Introduction
The biodiversity of Fiji is increasingly being recognized as a global conservation priority (Davis et al., Reference Davis, Heywood and Hamilton1996; Stattersfield et al., Reference Stattersfield, Crosby, Long and Wege1998; Myers et al., Reference Myers, Mittermeier, Mittermeier, da Fonseca and Kent2000; Government of Fiji, 2001; Ryan, Reference Ryan2001; Olson & Dinerstein, Reference Olson and Dinerstein2002; Masibalavu & Dutson, Reference Masibalavu and Dutson2006). Fiji's unusual biogeographical history (Van Balgooy, Reference Van Balgooy1971; Green & Cullen, Reference Green, Cullen and Coleman1973; Kroenke, Reference Kroenke, Keast and Miller1996; Hall, Reference Hall2002; Evenhuis & Bickel, Reference Evenhuis and Bickel2006; Heads, Reference Heads2006) has imparted an exceptionally diverse forest biota characterized by pronounced endemism at the level of species and higher taxa (Ash, 1982; Gibbons, Reference Gibbons1984, Reference Gibbons, Grigg, Shine and Ehmann1985; Ash & Vodonivalu, Reference Ash, Vodonivalu, Campbell and Hammond1989; Davis et al., Reference Davis, Heywood and Hamilton1996; Heads, Reference Heads2006) and the presence of numerous basal and primitive lineages (Ash, Reference Ash1992; Miller, Reference Miller1989; Hollingsworth, Reference Hollingsworth, Alberts, Carter, Hayes and Martins2004), unusual radiations (Camponotus and Pheidole ants; Sarnat, Reference Sarnat2006, Reference Sarnat2008) and Gondwanan elements (Bickel, Reference Bickel2006; Davis et al., Reference Davis, Heywood and Hamilton1996).
The Government of Fiji, landowners, NGOs and the private sector have committed to protecting Fiji's natural forest heritage through the enactment of Fiji's National Biodiversity Strategy and Action Plan (Government of Fiji, 2001), the Fiji Departments of Forestry and Environment's forestry certification programmes (Fiji Department of Forestry, 2007), and the ongoing establishment and recognition of a number of national and community-based forest protected areas (e.g. community-declared protected areas of Waisali, Bouma, Koroyanitu, Kilaka and Naicuvalevu). The conservation priorities for Fiji's forests identified in the Action Plan were based on a synthesis of existing conservation analyses (Lees, Reference Lees1989; Paine, Reference Paine1989; Tabunakawai & Chang, Reference Tabunakawai and Chang1991; Watling & Chape, Reference Watling and Chape1992; Government of Fiji, Reference Watling and Chape1993; Wright & Lees, Reference Wright, Lees, Keast and Miller1996) and consultations with biodiversity specialists (Government of Fiji, 1998a,b,c,d; Kretzschmar, Reference Kretzschmar2000).
The Action Plan promotes the ongoing refinement of conservation priorities as new data and analyses become available (Government of Fiji, 2001). As a contribution to this process we have synthesized biodiversity information and conservation analyses (Fiji Important Bird Areas, Masibalavu & Dutson, Reference Masibalavu and Dutson2006; Critical Ecosystem Partnership Fund's Fiji Profile, Olson & Farley, Reference Olson and Farley2004; Key Biodiversity Areas, Conservation International, 2005) that have become available since 2001 and re-evaluated Priority Forests for Conservation. The Fiji Department of Forestry's (2007) policy goal of 40% of all extant natural forest (corresponding to 20% of Fiji's original natural forests) to remain forested provided a percentage target. Given this target, we asked which specific forests should be designated as Protection Forests (i.e. forests where the protection of native forest cover in a relatively undisturbed state is the primary management goal; Tabunakawai & Chang, Reference Tabunakawai and Chang1991; Watling, Reference Watling1994) to achieve conservation and ecosystem service goals based on existing knowledge, principles of regional conservation strategies that have been evolving through similar efforts elsewhere (e.g. representation of distinct assemblages and habitats, maintaining ecological processes and ecosystem services, minimum-area requirements; Noss & Cooperrider, Reference Noss and Cooperrider1994; Dinerstein et al., Reference Dinerstein, Powell, Olson, Wikramanayake, Abell and Loucks2000; Groves, Reference Groves2003; Jennings et al., Reference Jennings, Nussbaum, Judd and Evans2003; Forest Stewardship Council PNG, 2006), and practical consideration of development trends and goals, and land use.
Methods
We employed a four-step approach to arrive at our recommended network of Priority Forests for Conservation:
(1) Information used in the preparation of the Biodiversity Strategy and Action Plan was reviewed. Data layers, such as existing protected areas, priority areas, natural forest cover and watersheds, were mapped. Using the geographical information systems MapInfo v. 8.5 (Pitney Bowes, Troy, USA) and ArcView v. 9.3 (ESRI, Redlands, USA) we overlaid watershed (Atherton et al., Reference Atherton, Olson, Farley and Qauqau2006) and topographic features with a map of remaining natural forests (Olson, Reference Olson2006) based on Landsat Thematic Mapper imagery (1991–1995) and Fiji Department of Forestry (1996) vegetation maps.
(2) Studies published or otherwise available after the Action Plan was prepared were evaluated for relevant information (Keppel, Reference Keppel2002; Olson et al., Reference Olson, Tuiwawa, Niukula, Bicoloa, Keppel and Naikatini2002; Barker, Reference Barker2003; Tuiwawa & Naikatini, Reference Tuiwawa and Naikatini2003a,Reference Tuiwawa and Naikatinib; Watling, Reference Watling2003; WCS, 2003; Farley et al., Reference Farley, Olson and Patrick2004; Olson & Farley, 2004; Yanega et al., Reference Yanega, Olson, Shute and Komiye2004; Atherton, Reference Atherton2005; Conservation International, 2005; Keppel, Reference Keppel2005a,Reference Keppelb,Reference Keppelc; Keppel et al., Reference Keppel, Cawani Navuso, Naikatini, Thomas, Rounds and Osborne2005, Reference Keppel, Rounds and Thomas2006; Atherton et al., Reference Atherton, Olson, Farley and Qauqau2006; Chape, Reference Chape2006; Evenhuis, Reference Evenhuis2006; Evenhuis & Bickel, Reference Evenhuis and Bickel2006; Heads, Reference Heads2006; Masibalavu & Dutson, Reference Masibalavu and Dutson2006; Monaghan et al., Reference Heads2006; Sarnat, Reference Sarnat2006, Reference Sarnat2008; Savu, Reference Savu2006; Jackson & Jit, Reference Heads2007; Keppel & Tuiwawa, Reference Keppel and Tuiwawa2007; Palmeirim et al., Reference Palmeirim, Champion, Naikatini, Niukula, Tuiwawa and Fisher2007; Rounds, Reference Rounds2007).
(3) Using a set of decision rules that considered natural habitats and ecological processes, representation of the full set of habitats and distinct species assemblages, and the realities of conservation (Table 1, Appendix 1), we assessed if there were any major gaps in how existing protected areas, priority areas and recently proposed priority areas (Olson & Farley, Reference Olson and Farley2004; Conservation International, 2005; Masibalavu & Dutson, Reference Masibalavu and Dutson2006) addressed these guidelines and decision rules. Generalized boundaries around Priority Forest blocks were then drawn using forest edges and watershed boundaries (Atherton et al., Reference Atherton, Olson, Farley and Qauqau2006) within contiguous forest if they corresponded to mataqali (i.e. traditional land ownership) and provincial boundaries. The emphasis on watershed boundaries is a practical consideration as landowners and logging companies can readily identify watershed boundaries when they are on the land. In some cases provincial and mataqali boundaries were used where they differed markedly from watershed boundaries.
(4) Priority areas not identified in previous analyses were highlighted and existing priorities confirmed, where appropriate.
We recognize that achieving agreement on the boundaries and management of protected areas often requires considerable negotiation with a variety of stakeholders, a process that will be influenced by many factors. Therefore, the generalized forest blocks delineated on our maps (Figs 1–3), and their biological justification, are more relevant to this process than the exact boundaries proposed.
The forest cover data we used date from 1991–1995 and there has been considerable logging activity since. We therefore assume that the current area, boundaries and distribution of forest cover are not exactly the same as those depicted on our maps but our collective field experience suggests that these changes are not of sufficient scale to alter our Priority Forest list. However, mapping of forest cover at a greater resolution, along with the distribution of major habitat types such as dry, transition and moist forest, is required. Current biogeographical knowledge is most comprehensive for birds and herpetofauna. Certain groups of plants, such as palms, are relatively well-studied, but the majority of taxa have not been the subject of extensive biogeographical study. Similarly, some genera and families of invertebrates and freshwater fish have recently been the subject of archipelagic-wide surveys (Sarnat Reference Sarnat2006, Reference Sarnat2008) but the majority of groups are understudied (Jenkins & Boseto, Reference Jenkins and Boseto2003; Evenhuis & Bickel, Reference Evenhuis and Bickel2006). The second largest island of Vanua Levu, in particular, is poorly surveyed for a wide range of taxa. Koro, Gau, Yasayasamoala and the southern Lau Group are also inadequately surveyed.
Results
Habitats and ecological processes
Forty areas (Table 2; Figs 1–3) were identified as Priority Forests for Conservation where protection should be the major management emphasis to achieve national-level conservation and ecosystem services goals. Together these areas cover 23% of the total land area and c. 58% of the remaining natural forest. Twenty-two large blocks (> 100 km2) of remaining natural forest were identified for protection. Two of the largest blocks of natural forest, western Serua and Wainimala, both remote areas (Olson et al., Reference Olson, Farley, Naisilisili, Raikabula, Prasad, Atherton and Morley2006), were not selected as they are experiencing intensive logging and may lose much of their conservation value in the near future. These areas should be re-evaluated if this situation changes.
We highlight as a priority two forested corridors that presently connect increasingly isolated Priority Forest blocks. These are: (1) the threatened Vunitorilau Corridor, which is the last undisturbed forest connection between the wider Sovi-Waimanu forests and the upland forests of Monasavu/Tomanivi, as well as the Serua forest blocks; and (2) the corridor between Monasavu and Tomanivi, which is disappearing rapidly. Forests of the Nakauvadra Range and Mt Evans Ranges on Viti Levu are increasingly isolated from the main forest block. Forest corridors linking the larger forests blocks of the dividing range of Vanua Levu are also of critical importance.
The relatively intact watersheds of Taveuni, southern Vanua Levu, eastern Kadavu (Koronibanuve), and south-west Viti Levu (Waimanu) were all selected. These areas, along with Gau, Tunuloa/Natewa and Nakorotubu, support some of Fiji's last forests that range from lowland to montane habitats. Several smaller islands are largely covered in native forest, including Sawa-i-Lau, western Macuata Island, Yaduataba, Namenalailai, Makodroga, Vatuvara, Vuaqava, Ogea Levu and Ogea Driki. Several larger forested watersheds that are adjacent to Fiji's high conservation value reefs (WWF, 2005; Atherton et al., Reference Atherton, Olson, Farley and Qauqau2006) were also selected (Nakauvadra, Nakorotubu, Navotuvotu, Kubulau, Vatuvonu, Delaikoro, Dogutuki, and Ovalau) to help protect these marine ecosystems.
Representation of distinct assemblages and habitats
The 40 areas represent all of the major islands and island groups, and major forest habitat types within the major islands. Mangroves, wetlands, Pandanus savannahs and littoral forests were not analysed. Most of the remaining fragments of tropical dry forest are encompassed. Tropical dry forest was formerly widespread on the leeward side of the larger islands but now occurs only in small remnant patches (Mueller-Dombois & Fosberg, Reference Mueller-Dombois and Fosberg1998).
All proposed biotic provinces and sub-provinces (see Appendix 1 for a description of identification and delineation; Fig. 4), except for Wainimala and western Serua, are represented. The four larger islands with notable bird endemism (Viti Levu, Vanua Levu, Taveuni and Kadavu) are represented, as are the majority of Important Bird Areas (Masibalavu & Dutson, Reference Masibalavu and Dutson2006). All biotic provinces for reptiles and amphibians are represented but the important site of Yanuya Island, Ono-i-Lau (the only known locality for Leiolopisma alazon) is not captured (Morrison, Reference Morrison2003a,Reference Morrisonb, Reference Morrison2005; Appendix 2: Fig. 5; biotic subdivisions may be less relevant for the relatively widespread Fijian herpetofauna; Morrison, Reference Morrison2005). Provisional biogeographic subdivisions for freshwater fish (Jenkins & Boseto, Reference Jenkins and Boseto2003; Appendix 2: Fig. 6) are all represented but two sites of special importance, the lower Sigatoka River and the Upper Lekutu River are not captured. All of the biotic provinces and sub-provinces for invertebrates (Appendix 2: Fig. 7) are represented except for the Wainimala Sub-province. Proposed biotic provinces for plants are all represented but the Wainimala Sub-province is not captured and two sites of special importance, central Kadavu and several sites with restricted-range palm species along the southern coast of Viti Levu and the southern Rewa River valley, are not captured (Fuller, Reference Fuller1997; Doyle & Fuller, Reference Doyle and Fuller1998; Zona & Fuller, Reference Zona and Fuller1999; Watling, Reference Watling2005; Appendix 2: Fig. 8).
Several areas known to have high alpha richness of plants (south-eastern Viti Levu: Tuiwawa & Doyle, Reference Tuiwawa and Doyle1998; Tuiwawa & Naikatini, Reference Tuiwawa and Naikatini2003b; Vatuvonu and Kubulau, Vanua Levu: Government of Fiji, 1994; Kretzschmar, Reference Kretzschmar2000) and invertebrates (Mt Evans/Koroyanitu/Abaca, Kubulau, Tomanivi/Wabu and Waimanu; Sarnat, Reference Sarnat2006; S. Prasad et al., unpubl. data) are priorities. Areas notable for concentrations of endemic plants and invertebrates are also represented, including Kadavu, the cloud and montane forests of central Viti Levu, Mt Evans Range (Thaman, Reference Thaman1996; Thaman et al., Reference Thaman, Whistler, Vodonaivalu and Tuiwawa1999), and south-eastern Viti Levu; hotspots within this area include Waimanu, Mt Voma, Korobasabasaga, Mt Naitaradamu, and eastern Serua (Watkins, Reference Watkins1994; Tuiwawa, Reference Tuiwawa1999). Other endemism hotspots include Taveuni, Gau and Koro. The latter two may have the most distinctive invertebrate fauna, in terms of percentage single-island endemism, of all the Fijian islands (Barker, Reference Barker2003; E. Sarnat, pers. comm.; S. Prasad et al., unpubl. data).
The largest known populations of the Critically Endangered Fijian crested iguana Brachylophus vitiensis are selected, representing the Viti Levu (Macuata Island), Mamanuca (Monuriki and Mono), Yasawa (Devuilau), and the presumptive Vanua Levu (Yaduataba) forms (Gibbons, Reference Gibbons1981; Laurie et al., Reference Laurie, Uryu and Watling1987; Harlow & Biciloa, Reference Harlow and Biciloa1999, Reference Harlow and Biciloa2000; Olson et al., Reference Olson, Tuiwawa, Niukula, Bicoloa, Keppel and Naikatini2002; Olson & Keppel, Reference Olson and Keppel2004; Harlow et al., Reference Harlow, Fisher, Tuiwawa, Biciloa, Palmeirim and Mersai2007). The broad distribution of the Priority Forests may also capture some of the poorly known geographical variation of the banded iguana Brachylophus fasciatus.
The unusual karst habitats of Sawa-i-Lau are represented but known karst habitats of Wailotua and several of the Lau Group islands are not (Heads, Reference Heads2006). The Priority Forests within the Lau and Yasayasamoala Groups are probably underrepresented in this analysis. Two of the largest extant sago palm Metroxylon vitiense swamps, Maratu and Wainikevu (Rounds, Reference Rounds2007), are included in the Serua and Waimanu forest blocks, respectively. Sago swamps on Vanua Levu are not represented, and the few documented swamps near Savusavu are small and threatened. Fiji's tropical dry forests are represented on Macuata Island, Yadua Taba, Rokosalase and Naicobocobo. Mangroves and seabird and sea-snake islands (Government of Fiji, 2001; Masibalavu & Dutson, Reference Masibalavu and Dutson2006) were not analysed.
Comparisons to previous priority-setting analyses
This analysis confirms existing priorities identified in the Fiji National Biodiversity Strategy and Action Plan (i.e. the Department of Environment's List of Sites of National Significance and Nature Reserves) and highlights several new areas of conservation importance, namely the Nakauvadra Range, the Waimanu region (south-east Viti Levu), eastern Serua, Macuata Island, Koro, Sawa-i-Lau, Nakorotubu, Yasayasamoala Group, Vatuvara, Kuata, Devuilau, Aiwa Levu and Aiwa Lailai, Vuaqava, and much of the remaining forest of southern Vanua Levu including Navotuvotu, Kubulau, Vatuvonu, Korolau (the larger forested landscape around Waisali Reserve), Dikeva and Sagani. Several of these new areas were previously identified in the Critical Ecosystem Partnership Fund's Fiji assessment (Olson & Farley, 2003; Conservation International, 2005), Important Bird Areas (IBAs would cover 39% of the remaining natural forest if implemented; Masibalavu & Dutson, Reference Masibalavu and Dutson2006) and Conservation International's (2005) Key Biodiversity Areas (KBA) draft analysis. The western and central portions of Serua (Serua Forest Wilderness and Upper Navua Gorge KBAs) and the Nausori Highlands KBA were not selected as Priority Forests because of the intensive logging in these areas.
Discussion
This Priority Forests for Conservation analysis confirms and complements previously identified priorities for the archipelago, including those of Fiji's National Biodiversity Strategy and Action Plan (Government of Fiji, 2001). There is strong consensus amongst Fiji's conservation community on important areas for protection, and the list and areas of proposed priorities continues to expand as knowledge of the archipelago's biodiversity increases and conservation strategies are refined. The Priority Forest map also addresses a broad range of conservation and ecosystem service goals. If all the Priority Forests and proposed marine priority areas (WWF, 2005) could be designated for protection, Fiji would have one of the most comprehensive and robust systems of protected areas (IUCN/UNEP, 1986; Rodrigues et al., Reference Rodrigues, Akçakaya, Andelman, Bakarr, Boitani and Brooks2004).
National development goals and the needs and aspirations of local people (Government of Fiji, 2004, 2005) may, however, preclude protection for the entire proposed Priority Forest network. However, our analysis suggests that the Government of Fiji's (Fiji Department of Forestry, 2007) policy goal of protecting 40% of Fiji's remaining forests may represent a minimum threshold, below which unacceptable losses of biodiversity may occur, both in forests and coastal marine ecosystems. Levels of 10–20% for protected areas that are commonly discussed in conservation negotiations (Desmet & Cowling, Reference Desmet and Cowling2004; Svancara et al., Reference Svancara, Brannon, Scott, Groves and Pressey2005) are inadequate to protect the full range of species in the archipelago. Recent biogeographical studies (Watling, Reference Watling2005; Heads, Reference Heads2006; Sarnat, Reference Sarnat2008) suggest that patterns of local endemism in Fiji may be more complex than previously documented, especially for invertebrates and plants. Multiple and widespread protected forests will be needed to represent the biota adequately.
Protecting < 40% of Fiji's remaining natural forest may also compromise important ecosystem services. Almost every remaining native forest has communities living within it, downstream, or utilizing coastal resources influenced by run-off. At a minimum, upper watersheds should all be protected, as indicated in Fiji's Forest Function map (Watling, Reference Watling1994), for their role in maintaining healthy freshwater and coastal fisheries, diminishing flood impacts, providing clean water, enhancing local rainfall and seasonal water availability, and as a source of non-timber forest products. Fiji's islands are sufficiently small that the effects of altering watersheds are immediately and dramatically felt in freshwater and coastal marine ecosystems, with cascading effects on subsistence and commercial fisheries and community health.
The current proliferation of high-impact logging operations in smaller coastal watersheds of Vanua Levu and Viti Levu is probably one of the major drivers of degradation in Fiji's freshwater and coastal resources (Atherton et al., Reference Atherton, Olson, Farley and Qauqau2006). Greater weight should be given to the critical role of intact forests in providing ecosystem services to Fiji's people and economy, and the substantial economic value of these services needs to be recognized and integrated into cost/benefit analyses and planning (Cambie & Ash, Reference Cambie and Ash1994; Government of Fiji, 1998b; Balmford et al., Reference Balmford, Bruner, Cooper, Costanza, Farber and Green2002). Fiji's efforts to develop a certification programme for forestry operations reflects a growing recognition of the importance of ecosystem services provided by protected forests (Fiji Department of Forestry, 2007).
If a protected area system based on the Priority Forest network is put in place then the entire forest product industry in Fiji could potentially receive a national-level forestry certification (i.e. landscape-scale certification; Farley et al., Reference Farley, Olson and Patrick2004), as long as standards relating to logging practices, social values, and High Conservation Value Forests (i.e. a certification category of the Forestry Stewardship Council's widely-adopted certification programme that emphasizes a narrower range of conservation considerations; Forest Stewardship Council, 2003; Jennings et al., Reference Jennings, Nussbaum, Judd and Evans2003) are upheld. While challenging to implement (Gullison, Reference Gullison2003; Lenoa, Reference Lenoa2003; Sesega, Reference Sesega2003), the economic and societal benefits of landscape-scale forestry certification may, in the long-term, outweigh those from certifying single operations (Pierce & Ervin, Reference Pierce and Ervin2003; Farley et al., Reference Farley, Olson and Patrick2004; Olson, Reference Olson2006).
The Government of Fiji continues to work towards the commitment it made in 2007 to protect 40% of its remaining natural forest. Our analysis recommends 58% protection to achieve conservation and ecosystem service goals. We, collectively, are thinking at a similar scale of protection, a pre-condition for success in the task that lies ahead, working with landowners and the forest industry to turn protection goals into reality.
Acknowledgements
We thank the Government of Fiji for their interest and support, the landowners for their interest, support and permission to work in their forests and for their hospitality and generosity. We are indebted to Fiji's Department of Environment (DOE), particularly to Manasa Sovaki and Epeli Nasome; the Department of Forestry (MAFF), particularly to Susana Tuisese, Inoke Wainiqolo, Joe Ceinturaga and Sanjana Lal; and Fiji Fisheries, particularly Aisake Batibasaga and Sunia Waiqanabete. We are grateful for the support and assistance of the Fijian Affairs Board, Provincial governments and traditional leaders and councils. We appreciate and recognize the following organizations for sharing data and expertise for our analyses: MAFF, DOE, Native Land Trust Board, The Pacific Islands Applied Geoscience Commission, Fiji Department of Lands and Survey, Fiji Hydrology, Mineral Resource Department, The National Trust for Fiji, University of the South Pacific, BirdLife, South Pacific Regional Herbarium, WWF and Wetlands International. This study was supported by a grant from the East Asia and Pacific Environment Initiative Program of the US Agency for International Development (Fiji Forestry Landscape Certification Project). Additional support came from the US Department of State, the US National Science Foundation Fiji Arthropod Survey Project, and the Schlinger Foundation. One of the authors, Lemeki Lenoa, passed away before completion of this study. We dedicate this paper to him, a great colleague and among Fiji's most committed and enlightened foresters and conservationists. We thank John Morrison, Neal Evenhuis, Dan Bickel, Eli Sarnat and two anonymous reviewers for reviewing earlier drafts.
Appendices 1–2
The appendices for this article are available online at http://journals.cambridge.org
Biographical sketches
The authors are naturalists, scientists, conservationists, natural resource managers, foresters, field researchers and conservation geographical information system specialists who share an interest in, and concern for, the future of Fiji's forests. They share a belief that a functional balance between conservation and sustainable development, livelihoods and quality of life for Fijians is attainable through science-based analyses, participatory planning and decision-making, respect for tradition, and partnerships and collaboration among all stakeholders. The decisions Fijians make today about preserving their natural wealth and heritage are critical because opportunities are likely to diminish in the coming decades if trajectories of forest loss continue. This analysis is a contribution from the authors' collective efforts to find the balance.