Maximizing conserved biodiversity: why ecosystem indicators and thresholds matter
Ecological Economics 38 (2001) 259– 274
www.elsevier.com/locate/ecolecon
ANALYSIS
Maximizing conserved biodiversity: why ecosystem
indicators and thresholds matter
Mark E. Eiswerth a,*, J. Christopher Haney b
a
Department of Applied Economics and Statistics, Uni6ersity of Ne6ada, Reno NV 89557, USA
b
Conser6ation Science Di6ision. The Nature Conser6ancy, 4245 N, Fairfax Dri6e, Suite 100, Arlington VA 22203, USA.
Received 05 June 2000; received in revised form 24 January 2001; accepted 30 January 2001
Abstract
Accounting for biodiversity is important in several different types of constrained choice problems, including public
and private decisions for habitat and species conservation, the establishment of recreational parks and natural areas,
mitigation banking, and natural resource damage assessment (particularly primary and/or compensatory restoration
planning and scaling). In such applications it is important to give careful consideration to (1) the choice of
biodiversity indicator(s) to be used, and (2) the role of discontinuous, nonlinear ecological processes in light of the
decisionmaker’s chosen time horizon. The former is important because the choice of indicator(s) can substantially
influence decisions about conservation priority-setting and planning. The latter is critical for the same reason,
notwithstanding that dynamic ecosystem processes have rarely been considered sufficiently, if at all, in such
applications (in part because the processes usually are poorly understood or measured). In this manuscript we use
avian diversity data, collected by one of the authors, from hardwood forest ecosystems in the eastern United States.
We couple these data with estimates of species prevalence factors to construct a case study of how indicator choice
and consideration of ecological thresholds influence the outcomes of biodiversity preservation problems. We show
that (1) the choice of indicator(s) is critical, (2) failure to account for nonlinear, threshold effects in an ecosystem’s
future progression alters preservation decisions and ignores important information, (3) the effect of choosing different
time horizons depends on the indicator used, and (4) for any given biodiversity indicator, dynamic solutions can
depend on the time horizon chosen but not necessarily in monotonic or simple fashion. Our case study highlights the
importance of further system-specific research on dynamic ecological progressions as well as uncertainty regarding
future supply and demand for ecosystem service flows. © 2001 Elsevier Science B.V. All rights reserved.
Keywords: Biodiversity; Conservation; Preservation; Habitat; Forests; Birds
1. Introduction
* Corresponding author. Tel.: + 1-775-3275085; fax: + 1-
Despite the existence of accepted general defini-
775-7841342.
tions of biodiversity, debate continues over just
E-mail address: eiswerth@unr.edu (M.E. Eiswerth).
0921-8009/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 1 - 8 0 0 9 ( 0 1 ) 0 0 1 6 6 - 5
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
260
2. Background
what biodiversity is, how it should be measured,
and why it is important. Ecologists have
2.1. Rele6ant literature
defined a number of different types, or levels,
of biodiversity, with an increasing consensus
Biodiversity has long been recognized to be a
that no one indicator can or should be relied
multidimensional attribute of natural systems,
upon to characterize it. Different measures
with scientists referring to different levels of bio-
provide different indications of the variety
diversity including ecosystem, species, and ge-
and integrity of ecosystems, however, and the
netic diversity (Office of Technology Assessment,
choice of measures to use in a given context
1988; McNeely et al., 1990; National Research
depends on the research or policy objectives at
Council, 1992). Several years ago, Ray (1988)
hand.
observed that an ‘‘accounting of species alone
In previous research, we compared the out-
can be highly misleading as a yardstick of diver-
comes from applying different biodiversity indi-
sity’’, which led him to emphasize the impor-
cators to constrained choice problems of
tance of higher-order taxonomic diversity.
ecosystem/habitat preservation (Eiswerth and Atkinson (1989) placed this consideration in
Haney, 1992; Haney and Eiswerth, 1992). In clear perspective by stating that ‘‘given two
more recent research, one of the authors col- threatened taxa, one a species not closely related
lected a substantial amount of plant and animal to other living species and the other a subspe-
data from hardwood forest ecosystems in the cies of an otherwise widespread and common
eastern United States. The data collection pro- species, it seems reasonable to give priority to
ject was designed to investigate the ecological the taxonomically distinct form.’’
importance of old growth via comparisons to Observations such as these have encouraged
younger seral (successional) stages of hemlock- the development of measures that use taxonomic
northern hardwood forest (Haney, 1994, 1995; information (May, 1990; Altschul and Lipman,
1990; Vane-Wright et al., 1991) or information
Haney and Schaadt, 1995). In this manuscript
from limited molecular sequences (Crozier, 1992;
we use a portion of these data to construct a
Faith, 1992). Researchers have also used genetic
case study of how the choice of biodiversity in-
distinctiveness data to indicate biodiversity, by
dicators may affect constrained choice problems,
incorporating genome-wide data and linking
for example, public decisions related to habitat
composite information about an organism’s en-
conservation, restoration, or mitigation activi-
tire genetic makeup to data on species richness
ties. In addition, this case study illustrates the
(Eiswerth and Haney, 1992). This is the kind of
dynamic considerations that are important to
information that can be useful in many contexts,
such decisions. The forest ecosystem we focus
including (but not limited to) the search for spe-
on is characterized by nonlinear changes over
cies that have pharmaceutical and other values
time in structure and function, with discontinu-
(e.g. Reid et al., 1993a; Simpson et al., 1994).
ities occurring as the ecosystem moves from one
In setting priorities for conservation, relevant
developmental stage to the next. As a result,
metrics may include combinations of indicators
biodiversity in this system is a discontinuous
that reflect both diversity and the amount of
function of time. This has implications for diversity at risk. For example, species risk fac-
problems in which the desired outcome tors can be combined with taxonomic distinc-
is to maximize the flow of future services pro- tiveness indicators to yield a layered proxy (e.g.
vided by biodiversity. We show how the dy- Haney and Eiswerth, 1992). Such layered indica-
namic solution to a biodiversity preservation tors illustrate how decisions comparing diversity
problem may depend significantly on the time among regions can change as more (and better)
horizon considered and the biodiversity indica- information is considered in addition to simply
tor used. species richness. Reid et al. (1993b) provided an
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 261
informative summary of a wide range of indicators tion banking, and (5) natural resource damage
useful for policymakers, including ones that em- assessment (NRDA), particularly primary and/or
body risk. Such indicators are important in applied compensatory restoration planning and scaling.
decision-making because direct measures of ecosys- Forest biodiversity receives wide attention be-
tem value are in most cases unavailable, insuffi- cause of the multiple ecological, social, and eco-
cient, or too expensive to develop using standard nomic values associated with forest ecosystems
valuation methods (King, 1997). Indicators that (National Research Council, 1998). Our case study
are easy to use, are applicable to large areas, and involving eastern forests is particularly relevant
have a close linkage with specific elements, pro- given that decision-makers are currently attempt-
cesses, or qualities of ecosystem integrity are likely ing to determine the optimal mix of management
to be the most useful (Bradford et al., 1998; Miller regimes for sustainable forests. For example, indi-
et al., 1998/1999). viduals in Maine recently expressed an interest in
To model ecological attributes of ecosystems purchasing lands from timber companies to create
realistically, it is necessary to consider dynamic a large reserve in which forests would stand undis-
thresholds and other nonlinear processes in system turbed (Northern Forest Alliance, 2000). In this
structure and function. Such dynamic processes are and related situations, one of the relevant choice
rarely considered sufficiently, if at all, in exercises problems is, or at least ought to be: ‘Given a set
such as habitat protection, restoration, or conser- of forest tracts and a budget constraint for preser-
vation planning. Nonlinear, threshold processes vation, what is the optimal mix of conservation
are considered even less frequently, in part because efforts (or more broadly, management regimes)
they usually are poorly understood or measured. that maximizes the preservation of biodiversity?’
The importance of such processes is sometimes at The answer depends on the way in which the
least recognized in the literature (e.g. King, 1997), problem is formulated and the characteristics of the
but to date their incorporation in decision-making candidate conservation areas. While this
is woefully inadequate. manuscript deals solely with indicators of biodiver-
sity rather than the broader (and more complex) set
2.2. Pertinent concepts and applications of potential indicators of all ecosystem functions
and services, we recognize that in many decision
Concepts about biodiversity that we explore in contexts such broader indicators are generally of
this manuscript include: (1) the choice of biodiver- interest. We focus on biodiversity per se as one
sity indicator does matter, and can drive conserva- characteristic of natural systems, and show that
tion decisions, (2) it is important to account for consideration of even one such characteristic is in
dynamic ecosystem processes, and decision rules itself a complex step.
that do so may yield quite different results from
those that do not, (3) for any given indicator of
biodiversity, investments in conservation may de- 3. Case study forest areas: characteristics and
pend on the time horizon considered, but not data
necessarily in monotonic fashion, and (4) the effect
on the dynamic solution of changes in the time This case study is based on avian data collected
horizon may depend upon the biodiversity indica- from over 20 study plots in hemlock-northern
tor used. hardwood forest. Numerical values for avian pop-
These concepts have relevance for a number of ulations and communities were obtained from field
different activities and decisions. Examples include: studies conducted in Clearfield, Potter, and McK-
(1) decisions related to the purchase of land for ean counties on the Allegheny Plateau, Pennsylva-
conservation easements, (2) the establishment of nia (unpubl. data, J.C. Haney, collected
new recreational parks or natural areas, (3) agency 1992–1994; Dessecker and Yahner, 1984). Cen-
priority-setting for habitat and/or species conserva- suses were conducted in each of five forest age
tion expenditures, (4) decisions involved in mitiga- classes: 4, 9, 50, 120, and 300+ years. Forest age
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
262
was computed as the time elapsed since the last first derived from random subsampling of study
stand-replacing disturbance (either catastrophic plots available from this forest type (N= 21).
windthrow or even-aged timber harvest). These Because there are other potential biases to c,
five classes are termed early, transitional, mid-suc- estimates also conformed to the following criteria:
cessional, late successional, and old growth, re- visiting or wandering bird species were eliminated;
spectively. Hemlock-northern hardwood forest data collection was standardized by sampling fre-
displays temporal discontinuities in vegetation quency (eight visits) and area (each plot was of
structure, threshold effects, and other nonlinear equal size — 6 hectares (James and Rathbun,
patterns in successional development (see, e.g. 1981); sampling was conducted wholly within a
Tyrrell and Crow, 1994). single habitat type; and study plots were located
Taxonomic groups can be used as indicators in within large tracts of consolidated forest that were
two fundamentally different ways: as proxies for not in close proximity to other habitats (Remsen,
biodiversity and as proxies for environmental con- 1994).
ditions. For a variety of reasons, focusing on a Following application of the criteria above, the
diverse taxon such as birds is useful since a num- resulting data were combined with other informa-
ber of structural and functional elements of the tion sources to develop multiple indicators of
environment are automatically integrated. As a biodiversity as well as biodiversity at risk. First,
group, birds require very diverse microhabitats numbers of bird species (S) and higher taxa (gen-
arising from structural attributes related to stand era [G], families [F]) were computed for each of
and floristic composition, snag availability, foliage the five forest age classes. Next, we calculated a
height diversity, horizontal complexity, core area, layered proxy (Sa) that combined species richness
and local moisture conditions (Wiens, 1989). Bird with local (physiographic province) population
communities also exhibit marked, well-docu- species prevalences derived from Breeding Bird
mented differences in assemblage structure associ- Atlas programs in nine contiguous states in the
ated with forest developmental sequences northeastern United States (Laughlin and Kibbe,
(Lanyon, 1981; Smith and MacMahon, 1981; 1985; Andrle and Carroll, 1988; Brauning, 1992;
May, 1982; Glowacinski and Weiner, 1983; Helle, Bevier, 1994; Buckelew and Hall, 1994; Foss,
1984). Compared to other taxonomic groups, 1994; Palmer-Ball, 1996; Robbins and Blom,
birds perform quite well as indicators of specific 1996; Nicholson, 1997). This layered proxy Sa was
environmental conditions (Morrison, 1986; computed as:
Croonquist and Brooks, 1991). However, because Si
Sa = % [1− LPi ]
a few species do not always serve as accurate (1)
i=1
substitutes for many others (Niemi et al., 1997),
we make no assumption that this single taxon where LPi denotes the prevalence factor for spe-
serves as a suitable proxy for other species group- cies i at the local (physiographic province) scale.
ings or biodiversity in general (but see Pharo et The prevalence factor from the Breeding Bird
al., 1999). Atlas data can assume any value between 0 and 1,
We used bird species richness derived from inclusive. For example, a value of 0.50 for local
breeding bird census methodology (Lowe, 1995) species prevalence means that the species is found
as the initial proxy for forest biodiversity. A on 50% of the land area at the level of the
number of approaches have been proposed to physiographic province studied (in this case, the
estimate total species richness, C, within an area Appalachian Plateau of Pennsylvania). As the
(Bunge and Fitzpatrick, 1993). For comparisons average prevalence of a collection of species rises,
across forest development (seral) stages, however, the value of Sa for the collection falls. Weighting
we required only a bias-free estimate of relative species richness in this manner thus provides us
species richness, c. This approach is equivalent to with an indication of not only (1) the number of
the data-analytic class of methods reviewed by species present in our study area, but also (2) the
Bunge and Fitzpatrick. Point estimates of c were subset of those species present that are not preva-
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 263
lent at a larger geographic scale. This metric of the five different forest age classes (seral
provides information somewhat similar to that stages). The values for each of these indicators, by
offered by specificity indicators reflecting the oc- forest stage, are shown in Table 1. Table 1 also
currence (abundance) of species within a given indicates the percentage of species that were
geographic space or ‘cluster’ of sites (Dufrene and uniquely detected within each seral stage. This
Legendre, 1997; Legendre and Legendre, 1998). illustrates that each forest seral stage displays its
Finally, we computed a similar indicator (Sb) own particular set of species.
by weighting species richness again, this time by Of course there are additional indicators that
regional population prevalence as calculated from one could develop and use. For example, one of
the Breeding Bird Atlas programs. Sb was com- the factors that a conservation planner may wish
puted as: to consider might involve the relative scarcity of
different forest types, in combination with the
Si
Sb = % [1− RPi ] (2) number of species unique to each type. Such a
i=1
metric would provide somewhat different infor-
mation when compared to indicators Sa and Sb.
where RPi denotes the prevalence factor for spe-
However, note that Sa and Sb do explicitly incor-
cies i at the regional scale. This indicator weights
porate the underlying relative scarcity of habitat
species richness to reflect those species present in
types that play host to each particular species
our study area that are not common at the re-
considered. These indicators do this by weighting
gional level (in this case, across the northeastern
each species by the percentage of land (on either a
United States). As the number of species in a
local or regional basis) on which the species is
forest age class that are not prevalent regionally
estimated to occur (and hence the percentage of
goes up, Sb rises as well.
land that currently provides habitat suitable to
The work described above yields multiple indi-
cators of diversity or diversity/prevalence for each each particular species). To the extent that a
Table 1
Indicators of biodiversity in Pennsylvania hemlock-northern hardwood forest plots of different seral stagesa
Forest seral stageb,c
Indicators
Early (15.2%) Transitional Mid-successional Late successional Old growth
(31.2%) (41.2%) (12.0%) (0.4%)
Total number of bird species 9 17 20 34 20
% Bird species uniquely detected in 22 24 10 29 40
seral stage
Total number of bird genera 9 17 16 25 15
Total number of bird families 2 9 8 11 10
Species richness weighted by 2.5 4.3 5.9 12.9 10.1
physiographic province (local)
population prevalence (Sa)
Species richness weighted by 2.6 5.0 7.2 15.9 11.5
regional population prevalence
(Sb)
a
Sources of data: J.C. Haney, unpubl. data collected 1992–1994; Dessecker and Yahner, 1984; Laughlin and Kibbe, 1985; Andrle
and Carroll, 1988; Brauning, 1992; Bevier, 1994; Buckelew and Hall, 1994; Foss, 1994; Palmer-Ball, 1996; Robbins and Blom, 1996;
Nicholson, 1997.
b
Early seral stage = stand age of 4 years; transitional = 9 years; mid-successional =approx. 50 years; late successional = approx.
120 years; old growth =300+ years.
c
The relative prevalence of each seral stage forest type found in Pennsylvania (as a percentage of total forestland) is given in
parentheses under the seral stage names. These relative prevalence values are derived from Alerich (1993).
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
264
particular species is associated with (unique to) however, at least one of the forest types contained
only one forest type, Sa and Sb embody the species not commonly found elsewhere, then Sa
relative scarcity of that forest type. and Sb would hold great meaning for objectives
As a second example, a relative measure such as such as preserving population sources, preventing
Sa/S, where S denotes number of species, may be further habitat fragmentation, increasing wildlife
of value in certain situations with particular con- corridors, and providing recreational services (i.e.
servation management objectives. Such a relative viewing rare species) even to visitors from far
measure could provide a higher indicator value away.
for a region that has very few species (e.g. 10) but
where a high percentage of those species are rare,
as compared to a region with many more species 4. Static approach and results
(e.g. 100) but relatively few rare ones. Some con-
4.1. Static illustration 1: choice of biodi6ersity
servation decision contexts may call for placing a
premium on rarity (and ignoring the absolute indicator can dri6e habitat rankings and thus
number of species) and in such cases a measure discrete choices regarding habitat conser6ation
such as Sa/S may be useful. At the same time, the
attraction of Sa by itself is that it does combine The simplest problem involves a discrete choice
two different kinds of information: species rich- problem of conservation. In such cases a decision-
ness and species rarity. maker may be interested in choosing a subset of
We do not attempt in this manuscript to iden- all geographic areas (one, in the simplest case) in
tify any one best indicator; indeed, indicators which to devote habitat conservation efforts. This
need to be matched carefully to management ob- situation may occur when available funding for
jectives since the choice of indicator will influence conservation is sufficiently constrained. It also
the decision outcome. Various alternative indica- may occur in processes that involve mitigation
tors, including but not limited to those in Table 1, banking or compensatory restoration for lost nat-
have different meanings with respect to conserva- ural resource service flows.
tion objectives and social/economic values. As an Table 1 reveals several points relevant to rank-
example, a skilled birdwatcher may attach a great ing our case study forest types. Perhaps the most
deal of importance to the sheer number of species apparent feature is that regardless of the indicator
that he or she is able to see, on average, upon chosen, the decision-maker would rank the late
visiting the forest. In contrast, an avid hiker or successional forest first in terms of biodiversity as
angler untrained in birdwatching may derive plea- well as biodiversity weighted by prevalence. This
sure from the incidental viewing of a wide variety forest stage dominates the others in species rich-
of birds while recreating, but may be unable to ness (34 species), higher taxa diversity (25 genera),
discern (or uninterested in noticing) differences species richness weighted by physiographic
among species that are closely related. Such an province (local) prevalence, and species richness
individual may attach more importance to the weighted by regional prevalence.
indicators in Table 1 that relate to the total Rankings of forest stages below the late succes-
number of bird genera or families, rather than sional are more problematic. If number of species
species richness. is used as a biodiversity indicator, the decision-
Similarly, the importance of indicators such as maker’s second choice for conservation efforts
Sa and Sb in comparison with the others depends could be either the mid-successional or old-
largely on the extent to which the conser- growth seral stage. If, however, the number of
vation planner’s objectives are tied to a broader genera were used as an indicator of higher taxo-
spatial (e.g. regional) context. If none of the spe- nomic diversity, the decision-maker would pick
cies under consideration is rare in terms of preva- the transitional age class over both the mid-suc-
lence within a larger spatial area, then the cessional and old growth as the second priority
meaning of Sa and Sb would be minimal. If, for conservation efforts.
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 265
Consideration of species prevalence factors from one another, or with differences in visitor
yields even more interesting results. First, on aver- profiles across the areas. Public preferences may
age, species found in the early, transitional, and also be of a type such that it is important that
mid-successional forest types have high prevalence natural attributes (such as species) can be enjoyed
values (relative to older forests) and correspond- in multiple areas, even when the areas are not that
ingly lower species rarity factors. In contrast, on far apart. If demand for natural areas and the
average, species found in older forests are less service flows (e.g. birdwatching) that they offer is
prevalent at broader spatial scales. If the indicator high relative to supply (this is the case in many
is defined as species richness weighted by preva- wildlife refuges today), then congestion comes
lence at either the physiographic province (Sa) or into play to make the presence of a species in one
regional (Sb) levels, then the old growth forest age area a poor substitute for its presence in another.
class becomes the clear second choice for conser- Consideration of risk and uncertainty provides an
vation efforts. additional basis for this assumption. As discussed
The usefulness of Sa and Sb as indicators is now in King (1997), uncertainty exists regarding the
clear, in that they attach a premium to forest age effects on ecosystems of future natural and an-
classes containing species that are not common. thropogenic changes. Since we do not know how
Such forest areas are potential population future natural changes or human activities close
‘sources’ (Pulliam, 1988) of species not prevalent to natural areas may affect their structure and
at broader scales. In Section 3 we also mentioned function, a motivation exists to expend conserva-
other possible indicators, including relative mea- tion efforts in multiple areas, even if they offer
sures such as Sa/S that attach complete impor- similar ecosystem services today.
tance to relative species rarity with no weight Given this assumption, a decision-maker that is
attached to the number of species. Such indicators concerned with habitat conservation in multiple
would give a higher priority to old growth forests. areas may wish to maximize the sum of biodiver-
sity across the areas, subject to a budget con-
4.2. Static illustration 2: choice of biodi6ersity straint for conservation efforts. Consider the
indicator can significantly influence the allocation following relationship between conservation ex-
of conser6ation expenditures among multiple penditures and an indicator of biodiversity:
geographic areas
bi = ki + fi (Mi ) (3)
In some instances a decision-maker may need where bi is the expected value of a biodiversity
indicator in area i, ki E 0 is the expected value of
to make decisions regarding the allocation of
habitat conservation efforts in multiple geo- a biodiversity indicator in area i given no conser-
graphic areas, rather than a discrete choice of vation expenditures in area i, Mi denotes conser-
which area(s) to conserve. Such decision-making vation expenditures in area i, and where
f %(Mi )\ 0, f %%(Mi )B 0. The function fi (Mi ) de-
requirements provide a richer context for analysis. i i
We assume that characteristics of one geo- notes the addition to the level of the biodiversity
graphic area are not substitutes for the same indicator expected to result from conservation
characteristics found in another geographic area. expenditures Mi.
In our case study this means, for example, that Eq. (3) and the equations that follow are writ-
the presence of a species in a forest of a given ten in the standard economic format of maximiz-
seral stage is not a substitute for its presence in ing a variable subject to a constraint on
another forest area of a different stage. This is not expenditures (M). However, the term M can also
a restrictive assumption, but rather is consistent be interpreted more broadly as a money metric
with a number of real-world contexts. For exam- equivalent of efforts devoted to conserving biodi-
ple, it is consistent with a situation in which a versity. Similarly, it is possible to interpret the
decision-maker is interested in devoting efforts to term b in (1) as a function of conservation efforts
natural or recreation areas some distance apart rather than expenditures. One reviewer of this
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
266
article has wisely pointed out that some important
conservation management steps may require less
out-of-pocket expenditure and more good (and
timely) planning, relative to alternative manage-
ment efforts. However, the standard economic
view is that such good and timely planning would
come at an opportunity cost, for example, hiring a
well-trained (and presumably well paid) ecologist
or wildlife biologist to spend part of his or her
time on the conservation planning process. For
this reason as well as ease of exposition, we refer
to M as conservation expenditure while realizing
that a more complex indicator of conservation
effort is also possible. Fig. 1. General shape of illustrative function linking expected
biodiversity to conservation expenditure.
Consider a case in which a decision-maker is
interested in two different geographic areas in
make the simplifying assumption that k1 = k2 = 0.
different forest age classes. If the decision-maker
Relaxation of this assumption could change the
is interested in allocating conservation expendi-
numerical solution to the problem, but would not
tures between these areas, a relevant constrained
change the flavor of the concepts and results upon
maximization problem is:
which we focus in this manuscript.
Max[k1 + f1(M1)+k2 +f2(M2)] Second, we assume for illustration that the
conservation expenditures necessary to set the ex-
s.t.:M1 +M2 =M (4) pected level of the biodiversity indicator equal to
the baseline (existing) level of biodiversity are
where M denotes the total resources available to
equal across the two geographic areas. For exam-
the decision-maker.
ple, if conservation expenditures involve purchas-
As an illustration of the way in which the
ing land, this assumption would denote that land
choice of biodiversity indicators affects the solu-
costs are equal for the two areas. For our illustra-
tion, we consider a specific case of the generalized
tions, we use a specification that is consistent with
problem. First, assume for simplicity that k1 =
these assumptions as well as the standard eco-
k2 = 0; that is, in both areas, the indicator of
nomic assumption of diminishing returns to
biodiversity is expected to be zero if conservation
expenditures:
efforts are zero. This is a special case of the more
general case ki E 0 and corresponds to a situation fi (Mi )= (b 0)(M) − 1/2(Mi )1/2 (5)
i
where a decision-maker is interested in protecting
where i denotes forest area i and b 0 the baseline
all or a portion of a land area from complete
development, e.g. total conversion of land into (current) level of biodiversity there. The general
housing subdivisions, a relevant scenario in many shape of this function is shown graphically in Fig.
parts of the United States. That is, the special case 1, which illustrates that additional conservation
is that if the decision-maker makes no conserva- efforts purchase a higher level of expected biodi-
tion expenditures, then complete habitat destruc- versity but at a diminishing rate. Though we
tion will occur. We certainly recognize that in assume for simplicity in our numerical analyses
that ki = 0, Fig. 1 depicts the more general case in
reality biodiversity does not necessarily equal zero
which ki \ 0 (some biodiversity will remain if no
even when land is completely developed. How-
ever, we have not collected data on biodiversity conservation efforts are undertaken).
for our case study bird species in a formerly The first-order condition, which gives the solu-
comparable area (e.g. close to our study sites) that tion to the constrained maximization problem
has been deforested and developed. Therefore we (noneconomists may see Chiang (1974) for an
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 267
introduction to constrained maximization) shown conservation planning. Some natural systems have
in Eq. (4), is: a very large number of species but relatively low
diversity at higher taxonomic levels. In contrast,
(b 0)(M1) − 1/2 =(b 0)(M2) − 1/2. (6) some systems, for example, some marine and
1 2
coastal ecosystems, are strikingly rich in their
To show the implications of using alternative endowment of diverse families with relatively few
indicators of biodiversity as input to the con- species representing each of those families (Ray,
strained maximization problem, we consider the 1988).
case in which forest areas 1 and 2 are currently in When species richness is weighted by regional
transitional and mid-successional stages, respec- population prevalence (to form the indicator Sb),
tively. The comparison between these two stages
the allocation of expenditures shifts substantially
is interesting because neither one dominates in
toward the older (mid-successional) forest. Using
terms of biodiversity.
this indicator, 68% of total conservation expendi-
The results from using alternative indicators are
tures will be targeted toward the mid-successional
summarized in Fig. 2. Clearly the choice of indi-
forest class. This outcome reflects the area’s abil-
cator can influence decisions on how to allocate
ity to act as a source for species that are not
efforts. In our illustration, the difference between
highly prevalent on a wider regional basis. As
using species richness and a higher taxa diversity
shown in Fig. 2, for the three indicators exam-
indicator is significant (33% more expenditures
ined, the outcome may range from a low of 44%
devoted to the transitional forest area using a
to a high of 68% of total available conservation
higher-taxa indicator rather than species richness).
expenditures being devoted to the older forest
For the subclass of problems where a decision-
area. The sensitivity of the solution to the choice
maker is interested in purchasing land or prevent-
of indicator illustrates the potential volatility of
ing development so as to preserve biodiversity,
decision-making processes to the types of infor-
even the differential found in our illustration
mation considered.
would lead to a difference in the portfolio of
forest areas that the planner chooses to buy/pro-
tect. In some cases, the choice between these two
indicators can have a substantial influence on
5. Dynamic approach and results
The relatively small subset of structural at-
tributes that exhibit temporal linearity, and the
threshold changes that occur in forests during
succession, create distinct stages in forest ecosys-
tems. To adequately characterize such ecosystems,
time-varying stages and threshold effects must be
taken into account. As in other ecosystems, diver-
sity in our case study system is time-scale depen-
dent, that is, dependent upon time from the most
recent disturbance.
As a result, it is important to consider not only
the current levels of diversity in particular areas,
but also the diversity levels that the areas can
potentially offer society in the future. The general
problem may be viewed as choosing management
options to maximize the expected ‘flow’ of diver-
sity from the present to some point in the future,
Fig. 2. Allocation of conservation expenditures: static frame-
subject to a budget constraint.
work.
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
268
Forests can experience abrupt structural seral stage. If the decision-maker is interested in
changes in either of two directions. Successional maximizing biodiversity over this period, then the
changes occur as the forest moves through growth relevant maximization problem is:
phases (seres), each consisting of varying intervals
T
when structure is relatively constant but where
Max % [ f 1(M 1)+ f 2(M 2)]
t t
rapid transition occurs between seres. Succession t=1
can also be reversed, and the entire set of ecolog-
s.t.: M 1 + M 2 = M
ical processes renewed, when catastrophic distur- (7)
bance (either man-made or natural) shifts forest
where M 1, M 2, and M may be thought of as the
structure back to an earlier sere. In the forests
used in our analysis, natural disturbances that discounted present values of the opportunity costs
completely remove canopy trees occur very rarely, of conservation that are incurred between now
about every 1200 years (Canham and Loucks, and period T (expressed in this way to simplify
1984; Frelich and Lorimer, 1991). At large land- the exposition). As with the static case, M is the
scape scales all successional stages can be main- total amount of resources available for conserva-
tion, and M 1 and M 2 are the amounts to be
tained in perpetuity, although not always in the
same amount or location (Shugart, 1984). In allocated to forest areas 1 and 2, respectively.
other words, by protecting relatively large areas of For most natural systems, scientists have not
forest, it is possible to ‘purchase’ increased levels collected continuous data on the ways in which
of certainty that a forest area will progress as various indicators of biodiversity change over
anticipated through its natural growth phases. time. At best, a limited set of observations may
In this section we provide illustrations of the exist for particular stand ages in forests, for exam-
relevance of natural dynamic processes. Section ple. In other cases, very little direct information is
5.1 illustrates the importance of recognizing that available. For our case study, we have the benefit
change may not be linear, and highlights the need of possessing standardized observations of bird
for better data on how and when ecosystems diversity in forests that are very similar (in terms
encounter thresholds. Section 5.2 shows that of climate, geographic zone, etc.) except that they
choice of time horizon and biodiversity indicator are of different ages.
may affect the dynamic solution to preservation, To illustrate the importance of knowing how
but not necessarily in the expected ways. Section natural systems evolve, suppose for a moment
5.3 illustrates how a dynamic approach may dif- that all that we knew about the problem was the
ferentiate natural areas that look equivalent from current number of species and higher taxa for
a static viewpoint. both areas, as well as the same information for
the late successional stage that both areas are
5.1. Dynamic illustration 1: the importance of expected to evolve into over the next 100 years.
Assume further that diversity is expected to in-
accounting for nonlinear, discontinuous ecological
crease in linear fashion over time in either forest
processes
area. In this case, the choice problem would in-
volve choosing M 1 and M 2 to maximize the sum
Consider the case in which a decision-maker
wishes to maximize the sum of a biodiversity of the areas under the (linear) biodiversity time
indicator across two different forest areas, the first paths in the two forests. Solution of this problem,
stand of 30 years and the second stand of 90 using species as a biodiversity indicator and the
years. We assume that, for each of the seral stages same diversity–expenditure functional forms
we examine, diversity is characterized by the ob- shown in Eq. (5), would provide the answer that
servations shown in Table 1. Suppose that the 42% of the available resources (opportunity costs)
decision-maker’s time horizon, T, is 100 years. It for conservation would be devoted to forest area
is expected that, by time T, both of these forest 1 (transitional), and 58% to forest area 2 (mid-
areas will have evolved to the late successional successional).
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 269
Table 2
Allocation of conservation expenditures among transitional and mid-successional forest areas, under alternative time horizons and
indicators of biodiversitya,b
Time horizon Number of species Number of families Species richness indicator
indicator (S) indicator (F) weighted by regional population prevalence (Sb)
Time horizon= 50 Area 1: 44% Area 1: 55% Area 1: 36%
Area 2: 56% Area 2: 45% Area 2: 64%
Time horizon= 100 Area 1: 37% Area 1: 47% Area 1: 29%
Area 2: 63% Area 2: 53% Area 2: 71%
Time horizon= 150 Area 1: 41% Area 1: 47% Area 1: 36%
Area 2: 59% Area 2: 53% Area 2: 64%
a
Forest area 1 is currently in the transitional seral stage with a stand age of 10 years. Forest area 2 is currently in the
mid-successional seral stage with a stand age of 50 years.
b
Each percentage in the table denotes the percentage of total conservation expenditures that will be devoted to a forest area,
according to the solution of the dynamic constrained maximization problem defined in the text.
5.2. Dynamic illustration 2: the choice of time
Now consider the problem given our knowledge
horizon and biodi6ersity indicator may ha6e a
that the biodiversity time path more closely re-
significant impact on the dynamic solution, but
sembles a step function than a linear function. It
not necessarily in monotonic fashion
is intuitively clear that forest area 2, currently at
stand age 90, will enter the late successional seral
Consider once again the allocation of conserva-
stage significantly sooner than forest area 1. Once
tion expenditures between a current transitional
the late successional stage is reached, the forest
forest area and a current mid-successional area.
area will exhibit higher levels of biodiversity as
Given the knowledge that the biodiversity time
measured by numbers of species, genera, or
path is subject to discontinuities as forests move
families. Therefore, one would expect that, if we
from one seral stage to the next, how does the
take account of the step function nature of the
choice of time horizon affect the solution to the
biodiversity time path, a premium would be
problem in Eq. (7)? And how does the choice of
placed on conservation efforts in forest area 2.
biodiversity indicator influence the result? We
Solution of the maximization problem accounting
consider three alternative time horizons (50, 100,
for a stepwise progression bears this out: using
and 150 years) and three alternative indicators (S,
species again as an indicator of biodiversity, the
F, and Sb). The solutions to the problem under
solution would involve only 33% of conservation
these conditions are shown in Table 2.
efforts in forest area 1 (vs. 42% assuming a linear
Three main points emerge from Table 2. First,
time path), with 67% of efforts now devoted to
conservation allocation outcomes vary signifi-
forest area 2.
cantly according to choice of indicator and time
The difference in solutions under linear and
horizon, from a low of 29% to a high of 55% of
step function approaches is perhaps not that strik-
total expenditures devoted to forest area 1. Sec-
ing for the particular example we have chosen,
ond, the dynamic solutions depend on the time
though it is significant. The salient point is that
horizon chosen but not necessarily in monotonic
the incorporation of information on threshold
fashion. For example, Fig. 3 illustrates how the
effects can affect the decision-making process.
allocation of expenditures varies according to T
Certainly there are cases in which accounting for
when species richness is used as an indicator of
these effects may have a substantial bearing on
biodiversity. The optimal percentage of resources
the planner’s decision, depending on the natural
to be devoted to forest area 1 (transitional) first
systems and time horizons considered.
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
270
50 to 100 (Table 2), for the same reasons de-
declines as the time horizon is increased from 50
scribed above for species. However, extension of
to 100 years, then rises as T goes from 100 to 150.
T from 100 to 150 leaves the solution unchanged
This is because the expected passage of both
with number of families as the indicator, unlike
forest areas into the late successional stage, which
the pattern under the species indicator. This is due
exhibits markedly high biodiversity, is considered
to the way in which species appear and disappear
to varying degrees according to the chosen time
as the forest moves through seral stages. Specifi-
horizon. With T=50, passage to the late succes-
cally, the number of species may increase or de-
sional stage is considered for neither forest area,
crease through time without there occurring a
and so the current levels of biodiversity largely
change in diversity as measured at higher taxo-
drive the result. With T = 100, passage of forest
nomic levels. The same kind of pattern can occur
area 2 to the late successional stage is taken into
for genetic diversity, i.e. if closely related species
account while that of forest area 1 is not. As a
appear or disappear through time, species diver-
result, a premium is attached to conserving forest
sity may change significantly while genetic diver-
area 2 and the percentage of total expenditures
sity does not.
devoted to it rises. With T = 150, the passage of
both forest areas to the late successional is consid-
5.3. Dynamic illustration 3: a dynamic approach
ered, and so emphasis shifts back toward a some-
may differentiate areas that are equi6alent from a
what higher level of emphasis on forest area 1.
static perspecti6e
While the shifts in expenditures for this illustra-
tion may not be dramatic, they are indicative of
Now consider the allocation of expenditures
the implications of choice of T for preservation
between two forest areas both currently in the late
decisions in general.
successional stage but that have different stand
Third, the influence of altering T depends on
ages. Specifically, consider forest areas 1 and 2,
the indicator of biodiversity that is used. For
which have stand ages of 125 and 250 years,
example, using number of bird families as an
respectively. Assume that these areas display simi-
indicator, conservation efforts devoted to forest
lar numbers of species, genera, and families. The
area 2 (mid-successional) increase as T goes from
main difference between them is that forest area 2
will evolve into an old growth forest 125 years
sooner than forest area 1.
Of course, if the two areas currently are similar
in terms of biodiversity, a static approach would
give them equal weight regardless of the indicator
used. However, one does not necessarily give
them equal weight if dynamics are taken into
account. The solutions to Eq. (7) for this problem
are shown in Table 3. With T= 50 years, the two
areas have equal weight because neither one will
have progressed out of the late successional seral
stage. With T= 100 or 150, however, the younger
forest area (1) will be accorded a significantly
higher percentage of conservation efforts (61 and
66% of the total for T=100 and T= 150, respec-
tively). As the time horizon increases, then, a
decision-maker interested solely in maximizing the
biodiversity indicator will attach more importance
Fig. 3. Allocation of expenditures may be nonmonotonic with
to conserving the younger late successional forest
respect to time horizon T: dynamic framework using species
area.
richness.
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 271
Table 3 argument against conserving old growth. How-
Allocation of conservation expenditures between two late suc-
ever, it does indicate strongly that decision-mak-
cessional forest areas of different stand age: dynamic
ers should clearly and deliberately prioritize
frameworka
conservation objectives on a site-specific basis, as
well as recognize that particular objectives may
Time horizon Relative expenditures by area, using
species richness indicator sometimes lead to decisions that run counter to
conventional wisdom.
Area of stand Area of stand
age =125 age =250
6. Conclusions
50 years 50% 50%
100 years 61% 39%
150 years 66% 34% We have used data from forest ecosystems to
illustrate several key concepts relevant to biodi-
a
Each percentage in the table denotes the percentage of
versity. First, the solution to a static biodiversity
total conservation expenditures that will be devoted to the
preservation problem may depend significantly on
corresponding forest area, according to solution of the dy-
namic constrained maximization problem defined in the text. the biodiversity indicator used. This is an impor-
tant concept for decision-makers to understand
This result may stand at odds with expecta- and assess, particularly at the site-specific level.
tions, given that the older forest area (2) will The use of alternative indicators to examine the
progress to old growth 125 years sooner, and multiple attributes of natural systems, and the
given the importance that society generally associ- extent to which those attributes are at risk, can
ates with old growth forest. The result is driven force a useful reexamination of conservation ob-
by the fact that the old growth seral stage is jectives. The choice of final indicators to use as
actually less diverse (as measured both by number guides may vary greatly from case to case and will
of species and number of higher taxa) than the depend on the context of the problem and the
late successional seral stage. Therefore, conserva- ecosystem services that are most highly valued by
tion decisions made solely on the basis of antici- the public.
pated biodiversity will tend to favor the late Second, for any given indicator, dynamic solu-
successional stage over old growth, and therefore tions may differ from the static solution, depend-
result in the conservation of younger forests. This ing on the time horizon chosen by the
is an issue that may arise in a number of different decision-maker. This forces a reexamination of
types of forest systems since evidence suggests the timeframes that we wish to take into account
similar patterns in a variety of forest types. when considering future streams of ‘biodiversity
Clearly, there may be other reasons to value old services’, or ecosystem functions and services
growth forest besides numbers of species or higher more broadly. This is a simple concept, but the
taxa (Brunson and Shelby, 1992). For example, existing literature does not adequately address it,
note that in Table 1 we show that a relatively particularly for cases in which ecosystems are
large number of the species found in old growth expected to display discontinuous processes in the
were uniquely detected in that forest type. Second, future. Our analysis also highlights the need for
a relative measure such as Sa/S (which prioritizes dynamically adaptive management, rather than a
areas solely according to the percentages of their long-term fixed formula for conservation, since
species collections that are not prevalent at a the portfolio of biodiversity and forest types will
broader scale) would attach high importance to continue to change as time passes.
old growth. Third, individuals may exhibit prefer- Third, for any given indicator, dynamic solu-
ences for recreation in old growth forest because tions can depend on the time horizon chosen, but
of factors totally unrelated to biodiversity. The not necessarily in monotonic fashion. This is a
counterintuitive result of this illustration certainly characteristic not common to well-behaved dy-
is not (and in no way is intended to be) an namic models and therefore merits special atten-
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
272
tion. Fourth, the effect of changes in the time goods) may affect future patterns of demand.
horizon on the dynamic solution is dependent on Such changes are equally uncertain.
the indicator used, which reinforces the need to Future research might assess the relative impor-
consider multiple proxies. tance of different sources of uncertainty in factors
This manuscript does not deal with choosing affecting supply and demand. This is likely to be
any one indicator over another, but rather em- quite site specific. In some instances, uncertainty
phasizes that the choice of indicator certainly does regarding future demand for environmental
matter and should be linked to conservation ob- amenities may swamp that connected with future
jectives. At the same time, the issue of indicator ecosystem processes. As mentioned above in con-
reliability will be important in actual decision- nection to forests, catastrophic disturbance (either
making applications. Reliability is largely a statis- man-made or natural) can reverse the successional
tical issue and depends on criteria such as process. In the forests we examined, however,
sensitivity, specificity, and predictability. A signifi- natural widespread disturbances rarely occur and
cant literature exists to help guide practitioners on many types of preservation efforts can effectively
this point (e.g. Murtaugh, 1996; Dufrene and insulate areas from major anthropogenic effects
Legendre, 1997; Legendre and Legendre, 1998). such as land development. Therefore, uncertainty
Our analysis is illustrative in nature in that it in ecosystem service supply may be small relative
relies upon an example relationship between con- to uncertainty in future demand for environmen-
servation effort and conserved biodiversity, rather tal amenities. In other types of systems, where
than an empirically estimated function between disturbance is more likely and vulnerability to
these two variables. The impact of conservation disturbance may be higher (some coastal ecosys-
effort on any given biodiversity indicator will vary tems may fall into this category), there may be
from site to site and potentially through time for substantial uncertainty in forecasting the supply
any particular site. Future ecological research to of ecosystem services for several years into the
examine the biodiversity ‘returns’ from increased future.
conservation activities, as well as the way that this
relationship varies by indicator, would be quite
useful. Acknowledgements
We intentionally have not incorporated uncer-
tainty in any fashion in this analysis, primarily to Financial support was provided by the Center
avoid detracting from the major points of interest. for Rural Pennsylvania (CRP), DuBois Educa-
Incorporation of this factor, however, represents tional Foundation Fund for Academic Excellence,
an important avenue for further research. For any Pennsylvania State University Research and De-
given natural area, uncertainty exists regarding velopment funds, the USDA Cooperative State
the future demand for and supply of various Research, Education, and Extension Service, and
ecosystem functions and services. Natural forces, a Challenge Grant from the Migratory Bird
as well as future anthropogenic change (e.g. Office, Region 5, US Fish and Wildlife Service
changes in patterns of adjacent human develop- (USFWS). D. DeCalesta, J. Palmer, and S. Stout
ment or changes in effects from pollutants trans- (US Forest Service), L. Lentz, J. Sowl, and D.
ported into the area), may change the supply of Wright (CRP), T. Mountain and D. Pence (US-
amenities that the ecosystem offers. However the FWS), C. Schlentner (Cook Forest State Park),
directions, magnitudes, and timing of such poten- and C. Schaadt provided logistic support, access
tial future changes, as well as the ways in which to study areas, or other assistance that greatly
the area will respond, are uncertain (King, 1997). facilitated this study. J. Lydic and R. Williams
In the same way, a host of factors (changes in performed many of the summary analyses. For
human population distributions, demographics, their help with the breeding bird censuses and
preferences for goods and services, and the rela- other field work, we thank B. Allison, J. Cheek,
tive prices of environmental amenities and other L. Hepfner, R. Kaufmann, J. Lydie, C. Schaadt,
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 273
Faith, D.P., 1992. Conservation evaluation and phylogenetic
J. Seachrist, J. Smreker, S. Weilgosz, S. Wetzel,
diversity. Biological Conservation 61, 1 – 10.
and R. Williams. We thank S. Ragland and three
Foss, C.R., 1994. Atlas of Breeding Birds in New Hampshire.
anonymous reviewers for comments on previous Audubon Society of New Hampshire. Arcadia-Chalford
versions of this manuscript. Publishing, Dover, NH.
Frelich, L.E., Lorimer, C.G., 1991. Natural disturbance
regimes in hemlock-northern hardwood forests of the Up-
per Great Lakes Region. Ecological Monographs 61, 145 –
164.
References
Glowacinski, Z., Weiner, J., 1983. Successional trends in the
energetics of forest bird communities. Holarctic Ecology 6,
Alerich C.L., 1993. Forest statistics for Pennsylvania — 1978 305 – 314.
and 1989. USDA Forest Service Research Bulletin NE-126. Haney, J.C., 1994. Winter bird populations as bioindicators of
Radner, PA. old-growth forest in eastern North America. Journal fur ¨
Altschul, S.F., Lipman, D.J., 1990. Equal animals. Nature Ornithologie 135 (3), 515.
348, 493 – 494. Haney, J.C., 1995. Policy aspects of old-growth forest reserves
Andrle, R.F., Carroll, J.R., 1988. The Atlas of Breeding Birds in the eastern U.S. In: Majumdar, S.K., Miller, E.W.,
in New York State. Cornell University Press, Ithaca, NY. Brenner, F.J. (Eds.), Forests — A Global Perspective. The
Atkinson, I., 1989. Introduced animals and extinctions. In: Pennsylvania Academy of Science, Pittsburgh, PA.
Western, D., Pearl, M. (Eds.), Conservation for the 21st Haney, J.C., Eiswerth, M.E., 1992. The plight of cranes: a case
Century. Oxford University Press, New York, pp. 54 – study for conserving biodiversity. Proceedings of the North
69. American Crane Workshop 6, 12 – 18.
Bevier, L.R., 1994. The Atlas of Breeding Birds of Connecti- Haney, J.C., Schaadt, C.P., 1995. Functional roles of eastern
cut. State Geological and Natural History Survey of Con- old-growth in promoting forest bird diversity. In: Davis,
necticut, Hartford, CT. M.B. (Ed.), Eastern Old Growth. Island Press, Washing-
Bradford, D.F., Franson, S.E., Neale, A.C., Heggem, D.T., ton, DC.
Miller, G.R., Caterbury, G.E., 1998. Bird species assem- Helle, P., 1984. Effects of habitat area on breeding bird
blages as indicators of biological integrity in Great Basin communities in northeastern Finland. Annales Zoologici
rangeland. Environmental Monitoring and Assessment 49, Fennici 21, 421 – 425.
1–22. James, F.C., Rathbun, S., 1981. Rarefaction, relative abun-
Brauning, D.W., 1992. Atlas of Breeding Birds in Pennsylva- dance, and diversity of avian communities. The Auk 98,
nia. University of Pittsburgh Press, Pittsburgh, PA. 785 – 800.
Brunson, M., Shelby, B., 1992. Assessing recreational and King, D.M., 1997. Leading indicators of ecosystem services
scenic quality. Journal of Forestry 90, 37 –41. and values, with illustrations for performing habitat equiv-
Buckelew, A.R., Hall, G.A., 1994. The West Virginia Breeding alency analysis. Invited paper, Conference on Lost Human
Bird Atlas. University of Pittsburgh Press, Pittsburgh, PA. Uses of the Environment, Washington, DC, 7 & 8 May,
Bunge, J., Fitzpatrick, M., 1993. Estimating the number of 1997.
species: a review. Journal of the American Statistical Asso- Lanyon, W.E., 1981. Breeding birds and old field succession
ciation 88, 364 – 373. on fallow Long Island farmland. Bulletin of the American
Canham, C.D., Loucks, O.L., 1984. Catastrophic windthrow Museum of Natural History 168, 5 – 57.
in the presettlement forests of Wisconsin. Ecology 65, Laughlin, S.B., Kibbe, D.P., 1985. The Atlas of Breeding Birds
803 – 809. of Vermont. University Press of New England, Hanover,
Chiang, A.C., 1974). Fundamental Methods of Mathematical NH.
Economics. McGraw-Hill, New York. Legendre, P., Legendre, L., 1998. Numerical ecology. Devel-
Croonquist, M.J., Brooks, R.P., 1991. Use of avian and opments in Environmental Modelling, vol. 20. Elsevier
mammalian guilds as indicators of cumulative impacts in Science, Amsterdam.
riparian-wetland areas. Environmental Management 15, Lowe, J.D., 1995. Resident bird counts 1994. Journal of Field
701 – 714. Ornithology 66(Suppl.), 3 – 4.
Crozier, R.H., 1992. Genetic diversity and the agony of choice. May, P.G., 1982. Secondary succession and breeding bird
Biological Conservation 61, 11 –15. community structure: patterns of resource utilization. Oe-
Dessecker, D.R., Yahner, R.H., 1984. Black cherry-maple cologia 55, 208 – 216.
clearcut. American Birds 38, 71 –72. May, R.M., 1990. Taxonomy as destiny. Nature 347, 129 – 130.
Dufrene, M., Legendre, P., 1997. Species assemblages and McNeely, J.A., Miller, K.R., Reid, W.V., Mittermeir, R.A.,
indicator species: the need for a flexible asymmetrical Werner, T.B., 1990. Conserving the World’s Biological
approach. Ecological Monographs 67, 345 –366. Diversity. International Union for Conservation of Nature
Eiswerth, M.E., Haney, J.C., 1992. Allocating conservation and Natural Resources, World Resources Institute, Con-
expenditures across habitats: accounting for inter-species servation International, World Wildlife Fund — US, and
the World Bank, Gland, Switzerland and Washington, DC
genetic distinctiveness. Ecological Economics 5, 235 –249.
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
274
Miller, B., Reading, R., Strittholt, J., Carroll, C., Noss, R., Ray, G.C., 1988. Ecological diversity in coastal zones and
Soule, M., et al., 1998/1999. 1999. Using focal species in the
´ oceans. In: Wilson, E.O. (Ed.), Biodiversity. National
design of nature reserve networks. Wild Earth 12, 81 –92. Academy Press, Washington, DC.
Morrison, M.L., 1986. Bird populations as indicators of envi- Reid, W.V., Laird, S.A., Meyer, C.A., Gamez, R., Sittenfeld,
ronmental change. In: Johnston, R.F. (Ed.), Current Or- A., Janzen, D.H., et al., 1993a. Biodiversity Prospecting:
nithology. Plenum Press, New York, pp. 429 –451. Using Genetic Resources for Sustainable Development.
Murtaugh, P.A., 1996. The statistical evaluation of ecological World Resources Institute, Washington, DC.
indicators. Ecological Applications 6, 132 – 139. Reid, W.V., McNeely, J.A., Tunstall, D.B., Bryant, D.A.,
National Research Council, 1998. Forested Landscapes in
Winograd, M., 1993b. Biodiversity Indicators for Policy-
Perspective: Prospects and Opportunities for Sustainable
makers. World Resources Institute, Washington, DC.
Management of America’s Nonfederal Forests. National
Remsen, J.V., 1994. Use and misuse of bird lists in community
Academy Press, Washington, DC.
ecology and conservation. The Auk 111, 225 – 227.
National Research Council, 1992. Conserving Biodiversity: A
Robbins, C.S., Blom, E.A.T., 1996. Atlas of Breeding Birds of
Research Agenda for Development Agencies. National
Maryland and the District of Columbia. University of
Academy Press, Washington, DC.
Pittsburgh Press, Pittsburgh, PA.
Nicholson, C.P., 1997. Atlas of the Breeding Birds of Tennessee.
Shugart, H.H., 1984. A Theory of Forest Dynamics. Springer,
University of Tennessee Press, Knoxville, TN.
New York.
Niemi, G.J., Hanowski, J.M., Lima, A.R., Nicholls, T., Wei-
Simpson, R.D., Sedjo, R.A., Reid, J.W., 1994. Valuing biodiver-
land, N., 1997. A critical analysis on the use of indicator
sity: an application to genetic prospecting. Resources for the
species in management. Journal of Wildlife Management 61,
Future Discussion Paper 94-20, Resources for the Future,
1240– 1252.
Northern Forest Alliance, 2000. Investing in the Northern Washington, DC.
Forest: Emerging Conservation Opportunities and FY 2001 Smith, K.G., MacMahon, J.A., 1981. Bird communities along
Appropriation Priorities. Northern Forest Alliance, Mont- a montane sere: community structure and energetics. The
pelier, VT. Auk 98, 8– 28.
Office of Technology Assessment, 1988. Technologies to main- Tyrrell, L.E., Crow, T.R., 1994. Structural characteristics of
tain biological diversity. Office of Technology Assessment, old-growth hemlock-hardwood forests in relation to stand
US Congress, Washington, DC. age. Ecology 75, 370 – 386.
Palmer-Ball, B., 1996. The Kentucky Breeding Bird Atlas. Vane-Wright, R.I., Humphries, C.J., Williams, P.H., 1991.
University Press of Kentucky, Lexington, KY. What to protect? — Systematics and the agony of choice.
Pharo, E.J., Beattie, A.J., Binns, D., 1999. Vascular plant
Biological Conservation 55, 235 – 254.
diversity and a surrogate for bryophyte and lichen diversity.
Wiens, J.A., 1989. The ecology of bird communities. Founda-
Conservation Biology 13, 282 –292.
tions and Patterns, vol. 1. Cambridge University Press,
Pulliam, H.R., 1988. Sources, sinks, and population regulation.
Cambridge.
American Naturalist 132, 652 –661.
.
www.elsevier.com/locate/ecolecon
ANALYSIS
Maximizing conserved biodiversity: why ecosystem
indicators and thresholds matter
Mark E. Eiswerth a,*, J. Christopher Haney b
a
Department of Applied Economics and Statistics, Uni6ersity of Ne6ada, Reno NV 89557, USA
b
Conser6ation Science Di6ision. The Nature Conser6ancy, 4245 N, Fairfax Dri6e, Suite 100, Arlington VA 22203, USA.
Received 05 June 2000; received in revised form 24 January 2001; accepted 30 January 2001
Abstract
Accounting for biodiversity is important in several different types of constrained choice problems, including public
and private decisions for habitat and species conservation, the establishment of recreational parks and natural areas,
mitigation banking, and natural resource damage assessment (particularly primary and/or compensatory restoration
planning and scaling). In such applications it is important to give careful consideration to (1) the choice of
biodiversity indicator(s) to be used, and (2) the role of discontinuous, nonlinear ecological processes in light of the
decisionmaker’s chosen time horizon. The former is important because the choice of indicator(s) can substantially
influence decisions about conservation priority-setting and planning. The latter is critical for the same reason,
notwithstanding that dynamic ecosystem processes have rarely been considered sufficiently, if at all, in such
applications (in part because the processes usually are poorly understood or measured). In this manuscript we use
avian diversity data, collected by one of the authors, from hardwood forest ecosystems in the eastern United States.
We couple these data with estimates of species prevalence factors to construct a case study of how indicator choice
and consideration of ecological thresholds influence the outcomes of biodiversity preservation problems. We show
that (1) the choice of indicator(s) is critical, (2) failure to account for nonlinear, threshold effects in an ecosystem’s
future progression alters preservation decisions and ignores important information, (3) the effect of choosing different
time horizons depends on the indicator used, and (4) for any given biodiversity indicator, dynamic solutions can
depend on the time horizon chosen but not necessarily in monotonic or simple fashion. Our case study highlights the
importance of further system-specific research on dynamic ecological progressions as well as uncertainty regarding
future supply and demand for ecosystem service flows. © 2001 Elsevier Science B.V. All rights reserved.
Keywords: Biodiversity; Conservation; Preservation; Habitat; Forests; Birds
1. Introduction
* Corresponding author. Tel.: + 1-775-3275085; fax: + 1-
Despite the existence of accepted general defini-
775-7841342.
tions of biodiversity, debate continues over just
E-mail address: eiswerth@unr.edu (M.E. Eiswerth).
0921-8009/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 1 - 8 0 0 9 ( 0 1 ) 0 0 1 6 6 - 5
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
260
2. Background
what biodiversity is, how it should be measured,
and why it is important. Ecologists have
2.1. Rele6ant literature
defined a number of different types, or levels,
of biodiversity, with an increasing consensus
Biodiversity has long been recognized to be a
that no one indicator can or should be relied
multidimensional attribute of natural systems,
upon to characterize it. Different measures
with scientists referring to different levels of bio-
provide different indications of the variety
diversity including ecosystem, species, and ge-
and integrity of ecosystems, however, and the
netic diversity (Office of Technology Assessment,
choice of measures to use in a given context
1988; McNeely et al., 1990; National Research
depends on the research or policy objectives at
Council, 1992). Several years ago, Ray (1988)
hand.
observed that an ‘‘accounting of species alone
In previous research, we compared the out-
can be highly misleading as a yardstick of diver-
comes from applying different biodiversity indi-
sity’’, which led him to emphasize the impor-
cators to constrained choice problems of
tance of higher-order taxonomic diversity.
ecosystem/habitat preservation (Eiswerth and Atkinson (1989) placed this consideration in
Haney, 1992; Haney and Eiswerth, 1992). In clear perspective by stating that ‘‘given two
more recent research, one of the authors col- threatened taxa, one a species not closely related
lected a substantial amount of plant and animal to other living species and the other a subspe-
data from hardwood forest ecosystems in the cies of an otherwise widespread and common
eastern United States. The data collection pro- species, it seems reasonable to give priority to
ject was designed to investigate the ecological the taxonomically distinct form.’’
importance of old growth via comparisons to Observations such as these have encouraged
younger seral (successional) stages of hemlock- the development of measures that use taxonomic
northern hardwood forest (Haney, 1994, 1995; information (May, 1990; Altschul and Lipman,
1990; Vane-Wright et al., 1991) or information
Haney and Schaadt, 1995). In this manuscript
from limited molecular sequences (Crozier, 1992;
we use a portion of these data to construct a
Faith, 1992). Researchers have also used genetic
case study of how the choice of biodiversity in-
distinctiveness data to indicate biodiversity, by
dicators may affect constrained choice problems,
incorporating genome-wide data and linking
for example, public decisions related to habitat
composite information about an organism’s en-
conservation, restoration, or mitigation activi-
tire genetic makeup to data on species richness
ties. In addition, this case study illustrates the
(Eiswerth and Haney, 1992). This is the kind of
dynamic considerations that are important to
information that can be useful in many contexts,
such decisions. The forest ecosystem we focus
including (but not limited to) the search for spe-
on is characterized by nonlinear changes over
cies that have pharmaceutical and other values
time in structure and function, with discontinu-
(e.g. Reid et al., 1993a; Simpson et al., 1994).
ities occurring as the ecosystem moves from one
In setting priorities for conservation, relevant
developmental stage to the next. As a result,
metrics may include combinations of indicators
biodiversity in this system is a discontinuous
that reflect both diversity and the amount of
function of time. This has implications for diversity at risk. For example, species risk fac-
problems in which the desired outcome tors can be combined with taxonomic distinc-
is to maximize the flow of future services pro- tiveness indicators to yield a layered proxy (e.g.
vided by biodiversity. We show how the dy- Haney and Eiswerth, 1992). Such layered indica-
namic solution to a biodiversity preservation tors illustrate how decisions comparing diversity
problem may depend significantly on the time among regions can change as more (and better)
horizon considered and the biodiversity indica- information is considered in addition to simply
tor used. species richness. Reid et al. (1993b) provided an
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 261
informative summary of a wide range of indicators tion banking, and (5) natural resource damage
useful for policymakers, including ones that em- assessment (NRDA), particularly primary and/or
body risk. Such indicators are important in applied compensatory restoration planning and scaling.
decision-making because direct measures of ecosys- Forest biodiversity receives wide attention be-
tem value are in most cases unavailable, insuffi- cause of the multiple ecological, social, and eco-
cient, or too expensive to develop using standard nomic values associated with forest ecosystems
valuation methods (King, 1997). Indicators that (National Research Council, 1998). Our case study
are easy to use, are applicable to large areas, and involving eastern forests is particularly relevant
have a close linkage with specific elements, pro- given that decision-makers are currently attempt-
cesses, or qualities of ecosystem integrity are likely ing to determine the optimal mix of management
to be the most useful (Bradford et al., 1998; Miller regimes for sustainable forests. For example, indi-
et al., 1998/1999). viduals in Maine recently expressed an interest in
To model ecological attributes of ecosystems purchasing lands from timber companies to create
realistically, it is necessary to consider dynamic a large reserve in which forests would stand undis-
thresholds and other nonlinear processes in system turbed (Northern Forest Alliance, 2000). In this
structure and function. Such dynamic processes are and related situations, one of the relevant choice
rarely considered sufficiently, if at all, in exercises problems is, or at least ought to be: ‘Given a set
such as habitat protection, restoration, or conser- of forest tracts and a budget constraint for preser-
vation planning. Nonlinear, threshold processes vation, what is the optimal mix of conservation
are considered even less frequently, in part because efforts (or more broadly, management regimes)
they usually are poorly understood or measured. that maximizes the preservation of biodiversity?’
The importance of such processes is sometimes at The answer depends on the way in which the
least recognized in the literature (e.g. King, 1997), problem is formulated and the characteristics of the
but to date their incorporation in decision-making candidate conservation areas. While this
is woefully inadequate. manuscript deals solely with indicators of biodiver-
sity rather than the broader (and more complex) set
2.2. Pertinent concepts and applications of potential indicators of all ecosystem functions
and services, we recognize that in many decision
Concepts about biodiversity that we explore in contexts such broader indicators are generally of
this manuscript include: (1) the choice of biodiver- interest. We focus on biodiversity per se as one
sity indicator does matter, and can drive conserva- characteristic of natural systems, and show that
tion decisions, (2) it is important to account for consideration of even one such characteristic is in
dynamic ecosystem processes, and decision rules itself a complex step.
that do so may yield quite different results from
those that do not, (3) for any given indicator of
biodiversity, investments in conservation may de- 3. Case study forest areas: characteristics and
pend on the time horizon considered, but not data
necessarily in monotonic fashion, and (4) the effect
on the dynamic solution of changes in the time This case study is based on avian data collected
horizon may depend upon the biodiversity indica- from over 20 study plots in hemlock-northern
tor used. hardwood forest. Numerical values for avian pop-
These concepts have relevance for a number of ulations and communities were obtained from field
different activities and decisions. Examples include: studies conducted in Clearfield, Potter, and McK-
(1) decisions related to the purchase of land for ean counties on the Allegheny Plateau, Pennsylva-
conservation easements, (2) the establishment of nia (unpubl. data, J.C. Haney, collected
new recreational parks or natural areas, (3) agency 1992–1994; Dessecker and Yahner, 1984). Cen-
priority-setting for habitat and/or species conserva- suses were conducted in each of five forest age
tion expenditures, (4) decisions involved in mitiga- classes: 4, 9, 50, 120, and 300+ years. Forest age
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
262
was computed as the time elapsed since the last first derived from random subsampling of study
stand-replacing disturbance (either catastrophic plots available from this forest type (N= 21).
windthrow or even-aged timber harvest). These Because there are other potential biases to c,
five classes are termed early, transitional, mid-suc- estimates also conformed to the following criteria:
cessional, late successional, and old growth, re- visiting or wandering bird species were eliminated;
spectively. Hemlock-northern hardwood forest data collection was standardized by sampling fre-
displays temporal discontinuities in vegetation quency (eight visits) and area (each plot was of
structure, threshold effects, and other nonlinear equal size — 6 hectares (James and Rathbun,
patterns in successional development (see, e.g. 1981); sampling was conducted wholly within a
Tyrrell and Crow, 1994). single habitat type; and study plots were located
Taxonomic groups can be used as indicators in within large tracts of consolidated forest that were
two fundamentally different ways: as proxies for not in close proximity to other habitats (Remsen,
biodiversity and as proxies for environmental con- 1994).
ditions. For a variety of reasons, focusing on a Following application of the criteria above, the
diverse taxon such as birds is useful since a num- resulting data were combined with other informa-
ber of structural and functional elements of the tion sources to develop multiple indicators of
environment are automatically integrated. As a biodiversity as well as biodiversity at risk. First,
group, birds require very diverse microhabitats numbers of bird species (S) and higher taxa (gen-
arising from structural attributes related to stand era [G], families [F]) were computed for each of
and floristic composition, snag availability, foliage the five forest age classes. Next, we calculated a
height diversity, horizontal complexity, core area, layered proxy (Sa) that combined species richness
and local moisture conditions (Wiens, 1989). Bird with local (physiographic province) population
communities also exhibit marked, well-docu- species prevalences derived from Breeding Bird
mented differences in assemblage structure associ- Atlas programs in nine contiguous states in the
ated with forest developmental sequences northeastern United States (Laughlin and Kibbe,
(Lanyon, 1981; Smith and MacMahon, 1981; 1985; Andrle and Carroll, 1988; Brauning, 1992;
May, 1982; Glowacinski and Weiner, 1983; Helle, Bevier, 1994; Buckelew and Hall, 1994; Foss,
1984). Compared to other taxonomic groups, 1994; Palmer-Ball, 1996; Robbins and Blom,
birds perform quite well as indicators of specific 1996; Nicholson, 1997). This layered proxy Sa was
environmental conditions (Morrison, 1986; computed as:
Croonquist and Brooks, 1991). However, because Si
Sa = % [1− LPi ]
a few species do not always serve as accurate (1)
i=1
substitutes for many others (Niemi et al., 1997),
we make no assumption that this single taxon where LPi denotes the prevalence factor for spe-
serves as a suitable proxy for other species group- cies i at the local (physiographic province) scale.
ings or biodiversity in general (but see Pharo et The prevalence factor from the Breeding Bird
al., 1999). Atlas data can assume any value between 0 and 1,
We used bird species richness derived from inclusive. For example, a value of 0.50 for local
breeding bird census methodology (Lowe, 1995) species prevalence means that the species is found
as the initial proxy for forest biodiversity. A on 50% of the land area at the level of the
number of approaches have been proposed to physiographic province studied (in this case, the
estimate total species richness, C, within an area Appalachian Plateau of Pennsylvania). As the
(Bunge and Fitzpatrick, 1993). For comparisons average prevalence of a collection of species rises,
across forest development (seral) stages, however, the value of Sa for the collection falls. Weighting
we required only a bias-free estimate of relative species richness in this manner thus provides us
species richness, c. This approach is equivalent to with an indication of not only (1) the number of
the data-analytic class of methods reviewed by species present in our study area, but also (2) the
Bunge and Fitzpatrick. Point estimates of c were subset of those species present that are not preva-
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 263
lent at a larger geographic scale. This metric of the five different forest age classes (seral
provides information somewhat similar to that stages). The values for each of these indicators, by
offered by specificity indicators reflecting the oc- forest stage, are shown in Table 1. Table 1 also
currence (abundance) of species within a given indicates the percentage of species that were
geographic space or ‘cluster’ of sites (Dufrene and uniquely detected within each seral stage. This
Legendre, 1997; Legendre and Legendre, 1998). illustrates that each forest seral stage displays its
Finally, we computed a similar indicator (Sb) own particular set of species.
by weighting species richness again, this time by Of course there are additional indicators that
regional population prevalence as calculated from one could develop and use. For example, one of
the Breeding Bird Atlas programs. Sb was com- the factors that a conservation planner may wish
puted as: to consider might involve the relative scarcity of
different forest types, in combination with the
Si
Sb = % [1− RPi ] (2) number of species unique to each type. Such a
i=1
metric would provide somewhat different infor-
mation when compared to indicators Sa and Sb.
where RPi denotes the prevalence factor for spe-
However, note that Sa and Sb do explicitly incor-
cies i at the regional scale. This indicator weights
porate the underlying relative scarcity of habitat
species richness to reflect those species present in
types that play host to each particular species
our study area that are not common at the re-
considered. These indicators do this by weighting
gional level (in this case, across the northeastern
each species by the percentage of land (on either a
United States). As the number of species in a
local or regional basis) on which the species is
forest age class that are not prevalent regionally
estimated to occur (and hence the percentage of
goes up, Sb rises as well.
land that currently provides habitat suitable to
The work described above yields multiple indi-
cators of diversity or diversity/prevalence for each each particular species). To the extent that a
Table 1
Indicators of biodiversity in Pennsylvania hemlock-northern hardwood forest plots of different seral stagesa
Forest seral stageb,c
Indicators
Early (15.2%) Transitional Mid-successional Late successional Old growth
(31.2%) (41.2%) (12.0%) (0.4%)
Total number of bird species 9 17 20 34 20
% Bird species uniquely detected in 22 24 10 29 40
seral stage
Total number of bird genera 9 17 16 25 15
Total number of bird families 2 9 8 11 10
Species richness weighted by 2.5 4.3 5.9 12.9 10.1
physiographic province (local)
population prevalence (Sa)
Species richness weighted by 2.6 5.0 7.2 15.9 11.5
regional population prevalence
(Sb)
a
Sources of data: J.C. Haney, unpubl. data collected 1992–1994; Dessecker and Yahner, 1984; Laughlin and Kibbe, 1985; Andrle
and Carroll, 1988; Brauning, 1992; Bevier, 1994; Buckelew and Hall, 1994; Foss, 1994; Palmer-Ball, 1996; Robbins and Blom, 1996;
Nicholson, 1997.
b
Early seral stage = stand age of 4 years; transitional = 9 years; mid-successional =approx. 50 years; late successional = approx.
120 years; old growth =300+ years.
c
The relative prevalence of each seral stage forest type found in Pennsylvania (as a percentage of total forestland) is given in
parentheses under the seral stage names. These relative prevalence values are derived from Alerich (1993).
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
264
particular species is associated with (unique to) however, at least one of the forest types contained
only one forest type, Sa and Sb embody the species not commonly found elsewhere, then Sa
relative scarcity of that forest type. and Sb would hold great meaning for objectives
As a second example, a relative measure such as such as preserving population sources, preventing
Sa/S, where S denotes number of species, may be further habitat fragmentation, increasing wildlife
of value in certain situations with particular con- corridors, and providing recreational services (i.e.
servation management objectives. Such a relative viewing rare species) even to visitors from far
measure could provide a higher indicator value away.
for a region that has very few species (e.g. 10) but
where a high percentage of those species are rare,
as compared to a region with many more species 4. Static approach and results
(e.g. 100) but relatively few rare ones. Some con-
4.1. Static illustration 1: choice of biodi6ersity
servation decision contexts may call for placing a
premium on rarity (and ignoring the absolute indicator can dri6e habitat rankings and thus
number of species) and in such cases a measure discrete choices regarding habitat conser6ation
such as Sa/S may be useful. At the same time, the
attraction of Sa by itself is that it does combine The simplest problem involves a discrete choice
two different kinds of information: species rich- problem of conservation. In such cases a decision-
ness and species rarity. maker may be interested in choosing a subset of
We do not attempt in this manuscript to iden- all geographic areas (one, in the simplest case) in
tify any one best indicator; indeed, indicators which to devote habitat conservation efforts. This
need to be matched carefully to management ob- situation may occur when available funding for
jectives since the choice of indicator will influence conservation is sufficiently constrained. It also
the decision outcome. Various alternative indica- may occur in processes that involve mitigation
tors, including but not limited to those in Table 1, banking or compensatory restoration for lost nat-
have different meanings with respect to conserva- ural resource service flows.
tion objectives and social/economic values. As an Table 1 reveals several points relevant to rank-
example, a skilled birdwatcher may attach a great ing our case study forest types. Perhaps the most
deal of importance to the sheer number of species apparent feature is that regardless of the indicator
that he or she is able to see, on average, upon chosen, the decision-maker would rank the late
visiting the forest. In contrast, an avid hiker or successional forest first in terms of biodiversity as
angler untrained in birdwatching may derive plea- well as biodiversity weighted by prevalence. This
sure from the incidental viewing of a wide variety forest stage dominates the others in species rich-
of birds while recreating, but may be unable to ness (34 species), higher taxa diversity (25 genera),
discern (or uninterested in noticing) differences species richness weighted by physiographic
among species that are closely related. Such an province (local) prevalence, and species richness
individual may attach more importance to the weighted by regional prevalence.
indicators in Table 1 that relate to the total Rankings of forest stages below the late succes-
number of bird genera or families, rather than sional are more problematic. If number of species
species richness. is used as a biodiversity indicator, the decision-
Similarly, the importance of indicators such as maker’s second choice for conservation efforts
Sa and Sb in comparison with the others depends could be either the mid-successional or old-
largely on the extent to which the conser- growth seral stage. If, however, the number of
vation planner’s objectives are tied to a broader genera were used as an indicator of higher taxo-
spatial (e.g. regional) context. If none of the spe- nomic diversity, the decision-maker would pick
cies under consideration is rare in terms of preva- the transitional age class over both the mid-suc-
lence within a larger spatial area, then the cessional and old growth as the second priority
meaning of Sa and Sb would be minimal. If, for conservation efforts.
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 265
Consideration of species prevalence factors from one another, or with differences in visitor
yields even more interesting results. First, on aver- profiles across the areas. Public preferences may
age, species found in the early, transitional, and also be of a type such that it is important that
mid-successional forest types have high prevalence natural attributes (such as species) can be enjoyed
values (relative to older forests) and correspond- in multiple areas, even when the areas are not that
ingly lower species rarity factors. In contrast, on far apart. If demand for natural areas and the
average, species found in older forests are less service flows (e.g. birdwatching) that they offer is
prevalent at broader spatial scales. If the indicator high relative to supply (this is the case in many
is defined as species richness weighted by preva- wildlife refuges today), then congestion comes
lence at either the physiographic province (Sa) or into play to make the presence of a species in one
regional (Sb) levels, then the old growth forest age area a poor substitute for its presence in another.
class becomes the clear second choice for conser- Consideration of risk and uncertainty provides an
vation efforts. additional basis for this assumption. As discussed
The usefulness of Sa and Sb as indicators is now in King (1997), uncertainty exists regarding the
clear, in that they attach a premium to forest age effects on ecosystems of future natural and an-
classes containing species that are not common. thropogenic changes. Since we do not know how
Such forest areas are potential population future natural changes or human activities close
‘sources’ (Pulliam, 1988) of species not prevalent to natural areas may affect their structure and
at broader scales. In Section 3 we also mentioned function, a motivation exists to expend conserva-
other possible indicators, including relative mea- tion efforts in multiple areas, even if they offer
sures such as Sa/S that attach complete impor- similar ecosystem services today.
tance to relative species rarity with no weight Given this assumption, a decision-maker that is
attached to the number of species. Such indicators concerned with habitat conservation in multiple
would give a higher priority to old growth forests. areas may wish to maximize the sum of biodiver-
sity across the areas, subject to a budget con-
4.2. Static illustration 2: choice of biodi6ersity straint for conservation efforts. Consider the
indicator can significantly influence the allocation following relationship between conservation ex-
of conser6ation expenditures among multiple penditures and an indicator of biodiversity:
geographic areas
bi = ki + fi (Mi ) (3)
In some instances a decision-maker may need where bi is the expected value of a biodiversity
indicator in area i, ki E 0 is the expected value of
to make decisions regarding the allocation of
habitat conservation efforts in multiple geo- a biodiversity indicator in area i given no conser-
graphic areas, rather than a discrete choice of vation expenditures in area i, Mi denotes conser-
which area(s) to conserve. Such decision-making vation expenditures in area i, and where
f %(Mi )\ 0, f %%(Mi )B 0. The function fi (Mi ) de-
requirements provide a richer context for analysis. i i
We assume that characteristics of one geo- notes the addition to the level of the biodiversity
graphic area are not substitutes for the same indicator expected to result from conservation
characteristics found in another geographic area. expenditures Mi.
In our case study this means, for example, that Eq. (3) and the equations that follow are writ-
the presence of a species in a forest of a given ten in the standard economic format of maximiz-
seral stage is not a substitute for its presence in ing a variable subject to a constraint on
another forest area of a different stage. This is not expenditures (M). However, the term M can also
a restrictive assumption, but rather is consistent be interpreted more broadly as a money metric
with a number of real-world contexts. For exam- equivalent of efforts devoted to conserving biodi-
ple, it is consistent with a situation in which a versity. Similarly, it is possible to interpret the
decision-maker is interested in devoting efforts to term b in (1) as a function of conservation efforts
natural or recreation areas some distance apart rather than expenditures. One reviewer of this
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
266
article has wisely pointed out that some important
conservation management steps may require less
out-of-pocket expenditure and more good (and
timely) planning, relative to alternative manage-
ment efforts. However, the standard economic
view is that such good and timely planning would
come at an opportunity cost, for example, hiring a
well-trained (and presumably well paid) ecologist
or wildlife biologist to spend part of his or her
time on the conservation planning process. For
this reason as well as ease of exposition, we refer
to M as conservation expenditure while realizing
that a more complex indicator of conservation
effort is also possible. Fig. 1. General shape of illustrative function linking expected
biodiversity to conservation expenditure.
Consider a case in which a decision-maker is
interested in two different geographic areas in
make the simplifying assumption that k1 = k2 = 0.
different forest age classes. If the decision-maker
Relaxation of this assumption could change the
is interested in allocating conservation expendi-
numerical solution to the problem, but would not
tures between these areas, a relevant constrained
change the flavor of the concepts and results upon
maximization problem is:
which we focus in this manuscript.
Max[k1 + f1(M1)+k2 +f2(M2)] Second, we assume for illustration that the
conservation expenditures necessary to set the ex-
s.t.:M1 +M2 =M (4) pected level of the biodiversity indicator equal to
the baseline (existing) level of biodiversity are
where M denotes the total resources available to
equal across the two geographic areas. For exam-
the decision-maker.
ple, if conservation expenditures involve purchas-
As an illustration of the way in which the
ing land, this assumption would denote that land
choice of biodiversity indicators affects the solu-
costs are equal for the two areas. For our illustra-
tion, we consider a specific case of the generalized
tions, we use a specification that is consistent with
problem. First, assume for simplicity that k1 =
these assumptions as well as the standard eco-
k2 = 0; that is, in both areas, the indicator of
nomic assumption of diminishing returns to
biodiversity is expected to be zero if conservation
expenditures:
efforts are zero. This is a special case of the more
general case ki E 0 and corresponds to a situation fi (Mi )= (b 0)(M) − 1/2(Mi )1/2 (5)
i
where a decision-maker is interested in protecting
where i denotes forest area i and b 0 the baseline
all or a portion of a land area from complete
development, e.g. total conversion of land into (current) level of biodiversity there. The general
housing subdivisions, a relevant scenario in many shape of this function is shown graphically in Fig.
parts of the United States. That is, the special case 1, which illustrates that additional conservation
is that if the decision-maker makes no conserva- efforts purchase a higher level of expected biodi-
tion expenditures, then complete habitat destruc- versity but at a diminishing rate. Though we
tion will occur. We certainly recognize that in assume for simplicity in our numerical analyses
that ki = 0, Fig. 1 depicts the more general case in
reality biodiversity does not necessarily equal zero
which ki \ 0 (some biodiversity will remain if no
even when land is completely developed. How-
ever, we have not collected data on biodiversity conservation efforts are undertaken).
for our case study bird species in a formerly The first-order condition, which gives the solu-
comparable area (e.g. close to our study sites) that tion to the constrained maximization problem
has been deforested and developed. Therefore we (noneconomists may see Chiang (1974) for an
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 267
introduction to constrained maximization) shown conservation planning. Some natural systems have
in Eq. (4), is: a very large number of species but relatively low
diversity at higher taxonomic levels. In contrast,
(b 0)(M1) − 1/2 =(b 0)(M2) − 1/2. (6) some systems, for example, some marine and
1 2
coastal ecosystems, are strikingly rich in their
To show the implications of using alternative endowment of diverse families with relatively few
indicators of biodiversity as input to the con- species representing each of those families (Ray,
strained maximization problem, we consider the 1988).
case in which forest areas 1 and 2 are currently in When species richness is weighted by regional
transitional and mid-successional stages, respec- population prevalence (to form the indicator Sb),
tively. The comparison between these two stages
the allocation of expenditures shifts substantially
is interesting because neither one dominates in
toward the older (mid-successional) forest. Using
terms of biodiversity.
this indicator, 68% of total conservation expendi-
The results from using alternative indicators are
tures will be targeted toward the mid-successional
summarized in Fig. 2. Clearly the choice of indi-
forest class. This outcome reflects the area’s abil-
cator can influence decisions on how to allocate
ity to act as a source for species that are not
efforts. In our illustration, the difference between
highly prevalent on a wider regional basis. As
using species richness and a higher taxa diversity
shown in Fig. 2, for the three indicators exam-
indicator is significant (33% more expenditures
ined, the outcome may range from a low of 44%
devoted to the transitional forest area using a
to a high of 68% of total available conservation
higher-taxa indicator rather than species richness).
expenditures being devoted to the older forest
For the subclass of problems where a decision-
area. The sensitivity of the solution to the choice
maker is interested in purchasing land or prevent-
of indicator illustrates the potential volatility of
ing development so as to preserve biodiversity,
decision-making processes to the types of infor-
even the differential found in our illustration
mation considered.
would lead to a difference in the portfolio of
forest areas that the planner chooses to buy/pro-
tect. In some cases, the choice between these two
indicators can have a substantial influence on
5. Dynamic approach and results
The relatively small subset of structural at-
tributes that exhibit temporal linearity, and the
threshold changes that occur in forests during
succession, create distinct stages in forest ecosys-
tems. To adequately characterize such ecosystems,
time-varying stages and threshold effects must be
taken into account. As in other ecosystems, diver-
sity in our case study system is time-scale depen-
dent, that is, dependent upon time from the most
recent disturbance.
As a result, it is important to consider not only
the current levels of diversity in particular areas,
but also the diversity levels that the areas can
potentially offer society in the future. The general
problem may be viewed as choosing management
options to maximize the expected ‘flow’ of diver-
sity from the present to some point in the future,
Fig. 2. Allocation of conservation expenditures: static frame-
subject to a budget constraint.
work.
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
268
Forests can experience abrupt structural seral stage. If the decision-maker is interested in
changes in either of two directions. Successional maximizing biodiversity over this period, then the
changes occur as the forest moves through growth relevant maximization problem is:
phases (seres), each consisting of varying intervals
T
when structure is relatively constant but where
Max % [ f 1(M 1)+ f 2(M 2)]
t t
rapid transition occurs between seres. Succession t=1
can also be reversed, and the entire set of ecolog-
s.t.: M 1 + M 2 = M
ical processes renewed, when catastrophic distur- (7)
bance (either man-made or natural) shifts forest
where M 1, M 2, and M may be thought of as the
structure back to an earlier sere. In the forests
used in our analysis, natural disturbances that discounted present values of the opportunity costs
completely remove canopy trees occur very rarely, of conservation that are incurred between now
about every 1200 years (Canham and Loucks, and period T (expressed in this way to simplify
1984; Frelich and Lorimer, 1991). At large land- the exposition). As with the static case, M is the
scape scales all successional stages can be main- total amount of resources available for conserva-
tion, and M 1 and M 2 are the amounts to be
tained in perpetuity, although not always in the
same amount or location (Shugart, 1984). In allocated to forest areas 1 and 2, respectively.
other words, by protecting relatively large areas of For most natural systems, scientists have not
forest, it is possible to ‘purchase’ increased levels collected continuous data on the ways in which
of certainty that a forest area will progress as various indicators of biodiversity change over
anticipated through its natural growth phases. time. At best, a limited set of observations may
In this section we provide illustrations of the exist for particular stand ages in forests, for exam-
relevance of natural dynamic processes. Section ple. In other cases, very little direct information is
5.1 illustrates the importance of recognizing that available. For our case study, we have the benefit
change may not be linear, and highlights the need of possessing standardized observations of bird
for better data on how and when ecosystems diversity in forests that are very similar (in terms
encounter thresholds. Section 5.2 shows that of climate, geographic zone, etc.) except that they
choice of time horizon and biodiversity indicator are of different ages.
may affect the dynamic solution to preservation, To illustrate the importance of knowing how
but not necessarily in the expected ways. Section natural systems evolve, suppose for a moment
5.3 illustrates how a dynamic approach may dif- that all that we knew about the problem was the
ferentiate natural areas that look equivalent from current number of species and higher taxa for
a static viewpoint. both areas, as well as the same information for
the late successional stage that both areas are
5.1. Dynamic illustration 1: the importance of expected to evolve into over the next 100 years.
Assume further that diversity is expected to in-
accounting for nonlinear, discontinuous ecological
crease in linear fashion over time in either forest
processes
area. In this case, the choice problem would in-
volve choosing M 1 and M 2 to maximize the sum
Consider the case in which a decision-maker
wishes to maximize the sum of a biodiversity of the areas under the (linear) biodiversity time
indicator across two different forest areas, the first paths in the two forests. Solution of this problem,
stand of 30 years and the second stand of 90 using species as a biodiversity indicator and the
years. We assume that, for each of the seral stages same diversity–expenditure functional forms
we examine, diversity is characterized by the ob- shown in Eq. (5), would provide the answer that
servations shown in Table 1. Suppose that the 42% of the available resources (opportunity costs)
decision-maker’s time horizon, T, is 100 years. It for conservation would be devoted to forest area
is expected that, by time T, both of these forest 1 (transitional), and 58% to forest area 2 (mid-
areas will have evolved to the late successional successional).
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 269
Table 2
Allocation of conservation expenditures among transitional and mid-successional forest areas, under alternative time horizons and
indicators of biodiversitya,b
Time horizon Number of species Number of families Species richness indicator
indicator (S) indicator (F) weighted by regional population prevalence (Sb)
Time horizon= 50 Area 1: 44% Area 1: 55% Area 1: 36%
Area 2: 56% Area 2: 45% Area 2: 64%
Time horizon= 100 Area 1: 37% Area 1: 47% Area 1: 29%
Area 2: 63% Area 2: 53% Area 2: 71%
Time horizon= 150 Area 1: 41% Area 1: 47% Area 1: 36%
Area 2: 59% Area 2: 53% Area 2: 64%
a
Forest area 1 is currently in the transitional seral stage with a stand age of 10 years. Forest area 2 is currently in the
mid-successional seral stage with a stand age of 50 years.
b
Each percentage in the table denotes the percentage of total conservation expenditures that will be devoted to a forest area,
according to the solution of the dynamic constrained maximization problem defined in the text.
5.2. Dynamic illustration 2: the choice of time
Now consider the problem given our knowledge
horizon and biodi6ersity indicator may ha6e a
that the biodiversity time path more closely re-
significant impact on the dynamic solution, but
sembles a step function than a linear function. It
not necessarily in monotonic fashion
is intuitively clear that forest area 2, currently at
stand age 90, will enter the late successional seral
Consider once again the allocation of conserva-
stage significantly sooner than forest area 1. Once
tion expenditures between a current transitional
the late successional stage is reached, the forest
forest area and a current mid-successional area.
area will exhibit higher levels of biodiversity as
Given the knowledge that the biodiversity time
measured by numbers of species, genera, or
path is subject to discontinuities as forests move
families. Therefore, one would expect that, if we
from one seral stage to the next, how does the
take account of the step function nature of the
choice of time horizon affect the solution to the
biodiversity time path, a premium would be
problem in Eq. (7)? And how does the choice of
placed on conservation efforts in forest area 2.
biodiversity indicator influence the result? We
Solution of the maximization problem accounting
consider three alternative time horizons (50, 100,
for a stepwise progression bears this out: using
and 150 years) and three alternative indicators (S,
species again as an indicator of biodiversity, the
F, and Sb). The solutions to the problem under
solution would involve only 33% of conservation
these conditions are shown in Table 2.
efforts in forest area 1 (vs. 42% assuming a linear
Three main points emerge from Table 2. First,
time path), with 67% of efforts now devoted to
conservation allocation outcomes vary signifi-
forest area 2.
cantly according to choice of indicator and time
The difference in solutions under linear and
horizon, from a low of 29% to a high of 55% of
step function approaches is perhaps not that strik-
total expenditures devoted to forest area 1. Sec-
ing for the particular example we have chosen,
ond, the dynamic solutions depend on the time
though it is significant. The salient point is that
horizon chosen but not necessarily in monotonic
the incorporation of information on threshold
fashion. For example, Fig. 3 illustrates how the
effects can affect the decision-making process.
allocation of expenditures varies according to T
Certainly there are cases in which accounting for
when species richness is used as an indicator of
these effects may have a substantial bearing on
biodiversity. The optimal percentage of resources
the planner’s decision, depending on the natural
to be devoted to forest area 1 (transitional) first
systems and time horizons considered.
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
270
50 to 100 (Table 2), for the same reasons de-
declines as the time horizon is increased from 50
scribed above for species. However, extension of
to 100 years, then rises as T goes from 100 to 150.
T from 100 to 150 leaves the solution unchanged
This is because the expected passage of both
with number of families as the indicator, unlike
forest areas into the late successional stage, which
the pattern under the species indicator. This is due
exhibits markedly high biodiversity, is considered
to the way in which species appear and disappear
to varying degrees according to the chosen time
as the forest moves through seral stages. Specifi-
horizon. With T=50, passage to the late succes-
cally, the number of species may increase or de-
sional stage is considered for neither forest area,
crease through time without there occurring a
and so the current levels of biodiversity largely
change in diversity as measured at higher taxo-
drive the result. With T = 100, passage of forest
nomic levels. The same kind of pattern can occur
area 2 to the late successional stage is taken into
for genetic diversity, i.e. if closely related species
account while that of forest area 1 is not. As a
appear or disappear through time, species diver-
result, a premium is attached to conserving forest
sity may change significantly while genetic diver-
area 2 and the percentage of total expenditures
sity does not.
devoted to it rises. With T = 150, the passage of
both forest areas to the late successional is consid-
5.3. Dynamic illustration 3: a dynamic approach
ered, and so emphasis shifts back toward a some-
may differentiate areas that are equi6alent from a
what higher level of emphasis on forest area 1.
static perspecti6e
While the shifts in expenditures for this illustra-
tion may not be dramatic, they are indicative of
Now consider the allocation of expenditures
the implications of choice of T for preservation
between two forest areas both currently in the late
decisions in general.
successional stage but that have different stand
Third, the influence of altering T depends on
ages. Specifically, consider forest areas 1 and 2,
the indicator of biodiversity that is used. For
which have stand ages of 125 and 250 years,
example, using number of bird families as an
respectively. Assume that these areas display simi-
indicator, conservation efforts devoted to forest
lar numbers of species, genera, and families. The
area 2 (mid-successional) increase as T goes from
main difference between them is that forest area 2
will evolve into an old growth forest 125 years
sooner than forest area 1.
Of course, if the two areas currently are similar
in terms of biodiversity, a static approach would
give them equal weight regardless of the indicator
used. However, one does not necessarily give
them equal weight if dynamics are taken into
account. The solutions to Eq. (7) for this problem
are shown in Table 3. With T= 50 years, the two
areas have equal weight because neither one will
have progressed out of the late successional seral
stage. With T= 100 or 150, however, the younger
forest area (1) will be accorded a significantly
higher percentage of conservation efforts (61 and
66% of the total for T=100 and T= 150, respec-
tively). As the time horizon increases, then, a
decision-maker interested solely in maximizing the
biodiversity indicator will attach more importance
Fig. 3. Allocation of expenditures may be nonmonotonic with
to conserving the younger late successional forest
respect to time horizon T: dynamic framework using species
area.
richness.
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 271
Table 3 argument against conserving old growth. How-
Allocation of conservation expenditures between two late suc-
ever, it does indicate strongly that decision-mak-
cessional forest areas of different stand age: dynamic
ers should clearly and deliberately prioritize
frameworka
conservation objectives on a site-specific basis, as
well as recognize that particular objectives may
Time horizon Relative expenditures by area, using
species richness indicator sometimes lead to decisions that run counter to
conventional wisdom.
Area of stand Area of stand
age =125 age =250
6. Conclusions
50 years 50% 50%
100 years 61% 39%
150 years 66% 34% We have used data from forest ecosystems to
illustrate several key concepts relevant to biodi-
a
Each percentage in the table denotes the percentage of
versity. First, the solution to a static biodiversity
total conservation expenditures that will be devoted to the
preservation problem may depend significantly on
corresponding forest area, according to solution of the dy-
namic constrained maximization problem defined in the text. the biodiversity indicator used. This is an impor-
tant concept for decision-makers to understand
This result may stand at odds with expecta- and assess, particularly at the site-specific level.
tions, given that the older forest area (2) will The use of alternative indicators to examine the
progress to old growth 125 years sooner, and multiple attributes of natural systems, and the
given the importance that society generally associ- extent to which those attributes are at risk, can
ates with old growth forest. The result is driven force a useful reexamination of conservation ob-
by the fact that the old growth seral stage is jectives. The choice of final indicators to use as
actually less diverse (as measured both by number guides may vary greatly from case to case and will
of species and number of higher taxa) than the depend on the context of the problem and the
late successional seral stage. Therefore, conserva- ecosystem services that are most highly valued by
tion decisions made solely on the basis of antici- the public.
pated biodiversity will tend to favor the late Second, for any given indicator, dynamic solu-
successional stage over old growth, and therefore tions may differ from the static solution, depend-
result in the conservation of younger forests. This ing on the time horizon chosen by the
is an issue that may arise in a number of different decision-maker. This forces a reexamination of
types of forest systems since evidence suggests the timeframes that we wish to take into account
similar patterns in a variety of forest types. when considering future streams of ‘biodiversity
Clearly, there may be other reasons to value old services’, or ecosystem functions and services
growth forest besides numbers of species or higher more broadly. This is a simple concept, but the
taxa (Brunson and Shelby, 1992). For example, existing literature does not adequately address it,
note that in Table 1 we show that a relatively particularly for cases in which ecosystems are
large number of the species found in old growth expected to display discontinuous processes in the
were uniquely detected in that forest type. Second, future. Our analysis also highlights the need for
a relative measure such as Sa/S (which prioritizes dynamically adaptive management, rather than a
areas solely according to the percentages of their long-term fixed formula for conservation, since
species collections that are not prevalent at a the portfolio of biodiversity and forest types will
broader scale) would attach high importance to continue to change as time passes.
old growth. Third, individuals may exhibit prefer- Third, for any given indicator, dynamic solu-
ences for recreation in old growth forest because tions can depend on the time horizon chosen, but
of factors totally unrelated to biodiversity. The not necessarily in monotonic fashion. This is a
counterintuitive result of this illustration certainly characteristic not common to well-behaved dy-
is not (and in no way is intended to be) an namic models and therefore merits special atten-
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
272
tion. Fourth, the effect of changes in the time goods) may affect future patterns of demand.
horizon on the dynamic solution is dependent on Such changes are equally uncertain.
the indicator used, which reinforces the need to Future research might assess the relative impor-
consider multiple proxies. tance of different sources of uncertainty in factors
This manuscript does not deal with choosing affecting supply and demand. This is likely to be
any one indicator over another, but rather em- quite site specific. In some instances, uncertainty
phasizes that the choice of indicator certainly does regarding future demand for environmental
matter and should be linked to conservation ob- amenities may swamp that connected with future
jectives. At the same time, the issue of indicator ecosystem processes. As mentioned above in con-
reliability will be important in actual decision- nection to forests, catastrophic disturbance (either
making applications. Reliability is largely a statis- man-made or natural) can reverse the successional
tical issue and depends on criteria such as process. In the forests we examined, however,
sensitivity, specificity, and predictability. A signifi- natural widespread disturbances rarely occur and
cant literature exists to help guide practitioners on many types of preservation efforts can effectively
this point (e.g. Murtaugh, 1996; Dufrene and insulate areas from major anthropogenic effects
Legendre, 1997; Legendre and Legendre, 1998). such as land development. Therefore, uncertainty
Our analysis is illustrative in nature in that it in ecosystem service supply may be small relative
relies upon an example relationship between con- to uncertainty in future demand for environmen-
servation effort and conserved biodiversity, rather tal amenities. In other types of systems, where
than an empirically estimated function between disturbance is more likely and vulnerability to
these two variables. The impact of conservation disturbance may be higher (some coastal ecosys-
effort on any given biodiversity indicator will vary tems may fall into this category), there may be
from site to site and potentially through time for substantial uncertainty in forecasting the supply
any particular site. Future ecological research to of ecosystem services for several years into the
examine the biodiversity ‘returns’ from increased future.
conservation activities, as well as the way that this
relationship varies by indicator, would be quite
useful. Acknowledgements
We intentionally have not incorporated uncer-
tainty in any fashion in this analysis, primarily to Financial support was provided by the Center
avoid detracting from the major points of interest. for Rural Pennsylvania (CRP), DuBois Educa-
Incorporation of this factor, however, represents tional Foundation Fund for Academic Excellence,
an important avenue for further research. For any Pennsylvania State University Research and De-
given natural area, uncertainty exists regarding velopment funds, the USDA Cooperative State
the future demand for and supply of various Research, Education, and Extension Service, and
ecosystem functions and services. Natural forces, a Challenge Grant from the Migratory Bird
as well as future anthropogenic change (e.g. Office, Region 5, US Fish and Wildlife Service
changes in patterns of adjacent human develop- (USFWS). D. DeCalesta, J. Palmer, and S. Stout
ment or changes in effects from pollutants trans- (US Forest Service), L. Lentz, J. Sowl, and D.
ported into the area), may change the supply of Wright (CRP), T. Mountain and D. Pence (US-
amenities that the ecosystem offers. However the FWS), C. Schlentner (Cook Forest State Park),
directions, magnitudes, and timing of such poten- and C. Schaadt provided logistic support, access
tial future changes, as well as the ways in which to study areas, or other assistance that greatly
the area will respond, are uncertain (King, 1997). facilitated this study. J. Lydic and R. Williams
In the same way, a host of factors (changes in performed many of the summary analyses. For
human population distributions, demographics, their help with the breeding bird censuses and
preferences for goods and services, and the rela- other field work, we thank B. Allison, J. Cheek,
tive prices of environmental amenities and other L. Hepfner, R. Kaufmann, J. Lydie, C. Schaadt,
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274 273
Faith, D.P., 1992. Conservation evaluation and phylogenetic
J. Seachrist, J. Smreker, S. Weilgosz, S. Wetzel,
diversity. Biological Conservation 61, 1 – 10.
and R. Williams. We thank S. Ragland and three
Foss, C.R., 1994. Atlas of Breeding Birds in New Hampshire.
anonymous reviewers for comments on previous Audubon Society of New Hampshire. Arcadia-Chalford
versions of this manuscript. Publishing, Dover, NH.
Frelich, L.E., Lorimer, C.G., 1991. Natural disturbance
regimes in hemlock-northern hardwood forests of the Up-
per Great Lakes Region. Ecological Monographs 61, 145 –
164.
References
Glowacinski, Z., Weiner, J., 1983. Successional trends in the
energetics of forest bird communities. Holarctic Ecology 6,
Alerich C.L., 1993. Forest statistics for Pennsylvania — 1978 305 – 314.
and 1989. USDA Forest Service Research Bulletin NE-126. Haney, J.C., 1994. Winter bird populations as bioindicators of
Radner, PA. old-growth forest in eastern North America. Journal fur ¨
Altschul, S.F., Lipman, D.J., 1990. Equal animals. Nature Ornithologie 135 (3), 515.
348, 493 – 494. Haney, J.C., 1995. Policy aspects of old-growth forest reserves
Andrle, R.F., Carroll, J.R., 1988. The Atlas of Breeding Birds in the eastern U.S. In: Majumdar, S.K., Miller, E.W.,
in New York State. Cornell University Press, Ithaca, NY. Brenner, F.J. (Eds.), Forests — A Global Perspective. The
Atkinson, I., 1989. Introduced animals and extinctions. In: Pennsylvania Academy of Science, Pittsburgh, PA.
Western, D., Pearl, M. (Eds.), Conservation for the 21st Haney, J.C., Eiswerth, M.E., 1992. The plight of cranes: a case
Century. Oxford University Press, New York, pp. 54 – study for conserving biodiversity. Proceedings of the North
69. American Crane Workshop 6, 12 – 18.
Bevier, L.R., 1994. The Atlas of Breeding Birds of Connecti- Haney, J.C., Schaadt, C.P., 1995. Functional roles of eastern
cut. State Geological and Natural History Survey of Con- old-growth in promoting forest bird diversity. In: Davis,
necticut, Hartford, CT. M.B. (Ed.), Eastern Old Growth. Island Press, Washing-
Bradford, D.F., Franson, S.E., Neale, A.C., Heggem, D.T., ton, DC.
Miller, G.R., Caterbury, G.E., 1998. Bird species assem- Helle, P., 1984. Effects of habitat area on breeding bird
blages as indicators of biological integrity in Great Basin communities in northeastern Finland. Annales Zoologici
rangeland. Environmental Monitoring and Assessment 49, Fennici 21, 421 – 425.
1–22. James, F.C., Rathbun, S., 1981. Rarefaction, relative abun-
Brauning, D.W., 1992. Atlas of Breeding Birds in Pennsylva- dance, and diversity of avian communities. The Auk 98,
nia. University of Pittsburgh Press, Pittsburgh, PA. 785 – 800.
Brunson, M., Shelby, B., 1992. Assessing recreational and King, D.M., 1997. Leading indicators of ecosystem services
scenic quality. Journal of Forestry 90, 37 –41. and values, with illustrations for performing habitat equiv-
Buckelew, A.R., Hall, G.A., 1994. The West Virginia Breeding alency analysis. Invited paper, Conference on Lost Human
Bird Atlas. University of Pittsburgh Press, Pittsburgh, PA. Uses of the Environment, Washington, DC, 7 & 8 May,
Bunge, J., Fitzpatrick, M., 1993. Estimating the number of 1997.
species: a review. Journal of the American Statistical Asso- Lanyon, W.E., 1981. Breeding birds and old field succession
ciation 88, 364 – 373. on fallow Long Island farmland. Bulletin of the American
Canham, C.D., Loucks, O.L., 1984. Catastrophic windthrow Museum of Natural History 168, 5 – 57.
in the presettlement forests of Wisconsin. Ecology 65, Laughlin, S.B., Kibbe, D.P., 1985. The Atlas of Breeding Birds
803 – 809. of Vermont. University Press of New England, Hanover,
Chiang, A.C., 1974). Fundamental Methods of Mathematical NH.
Economics. McGraw-Hill, New York. Legendre, P., Legendre, L., 1998. Numerical ecology. Devel-
Croonquist, M.J., Brooks, R.P., 1991. Use of avian and opments in Environmental Modelling, vol. 20. Elsevier
mammalian guilds as indicators of cumulative impacts in Science, Amsterdam.
riparian-wetland areas. Environmental Management 15, Lowe, J.D., 1995. Resident bird counts 1994. Journal of Field
701 – 714. Ornithology 66(Suppl.), 3 – 4.
Crozier, R.H., 1992. Genetic diversity and the agony of choice. May, P.G., 1982. Secondary succession and breeding bird
Biological Conservation 61, 11 –15. community structure: patterns of resource utilization. Oe-
Dessecker, D.R., Yahner, R.H., 1984. Black cherry-maple cologia 55, 208 – 216.
clearcut. American Birds 38, 71 –72. May, R.M., 1990. Taxonomy as destiny. Nature 347, 129 – 130.
Dufrene, M., Legendre, P., 1997. Species assemblages and McNeely, J.A., Miller, K.R., Reid, W.V., Mittermeir, R.A.,
indicator species: the need for a flexible asymmetrical Werner, T.B., 1990. Conserving the World’s Biological
approach. Ecological Monographs 67, 345 –366. Diversity. International Union for Conservation of Nature
Eiswerth, M.E., Haney, J.C., 1992. Allocating conservation and Natural Resources, World Resources Institute, Con-
expenditures across habitats: accounting for inter-species servation International, World Wildlife Fund — US, and
the World Bank, Gland, Switzerland and Washington, DC
genetic distinctiveness. Ecological Economics 5, 235 –249.
M.E. Eiswerth, J.C. Haney / Ecological Economics 38 (2001) 259–274
274
Miller, B., Reading, R., Strittholt, J., Carroll, C., Noss, R., Ray, G.C., 1988. Ecological diversity in coastal zones and
Soule, M., et al., 1998/1999. 1999. Using focal species in the
´ oceans. In: Wilson, E.O. (Ed.), Biodiversity. National
design of nature reserve networks. Wild Earth 12, 81 –92. Academy Press, Washington, DC.
Morrison, M.L., 1986. Bird populations as indicators of envi- Reid, W.V., Laird, S.A., Meyer, C.A., Gamez, R., Sittenfeld,
ronmental change. In: Johnston, R.F. (Ed.), Current Or- A., Janzen, D.H., et al., 1993a. Biodiversity Prospecting:
nithology. Plenum Press, New York, pp. 429 –451. Using Genetic Resources for Sustainable Development.
Murtaugh, P.A., 1996. The statistical evaluation of ecological World Resources Institute, Washington, DC.
indicators. Ecological Applications 6, 132 – 139. Reid, W.V., McNeely, J.A., Tunstall, D.B., Bryant, D.A.,
National Research Council, 1998. Forested Landscapes in
Winograd, M., 1993b. Biodiversity Indicators for Policy-
Perspective: Prospects and Opportunities for Sustainable
makers. World Resources Institute, Washington, DC.
Management of America’s Nonfederal Forests. National
Remsen, J.V., 1994. Use and misuse of bird lists in community
Academy Press, Washington, DC.
ecology and conservation. The Auk 111, 225 – 227.
National Research Council, 1992. Conserving Biodiversity: A
Robbins, C.S., Blom, E.A.T., 1996. Atlas of Breeding Birds of
Research Agenda for Development Agencies. National
Maryland and the District of Columbia. University of
Academy Press, Washington, DC.
Pittsburgh Press, Pittsburgh, PA.
Nicholson, C.P., 1997. Atlas of the Breeding Birds of Tennessee.
Shugart, H.H., 1984. A Theory of Forest Dynamics. Springer,
University of Tennessee Press, Knoxville, TN.
New York.
Niemi, G.J., Hanowski, J.M., Lima, A.R., Nicholls, T., Wei-
Simpson, R.D., Sedjo, R.A., Reid, J.W., 1994. Valuing biodiver-
land, N., 1997. A critical analysis on the use of indicator
sity: an application to genetic prospecting. Resources for the
species in management. Journal of Wildlife Management 61,
Future Discussion Paper 94-20, Resources for the Future,
1240– 1252.
Northern Forest Alliance, 2000. Investing in the Northern Washington, DC.
Forest: Emerging Conservation Opportunities and FY 2001 Smith, K.G., MacMahon, J.A., 1981. Bird communities along
Appropriation Priorities. Northern Forest Alliance, Mont- a montane sere: community structure and energetics. The
pelier, VT. Auk 98, 8– 28.
Office of Technology Assessment, 1988. Technologies to main- Tyrrell, L.E., Crow, T.R., 1994. Structural characteristics of
tain biological diversity. Office of Technology Assessment, old-growth hemlock-hardwood forests in relation to stand
US Congress, Washington, DC. age. Ecology 75, 370 – 386.
Palmer-Ball, B., 1996. The Kentucky Breeding Bird Atlas. Vane-Wright, R.I., Humphries, C.J., Williams, P.H., 1991.
University Press of Kentucky, Lexington, KY. What to protect? — Systematics and the agony of choice.
Pharo, E.J., Beattie, A.J., Binns, D., 1999. Vascular plant
Biological Conservation 55, 235 – 254.
diversity and a surrogate for bryophyte and lichen diversity.
Wiens, J.A., 1989. The ecology of bird communities. Founda-
Conservation Biology 13, 282 –292.
tions and Patterns, vol. 1. Cambridge University Press,
Pulliam, H.R., 1988. Sources, sinks, and population regulation.
Cambridge.
American Naturalist 132, 652 –661.
.