Efficiency of Incentives to Jointly Increase Carbon Sequestration And

PCA ETR:ITOUTR PERSPECTIVE INTRODUCTORY FEATURE: SPECIAL

Ecosystem services: From theory to implementation

Gretchen C. Daily*† and Pamela A. Matson‡ *Center for Conservation Biology (Department of Biology) and Woods Institute for the Environment, Stanford University, Stanford, CA 94305-5020; and ‡Environmental Earth System Science, Stanford University, CA 94305-2210

Around the world, leaders are increasingly recognizing ecosystems as natural capital assets that supply life-support services of tremendous value. The challenge is to turn this recognition into incentives and institutions that will guide wise investments in natural capital, on a large scale. Advances are required on three key fronts, each featured here: the science of ecosystem production functions and service mapping; the design of appropriate finance, policy, and governance systems; and the art of implementing these in diverse biophysical and social contexts. Scientific understanding of ecosystem production functions is improving rapidly but remains a limiting factor in incorporating natural capital into decisions, via systems of national accounting and other mechanisms. Novel institutional structures are being established for a broad array of services and places, creating a need and opportunity for systematic assessment of their scope and limitations. Finally, it is clear that formal sharing of experience, and defining of priorities for future work, could greatly accelerate the rate of innovation and uptake of new approaches.

ven in the face of intensifying conservation and human development, creasing carbon sequestration do not pressures and risks on the global and for incorporating material and in- necessarily increase species conservation environmental front, there is a tangible values of natural capital into (and vice versa). A clear finding is that growing feeling of Renaissance decision-making. Tallis et al. (10) ana- if payments for ecosystem services are inE the conservation community. This lyze World Bank projects with win–win not carefully designed, they may yield flows from the promise in reaching, to- objectives of alleviating poverty and pro- minimal gains in services of interest, gether with a much more diverse and tecting biodiversity, and find a success and may well harm the production of powerful set of leaders than in the past, rate of one in six. Using case studies, other services and biodiversity conserva- for new approaches that align economic they then propose a framework for an- tion. However, the authors demonstrate forces with conservation, and that ex- ticipating and improving the outcomes how new tools can enable good design plicitly link human and environmental of such projects. and progress toward multiple, poten- well-being (1). And this promise is flow- Ma¨ler et al. (11) review the history of tially competing objectives. ering thanks to substantial recent ad- green accounting and identify two major Naidoo et al. (13) attempt to quantify vances in key areas of inquiry, such as challenges to incorporating natural capital and map the production of ecosystem ser- ecology, economics, and institutions, systematically into economic accounts: (i) vices globally, to compare service produc- and their integration (2–5). the characterization of production function with priority sites for biodiversity Conservation efforts now are expanding tions for ecosystems, i.e., dynamic models conservation. They find that spatial con- into realms well beyond reserves, beyond that translate the structure and function cordance among different services and charity, and beyond biodiversity—and of ecosystems into the provision of ser- between ecosystem services and conserva- into the mainstream (6). While retaining a vices; and (ii) the development of institu- tion priorities varies widely. Nonetheless, core focus on protected areas designed to tions whose reach and strength is tightly their analysis permits clear identification sustain biodiversity, the new arenas of knitted to the estimation of accounting of areas in which payments for ecosystem conservation are much bigger and much prices for ecosystem services. Under weak services (PES) are more likely than else- more complex than the old. They encom- institutions, accounting prices will be low where to achieve biodiversity conservation pass new places dominated by human ac- (or even negative); as institutions improve, objectives. tivity, new revenue streams from public one expects (all else equal) accounting and private sectors, and new goals of eco- prices to increase. Challenges of Implementation system service provision. In fact, they en- Tallis et al. (10) and Ma¨ler et al. (11) The special issue then turns to policy compass important elements of tradi- both make compelling calls for intensive, design and implementation. Jack et al. tional, non-Western approaches (7, 8). interdisciplinary study of priority ecosys- (14) systematically review the history of Scholars and practitioners are seeking to tems and ecosystem service-oriented incentive-based mechanisms for environ- make conservation economically attractive projects, in which the potential for rapid mental policy, drawing lessons and in- and commonplace, routine in the deci- general advances in understanding is high. sights for the design of PES schemes. sion-making of individuals, communities, They also call for standardized techniques Such schemes compensate individuals or corporations, and governments (9). and metrics for valuing and monitoring communities for undertaking actions Here, we feature contributions that services. that increase the provision of ecosystem span the fundamental science of ecosys- services. The authors illustrate how the tem services through to the design, im- Modeling Provision of Ecosystem effectiveness of PES schemes is influ- plementation, and assessment of finance Services and Biodiversity Conservation enced by the biophysical, socioeconomic, and policy mechanisms and systems of The next two contributions take big political, and general dynamic context, governance. Each contribution is ori- steps in the directions suggested. Nelson giving concrete examples. ented around decisions, often cast in et al. (12) present a model that inte- Cowling et al. (15) go a step further, terms of tradeoffs among alternative grates the effects of policy on land-use proposing a pragmatic, operational future scenarios of change, whether in decisions and the resulting consequences natural resource management, popula- for the joint provision of ecosystem ser- tion, climate, or other key drivers. vices and biodiversity conservation Author contributions: G.C.D. and P.A.M. wrote the paper. across a landscape. They use data from The authors declare no conﬂict of interest. Linking Conservation and Development the Willammette Basin in Oregon, †To whom correspondence should be addressed. E-mail: We open with two pieces that set the United States, a very well studied re- [email protected]. stage, presenting frameworks for linking gion, and find that policies aimed at in- © 2008 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0804960105 PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9455–9456 model for achieving the safeguarding of and involvement, and maneuvering of frameworks and theory to practical in- ecosystem services in a given place. the political system at critical times. Al- tegration of ecosystem services into They focus on internalizing ecosystem though enabling legislation was essential, decision-making, in a way that is credi- service concerns into land- and water- it was not sufficient for shifting gover- ble, replicable, scalable, and sustain- use planning sectors, based on experi- nance toward adaptive co-management able. There remain many highly of the marine system. ence in South Africa. At the core of nuanced scientific challenges for ecolo- their model are three key elements, all Finally, we turn to China, with the world’s largest population and fastest gists, economists, and other social sci- very challenging but important to growing economy among major na- entists to understand how human achieve: socially relevant, user-inspired tions. Planned investments in ecosys- actions affect ecosystems, the provision research, stakeholder empowerment, tem service payments in China exceed of ecosystem services, and the value of and adaptive management embedded in 700 billion Yuan (1 US$ Ϸ 7.4 Yuan), those services. At least as demanding learning organizations. With these, one a magnitude matched by the ambition are the social and political challenges can establish the necessary enabling in their goals, the massive scales over associated with incorporating this un- conditions, windows of opportunity, which they operate, and their poten- derstanding into effective and enduring tially enormous impacts. Liu et al. (17) mechanisms for change, and outcomes institutions, to manage, monitor, and of effectiveness. review China’s foremost two ecosystem provide incentives that accurately re- Through a contrasting point of entry, service programs, the Natural Forest flect the social values of ecosystem ser- Olsson et al. (16) explore the strategies Conservation Program and the Grain vices to society. The candid analyses and actions that enabled the case of a to Green Program. To realize the po- recent transition to ecosystem-based tential for these programs to benefit presented here help light the way. management by the Great Barrier Reef China and the rest of the world, Liu et al. call for more systematic planning, ACKNOWLEDGMENTS. We are grateful to many Marine Park Authority in Australia. In diversified funding, effective compen- colleagues for insight and inspiration. We thank this transition, the focus of governance sation, integrated research, and com- especially the Beijer Institute of Ecological Eco- shifted from protection of selected indi- prehensive monitoring. nomics (of the Royal Swedish Academy of Sci- vidual reefs to stewardship of the much ences) and participants and supporters of the larger-scale seascape. Olsson et al. iden- Making Ecosystem Service Approaches Natural Capital Project, including the MacArthur Foundation, the Packard Foundation, the Re- tify management innovations that Operational sources Legacy Fund Foundation, the Winslow greatly improved coordination of the Radical transformations will be re- Foundation, Peter and Helen Bing, John and scientific community, public awareness quired to move from conceptual Tashia Morgridge, and Vicki and Roger Sant.

1. Hassan R, Scholes R, Ash N, eds (2005) Ecosystems and 8. Ostrom E (2005) Understanding Institutional Diversity USA 105:9495–9500. Human Well-Being (Island, Washington, DC). (Princeton Univ Press, Princeton). 14. Jack BK, Kousky C, Sims KRE (2008) Designing payments 2. Dasgupta P (2001) Human Well-Being and the Natural 9. Daily GC, Ellison K (2002) The New Economy of Nature for ecosystem services: Lessons from previous experi- Environment (Oxford Univ Press, Oxford). (Island, Washington, DC). ence with incentive-based mechanisms. Proc Natl Acad 3. Turner BL, II, et al. (2003) A framework for vulnerability 10. Tallis H, Kareiva P, Marvier M, Chang A (2008) An Sci USA 105:9465–9470. analysis in sustainability science. Proc Natl Acad Sci USA ecosystem services framework to support both practi- 15. Cowling RW, et al. (2008) An operational model for 100:8074–8079. cal conservation and economic development. Proc Natl mainstreaming ecosystem services for implementation. Acad Sci USA 105:9457–9464. 4. Salzman J (2005) Creating markets for ecosystem Proc Natl Acad Sci USA 105:9483–9488. 11. Ma¨ ler K-G, Aniyar S, Jansson Å (2008) Accounting for services: Notes from the ﬁeld. NYU Law Rev 80:870– 16. Olsson P, Folke C, Hughes TP (2008) Navigating the ecosystem services as a way to understand the require- 961. transition to ecosystem-based management of the ments for sustainable development. Proc Natl Acad Sci 5. Ruhl JB, Kraft SE, Lant CL (2007) The Law and Policy of USA 105:9501–9506. Great Barrier Reef, Australia. Proc Natl Acad Sci USA Ecosystem Services (Island, Washington, DC). 12. Nelson E, et al. (2008) Efﬁciency of incentives to jointly 105:9489–9494. 6. Goldman RL, Tallis H, Kareiva P, Daily GC (2008) Proc increase carbon sequestration and species conservation 17. Liu J, Li S, Ouyang Z, Tam C, Chen X (2008) Ecological Natl Acad Sci USA 105:9445–9448. on a landscape. Proc Natl Acad Sci USA 105:9471–9476. and socieoeconomic effects of China’s policies for 7. Berkes F, Folke C (1998) Linking Social and Ecological 13. Naidoo R, et al. (2008) Global mapping of ecosystem ecosystem services. Proc Natl Acad Sci USA 105:9477– Systems (Cambridge Univ Press, Cambridge, UK). services and conservation priorities. Proc Natl Acad Sci 9482.

9456 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0804960105 Daily and Matson PCA ETR:PERSPECTIVE FEATURE: SPECIAL

An ecosystem services framework to support both practical conservation and economic development

Heather Tallis*†, Peter Kareiva‡, Michelle Marvier§, and Amy Chang¶ *Department of Biology and Natural Capital Project, Woods Institute for the Environment, Stanford University, Stanford, CA 94305; ‡The Nature Conservancy, 4722 Latona Avenue NE, Seattle, WA 98105; §Environmental Studies Institute, Santa Clara University, Santa Clara, CA 95053; and ¶The Nature Conservancy, Santa Clara, CA 95053

Edited by Gretchen C. Daily, Stanford University, Stanford, CA, and approved February 8, 2008 (received for review October 5, 2007)

The core idea of the Millennium Ecosystem Assessment is that the human condition is tightly linked to environmental condition. This assertion suggests that conservation and development projects should be able to achieve both ecological and social progress without detracting from their primary objectives. Whereas ‘‘win–win’’ projects that achieve both conservation and economic gains are a commendable goal, they are not easy to attain. An analysis of World Bank projects with objectives of alleviating poverty and protecting biodiversity revealed that only 16% made major progress on both objectives. Here, we provide a framework for anticipating win–win, lose–lose, and win–lose outcomes as a result of how people manage their ecosystem services. This framework emerges from detailed explorations of several case studies in which biodiversity conservation and economic development coincide and cases in which there is joint failure. We emphasize that scientiﬁc advances around ecosystem service production functions, tradeoffs among multiple ecosystem services, and the design of appropriate monitoring programs are necessary for the implementation of conservation and development projects that will successfully advance both environmental and social goals. The potentially bright future of jointly advancing ecosystem services, conservation, and human well-being will be jeopardized unless a global monitoring effort is launched that uses the many ongoing projects as a grand experiment. poverty alleviation ͉ pro-poor conservation ͉ sustainable management

Poverty and environmental problems ment agencies that seek to also provide well-being and increasingly designs are both children of the same for the stewardship of nature under the projects in terms of the provision of mother, and that mother is mantra of sustainable development. ecosystem services. ignorance. These projects have variously been cate- Ali Hassan Mwinyi, gorized as integrated conservation– An Unclear Record for Projects that Seek Tanzanian President in 1998 development projects, community-based Both Conservation and natural resource management, and, Poverty Alleviation he Millennium Ecosystem As- more recently, pro-poor conservation There is no question that ecosystem ser- sessment (MA) contains a com- (3). Although these projects generally vices are now a focus for both scientists pelling argument that human lack a formal foundation of ecosystem and conservationists, but interest and well-being depends on the ser- widely held beliefs do not necessarily Tvices provided by nature, and that these service science, they all are motivated mean that data and results support the services have recently become so imper- by the general hypothesis that nature feasibility of helping both people and iled that we can expect negative feed- provides humans with benefits. In fact, biodiversity by maintaining ecosystem backs to people (1). In some ways there an analysis of the vision and mission services. What lessons have been is nothing new in this message: people statements of major environmental or- learned from the many projects already depend on nature, and people too often ganizations, including the major conser- conducted by conservation NGOs in damage nature, thereby endangering vation nongovernmental organizations which efforts have been made to both their own health and well-being. The (NGOs), found that nature is typically improve human well-being and protect novel contribution of the MA is its portrayed as resources necessary for hu- biodiversity? Answers are not as forth- championing of a new scientific focus, a man well-being and sustainable develop- coming as one would hope. For exam- focus on understanding how nature pro- ment (4). Increasingly, the language duces a wide array of ecosystem ser- used by conservation NGOs to discuss ple, in the most thorough systematic re- vices, quantifying the rate and value of the value of nature is becoming more view we know of regarding conservation the delivery of these services, and mod- explicit in its reference to ecosystem projects, researchers were stymied by eling the connections between ecosys- services. Examining the web domains of the lack of metrics of success for either tem services, human welfare, and four major conservation NGOs (Conser- conservation or poverty alleviation (5). economic systems (2). vation International: http://conservation. Indeed, the major conclusion of this re- Although the MA was a bold contri- org; The Nature Conservancy: www. view was that projects tended to rely on bution that exposed huge gaps in the nature.org; Wildlife Conservation overly simplistic definitions of both science of ecosystem services, the reality Society: http://wcs.org; and World Wild- biodiversity and poverty, and few is that both the conservation and eco- life Fund: www.worldwildlife.org), we nomic development communities have found that, on average (Ϯ1 SE), 7.8 Ϯ Author contributions: H.T., P.K., and M.M. designed re- embraced ecosystem services for at least 1.3% of the pages that mention biodi- search; H.T., P.K., M.M., and A.C. performed research; H.T., a decade, without explicitly labeling versity also specifically mentioned one P.K., M.M., and A.C. analyzed data; and H.T., P.K., and M.M. them as such. This melding of conserva- or more of the following terms: human wrote the paper. tion and development comes from two welfare, ecosystem service(s), human The authors declare no conﬂict of interest. distinct agendas: conservationists who well-being, drinking water (Google This article is a PNAS Direct Submission. seek to increase public support for search engine; August 23, 2007). Thus, †To whom correspondence should be addressed. E-mail: biodiversity protection by integrating the conservation community assumes a [email protected]. economic development, and develop- connection between nature and human © 2008 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705797105 PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9457–9464 40 A Framework for Analyzing Ecosystem Service Projects Aimed at Human 35 Well-Being and Biodiversity Protection 30 A number of authors have recently ar-

s gued that there are strong links between t 25

e ecosystem services and sustainable de- j oc r 20 velopment, particularly development ef- p

o forts that aim to reduce rural poverty f 15 % (8–10). We see two distinct routes by 10 which the science of ecosystem services can contribute to both nature conserva- 5 tion and sustainable development. First, 0 a thorough accounting of ecosystem ser- Pollution & Environ Water Biodiversity Land admin & Climate Other vices and a better understanding of how environ health policies & resources management change and at what rates ecosystems produce institutions management these services can be used to motivate payment for nature conservation. At Fig. 1. Shown are the percentage of projects listing each environmental subtheme for the 1,961 World least part of the generated funds can be Bank projects with a major theme of environment and natural resources management (ENRM) and an used to compensate people who suffer Ͼ approval date. The percentages sum to 100, because nearly half of the ENRM projects listed more than lost economic opportunities to protect one environmental subtheme. A maximum of ﬁve major themes or subthemes can be listed for each project. these services. For example, if rural poor are asked to take actions that reduce farm productivity to protect and projects directly examined the causal written between July 1998 and August regulate water supply, those farmers relationships responsible for the out- 2006 (when the format was consistent), could be compensated for the reduced comes observed. and (iv) stated environmental and pov- productivity they experience. When the Arguably, the most consistently col- erty alleviation goals and outcomes. benefits of natural ecosystems are ex- lected data regarding projects that seek We found that relatively few, only five plicitly quantified, those benefits are improvements in human well-being and of 32 projects (16%), had substantial more valued both by the people who the environment (or biodiversity) are gains in terms of their stated environ- directly interact with the ecosystems and found in the records of the World mental and poverty alleviation out- the governmental and other agencies Bank. Of the 11,155 economic develop- comes (Fig. 3). Thus, it is not impossible that would have to pay for substitute ment projects carried out by the World to make gains on both biodiversity and sources of these services if these ecosys- Bank since 1947 (www.worldbank.org; poverty fronts, but it is not easy. It is tems should become impaired. Appreci- accessed December 17, 2007; limited to noteworthy that nearly half of the World ating the value of ecosystem services closed and active projects with an ap- Bank projects with biodiversity protec- can motivate increased conservation in- proval date listed), 17.6% have had tion as a major goal had no objectives vestment to prevent having to pay for ‘‘environment and natural resources for poverty alleviation, which precluded substitutes later. This approach could be management’’ as a major theme. From them from being in our examination of characterized as a ‘‘government investment’’ approach because the payments a conservation perspective, it is note- win–win frequency. will generally come from beneficiaries worthy that 20% of the World Bank’s Unfortunately, this simple tally of outside of the local area, and a govern- environmentally oriented projects have World Bank scores cannot reveal what specifically included biodiversity pro- mental or other agency is typically strategies or factors tend to lead to fail- tection as a theme (Fig. 1). Moreover, responsible for collecting and redistrib- ure. Failures could be caused by poor the attention to biodiversity has been uting the funds. project design or intrinsic tradeoffs be- growing over the last 20 years, nearly Second, a focus on the conservation tween ecosystem services. For example, quadrupling in frequency (Fig. 2). of ecosystem services could improve the development of more productive agricul- Thus, the World Bank has two decades success of projects that attempt to both of experience supporting projects with ture, which is often a route by which conserve nature and improve the wel- the dual goal of economic development rural poverty is alleviated (7), may inex- fare of the rural poor by fostering mar- and biodiversity protection (6). In ad- orably harm biodiversity. Alternatively, kets for the goods and services that dition, the World Bank uses a categori- failures may be driven by external prob- local people produce or extract from cal scoring system by which it judges lems such as civil conflict or weak gov- ecosystems. These projects could be the degree to which each completed ernance that were not anticipated in characterized as more ‘‘community- project achieved its stated goals. One project design and have nothing to do based’’ because the goal is to foster the can imagine three general categories of with ecosystem services. Understanding more organic, or grassroots, develop- outcome for projects with both biodi- the factors that determine the outcome ment of cottage industries, such as eco- versity goals and economic develop- of projects with dual biodiversity and tourism, or the production of bushmeat ment or poverty alleviation goals: economic development goals is crucial. or nontimber forest products, that are win–win (both goals are met), lose– One path to understanding these out- enhanced by better protection of local lose (neither goal is met), and win–lose comes is to develop a framework for ecosystems. (only one of the goals is met). We assessing the connections between eco- Using this framework, we discuss the searched the World Bank database for system services and economic develop- factors that may have contributed to projects that (i) were approved be- ment on a project-by-project basis and failure or success for several projects tween 1993 and 2007, (ii) listed biodi- suggest indicators and metrics that could (Table 1). The scale of our analysis is versity as a theme, (iii) had available increase the likelihood of win–win local or project level. Often the hyper- an Implementation Completion Report outcomes. bole of ‘‘pro-poor conservation’’ might

9458 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705797105 Tallis et al. 45 Payment for Ecosystem Services in China. The government of China has imple- 40 pollution & env health mented several payment for ecosystem 35 service programs. One of the better known is the Sloping Land Conversion 30 Program. This program was motivated 25 by large floods on the Yangtze River in biodiversity 1999 that many presume were worsened 20 by sedimentation that has reduced the % of projects 15 flood mitigation potential of dams on climate change this river. The State Forestry Adminis- 10 tration linked sedimentation to erosion 5 from intensively farmed sloping lands in the upper reaches of the watershed. 0 The Sloping Land Conversion Pro- 83-89 90-94 95-99 00-04 05-current gram, also known as the Grain to Fig. 2. For the 1,961 World Bank projects with the ENRM major theme, the percentage of projects with Green Program, pays farmers in cash a pollution and environmental health theme has remained ﬂat or declined, whereas the percentage of or grain to abandon farming and re- projects with subthemes of biodiversity and/or climate change has increased. The numbers of projects with store forests on steep slopes along key the ENRM theme are: 1983–1989, 116; 1990–1994, 417; 1995–1999, 493; 2000–2004, 521; and 2005– rivers (refs. 13 and 14 and www.forest- present, 414. trends.org/documents/publications/ ChinaPES%20from%20Caro.pdf). The State Forestry Administration runs this be mistaken for a claim that conserva- and unsustainable land use practices program and projects that 14.6 million tion is a global strategy for poverty surround the reserve, leading to con- hectares in 24 provinces will be im- alleviation. It is not (11). The plight of tinuing declines in biodiversity and wa- proved through this program by 2010 urban poor and many of the world’s ter quality. (13). The upper watershed regions tar- poor has little to do with the production In 2000, The Nature Conservancy in geted by this program are often home to functions of ecosystem services. But for collaboration with the U.S. Agency for some of the most socially marginalized rural poor, at the local level, the status International Development worked to minority groups in the country, coinci- of ecosystem services can make a big establish a water fund that directs difference in their daily lives. dentally allowing social, economic, and money from water users to improve pro- ecological benefits to flow from this tection of the Condor Biosphere Re- Government Investments in program. serve (12). In 2004, the fund was worth Ecosystem Services To date, payments for ecosystem ser- $2.1 million, paid into by the Quito Mu- vices in this and other programs have When there is clear information that nicipal Water and Sewage Agency, the ecosystem services of obvious value (like come primarily from China’s central Quito Electricity Company, and the government. In 2004, 92% of the accu- clean water or flood control) are imper- Andina Beer Company (12). In this iled, governments often invest in their mulated value of the Sloping Lands Pro- case, the government and its municipal gram ($7.6 billion) was provided by the protection. The money for these invest- utilities chose to redirect current fees ments may come from charging benefi- national government. It is expected that and taxes rather than raise the price of the high costs of the program will soon ciaries for the use of ecosystem services, water. such as clean water, and then using the spur the development of privately The project reports successes on both payments either to improve enforcement funded payment schemes (14). As with social and ecological fronts. The nearly of protected areas or to compensate the Quito Water Fund, this program $5 million raised for conservation action those whose livelihoods are diminished was not motivated by biodiversity, but have been used to plant 3.5 million by the conservation efforts (12). These rather by interests in ecosystem services trees, hire nine new park guards that arrangements typically involve payments such as flood mitigation. If China had from beneficiaries outside of the imme- provide new jobs and increase enforce- not suffered severe floods in 1999 it is diate area. Here, we highlight two ment, build local capacity for monitor- unlikely that the Sloping Lands Program projects that demonstrate how ecosys- ing and conflict resolution, fund hydro- would have been launched; there first tem services can be used to motivate logic modeling and monitoring, and needed to be strong evidence of a con- payment for nature conservation. provide environmental education to nection between land use and degraded children (12). Financial support for con- ecosystem services. The Quito Water Fund. Quito, the capital servation came not from individuals city of Ecuador, houses 1.8 million peo- concerned about biodiversity or some Aligning Conservation of Ecosystem ple in a region of extremely high biolog- particularly charismatic species on the Services with Local Economic Activity ical diversity and endemism. In the most verge of extinction. Rather the support Many early projects that integrated strained regions around Quito, 63% of came from an appreciation of the role conservation and development used eco- water needs are not met, and in 1999, healthy forests play in supplying and tourism, which offers market-based in- most of the city’s 14 monitored water- regulating the availability of clean water. come and a natural alliance between sheds flowed with undrinkable water. A key to the success of this project may conservation of an ecosystem service The Condor Biosphere Reserve, situated have been a long record of flow and and economic development. Participants upstream of Quito, has the potential to sedimentation monitoring data collected in these programs typically receive funds alleviate social struggles around water by hydropower operations, which pro- directly from corporations or others in resources and environmental struggles vided a clear signal of a degrading eco- the private sector, rather than from around biodiversity loss. However, en- system service before any catastrophic governments or nongovernment organi- forcement of park restrictions was weak, event. zations. Here, we briefly outline two

Tallis et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9459 18 population and game species such as elephants, zebra, oryx, and springbok 16 have increased 600% in some places 14 (18). For the first time in 30 years, sea- 12 sonal migrations have resumed between Botswana and Namibia. The program 10 was a major development success, with 8 local incomes increased by a total of 6 $2.5 million in 2004 and Namibia’s net national income up by $9.6 million (18).

number of projects 4 Overall, 3,250 part-time and 547 full- 2 time jobs were created, and the fact that the majority of these jobs were obtained 0 by women meant that gender equity was no gain-no gain gain-no gain gain-gain substantially improved (18). acheivement of environmental & poverty alleviation Projects that Failed to Advance outcomes Conservation and Development Fig. 3. The World Bank database was searched for projects that (i) were approved between 1993 and The case studies described above indi- 2007, (ii) listed biodiversity as a theme, (iii) had available an Implementation Completion Report written cate that both conservation and en- between July 1998 and August 2006 (when the format was consistent), and (iv) stated environmental and hancement of human well-being can be poverty alleviation goals and outcomes. A total of 32 projects were found. Projects rated as having high accomplished as part of one project. It or substantial achievement of environmental and/or poverty alleviation outcomes are considered to have is important, however, to ask about fail- made a gain, whereas those rated as having modest or negligible achievement are considered to have ures. Failures are quite common. In a made no gain. random sample of 194 Implementation Completion Reports for World Bank programs that have successfully taken tures or entry costs. Some integrated projects, 16% were judged to be unsatis- this approach. conservation and development projects factory or highly unsatisfactory overall have found success in reducing these performance. Political or economic fac- The Il’Ngwesi Ecolodge. In Kenya, the barriers by granting local control of the tors outside the control of the project Maasai culture and highly diverse bio- management of biological resources to frequently contribute to their failure logical communities have been under communities or individuals. In Namibia, (Table 1). It would be extremely foolish threat from the same forces: cattle rus- a global biodiversity hotspot (the Cape to neglect the influence of external tling and poaching, alongside pressure Floristic region) overlaps with high forces and global markets and corrup- from the government to subdivide and hunting and poaching pressure, high tion and governance on the ability of ecosystem service projects to enhance develop land (15). Subdivision would poverty, and socially marginalized popu- human well-being. Here, we describe have led to the end of pastoral life for lations of indigenous people. Game pop- two examples of projects that have the Maasai and fragmentation of some ulations have declined, large-scale ani- failed on both environmental and pov- of the largest remaining tracts of wild- mal migrations have ceased, and black erty alleviation fronts. life habitat in Kenya. In 1995, 8,000 rhino populations have plummeted (18). members of a Maasai community at As part of the institutional change asso- West African Wildlife Project. Like Nami- Il’Ngwesi agreed to establish an ciated with Namibia’s independence, the bia’s Conservancy Program described ecolodge and promote tourism (15). As Nature Conservation Act was passed in above, the West Africa Pilot Community- a result, security in the region has in- 1996. The World Wildlife Fund became based Natural Resources and Wildlife creased, income from the ecolodge is involved in supporting communities who Management Project sought ‘‘a common funding the local school, poaching has wanted to enter into a ‘‘conservancy solution to both development and con- been halted, 13 of the region’s 19 large program’’ established by the act. servation concerns by involving local mammals (including the endangered Enrollment in the program gives com- communities in the sustainable, profit- Grevys zebra) have returned to the area, munities rights, for the first time, to able exploitation of wild resources and and some wildlife populations have in- huntable game and revenues from game assisting them to manage their wild land creased by as much as 500% (15). Obvi- products and tourism (18). The act has areas for their own economic benefit ously all ecotourism projects do not turn fundamentally changed the landscape in and for the benefit of biodiversity.’’ out so well, with failures commonly oc- Namibia. More than 60 communities This, however, is an example of an ef- curring because of either poor business now participate, supporting 31 conser- fort that failed to achieve both its bio- planning or poor ecological manage- vancies that cover some 70,000 km2,or diversity and development goals (19). ment. Nonetheless, ecotourism is well 17% of Namibia’s land area (18). In this Supported by the World Bank and initi- established as a joint development and case, new revenue streams were created ated in 1995, the project faced many conservation strategy, and the world’s by opening access to existing interna- difficulties including an initially very low two largest conservation NGOs (World tional markets. level of training among villagers, a Wildlife Fund and The Nature Conser- Again, this project has been a success deeply rooted mistrust of central gov- vancy) have developed special planning on both social and ecological fronts. ernment within local governing bodies, processes to help their field staffs pro- Many of the conservancies are on lands and the resumption in 2004 of civil con- mote ecotourism (16, 17). that now act as corridors between pro- flict in one of the participating nations, tected areas (18). Wildlife populations Cote D’Ivoire. Namibia’s Conservancy Program. In many have increased dramatically on conser- The project’s antipoaching program cases, ecotourism markets are not easily vancy lands. Namibia now houses the failed because of the lack of a legal ba- accessible because of regulatory struc- world’s largest free-roaming black rhino sis for villagers to apprehend poachers;

9460 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705797105 Tallis et al. Table 1. Examples of projects that have used ecosystem services to advance both conservation and development or poverty alleviation Project Conservation Development/poverty alleviation

Successes Quito’s Water Fund 3.5 million trees planted Alternative income, nine new jobs Nine park guards added Education Hydrology monitoring program started Clean water Conflict resolution training Technical capacity building China’s Sloping Lands Program 14.6 million hectares reforested (2010) Alternative income Targeted ethnic minority groups Flood control Kenya’s Il’Ngwesi Ecolodge Increasing wildlife populations Alternative income Poaching controlled Way of life Education (school funded) Security (poaching controlled) Namibia’s Conservancy Program Increasing wildlife populations Property rights Overgrazing controlled Income Landscape connectivity improved Cultural equality Gender equality Way of life South Africa’s Cape Peninsula Biodiversity Invasive plant eradication Improved infrastructure Project Antelope species reintroduction Income Increasing raptor populations Establishment of protected area Failures West African Wildlife Project Poaching Insufficient legal power Livestock encroachment Insufficient communication infrastructure No monitoring Civil unrest Insufficient capacity Governmental distrust India’s Shrimp Aquaculture Development Habitat destruction Unsustainable jobs Program Overharvest Unstable market

Azov-Black Sea Corridor Program Unenforced protected areas Insufﬁcient legal power Unsustainable agriculture Uninformed agricultural practices Habitat destruction Dysfunctional environmental institutions Kerinci Seblat Conservation and Poaching and wildlife decline Insecure land tenure Development Project Agriculture encroachment Insufﬁcient infrastructure Logging threat Unsustainable agriculture and nontimber forest products harvest Habitat destruction and fragmentation Weak and uncoordinated governance insufficient investment in high-quality example of a lose–lose situation in ecology of the farmed species sets up a weapons, which put village teams at a which neither people nor nature bene- conflict between shrimp pond develop- major disadvantage in conflicts with fited from a well intentioned project. ment, which is largely done by clearing poachers; and insufficient investment in The original idea of a project subsidizing mangroves, and shrimp production, a communications infrastructure, which shrimp aquaculture in Andhra Pradesh, which requires mangrove habitat for the caused slow response times. Moreover, India was to create economic opportu- provision of wild broodstock. weak and incomplete implementation of nity for local communities through Overdevelopment of the shrimp in- ecological and sociological monitoring shrimp farming, which can be enor- dustry in the Andhra Pradesh region has made it impossible to assess what was mously profitable. Unfortunately, the poised both economic and ecological working or not working. In the end, clearing of mangroves for shrimp farms systems on the brink of collapse. The there was no compelling evidence that can have negative impacts on the envi- area’s 61 hatcheries are suffering from a the wildlife or the local villages had ronment and the industry itself. widely fluctuating market with alternat- benefited from the project, and there In many parts of the world, including ing low demand and limited supplies of was anecdotal evidence that poaching Andhra Pradesh, shrimp aquaculture is broodstock (20). In fact, broodstock are and livestock encroachment had re- done with the shrimp species Penaeus at times so rare in this region that the sumed in wildlife zones. monodon (20). Because it is difficult to price for a single reproductive female bring this species to reproductive matu- shrimp can be as high as $2,000 (20). A Shrimp Aquaculture in Former Mangrove rity in captivity, hatcheries rely on wild careful assessment of the ecological and Forests. The clearing of coastal man- broodstock to maintain production (20). economic capacity of the system for grove forests for shrimp aquaculture in The wild shrimp populations, in turn, shrimp production could have identified coastal regions of Asia provides another rely on mangrove habitats. Thus, the the optimal number of farms the region

Tallis et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9461 could support at maximum capacity and a low broodstock prices. Operating at this optimal level would have saved subsidy funds that were used to develop the farms, provided sustainable jobs for the region, and retained more of the natural mangrove habitat that is key to storm protection (21) and the provision of other ecosystem services. This project failed in large part because the myriad b services provided by intact mangrove ecosystems, and the tradeoffs among these services, were not fully appreciated. Scientific Understanding that Could Improve the Likelihood of Win–Win Outcomes Projects that use ecosystem services to c simultaneously advance conservation and human agendas could benefit from improved scientific understanding of four key overriding issues: sustainable use of ecosystem services, tradeoffs among different services, the spatial flows of services, and economic feedbacks in ecosystem service markets. Bet- Fig. 4. ‘‘Tradeoff flowers’’ depicting alternative scenarios for ecotourism projects aimed at biodiversity ter quantification of ecosystem services protection and economic growth. (a) Unrestrained ecotourism can lead to infrastructure and human will reduce the transactional costs of traffic that degrades many ecosystem services, and ecotourism itself collapses. (b) Ecotourism develops policies based on ecosystem services, with good management of biodiversity and ecosystem services, so that income flows from tourism, will promote greater equity, and reduce biodiversity is enhanced, and ecosystem services are not lost. (c) Ecotourism develops and biodiversity is the risk of unintended consequences or protected in nature reserves, but the increase in roads and hotels undermines water quality and fisheries, surprises. Better quantification of eco- causing tradeoffs among ecosystem services and development. Which outcome is realized is largely a system services can also help to identify matter of a good management plans and making sure the intensity of human use is not too high. those situations where the economic benefits are sufficiently large to counter- growth rate. Because the success of a tion of both conservation and develop- balance money that can be made from project depends to a large extent on the ment agencies. For instance, habitat loss illegal abuses of an ecosystem (such as starting conditions and use levels, any and landscape change are documented poaching or illegal timber harvest). given strategy will not be equally effec- to have played a major role in the in- Any time a human endeavor seeks to tive in all places. Further, it is relatively creased prevalence of vector-borne dis- derive something of value from our nat- trivial to show that extreme impacts eases such as Lyme disease (24), West ural world, there is the possibility that arise from extreme actions, but we lack Nile virus, Chagas disease, and tick overuse or overexploitation could result the ecological production functions that borne encephalitis. Although claims are in failure. One of the most basic principles of ecology is the notion of a maxi- quantify the impacts of more incremen- commonly made about human health mum sustainable yield (MSY) derived tal changes. For example, it is well un- benefits from well managed natural sys- from intermediate levels of use. If this derstood that total deforestation yields tems (25), no conservation projects have level of use is exceeded, the ecosystem erosion, impaired water quality, and di- yet taken advantage of our growing un- or natural resource is degraded and minished flood control, but we do not derstanding of the link between land- both nature and humans lose (22). One have the ability to predict how the pro- scape change and human diseases or important lesson from the MSY concept vision of these same services would be health. Also largely absent from on-the- is that levels of extraction that provide affected by 10% versus 20% deforesta- ground ecosystem service projects are meaningful amounts of income may be tion (23). Given the extensive literature any of the supporting services such as feasible only under particular ecological describing dramatic changes in ecosys- soil formation, nutrient cycling, and pol- conditions. Consider, for example, a tem function in response to such things lination. We must simultaneously con- project similar to the conservancy pro- as complete deforestation or extreme sider multiple ecosystem services and gram in Namibia that aims to increase runoff of fertilizers into waterways, we multiple production functions, not just local income and wildlife populations by should be able to design projects that to obtain a more complete understand- encouraging a relatively low level of le- avoid such severe consequences. What is ing of the benefits and losses, but be- gal trophy hunting of kudu (Tragelaphus not well understood is how one might cause any one ecosystem service might strepsiceros), a large antelope. If the lo- pinpoint the extent to which a resource be related, either positively or nega- cal kudu population has a low popula- can be exploited without pushing an tively, to other services (26, 27). For tion growth rate, even a very low level ecosystem over its tipping point. example, developing ecotourism can of hunting (ecosystem service use) could Most real-world conservation, devel- bring income to local communities and lead to a decline in the population and opment, and ecosystem service projects this added revenue could foster im- a lose–lose outcome. Alternatively, the have paid attention to only a select few proved community stewardship of the same project could provide a win–win services that represent a narrow slice of natural features that attract ecotourists situation if it were initiated in an area the full spectrum provided by nature. (e.g., biodiversity for wildlife viewing or where the kudu population has a high Other services also deserve the atten- sports fisheries), providing a win–win

9462 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705797105 Tallis et al. Table 2. The relationship between nature and human well-being from the websites of major conservation NGOs NGO Relationship between nature and human well-being

Conservation International ЉThe rural poor within hotspots depend on the products of healthy ecosystems, harvesting wild plants for food, fuel, clothing, medicine and shelter. These services also help maintain energy and infrastructure activities that underpin economic development. For example, coral reefs, wetlands and mangroves can buffer beaches and prevent storm surges and coastal ﬂooding. The value of these ecological services is tremendous and we are only beginning to measure their signiﬁcance.Љ (http://web.conservation.org/xp/ CIWEB/strategies/humanwelfare) The Nature Conservancy ЉNature has tremendous assets that are key to human-well being and that have concrete economic valueѧit’s becoming increasingly clear that the goals of conservation and the goals of alleviating poverty and improving human health are deeply interwoven.Љ (www.nature.org/tncscience/bigideas/people/ art19846.html) World Wildlife Fund ЉThe world’s poorest people bear the brunt of forest loss, since forest resources sustain most of the 1.2 billion people in the world who live in extreme poverty. WWF is working locally, regionally and globally to address this threat and at multiple levels - with communities, governments and industry. In partnership we can ensure forests are protected for the people and species that depend on these habitats for their livelihoods.Љ (www.worldwildlife.org/forests) situation for nature and people (Fig. In addition to ecosystem services, one watershed to provide clean water 4b). On the other hand, excessive devel- there are economic feedbacks to con- for the masses? What happens when opment of infrastructure to support sider. Market expansion can affect that threshold is passed? Will construc- tourism, or excessive tourism activities prices, and there are often feedbacks tion of a water treatment plant remove such as hiking and fishing, can cause the between economic development, social the incentive for protection of the Con- decline of the very resources tourists change, and environmental condition. dor Biosphere Reserve? seek to enjoy, resulting in a lose–lose We see the challenges of dealing with Finally, good governance, rule of law, situation (Fig. 4a). For instance, rapid these complexities even in the examples and democracy may well be important development of trekking in Nepal has we have outlined as successes. As supply for achieving joint success in conserva- resulted in unsustainable overharvest of in any product increases, prices are tion and economic development. Statis- firewood for cooking, thereby damaging likely to fall. Because the goal of these tical analyses of economic growth in local ecosystems (28). Exactly how peo- projects is not to achieve an efficient poor countries hint at a connection be- ple manage ecosystems as they increase market, but rather to encourage a liveli- tween democracy and governance and the intensity of use in any natural sys- hood that is good for nature and people, economic success (32). It would not be tem can result in win–win, lose–lose, or low prices may cause failure. For in- surprising to find a similar link to win– tradeoff outcomes (Fig. 4c). Although stance, low prices for hunting and win ecosystem service projects. In Table scientists are now vigorously examining tourism in Namibia may make these 1, all of the win–win examples entail multiple ecosystem services, the practice livelihoods insufficient for well-being, strong functioning governments, whereas of conservation is still almost entirely and landowners may turn back to do- poor governance is implicated in three designed to focus on one service at a mestic livestock rearing or other more of the four lose–lose failures. We hy- time. Interlocking production models of destructive practices. The Namibian gov- pothesize that a thorough study of many the full suite of ecosystem services are ernment may want to limit the number case studies would uphold this trend. needed. of communities that can participate in If policy and financial incentives for the program so that prices remain high The Absence of Data Jeopardizes the conservation of ecosystem services are and participants receive enough income Success of Future Projects to be successful and equitable, we will to continue their investment in sound Increasingly, conservation organizations also need a solid scientific understand- game management. This example points and development groups are in the busi- ing of how services flow from one re- out the need to understand the focal ness of providing hopeful visions of peo- gion to another, what human groups market and the nature of demand for ple and nature jointly benefiting from benefit from ecosystem services, and the product of interest. conservation efforts (ref. 4 and see what groups or populations would need We also need to understand how the Table 2). However, the enthusiasm for to be compensated for protecting those growth of a market will affect the larger ecosystem services as a strategy for services. The science of ecology made social system and vice versa. For in- enhancing conservation support is far huge advances when it began to con- stance, expanding tourism in attractive outpacing credible evidence of what is sider dispersal and the importance of natural areas can stimulate immigration possible and how to best achieve the movement in governing the dynamics of of people hoping to benefit from ex- much desired win–win outcomes. Con- ecological communities (29). However, panded economic opportunities. This can servation that is justified on the basis of the science of ecosystem services has lead to a ‘‘tourism-income-population enhanced ecosystem services cannot af- not yet made this transformation, and as growth spiral,’’ which obviously in- ford to neglect rigorous evaluation of a result typically depicts ecosystem ser- creases pressure on local resources and both ecology and social well-being. In- vices as site-bound on static maps (ref. the environment (31) and can degrade deed, there are enough projects in place 30 and www.nj.gov/dep/dsr/naturalcap). the natural beauty that people initially around the world that if some simple Only when we have the scientific foun- traveled to view and enjoy. In Quito, the metrics were collected on each, it would dation to map the flows of ecosystem population continues to grow at 1.6% be possible to treat these efforts as a services, can we hope to meld conserva- per year (35). When will the needs for grand experiment. The natural science, tion with delivering benefits to the poor- residential development or demand on social science, and practitioner commu- est of the poor. the water supply outstrip the capacity of nities jointly need to establish a stan-

Tallis et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9463 dard set of measures and approaches for from heavy fertilizer application yielded (33, 34). This is a significant mistake, quantifying and monitoring ecosystem immediate economic benefits. However, and a mistake that is even more griev- service levels and values. Minimally, the dead zone in the Gulf of Mexico did ous in the case of ecosystem service projects should track what money funds not appear until 20 years after those projects. Whereas biodiversity cannot the protection of ecosystem services and initial gains in agricultural productivity. complain if it is not well served by a who provides it, who benefits from the Different ecosystems services will respond conservation project, people supposedly ecosystem services, and whether the on different temporal and spatial scales, benefiting from an ecosystem service project results in an impact on the deliv- and efforts to track interactions will have effort will be quick to ask for evidence ery of ecosystem services. to anticipate these different scales. of their enhanced well-being. Much of An added challenge with ecosystem Beyond the missed scientific opportu- the current enthusiasm for ecosystem service projects is that economic returns nity, conservation groups are risking service projects in the conservation and signals respond relatively quickly to damaged reputations because they have world is an act of faith. At some point, any action, whereas changes in ecosys- largely failed to deliver data that pro- however, that faith will need to be tem function may lag by decades. For vide evidence of a link between their backed up by irrefutable data showing example, the productivity gains in the actions and any improvement in the sta- that these projects benefit both people Mississippi River valley that resulted tus of biodiversity or ecosystem services and nature.

1. Millennium Ecosystem Assessment (2005) Ecosystems and 14. Scherr SJ, Bennett MT, Loughney M, Canby K (2006) 24. Ostfeld R, Keesing F (2000) Biodiversity and disease Human Well-Being: Synthesis (Island, Washington, DC). Developing Future Ecosystem Service Payments in risk: The case of Lyme disease. Conserv Biol 14:722– 2. Carpenter S, et al. (2006) Millennium ecosystem assess- China: Lessons Learned from International Experience 728. ment: Research needs. Science 314:257–258. (Forest Trends, Washington, DC). 25. Chivian E, Bernstein AS (2004) Embedded in nature: 3. Adams WM, et al. (2004) Biodiversity conservation and 15. Christ C (March 20, 2002) Dispatches: First-hand ac- Human health and biodiversity. Environ Health Per- the eradication of poverty. Science 306:1146–1149. counts from the field. The Grist (Seattle, WA). spect 112:A12–A13. 4. Campagna C, Ferna´ ndez T (2007) A comparative anal- 16. Denman R (2001) Guidelines for Community-Based 26. Bennett EM, Balvanera P (2007) The future of produc- ysis of the vision and mission statements of interna- Ecotourism Development (World Wildlife Fund Inter- tion systems in a globalized world. Front Ecol Environ tional environmental organizations. Environ Values national, Gland, Switzerland). 5:191–198. 16:369–398. 17. Drum A, Moore A (2005) Ecotourism and Development: 27. Heal GM, et al. (2005) Valuing Ecosystem Services: 5. Agrawal A, Redford K (2006) Poverty, Development, A Manual for Conservation Planners and Managers Toward Better Environmental Decision-Making (Natl and Biodiversity Conservation: Shooting in the Dark? (Nature Conservancy, Arlington, VA). Acad Press, Washington, DC). (Wildlife Conservation Society, New York). 18. World Wildlife Fund and Rossing Foundation (2004) 28. Epler Wood M (1998) The Nature Conservancy’s Tour- 6. MacKinnon K, Luz K, Sobrevila C, Wright E (2006) The Living in a Finite Environment (LIFE) Project. End of World Bank and Biodiversity 1988–2005: Mountains to ism and Protected Areas Publication Series (Nature Project Report for Phase II: August 12, 1999-September Coral Reefs (World Bank, Washington, DC). Conservancy, Arlington, VA). 30, 2004. Draft report, October 2004 (U.S. Agency for 7. McCulloch N, Weisbrod J, Timmer CP (2007) Pathways 29. Tilman D, Kareiva P (1997) Spatial Ecology: The Role of International Development, Washington, DC). Out of Poverty During an Economic Crisis: An Empirical Space in Population Dynamics and Interspecific Inter- 19. World Bank (2005) Implementation Completion Re- Assessment of Rural Indonesia (Center for Global De- actions (Princeton Univ Press, Princeton). port (TF-25143 TF-25142 TF-28302 TF-28303) on a velopment, Washington, DC), Working Paper 115. 30. Costanza R, et al. (2007) The Value of New Jersey’s Grant in the Amount of SDR 4.5 Million (US$7.0 8. Kareiva P, Marvier M (2007) Conservation for the peo- Ecosystem Services and Natural Capital (Department of Million Equivalent) to the Burkina Faso and Republic ple. Sci Am 297:50–57. Environmental Protection, Trenton, NJ). of Cote D’Ivoire for a West Africa Pilot Community- 9. Sachs JD, Reid WV (2006) Investments toward sustain- 31. Taylor J, Dyer G, Steward M, Yunez-Naude A, Ardila S Based Natural Resources and Wildlife Management able development. Science 312:1002. (2003) The economics of ecotourism: A Gala´ pagos Is- Project (World Bank, Washington, DC), Report 10. Kaimowitz D, Sheil D (2007) Conserving what and for 32218. lands economy-wide perspective. Econ Dev Cult whom? Why conservation should help meet basic hu- Change 51:977–997. man needs in the tropics. Biotropica 39:567–574. 20. Ro¨ nnba¨ ck P, Troell M, Zetterstro¨ m T, Babu D (2003) 32. Dasgupta P (2007) Economics: A Very Short Introduc- 11. Nordhaus T, Shellenberger M (2007) Breakthrough: Mangrove dependence and socio-economic concerns tion (Oxford Univ Press, Oxford). From the Death of Environmentalism to the Politics of in shrimp hatcheries of Andhra Pradesh, India. Environ 33. Ferraro PJ, Pattanayak SK (2006) Money for nothing? A Possibility (Houghton Mifflin, New York). Conserv 30:344–352. call for empirical evaluation of biodiversity conserva- 12. Krchnak KM (2007) Watershed Valuation as a Tool for 21. Danielsen F, et al. (2005) The Asian tsunami: A protection investments. PLoS Biol 4:482–488. Biodiversity Conservation (Nature Conservancy, Arling- tive role for coastal vegetation. Science 310:643. ton, VA). 22. Ludwig D, Hilborn R, Walters C (1993) Uncertainty, 34. Kremen C, Merenlender AM, Murphy DD (1994) Eco- 13. Sun C, Liqiao C (2006) A Study of Policies and Legisla- resource exploitation, and conservation: Lessons from logical monitoring: A vital need for integrated conser- tion Affecting Payments for Watershed Services in history. Science 260:17–36. vation development programs in the tropics. Conserv China (Research Center of Ecological and Environmen- 23. Tallis HM, Kareiva P (2006) Shaping global environmen- Biol 8:388–397. tal Economics Beijing and International Institute for tal decisions using socio-ecological models. Trends Ecol 35. Central Intelligence Agency (2007) The World Fact- Environment and Development, London). Evol 21:562–568. book (Washington, DC).

9464 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705797105 Tallis et al. Accounting for ecosystem services as a way SPECIAL FEATURE to understand the requirements for sustainable development

Karl-Go¨ ran Ma¨ ler*†, Sara Aniyar‡, and Åsa Jansson*‡

*Stockholm Resilience Center, Stockholm University, SE-106 91, Stockholm, Sweden; and ‡Beijer Institute of Ecological Economics, Royal Swedish Academy of Sciences, Box 50005, SE-104 05 Stockholm, Sweden

Edited by Gretchen A. Daily, Stanford University, Stanford, CA, February 21, 2008 (received for review, September 18, 2007)

Millennium Ecosystem Assessment documented the importance of production account in Standard National Accounts (SNA). We will ecosystem services. It is therefore important that these services are instead focus on the capital account according to refs. 2 and 3, included in our economic accounts (Standard National Accounts), as where it is shown that (i) it is theoretically impossible to develop an long as we believe that these accounts should tell us something about indicator like the green NNP for measuring sustainable develop- our wellbeing. This requires measures of the ecosystem assets and ment because of the changes over time in the prices of capital stocks their accounting prices. This article discusses how the concept of relative to the prices of consumption goods and services, and (ii)a inclusive wealth can be exploited for creating such accounts. wealth indicator is the appropriate measure to use as an indicator of sustainable development. ͉ ͉ wealth prices dynamics Already in the mid-1990s, estimates of the value of the change in capital stocks (called genuine investment) started to be published ince the introduction of the sustainable development concept in as a way to continue the work initiated by Pearce et al. (12). S1987 (1), this term has become a set of words very much used Currently, the World Bank continues publishing annual statistics on in environmental discussions. However, the meaning of the concept genuine savings (13). It has also published a thorough analysis of has remained opaque. Politicians and environmentalists have had wealth accounting and presents many empirical findings in ref. 14. their own interpretation and researchers and scientists have also Unfortunately, ref. 14 does not contain anything on ecosystem had theirs. Sustainable development was defined as ‘‘[. . .] devel- accounting, except for deforestation and damage from climate opment that meets the needs of the present without compromising change; and it does not take into account population changes. In ref. the ability of the future to meet their own needs’’ (1). Although this is not a very precise definition, it gives guidance on how one should 3, the theory is substantially extended by introducing population make it operational. For doing so, we will follow Dasgupta and changes in the framework. Lange’s studies of wealth in southern Ma¨ler (2) and Arrow, Dasgupta, and Ma¨ler (3). They interpret Africa consider man-made real capital, human capital, fisheries, SCIENCE sustainable development as the one where human welfare (or well diamonds, cattle, and water as the main assets (15). She shows SUSTAINABILITY being, we will use these two words interchangeably) is not going clearly how wealth statistics give a much different view of the down over time. Productive capacity of an economy is determined economic development than traditional gross domestic product by its capital stocks. These are man-made, human, and natural (GDP) statistics. In Arrow et al. (16), a group of ecologists and capital. Sustainable development requires that enough of these economists, based on ref. 3 and on the World Bank’s estimates of stocks are left to subsequent generations. The importance of capital genuine savings, investigated whether a selected number of coun- stocks is measured by their accounting prices. The methodological tries were on a sustainable development path. This work gives the apparatus for estimating accounting prices will depend, in general, background for this article. on the nature of the stock’s dynamics and on the nature of the institutional framework; we try to illustrate these issues in this Wealth as an Indicator of Sustainable Development article with a few examples. Let Cs ϭ (c1,s, c2,s,...,cm,s) be a list (or vector) of consumer goods This article gives a brief and consistent basis for accounting for and services in period s. The list must contain not only what we sustainable development focusing on ecosystem services. traditionally regard as consumer goods, but also environmental Brief Literature Overview amenities, public goods, etc. They are also included because all of Literature on so-called green accounting has a tradition going back them contribute to human well being in one way or another. to the early 1960s. It has focused on modifying the concept of a net We add the critical assumption that we have a forecast of the national product so it will reflect the flow of environmental future consumption vectors. Such a forecast obviously must depend damages (mainly from pollution) and depletion (mainly of nonex- on three factors: the present stocks of capital Ki,t (where the current haustible resources, although the depletion of forests is often period is t and i denotes the ith capital stock), a forecast of future included). By correcting for these factors a measure sometimes technologies, and a forecast of the institutions of the economy. called green net national product (NNP) has been calculated. We assume, as is standard in economics, that there is a utility In 1974, Weitzman (4) developed a general theory for the net function U(C1, C2,...,Cm) that describes the production of well national product as a welfare measure. Ma¨ler (5) extended Weitz- being in any given period. Social welfare is defined as the present man’s work to the environmental areas and introduced into the value of the stream of future utilities: theoretical model both the flow of environmental damage and damage to ecosystems. The work of Repetto and colleagues in Indonesia is one of the first attempts for adjusting NNP measures Author contributions: K.-G.M., S.A., and Å.J. designed research, performed research, and (6). Some of the early literature is contained in Ahmad, Serafy, and wrote the paper. Lutz (7), as well as in Lutz (8). These studies resulted later on in two The authors declare no conﬂict of interest. United Nations manuals (9, 10). For an overview of many appli- This article is a PNAS Direct Submission. cations of this approach, see Hecht (11). In this article we will not †To whom correspondence should be addressed. E-mail: [email protected]. follow these approaches, which more or less are based on the © 2008 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0708856105 PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9501–9506 ϱ conclusion is that the social value of investment in physical capital ͑ ͒ U Cs W ϭ ͸ [1] exceeds the value of the investment reported in SNA. l ͑ ϩ ␦͒sϪ1 Changes in human capital are not accounted for in Standard ϭ 1 s t National Income, whereas expenditures on education are accounted as consumption, for historical reasons. However, it is far In Eq. 1, Wt is the social welfare in period t. This representation of social welfare goes back at least to Ramsey (17), but Koopmans from certain that such expenditures reflect the social value of gave the rigorous motivation for it, in three articles (18–20). education. Instead, focus should be on the value of the output from Consumption goods are produced with capital stocks as input in the educational and research system. However, this is a rather the production process. Dasgupta and Ma¨ler (2) showed that there complicated, and because this article is on ecosystem services, we exist accounting prices p , p ,...,p on the capital stocks so that will avoid going into it. Interested readers are welcome to contact 1 2 n the authors for further information.

n Ecosystem Services Ϫ ϭ ͸ ͑ Ϫ ͒ ϩ W ϩ W p K ϩ K v [2] t 1 l i i,t 1 i,t t That ecosystem services are of extremely high importance for ϭ i 1 human well being has been shown convincingly (22), but it is still not Neglecting the last term, Eq. 2 says that the change in social welfare clear how to define units of ecosystems corresponding to capital between two time periods is equal to the sum over all capital stocks assets and how to measure them. of the value of changes in these stocks, when the value is calculated with the accounting prices. Thus, the economy is on a sustainable path if Classification of Ecosystem Services. The Millennium Ecosystem Assessment (MEA) (22) has also provided us with a useful classi- the change in welfare from one period to the next is always fication of ecosystem services: supporting, provisioning, regulating, nonnegative. and cultural. Although this classification is valuable, we have to This result depends on the nature of the accounting prices. modify it slightly to make it suitable for economic analysis. We have They are defined as the present value of the future perturbations reorganized the MEA classification so that provisioning and cul- of consumption because of a marginal change in the stocks today. tural services are merged into a new category, final services, and the They may sometimes coincide with market prices, but most often supporting and regulating services are merged into the category they are different. This is because of taxes, lack of markets for intermediary services. The reason for this is that both the cultural most natural capital, and other market failures. We will come and provisioning services are directly affecting human well being, across this later in our discussion of ecosystem services. whereas the two others are doing that only indirectly—they affect Mathematically, the accounting price on asset i at time t is defined the production of the final goods and services, as intermediary as goods do in national accounting. Another important issue on the classification of services is the ѨU͑C ͒ s distinction between private and public services. Public services are ϱ Ѩ Ki,t characterized by nonrivalry and nonexcludability. Nonrivalry im- p ϭ ͸ [3] i,t ͑ ϩ ␦͒sϪt plies that the use/consumption of a service by one individual does ϭ 1 s t not reduce the availability of it for another individual, for example, It is worth remembering that forecasted future consumption is a climate regulation. If climate is changed for one individual, it will function of the current capital stocks. In the social welfare equation also change for all others experiencing the same climate. Nonex- included above, the first term gives the ‘‘endogenous’’ change in cludability implies that it is impossible to exclude anyone from the use/consumption of the service. Climate is also an example of social welfare, that is, the change that is due to changes in resources nonexcludability. All other services are then called private services. inside the system. The last term, v , reflects changes in social welfare t Note that the distinction between private and public services has outside the studied system. For example, changes in a country’s nothing to do with who is responsible for management of the terms of trade is, for a small country, independent of changes inside production of the service (government or a private firm) or who is the country. The term will also reflect autonomous changes in distributing the service among households. It is the characteristics technology (i.e., technical changes that are independent of capital of the service itself that are important for the classification. Most accumulation in the country). of the final services are probably private. Food produced from Although both terms of trade and technical change can be quite agricultural lands or the oceans or the forests are private services, important for social well being, we will neglect these effects in this as also are biofuel, water, etc. However, most of the intermediary article, that is, we will neglect the ‘‘drift term.’’ For a motivation for services may be public. Climate has already been mentioned, but this, see refs. 2 and 21. However, the main reason for not including also disease and flood regulation are mainly public. Some water these terms is that we want to focus on how to include ecosystem purification services can be public, but others may be private. services in this framework. The distinction between public and private services is important because it makes a major difference for the economic analysis. If Man-Made Physical Capital and Human Capital most of the final services are private—as assumed previously—the The value of changes in ecosystem services is not in itself an value of most of these services is already included in GDP, but not indication of whether society is on a sustainable path or not. It is necessarily in the value of changes of ecosystem assets. Further- only after having integrated all assets to a whole that it is possible more, valuation techniques for the ecosystem services will be very to make judgments on sustainability issues. The two most important different. In general, it will be easier to value private than public assets, beside ecosystems, are man-made physical capital and goods. One reason is that private goods will frequently be sold and human capital. Man-made physical capital is included in conven- bought on a market that gives market values. This cannot happen tional national accounting and the prices used are the market prices. with public goods! The SNA excludes most public goods for that However, these market prices do not necessarily reflect the social reason, with the exception of government expenditures. value of the capital well. One reason is that environmental consequences from an investment are not necessarily reflected in the Ecosystem Size. We need a concept of ‘‘size’’ of the ecosystem and market prices and another is that the market prices are determined we need to know the dynamics of the system. The word size should by the net return after taxes on capital income. Tax revenues may not be interpreted literally. It will in general be a multidimensional have a social value and the market prices do not reflect it. The concept that characterizes the system at each moment of time. In

9502 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0708856105 Ma¨ ler et al. principle, all stocks that affect the growth of any other stock in It should be quite clear that the value of the accounting price producing ecosystem services must be included in our measures of depends on what we expect from the future. In the simplest case, SPECIAL FEATURE size. For example, consider a forest ecosystem. Two natural mea- when the fishery is operated optimally, the accounting price is sures of size is the area covered by the forest or the total biomass ѨC of the forest. However, there are other stocks that will be needed ϭ Ϫ pt qt Ѩ [7] for a proper accounting of the forest, such as the distribution of ht trees with respect to age, size, species, and space. Furthermore, the stocks of nutrients in the soil will influence the growth of the trees. That is, the accounting price for the fish stock is the present Similarly the stock of birds predating on tree pests will affect the marginal profit from harvesting one more unit of fish (in periods growth of biomass, as was shown by Ludwig et al. (23). For with positive harvest). However, the accounting price will be zero simplicity, most of our examples will have only one measure of size. when the fishery is an open-access fishery, which is the case when It is important to understand that ecosystems are production anyone can enter or leave without incurring a cost, as in the units and dynamic systems. Knowledge of the dynamics is Gordon–Schaefer model (27). Simply whenever there is an ex- essential in estimating the appropriate accounting prices. In the pected profit (or rent) from the fishery, fishermen will enter next section, we develop this idea with the help of some examples increasing the total catch, reducing the fish stock, and by that increasing the harvesting cost until the profit (rent) has completely Examples dissipated! The social value of an increase with one unit of the fish stock is in this case zero! This simple example tells us something In this section, we present some conceptual examples of estimating very important. The appropriate accounting prices for an asset accounting prices for ecosystem services, and in some of these depends on our beliefs of the future. This basic conclusion is most examples, we present some quantitative approximations of account- often neglected in valuation exercises. ing prices. However, the main objective here is to show the great As we have seen, our expectation for the future importance of an variety of methods that must be used for estimating the accounting ecosystem depends on our expectations on the institutional devel- prices. opment. Thus, institutional economics is tightly knitted to the estima- Ecosystems differ in their dynamics and this fact has profound tion of accounting prices for ecosystem services. Note that institutions implications for accounting price estimation. In the Standard here are something different from organizations. In brief, institu- National Accounts (SNA) the same problem does not occur at tions are the ‘‘rules of the game’’: legislation, social norms, markets, all or only to a small extent. This is because of the implicit etc. (28). With bad institutions managing ecosystems, the account- assumption that market prices capture those differences, and ing price will be low and even negative in some cases. However, if SNA is a framework to account for market transactions. institutions improve, because of policy reforms, for example, then Because there are no markets for many of the ecosystem services, the accounting prices will rise and accounting will show an increase the accountants must therefore estimate these prices from other in wealth per capita and therefore in human well being! data. Fortunately, a great number of techniques have been devel- Accounting looks at the marginal values, not at total values. It is oped during the past years for doing exactly that. The reader important to emphasize this because in marginal values we start interested in an overview of those techniques is referred to refs. 24 with an accounting price pi, and value a change in the ‘‘size’’ of the SCIENCE and 25. system as Pi(Ki,tϩ1 Ϫ Ki,t), that is, the change in the size is valued at SUSTAINABILITY a constant price. This is the marginal value of the change. Instead, Accounting Prices for a Fishery. Let us assume that a fishery can be the total value of the change can differ from the marginal value, approximated by the Schaefer model (26). This implies that there because the price may change due to changes in the size, thereby is one species only, the dynamics of which is given by the logistic generating changes in consumers’ and producers’ surpluses. This is model: also a source of errors in many empirical studies. In SNA, it is implicitly assumed that the values involved in transactions are x Ϫ ϭ ͩ Ϫ tͪ Ϫ marginal values. For example, the operating surplus is interpreted x ϩ x gx 1 h [4] t 1 t t K t as the return on the capital stock, although parts of it may be monopoly profits, not related to the capital stock. In Eq. 4, xt is the fish stock at time t, g is the stock’s intrinsic growth, K is the carrying capacity of the fishery, and ht is the Accounting Price for the Habitat Service Provided by a Mangrove catch at time t. Here, we assume that the carrying capacity K is Ecosystem to a Shrimp Population. We have modeled mangrove given (in the nest section, the carrying capacity of the fishery will habitat service to fisheries, as nutrient provider, in an extremely be a function of the size of the mangrove stock). The unit used simplified way just to show how an accounting price for that service for the fish stock is the biomass xt. may be calculated; with empirical data, this model would calculate Assuming that the utility function is equal to the net income the accounting price of the service. Therefore, it calculates the of the fishermen, the social welfare function is changes in the well being of fishermen that follows a marginal change in the stock of mangrove biomass. ϱ Ϫ ͑ ͒ The hypothetical case we modeled includes two biological pop- qshs C hs, xs W ϭ ͸ [5] ulations: mangrove forest and shrimp population. These two stocks t ͑1 ϩ ␦͒sϪ1 sϭt grow following a Schaefer model and the growth function of the shrimp stock depends on the mangrove biomass, which implies that In this last equation, qs is the price of the fish in period s, C(hs,xt) the ecosystem’s carrying capacity for shrimps, K, is not a parameter is the cost of catching hs, and ␦ is the social rate of time preference. but a variable. The net growth of the shrimp stock will then be a This social welfare function assumes that the value of catches equals function the shrimp’s biological growth minus the harvest, as usual, the income from sales of the catch minus the cost for effort. The but will also depend on the size of the mangrove biomass. The accounting price of this stock is given by Eq. 6: harvest is modeled as a Cobb–Douglas function of effort and the size of the shrimp stock. The effort is a function of capital and labor Ѩ Ѩ Ѩ Ѩ C hs C xs inputs to the fishery activities. ϱ ͫq Ϫ ͬ Ϫ ѨW s Ѩh Ѩx Ѩx Ѩx The model describes a situation in which the forest is small (4,000 ϭ t ϭ ͸ s t s t pt Ѩ ϩ␦ sϪ1 . [6] ha), which is the size of the mangrove forest that inspired this study, xt (1 ) sϭt ‘‘Los Olivitos’’ in Venezuela. The forest is modeled as effectively

Ma¨ ler et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9503 protected, so that there is no cutting and sales of mangrove timber; distribution in the fishing area). In this case, the accounting price is spatially homogeneous in quality; and is only used by a group of would be lower, showing once again the importance of including local fishermen having no power to change the price of shrimps. institutional considerations as well as technological specification in Fishermen’s access to the resource is regulated by a simple permit the estimation of accounting prices. system limiting the amount of boats in the fishing grounds. The dynamics of the two stocks and their interactions with the Boreal Forests. Boreal forests generate many different services. The fishermen’s economy is simulated with the software Stella for a most important service is probably the supply of timber, but besides period of 100 years. This complex system reaches equilibrium at that, forests are important for the hydrology, for the microclimate, about that time. We find in this way the values of the mangrove and for a large set of diverse services such as berry and mushroom and shrimp stocks when the system is in equilibrium. Assigning picking, recreation, fishing, hunting, reindeer, and cattle feeding, these values from equilibrium to the biological stocks, the model etc. is run again to evaluate the changes in fishermen’s well being Accounting price for timber. The traditional way of looking at timber derived from a marginal change (10 ha) in the mangrove stock. production is, in some ways, similar to the Beverton–Holt model The model calculates the stream of net income and then we in that there is an assumption of an exogenous growth function compare the net present value (NPV) of this stream before and of wooded biomass similar to the von Bertalanffy weight func- after having introduced a marginal change in the stock of mangrove. tion. However, the model has been extended to incorporate The discounting rate is set at 1.5%—approximately consistent with specific management practices such as fertilizing and thinning, the one used in the Stern review (29). The accounting price which influence the growth rate. We will neglect this in this obtained is 200 US dollars per hectare. An accounting price for the discussion. fish population can also be calculated with this model. The limiting factor for the growth of a forest is then the area it can grow on and its fertility. The major economic question is at Accounting Price for Plaice. In the Danish and Swedish fishery of which age to cut the trees. This question was answered in the plaice (a common flatfish mainly occurring in Europe) in Skagerrak mid-nineteenth century, when a German forester, M. Faustmann, and Kattegat, the plaice stock does not follow the Schaefer model, derived the equation by characterizing optimal rotation (for de- where food is the basic limiting factor. It seems that the limiting tailed analysis, see ref. 5): factors are suitable breeding habitats, and the area of these bottoms V͑T͒ Ϫ c varies with time because the bottoms can be covered by algae Ј͑ ͒ ϭ ͓ ͑ ͒ Ϫ ͔ ϩ V t r V T c r rT [10] resulting from eutrophication. Thus, the natural capital stock e Ϫ 1 representing the plaice fishery is this area of soft bottoms. Therefore, the Beverton–Holt model (30) seems better for T is the length of the rotation (the number of years from felling V T explaining the dynamics of the plaice stock (31). In this model, when the trees to the next felling of trees), ( ) is the net value of the stand of trees (where it is assumed that all trees have the same a young fish has been recruited to the stock, its growth will be age), VЈ(T) is the increase in net value due to future growth of completely exogenous and described by the von Bertalanffy weight the biomass, and r is the interest rate. function: V(T) Ϫ c is known as the stumpage value, representing the value w͑t͒ ϭ a͑1 Ϫ beϪct͒3 [8] of one hectare of forest land minus the cost of felling the trees on that hectare. The value includes the present value of future rota- In Eq. 8, w(t) is the average weight of fish at age t and a–c are tions of the forest. Thus, the stumpage value is the maximum value constants. The number of individuals at age t, N(t), is equal to a buyer would be willing to pay for the hectare with the present the recruitment at time 0 minus natural and fishing mortality. stand and is also the minimum price the present owner could accept The total plaice biomass at time t is thus: for selling the land. Thus, the stumpage value is the accounting price. B͑t͒ ϭ w͑t͒N͑t͒ [9] Sweden and many other countries have public statistics on stumpage values and it is therefore easy to include the timber value In contrast to the Schaefer model, the fishing manager must now of forests in the accounts. But this result is based on the implicit decide which generation is to be caught, that is, which mesh size to assumption that the only factor that affects the forest is humans. use in the nets to avoid catching fishes that are too young. Note that, Crepin (33), at the Beijer Institute, studied a forest consisting of once again, this decision will depend on the institutions governing birch trees, pine trees, and moose. Because of feedback between the the fishery. different species, the dynamics of the system now becomes non- The Beverton–Holt model is usually presented as a model with convex, implying that there are several different steady states, and continuous time, which makes it very difficult and sometimes because of this, the system is history-dependent. In such a system impossible to analyze mathematically, but Silva (32), at the the Faustmann analysis is no longer valid. Furthermore, a change Beijer Institute, rewrote the model into discrete time and solved in one of the stocks will imply changes in the production of the other the model by using the software GAMS (General Algebraic stocks and this will not be reflected in the stumpage value. Modeling System, GAMS Development Corporation) However, we do not know yet the empirical importance of this, and The social welfare function was in principle the same as discussed more studies are needed. Because of this and the difficulties of a in the two previous sections, that is, the present value of future net complete model of the boreal forests, we are using the stumpage incomes from the fishery. Assuming optimal management of the value in our Stockholm County Project. fisheries (including the unrealistic assumption that the fishing gears The discussion above has been limited to timber production but, are so precise as to select only one generation of fishes), Silva was as mentioned earlier, forests produce many other services. One is then able to calculate the change in social welfare from a change in pollination by wild pollinators, and we will address that in the next the size of the area of soft bottoms, that is, the accounting price of section. soft bottoms for use by the plaice for reproduction. However, there Another service is the role played by forests for microclimate is no reason to assume optimization. The model could, in principle, control and hydrology. There is no doubt that these are important be solved after various imperfections have been introduced. In fact, services of the Boreal forests. However, our knowledge about in one simulation (not included in the publication), Silva assumed mechanisms behind the generation of these services is very limited, that the selectivity in the gears was lacking so that the catch would and it is too early discuss how to include them in the accounting consist of a mix of different generations (corresponding to the age system. Finally, other provisioning services—berry picking, hunt-

9504 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0708856105 Ma¨ ler et al. ing, recreation, etc.—are much easier to include. For most of these increase in canola production. However, the increases in impacts on services, the land area seems to be one good measure of the size of the size of the natural or seminatural habitat seems to reach a SPECIAL FEATURE the stock. For hunting of moose, for example, one should also saturation point with regard to impacts on the size of the bee include the initial stock of moose. Valuation of these services for populations. If the habitat is smaller than saturation size, a decrease Sweden was done by Lars Hultcrantz (34). of habitat will result in lower bee population and therefore lower harvest of canola. Pollination of Cash Crops by Wild Pollinators. Many types of rapeseed (canola), a major cash crop in North America, are pollinator- Resilience. In all ecosystems, there are feedbacks between different dependent. For certain canola lines the seed weight per plant can components. Some of these feedbacks are positive, which implies increase Ͼ80% with pollination by bumblebees (35). The growing that an initial perturbation of the system will be amplified. Some- demand for urban development has significant impacts on terres- times, the positive feedback becomes active when the system trial ecosystems (36) and on habitat fragmentation (37), which reaches a particular state, and the result is that the system will flip represents a major threat to wild pollinators (38). In this context, it to a different state that may be very different from the initial state. is relevant to assess the pollination ecosystem service. In our The initial state is then a threshold or a bifurcation. If the initial Stockholm County Project we attempt to estimate the accounting state is judged to be better than the state the system would reach price for the pollination regulating service by calculating how the if it would switch, it is of importance to prevent it from reaching the pollination potential of canola can vary because of land use change threshold. The largest perturbation the system can absorb without in an urban development. flipping into a different state is known as resilience (48). It has been shown that the availability of mass flowering crops (as If there is no uncertainty about the dynamics of the system, we canola) has strong positive effects on bumblebee densities, and the can always manage the system to stay within the bounds of strongest correlation between the proportion of mass flowering resilience. However, we never have full information and it is better crops and bumblebee (Bombus terrestris, B. lucorum, B. lapidarius, to regard the system as a stochastic process. That implies that there B. pascuorum) densities was found for landscape sectors with a may be a positive probability that the system will reach the threshold 3,000-m radius (39). The bumblebees also require a 2% area of and flip to the nondesired state. This probability will be lower if the seminatural habitat within the circles surrounding the canola fields resilience increases. Therefore, it is essential to manage the resil- to obtain adequate nesting sites. ience. Furthermore, resilience should be regarded as capital stock By using a geographic information system (GIS; ArcView) and as it provides us with a kind of insurance against reaching a information on area and geographical location of canola fields, we nondesired state. could then place circles (3,000-m radius) around the canola fields As a stock, it has an accounting price and that price is roughly the of the study area (Stockholm County, Sweden) and calculate the change in the expected change in net present value of the expected pollination potential in each circle. By changing the land use future ecosystem services resulting from a marginal change in according to the regional development plan (40) of the study area resilience today (49). Thus, we can estimate the accounting price if (the Stockholm County, Sweden), we can then estimate the change we know the dynamics of the system and the statistics of the system in the pollination potential of the canola. The parameters on which (as a stochastic process). This has been applied in analyzing the SCIENCE our estimates of pollination potential changes are based are the Goulburn-Broken Catchment in Southeast Australia. This system proportion of mass flowering crops within the circle and the contains an area with extensive production of vegetables. The use SUSTAINABILITY minimum requirement of seminatural habitat. of irrigation has increased the salinity of the ground water so that, Because there is also a correlation between bumblebee density if the water table reached two meters under the surface, the saline and harvest index (41), the change in pollination potential can be water would be sucked up to the surface and the whole production linked to crop output. The change in crop output can then, in turn, of vegetables would collapse. Thus, the resilience in this system is be translated into monetary units through a market price method. the distance from the current level of the water table to the level two By using a similar approach, it has been shown (42) that forest- meters below the surface. The water table is affected by two factors: based pollinators increased coffee yield in plantations in Costa Rica precipitation and pumps that try to control the water table. Based by 20% and estimated that during 2000–2003 pollination services on historical data (which may no longer be relevant because of from two forest fragments translated into approximately US$60,000. global warming), the researchers estimated a probability distribu- Furthermore, the scales of operation of ecosystem services are tion for the level of the water table and then estimated the increase of essential consideration when valuing ecosystem services (43). in expected net income from the agriculture if the water table were The scale of operation of solitary wild bees (44) as well as some one meter lower. Unfortunately, there have been no other attempts long-tongued bumblebees (45) is in the realm of hundreds of to include resilience in accounting for ecosystem services. meters, as opposed to several thousand meters, as is the case for Conclusions the included generalist bumblebees; in our example, there are potentially several scales of operation to consider. From the brief presentations in the previous sections, it follows that The distribution of resources at the landscape scale is an impor- accounting for ecosystem services is very case sensitive, because: tant issue to consider in the context of mobile organisms contrib- Y Ecosystem dynamics varies from case to case. uting to ecosystem services (46). Landscape connectivity is needed Y The definition of stocks varies from case to case. for different pollinators and potentially also for relevant pest Y The nature of ecosystem services varies from case to case— control species (47), the freedom of choice to switch between sometimes private services, sometimes public services, and different crops, in the face of, for example, climate change, is sometimes a mixture. enhanced. This freedom allows adaptation to future environmental Y Institutions vary from case to case with implication for valu- and other changes and should also be considered an option value, ation. at least in part ascribed to the pollination service. The dynamics of the interactions between the wild pollinators It seems to be in startling contrast to the creation of the standard needs therefore at least two capital stocks: the size of the canola national accounts, because SNA includes almost only market plantation and the size of the natural and seminatural habitat. The transactions and very few imputations (assessing values to factors bee population seems to adjust very quickly to changes in the canola that are not transacted in markets) are needed, except for the public cultivation; thus there is a very fast positive feedback from increases sector. Industrial or infrastructural projects take time to carry in the canola area to the increase in stock of bees and the following through and they will have a long lifespan and complicated dynam-

Ma¨ ler et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9505 ics. Should not their complicated dynamics inhibit the construction Freeman (50), and Mcconnell and Bockstael (44); (ii) survey of SNA? The answer is no, because of the existence of market techniques: Carson and Hanemann (51); and (iii) dynamic model- prices. This complicated dynamics has been taken into account by ing of ecosystems: Xepapadeas (21), Carpenter et al. (52), Scheffer the managers of the projects and will therefore affect the demand (53), and Christiansen and Walters (54). and supply of goods and services now and in the future, and A strategy for future development of accounting for ecosys- therefore the prices. If the markets work perfectly, the prices will tems must include: correctly represent the social marginal costs and benefits of goods It is not possible to estimate accounting prices for every and services, and the data from transactions will reveal the true ecosystem in the world. Instead a strategy for future develop- values. ment must include: When we deal with ecosystem services, we the analysts and we the accountants must figure out the accounting prices from knowl- Y Selection of major ecosystems to be studied and included in edge of the working of every ecosystem. It is therefore—at least for the accounting framework now—impossible to design a standardized model for building a Y Establishment of the dynamics of the selected systems, as well wealth-based accounting system for ecosystems. We have to de- as possible velop such an accounting system by following a step-by-step path, Y Description of the institutions that control the system now and going from one ecosystem to another. are expected to do so in the future It has often been said that the major problem of including Y Development of the appropriate valuation techniques for each ecosystem services in national accounts is the difficulty of valuing chosen ecosystem service the services themselves. We do not believe that. Progress has been Y Standardization of methods over different ecosystems, as made, for example, on (i) valuation techniques: Bockstael and much as possible.

1. Brundtland G, ed (1987) Our Common Future (Oxford Univ Press, Oxford). 31. Clark C (2005) Mathematical Bioeconomics (Wiley, Hoboken, NJ). 2. Dasgupta P, Ma¨ ler KG (2000) Net national product, wealth, and social-well-being. J 32. Silva S (2007) Topics on the ecological economics of coastal zones. PhD thesis (Swedish Environ Dev Econ 5:69–93. University of Agricultural Sciences), 2007:73. 3. Arrow K, Dasgupta P, Ma¨ ler K (2003) Evaluating projects and assessing sustainable 33. Crepin AS (2002) Tackling the economics of ecosystems. PhD dissertation (Stockholm development. Environ Resour Econ 26:149–187. University), 2002:6. 4. Weitzman M (1976) On the welfare significance of national product in a dynamic 34. Hultkrantz L (1992) National account of timber and forest environmental resources in economy. Q J Econ 90:156–162. Sweden. Environ Resour Econ 2(3):283–305. 5. Ma¨ ler K-G (1991) National accounting and environ resources. Environ Resour Econ 35. Steffan-Dewenter I, Mu¨ nzenberg U, Bu¨ rger C, Thies C, Tscharntke T (2002) Scale- 1:1–15. dependent effects of landscape context on three pollinator guilds. Ecology 83(5):1421– 6. Repetto R, Magrath W, Wells M, Beer C, Rossini F (1989) Wasting Assets: Natural Resources 1432. and the National Income Accounts (World Resources Institute, Washington, DC). 36. McIntyre NE, Knowles-Yanez K, Hope D (2000) Urban ecology as an interdisciplinary 7. Ahmad YJ, El Serafy S, Lutz E (1989) Environmental accounting for sustainable devel- field: Differences in the use of ‘urban’ between the social and natural sciences. Urban opment (World Bank, Washington, DC). Ecosyst 4:5–24. 8. Lutz E, ed (1993). Toward Improved Accounting for the Environment (World Bank, 37. Sala OE, et al. (2000) Global biodiversity scenarios for the year 2100. Science 287:1770– Washington, DC). 1774. 9. United Nations (1993) Integrated Environmental and Economic Accounting. UN Doc. 38. Allen-Wardell GP, et al. (1998) The potential consequences of pollinator declines on ST/ESA/STAT/SER.F/61. the conservation of biodiversity and stability of food crop yields. Conserv Biol 12:8–17. 10. United Nations (2003) Integrated Environmental and Economic Accounting. UN Doc. 39. Westphal C, Steffan-Dewenter I, Tscharntke T (2003) Mass flowering crops enhance ST/ESA/STAT/SER.F/61/Rev. 1. pollinator densities at a landscape scale. Ecol Lett 6:961–965. 11. Hecht J (2005) National Environmental Accounting (Resources for the Future, Wash- 40. Stockholm Office of Regional Planning and Urban Transportation (2001) Regional ington, DC). Development Plan for Stockholm County. Available at: http:// 12. Pearce, et al. (1989) Blueprint for a Green Economy (Earthscan Ltd., London, UK). www.stockholmsregionen2030.nu/english/index.html. Last accessed March 16, 2008. 13. World Bank World Development Indicators. Available at: www.worldbank.org/data. 41. Steffan-Dewenter I (2003) Seed set of male-sterile and male-fertile oilseed rape Last accessed March 16, 2008. (Brassica napus) in relation to pollinator density. Apidologie 34:227–235. 14. World Bank (2006) Where Is the Wealth of Nations? (World Bank, Washington DC). 42. Ricketts T, Daily G, Ehrlich P, Michener C (2004) Economic value of tropical forest to 15. Lange GM (2004) Wealth, natural capital, and sustainable development: Contrasting coffee production. Proc Natl Acad Sci USA 101:12579–12582. examples from Botswana and Namibia. Environ Resour Econ 1:1–28. 43. Hein L, van Koppel K, de Groot RS, van Ierland E (2006) Spatial scales, stakeholders and 16. Arrow KJ, et al. (2004) Are we consuming too much? J Econ Perspect 18(3):147–172. the valuation of ecosystem services. Ecol Econ 57:209–228. 17. Ramsey FP (1928) A mathematical theory of saving. Econ J 38:543–549. 44. Mcconnell K, Bockstael N (2005) Valuing the environment as a factor of production. 18. Koopmans TC (1960) Stationary ordinal utility and impatience. Econometrica Handbook on Environmental Economics, eds Ma¨ ler K-G, Vincent JR (North–Holland, 28(2):287–309. Amsterdam), Vol 2, pp 621–670. 19. Koopmans TC, Diamond PA, Williamson RE (1964) Stationary utility and time perspec- 45. Walther-Hellwig K, Frankl R (2000) Foraging habitats and foraging distances of bum- tive. Econometrica 32:82–100. blebees, Bombus spp. (Hym.,Apidae), in an agricultural landscape. J Appl Ent 124:299– 20. Koopmans TC (1965) On the concept of optimal economic growth. Pontif Acad Sc 306. Scripta Varia 28:225–300. 46. Kremen C, et al. (2007) Pollination and other ecosystem services produced by mobile- 21. Xepapadeas A (2005). Economic growth and the environment. Handbook of Environ- organisms: A conceptual framework for the effects of land-use change. Ecol Lett mental Economics, eds Ma¨ ler K-G, Vincent JR (Elsevier, Amsterdam) Vol 3, pp 1219– 10:299–314. 1272. 47. O¨ berg S, Ekbom B, Bommarco R (2006) Influence of habitat type and surrounding 22. Millennium Ecosystems Assessment (2005) Ecosystems and Human Well-being Synthe- landscape on spider diversity in Swedish agroecosystems. Agric Ecosyst Environ sis (Island Press, Washington DC). 122:211–219. 23. Ludwig D, Jones DB, Holling CS (1978) Qualitative analysis of insect outbreak systems: 48. Holling CS (1973) Resilience and stability of ecosystems. Annu Rev Ecol Syst 4:1– The spruce budworm and forest. J Anim Ecol 47:315–332. 23. 24. Freeman AM (2003) The Measurement of Environmental and Resource Values (Re- 49. Ma¨ ler KG, Li C-Z, Destouni G (2007) Pricing resilience in a dynamic economy- sources for the Future, Washington, DC). environment system: A capital theoretical approach. Beijer Discussion Paper Series no. 25. Ma¨ ler KG, Vincent JR (2005) Handbook of Environmental Economics (Elsevier, Am- 208 (Beijer Institute, Stockholm). sterdam), Vol 2. 50. Bockstael N, Myrick Freeman A (2005) Welfare theory and valuation. Handbook on 26. Schaefer MB (1957) A study of the dynamics of the fishery for yellow in tuna in the Environmental Economics, eds Ma¨ ler K-G, Vincent JR (North–Holland, Amsterdam), Vol Eastern Tropical Pacific Ocean. Inter-Am Trop Tuna Comm Bull 2:246–268. 2, pp 517–570. 27. Gordon HS (1954) The economic theory of a common-property resource: The fishery. 51. Carson R, Hanemann M (2005) Contingent valuation. Handbook on Environmental J Pol Econ 62:124–142. Economics, eds Ma¨ ler K-G, Vincent JR (North–Holland, Amsterdam), Vol 2, pp 821–936. 28. North D, Thomas RP (1973) The Rise of the Western World: A New Economic History 52. Carpenter SR, Ludwig D, Brock WA (1999) Management of eutrophication for lakes (Cambridge Univ Press, Cambridge). subject to potentially irreversible change. Ecol Appl 9:751–771. 29. Stern N (2007) The Stern Review: The Economics of Climate Change (Cambridge Univ 53. Scheffer M (1998) The Ecology of Shallow Lakes, Population and Community Biology Press, Cambridge). Series 22 (Chapman & Hall, London). 30. Beverton RJH, Holt SJ (1957) Fisheries Investigations Series 2(19) (Ministry of Agricul- 54. Christiansen V, Walters C (2004) Ecopath with Ecosim: Methods, capabilities and ture, Fisheries and Food, London). limitations. Ecol Model 172:109–139.

9506 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0708856105 Ma¨ ler et al. Efficiency of incentives to jointly increase carbon SPECIAL FEATURE sequestration and species conservation on a landscape

Erik Nelson*†, Stephen Polasky‡, David J. Lewis§, Andrew J. Plantinga¶, Eric Lonsdorfʈ, Denis White**, David Bael‡, and Joshua J. Lawler††

*Department of Biology and Natural Capital Project, Woods Institute for the Environment, Stanford University, 371 Serra Mall, Stanford, CA 94305-5020; ‡Department of Applied Economics, University of Minnesota, 1994 Buford Avenue, St. Paul, MN 55108; §Department of Agricultural and Applied Economics, University of Wisconsin, 427 Lorch Street, Madison, WI 53706; ¶Department of Agricultural and Resource Economics, Oregon State University, Corvallis, OR 97331-3601; ʈDavee Center for Epidemiology and Endocrinology, Conservation and Science, Lincoln Park Zoo, Chicago, IL 60614; **U.S. Environmental Protection Agency, 200 Southwest 35th Street, Corvallis, OR 97333; and ††College of Forest Resources, University of Washington, Box 352100, Seattle, WA 98195

Edited by Gretchen C. Daily, Stanford University, Stanford, CA, and approved December 19, 2007 (received for review July 2, 2007)

We develop an integrated model to predict private land-use the incentive to manage land to provide ecosystem service and decisions in response to policy incentives designed to increase the biodiveristy conservation benefits because many of the benefits provision of carbon sequestration and species conservation across produced on their land accrue to others (i.e., the benefits are heterogeneous landscapes. Using data from the Willamette Basin, public goods). For example, carbon sequestration provides a Oregon, we compare the provision of carbon sequestration and global benefit by reducing atmospheric CO2 levels, while water species conservation under ﬁve simple policies that offer payments purification services help those downstream. In addition, the for conservation. We evaluate policy performance compared with provision of ecosystem services may depend on the spatial the maximum feasible combinations of carbon sequestration and pattern of land use, requiring coordinated effort across multiple species conservation on the landscape for various conservation landowners (6). Incentives for provision of ecosystem services budgets. None of the conservation payment policies produce and biodiversity conservation can be supplied by government increases in carbon sequestration and species conservation that programs such as the U.S. Department of Agriculture’s Conser- approach the maximum potential gains on the landscape. Our vation Reserve Program (CRP), which pays farmers to take land results show that policies aimed at increasing the provision of out of production and establish perennial plant cover to improve carbon sequestration do not necessarily increase species conser- habitat, reduce erosion, and supply other environmental bene- vation and that highly targeted policies do not necessarily do as fits. Currently the CRP pays farmers Ͼ$1.8 billion annually to well as more general policies. enroll 36 million acres in the program (7). Land trusts and conservation organizations are another important avenue for SCIENCE conservation payments ͉ ecosystem services ͉ landscape modeling ͉ providing ecosystem services and biodiversity conservation by SUSTAINABILITY private landowners ͉ land-use change securing conservation easements or buying land outright. Here we develop an integrated model that (i) predicts landowner cosystems provide a wide array of goods and services of value decisions as a function of existing market conditions and incentive- Eto people, as well as support for biodiversity. Human activ- based conservation payments and (ii) predicts the impact of land- ities, particularly regarding land use and land management, have use changes on ecosystem services and biodiversity conservation. altered ecosystems in fundamental ways with broad-ranging Our approach integrates the effect of policy on land-use decisions consequences for human well-being and the survival of other and the resulting consequences for the joint provision of ecosystem species. However, the effects of land-use and land-management services and biodiversity conservation across a landscape. decisions on ecosystem services and biodiversity often are not We use National Resources Inventory (8) data over several incorporated into decision-making, resulting in outcomes that time periods to estimate a model of land-use change on private land. Landowners choose to allocate land among various uses reduce the provision of ecosystem services and biodiversity based on current land use, predicted economic net returns to conservation in ways that harm both human well-being and each land use, land quality, and specific characteristics of the biodiversity. landowner (e.g., preferences or skill level). The empirical model One important step in improving decision-making is to pro- yields transition probabilities expressed as functions of net vide information about the effects of human decisions on returns and starting land use. This approach has the advantage ecosystem services and biodiversity. Although models exist that of being based on observed land-use decisions, and, because project the effects of land-use and land-management decisions transition probabilities depend on net returns, the effect of (hereafter referred to simply as land-use decisions) on individual incentive-based policies on landowner decisions can be simu- ecosystem services or specific taxonomic groups (e.g., refs. 1–3), lated by modifying net returns. Distributions of future land-use few landscape-scale assessments of multiple ecosystem services patterns in the Willamette Basin under alternative land conser- and biodiversity have been conducted (exceptions include refs. 4 and 5). Managing landscapes to deliver ecosystem services and species conservation requires the integration of spatially explicit Author contributions: E.N., S.P., D.J.L., and A.J.P. designed research; E.N., S.P., D.J.L., A.J.P., data and models from ecology, economics, and other disciplines. and D.B. performed research; E.N., S.P., D.J.L., A.J.P., E.L., D.W., D.B., and J.J.L. analyzed Information about ecosystem services and biodiversity con- data; and E.N., S.P., D.J.L., A.J.P., E.L., D.W., D.B., and J.J.L. wrote the paper. servation, although necessary, is not sufficient to generate The authors declare no conﬂict of interest. socially beneficial landscape-level management. Providing in- This article is a PNAS Direct Submission. centives to make decisions that reflect the value of ecosystem †To whom correspondence should be addressed. E-mail: [email protected]. services and biodiversity conservation is also important. Land- This article contains supporting information online at www.pnas.org/cgi/content/full/ use decisions are typically made by a large number of private 0706178105/DC1. landowners and public entities. Private landowners typically lack © 2008 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706178105 PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9471–9476 vation payment schemes, including a baseline with no conservation Basin [S.P., E.N., J. Camm, B. Csuti, P. Fackler, E.L., C. payments, are generated by running repeated simulations, with Montgomery, D.W., J. Arthur, B. Garber-Yonts, R. Haight, J. each simulation generating a land-use pattern consistent with the Kagan, A. Starfield, and C. Tobalske, unpublished data (avail- transition probabilities and the given conservation payment able from the author upon request) and see SI Fig. 5]. scenario. This approach explicitly accounts for the uncertainty We analyze the degree to which simple contracts or payment about individual landowner decisions. Further details on our schemes to convert land to conserved land use can be expected land-use change model are presented in Materials and Methods. to generate an efficient land-use outcome. We compare four The policy-generated and baseline land-use patterns are then targeted policies, under which only private landowners whose used as inputs into models that predict the provision of ecosys- lands meet certain criteria are eligible, with an untargeted policy tem services and biodiversity conservation. Modeling these in which all private landowners are eligible to participate. The factors often involves the use of sophisticated models that five conservation policies defined for privately owned parcels we require extensive data inputs, which makes them inapplicable in analyze are (i) All: any parcel is eligible for a conservation some situations. An important question we address is the degree contract; (ii) Rare Habitat: only parcels that could convert to to which policy recommendations are robust to changes in data particular types of rare natural habitat are eligible; (iii) Carbon: input requirements and model sophistication. We use models only parcels that could convert to conserved forest are eligible; with different levels of complexity and detail to explore this (iv) Riparian: only parcels with significant stream density are question. The first level of models uses readily available data and eligible; and (v) Species Conservation: only parcels identified as simple relationships to link land use to ecosystem services and important for vertebrate species conservation in the Basin [S.P., biodiversity. The second level requires more detailed, site- E.N., J. Camm, B. Csuti, P. Fackler, E.L., C. Montgomery, D.W., specific information and incorporates more complex relationships. J. Arthur, B. Garber-Yonts, R. Haight, J. Kagan, A. Starfield, To demonstrate the potential tradeoffs and synergies between and C. Tobalske, unpublished data (available from the author ecosystem services and biodiversity conservation on a landscape, upon request)] are eligible (see SI Fig. 6). The targeted policies we focus on carbon sequestration and terrestrial vertebrate mimic common targeting schemes used in incentive-based pro- species conservation. Carbon sequestration is an important grams of the U.S. Department of Agriculture, such as the CRP. ecosystem service for which data and models are readily avail- We compare the performance of targeted and untargeted land- able and for which markets for payment are emerging. Terres- use conservation payment schemes relative to the baseline trial vertebrate species conservation is an important objective of land-use patterns assuming no land-use conservation policy. We policy and is the main objective of many conservation organi- also compare the performance of targeted and untargeted zations. Details on the models we used to predict carbon incentive payment schemes relative to the efficiency frontier for sequestration and species conservation are presented in Mate- various levels of land-use conservation program budgets. rials and Methods. We use these models to compare efficient land-use patterns Results with those likely to emerge as a result of decisions by landowners Tradeoffs were found between carbon sequestration and species under various land-use conservation policy scenarios and con- conservation on efficiency frontiers in the Willamette Basin, as servation program budgets. Efficient outcomes occur when it is indicated by the negative slope of each efficiency frontier (Fig. not possible to increase one desired objective without simulta- 1, and see SI Tables 1–4). An efficiency frontier shows the neously decreasing another desired objective. In other words, all maximum amount of a given objective that can be attained on the potential ‘‘win–win’’ possibilities have been realized. We sum- landscape for a fixed level of the other objective and a given marize the set of efficient outcomes, with efficiency frontiers conservation program budget. Starting from an efficient land- that show the maximum feasible combinations of multiple use pattern, an increase in the production of one objective outputs that can be generated by the landscape. requires a decrease in the level of the other. Both species In principle, an efficient outcome can be implemented by conservation and carbon sequestration levels on the frontiers using a policy that aligns private landowner and social incentives. increase as the conservation budget increases because more In practice, however, several obstacles prevent the implemen- landowners can participate in the program. The shifting-out of tation of such a policy and a realization of efficient outcomes. the efficiency frontiers with increased conservation budget also First, because of unobservable landowner characteristics and reveals a tradeoff between species conservation/carbon seques- preferences, a government agency or conservation organization tration and commodity production. In other words, greater levels cannot predict with certainty which landowners will voluntarily of incentive payments corresponding to higher opportunity cost accept a contract or payment scheme. Therefore, it is not (e.g., value of foregone production) are required in order to shift possible to predict the policy-induced spatial pattern of the land to conservation use. Results obtained with the simple landscape with certainty (9). Second, because of landscape carbon and biodiversity models (Fig. 1A) show a greater tradeoff heterogeneity and spatial interactions, using incentive-based between carbon storage and species conservation. With the policies to generate an efficient land-use pattern may involve a complex biophysical models, we find significant portions of the complex set of various payments or contracts that go well beyond efficiency frontiers where one objective can be increased without the administrative ability of an agency or organization to im- significantly lowering the other (Fig. 1B). plement them. In addition, principles of fairness may limit In general, species conservation is maximized when landown- differential treatment of landowners, even though landowners ers who accept a conservation payment restore natural habitats contribute differently to the provision of desired outcomes. that are relatively rare on the current landscape, including oak We apply this suite of land-use change and landscape-level savanna, prairie, and emergent marsh. Carbon sequestration, on biophysical models by using data from the Willamette Basin, the other hand, is maximized when landowners who accept Oregon [see supporting information (SI) Fig. 4]. The Basin conservation payments restore conserved forests, including old consists of primarily forested slopes of the Coast and Cascade growth, mixed, and riparian forest. Although maximizing forest Mountain Ranges surrounding a relatively flat valley floor cover benefits some species (e.g., the spotted owl), it provides dominated by agriculture and urban centers. The valley floor and little benefit for the majority of the 37 species analyzed. The low foothills are largely privately owned, with public land different land-use patterns corresponding to different points on dominating the higher elevations. The terrestrial species con- the efficiency frontier are illustrated for the policy under which servation model uses 37 terrestrial vertebrate species determined all landowners are eligible for conservation payments, using the to be at risk or particularly sensitive to land-use change in the simple biophysical models (Fig. 2, and see SI Tables 5 and 6). On

9472 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706178105 Nelson et al. SPECIAL FEATURE

Fig. 1. Comparison of policy scenarios using the simple and complex biophysical models. (A) Increases in species conservation and carbon sequestration induced by each policy simulation and across each efficiency frontier, as compared with the base-case landscape with no conservation policy (the origin of each graph), using the simple species conservation and carbon sequestration models. Policy and efficiency results are shown for conservation program budget levels of $1 million, $5 million, and $10 million. Each point symbol shows the average response for the policy simulations across all 500 simulations. The leftmost graph shows the results when all private landowners are eligible for payments (the All policy scenario). The remaining graphs show results of targeted policies (see SI Table 1). The efficiency frontiers are shown as curved lines and assume that all private parcels eligible for a payment under the All policy scenario are eligible for a conservation payment when finding the frontier. The efficiency frontiers for the three budget levels are replicated on each policy graph (see SI Table 2). The baseline land-use pattern used to determine the origin of each policy graph had the mean service bundle level across all 500 simulated baseline land-use patterns. A 0.005 change on the x axis is equivalent to a 0.5% average change in the relative amount of a species’ habitat provided by a land-use pattern. A 0.005 change on the y axis is equivalent to 2.1 million metric tons of carbon. (B) We replicate the analyses shown in A by using the complex species conservation and carbon sequestration models (see SI Tables 3 and 4). The efficiency frontiers in B represent only portions of complete frontiers because of the difficulty in finding points on the frontier with the nonlinear species conservation model. A 0.02 increment on the x axis is equivalent to a 2% change in aggregate species persistence. A 0.02 increment on the y axis is equivalent to 1.8 million metric tons of carbon sequestration. the mapped portion of the valley floor, the land-use pattern that patterns have less agricultural land than the average baseline maximizes species conservation contains more oak savanna, land-use pattern on which no payments are offered for private prairie, marsh, and scrub–shrub lands (shown in yellow in Fig. 2), landowner conservation. SCIENCE whereas the landscape that maximizes carbon storage contains The land-use patterns induced by the five policy scenarios SUSTAINABILITY more managed forests (shown in blue in Fig. 2). Both land-use considered here produce only small increases in carbon sequestration and species conservation relative to the baseline landscape with no conservation policy (indicated by the circles in Fig. 1), as compared with the feasible increases shown by the efficiency frontiers (Fig. 1, and see SI Tables 1–4). These inefficiencies are evident for both the simple and complex biophysical models and across all three conservation program budget levels. The five policy scenarios generate very different mixes of outcomes relative to the baseline. Both the Rare Habitat and Species Conservation policies result in increases in species conservation but do not increase carbon sequestration, with actual reductions occurring under the Rare Habitat policy. By comparison, the Carbon policy, under which conservation contracts are only offered to landowners who can significantly increase the forest coverage on their land, results in increases in carbon sequestration but little increase in species conservation (simple models) or a decrease in species conservation (complex models). The Riparian policy provides little benefit in terms of either carbon sequestration or species conservation. Somewhat sur- prisingly, the untargeted policy (the All policy) increases species conservation as much or more than any of the targeted policies. There is considerable variation in the degree to which species Fig. 2. Land-use patterns on the $10 million efficiency frontier (simple conservation and carbon sequestration can be increased by a models). The two maps depict land-use patterns generated at two points policy for a given program budget level (Fig. 3, and see SI Tables along the $10 million budget efficiency frontier, using the simple carbon 7–10). The variation in results for a policy–budget combination, sequestration and species conservation models under the All policy scenario. using both simple and more complex models, is largely driven by The two maps at the top of the figure represent the entire study area at points the great variation in potential land-use patterns. Such variation 1 and 2 along the frontier. The blocks of numbers beside each Basin-wide map indicate the number of parcels in each land-use category on that map (see SI is indicative of the uncertainty with which land-use pattern can Tables 5 and 6). The corresponding maps directly below the Basin-wide maps be predicted using empirical models of voluntary decisions. show the land-use patterns in the area highlighted in gray on the small inset There is also greater variation in the species conservation results map. using the complex species conservation model where species

Nelson et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9473 Fig. 3. Variability in results. Results are shown for three simulations that represent the 5th percentile, mean, and 95th percentile in terms of species conservation and carbon sequestration values among the 500 simulations for the All policy scenario with a $10 million program budget (see SI Tables 7–10), using the simple (A) and complex (B) models. The origins used in the graphs are the same as those used in the graphs in Fig. 1 A and B, respectively. Each policy result dot, and its accompanying efﬁciency frontier, are functions of the same baseline land-use pattern. The baseline land-use patterns that form the basis of the 5th and 95th percentile results using the simple biophysical models are not the same patterns that form the basis of the 5th and 95th percentile results using the complex biophysical models. The baseline land-use pattern that forms the basis of the mean results in both models is the same pattern that was used to create the efﬁciency frontiers shown in Fig. 1. persistence probabilities increase nonlinearly with habitat and potential gain in species conservation. The remaining ineffi- spatial configuration matters. Both the nonlinear change and the ciency of incentive-based policies is because voluntary enroll- importance of spatial pattern make the species conservation ment does not ensure that the specific set of landowners neces- score much more sensitive to which landowners accept conser- sary to generate an efficient landscape pattern will enroll. vation payments and results in greater variation across land-use Nevertheless, many agencies view the budget as a measure of pattern permutations than is the case with the simple model that cost, and overpayments to landowners can have efficiency depends only on the amount of habitat. implications if there are costs to raising and administering program funds. Discussion Our analyses show that policies aimed at increasing the We have combined a model that predicts land-use decisions by provision of an ecosystem service can, but do not necessarily, multiple private landowners with models that predict the con- increase the provision of species conservation (and vice versa). sequences of these decisions on the provision of an ecosystem We found some cases in which a policy directed toward one goal service and on biodiversity conservation. Although more so- actually reduced the ability to attain the other goal. For example, phisticated models exist for the component parts [e.g., the the Rare Habitat scenario results in lower carbon sequestration CENTURY model for carbon sequestration (10, 11)], these than the case with no policy at all. If programs that pay for components have not been linked in a systematic fashion. We ecosystem services are not designed carefully, they may yield show that policies that pay private landowners to restore land to minimal gains in services of interest and may well result in harm natural cover increase the provision of an ecosystem service and to other services or biodiversity conservation. The consideration biodiversity conservation; however, these increases are less than of a greater range of ecosystem services in a conservation policy what is feasible for a given budget, as shown by the efficiency will only magnify the degree of potential tradeoffs. Another frontier. It is possible that more sophisticated policies, such as challenge to designing an efficient targeting scheme is the fact auctions in which landowners submit bids of their willingness- that the optimal conservation payment will vary across land- to-accept (WTA) land use change, may be able to close the gap owners and may be a function of the land-use decisions of each vis-a`-vis the efficiency frontier. landowner’s neighbors (as it is with the complex species conser- For a given policy scenario simulation, all landowners accept- vation model). Implementing a complicated incentive scheme ing conservation contracts receive the same payment even that accounts for such interactions would not be trivial. Finally, though their WTA varies significantly (see SI Fig. 7). The if a targeted policy does not accurately capture desired environ- inability to price-discriminate is one of the sources of the gap mental benefits, then results with targeting may be worse than between the policy-induced land-use patterns and those on the without targeting. Targeting reduces the number of eligible efficiency frontier. Overpayments to landowners, however, are landowners and, on average, raises the cost of a conservation simply transfers from the program administrator to the land- contract. Only through accurate targeting of conservation con- owners. The true social cost of the conservation policies is equal tracts to landowners providing high conservation benefits will to the sum of the WTA of landowners agreeing to conservation the gain of targeting be sufficient to outweigh the increase in contracts. Evaluating the efficiency of the incentive payments by contract cost. measuring social costs rather than budget costs (including trans- Fundamental differences in model assumptions and output fers) leads to a more optimistic interpretation of the efficiency metrics in the simple and complex species conservation models of our simple conservation policies. For example, under the make it difficult to directly compare their results. The simple Species Conservation policy, using the simple species conserva- species conservation model provides an estimate of the proportion model with a $10 million budget, the conservation policy tion of the landscape that provides habitat for the 37 terrestrial achieves 57% of the potential increase in species conservation. vertebrate species included in the analysis. This measure does However, using the sum of WTA as the measure of cost, the cost not take into account species’ area requirements for breeding of the contracts would be approximately $5 million (see SI Fig. and feeding activities or the spatial pattern of the landscape. The 8). Compared with what could be achieved at this program complex species conservation model estimates the probability budget level, the conservation policy achieves 81% of the that each of the 37 species would maintain a viable population

9474 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706178105 Nelson et al. in the Willamette Basin, taking into account both the species’ owner skills), we model the probability that a land parcel will change from one area requirements for breeding and feeding activities and the use to another conditional on observed net returns and model parameters. SPECIAL FEATURE spatial pattern of the landscape. A species that requires large These parameters are estimated by using a repeated sample of plot-level contiguous blocks of habitat in a landscape that contains many land-use decisions from the National Resources Inventory. The econometric analysis focuses on private land in the conterminous United States and on six small fragments of habitat may score well in the simple model, major land uses (crops, pasture, forest, urban, range, and land enrolled in the while the complex model produces a low score. In addition, it is CRP). The estimation yields land-use transition probabilities for all possible possible for relatively small land-use changes that result in starting and ending uses, conditioned on net returns and plot-level soil increases in key habitat to have a large impact on species characteristics. For this analysis, we extract the values of these conditioning conservation in the complex model that takes habitat configu- variables for the Willamette Basin to produce a set of transition probabilities ration into account. for each parcel. In this analysis, we investigate feasible combinations of an ecosystem service and species conservation that could be pro- Policy Simulation. Starting from an initial land-use pattern (circa 1990), we use duced on the Willamette Basin (‘‘supply side’’ analysis). This the econometrically estimated transition probabilities to predict changes in analysis shows tradeoffs between desired ends but does not the landscape over a 50-year time period (ref. 8, and see SI Text: Policy Simulation). Because changes on each land parcel are probabilistic, we simu- provide information about how to choose among these ends. late 500 baseline land-use patterns to generate a representative distribution Combining this analysis with non-market valuation studies of the range of potential landscape outcomes assuming no conservation (‘‘demand-side’’ analysis) would allow comparison of the value payments to private landowners. of different bundles of ecosystem services and species conser- We then use the estimated transition probabilities to generate a probabil- vation to maximize the value of outcomes generated by the ity distribution of WTA for each parcel of private land, where WTA is the landscape. It is relatively straightforward to value some ecosys- minimum amount of annual payment necessary for the landowner to agree to tem services, such as those that produce marketable commod- take their land out of its current production land use and switch to conser- ities (e.g., crops or timber). Studies of the potential losses vation. For each of the 500 baseline simulations, we generate a WTA value for each privately owned parcel by randomly drawing from the parcel’s WTA associated with climate change and rapidly emerging carbon probability distribution. Under any policy scenario simulation, an eligible markets offer some hope for accurately assessing the value of private landowner will accept a conservation contract if the contract’s annual carbon sequestration. At the other extreme, valuing species payment is equal to or exceeds the landowner’s WTA. We assumed that land existence or aesthetics is fraught with difficulty (e.g., ref. 12). placed into conservation is restored to the land’s pre-Euroamerican vegeta- Even without complete valuation of all environmental goods, tion cover for that parcel. A parcel that does not contract for conservation is however, presenting decision-makers with tradeoffs among ends assumed to remain in its baseline land use (see SI Text: Policy Simulation and provides information regarding the opportunity cost of achiev- SI Table 11). Because we have 500 baseline land-use patterns and 500 accom- ing particular environmental goals and, hence, a measure of the panying vectors of private landowner WTA values, we can generate 500 minimum value that those environmental goods must possess for policy-induced land-use patterns for each policy scenario and conservation program budget level. a particular policy to achieve a net gain in social welfare. We evaluate each of the five policy scenarios at three different annual Finally, predictions and policy advice generated by analyses budget levels: $1 million, $5 million, and $10 million. For each scenario and are only as good as the models and data on which they are based. budget combination, we solve for the annual per-acre contract price that SCIENCE In many ways, we are still at an early stage in the analysis of the would enroll just enough landowners to exhaust the budget (see SI Table 12). provision of ecosystem services and biodiversity conservation SUSTAINABILITY from landscapes, and much remains to be learned. Questions Carbon Sequestration Models. The simple version of the carbon model uses as remain regarding how much detail and complexity are required its biophysical basis Intergovernmental Panel on Climate Change default in order to inform policy. In our application, it appears that terrestrial carbon storage values as a function of land use (18). Storage values greater model complexity and data did not greatly change as a function of land use are specific to the Basin’s eco-region and climate modeling results or policy advice. Expanding the analysis to region. Carbon stored on the land is the sum of carbon stored in five pools: soil, below-ground biomass, above-ground biomass, deadwood and litter, and include more services risks going beyond what teams of analysts harvested wood products. The difference between the carbon stored on the know and understand. Linking many components into a com- initial and final land-use patterns represents the carbon sequestration that prehensive analysis also risks magnifying any errors present in occurs on the landscape over a 50-year time horizon. Because storage values the component analyses. The joint nature of the provision of are uncertain, we construct mean sequestration values to score land-use ecosystem services on landscapes, however, makes the push patterns (see SI Text: Simple Carbon Sequestration Model and SI Tables 13–18). toward comprehensive analysis of vital importance. Further In the more sophisticated version of the carbon model, we use detailed analysis, combined with monitoring and evaluation of the con- information specific to the Willamette Basin, including information about the sequences of past decisions, should offer more confidence in distribution of tree species, tree ages, and canopy densities across the landscape, along with characteristics of each parcel and tree stand allometric predictions and policy advice generated by landscape-level anal- tables (19), to predict annual carbon sequestration on each parcel over the yses such as the ones undertaken here. 50-year modeling time horizon. Annual carbon sequestration rates are a function of the mix of softwood and hardwood trees, tree stand age, canopy Materials and Methods density, soil type, elevation, initial and final land use on the parcel, and the The Parcel Map. The Willamette Basin of Oregon was divided into 10,372 timing of land-use transition (20). Annual changes in carbon sequestration are distinct parcels on the basis of a 1990 land-cover map (13). We deleted discounted at a 5% rate. Because the exact softwood and hardwood tree mix, water-covered parcels and parcels inside urban growth boundaries, leaving canopy density, carbon stock in the soil and other minor carbon pools, and 8,176 parcels. We obtained information on each of these parcels, including the timing of land-use transition on each parcel are unknown, we construct mean 1990 land-cover type, current ownership and conservation status (14), pre- sequestration values to score land-use patterns. The model’s final output is the Euroamerican vegetation cover (15), and other physical characteristics, such as amount of carbon sequestered across the whole landscape over the 50-year soil type (16) and location of perennial streams (ref. 17, and see SI Text: The modeling time horizon (see SI Text: Complex Carbon Sequestration Model and Parcel Map). SI Tables 19–24). Final landscape-level output from both carbon models is normalized by Econometric Model of Land-Use Change. An econometric land-use model (1) is dividing predicted 50-year carbon sequestration levels by the maximum po- used to quantify the relationship between private land-use decisions, the tential sequestration that could occur on the landscape over the same time economic net returns to alternative uses, and parcel-level characteristics (see period. SI Text: Econometric Model of Land-Use Change). Landowners are assumed to choose land uses to maximize the present discounted value of the stream of Species Conservation Models. The species conservation model translates land- expected returns to the land net of conversion costs. Because, in practice, use patterns into habitat maps for the 37 terrestrial vertebrate species mod- land-use decisions can also be influenced by unobserved factors (e.g., land- eled in the analysis. The simple species conservation model is solely a function

Nelson et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9475 of aggregate species habitat for breeding and feeding activities (see SI Text: scenario. The only change possible is from the baseline land use to the Simple Species Conservation Model, SI Table 25, and SI Fig. 5). For each conservation land use. For the efficiency frontier analysis, we assume that the generated land-use pattern, the simple model calculates the following ratio annual conservation payment paid to a landowner is equal the landowner’s for each species: the amount of breeding and feeding habitat area provided WTA. for the species vs. the maximum amount of breeding and feeding habitat area By varying the carbon sequestration goal, we can define the efficiency that could be provided if the landscape was completely managed for the frontier for a given program budget level [S.P., E.N., J. Camm, B. Csuti, P. benefit of the species. The overall species conservation score for a land-use Fackler, E.L., C. Montgomery, D.W., J. Arthur, B. Garber-Yonts, R. Haight, J. pattern is the average of all 37 species ratios. Time is not a factor in the simple Kagan, A. Starfield, and C. Tobalske, unpublished data (available from the species conservation model. author upon request)]. We then solve for the efficiency frontier for three In contrast, the complex model is spatially explicit and predicts species different budget levels. By increasing the budget, which represents greater persistence as a function of the amount and spatial pattern of breeding and opportunity cost in terms of foregone marketed commodity production, feeding habitat, as well as species’ breeding and feeding area requirements greater levels of species conservation and carbon sequestration can be ob- and dispersal abilities. The complex species conservation model determines tained (see SI Text: Efficiency Frontiers). the expected number of species that would persist on the landscape for an The efficiency frontiers in Fig. 1 indicate maximum joint production on the indefinite time period, given a land-use pattern (ref. 21, and see SI Text: landscape assuming a particular baseline land-use pattern; we did not find the Complex Species Conservation Model and SI Tables 25–27). The complex set of efficiency frontiers associated with each baseline land-use pattern and species conservation model results are normalized by dividing a land-use then take the average across all frontiers. pattern’s predicted species conservation amount by 37. ACKNOWLEDGMENTS. D.J.L., A.J.P., S.P., J.L.L., and D.W. gratefully acknowl- Efficiency Frontiers. An efficiency frontier plots combinations of outcomes edge conversations with the late Raymond O’Connor that provided the inspi- such that one outcome cannot be improved without reducing another out- ration for several components of this study. The authors thank Gardner Brown, Jeff Camm, Paul Fackler, Gretchen Daily, Peter Karieva, Volker Rade- come. We solve for a point on an efficiency frontier by finding the land-use loff, Taylor Ricketts, Heather Tallis, members of the Natural Capital Project, pattern that maximizes species conservation subject to the land-use pattern and participants at the 2007 Association of Environmental and Resource meeting a specified carbon sequestration value and a given policy budget. Economists Workshop for helpful comments. D.J.L. was supported by the Parcels eligible for a change when finding the efficiency frontier are the same Wisconsin Alumni Research Foundation, and A.J.P. was supported by the U.S. parcels that are eligible for a conservation contract under the All policy Department of Agriculture Forest Service Pacific Northwest Research Station.

1. Lubowski RN, Plantinga AJ, Stavins RN (2006) Land-use change and carbon sinks: 12. Bishop RC, Welsh MP (1992) Existence values in benefit-cost analysis and damage Econometric estimation of the carbon sequestration supply function. J Environ Econ assessment. Land Econ 68:405–417. Manage 51:135–152. 13. Oregon Natural Heritage Information Center (2000) Integrated Willamette Basin 2. Spies TA, et al. (2007) Potential effects of forest policies on terrestrial biodiversity in a Landcover GIS Coverage and Related Metadata (Oregon Natural Heritage Program, multi-ownership province. Ecol Appl 17:48–65. Portland, OR). 3. White D, et al. (1997) Assessing risks to biodiversity from future landscape change. 14. Oregon Natural Heritage Information Center (2005) Land Ownership and Land Stew- Conserv Biol 11:349–360. ardship for Oregon GIS Coverage and Related Metadata (Oregon Natural Heritage 4. Chan KMA, Shaw MR, Cameron DR, Underwood EC, Daily GC (2006) Conservation Program, Portland, OR). planning for ecosystem services. PLoS Biol 4:2138–2152. 15. Christy J, et al. (1998) Presettlement Vegetation for the Willamette Valley, Oregon 5. Naidoo R, Ricketts TH (2006) Mapping the economic costs and benefits of conservation. (Oregon Natural Heritage Program, Portland, OR). PLoS Biol 4:2153–2164. 16. US Department of Agriculture, Natural Resources Conservation Service (2001) National 6. Parkhurst GM, et al. (2002) Agglomeration bonus: An incentive mechanism to reunite SSURGO Database (US Dept of Agriculture, Washington, DC). fragmented habitat for biodiversity conservation. Ecol Econ 41:305–328. 17. Oregon Natural Heritage Information Center (2003) Willamette Basin Perennial 7. US Department of Agriculture Farm Service Agency (2007) CRP Enrollment as of Stream GIS Coverage (Oregon Natural Heritage Program, Portland, OR). September 2007 and October 2007 Rental Payments (US Dept of Agriculture, Wash- 18. Eggleston S, Buendia L, Miwa K, Ngara T, Tanabe K (2006) 2006 IPCC Guidelines for ington, DC). National Greenhouse Gas Inventories, Agriculture, Forestry, and Other Land Use (Inst 8. US Department of Agriculture, Natural Resources Conservation Service (2000) Sum- for Global Environ Strategies, Hayama, Japan), Vol 4. mary Report: 1997 National Resources Inventory (US Dept of Agriculture, Washington, 19. Smith JE, Heath LS, Skog KE, Birdsey RA (2006) Methods for Calculating Forest DC, and Statistical Laboratory, Iowa State University, Ames, IA). Ecosystem and Harvested Carbon with Standard Estimates for Forest Types of the 9. Lewis DJ, Plantinga AJ (2007) Policies for habitat fragmentation: Combining econo- United States. (US Dept of Agriculture, Forest Service, Northeastern Research Station, metrics with GIS-based landscape simulations. Land Econ 83:109–127. Newtown Square, PA), Gen Tech Rep NE-343. 10. Parton WJ, Stewart JWB, Cole CV (1988) Dynamics of C, N, P, and S in grassland soils: A 20. Adamus PR, Baker JP, White D, Santelmann M, Haggerty P (2000) Terrestrial Vertebrate model. Biogeochemistry 5:109–131. Species of the Willamette River Basin: Species-Habitat Relationships Matrix. (US 11. Schimel DS, et al. (1996) Climate and nitrogen controls on the geography and Environmental Protection Agency, Corvallis, OR). timescales of terrestrial biogeochemical cycling. Global Biogeochem Cycles 21. Polasky S, Nelson E, Lonsdorf E, Fackler P, Starfield A (2005) Conserving species in a working 10:677–692. landscape: Land use with biological and economic objectives. Ecol Appl 15:1387–1401.

9476 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706178105 Nelson et al. Global mapping of ecosystem services and SPECIAL FEATURE conservation priorities

R. Naidoo*†, A. Balmford‡, R. Costanza§, B. Fisher¶, R. E. Green‡ʈ, B. Lehner**, T. R. Malcolm*, and T. H. Ricketts*

*Conservation Science Program, World Wildlife Fund/United States, 1250 24th Street NW, Washington, DC 20037; ‡Conservation Science Group, Department of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, United Kingdom; §Gund Institute for Ecological Economics, University of Vermont, 617 Main Street, Burlington, VT 05405; ¶Centre for Social and Economic Research on the Global Environment, School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, United Kingdom; ʈThe Royal Society for the Protection of Birds, The Lodge, Sandy, Bedfordshire SG19 2DL, United Kingdom; and **Department of Geography, McGill University, 805 Sherbrooke Street West, Montreal, QC, Canada H3A 2K6

Edited by Gretchen C. Daily, Stanford University, Stanford, CA, and approved March 13, 2008 (received for review August 20, 2007)

Global efforts to conserve biodiversity have the potential to deliver In contrast, the spatial estimation of global ecosystem service economic benefits to people (i.e., ‘‘ecosystem services’’). However, values remains quite crude. Similar to initial estimates of species regions for which conservation benefits both biodiversity and richness, an early and controversial study on global ecosystem ecosystem services cannot be identified unless ecosystem services service values used localized, context-specific valuation studies can be quantified and valued and their areas of production to extrapolate economic values for the whole world (19). Ten mapped. Here we review the theory, data, and analyses needed to years after this study was published, global and regional efforts produce such maps and find that data availability allows us to to map ecosystem services continue to use these estimates quantify imperfect global proxies for only four ecosystem services. (20–22), despite the well known limitations (23). In addition, few Using this incomplete set as an illustration, we compare ecosystem studies have taken advantage of recent technical advances in the service maps with the global distributions of conventional targets selection of priority areas for biodiversity and adapted these for biodiversity conservation. Our preliminary results show that advances to cover ecosystem services (but see refs. 24–26). regions selected to maximize biodiversity provide no more eco- To move forward, global ecosystem service assessments must system services than regions chosen randomly. Furthermore, spa- generate better maps of where ecosystem services are produced, tial concordance among different services, and between ecosystem quantify the likelihood of land use conversion and its probable services and established conservation priorities, varies widely. impact on service provision, and understand the value and flow Despite this lack of general concordance, ‘‘win–win’’ areas— of benefits to nearby and distant human populations. This will regions important for both ecosystem services and biodiversity— require an extraordinary interdisciplinary effort (see Table 1) yet can be usefully identified, both among ecoregions and at finer is vital for informed decision-making. For example, payments for scales within them. An ambitious interdisciplinary research effort ecosystem services (PES), which involve those who benefit from is needed to move beyond these preliminary and illustrative an ecosystem service compensating those who provide the analyses to fully assess synergies and trade-offs in conserving service, will only benefit biodiversity conservation if there is SCIENCE biodiversity and ecosystem services. spatial congruence between important areas for biodiversity and SUSTAINABILITY ecosystem services and if there is congruence in the particular biodiversity ͉ carbon ͉ hotspots ͉ Global 200 ͉ conservation planning land use regimes that best deliver both targets. Because characterizing multiple ecosystem services and biodiversity across fforts to conserve wild nature have traditionally focused on the same region has only recently emerged as a field of study (27, Ebiodiversity: the variety of life on earth at scales from genes 28), these levels of congruence are poorly understood, and the to ecosystems (1). Recently, conservationists have become in- little quantitative evidence available to date has led to mixed terested in another aspect of conservation: the goods and conclusions (26, 29). services from ecological systems that benefit people (e.g., water The above represents an ambitious research agenda for at least purification, carbon sequestration, and crop pollination). These the coming decade, but how far can we get today? Here we ‘‘ecosystem services’’ are currently the focus of intensive re- explore this question by using available data of global extent and search, development, and policy attention (2–4). The Millen- find only four ecosystem services for which we could map proxies nium Ecosystem Assessment (5) documented the importance of at a global scale, and even those data are imperfect (Table 1). We ecosystem services to human well-being and showed that con- assess these maps against the elements necessary for compre- tinued supply of these services is threatened by unsustainable hensive mapping of global ecosystem services. Then, using this anthropogenic activities (5, 6). Conservation groups have begun incomplete set, we illustrate the potential synergies and conflicts to promote ecosystem services and the benefits that biodiversity between the conservation of ecosystem services and more programs confer on people (7), but there is little direct evidence conventional biodiversity targets. We use ecoregions as the of this effect beyond a few local case studies (8–10). To individual planning units because they are widely accepted as a effectively integrate ecosystem services into planned or existing useful template on which to base global-scale priority-setting conservation programs, we need to more broadly evaluate the (30) and because comprehensive data on vertebrate species spatial concordance between areas that produce ecosystem services and those that support biodiversity. Author contributions: R.N., A.B., R.C., B.F., R.E.G., T.R.M., and T.H.R. designed research; R.N., Such evaluation will require the best available data on the A.B., R.C., B.F., R.E.G., B.L., T.R.M., and T.H.R. performed research; R.N., A.B., R.E.G., T.R.M., distribution of both ecosystem services and biodiversity. On the and T.H.R. analyzed data; and R.N., A.B., R.C., B.F., R.E.G., B.L., T.R.M., and T.H.R. wrote the biodiversity side, the past 20 years have seen progress from paper. extrapolations and general guesses about biodiversity in biomes The authors declare no conflict of interest. such as tropical forests (11) to comprehensive, high-resolution This article is a PNAS Direct Submission. global datasets on species ranges of a number of taxonomic †To whom correspondence should be addressed. E-mail: [email protected]. groups (12–15). To accompany these improving data, research- This article contains supporting information online at www.pnas.org/cgi/content/full/ ers have developed sophisticated methods for prioritizing con- 0707823105/DCSupplemental. servation efforts (16–18). © 2008 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0707823105 PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9495–9500 Table 1. Elements required for mapping and valuation of ecosystem services in a spatial context Ecosystem service

Carbon Grassland production Element sequestration Carbon storage of livestock Water provision

Rate of service production Process model Extrapolated Statistical model Production map attributed observations upstream Flow of service away from Global flow Global flow Approximately zero flow Process-based hydrological production area model Presence of beneficiaries Global flow Global flow Restricted to livestock Initially mapped at point areas of use Economic value per unit Globally uniform Globally uniform Local net value of pasture Local net value of water to service to meat yield human uses Probability of system being Fine-scale conversion Fine-scale conversion Fine-scale conversion Fine-scale conversion converted to another state probabilities probabilities probabilities probabilities Change in service value if Difference in service Difference in service Difference in service Difference in service value converted value between value between value between between unconverted unconverted and unconverted and unconverted and and converted states converted states converted states converted states

Entries in bold are those we were able to address for the four services examined here. distributions have recently become available for all terrestrial and belowground carbon stored in various biomes, updated using ecoregions (15). We ask four main questions: (i) How well the Global Land Cover 2000 (GLC2000) land cover map (33) correlated across space are the four ecosystem services? (ii) How (Table 1, row 1). Because carbon storage avoids the prospect of do conservation priorities focused on biodiversity capture eco- further exacerbating global climate change, its remaining ele- system services, and vice versa? (iii) Which of the world’s ments in Table 1 are identical to those for carbon sequestration. terrestrial ecoregions represent ‘‘win–win’’ locations for biodi- The main limitations of this proxy measure are that it is based versity and ecosystem services (given current data on both) and on data that were originally published more than 20 years ago which represent trade-offs? and (iv) Do these global patterns and that the original measures only coincide with 18 biome types hold at within-ecoregion scales, where conservation investments that are mapped globally and assigned a single carbon value. are typically made? Nevertheless, this is still the only globally consistent dataset based on observations of carbon in vegetation biomass (N. Results Ramankutty, personal communication). Ecosystem Service Maps. We found only four ecosystem services Grassland production of livestock is the annual production of for which we could develop spatial proxies (maximum resolution livestock derived, at least in part, from grazing on unimproved 0.5°) to represent their global geographic distribution (Table 1). natural grasslands (Fig. S1C). We mapped livestock production Clearly, these four represent only a small subset of all important on natural grasslands by combining global data on livestock ecosystem services, but we were unable to locate global data for distributions (34), producer prices, and current and potential any others (we did not consider marine services, despite their vegetation (35, 36) (Table 1, row 1). Because the economic economic and nutritional importance). Below we define each benefits from livestock on grasslands are realized primarily at the service, describe the proxies we developed to map them, com- point of production (Table 1, row 3), there is essentially no pare these proxies with the ideal elements in Table 1, and further flow before the service reaches its primary beneficiaries highlight the major limitations of each proxy. See Materials and (Table 1, row 2). Limitations of these data include problems Methods for more details on proxy development. associated with mapping of pastures from remotely sensed Carbon sequestration is the net annual rate of atmospheric imagery (e.g., it is difficult to determine from remote sensing carbon added to existing biomass carbon pools, helping to slow whether grasslands have been ‘‘improved’’ by the introduction of global climate change. Our proxy for the spatial distribution of nonnative species) and the lack of spatially explicit weightings carbon sequestration across the globe was net carbon exchange that would reflect differences in the economic value of livestock (NCE) produced in simulations using the Terrestrial Ecosystem species in different regions of the world. Model (TEM) (31, 32) [supporting information (SI) Fig. S1A]. Water provision is water used for irrigation, industry, domestic Because this service is, in principle, enjoyed everywhere, irre- consumption, and livestock production. Our map of water spective of its point of production, spatially explicit models of provision (Fig. S1D) was based on the global hydrological model service flows (Table 1, row 2) to beneficiaries (Table 1, row 3) WaterGAP (37), which provides spatially explicit estimates of are unnecessary. However, mapping the actual benefits of carbon water availability and water use for various economic sectors sequestration would require additional spatial data on (i) the (Table 1, row 3). We attributed the flows of this water upstream probability that a given parcel of land of a given biome is from their points of use to their points of production (Table 1, converted from its current state and (ii) the differences in rates rows 1 and 2) by dividing the world into drainage basins (38) and of carbon sequestration between previous and new states (Table distributing the total volume of water used in a basin among all 1, rows 5 and 6). The main limitation of our proxy is that it is basin grid cells in proportion to a cell’s runoff level. The main model-based, not observational, and therefore depends heavily limitations of our proxy are that it ignores both spatial variation on the assumptions, input variables, and time scales that defined in water value (depending on scarcity and type of use) and, the particular TEM simulation we used (simulation S3 in ref. 31). crucially, changes in water provision resulting from land use Carbon storage is the amount of carbon stored in vegetation change (Table 1, rows 4 and 5). Hydrological responses to (both aboveground and belowground) and, therefore, an changes in vegetation are complex and controversial (39); our avoided flow of carbon into the atmosphere. Our proxy for global surface, therefore, shows only where any such changes carbon storage (Fig. S1B) was Olson’s classic estimates of above- would impact the most water as currently provided.

9496 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0707823105 Naidoo et al. Table 2. Correlations among log-transformed, per-unit-area ecosystem service production (ecoregions; Pearson correlation coefﬁcients 574 ؍ levels (n SPECIAL FEATURE Carbon Carbon Grassland Water storage sequestration production provision

Carbon storage 1 — — — Carbon sequestration 0.17 1 — — Grassland production Ϫ0.19 Ϫ0.02 1 — Water provision 0.002 Ϫ0.07 0.20 1

Comparisons Between Ecosystem Services and Biodiversity. To assess How effectively do existing priority areas conserve ecosystem the spatial concordance among ecosystem services at the global services? We asked this question for three contrasting global scale, we calculated mean per-unit-area ecosystem service pro- biodiversity priority schemes (Fig. S2): biodiversity hotspots duction for each ecoregion. Log-transformation of these aver- (41), high-biodiversity wilderness areas (HBWAs) (42), and ages, and subsequent Pearson correlation analyses, revealed Global 200 ecoregions (43). The relative performance of the little correspondence among services; no pair of services had a three prioritization schemes varied markedly across the four correlation coefficient Ͼ0.2 (Table 2). ecosystem services (Fig. 2). For carbon storage and sequestra- Next, we tested how well areas selected to maximize biodi- tion, HBWAs had the highest mean levels, whereas biodiversity versity capture ecosystem services, and vice versa. We used hotspots had the lowest levels. Hotspots were, in fact, net integer programming optimization methods (40) and ecoregion emitters of carbon during the 1980s. In contrast, water provision distribution data for mammal, bird, reptile, and amphibian and grassland production of livestock were highest in biodiversity species as our measures of biodiversity, conducting separate hotspots and lowest in HBWAs (Fig. 2). For all four services, analyses for each taxonomic group. We first maximized species Global 200 ecoregions offered intermediate levels of ecosystem representation, determining the set of ecoregions that together service provision that were near to global averages (Fig. 2). contained the most species for a given total area. We compared As a simple illustration of how ecoregions might be prioritized the resulting species accumulation curves with the equivalent for both biodiversity and ecosystem services, we plotted ecore- curves derived by choosing ecoregions to maximize total eco- gions on axes that quantify their rank importance for both system service provision. On average for all taxa, we found that attributes (Fig. 3). We defined biodiversity importance as the (for levels up to 90% of species representation) optimizing for number of endemic vertebrate species, adjusted for area. Com- individual ecosystem services conserved only 22–35% as many bining ecosystem services is difficult without valuations to weight species for a given area as did optimizing for species, that is, no each service relative to others, so, for the purposes of this more than were conserved by selecting ecoregions at random analysis, we defined ecosystem service importance by using

(Fig. 1A). carbon storage alone. We divided Fig. 3 into four quadrants, SCIENCE

We then conducted the converse analysis, selecting ecoregions based on median values for each variable. Despite a lack of SUSTAINABILITY to maximize ecosystem service provision and comparing these correlation between the two variables, many ecoregions lie in results with service levels captured while maximizing species. We quadrant 4, with importance for both biodiversity and ecosystem found that maximizing species representation for a given area services. These win–win ecoregions tend to be located in tropical captured only 17–53% of maximum ecosystem service provision, forested regions (Fig. S3), whereas ecoregions with low priority depending on which service was considered and at which area for both carbon storage and biodiversity are mostly in desert or limit the comparison was made. These levels of ecosystem tundra regions. service capture from species optimization were, again, no Finally, we used data from one specific ecoregion to examine greater than those from a random selection of ecoregions whether global patterns hold at the smaller scales where con- (Fig. 1B). servation decisions are typically made. Chan et al. (26) divided

Fig. 1. Percentage accumulation of species (A) and ecosystem services (B) as total area selected for conservation increases. Circles represent optimizations of species representation (mean results from separate optimizations of birds, mammals, reptiles, and amphibians; error bars ϭ 1 SD). Triangles represent optimizations of ecosystem services (mean results from separate optimizations of each of the four ecosystem services we considered). Shaded areas indicate 95% conﬁdence limits from 500 sets of ecoregions selected at random. Dashed vertical lines indicate area at which all vertebrate species are represented.

Naidoo et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9497 One of our most striking findings is simply how few ecosystem 8 C A services we were able to include in our analyses. This suggests 7 r y 1 that major new interdisciplinary efforts will be required in order . 0 /

x to quantify and map even a fraction of the most important a e ecosystem services—and to assess their coincidence with biodi- h d 5 / n 0 I

. versity—at global scales. 3456 05

C Learning what we could from the available data, we found that 2 t conservation priorities aimed solely at biodiversity do not con- 5 0 .

01 serve optimal levels of the ecosystem services we examined, and 0 − vice versa. This can be seen by using both theoretical planning Hotspots G200 HBWA Hotspots G200 HBWA algorithms (Fig. 1) and actual portfolios for biodiversity conservation (Fig. 2). More complete analyses, including additional 8 0 B . ecosystem services and other aspects of biodiversity, may show

0 D r

y better concordance. Nevertheless, our findings emphasize that 0 6 / 60 0 .

a such concordance cannot be assumed and needs to be tested 08 h m empirically. 4 k / 0 4 . c 0 The pattern of service capture among global conservation i C b t priority schemes (Fig. 2) reflects the distribution of service 0 u 20 . 0 c beneficiaries within these regions. The benefits of carbon storage

0 and sequestration are globally dispersed and do not depend on 0 00 .

02 local beneficiaries. Accordingly, these services were highest in Hotspots G200 HBWA Hotspots G200 HBWA the low-human-density HBWAs. On the other hand, water provision and grassland production of livestock, which benefit Fig. 2. Mean amount of ecosystem services per unit area provided by three people closer to the point of service production, were highest in global conservation priority schemes. (A) Carbon sequestration. (B) Carbon the densely populated biodiversity hotspots. The intermediate storage. (C) Grassland production of livestock. (D) Water provision. Horizontal levels of service provision for Global 200 ecoregions reflect that lines represent the global mean for each service. these regions include both wilderness areas and areas of dense human settlement. the California Central Coast ecoregion of the United States into Despite the overall lack of spatial concordance between Ϸ12,000 hexagonal planning units and assessed levels of biodi- biodiversity and ecosystem services (Table 2), we were still able to identify win–win areas: ecoregions important for both types versity and ecosystem services in each. This ecoregion occupies of targets (Fig. 3, quadrant 4). In these places, PES are more quadrant 4 in Fig. 3, identifying it as a global win–win for likely than elsewhere to achieve biodiversity conservation ob- conserving biodiversity and stored carbon. However, mapping jectives. Perhaps more importantly, Fig. 3 also identifies areas of the quadrant scores of individual hexagons within the ecoregion trade-off, where biodiversity conservation is crucial but is less indicates an abundance of both win–win and trade-off locations likely to be supported by PES (quadrant 3) or where PES will (Fig. S4, and see figure 2 in ref. 25 for similar results from more likely have lesser impact on global conservation (quadrant 1). sophisticated analyses). Mapping out all four quadrants (Fig. S3) illustrates that priorities for both ecosystem services and biodiversity can be located to Discussion guide conservation investment. The analyses presented here are clearly preliminary and are Although global analysis can inform broad-scale priorities, it intended primarily for heuristic purposes. However, they do is clear that actual conservation investments typically occur at suggest some early lessons and important research priorities. more local scales. We found that, although a California ecoregion falls within quadrant 4 in our global analysis (Fig. 3), less than one-quarter of the planning units within it represent ) h 0 1 4 win–wins at more local scales (Fig. S4). Hence, even win–win g 6 • • • i • • • • • • • • • • • • • • • • • • • • h • • • • • • • • ecoregions can contain areas that represent trade-offs between • • • •• • • • • • •

− • • • • • • • • • • • • • • • • • • • • • • • • • w • • • • • • • • • • conservation and ecosystem services. By extension, ecoregions • • • 0 • • • • • • • o • • • • • • • • • l • • 50 • • • • • • • • • ( • • • • • • •• • • • • • • •• in quadrants 1–3 (Fig. 3) will almost certainly contain local • • • • • • • • • •• s • • • • • • • • • • • • • • • • • • e • • • • • • • • • • • • • • • • • • win–win opportunities, especially when services that operate c • • • • • • • • • i • 0 • • • • • • • • • • • • • • • • • v • 40 • • • • • • • • • ••

r • • • • • • • over relatively small scales (e.g., pollination by wild insects) are • • • • • • • • • • • e • • • •• • • • •• • • • • • • • • s • • • • • • • • • • • • • • • • • considered. Although finer scale analyses will be essential for • • • • • • • • • • • • • • • • • • 0 • • • m • • • • • • • • • • • • • • 30 • • • e • • • • • • • • targeting specific conservation action, global analyses remain • • • t • • • • • • • • • • • • • • • • • • s • • • • • • • • • •• • • • • • • •

y • • • • • • • useful for identifying broad areas where taking such actions is • • • • • •• • • • • s • • • • • • • • • • • • • • 0 • • • • • • o • • • • • • • • • • • • 20 • • • most likely to benefit people. • • • c • • • • • • • • • • • • • • • • • • e • • • • • • • • • • • • • •• • •• • • • r • • • • • Data availability severely limited our analyses. For the four • • • • • • • • • • • • • • • o • • • • • • • • • • • f • •• • • • 0 • • • • • • • • • • • • • • ecosystem services we assessed, we were only able to assemble • • • • 10 • • • y • • • • • • • • • • • t • • • • • • • • • i • • • • • • • • • • • • r • • • • • • • • data on the first three elements that are needed for a compre- • • • • • • • • • • • • o • • • • • • • • •

i • • • • •• • • • • • • • r • 00 2 3 hensive assessment (Table 1, rows 1–3). Relatively speaking, P mapping of the biophysical production (row 1) and flows of 0 100 200 300 400 500 600 ecosystem services has perhaps seen the most progress, but, as Priority for species conservation (low−high) we have indicated, much remains to be done (44, 45). Identifying ecosystem service beneficiaries requires knowing who they are Fig. 3. Relationship between ecosystem service importance and conservation importance. Each axis is the rank order (low to high) of 574 terrestrial and where they live. Sociologists, anthropologists, and political ecoregions in terms of per-area carbon storage (y axis) or area-corrected scientists need to identify how social structure, cultural elements, number of endemic species [x axis; calculated as (No. endemic species/ and governance affect who benefits from ecosystem services and area0.25)]. Lines indicate median values for each variable. Diamond in quad- by how much (46). On the economics side (Table 1, row 4), no rant 4 represents the California Central Coast ecoregion. spatial data were available on economic values, preventing us

9498 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0707823105 Naidoo et al. from ‘‘monetizing’’ our maps. Although there is a rich literature closely approximate current carbon fluxes while also accounting for interan- on economic valuation of the environment, research on how nual variability. SPECIAL FEATURE values vary spatially has only recently begun to emerge (24). Beyond economic valuation, quantifying the net benefits of Carbon Storage. Olson (33) originally estimated biome-level carbon values in ecosystem conservation also requires spatially explicit data on preagricultural live vegetation through extensive field observations, inven- the probabilities of conversion to other land use types (Table 1, tories, and analyses of literature values. The updated dataset we used (33) mapped carbon values from Olson’s original biome types to land cover types row 5), coupled with information on likely rates of service of the GLC2000, hence providing an estimate of carbon stored in vegetation provision after conversion (Table 1, row 6). Different manage- circa the year 2000. Although the dataset enjoys widespread use in carbon ment regimes within the same land use type can also alter cycling research, its major deficiency is that whole biomes are assigned a single levels of service provision (47) and so are another important carbon value, neglecting major within-biome variability due to heterogeneity consideration. in climate, land use, soils, and topography. These data limitations likely affect our comparative results in several ways. For example, natural pastures are difficult to Grassland Production of Livestock. To map livestock production on natural distinguish from human-cleared areas by using remote sensing; pastures, we used recently developed 5Ј-resolution global maps of livestock therefore, we may have overestimated livestock production from distributions (34). These maps use regression-based methods to estimate the natural pastures in the biodiversity hotspots, with their prepon- expected density of cattle, sheep, goats, pigs, poultry, and buffalo across the earth’s surface. For each livestock type, we used these density estimates and derance of cleared lands (Fig. 2). More generally, the particular data on the mass of edible meat per animal (estimated by country from United services and species we considered may have led us to overes- Nations Food and Agriculture Organization data) to estimate the tons of meat timate how poorly priority areas for species conservation deliver produced in each cell. A global producer price was used to weight different services, and vice versa. Two of our four services (carbon storage livestock types; using these weights, an aggregate index of livestock produc- and sequestration) peak in tropical and temperate forests and tion was obtained by summing the weighted livestock meat weights. We then are, therefore, biased away from ecoregions with high vertebrate constructed a global map of natural pastures by combining a 5Ј-resolution endemism, which are often island or montane areas (41). potential vegetation dataset of savanna, grassland/steppe, and shrubland Similarly, given our endemics-based biodiversity measure, our biome types (35) and then masking out all known human-altered landscapes species data were limited to taxa with relatively large ranges, by using the GLC2000 (36). We intersected the livestock production index and the map of natural pastures to produce a global map of livestock production reducing the importance of broad areas of tropical forests. Both on natural pastures, which, unlike for the other services we consider, restricts of these data biases act to drive the curves in Fig. 1 further apart, the production domain to wild nature. We aggregated the map of livestock thus underestimating the degree to which planning for ecosystem production from 5Ј to 0.5° spatial resolution for consistency with the other services would capture biodiversity, and vice versa. For these ecosystem service datasets evaluated here. reasons, we reiterate that our analyses are only a first-cut attempt to illustrate the types of questions that could be asked, and Water Provision. We used a global hydrological model to map water provision approaches that could be used, once more comprehensive data for human consumptive use. The WaterGAP 2 model provides data at 0.5° on ecosystem services are available. spatial resolution on annual water availability (surface runoff and total dis- Despite the challenges discussed here, comparisons between charge) and consumptive water use per sector (e.g., industrial, domestic, biodiversity and ecosystem services have the potential to inform irrigation, and livestock), averaged over a 30-year period (1961–1990) (37). We SCIENCE summed consumptive water use across the four sectors to produce a spatially decision-making. To realize this potential, we need to learn how SUSTAINABILITY explicit map of total water use in biophysical units (cubic kilometers per year). to rigorously quantify and map more elements of more services. We then attributed the volume of water consumption back to its points of Such an improved understanding would allow us to operation- origin by using a basin-level perspective of water production. Drainage basins alize a principle that is becoming increasingly clear: that human for the globe were identified from a 0.5° global drainage direction map welfare is linked in diverse ways to biodiversity conservation and (DDM30) (38). We calculated the proportional contribution of each cell to the that sustainable development should involve managing for both. total water production of the basin in which it resides, calculated the amount of total water consumption for that basin, and then redistributed the total Materials and Methods consumption according to the proportion of basin-wide water production at Carbon Sequestration. We used NCE results from the TEM model (32) as a proxy each grid cell. By redistributing the volume of water consumption in this for carbon sequestration. TEM is a terrestrial biosphere model that simulates manner, we arrived at a coarse estimate of total water use attributed to point carbon exchange between the atmosphere and terrestrial biosphere on the of origin. basis of vegetation types, soils, climate, atmospheric CO2, and land use history. We used results from the S3 simulation of TEM (31), in which atmospheric CO2 Additional Materials and Methods. For further details on the methods and concentrations, cropland expansion, and climatic conditions were all varied materials used in this study, see SI Materials and Methods. simultaneously. The model was run to equilibrium to 1880 and then run transiently through 1992. In ref. 31, annual NCE was calculated as heterotro- ACKNOWLEDGMENTS. We thank Sam Buttrey, Dick Cameron, Karen Carney, phic respiration minus the sum of net primary productivity, carbon emissions Annie Claus, Jeff Hicke, Paul Jeferiss, Shuang Liu, Paul Morling, Nasser Olwero, from the conversion of natural vegetation to cropland, and carbon emissions Chris Potter, Navin Ramankutty, Ana Rodrigues, Rebecca Shaw, and Nikolai from the decay of forest and agricultural products (units ϭ t haϪ1 yrϪ1). Note Sindorf for critical comments and/or help with analyses that improved the C manuscript. This work was initiated at the ‘‘Valuing Wild Nature’’ workshop that for our analyses, we multiplied NCE from ref. 31 by Ϫ1, meaning that held at the University of East Anglia, March 12–16, 2006, sponsored by The positive numbers indicate a net sequestration of carbon from the atmosphere Royal Society for the Protection of Birds; the Department for Environment, to the biosphere, whereas negative numbers indicate a net release of carbon Food and Rural Affairs; and Natural England. Finally, we thank the Lever- from the biosphere to the atmosphere. We used average NCE for the 1980s hulme Trust and the Packard Foundation for supporting this work and ongo- (the most recent decade for which simulation data were available) to most ing follow-ups.

1. Brooks TM, et al. (2006) Global biodiversity conservation priorities. Science 313:58–61. 7. Odling-Smee L (2005) Dollars and sense. Nature 437:614–616. 2. Sachs JD, Reid WV (2006) Environment: Investments toward sustainable development. 8. Russ GR, Alcala AC, Maypa AP, Calumpong HP, White AT (2004) Marine reserve benefits Science 312:1002. local fisheries. Ecol Appl 14:597–606. 3. Wunder S (2007) The efficiency of payments for environmental services in tropical 9. Ricketts TH, Daily GC, Ehrlich PR, Michener CD (2004) Economic value of tropical forest conservation. Conserv Biol 21:48–58. to coffee production. Proc Natl Acad Sci USA 101:12579–12582. 4. Daily GC (1997) Nature’s Services: Societal Dependence on Natural Ecosystems (Island 10. Becker CD (1999) Protecting a garu´ a forest in Ecuador: The role of institutions and Press, Washington, DC). ecosystem valuation. Ambio 28:156–161. 5. Millennium Ecosystem Assessment (2005) Ecosystems and Human Well-Being: Synthe- 11. Erwin TL (1988) The tropical forest canopy: The heart of biotic diversity. Biodiversity,ed sis (Island Press, Washington, DC). Wilson EO (Natl Acad Press, Washington, DC), pp 123–129. 6. Balmford A, et al. (2002) Economic reasons for conserving wild nature. Science 12. Orme CDL, et al. (2005) Global hotspots of species richness are not congruent with 297:950–953. endemism or threat. Nature 436:1016–1019.

Naidoo et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9499 13. Ceballos G, Ehrlich PR, Soberon J, Salazar I, Fay JP (2005) Global mammal conservation: 31. McGuire AD, et al. (2001) Carbon balance of the terrestrial biosphere in the twentieth What must we manage? Science 309:603–607. century: Analyses of CO2, climate and land use effects with four process-based ecosys- 14. Stuart SN, et al. (2004) Status and trends of amphibian declines and extinctions tem models. Global Biogeochem Cycles 15:183–206. worldwide. Science 306:1783–1786. 32. Tian H, Melillo JM, Kicklighter DW, McGuire AD, Helfrich J (1999) The sensitivity of 15. Lamoreux JF, et al. (2006) Global tests of biodiversity concordance and the importance terrestrial carbon storage to historical climate variability and atmospheric CO2 in the of endemism. Nature 440:212–214. United States. Tellus Ser B Chem Phys Meteorol 51:414–452. 16. Wilson KA, McBride MF, Bode M, Possingham HP (2006) Prioritizing global conserva- 33. Gibbs HK (2006) Olson’s Major World Ecosystem Complexes Ranked by Carbon in Live tion efforts. Nature 440:337–340. Vegetation: An Updated Database Using the GLC2000 Land Cover Product (Oak Ridge 17. Pressey RL, Watts ME, Barrett TW (2004) Is maximizing protection the same as mini- Natl Lab, Oak Ridge, TN), Numeric Data Package (NDP)-017b. Available at http:// mizing loss? Efficiency and retention as alternative measures of the effectiveness of cdiac.ornl.gov/epubs/ndp/ndp017/ndp017b.html. proposed reserves. Ecol Lett 7:1035–1046. 34. Wint W, Robinson T (2006) Gridded Livestock of the World (FAO, Rome). 18. Margules CR, Pressey RL (2000) Systematic conservation planning. Nature 405:243–253. 35. Ramankutty N, Foley JA (1999) Estimating historical changes in global land cover: 19. Costanza R, et al. (1997) The value of the world’s ecosystem services and natural capital. Croplands from 1700 to 1992. Global Biogeochem Cycles 13:997–1027. Nature 387:253–260. 36. Bartholome E, Belward AS (2005) Glc2000: A new approach to global land cover 20. Sutton PC, Costanza R (2002) Global estimates of market and nonmarket values derived mapping from earth observation data. Int J Remote Sens 26:1959–1977. from nighttime satellite imagery, land cover, and ecosystem service valuation. Ecol 37. Alcamo J, et al. (2003) Development and testing of the WaterGAP 2 global model of Econ 41:509–527. water use and availability. Hydrol Sci J 48:339–348. 21. Li R-Q, et al. (2007) Quantification of the impact of land-use changes on ecosystem 38. Do¨ ll P, Lehner B (2002) Validation of a new global 30-min drainage direction map. J services: A case study in Pingbian County, China. Environ Monit Assess 128:503–510. Hydrol 258:214–231. 22. Turner WR, et al. (2007) Global conservation of biodiversity and ecosystem services. 39. Bruijnzeel LA (2004) Hydrological functions of tropical forests: Not seeing the soil for Bioscience 57:868–873. the trees? Agric Ecosyst Environ 104:185–228. 23. Bockstael NE, Freeman AM, Kopp RJ, Portney PR, Smith VK (2000) On measuring 40. Polasky S, Camm JD, Garber-Yonts B (2001) Selecting biological reserves cost- economic values for nature. Environ Sci Technol 34:1384–1389. effectively: An application to terrestrial vertebrate conservation in Oregon. Land Econ 24. Naidoo R, Ricketts TH (2006) Mapping the economic costs and benefits of conservation. 77:68–78. PLoS Biol 4:e360. 41. Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J (2000) Biodiversity 25. van Jaarsveld AS, et al. (2005) Measuring conditions and trends in ecosystem services hotspots for conservation priorities. Nature 403:853–858. at multiple scales: The Southern African Millennium Ecosystem Assessment (SAfMA) 42. Mittermeier RA, et al. (2003) Wilderness and biodiversity conservation. Proc Natl Acad experience. Philos Trans R Soc London Ser B 360:425–441. Sci USA 100:10309–10313. 26. Chan KMA, Shaw MR, Cameron DR, Underwood EC, Daily GC (2006) Conservation 43. Olson DM, Dinerstein E (1998) The Global 200: A representation approach to conserv- planning for ecosystem services. PLoS Biol 4:e379. ing the earth’s most biologically valuable ecoregions. Conserv Biol 12:502–515. 27. Schroter D, et al. (2005) Ecosystem service supply and vulnerability to global change in 44. Carpenter SR, et al. (2006) Millennium Ecosystem Assessment: Research needs. Science Europe. Science 310:1333–1337. 314:257–258. 28. Metzger MJ, Rounsevell MDA, Acosta-Michlik L, Leemans R, Schroter D (2006) The vulnera- 45. Kremen C (2005) Managing ecosystem services: What do we need to know about their bility of ecosystem services to land use change. Agric Ecosyst Environ 114:69–85. ecology? Ecol Lett 8:468–479. 29. Bohensky EL, Reyers B, Van Jaarsveld AS (2006) Future ecosystem services in a Southern 46. Chan KMA, et al. (2007) When agendas collide: Human welfare and biological con- African river basin: A scenario planning approach to uncertainty. Conserv Biol 20:1051– servation. Conserv Biol 21:59–68. 1061. 47. Steffan-Dewenter I, et al. (2007) Trade-offs between income, biodiversity, and eco- 30. Wikramanayake E, et al. (2002) Ecoregions in ascendance: Reply to Jepson and Whit- system functioning during tropical rainforest conversion and agroforestry intensifica- taker. Conserv Biol 16:238–243. tion. Proc Natl Acad Sci USA 104:4973–4978.

9500 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0707823105 Naidoo et al. Designing payments for ecosystem services: SPECIAL FEATURE Lessons from previous experience with incentive-based mechanisms

B. Kelsey Jacka, Carolyn Kouskya, and Katharine R. E. Simsa

Sustainability Science Program, Center for International Development, Harvard University, 79 John F. Kennedy Street, Cambridge, MA 02138

Edited by Gretchen C. Daily, Stanford University, Stanford, CA, and approved October 5, 2007 (received for review June 12, 2007)

Payments for ecosystem services (PES) policies compensate in- Incentive-based mechanisms dividuals or communities for undertaking actions that increase the provision of ecosystem services such as water puriﬁcation, Payments for ecosystem ﬂood mitigation, or carbon sequestration. PES schemes rely on sservice incentives to induce behavioral change and can thus be consid- Market Tradable ered part of the broader class of incentive- or market-based Charges Sesubsidi friction permits mechanisms for environmental policy. By recognizing that PES reductions programs are incentive-based, policymakers can draw on insights from the substantial body of accumulated knowledge Direct regulation about this class of instruments. In particular, this article offers a Public provision set of lessons about how the environmental, socioeconomic, political, and dynamic context of a PES policy is likely to interact Private contracts with policy design to produce policy outcomes, including envi- SCIENCES ronmental effectiveness, cost-effectiveness, and poverty Vroluntary andh educational app osac e ECONOMIC alleviation. Fig. 1. Locating PES as an incentive-based mechanism within a broader suite of environmental policy instruments. environmental policy ͉ incentive payments ͉ market-based instruments

cosystem services are the benefits that people derive from Recently, ‘‘payments for ecosystem services’’ (PES) has Eecosystems, including both commodities and regulating, sup- emerged as a policy solution for realigning the private and social porting, and cultural services (1, 2).b The type, quality, and quantity benefits that result from decisions related to the environment. of services provided by an ecosystem are affected by the resource The PES approach is based on a theoretically straightforward use decisions of individuals and communities. When the benefits of proposition: pay individuals or communities to undertake ac- an ecosystem service accrue mainly to those who make manage- tions that increase levels of desired ecosystem services. A formal ment decisions, as in the production of crops or livestock, private definition has been given by Sven Wunder (7): ‘‘A PES scheme, markets are likely to work relatively well in inducing service simply stated, is a voluntary, conditional agreement between at provision. However, when the benefits of an ecosystem service flow least one ‘seller’ and one ‘buyer’ over a well defined environ- primarily to others, such as with water purification or climate mental service—or a land use presumed to produce that ser- stabilization, public interests and the interests of the resource vice.’’d In the last decade or so, hundreds of new PES initiatives manager may be misaligned. This difference in private and social have emerged around the globe.e Costa Rica, Mexico, and China benefits, or the problem of ‘‘externalities,’’ results in a classic market all have initiated large-scale programs that give direct payments failure: individuals will tend to provide too little of the ecosystem to landowners for undertaking specific land use practices that service. This basic logic may explain much of the decline of could increase the provision of hydrological services, biodiver- important ecosystem services as a result of human pressures (2, 4). sity conservation, erosion prevention, carbon sequestration, or Potential policy solutions to externalities problems include scenic beauty (10–12). Some PES policies were initiated before public provision of goods and services, private contracts between the term ‘‘payments for ecosystem services’’ came into common the provider and the recipients, encouragement of voluntary efforts by firms and individuals, direct government regulation, Author contributions: B.K.J., C.K., and K.R.E.S. designed research, performed research, and and hybrid mechanisms such as government-supported trading wrote the paper. markets (see Fig. 1). Many government interventions to control The authors declare no conflict of interest. externalities have taken the form of command-and-control This article is a PNAS Direct Submission. regulation, which mandates that actors undertake specific ac- aTo whom correspondence may be addressed. E-mail: kelsey࿝[email protected], tions and applies sanctions if they do not comply. In contrast, carolyn࿝[email protected], and [email protected]. incentive-based policies address externalities by altering the bDefinitions of ecosystem services vary. Boyd and Banzhaf (3) distinguish between ecosys- economic incentives private actors face, while allowing those tem functions (the biological, chemical, and physical properties of ecosystems) and actors to decide whether and how much to change their behavior. ecosystem services (the aspects of ecosystems that are valued by humans). We use the term ‘‘ecosystem services’’ broadly to refer to both intermediate and final services. Most incentive-based mechanisms have been initiated through cIncentive-based mechanisms may also be referred to as ‘‘market-based instruments’’ public policies, although privately negotiated incentive-based because they rely on price signals, like those in private markets, to convey incentives for solutions are possible. Incentive-based mechanisms include behavioral change. charges (such as taxes, user fees, and deposit–refund systems), dMany projects that are called PES schemes fall short of this theoretical ideal definition in subsidies, tradable permits (including markets for pollution practice (8). reduction and tradable development rights), and market friction eA 2002 survey found examples of 287 ‘‘markets for environmental services’’ (9). reduction (e.g., liability rules and information programs) (5, 6).c © 2008 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705503104 PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9465–9470 example, payments might be offered as a lump sum for actions such Environmental Environmental context effectiveness as planting a buffer strip; as a set rate for a scaleable action, such Socio- as number of trees planted; through an allocation mechanism PES g economic policy Cost - such as a reverse auction ; or indirectly, through a system of context design effectiveness differential use taxes such that tax rates are lower for landholders who engage in desired land uses. PES policies may be funded by Political Equity taxes, by nongovernmental organization (NGO) funding acquired context from voluntary contributions, by direct fees on service consumers, Context or through other mechanisms. Some PES schemes take the form of dynamics tradable permit systems, such as wetland mitigation banking or tradable development rights; some are subsidies; and others, such Fig. 2. Context interacts with PES policy design to determine outcomes. as ecolabeling, work to reduce market friction by providing information about the origin of products. Despite this variation, PES policies share a common element: as with other incentive-based usage and yet are based on the same theory. For example, the approaches, PES policies work by changing incentives rather than U.S. Conservation Reserve Program, run by the U.S. Depart- by making explicit rules or directives. ment of Agriculture, has paid farmers to plant permanent In the sections that follow, we present lessons for PES policy vegetation on environmentally sensitive cropland since the mid- design based on previous experience with, and research on, 1980s (13). incentive-based approaches, grouping the lessons by their relation- PES schemes are similar in structure to other incentive-based ship to the environmental, socioeconomic, or political context of policies for achieving environmental goals, as highlighted in Fig. 1. implementation. We also discuss how changes in context over time Therefore, the accumulated experience with, and research on, can affect PES performance. incentive-based mechanisms provides relevant insights for both academics and practitioners interested in payment schemes for Environmental Context ecosystem services. In this article, we draw on the literature on Previous experience with incentive-based mechanisms has dem- incentive-based mechanisms for environmental policy to suggest onstrated that the properties of the ecosystem and/or pollutant lessons on how the socioeconomic, environmental, and political under consideration—the environmental context—influence how a context in which policies are implemented, together with policy policy should be designed and what type of outcomes should be design, influences the outcomes of PES schemes. expected (16, 17). Similarly, the properties of the ecosystem ser- Situating Policy Design in Context vice(s) being targeted in a PES scheme will interact with policy design to influence policy outcomes. Two lessons stand out based As illustrated in Fig. 2, the framework underlying the lessons on previous experience with incentive-based mechanisms. presented here is based on the assumption that context interacts with policy design and that together these determine policy When the Marginal Benefits from Service Provision Are Not Constant, outcomes. We address four aspects of context: the environmen- More Complex Incentive Schemes Are Needed to Achieve Environ- tal context, the socioeconomic context, the political context, and mental Effectiveness. In theory, incentive-based mechanisms can context dynamics. The policy outcomes we emphasize are envi- deliver the same environmental benefits as direct regulation. How- ronmental effectiveness, cost-effectiveness, and equity. To be ever, experience with pollution control mechanisms has demon- environmentally effective, a project must deliver a set level of strated that the design of environmentally effective policies is more environmental benefits, as defined by physical measurements. straightforward when marginal environmental benefits are constant To be cost-effective, a policy must achieve the same level of across abatement sources. When marginal benefits are constant, the environmental benefits at a lower cost than other possible first ton of pollution abatement provides the same benefit as the policies.f The costs of a PES scheme, from a social perspective, 100th ton of abatement, and this is the case regardless of the source include not only direct implementation costs, but also the or location of the abatement. Constant marginal benefits simplify transaction costs of the program and the costs of forgone the design of a policy because expected environmental benefits do alternative productive uses of the resource, often referred to as not depend on initial conditions or on which of the agents reduces ‘‘opportunity costs.’’ Transaction costs include the expense of pollution. Thus, a per-unit tax or simple trading scheme can be used negotiating contracts, performing scientific baseline studies, and to predictably reach a given environmental target. monitoring and enforcement. Finally, although many possible For many environmental problems, however, the marginal en- aspects of equity are important, we focus on poverty alleviation vironmental benefits from an additional unit of abatement are not because it is most frequently discussed in the emerging PES constant. Instead, they depend on source, location, and initial literature (8, 14, 15). We include equity as a relevant policy conditions. For example, toxic chemicals generally create health outcome, even though there are many cases of PES policies, effects that increase at an increasing rate with the amount of particularly in developed countries, where it is not an explicit exposure. When marginal environmental benefits are significantly goal of the program. different across sources, more complex incentive-based systems Given these potential goals for PES policies, the likelihood that all three are achieved will depend on the design characteristics of such as ambient permits, differential taxes, or trading zones are a PES scheme and the context in which it is implemented. Varia- needed to explicitly differentiate between polluters on the basis of tions in the structure of PES schemes include the form of the location or other characteristics (5, 17). incentive or payment, which services are provided, who the pro- Similarly, for PES policies, if the marginal environmental benefits viders are, who the implementers and intermediaries are, whether of a particular ecosystem service are not constant, simple PES incentives are given to individuals or communities, the eligibility schemes that do not account for how benefits change with different rules for participation, and how the payments are funded. For configurations of participants may not be environmentally effective. Many examples of nonconstant marginal benefits and threshold effects are found in ecological systems, including lakes, coral reefs, fThe criterion of cost-effectiveness takes as given a particular environmental goal (e.g., a level of benefits) and judges policies only on their cost side—by how cheaply a policy reaches that goal. Economic efficiency, on the other hand, compares benefits with costs gIn a reverse auction, landholders submit bids indicating how much compensation they and judges a policy by the net benefits, or total benefits minus total costs. require to undertake particular actions.

9466 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705503104 Jack et al. oceans, forests, and arid lands (18, 19). For example, the preser- costs. Likely sources of individual heterogeneity in the costs of vation of the habitat of a large predator might require a minimum providing ecosystem services include differences in the op- SPECIAL FEATURE area of land for species viability; below this level, preservation offers portunity costs of land use stemming from biophysical features no protection benefits for that species. If a PES scheme simply of the land and its location, as well as individual characteristics compensates for individual land use changes without considering of the landholders, such as education, risk aversion, and plot these irregularities, it may not achieve its environmental objective. size. The cost-effectiveness of PES policies, compared with a Several PES schemes have been developed (e.g., refs. 20 and 21) uniform set of regulations, will tend to be higher where there that take into account nonconstant marginal benefits, to avoid this is high variation in marginal provision costs across the problem. However, it is important to remember that increased population (24). complexity in design is likely to increase costs. When the Poorest Providers Are also Those with the Lowest Oppor- Assessments of the Final Ecosystem Services Depend on the Certainty tunity Costs and the Highest Service Provision Potential, PES Policies of the Relationship Between Proxies and Environmental Benefits. Are Most Likely to Help Alleviate Poverty. Incentive-based mecha- Measuring the environmental effects of a policy can sometimes be nisms target the providers who have lower opportunity costs. impossible or prohibitively expensive. Thus, incentive-based poli- Therefore, the potential for poverty alleviation as a coupled policy cies frequently tie the incentives to a proxy for environmental goal depends, in part, on the coincidence between opportunity cost benefits that is easy to measure and that relates to the level of and poverty. For instance, poorer farmers may tend to own benefits provided. Many air emissions estimates are based on marginal lands with higher slope and lower soil quality, in which models of material inputs and production processes, for example, case the opportunity cost of leaving the land in natural vegetation rather than on emissions measured in real time at the firm to increase ecosystem service provision may be lower. In this case, level (22). PES schemes have the potential to simultaneously direct payments Similarly, most PES schemes rely on observable proxies, such as toward the poor and toward the lowest cost providers of desired actions or outcomes (e.g., the presence of buffer strips or the ecosystem services. In addition, how the income of landowners amount of forest cover), because direct monitoring of ecosystem service outputs is difficult or costly. Devising appropriate proxies varies with production of environmental benefits is relevant to requires an understanding of how activities, such as planting trees, whether PES policies will be able to improve equity (15). When land that produces a high level of services is held by poor members of relate to ecosystem functions such as carbon storage and, ultimately, SCIENCES ECONOMIC to ecosystem services such as climate stabilization. Depending on society, then a PES approach may contribute to poverty reduction the type of ecosystem service, proxies may be relatively easy or by paying these landholders for the services they provide (25). difficult to use. The long-run viability of PES schemes may depend, However, PES schemes are likely to make a true improvement in in part, on advances in techniques for estimating ecosystem services poverty outcomes only if they pay landowners an amount substan- h from easily observable ecosystem properties. tially higher than they otherwise could have earned with the land. This implies a likely tradeoff between the cost-effectiveness of the Socioeconomic Context program and poverty alleviation.i The socioeconomic context—the distribution of resources, the price of goods and services, and other features of the economy and When Resources Are Owned by Many Small-Holders, Transaction Costs social system in which a policy occurs—can alter the impacts of a Will Possibly Be Higher, Implying a Tradeoff Between Cost-Effective- policy. Three lessons are given below. ness and Poverty Alleviation. To accurately assess the costs of a policy, transaction costs must be considered. In particular, imple- The Greater the Heterogeneity in Costs, the Greater the Potential for mentation, monitoring, and enforcement costs can be high under a PES Scheme To Be Cost-Effective Compared with a Command-and- incentive-based approaches if contracts or requirements are tai- Control Approach. The great promise of incentive-based instru- lored to individual users. All else being equal, contracting and ments for attaining environmental policy goals, such as pol- monitoring are cheaper when the number of agents is small. The lution control, is their potential to be cost-effective compared literature on incentive-based mechanisms frequently distinguishes with command-and-control solutions, by inducing an alloca- between point-source emissions, which involve identifiable sources tion of production or abatement that results in the least total with fixed locations, and area or nonpoint sources, which are cost. If, for example, pollution abatement costs vary, the lowest diffuse, mobile, or hard to identify. Monitoring and enforcement cost solution allocates emission reductions so that the marginal for point sources is usually much cheaper. PES schemes are often costs are equal across all producers. Any other allocation focused on nonpoint sources or on many individual landowners would require that some of the burden be shifted to a producer whose collective activities alter the levels of a given ecosystem with higher control costs. Greater heterogeneity among pro- service. This feature will increase policy costs. It is possible, ducers in terms of costs of abatement will generate higher however, that working with a third-party intermediary such as an savings compared with a command-and-control approach that NGO or a community could reduce the costs of working with a requires uniform abatement across producers or an approach large number of providers (27). that does not allow flexibility based on cost of control (23). PES schemes also have the potential to achieve a more Political Context cost-effective provision of ecosystem services relative to a Decades of experience with incentive-based instruments have mandatory approach that requires the same actions from all illustrated that the political influence of ‘‘winners’’ and ‘‘losers’’ landowners. By offering a set payment for service provision, from incentive-based policies will shape design and implementation individuals who can produce the ecosystem service at or below (28–30). Similarly, some types of PES policies may be more that price have an incentive to enroll in the program, whereas politically feasible than others, ultimately influencing the range of those providers who have a higher opportunity cost of enroll- ing do not. A reverse auction for PES contracts can also induce the cost-effective allocation of service provision. Society as a hIf poor landholders are more risk-averse, then they may demand greater compensation to whole gains the same amount of ecosystem services for less switch to unfamiliar land uses. On the other hand, the payments under a PES program may cost. However, whether a cost-effective allocation represents be relatively certain compared with other income. significant cost savings compared with a uniform requirement iAt least one study has found that PES payments constitute only a small fraction of from all landholders depends on the heterogeneity of provider landowners’ income, suggesting minimal equity impacts (26).

Jack et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9467 potential outcomes. Below, we highlight four lessons for PES Existing Subsidies That Are a Product of the Political Process May policies with respect to political context. Interfere with Effective Incentives. PES policies may be undermined by existing subsidy programs or tax regimes designed to encourage The Funding Available to ‘‘Buy’’ Services Depends not just on the resource use that is counter to the ecosystem service goals of the Latent Demand for Ecosystem Services, but also on the Structure of policy. In Indonesia, for example, the Rewarding Upland Poor for the Funding Mechanism. The rationale for a PES approach is that the Environmental Services (RUPES) program is working to provide recipients of the services have some measurable value or ‘‘willing- incentives to farmers to maintain jungle rubber mixed agroforestry ness to pay’’ for those services. However, converting that latent systems. At the same time, the government provides subsidies to demand into funding that reaches the suppliers of ecosystem farmers who clear land for conversion to rubber monoculture, services is a central challenge of PES schemes. which depletes environmental services (56). In some cases, elimi- When services are linked to an excludable good, such as (in some nating an existing subsidy on an environmentally bad behavior cases) drinking water, beneficiaries buy the service directly through might be as environmentally effective as creating a new incentive- the market. In other cases, where ecosystem services are nonex- based policy and might create fewer other distortions (33). cludable, such as climate stabilization or biodiversity, there will always be an incentive for the beneficiaries to not pay and to Nongovernmental Actors Will Be More Effective Where They Comple- ‘‘free-ride,’’ or hope to free-ride, on the benefits provided by others. ment Government Institutions. Although PES polices will almost This is a lesson stressed by economists: people are unlikely to pay certainly achieve better results in places with well functioning civil for something they can receive for free. It suggests that voluntary institutions, PES schemes driven by non-state actors may be able to approaches, such as donations to NGOs or the purchase of carbon partially compensate for weak state institutions. For instance, NGOs can provide much of their own monitoring and enforcement credits on the Chicago Climate Exchange, are unlikely to generate capacity. As another example, for suppliers to be willing to modify funding close to the level at which the services are valued. their land use practices to engage in a PES initiative, they must For ecosystem services that are public goods, then, compulsory perceive security in their ability to receive compensation for the mechanisms for demand generation or government payments for modification. Where this security is not provided by state legal provision will be necessary to overcome free-riding. Transferable institutions, it may be provided through informal institutions. PES development rights are one innovative way to pay for services (e.g., schemes may be able to take advantage of existing cooperative ref. 31). Under this approach, the developer pays to set aside land agreements between local communities, as examples from Bolivia in one location, and in exchange the government allows more suggest (34). Gaining trust through a participatory process may help intensive development elsewhere. Governments can also tax ben- some PES schemes reduce long-term monitoring and enforcement eficiaries to raise money for PES policies, although who is targeted costs and promote equity outcomes (35, 36). by the tax will clearly have distributional and political economy implications. Context Dynamics As environmental, socioeconomic, and political contexts change Although PES Schemes Can Be More Cost-Effective Overall, Their over time, the signals created by incentive-based mechanisms will Political Feasibility Depends on the Political Power of Those Who Bear also change. Possible future changes should be taken into account the Costs and Benefits. Despite the potential cost-effectiveness of when designing PES policies because these dynamic changes in taxes and tradable permits for pollution control, the adoption of context can alter how a policy performs, determining whether it is incentive-based approaches has been slow. In the United States, able to maintain a high degree of cost-effectiveness, environmental freely allocated tradable permits have been an exception, likely effectiveness, and equity over time.j because all stakeholders prefer this policy approach. The share of the burden paid directly by industry is low; environmental groups By Changing Prices, Incentive-Based Policies May Unintentionally are more satisfied with the fixed cap on pollution provided by Enhance the Profitability of an Environmentally Harmful Activity, tradable permits, as opposed to a tax; politicians benefit from the Undermining Environmental Effectiveness. Incentive-based mecha- fact that the allocation of permits can be arranged to satisfy nisms work by changing relative prices, making environmentally influential players; and permits create a barrier to market entry, beneficial activities more profitable and environmentally harmful protecting existing firms (29). activities more costly. However, the subsidy-like structure of many Ecosystem service providers are likely to prefer a PES policy over PES schemes carries with it many of the problems characteristic of traditional regulation because a PES approach offers compensation subsidies (38, 39). The pollution control literature has demon- for environmental improvements, and participation is voluntary. strated theoretically that a subsidy that provides firms with incen- However, although a PES approach is likely to be favored by the tives not to pollute could also make it more profitable in the long landholders eligible for payments, overall viability will be deter- run for some firms to enter the industry or to stay in the industry mined by the preferences and power of all relevant stakeholders, when they otherwise would not have (40). These firms produce including beneficiaries of the ecosystem service, policymakers, additional pollution, reducing the environmental effectiveness of financiers, community members, and program administrators. A the subsidy. nongovernmental entity may have a better chance at implementing Similarly, the additional environmental benefits provided by a a PES scheme because the funding comes from outside and relieves PES scheme may be compromised by new entry or other responses local communities and taxpayers. Preferences for the policy may be to subsidies over time. Paying farmers to keep land in forest on some determined not just by economic interests, but also by broader plots might increase the profitability of farming, leading to the concerns. For example, some PES projects in Bolivia met with clearing of additional plots (41, 42). Or, if landowners are credit- opposition both from those who saw them as limiting future constrained, receiving cash payments for good behavior on one economic development and from those who saw them as a privat- parcel of land may provide the income needed to begin an environmentally harmful use on another. To the extent that PES ization of nature (8). Political considerations are also likely to programs are small and do not change regional prices, or if there change the shape of policies during implementation. For example, is a fixed-factor of production, then this type of slippage or new in Mexico’s Payments for Hydrological Services program (32), funding targets were shifted away from key overexploited watersheds toward broader coverage, to more widely distribute program jOne approach to addressing the challenges presented by context dynamics has been benefits. discussed in the literature on adaptive management (37).

9468 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705503104 Jack et al. entry is less likely to occur (40). These secondary effects must be Price Changes That Increase the Overall Costs of the Policy Will Have taken into account when trying to measure the environmental Distributional Consequences and Could Compromise the Environmen- SPECIAL FEATURE benefits gained as a result of PES policies. tal Effectiveness of the Program. Prices could change in a way that Incentive-based mechanisms can also create the conditions for makes the costs of providing the environmental good more expen- ‘‘ransom behavior’’: threats or undesirable actions aimed at lever- sive with any possible method; the ultimate effects of such a change aging additional compensation (43). If pollution reduction require- depend on the structure of the policy. In a system of pollution taxes, ments for firms are assigned relative to a baseline, firms may an overall increase in the cost of abating pollution would lead to less deliberately increase pollution emissions to manipulate baseline pollution control as more firms prefer to pay taxes rather than emissions. Ransom behavior is also a major concern for PES abate. In China, pollution levies on industry have decreased in programs. The problem of ransom can be alleviated by basing effectiveness as the value of industrial output has increased, while policies on a clear historical baseline or by basing incentives on charges remain constant (51). In a system of tradable permits with levels of activities rather than on changes. Providing incentives for a fixed cap, when the cost of abatement goes up, the price of the levels, however, may create tradeoffs between avoiding ransom permits rises and firms bear a higher cost, but the total amount of behavior and paying landholders for activities that might have pollution control remains the same. occurred in the absence of the program (44). Changes in prices over time, particularly for agricultural goods, can have similar effects on PES schemes. Increases in agricultural Incentive-Based Policies That Encourage Innovation Will Be More output prices raise the opportunity cost of keeping land in natural Cost-Effective over Time. Incentive-based mechanisms have the vegetation. Both the budgetary costs (to the organization) and the potential to provide an incentive for firms to look for and adopt new true costs (the opportunity cost) may increase beyond original technologies that will lower the cost of protecting the environment expectations. The distributional and environmental effects depend in the long run. For instance, under a system of taxes or tradable on how the PES program is structured: if landowners are locked permits, an innovation that lowers the cost of abatement will into long-term contracts, then the environmental goal may be met, produce cost savings for the firm (45). Command-and-control but landowners will bear the increase in costs. If contracts are regulation, on the other hand, does not usually reward firms for short-term, then a budget increase may be necessary to sustain the reducing emissions beyond the target and, therefore, fosters less environmental effectiveness of the project. Private-sector pressures innovation (46, 47).k Existing literature demonstrates that innova- on the land also represent a distinctive threat. If timber companies SCIENCES tion and investment in new technology are most likely to occur when or oil palm plantations offer to buy a village’s land, even the ECONOMIC rewards are tied to marginal improvements in environmental best-designed PES scheme may be unable to compete with changes impacts and when flexibility is allowed in techniques for reducing in opportunity cost of this magnitude (52). pollution and in the timing of reductions, allowing firms to choose from a wider set of possible abatement options. The extent to which Conclusions innovation occurs is also likely to depend on agents’ perceptions of PES policies represent a growing trend in conservation policy. By the longevity of the incentive instrument (16). altering private incentives to induce desired outcomes, PES PES can also offer incentives to adopt or invent innovative schemes offer a direct, and possibly more equitable, method for approaches to providing ecosystem services at lower cost. However, achieving environmental outcomes than other approaches. How- because most PES policies base rewards on proxy actions rather ever, the context in which a PES initiative is implemented matters than on production of final ecosystem services, the incentive to greatly for effective policy design and the achievement of stated innovate may not be as direct. For example, the RUPES project in goals. We have argued that insights into how context matters can be Indonesia bases rewards to farmers on erosion-control activities on carried over from the existing literature on incentive-based ap- coffee farms, not on sedimentation loads in nearby streams. This proaches to environmental policy and applied to PES policy design type of system provides incentives to innovate over activities but and implementation. does not encourage innovative approaches for further reductions in The importance of context in achieving policy goals emphasizes sediment loads. Allowing flexibility in methods by basing rewards that no single policy is right for every scenario. Previous experience on reductions in sediment loads would encourage additional inno- with incentive-based approaches suggests it is unlikely a PES vation but would be more expensive to monitor and would force approach will always be able to simultaneously improve livelihoods, landowners to bear the risk that a given activity might not actually increase ecosystem services, and reduce costs. Potential tradeoffs reduce sediment loading. among these goals can be assessed reasonably well by considering the correlation between characteristics of poor landholders and Allowing Multiple Ways to Comply with an Incentive-Based Approach their land, characteristics of the costs and benefits of providing Will Increase Resilience to Price Changes That Affect the Production of ecosystem services, and the political feasibility of various policy Environmental Quality. Just as flexibility in methods for achieving options. environmental objectives can promote innovation, it can also allow The lessons also suggest other areas in which additional research firms to adapt to changes in prices, usually of inputs or technologies, is needed. Several PES projects that have been running in devel- which affect the cost of a particular method for pollution control oping countries for some time are starting to offer provocative (5). Similarly, when PES policies offer many ways of achieving findings about the use of PES mechanisms (53, 54). However, new service provision, participants will be more likely to withstand projects will only be able to learn from the successes and failures of changes in the relative prices of technologies. For instance, if many their predecessors if the manner in which outcomes relate to the different types of vegetation can be used for buffer strips, and there environmental, socioeconomic, and political contexts of the policy is an increase in the price of one species, landowners can switch to are systematically documented and compared across a range of a cheaper alternative and continue to provide the service. By cases.l With more long-run experience, rigorous program evalua- allowing a variety of ways to provide the same ecosystem service, tion will provide additional understanding of the effectiveness of either by increasing the range of allowable proxies or by directly different policy designs over time (55), as well as information on rewarding the ultimate service, participants are able to switch away how PES schemes respond to exogenous shocks. Collaborations from more expensive approaches in the face of price increases. between ecologists and economists can better specify the produc-

kOn pollution abatement and the inducement of technological change, see refs. 45 and 47. lSeveral inventories of PES schemes are under way, including efforts by The Natural Capital On technological change related to agriculture and natural resources, see refs. 48–50. Project and the Organization of American States.

Jack et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9469 tion function for ecosystem services. This information will improve Matson, Nancy Dickson, Darby Jack, and Jonathan Borck; members of the the design of input proxies and reduce the uncertainty surrounding Sustainability Science Program at Harvard University; and seminar participants at the 2006 Harvard–Stanford Ecosystem Services meeting, the 2007 RUPES environmental effectiveness. More research is also needed on how conference, and the 2007 Yale International Society of Tropical Foresters incentive-based mechanisms can account for potential tradeoffs meeting. Core funding was provided by a grant from The Ash Institute for and synergies in the production of multiple ecosystem services. Democratic Governance and The Alfred A. Taubman Center for State and Local Government at the John F. Kennedy School of Government, Harvard Additional analysis of large-scale PES policies can help us to University. Support is also acknowledged from the Sustainability Science understand the broader effects on the economy from scaling-up Program at the Center for International Development, the Kennedy School’s PES schemes (12, 13). Norberg–Bohm Fellowship (B.K.J.), the Teresa Heinz Scholar for Environmen- tal Research Grant (C.K.), and a National Science Foundation Graduate Re- search Fellowship (K.R.E.S.). The authors are Research Fellows in the Sustain- ACKNOWLEDGMENTS. The authors gratefully acknowledge inputs from two ability Science Program, Center for International Development, Harvard anonymous reviewers and from William Clark, Robert Stavins, G.C.D., Pamela University.

1. Daily GC (1997) Nature’s Services: Societal Dependence on Natural Ecosystems (Island 29. Keohane NO, Revesz R, Stavins RN (1998) Harvard Environ Law Rev 22:313–367. Press, Washington). 30. Oates WE, Portney PR (2003) in Handbook of Environmental Economics, eds Ma¨ ler KG, 2. Millennium Ecosystem Assessment (2005) Millennium Ecosystem Assessment Synthesis Vincent J (North–Holland/Elsevier Science, Amsterdam), Vol 1, pp 325–354. Report (Island Press, Washington). 31. McConnell V, Kopits E, Walls M (2003) How Well Can Markets for Development Rights 3. Boyd JW, Banzhaf HS (2005) Resources 158:16–19. Work? Evaluating a Farmland Preservation Program (Resources for the Future, Wash- 4. Heal G (2000) Nature and the Marketplace (Island Press, Washington). ington), Discussion Paper 03-08. 5. Stavins R (2003) in Handbook of Environmental Economics, eds Ma¨ ler K-G, Vincent JR 32. Alix-Garcia J, De Janvry A, Sadoulet E, Torres JM (2005) An Assessment of Mexico’s (Elsevier Science, Amsterdam), Vol 1, pp 355–435. Payment for Environmental Services Program (Comparative Studies Service, Agricul- 6. Panayotou T (1994) Economic Instruments for Environmental Management and tural and Development Economics Division, FAO, Rome). Sustainable Development (UN Environment Program/Nairobi, Geneva), Environmen- 33. Anderson K, McKibben WJ (2000) Environ Dev Econ 5:457–481. tal Economics Paper No 16. 34. Wunder S, Vargas MT (2005) Beyond ‘‘Markets’’: Why Terminology Matters (Katoomba 7. Wunder S (2007) Conserv Biol 21:48–58. Group’s Ecosystem Marketplace). Available at http://ecosystemmarketplace.net/pages/ 8. Robertson N, Wunder S (2005) Fresh Tracks in the Forest: Assessing Incipient Payments article.opinion.php?component࿝idϭ1252&component࿝version࿝idϭ3597&language࿝idϭ12. for Environmental Services Initiatives in Bolivia (Center for Intl Forestry Res, Jakarta, 35. McCann L, Colby B, Easter KW, Kasterine A, Kuperan KV (2005) Ecol Econ 52:527–542. Indonesia). 36. Birner R, Wittmer H (2004) Environ Plan C Gov Policy 22:667–685. 9. Landell-Mills N (2002) Philos Trans R Soc London Ser A 360:1817–1825. 37. National Research Council (2004) Adaptive Management for Water Resources Project 10. Pagiola S (2008) Ecol Econ 65:712–714. Planning (Natl Acad Press, Washington). 11. Mun˜ oz-Pin˜aC,et al. (2008) Ecol Econ 65:725–736. 38. Baumol WJ, Oates WE (1988) The Theory of Environmental Policy (Cambridge Univ 12. Xu Z, Xu J, Deng X, Huang J, Uchida E, Rozelle S (2005) World Dev 34:130–148. Press, Cambridge, UK). 13. Sullivan P, Hellerstein D, Hansen L, Johansson R, Koenig S, Lubowski R, McBride W, 39. Bramhall DF, Mills ES (1966) Water Resour Res 3:615–616. McGranahan D, Roberts M, Vogel S (2004) The Conservation Reserve Program: Eco- 40. Feng H, Kling CL, Kurkalova LA, Secchi S (2003) Subsidies! The Other Incentive-Based nomic Implications for Rural America (Economic Research Service, US Dept of Agri- Instrument: The Case of the Conservation Reserve Program (Center for Agricultural culture, Washington), Agricultural Economic Report No AER834. and Rural Dev, Iowa State Univ, Ames). 14. Alix-Garcia J, de Janvry A, Sadoulet E (2008) Environ Dev Econ, in press. 41. Wu J (2000) Am J Agric Econ 82:979–992. 15. Pagiola S, Arcenas A, Platais G (2005) World Dev 33:237–253. 42. Roberts MJ, Bucholtz S (2005) Am J Agric Econ 87:244–250. 16. Hahn RW, Stavins RN (1992) Am Econ Rev 82:464–468. 43. Salzman J (2005) NY Univ Law Rev 80:870–961. 17. Mendelsohn R (1986) J Environ Econ Manage 13:301–312. 44. Wunder S (2005) Payments for Environmental Services: Some Nuts and Bolts (Center 18. Scheffer M, Carpenter S, Foley JA, Folke C, Walker B (2001) Nature 413:591–596. for Intl Forestry Res, Jakarta, Indonesia), CIFOR Occasional Paper No 42. 19. Arrow K, Daily G, Dasgupta P, Levin S, Maler K-G, Maskin E, Starrett D, Sterner T, 45. Jaffe A, Stavins R (1995) J Environ Econ Manage 29:S43–S63. Tietenberg T (2000) Environ Sci Technol 34:1401–1405. 46. Jaffe AB, Newell RG, Stavins RN (2002) Environ Res Econ 22:41–69. 20. Pagiola S, Agostini P, Gobbi J, de Haan C, Ibrahim M, Murgueitio E, Ramı´rez E, Rosales 47. Jaffe AB, Newell RG, Stavins RN (2003) in Handbook of Environmental Economics, eds M, Ruı´z JP (2004) Paying for Biodiversity Conservation Services in Agricultural Land- Maler K-G, Vincent JR (North–Holland/Elsevier Science, Amsterdam), Vol 1, pp 461–516. scapes (World Bank Environ Dept, Washington), Environment Dept Paper No 96. 48. Sunding D, Zilberman D (2001) in Handbook of Agricultural Economics, eds Gardner 21. Eigenraam M, Strappazzon L, Lansdell N, Ha A, Beverly C, Todd J (2006) EcoTender: BL, Rausser GC (North–Holland/Elsevier Science, Amsterdam), Vol 1, pp 207–261. Auction for Multiple Environmental Outcomes (Dept of Primary Industry, Victoria, 49. Zilberman D, Khanna M, Lipper L (1997) Aust J Agric Res Econ 41:63–80. Australia). 50. Ruttan VW (2001) Technology, Growth, and Development: An Induced Innovation 22. US Environmental Protection Agency (2004) International Experiences with Economic Perspective (Oxford Univ Press, New York). Incentives for Protecting the Environment (Natl Center for Environmental Economics, 51. U.S. Environmental Protection Agency (2004) International Experiences with Economic Washington). Incentives for Protecting the Environment (Natl Center for Environmental Economics, 23. Newell RG, Stavins RN (2003) J Regul Econ 23:43–59. Washington). 24. Ferraro P (2007) Asymmetric Information and Contract Design for Payments for 52. Engel S, Palmer C (2006) For Policy Econ 8:434–446. Environmental Services (Andrew Young School of Policy Studies, Georgia State Univ, 53. Rosa H, Kandel S, Dimas L (2003) Compensation for Environmental Services and Rural Atlanta). Communities: Lessons from the Americas and Key Issues for Stregthening Community 25. Zilberman D, Lipper L, McCarthy N (2006) When Are Payments for Environmental Services Strategies (Prisma, San Salvador). Beneﬁcial to the Poor? (FAO, Rome), ESA Working Paper No 06-04. 54. Sanchez-Azofeifa GA, Pfaff A, Robalino JA, Broomhower JP (2007) Conserv Biol 26. Kosoy N, Martinez-Tuna M, Muradian R, Martinez-Alier J (2007) Ecol Econ 61:446–455. 21:1165–1173. 27. Kerr J, Meinzen-Dick R, Pender J, Suyanto S, Swallow B, Noordwijk MV (2005) Property 55. Ferraro PJ, Pattanayak SK (2006) PLoS Biol 4:482–488. Rights, Environmental Services, and Poverty in Indonesia (BASIS Collaborative Re- 56. Kartodihardjo H, Supriono A (2000) The Impact of Sectoral Development on National search Support Program/USAID, Univ of Wisconsin, Madison), BASIS Brief No 29. Forest Conversion and Degradation: The Case of Timber and Tree Crop Plantations in 28. Horan RD, Shortle JS (1999) Environ Res Econ 14:191–215. Indonesia (Center for Intl Forestry Res, Jakarta, Indonesia).

9470 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705503104 Jack et al. An operational model for mainstreaming ecosystem SPECIAL FEATURE services for implementation

Richard M. Cowling*†, Benis Egoh‡, Andrew T. Knight*, Patrick J. O’Farrell§, Belinda Reyers§, Mathieu Rouget¶, Dirk J. Rouxʈ, Adam Welz**, and Angelika Wilhelm-Rechman*

*Department of Botany, Nelson Mandela Metropolitan University, P.O. Box 77000, Port Elizabeth 6031, South Africa; ‡Centre for Invasion Biology, Department of Botany and Zoology, Stellenbosch University, Private Bag X1, Matieland 7602, South Africa; §Natural Resources and the Environment, Council for Scientiﬁc and Industrial Research, P.O. Box 320, Stellenbosch, 7599, South Africa; ʈNatural Resources and the Environment, Council for Scientiﬁc and Industrial Research, P.O. Box 395, Pretoria 0001, South Africa; ¶South African National Biodiversity Institute, Private Bag X101, Pretoria 0001, South Africa; and **Department of Botany, University of Cape Town, Rondebosch 7701, South Africa

Edited by Gretchen C. Daily, Stanford University, Stanford, CA, and approved November 29, 2007 (received for review July 12, 2007)

Research on ecosystem services has grown markedly in recent science (11), ecosystem service research needs to be user- years. However, few studies are embedded in a social process inspired, user-useful, and user-friendly. Although research-for- designed to ensure effective management of ecosystem services. implementation models exist for integrated natural resource Most research has focused only on biophysical and valuation management (7) and conservation planning (5), we know of no assessments of putative services. As a mission-oriented discipline, article that spells out pragmatically and comprehensively the ecosystem service research should be user-inspired and user- process for achieving the safeguarding of ecosystem services on useful, which will require that researchers respond to stakeholder the ground. Our article seeks to fill this gap. needs from the outset and collaborate with them in strategy To provide a real-world context, we have chosen to focus on development and implementation. Here we provide a pragmatic the internalization, or ‘‘mainstreaming’’ (12), of ecosystem ser- operational model for achieving the safeguarding of ecosystem vice concerns into the land-use (and water-use) planning sector. services. The model comprises three phases: assessment, planning, Land-use planning is a normative discipline (4) in the sense that and management. Outcomes of social, biophysical, and valuation it provides the legally entrenched norms and rules for making assessments are used to identify opportunities and constraints for decisions about how natural resources are to be used. In many implementation. The latter then are transformed into user-friendly parts of the world, governments are institutionally obliged to products to identify, with stakeholders, strategic objectives for iteratively conduct participatory, spatially explicit, land-use plan- implementation (the planning phase). The management phase ning aimed at integrating requirements for social, economic, and undertakes and coordinates actions that achieve the protection of environmental sustainability. Flaws notwithstanding (13), this ecosystem services and ensure the flow of these services to process provides a window of opportunity for mainstreaming beneficiaries. This outcome is achieved via mainstreaming, or ecosystem services into the activities of organizations that are incorporating the safeguarding of ecosystem services into the SCIENCE empowered to make routine decisions about the use of land and policies and practices of sectors that deal with land- and water-use SUSTAINABILITY planning. Management needs to be adaptive and should be insti- water resources (14, 15). tutionalized in a suite of learning organizations that are represen- We restrict ourselves to ecosystem services—defined as the tative of the sectors that are concerned with decision-making and end products of nature that benefit humans (16)—provided by planning. By following the phases of our operational model, natural and semi-natural habitats (wild nature). Thus, we do not projects for safeguarding ecosystem services are likely to empower consider agriculture or aquaculture ecosystems, acknowledging, stakeholders to implement effective on-the-ground management of course, that wild nature does provide services essential for the that will achieve resilience of the corresponding social-ecological success of these ecosystems. First, this article provides some systems. background on mainstreaming, a relative newcomer to the biodiversity lexicon. The second and substantive part provides a adaptive management ͉ land-use planning ͉ social–ecological systems ͉ pragmatic, operational model for guiding the things we need to stakeholder engagement do for implementing the safeguarding of ecosystem services. Our account draws on our collective experience over the past decade here has been an impressive growth in research on ecosystem in user-inspired research and implementation in the nature Tservices in recent years. However, few studies are embedded conservation and water sectors (e.g., refs. 14 and 17–20). in a social process designed to ensure effective on-the-ground What Do We Mean by Mainstreaming? management of areas that deliver ecosystem services. It is unlikely that the outcomes of technically sophisticated assess- In the context of natural resource management and conserva- ments published in scientific journals will lead to implementation, the objective of mainstreaming is to internalize the goals for tion via a ‘‘trickle-down’’ effect (1–3). As a mission-oriented, safeguarding resources into economic sectors and development pragmatic discipline (4), ecosystem service research should be models, policies, and programs, and therefore into all human geared for implementation, and scientists should assist this behavior (12). The concept is entrenched in several articles of the process by responding to stakeholder needs from the outset and Convention on Biological Diversity and is the explicit objective by becoming involved in the messy process of collaborating with of the Global Environmental Facility’s GEF-4 program, with its and empowering stakeholders in strategy development and particular emphasis on ecosystem services. implementation (1, 5–7). How to do this is the topic of this article. There are some excellent examples of research that have Author contributions: R.M.C. designed research; and R.M.C., B.E., A.T.K., P.J.O., B.R., M.R., resulted in the protection of ecosystem services (e.g., refs. 8–10). D.J.R., A.W., and A.W.-R. wrote the paper. But they are few and are cited repeatedly in the literature. Our The authors declare no conflict of interest. wish is that ecosystem service research does not become another This article is a PNAS Direct Submission. bandwagon driven by technological sophistication and charac- †To whom correspondence should be addressed. E-mail: [email protected]. terized by societal irrelevance. As a cornerstone of sustainability © 2008 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706559105 PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9483–9488 Project Phase Assessment Phase. The assessment is a structured process that

Assessment Planning Management provides knowledge useful for policies, strategies, and management but does not prescribe these (5, 14, 26). The assessment Empowered Implementation Resilient Mainstreaming seeks to answer questions inspired by the beneficiaries and

Learning social-ecological system Organization managers of ecosystem services; in our situation, it must provide Adaptive knowledge useful for mainstreaming ecosystem services into Management Strategy Status of local land-use planning. We identify three types of assessment: Opportunities social, biophysical, and valuation (Fig. 1). Involved and A key requirement of the assessment phase is the establishment Constraints Social of multidisciplinary and multisector teams (5, 27, 28). Although Assessment teams engage in empirical research, their activities should be Valuation Stakeholder collaboration coordinated by the goals defined for ecosystem services research, Biophysical which, in turn, are defined by the requirements of land-use plan- Informed Assessment Vulnerable ning, a normative discipline (4). This hierarchy of coordination Regional Local provides operational meaning to the notion of ecosystem service Spatial Scale research as a truly interdisciplinary activity (4). Teams should include researchers from the natural and social sciences and the humanities; scientists and managers from the natural resource management (water, fisheries, agriculture, for- Fig. 1. An operational model for implementing the safeguarding of eco- estry, conservation, etc.) and human well being (health, social system services. development, safety and security, land-use planning, etc.) sectors; and nongovernmental and other citizen-based organizations. In addition to data collected by using standard scientific Based on South African experience, there are four elements methods, assessment teams also should record tacit (or implicit) of a framework for achieving: mainstreaming: (i) prerequisites, and traditional knowledge because a great deal of useful infor- elements without which mainstreaming cannot happen; (ii) mation is associated with these informal systems (29, 30). stimuli (or windows of opportunity), elements external and Teamwork is both difficult and rewarding. It requires emo- internal to the sector that catalyze awareness of the need for tionally intelligent leadership, which is rare. There may be mainstreaming; (iii) mechanisms, the actual activities that seek confusion and contention about values assigned to nature: to effect mainstreaming; and (iv) outcomes, the measurable conservationists typically view nature as axiomatically ‘‘good,’’ indicators of mainstreaming effectiveness (20). The most fre- whereas other stakeholders perceive the value of nature in a quently cited prerequisites in these projects were democratic and relative sense (31). This kind of confusion needs to be managed accountable governance, awareness and knowledge, and orga- by effective leaders, as do power asymmetries and concealed nizational and institutional capacity. agendas. However, if properly managed, teamwork provides Mainstreaming is achieved primarily through behavior excellent opportunities for rapid, collaborative learning based on change. In the context of this article, it requires that the both explicit and tacit knowledge and for challenging or chang- safeguarding of ecosystem services is institutionalized in land- ing deeply entrenched world views or mental models (3, 30). use planning policies and is reflected in the day-to-day activities Social assessment. The social assessment should precede the bio- of this sector. physical one (Fig. 1), because it identifies the owners and beneficiaries of ecological functions that actually deliver services Operational Model and, hence, require biophysical assessments. It also identifies Here we discuss operational issues: the things that need to be markets for ecosystem services and other incentives for their done to mainstream ecosystem services. Fig. 1 shows the three safeguarding, as well as individual, institutional, and governance phases of the model (assessment, planning, and management) barriers to implementation (32). and their relationships to spatial scale, degree of stakeholder The social assessment should provide knowledge on the needs, engagement, and status of the social-ecological system—the values, norms, and behaviors of individuals, institutions, and integrated and interactive relationships between humans and organizations in the study area. In other words, it provides an understanding of how an area works in socioeconomic terms and ecosystem services (21, 22). Our operational model is based on why. Without the understanding of the social system provided by one devised by Knight et al. (5) for conservation planning. Any the social assessment, implementation is likely to be poorly project for safeguarding ecosystem services should strive to targeted. Specific issues requiring research will vary with con- arrive at the top right-hand corner in Fig. 1, where the adaptive text; however, knowledge of the spatial patterns of population management of the social-ecological systems associated with the density, human needs (for example, subsistence, protection, and defined suite of ecosystem services has been mainstreamed into identity), income distribution, current and future trends in land an appropriate land-use planning framework and governed by use, land prices, infrastructure, the social capital of natural learning organizations that are representative of, and supported resource management organizations, nature-related values, pref- by, the full range of stakeholders in the study area (a learning erences and ethics, and incentives for behavior change are likely organization is one skilled at creating and acquiring knowledge to emerge as important topics in most cases (26, 33–36). Wher- and modifying its behavior to reflect new insights) (23). Thus, ever feasible, data need to be captured spatially and matched to stakeholders are empowered to implement effective on-the- the scale used in the biophysical assessment (37). ground management of ecosystem services, and social-ecological Social assessments take time and can be costly. Adequate systems are resilient (they can absorb shocks and surprises) (1, budgets should be secured (38, 39). 22, 24). Getting there is a social process riddled with complexity, Biophysical assessment. Biophysical assessment provides knowl- contention, uncertainty, surprise, disappointment, and triumph. edge on the types and location of the biophysical features that It will take a long time—in many cases, decades—to achieve this provide ecosystem services, the spatial and temporal flows of goal (25). Below we describe some elements of this pathway to services in relation to beneficiaries, and the impacts of land and resilience by outlining the key actions associated with each of its water transformation on delivery (e.g., refs. 40–44). Heal (33) major phases: assessment, planning, and management. makes the important point that it is the biophysical rather than

9484 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706559105 Cowling et al. the valuation assessment that provide the knowledge-based case informed citizens. We recommend, where circumstances permit, for safeguarding services. For example, a simple model that encouraging stakeholders to reach consensus on assigning sub- SPECIAL FEATURE predicts the reduction of water supply below the projected jective values to ecosystem services. Such discourse-based ap- demand as a consequence of unchecked devastation by humans proaches (54, 55) enable social influence and consensus to define of a watershed (e.g., refs. 8 and 9) is likely to provide a more knowledge about the value of ecosystem services. As Starbuck compelling case to stakeholders for protecting the watershed (3) states: ‘‘Acceptance by people is crucial, because knowledge than dubious estimates of the reduction in the aggregated is what people say it is.’’ monetary value of all of the watershed’s goods and services Identify opportunities and constraints for implementation. The conclud- (most of which have no market value). ing stage of the assessment phase is a structured process in which Other than research on the links between biodiversity and all project participants identify opportunities and constraints for ecosystem services (45), very little research has been done on the implementing actions to safeguard ecosystem services. Because of ecology of ecosystem services. Kremen (40) provides a useful different value systems, research traditions, and mental models, this operational framework for studying ecological aspects of eco- process can be difficult (38, 39, 56); it requires excellent facilitation system services that we do not repeat here other than to and leadership. We cannot overstate the importance of this phase: reiterate—as others have done (e.g., refs. 43 and 46)—the the outcomes provide the bridge between assessment and planning importance of measuring the spatiotemporal scales over which by providing knowledge essential for identifying strategic imple- services operate. mentation objectives (Fig. 1). Of the published biophysical assessments, most focus on Identifying opportunities and constraints can be challenging mapping services, their flows, and the impacts of habitat trans- because of the complex outcomes of the three assessments. formation on these flows (e.g., 41–44). A feature of many studies There is a need to frame and depict in ways that harmonize with is the identification and mapping of natural features that have no stakeholders’ values, needs, and cognitive skills, the complexity direct beneficiaries or markets and in whose protection few of outcomes characterized by situations of ‘‘numerous possible people have an obvious interest. Invariably, these studies are not futures underpinned by numerous possible solutions’’ (57). Most user-inspired and lack social assessments for identifying the suite stakeholders are likely to lack the cognitive capacity to compre- of services that fulfill social needs, both presently and potentially. hend and absorb the significance of models that depict dynamic, In short, without beneficiaries, there are no services. long-term, continuous, and multiscale processes with complex An important component of the biophysical assessment is the feedback and uncertain outcomes (3, 29, 58). They find it much development of dynamic models of landscape change—the easier to relate to models that are described by discrete events, spatially explicit depiction of alternative futures (38, 47). These possibilities, pictures, emotions, and stories, and that provide products allow stakeholders to envision the consequences of prospects for harnessing their energies and skills (59–62). particular policy frameworks regarding land and water use. Scenario planning is one way that the assessment team can However, they need to be interpreted visually and depicted as display implementation opportunities and constraints in a man- plausible scenarios that stakeholders comprehend. We return to ner that is comprehensible to a broad range of stakeholders. This this very important point later on in this article. powerful tool deals with uncertainty by providing plausible, Valuation assessment. Just as the study of the nature-related values descriptive narratives or pathways to the future. Scenario plan- SCIENCE

(beliefs) that people hold is contentious, so is the study of the ning has a long history in business science where it has been used SUSTAINABILITY values they assign (in monetary or ranking terms) to nature (48). to challenge mental models, facilitate behavior change, promote There is a large and growing literature on the conceptual, collaborative learning, and confront tradeoffs (56). It also has technical, and operational aspects of the economic valuation of been used to good effect in the natural resource sector (7, 63, 64) ecosystem services (e.g., refs. 49–51). We do not review typolo- and was adopted by the Millennium Assessment (65, 66). In the gies and techniques for economic valuation but rather focus on context of our model, scenario assumptions are defined by several issues that are relevant for a socially engaged valuation implementation opportunities and constraints, and these are assessment. used to set parameters for spatially explicit models of alternative The valuation assessment is located at the intersection of the futures that can be depicted as maps and visual narratives (38). social and biophysical assessments (Fig. 1) and should be in- Scenarios can be especially effective when they capture alterna- formed by these (46). Most studies advocate, albeit with caution, tive futures visually and dramatically, in such a way as to reduce the monetary valuation of ecosystem services because ‘‘most stakeholder confusion by providing clarity about complex issues societies have an intuitive notion of economic value’’ (51) and it and vague language (67, 68). By providing compelling, positive provides ‘‘a metric than can be deployed across competing land alternatives to the status quo, scenarios can harness stakehold- uses’’ (52). Monetary valuation can be particularly effective in ers’ energies for strategy development and, thereby, overcome enabling informed tradeoffs in cost–benefit analyses, where the their sense of helplessness about the future (69–72). focus is on assessing the marginal change in the provision of an ecosystem service that has market value (e.g., amount of water Planning Phase. The second phase of the operational model is produced) relative to a competing land use that also is traded on planning, which is explicitly collaborative, involving all key the market (e.g., real estate) (33, 52). stakeholders, including researchers (Fig. 1). Collaborative plan- But the vast majority of services have no market price (33, 41, ning is a discourse-based process that comprises the identifica- 46, 52). To paraphrase Simpson (49), prices are not to be tion of a vision, a strategy to realize this vision, specific strategic confused with values, and prices are not the only values that are objectives, and instruments, tools, and organizations for imple- important. Nonmonetary units of value also can be used, for menting actions to achieve the objectives. example, cubic meters of clean water, jobs created, and lives Strategy development. The overall aim of this stage of the planning saved (27). Because money is the most commonly used inter- process is to collaboratively identify a set of strategic objectives and changeable commodity, valuation in monetary terms may send specific actions for the safeguarding of ecosystem services. These the message that a service is more easily replaced by human- objectives should seek to exploit the implementation opportunities manufactured providers than it actually is (53). and overcome the constraints identified in the assessment phase. Throughout the world, land-use decisions are seldom made on Scientists need to develop and present at the strategy workshops the basis of the outcomes of economic valuation studies; they products (for example, visually compelling scenarios and maps) that usually are made by officials and politicians—many of whom are are user-useful and user-friendly (5, 14). poorly informed—or, in functional democracies, by variously Strategy development is essentially a process for learning (56,

Cowling et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9485 73)—an opportunity for nonexperts to gain an understanding of Third, increased awareness and knowledge about environ- the issues at stake and for experts to appreciate the concerns and mental concerns, and even embracing pro-nature values, does contributions of other stakeholders, including decision makers not necessarily translate into adopting pro-nature behavior (62, and the socially marginalized. The involvement of nonexperts 86). Therefore, pragmatic solutions are required to overcome also is an important opportunity to engender pro-nature behav- the inertia in engendering pro-nature behaviors of individuals ior change: appropriately framed information and involvement and organizations that are required for mainstreaming. Social in a process of developing a strategy to achieve a mutually marketing is very promising in this respect: rather than attempt- desired state—the vision—can rapidly change people’s norms ing to understand the complex causes of behavior, it takes (62, 74, 75). It forces them to confront realities about unsus- existing behaviors as a given and then seeks to identify the tainable futures that will be harmful to themselves and their barriers to behavior change and to design specific incentive- offspring and to contribute by exploring possible solutions to based programs to overcome these barriers (86, 87). Incentives these problems (72, 75–77). relate to both internal barriers (e.g., absence of skills, opposing In the strategy process, scientists are enablers (5, 26). They values, and beliefs) and external barriers (e.g., inadequate need to frame issues clearly and communicate in simple and infrastructure and support). Social marketing has been ex- accessible language the benefits and costs of particular actions tremely successful in achieving behavior change in the health, and their associated uncertainties (58, 67, 78–80). Their role is social development, and waste management sectors but has yet to help stakeholders understand issues so as to avoid confusion to penetrate natural resource management and conservation and overcome helplessness (69, 71). sectors. Depending on the outcome of the assessment of gover- The strategic objectives need to be an unambiguous and nance and institutional capacities, it may be necessary to imple- tractable list of actions and behaviors that are clearly linked to ment programs of social marketing to bring about rapidly the instruments for implementation, which are supported by appro- desired levels of behavior change. priate institutions (5, 81). The instruments available will be Mainstreaming is an ongoing process that needs to be respon- context-specific and, because many instruments are complemen- sive to windows of opportunity and other unintended surprises tary, they should be identified as an optimal mix (5). They may arising from, among others, market emergence, infrastructure include financial incentives (e.g., direct and indirect payments development, and political changes and associated shifts in for service delivery), governance-based instruments (e.g., en- power regimes (20, 26). forcement of existing legislation, capacity-building, and the establishment of cooperative governance structures), and value- Management Phase. Management comprises the final phase of our based instruments (education and recognition) (28, 29, 38, 52, operational model for achieving resilience of the social- 82, 83). In the cases where markets exist for ecosystem services— ecological systems associated with ecosystem services. The over- for example, carbon sequestration, nature-based tourism, and all objective of this phase is to undertake and coordinate actions, water supply—institutions and organizations may need to be including additional research, that achieve the protection of established to capture the values of these (52, 83). biophysical features that provide ecosystem services and ensures Mainstreaming. Mainstreaming, the internalization of ecosystem the flow of services to beneficiaries. service safeguarding into the policies and practices of the Actions may include the implementation of social marketing land-use planning sector, is located at the interface of the projects, the restoration of vegetation for carbon credits, the planning and management phases of the operational model (Fig. 1). Optimal mainstreaming requires effective governance, orga- protection of watersheds key for water delivery, or the protection nizational and institutional capacity, and awareness of and a of viewsheds for nature-based tourism—it depends on what has comprehensive knowledge about the ecology and value of emerged from the assessment of implementation opportunities ecosystem services (20). The assessment and planning phases and constraints. We recommend, as others have done (1, 14), the provide knowledge about ecosystem services, increased aware- adoption of an adaptive management framework that embodies ness of the importance of these services among stakeholders an action-reflection cycle, or ‘‘learning by doing’’ (24, 30). In this (and may have already initiated a change in mental models or regard, the adoption of a quasi-experimental approach, whereby even behavior), and identify opportunities and constraints re- the effectiveness of interventions can be assessed relative to garding governance and capacity for implementation. situations where intervention is withheld (88), can be extremely The rationale, benefits, and mechanisms for safeguarding eco- effective in unraveling the complexities of social-ecological system services need to be mainstreamed into all of those sectors systems (14). Static products such as user-useful and user- that feed into land-use planning, e.g., water, forestry, agriculture, friendly maps of ecosystem services and guidelines for managing tourism, and urban planning. At least three things need to be them, which can be mainstreamed directly or via social market- considered when launching a mainstreaming initiative. ing into local integrative planning processes, potentially are very First, decision makers in all of the relevant sectors need to be useful (e.g., 14). made aware of the importance for sustaining society of safe- Adaptive management needs to respond effectively to the guarding ecosystem services and, where they exist, of their legal complex feedback, opportunities, and shocks that characterize mandates to do so, which is most effectively done by identifying social-ecological systems and provide insights that can be incor- ‘‘win–win’’ situations that address both natural resource and porated into the iterative processes of assessment and planning socioeconomic concerns (12). For example, this was done to (Fig. 1). Therefore, adaptive management needs to be institu- mainstream restoration projects in South Africa that delivered tionalized in a suite of learning organizations (5, 14, 28, 78), each on both ecosystem service and social equity goals (84, 85). focusing on a different ecosystem service. Such organizations Communication to decision makers must be effective (78, 80); it must be representative of the sectors that are concerned with often may be necessary to emphasize as compelling ‘‘sound land-use decision-making and planning and should foster a spirit bites’’ the immediate, social, and economic benefits of ecosystem of colearning, cogovernance, and accountability (22, 23, 56), service protection (59) rather than less certain benefits that may which is not always easy to achieve (19); key individuals and good only manifest in the longer term. leadership are of paramount importance for effective learning Second, new organizations and institutions will be required to organizations (22, 89). The learning organization should have address the tricky problem of coordinating governance across the authority to restrict access to ecosystem service providers, such a wide array of sectors (22). We discuss cooperative the wherewithal to offer incentives for their safeguarding, the governance in the next section of this article. capacity to monitor ecological and social conditions, the exper-

9486 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706559105 Cowling et al. tise to evaluate the outcomes of interventions, and sufficient priate for ecosystem service projects. Our operational model is flexibility to respond rapidly to changed circumstances (35, 57). a process that does have hallmarks for evaluation but is simply too complex and uncertain of outcomes to specify, with any SPECIAL FEATURE Conclusions and Caveats degree of realism, tangible outputs in short (1- to 5-year) At the core of our operational model are three elements: socially timeframes. relevant, user-inspired research, stakeholder empowerment, and The operational model specifies a process that engenders adaptive management embedded in learning organizations. The stakeholder collaboration and bottom-up decision-making, goal is the achievement of social and ecological resilience in an which is consistent with the notion that although most environ- uncertain world. mental problems are regional or global, the solutions are at the The activities prescribed by the model will not be easy to local and individual scales (94). However, there are many cases implement. Socially engaged, multi- and interdisciplinary re- where well intentioned, bottom-up projects fail because of search is relatively rare. Our process requires a fundamental failures of regional and global institutions to support their change, or transformation, in the way research generates knowl- outcomes (25, 95, 96). Bottom-up implementation needs to be complemented by the policies and practices of regional and edge (3, 4). Researchers will need to be responsive to stake- global trade and financial institutions (97). Of great importance holder needs, collaborate with many groups with values and is the incorporation of the value of ecosystem services into the norms foreign to their own, operate as facilitators of knowledge accounting systems of these institutions (16). transfer to stakeholders, and be prepared to engage time- Related to this is the need to project ecosystem services consuming processes that are not sympathetic to career aspira- research into the realm of transdisciplinarity by addressing tions and performance benchmarks predicated by the accumu- directly the values, ethics, and morals associated with individu- lation of publications in high-impact journals (7, 90). Moreover, als, organizations, and institutions (4, 92, 98). Do we want a the education philosophies of almost all universities are not world that promotes wealth accumulation and self-interest, or conducive to multi- and interdisciplinary research; instead, they one that fosters equity and common good? Questions such as encourage the atomization of disciplines and entrench the these raise issues about the kinds of economic systems we desire: boundaries between them (4, 58, 91). However, the recent ones based on perpetual growth or ones that strive for a steady emergence of sustainability science (11) is a very positive state (98, 99). Sadly, the prevailing consumerist economic par- development. The operational model presented here provides adigm, the high discount rates held by most humans, and their many opportunities for conducting research on the complex disconnect for the natural world (58, 69, 100) do not augur well problems inherent in managing social-ecological systems. Rec- for the radical transformations required to place the world on a ognition of the importance of this research through enhanced path to sustainability. Planning to ensure the persistence of funding and status can provide the impetus for its growth. ecosystem services in guaranteed to be an important and stim- Implementing the operational model for most projects will ulating challenge. take a lot of time (25) and incur large costs, especially transaction costs (92, 93). In developing countries, donor organizations fund ACKNOWLEDGMENTS. Kathy MacKinnon, Shirley Pierce, and Kent Redford provided useful comments on the manuscript. The National Research Foun- projects that are geared to specific deliverables subject to the dation, Council for Scientiﬁc and Industrial Research and the Nelson Mandela time-related tyrannies of log frames, which may not be appro- Metropolitan University provided ﬁnancial support. SCIENCE SUSTAINABILITY

1. Salafsky N, Margoluis R, Redford KH, Robinson JG (2002) Improving the practice of 19. Roux DJ, Rogers KH, Biggs HC, Ashton PJ, Sergeant A (2006) Bridging the science- conservation: a conceptual framework and research agenda for conservation science. management divide: Moving from unidirectional knowledge transfer to knowledge Conserv Biol 16:1469–1479. interfacing and sharing. Ecol Soc 11:4. 2. van Kerkhoff L, Lebel L (2006) Linking knowledge and action for sustainable devel- 20. Cowling RM, Pierce SM, Sandwith T (2002) in Mainstreaming Biodiversity in Devel- opment. Annu Rev Environ Resour 31:445–477. opment Case Studies from South Africa, eds Pierce SM, Cowling RM, Sandwith T, 3. Starbuck WH (2006) The Production of Knowledge, the Challenge of Social Science MacKinnon K (World Bank, Washington, DC), pp 143–153. Research (Oxford Univ Press, Oxford). 21. Carpenter SR, Folke C (2006) Ecology for transformation. Trends Ecol Evol 21:309–315. 4. Max-Neef MA (2005) Foundations of transdisciplinarity. Ecol Econ 53:5–16. 22. Folke C, Hahn T, Olsson P, Norberg J (2005) Adaptive governance of social-ecological 5. Knight AT, Cowling RM, Campbell BM (2006) An operational model for implementing systems. Annu Rev Environ Resour 30:441–473. conservation action. Conserv Biol 20:408–419. 23. Garvin DA (1993) Building a Learning Organization. Harv Bus Rev 71:78–91. 6. McNie EC (2007) Reconciling the supply of scientiﬁc information with user demands: 24. Holling CS (2001) Understanding the complexity of economic, ecological, and social An analysis of the problem and review of the literature. Environ Sci Policy 10:17–38. systems. Ecosystems 4:390–405. 7. Sayer JA, Campbell BM (2004) The Science of Sustainable Development: Local 25. Olsson P, et al. (2006) Shooting the rapids: Navigating transitions to adaptive gov- Livelihoods and the Global Environment (Cambridge Univ Press, Cambridge, UK). ernance of social-ecological systems. Ecol Soc 11:18. 8. van Wilgen BW, Cowling RM, Burgers CJ (1996) Valuation of ecosystem services. 26. Frost P, Campbell B, Medina G, Usongo L (2006) Landscape-scale approaches for Bioscience 46:184–189. integrated natural resource management in tropical forest landscapes. Ecol Soc 9. Becker CD (1999) Protecting a Garua forest in Ecuador: The role of institutions and 11:30. ecosystem valuation. Ambio 28:156–161. 27. Balmford A, Cowling RM (2006) Fusion or failure? The future of conservation biology. 10. Daily GC, Ellison K (2002) The Economy of Nature (Island Press, Washington, DC). Conserv Biol 20:692–695. 11. Clark WC, Dickson NM (2003) Sustainability science: The emerging research program. 28. Harwell MA (1998) Science and environmental decision making in South Florida. Ecol Proc Natl Acad Sci USA 100:8059–8061. Appl 8:580–590. 12. Cowling RM (2005) in Mainstreaming Biodiversity in Production Landscapes, eds 29. Silvano RAM, Udvardy S, Ceroni M, Farley J (2005) An ecological integrity assessment Petersen C, Huntley BJ (Global Environment Facility, Washington DC), pp 18–25. of a Brazilian Atlantic Forest watershed based on surveys of stream health and local 13. Dawe NK, Ryan KL (2003) The faulty three-legged-stool model of sustainable devel- farmers’ perceptions: Implications for management. Ecol Econ 53:369–385. opment. Conserv Biol 17:1458–1460. 30. Pfeffer J, Sutton RI (1999) Knowing ‘‘what’’ to do is not enough: Turning knowledge 14. Pierce SM, et al. (2005) Systematic conservation planning products for land-use into action. Calif Manage Rev 42:83–108. planning: Interpretation for implementation. Biol Conserv 125:441–458. 31. Callicott JB (2006) in Principles of Conservation Biology, eds Groom MJ, Meffe GK, 15. Murphy DD, Noon BR (2007) The role of scientists in conservation planning on private Carroll CR (Sinauer Associates, Sunderland, MA), pp 111–135. lands. Conserv Biol 21:25–28. 32. Cowling RM, Wilhelm-Rechmann A (2007) Social assessment as a key to conservation 16. Boyd J, Banzhaf HS (2005) The Architecture and Measurement of an Ecosystem success. Oryx 41:135–136. Services Index (Resources for the Future, Washington, DC), Discussion Papers dp-05- 33. Heal G (2000) Valuing ecosystem services. Ecosystems 3:24–30. 22. Available at www.rff.org/Documents/RFF-DP-05–22.pdf. 34. Ehrlich PR, Kennedy D (2005) Millennium assessment of human behavior. Science 17. Reyers B, et al. (2007) Developing products for conservation decision-making: lessons 309:562–563. from a spatial biodiversity assessment for South Africa. Divers Distrib 13:608–619. 35. Barrett CB, Brandon K, Gibson C, Gjertsen H (2001) Conserving tropical biodiversity 18. Knight AT, Driver A, Cowling RM, Maze K, Desmet PG, et al. (2006) Designing amid weak institutions. Bioscience 51:497–502. systematic conservation assessments that promote effective implementation: Best 36. Max-Neef MA (1991) Human Scale Development: Conception, Application and practice from South Africa. Conserv Biol 20:739–750. Further Reﬂection (Apex Press, New York).

Cowling et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9487 37. Vermaat JE, et al. (2005) Aggregation and the matching of scales in spatial economics 69. Hannon B (2005) Pathways to environmental change. Ecol Econ 52:417–420. and landscape ecology: Empirical evidence and prospects for integration. Ecol Econ 70. Reser JP, Bentrupperbaumer JM (2005) What and where are environmental values? 52:229–237. Assessing the impacts of current diversity of use of ‘environmental’ and ‘World 38. Hulse DW, Branscomb A, Payne SG (2004) Envisioning alternatives: Using citizen Heritage’ values. J Environ Psychol 25:125–146. guidance to map future land and water use. Ecol Appl 14:325–341. 71. Kaplan S (2000) Human nature and environmentally responsible behavior. J Soc Issues 39. Campbell LM (2005) Overcoming obstacles to interdisciplinary research. Conserv Biol 56:491–508. 19:574–577. 72. Redford K, Sanjayan MA (2003) Retiring Cassandra. Conserv Biol 17:1473–1474. 40. Kremen C (2005) Managing ecosystem services: What do we need to know about their 73. Chew BW, Leonard-Barton D, Bohn RE (1991) Beating Murphy’s Law. Sloan Manage- ecology? Ecol Lett 8:468–479. ment Rev 32:5–16. 41. Naidoo R, Ricketts TH (2006) Mapping the economic costs and benefits of conserva- 74. Monroe MC (2003) Two Avenues for Encouraging Conservation Behaviors. Hum Ecol tion. PLoS Biol 4:2153–2164. Rev 10:113–125. 42. Chan KMA, Shaw MR, Cameron DR, Underwood EC, Daily GC (2006) Conservation 75. Stern PC, Dietz T, Abel T, Guagnano GA, Kalof L (1999) A value-belief-norm theory of planning for ecosystem services. PLoS Biol 4:2138–2152. support for social movements: The case of environmentalism. Hum Ecol Rev 6:81–97. 43. Hein L, van Koppen K, de Groot RS, van Ierland EC (2006) Spatial scales, stakeholders 76. Mayton DM, Ball-Rokeach SJ, Loges WE (1994) Human values and social issues: An and the valuation of ecosystem services. Ecol Econ 57:209–228. introduction. J Soc Issues 50:1–8. 44. Guo ZW, Xiao XM, Li DM (2000) An assessment of ecosystem services: Water flow 77. Schultz KL, McClain JO, Thomas LJ (2003) Overcoming the dark side of worker regulation and hydroelectric power production. Ecol Appl 10:925–936. flexibility. J Oper Manag 21:81–92. 45. Dıáz S, Fargione J, Chapin FS, Tilman D (2006) Biodiversity loss threatens human 78. Kiker CF, Milon JW, Hodges AW (2001) Adaptive learning for science-based policy: well-being. PLoS Biol 4:1300–1305. The Everglades restoration. Ecol Econ 37:403–416. 46. Daily GC, et al. (2000) The value of nature and the nature of value. Science 289:395– 79. May RM (2006) Threats to tomorrow’s world. Address of the President, Lord May of 396. 47. Balvanera P, et al. (2001) Conserving biodiversity and ecosystem services. Science Oxford OM AC FRS, given at the anniversary meeting on 30 November 2005. Notes Rec 291:2047–2047. R Soc 60:109–130. 48. Kalof L, Sattersfield T (2005) The Earthscan Reader in Environmental Values (Earth- 80. Hobbs RJ (2006) Overcoming barriers to effective public communication of ecology. scan Publications, London). Front Ecol Environ 4:496–497. 49. Simpson RD (1998) Economic analysis and ecosystems: Some concepts and issues. Ecol 81. Gelderblom CM, et al. (2003) Turning strategy into action: Implementing a conser- Appl 8:342–349. vation action plan in the Cape Floristic Region. Biol Conserv 112:291–297. 50. Turner RK, Adger WN, Brouwer R (1998) Ecosystem services value, research needs, and 82. Ferraro PJ, Kiss A (2002) Direct payments to conserve biodiversity. Science 298:1718– policy relevance: A commentary. Ecol Econ 25:61–65. 1719. 51. Costanza R, Farber S (2002) Introduction to the special issue on the dynamics and value 83. Wunder S (2005) Payments for Environmental Services: Some Nuts and Bolts (Center of ecosystem services: Integrating economic and ecological perspectives. Ecol Econ for International Forestry Research, Bogor Barat, Indonesia), CIFOR Occasional Paper 41:367–373. 42. Available at www.cifor.cgiar.org/publications/pdf࿝files/OccPapers/OP-42.pdf. 52. Turner RK, et al. (2003) Valuing nature: Lessons learned and future research direc- 84. Mills AJ, et al. (2007) in Restoring Natural Capital Science, Business, and Practice, eds tions. Ecol Econ 46:493–510. Aronson J, Milton SJ, Blignaut J (Island Press, Washington, DC), pp 179–187. 53. Ekins P, Simon S, Deutsch L, Folke C, De Groot R (2003) A framework for the practical 85. Van Wilgen BW, Marais C, Magadlela D, Jezile N, Stevens D (2002) Win-win-win: South application of the concepts of critical natural capital and strong sustainability. Ecol Africa’s Working for Water programme, in Mainstreaming Biodiversity in Develop- Econ 44:165–185. ment: Case Studies from South Africa, eds Pierce SM, Cowling RM, Sandwith T, 54. Chee YE (2004) An ecological perspective on the valuation of ecosystem services. Biol MacKinnon K (World Bank, Washington DC), pp 5–20. Conserv 120:549–565. 86. McKenzie-Mohr D (2000) Promoting sustainable behavior: An introduction to com- 55. Wilson MA, Howarth RB (2002) Discourse-based valuation of ecosystem services: munity-based social marketing. J Soc Issues 56:543–554. Establishing fair outcomes through group deliberation. Ecol Econ 41:431–443. 87. Andreasen AR (1995) Marketing Social Change: Changing Behavior to Promote 56. Senge PM (1990) The leader’s new work: Building learning organizations. Sloan Health, Social Development, and the Environment (Josesey-Bass, Washington, DC). Management Rev Fall:440–463. 88. Ferraro PJ, Pattanayak SK (2006) Money for nothing? A call for empirical evaluation 57. Kinzig A, et al. (2003) Coping with uncertainty: A call for a new science-policy forum. of biodiversity conservation investments. PLoS Biol 4:482–488. Ambio 32:330–335. 89. Westley F, Vredenburg H (1997) Interorganizational collaboration and the preserva- 58. Penn DJ (2003) The evolutionary roots of our environmental problems: Toward a tion of global biodiversity. Organ Sci 8:381–403. Darwinian ecology. Q Rev Biol 78:275–301. 90. Daily GC, Ehrlich PR (1999) Managing earth’s ecosystems: An interdisciplinary chal- 59. Tallis HM, Kareiva P (2006) Shaping global environmental decisions using socio- lenge. Ecosystems 2:277–280. ecological models. Trends Ecol Evol 21:562–568. 91. Mascia MB, et al. (2003) Conservation and the social sciences. Conserv Biol 17:649– 60. Johns DM (2003) Growth, conservation, and the necessity of new alliances. Conserv 650. Biol 17:1229–1237. 92. Gatzweiler FW (2006) Organizing a public ecosystem service economy for sustaining 61. Jacobson SK, McDuff MD, Monroe MC (2007) Promoting conservation through the biodiversity. Ecol Econ 59:296–304. arts: Outreach for hearts and minds. Conserv Biol 21:7–10. 93. Mburu J, Birner R, Zeller M (2003) Relative importance and determinants of land- 62. Kollmuss A, Agyeman J (2002) Mind the Gap: Why do people act environmentally and owners’ transaction costs in collaborative wildlife management in Kenya: An empir- what are the barriers to pro-environmental behavior? Environ Edu Res 8:239–260. ical analysis. Ecol Econ 45:59–73. 63. Huntley BJ, Siegfried WR, Sunter C (1989) South African Environments into the 21st Century (Human and Rousseau Tafelberg, Cape Town, South Africa). 94. Mouratiadou I, Moran D (2007) Mapping public participation in the Water Frame- 64. Baker JP, et al. (2004) Alternative futures for the Willamette River Basin, Oregon. Ecol work Directive: A case study of the Pinios River Basin, Greece. Ecol Econ 62:66–76. Appl 14:313–324. 95. Balint PJ, Mashinya J (2006) The decline of a model community-based conservation 65. Bohensky EL, Reyers B, Van Jaarsveld AS (2006) Future ecosystem services in a project: Governance, capacity, and devolution in Mahenye, Zimbabwe. Geoforum Southern African river basin: A scenario planning approach to uncertainty. Conserv 37:805–815. Biol 20:1051–1061. 96. Flade M, Plachter H, Schmidt R, Werner A (2006) Nature Conservation in Agricultural 66. Millennium Ecosystem Assessment (2005) Ecosystems and Human Well-Being: Syn- Ecosystems (Quelle & Meyer, Wiebelsheim, Germany). thesis (Island Press, Washington DC). 97. Stiglitz JE (2002) Globalization and Its Discontents (WW Norton, New York). 67. Regan HM, Colyvan M, Burgman MA (2002) A taxonomy and treatment of uncertainty 98. Orr DW (2002) Four challenges of sustainability. Conserv Biol 16:1457–1460. for ecology and conservation biology. Ecol Appl 12:618–628. 99. Czech B (2006) If Rome is burning, why are we fiddling? Conserv Biol 20:1563–1565. 68. Sarewitz D (2004) How science makes environmental controversies worse. Environ Sci 100. Miller JR (2005) Biodiversity conservation and the extinction of experience. Trends Policy 7:385–403. Ecol Evol 20:430–434.

9488 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706559105 Cowling et al. Navigating the transition to ecosystem-based SPECIAL FEATURE management of the Great Barrier Reef, Australia

Per Olsson*†, Carl Folke*‡§, and Terry P. Hughes¶

*Stockholm Resilience Centre and §Department of Systems Ecology, Stockholm University, SE-106 91 Stockholm, Sweden; ‡Beijer Institute of Ecological Economics, Royal Swedish Academy of Sciences, P.O. Box 50005, SE-104 05 Stockholm, Sweden; and ¶Australian Research Council Centre of Excellence for Coral Reef Studies, James Cook University, Townsville QLD 4811, Australia

Edited by Gretchen C. Daily, Stanford University, Stanford, CA, and approved March 26, 2008 (received for review July 27, 2007)

We analyze the strategies and actions that enable transitions Different disciplines have studied pieces of the puzzle, for example, toward ecosystem-based management using the recent gover- organizational (28) and institutional aspects (29), but have rarely nance changes of the Great Barrier Reef Marine Park as a case analyzed broader social–ecological dynamics. study. The interplay among individual actors, organizations, and A literature on the role of leadership strategies in transitions to institutions at multiple levels is central in such transitions. A ecosystem-based management is emerging (30–33), focusing on the flexible organization, the Great Barrier Reef Marine Park Authority, relationship between social structures and human agency (human was crucial in initiating the transition to ecosystem-based man- capacity to make and impose choices). The scholarly debate in agement. This agency was also instrumental in the subsequent political science recognizes rigidity, veto points, and path depen- transformation of the governance regime and provided leadership dence as common characteristics of institutions and public policy- throughout the process. Strategies involved internal reorganiza- making (34). This includes sudden change and ‘‘punctuated equi- tion and management innovation, leading to an ability to coordi- librium’’ where long periods of stability and incremental change nate the scientific community, to increase public awareness of interact with abrupt, nonincremental, large-scale change (35–37). environmental issues and problems, to involve a broader set of Windows of opportunity offer possibilities for large-scale change stakeholders, and to maneuver the political system for support at (38, 39). Windows may open because of exogenous shocks and critical times. The transformation process was induced by increased crises, including shifts in underlying economic factors such as a pressure on the Great Barrier Reef (from terrestrial runoff, over- rapid rise in energy prices, a change in the macropolitical environ- harvesting, and global warming) that triggered a new sense of ment, new scientific findings, regime shifts in ecosystems, or rapid urgency to address these challenges. The focus of governance loss of ecosystem services (32, 39). Ingram and Fraser (40) use such shifted from protection of selected individual reefs to stewardship a punctuated equilibrium framework (37) to analyze policy inno- of the larger-scale seascape. The study emphasizes the significance vations in water management in California, where water manage- of stewardship that can change patterns of interactions among key ment and policy were locked into a highly engineered infrastructure actors and allow for new forms of management and governance to

that reinforced one policy and excluded others. A new awareness SCIENCE emerge in response to environmental change. This example illus- has now emerged among multistakeholders in California water SUSTAINABILITY trates that enabling legislations or other social bounds are essen- management. Policy and management have shifted and broadened tial, but not sufﬁcient for shifting governance toward adaptive to incorporate a wider array of state and federal agencies as well as comanagement of complex marine ecosystems. private and public organizations. This demonstrates how rigidity in policymaking can stifle innovations and capacity to deal with adaptive governance ͉ ecosystem services ͉ transformation crises (41). New frameworks are emerging for investigating the interplay he widespread degradation of marine ecosystems and their between long periods of stability and abrupt change in social– Tbiodiversity (1–3) results to a large extent from a failure of ecological systems and analyzing shifts toward ecosystem-based governance (4, 5). Traditional focus on single-species resources in management (9, 42, 43). In this context, there is an urgent need to fisheries and aquaculture has created organizational and institu- identify strategies that have enabled transitions in management tional structures with compartmentalized decision-making pro- from a conventional focus on a single resource or habitat to cesses, leading to narrow policy instruments that create incentives large-scale ecosystem-based management. Management brings to- for policies and actions that undermine sustainability (6–8). Such gether existing knowledge from diverse sources into new perspec- governance is ill prepared to respond to the complexity of dynamic tives for practice (44). Here, transition refers to a shift from one ecosystems or build an adaptive capacity for coping with change and management system to another, often a discontinuous shift to a new uncertainty (9–12). These approaches are often overwhelmed by management trajectory (45). We argue that, to understand such global economic drivers (13–15) and cannot address the complex transitions, for dealing with the degradation of marine ecosystems threshold dynamics of linked social–ecological systems (16–18). (2, 3), there is a need to address governance systems and investigate New and more effective governance systems are urgently needed. transformations from one governance system to another. By gov- The Millennium Ecosystem Assessment (19) highlighted the ernance systems we mean the interaction patterns of actors with importance of incorporating an understanding of ecosystem dy- conflicting objectives and the instruments chosen to steer social and namics into governance systems to build capacity for managing environmental processes within a particular policy area (18). Insti- ecosystem services. The search for better approaches to ensuring tutions are a central component (46–48), as are interactions sustainable outcomes has helped develop important principles and between actors and the multilevel institutional setting, creating protocols for ecosystem-based management of marine resources (4, 5, 20, 21). These acknowledge ecosystems as complex dynamic systems and address the mismatch between social systems and Author contributions: P.O. designed research; P.O. performed research; P.O., C.F., and ecosystem dynamics. Typically, prevailing approaches emphasize T.P.H. analyzed data; and P.O., C.F., and T.P.H. wrote the paper. spatial planning, usage zoning, and marine protected areas (22–27). The authors declare no conﬂict of interest. However, the burgeoning literature on ecosystem-based manage- This article is a PNAS Direct Submission. ment offers few empirically based insights into social–ecological †To whom correspondence should be addressed. E-mail: [email protected]. strategies that make transitions to such management possible. © 2008 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706905105 PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9489–9494 Table 1. Strategies used by GBRMPA for facilitating the transition to ecosystem-based management Strategies Actions Examples of barriers to change

Making internal Establishing Senior Managers Forum and four regional teams Resource constraints organizational Providing clear and transparent leadership at the relevant levels within Inability to innovate or deal with surprise changes the organization Lack of direction, shared vision, engagement, trust, leadership, Communicating a shared vision and goals cross-sector cooperation, and communication Having few leaders exacerbates vulnerability Bridging science and policy Drawing on existing networks of scientists, managers and industry to Science is fragmented promote dialogue Lack of scientiﬁc certainty Workshops and forums for synthesizing knowledge Different perceptions and views among scientists and managers, Communicating shared vision and goals lack of trust

Changing public perceptions Clear, simple, and tailored stakeholder information from a Different knowledge and interests among stakeholder groups communication professional Low awareness of problems, threats, and ecological interactions Visualizing the entire GBR as an interconnected ecosystem Creating a sense of urgency for conservation Facilitating community Building trust with communities: personal interactions and regional teams Lack of trust participation and Community information sessions Conflicting views among key actor groups, misinformation public consultation Recasting problems as opportunities Outreach to local communities difficult Periodic updates on the rezoning process Lack of leadership Innovative submission routines Gaining political support Prepared for change: politically expert staff, timing actions, having Change of people in power relevant information ready Lack of support from key politicians Briefing key players well before the new zoning plan Zoning plans can be stopped Allying with other key actor groups Opposing views Pollsters for leverage and monitoring public opinion complex relationships between people and ecosystem dynamics involved in the rezoning. Interviewees were asked to describe (18). The adaptive governance framework specifically addresses significant events during the rezoning process, triggers for initiating such dynamic interactions and the social–ecological capacity to zoning at the scale of the seascape, strategies and actions to make sustain ecosystem services in the face of uncertainty and change the shift, and barriers to change. The interviewees were asked to (49, 50). name key individuals instrumental in shaping change and navigat- The Great Barrier Reef Marine Park (GBRMP), the largest coral ing the transition. A ‘‘snowball sampling’’ technique was used for reef system in the world, has recently undergone a major rezoning selecting interviewees (54). and transformation in governance toward stewardship of the large- We used qualitative data analysis (56) to analyze the collected scale seascape, incorporating 70 bioregions (including many non- data. The conceptual framework, hypothesis, and the research reef habitats). The rezoning was an ambitious effort to better questions stated in the introduction were used to select, focus, and manage the Great Barrier Reef (GBR) and associated ecosystems organize the data. A first analysis of written sources and interviews and strengthen their resilience in the face of climatic change, i.e., identified important events and interacting structures and processes their ability to cope with disturbances and continue to generate that facilitated the shift. This information is presented in the essential ecosystem services (22, 51–53). sections that follow. The initial analysis also identified actions taken This article identifies social features and strategies that made it to deal with specific barriers to change. These insights guided the possible to shift the direction of an already-existing multilevel next step of the analysis: identifying individuals that could link governance regime toward large-scale ecosystem-based manage- actions to specific governance strategies and describe the motiva- ment. We hypothesize that achieving such a shift is more complex tion behind these strategies, how they were developed, and what than simply changing legislation, providing economic instruments, triggered them. Five overall strategies emerged from our analysis or introducing new restrictions on resource use. What triggered the described as Table 1 and in associated text. transition? Which were the significant events? Who were the key stewards? What were the barriers to the transition? What strategies Great Barrier Reef Marine Park and actions were used to overcome barriers and mobilize the shift? The Park covers 344,000 km2, an area almost the size of California. How did these strategies and actions help to change incentives, Coral-dominated assemblages form discontinuous fringing reefs on perceptions, institutions, and patterns of interactions between the mainland and inshore granitic islands. Most reefs are found actors? To investigate these issues, we analyze multilevel interac- 30–200 km offshore, on the middle and outer edge of the conti- tions among individuals, organizations, and institutions in relation nental shelf. Many other non-reef assemblages occur near the to ecosystem management of the Great Barrier Reef. Following on shore, on soft bottoms and in deeper water between the reefs (e.g., our earlier studies of transformations in social–ecological systems mangroves, seagrass beds, etc.). Like many other coral reefs, the (31), we were especially interested in how the rezoning was devel- Barrier Reef generates a multitude of essential ecosystem services oped and the preparations that helped seize a political window of (57). The GBR Marine Park contributes AU$6.9 billion annually to opportunity. the Australian economy, Ͼ85% of which is from the tourism We focus on the main agency involved in the rezoning, the Great industry (58). Barrier Reef Marine Park Authority (GBRMPA), which initiated The Australian federal government enacted The Great Barrier the shift. Data collection included an extensive review of published Reef Marine Park Act in 1975 in response to public concerns about and online sources, from 1975 to the present, to capture the history threats to the reef from oil drilling, mining, and unexplained of legislation, governmental agencies, public opinion, and how the outbreaks of coral-eating starfish (59, 60). The act established rezoning developed. Sources included scientific articles, legal doc- GBRMP, and in 1981 the Great Barrier Reef region was also uments, staff papers, annual reports, reports on the state of the reef, declared a World Heritage Area. The marine park is a multiple-use reviews, strategic plans, maps, and fact sheets, as well as the marine park, allowing a range of uses based on spatial zoning (22). GBRMPA web site (www.gbrmpa.gov.au). We also conducted 22 The seven marine zone types range from general use (the least open-ended, in-depth interviews (54, 55) with key informants restrictive zone, allowing most reasonable use) to preservation

9490 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706905105 Olsson et al. (very small ‘‘no-go’’ areas set aside as scientific reference areas). fisheries. The RAP process began in 1998–1999 with a mobilization Use and entry are allowed, either ‘‘as-of-right’’ or by permit in all of scientific expertise to identify and map habitat types. Panels of SPECIAL FEATURE zone types. The GBRMP generally extends inshore to the low-water experts compiled Ͼ40 existing data sets to characterize the biolog- mark but excludes some near-shore areas managed by the State of ical and physical diversity of the GBRMP (53). Geographic infor- Queensland. The State has created the Great Barrier Reef Coast mation systems-based tools and analytical methods identified and Marine Park to protect tidal lands and coastal waters. The State mapped 70 bioregions, of which 30 were reef bioregions and 40 Park complements the federal GBRMP by adopting similar zone were non-reefal. (A map of the 70 bioregions in the GBRMP is objectives and entry and use provisions. The comanagement be- available at www.gbrmpa.gov.au/࿝࿝data/assets/pdf࿝file/0016/ tween the federal and the state governments originates from a 7315/bioregions࿝2001࿝06.pdf.) formal agreement signed in 1979. This federal–state cooperation is Formal community participation in the RAP process took place important for enabling ecosystem-based management of the region. in two phases (66), with informal consultation throughout the planning phase. Over a 3-month period from May 2002, GBRMPA Zoning and Rezoning the Great Barrier Reef Marine Park. The Great sought formal community input for preparation of a Draft Zoning Barrier Reef Marine Park Act established the GBRMPA in 1976 Plan for the entire GBRMP. This resource-intensive process used and required the new agency to initiate zoning plans for the marine a range of techniques to ensure that all coastal communities park. The Authority grew in size and sophistication over time and adjoining the marine park were aware of the RAP. This first round Ͼ since 2000 has a staff of 130 and an annual budget close to AU$30 of formal public participation resulted in 10,190 written submis- million (GBRMPA annual reports). Between 1983 and 1988, each sions. By a combination of expert opinion, stakeholder involve- of the four sections of the GBRMP (the Far Northern, Cairns, ment, and analytical approaches, different options for no-take area Central, and the southern Mackay/Capricorn sections) were zoned networks were identified by GBRMPA using a variety of planning for the first time. No-take areas together accounted for 5% of the tools to integrate biophysical, social, and economic information (51, marine park, mainly in the remote Far Northern area and predom- 52, 67–69). This led to the completion of the Draft Zoning Plan and inantly covering coral reefs (reefs were considered the most im- focused the second formal community participation phase (66), portant habitats at that time but actually make up only 6% of the which resulted in 21,500 additional written submissions (70). entire GBRMP). In late 2003, the Draft Zoning Plan was revised to incorporate As scientific information accumulated from the 1970s onward, it information from the second consultation process, resulting in the became apparent that the Great Barrier Reef was showing signs of Revised Zoning Plan, which increased the percentage of no-take degradation, primarily from runoff of sediment from land, over- areas in the GBRMP by Ͼ6-fold to 33%, including at least 20% of harvesting, and more recently from global warming. Recent analysis each of the 70 bioregions. In December 2003 the new zoning plan of banded coral skeletons show that runoff of sediment from land was submitted to federal parliament and passed into law in July increased between 5- and 10-fold after about 1870, when European 2004. The Australian Government also agreed to a structural settlement and overstocking of semiarid river catchments began adjustment package providing compensation for those (such as (61). Recurrent outbreaks of crown-of-thorns starfish, which de- commercial fishers) adversely affected by the new zoning. stroy large amounts of coral, have affected Ͼ200 reefs since the early 1960s (62). One theory for these outbreaks is that enhanced Key Strategies Behind the Shift to Ecosystem-Based Management. SCIENCE Through the methods and data analyses outlined in the introduc- runoff of nutrients from land has shortened the starfish’s larval SUSTAINABILITY phase, leading to population explosions (63). Stocks of mega-fauna tion, we identified five general strategies that were used by species on the Great Barrier Reef including dugongs, turtles, and GBRMPA to implement RAP and the new zoning plan. These are sharks have fallen dramatically since European settlement (1). (i) internal organizational changes, (ii) bridging science and policy, Similarly, the size and densities of fish species targeted by recre- (iii) changing people’s perceptions, (iv) facilitating public consul- ational and commercial fisheries have declined in recent years. The tation and participation, and (v) gaining political support. Our biomass of coral trout is up to six times lower on heavily fished qualitative data analysis also revealed several actions taken to deal near-shore reefs compared with adjacent no-take areas established with specific barriers to change. The following section describes in the 1980s (64). Demographic and economic data gathered in the general strategies, actions, and examples of barriers, summarized in 1980s and 1990s showed rapid growth in human population, land Table 1. A common feature of GBRMPA’s strategy was anticipat- clearing, coastal development, tourist visits, and fishing pressure. ing and addressing potential barriers to the implementation of an Gradually, it became clear that the initial level of protection did not ecosystem-based approach. adequately protect biodiversity within the GBRMP or ensure that Internal organizational changes. The GBRMPA executive team during the entire ecosystem remained healthy, productive, and resilient RAP comprised the chair, Virginia Chadwick, and two executive (1, 2, 62). directors, to whom senior managers report. GBRMPA underwent In the late 1990s there was growing awareness among scientists major internal organizational changes from the late 1990s onwards. and reef managers that many biological communities on the GBR, Some changes were based on a 25-year strategic plan (71) and a such as inshore and deeper habitats, were poorly represented in 1997 report that suggested that GBRMPA should be organized existing no-take zones. They also realized that connectivity of larvae around four critical environmental issues (72). This was a turbulent and other poorly understood interactions between reef and non- time for GBRMPA, with a high staff turnover rate that facilitated reef habitats were important to maintain the resilience of the entire internal reorganization (73). The emerging concept of rezoning the ecosystem. Unprecedented regional bleaching occurred in the entire marine park initially occupied a small group, understaffed summer of 1997/1998, affecting large parts of the GBR and other and underfinanced. By the early 2000s, however, almost all of reefs in the Western Pacific and most of the tropical Indian Ocean GBRMPA was involved in the RAP process, led by the executive (65). It was a rude wake-up call to the dangers of global warming team (74). The executive team established a Senior Managers’ for coral reefs that required an urgent response. Forum to coordinate activities, enhance communication, solve In 1998, the GBRMPA initiated a major rezoning of the marine conflicts, tap into the expertise of senior managers to ensure a park called the Representative Areas Program (RAP) (22, 53) to shared policy direction, and advise the organization’s chair. As one systematically increase the protection of biodiversity within the of the executive directors describes: GBRMP by protecting representative examples of each type of I realized that, in house, [the RAP] was not something that habitat within a network of no-take areas. Focus was on protecting a section could handle . . . it was about the whole organi- biodiversity and maintaining ecosystem function and services zation. It was about the marine park. So everyone had to rather than on maximizing the yield of commercially important have ownership of it. And everyone was needed. We had

Olsson et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9491 to almost stop doing many other things, to do this. Oth- several minor setbacks regarding public consultation, some antic- erwise we couldn’t do it. So we formed what was called the ipated and others unexpected. The goals of RAP and the zoning Senior Managers’ Forum. plan were sometimes misunderstood, and some members of local In this way, GBRMPA used its organizational flexibility to communities were suspicious about GBRMPA’s agenda. On sev- establish and nurture an environment where creativity was encour- eral occasions misinformation about the intended location of aged and innovative solutions to problems could emerge. Impor- no-take zones (especially close to a town or the shore) resulted in tantly, this process was achieved without any additional funding and public meetings with large audiences including distressed, angry relied entirely on a flexible internal redeployment of staff. The individuals (75). GBRMPA turned these problems into opportu- Senior Management Forum unified internal management and nities for correcting misinformation and spreading key messages communicated a common vision throughout the organization, with about the RAP and rezoning and produced a ‘‘Correcting Misin- leadership of the Forum shared by both of the executive directors. formation’’ fact sheet. GBRMPA attended every public meeting The Senior Managers Forum led to the establishment of four they were invited to. Whenever GBRMPA was in charge of regional teams responsible for the comprehensive public consulta- organizing meetings they avoided large public meetings that could tions associated with the RAP. Using teams helped the GBRMPA be dominated by one or a few people and instead held several to avoid competition between sectors, increase internal collabora- hundred community information sessions in regional and local tion, and become more effective by pooling experiences and community centers. The reason for this is offered by a Senior resources. Manager: Bridging science and policy. The RAP process relied heavily on We thought that was a more effective way of interacting scientific expertise and a new synthesis of the best available data on with people because it gave everyone an even chance of species and habitats of the Great Barrier Reef. To harness this getting heard, where if you have a public meeting quite expertise, GBRMPA created new opportunities for interaction, often it is dominated by the loud and angry one. dialogue, and information sharing with researchers. This included establishing committees and panels, facilitating workshops, and Information sessions and follow-up meetings were a more effec- communicating to scientists GBRMPA’s overall vision and goals tive way to interact directly with all of the people present and build for the RAP and rezoning. For example, two independent advisory trust. Typically, four or five GBRMPA staff presented posters, committees (the Scientific Steering Committee and the Social, pictures, and other informative material and answered questions. Economic, and Cultural Steering Committee) were convened to Some communities had not had contact with GBRMPA for de- develop two sets of operating principles that guided the RAP cades. Periodic updates on the RAP process were produced from process (53). Scientists were encouraged to think beyond their May 2000 to November 2003 and were posted online (75). individual sample sites or specialized expertise to collectively reach For the second consultation phase, GBRMPA improved the a bioregional perspective on the GBR as a whole (53). This dialogue formulation and design of the submission packages distributed to was facilitated by a longstanding relationship between GBRMPA communities. New and innovative ways for dealing with the influx and researchers at universities, the Australian Institute of Marine of submissions were required because of the overwhelming re- Sciences, and the Cooperative Research Centre for the Great sponse from the public. The volume of submissions (Ͼ31,000 in Barrier Reef World Heritage Area (CRC Reef). total) came as a surprise to GBRMPA. A ‘‘factory’’ was set up for Changing people’s perceptions. Because of its iconic status, there was handling them, quickly allocating human and financial resources overwhelming support both nationally and locally for conserving within the organization without the need for additional external the Great Barrier Reef. GBRMPA tracked this support using funding. (Periodic updates on the progress of RAP are available at pollsters and used the information during RAP for political lever- www.gbrmpa.gov.au/corp࿝site/key࿝issues/conservation/rep࿝areas/ age. However, not everyone was aware of the threats to the reef or updates.) agreed with the proposed management changes (75). Some local Gaining political support. GBRMPA reports to the Australian Fed- recreational fishers in particular were vocal in their opposition to eral Minister for Environment and Heritage, whose support was no-take zones. Some within GBRMPA foresaw that the implemen- crucial for the RAP process. The rezoning legislation had to pass tation of the RAP and rezoning the marine park would require an through the two federal houses of Parliament. After the federal extensive communication strategy to bolster public support. Many election in November 2001, a new minister for environmental people still perceived the Great Barrier Reef as a pristine environ- issues, the Honorable David Kemp, was appointed in 2002. The new ment, protected from human impacts by its sheer size and relative minister, and ultimately the prime minister, had to be reassured that isolation. there was an adequate scientific basis for the zoning, that it could To address this issue, GBRMPA hired a highly skilled commu- be carried through to a successful conclusion within an acceptable nication officer to produce a ‘‘reef under pressure’’ information time frame, and that GBRMPA had the skills necessary to lead an campaign showing that the reef is no longer a pristine wilderness but extensive public consultation process. After several sessions with rather is subject to anthropogenic degradation caused by coastal the chair of GBRMPA, the minister was convinced that the RAP development, land use, shipping, tourism, and fishing. The cam- and the new zoning plan would be a major advance in conserving paign raised awareness and created a sense of urgency about the the biodiversity of the marine park that could be managed politi- need to better protect the GBRMP for future generations. The cally. In his interview for our study, the minister stated campaign included web sites, posters, pamphlets, and television I made a judgment about Virginias Chadwick’s capacity to advertisements showing well known local individuals advocating for lead the process. I came to the view that she could manage the need for change. The information campaign, tailored for a it . . . she had an experience in mapping the political range of audiences, was followed up by continuous polling to landscape and in talking with leaders of interest groups and monitor the change in people’s perceptions. in handling public meetings that were at times quite Facilitating community participation and public consultation. GBRMPA emotional. is required by the 1975 Act to inform the public about new zoning plans and provide two formal public consultation periods during The timing of submitting the plan to the Senate was crucial. The their preparation. Public consultation for the RAP greatly ex- chair of GBRMPA and the minister agreed that the new zoning ceeded the requirements of the Act and was by far the most plan should be submitted in December 2003 for it to become extensive in the history of the marine park. GBRMPA staff had to operational by July 2004, before the upcoming federal election the learn quickly, and several new methods were trialed. There were same year. The chair of the GBRMPA states

9492 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706905105 Olsson et al. I realized . . . that we needed 15 sitting days of Parliament, Similar studies have shown the essential role of shifting people’s perceptions. Such shifts are critical factors in altering the trajectory and a simple examination of the parliamentary sitting SPECIAL FEATURE timetable showed that if one wanted the 15 sitting days to of natural resource management (77, 78). In The Netherlands, a end, say, around midyear, then the plan had to be submit- recent shift to more integrated forms of water management dem- ted to Parliament before the Parliament got up for the onstrates that a change in people’s mental models, from ‘‘fighting Christmas break. If one didn’t do that . . . one would have the water’’ to ‘‘living with the water,’’ was critical for adaptive been looking at October, November, which would have management (79). In the Kristianstads Vattenrike Biosphere Re- been slap-bang in the middle of a federal election. serve in Sweden, a comparable shift in the perception of local politicians was critical in the transition to ecosystems-based man- This provided a narrow political window of opportunity that set agement (31). These examples and the GBRMPA case also point the time frame for GBRMPA. GBRMPA and the minister had to to the role of ecologists in social–ecological transformations (80) prepare a smooth passage of the plan through both houses of and the need to coordinate scientist interactions with each other, Parliament. Throughout the planning process, senior staff from the public, and politicians. GBRMPA made frequent trips to Canberra (a 5-hour journey by Our study points to the need for research on policy windows plane) and other key destinations to inform critical players such as and the ability to create the right links, at the right time, around governmental departments and agencies responsible for fisheries the right issues (30, 31). Well timed actions that change inter- and the environment, members of Parliament and senators (espe- action patterns among social actors can be crucial for governance cially those representing constituencies along the Queensland transformations (18). GBRMPA’s Senior Managers Forum, the coast), shipping interests, port authorities, and the Defense De- regional teams, and the community information sessions are partment. The political skills of GBRMPA’s executive team and the innovative evolving structures set up especially for the RAP to director for communication were important for navigating the improve interactions between individuals and divisions within political system. Senior scientists, conservation nongovernmental the GBRMPA and between the GBRMPA and other key actors organizations, and lobbyists for the tourism industry also played a across multiple levels. role in convincing politicians of the need to pass the reef legislation. Our study also shows the need to understand the dynamic Fishing interests were also politically active. processes that underlie the emergence of new forms of governance and management and the role of leadership in these processes. Discussion Initially, a small team working within the GBRMPA planned the The study of the Great Barrier Reef shows the critical role of rezoning of the entire marine park, which subsequently led to flexible governance systems that can deal with complex and dy- critical support from the Authority’s executive team for the major namic ecosystems by linking individuals, networks, organizations, rezoning effort. This happened in three stages, from (i) a relatively and institutions across multiple levels of human activity (49). We minor project within GBRMPA to (ii) incorporation across all parts used a broad analytical approach to identify strategies and actions of the Authority and status as an agency priority, to (iii) changing used by GBRMPA to overcome barriers to change and transform national legislation and influencing other areas in Australia (e.g., governance and management of the large-scale coral reef seascape. rezoning of Ningaloo Reef in Western Australia in 2004) and SCIENCE Interacting key strategies that emerged from our analysis are becoming a role model for policy development elsewhere. internal reorganization; coordination of scientists and other ex- The RAP process shows the role of skillful leadership (28) for SUSTAINABILITY perts; tailored information campaigns about the state of the GBR; moving between stages and across multiple organizational and community participation and public consultation; and actively political levels; first when the executive team allocated internal working to gain political support. resources for developing the RAP and second when the chair of The GBRMPA case illustrates that policy development and GBRMPA, Virginia Chadwick, won the critical support of David implementation are complex, highly dynamic, and sometimes Kemp, the federal minister, for proceeding with RAP and the new abrupt. Institutional inertia can develop into a major transition zoning plan. Leadership issues have been addressed in several within a fairly short period, in our case between 1998 and 2004, studies (81, 82) but not in the context of understanding transitions when a policy window of opportunity was effectively used. The to ecosystem-based management. RAP, and its use in the rezoning of the reef, is an innovative The GBRMPA case suggests that enabling legislation or other approach, developed within the GBRMPA, adopted at the highest social bounds is essential but not sufficient for achieving adaptive political level in Australia, and actively used to improve the gov- comanagement (83) of complex marine ecosystems. Bringing ernance of the GBR. Initially, there was the recognition that the together science and policy is another important but singly existing zoning network did not adequately protect the range of insufficient component of transitions and transformations. Crit- biodiversity of the reefs and hence could not maintain the GBR’s ical interacting strategies for transformations in social– resilience in the face of recurrent ecological disturbances. Com- ecological systems must be addressed to understand these shifts. bined with increased human pressures on the GBR, including the Additional empirical studies, case-study analyses, and compar- challenges of climate change, individual actors within the ative studies can develop a better understanding of strategies for GBRMPA were triggered to search for more holistic approaches to transformation in governance toward ecosystem-based manage- governance and management of this large marine ecosystem. The ment under various social–ecological conditions. For example, new, more sophisticated approach that emerged addressed both in contrast to the GBR case, marine zoning in the United States ecosystem dynamics and the intricate web of interactions between has been severely constrained because of inflexible institutions, social and ecological systems. lack of public support, difficulties developing acceptable legis- GBRMPA focused on communication and information through- lation, and failures to achieve desired results even after zoning out the RAP process, hiring a skilled communication officer to is established (8). Understanding successes and failures in ma- develop communication strategies and information. Combined rine governance systems is a first step in improving their adaptive with new scientific insights, this shifted the perception of the GBR capacity to secure ecosystem services in the face of uncertainty from a well protected pristine coral reef ecosystem to a vulnerable and rapid change. and complex seascape requiring active stewardship. Increased public and media interest helped to tip governance in the new ACKNOWLEDGMENTS. We thank the interviewees and many others who provided information and data for this study. We also thank two anonymous direction of ecosystem-based management with broad stakeholder reviewers whose comments helped improve the ﬁrst manuscript. Our collabora- engagement (76). tion has been funded through grants from the Australian Research Council’s

Olsson et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9493 Centre of Excellence for Coral Reef Studies and the Swedish Research Council for Program as well as support from Mistra, the Swedish Foundation for Strategic the Environment, Agricultural Sciences and Spatial Planning Centre of Excellence Environmental Research, to the Stockholm Resilience Centre.

1. Jackson JBC, et al. (2001) Historical overfishing and the recent collapse of coastal 45. Geels FW, Kemp R (2006) in Flexibility and Stability in the Innovating Economy, eds ecosystems. Science 293:629–638. McKelvey MD, Holme´ n M (Oxford Scholarship Online Monographs, Oxford), pp 227–257. 2. Hughes TP, et al. (2003) Climate change, human impacts, and the resilience of coral 46. North DC (1990) Institutions, Institutional Change and Economic Performance (Cam- reefs. Science 301:929–933. bridge Univ Press, Cambridge, UK). 3. Worm B, et al. (2006) Impacts of biodiversity loss on ocean ecosystem services. Science 47. Young OR (2002) in The Drama of the Commons, eds Ostrom E, Dietz T, Dolsak N, Stern 314:787–790. P, Stonich S, Weber EU (Natl Acad Press, Washington, DC), pp 263–291. 4. Costanza R, et al. (1998) Principles for sustainable governance of the oceans. Science 48. Ostrom E, et al., eds (2002) The Drama of the Commons (Natl Acad Press, Washington, DC). 281:198–199. 49. Dietz T, Ostrom E, Stern PC (2003) The struggle to govern the commons. Science 5. Young OR, et al. (2007) Solving the crisis in ocean governance: Place-based manage- 302:1907–1912. ment of marine ecosystems. Environment 49:20–32. 50. Folke C, Hahn T, Olsson P, Norberg J (2005) Adaptive governance of social-ecological 6. Ludwig D, Hillborn R, Walters C (1993) Uncertainty, resource exploitation, and con- systems. Annu Rev Environ Resour 30:441–473. servation: Lessons from history. Science 260:17–36. 51. Day J, et al. (2002) in Proceedings of the Ninth International Coral Reef Symposium 7. Hughes TP, Bellwood DR, Folke C, Steneck RS, Wilson J (2005) New paradigms for 2000, ed Moosa MK (Ministry of Environment, Indonesian Institute of Sciences, Inter- supporting the resilience of marine ecosystems. Trends Ecol Evol 20:380–386. national Society for Reef Studies, Jakarta, Indonesia), pp 687–696. 8. Crowder LB, et al. (2006) Resolving mismatches in U.S. ocean governance. Science 52. Lewis A, et al. (2003) in Coastal GIS 2003: An Integrated Approach to Australian Coastal 313:617–618. Issues, eds Woodroffe CD, Furness RA (Maritime Policy no. 14, Wollongong Papers, 9. Berkes F, Colding J, Folke C, eds (2003) Navigating Social–Ecological Systems: Building Wollongong Univ, Wollongong, Australia), pp 431–451. Resilience for Complexity and Change (Cambridge Univ Press, Cambridge, UK). 53. Fernandes L, et al. (2005) Establishing representative no-take areas in the Great Barrier Reef: 10. Wilson JA (2006) Matching social and ecological systems in complex ocean fisheries. Large-scale implementation of theory on marine protected areas. Conserv Biol 19:1733–1744. Ecol Soc 11:9. 54. Bernard HR (1994) Research Methods in Anthropology: Qualitative and Quantitative 11. Gunderson LH, Light SS (2006) Adaptive management and adaptive governance in the Approaches (Sage, Newbury Park, CA). Everglades ecosystem. Policy Sci 39:323–334. 55. Kvale S (1996) Interviews: An Introduction to Qualitative Research Interviewing (Sage, 12. Ostrom E (2007) Going beyond panaceas Special Feature: A diagnostic approach for Thousand Oaks, CA). going beyond panaceas. Proc Natl Acad Sci USA 104:15181–15187. 56. Miles MB, Huberman AM (1994) Qualitative Data Analysis: An Expanded Sourcebook 13. Naylor RL, Eagle J, Smith WL (2003) Salmon aquaculture in the Pacific Northwest: A (Sage, Thousand Oaks, CA), 2nd Ed. global industry with local impacts. Environment 45:18–39. 57. Moberg F, Folke C (1999) Ecological goods and services of coral reef ecosystems. Ecol 14. Berkes F, et al. (2006) Globalization, roving bandits, and marine resources. Science Econ 29:215–233. 311:1557–1558. 58. Access Economics (2007) Measuring the Economic and Financial Value of the Great 15. Deutsch L, et al. (2007) Feeding aquaculture growth through globalization: Exploita- Barrier Reef Marine Park, 2005–2006 (Great Barrier Reef Marine Park Authority, ࿝ tion of marine ecosystems for fishmeal. Glob Environ Change 17:238–249. Canberra, Australia, June 2007). Available at www.gbrmpa.gov.au/corp site/ ࿝ ࿝ ࿝ ࿝ ࿝ 16. Folke C, Pritchard L, Berkes F, Colding J, Svedin U (2007) The problem of fit between info services/publications/research publications/rp087/access economics report 0607. ecosystems and institutions: Ten years later. Ecol Soc 12:30. 59. Wright J (1977) The Coral Battleground (Thomas Nelson, Melbourne, Australia). 17. Brown K (2003) Integrating conservation and development: A case of institutional 60. Lawrence D, Kenchington R, Woodley S (2002) The Great Barrier Reef: Finding the misfit. Front Ecol Environ 1:479–487. Right Balance (Melbourne Univ Press, Melbourne, Australia). 18. Galaz V, Olsson P, Hahn T, Folke C, Svedin U (2008) Institutions and Environmental 61. McCulloch M, et al. (2003) Coral record of increased sediment flux to the Inner Great Change: Principal Findings, Applications, and Research Frontiers, eds Young OR, King Barrier Reef since European settlement. Nature 421:727–730. LA, Schro¨ der H (MIT Press), Cambridge, MA), pp 147–182. 62. Bellwood DR, Hughes TP, Folke C, Nystrom M (2004) Confronting the coral reef crisis. 19. Millennium Ecosystem Assessment (2005) Ecosystems and Human Well-Being: Synthe- Nature 429:827–833. 63. Brodie J, Fabricius K, De’ath G, Okaji K (2005) Are increased nutrient inputs responsible sis (Island, Washington, DC). for more outbreaks of crown-of-thorns starfish? An appraisal of the evidence. Mar Poll 20. Babcock EA, Pikitch EK (2004) Can we reach agreement on a standardized approach to Bull 51:266–278. ecosystem-based fishery management? Bull Mar Sci 74:685–692. 64. Russ GR, Alcala AC (2003) Marine reserves: Rates and patterns of recovery and decline 21. National Research Council (1999) Sustaining Marine Fisheries (Natl Acad Press, Wash- of predatory fish, 1983–2000. Ecol Appl 13:1553–1565. ington, DC). 65. Wilkinson C (2004) Status of Coral Reefs of the World (Australian Institute of Marine 22. Day JC (2002) Zoning—Lessons from the Great Barrier Reef Marine Park. Ocean Coast Science, Townsville, Australia). Available at www.aims.gov.au/pages/research/coral- Manage 45:139–156. bleaching/scr2004. 23. Lubchenco J, Palumbi SR, Gaines SD, Aandelman S (2003) Plugging a hole in the ocean: 66. Australian Government (2006) Review of the Great Barrier Reef Marine Park Act 1975: The emerging science of marine reserves. Ecol Appl 13:S3–S7. Review Panel Report (Government Department of the Environment and Heritage, 24. Guerry AD (2005) Icarus and Daedalus: Conceptual and tactical lessons for marine Canberra, Australia). Available at www.environment.gov.au/coasts/publications/gbr- ecosystem-based management. Front Ecol Environ 3:202–211. marine-park-act.html. 25. Arkema KK, Abramson SC, Dewsbury BM (2006) Marine ecosystem-based manage- 67. De’ath G (1999) Principal curves: A new technique for indirect and direct gradient ment: From characterization to implementation. Front Ecol Environ 4:525–532. analysis. Ecology 80:2237–2253. 26. Boyes SJ, Elliott M, Thomson SM, Atkins S, Gilliland P (2007) A proposed multiple-use 68. De’ath G, Fabricius KE (2000) Classification and regression trees: A powerful yet simple zoning scheme for the Irish Sea. Mar Pol 31:287–298. technique for ecological data analysis. Ecology 81:3178–3192. 27. Douvere F, Maes F, Vanhulle A, Schrijvers J (2007) The role of marine spatial planning 69. Ball I, Possingham H (2000) MarXan (V1.2) Marine Reserve Design Using Spatially in sea use management: The Belgian case. Mar Pol 31:182–191. Explicit Annealing: A Manual (Univ of Queensland, Brisbane, Australia). 28. Danter KJ, Griest DL, Mullins GW, Norland E (2000) Organizational change as a 70. Innes J, et al. (2004) Managing, Analysing and Presenting Public Submissions to component of ecosystem management. Soc Nat Resour 13:537–547. Achieve Marine Park Planning Outcomes: An Example from the Great Barrier Reef 29. Imperial MT (1999) Institutional analysis and ecosystem-based management: The Marine Park (Staff Paper, Great Barrier Reef Marine Park Authority, Townsville, institutional analysis and development framework. Environ Manage 24:449–465. Australia). Available at www.gbrmpa.gov.au/࿝data/assets/pdf࿝file/0005/8267/ 30. Westley F (2002) in Panarchy: Understanding Transformations in Human and Natural rap࿝public࿝submission࿝analysis.pdf. Systems, eds Gunderson LH, Holling CS (Island, Washington, DC), pp 333–360. 71. Great Barrier Reef Marine Park Authority (1994) A 25 Year Strategic Plan for the Great Barrier 31. Olsson P, Folke C, Hahn T (2004) Social-ecological transformation for ecosystem man- Reef World Heritage Area (Great Barrier Reef Marine Park Authority, Townsville, Australia). agement: The development of adaptive co-management of a wetland landscape in 72. Brown R (1997) Independent Review of the Great Barrier Reef Marine Park Authority. Southern Sweden. Ecol Soc 9:2. Available at www.gbrmpa.gov.au/࿝࿝data/assets/pdf࿝file/0011/2063/mp࿝007࿝full.pdf. 32. Olsson P, et al. (2006) Shooting the rapids: Navigating transitions to adaptive gover- 73. Great Barrier Reef Marine Park Authority (1999) Annual Report 1998–1999. Available nance of social-ecological systems. Ecol Soc 11:18. at www.gbrmpa.gov.au/corp࿝site/info࿝services/publications/annual࿝reports. 33. Fabricius C, Folke C, Cundill G, Schultz L (2007) Powerless spectators, coping actors, and 74. Great Barrier Reef Marine Park Authority (2003) Annual Report 2002–2003. Available adaptive co-managers: A synthesis of the role of communities in ecosystem manage- at www.gbrmpa.gov.au/corp࿝site/info࿝services/publications/annual࿝reports. ment. Ecol Soc 12:29. 75. Thompson L, Jago B, Fernandes L, Day J (2004) Barriers to Communication—How These 34. Duit A, Galaz V (2008) Governing complexity: Insights and emerging challenges.. Critical Aspects Were Addressed During the Public Participation for the Rezoning of Governance, in press. the Great Barrier Reef Marine Park (Staff Paper, Great Barrier Reef Marine Park 35. Baumgartner FR, Jones BD (1991) Agenda dynamics and policy subsystems. J Politics Authority, Townsville, Australia). Available at www.gbrmpa.gov.au/corp࿝site/ 53:1044–1074. management/zoning/zoning࿝publications. 36. True JL, Jones BD, Baumgartner FR (1999) in Theories of the Policy Process, ed Sabatier 76. Brock WA (2006) in Punctuated Equilibrium and the Dynamics of U.S. Environmental P (Westview, Boulder, CO), pp 97–115. Policy, ed Repetto R (Yale Univ Press, New Haven, CT), pp 47–77. 37. Repetto R, ed (2006) Punctuated Equilibrium and the Dynamics of U.S. Environmental 77. Trosper RL (2003) in Navigating Social-Ecological Systems: Building Resilience for Policy (Yale Univ Press, New Haven, CT). Complexity and Change, eds Berkes F, Colding J, Folke C (Cambridge Univ Press, 38. Kingdon J (1984) Agenda, Alternatives, and Public Policies (Little, Brown, Boston). Cambridge, UK), pp 328–351. 39. Farley J, et al. (2007) Opening the policy window for ecological economics: Katrina as 78. Huitema D, Kuks SMM (2004) in Integrated Governance and Water Basin Manage- a focusing event. Ecol Econ 63:344–354. ment: Conditions for Regime Change Towards Sustainability, eds Bressers JTA, Kuks 40. Ingram H, Fraser L (2006) in Punctuated Equilibrium and the Dynamics of U.S. SMM (Kluwer Academic, Dordrecht, The Netherlands), pp 61–100. Environmental Policy, ed Repetto R (Yale Univ Press, New Haven, CT), pp 78–109. 79. van der Brugge R, Rotmans J, Loorbach D (2005) The transition in Dutch water 41. Gunderson LH, Holling CS, Light S, eds (1995) Barriers and Bridges to the Renewal of management. Reg Environ Change 5:164–176. Ecosystems and Institutions (Columbia Univ Press, New York). 80. Carpenter SR, Folke C (2006) Ecology for transformation. Trends Ecol Evol 21:309–315. 42. Gunderson LH, Holling CS (2002) Panarchy: Understanding Transformations in Human 81. Young OR (1991) Political leadership and regime formation: On the development of and Natural Systems (Island, Washington, DC). institutions in international society. Int Org 45:281–308. 43. Chapin FS, III, et al. (2006) Policy strategies to address sustainability of Alaskan boreal 82. Manring SL (2007) Creating and managing interorganizational learning networks to forests in response to a directionally changing climate. Proc Natl Acad Sci USA achieve sustainable ecosystem management. Org Environ 20:325–346. 103:16637–16643. 83. Armitage D, Berkes F, Doubleday N (2007) Adaptive Co-management: Collaboration, 44. Westley F (1995) in Barriers and Bridges to the Renewal of Ecosystems and Institutions, Learning and Multi-level Governance (Univ of British Columbia Press, Vancouver, BC, eds Gunderson LH, Holling CS, Light S (Columbia Univ Press, New York), pp 391–427. Canada).

9494 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706905105 Olsson et al. Ecological and socioeconomic effects of China’s SPECIAL FEATURE policies for ecosystem services

Jianguo Liu*†, Shuxin Li*, Zhiyun Ouyang‡, Christine Tam§, and Xiaodong Chen*

*Center for Systems Integration and Sustainability, Department of Fisheries and Wildlife, Michigan State University, East Lansing, MI 48823-5243; ‡State Key Lab of Urban and Regional Ecology, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China; and §Department of Biology and Natural Capital Project, Woods Institute for the Environment, Stanford University, Stanford, CA 94305-5020

Edited by Gretchen C. Daily, Stanford University, Stanford, CA, and approved December 17, 2007 (received for review July 11, 2007)

To address devastating environmental crises and to improve human encompass 97% of China’s counties. Planned investment will well-being, China has been implementing a number of national exceed 700 billion yuan (at present, $1 U.S. ϭ 7.4 yuan). The policies on payments for ecosystem services. Two of them, the implementation of these programs is a milestone of China’s forest Natural Forest Conservation Program (NFCP) and the Grain to Green management; it marks the end of an era dominated by timber Program (GTGP), are among the biggest programs in the world production. because of their ambitious goals, massive scales, huge payments, and The NFCP and GTGP also have important global implications potentially enormous impacts. The NFCP conserves natural forests (1), although they were initially developed to address pressing through logging bans and afforestation with incentives to forest environmental problems in China. If implemented adequately and enterprises, whereas the GTGP converts cropland on steep slopes to sustainably, these two programs can generate many benefits to forest and grassland by providing farmers with grain and cash China and the rest of the world by addressing a wide array of subsidies. Overall ecological effects are beneficial, and socioeconomic environmental issues (e.g., biodiversity loss, climate change, deser- effects are mostly positive. Whereas there are time lags in ecological tification, droughts, floods, soil erosion, and water runoff) as well effects, socioeconomic effects are more immediate. Both the NFCP as socioeconomic challenges (e.g., poverty alleviation, social con- and the GTGP also have global implications because they increase flicts, and economic development) (1, 4, 5). vegetative cover, enhance carbon sequestration, and reduce dust to In this article, we provide an overview of the goals and payments other countries by controlling soil erosion. The future impacts of these for the NFCP and GTGP, illustrate their ecological and socioeco- programs may be even bigger. Extended payments for the GTGP have nomic effects as rigorously as possible from various sources of recently been approved by the central government for up to 8 years. literature, discuss future opportunities and challenges, and offer The NFCP is likely to follow suit and receive renewed payments. To recommendations to overcome their shortcomings and enhance make these programs more effective, we recommend systematic their potential. planning, diversified funding, effective compensation, integrated Natural Forest Conservation Program research, and comprehensive monitoring. Effective implementation SCIENCE of these programs can also provide important experiences and les- Background and Goals. According to the fifth national forest inven- sons for other ecosystem service payment programs in China and tory (1994–1998), the size of China’s natural forests was only 112 SUSTAINABILITY Ϸ many other parts of the world. million ha ( 70% of all forests), and most of these forests were degraded because of various human activities (e.g., pervasive conservation ͉ environment ͉ forests ͉ grassland ͉ sustainability logging). The overall goal of the NFCP is to protect and restore natural forests through such means as logging bans (Fig. 1). It is widely believed that achieving this goal can lead to many ecosystem ver the past three decades, China’s economy has grown the service benefits, such as soil erosion reduction, water retention, and Ofastest among all major nations. By contrast, China’s environ- flood control. To achieve this overall goal, the NFCP has also ment is increasingly deteriorating. For example, soil erosion is developed short-, medium-, and long-term goals as stepping stones. widespread, and ‘‘natural’’ disasters have caused devastating socio- The short-term goals (1998–2000) were to eliminate or reduce economic impacts (1). To mitigate the impacts of the degraded timber harvesting from natural forests and create alternative em- environment, China has been implementing large-scale conserva- ployment for traditional forest enterprises. The medium-term goals tion programs, including the Key Shelterbelt Construction Pro- (2001–2010) are to construct and protect forests for ecological gram, Beijing–Tianjin Sandstorm Control Program, Wildlife Con- benefits and to increase the capacity for timber harvesting from servation and Nature Reserve Development Program (2), and plantation forests. As a final goal (2011–2050), the NFCP aims to Forest Eco-Compensation Program (3). restore natural forests and meet domestic demand for timber in The severe droughts in 1997 and the massive floods in 1998 have plantation forests. prompted China to take two other unprecedented conservation The creators of the NFCP hoped to lower timber harvests in actions—the development and implementation of the Natural natural forests from 32 million m3 in 1997 to 12 million m3 in 2003, Forest Conservation Program (NFCP, also known as the Natural and plan to afforest 31 million ha by 2010 through mountain closure Forest Protection Program) and the Grain to Green Program (i.e., prohibition of human activities such as fuelwood collection and (GTGP, also known as the Sloping Land Conversion Program and grazing to allow regrowth) (12), aerial seeding, and artificial the Farm to Forest Program) (4–6). The NFCP conserves natural planting (4) (Fig. 2). The NFCP required that commercial logging forests through logging bans and afforestation with incentives to forest enterprises, whereas the GTGP converts cropland on steep slopes to forest and grassland by providing farmers with grain and Author contributions: J.L., S.L., Z.O., C.T., and X.C. designed research, performed research, cash subsidies. These actions resulted from the realization that analyzed data, and wrote the paper. those droughts/floods were at least partially caused by farming on The authors declare no conflict of interest. steep slopes and deforestation. This article is a PNAS Direct Submission. The NFCP and GTGP are two of the biggest programs offering †To whom correspondence should be addressed. E-mail: [email protected]. payments for ecosystem services in both China and worldwide in This article contains supporting information online at www.pnas.org/cgi/content/full/ terms of scale, payment, and duration (7–11). They are major 0706436105/DC1. components of China’s six key forest conservation programs, which © 2008 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706436105 PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9477–9482 Fig. 3. Cumulative amount of investment in NFCP (1998–2005). The dashed line indicates the goal for 2010. Data are from refs. 66 and 67.

observable indicators such as changes in harvested timber, newly forested area, and degree of soil erosion. By 2000, commercial harvesting of natural forests in 13 provinces had ceased, and the total area without logging had reached 8.9 million ha. The amount of timber harvested decreased 41% (from 18.5 to 11 million m3) in northeast China and Inner Mongolia Fig. 1. Current distribution of the NFCP and GTGP in China. Names of provinces, autonomous regions, municipalities, and two major rivers are between 1997 and 2003 (13). However, this change may be one shown. Data are from refs. 1 and 2 and supporting information (SI) Text. reason for increased timber imports from other countries. In 2002, the domestic production of commercial roundwood declined to 44.4 million m3, but the total import volume increased to 94.5 million m3 be completely banned in the upper reaches of the Yangtze and (14). In 2005, China imported 29.4 million m3 of logs, an increase Yellow rivers as well as in Hainan Province (Fig. 1) by 2000, and that of 10.4% from 2004 with imports from tropical forests accounting logging be substantially reduced elsewhere (4). for 7.4 million m3 (25% of total log imports). However, much of the The NFCP pilot study started in 12 provinces and autonomous imported timber to China was used to make products (e.g., furni- regions/municipalities in 1998. It was expanded to 18 provinces in ture) later exported to developed countries. 2000 (Fig. 1). The area under mountain closure and plantation increased rapidly over time and expanded to almost 11 million ha by 2005 (Fig. Payments. From 1998 to 2005, the NFCP received Ϸ61 billion 2). Native species (e.g., pine and China fir) are generally encour- yuan (Fig. 3). This investment was mainly used for payments to aged in the NFCP region although nonnative species (e.g., poplar cover economic losses of forest enterprises caused by the shift from and Hinoki cypress) were planted in some landscapes. Further- timber harvesting to tree plantations and forest management (2). more, only one or a few tree species were usually dominant in a The payments were linked to specific tasks: 1,050 yuan/ha for specific landscape, but there are efforts to diversify species com- allowing forest regeneration through mountain closure; 750 position (15). yuan/ha for aerial seeding; 3,000 and 4,500 yuan/ha for artificial Carbon sequestration has also increased. Between 1998 and 2004, ϭ 12 planting in the Yangtze and Yellow river basins, respectively; and 21.3Tg(1Tg 10 g) of carbon was sequestered in the new 10,000 yuan per worker for protecting 340-ha forest patches (4). plantations under the NFCP (of which, 6.4, 12.6, and 2.3 Tg of A total of 96.2 billion yuan has been designated for NFCP-related carbon were sequestered in northeastern China and Hainan Prov- activities from 2000 to 2010 (4), of which Ϸ50% has already been ince, the upper Yangtze River Basin, and the upper Yellow River spent (Fig. 3); Ϸ81.5% of this amount is anticipated from the Basin, respectively). In addition, wood production was reduced by 3 central government, and the remainder, from local governments. 9.7 million m , resulting in a reduction of carbon emissions of 22.8 Tg. The total carbon sequestered through the NFCP was 44.1 Tg Ecological Effects. Overall, progress has been made toward achiev- (1.2% of China’s CO2 emissions) from 1998 to 2004 (16). ing the goals of conserving and restoring natural forests. Most Soil erosion has been reduced, but the degree of reduction varied research and assessment efforts have focused on immediately among regions. Sample analyses of 22 counties in the upper and middle reaches of the Yangtze and Yellow river basins indicate that the area suffering from soil erosion declined by 6% from 1998 to 2003. In Sichuan Province, soil erosion was reduced by 1.5 billion tons (13). Wildlife habitat has also been improving. A long-term study in the Wolong Nature Reserve for the endangered giant pandas in Sichuan Province, which began 5 years before the NFCP started (17), demonstrates that illegal harvesting of natural forests has been rare (18) and that the panda habitat has been recovering since the NFCP’s inception (19).

Socioeconomic Effects. Whereas major types of NFCP impacts on forest enterprises had been anticipated before the NFCP started, the extent of some impacts has exceeded expectations. Further- more, there were unforeseen negative impacts on other industries and local governments. Fig. 2. Cumulative amount of land under the NFCP. The dashed line indicates Significant steps have been taken toward achieving the NFCP’s the goal for 2010. Data are from refs. 66 and 67. short-term goals in generating alternative jobs for those previously

9478 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706436105 Liu et al. in forest enterprises, eventually altering the employment and economic structure in forestry. Among the 1.2 million logging and SPECIAL FEATURE processing workers impacted by the NFCP, about two-thirds of them had retired or been transferred to other sectors by the end of 2002 (4). The dominant source of employment has shifted from logging to forest management and plantation farming. In the Chuannan Forestry Bureau and Ebian County of Sichuan Province, for instance, the percentages of staff in forest management increased from 0 and 13.1% in 1997 to 52.6% and 76.7% in 2001, respectively (20). Although there was a decline in income from forestry in some areas such as Longmenshan Township of Peng- zhou County in Sichuan Province, total income increased because of income from other sources such as tourism (21). As the central government anticipated, the economic structure in Fig. 4. Cumulative amount of land under the GTGP. The dashed line indi- many forest enterprises has changed from timber production only cates the goal for 2010. Data are from refs. 66 and 67. to multiple industries. Industries in China are classified in three groups: (i) ‘‘first industry’’ includes agriculture, forestry, animal husbandry, and fisheries; (ii) ‘‘second industry’’ includes mining; The GTGP aims to increase vegetative cover by 32 million ha by manufacturing; the production and supply of electricity, natural gas, 2010 (Fig. 4). Of this area, 14.7 million ha will be converted from and water; and construction; and (iii) ‘‘third industry’’ includes all cropland on steep slopes back to forest and grassland (2). Slope sectors that are not in the first or second industries (e.g., restaurants, steepness, Ն15° in northwestern China and Ն25° elsewhere (32), is hotels, and entertainment) (22). The average output of the third the main criterion by which plots are chosen for inclusion in the industry in 32 forest enterprises increased from 8.5% in 1997 to GTGP. The remaining portion of the 32 million ha of vegetative 20.1% in 2003 (13). However, for some enterprises, income from cover will be created by afforesting barren land. In addition to the the third industry decreased after the NFCP began. For example, primary goal of reducing environmental degradation, two associ- income from the third industry in the Chuannan Forestry Bureau ated goals with the GTGP are to alleviate poverty and to promote of Sichuan Province declined from 5.0 million yuan in 1997 to 1.1 local economic development (33). million yuan in 2001 because of a reduction in wood-related market The GTGP began its pilot study in three provinces (Sichuan, activities (23, 24). Shanxi, and Gansu) in 1999 (Fig. 1). It was expanded to 17 provinces Despite changes in economic structure, many forest enterprises in 2000 and finally to 25 provinces in 2002. The program focuses on have experienced hardship. Although retirees are covered by western China because of its ecological vulnerability since it con- government social security and pension systems (4), some enter- tains the headwaters of the Yangtze and Yellow rivers (Fig. 1) and prises cannot pay back their loans (25). By 2001, these loans reached accounts for Ϸ80% of the total area identified with soil erosion 12.9 billion yuan, and unpaid salaries amounted to 860 million yuan. problems (Ͼ360 million ha). Furthermore, most desertification In Sichuan Province, 1,172 wood-related industry enterprises and (174 million ha), three quarters of the cropland with a slope Ͼ25° SCIENCE

154,000 employees were impacted (26). Forestry workers depen- (600 million ha), and 60% of the population under the poverty line SUSTAINABILITY dent on income from timber harvesting suffered big economic are in western China (2, 31). losses. For instance, 55,000 people in Taijiang County of Guizhou Province lost Ϸ6 million yuan. This loss pushed some local forestry Payments. Under the GTGP, the government offers farmers 2,250 workers below the poverty line (27). and 1,500 kg of grain (or 3,150 and 2,100 yuan at 1.4 yuan per kg The NFCP has created budgetary burdens on some local gov- of grain) per ha of converted cropland per year in the upper reach ernments (26) because of partial funding responsibilities, resulting of the Yangtze River Basin and in the upper and middle reaches of in declines in local revenues (27, 28). For instance, from 1998 to the Yellow River Basin, respectively. In addition, 300 yuan/ha per 2001, the revenues in Yanbian County, Ebian County, Yanbian year for miscellaneous expenses and a one-time subsidy of 750 Forestry Bureau and Chuannan Forestry Bureau of Sichuan Prov- yuan/ha for seeds or seedlings are provided (34, 35). The duration ince decreased by 9.7, 2.8, 3.7, and 32.0 million yuan, respectively; of subsidies depends on the outcome of cropland conversion: 2 whereas matching funds from local governments accounted for years if the cropland is converted into grassland, 5 years if converted 13%, 44%, 21%, and 0.3% of the total investment from the central into economic forests by using fruit trees, or 8 years if converted to government (20). Taijiang County of Guizhou Province was unable ecological forests by using tree species such as pine and black locust to pay back a loan of 15.2 million yuan for developing a timber base (35). Furthermore, no taxes on the converted cropland are col- and had no funds for seedlings because of a decrease in commercial lected (4). timber revenues (27). However, in other places, such as those By the end of 2005, Ͼ90 billion yuan had been invested in the reported in the three case studies in the northeast and southwest GTGP (Fig. 5). The GTGP began to receive more cumulative (29), subsidies from the central government and other sources (e.g., investment than the NFCP did in 2004 when its budget amounted tourism) have offset losses in timber revenues. to Ϸ10 billion yuan more than that of the NFCP (Figs. 3 and 5). The planned total investment in the GTGP will reach 220 billion yuan Grain to Green Program by 2010 (Fig. 5). Background and Goals. Conserving natural forests through the NFCP is an important way to reduce soil erosion, but the most Ecological Effects. Similar to those of the NFCP, measures of the important contributor to this erosion is farming on steep slopes GTGP ecological effects are generally those immediately observ- (30). In the Yangtze and Yellow river basins alone, nearly 4.3 able: the amount of land converted and afforested, and the changes million ha of cropland were on slopes of Ն25°. To complement the in vegetative cover, water surface runoff, and soil erosion. Ecosys- effort of the NFCP, China initiated the GTGP, another large tem service changes on large scales, such as flood control, are ecosystem service payment policy, in 1999. Compared with the mainly inferred from changes in immediately observable factors. NFCP, the GTGP started a year later but is broader in terms of By the end of 2006, the GTGP had converted almost 9 million ha geographic extent (Fig. 1). The grain oversupply in the late 1990s of cropland into forest/grassland and had afforested 11.7 million ha as well as China’s increasing financial capability provided a stable of barren land (Fig. 4). In Guizhou Province alone, the forested area foundation for implementing the GTGP (31). increased by 952,000 ha (5.5%) between 2000 and 2005 (36). The

Liu et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9479 Socioeconomic Effects. The GTGP has generated more positive socioeconomic impacts than the NFCP. Unlike the NFCP, which has cut off income from timber harvesting for many forest workers, the GTGP has helped alleviate poverty. The GTGP has directly benefited 120 million farmers in Ͼ30 million households nationwide, whereas the NFCP has directly affected only hundreds of state-owned forest enterprises and indirectly impacted numerous households. The GTGP has improved the socioeconomic well- being of participating households in most areas (44). The vast majority of households surveyed were happy with the GTGP (33, 45). This program has helped numerous farmers to change their income structure by shifting from farming to other activities. In Fig. 5. Cumulative amount of investment in the GTGP from 1999 to 2005. Wuqi County of Shaanxi Province, for instance, 15,000 farmers Because data on separate investment in the GTGP for 1999 and 2000 are not switched from farming to mainly construction, transportation, and available, the combined amount of investment during these 2 years is shown in restaurant businesses between 1998 and 2003 after reforesting 2000. The dashed line indicates the goal for 2010. Data are from refs. 66 and 67. 103,700 ha of their cropland (46). The GTGP has generated a large number of surplus laborers and prompted many of them to seek jobs total amount of converted cropland nationwide is higher than the in cities, contributing to and facilitating the surge in migrant labor across China. For example, in Guizhou Province, the number of quota for conversion set by the central government (4). Further- migrant workers increased 48% (from 2.2 to 3.1 million) between more, the total amount of GTGP land has exceeded the total 2000 (before the GTGP) and 2005 (36). In Yiyang County of Jiangxi amount of NFCP land since 2002, with the difference increasing Province, the proportion of income earned by migrant workers over time (Figs. 2 and 4). The statistics of the State Forestry increased from one-third in 2000 to one-half in 2002 (47). Administration suggest that forest cover within the GTGP region Like the NFCP, the GTGP has created financial burdens for has increased 2% during 8 years. many local governments. Because no taxes on the converted The GTGP reduces surface runoff and soil erosion. In Hunan cropland have been collected since the program’s inception and Province, for instance, between 2000 (when the program began) agricultural taxes on the cultivated land are now also exempted, and 2005, soil erosion declined by 30%, and surface runoff dropped local governments lose tax revenues (4, 48). The central government Ϸ20% (37). In 14 counties of Sichuan Province from 1998 to 2003, provides only partial subsidies to local governments and has stated the forested area increased by 113,100 ha (12.7% of forested area that other expenses for the GTGP implementation (e.g., monitor- in 14 counties in 2003), and the area affected by soil erosion was ing and grain transportation) must be covered by them (24). The reduced 10% (from 13,528 to 12,171 km2) (38). In Zigui County of degree of loss depends on the region. In Kangding County of the Three Gorges Reservoir Region of Hubei Province (Fig. 1), Sichuan Province, for example, local government income decreased 3,085 ha of cropland (8.1% of total cropland in Zigui County) were 28% to 15 million yuan during 1999–2001 (49). converted to forest in 2000, lowering soil erosion by 54,900 tons/year The economic value of the ecosystem services produced by the during 2000–2005. Five years after GTGP implementation, con- GTGP is estimated to be very large, but the estimation methods are verted plots reduced surface runoff by 75–85% and soil erosion by somewhat controversial. For instance, the total economic value of 85–96% compared with cropland on steep slopes without the ecosystem services 4 years after GTGP implementation in the GTGP (39). 55,300-ha GTGP land of Zhangjiajie in Hunan Province was 428 The program also improves the physical properties of soil struc- million yuan, which is 11 times higher than the amount of direct ture and reduces nutrient loss for maintaining soil fertility and income from the same land in 2000 (before the GTGP). This value lowering river sediments. In the Chaigou Watershed of Wuqi includes ecosystem productivity, tourism, biodiversity, water and County of Shaanxi Province, the average soil moisture and mois- soil conservation, and pollution reduction (50). In Wuqi County of ture-holding capacity in GTGP plots after 5 years were 48% and Shaanxi Province, the estimated economic value from the first 6 55% greater, respectively, than those in non-GTGP plots (40, 41). years of the GTGP was 2.48 billion yuan (51). In Guizhou Province, loss of phosphorus was 35–53% less in GTGP Opportunities, Challenges, and Recommendations plots after 2–4 years than in non-GTGP plots (41). The GTGP conserves water resources and reduces desertifica- Both the NFCP and the GTGP have led to a series of accomplish- ments, but the originally planned duration of subsidies is too short tion. For example, 516,000 m3 of water were saved in 2003 through for forests to recover fully or for trees to grow large enough to yield reduced irrigation on 4,300 ha of GTGP land in Minqin County of sufficient harvest and income to offset losses from the converted Gansu Province (42), an area where the rate of desertification has land (2). Many studies have indicated that if subsidies end, it is dropped (42) because tree stems and leaves can absorb dust in the possible that some of the converted forest and grassland will be air, reduce wind speed on the soil surface by 30–50%, and increase converted back to cropland (45) and natural forests will be logged air humidity by 15–25% (42). again (52). Considering these and other factors, the central gov- Although vegetative cover and forested area have increased (36, ernment has recently extended the GTGP for another cycle of 2–8 40), diversity of tree species chosen for the GTGP is typically low, years. The years of extension are exactly the same as those in the and the tree species planted may not be the same as the original initial program: 2, 5, and 8 years if the cropland is converted into local species. Although the specific species planted (e.g., black grassland, economic forests, and ecological forests, respectively. locust, larch, and poplar) may vary across the GTGP region, GTGP The annual payments during the extension will be half of the land in many places is often dominated by a single or a few tree amounts in the initial program (53), but the 300 yuan/ha per year species. For instance, in Jiangxi Province, 60% of the converted for miscellaneous expenses will remain the same (53). The planned land in 2006 was planted with Oil Camellia. In Henan Province investment in the GTGP for the second cycle is Ϸ210 billion yuan, during 2000–2005, poplar accounted for 40% of the reforested area, and the total investment in the GTGP will be Ͼ430 billion yuan. It whereas other species accounted for Ͻ2% and fruit trees were is possible that the NFCP also will be extended to achieve the goals planted on the remaining area. established for the program by 2010 (Figs. 2 and 3) and 2050.

9480 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706436105 Liu et al. Although continuing these programs provides good opportuni- The central government has included eco-compensation in Chi- ties for restoring and conserving ecosystem services, there are also na’s 11th Five-Year Plan (2006–2010). The NFCP and GTGP SPECIAL FEATURE many challenges and unexpected outcomes as discussed above and should be incorporated into the eco-compensation system to pro- below. For instance, large gaps still exist between the achievements vide a legal basis for appropriate long-term payments to farmers so so far and the goals set for 2010 (Figs. 2–5). However, the that positive ecological effects of these programs can be sustained. experiences and lessons learned (13) from the policies on payments Offering job information and training to farmers and employees for ecosystem services in the past several years have laid a good of forest enterprises to develop new skills are indirect compensation foundation for their continuation and expansion. It is our hope that approaches and are sometimes more effective than grain and cash the following recommendations will make future endeavors more (36). Providing households with off-farm skills can help transform successful. farmers and employees of forest enterprises from directly depending on the land to finding off-farm jobs or creating new businesses, Systematic Planning. It would be more productive to carry out thus ultimately changing household economic structure and reduc- systematic planning at multiple government levels. This requires ing dependence on compensation. overall strategic planning at the central government level and detailed planning at the local government level with better inte- Interdisciplinary Research. Whereas many studies have assessed the gration. Instead of taking the traditional top-down approach, more ecological and socioeconomic effects of the NFCP and GTGP, they input and feedback from local people affected by the policies should are mostly scattered, fragmented, short-term, and opportunistic. be actively sought and incorporated into the decision-making For example, although both the NFCP and GTGP are concurrently process. implemented in many regions (Fig. 1), they are rarely discussed Complementing the GTGP and NFCP with other conserva- together (4, 54, 57). Little is known about their interactive effects. tion programs would improve the efficiency and effectiveness of Although both the NFCP and GTGP have significant ecological all programs. Other conservation programs include the other and socioeconomic consequences, they are often evaluated by four forest-related programs, the recently enacted forest eco- ecologists and social scientists separately. As these policies affect compensation program for conserving forests (75 yuan/ha per both ecosystems and humans, treating them as part of coupled year) in ecologically sensitive areas (e.g., along river banks) (3), human and natural systems (58, 59) would produce new insights into and possible future ecosystem service payment programs. In the complexity of the policies and their impacts. addition, linking the GTGP and NFCP with economic develop- The future successes of the NFCP and GTGP could benefit from ment programs can help generate more alternative off-farm a national network of interdisciplinary research on ecosystem income and thus reduce pressure on vulnerable ecosystems. services, with a particular focus on the NFCP and GTGP. The To make systematic planning sustainable and effective, laws network would coordinate and promote integrated social and should consider all ecosystem services, their interrelationships, ecological research on important questions from local to national and coordinated management. Although China already has levels. In addition to temporal comparisons (i.e., before/after the many laws related to natural resource management, most of NFCP and GTGP), using intervention analysis (60) would allow for them were developed piecemeal and have resulted in conflicts

more rigorous experimental approaches (e.g., treatment, with the SCIENCE among laws for forests and other resources (e.g., grasslands, soil, NFCP and/or GTGP, vs. control, without the NFCP and GTGP) to and water) (1, 54). SUSTAINABILITY evaluate outcomes of these programs. Spatially explicit modeling tools can help evaluate long-term ecological and socioeconomic Diversified Funding. So far, the NFCP and GTGP have been largely financed by the central government and have caused financial impacts under various policy scenarios (61, 62). hardships for some local governments. Establishing endowments for ecosystem services would be helpful for sustainability of the Comprehensive Monitoring. Comprehensive monitoring can help NFCP and GTGP, although annual government funds and contri- provide timely feedback for adjusting and refining large programs butions from other stakeholders are still important for continuation such as the GTGP and NFCP. Many advanced tools are available of these policies. for comprehensive monitoring. High-resolution remote-sensing Market-based mechanisms (55) should also be explored with data (e.g., IKONOS and QuickBird) can be useful (63–65), because assistance and support from the central government and other they can detect many ecological effects across large areas efficiently stakeholders (54). The NFCP and GTGP have many beneficiaries and quickly. Also, frequent surveys of various stakeholders can who could contribute to the payments, including hydropower generate timely information on socioeconomic effects. In some plants, insurance companies for flood and drought disasters, people NFCP and GTGP regions, some local officials have exaggerated and businesses in the lower reaches of the Yangtze and Yellow river reports on the amounts of cropland converted into forest or basins, and even other countries (e.g., Japan, Korea, and the United grassland to receive higher payments. A combination of remote- States) that benefit from an increase in vegetative cover to thwart sensing data, social surveys, or third-party involvement could help sandstorms originating in western China and improve carbon ensure accurate reporting. A web-based nationwide database would sequestration (56). facilitate the synthesis and dissemination of all relevant information (including detailed research and monitoring methods) for adaptive Effective Compensation. The amount and duration of compensation management of these programs. should be determined by multiple factors, including ecosystem values, risks to ecosystem services, basic needs of affected stake- Concluding Remarks holders, and benefits and costs across space and time. Current Despite the relatively short time since the NFCP and GTGP began, payments for the two programs are relatively uniform across space both programs have already demonstrated substantial ecological (33, 34), although there are large variations in the costs of imple- and socioeconomic impacts. Some goals (e.g., converting cropland menting the programs in different regions. In the past, annual to forest/grassland) have been overachieved, some (e.g., the logging subsidies remained basically constant despite increases in market ban) have been achieved, and some are still in progress. Although prices of agricultural products over time. Market prices of agricul- the programs have produced many positive ecological and socio- tural products also affect the opportunity cost of GTGP land. For economic outcomes, they have also generated some negative con- example, in Xiqu Township, Minqin County of Gansu Province, sequences. Whereas some socioeconomic effects are negative in the farmers lost 3,852–4,000 yuan/ha partially because of increased short run, structural changes in forestry and agriculture may prices for agricultural products in 2003 (42). ultimately benefit forest workers, farmers and other stakeholders.

Liu et al. PNAS ͉ July 15, 2008 ͉ vol. 105 ͉ no. 28 ͉ 9481 The impacts of these programs will be larger in the future as they Additional references can be found in SI Text. continue and as ecosystems recover. To make the GTGP and NFCP more successful, it is essential to ACKNOWLEDGMENTS. We thank Gretchen Daily and Pamela Matson for their invitation to write this paper. We also thank Joanna Broderick, develop and adopt new strategies to overcome their shortcomings Gretchen Daily, and two anonymous reviewers for their constructive com- and enhance their potential. These two programs provide impor- ments on an earlier draft. Financial support was provided by the National tant insights regarding opportunities and challenges in the devel- Science Foundation (Grants 0216450 and 0709717), National Aeronautics opment, implementation, and sustainability of similar ecosystem and Space Administration (Grants NNG06GI29G and NNX08AL04G), Na- tional Natural Science Foundation of China (Grant 40621061), National Key service payment programs, at present and in the future, both inside Basic Research Program of China (Grant 2006 CB403402), and Michigan China and around the world. Agricultural Experimental Station.

1. Liu J, Diamond J (2005) China’s environment in a globalizing world. Nature 435:1179– 36. Yang S (2006) Policy recommendations for Grain to Green Program in the ‘‘11th 1186. Five-Year Plan.’’ For Econ 9:7–10 (in Chinese). 2. Ouyang Z, ed (2007) Ecological Construction and Sustainable Development in China 37. Li D, Bo F, Tao J (2006) Achievements in and strategies for Grain to Green Program in (Science Press, Beijing) (in Chinese). Hunan Province. Hunan For Sci Technol 33:1–5 (in Chinese). 3. Ministry of Finance of China, State Forestry Administration of China (2007) Notice on 38. Bao J, Tang D, Chen B (2005) Socioeconomic effects of Grain to Green Program in Managing Forest Eco-Compensation Funds (State Forestry Administration of China, Sichuan Province. Sichuan For Explor Des 1:26–32 (in Chinese). Beijing) (in Chinese). 39. Wang Z, et al. (2007) Effects of Grain to Green Program on soil and water conservation 4. Xu J, Yin R, Li Z, Liu C (2006) China’s ecological rehabilitation. Ecol Econ 57:595–607. in Zigui County of the Three Gorges Reservoir Region. Sci Soil Water Conserv 5:68–72 5. Loucks C, et al. (2001) Giant pandas in a changing landscape. Science 294:1465. (in Chinese). 6. Zhang P, Zhou X, Wang F (1999) Introduction to Natural Forest Conservation Program 40. Liang W, Bai C, Sun B, Hao D, Qi J (2006) Soil moisture and physical properties of regions (Chinas Forestry Publishing House, Beijing) (in Chinese). under Grain to Green Program in the Gullied Rolling Loess Area. Soil Water Conserv 7. Daily GC, ed (1997) Nature’s Services: Societal Dependence on Natural Ecosystems China 3:17–18 (in Chinese). (Island Press, Washington DC). 41. Liu F, et al. (2002) Role of Grain to Green Program in reducing loss of phosphorus from 8. Daily GC, Ellison K (2002) The New Economy of Nature (Island Press, Washington, DC). yellow soil in hilly areas. J Soil Water Conserv 16:20–23 (in Chinese). 9. Li S, Zhai H (2002) A comparative study on forestry ecological projects in the world. Acta 42. Ma Y, Fan S (2005) Ecological-economic effects of Grain to Green Program in deserti- Ecol Sin 22:1976–1982 (in Chinese). fication areas. J Nat Resour 20:590–596 (in Chinese). 10. Wunder S (2005) Payments for Environmental Services (Center for International 43. Hou J, Zhang S (2002) Evaluating effects of Grain to Green Program in the Loess Plateau Forestry Research, Jakarta, Indonesia). Area. Bull Soil Water Conserv 22:29–31 (in Chinese). 11. Ferraro PJ, Kiss A (2002) Direct payments to conserve biodiversity. Science 298:1718–1719. 44. Xu Z, et al. (2006) Grain for green versus grain: Conflict between food security and 12. Editorial Department of Chinese Agriculture Encyclopedia (1989) Entry for “forestry.” conservation set-aside in China. World Dev 34:130–148. Chinese Agriculture Encyclopedia (Agriculture Publishing House, Beijing) (in Chinese). 13. Zhang Z (2006) Review of the 10th Five-Year Plan and perspective on the 11th Five-Year 45. Hu C, Fu B, Chen L, Gulinck H (2006) Farmers’ attitudes towards the Grain-for-Green Plan in relation to Natural Forest Conservation Program. For Econ 1:49–52 (in Chinese). programme in the loess hilly area, China. Int J Sustainable Dev World Ecol 13:211–220. 14. State Forestry Administration of China (2003) China Forestry Development Report 46. Ge W, Li L, Li Y (2006) On sustainability of Grain to Green Program. For Econ 11:33–49 (China Forestry Publishing House, Beijing) (in Chinese). (in Chinese). 15. An J (2002) Strategies and suggestions for conserving natural forests in Qilian Moun- 47. Wang H, Yan C, Jiang L (2007) A comparative study of effects of Grain to Green Program tain. For Gansu 6:16–17 (in Chinese). on farmers’ income in central and western regions of China. Acta Agric Univ Jiangx- 16. Hu H, Liu G (2006) Carbon sequestration of China’s Natural Forest Conservation iensis 29:318–322 (in Chinese). Program. Acta Ecol Sin 26:291–296 (in Chinese). 48. Huang Q (2007) On Grain to Green Program. Ecol Econ 2:60–63 (in Chinese). 17. Liu J, et al. (2001) Ecological degradation in protected areas. Science 292:98–101. 49. Dong J (2003) A comparative study of values between sloping fields and forestland. 18. Bearer S, et al. (2008) Effects of timber harvesting and fuelwood collection on giant China Popul Resour Environ 13:81–83 (in Chinese). panda habitat use. Biol Conserv, in press. 50. Zhuang D, Tang X (2006) Ecological-economic effects of Grain to Green Program in 19. Vina A, et al. (2007) Temporal changes in giant panda habitat connectivity across Zhangjiajie. J Mountain Sci 24:373–377 (in Chinese). boundaries of Wolong Nature Reserve, China. Ecol Appl 17:1019–1030. 51. Lai Y, Zhu Q, Zhang Y, Qin W, Li W (2006) Valuing ecological effects of Grain to Green 20. Liu C, Meng Q, Li Y, Lu J (2005) A case study on ecological and socioeconomic benefit Program in Wuqi County. J Soil Water Conserv 20:83–87 (in Chinese). evaluation of Natural Forest Conservation Program in Sichuan. Acta Ecol Sin 25:428– 52. Shen Y, Liao X, Yin R (2006) Measuring the socioeconomic impacts of China’s Natural 434 (in Chinese). Forest Protection Program. Environ Dev Econ 11:769–788. 21. Qiao R, Gao J, Zhang A (2006) Effects of Natural Forest Conservation Program on 53. State Council of China (2007) State Council’s Notice on Improving Grain to Green farmers’ income. Res Agric Modernization 27:40–43 (in Chinese). Program (State Forestry Administration of China, Beijing) (in Chinese). 22. National Bureau of Statistics of China (2003) Classification of Industries (Beijing) (in 54. Xu J, Katsigris E, White T, eds (2002) Implementing the Natural Forest Protection Program Chinese). and the Sloping Land Conversion Program (China Forestry Publishing House, Beijing). 23. Liu C (2004) The effects of Natural Forest Conservation Program on farmers’ income 55. Pagiola S, Bishop J, Landell-Mills N, eds (2002) Selling Forest Environmental Services and animal husbandry. For Sci Technol Manage 3:14–17 (in Chinese). (Earthscan, London). 24. Liu Y, Zhou Q (2005) Shortcomings in the incentives of Grain to Green Program. China 56. Huang F, Kang M, Zhang X (2002) Economic compensation strategies in the process of Popul Resour Environ 15:104–107 (in Chinese). converting cultivated land back into forests/grasslands. Acta Ecol Sin 22:471–478 (in Chinese). 25. Huang Y (2005) Countermeasures to the main problems in Natural Forest Conservation 57. Weyerhaeuser H, Wilkes A, Kahrl F (2005) Local impacts and responses to regional Program in Guizhou. Guizhou For Sci Technol 33:49–51 (in Chinese). forest conservation and rehabilitation programs in China’s northwest Yunnan prov- 26. Zhou X (2006) Countermeasures to problems in implementing Natural Forest Conser- ince. Agric Syst 85:234–253. vation Program in Sichuan Province. For Resour Manage 2:19–23 (in Chinese). 58. Liu J, et al. (2007) Complexity of coupled human and natural systems. Science 317:1513– 27. Yang L (2004) Natural Forest Conservation Program and forest eco-compensation 1516. system. J Mountain Agric Biol 23:158–163 (in Chinese). 59. Liu J, et al. (2007) Coupled human and natural systems. Ambio 36:639–649. 28. Wei Z (2004) An analysis of social and economic benefits of Natural Forest Conservation 60. Stewart-Oaten A, Bence JR (2001) Temporal and spatial variation in environmental Program in Xiaolongshan Forest Bureau. J Northwest For Univ 19:156–160 (in Chinese). impact assessment. Ecol Monogr 71:305–339. 29. Liu C (2002) An Economic and Environmental Evaluation of the Natural Forest 61. Liu J (1993) ECOLECON: A spatially explicit model for ECOLogical-ECONomics of species Protection Program (State Forestry Administration, Center for Forest Economic Devel- conservation in complex forest landscapes. Ecol Modell 70:63–87. opment and Research, Beijing), Working Paper. 30. Cha S, Li Q (1998) Water and soil loss of sloping field in the Three Gorges Area. 62. An L, Linderman M, Shortridge A, Qi J, Liu J (2005) An agent-based spatial model for Agro-Environ Dev 2:31–34 (in Chinese). cross-discipline and cross-scale integration. Ann Assoc Am Geogr 95:54–79. 31. Wang R, Chen K (2006) Analysis of the current status and problems in Grain to Green 63. Du L (2006) Remote sensing technology and application perspectives for monitoring Program in China. Chinese Agric Sci Bull 13:188–192 (in Chinese). Grain to Green Program. Remote Sens Technol Appl 21:477–482 (in Chinese). 32. Uchida E, Xu J, Rozelle S (2005) Grain for green: Cost-effectiveness and sustainability 64. Huang J, Ju H, Zhao F, Yue Y, Zhang L (2004) Application of high resolution Quickbird of China’s conservation set-aside program. Land Econ 81:247–264. satellite data in Grain to Green Program. For Res 17:167–271 (in Chinese). 33. Xu J, Cao Y (2002) On sustainability of converting farmland to forests/grasslands. Int 65. Shi L, Du J, Tan B (2006) Application of SPOT-5 data in monitoring Grain to Green Econ Rev Z2:56–60 (in Chinese). Program. For Resour Manage 2:84–87 (in Chinese). 34. Feng Z, Yang Y, Zhang Y, Zhang P, Li Y (2005) Grain-for-green policy and its impacts on 66. State Forestry Administration of China (2006) Bulletin of the Six Major Forestry Projects grain supply in West China. Land Use Policy 22:301–312. in 2005 (State Forestry Administration of China, Beijing) (in Chinese). 35. Xu J, Dao R, Xu Z (2004) Grain to Green Program: Effectiveness, effects of structural 67. State Forestry Administration of China (2000–2007) China Forestry Development adjustment, and economic sustainability. China Econ Q 4:139–162 (in Chinese). Report (China Forestry Publishing House, Beijing) (in Chinese).

9482 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0706436105 Liu et al. Field evidence that ecosystem service projects support biodiversity and diversify options Rebecca L. Goldman*†, Heather Tallis‡, Peter Kareiva§, and Gretchen C. Daily‡

*Interdisciplinary Program in Environment and Resources, Department of Biological Sciences, 371 Serra Mall, Stanford University, Stanford, CA 94305-5020; ‡Department of Biological Sciences, 371 Serra Mall, Stanford University, Stanford, CA 94305-5020; and §The Nature Conservancy, 4722 Latona Avenue NE, Seattle, WA 98105

Edited by Harold A. Mooney, Stanford University, Stanford, CA, and approved April 24, 2008 (received for review January 8, 2008) Ecosystem service approaches to conservation are being champi- ating protected areas). There is great need for such understand- oned as a new strategy for conservation, under the hypothesis that ing, given the rapid development and deployment of ecosystem they will broaden and deepen support for biodiversity protection. service approaches globally. Where traditional approaches focus on setting aside land by We conducted a case study of The Nature Conservancy purchasing property rights, ecosystem service approaches aim to (TNC), the world’s largest conservation organization investing engage a much wider range of places, people, policies, and finan- more than $700 million annually in conservation in more than 30 cial resources in conservation. This is particularly important given countries on five continents. Using only TNC projects provides projected intensification of human impacts, with rapid growth in a relatively large and homogeneous sample of conservation population size and individual aspirations. Here we use field efforts, all following the same methodology and all reporting research on 34 ecosystem service (ES) projects and 26 traditional their project design with the same terminology and framework biodiversity (BD) projects from the Western Hemisphere to test (‘‘conservation by design’’; see ref. 19). Additionally, TNC whether ecosystem service approaches show signs of realizing maintains a project database that can be searched by geographic their putative potential. We find that the ES projects attract on region so sampling is unbiased without regard to success or other average more than four times as much funding through greater attributes. corporate sponsorship and use of a wider variety of finance tools We developed a database (supporting information (SI) Ap- than BD projects. ES projects are also more likely to encompass pendix) of biodiversity-oriented (BD) and ecosystem service- working landscapes and the people in them. We also show that, oriented (ES) projects focusing on the Western Hemisphere despite previous concern, ES projects not only expand opportuni- because of the longevity of TNC investments in this area—the ties for conservation, but they are no less likely than BD projects to United States since 1954 and Latin America for almost 30 yr. include or create protected areas. Moreover, they do not draw (Only within the last decade has TNC started working beyond down limited financial resources for conservation but rather en- the Western Hemisphere.) TNC projects are sustained efforts at gage a more diverse set of funders. We also found, however, that protecting species, habitats, or community types following ex- monitoring of conservation outcomes in both cases is so infrequent plicit conservation objectives. All projects entail a written de- that it is impossible to assess the effectiveness of either ES or BD scription of the strategies used to accomplish conservation, approaches. ranging from the establishment of a nature reserve or the purchase of conservation easements to advocacy for local zoning conservation ͉ conservation organizations ͉ protected area ͉ laws that restrict certain land uses. All have biodiversity goals or working landscapes focal species or habitats (etc.) they aim to protect. For our study, we defined projects with only biodiversity goals as ‘‘BD projects.’’ ‘‘ES projects’’ are those that have biodiversity cosystem service approaches to conservation offer a prom- goals, but in addition have an explicit goal or strategy of at least Eising way to align conservation and production, simulta- one ecosystem service such as water purification, carbon seques- neously enhancing human well being and protecting Earth’s tration, and opportunity for ecotourism. To be an ES project, biodiversity and life-support systems (1–10). Developing market- reference to an ecosystem service must be explicit. This is not an based mechanisms for ecosystem services by ascribing them outcome-based definition but rather a process-based one. value, both economic and social, may help diminish poverty and TNC works via partnership: the project participants in our improve human welfare (11–15). These approaches offer hope database include 8 private corporations; 32 federal, 29 state, and for making conservation mainstream, by enlisting the support of

24 local government agencies; 67 nongovernmental organiza- SCIENCE a greater number and variety of funders and partners, by tions (NGOs); 12 universities; and 2 private landowners. The SUSTAINABILITY spanning the continuum of ‘‘wild’’ to human-dominated places, NGOs range from national (e.g., Tro´pico in Bolivia) to multi- and by broadening the financial and institutional approaches lateral organizations (e.g., the World Bank). Thus, a focus on used for conservation. At the same time, there is a risk that by TNC actually draws from the ES and BD conservation efforts of straying from a pure focus on nature reserves and biodiversity, a global array of institutions. conservation projects that address ecosystem services may de- We collected information about project strategy and structure tract from biodiversity protection (16). through a series of semistructured, open-ended interviews Previous research and reviews have demonstrated the use of [rather than surveys because interviews are more appropriate for specific tools for including ecosystem services in conservation (e.g., 11, 17, 18). Here, we provide the first quantitative comparison of conservation projects focused in part on ecosystem Author contributions: R.L.G. and H.T. designed research; R.L.G. performed research; R.L.G., services (ES) and those oriented more traditionally around H.T., and P.K. analyzed data; and R.L.G., H.T., P.K., and G.C.D. wrote the paper. biodiversity (BD) alone. Our aim is to test whether ecosystem The authors declare no conflict of interest. service projects attract new and more diverse financial support This article is a PNAS Direct Submission. and to explore other differences between these two project types. †To whom correspondence should be addressed. E-mail: [email protected]. In particular, we ask whether ecosystem service projects expand This article contains supporting information online at www.pnas.org/cgi/content/full/ conservation options (finance tools, actions, and landscapes) 0800208105/DCSupplemental. without neglecting traditional approaches (maintaining and cre- © 2008 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0800208105 PNAS ͉ July 8, 2008 ͉ vol. 105 ͉ no. 27 ͉ 9445–9448 Fig. 1. Project funding information. Corporate funders are private, for-profit organizations. Federal, state, and local funders refer to levels of government. Nonprofit donors are nongovernment organizations. Education donors have a connection to a university, and private individuals refer to private landowner donations. (a) Proportion of funder types by project type. On average, 27% of funders are corporate for ES projects whereas only 8% are for BD projects (P ϭ 0.010, n ϭ 57 using 9,999 permutations in a randomization experiment). (b) The subset (17 ES, 16 BD) of projects with exact funding information and the average percentage of overall funding coming from the different funding sources. ES projects have significantly more revenue sources overall, particularly from corporate sources, and on average, an ES project receives 42% of its funding from corporate sources compared with an average of 14% for BD projects (P ϭ 0.023, n ϭ 33). *, P Ͻ 0.05; **, P Ͻ 0.01. The error bars represent SD. capturing details and nuances in case studies (20)] of TNC 5.80, P ϭ 0.02); the difference is even more dramatic outside the personnel and partners to determine characteristics of on-the- United States where 50% of BD projects used no finance tool 2 ground implementation of ES and BD projects. Attributes compared with 5% of ES projects (Pearson ␹ 1 ϭ 8.89, P ϭ included goals (both broad and specific goals such as species, 0.003) (Fig. S1). ES projects also draw on a broader portfolio of habitats, and services targeted), partners, funding, landscapes finance tools. Traditionally, conservation organizations have encompassed, activities promoted, and monitoring involved (SI relied on land purchase and purchase of property rights (ease- Appendix). All recorded information was checked by the inter- ments) (21), both of which are expensive and can be restrictive viewee in the database. We sampled 60 projects: 34 ES and 26 outside the United States where many such organizations cannot BD with a relatively similar geographic breakdown to account own property. BD projects employ land purchasing 1.5 times and 2 for differences in project implementation based on geography. easements almost 2.5 times more often (Fig. 2; Pearson ␹ 1 ϭ For ES projects, 35% were in South America, 24% in Meso- 7.27, P ϭ 0.007) than ES projects. Additionally, ES projects use America and Caribbean region, and 41% in the United States; markets (e.g., creating carbon credits, mitigation banking, or- BD projects were 28%, 16%, and 56%, respectively. We had a ganic products) and user access fees (e.g., water use or ecotour- 74% response rate of contacts responding to our interview ism) significantly more frequently than BD projects (Fig. 2; 2 2 request. Our initial sampling aimed for 30 projects of each type, Pearson ␹ 1 ϭ 3.79, P ϭ 0.052 and ␹ 1 ϭ 7.06, P ϭ 0.008, but four of the BD projects sampled turned out to be ES projects respectively). during interviewing. For the subset of ES (n ϭ 17) and BD (n ϭ 16) projects for which we received exact funding information, ES projects, on average, obtain more than four times the revenue of BD projects (one-way ANOVA, F1,33 ϭ 5.57; df ϭ 1; P ϭ 0.02). Looking at our full sample of projects, ES projects also have significantly more funders [3.16 Ϯ 2.03 (mean Ϯ SD)] than do BD projects 2 (2.24 Ϯ 1.88) (Wilcoxon–Mann–Whitney U test; ␹ 1 ϭ 6.32; P ϭ 0.012). Although we cannot determine why ES projects received more money from more sources, it is evident that ES projects are more successful at securing corporate funding. On a per-project basis, ES projects have at least one corporate funder nearly three times more often than BD projects (P ϭ 0.009, n ϭ 57). ES projects also engaged significantly more corporate funders overall (27% of funders) as compared with BD projects (7% of funders) (Fig. 1a; P ϭ 0.010, n ϭ 57). In addition, again focusing on projects Fig. 2. Greater diversity and number of finance tools are used in ES projects. with exact financial breakdowns, on average ES projects received Easement purchase involves purchase of particular property rights. The other much more (42%) of their funding on a per project basis from markets category includes carbon credits, mitigation banking, habitat bank- corporate sources compared with BD projects (14%) (Fig. 1b; ing, and specialized-product markets (e.g., organic products). Taxes and sub- P ϭ 0.023, n ϭ 33). sidies involve creating new as well as redistributing existing ones. User fees are In addition to funding sources, the actual finance tools (e.g., for water or ecotourism. Projects that use no finance tools are also indicated. purchasing land rights, selling carbon credits, designing new More than 50% of BD projects use land and easement purchasing whereas subsidies, etc.) used as incentives to advance conservation are Ͻ40% of ES project purchase land and Ͻ25% purchase easements. Targeting ecosystem services opens up the ability to use finance tools such as markets important. When no finance tools are used, conservation is (almost 40% of ES projects use a market) and user access fees (almost 25% use imposed without any compensation to affected parties. A total this finance tool). In general, ES projects use more finance tools (Ͼ97% use at absence of finance tools is almost eight times more common in least one tool) than BD projects (only Ϸ75% use one, mostly in the United 2 BD than in ES projects (Fig. 2; 23% versus 3%, Pearson ␹ 1 ϭ States). *, P Ͻ 0.05; **, P Ͻ 0.01.

9446 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0800208105 Goldman et al. Fig. 4. Types of landscapes and landowners included in ES and BD projects. Projects that initiate sustainable agriculture introduce sustainable techniques (such as contour farming, no till, organic crops). Projects that maintain private landownership involve no private land purchasing within the project area. Fig. 3. Types of preservation activities used in ES and BD projects. There is no Land starts as private and remains private. Significantly more ES projects significant difference between investments in enforcing legal protection of target agricultural landscapes (75% compared with 46% of BD projects) previously designated protected areas, designating new protected areas for initially. More ES projects also work to maintain human use by initiating conservation, and hiring park guards (77%, 56%, 19%, respectively, of ES sustainable agriculture (44%) and maintaining private landownership (50%) projects; 59%, 73%, 21%, respectively, of BD projects) to ensure protected 2 2 as opposed to BD projects (Ͻ20% initiate sustainable agriculture and only areas are properly managed (Pearson ␹ 1 ϭ 1.88, P ϭ 0.17; ␹ 1 ϭ 2.17, P ϭ 0.14; 2 23% maintain private ownership). *, significant P Ͻ 0.05. ␹ 1 ϭ 2.24, P ϭ 0.13, respectively).

2 Both ES and BD projects employ a wide range of institutional project area, aiming to prevent direct human use (Pearson ␹ 1 ϭ tools (defined as changing or creating policy or altering a law). 4.11; P ϭ 0.043). Both ES projects and BD projects work with These institutional tools include creating or redistributing a tax, private landowners, but significantly more ES projects keep land subsidy, and/or a fee. They also encompass legal alteration of in private ownership rather than negotiating a land purchase 2 ownership rights, development rights, or administration rights. (Pearson ␹ 1 ϭ 4.52; P ϭ 0.034) (Fig. 4 and SI Appendix). Both Within the Unites States, BD projects rely heavily on land BD and ES projects invest in protection, but ES projects are purchase and do not emphasize institutional change (only 30% significantly more likely to invest in working landscapes, invest- of BD projects in the United States work to affect policy). In ments critical for conservation success. contrast, significantly more ES projects, 70% in total, alter ES projects are being implemented exactly as one might expect 2 an institutional policy (Fig. S1; Pearson ␹ 1 ϭ 5.14, P ϭ 0.023). In given the link between ecosystem services and human well being. the United States, some of these institutional policy changes Compared with BD projects, ES projects are obtaining greater in ES projects include changing ownership rights of a dam to revenue overall and more funding from corporate sources, using allow decommissioning, selling rights to carbon credits, and a wider variety of financial incentives, and giving much more being granted access rights to particular forests. In Latin Amer- attention to private lands, especially those used for agriculture. ica, one recent example is changing administration rights but not This final point emphasizes the essential role ES projects will ownership rights of a park from public control to nonprofit likely play in the future because agricultural lands are projected control. Affecting policy will likely lead to project longevity to expand by at least 120 million hectares by 2030 (26). Pursuing because it institutionalizes conservation efforts. conservation in pristine and working landscapes will be essential, Given the changes ES projects bring to conservation practice, and ecosystem services give us a tool to use in both. it is important to ask whether these projects fail to address the The ultimate question is whether either of these strategies is conservation of protected areas. The answer is no. We consid- enhancing biodiversity—the mission of TNC and many other ered enforcing legal protection, designating new protected areas, conservation organizations. Our sample of 60 BD and ES or hiring park guards as means of protected area conservation projects revealed that Ͻ20% of all projects are doing compre- and found no significant difference between the proportions of hensive monitoring (systematic monitoring geared to at least two projects engaging in these strategies (Fig. 3). Overall, 76% of ES desired outcomes/goals). Most of the biodiversity monitoring projects and 92% of BD projects engage in at least one protected involved sampling the change in abundance of a particular SCIENCE area strategy.

species of rare plant or animal over time. In addition, only 26% SUSTAINABILITY Although ES and BD projects have similar commitments to of ES projects are monitoring any ecosystem service outcomes, protected areas, ES projects differ in also pursuing conservation and the vast majority of this monitoring involves ensuring that outside reserves, a strategy limited by both political and socioeconomic constraints (22). Conservation outside reserves is also carbon projects meet their targets, and nothing beyond that. critically important given continuing rapid growth of both the Thus, we do not have the data to assess whether either ES or BD human population and human-dominated lands (2) and has been projects are delivering their conservation promises. the subject of a larger debate (for example, see ref. 23). The absence of measures for monitoring conservation projects Agricultural and pasture lands represent Ϸ40% of global land is widely appreciated (27), but ours is one of the first quantitative surface (24), and these lands can provide important contribu- reports of what proportion of projects do monitor. Although the tions to biodiversity protection (25). It is striking the extent to data we report show that ES projects are broadening the scope which ES projects target agricultural lands compared with BD and support for conservation, the question of outcomes is 2 projects (75% and 46%, respectively; Pearson ␹ 1 ϭ 5.83, P ϭ critical. In many cases public funds and taxes are being used to 0.012) (Fig. 4). Approximately 44% of ES projects initiate finance ecosystem service projects with the promise of better sustainable agriculture (e.g., introduce contour plowing, con- services being delivered to people. If we fail to monitor and servation tillage, organic farming, etc.) as a strategy for achiev- evaluate the delivery of ecosystem services, we risk alienating the ing sustainable human activity within the project area. By new support base that ecosystem service approaches are bringing contrast, Ͻ20% of BD projects maintain any agriculture in the to conservation.

Goldman et al. PNAS ͉ July 8, 2008 ͉ vol. 105 ͉ no. 27 ͉ 9447 Materials and Methods involved; financial, institutional, and social tools used; on-the-ground conser- Data Sampling. From November 2006 to July 2007, we interviewed relevant vation activities in use; details on who pays and who receives payment for personnel (such as project managers and field coordinators) at The Nature ecosystem services, when relevant; types of monitoring and policy analysis Conservancy (TNC) both in person and over the phone to sample a variety of used; and lessons learned. Interviews were conducted in person (27 projects) ecosystem service and biodiversity projects. When TNC’s role in a project was or over the phone (33 projects) and lasted anywhere between 45 min and 6 h relatively peripheral, we talked with key partners. We conducted Ͼ70 inter- (if a tour of the project site was involved). Interview transcripts based on notes views and sampled 60 projects: 34 ES projects and 26 BD projects. We limited taken during the interview were used to upload the information into the our sampling to three TNC regions: South America, Meso-America, and the database. R.L.G. was responsible for all interviews, note taking, and data entry Caribbean (which includes Mexico), and the United States. thereby eliminating inconsistencies in the data collection process. For ES projects, we tried to sample comprehensively because, compared The database was designed for both quantitative and qualitative data. with BD projects, there are relatively few. TNC traditionally and presently Quantitatively, the database consists of a series of checkbox sets defining a focuses on biodiversity projects with hundreds of such projects ongoing. ES range of possible outcomes for a given project characteristic (see Checkbox projects, on the other hand, are far fewer, and we attempted to sample as Definitions in SI Appendix). For example, we created a list of possible threats many of such projects as exist in the Western Hemisphere. Most ES projects at a project could combat (based on the International Union for Conservation of TNC started within the last decade and many in the last 5 yr. To comprehen- Nature threat list that TNC uses for their project definitions) and incorporated sively sample ES projects we used email correspondence with each regional this list as a checkbox set within the database where for each project we director (as defined by TNC) asking for project managers to contact us if they ‘‘checked’’ all applicable threats. These sets were coded to record presence/ absence of project characteristics as well as summations of various attributes had a project in the region they considered an ES project. We also then used (e.g., total number of corporate donors). a snowballing technique, where project managers would tell us about other Qualitatively, the database has a number of ‘‘fill-in’’ boxes in which text can projects they knew of, to acquire other ES projects that did not appear in the be entered to capture the more nuanced, unique features of each project (see first sampling. The request was sent out with an outline of our definition of an Fill-In Box Definitions in SI Appendix). All checkboxes and fill-in boxes were ES project. We sent around this request twice to each region. completed by using interview transcripts. To sample the vast array of BD projects, we used TNC’s basic repository of After completion of interviews in July 2007, each project entry was sent to projects to randomly select 40 projects all of which had a project start date of the interviewee(s) for him/her to approve and correct, as appropriate. Based 1990 or more recently as this was as old as the oldest ecosystem service project. on these responses, the database was updated and changed to better reflect We used a random number generator to select projects. We used a stratified the actual project. random sampling technique to ensure relevant and accurate coverage of projects in the three main regions of study and to ensure a similar geographic breakdown of projects to the ES projects. We originally tried to sample the Data Analysis. Categorical presence/absence data were analyzed by using the 2 same number of ES and BD projects. Because we found 30 ES projects, we Pearson ␹ test appropriate for this sample size (28) in JMP version 5.0.1.2 (29). randomly selected 30 BD projects. Upon conducting interviews, however, we Summary data of total project funding by project type used a one-way ANOVA found that four projects that appeared to be BD projects were actually ES after a log transformation to normalize the data (29). We assessed the projects by our definition. independence of our project types based on likelihood of corporate funding For each project, we conducted a semistructured interview based on a and amount of total funding from corporate sources using a one-tailed database we designed to assess, analyze, and compare the projects across a probability distribution permutation test with 9,999 permutations coded in variety of different characteristics (see Interview Protocol in SI Appendix). We MATLAB following Sokal and Rohlf (30). chose an interview approach rather than a survey approach to enable an in-depth and rigorous comparison of the two conservation approaches (18). ACKNOWLEDGMENTS. We thank all TNC staff and partners who helped make Semistructured interviews help minimize the amount of information left up to this research possible. We thank K. Brauman, B. Brosi, J. Goldstein, H. Mooney, and L. Pejchar for their feedback on previous versions of this manuscript. We interpretation. We were able to explain terminology and ensure that inter- also thank two anonymous reviewers for very helpful comments. We thank viewees were answering the intended question. Teresa Heinz Environmental Scholars fellowship for funding this research and We sampled for information about project goals; targets (species, habitat, the National Science Foundation Graduation Research Fellowship for support- ecosystem services); threats; partners and partner motivations; funding infor- ing R.L.G. Additionally, we thank P. and H. Bing, V. and R. Sant, and T., W., and mation (who and why); types of landowners, land uses, and land covers K. Wirth for their generosity in supporting The Natural Capital Project.

1. Chapin, FS III, et al. (2000) Consequences of changing biodiversity. Nature 405:234–242. 16. Schwartz M, et al. (2000) Linking biodiversity to ecosystem function: Implications for 2. Millenium (2003) Ecosystem and human well-being. Millennium Ecosystem Assessment conservation ecology. Oecologia 122:297–305. (MA) (Island Press, Washington, DC) pp 1–235. 17. Gouyon A (2003) Rewarding the Upland Poor for Environmental Services: A Review of 3. Matson P, Parton W, Power A, Swift M (1997) Agricultural intensification and ecosys- Initiatives from Developed Countries [World Agroforestry Center (ICRAF), Bogor, tem properties. Science 277:504–509. Indonesia]. 4. Janzen D (1998) Gardenification of wildland nature and the human footprint. Science 18. FAO (Food and Agriculture Organization) (2004) Payment schemes for environmental 279:1312–1313. services in watersheds. Proceedings of a Regional Forum (Regional Office for Latin 5. Heal G (2000) Nature and the Marketplace: Capturing the Value of Ecosystem Services American and the Caribbean, Arequipa, Peru). Available at http://www.fao.org/ (Island Press, Washington, DC). docrep/006/y5305b/y5305b01.htm. 6. Singh S (2002) The biodiversity crisis: A multifaceted review. Curr Sci 82:1331–1335. 19. The Nature Conservancy (2006) Conservation by Design, A Strategic Framework for 7. Balvanera P, et al. (2001) Conserving biodiversity and ecosystem services. Science Mission Success. Available at http://www.nature.org/aboutus/howwework/cbd/files/ 291:2047–2048. cbd.pdf. 8. Daily G, Ellison K (2002) The New Economy of Nature: The Quest to Make Conservation 20. Yin R (2003) Case Study Research: Design and Methods (Sage Publications, Thousand Profitable (Island Press/Shearwater Books, Washington, DC). Oaks, CA), 3rd Ed. 9. Swift MJ, Izac A-M N, van Noordwijk M (2004) Biodiversity and ecosystem services in 21. Merenlender A, Huntsinger L, Guthey G, Fairfax S (2004) Land trusts and conservation agricultural landscapes – are we asking the right questions? Agric Ecosyst Environ easements: Who is conserving what for whom? Conserv Biol 18:65–76. 104:113–134. 22. Polasky S, Nelson E, Lonsdorf E, Fackler P, Starfield A (2005) Conserving species in a 10. Chan K, Shaw M, Cameron D, Underwood E, Daily G (2006) Conservation planning for working landscape: Land use with biological and economic objectives. Ecol Appl ecosystem services. PLoS 4:2138–2152. 15:1387–1401. 11. Landell-Mills N, Porras I (2002) Silver bullet or fool’s gold? A global review of markets 23. Schwartzman S, Nepstad D, Moreira A (2000) Arguing tropical forest conservation: for forest environmental services and their impacts on the poor, Instruments for People versus parks. Cons Biol 14:1370–1374. Sustainable Private Sector Forestry Series (International Institute for Environment and Development, London). 24. Foley J, et al. (2005) Global consequences of land use. Science 309:570–574. 12. Pagiola S, Bishop J, Landell-Mills N, eds (2002) Selling Forest Environmental Services: 25. Tilman D, Cassman KG, Matson PA, Naylor R, Polasky S (2002) Agricultural sustainability Market-Based Mechanisms for Conservation and Development (Earthscan, London). and intensive production practices. Nature 418:671–677. 13. Pagiola S, Arcenas A, Platais G (2002) Can payments for environmental services help 26. Bruinsma J, ed (2003) World Agriculture: Towards 2015/2030, an FAO Perspective reduce poverty? An exploration of the issues and the evidence to date from Latin (Earthscan, London). America. World Devel 33:237–253. 27. Ferraro PJ, Pattanayak SK (2006) Money for nothing? A call for empirical evaluation of 14. Greig-Gran M, Porras I, Wunder S (2005) How can market mechanisms for forest biodiversity conservation investments. PLoS 4:482–488. environmental services help the poor? Preliminary lessons from Latin America. World 28. Agresti A (2002) Categorical Data Analysis (Wiley, New Jersey), 2nd Ed. Devel 33:1511–1527. 29. SAS Institute (1989–2000) JMP (SAS Institute, Cary, NC), Version 5.1. 15. Pfaff A, et al. (2007) Will buying tropical forest carbon benefit the poor? Evidence from 30. R. Sokal R, Rohlf F (1995) Biometry: The Principles and Practice of Statistics in Biological Costa Rica. Land Use Policy 24:600–610. Research (W.H. Freeman, New York), 3rd Ed.

9448 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0800208105 Goldman et al.