Contents - Next

This is the old United Nations University website. Visit the new site at


This study examines the diverse traditional systems of agroforestry that have evolved over millennia in the Pacific Islands. Attention is also given to present-day urban agroforestry, to agroforestry as practiced in conjunction with monocultural cash-cropping, and to the modern agroforestry projects promoted by governments and international funding agencies. The study is based on an aggregate of several decades of research on Pacific Island agro-ecosystems by the five contributors. From 1982 through 1989, field research for the study was funded by the United Nations University.

We intend the study to be a contribution to an increased understanding of the diversity and utility of traditional agroforestry systems in the Pacific. Our further purpose is to provide an inventory of Pacific trees and a description of Pacific Island agroforestry systems that might serve as bases for further development or re-development of Pacific agroforestry in the face of pressures for agrodeforestation.

We would like to thank the United Nations University for funding the research project "Pacific Island Agrosilviculture Systems: A Basis for Sustained Development and Environmental Protection" and also for providing funds for final bibliographic work and consultation between the editors. We are also grateful to Lee MacDonald, UNU Programme Officer at the time of the initiation of the project, to Kathleen Landauer, who served in the same position later, and to Professors Walter Manshard and Roland Fuchs of the UNU for their patience, understanding, and support over the long life of the project. We would also like to thank the many practicing agroforesters (professionals and Pacific Island villagers and farmers) and other per sons throughout the Pacific who have so freely shared their knowledge. Thanks are also due to the University of the South Pacific for its continued assistance and institutional support. Particular thanks to Sheila Singh, Sharon Bing, Iliana Seru, Marica Bolabola, and Sharon Smith-McGowan, who typed the numerous drafts with their daunting lists of scientific and vernacular plant names.

W.C. Clarke and R.R. Thaman, The University of the South Pacific, Suva, Fiji

1 Introduction

Context of the study
Geographical background
Definition of terms
Deforestation and agrodeforestation in the Pacific
Organization of the study

Context of the study

In much of the tropical world, agriculture and forests compete for the use of land, with agriculture usually the winner. The defeat of forests in this competition is the continuation, at a faster rate than ever before, of one of the most ancient relationships between humanity and the environment - the replacement of wildlands with human domesticates, including humans themselves (Janzen 1990, xi). Now, because a near total loss of tropical forest is foreseen and because newly cleared as well as older agricultural lands suffer increasing degradation, governments, funding agencies, and scientists seek ways to rehabilitate the damaged interface of agriculture and forests and to prevent further deterioration.

To achieve these goals, attention has turned strongly to "agroforestry," a term and concept that became widely known after the publication in 1977 of the seminal work Trees, Food and People: Land Management in the Tropics by Bene, Beall, and Cote, who recommended the establishment of what became the International Council for Research in Agroforestry (ICRAF). Since then, interest in agroforestry has greatly increased, and the concept has now been brought into prominence under the banner of "sustainable development." Researchers now specialize in the discipline of agroforestry; many conferences have focused on agroforestry; and a rapidly growing literature has been published - e.g., the ICRAF journal Agroforestry Systems, many articles in other journals, and at least 16 books, manuals, or conference proceedings in the past 6 years (Baumer 1990;

Beets 1989; Buck 1989; Budd et al. 1990; Gholz 1987; Gregersen et al. 1989; Kartasubrata et al. 1990; MacDicken and Vergara 1990; Nair 1989b; Raintree 1987a, 1987b; Rocheleau et al. 1989; Steppler and Nair 1987; Vergara and Briones 1987; Withington et al. 1988; Wood and Burley 1991).

Examination of this literature reveals that there are two distinct, although not always entirely separated, approaches to agroforestry: the modern, institutional; and the traditional, or indigenous. Institutional agroforestry relies on modern agronomic science and field experimentation for such purposes as assessing yields of crops, trees, and animals in varying combinations and under varying conditions; determining competition or complementarily of system components; seeking maximization of aggregate production; or determining which leguminous tree species produce the green manure supportive of the greatest nitrogen mineralization. The approach that focuses on traditional or indigenous agroforestry (Denevan et al. 1984) arises from cultural geography and ecological anthropology; it seeks to record the attributes of traditional or non-institutionalized agroforestry systems that are in use now and that may have been practiced by tropical peoples for centuries or millennia. Some cultural-ecological studies derive purely from ethnographic interest but most are now also motivated by the belief that the traditional or non-institutionalized systems are worthy of preservation and dispersal. The present study, written by geographers, mainly follows the cultural-ecological approach rather than the quantitatively analytic agronomic approach, but our interest extends beyond the ethnographic to the applied, for we believe that the past and present agroforestry systems that have been empirically developed by Pacific peoples have much to offer in the search for sustainable development.

A merit of the indigenous systems is that they are based on indigenous trees, which are familiar and useful to local people and already adapted to local climatic and soil conditions (Montagnini 1990, 5051). Further, though any particular local agroforestry system probably yields less than its potential maximum, its technologies are adapted to local conditions by virtue of empirical experimentation and have already been adopted by local people. In many cases, the system could be improved by tinkering based on analytic agronomic knowledge but it does not have to be invented anew. It is in place and it works.

The value of the indigenous systems is acknowledged by most agroforestry researchers, and the antiquity of agroforestry among tribal and peasant peoples is widely recognized (e.g., King 1989). One of ICRAF's projects, launched in the early 1980s, was a systematic inventory of agroforestry systems (Nair 1987): an effort that showed "there was a bewildering array of agroforestry systems worldwide" with more than 2,000 species of multi-purpose trees in use (Steppler 1987, 15). None the less, most government-supported, aid-funded projects in agroforestry follow the institutional approach, which usually requires the introduction of unfamiliar, non-indigenous trees and associated slow, complex experimentation. In discussing the difficulty of gaining scientific understanding of the large number of agroforestry systems, Steppler (1987, 17) noted:

Any one system undergoing experimentation would include, at a minimum, a tree species and a crop species. Each of these could have variation in genotype and management such as spatial arrangement, maturity type for crop and harvesting methods (e.g., lopping and coppicing timing for the tree). It quickly becomes apparent that we are dealing with a multifactor design with many combinations. As we add species of trees and/or crops or introduce animals, the experiment grows in size logarithmically.... The other dimension to the problem is the fact that we have combined long-lived woody perennials with annuals, short-lived perennials and/or animals. Ideally, experiments should continue for the life of the longest-lived component; this could be upwards of 40 years and we cannot wait that long. Thus, we must also devise tests and methods of prediction that will have acceptable levels of confidence in predicting long-term effects.

These sorts of difficulties suggest to some planners and students of agroforestry that, while there is a need to build institutionalized analytic capacity, it should be recognized that "many of the problems which are addressed by agroforestry do not have clear and explicit objectives associated with them. That is, the task is not necessarily one of solving a problem, but rather one of establishing a context in which social learning can take place" (Budd et al. 1990, 332-333).

The existence of so many site-specific indigenous agroforestry systems and the information that is available about them makes it less necessary to seek "valid" scientific findings by means of long experimentation with a fixed set of variables according to a strict problem-solving paradigm. As Nair (1984, 73) says, such luxuries cannot be afforded in agroforestry research at this stage. Existing indigenous systems already satisfy, to varying degrees, the three basic criteria that Raintree (1990, 58) argues should be built into the design of a good agroforestry system: productivity, sustainability, and adoptability. That the systems have remained in use shows that local farmers are convinced of the adequacy of their production in relation to inputs. The presence of trees in the system lengthens the time-span of production and protection and provides an inherent basis for sustainability. That the local people have already adopted the system meets the requirement - increasingly recognized as basic to the success of any project to ameliorate living conditions - that the intended users accept the project and be active participants in its design, trial, evaluation, and redesign. The indigenous systems have already been well tested by local farmers from season to season and generation to generation and so provide a strong locally based framework for sound management and incremental agro-ecological innovation (Gliessman 1990, 36; Richards 1985).

In their review of institutional agroforestry in the South Pacific region, Vergara and Nair (1985, 377) comment that, "in general, trees may be considered suitable to agroforestry if they complement and support rather than compete with the interplanted food crops.... Unfortunately, out of the over 2000 species that satisfy these characteristics, only a handful have been tested and used in agroforestry, such as Leucaena, Albizia, Gliricidia, and Calliandra. The rest remain untried and therefore their potentials unrealized."

In contrast, our point is that in the Pacific- and elsewhere, as in the Peruvian Amazon (Denevan and Padoch 1988; Denevan et al. 1984) - many or most of the indigenous trees suitable to agroforestry have already been tried and their potentials at least partially realized in traditional systems, even if they have not yet been the subject of institutional agronomic experiments.

Our purpose in this book is to tap this already-existing knowledge by offering details on the wide range of non-institutional, informal agroforestry systems practiced in the Pacific Islands, now or previously. We hope this information will be of applied use in the further adaptation and diffusion of agroforestry practices, whether institutional or developed informally. We do include some details about institutional agroforestry developments in the Pacific (chapter 9), but emphasize the traditional or indigenous systems because we believe that Pacific landscapes themselves and the agroforestry systems they already contain are the most useful drawing-board from which to move agroforestry into the landscapes of the future.

Geographical background

The Pacific Islands are, for the purposes of this study, defined as the islands of Melanesia, Micronesia, and Polynesia (excluding Hawaii and New Zealand) (see map p. 5). The study area includes the large continental island of New Guinea in the west and extends to the small atolls and recent volcanic islands of the central and eastern Pacific, where traditional agroforestry systems remain common in unchanged or only slightly modified forms. Some of the political units are single islands rather than island groups (e.g., Nine and Nauru), or territories rather than independent countries (e.g., New Caledonia, French Polynesia, American Samoa, and Guam). For simplicity's sake, they will be included in discussion of "groups" or "countries."

The Pacific islands

The study area includes a diversity of island types:

  1. Continental islands, such as New Guinea and New Caledonia, which are composed of geologically-ancient sedimentary, metamorphic, and igneous rocks of continental origin.
  2. Andesitic-arc islands, such as most of the islands of the Solomon Islands, Vanuatu, Fiji, Guam, and the Northern Marianas, as well asthe eastern islands of Papua New Guinea (e.g., Manus, New Britain, and Bougainville) and the more recent volcanic islands of Tonga (e.g., Niuatoputapu, Niuafo'ou, Kao, and Tofua). Andesitic-arc islands have been formed by recent andesitic volcanic activity in proximity to the subduction zone (previously, but now inappropriately, referred to as the Andesite Line), where the newer crust of the Pacific Plate and the older crust of the Indo-Australian Plate come together, with one being forced (subducted) under the other.
  3. Basaltic volcanic ("hot-spot") islands, which are high oceanic islands such as the Samoas, the Cook Islands, Tahiti in French Polynesia, the Hawaiian Islands, and Pohnpei (Ponape) and Kosrae in the Federated States of Micronesia. Hot-spot islands have been formed as the result of the extrusion of magma through cracks or rifts in the Pacific Plate as it moves over "hot spots" in the Earth's mantle.
  4. Raised limestone islands such as Nauru and Nine, the islands of Vatulele in Fiji, Aniwa in Vanuatu, Ouvea in the Loyalty Islands of New Caledonia, most of the Tongan group, and portions of Mangaia and Atiu in the southern Cook Islands and Vanuabalavu in Fiji. These islands, formed of old-reef and foraminiferous limestone, have experienced considerable uplift relative to sea level.
  5. Coral atolls is a term that technically refers to the roughly circu lar coral-reef structures at, or just below, or slightly above, sea-level that usually support scattered reef islets (motu) and that surround a central lagoon. Most of the islands in the Marshall Islands, Kiribati, Tokelau, and the Tuamotus in French Polynesia are atolls. Often included in the category are reef islets or "table reefs," such as Arorae and Tamana in Kiribati and Nukulaelae in Tuvalu, which have no lagoons and should probably be categorized as raised coral-limestone islands. Reef islets would also include islets on barrier reefs surrounding larger islands, but clearly separated from the main island by a lagoon. Similarly, some barrier reefs and associated reef islets, such as those around the central volcanic peaks and lagoons of the islands of Aitutaki in the Cook Islands and Bora Bora in French Polynesia, take the form of atolls. These islands are commonly referred to as "almost atolls" because remnants of the volcanic peak that forms the foundation of all atolls still remain emergent above sealevel in the central lagoon.

It must be stressed that, just as there are continental or andesitic islands very far from continental shores in the Pacific (e.g., Easter Island), there is also basaltic volcanic activity close to subduction zones that cannot be reconciled with the hot-spot model and remains largely unexplained (e.g., Taveuni in Fiji). Similarly, raised limestone islands, atolls, and reef islets can be found on both sides of the subduction zone, thus adding considerable ecosystemic and environmental diversity, with most island groups including more than one island type (table 1).

There is also great geographical and demographic diversity among the islands. Easter Island, Guam, Kosrae, Nauru, and Niue consist of a single small island; Fiji, Tonga, French Polynesia, and Hawaii consist of hundreds of large and small, widely dispersed islands; Papua New Guinea and Irian Jaya share the very large, high continental island of New Guinea, and both include many smaller offshore islands. Total land areas vary from 10 to 26 sq km for groups of low-lying, coral-limestone islands like Tokelau and Tuvalu to over 400,000 sq km for the continental island areas of Irian Jaya and Papua New Guinea (Thaman 1988a).

Population densities for entire groups range from just over 1 person per sq km for the Galapagos and Pitcairn Island and 2.5 for Irian Jaya, to almost 300 or more for Nauru, Truk, and Tuvalu. If the "most populous islands" are considered, the figures jump to over 100 persons per sq km for four islands, and over 200 for three islands;

Table I Types of islands up selected island nations and territories of the tropical Pacific Ocean. (Some individual islands may be composite, combining more than OK island type) and are 421 for Koror in Palau, 757 for Funafuti in Tuvalu, 1,179 for Majuro in the Marshall Islands, and 2,190 for Tarawa in Kiribati. The estimated population for Betio Islet of Tarawa atoll is expected to reach 34,066 by 1993, which will give it a population density of 4,705 per sq km, thus rivalling the population densities of Hong Kong (Carter 1984, 231). If we consider Ebeye, one of some 90 islets of Kwajalein Atoll in the Marshall Islands, to which people have been relocated by the US military to free the atoll's lagoon for intercontinental ballistic-missile testing, the population density sky-rockets to 25,000 per sq km (Keju and Johnson 1982)!


Nation or territory Continental Andesitic arc High basaltic Raised limestone Coral Atoll
Irian Jaya + +   +  
Papua New Guinea + +   + +
Torres Strait + +   +  
New Caledonia + +   + +
Solomon Islands   +   + +
Vanuatu   +   + +
Fiji   + + + +
Tonga   +   +  
Wallis and Futuna     + +  
Western Samoa     +    
American Samoa     +   +
Nine       +  
Tuvalu       + +
Tokelau         +
Cook Islands     + + +
French Polynesia     + + +
Pitcairn Island     +   +
Hawaii     +   +
Easter Island     +    
Galapagos Islands     +    
Palau       +  
Guam   +   +  
Northern Marianas   +      
Yap     +   +
Truk     +   +
Pohnpei     +   +
Kosrae     +    
Nauru       +  
Marshall Islands         +
Kiribati       + +

Sources: Personal observation; Carter 1981, 1984; Dahl 1980; Klee 1980b; Thaman 1988a.

This range of diversity in island types and population densities when combined with differences in climate, geological resources, topographical features, soil types, water availability, flora and fauna, and culture - goes a long way to explain the diversity of agroforestry systems found in the Pacific Islands. The more specific nature of the physical and biological resources of individual island ecosystems, and the extent to which they are currently protected or endangered by exploitation, have been comprehensively analysed by Arthur Dahl (1980) in his Regional Ecosystems Survey of the South Pacific Area, which contains descriptions and the conservation status of all marine and terrestrial ecosystems and physical and biological features or resources of particular ecological and cultural importance; as well as lists of rare, endemic, or endangered species; existing and proposed conservation legislation; and existing, proposed, and recommended reserves for each island group.

It is beyond the scope of this study to present such detailed information, although some of it will be considered in the case-studies of individual agroforestry systems. For detailed data on various aspects of island groups see Bakker (1977a, 1977b), Brookfield with Hart (1971), Carter (1981, 1984), Dahl (1980), Douglas and Douglas (1989), McArthur (1967), Thaman (1988e), Ward and Proctor (1980), and Winslow (1977).

Definition of terms

As noted earlier in this chapter, agroforestry is a new name for an old practice. As the word and concept became widely accepted in international land-use circles, many definitions of the term were put forward, as described in detail by Nair (1989a). The definition of agroforestry that ICRAF has used since the early 1980s is as follows (Lundgren 1987, 48):

Agroforestry is a collective name for all land-use systems and practices in which woody perennials are deliberately grown on the same land management unit as crops and/or animals. This can be either in some form of spatial arrangement or in a time sequence. To qualify as agroforestry, a given land use system or practice must permit significant economic and ecological interactions between the woody and non-woody components.

Of the many other definitions of "agroforestry," one of the most comprehensive is that of King and Chandler (1978) in an early ICRAF publication The Wasted Lands. Recently reproduced by Nair (1989a, 13), the definition reads: "Agroforestry is a sustainable land-management system which increases the overall yield of the land, combines the production of crops (including tree crops) and forest plants and/or animals simultaneously or sequentially, on the same unit of land, and applies management practices that are compatible with the cultural practices of the local population."

Along similar lines, for the purposes of this report, "agroforestry" is defined as: "The deliberate incorporation of trees into, or the protection of trees within, an agro-ecosystem in an effort to enhance its short- and longterm productiveness, its economic and cultural utility, and its ecological stability."

In this context, an "agroforestry system" is defined as: "Any agricultural system (agro-ecosystem) in which planted or protected trees are seen as economically, socially, or ecologically integral to the system."

These non-restrictive and functional definitions have been selected because they can cover the great diversity and functional utility of existing Pacific Island agroforestry/agricultural systems, which range from home-garden or household and squatter-garden agroforestry in both urban and rural areas to deliberate intercropping and the protection of trees and tree-like perennials in gardens and pastures and the planting of woodlots and protection of inland and coastal forest stands (which are seen as part of integrated agroecosystems) in sparsely populated rural areas.

Finally, the new term, "agrodeforestation," is introduced, defined by Thaman (1988b, 1988c, and 1989a) as: "The removal of trees or the deemphasis on the planting and/or protection of trees in agroecosystems."

Contents - Next