Chapter 4 :Growth of Agricultural Production in Developing Countries

Contents - Previous - Next

4.1 Introduction
4.2 Agricultural land and irrigation
4.3 Land-yield combinations, major crops
4.4 Agricultural research and modern varieties
4.5 Fertilizers and plant protection agents
4.6 The livestock production

4.1 Introduction

This chapter discusses the main agronomic factors underlying the projections of production. Section 4.2 presents an evaluation of availabilities of land with agricultural potential, how much of it is used now and how much may be used in 2010, for both rainfed and irrigated crop production. Section 4.3 presents for the major crops the combinations of increases in land use, the cropping intensities and the yields that may be achieved in the future and would be compatible with the projected growth of the crop sector. Section 4.4 discusses the role of modern varieties and related agricultural research for sustaining the growth of yields. Section 4.5 deals with prospects for fertilizers and plant protection. Section 4.6 presents the main livestock sector parameters underlying the projections of production. Several themes related to this chapter, particularly issues of environment and sustainability, are the subject of Chapters 11 to 13. Unfortunately, China could not be dealt with in the same detail as the other developing countries since data on land with crop production potential and on cropping patterns by agroecological land class are missing.' China therefore is covered in the same detail as other countries only in Section 4.6 on livestock production.

4.2 Agricultural land and irrigation

The overall situation

Land currently used in crop production in the developing countries (excluding China) amounts to some 760 million ha, of which 120 million ha are irrigated, including some 35 million ha of arid and hyperarid land made productive through irrigation. These 760 million ha represent only 30 percent of the total land with rainfed crop production potential which is estimated to be 2570 million ha, including the 35 million ha of irrigated hyperarid land. The remaining 1.8 billion ha would therefore seem to provide significant scope for further expansion of agriculture. However, this impression is severely redimensioned if a number of constraints are taken into account, as follows:

1. About 92 percent of the 1.8 billion ha of land with rainfed crop production potential but not yet so used is in sub-Saharan Africa (44 percent) and Latin America/Caribbean (48 percent). At the other extreme, there is little land for agricultural expansion in South Asia and the Near East/North Africa.

2. Two-thirds of the 1.8 billion ha of land not in crop production is concentrated in a small number of countries, e.g. 27 percent is in Brazil, 9 percent in Zaire and another 36 percent in 13 other countries (Indonesia, Sudan, Angola, Central African Republic, Mozambique, Tanzania, Zambia, Argentina, Bolivia, Colombia, Mexico, Peru and Venezuela).

3. A good part of this land "reserve" is under forest (at least 45 percent, but probably much more) or in protected areas and therefore it should not be considered as a reserve readily available for agricultural expansion.

4. A significant part (72 percent, see Table 4.2) of the agricultural land of the two regions (sub-Saharan Africa and Latin America/Caribbean) which have 92 percent of the total "reserve" suffers from soil and terrain constraints. This is a much higher share than encountered in the other regions. Overall, some 50 percent of the 1.8 billion ha land "reserve" is classified in the categories "humid" or "marginally suitable for crop production" (see below). Only 28 percent of land presently in use is in these two categories. It follows that the land in the "reserve" is of generally inferior quality compared with that presently in agricultural use. However, given the widely differing land scarcities among countries, land in agricultural use in any one country can very well be of inferior quality compared to land not-in-use in another country. Therefore, the preceding statement need not apply to any given group of countries considered together. For example, land-in-use in countries X and Y together may be inferior to land in "reserve" if X has good land in "reserve" and Y has poor land-in-use and no "reserve". It is also possible for land-in-use within individual countries to be inferior to that in "reserve" due to health, infrastructural or institutional constraints affecting accessibility of the better quality land in "reserve".

5. Finally, human settlements and infrastructure occupy some of the land with agricultural potential, roughly estimated at some 3 percent of it. This proportion would increase in the future, perhaps to 4 percent by 2010.

It is against this background that the prospects that more land will come into crop production use in the next 20 years must be examined. A process of expansion into new land has characterized the evolution of agriculture in the past and there is no reason to think that it will not be present in the future in the countries in which a combination of potential and need so dictates. The fact that there is little scope for agricultural land expansion in many countries should not lead one to conclude that this applies to the developing countries as a whole. In what follows, an attempt is made to project how much new land may be brought into crop production by the year 2010. Potential and need are the main factors that will determine the rate of expansion. The first step is to estimate the potential. This is done using the geo-referenced agroecological zones (AEZ) data base of FAO.

Estimating the extent of land with crop production potential

For each developing country covered in this study (excluding China) an evaluation was made of the suitability of land for growing 21 crops under rainfed conditions and various levels of technology. The method is described briefly below:

1. The raw material for the evaluation consists of two geo-referenced data-sets: (a) the inventory of soil and land form characteristics from the FAO)

Unesco Soil Map of the World (SMW); and (b) the inventory of climate regimes in which data on temperature, rainfall, relative humidity, wind speed and global radiation are used together with information on evapotranspiration, to characterize the thermal regimes and length of growing periods (LGP), i.e. the length of the period during a year (in number of days) when moisture availability in the soil permits crop growth. The two digitized inventories were overlaid in FAO's Geographic Information System (GIS) to create a land resources inventory composed of thousands of agroecological cells which are pieces of land of varying size with homogeneous soil, land form and climate attributes.

2. Each agroecological cell with given soil, terrain and LGP characteristics was tested on the computer for its suitability for growing each of the 21 crops under three levels of technology. The latter are: low, using no fertilizers, pesticides or improved seeds, equivalent to subsistence farming; intermediate, with some use of fertilizers, pesticides, improved seeds and mechanical tools; high, with full use of all required inputs and management practices as in advanced commercial farming. The resulting yields for each cell, crop and technology alternative were then compared with those obtainable under the same technology and LGP characteristics on land without soil and terrain constraints (termed the maximum constraint-free yield-MCFY).4 Any piece of land (agroecological cell) so tested, or part thereof, is classified as suitable for rainfed crop production if at least one of the crops could be grown under any one of the three technology alternatives with a yield of 20 percent or more of the MCFY for that technology. If more than one crop met this criterion, the amount of land classified as suitable was determined on the basis of the crop which utilized the greatest part of the land in the cell. Any piece of land where none of the 21 crops met this criterion was classified as not suitable (NS) for rainfed crop production. It is noted, however, that land which is classified as NS on the basis of this evaluation is sometimes used for rainfed agriculture in some countries, e.g. where steep land has been terraced or where yields less than 20% of the MCFY are acceptable under the local economic and social conditions. It is for these reasons that the reported land in agricultural use in some countries exceeds the areas evaluated here as having rainfed crop production potential (see country data in Appendix 3).

Table 4.1 Land with rainfed crop production potential, developing countries (excluding China)

3. The land classified as having potential for rainfed crop production is further classified into three suitability classes on the basis of the obtainable yield as a percentage of the MCFY, as follows: very suitable (VS) at least 80 percent; suitable (S) 40 to 80 percent; marginally suitable (MS) 20 to 40 percent. Not suitable (NS) land is that for which obtainable yields are below 20 percent of MCFY for all crops and technology levels.

The estimates of the land with rainfed crop production potential were subsequently aggregated for the purposes of this study into seven rainfed land classes (denoted as land classes AT1 to AT7) defined as in Table 4.1. Data by region are shown in Table 4.4 and the geographical location of the land is shown in the maps in the annex to this chapter.

As was noted above, significant parts of the rainfed land with crop production potential are subject to terrain and soil constraints (Table 4.2). This land is classified as having crop production potential if the constraints are not prohibitive, i.e. so long as it could produce a yield of 20 percent or more of the MCFY for at least one of the crops under any one of three technology variants. The importance of the various constraints differs among regions. For example, in the Near East/North Africa region, rainfed land with agricultural potential is largely located in the mountainous areas, where the precipitation is sufficient for cultivation, though the "steep slopes" constraint affects a relatively high share (24 percent) of the potential agricultural land. This constraint affects also relatively high shares of the land in East Asia and South Asia. In sub-Saharan Africa and Latin America and the Caribbean, 40 45 percent of the land with rainfed crop production potential has soils with low natural fertility; and in sub-Saharan Africa a high share of the agricultural land is subject to the constraint "sandy or stony soils".

Table 4.2 Share of land with terrain and soil constraints in total rainfed land with crop production potential (%)

*Salinity, sodicity and excess of gypsum. Individual constraints are non-additive, i.e. they may overlap.

Competing uses of agricultural land: forest, protected areas, human settlements and infrastructure

As noted earlier, not all the land with crop production potential shown in Table 4.1 is, or should be considered as, available for agricultural expansion. As far as some speculative estimates could be made,7 perhaps some 94 million ha of land of all types are occupied by human settlements and infrastructures in the developing countries (excluding China), i.e. some 33 ha per 1000 persons.

With population growth, more land will be converted to human settlements and infrastructure, though such use will probably increase less rapidly than total population. The rough estimates used here imply conversion of land to such uses in the period to 2010 at the rate of about 21 ha per 1000 persons increase in total population. However, there are wide differences among countries, depending on overall population densities. For example, in Canada, a low population-density country, it is estimated that population growth in the "urban-centred regions" in 1981-86 resulted in land expansion for such purposes at an average rate of 64 ha per 1000 persons increment in the population of these regions, of which 59 percent was land defined as "with prime capability to produce crops". The situation is very different in the land scarce countries. A recent UN report contains the following statement referring to India: "At the beginning of the 1980s it was estimated that between 1980 and 2000 about 600 000 ha of rural land would have to be converted to urban use" (UN/ESCAP, 1993). India's urban population was projected at the time to increase by 170 million. This implies 3.5 ha of rural land conversion per 1000 persons increase in the urban population. This is a very small number. It probably refers to conversion of agricultural land only.

Of the total of 94 million ha of land of all types currently estimated to be under human settlements and infrastructure, about 50 million ha are probably included in the land evaluated in the preceding section as having potential for rainfed crop production but not so used at present (land balance). This is a comparatively small proportion, around 2.8 percent of the total land balance. It may grow to account for some 4 percent of the balance by 2010. However, the proportions are much higher in the land-scarce regions. For example, in South Asia some 45 percent of the land balance is probably occupied by human settlements and it could grow to 66 percent by 2010. Therefore, population growth in the land-scarce regions is a significant factor in reducing the net land area available for agricultural use. The relevant estimates by region are shown in Table 4.3.

As noted in Chapter 2, data on extent of forest cover from the 1990 Forest Resources Assessment-Tropical Countries (FRA90, FAO, 1993; see also Chapter 5) are available for only 69 of the developing countries of this study (see list of countries in Appendix 1). These 69 countries account for the great bulk of forest of the tropical countries, 1690 million ha out of a total of 1756 million ha. They also account for the bulk (1724 million ha) of the 1.8 billion ha of land with rainfed crop production potential not yet in agricultural use (the land balance) in the developing countries (excl. China). The extent of overlap between the forest area and the land balance could not be ascertained because the FRA90 provided the land data only in tabular form by administrative unit. An estimate of the minimum overlap was calculated indirectly by first estimating the area of forest that could exist on the land classified as not suitable for crop production (class NS, as discussed above). If that part of the NS land which was good for forest (916 million ha in the 69 countries) was indeed under forest, then the difference (774 million ha of tropical forest area) must be by definition on the 1.8 billion ha of land with crop production potential. This is an estimate of the minimum overlap. The real overlap is probably much larger. Still, the estimate of the minimum overlap is a useful figure for putting the analysis of competition between agriculture and the forest in perspective. The relevant estimates are shown in Table 4.3.

Table 4.3 Land occupied by human settlements, under forest and in protected areas (million ha)

Note: see Appendix 1 on country classification for a list of countries for which data on forest areas and protected areas are available. The estimation of the overlap between the forest and protected areas with the land with agricultural potential not in agricultural use (land balance) was made only for the countries with data for forest and protected areas.
*NB Areas under human settlements, forests and protected areas can be overlapping.
^†The 69 countries account for 1724 million ha (95%) of the total 1988/90 land balance (1816 million ha).
^‡The 63 countries account for 1675 million ha (92%) of the total 1988/90 land balance (1816 million ha).

Finally, part of the 1.8 billion ha land balance with rainfed crop production potential may not be used for expansion of agriculture because it is in areas legally defined as protected (national parks, conservation forest and wildlife reserves). The relevant data are available for 63 of the developing countries of this study (excl. China). These 63 countries account for 92 percent of the 1.8 million ha land balance. Their protected areas are 385 million ha. The relevant data are geo-referenced and were overlaid on those of the land balance. It results that some 200 million ha of protected areas are located on the land classified as having rainfed crop production potential, covering 12 percent of the land balance of the 63 countries. The relevant data by region are shown in Table 4.3. Agriculture, among other economic activities, is prohibited by law in the protected areas but law enforcement is weak in some countries with the result that some farming activity takes place in them, though the degree of encroachment is not known.

In conclusion, the data and estimates of Table 4.3 give an idea of the main competing, and often overlapping, uses of land with rainfed crop production potential that play a role in limiting the extent to which new land may be brought into cultivation in the future.

Future expansion of land in crop production, rainfed and irrigated

Land in crop production in the developing countries, excl. China, may expand from the 760 million ha in 1988/90 to 850 million ha in 2010, an increase of 90 million ha or about 5 percent of the 1.8 billion ha land balance. The bulk of the increase would be in sub-Saharan Africa (42 million ha or 5 percent of its land balance) and in Latin America/Caribbean (27 million ha or 3 percent of its land balance). The rest of the increase would be mainly in East Asia, and very little of it would be in South Asia and the Near East/North Africa (see Table 4.4).

This rate of expansion of land in crop production has been derived for each of the land classes of Table 4. I taking into account the following factors in each country: (a) the land use data or estimates in the base year 1988/90 (land in crop production and relative cropping intensities, as well as the harvested areas and yields by crop, all by land class, including irrigated land); (b) the production projections for each crop; (c) the likely increases in yields by crop and land class; (d) the possible increase in irrigation which increases yields and cropping intensities on previously rainfed land and brings otherwise unusable hyperarid and arid land into use; (e) the changes in the cropping intensities through which the land use by crop (defined in terms of harvested area) is translated into physical area (hereafter called arable land); and (f) the land balances derived as described in the preceding section. The values for each of these parameters are projected on the basis of, essentially, expert judgement in several rounds of iterations subject to overall accounting consistency in the land accounts and for the levels of production, consumption and trade for each commodity, country and the world as a whole, as explained in Appendix 2.

The configurations of the main parameters underlying the conclusion that arable land in crop production would need to, and could, increase by 90 million ha are given in Tables 4.4 (land balances) and 4.5 (cropping intensities and irrigation). The other main parameter of the projections (growth of yields) is discussed in a later section of this chapter. The following observations may be made:

1. Although the arable land may expand by 90 million ha, the harvested area could increase by 124 million ha because cropping intensities would rise. This rise is projected to be from 79 percent in 1988/90 to 85 percent in 2010 on the average for all land classes (Table 4.5). The trend for the cropping intensities to rise and for fallow periods to become shorter is a well established phenomenon (though there are no systematic comprehensive historical data on this variable), accompanying the process of agricultural intensification and reflecting among other things increases in population densities and the rising share of irrigation in total land use. Cropping intensities have also been rising in rainfed agriculture and are projected to continue to do so at rates differing among regions and land classes. The rise in cropping intensities has been one of the factors responsible for increasing the risk of land degradation and threatening sustainability when not accompanied by technological change to conserve the land, including adequate and balanced use of fertilizers to compensate for soil nutrient removal by crops. It can be expected that this risk will continue to exist because in many cases the socioeconomic conditions would not be favourable for promoting the required technological change to ensure sustainable intensification of land use.

2. Irrigated land in the developing countries may expand by 23 million ha or by 19 percent in net terms, i.e. assuming that losses of existing irrigated land (due to, for example, water shortages or degradation due to salinization) will be compensated, e.g. through rehabilitation or substitution of new areas for the lost ones. It has not been possible to project the rate of irrigated land losses. The few existing historical data on such losses are too uncertain and anecdotal and do not provide a reliable basis for drawing inferences about the future. If it is assumed that 2.5 percent of existing irrigation must be rehabilitated or substituted by new irrigation each year (that is, if the average life of irrigation schemes were 40 years) then the total irrigation investment activity over the period of the study in the developing countries (excl. China) must encompass some 85 million ha, of which over 70 percent would be for rehabilitation or substitution and the balance for net expansion.

3. The projections of irrigation used here reflect a composite of information on existing irrigation expansion plans in the different countries, potentials for expansion and need to increase crop production. The projections include some expansion in informal (community managed) irrigation, which is important in sub-Saharan Africa. Cropping intensities on irrigated land would continue to grow, particularly in the land-scarce regions. This would result in the harvested irrigated area increasing by 45 million ha, compared with the 23 million ha projected for the arable (physical) area in irrigation. The projected increase in arable irrigated land is well below that of the preceding 20 years when it was 40 million ha (Table 4.5). It is even lower when considered in relative terms, with the projected growth rate being 0.8 percent p.a., compared with 2.2 percent in the 1970s and 1.9 percent p.a. in the 1980s. The projected slowdown reflects the increasing scarcity of water resources, the rising costs of irrigation investment and the projected lower rate of agricultural production growth.

4. Account must be taken of the fact that irrigation expansion does not always subtract from the stocks of land which are suitable for rainfed agriculture, e.g. when irrigation brings arid and hyperarid land into agricultural use. These areas are included in the broader estimate of "land with agricultural potential" if they are presently irrigated (see Tables 4.1, 4.4). In some regions and countries, irrigated arid and hyperarid land forms an important part of total land presently in use (about one-fifth in Near East/North Africa). Overall, it adds another 36 million ha to the estimate of land with agricultural potential in developing countries. It is projected that another 2 million ha of this type of land could be irrigated by 2010 (Table 4.5).

5. The projected expansion of irrigation will probably use up a small part of the total existing (but unknown) potential for such expansion. It was not found practicable for this study to attempt to make an estimate of ultimate potential for irrigation expansion similar to that made for the land with rainfed crop production potential. To estimate such potential the existing data on water resources need to be interfaced with those on land characteristics and also socioeconomic factors need to be taken into account. On each of these counts there are numerous practical and conceptual difficulties. The definition of water resources available for irrigation and their quantification are subject to widely differing interpretations. Such resources can be quantified meaningfully only on the basis of physical hydrological units (watersheds or aquifers) which can cut across international boundaries. The water of any given hydrological unit cannot be automatically allocated to a specific area of land, because it can be carried over considerable distances and there may be international agreements required, or in existence, defining the allocation rules. Furthermore, since surface and groundwater are linked, it is difficult to assess these two resources independently. Moreover, groundwater resources are not adequately known in all countries, precluding systematic, reliable assessment. The assessment of available water resources is complicated further by inter-basin transfers, inter-annual variability and the use of fossil water (groundwater resources which are not renewable). Perhaps more important still are the socioeconomic aspects of the problem, because the potential for irrigation expansion can only be defined in terms of its economic and social benefits and costs, including environmental costs.

Projected land expansion in relation to past trends

It is clear from the preceding discussion that the projections of land use were not derived as extrapolations of historical trends. Even if one had wanted to extrapolate these trends (and this is not the case here), the historical time-series data in terms of land classes for each country are not available. Even the existing historical data on total land in crop production use (defined as "arable land and land in permanent crops") do not constitute a sufficiently reliable basis for analysing the historical evolution of this variable in a number of countries. This conclusion is based on a comparison between the total harvested area resulting from the summation of the reported harvested areas by crop and the total arable land reported in the land use statistics. The implicit cropping intensities are often not realistic for the type of land in agricultural use. This is particularly the case with the historical data of some sub-Saharan African countries with large shares of their agricultural land in the semi-arid and marginally suitable categories. Country experts and agronomists recommend that in such cases, if the implicit cropping intensities are unrealistically high, they should be adjusted downwards by increasing the arable land estimates to reflect more closely what they believe is the actual situation.

Adjustments to cropping intensities were made for the base year 1988/90 and they yielded the above-reported estimate of 757 million ha of arable land in crop production use. The unadjusted figure was 669 million ha. For 42 of the 91 countries the adjustment exceeded 20 percent, but these were mainly the smaller countries and accounted for 204 million ha of the reported total of 669 million. After adjustment their arable land estimate for 1988/90 was raised to 308 million ha out of the adjusted total of 757 million ha.

It follows that for this "heavily adjusted" group of countries the historical data are not sufficiently reliable to provide a yardstick against which to compare the projections. But for the remaining 49 countries, whose base year data were adjusted by less than 20 percent or not at all, and which account for about 60 percent of the 757 million ha of 1988/90, the historical evidence is more relevant. For this sub-group, the projected rate of arable land expansion is 0.4 percent p.a. This is the same rate as in 1980-91 and lower than those of the earlier decades (Table 4.6, Row 8). The conclusion is that the growth rate may not continue to fall as in the past, though at 0.4 percent p.a. it will be very low, implying a 10 percent increase over 20 years for this group of countries. This result is compatible with the projections of production and with the prospect that yields would grow at a lower rate than in the past, when their rapid growth made possible high growth rates of production while the expansion rates of arable land were falling.

Table 4.4 Land with crop production potential, land-in-use and land balances, developing countries (excluding China) (million ha)

Table 4.5 Arable land in use cropping intensities and harvested land,* developing countries. excluding China (million ha)

Table 4.6 Arable land-in-use, comparison of projections with the historical data (before and after adjustment)

There is not much more one can say except perhaps that for three out of the five developing regions the projected growth rates of land expansion are lower than those derived from the unadjusted data of the historical period. For South Asia the projected growth rate is not lower than in the 1980s, but remains at only 0.1 percent p.a. This leaves sub-Saharan Africa as the only region for which future agricultural land expansion is projected to be more rapid than in the past, though, as noted, the historical data are not a good guide to what has been happening in the region. Sub-Saharan Africa is also the only region for which a higher growth rate of agriculture is projected compared with the past. Given the unfavourable agroecological conditions prevailing in part of the region for rapid yield increases of many crops, it follows that a higher rate of land expansion than in the past will be required to support a significant acceleration in the growth of production.

Table 4.6 presents an overall picture of the world agricultural land use, including China, the residual (non-study) developing countries and the developed ones. The near constancy of agricultural land use in the developed countries is noted, as are the uncertainties regarding the land use data of China. Assuming some decline of agricultural land use in the developed countries (for which no land projections were made) it can be hypothesized that there will be only modest expansion of land in agricultural use for the world as a whole.

Contents - Previous - Next