The Agricultural Economy Model (AEM) forms part of the Terrestrial Environmental System (TES) and computes the regional demand for food and feed crops and timber. Production required is determined by the sum of domestic regional demand and net trade. The output of the AEM, including net trade, is used in the Land-Cover Model (LCM) to simulate the change in land use and land cover in each region required to meet the demand for food, feed and timber. The demand for modern and traditional biofuels (i.e., fuelwood and charcoal) is determined by the TIMER energy model, which is part of the Energy/Industry System (EIS).
Although general statements can be made on food security on the basis of the results of the AEM, a detailed simulation of global food demand and supply is not a primary objective of the AEM (or TES).
The model input and output variables of AEM are listed below.
| Model input (food / feed) | Income per capita |
| Land-use intensities | |
| Maximum availability of land in the region considered | |
| Model input (timber) | Value added for the industrial sector |
| Current forest area in the region considered | |
| Assumptions | Preference levels (i.e., food demand per capita in the case of no constraint on production) |
| Productivity of animals for five animal categories | |
| Slaughtering of animals (i.e., off-take rate, the ratio of slaughtered animals : total population by animal category) | |
| Feed efficiency for pigs and poultry | |
| Composition of the feed (grass and fodder species, food crops, residues) | |
| Fraction of dairy cattle pregnant | |
| Number of horses, mules and asses | |
| Self-sufficiency ratios, which determine food trade | |
| Model output for each region | Demand for basic products (disaggregated to six crop groups) |
| Demand for affluent products (disaggregation to five animal product types and oilcrops is scenario driven) | |
| Demand for feed products (seven food crops, residues, grass and fodder species) | |
| Demand for wood products (pulpwood and particles; saw logs, veneer and other industrial roundwood) |
|
|
|
Two broad categories of food products are defined in IMAGE 2.2:
The core of the AEM consists of regional utility functions; these yield a utility value for a given diet composed of basic and affluent products. The maximum utility is achieved when the demand equals the so-called 'preference level'. Preference levels are the daily per capita consumption in the case of no constraint on production for a given income level.
The overall shape and steepness of the utility function is determined by the values of preference levels and weighting constants indicating the eagerness to consume food products at their preference level. The weighting constants and preference levels have fixed values for each region; this is based on an historical analysis, as described in detail in Strengers (2001). In some scenarios, preference levels decrease after 1995.
The AEM optimizes the utility function in terms of land 'budgets', which are expressed in m2 per capita. The maximum budget is the amount of land needed to produce the preferred diet (i.e., production equals preference levels for both basic and affluent products). The actual budget is a function of income, land-use intensity, income elasticities and the 'half-life' parameter.
Food products are valued by using 'land-use intensities' as surrogates of food prices. Land-use intensities are equal to the amount of land needed to produce 1 Kcal of the product under consideration. The half-life determines (in combination with the income elasticities) the rate at which the actual budget approaches the maximum budget with rising income.
The general behaviour of the utility functions is that:
The land-cover model requires projections for the 13 individual food products to compute the total area of agricultural land needed. The broad categories of basic and affluent products are therefore disaggregated on the basis of fractions for the 13 food products. These fractions are exogenous to the model and can change over time, depending on the scenario considered. For example, additional assumptions are made for oilcrops and milk (included in the group of affluent products) to prevent unrealistic future levels of intake, and for the consumption of pork in North Africa and Middle East, and for beef in South Asia (see scenario assumptions land use).
Net trade is added to the demand for each product to obtain the required production of crops and animal products in each region; this is subsequently used in the land-cover model. In regions where the production capacity is insufficient to meet the domestic demand, import complements the domestic production, while export leads to extra demand for agricultural land in the exporting region.
Trade scenarios are prescribed using regional self-sufficiency ratios (SSR). SSR is the regional production divided by the regional consumption, hence exporting regions have an SSR > 1. The exporting regions drive world trade, which means that the total amount of exported food is computed from the SSR values of exporting regions. Total exports are added to a 'global basket', available to the importing regions. The SSR values of importing regions are used as scaling factors to allocate food from this global basket, so that global exports equal global imports. Future net trade of food is defined by the desired self-sufficiency ratio (DSSR) (see scenario assumptions on land use).
The general rule for cases where the total demand for agricultural products cannot be satisfied due to land scarcity is that the consumption of basic products is increased at the cost of affluent products (see agricultural economy model). Due to the modelling approach, in some situations this rule is not applied, and the human consumption of food crops is lowered. This occurs especially when the export of animal products is high compared to the domestic demand. This leads to a strong demand for feed crops (which may be comparable in size to the human demand for food crops) leading to periodic scarcity of arable land. Since this problem causes fluctuations of only a few percent, we consider the errors acceptable in view of the scale of simulations and the aims of IMAGE 2.2.
|
|
|
In IMAGE 2.2 the following categories of animal feed are distinguished:
The demand for animal feed is computed on the basis of the production of meat and milk. For cattle, the total feed demand is calculated on the basis of the energy requirements for maintenance, obtaining feed, growth, lactation, animal traction and calving. Feed requirements for dairy and non-dairy cattle increase along with increasing animal productivity. For the other animals the total feed requirement is calculated from feed efficiencies, i.e. the amount of feed required to produce 1 kg product.
The composition of the feed depends on the animal category considered. Grazing animals such as cattle, goats and sheep depend mainly on pasture and fodder species, while pigs and poultry rely primarily on crops.
For the historical period the composition of the feed was calibrated against data from the literature for various regions. After 1995 the feed mix is scenario-driven; here, the importance of food crops in the animal diet increases at the cost of pasture and fodder species and crop residues, along with increasing intensity of production on the basis of recent trends observed (Alexandratos, 1995; de Haan et al., 1999; FAO, 1996). A detailed description of the treatment of feed composition can be found in the appendix. The resulting feed requirements can be found in the description on animal husbandry.
|
|
|
The following categories of wood products are distinguished:
The demand for fuel wood and charcoal is assumed to be a fixed fraction of the demand for traditional biofuels, as computed by the energy model (TIMER). Production of the other wood products is calculated on the basis of a statistical relationship (described in detail by Alcamo et al. (1998) with updated parameters based on historical data from FAO (1999)) between wood production, population growth, industrial value added and the availability of forests. This relationship is based on the following assumptions:
Wood volumes are converted to weights on the basis of FAO conversion factors. The treatment of forests after wood extraction is described in the landcover model.
Additional scenario assumptions for the future production of wood products are presented in scenario assumptions on land use.
|
|
|
The basis of the calculation of feed requirements in IMAGE 2.2 is the FAO (1999) data on livestock production, including data for meat, milk and eggs from the various animal categories. In IMAGE 2.2 a description at the level of production systems is made. For reasons of simplification we have grouped a number of animals and products. Sheep and goats are considered as one group of animals, and poultry and eggs are in one group of products. Meat from buffaloes is included in the products of the non-dairy cattle category, and milk from buffaloes is included in the products of the dairy cattle category.
Subsequently the following aspects will be discussed:
The calculation of total feed required in dairy and non-dairy cattle production is described by (Alcamo et al., 1998) on the basis of (EPA, 1994). In this approach the energy requirements were divided into six components: maintenance, collection of feed, growth, lactation, work and calving. With this approach the energy requirement per animal increases along with animal productivity, but the energy needed for the production of 1 kg of meat or milk decreases (Table 1).
|
Table 1. Total feed intake for five animal categories and five world regions for 1970 and 1995. |
|||||
| Animal category (product) | USA | OECD Europe | South America | South Asia | East asia |
|
|
|||||
|
Non-dairy cattle (beef) |
30 | 22.7 | 62.8 | 99 | 327.8 |
|
Dairy cattle (milk) |
1.1 | 1.2 | 2.9 | 2.2 | 2.9 |
|
Pigs (pork) |
8 | 8 | 8 | 8 | 11 |
|
Sheep and goats (mutton and goat meat) |
68.3 | 92 | 152.4 | 89.8 | 246.1 |
|
Poultry (poultry and eggs) |
4 | 4 | 5 | 5 | 5 |
|
|
|||||
|
Non-dairy cattle (beef) |
25.6 | 23.9 | 63.6 | 72 | 59 |
|
Dairy cattle (milk) |
1 | 1 | 2.7 | 1.7 | 1.9 |
|
Pigs (pork) |
6.2 | 6.2 | 7.1 | 7.1 | 7.7 |
|
Sheep and goats (mutton and goat meat) |
61.2 | 78.5 | 159.6 | 67 | 69.2 |
|
Poultry (poultry and eggs) |
3.1 | 3.1 | 4.4 | 4.4 | 4 |
| Units: kg dry matter (DM) per kg of productColor | |||||
The feed requirements of the other animal categories are described on the basis of an extensive literature data. Similar to Alcamo et al. (1998) the feed requirements for sheep and goats are based on the overall technological level of the region, with typical numbers derived from Kassam et al. (1991). For pig and poultry production systems we based our estimates of energy efficiency on data from literature (Bos, 1996; De Wit et al., 1996; Simpson et al., 1994), production systems analysis and regional data. The lowest efficiency of 11 kg dry matter (DM) per kg of pork produced is assumed for China for 1970 based on Simpson et al. (1994), where high feed requirements are based on low productivity per animal and the poor feed quality (Table 1). On the other extreme the highest efficiency is seen in regions with industrial pork production, such as OECD Europe, of about 6 kg of DM per kg of pork produced. Other regions are assigned values between these two extremes on the basis of their meat production per animal and the overall technological level. Along with economic growth these values are assumed to decrease, eventually to the level of high-income countries of about 6.0 kg DM per kg pork. This is assumed to be the optimum energy efficiency with high feed quality. See also scenario assumptions land use.
For poultry production the highest feed requirement for 1970 of 5 kg DM per kg of meat and eggs produced is assumed for regions with low technological level, and higher efficiencies and lower feed requirement (4 kg DM per kg of meat and eggs produced) in regions with high technological levels (Table 1). Along with economic growth these values are assumed to decrease to values of between 3.0 and 3.5 kg DM per kg product, which is assumed to be the optimal efficiency. See also scenario assumptions land use.
|
|
|
For the period 1970-1995 we used the FAO (1999) data on feed use of crops, crop by-products and animal products (mainly milk). A number of assumptions are made concerning the other types of feed used. Grazing animals including cattle, sheep and goats, are partly fed with crops and crop by-products and residues. Pigs and poultry are purely fed with crops, crop by-products and residues. A fraction of the feed stems from grass and fodder crops (grass, fodder and forage crops such as fodder maize and alfalfa), depending on the animal category and region considered.
For sheep and goats we assumed that 95-100% of the intake is from grass and fodder crops. For cattle we used data from Safley et al. (1992) on the fraction of animal waste excreted in pastures and paddocks to roughly estimate the fraction roughage in the diet. For many regions with extensive grazing and very high fraction of grass and fodder species of 85-95% in the diet (Central America, Latin America, Western Africa, Eastern Africa, Southern Africa, and Oceania), and for Asian countries (South Asia, Souteast Asia) these data had to be modified only slightly. For regions with more intensive production systems we used a different approach. Here the estimates of Van Der Hoek and Bouwman (1999) formed a basis. A fraction grass and fodder crops in the diet of 60-80% was assumed for western European countries, Canada, U.S.A., North Africa and Middle East, Eastern Europe and the former U.S.S.R. (Table 2).
|
Table 2. Percentage of grass and fodder species, crops + byproducts and residues in the diet of dairy cattle for five world regions for 1970 and 1995. |
|||||
|
Feed category |
|
|
|
|
|
| 1970 | |||||
|
Grass and fodder crops |
70
|
80
|
83
|
20
|
75
|
|
Crops + byproducts |
21
|
12
|
9
|
6
|
10
|
|
Residues |
9
|
8
|
8
|
74
|
15
|
| 1995 | |||||
|
Grass and fodder crops |
61
|
74
|
88
|
9
|
74
|
|
Crops + byproducts |
27
|
13
|
6
|
6
|
12
|
|
Residues |
12
|
13
|
6
|
85
|
13
|
The difference between the total feed requirement and the consumption of grass and fodder crops consists of crops and animal products, crop residues and other feedstuffs. By trial and error and comparison with literature data we derived the fraction of feed stemming from crop residues for each region. We assumed that in less industrialized countries pigs and poultry consume 50-60% crop residues, and 40-50% crops (Table 3). In industrialized countries pigs and poultry consume 25-35% residues and other feedstuffs, and the complement consists of crops + by-products. Sheep and goats consume 0-5% crops and crop by-products. The demand for crops + by-products and residues for cattle is calculated as the difference between the total consumption of crops + by-products and residues given by FAO (1999) and the demand for pigs, poultry and sheep and goats discussed above. The remainder of the feed demand of cattle is assumed to consist of residues and other feedstuffs.
|
Table 3. Percentage of residues in the diet of the major animal categories for five world regions for 1970 and 1995. |
|||||
|
Animal category |
USA | OECD Europe | South America | South Asia | East Asia |
| 1970 | |||||
|
Non-dairy cattle |
9 | 8 | 6 | 74 | 15 |
|
Dairy cattle |
9 | 8 | 8 | 74 | 15 |
|
Pigs |
25 | 25 | 50 | 60 | 60 |
|
Sheep and goats |
3 | 3 | 0 | 5 | 0 |
|
Poultry |
25 | 25 | 50 | 60 | 60 |
| 1995 | |||||
|
Non-dairy cattle |
13 | 13 | 5 | 85 | 13 |
|
Dairy cattle |
13 | 13 | 6 | 85 | 13 |
|
Pigs |
25 | 25 | 41 | 53 | 55 |
|
Sheep and goats |
3 | 3 | 0 | 5 | 0 |
|
Poultry |
25 | 25 | 41 | 53 | 55 |
|
|
|
After harvesting of agricultural crops a considerable amount of residues remain, both above- and below ground. Aboveground residues have various uses, including use as fuel and feed (straw, stubble), and a part is burnt in the field; the remainder is eventually incorporated in the soil where it is decomposed.
Crop residues have been estimated on the basis of the production data (FAO, 1999) and the crop characteristics presented in Table 4. There is considerable uncertainty involved in these ratios (Smil, 1999), since they vary among cultivars as well as for the same cultivar grown in different environments and under different management regimes. In addition, it appears that harvest indices have increased in recent decades, while total plant production stayed more or less stable. Hence, increasing harvest indices formed one of the major factors causing yield increases (Donald and Hamblin, 1976; Hay, 1995).
We derived residues by applying ratios for straw:harvested parts (Table 4), and below ground parts:harvested parts. These ratios are calculated from the apparent harvest index from FAO (1981). These estimates result in global totals which are consistent with Smil (1999).
|
Table 4. Apparent harvest index1 and some other characteristics of the crop groups. |
|||||
|
Crop group |
|
|
|
|
|
|
Temperate cereals |
0.4 | 0.8 | 0.88 | 0.021 | 0.006 |
|
Rice |
0.45 | 0.8 | 0.88 | 0.014 | 0.006 |
|
Maize |
0.35 | 1 | 0.88 | 0.016 | 0.006 |
|
Tropical cereals |
0.35 | 1 | 0.85 | 0.017 | 0.006 |
|
Pulses |
0.35 | 0.8 | 0.85 | 0.04 | 0.025 |
|
Roots & tubers |
0.55 | 0.5 | 0.2 | 0.016 | 0.005 |
|
Oil crops |
0.35 | 0.8 | 0.92 | 0.033 | 0.025 |
|
Other |
0.25 | 1.8 | 0.25 | 0.02 | 0.007 |
|
1 Apparent harvest index is the economic yield of a crop as a proportion of total plant dry matter including root biomass, see e.g. Hay (1995); Schapaugh and Wilcox (1980); Walker and Fioritto (1984). |
|||||
Specific assumptions were required for agricultural residues. At present, the assumed regional fraction of aboveground residues burnt in the field or used as fuel varies from 5 to 40%. These estimates were based on Smil (1999) and country data, while other estimates were made to be consistent with amounts of residues for energy generation presented by Hall et al. (1994).
In the high-income countries 20-40% of available residues is used for animal feed, while in less industrialized countries and Japan this is 50-60%. One exception is the region South Asia, dominated by India, where the amount of residues required exceeds the available quantity (see below).
|
|
|
The calculated total grass consumption for Europe of about 350 Mton yr-1 in 1990 (Table 5) implies a consumption of close to 6 ton ha-1. This is about 60% of the dry matter production in intensively managed grasslands of 8-10 ton ha-1, which is a realistic estimate. Based on the total energy requirement in 1990 of 600 Mton yr-1, and data on feed use of crops (145 Mton yr-1) and animal products (mainly milk), the use of crop residues must be about 80 Mton yr-1. In addition to the total feed requirement we have data on the use of concentrates for different animals. Our estimate for the use of crops and crop by-products of close to 10% of the dry matter intake compares well with the estimates of Van Der Hoek and Bouwman (1999).
|
Table 5. Use of different animal feedstuffs for five world regions in 1990. The world total is compared with estimates by FAO. |
|||||||
|
|
|
|
|
|
|
|
|
| Crops + by-products total | 156.2 | 51.7 | 145.6 | 29.1 | 140.7 | 887.8 | 1049 |
|
Of which:
|
|||||||
|
84.6 | 27.4 | 38.4 | 23.5 | 22 | 375.9 | 442 |
|
35.3 | 8.1 | 75.9 | 1.2 | 92.7 | 308.9 | 358 |
|
0.3 | 0.0 | 2.5 | 0.0 | 0.0 | 9.0 | 12 |
|
36.1 | 16.2 | 28.9 | 4.4 | 26 | 194 | 237 |
| Animal products | 0.5 | 0.3 | 2.1 | 0.3 | 0.1 | 6.7 | - |
| Residues | 64.3 | 58.4 | 83.5 | 331.3 | 173.6 | 992.9 | - |
| Grass | 222.4 | 632.6 | 362.5 | 139 | 266.6 | 2943.9 | - |
| Total | 443.4 | 742.9 | 593.8 | 499.7 | 581 | 4831.3 | - |
|
Data are in million ton DM per year |
|||||||
For East Asia literature data on use of food crops (140-150 Mton DM yr-1), animal products (negligible), and total demand (553 Mton DM yr-1) (Simpson et al., 1994) closely match our results (Table 5) (respectively 140, 0, and 580 Mton yr-1 for crops, animal products and total demand). According to our calculations the grazing demand is 265 Mton DM yr-1. With the grassland area of 520 Mha the grass consumption per ha is about 0.5 ton ha-1, which is consistent with the results of Van Der Hoek and Bouwman (1999). However, (Simpson et al., 1994) presented a quantity of about 200 Mton DM yr-1 of 'unaccounted for' animal feed. In our calculations this is divided over residues and grass in about equal parts. Concentrates make up about 7% of the diet of ruminants in East Asia, which is close to the 6% estimated by Van Der Hoek and Bouwman (1999).
For South Asia the literature data from Singh et al. (1997) agree with our results. However, as stated above, the use of residues (330 Mton DM in 1990) exceeds the quantity of available aboveground residues (~400 Mton, of which close to 140 Mton is used as fuel or burnt in the field). The fact that the available residues in our calculations are much less than the estimate of Singh et al. (1997) supports our results for a lower contribution of residues and a higher one for grazing. The calculated grass consumption per ha in India is of the same level as that in Europe, which may not be realistic. Apparently, in India there is another source of animal feed not accounted for in both our calculations and Singh et al. (1997). Candidates for such unaccounted feed resources may be scavenging (e.g., roadside grazing) and household and industrial wastes.
For other regions the different parameters were selected on the basis of a grass consumption per ha of grazing land in the order of 0.4-1.0 ton DM ha-1 yr-1, and ratios of residues used : available residues and residues used: feed use of crops in line with the data for China, India and OECD Europe.
The global utilization of crops and crop-by products as concentrates by animal category is consistent with estimates of de Haan et al. (1999) (see Table 5), although our estimates are lower for all animal categories. This is probably the result of differences in definition of 'concentrates' used by de Haan et al. (1999) and 'crops + by-products' used in this study. Part of the residues and grass and forage consumption calculated here is probably included in the definition of concentrates used by de Haan et al. (1999).
The global results also agree with Van Der Hoek and Bouwman (1999). The use of concentrates (in dry matter) for ruminants is about 10%, which is consistent with the 20% of N in concentrates of Van Der Hoek and Bouwman (1999). For pigs and poultry, which consume mostly concentrates, the dry matter and N fraction should be similar (60% in our results, vs. 50% in Van Der Hoek and Bouwman (1999). Our estimate of global annual grass and fodder consumption of ~3000 Mton DM (Table 5) is very close to the estimate of ~3400 Mton DM presented by Van Der Hoek and Bouwman (1999). The calculated global use of crops to feed animals is 30% of global crop production, which is consistent with the conclusions of Smil (1999). Finally, the global use of crop residues as animal feed is about 25% of available residues, also in line with Smil (1999).
|
|
|