Optimal Allocation of Resources among Threatened Species: a Project Prioritization Protocol

Abstract

Conservation funds are grossly inadequate to address the plight of threatened species. Government and conservation organizations faced with the task of conserving threatened species desperately need simple strategies for allocating limited resources. The academic literature dedicated to systematic priority setting usually recommends ranking species on several criteria, including level of endangerment and metrics of species value such as evolutionary distinctiveness, ecological importance, and social significance. These approaches ignore 2 crucial factors: the cost of management and the likelihood that the management will succeed. These oversights will result in misallocation of scarce conservation resources and possibly unnecessary losses. We devised a project prioritization protocol (PPP) to optimize resource allocation among New Zealand's threatened-species projects, where costs, benefits (including species values), and the likelihood of management success were considered simultaneously. We compared the number of species managed and the expected benefits gained with 5 prioritization criteria: PPP with weightings based on species value; PPP with species weighted equally; management costs; species value; and threat status. We found that the rational use of cost and success information substantially increased the number of species managed, and prioritizing management projects according to species value or threat status in isolation was inefficient and resulted in fewer species managed. In addition, we found a clear trade-off between funding management of a greater number of the most cost-efficient and least risky projects and funding fewer projects to manage the species of higher value. Specifically, 11 of 32 species projects could be funded if projects were weighted by species value compared with 16 projects if projects were not weighted. This highlights the value of a transparent decision-making process, which enables a careful consideration of trade-offs. The use of PPP can substantially improve conservation outcomes for threatened species by increasing efficiency and ensuring transparency of management decisions.

Full text

Contributed Paper
Optimal Allocation of Resources among Threatened
Species: a Project Prioritization Protocol
LIANA N. JOSEPH,∗‡ RICHARD F. MALONEY,† AND HUGH P. POSSINGHAM∗
∗The Ecology Centre, School of Integrative Biology, University of Queensland, St Lucia 4072 Australia
†Threatened Species Development, Ecosystems and Species Unit, Research and Development Group, Department of Conservation,
Christchurch, New Zealand
Abstract: Conservation funds are grossly inadequate to address the plight of threatened species. Government
and conservation organizations faced with the task of conserving threatened species desperately need simple
strategies for allocating limited resources. The academic literature dedicated to systematic priority setting
usually recommends ranking species on several criteria, including level of endangerment and metrics of species
value such as evolutionary distinctiveness, ecological importance, and social significance. These approaches
ignore 2 crucial factors: the cost of management and the likelihood that the management will succeed.
These oversights will result in misallocation of scarce conservation resources and possibly unnecessary losses.
We devised a project prioritization protocol (PPP) to optimize resource allocation among New Zealand’s
threatened-species projects, where costs, benefits (including species values), and the likelihood of management
success were considered simultaneously. We compared the number of species managed and the expected
benefits gained with 5 prioritization criteria: PPP with weightings based on species value; PPP with species
weighted equally; management costs; species value; and threat status. We found that the rational use of cost and
success information substantially increased the number of species managed, and prioritizing management
projects according to species value or threat status in isolation was inefficient and resulted in fewer species
managed. In addition, we found a clear trade-off between funding management of a greater number of the
most cost-efficient and least risky projects and funding fewer projects to manage the species of higher value.
Specifically, 11 of 32 species projects could be funded if projects were weighted by species value compared with
16 projects if projects were not weighted. This highlights the value of a transparent decision-making process,
which enables a careful consideration of trade-offs. The use of PPP can substantially improve conservation
outcomes for threatened species by increasing efficiency and ensuring transparency of management decisions.
Keywords: conservation planning, conservation priorities, cost–benefit analysis, probability of success, species
management, species values, threat status
Asignaci´
on ´
Optima de Recursos entre Especies Amenazadas: un Protocolo de Priorizaci´
on de Proyectos
Resumen: Los fondos para la conservaci´
on son insuficientes para atender la dif´
ıcil situaci´
on de las especies
amenazadas. Las organizaciones gubernamentales y de conservaci´
on encargadas de la tarea de conservar
especies amenazadas desesperadamente requieren de estrategias simples para la asignaci´
on de recursos limi-
tados. La literatura acad´
emica dedicada a la definici´
on sistem´
atica de prioridades generalmente recomienda
clasificar a las especies considerando varios criterios, incluyendo el nivel de peligro y medidas del valor de las
especies como la unicidad evolutiva, la importancia ecol´
ogica y el significado social. Estos m´
etodos ignoran
dos factores cruciales: el costo del manejo y la probabilidad de que el manejo sea exitoso. Estas omisiones
resultar´
an en la asignaci´
on err´
onea de recursos de conservaci´
on escasos y, posiblemente, en p´
erdidas innece-
sarias. Dise˜
namos un protocolo de priorizaci´
on de proyectos (PPP) para optimizar la asignaci´
on de recursos
entre los proyectos sobre especies amenazadas de Nueva Zelanda, en el que se consideraron simult´
aneamente
los costos, los beneficios (incluyendo valor de las especies) y la probabilidad de ´
exito del manejo. Comparamos
‡email l.joseph@uq.edu.au
Paper submitted November 19, 2007; revised manuscript accepted August 4, 2008.
328
Conservation Biology, Volume 23, No. 2, 328–338
C
2008 Society for Conservation Biology
DOI: 10.1111/j.1523-1739.2008.01124.x

Joseph et al. 329
el n´
umero de especies manejadas y los beneficios esperados obtenidos con cinco criterios de priorizaci´
on: PPP
con ponderaciones basadas en el valor de las especies; PPP con ponderaciones equitativas de especies; costos de
manejo; valor de las especies y estatus de amenaza. Encontramos que el uso racional de la informaci´
on de
costo y ´
exito increment´
o sustancialmente el n´
umero de especies manejadas, y la priorizaci´
on de proyectos
de manejo de acuerdo con el valor de las especies o el estatus de amenaza fue insuficiente y result´
oen
menos especies manejadas. Adicionalmente, encontramos una clara desventaja entre el financiamiento de la
mayor´
ıa de los proyectos m´
as rentables y menos riesgosos y el financiamiento de menos proyectos para mane-
jar las especies de mayor valor. Espec´
ıficamente, 11 de 32 proyectos pudieran ser financiados si los proyectos
fueran ponderados por el valor de las especies en comparaci´
on de 16 proyectos si no hubiera ponderaci´
on.
Esto resalta el valor de un proceso transparente de toma de decisiones, que permite una consideraci´
on
cuidadosa de los pros y contras. El uso de PPP puede mejorar sustancialmente los resultados de la conser-
vaci´
on de especies amenazadas al incrementar la eficiencia y asegurar la transparencia de las decisiones de
manejo.
Palabras Clave: an´
alisis de costo-beneficio, estatus de amenaza, manejo de especies, planificaci´
on de la con-
servaci´
on, prioridades de conservaci´
on, probabilidad de ´
exito, valores de las especies
Introduction
The resources available for conserving the world’s bio-
diversity are grossly inadequate for the task. The exist-
ing expenditure is just a fraction of the predicted to-
tal cost of maintaining global biodiversity (James et al.
2001; Balmford et al. 2003). Currently, only a small pro-
portion of species recognized as threatened with ex-
tinction are managed for recovery (Baillie et al. 2004).
In the years 1989–1991, 54% of U.S. funding was de-
voted to conservation of just 1.8% of all U.S. threatened
species (Metrick & Weitzman 1996). Similarly, in 2006
only 22% of New Zealand’s threatened species were man-
aged, and many of these were inadequately managed to
ensure persistence (Joseph et al. 2008). Given this short-
fall in available resources, it is essential that the capital
available for management of threatened species be spent
wisely.
Formal methods for setting threatened-species priori-
ties are often conducted by identifying species that are
highly threatened or valuable and ranking them accord-
ing to one or both criteria. The management of species
near the top of the rank-ordered list is then funded (Miller
et al. 2006). A number of frameworks exist for ranking
species based on criteria such as the level of endanger-
ment (Master 1991; Carter et al. 2000), evolutionary dis-
tinctiveness (Faith 1994; Vane-Wright et al. 1994; N. J.
Isaac, S. T. Turvey, B. Collen, C. Waterman, and J. E.
Baillie. 2007. Mammals on the EDGE: conservation pri-
orities based on threat and phylogeny. Public Library of
Science ONE 2:e296.), a combination of these and so-
ciopolitical significance (Rodr´
ıguez et al. 2004), ecologi-
cal importance, and potential for recovery (Marsh et al.
2007). Understanding how to value species and the ur-
gency required for management is essential for making
good decisions. Nevertheless, any method that considers
only these criteria and ignores financial and technical con-
straints will result in fewer species being managed and
potentially unnecessary extinctions (IUCN 2001; Possing-
ham et al. 2002; Mace et al. 2006; Miller et al. 2006).
A species of low threat status or value may be cheap
to secure, whereas management of a highly threatened
or valuable species may be costly and have little po-
tential to reduce extinction risk. By failing to consider
the cost of management, the technical capacity to man-
age, and the potential for species recovery, standard ap-
proaches to priority setting make the implicit assump-
tion that the conservation budget is large enough to
fund all projects. This incorrect assumption results in
misallocation of scarce conservation resources and po-
tentially more extinctions. Simultaneously considering
species value and financial and technical constraints en-
sures that cost-efficient management is given priority and
will maximize conservation outcomes.
The Noah’s Ark framework (Metrick & Weitzman 1998;
Weitzman 1998) provides a cost-efficient solution to
the problem of threatened-species resource allocation.
The framework considers the benefits (i.e., increase in
the probability of a species persisting under a recovery
project), costs of the recovery project, species contri-
bution to diversity (i.e., distinctiveness), and value of
the species (i.e., utility of a species). On the basis of
mathematical theory (i.e., knapsack theory, Kellerer et al.
2004), with a slightly flexible total budget of a reasonable
size, the Noah’s Ark framework presents an optimal or
near-optimal solution (i.e., a greedy solution, Martello &
Toth 1990; Weitzman 1998; Hartmann & Steel 2006).
Although the Noah’s Ark framework provides an ap-
propriate methodology for considering costs and ben-
efits of conservation, it fails to consider a factor that is
crucial for making sound decisions to maximize conserva-
tion outcomes: the probability that the management will
succeed. A management action that is likely to succeed
should be given a higher priority than an action that is
Conservation Biology
Volume 23, No. 2, 2009

330 Cost-Efficient Conservation Priorities
likely to fail. Managers often have a good understanding
of their confidence in management projects, but these
judgments have never been included in a formal decision-
making process. Thus, we developed the Noah’s Ark
framework to include the likelihood that management
will succeed.
In addition, we extended the Noah’s Ark concept and
applied it to a genuine management problem. Only theo-
retical examples are described in the literature (Weitz-
man 1993). We developed the theoretical framework
into an operational Project Prioritization Protocol (PPP)
that requires setting conservation objectives (Nicholson
& Possingham 2006) and targets (see Sanderson 2006),
clarification of realistic working definitions of benefits,
costs and values, and careful consideration of the effects
of data limitations. We applied the PPP to the problem of
minimizing the number of extinctions of species over 50
years with a fixed budget. We used the framework in a
case study of the allocation of conservation resources for
managing the threatened species of New Zealand, includ-
ing a representative range of terrestrial and freshwater
taxonomic groups.
Our aim was to investigate efficiency of the PPP in
choosing the optimal set of management actions to max-
imize the overall benefit to a region’s threatened species
under financial constraints. We compared 5 priority-
setting methods: PPP with weightings based on species
value, PPP with species weighted equally, management
costs, species value, and threat status. We examined their
efficiency with 4 metrics of success: number of species
managed; expected benefit gained by managing the set
of species; summed value of the set of species managed;
and summed expected value gained by managing the set
of species. In summary we assessed the gains in efficiency
in spending afforded by properly incorporating informa-
tion about species values and the management costs, ben-
efits, and likelihood of management success.
Methods
Case Study: Management of the Threatened Species of New
Zealand
New Zealand is experiencing a massive biodiversity cri-
sis. Over the last 100 years, there have been 33 docu-
mented extinctions, including 16 birds, 9 terrestrial inver-
tebrates, and 6 vascular plants (King 1984; Hitchmough
et al. 2005). Of the more than 90,000 indigenous species,
about 2,000 are listed as threatened and another 3,000 are
data deficient under New Zealand’s threatened-species
classification system (Hitchmough et al. 2005). The De-
partment of Conservation has a total budget of approxi-
mately NZ$32 million/year (Department of Conservation
2006) specifically to improve the status of threatened
species.
Project Prioritization Protocol
The PPP consisted of 9 steps: (1) define objectives, (2)
list biodiversity assets (in this case, threatened species),
(3) weight assets, (4) list management projects, (5) cal-
culate the costs of each project, (6) predict the benefit
to species generated by each project, (7) estimate likeli-
hood of success, (8) state constraints, and (9) combine
information on costs, values, benefits and likelihood of
success to rank projects according to benefits per unit
dollar and choose set of projects.
STEP 1: DEFINE OBJECTIVES
To optimally allocate resources among management
projects, it is essential to clearly state the objectives (Poss-
ingham et al. 2001; Sanderson 2006). The general goal set
by New Zealand’s Department of Conservation Threat-
ened Species Program is to improve security of the great-
est possible number of unique species. The term security
refers to initial security from extinction in the wild. Secu-
rity is achieved when available evidence indicates there
is a viable population (i.e., one or more spatially discrete
populations) that is stable and will be able to recover in
the future once key agents of decline have been removed
or mitigated. The uniqueness of a species is measured as
its taxonomic distinctiveness (see step 3).
STEP 2: LIST BIODIVERSITY ASSETS
Biodiversity assets are the components of biodiversity
that one wishes to secure from extinction. In our case
study, biodiversity assets were species that have ≤95%
probability of being secure in 50 years if they are not man-
aged. We assumed that species that are not likely to be se-
cure in 50 years are those that are currently listed in one of
the following threat categories of New Zealand’s threat-
ened species classification system (Hitchmough et al.
2005): nationally critical (NC), nationally endangered
(NE), nationally vulnerable (NV), serious decline (SD),
gradual decline (GD), sparse (S), and range restricted
(RR).
To illustrate our protocol, we used 32 species listed
on New Zealand’s list of threatened species that repre-
sented a broad range of threat categories (15 NC, 10 NE,
1 NV, 2 SD, 2 GD, 1 S, and 1 RR), taxonomic groups
(4 frogs, 1 bat, 7 birds, 3 freshwater fish, 3 reptiles, 4 ter-
restrial invertebrates, and 10 vascular plants), measures
of taxonomic distinctiveness (number of close relatives,
see step 3), and potential management actions (e.g., cap-
tive breeding and translocation, fencing, predator con-
trol). The common and scientific names of the case-study
species are in Table 1.
STEP 3: WEIGHT ASSETS
Social, political, or biological values may be incorporated
into PPP by weighting species (see step 9). Species may
Conservation Biology
Volume 23, No. 2, 2009

Joseph et al. 331
Table 1. A list of the project parameters (benefit, B;cost,C; and probability of success, S) and species parameters (taxonomic distinctiveness, W; and threat status) that were used to calculate
weighted and unweighted project efficiency.
Weighted Unweighted
efficiency efficiency Taxonomic Probability
Common name Scientific name ∗1e12 ∗1e9 Benefit distinctiveness Cost ($) of success Threat status
Dactylanthus Dactylanthus taylorii 218502 927 0.70 0.236 755,103 1.00 4 serious decline
Maud Island frog Leiopelma pakeka 47849 550 0.70 0.087 1,273,311 1.00 2 nationally endangered
Shrubby tororaro Muehlenbeckia astonii 43206 473 0.40 0.091 760,615 0.90 3 nationally vulnerable
Hamilton’s frog Leiopelma hamiltoni 39012 448 0.60 0.087 1,338,631 1.00 1 nationally critical
North Island brown kiwi Apteryx mantelli 26854 120 0.95 0.224 7,910,292 1.00 4 serious decline
Climbing everlasting daisy Helichrysum dimorphum 22112 942 0.35 0.023 371,438 1.00 2 nationally endangered
Hochstetter’s frog Leiopelma hochstetteri 22099 254 0.10 0.087 354,472 0.90 6 sparse
New Zealand shore plover Thinornis novaeseelandiae 15296 355 0.40 0.043 855,261 0.76 1 nationally critical
Pittosporum patulum 12902 294 0.95 0.044 3,229,002 1.00 2 nationally endangered
Oreomyrrhis sp. 12746 869 0.70 0.015 724,823 0.90 2 nationally endangered
nov [ =O.aff rigida]
Pachycladon exilis 12573 515 0.95 0.024 884,866 0.48 1 nationally critical
Archey’s frog Leiopelma archeyi 11168 128 0.70 0.087 4,909,753 0.90 1 nationally critical
Canterbury mudfish Neochanna burrowsius 8425 173 0.95 0.049 891,340 0.16 2 nationally endangered
Carmichaelia hollowayi 3721 637 0.70 0.006 751,950 0.68 1 nationally critical
Poa spania 3370 743 0.70 0.005 847,626 0.90 1 nationally critical
Chatham Island Haematopus chathamensis 2768 223 0.95 0.012 1,798,486 0.42 1 nationally critical
oystercatcher
Kaki Himantopus novaezelandiae 1962 102 0.95 0.019 8,851,848 0.95 1 nationally critical
Cardamine cf. bilobate 1647 216 0.95 0.008 874,373 0.20 1 nationally critical
Black robin Petroica traverse 1311 107 0.40 0.012 1,534,592 0.41 1 nationally critical
Pygmy button daisy Leptinella nana 704 86 0.60 0.008 297,571 0.04 2 nationally endangered
Cook Strait giant weta Deinacrida rugosa 592 509 0.40 0.001 785,586 1.00 7 range restricted
Big-nose galaxias Galaxias macronansus 525 28 0.95 0.019 2,047,005 0.06 5 gradual decline
Long-tailed bat Chalinolobus tuberculate 332 31 0.95 0.011 6,210,151 0.21 2 nationally endangered
Carabid beetle Zecillenus tillyardi 325 117 0.70 0.003 1,242,714 0.21 1 nationally critical
Lowland long jaw galaxid Galaxias cobitinis 283 15 0.95 0.019 1,757,261 0.03 1 nationally critical
Orange-fronted parakeet Cyanoramphus malherbi 266 6 0.95 0.041 14,453,678 0.10 1 nationally critical
Mohua Mohoua ochrocephala 188 26 0.40 0.007 5,762,489 0.38 2 nationally endangered
Grand skink Oligosoma grande 159 58 0.95 0.003 7,754,153 0.48 1 nationally critical
Otago Skink Oligosoma otagense 159 58 0.95 0.003 7,754,153 0.48 1 nationally critical
Short horned grasshopper Sigaus minutus 49 33 0.90 0.001 3,795,354 0.14 5 gradual decline
Chevron skink Oligosoma homalonotum 35 13 0.70 0.003 7,888,632 0.14 2 nationally endangered
Robust grasshopper Brachapsis robustus 19 8 0.90 0.002 5,636,964 0.05 2 nationally endangered
Minimum 19 6 0.10 0.0012 297,571 0.03 7 range restricted
Maximum 218502 942 0.95 0.2357 14,453,678 1.00 1 nationally critical
Mean or median 2365∗150∗0.73 0.0167∗3,259,484 0.56 2 nationally endangered∗
(denoted by ∗)
Coefficient of variation 0.33 1.06 0.67
Fold difference 11661 146 10 203 49 36 7
Conservation Biology
Volume 23, No. 2, 2009

332 Cost-Efficient Conservation Priorities
be weighted on the basis of factors such as cultural signif-
icance, economic importance, evolutionary significance,
ecological function, and endemicity, or species may be
treated equally (i.e., no weights). If more than one type
of value is used to weight species, the values can be com-
bined by summing their weighted scores. The Noah’s
Ark framework combines 2 species values: distinctive-
ness and species utility. We considered only distinctive-
ness and assumed the utility of species was the same for
all species (i.e., equal to zero).
The weighting factor, Wi, quantified the taxonomic
distinctiveness of species i, where distinctiveness was
inversely related to the number of relatives of species
i. For example, a species that has many close relatives
(e.g., robust grasshopper) was given a small weight,
whereas a species that has few close relatives (e.g.,
North Island Brown Kiwi) received a large weight. We
expressed taxonomic distinctiveness as the inverse of
the product of the number of branches at the genus,
family, and order nodes (i.e., a modified version of the
method of Daniels et al. [1991]). We believe taxonomic
distinctiveness is well characterized for our purposes by
this method because it delivers a metric that is contin-
uous and encapsulates diversity at the species, genus,
and family levels. This method is rapid and inexpensive
and can be applied to a simple, widely available, and
partially resolved taxonomic system, such as Linnaean
taxonomy.
By weighting species by their taxonomic distinctive-
ness, we sacrificed species richness for taxonomic di-
versity (Solow et al. 1993). We maintained the order of
species while reducing the magnitude of the weighting
so that less emphasis was placed on taxonomic diver-
sity and more on species richness by taking the xth root
of measure of distinctiveness (Solow et al. 1993). The
choice of xdepended on the relative value of taxonomic
diversity and species richness; we used x=2because
it provided a balance between conserving the greatest
number of unique species and simply the greatest quan-
tity of species. Therefore, we measured the taxonomic
distinctiveness for species, Wi,asthexth root of the in-
verse of the product of the number of branches, b,atthe
genus, family, and order nodes, n:
Wi=x




1

n
b,(1)
where x= 2. We used the on-line version of the An-
nual Checklist (http://www.catalogueoflife.org/annual-
checklist/2007/) of the Catalogue of Life (Bisby et al.
2007) to estimate the number of branches at each node.
The Catalogue of Life is incomplete and some of the
species or genera are not listed. When a species or genus
of interest was absent from the list, we added this taxon
to the total number of branches.
STEP 4: LIST MANAGEMENT PROJECTS
We asked experts to choose an appropriate project for
each species. A project was the minimum set of all neces-
sary actions for obtaining a reasonable (≥95%) probability
of securing the species over 50 years. Projects had 4 com-
pulsory components (outcome monitoring, services and
support, project management, and infrastructure) and at
least one optional intervention (e.g., captive breeding,
translocation, pest animal control, weed control, legal
actions, and education). Experts clearly described a pre-
cise location, intensity, and duration of management for
each action. Because there was a one-to-one match of
projectsandspecies,weusedtheindexifor both.
STEP 5: ESTIMATE COST
The cost of each project, Ci, was estimated each year
and converted to current dollar values. Costs included
all future outlays, whereas past outlays (e.g., the cost
of building captive-breeding facilities that are now avail-
able for use) were not considered. We calculated the net
present value of the cost to fund each project over 50
years as
Ci=
50

t
ci,t
(1 +r)t,(2)
where cis the cost of project iin year tand ris a discount
rate derived from the present value of future spending
(i.e., discount factor). For example, if it cost $10,000/year
to control stoats now, $395,881 in today’s dollars (r=
0.01) would be needed for stoat control over 50 years.
Effectively, our choice of the discounting rate (e.g., r
>0) resulted in the cost of actions in the future being
cheaper in today’s dollars. In the case in which an action
benefited more than one species, the cost of the action
among the species projects could be shared; thus, actions
that benefited multiple species were favored.
STEP 6: ESTIMATE BENEFITS
The impact of project ion the species probability of being
secure was recorded as the biodiversity benefit, Bi.Se-
curity probabilities are related to extinction probabilities
and may be calculated from stochastic population mod-
els (Burgman et al. 2001; Drechsler & Burgman 2004) or
estimated directly with expert knowledge (Cullen et al.
2001, 2005). Because knowledge available for many of
the species is not sufficient to estimate parameters for
population viability models, we used experts to estimate
security probabilities directly. The biodiversity benefit of
project i,Bi, is the difference between the probability of
the species being secure in 50 years with (Pi) and without
(P0) management:
Bi=Pi−P0.(3)
Conservation Biology
Volume 23, No. 2, 2009

Joseph et al. 333
STEP 7: ESTIMATE LIKELIHOOD OF SUCCESS
We asked experts to state the probability that each
project, i, could be implemented successfully, Mi,and
the probability that, if implemented successfully, it would
be reasonably (≥95%) successful in securing the species,
Ni. Only actions that would have a direct impact on the
probability of security (e.g., translocation) were assessed,
whereas essential actions that would not have direct ef-
fects (i.e., compulsory actions such as infrastructure and
service support) were assumed to have 100% probabil-
ity of succeeding. Total probability of success of each
project, Si, was Si=MiNi.
To aid in estimation of the probabilities of implemen-
tation success, we asked experts to provide probabil-
ities on a series of possible limitations to the success
of the projects. Factors that may contribute to failure
of the project included operational (e.g., isolation, dif-
ficulty moving around terrain, capacity, unpredictable,
prohibitive weather), legal (e.g., resource-management
acts, building codes, health and safety, animal ethics),
and political and social (e.g., public attitude, conflict-
ing land use, iwi views, access permission) constraints.
Experts estimated the probability of successfully imple-
menting each action of project i, and we combined the
probabilities to calculate the total probability of success-
ful implementation of the full project, Mi:
Mi=
J

j=1
Kij,(4)
where Kij is the probability that action jof project i
works.
Experts directly estimated the probability of technical
success, Nj, of each action in the project through demon-
strated evidence and confidence in project choice. Ex-
perts based their estimates on knowledge of the propor-
tion of successes and failures of past implementation and
said whether their opinions were based on information
relevant to the same or other species, habitats, threats,
and scales. We combined the probability of technical suc-
cess for each action to calculate the total probability of
technical success for the ith project, Nias follows:
Ni=
J

j=1
Lij,(5)
where Lij is the probability that action jof project isuc-
ceeds.
STEP 8: STATE CONSTRAINTS
The primary constraint on the resource-allocation prob-
lem was the total budget available for management of
threatened species. We focused on 32 species, and the
total budget was proportional to the budget available for
all 2000 threatened species. The total budget available
for the management of 2000 threatened species in the
2005–2006 financial year (i.e., $32,000,000) was equiva-
lent to an annual budget of $512,000. Allocation of this
amount among 32 species projects resulted in a budget
of $20,269,096 for the 50-year period (with r=0.01
discounting).
STEP 9: CHOOSE SET OF PROJECTS
The Noah’s Ark framework ranks species or species
projects on the basis of ranking criterion, R,whichis
a cost-efficiency metric:
Ri=Wi×pi
Ci
,(6)
where pis analogous to our biodiversity benefits, B,
and is defined as the change in survivability of a species
iand Wis the sum of distinctiveness and species utility.
We modified the cost-efficiency measure to include the
likelihood of success of a project and called the modified
metric the project efficiency, E,ofprojecti.TheEiwas
calculated as:
Ei=Wi×Bi×Si
Ci
,(7)
where Wiis the species weights, Biis the biodiversity
benefits, Siis probability of success, and Ciis the cost of
project i.
COMPARISON OF PRIORITY SETTING METHODS
We used 4 metrics to measure efficiency: number of
species managed; expected benefit gained by managing
the set of species (i.e., sum of the product of the bene-
fit and the probability of success); uniqueness of the set
of species managed (i.e., summed uniqueness values);
and expected uniqueness gained by managing the set
of species (i.e., the sum of the product of the benefit,
the probability of success, and the taxonomic distinctive-
ness). We compared the efficiency of prioritizing man-
agement projects by their project efficiency with species
weighted by taxonomic distinctiveness with 4 other crite-
ria: project efficiency when species are weighted equally,
cost, taxonomic distinctiveness, and threat status.
Project efficiency with species weighted by taxonomic
distinctiveness was calculated with the methodology de-
scribed earlier (hereafter referred to as weighted project
efficiency). We ranked species projects in order of de-
creasing weighted project efficiency values: small val-
ues were ranked lower than high values. The project
efficiency when species were weighted equally was cal-
culated in the same manner; however, species weights
were considered equal to each other and equal to one
(hereafter referred to as unweighted project efficiency).
Conservation Biology
Volume 23, No. 2, 2009

334 Cost-Efficient Conservation Priorities
We ranked species projects in decreasing order of the
unweighted project efficiency value. The procedure for
calculating project costs is outlined earlier. We ranked
projects in increasing order of price. Taxonomic distinc-
tiveness was calculated with the methods described ear-
lier. Projects were ranked on the basis of the taxonomic
distinctiveness of each species: species with few close rel-
atives were ranked higher than those with many close rel-
atives. To rank projects by species threat status, we used
the threat categories and species status of New Zealand’s
Classification Species List (Hitchmough et al. 2005). In
this system species are listed in 1 of 8 categories of threat.
We assumed an order of decreasing extinction risk from
NC, NE, NV, SD, GD, S, and RR to not threatened. We
ranked projects in order of decreasing extinction risk of
the species. To distinguish among species within threat
categories, we ranked projects by their cost: inexpensive
projects were given a higher ranking. We selected cost
of management because it is an intuitive and important
consideration and information required to rank manage-
ment projects by cost is readily available (compared with
probability of success or benefits).
Table 2. Species projects ranked with the weighted project efficiency (columns 1 & 2) and priority ranks obtained with the 4 other priority-setting
methods: unweighted project efficiency, cost, taxonomic distinctiveness, and threat status.
Weighted Unweighted Cost Distinctiveness Threat
Species efficiency rank efficiency rank rank rank status rank
Dactylanthus 1∗2∗6∗127
Maud Island frog 2∗6∗15∗420
Shrubby tororaro 3∗9∗7∗326
Hamilton’s frog 4∗10∗16∗57
∗
North Island Brown Kiwi 5∗18 30 2 28
Climbing everlasting daisy 6∗1∗3∗13 17
Hochstetter’s frog 7∗13∗2∗631
New Zealand Shore Plover 8∗11∗10∗10 3∗
Pittosporum patulum 9∗12∗21 9 21
Oreomyrrhis sp.nov(=O. aff rigida)10
∗3∗4∗17 18
Pachycladon exilis 11∗7∗12∗12 5∗
Archey’s frog 12 17 23 7∗11∗
Canterbury mudfish 13 16∗13∗8∗19
Carmichaelia hollowayi 14 5∗5∗24 1∗
Poa spania 15 4∗9∗25 2∗
Chatham Island Oystercatcher 16 14∗19∗18 10∗
Kaki 17 21 31 14 14
Cardamine cf. bilobata 18 15∗11∗22 4∗
Black Robin 19 20 17∗19 8∗
Pygmy button daisy 20 22 1∗21 16
Cook Strait giant weta 21 8∗8∗32 32
Big-nose galaxias 22 27 20∗15 29
Long-tailed bat 23 26 26 20 24
Carabid beetle 24 19 14∗26 6∗
Lowland long jaw galaxid 25 29 18∗16 9∗
Orange-fronted Parakeet 26 32 32 11 15
Mohua 27 28 25 23 23
Grand skink 28 23 27 27 12
Otago skink 29 24 28 28 13
Short-horned grasshopper 30 25 22 31 30
Chevron skink 31 30 29 29 25
Robust grasshopper 32 31 24 30 22
∗Species selected with a fixed budget of $20,269,096.
Results
The management project parameters (costs, benefit, and
probability of success) and species parameters (taxo-
nomic distinctiveness and threat status) spanned a wide
range and generated informative project efficiency values
for species management projects (Table 1).
Order of Species Projects
The order of species projects varied considerably for each
of the 5 priority-setting methods (Table 2). For the pro-
jected available budget of $20,269,096, a different set of
projects was selected by each of the priority-setting meth-
ods. The first 4 methods (both project efficiency, cost and
distinctiveness methods) selected comparatively similar
sets of projects, whereas the threat-status ranking crite-
rion’s set of projects was noticeably different. For threat-
status ranking, 3 projects were selected in common with
both of the project efficiency and cost methods, whereas
these 3 methods shared 9 of a possible 11 projects in
common.
Conservation Biology
Volume 23, No. 2, 2009

Joseph et al. 335
Figure 1. Returns on spending budgets up to $50 million over 50 years for each of the 5 priority-setting methods:
project efficiency with species weighted by taxonomic distinctiveness (weighted E); project efficiency when species
are weighted equally (unweighted E); cost; taxonomic distinctiveness (distinctiveness); and threat status (threat).
Returns are measured as (a) number of species managed, (b) expected benefit gained by managing the set of
species (i.e., summed product of benefit and probability of success), (c) uniqueness of the set of species managed
(i.e., summed taxonomic distinctiveness), and (d) expected uniqueness gained by managing the set of species (i.e.,
product of the benefit, probability of success, and taxonomic distinctiveness).
Number of Species Funded
The number of species projects that may be funded with
a range of budgets (Fig. 1a) and the projected budget
(Table 3) differed for each of the 5 priority-setting meth-
ods. The rank order produced by prioritizing projects
Table 3. The returns on the spending of $20,269,096 over 50 years for each of the 5 priority-setting methods.∗
Efficiency Weighted Unweighted Taxonomic Threat
metrics efficiency efficiency Cost distinctiveness status
Number of species managed 11 16 20 8 11
Expected benefit gained (B∗S) 6.09 7.39 6.86 4.18 4.02
Uniqueness of species managed (W) 0.96 0.82 0.83 0.95 0.31
Expected uniqueness gained (W∗B∗S) 0.61 0.42 0.38 0.59 0.15
∗Returns are measured as number of species managed; expected benefit gained by managing the set of species (i.e., the summed product of
benefit, Band probability of success, S); uniqueness of the set of species managed (i.e., summed taxonomic distinctiveness, W); and expected
uniqueness gained by managing the set of species (i.e., the summed product of the benefit, B; probability of success, S; and taxonomic
distinctiveness, W).
on the basis of cost always resulted in the most projects
being funded for any given budget. Ranking projects
with the unweighted project efficiency criterion was
the second-most efficient method. Ranking species by
taxonomic distinctiveness always resulted in the fewest
species managed.
Conservation Biology
Volume 23, No. 2, 2009

336 Cost-Efficient Conservation Priorities
Expected Benefit Gained
The expected benefit gained by funding projects with
the projected budget (Table 3) and across the range of
budgets (Fig. 1b) differed among the 5 methods. The rank
order generated by prioritizing projects on the basis of
the unweighted project efficiency metric was always the
most efficient because it generated the highest expected
benefit for a given budget. Prioritizing projects on the ba-
sis of threat status or taxonomic distinctiveness produced
the lowest expected benefit.
Uniqueness of Species Managed
Interestingly, ranking projects by taxonomic distinctive-
ness did not produce a rank order of species with the
highest uniqueness for the majority of budgets (Fig.
1c), including the projected available budget (Table 3).
For some budgets, 3 of the priority selection methods
(cost, weighted and unweighted project efficiency crite-
ria) generated a rank order that selected a set of species
with a higher summed uniqueness value than ranking
by taxonomic distinctiveness. Generally, ranking species
projects by the weighted project efficiency produced a
rank order with the highest summed uniqueness. Con-
versely, ranking species by threat status consistently se-
lected a set of species with the lowest summed unique-
ness.
Expected Uniqueness Gained
Across the range of budgets, the expected gains in
uniqueness was maximized if projects were ranked by the
weighted project efficiency (Table 3 & Fig. 1d). Ranking
projects on the basis of taxonomic distinctiveness gen-
erated the second-highest expected gains in uniqueness.
For larger budgets, the unweighted project efficiency cri-
terion was equally as efficient as the weighted index;
however, for smaller budgets, this method generated
lower expected gains in uniqueness. Ranking projects
by the species threat status performed poorly across all
budgets because this ranking method produced very low
expected gains in uniqueness.
Discussion
The PPP provided a means for prioritizing management to
maximize conservation outcomes for threatened species
when budgets were limited. Efficiency in spending was
improved by considering project costs, benefits, and like-
lihood of success. The unweighted project efficiency cri-
terion resulted in the maximum benefits gained for all
budgets. The expected gains in value (i.e., uniqueness)
from projects were maximized by simultaneously con-
sidering values, benefits, costs, and the probability of
management success.
Inadequate Priority-Setting Methods: Threat Status and
Values
Methods for prioritizing projects that focus exclusively
on threat status resulted in management of fewer species
and procurement of less benefit to threatened species in
general. The projected available budget for threatened
species management was enough to fund only 11 species
projects when ranked by threat status, compared with
the 16 projects that would be funded if species were
ranked with the unweighted project efficiency ranking
criterion. In addition, ranking species by threat status ig-
nored species distinctiveness and hence selected projects
for a set of species that were not highly valued.
Likewise, to prioritize projects efficiently, it is not suf-
ficient to consider a species value, such as taxonomic
distinctiveness, in isolation. Ranking species by their
taxonomic distinctiveness resulted in appreciably fewer
projects (8 projects) being funded with the available
budget when compared with either of the project ef-
ficiency ranking methods (weighted project efficiency
=11 projects and unweighted project efficiency =16
projects). In addition, the expected benefit gained was
substantially lower than the unweighted project effi-
ciency method (taxonomic distinctiveness =4.18, un-
weighted project efficiency =7.39). Significantly, for the
majority of budgets, ranking projects by taxonomic dis-
tinctiveness did not maximize the summed taxonomic
distinctiveness for the set of priority projects; in many
cases, ranking projects by their cost or unweighted
and weighted project efficiency scores performed bet-
ter. Nevertheless, the expected uniqueness gained when
ranking species on taxonomic distinctiveness was only
slightly lower than the optimal choice of ranking with
weighted project efficiency for all budgets. Prioritizing
projects by species distinctiveness produced gains in
uniqueness by giving high priority to a few species with
high taxonomic distinctiveness rather than, as with the
weighted project efficiency method, a greater number of
species that are less taxonomically distinct.
Values and Objectives
A trade-off exists between funding the management of
a greater number of the most cost-efficient and least
risky projects and funding fewer projects to manage
the species of higher value (i.e., uniqueness). The un-
weighted project-efficiency method resulted in the man-
agement of 5 additional species because it ignored
species distinctiveness (i.e., unweighted project effi-
ciency =16 species compared with weighted project
efficiency =11 species). Five of 32 species is 15.6% of
the case-study species; this is a considerable proportion,
equivalent to 312 of the 2000 species listed as threatened
in New Zealand.
Specifically, the unweighted project-efficiency method
ranked the species such that 6 additional species,
Conservation Biology
Volume 23, No. 2, 2009

Joseph et al. 337
including 3 plants (Carmichaelia hollowayi,Cardamine
cf. bilobata,andPoa spania), a bird (Chatham Island
Oystercatcher), a fish (Cantebury mudfish), and an inver-
tebrate (Cook Strait giant weta) could be funded at the
expense of the highly taxonomically distinct North Island
brown kiwi. The additional 6 species were not individ-
ually taxonomically distinct (median =0.007), but they
were all relatively cheap to manage (mean =$991,560,
CV =0.40) and had high benefits (mean =0.78, CV =
0.28) or a high potential for management success (mean
=0.56, CV =0.63).
To maximize the expected value (in this case, taxo-
nomic distinctiveness) of a set of species, it is essential to
weight management benefits by the species value. Nev-
ertheless, to what degree should efficiency (e.g., number
of species managed or total expected benefit) be sacri-
ficed for the preferential allocation of resources to highly
valued species? The North Island Brown Kiwi has high
taxonomic distinctiveness (W=0.224); hence, it is highly
valued by the weighted project efficiency criterion. Nev-
ertheless, is funding this species worth sacrificing 6 other
species, even if each is individually relatively taxonom-
ically indistinct? Managers need to think clearly about
their intention for weighting species by their uniqueness
or any other value and the ramifications of the form of
metric selected. The PPP provides a transparent approach
to project selection that promotes scrutiny of decisions
such as this.
Department of Conservation policy
New Zealand’s Department of Conservation is using the
PPP to enable efficient allocation of resources among
management projects for its work on threatened species
and ecosystems. Initial prioritization is aimed at maximiz-
ing the number of species secured from extinction. Ad-
ditional objectives include restoring threatened species
to self-sustaining populations and maximizing ecolog-
ical integrity. They have used PPP to assess manage-
ment projects and generate a rank-ordered list of species
projects to guide senior managers in resource allocation
decisions for more than 2000 species. The process has
taken <1.5 years and 105 experts have been consulted.
We do not believe this process is beyond the capabil-
ities of any national government department, including
mega-diverse countries, where data are limited. If there is
enough data to list the species on threatened species lists,
then there should be enough data to rank management
projects with project prioritization protocol.
Conclusion
Our protocol for prioritizing management actions inte-
grates biological data, financial and technical constraints,
and social and/or biological values. The framework pro-
vides a systematic, transparent, and repeatable method
for prioritizing actions to minimize the number of extinc-
tions. Clearly stating the steps used to make decisions
presents an opportunity to scrutinize and improve the
decision-making process; initiates a forum for the explicit
examination of management principles and limitations,
including the development of unambiguous working ob-
jectives; and reveals knowledge gaps and uncertainty in
the system. We demonstrated that to select management
actions that maximize conservation outcomes, it is insuf-
ficient to prioritize species based solely on threat status or
species value. Correspondingly, return on investment of
conservation dollars is substantially improved by incorpo-
rating management costs, benefits, and likelihood of man-
agement success. Consequently, the number of species
managed and the expected overall benefit to threatened
species is increased remarkably.
Acknowledgments
We thank S. O’Connor, P. Cromarty, A. Holzapfel, L. Hart-
ley,J.Terry,B.Kappers,J.Davis,N.Singers,L.Adams,
E. Edwards, G. Taylor, J. Reardon, T. Stephens, D. God-
den and the more than 100 threatened species experts
for their contributions and support. This work was sup-
ported by the New Zealand Department of Conserva-
tion and the Applied Environmental Decision Analysis
research hub that is funded by the Commonwealth Envi-
ronment Research Facilities program.
Literature Cited
Baillie, J. E. M., C. Hilton-Taylor, and S. N. Stuart. 2004. 2004 IUCN Red
List of threatened species: a global species assessment. Pages xxiv–
191. IUCN, Gland, Switzerland, and Cambridge, United Kingdom.
Balmford, A., K. J. Gaston, S. Blyth, A. James, and V. Kapos. 2003. Global
variation in terrestrial conservation costs, conservation benefits, and
unmet conservation needs. Proceedings of the National Academy of
Sciences of the United States of America 100:1046–1050.
Bisby, F., Y. Roskov, M. Ruggiero, T. Orrell, L. Paglinawan, P. Brewer,
N. Bailly, and J. van Hertum. 2007. Species 2000 & ITIS catalogue of
life: 2007 annual checklist. CD-ROM. Species 2000, Reading, United
Kingdom.
Burgman, M. A., H. P. Possingham, A. J. Lynch, D. A. Keith, M. A.
McCarthy, S. D. Hopper, W. L. Drury, J. A. Passioura, and R. J.
Devries. 2001. A method for setting the size of plant conservation
target areas. Conservation Biology 15:603–616.
Carter, M. F., W. C. Hunter, D. N. Pashley, and K. V. Rosenberg. 2000.
Setting conservation priorities for landbirds in the United States: the
partners in flight approach. Auk 117:541–548.
Cullen, R., G. A. Fairburn, and K. F. D. Hughey. 2001. Measuring the
productivity of threatened-species programs. Ecological Economics
39:53–66.
Cullen, R., E. Moran, and K. F. D. Hughey. 2005. Measuring the success
and cost effectiveness of New Zealand multiple-species projects
to the conservation of threatened species. Ecological Economics
53:311–323.
Conservation Biology
Volume 23, No. 2, 2009

338 Cost-Efficient Conservation Priorities
Daniels, R. J. R., M. Hegde, N. V. Joshi, and M. Gadgil. 1991. Assigning
conservation value: a case-study from India. Conservation Biology
5:464–475.
Department of Conservation. 2006. Annual report for the year ended
30 June 2006. Presented to the House of Representatives pursuant
to section 44(1) of the Public Finance Act 1989. Department of
Conservation, Wellington, New Zealand.
Drechsler, M., and M. A. Burgman. 2004. Combining population via-
bility analysis with decision analysis. Biodiversity and Conservation
13:115–139.
Faith, D. P. 1994. Phylogenetic diversity: a general framework for the
prediction of feature diversity. Pages 251–268 in P. L. Forey, C. J.
Humphries, and R. I. Vane-Wright, editors. Systematics and conser-
vation evaluation. Clarendon Press, Oxford, United Kingdom.
Hartmann, K., and M. Steel. 2006. Maximizing phylogenetic diversity
in biodiversity conservation: greedy solutions to the Noah’s Ark
problem. Systematic Biology 55:644–651.
Hitchmough, R., L. Bull, and P. Cromarty. 2005. New Zealand clas-
sification system lists. Pages 1–194. Department of Conservation,
Wellington, New Zealand.
IUCN (World Conservation Union). 2001. IUCN Red List categories
and criteria. Version 3.1. IUCN Species Survival Commission, IUCN,
Gland, Switzerland, and Cambridge, United Kingdom.
James, A., K. J. Gaston, and A. Balmford. 2001. Can we afford to conserve
biodiversity? BioScience 51:43–52.
Joseph, L. N., R. F. Maloney, S. M. O’Connor, P. Cromarty, P. Jansen,
T. Stephens, and H. P. Possingham. 2008. Improving methods
for allocating resources among threatened species: the case for
a new national approach in New Zealand. Pacific Conservation
Biology 3: in press.
Kellerer, H., U. Pferschy, and D. Pisinger. 2004. Knapsack problems.
Springer-Verlag, Berlin.
King, C. 1984. Immigrant killers: introduced predators and the con-
servation of birds in New Zealand. Oxford University Press,
Auckland.
Mace, G. M., H. P. Possingham, and N. Leader-Williams. 2006. Prioritiz-
ing choices in conservation. Pages 17–34 in D. W. MacDonald and
K. Service, editors. Key topics in conservation biology. Blackwell
Publishing, Malden, Massachusetts.
Marsh, H., A. Dennis, H. Hines, A. Kutt, K. Mcdonald, E. Weber, S.
Williams, and J. Winter. 2007. Optimizing allocation of management
resources for wildlife. Conservation Biology 21:387–399.
Martello, S., and P. Toth. 1990. Knapsack problems: algorithms and
computer implementations. Wiley, Chichester, New York.
Master, L. L. 1991. Assessing threats and setting priorities for conserva-
tion. Conservation Biology 5:559–563.
Metrick, A., and M. L. Weitzman. 1996. Patterns of behavior in endan-
gered species preservation. Land Economics 7:1–16.
Metrick, A., and M. L. Weitzman. 1998. Conflicts and choices in biodiver-
sity preservation. The Journal of Economic Perspectives 12:21–33.
Miller, R. M., et al. 2006. Extinction risk and conservation priorities.
Science 313:441.
Nicholson, E., and H. P. Possingham. 2006. Objectives for multiple-
species conservation planning. Conservation Biology 20:871–881.
Possingham,H.P.,S.J.Andelman,B.R.Noon,S.Trombulak,andH.
R. Pulliam. 2001. Making smart conservation decisions. Pages 225–
244 in M. E. Soul´
e and G. H. Orians, editors. Conservation biology:
research priorities for the next decade. Island Press, Washington,
D.C.
Possingham,H.P.,S.J.Andelman,M.A.Burgman,R.A.Medellin,L.
L. Master, and D. A. Keith. 2002. Limits to the use of threatened
species lists. Trends in Ecology & Evolution 17:503–507.
Rodr´
ıguez, J. P., F. Rojas-Su´
arez, and C. J. Sharpe. 2004. Setting priorities
for the conservation of Venezuela’s threatened birds. Oryx 38:373–
382.
Sanderson, E. W. 2006. How many animals do we want to save? The
many ways of setting population target levels for conservation. Bio-
Science 56:911–922.
Solow, A., S. Polasky, and J. Broadus. 1993. On the measurement of
biological diversity. Journal of Environmental Economics and Man-
agement 24:60–68.
Vane-Wright, R. I., C. R. Smith, and I. J. Kitching. 1994. Systematic as-
sessment of taxic diversity by summation. Pages 309–326 in P. L.
Forey, C. J. Humphries, and R. I. Vane-Wright, editors. Systemat-
ics and conservation evaluation. Clarendon Press, Oxford, United
Kingdom.
Weitzman, M. L. 1993. What to preserve? An application of diversity
theory to crane conservation. The Quarterly Journal of Economics
108:157–183.
Weitzman, M. L. 1998. The Noah’s Ark problem. Econometrica 66:1279.
Conservation Biology
Volume 23, No. 2, 2009