Overstory #224 - Global Extent and Geographical Patterns of Agroforestry
Agroforestry, the inclusion of trees within farming systems, has been a traditional landuse developed by subsistence farmers throughout most of the world. In the last 40 years it has also become a subject for systematic study and improvement, and a livelihood option promoted by landuse managers and international development efforts. It has come to the attention of global analysts and policy makers, for example UNFCCC (2008) and MEA (Hassan et al 2005), and has been recognized in regional and national development plans (NEPAD 2003) and is an obvious component of many farming systems.
Agroforestry systems range from subsistence livestock silvo-pastoral systems to home gardens, on-farm timber production, tree crops of all types integrated with other crops and biomass plantations within a wide diversity of biophysical conditions and socioecological characteristics. The term has come to include the role of trees in landscape level interactions, such as nutrient flows from forest to farm, or community reliance on fuel, timber, or biomass available within the agricultural landscape.
Despite its ubiquity and apparent importance, is hard to find data on the actual extent of agroforestry around the world. The lack of data and more fundamental misconceptions of what agroforestry is, has led to an assumption that it is globally of little importance, even by people who should know better: "During preparation of the IAAST report, USA referees said that everyone knew there were only 50,000 ha of agroforestry in the world and that they were a failure" (Roger Leakey, personal communication). Such misunderstandings lead to suboptimal policy decisions, and can best be reversed by providing objective, data-based measures of the extent of agroforestry.
Understanding the extent and distribution of trees on agricultural land, at the landscape level, including the numbers and characteristics of farmers and farming communities within those landscapes, can help to assess the importance and role of agroforestry both to the livelihood of farming communities as well as to overall global agricultural production. Further, understanding the geographic, ecological, and demographic distribution of agroforestry related land uses can also highlight those areas where increased tree densities could make a greater contribution to livelihoods or landscapes.
We set out with the aim of answering the basic question:
1. How much agroforestry land is there and where is it?
Once we had a viable method, described below, we realized it would provide much richer data that could answer further questions:
2. How many people are associated with agroforestry?
3. What patterns can be seen in the density of people in agroforestry land? What patterns of tree cover can be seen across different densities of people?
4. How are the patterns of tree cover, population density, and their interactions affected by climate and basic ecology?
Measuring agroforestry extent
For many years the term "agroforestry" was applied to particular arrangements of trees in crop and animal production systems. This view was summarised as follows:
"Agroforestry is a collective name for land-use systems and technologies, where woody perennials (trees, shrubs, palms, bamboos, etc.) are deliberately used on the same land management unit as agricultural crops and/or animals, either in some form of spatial arrangement or temporal sequence. In agroforestry systems there are both ecological and economical interactions between the different components." (ICRAF, 1993)
Based on this view, several authors have produced estimates of the extent of particular systems. IAASTD (2008) listed those in Table 1. Nair and Nair (2003) estimated the extent of alleycropping, silvopasture, windbreaks, and riparian buffers in the USA as 235.2 M ha. Kumar (2006) estimated the area of homegardens in South- and southeast Asian homegardens as 8.0 M ha, and Reisner et al (2007) estimated European silvoarable systems to cover 65.2 M ha.
The problem with trying to produce a more encompassing estimate of the extent of agroforestry systems was summarized by Nair et al (2009):
"A major difficulty in estimating the area under agroforestry is lack of proper procedures for delineating the area influenced by trees in a mixed stand of trees and crops. In simultaneous systems, the entire area occupied by multistrata systems such as homegardens and shaded perennial systems and intensive tree-intercropping situations can be listed as agroforestry. However, most of the agroforestry systems are rather extensive, where the components, especially trees, are not planted at regular spacing or density; for example, the parkland system and extensive silvopastures. The problem is more difficult in the case of practices such as windbreaks and boundary planting where although the trees are planted at wide distances between rows (windbreaks) or around agricultural or pastoral parcels (boundary planting), because the influence of trees extends over a larger than easily perceivable extent of areas... The problem has a different dimension of difficulty when it comes to sequential tropical systems such as improved fallows and shifting cultivation. In such situations, the beneficial effect of trees and other woody vegetation (in the fallow phase) on the crops that follow them (in the cropping phase) is believed to last for a variable length of time (years)."
Nair et al (2009) go on to make an estimate of 823 M ha globally under agroforestry and silvo-pastoral systems. Of these, 307 M ha are agroforestry. However their estimate comes from taking the FAO estimate of agriculture land multiplied by an estimate of 20% covered by agroforestry. The value of 20% is not based on any objectively measured data. Another estimate of global agroforestry extent which is widely quoted is Dixon (1995), who suggests 585-1215 M ha of agrosilvopastoral and agroforestry systems in Africa, Asia, and the Americas. However, this is an estimate of the area they judge technically suitable for these systems, not occupied by them.
The current view of agroforestry is not as a collection of technologies, but of trees included in agricultural landscapes. For example, Schroth and Sinclair (2003) note that agroforestry is increasingly recognized for its ecological and economic interactions at the landscape scale.
This changes the measurement problem considerably, for we have global databases which can be combined and interpreted to generate relevant information. Three data sources are used:
1. Global land use. Spatial data layers exist which classify any pixel as agricultural or some other land use.
2. Global tree cover. Remotely sensed data has been interpreted to give estimates of % tree cover in a pixel.
3. Global population. Spatially disaggregated population layers are available which give an estimate of population in any pixel and can be used to measure the extent of agroforestry in terms of population.
Details of the data sources are given in the Methods section and Appendix 1 of the Original Source (see below). "Landscape scale" is not precisely defined. However, each of the above data sources is available at 1 km x 1 km resolution. This corresponds roughly to a common notion of landscape scale. Thus we look at the 1 km x 1 km pixels that are classified as "agriculture" and find the percent tree cover in each. This varies from 0%, clearly not agroforestry, up to close to 100%, though most pixels with high tree cover have not been classified as agricultural. It is not necessary to choose a cut off value for tree cover below which we do not consider the landscape as being an agroforestry landscape. Most results can be presented as a continuum of patterns from low to high tree cover, a continuum which represents reality better than any arbitrary cut off.
We can then assume the population estimated as living in the 1 km x 1 km pixel is in some sense "connected with" that agricultural landscape and its trees. While we do not know the extent to which those people depend on the agroforestry landscape, it is reasonable to assume that at the 1 km scale they are influenced by and influence that landscape.
The limitations of this approach are numerous, with the major ones being outlined in the Original Source (see below). However, it appears to be a step forward compared with other estimates to date, particularly when it is understood as a global assessment. We do not expect the results of every pixel to closely match an observation on the ground, but expect the broad patterns to be revealing.
We believe the following key messages can be gleaned from this analysis.
A. Tree cover is a common feature on agricultural land. It is therefore essential that this is recognized by all involved in agricultural production, planning and policy development. Agroforestry, if defined by tree cover of greater than 10% on agricultural land, is widespread, found on 46% of all agricultural land area globally, and affecting 30% of rural populations. Based on our datasets, this represents over 1 billion hectares of land and 558 million people. Agroforestry is particularly prevalent in southeast Asia, Central America, and South America with over 80% of area under agroforestry.
B. It is not possible to describe the resulting patterns as "good", "bad", appropriate or inappropriate tree cover. We did not analyse the costs and benefits associated with these agroforestry lands, nor the implications of a change in tree cover. However, the existence of extensive areas of agroforestry even in arid areas shows that such systems are viable in some sense.
C. There is large variation in tree cover in agricultural land. From continental scale down to the smallest detectable in this analysis (1 km2) there is variation in tree cover in agricultural lands. But some major trends stand out.
D. There is a strong association between aridity and tree cover. The more humid the climate, the higher the level of tree cover. The results from South East Asia, Central America, and South America are examples of this relationship. However, there are still many exceptions to this rule -- high tree cover found in more arid zones and low tree cover found in more humid zones -- that are thus explained by other factors.
E. There is no general tradeoff in agricultural landscapes between people and trees. Within aridity classes and continents, there are distinct patterns in the relationship between trees and people, but these do not generally correspond to either a negative or positive correlation, except in the very low or high range of tree cover.
F. Large scale tree cover patterns cannot be fully explained by aridity, population density or region. This points towards the importance of other factors like tenure, markets, or other policies and institutions in affecting incentives for tree planting and management, as well as the historical trajectory that has lead to the current pattern.
G. Tree cover patterns and relationships to other variables like aridity or population vary considerably across sub-continent. "Global" level results are rarely replicated in any specific subcontinent and hence may not be practically applied. Further investigations at finer regional scales are likely therefore to prove even more illuminating in terms of understanding where on the landscape agroforestry is practiced.
Dixon RK .1995. Agroforestry systems: sources or sinks for greenhouse gases? Agroforestry Systems 31, 99-116.
Hassan R, Scholes R, and Ash N (eds) 2005. Ecosystems and Human Well-being: Current State and Trends, Volume 1. Island Press, Washington.
IAASTD. 2008. Agriculture at a Crossroads: Global Report. International Assessment of Agricultural Knowledge, Science, and Technology for Development. Washington DC: Island Press.
ICRAF. 1993. International Centre for Research in Agroforestry: Annual Report 1993. Nairobi, Kenya. pp 208.
Kumar, B. M. 2006. Carbon sequestration potential of tropical homegardens, in Kumar, B. M., Nair, P. K. R. (eds.): Tropical Homegardens: A Time-Tested Example of Sustainable Agroforestry. Advances in Agroforestry 3, Springer, Dordrecht, the Netherlands, pp. 185– 204.
Nair PKR, BM Kumar and VD Nair. 2009. Agroforestry as a strategy for carbon sequestration. J. Plant Nutr. Soil Sci. 172, 10–23.
NEPAD. 2003. Action Plan for the Environment Initiative. New Partnership for Africa’s Development, Midrand, South Africa.
Reisner, Y., de Filippi, R., Herzog, F., Palma, J. 2007. Target regions for silvoarable agroforestry in Europe. Ecol. Eng. 29, 401–418.
This article was excerpted with the kind permission of the publisher from:
Zomer RJ, Trabucco A, Coe R and Place F. 2009. Trees on Farm: Analysis of Global Extent and Geographical Patterns of Agroforestry. ICRAF Working Paper no. 89. World Agroforestry Centre, Nairobi, Kenya. http://www.worldagroforestry.org/downloads/publications/PDFs/WP16263.PDF
About the Authors
Robert J. Zomer
International Centre for Integrated Mountain
G.P.O. Box 3226, Khumaltar, Kathmandu, Nepal
(formerly based at ICRAF)
Forest Ecology and Management
Division Forest, Nature and Landscape -
Celestijnenlaan 200E, BE - 3001 Leuven, Belgium
World Agroforestry Centre (ICRAF)
P.O. Box 30677, 00100
World Agroforestry Centre (ICRAF)
P.O. Box 30677, 00100
Related Editions to The Overstory
- The Overstory #214--Indicators and guidelines for landscape assessment and planning for agroforestry
- The Overstory #188--Helping forests to help themselves
- The Overstory #167--Riparian buffer zone restoration for food security
- The Overstory #149--Live Fences, Isolated Trees, and Windbreaks: Tools for Conserving Biodiversity
- The Overstory #142--Urban Trees and Forests
- The Overstory #109--Cultural Landscapes
- The Overstory #105--Complex Agroforests
- The Overstory #98--Integrating Forestry into Farms
- The Overstory #95--Agroecology
- The Overstory #93--Trees, Forests and Sacred Groves
- The Overstory #88--Revegetation Planning for Farm Forestry
- The Overstory #77--Tropical Forest Conservation
- The Overstory #73--Buffers, Common-Sense Conservation
- The Overstory #66--Carbon Sequestration