Previous Page Table of Contents Next Page


2.1 Objective and Scope

The objective of this study is to develop a classification and characterization of the world's livestock systems enabling detailed studies of livestock environment interactions by livestock systems and by impact domains. In order to put interactions between livestock and the environment in a system, regional and global perspective, livestock production systems must be defined, described and put in a geographic context. This is done by providing quantitative estimates of the importance of each system globally and by region in terms of their resource base, human population affected, livestock numbers and outputs.

More specifically, this study aims at:

  1. Delineating and defining elements of a classification of livestock production systems.

  2. Quantitatively and qualitatively describing each livestock production system in terms of feed and livestock resources; livestock commodities produced; production technology; product use and livestock functions; area covered; geographic locations; and human populations supported.

  3. Providing insights into the importance of livestock systems across world regions and agro-ecological zones and related trends in order to provide orientation to decision-makers involved in livestock development.

Apart from its use for the environmental impact assessment by production system, the results of the classification and subsequent quantification exercise are thought to be of use beyond the immediate purpose as an analytical framework of the livestock-environment study. This paper thus exposes the results of the exercise to a wider public to be used in priority setting and as a basis for a general discussion on livestock development.

The study covers livestock production systems involving the following animal species: cattle, buffalo, sheep, goats, pigs and chicken. By neglecting a series of smaller animal species and game animals, the analysis underestimates the availability of animal protein, particularly in more rurally based developing countries.

In geographic terms, systems are grouped according to the following regions: sub-Saharan Africa (SSA), Asia, Central and South America (CSA), West Asia and North Africa (WANA), Organization for Economic Cooperation and Development (OECD) member countries, excluding Turkey, which was included in WANA, eastern Europe and Commonwealth of Independent States (CIS), other developed countries (Israel and South Africa). The study covers 150 countries comprising over 98 percent of the world production of the commodities concerned. A list of the countries included and their allocation to country groups is presented in Annex 1.

Livestock production systems are considered a subset of the farming systems. A review of the literature (Ruthenberg, 1980; Jahnke, 1982; Humphrey, 1980; De Boer, 1992; Wilson, 1994) revealed that most farming systems classifications are not backed by quantitative criteria, which would enable cases to be clearly allocated to one class. These classifications are closer to typologies. No attempt at developing a classification of world livestock systems by using quantitative statistical methodologies (cluster analysis and related methodologies) could be located in the literature. This probably relates to the lack of appropriate data sets for such approaches at a global scale.

A preliminary classification uses the following criteria in the sequence given: integration with crops, relation to land, agro-ecological zone, intensity of production, and type of product.

Operational considerations related to the number of different systems to be handled throughout the whole Livestock Environment Study led to the decision to limit the classification to ten systems by retaining only the first three classification criteria. The complete classification structure is outlined in Figure 1. In addition, the landless system group was split into landless ruminant and landless monogastric systems, bringing the total number to eleven.

Fig. 1: Classification of world livestock production systems

Fig. 1

Systems' boundaries need to be defined to clearly allocate cases to systems. These boundaries are defined using sources of the dry matter fed, percentages of total value of output and climatic criteria. The resulting classification is presented in Chapter 2.3.

These variables are not reported on a consistent basis for existing production systems which prevents their use for the actual quantification of systems' importance. Instead, this has been done using proxy variables. The quality of these proxies could not be tested empirically by quantitative methods. However, the results were validated to a certain degree by experts, knowledgeable in the different regions and production systems, who checked empirical results.

Efforts to quantify resources involved in different livestock systems at a global and regional level have been very limited in the past. The Winrock study commissioned by TAC (Winrock International, 1992) basically classified selected countries in relation to indicators of the relative importance of crops versus livestock and then linked these systems to the main livestock and crop species. The approach covered only a subset of the developing countries (about 70), excluding a number of large heterogeneous countries such as Brazil, Argentina and Mexico in Latin America.

TAC (1992) has allocated stock numbers and animal production to agro-ecological zones of the developing world. This is based on previous work by FAO on the land resource base of the world (FAO, 1985). Data on arable and irrigated land as well as population by each of the nine agroecozones (AEZs) of the developing world, by country, was used by TAC to estimate relative importance of individual crops and animal products by AEZs as the main input into TAC's research priority setting exercise. Production systems were not considered explicitly by this study.

Given the resources available for this exercise, digitalizing actual livestock distributions on a global scale and overlaying this information with that on land resources available in FAO's Information System for Agricultural Statistics (Agrostat) (REF) was not feasible. Furthermore, such information would not have been sufficient for classifying world livestock populations into usefully defined production systems. This recognition led to following and expanding the approach developed by TAC. The major extensions were to expand from the developing country coverage to a global perspective and to further break down agro-ecological zones into production systems, mainly based on the proportion of land used in each AEZ. In addition, data were updated to the period 1991/93 and a number of productivity parameters were calculated from the data generated.

Figure 2 describes the flow of the data processing required to arrive at the final tables (Annex 3) which describe the resource base, animal numbers, output, selected productivity indicators and geographical distribution for each system.

Fig.2: Data processing for system quantification

Fig. 2

2.2 Approach and Data

The calculations performed involve a series of spreadsheets which sequentially allocate items (e.g. stock numbers of certain species, production levels, etc) available at the national level from FAO/Agrostat to specific cells with defined attributes.

At the first stage, national totals were assigned to one or more of the AEZs of the country in question using decision rules. For land-based production, for example, that is the proportion of arable land in each AEZ and for landless production, the prorating factor became the population in each AEZ, in proportional terms. The world's land surface was classified into ten agro-ecological zones.

The next stage was the overlay with classification criteria defining the attributes of the farming system such as mixed versus grazing, mixed rainfed versus mixed irrigated. This classification was based on decision rules related to the share of arable as compared to grazing land and to the share of irrigated versus non-irrigated arable land.

Data for each item was then aggregated across specified AEZs to arrive at climatically broader and less numerous systems, for example, humid + sub-humid tropics and subtropics. Data on different dimensions of a livestock production system were extracted from the item-oriented spreadsheets to produce the system description spreadsheets.

Potentially, the data structure of the study comprises a maximum of ten AEZs in each country. No country covers all ten AEZs and there are about 224 cells containing information for each item. The main strength of the study is that, in spite of the many decision rules used to allocate values to cells, actual data on land resources, stock numbers reported by the 150 countries, generates a high variability which is then grouped.

The data sources used are listed in detail in Annex 2. The groupings chosen are the result of a series of compromises given the very important changes in the geopolitical structure. To maintain compatibility with TAC, developing countries are grouped in the same way. Developed countries are grouped into OECD member countries, eastern Europe and CIS, and Other Developed Countries. Given the difficulties in obtaining data for many of the new nations, former USSR, and the former Republic of Yugoslavia are kept as geographical units.

The way in which values are allocated to AEZs and systems is a central aspect of the methodology employed, in view of the fact that actual data for each AEZ are not available and national data have to be prorated. Following the reasoning applied by the TAC study (1992), the distribution of ruminant populations are considered to be reasonably closely linked to the land resource base (with the exception of the landless systems) which is constituted of proportions of land use classes in each AEZ. The proxy available is arable land, even though some weighted index based on both arable and grazing land would be preferable. Grazing land information is only available from FAO/Agrostat (1994) at the country level and is thus also allocated to AEZs using the arable land prorating factors.

Since monogastrics are less dependent on the land resource base they are distributed across AEZs using population as the prorating factor. Individual cells of the country by AEZ matrix are then assigned to mixed or grazing systems. This is done using the information on the relation of arable to total farming land (computed as arable plus grazing land). Where arable land comprises more than 15 percent of the total, the system is considered mixed, given the importance that crop residues and by-products can be expected to play in ruminant nutrition. The potential variability of this index is somewhat limited for the case of countries having more than one AEZ. Since grazing land is prorated using the arable land distribution, by definition all AEZs in the same country will have the same ratio and thus fall into the same category.

Among the cells defined as mixed, the share of irrigated land in total arable land is used as the criterion to separate mixed irrigated from mixed rainfed systems. The threshold level is set at 25 percent. In this case, as both variables are truly exogenous for each cell within the same country, different regions within a country can be assigned to different systems.

Given the intrinsic weakness of the procedure of allocation to systems, in the case of large countries with diverse ecologies, subnational statistics were consulted to manually allocate data of each administrative unit to a production system. This type of analysis could only be done for the major countries: China and India in Asia, Nigeria, Sudan, Ethiopia in sub-Saharan Africa, Brazil, Mexico and the United States in the Americas. These countries represent a very substantial share of the world livestock economy; they comprise about 40 percent of the world's beef and veal production. This information is superimposed on the automatically generated data.

For the landless monogastric systems, attempts to use feed balances to allocate production to either landless or land-based mixed systems resulted not viable because of inconsistencies and lack of data. Given that large landless systems rely on efficient access to large numbers of consumers, data on urbanization and per caput income are used to estimate the extent to which landless production systems for these species prevail. The rest of the production was considered to be distributed across AEZs in a given country following the population distribution. Within the AEZ the production was allocated to the system predominant in that cell.

The magnitude of the landless monogastrics systems is estimated at the country level in the following way:

  1. For pigs, pig meat production is weighed with the degree of urbanization (expressed as a fraction) multiplied by a factor of 0.5, i.e. 0.5 × urbanization × national output. This factor is raised to 0.7 for countries with per capita incomes above US$ 6,000 per annum.

  2. For poultry meat and eggs, the straight urbanization factor is used to estimate the production volume from landless systems (urbanization × national output). For countries with per capita incomes above US$ 6,000 per annum, the factor is increased by adding 0.5 × (100-urbanization) to the initial urbanization factor. Thus, levels of landless production share are higher than urbanization, while asymptotically reaching 100 percent.

For landless ruminant systems, statistics on the magnitude of landless ruminant production are only available for the United States. For other countries, herd size reports are used for an initial estimate, checked against informal sources of information.

Figure 3 gives an overview of the data flow for the quantification of livestock production systems using cattle numbers as an example.

Productivity indices: All indices are computed at the final stage using regional totals of each variable included in the calculations. They are thus weighted values.

Fig.3: Data Flows for the Quantification of Livestock Production Systems - example: Distribution of Cattle


Animal units are computed assuming that eight head of small ruminants head are equivalent to one animal unit represented by a large ruminant. A study of a global nature must take into account the large differences in mature body weight when computing animal units, using carcass weights as a basis. Weighted means for each of the country groupings are calculated. The largest values are reported for OECD member countries. This value is set to one and values for other regions are calculated as fractions of that. Animal unit numbers by region are then weighted by this adjustment factor. Data in the report thus refer to animal units of the size of the average head of cattle in OECD member countries. This implies that differences in herd structure within species across countries are not taken into account. Empirical evidence suggests that production systems of different intensity levels do not differ too much in herd structure.

To calculate growth rates of stock numbers and production systems over the past decade, the same commodities are used as for the calculations of livestock production systems described above. The values for stock numbers and output figures for livestock production have been extracted from FAO/Agrostat as 1981–83 averages.

As regards the calculation of growth rates for landless production systems, monogastric landless production are calculated based on urbanization rate and GDP per caput for the period 1981–83. Average values for 1980 are taken from the UN-report “World Urbanization Prospects 1990”. (Ref) Since GDP per caput for the base period is only given for the year 1991, figures are taken accordingly for 1981 from the World Bank. The GDP threshold-value for 1991 (US$ 6,000) is deflated to 1981 terms (US$ 4,045) using the deflators given by the International Monetary Fund (1992).

Landless ruminant production is computed for the 1981–83 average by using the annual growth rate of stock numbers or production per country as a measure. The following formulae are utilized.

In case of a negative growth rate over the decade:

Landless 1981/83 = landless 1991/93 ÷ (growth rate ÷ 2 + 1)10

In case of a positive growth rate over the decade:

Landless 1981/83 = landless 1991/93 ÷ (growth rate × 2 + 1)10

This assumes that the growth of the landless sector is twice that of the traditional sectors. The landless stock numbers and outputs were deducted from the total figures to generate data for landbased systems.

Landbased ruminant stocks and outputs are combined with figures of stocks and outputs per AEZs, and prorated according to arable land in each AEZ. These are successively split into data for grazing, mixed rainfed and mixed irrigated systems using the decision rules described above. Adding them gives the LPS-data for 1981–83. Data for landbased monogastric productions are prorated to AEZs using the population distribution as indicator.

For those eight large countries where an overlay is introduced, different data sets are used: (value 1991/93 ÷ value 1981/83)1 ÷ n - 1, n being ten years.

Statistical reports did not present information by production systems, but by commodities, resources, etc. This implied that links, particularly with the land base, crops, etc. had to be undertaken using simple decision rules.

These problems were compounded by the fact that data was for national aggregates but that these masked very important differences within countries. This problem was circumvented by obtaining sub-national data for the major countries and allocating them to ecological zones.

Landless systems presented similar problems as they are not reported separately in most national statistics. Qualified informants were used for landless ruminant systems. A simple mathematical model linking landless pig and poultry production to urbanization and GDP per capita was developed for monogastric systems. Clearly these aspects merit refinement, should better data become available.

2.3 Definitions Used

Decision units: The farm is usually the unit making resource allocation decisions. In certain environments different actors have control over different resources used in the same production process. In these cases the unit of analysis is the group of people making these decisions rather than the individual. Examples are grazing systems with private ownership of livestock and communal grazing or the close interaction between a livestock keeper and an agriculturalist, jointly utilizing land and fodder resources.

Farming systems: Groups of farms which have a similar structure and function and can be expected to produce on similar production functions (Ruthenberg, 1980).

Livestock systems: A subset of the farming systems, including cases in which livestock contribute more than 10 percent to total farm output in value terms or where intermediate contributions such as animal traction or manure represent more than 10 percent of the total value of purchased inputs.

Livestock units (LU): To allow for the calculation of total stocking rates the following conversion factors are used:

1 head of cattle or buffalo1 LU
1 sheep or goat0.125 LU

Given the variability of body sizes of the main animal species across geographical regions, animal units were standardized for comparisons across the world. The weighted average carcass weight of cattle is used as a proxy for animal size. The highest weight, the one found for OECD member countries, was set to one and the factors for other regions computed accordingly.

Actual factors used were:

Sub-Saharan Africa (SSA)0.46
Central and South America (CSA)0.75
West Asia and North Africa (WANA)0.42
OECD member countries1.00
Eastern Europe and CIS0.73
Other Developed Countries0.82

All indices related to animal units refer to temperate animal units, i.e. units which are substantially larger than most tropical ones. This must be taken into account when comparing with results of other studies.

Agro-ecological classification: Based on length of growing period (LGP), which is defined as the period (in days) during the year when rainfed available soil moisture supply is greater than half potential evapotranspiration (PET). It includes the period required to evapotranspire up to 100 mm of available soil moisture stored in the soil profile. It excludes any time interval with daily mean temperatures less than 5°C.

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

Tropical highland areas and temperate regions are defined by their mean monthly temperature.

Temperate: One or more months with monthly mean temperature, corrected to sea level, below 5°C.

Tropical highlands: Tropical areas with daily mean temperature during the growing period in the range 5–20°C.


Solely Livestock Systems (L): Livestock systems in which more than 90 percent of dry matter fed to animals comes from rangelands, pastures, annual forages and purchased feeds and less than 10 percent of the total value of production comes from non-livestock farming activities.

Landless Livestock Production Systems (LL): A subset of the solely livestock systems in which less than 10 percent of the dry matter fed to animals is farm produced and in which annual average stocking rates are above ten livestock units (LU) per hectare of agricultural land. The following additional differentiation is made:

Grassland Based Systems (LG): A subset of solely livestock systems in which more than 10 percent of the dry matter fed to animals is farm produced and in which annual average stocking rates are less than ten LU per hectare of agricultural land.

Mixed Farming Systems (M): Livestock systems in which more than 10 percent of the dry matter fed to animals comes from crop by-products, stubble or more than 10 percent of the total value of production comes from non-livestock farming activities.

Rainfed Mixed Farming Systems (MR): A subset of the mixed systems in which more than 90 percent of the value of non-livestock farm production comes from rainfed land use, including the following classes.

Irrigated Mixed Farming Systems (MI): A subset of the mixed systems in which more than 10 percent of the value of non-livestock farm production comes from irrigated land use, including

Previous Page Top of Page Next Page