Water resources

The section on rainfall (above) showed that the mean annual rainfall of the basin varies considerably (200-1 500 mm) and that the bulk of the basin receives less than 500 mm/year (Figure 7). Rainfall is highly seasonal with 95 percent occurring between October and April. The rainy season is short with the annual number of rain days seldom exceeding 50.

Figure 25 shows the major rivers and streams within the Limpopo River Basin. Table 5 lists the main subcatchments constituting the basin and the area, mean annual precipitation (MAP) and mean annual evaporation (MAE) of each.

Table 5 shows that the Elephants (known as Olifants in South Africa) subcatchment (Figure 25) ranks first in terms of area, covering 17 percent of the basin. It receives the third-highest rainfall and has the third-lowest evaporation. Most of this sub-basin (84 percent) is located in South Africa, including the high-rainfall parts. To the north, in Zimbabwe, there is a short divide with the Zambezi River Basin near Bulawayo, and further east along the watershed, with the Save River.

The southeastern Limpopo River Basin borders the Incomati River (with the upstream Komati and Sabie tributaries), whose basin covers the southern half of the Kruger National Park and the adjacent Nelspruit area. To the south and southwest, the boundary is shared with the watershed of the Orange River, which flows into the Atlantic Ocean, with the Vaal River as the nearest mostnorthern main tributary.

TABLE 5
Rainfall and evaporation figures for major subcatchments (Figure 25)

The western boundary of the Limpopo River Basin borders on the internal drainage system of the central Kalahari Desert and the Okavango Delta. After swinging eastward between Limpopo Province in South Africa and southern Zimbabwe, the Limpopo River receives the Shashe River and flows about 240 km to Mozambique, where it reaches the fall line. In this zone, the river drops about 250 m, with most of the drop concentrated in 43 km of rapids, especially those at Malala, Molukwe and Quiqueque. The Limpopo River is unnavigable until its confluence with the Elephants River, 209 km from the coast in the Indian Ocean. Though partially blocked by a sandbar at its outlet, coastal steamers can enter the river at high tide.

Other tributaries are the Changane River (left bank and downstream of Chokwé), with an area of 43 000 km², and the Lumane River (right bank), with an area of 1 030 km². Both the Changane and Lumane Rivers are entirely in Mozambique, but have very low runoff coefficients and long periods with no discharge at all.

Surface water resources

FAO (1997) has provided an overview of the hydrology of the Limpopo River Basin. The main river can be divided into the following logical reaches:

• Upper Limpopo River, down to the Shashe confluence at the South Africa-Botswana-Zimbabwe border; runoff from South Africa and Botswana.

• Middle Limpopo River, between the Shashe confluence and the Luvuvhu confluence at the South Africa-Zimbabwe-Mozambique border (at Pafuri); runoff from Botswana (Shashe), Zimbabwe and South Africa.

• Lower Limpopo River, downstream of Pafuri to the rivermouth in the Indian Ocean; runoff from Zimbabwe (Mwenezi), South Africa and Mozambique.

Several tributaries originate in Botswana, the most important being the Shashe River, which forms the border between Botswana and Zimbabwe before flowing into the Limpopo River. Other major tributaries, ranked according to decreasing mean annual runoff (MAR), are the Motloutse, Lotsane, Notwane, Bonwapitse and Mahalapswe Rivers. The main tributaries within Zimbabwe are the Shashe and Umzingwane Rivers, with the Mwenezi and Bubi Rivers as other major tributaries. The Mwenezi River originates in Zimbabwe, but joins the Limpopo River in Mozambique.

The most important tributary within South Africa is the Elephants River, which originates between Johannesburg and Witbank and flows into the Limpopo River in Mozambique. Important tributaries of the Elephants River are the Shingwedzi and Letaba Rivers. Major tributaries of the Crocodile-Limpopo part of the Limpopo River Basin, ranked according to decreasing MAR, are: Luvuvhu (which joins the Limpopo River at Pafuri), Mokolo, Mogalakwena, Marico, Lephalala, Nzhelele, Sand and Matlabas Rivers.

The part of the Limpopo River Basin in Mozambique is estimated to contribute 10 percent of the total MAR runoff of the river. The Limpopo River, which was initially a perennial river in Mozambique, can actually fall dry for up to a period of eight months per year, mainly as a consequence of abstractions in the upper catchment area (GOM-DNA, 1995). Downstream of Chokwé, the Changane River (an intermittent tributary) joins the Limpopo River. Although it drains 43 000 km², it has a very low runoff coefficient and periods with no discharge at all. The Lumane River, the last of the most important tributaries, originates in Lake Pave and receives water from the sandy hillsides and, therefore, flows permanently.

Hydrological studies of the Limpopo River Basin

Görgens and Boroto (1999) mention the following hydrological studies of parts of the Limpopo River Basin (year and executing countries in brackets):

• Limpopo Water Utilization Study (1989, Botswana);

• Botswana National Water Master Plan Study (1990, Botswana);

• Joint Upper Limpopo Basin Study (JULBS) (1991, Botswana and South Africa);

• various studies on a number of tributaries (Botswana and South Africa);

• Surface Water Resources of South Africa (South Africa);

• Zimbabwe National Master Plan for Rural Water Supply and Sanitation (1986, Zimbabwe);

• Monografia hidrografica da bacia do rio Limpopo (1996, Mozambique);

• Hydrological Modelling of the Limpopo Main Stem (1999, South Africa).

According to Görgens and Boroto (1999), hydrologically, the JULBS study was particularly significant, as it revealed the existence of significant transmission losses, attributable to alluvial channel and floodplain recharge and channel evaporation, as well as riparian consumptive use by the well-established riparian bush.

Table 6 shows the catchment area of each tributary of the Limpopo River as well as the naturalized MAR and the denaturalized MAR for each tributary, as derived from the above sources. Different studies have employed different hydrological estimation techniques and covered various long-term periods and denaturalization horizons. Thus, the comparison of MARs is merely indicative.

Pallett (1997) has estimated the total natural runoff of the Limpopo River at more than 5 500 million m³. Recent figures for South Africa (GOSA-DWAF, 2003a-d) indicate a figure of about 8 000 million m³. Entering Mozambique, the main river has an average natural MAR of 4 800 million m³ (FAO, 1997). According to Görgens and Boroto (1999), the current MAR at its mouth is about 4 000 million m³, almost 2 000-4000 million m³ less than the estimated natural MAR.

From Table 6, the following observations can be made:

• The Elephants River, which joins the Limpopo in Mozambique, has the largest catchment area and is also the largest contributor of flow to the Limpopo River. The Massingir Dam in Mozambique is located on the Elephants River.

• The Luvuvhu River is the tributary with by far the highest unit runoff and also has a high ratio of denaturalized to naturalized MAR (86 percent), indicating a relatively low level of development in the catchment. The water of the Luvuvhu River also flows directly into Mozambique at Pafuri.

• The Crocodile River, which is the tributary with the second-largest catchment area, has a low ratio of denaturalized to naturalized MAR (43 percent), indicating a high level of development in its catchment.

The naturalized MARs of the Shashe, Umzingwane, Bubi and Mwenezi Rivers are not known. However, the level of development in the Umzingwane is known to be high because a large number of small to large dams have been constructed in its catchment. Some dams have also been built in the Shashe catchment.

TABLE 6
Characteristics of the Limpopo River Basin from upstream to downstream

Notes:

* Denaturalized MAR will change when utilization of the Letsibogo Dam increases.

** According to Görgens and Boroto (1999), the MAR for Zimbabwe is the denaturalized MAR; according to GOZ-MRRWD-DWD (1984), the given MAR is the naturalized MAR.

Sources: Görgens and Boroto (1999); GOSA-DWAF (1991); GOSA-DWAF (2003 a-d); GOB-MMRWA (1992); GOZ-MRRWD-DWD (1984); FAO (1997).

TABLE 7
Characteristics of the Botswana sub-basins of the Limpopo River Basin

Notes:

* 55 = Bonwapitse + Mahalapswe; 195 = Lotsane + Motloutse.

** Denaturalized MAR (111.1) will change when utilization of the Letsibogo Dam increases.

*** According to Görgens and Boroto (1999).

Sources: GOSA-DWAF (1991), GOB-MMRWA (1992), Görgens and Boroto (1999).

Botswana

The Botswana part of the basin feeds into the upper reach of the Limpopo River. It consists mainly of the following sub-basins:

• Notwane,

• Bonwapitse and Mahalapswe,

• Lotsane and Motloutse,

• Shashe (part also in Zimbabwe).

The catchments of some sub-basins are difficult to define, as the topography is very flat towards the Mkgadikgadi Pans and the Central Kalahari. The development of surface water resources is also complicated by the semi-arid environment where potential evapotranspiration is about four times higher than the rainfall and where streamflow records are relatively short and of poor quality. This explains why estimates of catchment areas and related runoff figures differ between one study and the other. Table 7 lists relevant water catchment information from various sources.

With the exception of the Shashe River, most of the available water resources are highly developed. The demand for water is increasing rapidly because of rapid urbanization and industrial development. The total water demand of Botswana was estimated at 193.4 million m³ for 2000. Of this total, 24 percent goes to urban centres, 23 percent to livestock, 18 percent to mining and energy, 15 percent to irrigation and forestry, 11 percent to major villages, 5 percent to rural villages, 3 percent to wildlife and 1 percent to settlements (GOB-MFDP, 1997). Owing to high water demands, most of the subcatchments have a water deficit and rely on water importation and water saving techniques to meet demand. The potential storage capacity of dams in Botswana is estimated to be almost 300 million m³, but this potential is not regularly realized.

The Notwane sub-basin

The Notwane River rises on the edge of the Kalahari sandveldt and flows northeast until it reaches the Limpopo River some 50 km downstream of the confluence of the Limpopo River with the Marico River. About one-third of Botswana’s 1.6 million population reside in the Notwane Basin, which includes the urban centres of Gaborone, Molepolole, Mochudi, Kanye, Lobatse and Jwaneng.

Domestic water needs dominate water use in the Notwane sub-basin and demands are growing rapidly. The large urban centres account for more than 60 percent of the domestic water demands. Gaborone consumes 50 percent of all urban use (i.e. 30 percent of national domestic water use) and this is expected to increase significantly because of rapid urbanization (i.e. up to 40 percent of national domestic demand by 2020).

In order to address inequalities of water abstraction, the South African Department of Water Affairs and Forestry is planning the transfer of 124 million m³ of water per year from the Crocodile West and Marico Water Management Area to Gaborone in Botswana (GOSA-DWAF, 2003c).

The Gaborone and Bakaa Dams are located on the Notwane River and are the main sources of domestic water supply to Gaborone and surroundings. The Nywane Dam serves Lobatse, which is also linked by a pipeline to the Gaborone Dam. The catchment contains about 200 small dams which result in an estimated 25-percent reduction in runoff, serving primarily livestock needs. In most years, the catchment has a water deficit, and water importation from South Africa from another part of the Limpopo catchment (i.e. from the Molatedi Dam on the Marico River) to Gaborone is one of the mitigating measures. The completion of the North-South Water Carrier, which is designed to transport water from the Shashe Dam near Francistown to Gaborone, will bring much needed relief to the stressed local surface water and groundwater resources, and secure water resources up to 2020 (Pallett, 1997).

The Bonwapitse and Mahalapswe sub-basins

The Bonwapitse River has its upper catchment in the Kalahari sandveldt and is seldom in flow. The Mahalapswe River also contributes very little to the flow of the Limpopo River and normally does not have surface runoff during the winter. However, water is stored in the sand bed and this is an important source of domestic water for small communities and their livestock along the river reaches. No large dams have been constructed and no potential exists. Small dams provide suitable water resources for stock watering, small-scale irrigation (horticulture) and in some cases also serve domestic uses of small villages.

The Lotsane and Motloutse sub-basins

Both rivers flow mainly during summer rainfall and have limited development potential owing to the relatively flat terrain and restricted options for the building of larger dams. Development potential is suited primarily for small communities, livestock and small-scale irrigation. Communities along the river reaches often access water stored in the sandy riverbed or small dams and depend on this for most of the winter months. However, investigations for the Botswana National Water Master Plan (BNWMP) have shown that the sand volume of riverbeds is generally limited in depth and, hence, does not support a major abstraction capability (GOB-MMRWA, 1992). A total of 366 small dams have been constructed in Central Region, mostly located in the Lotsane and Motloutse catchments.

The Motloutse River has an MAR of 111 million m³/year. Construction of the Letsibogo Dam is probably the only significant development and will serve primarily the industrial town of Selebi-Pikwe and surrounding local needs, including potential irrigation (Box 8). Its potential contribution to the North-South Water Carrier is uncertain.

The Shashe sub-basin

The Shashe sub-basin shows potential for development. To date, there are 146 small dams in the Francistown region, and the Ministry of Agriculture is investigating various small- to medium-sized irrigation schemes. The existing Shashe Dam serves Francistown and Selebe-Pikwe. During Phase-I of the North-South Water Carrier Project, the Shashe Dam will also transfer water via Selebe-Pikwe to Gaborone, supplementing local water resources of the main towns on-route including Palapye, Mahalapye, Palla Road and Mmamabula. Construction of the Letsibogo Dam on the Motloutse River will relieve the Shashe Dam, to serve primarily Francistown and growing urban and peri-urban areas in the region.

Further phases of this project will construct and link the Lower Shashe Dam and duplicate the pipeline from Selebe-Pikwe to Gaborone providing for all water demands until 2020, given current water use trends (IUCN, 1999). Thereafter, Botswana will have to resort to international water sources from the Limpopo, Ramokgwebana and Zambezi Rivers (GOB-MFDP, 1997). Three possible dam sites in the Limpopo River are being investigated, including the Pont Drift Dam, the Martin’s Drift Dam and the Cumberland Dam. Several committees are responsible for joint planning and decision-making (see Chapter 3).

South Africa

The South African part of the Limpopo catchment feeds into all three river reaches mentioned in the beginning of this section, and can be grouped into two major components (Table 8):

• Crocodile/Limpopo River Basin up to the confluence with the Luvuvhu River (near Pafuri) at the border with Zimbabwe, South Africa and Mozambique;

• Elephants River Basin, which leaves South Africa through the Kruger National Park and joins the Limpopo River in Mozambique.

Storage capacity in the Limpopo River Basin

Effective storage and bulk distribution of water is located mainly in the upper part of the Crocodile River and the upper and middle parts of the Elephants River. Only a few additional development options exist within the Limpopo River, which are basically the Pont Drift, Martin’s Drift and Cumberland dam sites. The present infrastructure is limited to run-of-river abstractions by irrigation farmers in a narrow band around the main stem of the Limpopo River, primarily along the South African side of the river.

TABLE 8
Characteristics of major South African sub-basins of the Limpopo River Basin

Source: Görgens and Boroto (1999); GOSA-DWAF (2003a-d).

Some 100 large dams exist of which about 40 are categorized as major dams with a capacity of more than 2 million m³. The total capacity is almost 2 500 million m³. Of these, the following dams are key sources for domestic water (capacity in brackets):

• Roodeplaat (41.2 million m³), Vaalkop (56.0 million m³), Roodekoppies (103.0 million m³) and Klipvoor (42.1 million m³) in the Crocodile catchment;

• Molatedi (201.0 million m³) on the Marico River (water supply to Gaborone, Botswana);

• Witbank (104.0 million m³), Middelburg (48.1 million m³), Loskop (362.0 million m³) and Arabie (99.0 million m³), and the Phalaborwa barrage on the Elephants River;

• Ebenezer (69.1 million m³) and Tzaneen (157.0 million m³) on the Letaba River (serving Polokwane, Tzaneen and surroundings);

• Vondo (30.5 million m³) and Albasini (28.2 million m³) on the Luvuvhu River.

Some other major dams, used mainly for irrigation, but with capability for domestic use include:

• Hartebeespoort (186.0 million m³) on the Crocodile River (currently experiencing water quality problems);

• Marico (27.0 million m³) on the Marico River;

• Mokolo (145.0 million m³) on the Mokolo River (currently serving irrigation and the Matimba power station);

• Blyderivierpoort (55.2 million m³) and Rooipoort (proposed) on the Elephants River;

• Glen Alpine (20.0 million m³) and others on the Mogalakwena River;

• Nzhelele (55.3 million m³) on the Nzhelele River.

The future construction of the Rooipoort Dam on the Elephants River and a proposed dam on the Steelpoort River will make the system highly regulated and it will probably exceed its full capacity by about 2020. However, if more emphasis is put on water demand rather than water supply management, the construction of the dams can be postponed for a number of years. Most of the water-needy population is located far from the existing dams and, thus, costly distribution networks will be required in order to include them in supply systems. The most suitably located dams and their future extended supply capacity include:

• increased height for the Arabie Dam and the future Rooipoort Dam until 2020 and possibly 2035;

• proposed new dam on the Steelpoort River until 2020 and possibly 2030;

• the Middle Letaba Dam until 2020 and possibly 2030;

• the Vondo Dam is already close to its limit and the Nzhelele Dam has only a limited capacity;

• the Glen Alpine Dam can also only serve the immediate surroundings until 2020.

Water is imported from the Vaal River catchment (Orange River Basin) to urban areas (Johannesburg and Pretoria) in Gauteng Province in order to augment local water resources (Box 9). A significant portion of the return flows from these water uses enters the Limpopo catchment and as such supplements the capacity.

However, recent surveys by the Department of Water Affairs and Forestry (DWAF) make it evident that most of the rural households in the Limpopo River Basin in South Africa cannot be served by surface water only. Present indications are that up to 52 percent will use only groundwater, and the majority of remaining communities will use a combination of groundwater and surface water (GOSA-DWAF, 1999a-c).

Water management areas

In preparation of a national water resource strategy, the country was subdivided into 19 water management areas. Of these, the following four constitute the Limpopo River Basin in South Africa (Table 9).

Crocodile (West) and Marico water management area

Particularly evident from Table 9 is the overriding importance of water transfers into this water management area. In total, nearly 45 percent of the current water available in the water management area is supplied by transfers from the Upper Vaal water management area and beyond. Almost 30 percent of the total water available for use is from effluent return flows, most of which results from water transferred to the large urban and industrial centres in the water management area.

Also significant is the contribution of groundwater, representing about 40 percent of the yield available from the water resources naturally occurring in the water management area.

DWAF plans the transfer of 124 million m³ of water per year from the Crocodile (West) and Marico water management area to Gaborone in Botswana.

Limpopo water management area

In the Sand subarea, groundwater is of overriding importance, while the contribution of groundwater to the total water available in the water management area is among the highest of all water management areas. However, well over half of the available water originates from surface resources, which require careful and efficient management. Water transfers into the water management area serve to augment supplies to the larger urban and industrial areas as well as some mining developments, and are vital to the economy of the water management area. Also noticeable is the volume of return flows estimated to be available for reuse, the quantification of which requires improvement.

Elephants water management area

Large quantities of water are also transferred into the Upper Elephants subarea. These constitute about 22 percent of the total water available in the water management area. Of note is the significant contribution of groundwater, which constitutes nearly 20 percent of the water naturally occurring in the water management area. Usable return flows also represent a substantial proportion of the water available for use in the water management area. However, there is particular uncertainty about the quantity of return flow from irrigation water use, which may have an important impact on the total water availability in the water management area.

Luvuvhu and Letaba management area

Surface water is the dominant source of supply in four of the five subareas. The only exception is the Shingwedzi subarea where more than half of the water available is abstracted from groundwater, while water is also transferred into the subarea from the Luvuvhu River catchment. Also noticeable is the volume of return flows estimated to be available for reuse, the quantifications of which require improvement.

TABLE 9
South African water management areas and sub-basins of the Limpopo River Basin: available water in 2000

Source: GOSA-DWAF (2003a-d).

The quality of the water from the Limpopo River poses serious problems during periods of low flow, in particular upstream of the confluence with the Shashe River (P. Nell, personal communication, 1999). Water pumped from the riverbed from deeper than 1-2 m is very costly and rapidly becoming of inferior quality owing to salinity, herbicides, toxic elements and heavy metals such as boron.

Zimbabwe

Zimbabwe is divided into six hydrological zones (A-F). The Limpopo catchment corresponds to Zone B, which has 30 hydrological subzones, covering a total area of 62 541 km² (16 percent of Zimbabwe’s land area). The catchment has an MAR of 1 157 million m³ or 19 mm, which is less than 6 percent of the total MAR of the country (GOZ-MRRWD-DWD, 1984). Table 10 summarizes the characteristics according to the main tributaries.

The Limpopo River Basin has a highly variable and unreliable flow, and consequently an unreliable water supply. The rivers are intermittent with peak flows in February followed by low flow from May to early November. The reliability of runoff can be indicated by the percent CV, which is the difference between the highest runoff and the lowest runoff of the catchment over time, expressed as a percentage of the MAR. The higher is the percent CV, the less reliable is the runoff. For the Limpopo catchment in Zimbabwe, the CV is 130 percent, which reflects a low reliability, hence the greater the risk of water shortage in the area, unless groundwater sources exist (GOZ-MRRWD-DWD, 1984).

TABLE 10
Characteristics of the Zimbabwe part of the Limpopo River Basin

* According to Görgens and Boroto (1999), the MAR given in column (3) is the denaturalized MAR, but according to GOZ-MRRWD-DWD (1984), the MAR given in (3) is the natural MAR (see also Table 7, where it is put as denaturalized MAR).

** There is a discrepancy between the area of the catchment given above (62 541 km²) and the area according to FAO (1997) (51 467 km²). An explanation may be that the FAO study had included part of the Limpopo River Basin in the adjacent Save or Zambezi basins, which should be corrected

Source: GOZ-MRRWD-DWD (1984).

The priority of water allocation in this catchment goes to domestic and industrial purposes, followed by mining and finally agriculture. The city of Bulawayo is supplied by the Mzingwane, Inyankuni and Ncema dams; Gwanda by the Mtshabezi Dam, Kezi by the Shashani Dam, Mwenezi and Rutenga by the Manyuchi Dam. Beitbridge receives its water directly from the Limpopo River (GOZ-MRRWD, 1999). Most irrigation schemes also receive their water from dams. However, in the schemes developed along the major rivers, sand abstraction is practised. The river sand acts as the aquifer into which boreholes are sunk in order to abstract irrigation water. The water is normally found at 3-10 m below the riverbed.

Storage capacity in the Limpopo River Basin

There are 2 168 dams in the Zimbabwean part of the Limpopo River Basin. However, regardless of the potential available in dams and rivers, the water demand of the rural communities is rarely satisfied. This can be explained by a lack of the costly infrastructure needed to bring the water to these communities. In Zimbabwe, the total capacity of the dams has fallen by about 29 million m³ in the last three years as a result of siltation (Pallett, 1997; GOZ-MRRWD-DWD, 2000).

Currently, of the total MAR, almost 99 percent of the water is already being harnessed/stored. The potential for development exists in places where the MAR is larger than the storage and/or flow rights. However, where the MAR is less than storage, it means that the subcatchment has been developed or already has more dams constructed than can be filled up with the MAR. These dams can still fill up during wet years with a higher runoff than that reflected by the MAR. This is significant when the total commitment is larger than the MAR. Total commitment includes both storage and flow rights. Where a subcatchment has no dams in it, the flow rights will exceed the storage rights.

Mozambique

To a large extent, Mozambique’s water resources are conditioned by the fact that they form part of international river basins, where neighbouring countries upstream are increasingly exploiting available water resources. Such action is claimed to exacerbate downstream problems of water shortages and drought in Mozambique.

An extensive number of studies and reports have provided assessments of national water resources (GOM-DNA, 1986, 1998) and other hydrological assessments (MacDonald and Partners, 1990), including irrigation development in the Limpopo River Basin (Sogreah, 1993).

Using data cited in Sogreah (1987), the basin at the Chokwé station (Figure 5) covers 342 000 km² and the MAR is 5 280 million m³, calculated over a period of 34 years (from October 1951 to September 1985). Figures given above reveal that about 4 800 million m³ enters Mozambique, which means that less than 10 percent is generated within Mozambique. The area of the Limpopo River Basin within Mozambique is 84 981 km², which is about 11 percent of the total area of the country and 21 percent of the total area of the basin (Table 6).

The main tributary, the Elephants River, has a basin of 70 000 km², most of which is in South Africa (68 450 km²). The Elephants River is regulated by the Massingir Dam, with a capacity of about 2 200 million m³. At the Massingir Dam site, the MAR is 1 800 million m³, calculated over the same period of 34 years. The Massingir Dam controls 34 percent of the total flows at Chokwé. Mihajlovich and Gomes (1986) estimated that the annual volume of water entering the station of Mapai on the Limpopo River (upstream of the confluence of the Limpopo and Elephants Rivers) is 3 510 million m³, calculated over a period of 32 years, representing 65 percent of the total flows at Chokwé. However, the discharge is very irregular and may be practically zero in winter.

Downstream of Chokwé, the Changane River (an intermittent tributary without regulation structures), drains a basin covering 43 000 km², but it has a very low runoff coefficient and long periods with no discharge at all. The Lumane River, the last of the most important tributaries, originates in Lake Pave and receives regular inflows from the sandy hillsides. It has a discharge of about 10 m³/s. Table 11 shows the flow regime of the Elephants and Limpopo Rivers in Mozambique.

Sogreah (1993) noted that more than 75 percent of the annual volume occurs during only three months (January-March), which is quite extreme even in comparison with other rivers in the south of Mozambique, and that the whole dry semester (May-October) represents less than 10 percent of the annual flow.

The annual inflow CV is 1.07 for the Limpopo River at Mapai and 0.61 for the Elephants River at Massingir (comparable with those of the Sabie, Incomati and Umbeluzi). In addition to the irregularity of natural inflows, which calls for the construction of dams in order to guarantee a regulated annual volume, another problem lies in the form of offtake in the neighbouring countries. Many dams have been built in Zimbabwe and South Africa and the effect of water storage in these countries may be seen in recent years in terms of change in the discharge recession curves during the dry season.

TABLE 11
Flows measured at the Massingir and Chokwé stations

Offtake in countries upstream of Mozambique may have a favourable effect in reducing floods, but it is detrimental in low-water periods, when water requirements are most acute. Citing Sogreah (1993), South Africa and, to an unknown extent, Botswana and Zimbabwe are extracting considerable quantities of water from the Limpopo River Basin, estimated at 1 173 million m³ in 1980 (GOSA-DWA, 1985). This figure was projected to increase to 1 385 million m³ in 1990 and 1 723 million m³ in 2000. The corresponding figures for the Elephants River Basin were 1 038, 1 188 and 1 254 million m³, but their origin is questionable as they were presented within a negotiation framework.

The effect of the increase in abstractions is already apparent in the dry season in the Limpopo River, where it has become normal to find it completely dry for some months each year. Sogreah (1993) used recession curves of different years to demonstrate an increase in their gradient in more recent years, and developed a number of scenarios for analysis of water abstractions and their impact in Mozambique.

Between October 1981 and September 1986, the mean inflow at Massingir was 792 million m³, which is only 44 percent of the mean over a period of 34 years. The reason for this is not offtakes in South Africa, but rather climate conditions. The mean inflow of the five earliest years (1961-66) was 834 million m³. Conversely, the mean inflow in 1973-1978 was 2 976 million m³, which is 165 percent of the 34-year mean. The possibility of several successive dry years occurring means that large-capacity dams will have to be built and will enable a maximum of only 60 percent of the natural inflow to be regulated (Sogreah, 1987).

Groundwater resources

Groundwater is used extensively in the region, mainly for irrigation and rural supplies. Communities are often located at significant distances from river reaches and depend solely on groundwater resources for survival.

Botswana

Most of the rural population is located far from surface water resources and depends mainly on groundwater resources. Traditionally, most major villages have also used groundwater, which is now being augmented by local or regional surface water supplies, including the North-South Water Carrier. Most smaller rural villages can derive their domestic water needs from groundwater without depleting the available resource. However, in periods when rivers and local dams are dry, these groundwater resources are overexploited to also serve livestock watering needs. Groundwater resources at Jwaneng and Orapa are being overexploited resulting in so-called groundwater mining, which is depleting available groundwater reserves.

Botswana’s groundwater is characterized by very low recharge rates, low probability for high-yielding boreholes and relatively high salinity. In the Limpopo catchment of Botswana, the mean recharge to groundwater ranges from 1 to 3 mm/year in the Kalahari and northwestern parts of Central Region to 5-9 mm/year for most of the eastern areas, except for the Tuli Block. Extended drought periods affect the reliability of these sources and require active monitoring and management. The low recharge precludes any large-scale development of groundwater because it would lead to unacceptably high rates of groundwater exploitation (mining) and subsequent damage to the resource.

The median yield from successful boreholes is moderate to low ranging from 3 to 6 m³/h for most of the Limpopo catchments in Botswana. The best production areas are between Gaborone and Mahalapye and in smaller areas around Ramotswa and Palapye. They have been defined into 10 production areas, including well fields at Palla Road, Ramotswa, Lobatse, Ramonnedi, Molepolole, Mochudi, Palapye, Serowe, Paje and Shashe. The Palla Road and Ramotswa well fields have the highest capacity and will probably be linked as sources to the North-South Water Carrier Project. Dolomite aquifers in the Kanye, Sekoma, Molopo and Mashaneng areas have significant storage capacity and high yields.

Sandstone formations, which are linked to the central Kalahari, have the largest recharge areas and total storage, but often have high salinity and require large numbers of boreholes. Occasionally, sand rivers provide good recharge and storage capacity for local extraction, but generally they have inadequate capacity for regional uses. The main aquifers and their related storage and yield capacities are listed in the national water master plan (GOB-MMRWA, 1992). Subsequent national development plans (GOB-MFDP, 1997) provide upgraded information based on groundwater investigations and compliance monitoring.

Groundwater quality is often deficient with high salinity and excess concentrations of fluorides, nitrates, and other harmful elements in some regions. The total dissolved solids range from 1 000 to 1 500 mg/litre for most of the Limpopo catchments in Botswana, and increased levels of nitrates are occurring near irrigation and within settlement areas.

Conjunctive use of surface water and groundwater is essential for sustaining water quantity and quality requirements of users. Monitoring programmes are being implemented to protect and manage groundwater and surface water against pollution and overexploitation. Active management of water demand and water quality is critical to managing drought and the impact of drought in Botswana.

However, the management of groundwater is complicated by the “common pool” problem. While individual use or misuse may not result in a significant problem, the combined impact is often unacceptable. It is then difficult to determine who is responsible and how the situation is to be regulated.

South Africa

Rural communities and irrigation farming make extensive use of groundwater, extracting a total of about 850 million m³/year. It is estimated that more than 55 percent of rural communities are supplied from groundwater as their only source. Most of the remaining communities use a combination of groundwater and surface water (GOSA-DWAF, 1999a-c).

Dolomite aquifers occur in the Crocodile River Basin and the Blyde River area, but are generally distant from the needy rural communities. Most rural communities are located on minor aquifer types with an average borehole yield of about two litres per second. Communities north of Soutpansberg are located on poor aquifer types yielding less than 1 litre/s. The water they provide often fails to meet domestic water quality standards because of high salinity. Only the southeastern parts of the former Bolobedu area of Lebowa have reasonable groundwater potential and quality.

On average, 5-10 boreholes need to be drilled for each community and they can generally serve only communities of fewer than 2 000 people. There are indications that up to 27 percent of boreholes have a water quality that is marginal or poor for domestic use and causing it to have a number of limitations for crop irrigation. Key problem areas are: the Springbok Flats and surroundings, where high fluoride concentrations are common; the area north of Soutpansberg, where high solute concentrations are found; and areas around Dendron (Box 10) and along the main stem of the Limpopo River, where high evaporation influences the salinity level of the water.

Zimbabwe

The Limpopo River Basin area in Zimbabwe is not well endowed with groundwater. Most of the wells in the communal areas that are used for household purposes, caring for livestock, and watering of gardens, run dry long before the rainy season starts. However, in addition to a few groundwater aquifers, the basin also has subsurface water stored close to the surface in a few dambo (wetland) areas. True dambos no longer exist in most parts of Matabeleland South Province, primarily because of land degradation over the years from prolonged droughts and overstocking. However, vegetable production on seasonal wetlands does occur on communal lands near Matopo, Esigodini, Godlwayo and some other areas (DANIDA, 1990).

Data on the exact extent and quantity of groundwater are not available. Areas with reasonable groundwater reserves, both in terms of quantity and quality, are found in two areas (Figure 5). The first is around Esigodini, south of Bulawayo in the Umzingwane River Basin, where the water is used for the production of vegetables and fruit under irrigation. The second is in the Malipati area at Manjinji, near the southern reaches of the Mwenezi River, where there is potential to irrigate up to 1 000 ha, out of which only a very small portion is already irrigated.

The water at both Esigodini and Malipati occurs at shallow depth (20-30 m) and is of good agricultural quality. Groundwater also occurs along the Limpopo River and below the riverbed sand aquifer (3-10 m deep). The quality of water east of Beitbridge around Grootvlei is low owing to salinity, although the quantities are good (GOZ-AGRITEX, 1990). An irrigation scheme established before independence has been abandoned because of poor water quality.

Mozambique

Existing data on groundwater resources relate to information from wells (at time of construction) and to geological information. According to Sogreah (1993), the groundwater potential in the Limpopo River Basin area is limited, particularly because of the high mineralization of many of the aquifers. According to another synthesis report on water resources in Mozambique (GOM-DNA, 1999), groundwater in the vast interior area of Gaza Province is unfit for consumption because of high levels of salinity.

Six different zones are considered in characterizing the groundwater potential in the Limpopo River Basin in Mozambique:

• Dune area: a 40-60-km wide strip along the coast. Productivity is considered low to medium. Quality is good because of to the high recharge rate of 50-200 mm/year. Recent studies estimate the exploitable amount of groundwater to be about 5-10 m³/h per km².

• Alluvial valleys: formed by the incised main valleys of the Limpopo and Elephants Rivers. Productivity is high, but water quality is a major problem because the rivers drain the adjacent plains that have highly mineralized groundwater. Fresh groundwater occurs where the surface waters of the rivers replenish the aquifers directly, but care is needed to prevent overexploitation and avoid the risk of salinization of the aquifers.

• Old alluvial plains: bordering the dune area. This region does not provide any potential for groundwater exploitation as it is highly mineralized.

• Erosion plains and erosion valleys: a shallow eluvial cover of sandy clays over the entire inland area. Productivity is low in general but calcareous sandstones have higher specific yields. Water quality is usually poor, with exceptions found along water lines and local depressions that are recharged from the temporary rivulets.

• Deeper aquifer: found in the medium and lower Limpopo River Valley at depths ranging from 80 m at Mabalane to 200 m at Xai-Xai. The total exploitable groundwater in this aquifer, which seems to be enclosed by a saline cover and a brackish base, has been estimated at 300-600 m³/h.

• Lebombo Range: the rhyolites of the Lebombo Range have very low productivity. Very few wells have been drilled in this region and the failure rate is high.

The aquifers related to the sedimentary post-Karoo formation deliver sufficient water of suitable quality for irrigation. This formation generally runs parallel to the course of the main rivers south of the Save River. The more recent deposits of the coastal dunes also provide water of good quality, but the yield is small.

Many reports conclude that large-scale groundwater abstractions in the Limpopo River Basin are very limited as a consequence of low productivity and poor water quality. There exists a deep aquifer between 250-350 m, which may be continuing to the south, but exploitation of this source is not economically feasible. Water quality becomes progressively worse downstream of Chokwé and the confluence with the Changane River. Only the dune unit can be used for small- and medium-scale abstractions without restrictions posed by water quality. For irrigation purposes, groundwater safe yields are too small and can be ignored.

Another source of water is that from the sandy hillsides in the lower Limpopo area, which is being partially used for the irrigation of the Machongos. Machongo is the local name for a type of hydromorphic soil (a kind of peat soil wetland) with very high organic matter content, part of which is in a very coarse form. While not directly under the irrigation subsector, subsurface irrigation on machongos is practised in Xai-Xai District. Machongos are found mostly along the sea coast, in the valleys of the main rivers (at the junction of the valley with the higher surrounding ground), or associated with smaller streams where the flow of water is seasonally impeded.

Interbasin and intrabasin water transfers

A number of water transfer schemes have been developed or proposed in order to address the relatively severe water shortage in the Limpopo River Basin while maintaining the current emphasis on water supply management rather than water demand management. Transfers of water may be made from one sub-basin of the Limpopo River Basin to another, within one country or between countries. This is termed intrabasin water transfer. Water can also be transferred between the Limpopo River Basin and other basins, within one country or between countries. This is termed interbasin water transfer. The Limpopo has four interbasin transfer schemes and two intrabasin schemes (Table 12).

In South Africa, water is imported from the Usutu, Vaal and Komati Rivers to serve the high water quantity and quality demands of the power stations in the Upper Elephants River Basin. In addition, continued importing of water from the Orange and Vaal River system (Box 9) and greater reuse of return flows in the Crocodile River could create surplus water in the Crocodile River, which could then be exported to supplement the Mogalakwena, Mokolo and Elephants basins. However, these return flows affect the quality of the water, and it becomes increasingly important to manage the resulting water quantity and quality in an integrated way in order to prevent environmental impacts and to ensure compliance with other water user requirements.

These transfer schemes are the subject of intense debate because of their high cost and potentially negative impacts on the environment and ecological balance. Transfer schemes affect river basin planning, water quantity, water quality, land, aquatic systems, terrestrial systems and socio-economic issues in the countries sharing the receiving or supplying basins. This requires all countries affected directly and indirectly to be involved from the outset in the planning and decision-making. However, the existing treaties on the Limpopo River Basin are either bilateral or trilateral agreements and, as yet, no agreement has been established and signed by all four countries sharing the basin.

TABLE 12
Interbasin and intrabasin water transfers related to the Limpopo River Basin

¹ Countries initiating and implementing the transfer scheme.

² Countries sharing involved basin and which consequently are affected by the scheme.

Issues such as interbasin water transfers are meant to be regulated by the protocol on shared watercourse systems in the SADC region, which came into force in 1998 after ratification by the required two-thirds majority of the SADC member states (SADC, 1998). The introduction of proper water management demand systems could postpone future transfer schemes and overabstraction from international rivers. However, the protocol fails to incorporate water demand management as an explicit strategy. Notwithstanding this, the agreement supports the requirement that national resources be used as efficiently as possible prior to international abstractions (IUCN, 1999).

Even though the Limpopo River Basin lacks a single comprehensive treaty, there is the firm commitment to cooperate through the Limpopo Basin Permanent Technical Committee, established in Harare in 1986, which includes all four countries. Negotiations are currently underway for this committee to become the Limpopo Basin Commission (LIMCOM). The draft agreement is under preparation at present and is discussed in more detail in Chapter 3.

Land and vegetation classification and assessment

Agro-ecological zoning and land evaluation

FAO’s AEZ has been developed to assist with land resources assessment for better planning management and monitoring of these resources (FAO, 1996). The AEZ system includes:

• inventory of land resources and land utilization systems;

• evaluation of land suitability and productivity;

• mapping of AEZ, land suitability, problem soils, etc.;

• land degradation and population-supporting capacity assessment.

The AEZ methodology was developed in the 1970s and is applied as a system to evaluate land for rainfed and irrigated agriculture, forestry and extensive grazing. The AEZ concept involves the combination of layers of spatial information, such as topography, physiography, soils, climate, catchments, land cover, production systems and population, combined and analysed using a geographical information system (GIS). AEZ is applied widely across the globe, including in countries in southern Africa. However, a regional AEZ map of southern Africa does not exist.

AEZ is an established reference system in Botswana, and is also used in multiple applications, including the production of land suitability maps. The general soil map of Botswana (De Wit and Nachtergaele, 1990), in combination with climate data, crop requirements and other information, has provided the basis for the national land suitability map for rainfed crop production (Radcliffe, Tersteeg and De Wit, 1992). In several areas of the northern part of the Limpopo River Basin, detailed land evaluation studies have determined land suitability for rainfed and irrigated crop production (De Wit and Cavaliere-Parzanese, 1990; De Wit and Moganane, 1990).

AEZ has been used in Mozambique since the 1970s, and a national AEZ map at a scale of 1: 2 000 000 is available (not digital). Following independence in 1975, a handful of foreign consulting firms, contracted by national directorates and secretariats, acted as the principal providers of ad hoc land resources information and their assessment for agricultural potential, applying their own methodologies and procedures. In the last 20 years, a series of multilateral and bilateral programmes (FAO, the Netherlands and the Soviet Union) have upgraded the capacity in soil surveying and land evaluation in the National Directorate for Geography and Cadastre (DINAGECA) and the INIA.

Voortman and Spiers (1981) produced a qualitative national assessment of land resources and their suitability for rainfed production in a series of five maps (mean annual rainfall; vegetation and potential utilization; agroclimate zones; terrain limitations for rainfed agriculture; and agricultural land suitability) at a scale of 1:4 million. Further systematic land resource inventories and associated studies were conducted by FAO in 1982, which compiled information nationally on land suitability for eight rainfed crops in a six volume report by Kassam et al. (1982). Based on the AEZ methodology of FAO (1978), this study determined the agroclimatic suitability of the major rainfed crops for each growing period zone, to arrive at the land suitability classification. It produced the National Land Resources Map at a scale of 1:2 million and determined climate suitability for maize, sorghum, millet, wheat, soybean, groundnut, cassava and cotton, each mapped at a scale of 1:5 million. The inventory was updated by Snijders (1986).

Another FAO project conducted land resource assessments at regional scale involving systematic soil surveys and land evaluation. This included an assessment of Gaza Province, in which soils, geomorphology and terrain were mapped at a scale of 1:250 000. The project also introduced the automated land evaluation system (ALES) to handle the 125 soil mapping units in Mozambique, resulting in individual land evaluations that are crop, area and land-user specific.

AEZ according to FAO standards has not yet been determined in South Africa. The closest related system currently in use is that of land type maps showing climate zones. The Institute for Soil, Climate and Water of the ARC uses statistics from weather stations to determine the climate zones and to develop ten-daily vegetation greenness maps using a normalized difference vegetation index (NDVI), which form part of the drought management system. The introduction of AEZ in South Africa is recommended, as it would also provide essential linkages with other global land resource approaches (Van Der Merwe et al., 2000).

Venema (1999) used the AEZ concept to subdivide the five natural regions of Zimbabwe into 18 provisional agro-ecological zones. The zones reflect rainfall probability, LGP, and predominant soil type. The system is based on the long-established system of the five natural regions of Zimbabwe defined by the Agricultural, Technical and Extension Service (AGRITEX), based on mean annual rainfall, rainfall distribution and altitude (GOZ-Surveyor-General, 1998).

Bernardi and Madzudzo (1990) distinguished six agroclimatic zones, based on the ratio of mean annual rainfall at the 80 percent probability level and the calculated or extrapolated average annual ET_o. In the moist areas of Zimbabwe, the AGRITEX natural regions are similar to these agroclimatic zones. However, in the dry southeast lowveldt, Natural Region V has been split into two parts in order to create Agroclimatic Zone VI in the extreme south, representing the most arid climate of Zimbabwe.

Land cover and vegetation classification

Currently, individual land cover exercises are difficult to compare as countries use different categories and legends, based on a variety of definitions. Descriptions of land cover are usually dominated by forest types, such as woodland, bushland, savannah, and wooded and open grassland. There are also differences between the database structures and descriptions, and as these have been defined without coordination, the exchange and comparison of data from different countries is difficult.

At a generalized scale, the World Resources Institute, has published generalized land cover data (GLCCD, 1998; Loveland et al., 2000) and other information derived from the Water Resources eAtlas (World Resources Institute, 2003) for a number of Basins, including the Limpopo River Basin (Box 11). This information shows the Limpopo River Basin to be occupied mainly by savannah, cropland/natural vegetation mosaic, some grassland and urban/industrial areas (Figure 26).

Apart from the coverage of the Eastern Africa Region by UNESCO’s International Classification and Mapping of Vegetation in 1973, no regional vegetation coverage using a standardized classification and a larger scale is presently available. Using Landsat images and aerial photos, the Southern Africa Regional Office of the World Wildlife Fund (WWF) has done some mapping of land cover and land use change in southern Africa. The USGS-EROS Global Land Cover Project is also developing more standardized datasets on land cover using satellite imagery.

Source: World Resources Institute (2003).

Source: World Resources Institute (2003).

Botswana

Information on forest cover in Botswana is often conflicting, depending on the definition applied to forest, woodland, etc. The land systems map of Botswana gives detailed information on the occurrence of woodlands, savannahs and grassland, including subdivision into density classes (GOB-MOA, 1990). The provisional vegetation map of Botswana by Weare and Yalala (1971) is the first detailed national vegetation map of Botswana. The most important units distinguished from northeast to southwest along the Limpopo River are:

• arid sweet bushveldt;

• mopane bushveldt;

• tree savannah (Acacia nigrescens/Combretum apiculatum);

• arid sweet bushveldt;

• semi-sweet mixed bushveldt.

Timberlake (1980) described and mapped the vegetation of southeast Botswana, distinguishing five main types:

• sandveldt tree and shrub savannah;

• hardveldt woodland and tree savannah;

• woodland on hills;

• shrub savannah on clay and calcrete;

• riverine woodland.

Timberlake’s mapping approach was used during the soil mapping of the hardveldt (e.g. Remmelzwaal, 1989; Moganane, 1990), providing a better relationship with soil patterns as compared with the Weare and Yalala mapping. All soils reports of the standard 1:250 000 Botswana map series contain detailed descriptions of the vegetation. The overall descriptions were then combined in a national vegetation map of Botswana at a scale of 1:2 million, showing a strong relationship with underlying soil patterns (Bekker and De Wit, 1991). The four dominant vegetation associations along the Limpopo catchment from north to south are:

• Colophospermum mopane with Acacia nigrescens;

• Combretum apiculatum with Acacia nigrescens;

• Terminalia sericea with Acacia tortilis and A. erubescence (in sandy areas);

• Pelthophorum africanum with Acacia tortilis.

Currently, the Botswana Range Inventory and Monitoring Project (BRIMP) is re-mapping and sampling the vegetation of Botswana. About 20 percent of the task is complete. Ground data on species composition and cover is captured in GIS format.

Mozambique

Baseline information on the country’s natural forest and woodland resources was first provided by a 1980 reconnaissance forest inventory, carried out in 1979-80 under an FAO-UNDP forestry sector project. At this time, the forest area of Gaza Province was inventoried at 1.3 million ha (Macucule and Mangue, 1980).

The 1980 inventory was updated in the early 1990s using visual interpretation of 1990-91 colour composite images (1:250 000) and Landsat TM (1:1 million), together with field truthing in all provinces. It produced a national land cover and land use map at a scale of 1:1 million, with quantitative data on the extent of the forest resource base by major forest type and land use classes (using the AFRICOVER legend), the extent of agricultural encroachment within the natural woody vegetation, as well as an estimate of deforestation rates. Table 13 provides data on land cover in Gaza Province according to the updated national forest inventory.

TABLE 13
Forest type and land use class in Gaza Province, Mozambique

Source: Saket (1994).

Overall, Gaza Province showed a low rate of deforestation (0.92 percent over 18 years), even though clearings of natural vegetation occurred widely along the Limpopo Corridor where many villages were created after 1972, and around the districts of Chicualacuala and Massangena. In contrast to other provinces, woody vegetation had recovered over wide areas in Chigubo, Dindiza, Nalazi and Changane districts and along the western border with Inhambane Province. Deforestation is most severe in Xai-Xai District, caused essentially by inappropriate agricultural practices, the collection of woodfuels and building materials from woodlands, and the high frequency of extensive forest fires. The forest inventory determined the allowable cut at 13 141 m³ from an area of productive forest of 1 437 162 ha.

Current initiatives involve the mapping of land use and land cover for the whole of Mozambique at a scale of 1:250 000 and eight selected districts at a scale of 1:50 000 scale (Desanker and Santos, 2000), using the FAO AFRICOVER legend.

The National Remote Sensing Centre (CENACARTA), created in 1989 under a French aid programme, is the supplier of satellite images in Mozambique (SPOT, Landsat, Radarsat, ERS and SPIN). Mozambique does not have a receiving station, but the CENACARTA has agreements with image suppliers in South Africa, such as the Satellite Applications Centre (SAC) of the Council for Scientific and Industrial Research (CSIR) in South Africa. The CENACARTA has technical capacities to process images and supply them in analogue and digital format.

The main biome in Gaza Province is the dry/eutrophic savannah (Figure 27), which lies between the 400-and 600-mm rainfall isohyets. It is characterized by the dominance of Acacia spp. and Colophospermum mopane on heavier-textured, base-saturated soils, and Caesalpinoideae and Combretaceae on leached, sandy and lighter-textured soils. The Miombo savannah woodlands, which typify the moist/dystrophic savannah, reach their southern limit as a whole system north of the Limpopo River estuary. Three main ecoclimate zones can be distinguished from upstream to downstream:

• From the border to 100 km downstream of the confluence with the Elephants River, the sand plateau is dominated by a vegetation of very dense arid sand thicket communities or by woodlands dominated by multistemmed short trees. On the slopes of the plateaus, Colophospermum mopane and Acacia exuvialis dominate an open woodland community, with woodlands of Acacia, Commiphora and Terminalia at lower elevations. Floodplains support woodland and open woodlands of Acacia (xanthophloea, tortilis and nilotica) with shrub-thickets of Salvadora persica. Acacia xanthophloea, a flat-topped tree with yellow bark, grows where its roots can find water easily. It is, as is Salvadora persica, an indicator of saline-alkaline soils on flat seasonally waterlogged soils close to the riverbanks. Acacia tortilis is a salt-tolerant tree found on the more permeable soils.

• In the more subhumid zone downstream to about 100 km inland from the sea, there are very open woodlands with short dense thorn thickets in the bottomlands; a result of the continued removal of woody elements for fuelwood in this region. The floodplain supports woodlands of Acacia. It is within this region that the transition from a xeric to moist sand thicket and woodland takes place. On the levee deposits along the Limpopo River, there are dense thickets of Ficus, Diospyros and Zanthocercis. The near floodplain supports Acacia or open woodlands. Around the Chokwé scheme, there are considerable reductions in the woody biomass, except in the western part where very dense thickets and woodlands of Acacia remain. A survey by Timberlake, Jordáo and Serno (1986) showed a close association of vegetation with soil type and drainage, determined primarily by soil texture, and by period of water inundation. Poorly drained soils support open grassland vegetation with few shrubs or trees, while the grass species present on the plains appear to depend on length of inundation by water. Sandy loam soils support an Acacia woodland, often associated with reasonably high soilcalcium levels. Deep sandy soils (Arenosols) support woodlands of broad-leaved deciduous tree species.

• In the coastal area characterized by a humid tropical climate, the present composition and the physiognomy of the vegetation patterns reflect anthroprogenic changes to the original mixed forest-woodland-grassland physiognomy of many species and facies.

Tique (2000) characterized the vegetation in Chicualacuala District as being dry deciduous tree savannah dominated by Androstachys johnsonii and deciduous tree savannah at medium and low altitudes, which is a secondary formation of Colophospermum mopane and other low-altitude tree savannah, occasionally mixed with smaller trees such as Sclerocarya caffra, Kirkia acuminata and Combretum spp. There is also a strong relationship between soil types and vegetation patterns on the sandy plains. Terminalia sericea and Rhigozum sp. are the dominant species on the very deep yellowish-brown sands (Ferralic Arenosols), along with Androstachys johnsonii and some Commiphora, Grewia and Combretum. In soils with a high clay content, Colophospermum mopane is dominant.

South Africa

The South African National Land Cover Database Project (NLC) was completed in 1999 (Thompson, 1999; Fairbanks et al., 2000), producing a standardized land cover database for all of South Africa, Swaziland and Lesotho (see Figure 28 for transformed land cover types in the South African part of the Limpopo River Basin). The project utilized Landsat TM satellite imagery captured in 1994-95. The product is designed for 1:250 000- scale mapping applications, intended to provide national baseline information on land cover. Georeferenced land use types at a national scale are inferred from the land cover map, to be read in conjunction with information from Statistics South Africa and the Department of Agriculture. Data sources are available, at best, at magisterial district level.

The classification scheme is based on clear and unambiguous terminology and class definitions, designed to ensure data standardization and to conform to internationally accepted standards and conventions, such as in use for the Vegetation Resource Information System (VEGRIS) of Zimbabwe, and the FAO AFRICOVER project.

One advantage of the NLC classification is that it shows simultaneously the major land cover, based primarily on broad structural vegetation, but also incorporating land use components. There are three levels of classification and the second level includes useful subclasses based on degradation and different crops.

On the South African side, the Limpopo River Basin falls largely within the savannah biome (commonly known as bushveldt), which is the largest biome in southern Africa, occupying 46 percent of the southern African area. The vegetation is well summarized in Low and Rebelo (1996), from which extracts have been included below.

A grassy ground layer and a distinct upper layer of woody plants characterize the vegetation. A major factor delimiting the biome is the lack of sufficient rainfall, which prevents the upper layer from dominating, coupled with fires and grazing, which keep the grass layer dominant. Several variations in vegetation occur within the basin.

The mopane bushveldt dominates the undulating landscapes from Kruger National Park to Soutpansberg in Limpopo Province. The vegetation is characterized by a fairly dense growth of mopane (Colophospermum mopane) and a mixture of other tree species. The shrub and grass layers are moderately well developed.

The Soutpansberg arid mountain bushveldt is restricted to the dry, hot, rocky, northern slopes and summits of the Soutpansberg Mountains (Low and Rebelo, 1996). There is a distinct tree layer, which is characterized by Kirkia acuminata, Englerophytum magalismontanum and other tree species. The shrub layer is moderately developed and the grass layer is poorly to moderately developed. The Waterberg moist mountain bushveldt, although related to the Soutpansberg arid mountain bushveldt, occurs on the sandstone and quartzite soils on the rugged and rocky Waterberg Mountains. The tree layer is well developed with a moderately developed shrub layer and a moderately- to well-developed grass layer.

The clay thorn bushveldt is distributed widely on the flat plains that have black or red vertic soils southeast of Potgietersrus and Nylstroom in Limpopo Province and similar habitats in North West Province. Vegetation is dominated by Acacia species and turf grasses (Ischaemum afrum).

Sweet bushveldt occurs on deep greyish sand overlying granite, quartzite or sandstone in the dry and hot Limpopo River Valley and the associated valleys of tributary rivers in northwest Limpopo Province. The vegetation structure is mostly short and shrubby. Trees such as Terminalia sericea, Rhigozum obovatum and Acacia tortilis dominate sandy areas. Grasses dominate the herbaceous layer. The mixed bushveldt vegetation type found on undulating to flat plains varies from a dense, short bushveldt to a rather open tree savannah covering most of Limpopo Province and northern North West Province. On shallow soils, Combretum apiculatum dominates the vegetation. Grazing is generally sweet and the herbaceous layer is dominated by grasses such as Digitaria eriantha on shallow soils and Terminalia sericea on deeper soils.

The mixed lowveldt bushveldt is found on the sandy soils of the undulating landscapes of Limpopo Province and Mpumalanga Province on the eastern boundary of the country. Vegetation is usually dense bush on the uplands, open tree savannah in the bottomlands, and dense riverine woodland on riverbanks. This bushveldt is confined to frost-free areas. In general, vegetation has been damaged severely in vast areas and has in some cases been destroyed almost completely by overgrazing and injudicious utilization.

Zimbabwe

In Zimbabwe, the VEGRIS supports a land use database for the whole country, derived from SPOT imagery of 1992 and ground truthing. This database also contains topographic information, but no detailed species data are captured. Recently, the Zimbabwe land reform programme has stimulated the digitization of commercial farm boundaries at a scale of 1:250 000 in order to produce a national land inventory.

Land cover information is available from the woody cover map of Zimbabwe (GOZ-FC, 1998), which distinguishes the following main categories of vegetation structure:

• cultivation (27.5 percent);

• woodland (53.2 percent);

• bushland (12.7 percent);

• wooded grassland/grassland (4.8 percent);

• plantation forestry (0.4 percent).

The remaining 1.4 percent consists of water bodies, settlements, rock outcrops and natural moist forests.

Two woodland types predominate in the Limpopo River Basin: mopane woodlands and Acacia-Combretum-Terminalia woodlands. Mopane woodlands are quite widespread in Zimbabwe and are associated with low-altitude, hot areas with sodic or alluvial soils. Colophospermum mopane is the dominant species and constitutes 18.5 percent of the 101 500 ha under mopane woodland. It is used for craftwork, fuelwood, poles, railway sleepers and parquet floors.

Acacia-Combretum-Terminalia woodland type is found in the vleis in the drier parts of the country. Acacias tend to form the dominant component of these woodlands. In terms of population pressures, these woodlands are second to the miombo woodlands. Acacias provide nutritious fodder, improve soil fertility and rehabilitate degraded sites and fix sand dunes. On the other hand, Terminalia tends to dominate burnt sites and is important for fuelwood, poles, tools and wagon draught shafts.

The southeast middleveldt is an area of transition with low mopane woodland in the drier areas (less than 500 mm mean annual rainfall) and Terminalia sericea open woodland in the slightly wetter areas. Julbernardia globiflora is found locally on high ground, and Brachystegia glaucescens on outcrops of granite and gneiss. An association of Colophospermum-Combretum-Acacia is common in lower slope positions. Acacia spp. are dominant on the few areas of red clay soils derived from schist.

Colophospermum mopane is the dominant species in the southeast lowveldt, where it forms an open tree savannah. Commiphora tree savannah is the typical vegetation type on shallow soils over basalt, which includes Combretum apiculatum, Boscia albitrunca, Adansonia digitata and Colophospermum mopane. Mopane shrub savannah dominates moderately deep vertic soils on basalt. The Karoo sandstone areas near the border with Mozambique are characterized by Guibourtia conjugata tree savannah and Guibourtia conjugata/Baphia massaiensis woodland thicket, with Androstachys johnsonii thicket occurring locally.