Module 4: Benchmark site selection and description for ecoregional research in crop–animal systems

D. Pezo

ILRI, c/o IRRI DAPO Box 7777 Metro Manila, The Philippines

4.1 Performance objectives

4.2 The benchmark site concept

4.3 Site selection

4.4 Site and systems description

4.5 GIS: A tool for benchmark site selection and description

4.6 References

4.1 Performance objectives

Module 4 is intended to enable you to:

1. Discuss the importance of having a benchmark site, and its relationship with a recommendation domain.

2. Identify the criteria used for the selection of a benchmark site.

3. Establish a procedure for site selection based on the goals and objectives of a project.

4. Describe the main purposes of site description.

5. List the most relevant agro-climatic attributes for site selection and description.

6. Identify three common animal production systems in your country, and classify those  according to the livestock classification system shown in Box 4.2.

7. Discuss the relevance of secondary information, participatory rural appraisal and static and dynamic surveys for the collection of socio-economic data needed for site description.

8. Explain why gender issues should be considered in site description reports.

9. Describe how geographic information systems (GIS) can be used for site selection and description.

4.2 The benchmark site concept

The selection of experimental units is a key step in any research activity, since it defines the limits for the inference and subsequent extrapolation of results. The nature of this experimental unit is a function of the level of the agricultural systems hierarchy the researcher wants to impact, and the type of study he/she is involved in. Figure 4.1 illustrates the type of unit of analysis for different studies, going from an enterprise within a farm in case a specific management option is studied (e.g. treatment of rice straw with urea) to a geographic region when studies are conducted at a macro-level (e.g. impact of a given policy facilitates access to smallholder livestock farmers to local markets).

In eco-regional research, researchers have to deal with units of study sufficiently large to measure the impact of interventions on the natural resource base and on the wellbeing of the communities present in a given area. These units are known as benchmark sites. The geographical units selected as benchmarks sites must be representative of a larger geographical entity, the recommendation domain¹, and should be large enough to capture the diversities within such a geographical entity, but small enough to be manageable by the research team. In many cases it has been suggested that an adequate unit for eco-regional research is a watershed, but could be part of it, if it responds to the concepts of representativeness and practicality stated above. Also, in some cases more than one benchmark site may be needed to capture the diversity of interests in a project.

1. A recommendation domain or domain of adaptation is defined as a geographical entity -not necessarily continuous in space- or a group of farmers with similar resources, opportunities and limitations, where the results obtained through a given research effort can be extrapolated. 

Source: Li Pun et al. (1998).

Figure 4.1. Levels in the agricultural systems hierarchy.

A benchmark site can fulfill different purposes. It is the area where eco-regional research is targeted and carried out and therefore is the primary focus for the transfer of those technological interventions identified as promising. However, the benchmark site is supposed to be representative of a recommendation domain, so results obtained there can be extrapolated to analogous sites. A benchmark site can also serve to monitor how exogenous factors (e.g. market opportunities, infrastructure and policies) affect the production and productivity of agricultural systems in a given community or group of communities. Furthermore, a benchmark site is an appropriate unit to evaluate the extent to which a given technological option could affect the environmental quality of a community or a watershed.

The need for a benchmark site to be representative of a recommendation domain is more than just to extrapolate research results. It allows the researches to determine if the response to a given intervention is similar in different sites (e.g. across-site comparisons) within a given recommendation domain, or even in different recommendation domains. In addition, the representativeness of a benchmark site is needed to understand the social and decision-making processes related to technology adoption, the links between technology and human well being, nutrition, health and income.

4.3 Site selection

Site selection is the first step in systems research and this decision has to be carefully taken. Experience has demonstrated that a wrong choice of the project site(s), weak institutional linkages and poorly managed relationships with the local authorities, community leaders and farmers at the beginning of a project, usually lead to failures in the implementation of research and development projects (Villar et al. 1999).

Eco-regional research requires relatively long-term commitment from government agencies and this is costly; therefore the area selected should be a priority for the government, either at national or regional level. Furthermore, for strategic and operational purposes it is important to consider the active presence of R&D institutions in the area, and reasonable infrastructure (Frio and Bhasayavan 1988). The R&D institutions are needed to provide logistic and technical support, and the infrastructure that facilitates transportation of the research team to provide and deliver the results obtained to different stakeholders.

The sites selected should match the goal and objectives of the project, therefore before initiating any selection effort the team has to identify the key attributes that define the system(s) targeted by the project. For example, the goal for an ILRI project entitled `increasing the contribution of livestock to increase productivity of crop–livestock systems in South-East Asia', funded by the Asian Development Bank (ADB), was defined as follows: 'To increase the productivity and economic viability of market-oriented smallholder crop–animal systems in rainfed ecosystems, and protect the natural resource base that supports them.'

In that case, it is clear that the sites to be selected must be representative of the rainfed ecosystems, with a predominance of smallholders practising mixed (crop–animal) farming systems, which can be improved through technological or policy interventions, and with actual or potential easy access to markets.

Within the context defined by the goals and objectives set for a given project, the selection of benchmark sites must consider representativeness of major recommendation domains, in terms of their biophysical attributes and socio-economic conditions. In the first group are included agro-climatic factors, and those characteristics that define the prevalent production systems. Some of the elements considered in each category for the selection of a benchmark site are shown in Table 4.1.

Table 4.1. Some attributes considered for the selection of benchmark sites, farms/benchmark sites and enterprises/farms for crop–animal systems research.

As research is conducted on farm and within this in plots and/or with animals, then the concept of representativeness should be also considered for the selection of these types of units where interventions are tested. Some of the attributes used to define representativeness at the farm and enterprise levels for research activities on crop–animal systems within an eco-regional approach are also listed in Table 4.1. However, representativeness at one level of the hierarchy does not necessarily guarantee the same at a lower hierarchical stage. For example, a farm could be representative of a population of farms for a given type of study, in terms of the type of crop–animal system practiced, but its fields or animals are not automatically representative of the majority of fields or animals present in the benchmark site.

The information needed for site selection and description is similar, and in general terms the methods used for its procurement could also be the same as described for farms or enterprises within farms, but there are differences in the level of detail required for each purpose, being less for site selection. A brief description of the procedures used to obtain the information needed for site selection and description and the attributes used is given here. These will be discussed in more detail in the following sections of this module and as well as in the following modules.

Secondary information (formal publications, grey literature reports, databases, aerial photographs traditional and GIS maps, and other information services), interviews with key informants and the use of participatory rural appraisal (PRA) techniques are the common sources of the information required for site selection. This order should be kept in mind, because frequently the sites have already been characterised, and the communities living there have already been investigated, therefore the team should concentrate on getting the information that is missing after contrasting with the project's objectives. To obtain good quality information from the interviewees, avoid asking questions already asked in other circumstances, unless important changes have occurred.

During the initial stages of site selection, gather information on policies and national programme related to rural development, land tenure, crop and livestock production, natural resource management and conservation, credit and subsidies. In addition, collect information on those policies which could affect either the particular region where the candidate site is located, or nationally (Frio and Bhasayavan 1988). A historical analysis of the systems practised should also be carried out and, if changes occurred, the driving forces behind them should be determined.

Most of the agro-climatic attributes listed in Table 4.1 can be obtained from secondary sources, but some can be confirmed or rectified in situ using key informants and/or through reconnaissance visits which are parts of the participatory rural appraisal techniques. Some of the attributes that characterise the production systems and some of the socio-economic factors included in Table 4.1, can generally be obtained from secondary sources, including census data and GIS maps. However, interviewing key informants or having focus group discussions can help get more detailed and up-dated information.

As an illustration of the information gathered and the rationale used to select benchmark sites, indicates some characteristics for five of the seven sites in the Philippines, analysed as candidates for the above-mentioned project (Villar et al. 1999).

The location of the site with respect to the research team base is also important, because it has implications for the operation of the project, given the costs associated to transportation of the research team. However, if other criteria strongly favour a given site and if accessibility is adequate, distance and costs should not be a problem. For example some of the sites listed in Table 4.2 (e.g. Leyte and Mindanao) are located on islands other than the one where the project personnel was based. Therefore, the research team needed to fly at least at one hour to reach the nearest airport, and travel for a few hours to get to the sites. This was one of the reasons why the two locations were not selected.

All sites are representative of the rainfed ecosystems, which was one of the conditions for the pre-selection of benchmark sites to be evaluated, as defined in the project goal. For eco-regional research it is important to consider how interventions practised in a given site may affect neighbouring sites, or the interactions between agro-ecological zones in a given site. In that respect, those sites including at least two agro-ecological zones are preferred. Except for the site in Mindanao that was strictly rainfed uplands, the rest have two or more agro-ecological zones represented.

Rainfall distribution (expressed as length of the dry season in Table 4.2) influences the type of crop–animal systems practised, not only in terms of the crops grown and cropping patterns, but also the use of crop residues for animal feeding. In general, the dependence of animal on crop residues tends to be stronger in those areas with a longer dry season (Quiroz et al. 1997). The sites in Pangasinan and Batangas have the longest dry season, and therefore the harshest nutritional conditions for the animals.

Information on the crops and animal species raised gives an idea of the type of systems and the type of feed resources that could be derived from the crop component. For example the sites in Leyte and Quezon were representative of the plantation tree crops-ruminant system, whereas in the rest the annual crops-animal system was the most commonly practised. Under the animal species heading there is also a reference to the main purpose of raising animals, which is important for site selection since the project goal states that the focus will be on market-oriented smallholder mixed systems. Information regarding the relative contribution of livestock to income is an indication of this.

Table 4.2. Characteristics of potential benchmark sites for a crop–animal systems project in The Philippines.

* Soil erosion, soil mining (short fallow interval), compaction, deforestation, water pollution etc.

The identification of the main constraints helps establish if the systems could be improved through technological or policy interventions. The best possibilities for short- or medium-term impact can be determined based on the resources available and the duration of the project. For example, an important constraint for the site in Mindanao is the availability of ruminant animals. The solution to this problem is beyond the scope of the project, as it requires the importation of animals to the site, and eventual distribution to the smallholders, under an Animal Dispersal Scheme that operates in the Philippines. Another example is the case of Quezon where farmers face problems in milk marketing, a problem that has to be solved before farmers can adopt technological interventions designed to increase milk production. For the sites in Leyte and Quezon, land tenure is also an important constraint as many of the farmers are tenants. Therefore, it is less likely that farmers will pay attention to those alternatives designed to improve or conserve the natural resource base, unless these have an impact on the availability of portable water or fuel wood or on any other resource that in the short term could benefit the household.

Additional comments related to policy and institutional issues, which may constitute either a constraint or an opportunity, are important as determinants of the selection of a given site. For example, because the sites in Quezon and Batangas are partially or totally included in an industrialisation development plan, opportunities for adoption of technology may be limited or at least the impacts will be short-lived. Furthermore, if there is available technology that has not been transferred, although it has proved valuable, this indicates that there are institutional weaknesses in the link between research and extension or in either one of the processes, which may limit the potential success of the project. The latter was one of the reasons why the site in Leyte was discarded. The site in Pangasinan has in its favour the importance of livestock at provincial and district level (e.g. animal density) and the existence of well-structured market opportunities for animals.

Once information like the one described in previous paragraphs has been gathered, the research team analyses the available data and decides which site to select. This decision must be taken by consensus. In the project in The Philippines illustrated in Table 4.2, the site chosen was Unmingan District, in the province of Pangasinan.

4.4 Site and systems description

The selection of the site(s) where a project will be implemented requires a certain level of characterisation or description, but once the site(s) has (have) been selected, more in-depth information is needed. The procedures used for gathering data could be the same as those described in previous paragraphs, basically considering a single visit to the site and key informants. However, to get more detailed and reliable information, especially related to variations during the year in the availability and use of resources, and on strategies used by farmers to overcome limitations, the use of the dynamic survey technique is needed. The same is true to obtain a clear and reliable appreciation of the farmers' goals, expectations, and motivations. A proposal of flow of information for site description, indicating the type of mechanisms used for data collection is illustrated in Figure 4.2.

Site description is a multi-purpose activity in farming systems research. It could have some or all the purposes listed below:

• to define the agro-ecological and socio-economic boundaries of the site
• to understand the prevalent socio-cultural processes and norms
• to identify the main components of the most relevant farming systems practised and their interactions
• to define the technological and socio-economic constraints, elucidate their causes and determine the potential of those systems to be improved through research, extension and/or political decisions
• to identify recommendation domains
• to provide information that could be used as a reference for ex ante and ex post impact assessment studies
• to make a preliminary identification of farmers, fields and animals for testing potential interventions
• to identify needs for more in-depth monitoring.

Source: Adopted from Leon-Velarde and Quiroz (1998).

Figure 4.2. Flow of information for site description.

In general terms, the attributes used for site description and for site selection are classified in two main groups: The biophysical and the socio-economic attributes. In turn the biophysical attributes are divided into agro-climatic and production systems characteristics.

4.4.1 Attributes

The agro-climatic attributes used for site description, the procedures used for data collection and the frequency with which these can be obtained are summarised in Table 4.3. Some of the agro-climatic attributes used for that purpose are:

• climatic parameters (includes geographic location and altitude because of their relationship with temperature regimes and averages)
• soil attributes (physical, biological and fertility)
• vegetation cover and land use
• water availability and quality.

Some of these attributes are correlated, but all of them influence crops, animals or both. For example, there is an inverse relationship between elevation and temperature (e.g. mean temperature decreases 0.6°C per 100 m increase in elevation). Also these two influence the type of predominant vegetation, not only in terms of biomass availability, but also its diversity. The same also have implications on the type of crops that can be grown in a given site, as well as the animal species and genotypes that can be raised.

Climate

The climate of a given site or region can be defined by a set of parameters, such as temperature (mean, maximum and minimum), rainfall (total and distribution), radiation and wind velocity.

All these have a direct or indirect influence on the adaptation of crops and animals to a given environment, the type of agricultural systems practised in a given site, and on the potential productivity. Besides the parameters listed above, which can be easily obtained from meteorological databases, it is also important to define the risks and frequency of occurrence of deleterious climatic events (e.g. typhoons and droughts).

Table 4.3. Source of information and frequency of agro-climatic attributes used for site description.

To integrate relevant climatic parameters, efforts have been made to develop several climatic classification systems. Among these, the system proposed by Köppen (1936) has been the most widely used with several modifications (White et al. 1998). Köppen's (1936) system considers the following attributes:

• average temperature in the warmest month
• average temperature in the coldest month
• average thermal amplitude between the warmest and coldest months
• number of months with average temperature exceeding 10°C
• total rainfall
• presence of rains during the winter or summer seasons.

A description of the different categories considered in Köppen's Climatic Classification System can be obtained through the Internet at:

http://www.fao.org/WAICENT/FAOINFO/SUSTDEV/Eldirect/climate/Elsp0002.htm

Readers can also download maps for different regions of the world based on this system through the above mentioned website (an example is given in Figure 4.3).

One of the major criticisms of Köppen's Climatic Classification System is that it is static and descriptive (Le Houérou et al. 1993; White et al. 1998), since it is based exclusively on meteorological data, not relating properly to the needs of crops. Based on this, other agroclimatic classification systems incorporating the concept of length of the growing period (LGP) have been proposed (FAO 1996; Seré et al. 1996).

The LGP concept is an effort to match rainfall and ambient temperature with the conditions required for plants to grow. It assumes that the rainfed available soil moisture has to be at least 50% of the potential evapotranspiration (PET) for crop growth to be achieved, and that the mean daily temperature during the growing period has to exceed 5°C (Seré et al. 1996) or 6.5°C (FAO 1996).

Based on the length of the growing period, four bioclimatic categories have been identified:

• Arid: LGP less than 75 days
• Semi-arid: LGP in the range of 75 to 180 days
• Sub-humid: LGP in the range of 181 to 270 days
• Humid: LGP greater than 270 days.

Figure 4.3. Application of the Koeppen's climate classification on a world-wide basis.

Furthermore, considering that cool temperatures do occur in the tropical highlands, Seré et al. (1996) suggested that the difference between these and temperate regions should be based on the mean monthly temperature corrected to sea level. Based on this criterion, a temperate area is the one where at least one month has a mean temperature below 5°C (corrected to the sea level), whereas in the tropical highlands the equivalent daily mean temperature during the growing period is in the range of 5–20°C.

Some adjustments to the concept have been proposed (e.g. the rainfall can be in the form of snow), but still more are needed especially for tropical conditions. For example, the 5°C-temperature threshold is adequate for those crops growing in temperate areas, and in the tropical highlands, but not for the ones adapted to the tropical lowlands. For example, tropical grasses with the C4 photosynthetic pathway stop growing when average temperature is below 15°C (Pezo et al. 1992), therefore the lowest limiting temperature used in the LGP definition is far below the critical level for many of the most important tropical crops and forages.

Regardless of future refinements on the definition of the LGP, it is a valuable criterion because of its relationship with crop production and distribution (including pasture species), and the consequent implications on the feeding strategies applied. A map illustrating the distribution of the different LGP categories in South and South-East Asia is shown in Figure 4.4. Notice that the subhumid and humid conditions are the most common in South-East Asia.

There have been other efforts to correlate climatic parameters with plant and animal requirements and constraints, such as the Agro-bioclimatic Classification System proposed by Le Houérou et al. (1993) for Africa. The system combines a large number of climatic, biological, agronomic and geographical criteria, leading up to 200 combinations in Africa. This makes the system almost impractical in most situations, especially if the same is going to be extrapolated to other regions of the world, because of the considerable need of resources to evaluate all the criteria used for classification purposes (White et al. 1998).

Figure 4.4. Map of South-East and South Asia illustrating zones based on the length of the growing period.

Soils and water

Soil characteristics and water availability influence on the type of crops and forages that can be grown at a given site, and their potential (yields), and indirectly animal productivity. Although detailed soil analysis is not required for site selection, the use of available information on soil classification and land capability is recommended. Once the site has been selected, more detailed information on soils will be needed especially for the design of potential interventions.

Figure 4.5. Application of the FAO-UNESCO soil classification on a world-wide basis.

Many soil classification systems have been proposed, but the most widely used are the USDA (United States Department of Agriculturre) Soil Taxonomy (Buol et al. 1973) and the FAO-UNESCO Soil Map of the World. Several authors (Sanchez 1976; Landon 1984) have cited similarities between both systems and with others. Maps showing the world-wide distribution of the different FAO-UNESCO Soil Classes (Figure 4.5) are now available on CD-ROM², but also can be reviewed on the Internet at:

http://www.fao.org/ag/AGL/agll/wrb/wrbmaps/htm/domsoi/htm

2.The CD-ROM can be bought locally through FAO sales agents or directly from: Sales and Marketing Group Information Division, FAO, Via delle Terme di Caracalla, 00100 Rome, Italy. Fax: +39 06 5705 3360 or Publications-sales@FAO.ORG.

Usually those maps are at a broader level of resolution and therefore it is difficult to properly identify the boundaries of a given site, making those of limited application for benchmark site selection. Consequently, if detailed information on soil classification is going to be used, it is necessary to interact with qualified informants with a soil science background, and even better if they are part of the multi-disciplinary team. Moreover, the soil classification systems used in a given country may be different to the ones described, but a local soil scientist could help to clarify equivalents between the system commonly used in the country and those internationally applied.

Source: Adapted from Sanchez (1976).

Figure 4.6. Soil associations along the slope and base of a volcano in Indonesia.

In a watershed that is a potential candidate to become a benchmark site, different soil associations can be observed in the landscape, and these aspects and a soil scientist can more properly manage their implications in an eco-regional approach. For example, Figure 4.6 shows the case of a site in Indonesia influenced by volcanic activity. Andosols (Andepts), soils formed by deposition of volcanic ashes, dominate the upper parts of the watershed, since high rainfall and low temperature limit weathering of volcanic ashes. In contrast, at intermediate elevations Inceptisols (Tropepts) are present because the predominant environmental conditions favour some degree of soil weathering. However, in the lower part of the watershed highly weathered Ultisols (Humults and Ustults) are found with Vertisols (Usterts) located in a land depression.

Other criteria to be considered for site selection, but especially for site description, are the ones used as a basis of the USDA Land Capability Classification System (Landon 1984), which are described in Box 4.1. The system has been designed primarily for soil conservation to determine the maximum intensity of land use consistent with low erosion risks and sustained productivity. The system recognises eight classes (Figure 4.7); the first four comprise land that is suitable for cultivation, varying from the one without restrictions (Class I) to that with very severe limitations that restricts the range of crops or require special conservation practices or both (Class IV). The following three classes are suited for grazing, from intense (Class V) to limited (Class VII), whereas land of Class VIII is only for conservation purposes (natural forest, wildlife and recreation). For eco-regional research it is important to contrast actual and potential land use and correlate it with the environmental problems detected in a given site.

For water sources, availability and quality, access and uses need to be defined. The seasonal variations of this resource need to be documented (León Velarde and Quiroz 1998), especially in less endowed areas where water is scarce (e.g. subhumid, semi-arid and arid ecosystems). In all cases it is important to differentiate between the water used for household consumption from that used in agriculture.

Note. The intensity with which each land capability class can be used with safety, and the limitations acting, increase as one moves from land capability Class I to Class VIII.

Source: Landon (1984).

Figure 4.7. Relationships between the USDA( United States Department of Agriculture) land capability classes  and potential utilisation of land

4.4.2 Production systems characteristics

An agricultural production system is characterised by its components (e.g. soils, crops, animals, fish pond, home garden and forest), inputs (e.g. fertilisers, medicines and supplements), outputs (milk, meat, manure and draft power) and the interactions and processes that occur within the system, either naturally or as a result of the farmer's management decisions. The functioning of a given system is influenced not only by the biophysical relationships among components, but also by the social, economical, political and cultural conditions under which are practised. Only those attributes of biophysical nature that characterise a production system in this section. Those criteria used for the selection and description at a macro-level (e.g. benchmark site) will be emphasised, recognising that those used for detailed characterisation at the farm/enterprise level are discussed in Module 5.

A first approach to characterising a given site in terms of the prevalent livestock based production system is to identify which are the types most commonly practised and rank them in order of importance. The Livestock Classification Systems proposed by Seré et al. (1996) is recommended for this since the classes included in it (Box 4.2) are comprehensive. Using too many categories may hinder the identification of the predominant livestock based systems. The rationale for classification is that it allows a clear identification of the components present and their relative contribution to the total farm production value. In addition, at the subclass level the bio-climatic element is taken into consideration.

Once the most common livestock based system(s) has (have) been identified, attention should be given to the inflows (e.g. resources and inputs used), the outflows (e.g. products and services derived), and to the most relevant interactions within and between crops, animals and other components (Table 4.4). It must be accepted that the level of detail needed for the description of a site and the production systems practised, as well as the reliability of the information obtained, tend to improve with time as a result of increased confidence of farmers in the research team. Furthermore, the procedures used to get such information allow for a better understanding of the production systems, as well as for verifying whatever was obtained in previous stages (Frio and Bhasayavan 1988).

Table 4.4. Sources of information and frequency of production systems attributes used for site description.

For example, can be relevant for the initial characterisation of the production systems, in terms of available resources and inputs, outputs and flows (to, within and from the system). The static survey will allow the identification of resources, inputs and outputs for each component and the household and the interactions (e.g. complementarity and competition) among components (Table 4.4). The dynamic survey serves to identify the use of inputs, their distribution and the direction of interactions over time.

4.4.3 Socio-economic conditions conditions

The attributes used to define conditions are numerous and diverse; therefore to avoid gathering and storing irrelevant data, with the costs associated with these activities. For that purpose, it is recommended that the researchers review the objectives of the site description stage. Once the benchmark site(s) has (have) been identified and characterised, specific studies can be carried out to gather more information on the systems.

As stated at the beginning of this module, site characterisation is done to define the scenario under which the prevalent systems work, to determine the constraints and elucidate their causes, and to provide information for ex ante and ex post assessments. Using the objectives and hypotheses as the starting point, the team can define which data are needed, the sources and the most appropriate methods for data collection, including the field instruments and analytical procedures (Jabbar et al. 1997). Some of the socio-economic variables needed for site description, procedures that can be used for data collection, and suggestions about the frequency of data collection are summarised in Table 4.5.

Table 4.5. Source of information and frequency of collection of socio-economic attributes 
used for site description.

A review of secondary information and the use of GIS maps can provide information on attributes such as human population and density, per capita gross domestic product (GDP), average farm size, livestock population and density, area dedicated to different crops and their average yields. Frequently, this type of information is only available for the geographic or political division units (e.g. provinces and districts) used in censuses, but not for units that are relevant in eco-regional research (e.g. watersheds). This does not mean that the information described before is useless, but the research team needs to generate additional data according to the definition of the benchmark site, especially from key informants. Secondary information is also the source of data on infrastructure facilities (e.g. roads, irrigation, market, product storage and/or transformation) and institutional matters (e.g. policies, farmers groups, research and development services, and credit), but researchers should complement these sources with the use of PRA, static and even dynamic surveys.

Those attributes related to household composition, its characteristics and activities (e.g. age and level of education of the farmer and the rest of household members, role of each in farm activities, and labour supply and use) can be initially obtained through PRA. But, more reliable information, and/or greater coverage, can be obtained using static and dynamic surveys. In addition, the quality and reliability of quantitative economic information (e.g. costs of inputs, total costs of production, market prices and on- and off-farm per capita income) improve when informants develop more confidence in the research team. Therefore, most of the time, researchers should not look for that type of information in the first visit or in group-meetings. It is better to include those in static or dynamic surveys.

To determine the attributes of the communities and individuals (e.g. beliefs, indigenous knowledge, expectations, attitudes and motives) the research team must get some information before visiting the sites, either through the review of secondary information or interviewing key informants. This could also help define the appropriate means of approaching the target population. The collection of data starts with the application of PRA techniques, but a better understanding of most of these attributes is obtained after the research team and farmers have interacted for some time, as happens with dynamic surveys.

There is often a tendency to interview only male heads of households for site/farm , and this even applies for the whole farming systems research process, ignoring that in many situations women are the household heads with complete responsibility for farm management. Most of the time, in low-income households women are co-breadwinners and participate effectively in the farm decision making process; most frequently, they contribute a larger share of their income than men to satisfy the basic family maintenance (Paris 1995). If women are not involved in the characterisation of sites and farming systems, the understanding of farm activities will be incomplete especially for those components that women manage directly (e.g. the small animal enterprises). Also, a biased appreciation and/or ranking of problems and opportunities will be obtained, as there are gender differences in the use, access to and control of resources and in personal preferences and responsibilities at the farm and household level (Paris 1995).

PRA is the first approach for identifying socio-economic and technological constraints, and to identify opportunities to overcome limitations in the prevalent systems. However, the outcomes of this effort should be considered a starting point for setting a research and development plan with a participatory approach, recognising that the research team will be able to go in-depth and define more precisely a research agenda, after applying the static and dynamic surveys. The research team should function only as facilitators in the identification, classification and prioritisation of constraints, and in the definition of their causes.

As an example of the results obtained in this type of activity, the constraints identified by a research team after interviewing farmers practising mixed (crop–animal) farming, government agricultural technicians and other key informants (Villar et al. 1999) are shown in Box 4.3. Later, based on the primary data obtained from key informants and the secondary data available for the area, the research team constructed a problem tree analysis chart (Figure 4.8). This chart allows the identification of the root causes of those problems and the interventions needed to overcome limitations (Box 4.4). Note that all the interventions identified are not necessarily researchable; many of them may be solved through policies and/or administrative decisions.

4.5 GIS: A tool for benchmark site selection and description

Geographic information systems (GIS) is a computer-based information management technology used for the integration of spatial/geographic data or geographically-referenced data, with non-spatial data or attributes associated with a geographical entity (e.g. population density by districts). The outputs of this tool are either graphic data (e.g. map, graphics or scanned images) or non-graphic data (e.g. tables, lists, numbers or text).

The reader can find in a publication a map showing the distribution of the in South and South-East Asia (e.g. Figure 4.4), which is an output of a GIS work, but has no idea how researchers and technicians produced it. To explain simply how a GIS map is prepared, lets say that a geographic map, an aerial photograph or a satellite image was obtained, but cannot be stored as such in the database. To do so, it needs to be transformed into analog form through encoding, digitising or scanning, and only in this form can it be eventually used in GIS. Those attributes associated to geographical entities are collected from primary or secondary sources, input, organised and also stored in the database. The stored data is then processed and analysed using specific GIS software which overlays two or more sets of data, which could be different maps, or maps with geographically associated attributes.

Source: Villar et al. (1999).

Figure 4.8. Problems tree analysis for the constraints faced by smallholders practising crop–animal systems in rainfed areas of Pangasinan Province, The Philippines.

Many software packages available in the market also allow the preparation of maps with a composed attribute (e.g. index) generated by algebraic operation of data available in different sets. For example, the database may have information on single parameters such as temperature, rainfall and soil texture or based on a thematic property (e.g. soil type, climatic and hydrological conditions), which are used to generate a composed attribute or index. The new composed attribute can be used to prepare a GIS map that shows the distribution of those attributes in different regions (e.g. geographical distribution of agro-ecological zones). Moreover, these data sets could be contrasted with a set of criteria available in the literature and/or provided by a group of experts, and GIS could generate information on the suitability of a geographical entity for a given purpose, let's say for different animal production activities. An example of this is shown in Figure 4.9.

Sources: C.T. Hoan, personal communication.

Figure 4.9. An example of the use of GIS for zonation and land evaluation for animal production purposes.

Source: C.T. Hoan, personal communication.

Figure 4.10. An example of the use of land unit characterisation as a basis for the definition of Recommendation Domains.

The applications of GIS are quite diverse and cover a continuum of opportunities in ecoregional research, from geographic data management to decision support systems. GIS can be used as a tool for inventory purposes, providing easy access to information (in graphic or non-graphic forms) on the biophysical and socio-economic attributes associated with different geographical entities. If the attributes needed for site selection or description are already in the database, GIS outputs save the research team time that would be spent gathering and analysing information available from census date, soil classification studies, meteorological reports etc. Sometimes census information cannot be used directly either because the unit of intervention of the project does not correspond to the geographical unit used in the census, or the details provided are not enough. For example, in census data there is information on the population of cattle, but these are not sorted by purpose (e.g. beef, draft and dairy); therefore, if that classification is needed, then the information available has to be contrasted and complemented with expert knowledge (Gijsman and Kerridge 1998).

GIS can also be used as an analytical tool, combining different attributes and/or contrasting those with a set of criteria based on the project goal and objectives, it can be helpful for site selection and description, generating more integral attributes (e.g. land suitability, market accessibility and land use changes). If appropriate information is available in the database, it can also serve as a basis for defining recommendation domains (Figure 4.10).

The use of GIS as a management tool is beyond site selection and description, but can use the elements of site description to set scenarios for analysing different options. In that case, GIS is complemented with simulation models, expert systems and multiple-goal analysis, and with information on the choices of stockholders. An example of this approach is the work done by the Systems Research Network for Ecoregional Land Use Planning (SysNet Project), led by the International Rice Research Institute (IRRI), which is looking for optimal land use options for different regions in Asia (Roetter and Hoanh 1998). Another effort in this direction is that of the Consortium for Sustainable Development of the Andean Region, led by the International Potato Center (CIP). They are trying to improve the predictive ability of available crop, pasture and animal production models, through linking remote sensing and GIS information with dynamic simulation models. Eventually, it will allow researchers to quantify climatic and production risks associated with the different interventions in the systems they are dealing with (Quiroz et al. 1998).

4.6 References

Buol S.W., Hole F.D. and McCracken R.J. 1973. Soil genesis and classification. Iowa State University Press Ames, Iowa, USA. 360 pp.

FAO (Food and Agriculture Organization of the United Nations). 1996. Agro-ecological zoning-guidelines. FAO Soils Bulletin 73. FAO, Rome, Italy. (http://www.fao.org/docrep/W2962E /W2962E00.htm)

Frio A.L. and Bhasayavan N. 1988. Target area and research site selection and description for crop–animal systems research. In: Proceedings of the crop–animal systems research workshop held in Serdang, Malaysia, 15-–9 August, 1988. IRRI (International Rice Research Institute), Los Baños, The Philippines. pp. 473–514.

Gijsman A. and Kerridge P. 1998. Integrating experiments with agronomic models and geographic information systems to better target research and the extension of research results. In: Thornton P.K. and Odero A.N. (eds), Proceedings of the workshop on eco-regional research at ILRI, Addis Ababa, Ethiopia, 5–8 October 1998. ILRI (International Livestock Research Institute), Nairobi, Kenya. pp. 83–85.

le Houérou H.N. Popov G.F. and See L. 1993. Agro-bioclimatic classification of Africa. FAO Meteorological Series 6. FAO (Food and Agriculture Organization of the United Nations), Rome, Italy. 227 pp.

Jabbar M., Tambi E. and Mullins G. 1997. A methodology for characterizing dairy marketing systems. Market-Oriented Smallholder Dairy Research Working Document 3. ILRI (International Livestock Research Institute), Nairobi, Kenya. 62 pp.

Landon J.R. 1984. Booker tropical soil manual: A handbook for soil survey and agricultural land evaluation in the tropics and subtropics. BAI (Booker Agricultural International), New York, USA. 450 pp.

León Velarde C.U. and Quiroz R.A. 1998. Crop–livestock systems research in the Andean region: Eco-regional approach, methods and procedures. In: Thornton P.K. and Odero A.N. (eds), Proceedings of the workshop on eco-regional research at ILRI, Addis Ababa, Ethiopia, 5–8 October 1998. ILRI (International Livestock Research Institute), Nairobi, Kenya. pp. 27–42.

Li Pun H.H., Jabbar M. and Thornton P.K. 1998. Eco-regional research at ILRI: Background. In: Thornton P.K. and Odero A.N. (eds), Proceedings of the workshop on eco-regional research at ILRI, Addis Ababa, Ethiopia, 5–8 October 1998. ILRI (International Livestock Research Institute), Nairobi, Kenya. pp. 1–15.

Paris T.R. 1995. Gender analysis in crop–animal research. In: Devendra C. and Sevilla C. (eds), Crop–animal interaction. Discussion Paper Series 6, IRRI (International Rice Research Institute), Manila, The Philippines. pp. 501–522.

Pezo D., Romero F. and Ibrahim M. 1992. Producción, manejo y utilización de los pastos tropicales para la producción de leche y carne. In: Fernández-Baca S. (ed), Avances en la producción de leche y carne en el trópico americano. FAO (Food and Agriculture Organization of the United Nations), Oficina Regional para América Latina, Santiago, Chile. pp. 47–98.

Quiroz R., Bowen W. and Gutarra A. 1998. Integrating remote sensing and dynamic models to assess pasture and livestock production at the eco-regional level: Developments in the Altiplano. In: Thornton P.K. and Odero A.N. (eds), Proceedings of the workshop on eco-regional research at ILRI, Addis Ababa, Ethiopia, 5–8 October 1998. ILRI (International Livestock Research Institute), Nairobi, Kenya. pp. 97–103.

Quiroz R.A., Pezo D.A., Rearte D.H. and San Martín F. 1997. Dynamics of feed resources in mixed farming systems of Latin America. In: Renard C (ed), Crop residues in sustainable mixed crop/livestock farming systems. CAB (Commonwealth Agricultural Bureau) International, Wallingford, UK. pp. 149–180.

Roetter R. and Hoanh C.T. 1998. The systems research network for eco-regional land-use planning in tropical Asia: Progress and outlook. In: Proceedings of the Methodological Research at the Eco-regional Level Review Workshop, The Hague, The Netherlands, 20–22 April 1998. ISNAR (International Service for National Agricultural Research), The Hague, The Netherlands. pp. 21–37.

Sanchez P.A. 1976. Properties and management of soils in the tropics. Wiley, New York, USA. 618 pp.

Seré C., Steinfeld H. and Goeneweld J. 1996. World livestock production systems: Current status, issues and trends. FAO Animal Production and Health Paper 127. FAO (Food and Agriculture Organization of the United Nations), Rome, Italy.

Villar E.C., Lanting E.F. and Palacpac-Alo A.M. 1999. Crop–animal systems: The Philippines scenario. In: Devendra, C. (ed), Proceedings of the planning workshop in the crop–animal systems projict, held at IRRI, Los Baños the Philippines, 1–4 June 1999. IRRI (International Rice Research Institute), Los Baños, the Philippines. pp. 44-80.

White D., Zuo H. and Lubulwa G. 1998. Global agro-climatic classifications, with emphasis on Asia. In: Thornton P.K. and Odero A.N. (eds), Proceedings of the workshop on eco-regional research at ILRI, Addis Ababa, Ethiopia 5–8 October 1998. ILRI (International Livestock Research Institute), Nairobi, Kenya. pp. 155–173.