See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/37626421

Making smart conservation decisions

Article · January 2001

Source: OAI

CITATIONS READS
257 2,065

5 authors, including:

Hugh P Possingham Sandy Andelman

University of Queensland Wildlife Conservation Society
1,114 PUBLICATIONS 88,273 CITATIONS 140 PUBLICATIONS 19,548 CITATIONS

SEE PROFILE SEE PROFILE

Barry R Noon Stephen C Trombulak

Colorado State University Middlebury College
187 PUBLICATIONS 10,450 CITATIONS 62 PUBLICATIONS 6,199 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by H. Ronald Pulliam on 01 June 2014.

The user has requested enhancement of the downloaded file.

Possingham et al. Conservation Decisions September 2000

Making Smart
Conservation Decisions

(To appear as a chapter in “Research priorities for nature conservation” in 2001 – a book
sponsored by the Society for Conservation Biology and edited by Soule, Orians and Kohm)

1,2
H. P. Possingham ,

S. J. Andelman2,

B. R. Noon3,
4
S. Trombulak, and

H.R Pulliam5.

1
Centre for Conservation Biology, The University of Queensland, Brisbane QLD 4072
AUSTRALIA.
2
National Center for Ecological Analysis and Synthesis, Santa Barbara, 735 State St, CA
93101, USA.
3
Department of Fishery and Wildlife Biology, Colorado State University, Fort Collins, CO
80523, USA
4
Department of Biology and Environmental Studies, Middlebury College, VT 05753, USA
5
Institute of Ecology, University of Georgia, Athens, GA 30602, USA

1
Possingham et al. Conservation Decisions September 2000
Introduction
In little more than a decade, conservation biology has generated a large and growing
body of theory aimed at predicting and managing the impacts of anthropogenic activities on
populations, species and ecosystems (e.g., Meffe and Carroll 1997, Ferson and Burgman
2000). However, to succeed in its mission, conservation biology must do more than
generate theories and principles. Conservation biologists must use their theories to deliver
effective, science-based decision tools for practical use by managers and policy makers.
Furthermore, to demonstrate the efficacy of conservation biology theory and applications to
skeptical managers, conservation biologists must also establish feasible yet reliable
monitoring programs to detect key trends and effects and determine the effectiveness of
decisions (Kareiva et al. 1999; Simberloff 1999). Transforming pure science - whether
theoretical or empirical - into information that can actually be used by managers and
decision-makers to address the sorts of conservation problems outlined throughout this
volume remains a major challenge in conservation biology. To meet this challenge, the
next generation of conservation biologists will need to embrace more economics, more
management science, more decision theory and more operations research.

This chapter has two broad objectives. First, we outline the key elements for how to
make smart conservation decisions and measure their benefits – an outline that is applicable
to the sorts of conservation problems discussed in this book. With this framework in mind,
we then highlight priorities and directions for future research.

We begin by discussing some specific problems generic to species and ecosystem

management that have resisted rational, science-based solutions. These examples illustrate
some of the gaps in traditional approaches to conservation biology research and provide a
platform to illustrate the benefits of decision-theory approaches. Next we give examples of
the application of decision theory to solving conservation problems. While decision-
making methods can account for uncertainty and errors, monitoring the results of
conservation actions is essential for improved decision-making in the future. Ultimately
our objective is to encourage the development of active adaptive management programs on
real conservation problems (Parma et al. 1998). Within an adaptive management
framework, once an action has been initiated, a monitoring program must be implemented
that evaluates and measures the performance of that action. The performance evaluation
then informs the decision-making process by reducing uncertainty and ignorance. In this
paper we use the term monitoring to refer to a repeated assessment of some environmental
attribute for the purpose of detecting change within a defined area over time (Thompson et
al. 1998). Specifically, the type of monitoring we describe is “performance evaluation”.
That is monitoring explicitly designed to assist decision-making and management. We
conclude with an overview of research opportunities for monitoring and decision-making,
emphasizing the challenge of developing robust rules of thumb for making decisions.

Two common conservation decisions where conservation biology has failed

to provide useful advice

Where in space should habitat be protected from destruction, and where should it be
restored? This first question is fundamental to conservation and conservation biology
should provide the answer. There is a range of theories that generally address these
questions, beginning with “island biogeography theory” (MacArthur and Wilson 1962), and

2
Possingham et al. Conservation Decisions September 2000
more recently, with “metapopulation theory” (Hanski 1999) and “source–sink theory”
(Pulliam 1988). Let us see what answers these theories provide to this question.

Island biogeography theory generated a series of rules, most notably:

• Big reserves are better than small reserves; and

• Connected (or close) reserves are better than unconnected reserves.

Metapopulation theory tells us that there are several ways to increase the likelihood that a
metapopulation will persist:

• Decrease local extinction rates;

• Increase between patch colonization rates;

• Increase the number of suitable patches (or, conversely, minimize patch loss); and

• Increase the number of occupied patches.

Source–sink theory emphasizes the importance of identifying habitat where population

growth rates are consistently positive hence the rule:

• Protect source populations and ignore sink populations.

We also have the empirically derived rule:

• Reserves with a low edge to area ratio are better than reserves with a high edge to area
ratio.

All these rules have been useful in providing general conservation principles, but
their use is more limited for dealing with specific conservation and management problems.
To provide useful practical advice, conservation biologists must have a clear understanding
of the questions that managers want answered. For example, consider the question of
where it is best to restore habitat. A land manager interested in habitat restoration will
typically have a finite set of resources and financial constraints. She may envisage
revegetating 10% of an existing site. The question for conservation biologists is: which
10% should be restored to maximize biodiversity benefits?

Let us assume there are two existing patches of native vegetation on the site, one
larger than the other. The specific question then is: to which of the following options
should revegetation efforts be directed?

1. Make the existing bigger patch bigger;

2. Make the existing smaller patch bigger;

3. Connect the two existing patches with a broad corridor;

4. Make one new large patch; or

5. Make many new small patches evenly scattered over the property.

3
Possingham et al. Conservation Decisions September 2000
Island biogeography theory and metapopulation theory provide few insights to help
us choose amongst these options, except, perhaps, ruling out option 5. Although without
further information about the distribution of species and communities across the property,
we might not want to eliminate this option. The reason these conservation biology theories
have limited utility for such practical decisions is that the general principles have not been
couched within a decision-making framework. The question of where to focus restoration
efforts raises the immediate question of what biodiversity is the land manager trying to
maintain or recover? What is the objective? A simple question that is rarely answered.

Even when an explicit objective is specified, none of our theories provide explicit
predictions as to how the objective is most likely to be met. If the biodiversity objective is
to minimize the likelihood a species becomes extinct, we may test alternative habitat
reconstruction scenarios in a population viability model (something which is rarely done
given the lack of data and the demands on the time of managers). If the biodiversity
objective were to restore a representative sample of habitats, then information on soil types
and topography would be vital. Given the vast conservation literature on habitat
fragmentation generated over the past decade, it is an embarrassment that so little of this
research (empirical or theoretical) provides answers to the most basic practical questions
concerning optimal habitat reconstruction for biodiversity conservation. The theories
generate a list of generally useful things we might do, but they offer little insight into how
best to choose among the alternative options (Possingham 1997).

The example provided is deliberately simple. Most conservation problems are more
complex, and a land manager may envisage spending her conservation dollars in other
nature conservation activities, such as control of introduced predators and weeds in the
existing habitat patches. The question then becomes: What fraction of our conservation
effort should be directed to each of several management activities: revegetate, predator
control, and weed control? What conservation biology theory helps us juggle these
alternatives?

A second common question in conservation biology relates to the problem of

reintroductions and captive breeding. There is a large amount of science focused on
reintroduction and captive breeding (Ralls and Ballou 1986), but few protocols exist for
answering basic questions like:

• When, in the decline of a threatened species, should we initiate a captive breeding

program?

• How many individuals should be brought in to found a captive breeding program?

• Which individuals should comprise the initial captive population (i.e., what combination
of ages and sexes)?

• When should individuals be released into the wild, and how many?

Again, the existing theory provides general rules such as: reintroduction of a large
number of individuals will typically be better than a smaller reintroduction. Empirical
reintroduction research yields essential data and experience, but provides only a limited
framework for extending experience with a particular species or site to new species or
uncertain circumstances. Maguire et al. (1987), Maguire and Servheen (1992) and Lubow
4
Possingham et al. Conservation Decisions September 2000
(1996) addresses some of these reintroduction/captive breeding questions for specific
circumstances using decision theory tools, but they are lonely beacons.

What is decision theory and how has it been used in conservation?

Decision theory is a framework within which people responsible for management

attempt to achieve explicitly stated objectives while acknowledging the levels of
uncertainty involved with the decision process (Clemen 1996). Other professionals, such as
engineers (Kulkarni et al. 1993, Ravirala & Grivas 1995, Teng & Tzeng 1996), and
financial advisers (Beenhakker 1975), use this sort of theory to make decisions. In some
cases decision theory relies on complex mathematical tools that go under the generic name
of “mathematical programming methods.” However decision theory can be much broader
and often includes qualitative tools such as “ecological risk assessment” or “multi-criteria
decision analysis.” The common thread within the theory is a disciplined protocol for
problem solving that includes the following seven steps (Shea et al 1998):

1. Specify the management objectives, or at least list the indicators of policy performance
(e.g., minimize the risk that numbats become extinct in the next 50 years). Where there
are multiple objectives, utility theory (Maguire and Servheen 1982, Guikema and Milke
1999) is employed to deal with the problem of how to maximize multiple objectives that
are measured in different currencies (e.g. money, biodiversity loss and risk to human
health).

2. List the management options and express them as control variables (e.g., release x
animals in year t, expend y$ on baiting predators, etc.).

3. Specify the variables that describe the state of the system (e.g., population size, predator
abundance).

4. Develop a conceptual model of the dynamics of the system being managed, and if
possible develop equations to describe the dynamics of the state variables (e.g.,
equations for numbat and predator population co-dynamics). This step will often
involve collaboration between a field biologist and ecological modeler.

5. Specify constraints that bound the decision variables and state variables (e.g., a recovery
program budget, the social acceptability of predator reduction strategies, etc.).

6. Specify the range of uncertainty for all the parameters.

7. Find solutions to the problem. Once the problem is defined by these components the
manager may often need a decision-making protocol and/or mathematical programming
package to find the best, or at least some good, solutions.

To illustrate how decision theory thinking might be integrated with traditional

conservation biology approaches, we now present three examples.

Example 1: Optimal fire management for a large conservation park

Managing disturbance processes in conservation areas is a contentious issue

throughout the world. In many large and semi-natural terrestrial systems fire is the main

5
Possingham et al. Conservation Decisions September 2000
force that disturbs vegetation and it is often managed with fuel reduction burning, attempts
to reduce ignition frequency, and prescribed burning to increase fire frequency. The pro-
and anti-fire advocates often take highly polarized positions based on preconceived ideas,
conflicting goals, and non-predictive mental models of an area’s natural history. The fear
of making the wrong decision invariably means we make the decision to do nothing.

Here we summarize Richards et al.’s (1999) use of decision theory to determining

an optimal fire management strategy for a large conservation reserve. The description of
the problem fits into the seven-point framework above.

1. Statement of objective(s). We set the broad objective of maintaining a balance of

the different successional states. Specifically, Richards et al. (1999) set the target that at
least 20% of the park should be in each of the three successional states of the vegetation -
early, mid and late. (This is a surrogate objective for the more fundamental objective of
minimizing species loss. However there are so many species, and limited information on
each, that the ecosystem-level objective is expedient and more likely to be implemented and
monitored).

2. List of management options. Each year the park manager has three options: do
nothing, attempt to stop all wild fires, and prescribe burns in some of the park.

3. State variables. The state of the system is the percentage of the park in each of the
three successional states - early, mid and late.

4. State dynamics. Using basic probability theory, we constructed transition

probabilities that predicted the chance of the park moving from any one state to any other
state in a single year, given that the manager had chosen a particular management option for
that year. The two forces that drove the dynamics were fire and vegetation succession, and
the final state dynamics model is a Markov chain model.

5. Constraints. We did not constrain the ability of the manager to enact any of the
three management options, although we did assume that the fire suppression option would
not be completely effective.

6. Uncertainty. We varied the definition of early, mid and late successional habitat,
and the relationship between fire frequency and habitat state, to do a limited test of the
robustness of our conclusions. For more detailed fire management a whole host of
uncertainties regarding ability to control fire, risk to life and property, and even variability
in the political climate, could be used to enhance the robustness of the results.

7. Solution Methods. We used Stochastic Dynamic Programming, a common

mathematical programming method (Intrilligator 1971), to find the optimal solution.

The main outcome is a state-dependent fire management diagram that tells a

manager what to do given a particular state of the park (see Fig. 1, from Possingham and
Shea 1999). Where the resources to explore options for a particular park are limited, Figure
1 could be thought of as a rough rule of thumb for fire management in a large semi-natural
park, although its robustness would require considerable further testing (see point 6 above).

6
Possingham et al. Conservation Decisions September 2000
Unexpectedly, we found that a manager often needs to do something even when the
park is in a “good” state, and sometimes a manager should do nothing when the park is in a
“bad” state. Exactly when a manager should take the action of “Do Nothing” is not easy to
predict without a model. The stochastic dynamic programming method is able to accurately
integrate decisions in a stochastic world over a longer time frame than human intuition.
There is no fixed optimal policy - it depends on the state of the park.

One question raised by the work was the social and economic cost of adopting
different policies. This was not dealt with in detail, yet may be important in some cases.
Utility theory would need to be applied to generate relationships between different “values”
- conservation value, economic cost, and social perception of fire. This issue is discussed in
the context of another example below.

100% Figure 1. A state-dependent decision diagram

for fire management in a large conservation
reserve. The result is based on an objective of
trying to keep at least 20% of the park in each
of the three successional states, early, mid and
late. Do nothing means that fires are neither
started or extinguished.

Percent of early-
successional
habitat in the park Set of “good” states
Control
wildfires
Prescribe
burn
Do nothing
Do nothing
0%
0% 100%
Percent of mid-successional habitat in the park

Example 2: Conservation Program Planning (Guikema and Milke)

Guikema and Milke (1999) consider the problem of allocating funds between
different conservation projects. They use decision analysis to show how an agency would
go about fund allocation. This use of decision theory is at quite a different scale to the
example above, and relies heavily on resource allocation theory developed in other
industries. The work is briefly described below.

Envisage an agency that needs to allocate funds between a variety of projects. For
example, it may be a federal agency that must decide how to allocate funds among 100

7
Possingham et al. Conservation Decisions September 2000
projects on different threatened species, or it may be a non-governmental organization that
must decide how much money to invest in acquisition of new protected areas, vs. how
much to invest in management and restoration of existing reserves. Guikema and Milke
(1992) summarize the conservation planning procedure process using a single diagram (Fig
2).

OBJECTIVE
MODEL

UTILITY UNCERTAINTY
MODEL MODEL

CHOICE MODEL

OPTIMIZATION SENSITIVITY
ROUTINE and ANALYSIS

Figure 2. Components of the proposed conservation program planning procedure

(Guikema and Milke 1999).

Again we frame the problem in our seven-step outline.

1. Statement of objective(s). A typical agency-based management objective will represent

a balance between recreation and biodiversity objectives, such as: number of visitor days,
level of visitor satisfaction, species protection, and ecosystem protection. Much of
Guikema and Milke (1999) is devoted to defining a utility function for each objective and
combining them into one objective function. They show how some probability is required
to modify the value of different objectives where the chance they will occur is uncertain.
For species protection this might require some Population Viability Analysis (Beissinger
and Westphal 1998). For an objective such as number of visitor days this may require some
expert judgment and statistical modeling.

2. List of management options. The decision is how to allocate the total budget among
the projects. The control variables will be the amount of money allocated to each project.

3. and 4. State variables and their dynamics. The relationship between the funds
allocated to each project and the achievement of each objective needs to be modeled. Often
this requires sub-models that may include expert opinion. These sub-models will

8
Possingham et al. Conservation Decisions September 2000
incorporate definitions of state dynamics, state variables and the impact of funding
allocation on the state variables and dynamics.

5. Constraints. The primary constraint is the total budget. The sum of the allocated funds
must be less than, or equal to, the total budget. The user may wish to allocate varying
amounts of funds to each project or just allow a project to occur, or not, with one fixed
budget.

6. Uncertainty. Systematically changing uncertain parameters would involve modifying

the weights given to different objectives, or the parameters determined by expert opinion,
etc.

7. Solution Methods. In this case, the problem is linear (the objective function is a linear
combination of the control variables, as are the constraints) so many standard mathematical
programming packages can solve this problem quickly (even with a spreadsheet).

Guikema and Milke (1999) focussed most of their discussion on step 1 - framing the
objective using multi-component decision analysis. This task in government agencies is
extremely complex and should include a wide range of social and human management
issues that we have glossed over. While the scale and scope of this problem is quite
different from example 1, we can apply the same seven-step framework.

Example 3: Reserve system design

There is now a large literature on optimal reserve system design (Margules and
Pressey 2000). Some of the problem formulation approaches and mathematics involved are
reviewed in Possingham et al. (2000). Most ecologists are aware of the problem that is
outlined in the seven-step decision theory framework below. The social and economic
aspects of reserve system design are in most need of further research.

1. Statement of objective(s). The objective is to adequately represent a suite of

biodiversity features (usually species or habitat types) in a reserve system for the
minimum cost. The minimum cost could be expressed in terms of the area, land cost,
opportunity cost for other uses, management cost and/or boundary length of the reserve
entire system eventually chosen.

2. List of management options. The control options are to acquire, or not, each parcel of
land or sea (called a planning unit) for the reserve system. There is a control variable for
each planning unit. Current interest in off-reserve conservation and marine zoning
would mean that the list of options for any parcel of land or sea should be broader and
include non-purchase options (like covenanting) or partial protection (through zoning).

3. State variables. This is usually a static problem where the key parameters are whether
or not a biodiversity feature occurs in a planning unit.

4. State dynamics. There are normally no dynamics in reserve design problems, although
this is an area of active research.

5. Constraints. The primary constraints are the biodiversity constraints. Usually the
problem is formulated so that all biodiversity feature targets (e.g., conserve 15% of

9
Possingham et al. Conservation Decisions September 2000
habitat A, or 500 individuals of species B) must be met. Planning units are constrained
to one of a few states: already in the reserve system, available for conservation
designation, or unavailable. Where we are interested in multi-use zoning of a region
(e.g., the Great Barrier Reef Marine Park) then a planning unit could be allocated to one
of many zones.

6. Uncertainty. What happens if some of the GIS data layers have errors, or species lists
are incomplete and inaccurate? Recent work (Andelman and Meir, in press; Meir and
Andelman, in press) has been carried out on the impacts of uncertain, incomplete or
unreliable data.

7. Solution Methods. These sorts of problems often become so large that classical
mathematical programming methods fail. Simple heuristic algorithms or less classical
optimization methods like simulated annealing have been used (Possingham et al. 2000).

Where reserve system design is implemented, it quickly becomes apparent that

society is not a single entity with a single value system. Whereas a conservationist may
value a particular site because it contains habitat for an endangered species, a timber
company may value that site because of the potential revenue that might be generated from
harvesting trees, while a horse-riding group may value that site for its recreational values.
Perhaps one of the greatest benefits of decision theory is that it provides a protocol to help
distinguish among, and integrate, various goals held by multiple sectors of the public.

For example, for the Great Barrier Reef Marine Park we are working on designing a
reserve system to conserve representative samples of about 80 marine habitats and reefs in a
fully protected zone. If conserving biodiversity is all that matters, this problem is hard, but
not as hard as a problem that takes into account other values. Real world marine reserve
design has to include the interests of others: commercial fishers, recreational fishers, the
tourist industry, etc. Another complication for reserve system design is multiple zoning,
where any planning unit needs to be assigned to one of several zones, each of which has
implications for conservation, scientific, social and economic use. Multiple zoning
problems raise many questions. From the biodiversity perspective, would two reefs closed
for fishing but not tourism compensate for one reserved reef opened up for fishing? A
question like this is hard to answer for a single species, let alone for biodiversity in its
entirety. We are a long way from developing fully integrated and adaptive reserve system
planning tools, yet most countries need them now.

The reader will have noticed that the seven steps to decision-making fall short of
taking action. Two further essential steps now need to be considered, enacting the
consequences of the decision-making and then monitoring their consequences. Monitoring
and consequent re-evaluation and reformulation of our underlying data and models should
lead us in to the cycle of active adaptive management.

Monitoring and Performance Evaluation

Once a decision has been made, a monitoring program needs to be implemented,

with the aim of evaluating and refining the solution or management action chosen. Ideally
the monitoring targets biodiversity assets and processes of greatest concern and uncertainty.

10
Possingham et al. Conservation Decisions September 2000
Although monitoring activities are widespread in conservation biology and applied ecology,
data from existing monitoring programs are under-used for a number of reasons:

• Many monitoring programs are poorly designed (e.g., the questions they are intended to
answer are not well defined, and/or the methods selected are inappropriate to the
objectives), resulting in sub-optimal data;

• The spatial and temporal scales of monitoring activities are frequently inappropriate to
the scales at which inferences from those monitoring activities are made;

• Statistical and technological tools for analyzing monitoring data remain relatively
primitive;

• There is little known about the population and natural history needs of the organisms of
concern, and

• There is a perception, particularly among agencies, that data collection, rather than
analysis and communication of trends and patterns in those data, is the goal of
monitoring.

Detecting ecologically meaningful trends, and distinguishing the effects of human

activities from those of natural disturbances – particularly at large scales - are notoriously
difficult (Dayton et al. 1998; Bradshaw et al., in review). Recent controversies over the
magnitude and causes of declines of neotropical migratory songbirds (e.g., James et al.
1996; Thomas 1996) and amphibians (Blaustein et al. 1994; Alford and Richards 1999 and
references therein) illustrate some of these difficulties. Furthermore, historically,
monitoring has not been viewed as a legitimate research activity by many academic
institutions, and maintaining adequate funding for monitoring over the requisite temporal
and spatial scales has been difficult. Nevertheless, there is no way around the fact that
successful management of biodiversity requires substantial monitoring.

Just as the common thread within decision theory involves a disciplined protocol
consisting of seven steps, designing an effective monitoring program also involves a series
of seven systematic steps (Mulder et al. 1999):

1. Specify goals

2. Identify stressors

3. Develop conceptual Model

4. Select Indicators

5. Establish Sampling Design

6. Define Methods of Analysis

7. Ensure Link to Decision-Making

Below we briefly discuss each of these stages, highlighting key steps, as well as
areas where more research is needed.

11
Possingham et al. Conservation Decisions September 2000
Specify Goals

As discussed in the context of decision-making, many conservation projects go

wrong at this step. The same is true of monitoring projects. We cannot emphasize enough
the importance of providing an explicit statement of the questions the monitoring program
is intended to answer. Without such a statement, it is impossible to evaluate the
effectiveness of the monitoring program.

Identify Stressors and Develop Conceptual Model

A conceptual model provides an overview of scientific understanding of how the

ecosystem or population works and the state variables that describe the system. This
knowledge is used to determine what measures of system performance are likely to be
useful. The conceptual model should also include the effects of natural and human-induced
stressors on the system. Measurements and inferences to biological systems are affected by
the scale of observation, thus the temporal and spatial scales at which processes operate and
populations and communities respond must be estimated and clearly identified in the
conceptual model (Noon et al. 1999).

Select Indicators and Establish Sampling Design

No monitoring design can feasibly encompass all ecological processes and species
(Bradshaw et al. 2000). Therefore, the design of monitoring programs requires careful
consideration of candidate species and processes for measurement. Selection of variables
should be made in the light of the overall goals and underlying conceptual model of the
ecosystem of interest (NRC 2000). Although biodiversity has many dimensions, since we
are limited in the number of things we can measure, and there is pressure for quick answers,
it is popular to rely on monitoring surrogates such as indicators, keystones, or umbrella
species (Landres et al. 1988, Simberloff 1998, Andelman and Fagan 2000).

Despite their popularity, particular umbrella and indicator species have often been
chosen as management and monitoring surrogates without adequate testing of their
performance (Simberloff 1998; Caro and O’Doherty 1999). This, in turn has led to
confusion about when a species is being monitored for its own sake, and when a species can
appropriately function as an accurate indicator of other organisms or environmental
attributes. Where the performance of surrogates has been evaluated, results suggest that the
utility of existing approaches to selecting surrogates is limited and almost certainly scale-
dependent (Andelman and Fagan 2000; Williams et al. in press). Another analysis,
focusing on the Northern Spotted Owl, suggests that this species may act as an adequate
habitat surrogate for many other species (Noon and Bingham, in press).

Despite concerns, the fact remains that it is impossible to assess the status and trend
of all species of interest and that some surrogate-based approach is necessary. In addition,
some land management agencies, such as the U.S. Forest Service, are legally required to use
“management indicator species” to assess the impacts of any proposed management action
on the system as a whole. This reality lead a recent Committee of Scientists to propose the
use of focal species, defined as those species whose status and trend allow insights to the
integrity of the larger ecological system to which they belong. However, the Committee
noted that currently there is no definitive algorithm for unambiguously identifying focal
species. This remains a priority area for accelerated ecological research.
12
Possingham et al. Conservation Decisions September 2000
Define Methods of Analysis and Ensure there is a Link from Monitoring to
Decision-making

For monitoring programs to be useful they must be efficient, informative and

reliable (Thomas 1996). For example, given an objective of detecting a population decline,
and a fixed level of effort or resources, how many sites should be monitored? How many
samples are required to detect a trend in population size when there is uncertainty in
observations and significant environmental variation? How likely is it that a particular
sampling procedure will detect a rare species?

Most frequently used statistical techniques do not answer these questions because
they are rooted in the tradition of null-hypothesis testing. Typical management programs
hinge on the implicit assumption that if no problem is observed, none exists. The burden of
proof rests with monitoring programs. If monitoring programs fail to detect an impact or a
rare species, these impacts are assumed to be absent. In these circumstances, the reliability
of the monitoring system becomes critically important. This reliability, in turn, depends on
statistical power, the ability of a method to detect real outcomes, often against a background
of natural environmental variation, measurement error, and ignorance of biological
processes.

Given the importance of reliable monitoring, the persistent failure of conservation

biologists and ecologists to explicitly incorporate reliability considerations in the design of
monitoring programs is notable (Fairweather 1991, Mapstone 1995). The task of
stipulating an appropriate level of statistical power and an acceptable effect size is not
simply a statistical decision. It entails judgments about the biological importance of an
effect (Mapstone 1999). These judgments may, in part, be influenced by the social,
political and economic implications of the ensuing management or policy decisions. Such
considerations are critically important because analysis indicates that without reliability
calculations, experimenters often are overly optimistic about the reliability and
representativeness of their samples. The risk of false optimism may be compounded by
experimental designs that fail to account for the independence of replicates (Hulbert 1984).

These observations highlight a general failing in the interpretation of scientific data.

The implications of this failing are especially important in conservation biology, where the
consequences of wrong decisions may be irreversible (i.e. destruction of critical habitat or
extinction of species). Thus, conservation biologists should seek to apply methods that
minimize incorrect decisions. Such methods are readily accessible (Fairweather 1991,
Thomas 1996), yet seldom applied.

One potential explanation for this lack of application suggests that different sorts of
research may be needed if we are to improve the effectiveness of monitoring programs and
the linkages between monitoring, decision-making, conservation management and policy.
Research on human judgment under uncertainty suggests that humans have a widespread
tendency to perceive patterns where none exist (Redelmeier and Tversky 1996). Thus,
efforts to link monitoring programs to conservation decision-making may need to explicitly
consider such fundamental human psychological constraints.

Recently, there has been a chorus of calls for establishment of global monitoring of
biodiversity trends. While well intentioned, such programs have the potential to waste time

13
Possingham et al. Conservation Decisions September 2000
and use limited resources. Continued use and advocacy of unproven methods, regardless of
scale, serve only to undermine the credibility of conservation biology as a legitimate
scientific discipline, in academic, policy, and other public fora. More seriously, given the
magnitude of current rates of human exploitation of biodiversity, a consensus is needed as
to priority questions monitoring should aim to answer. Once those questions have been
identified, the development of credible, extensive, and selectively intensive monitoring is
essential, to provide warnings and adequate opportunities to avoid biotic catastrophe
(Beddington 1995).

Future directions for research

The examples provided here illustrate a few specific ways in which decision theory
and monitoring can apply to conservation on the ground. The application of decision theory
to identifying solutions and evaluating alternatives is limited only by the imagination of
conservation practitioners. Virtually every practical, real-world problem in nature
conservation would benefit from application of a decision theory and adaptive management
approach. Several aspects of decision theory as applied to conservation problems would
benefit from further research to strengthen its potential as a practical tool for resource
managers and conservation biologists. These include:

1. Develop ways to assess social values and use them to define composite utilities to allow
for the complex integration of ecology, social sciences and economics in specific
situations. This has rarely been achieved and most of the examples described here have
dealt with simple biodiversity objectives, or multiple biodiversity objectives, while
blithely ignoring the realities of real world decision-making. Multi-disciplinary
research in this area would be fruitful, yet the tradition of specialization within both the
natural and social sciences may mitigate against tackling these hard problems.

2. Develop models that can tractably evaluate multiple biological and social objectives
that are measured in different currencies (e.g., representation, probabilities of
extinction, frequency and magnitude of disturbance events) and that may also interact
with one another (e.g., cascading effects through food webs, differential influence of
habitat fragmentation on population densities of species). With such models, decision
theory can be used to assess optimal strategies for achieving multiple objectives that
collectively are considered to be integrated indices of, for example, ecological health or
biological integrity. In this way, conservation strategies can begin to move away from
its tendency toward single-species management.

3. Develop models that allow for the assessment of the trade-offs involved for a wider, but
more realistic, range of management options than has been accomplished to date. For
example, with respect to the implementation of an ecological reserve system, such a set
of realistic options might include acquisition of a new reserve for the system, enlarging
an existing reserve, or restoring an existing reserve to improve its quality. Moreover,
the value of each of these management options may change depending on the timing of
implementation. These same approaches could then be used to evaluate a wider range of
management options for an existing reserve, independent of other reserves in the
system, such as revegetation, species reintroduction, eradication of exotics, road
closure, and restoration of disturbance regimes.

14
Possingham et al. Conservation Decisions September 2000
4. Explicitly test a specific decision theory model, both in terms of whether the
hypothetical optimal solution generally performs better than random, status-quo or no
management. This would be impossible for some objectives, like minimizing
extinction probabilities. However, it would be possible for testing management
activities that are repeated frequently in space and time, like weed management.

5. Develop the tools that can tractably evaluate the relative sensitivity of the solutions to
uncertainty in the true state of the system and the dynamics of the state variables. This
is important for giving resource managers insight to how robust predicted solutions are
and where priorities should be placed for collecting additional data.

6. Develop user interfaces that will allow resource managers to change models and
quickly generate new solutions. Those who will use the models to generate solutions
will rarely be the ones to develop the models in the first place, yet they need to be able
to alter models to input new goals, new management options, new values for state
variables, new descriptions of the dynamics among state variables, and new constraints.
This happens not only because managers want to evaluate alternative scenarios and the
robustness of solutions, but also because the political and economic realities of
management never remain constant.

7. Develop and test rules for the selection of surrogate and indicator species. Since we
can feasibly measure only a handful of environmental attributes, we urgently need a set
of robust rules to help determine which attributes will be most informative.

8. Develop and test rules of thumb for conservation practice and implementation of active
adaptive management. Conservation practitioners must make many decisions and their
resources for modeling, experimental design and evaluation are limited. While it is
important for conservation biology researchers to provide guidance on best practice in
decision-making and monitoring, we must also deliver robust rules of thumb that can be
generically applied and improve the consequences of conservation management in the
short term.

Acknowledgements
We are very grateful for the ideas of fellow workshop participants, the editors, and others
whom have commented on the paper: Dee Boersma, Drew Tyre and Bill Murdoch.

References
Alford, R.A. and S. Richards. 1999. Global amphibian declines: a problem in applied
ecology. Annual Review of Ecology and Systematics 30: 133-165.

Andelman, S.J. and W.F. Fagan. 2000. Umbrellas and flagships: efficient conservation
surrogates or expensive mistakes? Proceedings of the National Academy of Sciences
97:5954-5959.

Andelman, S.J. and E. Meir. In press. Breadth is better than depth: Biodiversity
requirements for adequate reserve networks. Science.

Beddington, J. 1995. Fisheries – the primary requirements. Nature 374:213-214.

15
Possingham et al. Conservation Decisions September 2000
Beenhakker, H.L. 1975. Capital investment planning for management and engineering.
Rotterdam: Rotterdam University Press.

Beissinger, S. R. and Westphal, M. I. 1998. On the use of demographic models of

population viability in endangered species management. J. Wildlife Management 62:821-
841.

Blaustein, A., D. Wake and W. Sousa. 1994. Amphibian declines: judging stability,
persistence and susceptibility of populations to local and global extinction. Conservation
Biology 8: 60-71.

Bradshaw, G.A., M.-J. Fortin, and H.C. Biggs. In review. Large-scale monitoring of
terrestrial and aquatic systems: matching what science can do with what society wants.
Conservation Biology.

Caro,T.M. and G. O’Doherty. 1999. On the use of surrogate species in conservation

biology. Conservation Biology 13:805-814.

Clemen, R.T. 1996. Making Hard Decisions: An Introduction to Decision Analysis.

Duxbury Press, Pacific Grove, CA.

Dayton, P. K., Tegner, M. J., Edwards, P. B. and Riser, K. L. 1998. Sliding baselines,
ghosts, and reduced expectations in kelp forest communities. Ecological Applications
8:309-322.

Fairweather, P. G. 1991. Statistical design and power requirements for environmental

monitoring. Australian Journal of Marine and Freshwater Research 42:555-567

Ferson, S. and M.A. Burgman. 2000. Quantitative Methods for Conservation Biology.
Springer-Verlag.

Guikema, S. and Milke, M. 1999. Quantitative decision tools for conservation programme
planning: practice, theory and potential. Environmental Conservation 26:179-189.

Hanski, I. 1999. Metapopulation Ecology. Oxford University Press, Oxford.

Hulbert, S.H. 1984. Pseudoreplication and the design of ecological field experiments.
Ecological Monographs 54:187-211.

Intriligator, M. D. 1971. Mathematical optimisation and economic theory. Prentice-Hall,

Englewood Cliffs, NJ.

James, F.C., C.E. McCullough and D.A. Wiedenfield. 1996. New approaches to the
analysis of population trends in land birds. Ecology 77:13-27.

Kareiva, P.K., S.J. Andelman, D. Doak, B. Eldred, M. Groom, J. Hoekstra, L. Hood, F.

James, J. Lamoreux, G. LeBuhn, C. McCulloch, J. Regetz, L. Savage, M. Ruckelshaus, D.
Skelly, H. Wilbur, K. Zamudio, and the NCEAS HCP Working Group. 1999. Using science
in habitat conservation plans. NCEAS and AIBS, 93 pages.

16
Possingham et al. Conservation Decisions September 2000
Kulkarni, R.B., R.L. Burns, J. Wright, B. Apper, T.O. Baily and S.T. Noaek. 1993.
Decision analysis of alternative highway alignments. Journal of Transportation
Engineering 119:317-332.

Landres, P., J. Verner and J. Thomas. 1988. Conservation Biology 2:316-328.

Lubow, B. C. 1996. Optimal translocation strategies for enhancing stochastic

metapopulation viability. Ecological Applications 6:1268-1280.

MacArthur, R.M. and E.O. Wilson. 1967. The theory of island biogeography. Princeton
University Press.

Maguire, L. A., Seal, U. S. and Brussard, P. F. 1987. Managing a critically endangered

species: the Sumatran Rhino as a case study. Pages 141-158 in “Viable populations for
conservation.” Ed M. E. Soule, Cambridge University Press, Cambridge, UK.

Maguire, L.A. and Servheen, C. 1992. Integrating biological and social concerns in
endangered species management: augmentation of grizzly bear populations. Conservation
Biology 6:426-434.

Mapstone, B. D. 1995. Scalable decision rules for environmental impact studies – effect
size, type-I and type –II errors. Ecological Applications 5:401-410.

Margules, C. R. and Pressey, R.L. 2000. Systematic conservation planning. Nature

405:243-253.

Meffe, G.K. and C.R. Carroll. 1997. Principles of Conservation Biology, 2nd edition.
Sinauer and Associates.

Meir, E. and S. J. Andelman. In press. Biodiversity data requirements for adequate reserve
system design. Ecological Applications..

Mulder, B.S., B.R. Noon, C.J. Palmer (and others), tech. coords. 1999. The strategy and
design of the effectiveness monitoring program for the Northwest Forest Plan General
Technical Report. PNW-GTR.

Noon, B.R., and B.B. Bingham. In press. The Northern Spotted Owl as an umbrella species
for late seral forests of the Pacific Northwest. Single-species Approaches to Ecosystem
Conservation, P. Foster-Tukey, P. Kelly, and R. Medellin, eds. Island Press.

Noon, B.R., T.A. Spies, and M.G. Raphael. 1999. Conceptual basis for designing an
effectiveness monitoring program. Pp. 17-39 In: Mulder, B.S., B.R. Noon, C.J. Palmer (and
others), tech. coords. The strategy and design of the effectiveness monitoring program for
the Northwest Forest Plan. USDA, Forest Service General Technical Report. PNW-GTR.

NRC. 2000. Ecological Indicators for the Nation. National Academy Press, Washington,
D.C.

Parma, A., Amarasekare, P., Mangel, M., Moore, J., Murdoch, W. W., Noonburg, E.,
Pascual, M. A., Possingham, H. P., Shea, K., Wilcox, W. and Yu, D. 1998. What can
adaptive management do for our fish, food, forests and biodiversity. Integrative Biology

17
 Possingham et al. Conservation Decisions September 2000
1:16-26.

Possingham, H. P. 1997. State-dependent decision analysis for conservation biology. Pages

298-304 in The ecological basis of conservation: Heterogeneity, ecosystems and
biodiversity Eds S.T.A. Pickett, R.S. Ostfeld, M. Shachak, and G.E. Likens, Chapman and
Hall, New York.

Possingham, H. P., Ball, I. R. and Andelman, S. 2000. Mathematical methods for reserve
system design. Pages 291-306 In: Quantitative methods for conservation biology. Ferson,
S. and Burgman, M. (eds). Springer-Verlag, New York.

Pulliam, H.R. 1988. Source, sinks and population regulation. The American Naturalist
132:652-661.

Ralls, K. and Ballou, J. 1986. Captive breeding programs for populations with a small
number of founders. Trends in Ecology and Evolution 1:19-22.

Ravirala, V. and D.A. Grivas. 1995. Goal-programming methodology for integrating

pavement and bridge programs. Journal of Transportation Engineering 121:345-351.

Redelmeier DA, Tversky A 1996. On the belief that arthritis pain is related to the weather.
Proc. Nat. Acad Sci (USA) 93:2895-2896.

Richards, S.A., Possingham, H.P. and Tizard, J. 1999. Optimal fire management for
maintaining community diversity. Ecological Applications 9:880-892.

Shea, K., Amarasekare, P., Mangel, M., Moore, J., Murdoch, W. W., Noonburg, E., Parma,
A., Pascual, M. A., Possingham, H. P., Wilcox, W. and Yu, D. 1998. Management of
populations in conservation, harvesting and control. Trends in Ecology and Evolution
13:371-374.

Simberloff, D. 1998. Flagships, umbrellas and keystones: Is single species management

passe in the landscape era? Biological Conservation 83:247-257.

Simberloff, D. 1999. The role of science in the preservation of forest biodiversity. Forest
Ecology and Management 115:101-111.

Teng, J. and G. Tzeng. 1996. A multi-objective programming approach for selecting non-
independent transportation investment alternatives. Transportation Research 30B:291-307.

Thomas, L. 1996. Monitoring long-term population change: why are there so many analysis
methods? Ecology 77:49-58.

Thompson, W.L., G.C. White, and C. Gowan. 1998. Monitoring Vertebrate Populations.
Academic Press, New York.

Williams, P.H., N.D. Burgess, and C. Rahbek. In press. Flagship speciesm ecological
complementarity, and conserving the diversity of mammals and birds in sub-Saharan
Africa. Animal Conservation.

18

View publication stats