Agricultural statistics and environmental issues 1 The article that follows provides an example of how agriculture-related statistics can be used in an integrated fashion to examine developments occurring in subject areas that in a narrow sense may be considered outside the domain of agriculture per se, in this case the environment. In developing projections on the release of atmospheric emissions of ammonia from animal waste and fertilizer use, the author has drawn extensively on the statistical data base of the Food and Agriculture Organization of the United Nations. Currently there is concern that activities related to food production may be inadvertently changing the environment. In intensive production systems which may become increasingly important in developing countries the primary environmental concerns arise from animal excreta. Land scarcity, bought-in feed, manure disposal problems, concerns about water quality, disease and odours, and animal welfare, define the basic parameters of livestock-environment interactions. While excessive fertilizer use presents a problem in some irrigated areas in developing countries, in other regions problems arise from under use of nutrients. To illustrate how statistical information can be used for environmental reconnaissance studies, examples will be given of inventories and scenarios of future atmospheric emissions of ammonia (NH 3 ) associated with animal excreta and the use of synthetic fertilizers. These scenarios are based on long-term projections of agricultural production and land use. Atmospheric ammonia is generated by a variety of sources. Emission from decomposing animal and human excreta, inadvertent losses during production and application of fertilizers and losses from biomass burning make up perhaps half of the global emission of NH 3 (see Table 1). Despite its short residence time of about 10 days, NH 3 is the third most abundant nitrogen gas in the atmosphere (after N 2 and N 2 O), and it is an important atmospheric pollutant with a wide variety of impacts. Most atmospheric NH 3 is returned to the surface by deposition (primarily in rain). The re-deposited NH 3 plays an important role in soil acidification, and in the agricultural and general biospheric N-cycle through its contribution to soil nitrogen (N). In the soil, part of the deposited NH 3 is converted to NO and N 2 O during nitrification and to N 2 O (and N 2 ) by subsequent denitrification (Firestone and Davidson, 1989). Ammonia is also involved in the Earth's radiative balance by its role in aerosol formation and by its transformation to N 2 O in the atmosphere (Dentener and Crutzen, 1994). During past decades the extent of permanent pasture areas underwent a slow decrease in the developing countries. Although the statistical data on permanent pastures may not coincide exactly with grazing land, the decline in pasture area may well reflect the tendency to rely less on grazing and more on fodder crops and feed concentrates, as noted by Alexandratos (1995). The FAO study, "World agriculture: towards 2010," posits increases in developing countries in both population and per capita food consumption, particularly the consumption of livestock products, along with economic development (see Table 2). This expansion in livestock production points to an increased dependency on feed, which, in turn, means rising demand for land to produce feedstuffs, including cereals, starchy feeds such as cassava, and protein-rich feedstuffs such as alfalfa and oilseed cakes. Because about half of the synthetic fertilizers is used on cereals (FAO/IFA/IFDC, 1994), livestock production also induces an increase in 1 Based on the results of the study "Long-term scenarios of livestock-crop-land use interactions for the assessment of environmental indicators in developing countries", FAO, Rome 1995.
demand for fertilizers. A relative increase of pork and poultry production which are more feed intensive than beef production will accelerate this trend. According to the demand and production scenario for meat and cereals for the period 1990-2025, the use of cereals for feed in developing countries will increase from 16% of domestic demand at present to 26% in 2025 (see Table 2). The rate of increase is expected to be faster in the years 2010-2025 than in the period 1990-2010. Emissions of NH 3 from domestic animals are estimated based on several factors: worldwide animal populations; estimates of N excretion; and the type of waste management and associated NH 3 emission rate for the different animal categories. A number of factors determine the amount of N excreted by animals. These include the amount, digestibility and N content of the animal feed, and the retention of N in meat or milk. Generally the retention is 5-20%, the remainder being excreted via dung and urine. The amount and type of feed are the key factors determining N excretion. Total feed use by country for the various commodities as derived indirectly from the country supply/utilization accounts, suggests that in developing countries at present about 1.8 kg of cereals, 1 kg of starchy foods and 0.1 kg of oilseed proteins are used to produce one kg of feed-intensity-weighted livestock product (feed intensities in Table 2, note c). Accurate country data on feed use by animal category are not available, and the composition of the diet is highly variable because different mixes of feed and fodders are substitutable, e.g. of cereals, oilseed cakes, fodder legumes and crop residues. An additional problem is that crops which are not harvested for their grain, such as roots, tubers and fruit, but which are fed to animals as roughage, are not consistently included in the feed statistics. Hence, the available statistical data do not allow the estimation of N excretion rates directly from feed intake. Body weight and milk production largely determine total feed intake (Intergovernmental Panel on Climate Change, IPCC, 1995). Estimates of the current body weights averaged over the total population of animal categories are available for continents (IPCC, 1995). Based on estimated herd compositions, animal weights and milk production, the current N excretion can be deduced (Bouwman et al., 1995). In order to approximate future changes in body weights and excretion, carcass weights were used as an indicator for beef cattle, with a linear increase from current conditions towards the Western European values. For dairy cattle, the annual milk production per head is taken as an indicator for the N excretion, with a similar linear growth towards the European values. For the other animal categories, constant N excretion rates were used, as these rates are not as variable for total regional populations as the N excretion by cattle (Bouwman et al., 1995). Although the assumption that body weight and carcass weight are proportional may not be completely correct for nondairy cattle, the relation used describes the increasing efficiency of N use generally observed as animal productivity increases. The NH 3 volatilization rates depend on the type of animal waste management (see Table 3). The percentage of NH 3 loss from animal excreta in developing countries is very close to that in developed countries (Lee et al., in prep.). This indicates that for large regions of the world the productivity level and the animal waste management have minor influence on the NH 3 losses. Therefore, in the emission scenario the NH 3 loss rates are assumed to be constant in time. Combining the scenario of animal populations with estimates of excretion and emission rates resulted in estimated annual emissions of 15 million tons of NH 3 -N for all developing countries including China and a global emission of 23 million tons of NH 3 -N for 1990 (see Table 3). The increase in animal excretion and associated NH 3 production to 24 million tons of NH 3 -N per year in the developing countries by 2025 is less rapid than the growth of livestock production, due to increasing efficiency of N use. Concentration of livestock production, which may occur along with intensification, may, however, lead to increasing NH 3 emission densities and associated adverse environmental consequences. Reductions of ammonia emission from the animal waste application to arable land can be achieved by incorporation of the waste, thus preventing NH 3 evolution and increasing the N recovery rate.
To develop scenarios of fertilizer use, the concept of fertilizer intensity (analogous to feed intensity) was used. The fertilizer intensity is the fertilizer input expressed as the amount of N + P 2 O 5 + K 2 O as a fraction of total weighted biomass production (Table 4). Fertilizer intensity is not identical to nutrient intensity, because only synthetic fertilizers are considered, and animal excreta, crop residues, N deposition, N fixation and other minor sources of nutrients are omitted. The fertilizer intensity was correlated with the total weighted biomass production from all crops per unit of harvested area. Regression analysis of the 3-year average regional data for 1961/63, 1979/81, and 1989/91 yielded a value of r2 of close to 0.8. The fertilizer intensity varies from one region to another, reflecting differences in the mix of agricultural products and differences in crop production systems (yield levels, incorporation of legumes in rotations, recycling of organic materials, management of animal excreta, etc.). Nevertheless, the set of data shows a satisfactorily coherent pattern, whereby production appears to become more dependent on inputs from synthetic fertilizers at increasing yield levels. In the developed regions, fertilizer intensity has been decreasing since 1980 or so (see Table 4). In North America in particular there has been a strong decrease in fertilizer intensity along with rising crop yields. This may be caused by an increasing proportion of soybeans or other leguminous crops in crop rotations. Because leguminous species fix atmospheric N, the crop following the legume may benefit from the N remaining in residues and soil. The overall result of this development may be a region-wide decrease in synthetic fertilizer intensity. One of the scenarios of fertilizer use is based on the assumption that developing countries will increase their fertilizer intensity to a maximum value equal to the 1990 European average of 0.056 (kg [N + P 2 O 5 + K 2 O] per kg biomass); a different scenario (not presented here) takes the 1990 world average of 0.046 as the maximum fertilizer intensity (see Table 4). Using a ceiling for the fertilizer intensity suggests that in future, fertilizer use efficiency would need to be increased along with crop yields. The regression equation and ceilings for the fertilizer intensity coupled with the scenario of crop production and harvested areas were used to obtain the fertilizer use scenario (see Table 5). The results for the developing countries indicate considerable increases in both total fertilizer use and application rates. This reflects the combination of increasing crop production and rising yields which the scenario assumes (see Table 2 for the cereals yield scenario). According to the scenario estimate for 2025, the average fertilizer application per hectare in developing countries would be slightly higher than the current rate in developed countries (Table 5). On the basis of the fertilizer nutrients scenario and an assumed constant N fraction, the N fertilizer use in the developing countries will increase from the current 41 million tons of N per year (see Table 6) to 77 million tons of N by 2010 and 105 million tons of N by 2025. The estimated global loss of NH 3 -N from current synthetic N fertilizer use is about 9 million ton N per year. Most of this loss occurs in the developing countries (7.5 million ton N per year). It is apparent that future NH 3 emissions depend very much on the type of fertilizer applied. The mix of N fertilizer types may change in the future, so the average NH 3 loss rate may also change as a consequence. Here, it is assumed that all developing countries achieve a reduction of NH 3 losses from the current rate of 18% to 5%, equal to that estimated for 1990 for developed countries (Table 6). The resulting projected annual NH 3 loss from fertilizers in developing countries for 2025 would then be lower than the current losses, despite the increasing N fertilizer use. Acknowledgements The author is grateful for ideas and suggestions on the development of the scenarios by J. Bruinsma (FAO) and on animal production and excretion by M. Sanchez (FAO). Thanks are also due to K. van der Hoek (RIVM, Netherlands) for making available the estimates of animal excretion and ammonia emissions. The author remains responsible for the contents of this paper and the underlying study.
Table 1. Global emissions of atmospheric ammonia (NH 3 ) in million tons NH 3 -N per year. Domestic animals 23 Synthetic fertilizers 9 Undisturbed ecosystems 10 Biomass burning 1 Human excrement 4 Sea surface 13 Coal combustion 2 Automobiles 0.2 Total emission 63 a Adapted from Schlesinger and Hartley (1992). a Global annual deposition (primarily in rain) amounts to about 60 million tons of NH 3 -N per year.
Table 2. Historical statistical 3-year average data and projections for the years 2010 and 2025 of food demand per caput, total domestic demand, self sufficiency ratio and production for total meat and cereals for all developing countries including China. For cereals total per caput demand (including feed and other uses), the use as animal feed, and average yields are also presented. 1961/63 1969/71 1979/81 1989/91 2010 2025 POPULATION 10 6 inhabitants 2116 2583 3228 3998 5619 6802 MEAT Demand per caput (kg/cap.) a 9 11 13 17 25 30 Total demand (10 6 ton) 19 28 43 69 143 204 SSR 105 105 101 101 100 100 Production (10 6 ton) 20 29 44 70 143 203 CEREALS Cereals total (kg/cap.) 171 191 219 237 257 278 Cereals food (kg/cap.) a 133 146 161 168 174 177 Domestic demand (10 6 ton) 361 494 706 940 1443 1874 SSR 97 97 92 92 90 88 Production (10 6 ton) 351 480 648 862 1296 1655 Use as animal feed (10 6 ton) 32 56 107 154 327 490 Use as animal feed (%) b 9 11 15 16 23 26 Feed intensity (kg/kg) c 1.5 1.8 2.1 1.8 1.9 1.9 Yield (ton/ha) 1.0 1.2 1.6 2.0 2.8 3.4 All projections 1990-2010 are updated from the FAO study "World Agriculture: Towards 2010" (Alexandratos, 1995) using the UN (1994) scenario of population. Self-sufficiency ratios were assumed constant in 2010-2025, except for Near East and North Africa, where self-sufficiency ratios for cereals were lowered to prevent the arable and irrigated land areas from exceeding the potential areas. a Projections of food consumption are based on estimated income elasticities of demand adapted from Zuidema et al. (1994) with data from the FAO study "World Agriculture: Towards 2010" (Alexandratos, 1995). b feed use as % of domestic demand. c kg cereals / kg feed intensity weighted livestock production; the latter is computed as 0.3 (beef + mutton) + 0.1 (milk) + 1.0 (pork + poultry meat + eggs). This weighting is required since data on feed use for individual animal categories is not available from statistics.
Table 3. Historical 3-year average and projected ammonia emission from animal excreta for developing regions and the developed countries and projections for 2010 and 2025. Emissions in million tons NH 3 -N per year. 1961/63 1969/71 1979/81 1989/91 2010 2025 East Asia incl. China 1.5 2.5 2.9 4.0 6.6 8.4 South Asia 2.8 3.0 3.5 3.9 5.3 6.1 Near East in Asia 0.5 0.5 0.6 0.6 0.8 1.1 North Africa 0.1 0.2 0.2 0.3 0.4 0.4 Sub-Saharan Africa 1.2 1.5 1.8 2.1 2.7 3.6 Latin America 2.1 2.5 3.3 3.7 4.4 4.4 Developing 8 10 12 15 20 24 Developed 8 World 23 The 1989/91 estimates are based on Lee et al. (in prep). The regional NH 3 emission is the sum of emissions from excreta calculated for dairy cattle, nondairy cattle (= total cattle - dairy cattle), buffaloes, pigs, sheep, goats, poultry, and camels. Emissions are calculated as: number of heads x N excretion per head x emission factor. The emission factor indicates the fraction of N in the excreta that volatilizes as NH 3. Current N excretion is from Bouwman et al. (1995). Future excretion for dairy cattle is estimated from the scenario of milk production per head for dairy cattle and carcass weight (as a correlate for body weight) for nondairy cattle. Excretion for other animals assumed to be constant (see text). The emission factors used are 15-20% for stable and storage conditions, 25% for NH 3 loss for manure application as fertilizer, and 10-15% for grazing conditions. The resulting estimates for the NH 3 losses from animal waste vary from 40% for poultry, 20-30% for cattle, 20% for camels and buffaloes, and 12% for sheep and goats.
Table 4. Historical 3-year average regional overall yields (Y) and fertilizer intensities (FI) for the developing countries and some other regions. '69/71 '79/81 '89/91 All developing incl. China Y a 1.6 2.0 2.5 FI b 0.014 0.028 0.036 Europe c Y 3.1 3.7 4.3 FI 0.070 0.076 0.056 Developed countries Y 2.3 2.7 3.1 FI 0.061 0.069 0.060 World Y 1.8 2.2 2.7 FI 0.035 0.046 0.046 a Overall yield, Y (ton/ha) = P / A, where P = total crop production, presented as the sum of 1.0 x (cereals, incl. unmilled rice) + 0.10 x starchy foods + 1.0 x (vegetable oils + sugar + pulses + other crops) (1000 ton); A = total harvested area (1000 ha). b Fertilizer intensity, FI (ton/ton biomass) = FERT / P; FERT = (ton N + P 2 O 5 + K 2 O). c The estimate of FI for Europe, as calculated from the total fertilizer use, may be incorrect because in this region part of the synthetic fertilizers is applied to grasslands (FAO/IFA/IFDC, 1994).
Table 5. Historical 3-year average synthetic fertilizer use for developing and developed countries and the world, and the projected use for developing countries for the period 2010 and 2025. '69/71 79/81 89/91 2010 2025 Developing countries Total NPK a 14 38 65 123 169 NPK use/ha b 22 57 89 147 191 Developed countries Total NPK 55 76 72 NPK use/ha 134 174 176 World Total NPK 68 114 137 NPK use/ha 65 100 121 a Total NPK = NPK use in million tons N + P 2 O 5 + K 2 O b NPK use/ha = kg N + P 2 O 5 + K 2 O per hectare of harvested land.
Table 6. Synthetic N fertilizer use and NH 3 losses for 1990 and projected loss rates for 2010 and 2025 for developing regions, developed countries and the world. Region N fertilizer NH 3 loss NH 3 loss use in 1990 a 1990 b 1990 2010 c 2025 c (Mton N/year) (%/year) (Mton N/year) East Asia incl. China 23.7 18 4.3 1.9 2.4 South Asia 9.8 21 2.1 1.1 1.6 Near East in Asia 2.4 13 0.3 0.3 0.4 North Africa 1.1 12 0.1 0.1 0.1 Sub-Saharan Africa 0.7 10 0.1 0.1 0.2 Latin America 3.8 15 0.6 0.4 0.5 Developing 41 18 d 7.5 e 3.9 5.3 Developed 34 5 1.7 1.7 World 76 12 9.2 3.8 a Fertilizer use for 1990 by type of N fertilizer from IFA (1994), complemented with FAO data. Emissions are calculated from the scenario of fertilizer presented in Table 5, whereby the fraction N of total NPK fertilizer is the average 1960-1990 level. b Estimates of 1990 NH 3 losses from Asman and Bouwman (1995). c The projections for 2010 and 2025 are based on an assumed NH 3 -N loss of the N fertilizer applied of 5%, equal to the loss rate in developed countries in 1990. d The NH 3 -N loss rates are related to the type of fertilizer and to the climatic conditions. In addition, the NH 3 emission may be higher in wetland rice cultivation than in dryland fields. In developing countries 56% of the N fertilizer used is in the form of urea (IFA, 1994). Asman and Bouwman (1995) indicated that NH 3 losses from urea may be 25% in tropical regions and 15% in temperate climates. Another 21% of the N fertilizer used in developing countries is in the form of ammonium bicarbonate (Asman and Bouwman, 1995), which is a highly volatile compound. Urea is less volatile than ammonium bicarbonate, because in the soil urea is converted to ammonium bicarbonate by the enzyme urease, which takes about 2-3 days. The NH 3 loss from direct use of ammonium bicarbonate as fertilizer may be 30% in the tropics and 20% in temperate zones. Contrary, the NH 3 loss from injected anhydrous ammonia, which is widely used in North America (IFA, 1994), is only 4% (Asman and Bouwman, 1995). e Annual loss of 7.5 million ton N is significant. For example, for developing countries the difference between 18% and 5% N loss (5% is the current loss rate in developed countries) accounts for about 5 million tons N. Such a saving of N fertilizer represents a value of about US_ 2 billion (using the current price of urea of US_ 400/ton N); the saving of 5 million ton N is the equivalent of the N content of 250 million ton rice, about equal to the total Chinese rice consumption in 1990. A.F. Bouwman was a visiting scientist to FAO in 1995-96