209
PERSPECTIVE / PERSPECTIVE
The opportunity cost of information: an economic
framework for understanding the balance
between assessment and control in sea lamprey
(Petromyzon marinus) management
Eli P. Fenichel and Gretchen J.A. Hansen
Abstract: Fisheries managers must make trade-offs between competing management actions; however, the inherent tradeoffs associated with information gathering are seldom explicitly considered. Incorporating economics into management decisions at the outset can aid managers in explicitly considering the trade-off between collecting more information to guide
management and taking management actions. We use control of the invasive sea lamprey (Petromyzon marinus) in the
Laurentian Great Lakes to illustrate how budget constraints shape this trade-off. Economic theory is used to frame previous empirical work showing that reducing the allocation of resources to conducting assessment, and thereby freeing resources for treatment, would result in a greater reduction of sea lamprey populations — the overarching management
objective. The optimal allocation of resources between assessment and control depends on the total budget, the relative
cost of each management activity, the marginal reduction in uncertainty associated with increased assessment, and the marginal effectiveness of increased treatment. Formal incorporation of prior information can change the optimal allocation of
resources. The approach presented here is generally applicable to a wide range of fishery management and research questions.
Résumé : Les gestionnaires des pêches sont appelés à faire des compromis entre des activités concurrentes de gestion; ils
ne tiennent cependant que rarement compte de façon explicite des compromis inhérents à la cueillette de renseignements.
L’incorporation dès le départ d’une dimension économique dans les décisions de gestion peut aider les gestionnaires à envisager explicitement le compromis entre la cueillette additionnelle de données pour guider la gestion et la mise en opération d’activités de gestion. Le contrôle de la grande lamproie marine (Petromyzon marinus) qui a envahi les Grands Lacs
laurentiens nous sert à illustrer comment les contraintes budgétaires modulent ce compromis. La théorie économique permet d’encadrer le travail empirique antérieur afin de démontrer qu’une diminution des ressources allouées aux évaluations,
qui aurait ainsi libéré des ressources pour le traitement, aurait mené à une plus grande réduction des populations de grandes lamproies marines — l’objectif prépondérant de la gestion. L’allocation optimale des ressources entre l’évaluation et
le contrôle dépend du budget total, du coût relatif de chacune des activités de gestion, de la réduction marginale de l’incertitude reliée à une évaluation additionnelle et de l’efficacité marginale d’un traitement supplémentaire. Une incorporation
formelle de renseignements a priori peut modifier l’allocation optimale des ressources. La méthodologie présentée ici peut
s’appliquer de manière générale à une gamme étendue de questions de gestion et de recherche relatives à la pêche.
[Traduit par la Rédaction]
Introduction
Efficient resource management requires that management
objectives be achieved at least cost. To this end, managers
aim either to minimize costs while achieving an objective
or to maximize the production of a ‘‘good’’ (e.g., fish
stocked, pests killed, or ecosystem services provided), subject to a set budget. Explicitly considering costs of achieving
ecological objectives requires integrating economics in its
Received 11 March 2009. Accepted 23 September 2009. Published on the NRC Research Press Web site at cjfas.nrc.ca on 17 December
2009.
J21100
E.P. Fenichel.1 Arizona State University, School of Life Sciences, Box 874501, Tempe, AZ 85287, USA.
G.J.A. Hansen.2 Quantitative Fisheries Center and Department of Fisheries and Wildlife, Michigan State University, 13 Natural
Resources Building, East Lansing, MI 48824, USA.
1Corresponding
2Present
author (e-mail: [email protected]).
address: Center for Limnology, University of Wisconsin – Madison, 680 North Park Street, Madison, WI 53706, USA.
Can. J. Fish. Aquat. Sci. 67: 209–216 (2010)
doi:10.1139/F09-163
Published by NRC Research Press
210
most fundamental role as a decision science with ecology
and natural resources management. Economics is the study
of resource allocation and frequently has been used in fisheries management (e.g., Clark 2005; Costello et al. 2008).
Costs are a fundamental part of natural resource management planning (Polasky 2006), and recently there has been
a push for economics to play a stronger role in natural resource management (Shogren et al. 1999). A key component
of an economic analysis is to measure costs in terms of the
lost opportunities that occur when resources are allocated to
a certain activity.
Trade-offs between the assessment and the management
of a system are common, but seldom explicitly considered
(Hansen and Jones 2008a). Information is required to manage fisheries efficiently, and researchers and managers alike
often call for more information to guide decision-making.
These calls for more information may fail to consider the
trade-off between collecting more information (i.e., research
and assessment) and taking other management actions.
Under fixed budget conditions, any investment in information gathering detracts from resources available for other
management actions. Accordingly, it can be unclear how
many resources should be allocated to collecting information
versus other areas of management.
Hansen and Jones (2008a) discuss the trade-off between
information gathering and management in aquatic systems
and encouraged an explicit consideration of this and other
trade-offs when making resource management decisions. In
some cases, it may be important to delay management actions in favor of data collection (see Fenichel et al. (2008)
for an example of when this might be preferable); in other
cases, uncertainty and variability are sufficiently high that
no reasonable amount of research yields the level of knowledge required to act with confidence (Ludwig et al. 1993;
Johannes 1998). At some point, resources used for learning
and assessment would provide higher returns, in terms of
the management objective, if allocated to other management
activities. Hansen and Jones (2008b) found that control of
Great Lakes sea lampreys (Petromyzon marinus) could be
improved by allocating fewer resources to assessment activities and more to chemical control.
In this article, we provide theoretical motivation for Hansen and Jones’ (2008b) findings that likely applies to other
fishery management cases. We explicitly incorporate the
costs of assessment and treatment and apply an economic
production framework to explore the theoretical basis of the
trade-off between the two. We represent the costs of management activities as forgone opportunities; resources used
to fund one activity are not available to fund others under a
limited budget. The cost of each activity includes the value
of lost alternatives, not just the dollars spent. We use Great
Lakes sea lamprey control to broadly illustrate the trade-off
between gaining information (assessment) and taking a management action (treating streams with lampricide) and discuss conditions that favor each activity.
Sea lamprey background
Sea lampreys are invasive to the North American Great
Lakes, and their impact on native fishes is well documented
(e.g., Smith and Tibbles 1980; Heinrich et al. 2003). An in-
Can. J. Fish. Aquat. Sci. Vol. 67, 2010
tensive control program targets Great Lakes sea lampreys to
reduce sea lamprey populations and achieve fish community
objectives (Great Lakes Fishery Commission (GLFC) 2001).
Sea lamprey control is chiefly achieved through the periodic
treatment of streams with chemical lampricides that target
stream-dwelling larvae and recently metamorphosed larvae
(known as transformers), before they enter the Great Lakes
and become parasites (Smith and Tibbles 1980). Lampricide
is applied to streams on a cycle corresponding to the streamspecific duration of the sea lamprey larval phase. Natural
demographic variation makes it impossible to perfectly predict when each stream will require treatment; therefore, candidate streams are assessed annually to determine the larval
population and size structure (Slade et al. 2003). Assessments are used along with a model-based decision tool to
predict stream-specific transformer abundances in the following year (Christie et al. 2003). Streams are prioritized
for treatment based on the predicted number of transformers
killed per stream treatment cost. The operational goal of sea
lamprey management could be described as maximizing the
number of transformers killed given the current control
budget. Care is also taken to minimize nontarget effects of
lampricide (GLFC 2001); following Hansen and Jones
(2008b), we abstract from this issue.
A trade-off exists in sea lamprey control between allocating resources to stream assessment and stream treatment.
From 1995–2007, streams were assessed using quantitative
assessment sampling (QAS), a resource-intensive sampling
regime intended to provide accurate and absolute estimates
of larval abundance and size structure (Christie et al. 2003;
Slade et al. 2003; but for critiques, see Steeves 2002 and
Hansen et al. 2003). Resource-intensive assessments reduce
resources available to treat streams after they are ranked, all
else equal. Hansen and Jones (2008b) developed a less resource intensive alternative, rapid assessment (RA). RA provides an index of stream-specific lamprey populations. It is
less precise than QAS but frees resources to treat additional
streams. Hansen and Jones (2008b) found that RA outperformed QAS in terms of the number of sea lampreys killed
with identical overall program expenditures. In 2008, RA
was implemented for assessing and ranking Great Lakes
streams.
Trade-offs between information and action
We use the sea lamprey control example to illustrate the
trade-off between collecting information and actively managing ecological systems. We assume a fixed budget for sea
lamprey control and that the program’s operational objective
is to maximize the number of sea lamprey transformers
killed. This goal is achieved in two steps: (i) find transformers through assessment (a), and (ii) kill larvae and transformers with lampricide treatment (t). The two assessment
methods, QAS and RA, represent two different levels of assessment effort. QAS is more expensive, in terms of dollars,
but is assumed to have the potential to yield more (or higher
quality) information about the location of transformers. Determining which assessment method is ‘‘better’’ is a question
of maximizing information about transformer location conditional on the assessment budget. At the same time, the
size of the assessment budget depends on the resources allocated between treatment and assessment. The proportion of
Published by NRC Research Press
Fenichel and Hansen
the total budget that is optimally allocated to assessment depends on the total budget and the relative cost, in terms of
forgone opportunities, of treatment and of each form of assessment.
Assume that under foreseeable future conditions, the number of sea lamprey transformers killed in one year reduces
the number of parasites and spawning adults but does not affect the supply of larvae and transformers in subsequent
years due to high levels of compensation and densityindependent variation in recruitment (Jones et al. 2003;
Dawson 2007). Furthermore, assume that the information
gained from assessment in one year is not used in following
years. This latter assumption is a simplification but is justifiable if the information from prior years is not formally combined with the new information to inform management. In
this case, the decision-making process treats the system as
static.
Economic production theory relates to how a firm, or
management agency, can maximize production of a good
(e.g., dead transformers), which requires certain inputs (e.g.,
information from assessment and streams treated with lampricide), subject to a budget constraint. The price of each input is explicitly considered.
First, assume that a single assessment strategy is available. Both assessment and treatment likely have diminishing
marginal returns. No realistic assessment technique provides
complete information regarding which streams should be
treated, and therefore any assessment technology asymptotically approaches a maximum level of achievable information. Similarly, as more resources are allocated to treatment,
fewer transformers are killed per dollar. Increasing either
treatment or assessment leads to more transformers killed,
but given decreasing returns to scale for both inputs, the
number of transformers killed does not increase linearly
with an increase in either treatment or assessment.
Treatment and assessment are substitutes in production.
Killing transformers does not require a fixed ratio of treatment to assessment; rather, less treatment can be made up
for, to some extent, with more assessment (and vice versa).
Applying treatment in a small number of ‘‘correct’’ spots
kills as many transformers as treating a greater number of
‘‘incorrect’’ spots. Graphically, this is shown with an ‘‘isokill’’ curve or more generally with an iso-quant curve
(Fig. 1). Iso-kill curves represent combinations of assessment and treatment that result in the same number or level
of transformers killed, k. More transformers are killed for
iso-kill curves located higher and further to the right on the
figure. The exact shape of the iso-quant curves is an empirical question and an expanding area of research (e.g., Nelson
et al. 2009).
The iso-kill curves that represent sea lamprey transformer
kill levels can take on one of two qualitative forms. For the
first form, both treatment and assessment must be greater
than zero to kill sea lamprey transformers (e.g., locating
streams). This implies that one input can never fully replace
the other. The principles of integrated pest management
(IPM) suggest this form, because IPM requires quantitative
data to allocate control resources (Maitland 1980; Slade et
al. 2003). Iso-kill curves are shown in which some level of
both assessment and treatment is necessary (Fig. 1a). However, this is not necessarily the case. Treatments could be
211
applied randomly to subsets of streams in the absence of assessment and some sea lamprey transformers would still be
killed. Figure 1b illustrates potential iso-kill curves for the
case in which it is possible to kill sea lamprey with zero assessment.
The ratio at which one input can be substituted for another input (e.g., assessment for treatment) to achieve the
same level of production (e.g., dead transformers) is the
marginal rate of technical substitution (MRTS). The MRTS
can be derived by considering the marginal product, or contribution, of small increases in each input. The marginal
product of treatment is the change in the number of transformers killed for per unit treatment, qk/qt. The marginal
product of assessment is the change in the number of transformers killed per unit assessment, qk/qI qI/qa, where I is
the level of information generated by assessment. The ratio
of the marginal product of assessment to the marginal product of treatment, (qk/qa)/(qk/qt) = qt/qa, is equal to the
MRTS, which is the negative of the slope of the iso-kill
curve, –dt/da.
It is necessary to know the prices of assessment and treatment and the total sea lamprey control budget, b, in addition
to the MRTS to find the optimal balance between assessment and treatment. The budget is exclusive; resources spent
on assessment cannot be used for treatment. Define a budget
line, b = ptt + paa, where pt and pa represent the respective
price of treatment and assessment (in dollars). If the entire
budget were spent on assessment, b/pa assessments would
be possible. If the entire budget were spent on treatment,
b/pt treatments would be possible. The budget line can be
graphed on the same axes as the iso-kill curves and has a
slope of –pa/pt (Fig. 1). This ratio of prices of inputs pa/pt
represents the opportunity cost of one input in terms of the
other, because it represents what must be forgone in terms
of other management opportunities.
At the maximum achievable kill level, the MRTS and
price ratio will be exactly equal (Fig. 1). The greatest number of transformers that can be killed for a given budget is
achieved when the budget line is tangent to the iso-kill
curve (Fig. 1). The optimal level of assessment and treatment is the combination of assessment and treatment at the
tangency. Other combinations of assessment and treatment
effort are possible under the same budget but result in lower
numbers of transformers killed.
If some transformers can be killed with no assessment,
then it is possible for an optimal allocation that only funds
treatment (Fig. 1b). If the price of assessment, pa, is large
enough relative to pt, then a tangency cannot occur (Fig. 1b,
budget b). Here we would have to give up too much treatment to afford any assessment. Even if the price of assessment is less than treatment, assessment is relatively costly
in terms of forgone treatment opportunities. However, if the
budget is increased, it may become optimal to invest in assessment, even though some lamprey transformers could be
killed with no assessment (Fig. 1b, budget b’).
Trade-offs between assessment protocols
When more than one assessment method is available, it is
important to consider the trade-off between methods. Consider the choice between QAS and RA. Assume that QAS
has a higher maximum level of achievable information than
Published by NRC Research Press
212
Can. J. Fish. Aquat. Sci. Vol. 67, 2010
(a)
b
pt
k′
Treatment
k
b
pa
(b)
pt
k ′′
b
RA
QAS
Resources used in assessment
k ′′
b′
Fig. 2. The relationship between two larval lamprey assessment
methods: QAS, which potentially provides more information but
requires more units of assessment to do so (black line); and RA,
which can provide more information at low levels of assessment
but cannot provide as much information as QAS at high levels of
assessment (shaded line). The vertical line represents the switching
point between RA and QAS.
Information acquired
Fig. 1. Iso-kill curves (solid curves), assuming a single type of assessment and a budget constraint (broken line). Along a given isokill curve, the same number of sea lamprey transformers are killed,
with kill level k < k’ < k@. The budget constraint line (broken line)
is a graphical representation of the sea lamprey control budget, b,
with b’ > b and treatment and assessment prices pt and pa, respectively. A budget line slope < –1 indicates that treatment has a
higher unit price than assessment. The optimal allocation of assessment and treatment is where the budget line is tangent to the highest achievable iso-kill curve (k’ in panel a). Various combinations
of assessment and treatment could kill fewer sea lamprey transformers (k), but higher levels of killed sea lamprey transformers (k@)
are not achievable given the budget, b. (a) The case in which both
treatment and assessment are necessary to kill sea lamprey transformers; (b) the case in which some transformers could be killed
with zero assessment.
k′
pt
k
b
b′
pa
Assessment
pa
RA, but the level of information increases more rapidly for
each unit of RA at lower levels of assessment (Fig. 2). If assessment expenditures are low (area to the left of the vertical line in Fig. 2), then RA provides more information about
which streams to treat, but QAS provides more information
as assessment expenditure increases (area to the right of the
vertical line in Fig. 2). Therefore, under certain budgetary
allocations, RA provides more information, whereas under
other allocations, QAS provides more information. Alternative treatment protocols could pose similar trade-offs.
The iso-kill curves in Fig. 3 qualitatively show combinations of treatment and assessment effort associated with
QAS (black curves) and RA (shaded curves) that result in
the same total number of sea lamprey transformers killed.
The RA and QAS iso-kill curves are shaped differently for
the same kill level because QAS and RA produce different
amounts of information per dollar spent. QAS and RA result
in different MRTSs. Assume that the iso-kill curves associated with QAS and RA for a given level of transformers
killed intersect each other. This implies that under some
conditions, QAS is preferred, and under others, RA is preferred. However, it is possible that QAS or RA could always
be better at producing information. In such a case, the manager only has one trade-off to consider — the trade-off between treatment and assessment. With a small budget
allocation to assessment, an approach employing RA can
kill more transformers than an approach employing QAS
for the same budget. Notice that the RA iso-kill curve is below the QAS curve to the left of the intersection point
(Fig. 3). In this case, the opportunity cost of treatment, in
terms of foregone assessment, is relatively small; relatively
few assessment units must be forgone to gain a treatment
unit. An additional unit of treatment has a greater effect on
the number of sea lamprey transformers killed than would
an additional unit of assessment. Conversely, if the opportunity cost of assessment were high in terms of forgone treatment opportunities, then an additional unit of assessment
does not offset the reduction in killed transformers from forPublished by NRC Research Press
Fenichel and Hansen
Fig. 3. Comparison of two QAS (black) and their respective RA
(shaded) iso-kill curves at kill levels k and k’, where k’ > k. The
point k~ indicates where a combination of treatment and assessment
results in the same number of killed transformers with either QAS
0
~ The QAS and RA lines that cross at k~ or k~ 0 have the
or RA, k~ > k.
same kill levels, respectively. The point at which the budget line
(broken line) for budget level b is tangent to the highest achievable
iso-kill line is the optimal combination of assessment and treatment. (a) The budget line for assessment price pa is illustrated. The
budget constraint is tangent to the RA iso-kill curve, k, but does not
intersect the QAS iso-kill curve associated kill level k, indicating
that RA is the preferred assessment method given assessment cost
pa; (b) the budget line (broken line) for assessment price pa 0 is illustrated, where pa 0 < pa and the arrow is used to indicate that b=pa 0
is farther to the right on the x axis. In this scenario, the budget
constraint is tangent to the QAS iso-kill curve at kill level k’ but
does not intersect the k’ RA iso-kill curve. Therefore, QAS is the
preferred assessment method at assessment price pa 0 . The price of
treatment is pt and does not change across the two panels.
going a unit of treatment. Finally, when fewer assessment
units are employed, RA collects information ‘‘faster’’
(Fig. 2), and RA is preferred.
The relative cost of each activity determines the circumstances under which each assessment method is preferred. If
the budget line is tangent to the iso-kill curve at the point at
which the QAS and RA iso-kill curves cross each other,
then the optimal allocation to assessment and treatment is
the same regardless of the assessment method. Call this
~ Two kill levels, k < k’, are illustrated (Fig. 3), with
point k.
the iso-kill curves k and k’ labeled by their respective points
0
k~ and k~ . Moving from left to right, the RA iso-kill curve
213
starts below and to the left of the QAS iso-kill curve for the
same kill level. After the curves cross, the RA iso-kill curve
is above and to the right of the QAS iso-kill curve. The isokill curves also become ‘‘flatter’’, implying that increasing
assessment has a decreasing ability to offset forgone treatment. This is especially important in the region to the right
~ where QAS maintains a greater ability to offset forof k,
gone treatment in this region. Consider a system in which
the assessment–treatment price ratio, pa/pt, is greater than
the marginal effect of increasing treatment on the MRTS at
the point at which both allocation to treatment and assessment would be the same under both QAS and RA
(qMRTSQAS/qa at k~ < pa/pt). In this case, RA is preferred
(Fig. 3a). However, as the price of assessment, pa, decreases
to pa’, the slope of the budget line decreases in absolute
value (Fig. 3b). This has two effects. First, assuming that
some assessment is optimal if the assessment price is pa’,
then it makes sea lamprey managers wealthier, and they can
kill transformers regardless of allocations between assessment and treatment. The second effect is on the substitution
between treatment and assessment. Assessment becomes less
expensive, and the proportion of the budget allocated to assessment increases. If this substitution effect is great
enough, then the preferred assessment approach could
change (Fig. 3b). Decreases in the price of treatment (pt)
have the reverse effect.
The condition pa < pt does not guarantee that QAS is preferred. It is the prices relative to the marginal products —
the opportunity cost, not the absolute price — that is important. For the sea lamprey case and for other real management problems, assessment likely has a lower price than
action, but assessment may have greater opportunity costs.
The preferred assessment method and optimal allocation between assessment and treatment depends on the marginal
products of assessment and treatment, the relative price of
each activity, the information gathered for each unit of each
assessment method, and the total budget. When all of these
factors are considered, the optimal budget allocation to each
activity can be determined.
Accounting for prior information
Prior information may influence management decisions,
but neither QAS nor RA explicitly incorporates prior information. Both QAS and RA could be modified to explicitly
account for information collected in previous time periods,
and additional alternative assessment procedures could be
developed that explicitly include prior information (see Anderson 2006).
Trade-offs between optimal learning and action are an
emerging area of research (e.g., Springborn and Costello
2010). To fully characterize the trade-offs between action
and learning over time requires stochastic dynamic programming (Clark 2005) or approximations to stochastic dynamic
programming (Powell 2007). The inclusion of prior information necessarily alters the marginal production of information over time. A full exploration of these ideas is beyond
the scope of this article; however, we make some qualitative
comments about how the formal incorporation of prior information could affect the relationships described above.
Assume that prior information, I, can be used to inform
current decisions and that the prior information is not misPublished by NRC Research Press
214
leading. Any combination of I and current assessment must
provide at least as much information as the current assessment alone, i.e., Ci(a, I) ‡ Ci(a), where C quantifies total
new information and i indexes the assessment approach
(e.g., QAS or RA). This may take the form Ci(a, I) =
Ci(a) + Fj(I), implying that information is a function of the
information provided by the current assessment, Ci(a), plus
some function, Fj, that fully characterizes the usefulness of
the prior information available in the current period, where
j indexes the source of prior information. Prior information
may have been collected through previous RA or QAS, a
mix of RA and QAS, other assessment types, and expert
opinion. The functional form of F is unknown and may depend on the type and quality of information and how it was
gathered.
It is possible to modify the MRTS to MRTS’ to account
for prior information. Assuming that the system is at equilibrium so that assessment effort is constant over time, at =
at–1 = a, and assessment is conducted such that information,
or the level of knowledge about the system, is just kept current (i.e., usefulness of prior information does not change
over time, Fj,t(Ct–1) = Fj,t–1(Ct–2) = F, where t indicates the
time period). Furthermore, the biological nature of the system is approximately constant. If F > 0, then we generally
expect MRTS’ < MRTS, which may be interpreted as a decrease in the marginal product of new assessment. This is
analogous to a Bayesian updating process with a strong
prior. This means that treatment becomes less costly in
terms of forgone assessment, because the marginal contribution of new assessment to killing transformer sea lamprey is
less. Therefore, an additional unit of assessment is relatively
more costly in terms of forgone treatment. These conditions
favor a less expensive and less informative assessment
method. This results in a substitution effect from assessment
to treatment, all else being equal. However, there is also an
‘‘income’’ effect: treatment is effectively less costly when
prior information is used, and there are more resources to
~ it
allocate to both treatment and assessment. At the point k,
is possible that this income effect outweighs the substitution
effect, providing a wider range of conditions where QAS
would be preferred.
When the system is out of equilibrium, the MRTS’ will
change over time, and a dynamic programming approach
will be necessary. Allocations to assessment will have a current period production component and an investment component in future production of k. In this case, the usefulness
of information will change as assessment changes. Managers
will need to be concerned with allocations aimed at accumulating information over time in addition to trade-offs between assessment and treatment. Additional optimality
conditions are necessary to inform efficient intertemporal allocation (Clark 2005). Given the nonlinear nature of the production function, k, and the likely nonlinear nature of any
realistic specification of F, the intertemporal investment
problem is complex. The approach path to any equilibrium
will involve nonlinear dynamics. Assuming that F is monotonically increasing (i.e., prior information builds upon itself
and never decreases) implies a flattening of the MRTS’. For
constant relative costs for treatment and assessment, this
flattening may imply a switch over time from resourceintensive assessment methods (i.e., QAS) to less resource-
Can. J. Fish. Aquat. Sci. Vol. 67, 2010
intensive methods (i.e., RA). An optimal strategy of
switching technologies can result because early investments
in information carry value into the future, and such investments in information may help managers approach equilibrium faster. This depends on the nature of Fj(I). The
complexity of such problems makes it all the more important that trade-offs are explicitly considered.
Discussion
In this article, we use an economic framework to explain
the findings of Hansen and Jones (2008b) that sea lamprey
control is improved by allocating fewer resources to assessment, thereby freeing resources for chemical treatments. We
show how the optimal allocation to assessment depends on
the relationship between assessment and information, the total management budget, the cost of assessment relative to
other management actions, and the marginal products of assessment and treatment. Increasing the information obtained
per unit of assessment, the effectiveness of treatment, or the
price of treatment or assessment changes the optimal balance between the two management actions. For example,
we demonstrate graphically that a decrease in the price of
assessment could result in more transformers killed overall
and a change in the preferred assessment method from RA
to QAS.
The relative performance of an assessment technique can
also change when prior information is formally incorporated
into decision-making. Formally combining prior knowledge
with current assessments could benefit the management of a
variety of other systems, reduce overall costs, and increase
the return on fishery management investments. Bayesian decision analysis (e.g., Ellison 1996; Stauffer 2008) and expert
calibration (Lele 2004) quantitatively incorporate prior information. Use of traditional knowledge (e.g., Aswani and
Lauer 2006) or expert opinion could also increase the effective level of assessments.
The QAS method was developed in 1995 to guide lampricide treatment decisions (Slade et al. 2003). Following a
decade of QAS surveys, an alternative assessment method
that used fewer resources was developed, RA, thereby increasing resources available for lampricide control and presumably killing more sea lampreys (Hansen and Jones
2008b). The development of RA was made possible by the
data collected through QAS. Sea lamprey managers
switched from QAS to RA to prioritize streams for chemical
treatment beginning in 2008, and this switch matches the rational decision process suggested by economic theory. The
GLFC did not use an economic model to guide these assessment decisions but has supported an adaptive learning and
research process. The similarity of the assessment decisions
made by the GLFC and predictions from economic theory
provides support for this adaptive learning approach but
also suggests that the adaptive learning approach could be
accelerated by formally incorporating economics.
In the case of sea lamprey management, there exists a direct and easily illustrated connection between the knowledge
gained from assessment and the application of alternative
management actions (lampricide treatments) that leads to
the realization of management goals. In other systems, the
connection may be less clear, in particular if assessment
Published by NRC Research Press
Fenichel and Hansen
and management actions are funded by separate agencies.
However, the same trade-off between assessment and management actions exists at some level in every system. State
and federal management agencies face similar trade-offs
every year when allocating limited budgets between monitoring, research, direct management interventions (e.g.,
stocking or culling), and enforcement.
Recommendations for more research are common in a variety of managed natural resource systems. All else being
equal, more information is always a welcome contribution
to fisheries management. However, all else is seldom equal,
and increased information comes at a cost. These costs may
be highly nontrivial when increasing research in one area
means diverting resources from on-the-ground management
or research in other areas. Furthermore, these costs should
be considered at the outset of management planning, not as
a final filtering step (Naidoo et al. 2006; Fenichel et al.
2010). We do not intend to imply that research and assessment are overvalued or overfunded; however, decisions to
fund these and other management activities should be made
with an acknowledgement of the trade-offs between potentially competing programs. The analysis presented here illustrates how an economic framework can be used to evaluate
assessment–management trade-offs and optimally allocate a
limited budget.
Increased information about a system may come at a cost
of decreasing other management actions. Of course, decreased short-term performance to achieve long-term goals
may be a reasonable trade-off. Indeed, this is the underlying
idea behind adaptive management (Walters 1986). Fisheries
science has moved towards explicit consideration of tradeoffs, in general, with increasing use of decision analysis and
related techniques to analyze intertemporal trade-offs (e.g.,
Robb and Peterman 1998; Punt 2006; Fenichel et al. 2008).
However, these techniques do not automatically account for
the budget constraints that managers always face. Distinct
but related literature has argued for increased assessment of
the potential payoff from research by considering the expected value of the information (e.g., McDonald and Smith
1997; Kim et al. 2003), whereas Gerber et al. (2005) focus
on assessing the marginal product of assessment. Such analyses are important in informing management trade-offs but
do not alleviate the need to explicitly consider budget constraints.
Acknowledgements
Funding for this study was provided by the Great Lakes
Fishery Commission and the Quantitative Fisheries Center
at Michigan State University. We thank Michael Jones, Don
Jackson, John Holmes, Dieter Busch, and two anonymous
reviewers for comments on an earlier draft. The US Fish
and Wildlife Service and Fisheries and Oceans Canada sea
lamprey assessment staff provided valuable insight into the
sea lamprey control program. This is contribution number
2009-21 of the Quantitative Fisheries Center at Michigan
State University.
References
Anderson, G.J. 2006. Improving larval sea lamprey assessment in
the Great Lakes using adaptive management and historical re-
215
cords. M.Sc. thesis, Department of Fisheries and Wildlife, Michigan State University, East Lansing, Michigan.
Aswani, S., and Lauer, M. 2006. Benthic mapping using local aerial photo interpretation and resident taxa inventories for designing marine protected areas. Environ. Conserv. 33(03): 263–273.
doi:10.1017/S0376892906003183.
Christie, G.C., Adams, J.V., Steeves, T.B., Slade, J.W., Cuddy,
D.W., Fodale, M.F., Young, R.J., Kuc, M., and Jones, M.L.
2003. Selecting Great Lakes streams for lampricide treatment
based on larval sea lamprey surveys. J. Gt. Lakes Res. 29(Suppl.
1): 152–160.
Clark, C.W. 2005. Mathematical bioeconomics optimal management of renewable resources. 2nd ed. John Wiley & Sons, Hoboken, New Jersey.
Costello, C., Gaines, S.D., and Lynham, J. 2008. Can catch shares
prevent fisheries collapse? Science (Washington, D.C.),
321(5896): 1678–1681. doi:10.1126/science.1159478. PMID:
18801999.
Dawson, H.A. 2007. Recruitment dynamics of Great Lakes sea
lamprey (Petromyzon marinus) populations and implications for
integrated pest management. Ph.D. dissertation, Michigan State
University, East Lansing, Michigan.
Ellison, A.M. 1996. An introduction to Bayesian inference for ecological research and environmental decision-making. Ecol. Appl.
6(4): 1036–1046. doi:10.2307/2269588.
Fenichel, E.P., Tsao, J.I., Jones, M.L., and Hickling, G. 2008. Real
options for precautionary fisheries management. Fish Fish. 9:
121–137.
Fenichel, E.P., Horan, R.D., and Hickling, G.J. 2010. Bioeconomic
management of invasive insect-vectored diseases. Biol. Invasions. In press.
Gerber, L.R., Beger, M., McCarthy, M.A., and Possingham, H.P.
2005. A theory for optimal monitoring of marine reserves. Ecol.
Lett. 8(8): 829–837. doi:10.1111/j.1461-0248.2005.00784.x.
Great Lakes Fishery Commission (GLFC). 2001. Strategic vision of
the Great Lakes Fishery Commission for the first decade of the
new millennium. Available at http://www.glfc.org/pubs/
SpecialPubs/StrategicVision2001.pdf [accessed 3 May 2009].
Hansen, G.J.A., and Jones, M.L. 2008a. The value of information
in fishery management. Fisheries, 33(7): 340–348.
Hansen, G.J.A., and Jones, M.L. 2008b. A rapid assessment approach to prioritizing streams for control of Great Lakes sea
lampreys (Petromyzon marinus): a case study in adaptive management. Can. J. Fish. Aquat. Sci. 65: 2471–2484. doi:10.1139/
F08-153.
Hansen, M.J., Adams, J.V., Cuddy, D.J., Richards, J.M., Fodale,
M.F., Larson, G.L., Ollila, D.J., Slade, J.W., Steeves, T.B.,
Young, R.J., and Zerrenner, A. 2003. Optimizing larval assessment to support sea lamprey control in the Great Lakes. J. Gt.
Lakes Res. 29(Suppl. 1): 766–782.
Heinrich, J.W., Mullett, K.M., Hansen, M.J., Adams, J.V., Klar,
G.T., Johnson, D.A., Christie, G.C., and Young, R.J. 2003. Sea
lamprey abundance and management in Lake Superior, 1957–
1999. J. Gt. Lakes Res. 29(Suppl. 1): 566–583.
Johannes, R.E. 1998. The case for data-less marine resource management: examples from tropical nearshore finfisheries. Trends
Ecol. Evol. 13(6): 243–246. doi:10.1016/S0169-5347(98)013846.
Jones, M.L., Bergstedt, R.A., Twohey, M.B., Fodale, M.F., Cuddy,
D.W., and Slade, J.W. 2003. Compensatory mechanisms in
Great Lakes sea lamprey populations: implications for alternative control strategies. J. Gt. Lakes Res. 29(Suppl. 1): 113–129.
Kim, J.B., Hobbs, B.F., and Koonce, J.F. 2003. Multicriteria Bayesian analysis of lower trophic level uncertainties and value of
Published by NRC Research Press
216
research in Lake Erie. Hum. Ecol. Risk Assess. 9(4): 1023–
1057. doi:10.1080/713610022.
Lele, S.R. 2004. Elicit data, not prior: on using expert opinion in
ecological studies. In The nature of scientific evidence. Edited
by M.L. Taper and S.R. Lele. The University of Chicago Press,
Chicago, Illinois. pp. 410–431.
Ludwig, D., Hilborn, R., and Walters, C. 1993. Uncertainty, resource exploitation, and conservation: lessons from history.
Science (Washington, D.C.), 260: 17, 36.
Maitland, P.S. 1980. Assessment of lamprey and fish stocks in the
Great Lakes in relation to control of the sea lamprey (Petromyzon marinus): report from the SLIS Assessment Measurements Task Force. Can. J. Fish. Aquat. Sci. 37(11): 2197–2201.
doi:10.1139/f80-262.
McDonald, A.D., and Smith, A.D.M. 1997. A tutorial on evaluating
expected returns from research for fishery management using
Bayes’ theorem. Nat. Resour. Model. 10: 185–216.
Naidoo, R., Balmford, A., Ferraro, P.J., Polasky, S., Ricketts, T.H.,
and Rouget, M. 2006. Integrating economic costs into conservation planning. Trends Ecol. Evol. 21(12): 681–687. doi:10.1016/
j.tree.2006.10.003. PMID:17050033.
Nelson, E., Mendoza, G., Regetz, J., Polasky, S., Tallis, H., Cameron, D.R., Chan, K.M., Daily, G.C., Goldstein, J., Kareiva,
P.M., Lonsdorf, E., Naidoo, R., Ricketts, T.H., and Shaw, R.
2009. Modeling multiple ecosystem services, biodiversity conservation, commodity production, and tradeoffs at landscape
scales. Front. Ecol. Environ, 7(1): 4–11. doi:10.1890/080023.
Polasky, S. 2006. You can’t always get what you want: conservation planning with feedback effects. Proc. Natl. Acad. Sci.
U.S.A. 103(14): 5245–5246. doi:10.1073/pnas.0601348103.
PMID:16567653.
Powell, W.B. 2007. Approximate dynamic programming, solving
the curses of dimensionality. John Wiley & Sons, Hoboken,
New Jersey.
Punt, A.E. 2006. The FAO precautionary approach after almost
10 years: have we progressed towards implementing simulation-
Can. J. Fish. Aquat. Sci. Vol. 67, 2010
tested feedback-control management systems for fisheries management? Nat. Resour. Model. 19(4): 441–464.
Robb, C.A., and Peterman, R.M. 1998. Application of Bayesian decision analysis to management of sockeye salmon (Oncorhynchus nerka) fishery. Can. J. Fish. Aquat. Sci. 55(1): 86–98.
doi:10.1139/cjfas-55-1-86.
Shogren, J.F., Tschirhart, J., Anderson, T., Ando, A.W., Beissinger,
S.R., Brookshire, D., Brown, G.M., Coursey, D., Innes, R.,
Meyer, S.M., and Polasky, S. 1999. Why economics matters for
endangered species protection. Conserv. Biol. 13(6): 1257–1261.
doi:10.1046/j.1523-1739.1999.98414.x.
Slade, J.W., Adams, J.V., Christie, G.C., Cuddy, D.W., Fodale,
M.F., Heinrich, J.W., Quinlan, H.R., Weise, J.G., Weisser, J.W.,
and Young, R.J. 2003. Techniques and methods for estimating
abundance of larval and metamorphosed sea lampreys in Great
Lakes tributaries, 1995–2001. J. Gt. Lakes Res. 29(Suppl. 1):
137–151.
Smith, B.R., and Tibbles, J.J. 1980. Sea lamprey (Petromyzon marinus) in Lakes Huron, Michigan, and Superior: history of invasion and control, 1936–78. Can. J. Fish. Aquat. Sci. 37(11):
1780–1801. doi:10.1139/f80-222.
Springborn, M., and Costello, C. 2010. Optimal random exploration
for trade-related non-indigenous species risk. In Bioinvasions
and globalization: ecology, economics, management, and policy.
Edited by C. Perrings, H. Mooney, and M. Williamson. Oxford
University Press, Oxford, UK. pp. 127–144. In press.
Stauffer, H.B. 2008. Application of Bayesian statistical inference
and decision theory to a fundamental problem in natural resource science: the adaptive management of an endangered species. Nat. Resour. Model. 21: 264–284.
Steeves, T.B. 2002. Uncertainty in estimating sea lamprey (Petromyzon marinus) abundance in Great Lakes tributaries. M.Sc.
thesis, Department of Fisheries and Wildlife, Michigan State
University, East Lansing, Michigan.
Walters, C. 1986. Adaptive management of renewable resources.
MacMillan Publishing, New York.
Published by NRC Research Press