The complex agroforests of Indonesia are indigenous systems invented over generations by local people living at the margins of tropical rainforests in Sumatra (Torquebiau, 1984). After slash-and-burn of a primary forest, food crops are planted along with coffee, pepper, fruit trees (Lansium domesticum-duku, Durio zibethinus-durian), and the resin-producing damar tree (Shorea javanica). The trees eventually shade out the crops, occupy different strata, and produce high-value products such as fruits, resins, medicinals and high-grade timber (de Foresta and Michon, 1994). Biophysical scientists have studied the productivity and ecological dimensions of these systems (Michon and de Foresta, 1996). Villagers in Krui, Lampung Province, who live off these complex agroforests obviously have a higher standard of living than those that grow only crops (Bouamrane, 1996). Detailed explanations of other best-bet agroforestry activities are described elsewhere, including activities that start from secondary forest fallows, reclamation of abandoned Imperata grasslands (Garrity, 1997; Friday et al., 1999), or degraded pastures in the Amazon.
Use and Potential
The greatest potential area for expanding agroforestry practices and other forms of land-use intensification is in areas considered "degraded" at the margins of the humid tropics, such as many secondary forest fallows, Imperata grasslands, and degraded pastures. These areas amount to about 250 Mha, or 42 percent of the total deforested area of the humid tropics (Sanchez et al., 1994). A major advantage of such lands is their proximity to roads and urban areas because they often were the first ones to be cleared. We assume that 3 percent of these lands (7.5 Mha) plus 20 percent of the 15 Mha annually deforested (3 Mha)-a total of 10.5 Mha-can be put into agroforestry yearly with enabling government policies such as those described by Fay et al. (1998) and Tomich et al. (1998).
Changes in time-averaged aboveground and soil carbon can be measured as described elsewhere in this Special Report. Newly developed allometric equations for estimating biomass of shrubs and small trees provide an important tool (Palm et al., 2000).
Current Knowledge and Scientific Uncertainties
Data by Palm et al. (2000) can be used to calculate the difference between carbon lost when moist tropical forests are transformed into agroforests and carbon lost when forests are transformed into croplands or pastures. In this case, the estimate is 46 t C ha-1. The largest uncertainty is the area that will benefit from such practices. Other human-induced activities at the forest margins that do not include trees (e.g., active slash-and-burn cropping phase, pastures, and grasslands) are sources rather than sinks of carbon.
Monitoring, Verifiability, and Transparency
A 10-year period is recommended to assess the impact on soil carbon stocks. Direct measurement (time-averaged) of aboveground and soil stocks should be used for monitoring. Combining present algorithms for estimating biomass in shrubs and small trees (Palm et al., 2000), standard soil carbon sampling, and GIS techniques appears feasible. The level of carbon stock change can be readily verified through land-based techniques described above. Assumptions and methodologies associated with this practice can be explained clearly to facilitate replication and assessment. Scientific methods are open to review and replicable over time.
Aboveground carbon stocks can be rapidly eliminated by shifting from this practice to slash-and-burn, followed by cropland or pasture. About half of the carbon stored in the soil is likely to have a turnover rate of >20 t C yr-1. Stopping the activity would lead to an estimated loss of soil carbon of 50 percent in about 5 years.
Agroforestry is an economic activity that helps to reduce or eliminate poverty at the forest margins if high-value products (fruits, resins, medicines, and high-grade timber) are produced. Agroforestry also facilitates land tenure because it encourages settled farming systems (Fay et al., 1998). Tree-based systems can serve as a methane sink (Murdiyarso et al., 1994; Palm et al., 2000). Methane emitted by 1 ha of paddy rice can be absorbed by 24 ha of agroforests if the patches are close enough in a landscape mosaic. There seems to be no difference in N2O emissions between the original forest, agroforests, cropland, or grassland at the humid tropical forest margins (Palm et al., 2000).
Agroforestry systems are more biodiverse above ground than crops, grasslands, and secondary forest fallows in the humid tropics (Gillison, 1999). Differences in below-ground biodiversity seem less important (Swift, 1999). Plant diversity in mature complex agroforests of Indonesia is on the order of 300 species per hectare, which approximates that of adjacent undisturbed forests (420 plant species per hectare). The diversity of bird species in these agroforests is approximately half that of the original rainforest, and almost all mammal species are present in the agroforest (de Foresta and Michon, 1994). This biodiversity is possible because such agroforests, which are composed of hundreds of small plots that are managed by individual families, occupy contiguous areas of several thousand hectares in Sumatra. Such agroforestry "corridors" are an important tool for avian biodiversity conservation in Central America (Current et al., 1998).
In view of human migrations to the forest margins, the optimal tradeoffs between carbon capture and economic and social benefits are an important policy determination. Examples of such tradeoffs are described by Gokowski et al. (1999), Vosti et al. (1999), and Tomich et al. (1998, 1999).
Relationship to IPCC Guidelines
Changes in woody biomass stocks associated with agroforestry practices can be included by using the procedures for estimating "changes in forest and other woody biomass stocks." The Guidelines do not provide specific default values for agroforest systems, however. Changes in soil carbon stocks can also be estimated using the Guidelines, although specific examples and default values for agroforest systems are not provided. Soil carbon stock change estimates would require values for "Input Factors," and potentially "Tillage Factors," related to the level of management of agroforests. For land-use conversions to agroforestry, "Base Factors" exist in the Workbook for conversions from forest and from cultivated land.
Other reports in this collection