Ecoregions in Context: a Critique with Special Reference to Indonesia

PAUL JEPSON* AND ROBERT J. WHITTAKER School of Geography and the Environment, University of Oxford, Mansfield Road, Oxford OX1 3TB, United Kingdom

Abstract: World Wildlife Fund–United States ( WWF ) is promoting an ecoregional framework internation- ally as a new hierarchical approach to organizing and prioritizing conservation efforts. We assessed WWF ecoregions against existing frameworks: (1) the Dasmann-Udvardy ( World Conservation Union [IUCN]) Bio- geographical Representation Framework, (2) the Bailey Ecoregional Framework (U.S. Forest Service), and (3) the hotspot approach, as exemplified by the BirdLife Endemic Bird Area Approach and the WWF–IUCN Centres of Plant Diversity Program. We examined the genealogy of the schemes from three perspectives: meth- odological explicitness, transparency and repeatability, and whether the WWF–ecoregions system improves on existing schemes. We considered Indonesia as a case study and assessed the efficacy of each system in the Indonesian context. The existing planning frameworks achieved their objective; in general had explicit, trans- parent, and repeatable methods; and, in the case of the Dasmann-Udvardy system, attained an institutional reality in Indonesia. The central purpose of the WWF–ecoregions framework is the same as the 25-year-old Dasmann-Udvardy system, and at the coarsest spatial scales it relies on similar spatial delineators ( biomes and faunal regions). The WWF methodology, however, employs a gestalt approach to defining ecoregion boundaries. In the Indonesian context the resulting map appears problematic both in terms of the underlying rationale of the ecoregion approach and in terms of apparent conflict with preexisting protected-area design. We suggest, insofar as refined planning frameworks are needed, that an alternative route that builds on rather than competes with existing approaches would be to combine at the mesoscale the landform delinea- tors that characterize the Bailey ecoregion system with the existing macroscale ecoclimatic and biogeographic delineators of the Dasmann-Udvardy system. We question the investment in developing and promoting the WWF–ecoregion scheme in Indonesia when the existing Dasmann-Udvardy system, used in conjunction with hotspot studies, provides a seemingly adequate system and when the reserve system itself is under consider- able pressure.

Ecoregiones en Contexto: una Critica con Especial Referencia a Indonesia Resumen: El Fondo Mundial para la Vida Silvestre ( WWF) de los Estados Unidos está promoviendo interna- cionalmente un marco de trabajo ecoregional como una nueva aproximación jerárquica a la organización y priorización de los esfuerzos de conservación. Evaluamos las ecoregiones de WWF contra marcos de tra- bajo existentes: 1) el marco de trabajo de representación Biogeográfica de la Dasmann-Udvardy (Unión Mundial para la Conservación (IUCN), el marco de trabajo ecoregional Bailey (Servicio Forestal de USA) y 3) la aproximación de la regiones problemáticas ejemplificado por la estrategia de Áreas para Aves Endémi- cas y el programa de centros para la diversidad de plantas de WWF/IUCN. Examinamos la genealogía de los esquemas desde tres perspectivas: nivel de claridad en la metodología, transparencia y repetibilidad y si el sistema de ecoregiones de la WWF mejora los esquemas existentes. Consideramos a Indonesia como un caso de estudio y evaluamos la eficacia de cada sistema en el contexto de Indonesia. Los planes de trabajo exis- tentes alcanzaron su objetivo; en general tuvieron métodos explícitos, transparentes y repetibles; y en el caso del sistema Dasmann-Udvardy alcanzó una realidad institucional en Indonesia. El propósito central de los marcos de trabajo de las ecoregiones de WWF es el mismo que el sistema de hace 25 años de Dasmann- Udvardy y a nivel de escalas espaciales amplias está basado en delineadores espaciales similares ( biomasa y

*email [email protected] Paper submitted September 17, 1999; revised manuscript accepted May 9, 2001. 42

Conservation Biology, Pages 42–57 Volume 16, No. 1, February 2002

Jepson & Whittaker Ecoregions in Context 43 regiones faunísticas). Sin embargo, la metodología de WWF emplea una metodología de configuración de el- ementos separados (gestalt) para definir los límites de las ecoregiones. En el contexto de Indonesia, los mapas resultantes parecen ser problemáticos tanto en términos de la racionalidad subyacente de la metodología de ecoregión y en términos de un aparente conflicto con los diseños de las áreas protegidas existentes. Sugerimos que hasta el momento se necesitan planes refinados de marco de trabajo. Una ruta alternativa que con- struya, y no que compita con las metodologías existentes sería la combinación a nivel de mesoescala de los delineadores de contornos que caracterizan el sistema de ecoregión Bailey con la macroescala ecoclimática existente y los delineadores biogeográficos del sistema Dasmann-Udvardy, usados en conjunción con los estu- dios de regiones problemáticas. Esto provee un sistema aparentemente adecuado cuando el sistema de reser- vas se encuentra bajo una considerable presión.

Introduction nesia captures a wide range of biotic variation (it spans two zoogeographical regions, dry and ever-wet tropics, Over the last 30 years, a variety of spatial frameworks and its ecosystem variation ranges from tropical glacier has been developed for the purpose of guiding conser- to mangrove). vation action internationally. The most recent is the To meet the aims outlined above, we describe (in ecoregional approach developed by World Wildlife chronological order) the three most prominent catego- Fund–United States ( WWF) (Dinerstein et al. 1995; Ol- ries of spatial planning frameworks developed at the glo- son & Dinerstein 1998). It is being adopted and pro- bal scale: biogeographical provinces, hotspots, and moted widely by the WWF family of agencies (WWF In- ecoregions. We summarize each framework with re- ternational, WWF national organizations, and WWF spect to their aims, rationale, and context of development country representative offices) and by international pro- and then assess their efficacy with reference to Indonesia’s grams of the U.S.–based, nongovernmental organization terrestrial ecosystems. Our assessment is concerned prin- The Nature Conservancy (TNC). The combination of ad- cipally with generic issues rather than specific boundary vocacy power, human and financial resources, and inter- questions in Indonesia. As far as we have been able to national project portfolios possessed by these two inter- establish, no comparative overview of these different national organizations make it likely that the ecoregional schemes has been published, and we therefore hope framework will be adopted and used by other agencies, this contribution will stimulate debate among conserva- including the Global Environment Facility (GEF) and tionists in general, not merely those directly involved in government conservation agencies in developing coun- planning within the Indo-Malayan realm. tries. In other areas of natural resource management, misunderstanding of alternative spatial planning frame- works has resulted in inconsistency in their use and ulti- Biogeographical Representation mate effectiveness (Omernik & Bailey 1997). Those in- volved in conservation planning on the ground therefore The Dasmann-Udvardy Framework need to know what this scheme brings with it that pre- existing schemes do not. A central concern of the IUCN since its creation has been Our review has three aims: (1) to compare the WWF the need to establish a worldwide network of natural re- ecoregions with existing spatial frameworks for terres- serves encompassing representative areas of the world’s trial conservation planning with a global perspective; (2) ecosystems. In the 1960s there was widespread support to assess scientific explicitness, transparency, and re- for this “representation principle.” In response, Dasmann peatability of methods; and (3) to ask whether the WWF (1972, 1973) prepared for IUCN a hierarchical system that ecoregions framework improves upon existing frame- defines and classifies natural regions for the purpose of works. To address these points we reviewed and at- conservation. His aim was to provide a system that gave tempted to define the purpose for which the various equal emphasis to the IUCN’s interests in conserving schemes were devised. natural ecosystems and vegetation types and the conser- We considered Indonesia as a case-study country with vation of species. His solution was to establish a system which to explore these questions. Indonesia is a suitable of classification of communities based on ecoclimatic fea- choice because (1) it is one of the most biodiverse areas tures but emphasizing taxonomic differences (Table 1). on Earth; (2) the government has consistently been at At the top level in the hierarchy, Dasmann (1972) chose the forefront in adopting new spatial conservation plan- the biome system (e.g., tundra, taiga, deciduous broad- ning frameworks; and (3) as an archipelago comprising leaved forest) of Clements and Shelford (1939) because large continental-shelf islands and oceanic islands, Indo- it is readily applicable globally, takes into account both

Conservation Biology Volume 16, No. 1, February 2002

44 Ecoregions in Context Jepson & Whittaker continued conservation work on the basis of geopolitical boundaries to planning within ecologically derived areas establishing protected areas and achieving sustainable management in the non-reserve matrix freshwater, and marines ecosystems harboring globally important biodiversity and ecological processes type taxonomic and ecoclimatic: major ecosystem type (Dinerstein et al. 1995) dispensed with in subsequent schemes; no classification cited but biogeographic zones listed for Latin and North America (Dinerstein et al. 1995; Ricketts et al. 1999) equivalent to Wallace (1876) modified by other biophysical characteristics: in Latin America follows various preliminary schemes (Dinerstein et al. 1995), in North America follows Küchler (1975), in Indonesia follows Whitmore (1984) based on van Steenis (1957) move away from organizing support two-pronged strategy of promote conservation of terrestrial, biogeographical zone & major ecosystem major habitat type (9 divisions) climatic, Ecoregions WWF ecoregions Ecoregions WWF as an integrated entity but is still suitable for multipurpose applications goal; Omernik: effective management of water quality ensuring that all land uses coincidentally sustain resource productivity and maintain ecosystem process and function generated by overlaying isotherm patterns ( James 1959) and moisture James 1959)limits (Schott, in level II ecoregion (Omernik) climatic, following Köppen (1931) and Trewartha (1968) climate classification systems and dominant potential vegetation (Küchler 1964, 1970) Bailey: assist land managers to meet this optimal of land, defined as optimal management domain (Bailey) climatic: four zones ecoregion divisions & provinces (Bailey), R. G. Bailey, C. Omernik D. M. Olson, E. Dinerstein a management of protected areas in EBAs unique kinds of organism for priority assignment of conservation action M. J. Crosby select a meaningful scale develop a system that classifies land designate and/or strengthen identify areas richest in Biogeographic province Endemic bird areas (EBAs) taxonomic differences of ecosystems network of natural reserves species diversity climatic & taxonomic: biomes climatic & taxonomic: biomes (Clements & Shelford 1939) subdivided by faunal regions ( Wallace 1876 ) (7 divisions) eco-climatic & taxonomic: climax vegetation ( Weaver & Clements 1938) subdivided by percent faunal similarity (Hagmeier & Stults 1964) give equal stress to structural and biogeographic realm (region) biogeographical (biotic) province 2 ) km 8 –10 4 2 b (10 km 6 7 10 10 consideration Macro ecosystem (level II ) global (level I) Dates of development 1970–1975Approach 1989–1992 1985–1996 1991–present Key design Operational purpose guide establishment of worldwide Table 1. Comparative analysis of four spatial conservation planning frameworks. Developing agencyLead authors World Conservation UnionGeneral goal BirdLife International R. F. Dasmann, M. D. UdvardyStattersfield, J. A. Bibby, J. C. conserve global habitat and U. S. Forest Service World Wildlife Fund–U.S.

Conservation Biology Volume 16, No. 1, February 2002

Jepson & Whittaker Ecoregions in Context 45

plants and animals, and broadly conforms to observable reality in areas not greatly modified by humans. Because the biome approach emphasizes ecological similarities at the expense of taxonomic difference, Dasmann (1972) divided the biomes of the world into regional subdivisions based on Wallace’s (1876) faunal regions (e.g., Palearctic, Ethiopian, Nearctic) and additional tran- sitional areas and biotic subdivisions that had long been accepted by biogeographers. These he termed “biotic realms” (e.g., Indo-Malayan realm). Macro-scale (level II) units were termed “biotic provinces” and delineated by subdividing a physiognomically defined climax vegeta- America adopts Omernik with some modifications; in Latin America adopts various national schemes; in Indonesia Whitmore divisions and biounts merged and modified by EBA boundaries with no existing scheme (i.e. outside North America) conducts gestalt synthesis of various schemes and criteria tion type at the level of the vegetation formation of ecoregion (35–40 divisions) in North limited: purpose all-embracing; in regions Weaver and Clements (1938) on the basis of a distinctive fauna (Dasmann 1973). Faunal distinctiveness is assessed by comparing the number of species in common be- tween areas divided by barriers that could have some s of Plant Diversity Program. To date this has had limited conceivable distributional significance. Based on a re- view of various North American schemes (Dasmann cites Goldman & Moore 1945; Blair 1950; Miller 1951; Hall & Kelson 1959; Hagmeier 1966) Dasmann consid-

Ecoregions WWF ecoregions Ecoregions WWF ered areas with 65% of their species in common to be separate faunal provinces. Sixty-five percent is arbitrary, but because it is about two-thirds of the total species compliment, it constitutes a simple fraction of intuitive

ecoregion (Omernik) landform (geology and topography) informed by Hammond’s (1954, 1964) landform classification scheme; Omernik divi- sions also informed by land-use pattern (Anderson 1970) and various soils maps to measure; methods difficult under- stand; robust classifications combined, but decisions on divisions based subjective grounds value. Dasmann (1972) recognized high mountains and landscape mosaic (Bailey), level III adequate: defined purpose but difficult mountainous islands (azonal features) as special situa- a tions because vegetation and biota are likely to change markedly within short distances due to steep environ- mental gradients. Arbitrarily, he defined mountain ranges and island groups (e.g., Lesser Sundas) as separate biotic (Terborgh &

2 provinces embedded within the system of province range criterion 2 boundaries derived from his zonal methodology. His pro- visional list of biotic provinces (including Australisia and the Antarctic) totaled 198. 50,000 km

Winter 1983) methods easy to under- stand but based on arbitrary 50,000-km and assumes mesoscale congruence with other taxa overlay distributions of bird species with ranges of To his chagrin, Dasmann (1973) found that the bio- good: defined purpose, EBA (24 units) taxonomic: geographer Udvardy (1969) had already published a de- tailed review of statistical methods for distinguishing bi- otic provinces, and that Hagmeier and Stults (1964) and Hagmeier (1966) had made more exhaustive compari- sons than his but had arrived at the same conclusion in adopting the 65% similarity criterion. The IUCN then commissioned Udvardy to develop and refine Das- mann’s system. Udvardy adjusted Dasmann’s terminol- ogy so that the highest level (biotic region) became the “biogeographic realm” and the second level (biotic prov- Biogeographic province Endemic bird areas (EBAs) inces) became “biogeographic provinces.” In substance, to understand methodology with simple algorithm based on species number data, but analysis confined to bird and mammal data (40 divisions) taxonomic: same percent faunal similarity algorithm applied by MacKinnon & Wind (1981) to smaller geographical units (Fig. 1a)

biogeographical unit (biounit) Udvardy’s report (1975) was a reaffirmation of the scien-

2 tific merits of Dasmann’s system. The resulting frame- work we call the Dasmann-Udvardy system. ) km 7 –10 2 Assessment of Biogeographic Representation in the

(10 Indonesian Context 3

10 The development of the biogeographic province frame- Meso ecosystem (level III) A global hotspot approach for plants has been developed by the World Conservation Union and Wildlife Fund in their Centre Summary of terminology, hierarchical determinators, and foundation studies at three spatial scales. a b Explicitness excellent: defined purpose, easy effect and therefore is not included in this table. Table 1. (Continued) work coincided with the implementation in Indonesia of a

Conservation Biology Volume 16, No. 1, February 2002

46 Ecoregions in Context Jepson & Whittaker two-phase (1974–1982) Food and Agriculture Organiza- major biogeographic regions in the country and repre- tion (FAO) project that helped Indonesia establish national sentative systems of reserves in each are identified; (2) parks and expand the protected-area network (Blower within each biogeographic division, priority is given to 1973; FAO 1977; Sumardja 1981). The national conserva- establishment of a major ecosystem reserve to include tion plan that was prepared to guide this process (MacKin- continuous habitat types and, if possible, the richest ex- non & Artha 1982) was the first application of the global ample of those habitats; (3) these large reserves are aug- Dasmann-Udvardy system on a national scale. mented with smaller reserves to protect special or In practice, it was realized that the biogeographic unique additional habitat types or to cover regional vari- provinces were at too coarse a scale to capture the bio- ations; and (4) small reserves are included to protect geographic variation in Indonesia, particularly in the specific sites of special beauty or interest. MacKinnon transitional island region of Wallacea (bridging the Ori- was subsequently commissioned by the IUCN to pre- ental and Australasian faunal realms). In response, Mac- pare similar reviews for the Indo-Malayan and Afrotropi- Kinnon and Wind (1981) applied the Dasmann-Udvardy cal realms, wherein he applied the same biogeographic algorithm based on smaller geographic units to distin- spatial system and reserve design principles (MacKinnon guish between main regions within the larger islands & MacKinnon 1986a, 1986b). These principles are re- and island groups. This created a third tier in the hierar- tained in the latest review of Indo-Malayan protected- chy, which was termed the biogeographical unit or “bio- area systems (MacKinnon 1997). unit” (Table 1; Fig. 1). The subdivision into biounits further emphasizes taxo- The national conservation plan for Indonesia (MacKin- nomic differences, but at the reserve selection level this non & Artha 1982) proposed a representative network is balanced with MacKinnon and Artha’s (1982) second of reserves based on the following principles: (1) the and third design principles, which emphasize habitat

Figure 1. Dasmann-Udvardy biogeographic provinces for Indonesia, with the third-level “biounits” added to the sys- tem by MacKinnon and Wind (1981) overlain on the natural habitat-type boundaries redrawn with permission from MacKinnon (1997). Biogeographic provinces and biounits: 21, Sumatra (21a, south Sumatra; 21b, north Sumatra; 21c, Mentawi Islands; 21d, Nias and Batau islands; 21e, Simeuleu Islands; 21f, Enggano Island; 21g, Lingga Archipelago); 22, Java (22a, West Java; 22b, east Java; 22c, Bali Island); 25, Borneo (25b, southwest Borneo; 25e, central mountains; 25f, east Borneo; 25g, east Borneo; 25h, northwest Borneo); 24, Sulawesi (24a, central Su- lawesi; 24b, north Sulawesi; 24c, south Sulawesi; 24d, southeast Sulawesi; 24e, northeast Sulawesi; 24f, Sangihe-Ta- laud Islands); 23, Lesser Sundas; (23a, north Nusa Tenggara; 23b, Sumba Island; 23c, Timor and Wetar islands; 23d, Tanimbar); 13, Moluccas (13a, n. Maluku Islands; 13b, Obi; 13c, Buru; 13d, Ceram & Ambon; 13e, Kai Islands); P3, New Guinea; (P3a, Aru Islands; P3b, Western Islands; P3c, Geevlink Bay islands; P3d, Vogelkop; P3e, northwest New Guinea; P3f, southwest New Guinea; P3g, Snow Mountains; P3h, Star mountains; P3l, Trans-Fly).

Conservation Biology Volume 16, No. 1, February 2002

Jepson & Whittaker Ecoregions in Context 47 representation (above). This is consistent with the aim of number of species in an area that occur nowhere else), the Dasmann-Udvardy system to give equal emphasis to and there were many prognoses of an impending mass structural and taxonomic features. Subsequent pro- extinction of species (Myers 1979; Ehrlich & Ehrlich tected-area reviews and biodiversity conservation strate- 1981; Wilson 1988). These prognoses were based in gies in Indonesia have adopted the same system and have part on the simple observation that deforestation in the confirmed the network of 80 key reserves to meet repre- tropics was progressing at a rapid rate and was linked to sentation goals (Regional Physical Planning Programme the “relaxation effect.” The effect is predicated on Mac- for Transmigration 1990; Government of Indonesia 1991; Arthur and Wilson’s (1963, 1967 ) equilibrium theory of BAPPENAS 1993; KLH 1993). Since commencement of island biogeography and postulates that species number the FAO conservation project in 1974, Indonesia’s pro- will inevitably re-equilibrate to a lower number if habitat tected-area network has been expanded from 170 re- area is reduced and isolation of patches increased (re- serves covering 3.3 million ha (Sinaga, unpublished data) viewed by Whittaker 1998). to 384 reserves covering 22.3 million ha ( PHPA 1999); The notion of an impending “extinction crisis” led to a with the exception of East Kalimantan, the Moluccas, sensible and obvious desire to target scarce conserva- and Nusa Tenggara, most of the 80 key reserves have tion resources and to give priority to establishing new been designated. Furthermore, it is only now, 20 years af- reserves first in regions that (1) are exceptionally rich in ter the design of the network, that forest loss in the few species and/or unique species and (2) are under threat. biounits still without reserves necessitates significant re- Myers’s (1988, 1991) “hotspot analysis” was the first visions to the reserve configurations originally proposed. such global study (updated by Myers et al. 2000). Al- Many areas of economically valuable lowland habitats though at too broad a scale (18—now 25—hotspots in were excised from proposed reserves at the time of the world) to be of much practical use (Long et al. 1996), gazettement, but we consider this a reflection of the ten- the idea inspired more detailed studies, notably the dency of all governments to allocate land with limited ag- BirdLife International Endemic Bird Area (EBA) Approach ricultural potential for biodiversity conservation, rather International Council for Bird Preservation ([ICBP] 1992), than a weakness with the system or prioritization per se. the IUCN–WWF Centres of Plant Diversity (IUCN & WWF Thus, in Indonesia, at least, the Dasmann-Udvardy system 1994), and, recently, the Global 200 ecoregions (Olson has worked (Table 2). This assessment sets aside the vital & Dinerstein 1998). issues of and threats to reserve integrity. The Dasmann-Udvardy system has the merit of a trans- parent and repeatable methodology with a genealogy The IUCN–WWF Centres of Plant Diversity Approach that goes back to such authorities as A. R. Wallace and F. E. Clements. The delineation of biogeographic prov- The IUCN–WWF Centres of Plant Diversity Program (IUCN inces and units is open to review as distributional data & WWF 1994) sought to identify sites around the world sets on faunal groups other than birds and mammals are of greatest importance for plant conservation. Criteria completed, or after changes in taxonomy. Furthermore, were developed for selecting sites that (1) are obviously the method provides for finer-scale subdivisions (Mac- rich in species, (2) are rich in endemic species, (3) are Kinnon & Wind 1981). threatened and/or (4) contain a diverse range of habi- The seven Dasmann-Udvardy biogeographic provinces tats, (5) have a gene pool of species useful to humans, of Indonesia accord with the main geographic, cultural, and (6) contain species adapted to particular edaphic and economic developmental regions of Indonesia, and conditions. Sites were nominated and criteria applied on they have been adopted widely as a framework for un- the basis of expert review. derstanding biological variation throughout the archipel- The Centres of Plant Diversity scheme is neither ago. They are taught in schools, they define coverage of widely known nor widely used in Indonesia. Its strength volumes in the ecology of Indonesia series ( Whitten et lies in it is ability to pinpoint areas and features of con- al. 1987, 1996, 2000; MacKinnon et al. 1996; Monk et al. servation importance with scattered (azonal) distribu- 1997 ), and they provide the planning units for several tions—notably limestone massifs—which zonal strategies and overviews and the geographic units in var- schemes, such as those described in this paper, do not ious tourist guidebooks. In short, biogeographic prov- identify. The main weakness of the approach lies in the inces have attained a social reality in Indonesia (Searle comparative incompleteness and unevenness of plant 1996 ) (Table 2). distributional data relative to those for mammals and birds: botanists tend to have collected in places of easy access. As a result, sites identified may reflect collecting Hotspot Approaches effort rather than truly exceptional levels of species rich- During the late 1970s, two attributes of biodiversity ness or endemism (cf. Nelson et al. 1990), and reserves started to attract particular attention: species richness come out as priority sites because this is where people (the number of species in an area) and endemism (the have collected.

Conservation Biology Volume 16, No. 1, February 2002

world lacking detailed preexisting schemes, but we question the merit of promoting this approach over existing schemes in Indonesia could compromise goals by undermining the established approaches already advanced in meeting comparable goals introduced new generation of conservationists from social or economic backgrounds to notions of ecological repre- sentation; instrumental in extending spatial prioritization approach to the marine realm understand its meaning or utility may be beneficial in areas of the none yet term gaining recognition, but few species variation means this system is unsuitable for biodiversity conservation planning at global level to mesoscale; focus on landform below the mesoscale offers opportunity to combine concerns relating to social justice and ecological integrity with traditional habitat and species- preservation concerns in conservation; landform could form a fourth-level determinator in the Dasmann-Udvardy system familiar with climatic, vegetation, and landform variables used in delineation of physiographical types and land suitability for development (see RePPProT 1991) lack of consideration given to not applicable appears to add nothing new and limited understandingnot applicable none yet none yet of any substance; has resource-management disciplines similar concepts and classifications microecosystem centers of endemism microecosystem centers of endemism is likely to increase in value line with increasing landscape change and human population growth; hotspot approaches such as EBAs are essential supplements to “representative” approaches for macroscale conservation planning and strategizing useful adjunct to the biogeographic province approach conservation importance of the small island regions of Lesser Sundas and Maluku; main factor in location and/or designation of several new reserves and in promoting field research to fill distributional lacunae strengthened local pride and sense of place on islands where EBAs are part of conservation planning framework of unique birds in EBAs is proving a powerful means of generating popular interest in conservation ornithologists and specialist conservation planners biodiversity planners and in local planning authorities of Maluku and East Nusa Tenggara knowledge of location meso- or achieving purpose and providing created a policy awareness of the at an operational level the message well known among the few scientific well known among specialist conservation’s joint concerns for preservation of ecosystem types and species variation, creation of reserves to meet these concerns is expected to stand the test of time; Dasmann- Udvardy system a good hierarchical system for conservation planning down to the mesoecosystem scale extent the representation principle in national policy; government policy to establish a “minimum set” of 80 major ecosystem reserves; nearly 200 new reserves covering 20 million ha added to the existing reserves system; desire to support and participate in the above program led to five major international conservation NGOs establishing programs in Indonesia among Indonesians and related to the major island and hence cultural groups macrolevel ecological variation in schools and universities government and donor conservation strategy and policy documents principle of representation, national conservation plan enshrined established framework for future value outcomes credentials Assessment of Overall assessment achieved purpose to a remarkable Tangible conservation Public awareness seven provinces well known Scientific credentials basic framework for teaching Table 2. Efficacy of four spatial conservation planning frameworks in Indonesia. CriteriaPolicy & development Biogeographic province Endemic bird areas (EBAs) Ecoregions World Wildlife Fund ecoregions

Jepson & Whittaker Ecoregions in Context 49

BirdLife International’s Endemic Bird Area Approach are located in the complex island region of Wallacea. To identify priority islands and reserves for conservation, Contemporaneous with the Centres of Plant Diversity BirdLife identified sub-EBAs by island(s) and by habitat program, BirdLife International developed its Endemic by creating simple matrices of species presence against Bird Area Approach ( ICBP 1992; Stattersfield et al. 1998), island/island group and habitat type (Sujatnika et al. 1995). which is the most complete example of a global hotspot The EBA approach assisted in implementation of the analysis (Long et al. 1996). This approach is based on the national conservation plan (based on the Dasmann- belief that “first priority must be assigned to programs to Udvardy system) by prioritizing the proposed reserves conserve areas richest in unique kinds of organism” ( Ehr- sited in areas of concentrated endemism. Prior to the es- lich 1988) and was inspired by pioneering studies of bird tablishment of the PHPA–BirdLife Indonesia Programme, distribution in Africa (Hall & Moreau 1962), Colombia, conservation efforts in Indonesia were focused on the and Ecuador (Terborgh & Winter 1983). The last study 2 large land masses of Sumatra, Kalimantan, Java, Irian Jaya, mapped bird species with ranges of 50,000 km (an arbi- and Sulawesi (mainly in Minahasa). The Lesser Sunda trary size) to locate areas of concentrated endemism that and Moluccas biogeographic provinces received little at- would be optimal for designation as reserves. The tention, despite their long-recognized scientific and con- BirdLife Biodiversity Project applied Terborgh and Win- 2 servation importance (e.g., Harper 1945). Between 1992 ter’s 50,000-km range criterion worldwide. For bird spe- and 1999, and as a result of the BirdLife initiative, two new cies meeting this criterion (2649 species, following Long national parks were designated on Sumba ( Jepson et al. et al. 1996), project researchers conducted a compre- 1996), four new key reserves were designated on Timor, hensive literature review and compiled a database of a major GEF–financed project was prepared to establish geographically referenced distributional records. Records five new reserves in Maluku (since cancelled because of were plotted and distributions of species overlaid. Areas political and social instability), and proposals for desig- where two or more such species co-occurred (an arbi- nating new reserves in Flores and Sumbawa have been trary choice) were termed endemic bird areas. In regions prepared ( Trainor et al. 2000; P.J. et al., unpublished with complex distributional patterns, a divisive cluster data). The program inspired 10 university biological ex- analysis of multivariate distributional data summarized by peditions to remote islands and the “Action Sampiri” grid square was performed to aid in identification of “nat- conservation project on Sangihe-Talaud (Riley 1997). In ural” groupings of species (ICBP 1992; Long et al. 1996). Indonesia the EBA approach is succeeding in directing Although this is a repeatable methodology, the use of ar- new conservation effort to centers of avian endemism. bitrary criteria (e.g., size of range, number of endemics, When EBAs are assessed against the hotspot ap- degree of range overlap), as with all such schemes, inevi- proach’s dual criteria of richness/endemism and threat, tably has a bearing on precisely which areas are selected. a weakness is evident: the areas mentioned in the preced- Like the Dasmann-Udvardy system, the EBA approach ing paragraph (e.g., Timor, Sumba, Flores) are less threat- has the merit of an explicit purpose and a transparent ened than areas in west Indonesia that are not included and repeatable methodology. The database on which within EBAs. For example, with 164 endemic bird spe- EBA boundaries were devised is freely available, and the cies (Wells 1985), the lowland ever-wet forest of the method could be developed to identify centers of ende- Sunda shelf constitutes a center of endemism, but many mism at different taxonomic levels or (with greater ef- of these species have ranges above the 50,000-km2 fort) using a different range-size threshold. This allows threshold for EBAs. These forest ecosystems are being conservation recommendations derived from EBAs to be converted to agriculture and estate crops at an alarming independently assessed, revised, or refined. rate. On Sumatra, there is little intact lowland forest re- maining outside reserves (Laumonier 1997), and even within reserves this habitat is being degraded by illegal Assessment of Endemic Bird Areas in Indonesia logging and fire. Yet EBAs in west Indonesia are all in BirdLife originally identified 221 EBAs worldwide, 24 of mountainous regions that are relatively intact and se- which are in Indonesia (ICBP 1992; Sujatnika et al. cure. The new reserve currently being advocated in In- 1995). In response, BirdLife launched its Indonesia pro- donesia is Sebuku-Sembukang in East Kalimantan, but gram in partnership with the Directorate General of For- this priority was identified on the basis of a simple analy- est Protection and Nature Conservation (PHPA) in 1992, sis of gaps in representation of habitats and species of in- with the purpose of securing the designation of reserves ternational conservation interest (Momberg et al. proposed under the national conservation plan in prior- 1998a), not with the EBA approach. ity EBAs ( Jepson 1995). The EBAs are not designed in a Endemic bird areas were promoted as centers of unique hierarchical system. As biogeographic entities, however, biodiversity and thus as indicators of likely areas of ende- EBA boundaries in Indonesia nest within biogeographic mism in other taxa (ICBP 1992). Although congruence of provinces and accord closely with the MacKinnon and bird endemism with centers of endemism in other groups Wind (1981) biounits ( Jepson & Sujatnika 1997). Many generally obtains at the macroscale, it often fails at finer

Conservation Biology Volume 16, No. 1, February 2002

50 Ecoregions in Context Jepson & Whittaker scales of analysis (cf. Bush 1994). At the scale of national sustain resource productivity and maintain ecosystem pro- conservation planning in Indonesia, expert consultation cesses and functions. suggests that existing data support the notion of congru- Bailey’s system aims to delineate at a given level bound- ence only for some small-mammal groups (D. Kitchener, aries of ecosystems that control the process and func- personal communication) and swallowtail butterflies (P.J., tion of ecosystems at the next level down. He adopts the unpublished data). In the experience of the senior author, “controlling factor method” (Bailey 1996 ) by which a overstating the scope of the approach in the early 1990s spatial hierarchy is constructed by successive subdivi- was counterproductive because this drew criticism from sion of large ecosystems on the basis of controlling factors other conservation specialists when the approach was pre- operating at different scales. Ecoregions are delineated at sented in conservation fora. As a result, policy makers in In- three levels. At the macroscale, climate is considered the donesia were wary of its merits and came to perceive EBAs principal controlling factor. In the top tier, four domains, as a specific approach for endemic bird conservation. This or ecoclimatic zones of the Earth, are delimited—humid goal was perceived as a relatively unimportant component tropical, humid temperate, polar, and dry—by simple of the wider Indonesian biodiversity discourse because en- overlay of global thermal and moisture patterns ( James demic birds lack a clear utility value. 1959). These domains are subdivided on the basis of Despite these shortcomings, the EBA approach became Köppen’s system of climatic classification (as modified established in Indonesia as a distinct system complimen- by Trewartha (1968) into second-tier ecoregions, termed tary to the biogeographic-region system. The value of divisions, of which there are 31 globally. Bailey’s divisional EBAs lies in drawing attention to the existence of discrete system distinguishes between zonal and azonal ecore- centers of avian endemism and in generating local sup- gions. Azonal ecoregions are, for example, wetlands or port for reserve designation. This last point is worth elab- alpine ecosystems that can occur in any zone where the oration. In Indonesia, approval by the provincial governor appropriate geomorphology occurs; for instance, the and district officer is required for reserve designation to zonal “icecap division” is matched by the azonal “icecap progress. BirdLife employees have found that such offi- regime mountains.” At the macroscale, Indonesia has cials respond with pride and interest when informed that four divisions: savanna, savanna regime mountains, rain- their territory supports an assemblage of bird species forest, and rainforest regime mountains. The divisions found nowhere else on Earth, and this has been instru- may also in turn be subdivided into “provinces” on the mental in securing their support for the new reserves. basis of macrofeatures of the vegetation that reflect more refined climatic differences. This province level is referenced but not presented by Bailey (1996). Ecoregional Approaches Bailey (1996) proposes further refinements to his The term ecoregion was introduced into the arena of scheme, as follows (Table 1). At the mesoscale, he con- U.S. land-management planning and conservation by R. G. siders landform the principal determinant of potential Bailey (e.g., 1983, 1996) and the U.S. Forest Service (1993) vegetation and uses Hammond’s (1954, 1964) scheme of ECOMAP project, who developed and refined a hierar- land classification, informed by Küchler’s (1964, 1970) chical system for the purpose of optimizing land-man- maps of potential vegetation in the United States to deter- agement goals within the United States. Another impor- mine the limits of various mesoecosystems, termed land- tant contributor to the development of the ecoregion scape mosaics. This is the third level of the Bailey concept is J. M. Omernik (1987, 1995). Internationally, scheme. These are further subdivided into smaller micro- Bailey (e.g., 1989) extended his ecoregions approach to ecosystems based on edaphic factors. Finally, Bailey pro- the world at the macroecosystem level of resolution. poses dividing all contemporary ecosystems into four The conservation science program of WWF–United States classes, reflecting the degree of human transformation, subsequently developed a mesoscale ecoregional classi- based on the system of Milanova and Kushlin (1993). fication for Latin America (Dinerstein et al. 1995) and Omernik’s system is essentially the same. The same de- embarked on a project to develop ecoregion maps for lineators are used at the macroscale to produce what he the rest of the world (Dinerstein 1999). terms a level II ecoregion. In Omernik’s scheme, level III ecoregions divisions are informed by land-use pattern (Anderson 1970) and various soil maps in addition to Ham- Ecoregions: Bailey and Omernik Frameworks mond’s land-use and Küchler’s vegetation maps (Omernik Unlike the Dasmann-Udvardy system, the Bailey (1996, 1987). The decision criteria for combining these data lay- 1998) and Omernik (1987, 1995) hierarchical ecosystem ers in both schemes are subjective ( Wright et al. 1998). classification systems do not aim to incorporate taxonomic distinctions but rather focus on characteristics of ecosys- WWF Ecoregions tem structure. This reflects their purpose, namely the opti- mal management of land and water (Omernik & Bailey The WWF ecoregions are more inclusive in purpose than 1997), defined as ensuring that all land uses coincidentally those of the Bailey and Omernik systems. The WWF ecore-

Conservation Biology Volume 16, No. 1, February 2002

Jepson & Whittaker Ecoregions in Context 51 gions aim to create a superior biogeographic unit for con- “first cut” boundaries following consultation with re- servation planning at regional scales to meet the four main gional experts. goals of biodiversity conservation, as defined by Noss The intent of the WWF ecoregion approach is to map (1992) and Noss and Cooperrider (1994): (1) representa- boundaries of ecosystems at the mesoscale. It employs tion of all distinct natural communities within a network of map overlay methods at the macroscale. At the meso- protected areas, (2) maintenance of ecological and evolu- scale, it employs a gestalt approach in which regional tionary processes, (3) maintenance of populations of spe- boundaries are drawn intuitively around areas that ap- cies, and (4) conservation of large blocks of natural habitat. pear homogenous (Bailey 1996), combining various ex- Ecoregion-based conservation (the practical application of isting zonal and azonal schemes not necessarily with the the framework) aims to promote these goals by employing same genealogies. A key assumption at the mesoscale is a “two pronged strategy of establishing protected areas and that boundaries of original vegetation types equate to achieving sustainable management of the lands and waters boundaries within which key ecological flows and link- outside protected areas” (Ricketts et al. 1999). ages operate internally, as opposed to externally. The WWF ecoregion approach combines two ele- This is difficult to substantiate with certainty because ments: (1) mapping of regional patterns of biodiversity the terminology adopted in the WWF ecoregion frame- (i.e., delineation of ecoregion boundaries) and (2) prior- work is vague. First, key ecobiogeographical terms— itization of ecoregions for conservation action (Wikra- biome, bioregion, major habitat type—are not linked to manayake et al. 2001). Here, we focus on the first ele- foundational definitions, which introduces a degree of ment. The second element, a form of hotspot analysis, subjectivity and confusion. Second, macroscale units uses transparent and repeatable indices of biological dis- such as bioregion are frequently defined in terms of tinctiveness and conservation status (Dinerstein et al. amalgamation of ecoregions, which is counter to normal 1995; Ricketts et al. 1999; Wikramanayake et al. 2001) to practice in the field and introduces circularity. Third, identify priority ecoregions known as the Global 200 new methodological terms are introduced but not de- ecoregions (Olson & Dinerstein 1998; WWF 2000). In fined, which obscures method. Use of the term stratify several cases, ecoregions have been aggregated to create is a case in point. Stratify should refer to layering more a single Global 200 ecoregion. The basis of aggregation than once, but in WWF ecoregion discourse it means fit- is not evident in the literature cited. ting clusters of ecoregion boundaries within boundaries An ecoregion is defined in the WWF scheme as “an eco- of major habitat types. system of regional extent” (Dinerstein et al. 1995), which we take to mean a mesoscale ecosystem (102–107 km2) Assessment of the WWF Ecoregions for Indonesia that controls and is defined by smaller constituent ecosys- tems. The WWF ecoregion framework is conceived as “a Our assessment of WWF ecoregions is based on the In- hierarchy based on habitat types” (Dinerstein et al. 1995). donesia sections of Figs. 2.1.b and 2.1c of Wikramanay- Dinerstein et al. (1995, their Figs. 1 & 2) place major eco- ake et al. (2001), reproduced here as Fig. 2. This repre- system types at the top level. These are subdivided into sents a refinement of “first cut” ecoregion boundaries major habitat types said to equate broadly with biomes that have circulated in Indonesia conservation-planning (Olson & Dinerstein 1998) and are overlaid with biogeo- circles since 1996 and were used by Yayasan WWF– graphic realm (e.g., Nearctic, Indian Ocean) to delineate Indonesia to reorganize and develop new strategic direc- third-level ecoregion boundaries. In subsequent studies tions (see Momberg et al. 1998b, their Map 1). (Ricketts et al. 1999, their Fig. 2.3; Wikramanayake et al. At the time of their introduction to Indonesia, the Das- 2001, their Box 2.1), the hierarchy is represented more mann-Udvardy and EBA schemes were the first of their simply as biogeographic realm (they use the term zone) type and guided new conservation programs. In con- subdivided by major habitat type (biome). Confusingly, trast, the WWF ecoregions represent an alternative scheme the Global 200 accounts (Olson & Dinerstein 1998; WWF intended to improve the performance of existing conser- 2000) state the reverse, major habitat types subdivided by vation programs. Our assessment therefore focuses on biogeographic realm. whether ecoregions compliment or improve upon exist- In practice, WWF ecoregions are delineated by com- ing frameworks. bining boundaries of existing regional schemes. For According to Wikramanayake et al. (2001), the ecore- Latin America, various schemes were combined (Diner- gion delineation in the Asia Pacific is similar to the Das- stein et al. 1995); for North America, Omernik’s (1987 ) mann-Udvardy hierarchy. MacKinnon’s (1997 ) maps of ecoregion maps were adopted (Ricketts et al. 1999); for biounits and original forest vegetation were used as the the Asia-Pacific volume, MacKinnon’s biounit and origi- general guide, modified by EBA boundaries and the com- nal forest cover maps (Fig. 1) were combined ( Wikra- ments of three regional experts, A. J. Whitten, T. C. manayake et al. 2001), all with modifications. Modifica- Whitmore, and D. Madulid. Comparison of Figs. 1 and 2 tions to existing schemes are described in ecoregional shows that ecoregion boundaries correspond to a syn- accounts and were made on the basis of refinement of thesis of MacKinnon (1997, his Map 2) in western Indo-

Conservation Biology Volume 16, No. 1, February 2002

52 Ecoregions in Context Jepson & Whittaker

Figure 2. Endemic bird areas of Indonesia, redrawn with permission from International Council for Bird Preser- vation (1992): 1, Sumatra; 2, Enggano; 3, Bornean mountains; 4, Java and Bali forests; 5, Javan coastal zone; 6, Sulawesi lowlands; 7, Sulawesi mountains; 8, Sangihe-Talaud; 9, north Nusa Tenggara; 10, Sumba; 11, Timor and Wetar; 12, Banda Sea Islands; 13, Seram; 14, Buru; 15, Banggai and Sula; 16, north Molucca (including Halma- hera); 17, west Papuan lowlands; 18, west Papuan mountains; 19, Geelvink islands; 20, north Papuan lowlands; 21, north Papuan mountains; 22, central Papuan ranges; 23, south Papuan lowlands; 24, Trans-Fly. nesia, and EBA boundaries (Fig. 2) in Wallacea and to a The use of major habitat type in Indonesian ecoregion lesser extent western New Guinea (Irian Jaya). There names is consistent with WWF’s desire to stress habitat are several modifications to this broad picture, differing representation. In reality, WWF ecoregion boundaries in significance, that are described but not fully justified simplify MacKinnon’s map (1997, Fig. 2) of vegetation by Wikramanayake et al. (2001). types, particularly in the Wallacea region. MacKinnon’s The aim of delineating mesoecosystems (as defined habitat type boundaries (Fig. 1) are based on Whitmore’s above) is not consistently achieved in Indonesia, as illus- (1984) vegetation formations. Spatially these may be (1) trated by three cases. (1) The 1.8 million-ha tectonic basin large, homogenous areas such as lowland everwet for- characterized by the Middle Mahakam wetland system est, (2) smaller scattered patches such as limestone for- (Voss 1983) in East Kalimantan is unquestionably a dis- ests, or (3) linear features such as beach forests. Homoge- tinct mesoecosystem or ecoregion in the Bailey-Omernik nous-area formations appear at macro- and mesoscales, sense. Yet Wikramanayake et al. (2001) divide it into a scattered formations at the mesoscale, and some linear complex of three ecoregions on the basis of dominant formations only at finer scales. As a result, WWF ecore- vegetation formations (heath forest, peat swamp forest, gions emphasize homogeneous major habitat types but ig- and freshwater swamp forests). These are constituent ec- nore scattered and linear formations, or treat these as osystems. (2) Two small island systems (Sangihe-Talaud distinct habitat types within an ecoregion. Moreover, and Bangai-Sula) are combined with areas below 1000 m Whitmore’s (1984) scheme is an ever-wet forest vegetation on Sulawesi in a single ecoregion. It is inconceivable that classification: dry-tropical forest formations have not yet there are distinct ecological flows and linkages internal to been systematically classified in Southeast Asia. Conse- and controlling an area with this pattern. (3) Montane quently, WWF ecoregions underrepresent dry-tropical ecoregions are delineated on the basis of the 1000-m con- forest habitats that are priorities for conservation action tour because it approximates a change in forest type. globally (Green et al. 1996). Thus, Indonesian ecoregions There are several problems with this: for example, the ac- arguably do not place greater emphasis on habitat repre- tual elevation of forest-type change varies substantially as sentation and may reflect expediency—which existing a function of the Massenerhebung (mass-elevation) ef- classifications can be mapped at the desired scale—at the fect ( Whitmore 1984; Richards 1996); faunal assemblages expense of other ecological and biogeographical patterns. that are the target of conservation action are not confined Indonesia already has a protected-area network in distribution to above or below a single elevational planned and justified by the Dasmann-Udvardy frame- threshold (Md. Nor 2001); and factors controlling ecosys- work operationalized by MacKinnon. In practice, con- tem function generally coincide with mountain landscape servation planners in Indonesia capture habitat variabil- units rather than being bounded by an arbitrary elevation. ity in reserve networks by listing and locating vegetation

Conservation Biology Volume 16, No. 1, February 2002

Jepson & Whittaker Ecoregions in Context 53

Figure 3. World Wildlife Fund ecoregions of the Indo-Pacific, redrawn with permission from Wikramanayake et al. (2001): 82, Sumatran lowland rainforests; 83, Sumatran montane rainforests; 84, Mentawi Islands rainforest; 85, Sumatran peat-swamp forests; 86, Bornean peat-swamp forests; 88, Sumatran freshwater swamp forests; 89, Southern Borneo freshwater swamp forests; 90, Sundaland heath forests; 93, western Java rainforests; 94, eastern Java-Bali montane rainforests; 95, Borneo montane rainforest; 96, Borneo lowland rainforests; 105, Sumatran tropical pine forests; 107, Sunda shelf mangrove; 109, Sulawesi lowland rainforest; 110, Sulawesi montane rain- forest; 111, Lesser Sundas deciduous forests; 112, Timor and Wetar deciduous forests; 113, Sumba deciduous for- ests; 114, Halmahera rainforest; 115, Buru rainforest; 116, Seram rainforest; 117, Banda Sea Islands moist decidu- ous forests; 118, Vogelkop montane forests; 119, Vogelkop-Aru lowland forests; 120, Biak-Numfoor rainforests; 121, Japen rainforests; 122, northern New Guinea montane rainforests; 123, northern New Guinea lowland rainforest and freshwater swamp forest; 125, central Range montane rainforests; 127, southern New Guinea freshwater swamp forests; 128, southern New Guinea lowland rainforest; 129, New Guinea mangroves.

formations and associations occurring within each bio- tem reserves. This creates two alternative ecological unit at the time a reserve is designed. This is expedient management units, the ecoregion and the major ecosys- because it avoids the need for a hierarchical system of tem reserve. Embarking on such major reorganization at vegetation, which would be required to capture the a time when ecosystem reserves in Indonesia are under massive vegetation heterogeneity in Indonesia and which siege from exploitation interests is inadvisable. would be a complex task to attempt. Replacement of existing conservation planning frame- We are unclear how introduction of WWF ecoregions works with ecoregions could have important implica- will strengthen the design of a protected-area system in tions for the targeting of future conservation resources. Indonesia. As currently presented, ecoregions risk weak- For example, representation of the Sulawesi lowlands as ening the system by implying that past planning was homogenous ( Wikramanayake et al. 2001) may suggest based on inferior science or information and by break- that targeting resources to any peninsula is equally good, ing the link between spatial frameworks and clear and whereas in reality the northern (Minahasa) peninsula is practical reserve-design principles. In addition, there is a the highest priority for conservation of large mammals more subtle issue. The goal of conserving large blocks of and endemic species ( Whitten et al. 1987; Sujatnika et natural habitats has been pursued in Indonesia by advo- al. 1995). Creating mangrove ecoregions in the Sunda- cating major ecosystem reserves. These are justified on shelf and New Guinea bioregions but not in Wallacea the basis of watershed protection, the conservation im- may lead some to conclude that Wallacea lacks man- portance of intact ecological gradients, and area require- groves worthy of conservation investments, which is not ments of viable populations of megafauna. In short, the case. The ecoregion complexes in east and south these reserves and surrounding forest landscapes have Kalimantan could create the misleading impression that been presented as functioning ecological entities (e.g., investing in the provinces where they occur will con- the Leuser Ecosystem). Ecoregions are also so pre- serve more biodiversity than similar investment in ap- sented, but delineation on the basis of habitat bound- parently simpler regions such as the Lesser Sundas or Su- aries results in the division of at least 11 major ecosys- lawesi.

Conservation Biology Volume 16, No. 1, February 2002

54 Ecoregions in Context Jepson & Whittaker

Given these problems, and the fact that in their North tant features of systematic conservation planning tools American scheme WWF has adopted geomorphology that guard against planning decisions having more to do and topography as delineators and in the names of the with political, organizational, and technological expedi- ecoregions, we are unclear why they have not done the ency than persistence of biodiversity (Pressey 1999). same in Indonesia and the Asia-Pacific realm. Adoption We have shown that preexisting schemes generally of landform as a primary mesoscale ecoregion delinea- meet these criteria. By adopting an existing ecoregion tor, as recommended by Bailey, would have made a dis- scheme (Omernik 1987 ), the WWF ecoregions of North tinct and useful contribution and avoided several of the America (Ricketts et al. 1999) appear to meet these cri- problems described above. A landform analysis of Indo- teria (but see Wright et al. 1998). nesia was conducted by RePPProT (1990) and is avail- We argue that the WWF ecoregion methodology is un- able. clear in important respects and that the assumption that major habitat boundaries equate to mesoscale ecosystem boundaries is flawed in several instances. We are con- Discussion cerned that ecoregion delineation on the basis of gestalt synthesis and “expert” review may not meet our three The first aim of our paper was to place WWF ecoregions methodological criteria well enough for ecoregions to in context by assessing how the alternative systems dif- form the foundation of global and national conservation fer. The Dasmann-Udvardy biogeographic framework planning decisions. Specifically, the claim that “several” and Bailey-Omernik ecoregion frameworks are distinct other regional-scale assessments (i.e., Dinerstein et al. in terms of methodology (captured in the terms used) 1995; Ricketts et al. 1999) use the same ecoregion delin- and purpose. The former combines ecoclimatic and tax- eation scheme cannot be justified. Whereas in North onomic delineators with the purpose of achieving bio- America the WWF adopted the Bailey-Omernik scheme logical representation in a global reserve network, and it (Ricketts et al. 1999), in Indonesia it adopted the Das- emphases taxonomic delineators. The latter emphasizes mann-Udvardy scheme as modified by MacKinnon ( Wik- ecoclimatic indicators to achieve sustainable land use. ramanayake et al. 2001). Consequently, prioritization The WWF ecoregions approach aims to do both and among ecoregions globally (Olson & Dinerstein 1998) is adopts methodological aspects of each spatial frame- suspect because similar regions are not compared glo- work. At level 3, the mesoscale, the Dasmann-Udvardy bally. approach as extended by MacKinnon employs taxo- The development of systematic approaches is produc- nomic delineators, the Bailey and Omernik ecoregion ing powerful decision-support tools and has the poten- scheme employs landform and topographic delineators, tial to improve the public accountability of conservation and the WWF ecoregions approach uses a gestalt synthe- agencies. One measure of explicitness and transparency, sis of available regional schemes, which unfortunately however, is consistency in terminology, which is poor have uneven geographical coverage. in this field. Within our comparatively brief review, we It is important to distinguish between systematic plan- have encountered an enormous variety of terms ( Table ning frameworks and strategic planning approaches. 1). The potential for confusion and inefficiency in devel- The existing schemes are primarily spatial planning opment of conservation planning and implementation frameworks that represent three basic conservation ap- appears considerable. We therefore call for greater stan- proaches: (1) representation of biodiversity attributes in dardization of terminology in the field and for explicit networks or reserves (Dasmann-Udvardy system), (2) referencing of terminology and delineators to existing protection of special elements such as centers of species schemes and classifications. endemism or richness (EBAs, centers of plant diversity), Our third goal was to ask if WWF ecoregions improve and (3) land-use planning within ecologically defined ar- upon existing schemes, for which we took Indonesia as eas (Bailey-Omernik ecoregions). The zonal schemes as- a case study. We focused on the utility of the schemes sume that the actions for which they are designed (des- for guiding tangible, on-the-ground conservation out- ignation of reserves, optimal management of land) will comes. Our assessment (Table 2) is that the existing be pursued within each spatial unit. The azonal frame- schemes of Dasmann-Udvardy (adapted by Mackinnon works assume that each hotspot merits conservation ac- and colleagues), complimented by EBA and reserve-design tion. In this sense, the distinctive contribution of WWF principles, for all their individual weaknesses, have pro- ecoregions is the application of strategic-planning crite- vided a successful mesoscale conservation planning ria to a zonal framework. The value of the WWF contri- framework. They have informed and motivated a major bution therefore depends on the scientific merits of expansion of the protected-area network, established a both azonal and zonal aspects. Our critique has consid- biogeographic perception of space in government de- ered only the zonal framework. partments, and educated Indonesian society. Our second concern is scientific explicitness, trans- The WWF ecoregions “seek to advance biodiversity parency, and repeatability of methods. These are impor- conservation planning beyond previous approaches,

Conservation Biology Volume 16, No. 1, February 2002

Jepson & Whittaker Ecoregions in Context 55 such as hotspots” (Dinerstein et al. 1995:2) to “achieve sity in Indonesia means that the overlay maps of physi- representation of habitat types on a global scale” (Olson ography, geology, and land use necessary to the applica- & Dinerstein 1998:502) and to move conservation be- tion of Bailey’s third-level delineators, are well- yond “reactive efforts and into a realm of visionary, pro- developed and generally available. Unfortunately, WWF active approaches at regional and landscape scales”—or ecoregions do not make this step but instead appear to “big conservation” as they now define it ( Wikramanayke combine existing frameworks based on a gestalt meth- et al. 2001). Our review demonstrates that systems to odology that renders their output less explicit. achieve representation predate hotspot approaches and Given that the existing schemes, despite their flaws, that conservation planning to achieve habitat represen- have achieved their purpose in Indonesia and are trans- tation has been the norm in Indonesia for 25 years. Con- parent in their methods and operational criteria, we servation planners have long thought beyond political question the wisdom of introducing the WWF ecore- boundaries and at landscape scales. The Bentuang-Kari- gions system. The tangible additional benefits that it mun transnational reserve (designated in 1995) and the promises—extension to the aquatic and marine realms Kerinci-Seblat National Park (declared in 1982), which and introduction of an ecologically based planning spans four provincial (state) boundaries, are testimony framework to a new generation of decision-makers—ap- to this point. The 1982 Indonesian national conservation pear limited in the Indonesian context and may be as plan was proactive and visionary. Furthermore, deliver- readily attainable within the existing conservation plan- ing on this vision, which is already advanced, is the best ning framework. Moreover, there is a risk that the intro- hope of meeting biodiversity conservation goals. This is duction of a new scheme may undermine and delay im- especially so given that spatial planning in the wider plementation of existing schemes that are well along in landscape may be obsolete for the next decade or more achieving comparable goals. because Mafia-like networks involved in illegal logging In other parts of the world where long-established and land grabs constitute the de facto institutional ar- conservation planning frameworks are lacking, there may rangement for natural resource management in many be a stronger case for introducing the WWF ecoregions newly decentralized districts ( Jepson et al. 2001). Fur- framework, providing that the methodological concerns thermore, given that conservation resources are scarce, raised here (and see Wright et al. 1998) are addressed. ever more detailed ecoreogional work may take funds from field conservation. Nevertheless, we believe there is scope to modernize Acknowledgments the Dasmann-Udvardy system. We suggest that Bailey’s third-tier delineator of landform mosaic or landscape We thank C. Bibby, W. Duckworth, A. Hamilton, S. level can readily be incorporated into the Dasmann- Schmitt, and J. Tordoff for discussion of earlier versions Udvardy system as a third- or fourth-level delineator (Ta- of this paper, and R. E. Pressey, T. C. Whitmore, A. J. ble 1), and we see five principal benefits of doing this. Whitten, and three anonymous reviewers for construc- First, professional conservation planners in Indonesia tive comments during the review process. The views ex- use several maps (and systems) in conjunction with one pressed are our own. We thank BirdLife International, another. A landform-based map would provide a wel- the Asian Conservation Bureau, and World Wildlife come new resource to replace existing physiographic Fund–United States for permission to reproduce the maps. maps (e.g., RePPProT 1990) that are sometimes found lacking for conservation planning purposes. Second, Literature Cited landscape units are suitable for biogeographical analysis at increasingly discrete scales. Third, landscape provides Anderson, J. R. 1970. Major land uses (map, scale 1:750,000). Pages a framework for considering ecological linkages and pro- 158–159 in The national atlas of the United States of America. U.S. Geological Survey, Washington, D.C. cesses at the scale at which they operate (Brunckhorst Bailey, R. G. 1983. Delineation of ecosystem regions. Environmental & Rollings 1999). Fourth, humans change landscapes, Management 19:239–248. and landscapes embody culture, so threats to biodiver- Bailey, R. G. 1989. Explanatory supplement to ecoregions map of the sity operate at this level. This dynamic is a basic idea in continents. Environmental Conservation 16:307–309. emerging integrated implementation approaches such Bailey, R. G. 1996. Ecosystem geography. Springer-Verlag, New York. Bailey, R. G. 1998. Ecoregions: the ecosystem geography of the oceans as bioregional management (Miller 1995). Finally, all and continents. Springer-Verlag, New York. agencies involved in natural resource or land manage- BAPPENAS. 1993. Biodiversity action plan for Indonesia. Ministry of ment (and society as a whole) perceive large areas as National Development Planning, National Development Planning collections of landscapes, thereby providing the basis Agency, Jakarta. for the interagency cooperation necessary for successful Blair, W. F. 1950. The biotic provinces of Texas. Texas Journal of Sci- ence 2:33–117. conservation planning and implementation. Blower, J. 1973. Report on nature conservation and wildlife manage- Ironically, the fact that state-planned conversion of ment in Indonesia. Country report WS/E2273. Food and Agricul- natural landscapes constitutes a major threat to biodiver- ture Organization of the United Nations, Rome.

Conservation Biology Volume 16, No. 1, February 2002

56 Ecoregions in Context Jepson & Whittaker

Brunckhorst, D. J., and N. M. Rollings. 1999. Linking ecological and so- Jepson, P., J. K. Jarvie, K. Mackinnen, and K. A. Mon. 2001. The end cial functions of landscapes. I. Influencing resource governance. for Indonesia’s lowland forests? Science 292:859–860. Natural Areas Journal 19:57–64. KLH. 1993. Indonesian national strategy on the management of biolog- Bush, M. B. 1994. Amazonian speciation: a necessarily complex model. ical diversity. State Ministry of Environment, Jakarta. Journal of Biogeography 21:5–17. Köppen, W. 1931. Grundriss der Klimakunde. Walter de Gruyter, Berlin. Clements, F. E., and V. E. Shelford. 1939. Bio-ecology. Chapman & Küchler, A. W. 1964. Potential natural vegetation map (scale 1: Hall, New York. 7,500,000). Pages 89–91 in The national atlas of the United States. Dasmann, R. F. 1972. Towards a system for classifying natural regions U.S. Geological Survey, Washington, D.C. of the world and their representation by national parks and re- Küchler, A. W. 1970. Potential natural vegetation map of the contermi- serves. Biological Conservation 4:247–255. nous United States. Map and illustrated manual. American Geo- Dasmann, R. F. 1973. A system for defining and classifying natural re- graphical Society, New York. gions for the purposes of conservation. World Conservation Union, Küchler, A. W. 1975. Vegetation maps of North America. University of Morges, Switzerland. Kansas Libraries, Lawrence. Dinerstein, E. 1999. Preface. Page xxi in Ricketts et al., editors. Terres- Laumonier, Y. 1997. The vegetation and physiography of Sumatra. Klu- trial ecoregions of North America. A conservation assessment. Is- wer, Dordrecht, The Netherlands. land Press, Washington, D.C. Long, A. J., M. J. Crosby, A. J. Stattersfield, and D. C. Wege. 1996. To- Dinerstein, E., D. M. Olson, D. J. Graham, A. L. Webster, S. A. Pimm, M. wards a global map of biodiversity: patterns in the distribution of A. Bookbinder, and G. Ledec. 1995. A conservation assessment of restricted-range birds. Global Ecology and Biogeography Letters 5: the terrestrial ecoregions of Latin America and the Caribbean. The 281–304. World Bank, Washington, D.C. MacArthur, R. H., and E. O. Wilson. 1963. An equilibrium theory of in- Ehrlich, P. R., and A. H. Ehrlich 1981. Extinction: the causes and conse- sular zoogeography. Evolution 17:373–387. quences of the disappearance of species. Random House, New York. MacArthur, R. H., and E. O. Wilson. 1967. The theory of island bioge- Ehrlich, P. R. 1988. Tomorrow’s world: why saving biodiversity is to- ography. Princeton University Press, Princeton, New Jersey. day’s priority. World Birdwatch 10:6–9. MacKinnon, J., editor. 1997. Protected area systems review of the Food and Agriculture Organization of the United Nations ( FAO). 1977. Indo-Malayan realm. The Asian Bureau for Conservation Limited, Nature conservation and wildlife management. National parks de- Canterbury, United Kingdom. velopment project Indonesia: project findings and recommenda- MacKinnon, J., and M. B. Artha. 1982. National conservation plan for tions. Interim report DP/INS/73/013, Rome. Indonesia, Volumes I–VIII. Field reports 17, 18, 34, 35, 36, 39, 44 Goldman, E., and R. Moore. 1945. The biotic provinces of Mexico. FO/INS/78/061. Food and Agriculture Organization of the United Journal of Mammalogy 26:347–360. Nations, Rome. Government of Indonesia. 1991. Indonesia tropical forestry action MacKinnon, J., and K. MacKinnon. 1986a. Review of the protected plan. Ministry of Forestry, Jakarta. area systems of the Afrotropical realm. World Conservation Union Green, M. J. B., M. G. Murray, G. C. Bunting, and J. R. Paine. 1996. Pri- and United Nations Development Program, Gland, Switzerland. orities for biodiversity conservation in the tropics. Biodiversity bul- MacKinnon, J., and K. MacKinnon. 1986b. Review of the protected ar- letin 1. World Conservation Monitoring Centre, Cambridge, United eas system in the Indo-Malayan Realm. World Conservation Union, Kingdom. CNPPA, Gland, Switzerland and United Nations Development Pro- Hagmeier, E. M. 1966. A numerical analysis of the distributional pat- gram, Cambridge, United Kingdom. terns of North American mammals. II. Re-evaluation of the prov- MacKinnon J., and J. Wind. 1981. Birds of Indonesia. Special report F0/ inces. Systematic Zoology 15:279–299. INS/78/061. Food and Agriculture Organization of the United Na- Hagmeier, E. M., and C. D. Stults. 1964. A numerical analysis of the dis- tions, Rome. tributional patterns of North American mammals. Systematic Zool- MacKinnon, K., G. Hatta, H. Halim, and A. Mangalik. 1996. The ecol- ogy 13:125–155. ogy of Kalimantan. Periplus Editions, Singapore. Hall, E. R., and K. Kelson. 1959. The mammals of North America. 2 vol- Md. Nor, S. 2001. Elevational diversity patterns of small mammals on umes. Ronald Press, New York. Mount Kinabalu, Sabah, Malaysia. Global Ecology and Biogeogra- Hall, B. P., and R. E. Moreau. 1962. A study of rare birds in Africa. Bul- phy 10:41–61. letin of the British Museum of Natural History 8:313–378. Milanova, E. V., and A. V. Kushlin. 1993. World map of present day Hammond, E. H. 1954. Small-scale continental landform maps. Annals landscapes: an explanatory note. United Nations Environmental of the Association of American Geographers 44:33–42. Program, Moscow State University, Moscow. Hammond, E. H. 1964. Classes of land-surface form in the forty eight Miller, A. 1951. An analysis of the distribution of the birds of Califor- states, U.S.A. Annals of the Association of American Geographers nia. University of California Publications on Zoology 50:531–644. 54:map supplement 4 (scale 1:5 million). Miller, K. 1995. Balancing the scales: guidelines for increasing biodi- Harper, F. 1945. Extinct and vanishing mammals of the old world. Amer- versity’s chances through bioregional management. World Re- ican Committee for International Wild Life Protection, New York. sources Institute, Washington, D.C. International Council for Bird Preservation ( ICBP). 1992. Putting Momberg, F., P. Jepson, and H. van Noord. 1998a. Justification and biodiversity on the map: priority areas for global conservation. preliminary boundary proposals for a Sebuku-Sembukang reserve ICBP, Cambridge, United Kingdom. in East Kalimantan, Indonesia. PHPA/WWF Indonesia technical James, P. E. 1959. A geography of man. Ginn, Boston memorandum 1. 2nd edition. Yayasan WWF–Indonesia, Jakarta. Jepson, P. 1995. The PHPA/BirdLife International–Indonesia Pro- Momberg, F., P. Jepson, and H. van Noord. 1998b. Kalimantan biodi- gramme: goals and approaches. Kukila 7:90–98. versity assessment. Final report. Yayasan WWF–Indonesia, Jakarta. Jepson, P., and Sujatnika. 1997. Avian distribution patterns in Nusa Monk, K. A., Y. de Fretes, and G. Reksodiharjo-Lilley. 1997. The ecol- Tenggara and Maluku. Pages 350–351 in K. A. Monk, Y. De Fretes, ogy of Nusa Tenggara and Maluku. Tuttle, Boston. and G. Reksodiharjo-Lilly, editors. The Ecology of Nusa Tenggara Myers, N. 1979. The sinking ark: a new look at the problem of disap- and Maluku. Periplus Editions, Singapore. pearing species. Pergamon, Oxford. Jepson, P., A. B. Ora, R. Suparman, and R. Wahyu. 1996. Evaluation of Myers, N. 1988. Threatened biotas: “hotspots” in tropical forests. Envi- a protected area network for the conservation of forest values on ronmentalist 8:1–20. Sumba, East Nusa Tenggara. PHPA/BirdLife International–Indonesia Myers, N. 1991. The biodiversity challenge: expanded “hotspots” anal- Programme. Report. BirdLife International, Bogor, Indonesia. ysis. Environmentalist 10:243–256.

Conservation Biology Volume 16, No. 1, February 2002

Jepson & Whittaker Ecoregions in Context 57

Myers, N., R. A. Mittermeier, C. G. Mittermeier, G. A. B. da Fonseca, Trewartha, G. T. 1968. An introduction to climate. 4th edition. and J. Kent. 2000. Biodiversity hotspots for conservation priorities. McGraw-Hill, New York. Nature 403:853–858. U.S. Forest Service (USFS). 1993. ECOMAP. National hierarchical frame- Nelson, B. W., C. A. C. Ferreira, M. F. de Silva, and M. L. Kawasaki. work of ecological units. USFS, Washington, D.C. 1990. Endemism centres, refugia and botanical collection density Udvardy, M. D. F. 1969. Dynamic zoogeography, with special refer- in Brazilian Amazonia. Nature 345:714–716. ence to land animals. Van Nostrand Reinhold, New York. Noss, R. F. 1992. The wildlands project and conservation strategy. Udvardy, M. D. F. 1975. A classification of the biogeographical provinces Wild Earth (special issue):10–15. of the world. World Conservation Union, Morges, Switzerland. Noss, R. F., and A. Y. Cooperrider. 1994. Saving nature’s legacy: pro- van Steenis, C. G. G. J. 1957. Outline of vegetation types in Indonesian tecting and restoring biodiversity. Island Press, Washington, D.C. and some adjacent regions. Proceedings of Pacific Science Congree Oates, J. F. 1999. Myth and reality in the rainforest: how conservation 8:61–97. strategies are failing in West Africa. University of California Press, Voss, F. 1983. Atlas: East Kalimantan, Indonesia. East Kalimantan Berkeley. Transmigration Area Development Project (TAD), Department of Olson, D. M., and E. Dinerstein. 1998. The Global 200: a representa- Manpower and Transmigration, Jakarta, and German Agency for tion approach to conserving the Earth’s most biologically valuable Technical Cooperation, Hamburg. ecoregions. Conservation Biology 12:502–515. Wallace, A. R. 1876. The geographical distribution of animals. 2 vol- Omernik, J. M. 1987. Ecoregions of the conterminous United States. umes. Macmillan, London. Annals of the Association of American Geographers 77:118–125. Weaver, J., and F. E. Clements. 1938. Plant ecology. 2nd edition. 2 vol- Omernik, J. M. 1995. Ecoregion: a framework for managing ecosys- umes. Macmillian, London. tems. George Wrights Society Forum 12:35–50. Wells, D. R. 1985. The forest avifauna of western Malasia and its conser- Omernik, J. M., and R. G. Bailey. 1997. Distinguishing between water- vation. Pages 213–232 in A. W. Diamond and T. E. Lovejoy, editors. sheds and ecoregions. Journal of American Watershed Resources Conservation of tropical forest birds. Technical publication 4. Inter- Association 33:935–949. national Council for Bird Preservation, Cambridge, United Kingdom. PHPA. 1999. Protected area database. Sub-dit Pemolahan, National Wells, M. K., S. Guggenhiem, A. Khan, W. Wardojo, and P. Jepson. Conservation Information Centre, Bogor, Indonesia. 1999. Investing in biodiversity: a review of Indonesia’s integrated Pressey, R. L. 1999. Editorial: systematic conservation planning for the conservation and development projects. The World Bank, Wash- real world. Parks 9:1–6. ington, D.C. Regional Physical Planning Programme for Transmigration. 1990. The Whitmore, T. C. 1984. Tropical rain forests of the far east. Clarendon land resources of Indonesia: a national overview (and accompany- Press, Oxford, United Kingdom. ing atlas). Land Resources Department, Natural Resources Insti- Whittaker, R. J. 1998. Island biogeography: ecology, evolution, and tute, Overseas Development Administration, London, and Ministry conservation. Oxford University Press, Oxford, United Kingdom. of Transmigration, Jakarta. Whitten, A. J., M. Mustafa, and G. S. Henderson. 1987. The ecology of Richards, P. W. 1996. The tropical rain forest: an ecological study. 2nd Sulawesi. Gadjah Mada University Press, Yogyakarta, Indonesia. edition. Cambridge University Press, Cambridge, United Kingdom. Whitten, A. J., R. E. Soeriaatmadja, and S. A. Afiff. 1996. The ecology of Ricketts, T. E., E. Dinerstein, D. M. Olson, C. J. Loucks, W. Eichbaum, Java and Bali. Tuttle, Boston. D. DellaSala, K. Kavanagh, P. Hedao, P. T. Hurly, K. M. Carney, R. Whitten, A. J., S. J. Damanik, J. Anwar, and N. Hisyam. 2000. The ecol- Abell, and S. Walters. 1999. Terrestrial ecoregions of North Amer- ogy of Sumatra. 2nd edition. Tuttle, Boston. ica: a conservation assessment. Island Press, Washington, D.C. Wikramanayake, E. D., E. Dinerstein, C. Loucks, D. Olson, J. Morrison, Riley, J. 1997. Conservation on the Sangihe and Talaud islands, Indone- J. Lamoreux, M. McKnight, and P. Hedao. 2001. Terrestrial ecore- sia. Bulletin of the Oriental Bird Club 25:19–23. gions of the Indo-Pacific: a conservation assessment. Island Press, Searle, J. 1996. The construction of social reality. Penguin Books, London. Washington, D.C. Stattersfield, A. J., M. J. Crosby, A. J. Long, and D. C. Wedge. 1998. Endemic Williamson, D., editor. 1996. Conservation by design: a framework for bird areas of the world: priorities for biodiversity conservation. Conser- mission success. The Nature Conservancy, Washington, D.C. vation series 7. BirdLife International, Cambridge, United Kingdom. Wilson, E. O., editor. 1988. Biodiversity. National Academy Press, Sumardja, E. 1981. First five national parks in Indonesia. Parks 6:1–4. Washington, D.C. Sujatnika, P. Jepson, T. R. Soehartono, M. J. Crosby, and A. Mardiastuti. World Conservation Union and World Wildlife Fund–United Kingdom 1995. Conserving Indonesian biodiversity: the endemic bird area ( IUCN & WWF ). 1994. Centres of plant diversity: a guide and strat- approach. PHPA/BirdLife International-Indonesia Programme, Jakarta. egy for their conservation. IUCN, Gland, and WWF, Godalming, Terborgh, J., and B. Winter. 1983. A method for siting parks and re- Switzerland. serves with special reference to Columbia and Equador. Biological World Wildlife Fund ( WWF ). 1998. Proceedings: ecoregion-based Conservation 27:45–58. conservation workshop. WWF, Washington, D.C. Trainor, C., W. Prayitno, D. Lesmana, and A. Gatur. 2000. Searching for World Wildlife Fund ( WWF ). 2000. The global 200 ecoregions: a a future: the biological and cultural importance of the Mbeliling user’s guide. WWF, Washington, D.C. forest for biodiversity conservation on flores. Report 10. BirdLife Wright, R. G., M. P. Murray, and T. Merrill. 1998. Ecoregions as a level International, Bogor, Indonesia. of ecological analysis. Biological Conservation 86:207–213.

Conservation Biology Volume 16, No. 1, February 2002