A procedure for range resources assessment: a Swaziland example

J. Sweet

The author can be contacted at his permanent address: 8 Boyne House, 55 Blackwater Rd, Eastbourne BN20 7DL, UK. Tel.: 1323 645993. Fax: 1323 737930.

UNE M�THODE D'�VALUATION DES RESSOURCES EN PARCOURS: L'EXEMPLE DU SWAZILAND

L'auteur d�crit une m�thode d'�valuation des ressources en parcours ainsi que des possibilit�s pastorales et pr�sente une �tude de cas effectu�e au Swaziland. La m�thode, qui est compl�mentaire de celle utilis�e pour une �valuation des sols dans le but de d�terminer les possibilit�s de production v�g�tale, est fond�e sur une stratification primaire en zones agro�cologiques (ZAE) selon la g�omorphologie et le climat, et sur une classification secondaire en unit�s de v�g�tation en fonction de la nature des sols, de la topographie et du couvert v�g�tal. On cartographie les diff�rents modes d'utilisation des sols et de faire-valoir et l'on calcule les superficies dans chaque unit� de v�g�tation. Apr�s compilation des donn�es disponibles concernant le cheptel, on proc�de � une estimation des taux de charge selon le mode de faire-valoir dans chaque ZAE ou chaque unit� de v�g�tation. Un mod�le local permettant de d�terminer la capacit� de charge en fonction de la pluviosit� est mis au point � partir d'un mod�le empirique, puis modifi� en tenant compte de l'exp�rience acquise au plan local � propos des taux de charge et de l'�tat des parcours. On applique des facteurs de correction refl�tant les conditions locales existantes pour ce qui concerne la texture des sols, la pente, le couvert ligneux et l'�tat des parcours. En dernier lieu, on cr�e une base de donn�es informatis�e sur les ressources en parcours aux fins de stockage, d'analyse, d'extraction et de pr�sentation des informations disponibles.

PROCEDIMIENTO PARA LA EVALUACION DE LOS RECURSOS DE PASTIZALES: EL EJEMPLO DE SWAZILANDIA

Se describe un procedimiento para la evaluaci�n de los recursos de pastizales y el potencial de pastoreo y se presenta el estudio de un caso en Swazilandia. El procedimiento es complementario del utilizado para la evaluaci�n del potencial de producci�n de los cultivos en las tierras, y su fundamento es la estratificaci�n primaria de las zonas agroecol�gicas, basada en la fisiograf�a y el clima, y la clasificaci�n secundaria de las unidades de vegetaci�n, basada en los suelos, la topograf�a y la vegetaci�n. Se cartograf�an los tipos pertinentes de utilizaci�n y tenencia de la tierra y se calculan las zonas de cada unidad de vegetaci�n. Se recopilan los datos sobre la poblaci�n pecuaria y se estima la densidad de pastoreo en funci�n de la tenencia de la tierra en cada zona agroecol�gica o unidad de vegetaci�n. Se expone un modelo de la capacidad de carga local en funci�n de las precipitaciones, obtenido a partir de un modelo emp�rico seleccionado y ajustado de acuerdo con la experiencia local en cuanto a la densidad de pastoreo y las condiciones de los pastizales. Luego se aplican factores de correcci�n para las condiciones predominantes en el lugar en cuanto a la textura del suelo, la pendiente, la cubierta le�osa y las condiciones de los pastos. Se establece una base de datos informatizada sobre los recursos de pastizales para el almacenamiento, an�lisis, b�squeda y presentaci�n de datos.

The effective planning and management of range resources requires a sound information base, which should include:

• environmental data - physiography, climate and vegetation;
• current usage - land use, land tenure, livestock numbers and livestock distributions;
• present status - range condition and soil erosion.

For a number of years there had been growing concern that Swaziland's range resources were becoming increasingly overstocked, resulting in range degradation and soil erosion, particularly in the communal areas. In 1992, the Swaziland Government asked FAO for a survey on and assessment of the status of the nation's range resources, for the formulation of guidelines for reasonable stocking rates in the different ecological zones and for recommendations on grazing management and improvement.
Consequently, a comprehensive range resources survey and assessment was undertaken in 1994 (Sweet and Dlamini, 1994) through the FAO/UNDP Project TCP/SWA/2353. This study was conveniently able to build on the physiographic and climatic characterization work already completed under the FAO/UNDP Project SWA/89/001 for producing guidelines on crop production potentials.
The procedure is described below in general terms and the Swaziland case is used for illustrative purposes. In other situations, some steps might be omitted, if considered unnecessary or if requisite data are unavailable. It is primarily applicable for planning at national, provincial and district levels, which encompass different agroclimatic zones but, once developed from a broad database, it can be applied down to the production unit level. The full procedure does, however, require reference to previously derived relationships between annual forage production and annual rainfall (Rutherford, 1978; Le Houerou, 1984) in comparable vegetation types as well as to an adequate information base for physiographic and climatic (rainfall) zoning and to some reliable stocking rate data, taken from different zones in the study area over a number of years (e.g. from commercial ranches or research stations) and relevant to assessments of range conditions and trends.
The principal stages and steps of the procedure may be listed as follows:

Range and livestock inventory

1. Primary stratification of the area into agro-ecological zones (AEZ) and secondary stratification into agro-ecological units (AEU), based on physiographic and climatic data. Calculation of the areas of each AEZ and AEU.
2. Mapping of the principal forms of current land use, for example by interpretation of satellite imagery and calculation of the area of each land-use type in each AEZ or AEU.
3. Mapping of relevant types of land tenure, such as state land, private land (freehold and/or leasehold) and communal areas. Calculation of the grazing area of each land tenure type in each AEZ or AEU.
4. Mapping and description of vegetation, using the AEU boundaries as vegetation boundaries. First stage classification of major vegetation types and second stage classification of vegetation units with similar inherent grazing potential.
5. Calculation of the total areas of each vegetation unit and of the percentages of area under the different forms of land use and land tenure.

Assessment of grazing potential

6. Compilation and assessment of previous estimates of carrying capacity for the major vegetation types or AEZs.
7. Compilation and analysis of stocking rate data. Determination of livestock distribution and stocking rates by land tenure in AEZs or vegetation types.
8. Referral to empirical models of primary production and carrying capacity derived in similar environments, particularly linear relationships between annual forage production and annual rainfall.
9. Development of a local carrying capacity model based on a selected empirical model and adjusted according to local knowledge of stocking rates and range condition. Calculation of preliminary rainfall-related carrying capacities for each vegetation unit.
10. Formulation of correction factors to make allowance for prevailing site conditions of soil texture, slope, woody cover and range condition.
11. Refinement of the model, after incorporation of correction factors, by comparison with known stocking rates. Calculation of corrected carrying capacity estimates for each vegetation unit.

Establishment of a range resources database

12. Establishment of a computerized range resources database for the storage, analysis, retrieval and presentation of data.

PROCEDURE

Range resource inventory

The first step is to decide on a stratification appropriate to the physiographic, climatic, environmental and/or vegetation data available for the study area. Ideally, physiography (landforms, geology and soils) or readily discernible terrain characteristics (e.g. topography) should be used for primary stratification; then climatic data, notably rainfall, are superimposed to demarcate agro-ecological zones or units, depending on the degree of physiographic subdivision.

• In Swaziland, a country with considerable variation in altitude, the landforms and elevation levels were used as the major elements of physiographic classification, with geology, soils and topography as subordinate factors (Remmelzwaal, 1993). Moisture zones and thermal zones were determined from long-term meteorological data (Van Waveren and Nhlengetfwa, 1992) and related to the physiographic units to produce a map of agro-ecological units (Remmelzwaal and Van Waveren, 1994). Delineated units were digitized and their areas calculated by a geographic information system (GIS). Six physiographic zones and 102 AEUs were identified. The general vegetation types of the physiographic zones were noted and the characteristics of each zone summarized as in Table 1. Agro-ecological zones were broadly coincident with the physiographic zones.

1
Physiographic zones of Swaziland
Zones g�omorphologiques du Swaziland
Zonas fisiogr�fica de Swazilandia

In Step 2, appropriate categories of present land use are decided, for example rangeland grazing, improved pasture, large-scale cropping and small-scale mixed agriculture. These categories are then mapped from available data sources, such as interpretation of satellite imagery or aerial photographs. The boundaries are superimposed on to the AEUs and AEZs to calculate the areas of each type of land use within each ecological unit.

• Nine categories of land use were selected for Swaziland. Much of the grazing land in the communal areas exists in arable/settlement/grazing mosaics which could not be mapped separately at the mapping scale used (1:250 000), so categories with different proportions of grazing land (<25, 25-50, 50-75 and >75 percent) were identified (Table 2).

2
Land-use areas of agro-ecological zones
Diff�rents modes d'utilisation des sols dans les zones agro�cologiques
Tipos de utilizaci�n de la tierra en las zonas agroecol�gicas

¹ TDL = Title Deed Land.
Note: For zone abbreviations, see Table 1.

• urban category included areas of residential, industrial and recreational use as well as major water bodies; it totalled an insignificant 0.7 percent of the whole country and was ignored in subsequent calculations involving land use. The forestry category represented plantation forests only. The wildlife category included wildlife and nature reserves and reserved hunting areas in which domestic stock were prohibited.

In Step 3, the categories of land tenure which are considered to influence range development or grazing management are identified and mapped. This step requires reference material (ideally an existing map of land tenure types) or local knowledge. The pattern of land use determined in Step 2 can also provide a useful indicator of land tenure.
The total area available for grazing in each category of land tenure within each AEZ or AEU is then calculated. At this stage, the estimates of areas available for grazing include relatively inaccessible and unutilized areas such as thick bush, riverine forest and steep slopes, but these factors are taken into account in the delineation of vegetation units and assessment of grazing potentials (Steps 4 and 10).

• In Swaziland, livestock ownership objectives, stocking rates and grazing management practices are strongly related to land tenure. There are two main types of grazing land tenure and usage: Swazi Nation Land (SNL), which is mainly communal grazing, and Title Deed Land (TDL), which is mainly private ranching. An outdated map of land tenure was supplemented by local knowledge and by the patterns of land use determined from satellite imagery in Step 2.
• actual areas of SNL grazing in each mixed use category in Table 2 were estimated by multiplying the area of the mixed category by the mean percentage of grazing in that category. Thus, the total area of the first mixed category was multiplied by 0.125 (the mean of 0 percent and 25 percent), and so on. The products were then summed for estimates of the total areas of SNL grazing land in each AEZ.
• total (grazing + non-grazing) area of SNL was also considered important because the bulk of the non-grazing part is cropland or fallow and available for dry season usage by livestock, and therefore the overall total was considered a more appropriate figure for calculating SNL stocking rates than that of the grazing area alone. The areas and percentages of SNL grazing, TDL grazing and total SNL were summarized as in Table 3.

3
Grazing and cropping areas available to livestock in each agro-ecological zone
Superficies herbag�res et cultiv�es accessibles au b�tail dans chaque zone agro�cologique
Superficie de pastoreo y de cultivo disponible para el ganado en cada zona agroecol�gica

TDL = Title Deed Land; SNL = Swazi Nation Land.
Note: For zone abbreviations, see Table 1.

Step 4 is the mapping and description of the vegetation types. Where available, existing vegetation maps should be used as a starting point but additional fieldwork and, possibly, interpretation of satellite imagery or aerial photos, is likely to be necessary. An appropriate classification system is selected and all AEUs characterized accordingly, so that equivalent AEUs sharing the same vegetation type are classified as the same vegetation unit (VU). If desired, the nomenclature can be modified after fieldwork is complete, to reflect the physiography of the VUs better.
The objective is to distinguish units which are functionally different in respect of their inherent grazing potential rather than their botanical composition only. The vegetation units are based on climatic, topographic and soil characteristics as well as on vegetation, and the same plant species may occur in more than one VU.

• In Swaziland, a total of 22 VUs were mapped and described using a combination of satellite imagery interpretation, reference to previous vegetation mapping and fieldwork. It was not possible to classify the VUs according to characteristic grass or woody species owing to the occurrence of the same species in more than one unit. Therefore, a classification was adopted in which the physiography and topography were incorporated, for example high veld plateau grassland, lower middle veld plains, broad-leaved savannah, etc. Each unit was also ascribed an identification letter and number denoting the physiographic zone and the serial number of the vegetation unit in that zone, for example H1 for the first high veld unit and UM3 for the third upper middle veld unit.
• climatic, topographic and soil-related characteristics of each VU were summarized as illustrated for selected units in Table 4. More detailed descriptions of each unit were prepared, including altitude, topography, soils and climatic zones, characteristic grasses, trees and shrubs and the changes in grass composition which tend to occur with overgrazing. The present status of each unit was described in the context of land use, land tenure and range condition, and the observed trends were recorded. Finally, the grazing and browsing potentials were summarized and stocking rates (with corrections for soils and slopes) were recommended (from Step 11).

4
Summary of vegetation units
R�capitulatif des unit�s de v�g�tation
Resumen de las unidades de vegetaci�n

¹ Soil classification is after Murdoch (1970).

In Step 5, the total areas of each unit and the percentages (or actual areas) under each form of land use are calculated. Estimates of the grazing areas according to land tenure can similarly be derived for each VU.
For the purpose of stocking rate and carrying capacity assessments in communal areas, the postharvest availability of crop and fallow land should be considered. However, the ratio of cultivation area to grazing land is also important in determining whether feed availability is more likely to be a constraint during the growing season (summer) or the dry season (winter).

• The total areas of each VU and the areas under each form of land use were summarized as illustrated in Table 5. For simplicity in stocking rate and carrying capacity assessments, all non-grazing areas of SNL were considered to be crop or fallow land equivalent in winter to grazing land. In Swaziland, stock densities are closely related to arable concentrations. On the basis of a maximum of four months per year of useful access to crop residues, fallow lands and interstitial grassy patches within the cropping areas, it followed that the ratio of summer grazing area to winter feed area should be at least 2:1 (or 66 to 33 percent) in order to avoid the risk of grazing shortages in summer. The percentages of grazing land and arable area for SNL in each VU were determined and the ratios checked as a possible indication of summer feed shortage and consequent overgrazing.

5
Land use in vegetation units
Modes d'utilisation des sols par unit� de v�g�tation
Utilizaci�n de la tierra en las unidades de vegetaci�n

TDL = Title Deed Land.

Assessment of grazing potential

Step 6 is the compilation and assessment of previous estimates of carrying capacity for the study area. These are valuable for helping to refine or validate the carrying capacity model and can come from a variety of sources. There may have been previous studies or recorded estimates. If not, information on acceptable stocking rates in the main physiographic or vegetation zones should be collected from knowledgeable individuals, such as good ranchers.

• In Swaziland, over the past 22 years there had been five reports which included carrying capacity estimates, but only one was based on detailed fieldwork. All the estimates were incomplete for the six AEZs and they varied considerably in their recommendations. In addition to referring to these reports, visits were made to a number of commercial ranches for range condition assessments and discussions on stocking rates. Further discussions were held with other knowledgeable individuals within Swaziland and at pasture research stations in neighbouring South Africa.

Step 7 is the compilation and analysis of all available livestock population data for the current year and for previous years of particular interest for the purpose of building up a detailed picture of livestock distribution and stocking rates and obtaining an indication of population trends. These data will be used for comparison with carrying capacity estimates. The comprehensiveness and locational precision of the population data will determine the maximum level (e.g. AEZ or VU) to which stocking rate estimates can reasonably be applied. Records from freehold/leasehold (private and commercial) enterprises are normally readily obtainable. Dipping records, where they exist, are likely to be the prime source of information on stock numbers and distributions for communal areas.

• Swaziland has approximately 800 registered dip tanks (plunge dips or spray races) at which all cattle in the country are mandatorily dipped at set intervals. All ruminant and equine stock in the areas served by the dip tanks are supposed to be recorded. With the help of the dip tank inspectors, each dip tank in the records was allocated to an AEZ, the maximum locational accuracy possible at the time. Each was also allocated to the appropriate category of tenure.
• for three particular years were selected and entered into a spreadsheet: 1992 (to represent peak livestock numbers before the 1992/93 drought mortality), 1993 (the post-drought remainder) and 1982 (ten years prior to the drought). The numbers of each livestock species were converted into standard livestock units (LU) of 450 kg of liveweight to enable a comparison of overall stocking rates between years and between AEZs. The results of the stocking rate calculations were summarized as illustrated in Table 6.

6
Stocking rates by land tenure in the agro-ecological zones
Taux de charge par mode de faire-valoir dans les diff�rentes zones agro�cologiques
Densidad de pastoreo por tipos de tenencia de la tierra en las zonas agroecol�gicas

TDL = Title Deed Land, grazing area only; SNL = Swazi Nation Land, grazing area only; SNL+ = total SNL area.
Note: For zone abbreviations, see Table 1.

In Step 8, reference is made, through literature and research stations, to previous studies of relationships between annual forage production and annual rainfall in similar agro-ecological or vegetational zones. A number of linear relationships between annual forage production and annual rainfall have been demonstrated, particularly in regions with less than 500 to 600 mm annual rainfall, but at higher rainfalls (>800 to 1 000 mm) there are departures from linearity (Rutherford, 1978). These effects, however, are reduced by the diminishing marginal response in forage production and carrying capacity to successive equal increments of rainfall. For example, a 100 mm increase in annual rainfall from 200 to 300 mm effects a 50 percent increase in forage production, but a change from 1 000 to 1 100 mm effects only a 10 percent increase.
The relationships are influenced by vegetation type, range condition and soil fertility. It is therefore desirable to use more than a single model for a variety of soil and vegetation types but the level of information detail required is seldom available and, in practice, an allowance for soil and vegetation characteristics can be incorporated into the site correction factors (Step 10).
The majority of the relationships are of the form:

y = a + bx

where y is the herbage annual dry matter production (kg/ha), x is the annual rainfall (mm),a and b are constants.

However, they can be simplified to straightforward expressions of kg of annual dry matter production per mm of annual rainfall (the water-use efficiency).

• For Swaziland, extension recommendations from the neighbouring Natal Province of South Africa were used. These were based on annual dry matter production varying from 1 kg/mm for range in critical condition to 5 kg/mm for range in excellent condition, with a median of 3 kg/mm for range in average condition (Camp and Smith, 1991).

In Step 9, a preliminary carrying capacity model is developed from the selected water-use efficiency value. The objective is a direct relationship between annual rainfall and recommended stocking rates (carrying capacity) but the intermediate steps of estimating the annual dry matter intake requirements per LU, the edible fractions and the acceptable plant removal fractions need to be included unless usable direct relationships already exist. These intermediate steps are very much a means to an approximate end and their accuracy is not critical, provided calibration of some of the resultant carrying capacity estimates is possible from stocking rate data. To allow for interannual rainfall variation, the minimum annual rainfall that can be expected with a certain probability (e.g. 70 or 80 percent) can be used in the calculations.

• In Swaziland, the Natal extension recommendations of 3 kg/mm water use efficiency for average rangeland, 3 500 kg annual dry matter intake per LU, and a utilization factor of 70 percent of the rainfall-related annual forage production were used as a starting point. These figures were combined to produce an estimate of carrying capacity (LU per hectare) per mm of rainfall (the stocking rate constant), as shown in the steps below:
• (ha/LU) = 3 500/(0.7 x 3) = 1 666 ha for 1 mm rainfall
• the area required per animal is inversely related to rainfall; hence, to obtain a positive correlation, CC must be expressed in LU/ha.
• (LU/ha) = 1/(3 500/0.7 x 3) = 0.0006 per mm
• was the stocking rate constant. To return to expression of stocking rates and carrying capacities in ha/LU, the reciprocal was used again.
• (ha/LU) = 1/(mm Rainfall x 0.0006)
• was the preliminary model, and the formula was entered into a spreadsheet to calculate the rainfall-related carrying capacity of each VU. A dependable rainfall based on an 80 percent probability was used in the model. This had been determined for each moisture zone (Step 1) and was easily extracted for each VU. Where a VU spanned more than one moisture zone, the full range of dependable rainfalls was used.

Step 10 is the formulation and application of correction factors to make allowances for the influence of prevailing site conditions of soil texture, slope, woody cover and range condition on recommended stocking rates in each VU. They should be additive so that they can be applied separately or in combination. Normally they imply a constraint and are used to increase the area required per animal above the rainfall-related estimate for average conditions. However, upward correction for above-average soil fertility or range condition is also possible. The slopes and soils are permanent features so the appropriate corrections can be applied to the whole of the VU. However, range condition and bush density are normally management-related and the corrections should be applied only to the locations where they are applicable. The site correction factors developed for Swaziland are shown below.

The correction factors were applied by multiplying the number of ha per LU by [1 + the sum of the correction factors]. For example, applying a total correction factor of 0.6 to a standard stocking rate (SR) of 2 ha/LU

Adjusted SR = standard SR x [1 + correction factor]
                    = 2 x 1.6
                    = 3.2 ha/LU

The final step in development of the model is refinement, by minor adjustments of the stocking rate constant, to improve the matching of the model estimates (with correction factors applied) to known appropriate stocking rates or reliable previous carrying capacity estimates in the region. The present procedure of classifying VUs is likely to produce more units than previously used in carrying capacity estimations, hence selective matching is necessary. The product of this step is the working model, which is used for planning purposes. However, the stocking rate constant and any of the correction factors can be modified according to experience to improve the model further.
From the working model, the rainfall-related recommended stocking rates for each VU are listed and the correction factors for slope and soil applied to give the preliminary adjusted stocking rate recommendations. These represent the current best estimate of grazing potential for each unit, assuming reasonable range condition and mod-erate bush density, and can be circulated to all planning sections. For a site-specific application of the estimates, the local site conditions of range condition and woody plant cover must be included. Where a grazing area overlaps two or more VUs, the overall stocking rate (total number of livestock recommended) can be determined from a pro rata allocation. The nature of the VU classification (Table 4) is highly amenable to the selective aggregation of units sharing similar characteristics, should the individual VUs be considered too small for particular planning purposes.

• In Swaziland a stocking rate constant of 0.00065 was found to give the best fit with ground data from different AEZs. For three particular VUs of hillside and escarpment bush, the bush cover was considered to be an inherent feature of the VUs (rather than management-related) and so the bush density correction factors were directly incorporated in the working model. The stocking rate recommendations, with preliminary adjustments applied, were summarized as illustrated in Table 7.

7
Stocking rate (SR) recommendations for each vegetation unit (VU)
Recommandations concernant le taux de charge pour les diff�rentes unit�s de v�g�tation
Recomendaciones sobre la densidad de pastoreo para cada unidad de vegetaci�n

Correction factors for local site conditions of current range condition and bush density must be included for the final adjusted stocking rate recommendations.

Establishment of a range resources database

Step 12 is the establishment of a computerized information system for the storage, analysis, retrieval and presentation of the range resources data. Ideally this should include a GIS for interactive processing of different thematic layers. The combined system should enable easy updating or modification of data as new information becomes available or situations change.
In Swaziland, the range resources data were incorporated into the information system developed by the FAO Land Use Planning Project SWA/89/001, comprising the following components:
A database management system (D Base III+ compatible) for entry, storage, processing and retrieval of data and its presentation in tabular form. The range and livestock databases included:

• allocation of AEUs to VUs;
• characteristic grasses, trees and shrubs of each VU;
• increaser grasses in each VU;
• present range condition assessment in SNL and TDL in each VU;
• present bush density assessment in SNL and TDL in each VU;
• allocation of dip tanks to AEZ and to tenure;
• dip tank data for all domestic livestock for the years 1982, 1992 and 1993, with totals for each species and for total livestock units, according to tenure and AEZ;
• stocking rates by tenure for each main AEZ;
• a rainfall-dependent stocking rate model with correction factors for slopes, soils, woody cover and range condition and stocking rate recommendations for each VU.

A raster-based GIS package, IDRISI, for spatial analysis and map production. The thematic layers relevant to range resources included:

• moisture zones (dependable rainfall and length of growing period);
• thermal zones (mean temperatures);
• physiography (landform, geology, topography, soil characteristics);
• agro-ecological units (combinations of moisture, thermal and physiographical layers);
• vegetation units (combinations of climatic, physiographic and vegetation characteristics);
• land use;
• land tenure.

A typical scene in the highlands of Swaziland
Une sc�ne typique des hautes terres du Swaziland
Escena t�pica en las tierras altas de Swazilandia

A Swazi herder viewing the land while taking care of his animals
Un gardien de troupeau swazi contemple l'horizon en surveillant ses animaux
Un pastor swazi observa las tierras mientras cuida de sus animales

DISCUSSION AND CONCLUSIONS

Range resource assessments tend to be conducted in isolation from land evaluations for other forms of agricultural development, even where grazing areas and arable areas are interspersed. Two consequences of traditional procedures are:

• there is often an inadequate information base for planning when there are changes in land use, for example from grazing to cropping;
• carrying capacity estimates tend to be related to broad vegetation types and to take inadequate account of local topography, soils, range condition and bush density.

With a little foresight and planning, a more systematic procedure could be followed to provide a framework for multipurpose land evaluation. The basic unit appropriate for most purposes of land evaluation is the AEU, which incorporates physiographic and climatic data to the stratification level of local topography and soil types, combined with rainfall and temperature means.
Vegetational characterization (and aggregation) of the AEUs results in VUs, which are the smallest appropriate mapping units portraying similar characteristics of physiography, climate and vegetation and, hence, inherent grazing potential. The VUs are thus the most appropriate units for the estimation of carrying capacities and for range resource inventory. A suggested hierarchical structure for deriving VUs is shown in Figure 1.

1
Suggested hierarchical structure for vegetation units
Structure hi�rarchique sugg�r�e pour les unit�s de v�g�tation
Estructura jer�rquica propuesta para las unidades de vegetaci�n

The procedure outlined in this paper presents a logical framework of land evaluation for rangeland resources, building on the stratification required for assessment of crop production potentials. It is suggested that this procedure could be usefully applied more widely in development projects, particularly in areas of high physiographic and topographic variability.

Bibliography

Camp, K.G.T. & Smith, J.M.B. 1991. Veld management planning in Natal. Extension bulletin. Pietermaritzburg, Natal Region Department of Agricultural Development, South Africa.
Le Houerou, H.N. 1984. Rain use efficiency: a unifying concept in arid land ecology. J. Arid Environ., 7: 213-247.
Murdoch, G. 1970. Soil and land classification in Swaziland. Mbabane, Swaziland Ministry of Agriculture.
Remmelzwaal, A. 1993. Physiographic map of Swaziland. SWA/89/001 Field Document No. 4. Mbabane, FAO/UNDP/Swaziland Ministry of Agriculture and Cooperatives.
Remmelzwaal, A. & Van Waveren, E.J. 1994. Agro-ecological analysis of Swaziland. Part A. Land resources: The agro-ecological map. Mbabane, FAO/UNDP/Swaziland Ministry of Agriculture and Cooperatives.
Rutherford, M.C. 1978. Primary production ecology in southern Africa. In M.J.A. Werger, ed. Biogeography and ecology of southern Africa. Vol. 1. The Hague, W. Junk.
Sweet, R.J. & Dlamini, S. 1994. Range resources and grazing potentials in Swaziland. TCP/SWA/2353 Field Document No. 1. Mbabane, FAO/UNDP/Swaziland Ministry of Agriculture and Cooperatives.
Van Waveren, E.J. & Nhlengetfwa, J.V. 1992. Agroclimatic characterization of Swaziland. SWA/89/001 Field Document No. 2. Mbabane, FAO/UNDP/Swaziland Ministry of Agriculture and Cooperatives.