the model achieve 95% accuracy and that a precautionary interim rule, which overestimates
Key Points
fish presence, would be followed during model development.
The burden of proof can swing dramatiA multistakeholder science group has been working on the
cally in a political process.
model since 2000, but their modeling has not been able to meet
DSS can improve the accuracy of predicthe 95% accuracy threshold in all areas of the state because of
tions; however, this inevitably threatens
geomorphic variability and the limited resolution of the topothose parties whose interests are favored
graphic data. Debate on the further development and potential
by the existing burden of truth.
use of the water-typing model continues, and the interim rule
Basing model acceptance on a standard of
remains in force.
accuracy (especially a high absolute one)
appears to be a common sense approach,
but parties can use technical debate to
The extent of the modeling task was large, the whole state
delay implementation on what ultimately
of Washington, but the analytical complexity of the modeling
must be a political decision.
process was deliberately kept low (at least conceptually). The
model results were simply fish presence versus absence, which
was based on only four geographic attributes: basin size, elevation, downstream gradient, and mean annual precipitation. The
model was designed to represent just one snapshot in time. Analytical complexity was
limited by limiting the number of models developed to two (east/west) and using only
a small subset of possible habitat variables.
Social Complexity: High (5)

The social complexity of the case was very high because it occurred in an open
stakeholder context involving direct allocation of resources. The purpose of the model
was to set regulations, which have large, direct economic and biological consequences.
Social complexity was limited by asking scientists to set aside their organizational
goals in the scientific work groups.
More Information
http://www.dnr.wa.gov/forestpractices/watertyping/ (Last accessed July 3, 2006)
81
Local
Summit County Lower Blue Subbasin Master Plan
DSS used: GIS (System for Conservation PlanningSCoP)
Description
Much of the responsibility for land use planning in the U.S. falls to county-level
government. There are, however, few written examples of DSS use related to biodiversity issues at the county level. One exception is the work of Tom Hobbs and David
Theobold at Colorado State University. One of their examples is a collaboration with
Summit County, Colorado, which is located about 60 miles west of Denver and is
the home of the mountain resorts of Breckenridge, Vail, and
Keystone. In terms of population, it has been one of the fastest
Key Points
growing counties in the nation (99.5% increase from 1990
There is a gap between the general scien2000). The White River National Forest occupies over 80% of
tific principles established on landscape
the total land area in the county, and considerable develophabitat evaluation and the specificity
ment has occurred in forested areas or on private urbanized
needed to implement them in a land use
lands that are forested and adjacent to federal lands (i.e., the
plan; expert interpretation is needed.
wildland/urban interface). As is common in many U.S. counties, a citizen committee updates a master plan for the county
Citizen involvement is needed to make
every few years. These plans do not directly set regulations,
the many value-based judgments needed
but rather provide guidance in setting legal standards, such as
in conservation planning.
zoning regulations. In Summit County, the comprehensive plan
Models used must be simple enough for
is further subdivided into four subbasins, of which the Lower
stakeholders to understand.
Blue subbasin is the least developed to date.
Models tend to become complex as they
Dr. Hobbs championed the need to better integrate biodiare customized for a particular use; in
versity information into county-level planning and received
order to be transferable to other locales,
funding in 1994 to develop such a system from the Great
they must be stripped to a few core eleOutdoors Colorado fund (state lottery money) and the Coloments.
rado Division of Wildlife. County commissioners from Sum Landowners value plans and regulations
mit and Larimer expressed interest and provided support for
that are constant and predictable, which
implementation in their counties. In addition to the computer
conflicts with the learn as you go proprogrammers and scientists, Hobbs assembled a collaborative
cess of adaptive management favored by
design group consisting of a county commissioner, a planner,
ecologists.
a developer, a land owner, a wildlife manager, and some environmental advocates. The system was built using an iterative
process of collecting ideas from the design group, constructing
prototypes, and obtaining feedback. Theobald et al. (2000) describe one of the lessons
learned in the collaborative design process:
Scientists on our design team originally advocated development of generalized population viability models as a way to analyze the consequences of development of a patch of
habitat. However, the citizen participants found this approach to be obtuse and excessively technical, requiring them to take on faith the validity of models produced by experts. There was a strongly expressed sentiment among these nontechnical members of
our design team that they must be able to explain any analysis we used in a reasonable
way to their fellow citizens, without relying on outside technical expertise to establish
the credibility of the analysis.
82
Additionally, they found a gap between the generalized theories that scientists
work with and the more specific information needed in local land-use planning.
Bridging this gap required experts willing to make difficult judgments or assumptions.
Many of the assumptions and parameters also involved value judgments, which were
most appropriately derived from the stakeholders. They also encountered difficulties
in bridging the differences in time and space as they relate to ecological processes
(long times and large areas) versus county planning processes (shorter times and
smaller areas). Most biodiversity data is collected at the state level, so the level of detail
is often less than ideal for local planning.
The resulting maps were used by the county in the update of their Lower Blue
Master Plan. By the end of the process, the SCoP tool had become too complex to be
easily transferable to other counties, and there was less political support at the state
level for such planning related to private lands. Some of the ideas were incorporated
into a statewide service operated by the Division of Wildlife called the Natural Diversity Information Source (NDIS). NDIS provides basic county-level statistics, species
status lists, and internet maps of historical land use development trends. It does not,
however, provide the type of species distribution and future build-out analyses that
formed the core of the Blue Subbasin analysis.

The analysis looked at a midsized area (178,400 acres/722 km2) at a fairly high level
of resolution (much of the data was from 30-m2 satellite imagery or 1:24,000 scale maps).
The analysis was simplified by looking, not at changes over time, but rather the maximum
build out estimate under nine development scenarios. Biodiversity was simplified to four
measures: rare vegetation types, habitat for species of special concern (federal and state),
neighborhood species richness, and economically important species habitat.
Social Complexity: Medium-high (4)

Social complexity was relatively high because it was a public planning process with
a likely effect (although not direct regulation) on private properties. Although the
design group was diverse, it was also small and necessarily simplified the full range of
possible perspectives.
More Information
Natural Diversity Information Source (NDIS). Online: http://www.ndis.nrel.colostate.
edu/ (Last accessed July 31, 2006)
Theobald, D.M., and N.T. Hobbs. 2002. A framework for evaluating land use planning
alternatives: protecting biodiversity on private land. Conservation Ecology 6(1):5.
Online: http://www.consecol.org/vol6/iss1/art5 (Last accessed July 31, 2006)
Theobald, D.M., N.T. Hobbs, T. Bearly, J.A. Zack, T.Shenk, and W.E. Riebsame. 2000.
Incorporating biological information in local land-use decision making: designing
a system for conservation planning. Landscape Ecology 15(1):355.
Duerksen, C.J., N.T.Hobbs, D.L. Elliott, E. Johnson, and J.R. Miller. 1996. Managing
development for people and wildlife: a handbook for habitat protection by local governments. Clarion Associates of Colorado, LLC and Colorado Division of
Wildlife, Denver, CO. Online: http://www.ndis.nrel.colostate.edu/handbook/
handbook.html (Last accessed July 31, 2006)
83
Baltimore Reservoirs Forest Conservation Plan

DSS used: NED-1 and ArcView GIS
Description
The city of Baltimore, Maryland, used a combination of computer-based tools,
primarily the ArcView geographic information system (GIS) and the NED-1 system,
to analyze risks to the long-term sustainability of their reservoir lands and to develop
and evaluate alternative scenarios for management of the lands. While maintaining
water quality was the primary goal, the second and third goals were maintaining and
enhancing the forest habitat as a contribution towards regional biodiversity. NED-1
inventories incorporated data needed to evaluate wildlife habitat composition and
structure and the quality of habitat along first- and second-order streams. While
providing a platform for the management and analysis of data
Key Points
on numerous key abiotic and biotic forest characteristics, the
NED-1 decision support software did not provide a mechanism
NED provided the best data structuring
for evaluating the relationships of these landscape elements.
mechanism the analyst could find, but it
The need to understand how landscape context and current
still left out major ecological elements, such
ecological processes were shaping the forest required a synas streams, roads, nutrient movement, and
thesis of tools and often required stepping outside the decidisturbance regimes.
sion support mechanism for critical answers to conservation
GIS provided a platform that was generic
problems.
enough to integrate these other ecological
elements, but, because it was generic, much
Analytical Complexity: Medium-low (2.5)
work was required to model them from
The spatial complexity of the analysis was relatively low: it
scratch.
covered 17,580 acres divided into 836 stands. Fourteen types of
The analyst emphasized that the final
forest plant communities were distinguished, and several eleproduct needed to be more than a report;
ments of forest habitat structure were analyzed (vertical canopy
it needed to include an information system
structure, interior habitat, coarse woody debris). Simplifying
that could be transferred to the City.
assumptions included using only these coarse filter measures
The analyst thought that communication
of biodiversity; no individual species needs were tracked, and
was best if organic and flexible, and he conno temporal aspect of the analysis was mentioned, so it appears
sidered the term communication strategy
it was based only on the current inventory.
to imply selling a predetermined solution.
Social Complexity: Medium-high (4)
The best approach was to attend citizen
advisory group meetings from the start,
The social complexity of the case was rated medium-high
express openness to their ideas, and volunbecause the report was expected to significantly influence the
teer to attend their interest group meetings
management of these public lands. Several local recreation
(which minimized the need for advisory
groups (bird watchers, mountain bikers, hunters, boaters) had
members to be messengers).
considerable interest in the lands. Rather than relying primarily on large, formal public forums, the lead planner appears to
have reduced the potential conflict by attending the various meetings of individual
interest groups to explain the effort and solicit their input more informally.
More information
http://cityservices.baltimorecity.gov/dpw/waterwastewater03/watershed_fcp/ (Last
accessed November 11, 2005)
84
Private
International Papers Forest Patterns System
DSS used: Forest Patterns (internally developed, based on ArcGIS and Microsoft Access)
Description
IP developed their Forest Patterns system to help them manage at the landscape
level and comply with environmental laws and the sustainable forestry certification
standards. The program tracks a hierarchy of land uses beginning with three broad
tiers of management: timber production, conservation, and nonforest. It contributes
to the conservation of biodiversity via management of landscape units (typically
40,000 to 60,000 acres). Landscape units can be assessed to determine structure and
forest cover type gaps or surpluses when compared with regional vertebrate landscape
scale models developed by the U.S.D.A. Forest Service.
Key Points

Forest Patterns is implemented at the regional division scale,
so the Northeast system covers a relatively large 1.6 million
acres. The relative biophysical complexity used in the system,
however, is moderate to low. To simplify the analysis, the system uses coarse filter habitat indicators, does not model habitat
over time, and does not explicitly try to estimate or characterize uncertainties. Analytical complexity was limited by using
a relatively small number of habitat classes and assuming that
the habitat targets will sustain biodiversity over time (a coarse
filter approach).
Social Complexity: Medium-low (2)

Top management support is critical.
A diverse group and external reviewers

should be involved in the system design.
Keeping data up to date takes a large com mitment.
Regular, direct contact with state agency

personnel should be maintained for up dating data.
A flexible framework should be created
that can be adapted to the needs of differ ent regions.
Training and a simple software interface

should be provided to make the system
relevant and usable for the foresters making decisions on the ground.
The systems original purpose was to address high social

complexity in the form of a distrustful public; however, conflict in
the immediate operating environment is relatively low because the
development and use of the system is controlled by one company. It is also used as documentation and a way to communicate
practices in governmental regulation and forest certification
processes, which involves the more complex social scenario of
communication between two organizations. Yet there were no mentions of the systems
use in conflict situations. The principal day-to-day use of the system is in coordinating
multiple decision makers within a single organization. Although participation in system
design and review was extended beyond the organization, social complexity was limited
to technical specialists, rather than trying to directly integrate public input.
More Information
http://ipaper.com/Our%20Company/Environment/EnvironmentalStewardship.html
(Last accessed May 23, 2006)
85
SFI Certification
DSS used: GIS
Description
The Sustainable Forestry Initiative Standard (SFIS) is a form of self-regulation initiated by the forest industry. All companies belonging to the American Pulp and Paper
Association are required to undergo SFIS certification. As seen in the IP case above, DSS
can contribute significantly towards the certification process. In contrast to the previous
case, here we look at DSS use from the perspective of those doing the certification (regulators), rather than the landowners. Under SFIS objective 4, companies are required to
have programs to promote biological diversity at stand and landscape levels. A review
of several certification summary reports and interviews with two certification specialists
revealed that there is no one standardized procedure for these biodiversity analyses. It is
up to each company to devise methods and each certifier to judge their acceptability.
Key Points
The most visible emerging trends are collaboration with and borrowing techniques from The Nature Conservancys (TNC) ecoregional analyses and the use of
Few details are publicly
a global species ranking system devised by NatureServe and its network of Natural
available on how compaHeritage programs. Companies are required to have plans to conserve native bionies do SFIS biodiversity
logical diversity (ecological communities and individual species) in general, and
analyses; they are only
to locate and protect known sites associated with viable occurrences of critically
required to publicize brief
imperiled (NatureServe rank G1) and imperiled (rank G2) species and communisummaries of their certities. These requirements are most often met by developing customized GIS analyfication processes.
ses that combine the companys forest inventory data with other biophysical layers
Companies appear to
(e.g., slope, soils) to identify important biodiversity areas. These GIS screens are
have been largely devisused to identify both ecological communities (coarse filter) and individual species
ing their own individual
(fine filter) habitat needs. NatureServe recently came out with a new DSS named
methods to meet biodiVista, which could help companies integrate economic and biodiversity values to
versity analysis requireprioritize conservation areas. The SFIS 20052009 standards now require expertise
ments, usually using a
in forest modeling on the certification teams, which may begin to raise the bar
standard GIS package.
on expectations for habitat analysis (especially simulations into the future).
Companies should be
able to continue to build
Analyses done for SFI certification have been fairly simple to date. Geographic
on information and
areas
are usually confined to one companys land holdings, although they may reftools provided by TNC,
erence broader analyses done by other organizations (TNC, states). Little evidence
NatureServe, and the
of habitat analyses over time was found. The characterization of biodiversity was
state Natural Heritage
moderately complex, as it included both coarse and fine filter measures.
programs; NatureServes
Vista DSS can help
Social Complexity: Medium-low (2)
integrate economic and
Social complexity is relatively simple and was ranked as medium-low because
biodiversity values.
the process mainly consists of developing standards internally within one company.
Standards and implementation are reviewed by a third party certification team;
however, the level of conflict is typically much less than for enacting public regulations.
More information
http://www.sfiprogram.org/ (Last accessed July 14, 2006)
http://www.natureserve.org/ (Last accessed July 14, 2006)
http://conserveonline.org/workspaces/ecotools (Last accessed July 14, 2006)
86
Vermont Consulting Foresters

DSS used: NED-1
Description
In the Phase I review of available decision support tools, NED was the system
most oriented towards small landowners. In an interview with the software developer,
he noted that few small landowners appear to use it themselves; rather, the main users
seem to be consulting foresters. Three foresters were contacted and interviewed about
their use of NED with small landowners. The NED system contains a
wildlife module that uses a forest inventory to estimate habitat types
Key Points
and qualitative likelihood of wildlife presence/absence (based on
DeGraaf and Yamasaki 2001). None of the consultants interviewed
The main appeal of the software has
has used the NED wildlife module with clients, however; instead they
been to calculate inventories from
simply use their own knowledge to advise landowners on wildlife
sample cruises.
issues.
Many landowners are interested in
wildlife, but the qualitative evaluaAnalytical Complexity: Low (1)
tion provided by the software does
Analytical complexity was rated low because the spatial complexnot provide any value-added to
ity was limited by the small properties involved. The DSS was generthe consulting foresters who use the
ally only used to structure the current forest inventory, rather than
system; it is easier for them to simto project forest growth over time. As mentioned above, the DSS was
ply draw on their own knowledge.
not used to assess biodiversity indicators, and the type of analysis
needed for the situation was a simple, qualitative assessment of wildlife effects of different forest management options.
Social Complexity: Low (1)

Social complexity was low because the decisions involved only the goals of individual landowners.
More Information
http://www.fs.fed.us/ne/burlington/ned/index.htm (Last accessed July 5, 2006)
Degraaf, R.M. and M. Yamasaki. 2001. New England Wildlife. Habitat, Natural History, and Distribution. University Press of New England, Hanover, New Hampshire.
87
Appendix B. DSS Abbreviations and References

Category
Biodiversity

Forestry

General

RegAssess

Abbreviation
BMAS

CAPS
C-Plan
MARXAN / SPEXAN
PATCH
RAMAS
Refuge GAP
ResNet & Surrogacy
Sites
Vista
FVS
Habplan
Harvest
LANDIS
LANDSUM
LMS
MAGIS
NED
RELM
RMLANDS
SIMPPLLE
Spectrum
TELSA
VDDT
Woodstock
DEFINITE
EMDS
Netica

CLAMS
LUCAS
MRLAM
Restore
WBAFA
Full name
Biodiversity Management Area Selection

Conservation Assessment and Prioritization System
C-Plan
MARXAN / SPEXAN
Program to Assist in Tracking Critical Habitat
RAMAS
Refuge GAP
ResNet & Surrogacy
Sites/Site Selection Module
NatureServe Vista
Forest Vegetation Simulator
Habplan
Harvest
LANDIS
Landscape Successional Model
Landscape Management System
Multiple-resource Analysis and Geographic Information System
NED
Regional Ecosystem and Land Management Decision Support System
Rocky Mountain Landscape Simulator
Simulating Patterns and Processes at Landscape Scales
Spectrum
Tool for Exploratory Landscape Scenario Analyses
Vegetation Dynamic Development Tool
Woodstock, Spatial Woodstock & Stanley
DEFINITE
Ecosystem Management Decision Support
Netica

Coastal Landscape Analysis and Modeling System
Land-Use Change and Analysis System
Multi-Resource Land Allocation Model
Restore
88
Website or reference
Fischer, D. and Church, R. 2003. Clustering and compactness in reserve site selection: an
extension of the Biodiversity Management Area Selection model. Forest Science 49(4): 555565.
www.umass.edu/landeco/research/caps/caps.html
www.uq.edu.au/~uqmwatts/cplan.html
www.ecology.uq.edu.au/marxan.htm
www.epa.gov/wed/pages/models/patch/patchmain.htm
www.ramas.com
www.sdvc.uwyo.edu/wbn/refuge/
http://uts.cc.utexas.edu/~consbio/Cons/Labframeset.html
www.biogeog.ucsb.edu/projects/tnc/toolbox.html
www.natureserve.org/Vista
www.fs.fed.us/fmsc/fvs/
http://ncasi.uml.edu/projects/habplan/
www.ncrs.fs.fed.us/4153/harvest/harvhome.asp
http://landscape.forest.wisc.edu/projects/landis.htm
www.landfire.gov/NationalProductDescription20.php
http://lms.cfr.washington.edu
www.fs.fed.us/rm/econ/magis
www.fs.fed.us/ne/burlington/ned
www.fs.fed.us/institute/planning_center/plan_relm.html
www.umass.edu/landeco/research/rmlands/rmlands.html
www.fs.fed.us/rm/missoula/4151/SIMPPLLE
www.fs.fed.us/institute/planning_center/plan_spectrum.html
www.essa.com/downloads/telsa/
www.essa.com/downloads/vddt/
www.remsoft.com
http://tinyurl.com/2prcxe
http://www.institute.redlands.edu/emds/
www.norsys.com
www.spiritone.com/~brucem/bbns.htm
www.fsl.orst.edu/clams
www.cs.utk.edu/~lucas
http://www.fvle.org/
http://biosys.bre.orst.edu/restore/default.cfm
http://oregonstate.edu/Dept/pnw-erc/

Conserving Creatures of The Forest

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Conserving Creatures of The Forest

Transféré par

Droits d'auteur :

Formats disponibles

Conserving Creatures of

Chapter 1. Introduction Norm Johnson and Sally Duncan

Chapter 2. How Do We Make Decisions? Denise Lach and

Chapter 3. Conserving Biodiversity: The Challenge

Chapter 4. DSS: How Might They Help? How Might

Chapter 5. Tips For Successful DSS Use: Its All

Chapter 6. Shopping for the System that Meets

Appendix A. Case Study BriefsSean Gordon

Appendix B. DSS Abbreviations and References

This research was supported by a grant from the National

ver the last 30 years, we

complex decisions in the context of

clear-eyed approach to whether such

sion support systems.

out of past choices and understand

Why are biodiversity

suming the needs of other species

which the decision is made is critical.

strate biodiversity conservation to

Very little has been published

lem solving. Most of the literature

DSS: How might they

A DSS can help by

enabling consideration of detailed and complex information and relationships

A DSS can hurt by

committing the organization

ison is that decision support systems

our points about organizing group

Tips on successful use of

Who should be involved?

and resources. Our experience and

Who will do the DSS modeling?

they might also be able to play

How Should the DSS be used?

with means to check the results, both

in terms of the coarse and fine filter

Focus Major Function

three major aspects of DSS functionality. First, we compared the outputs

mutiple spatial scales, and the support

ver the last 30 years, we

ity, and no decision about conserving

decision we make in our lives to a detailed evaluation of goals, alternatives,

Some Rules for

are computerized decision aids to

More generally, we have decision

Thus, our philosophy on using

Firm deciding where to

Chapter 2. How Do We Make Decisions?

Classic decision models are

conservation of biodiversity may have

the premise that effective decisions

and expert in, the strategies that we

doing when theyre quickly searching

Forest Creatures, Old Growth, and Timber Harvest

Group Decision Making

points was reduced, sundering the

characterized as individuals looking

Worldviews and Decision

worldview for individualists who

Water Rights and Learning to Share

of the problems we face and the

us are unable to resolve. Typically,

with a satisfying solution and move