These management decisions are strongly influenced by the particular farming systems in which the household operates, as well as its individual circumstances. In fact, these two factors account for most of the observed variation in such decisions. Circumstances vary significantly even within any given broad farming system and also over the family life cycle of households. Within the one village there is generally a range of farm and household sizes that may still produce a similar range of agricultural products, albeit using different technologies. An understanding of the wide spectrum of farming systems that exist across the developing world can contribute to effective rural development strategies and agricultural development policies.

Farm household systems and their immediate external rural environment, including local effects of policies and institutions, markets and information linkages, are interdependent and over time co-evolve in response to changes in population, markets, technologies, policies, institutions and information flows.1

Differences in the farming system account for a major part of the variation in farm management decisions. By adopting a livelihood approach to define farming systems-recognizing multiple sources of livelihoods, including cash crops, aquaculture, self-consumption and off-farm income-it is possible to consider the local institutional environment, including farm- gate price ratios, local markets, credit availability to farmers and arrangements for resource-sharing as an integral part of the local system. In this way, a large proportion of the variation within any particular country or region is represented by a classification of farming systems.

An ideal framework would allow for the tremendous diversity of agricultural settings to be simplified and codified, without eliminating important differences that need to be taken into account by development practitioners, and would be hierarchical in order to meet the different needs of decision makers at different levels.

A recent FAO/World Bank study identified generic farming systems categories and, at a lower level of aggregation, broad farming systems, defined as populations of individual farm household systems with broadly similar resources, livelihoods and vulnerabilities, similar opportunities and constraints, and for whom similar development strategies and interventions might be appropriate. The broad farming systems defined in the study encompass many millions of households.

To develop the farming systems knowledge base, the study team blended information from global Geographic Information Systems (GIS), farming system studies, decentralized administrative data and expert knowledge. After the global forces driving change in farming systems were identified, small multidisciplinary teams identified the characteristics and extent of each farming system zone. For this purpose, the teams used the FAO agro-ecological zone (AEZ) maps as a base and added other GIS layers as relevant, including irrigation, environmental constraints, cultivated extent, livestock (in some regions) and human population. In addition, decentralized administrative data from selected units were tabulated within each zone, supplemented by estimates from local farming system studies, where available.

The teams identified the specific trends, emerging constraints and strategic development priorities for each farming system. The results were presented in regional stakeholder consultations and the feedback incorporated in the analysis. In addition, the analysis was extended in two ways: estimating the relative importance of household strategies ("backcasting" from the target of halving the number of poor people by 2015), and consolidating the findings across all regions. The analysis drew on the knowledge of more than 50 experts with practical development experience from a wide variety of disciplines.

There are eight generic farming system categories defined across the developing regions of the world. They are:
* irrigated smallholder;
* wetland rice-based;
* rainfed in humid areas;
* rainfed in steep and highland areas;
* rainfed in dry or cold areas;
* dualistic with both large-scale commercial and smallholder farms;
* coastal artisanal fishing mixed; and
* urban-based.

Within these categories, a total of 72 broad farming systems were identified and mapped (varying from 8 to 16 per region). Because equivalent farming systems exist in different regions, the total number of different farming systems at the global level is 44. In each region there are more than a dozen thematic layers which have been overlaid on the farming systems maps-including AEZ, rainfall, environmental constraints, altitude, cultivated extent, livestock population, human population-resulting in more than 100 regional maps. The six regional maps and systems were subsequently consolidated, taking into account the equivalence of some farming systems in different regions, to produce a global map with 44 farming systems (see below).2

Note: This map of farming systems of developing regions is for illustrative purposes only. Source: Data acquisition and spatial analysis team, Global Farming Study,FAO/World Bank Boundaries and delineations used on this map do not imply official endorsement or acceptance by the United Nations.

The value of any new tool or perspective must lie in its ability to provide additional insights to those utilizing it. As the initial study was undertaken at a global level, considering regionally defined farming systems, the focus until recently has been primarily on its application as a framework for large-scale regional analysis and strategy development. Two broad approaches have been used. In the first, data developed from other technical fields has been overlaid on or compared with farming systems occupying the same areas to see possible relationships and extrapolate data across systems. This has been the case, for example, for studies on disease occurrence and carbon sequestration.

Another approach has been to use defined farming systems as guidelines for developing data and approaches, as in the development of a long-term strategy for the management of African forestry resources, looking at land use pressure and predicted intensification in farming systems across the continent, and using these as contributions to identifying different levels of pressure of forest resources in future decades. For both of these approaches, the farming systems perspective can help determine regional priorities for rural investment and research, the identification and dissemination of best practices across a farming system, and for monitoring and impact assessment.

At the regional level, the process of identifying and disseminating best practices can be effectively supported through a knowledge of farming systems, as best practices are normally specific to broadly similar sets of circumstances, such as natural resources, production patterns, market opportunities and population pressure, which are captured by farming system delineations. These best practices may encompass such areas as technology adoption, services (marketing, financial, etc.), and institutional and policy arrangements. In the longer run, more geographically specific applications of the farming systems perspective will prove to be equally or perhaps even more important, providing insights at sub-regional, national or even sub-national levels. However, in order for the farming systems approach to yield useful results at these finer scales, the delineations of the original study will have to be further refined.

This would require the definition of sub-systems, which not only take greater account of climatic, natural resource and population variations within a single system but also reflect the varying influence of factors that may change across political boundaries, such as government policies and institutional development.3 In particular, statistical and biophysical data collected in censuses and rural surveys would have to be re-aggregated to reflect farming systems rather than political boundaries. The increasing use of global farming system references for such data will make this increasingly easy to achieve in the future.

Source: FAO/World Bank Global Farming Study Boundaries and delineations used on this map do not imply official endorsement or acceptance by the United Nations.

Key applications are likely to fall into three broad categories:

• input to strategy development, in a similar manner to those developed on a regional basis;
• identification and development of rural investments; and
• use as tools in developing operational guidelines and support materials.

Examples where improved farming systems data would provide significant benefits include national rural development strategies, such as the Poverty Reduction Strategy Papers and other documents prepared by the ministries of agriculture, natural resources and similar agencies, as well as guidance and discussion documents prepared by international financing institutions-the World Bank Country Assistance Strategy and the International Fund for Agricultural Development's Country Strategic Opportunities Papers. The identification and design of rural investments comprise a logical next step to these strategy exercises, and provide a similar demand for an improved perspective on patterns of household livelihoods in these areas.

In the absence of well developed national and sub-national farming systems definitions, the range of operational guidelines that may be of value is more difficult to predict. However, it is clear that extension and advisory services could benefit heavily from the ability to target technology menus, technical assistance packages, field trials and similar activities by farming system.

These applications are also likely to be of interest to the private sector and other non-governmental users. But the private sector, including the fertilizer industry, can also draw useful information from the data already available, for example by looking at intensification and diversification levels by system to identify areas of likely future strong demand for fertilizers. Globally, more than 1.5 billion hectares of cultivatable land fall within 15 predicted high- growth systems, encompassing an agricultural population of over 500 million people, while more than 2 billion hectares are classified within low intensification systems. Of the high-growth areas, the majority by area are estimated to fall within Latin America, which accounts for more than half, although the bulk of the population living in high-growth systems, in particular the rice-wheat system, can be found in South Asia.

Notes

1 Important linkages include labour markets, as well as off-farm employment, capital markets, informal safety nets, information exchange and social networks.
2 Due to its complexity, the map of Farming Systems of Developing Regions in this article is for illustrative purposes only. It should be noted that the farming systems of the OECD countries have not yet been defined or classified in this system (although national classifications do exist). Maps, with their classification keys, are available through the FAO website (http://www.fao.org/ farmingsystems/).
3 Regional-level farming systems tend to use the predominant or "average" policy and institutional environment in defining system characteristics and trends.