Myburgh & bredenkamp.qxd Macro-channel riparian vegetation of the Olifants River System in the Grassland Biome, Mpumalanga W.J. MYBURGH and G.J. BREDENKAMP Myburgh, W.J. and G.J. Bredenkamp. 2004. Macro-channel riparian vegetation of the Olifants River System in the Grassland Biome, Mpumalanga. Koedoe 47(1): 41–54. Pretoria. ISSN 0075-6458. The vegetation associated with the macro-channel of the Olifants River System was investigated to identify plant communities mappable at a spatial scale of 1:250 000. The results obtained by using the PHYTOTAB PC-classification and mapping program package, revealed eight distinct plant communities associated with the macro-channel of this river system that occurs within the Grassland Biome. Key words: macro-channel, Olifants River System, plant communities, riparian vegeta- tion. W.J. Myburgh , Private Bag X680, Department of Nature Conservation, Tshwane University of Technogoly, Pretoria, 0001 Republic of South Africa; G.J. Bredenkamp, Department of Botany, University of Pretoria, Pretoria, 0002 Republic of South Africa. ISSN 0075-6458 41 Koedoe 47/1 (2004) Introduction The Olifants River, the second largest river in the former Transvaal (now the Limpopo Province and Mpumalanga), is one of the most severely polluted rivers in the region (Olifants River Forum 1995). This project was initiated, inter alia, to obtain a holistic perspective of the current state of the vegeta- tion associated with the macro-channel, including the extent and distribution of exotic vegetation within the macro channel of the river system (Myburgh 2000, 2001). In the study, which was conducted over a period of four years, the vegetation was mon- itored prior to and after flooding events. This report focuses on the species composition and structure of the plant communities asso- ciated with the macro-channel of the Olifants River System within the Grassland Biome. Study area The Grassland Biome section of the Olifants River (Fig. 1) stretches downstream from the river's origin in the vicinity of the town Breyten, to the Witbank Dam (Rutherford & Westfall 1986). The vegetation of this area is broadly referred to as Bankenveld (Veld Type 61) (Acocks 1988). The width of the macro-channel varies from only a few metres (< 5 m), with no distinct riparian vegetation zone on the river banks, to 32 m wide with a well-developed riparian zone quite distinct from the adjacent zonal grassland. This section of the Olifants River flows through a landscape that may, in gen- eral, be described as flat to rolling plains dominated by grassland, with woody plant species restricted to the rocky river banks. The catchment area surrounding the macro- channel is mainly used for agricultural and industrial purposes, with mining activities being the most prominent. Methods Various researchers (Vannote et al. 1980; Ward & Stanford 1983; Naiman et al. 1988; Townsend 1989) indicated that riparian vegetation changes down- stream from the origin of a river system, as the macro-channel develops and changes. In this study, it was decided to use a spatial scale of 1:250 000, in order to include the Olifants River System from its origin to the western border of the Kruger National Park. Geological data were the most practical empir- ical data set available at this particular scale and were Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 41 therefore used as baseline template to stratify the landscapes through which the Olifants River flows. The width of the riparian zone varies in correspon- dence to the slope and height of the banks at any given locality (Nilsson et al. 1994). An area-based vegetation sampling method was used to survey the vegetation. Variable-sized belt transects were placed within the stratified units. The number of sample plots placed in each stratified unit was calculated proportionally depending on the length of the unit. The width and length of the belt transects were influ- enced by the width of the channel bank and the species richness and growth forms present at a spe- cific site. Floristic and habitat data were collected at 18 geo-referenced localities representing 36 relevés (>200 m²), since both macro-channel banks were surveyed as separate relevés. The floristic data recorded at each of these relevés included species composition, growth forms and canopy cover using the Plant Number Scale (Westfall & Panagos 1988). The floristic data were analysed using the PHY- TOTAB-PC computer program package (Westfall 1997). Results and discussion A total of eight clearly recognisable plant communities, representative of the vegeta- tion of this section of the macro-channel of the Olifants River System (Myburgh 2000, 2001), were identified, mapped and described Koedoe 47/1 (2004) 42 ISSN 0075-6458 Fig. 1. Distribution of the plant communities of the Olifants River associated with the Grassland Biome. Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 42 at a spatial scale of 1:250 000 (Fig. 1, Table 1). 1. The Eragrostis plana-Monopsis decipiens grassland The Eragrostis plana-Monopsis decipiens grassland is represented by two relevés only. This grassland community represents pri- marily the grassland along the macro-chan- nel banks of that section of the Olifants River that stretches from its origin to the Nooitgedacht farm (Fig. 1). The distribution is limited to the areas associated with the Ea Land Type (Land Type Series 1985a), which are characterised by rocks from the Karoo Sequence, at an altitude of about 1 750 m a.s.l. The soil depth at the top of the banks varies between 600 mm and 640 mm. This section of the river has a single active channel in the macro-channel floor that is 1–6 m wide (Fig. 2). The tops of the banks are flat, with a slope gradient of ≤1º, and there is no clear riparian zone. The vegeta- tion on top of the banks may be described as typically terrestrial Eragrostis plana grass- land and it cannot be visually distinguished from the surrounding grassland. There are no ISSN 0075-6458 43 Koedoe 47/1 (2004) Table 1 Mean canopy cover (M.C.C) of the plant communities identified in the Grassland Biome section of the Olifants River Plant Mean canopy cover (%) Total community Trees Shrubs Dwarf Forbs Grasses M.C.C. number shrubs (%) 1 - - - 3 30 33 2 - - - 10 32 42 3 - - 7 4 26 37 4 - <1 <1 5 50 56 5 - - 1 5 33 39 6 <1 4 2 9 23 38 7.1 - 2 <1 6 39 48 7.2 12 9 2 21 24 68 8 4 <1 <1 12 46 62 Fig. 2. River profile of the Eragrostis plana-Monopsis decipiens grassland Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 43 stones or rocks in the channel bed or on the banks. This grassland is characterised by two diag- nostic species (Table 2). On average, the canopy cover of both the forb Monopsis decipiens and the sedge Cyperus esculentus is <1 %. Dominant grasses are Eragrostis plana, Eragrostis curvula and Cynodon dactylon. The total average canopy cover of this grassland is 33 % (Table 1). 2. The Themeda triandra-Fingerhuthia sesleriiformis grassland The Themeda triandra-Fingerhuthia sesleri- iformis grassland is represented by four relevés and describes the floristics of both the macro-channel banks and the vegetation of the surrounding terrestrial grassland. As is the case with plant community 1, this plant community (Fig. 1) is associated with the Ea Koedoe 47/1 (2004) 44 ISSN 0075-6458 Table 2 Diagnostic species of the Eragrostis plana-Monopsis decipiens - grassland (M.C.C. - mean canopy cover) Species Growth Consistency M.C.C. name form (%) (%) Monopsis decipiens forb 100 <1 Cyperus esculentus grass 50 <1 Land Type (Land Type Series 1985a) at an altitude of about 1 750 m a.s.l., but here the rock is typically dolerite. The soil depth varies between 700 mm and 1 000 mm, and the soil has a high clay content throughout (>55 % clay). This section of the Olifants River has a sin- gle active channel that is < 17 m wide (Fig. 3). The river has cut into the landscape and consequently has steep macro-channel banks of about 2–3 m high. The vegetation above the macro-channel banks may be described as typical terrestrial Themeda triandra grassland on flat plains. The pres- ence of large, isolated sheets of rock charac- terises parts of the active channel bed. The grass species, Fingerhuthia sesleri- iformis, Brachiaria eruciformis, Aristida bipartita, and the forb species, Hemizygia sp., Haplocarpha scaposa, Falckia oblonga, Ledebouria sp., Anthospermum pumilum subsp. rigidum, Scabiosa columbaria and Jamesbrittenia montana are diagnostic for this community (Table 3). Grass species, such as Themeda triandra and Eragrostis plana, and the forb Haplocarpha scaposa are dominant and have the highest average canopy cover. These plant species also occur in the adjacent terrestrial veld. Fingerhuthia sesleriiformis, Brachiaria eruciformis, Cirsi- um vulgare, Tagetes minuta and Cyperus longus var. tenuiflorus are mostly limited to Fig. 3. River profile of the Themeda triandra-Fingerhuthia sesleriiformis grassland. Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 44 the macro-channel banks and active channel bed. A woody component is absent in this grassland community. The total average canopy cover of the Themeda triandra-Fin- gerhuthia sesleriiformis grassland is 42 % (Table 1). 3. The Clutia natalensis-Panicum dregeanum grassland The Clutia natalensis-Panicum dregeanum grassland is represented by two relevés and occurs on both the macro-channel banks and the terrestrial area above the banks. This plant community is situated in the vicinity of the Frischgewaagd farm (Fig. 1). It is a grassland community found where the Olifants River runs through the Bb Land Type (Land Type Series 1985a), associated with the Karoo Sequence and aeolic sand at an altitude of about 1 650 m a.s.l. The deep (≥ 1 200 mm) soil on the macro-channel banks is typically a sand-deposit covering clayey subsoil (>55 % clay). ISSN 0075-6458 45 Koedoe 47/1 (2004) Table 3 Diagnostic spesies of the Themeda triandra-Fingerhuthia sesleriiformis grassland (M.C.C. - mean canopy cover) Species name Growth form Consistency M.C.C. (%) (%) Fingerhuthia sesleriiformis grass 100 3 Hemizygia sp. forb 100 <1 Haplocarpha scaposa forb 75 3 Falcforbia oblonga forb 75 1 Ledebouria sp. forb 75 <1 Brachiaria eruciformis grass 75 1 Anthospermum pumilum subsp. rigidum forb 50 <1 Scabiosa columbaria forb 50 >1 Jamesbrittenia montana forb 50 <1 Aristida bipartita grass 50 2 Fig. 4. River profile of the Clutia natalensis-Panicum dregeanum grassland. Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 45 The macro-channel in this section of the river is, on average, 14 m wide and the steep macro-channel banks are similar to the Themeda triandra-Fingerhuthia sesleri- iformis grassland (Fig. 4). The surrounding grassland above the banks is situated on flat to convex plains, with a clear riparian zone limited to the macro-channel banks. This grassland community is characterised by the presence of a woody component, with the dwarf shrub species Clutia natalensis, Artemisia afra and Gomphocarpus frutico- sus limited to the steep macro-channel banks, while the reed Phragmites australis and forb Persicaria lapathifolia are found at the waterside in the active channel. Koedoe 47/1 (2004) 46 ISSN 0075-6458 Table 4 Diagnostic species of the Clutia natalensis-Panicum dregeanum grassland (M.C.C. - mean canopy cover) Species name Growth Consistency M.C.C. form (%) (%) Panicum dregeanum grass 100 2 Artemisia afra dwarf 100 7 shrub Oenothera erythrosepala forb 100 <1 The grass Panicum dregeanum, forb Oenothera erythrosepala, and dwarf shrub Artemisia afra are diagnostic for this com- munity (Table 4). Dominant dwarf shrub species are Clutia natalensis and Artemisia afra, while Themeda triandra, Eragrostis curvula, Setaria sphacelata var. sphacelata, Andropogon appendiculatus and Hyparrhe- nia tamba mostly dominate the grass layer. The total average canopy cover of this grass- land is 37 % (Table 1). 4. The Heteropogon contortus-Cyperus longus var. tenuiflorus grassland The Heteropogon contortus-Cyperus longus var. tenuiflorus grassland is represented by seven relevés and occurs on both the steep macro-channel banks and the terrestrial area above the banks. This plant community (Fig. 1) is associated with that section of the Olifants River that is situated in the Bb Land Type (Land Type Series 1985a) with sand- stones and shales from the Karoo Sequence. Deep soil (1 000–1 200 mm) is common and the soil texture varies from sandy loam (11 % to 15 % clay) to clayey soil (>55 % clay). The width of the macro-channel varies from 15–27 m (Fig. 5). The river has cut deeply into the landscape, and steep macro-channel Fig. 5. River profile of the Heteropogon contortus-Cyperus longus var. tenuiflorus grassland. Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 46 banks, up to 3.5 m high, are common. The landscape above the banks is flat to convex and has slopes with gradients of 1–8 º. Riparian vegetation is limited to the macro- channel banks. Woody vegetation is mostly limited to rocky areas along the macro-chan- nel banks. The active channel bed has alluvial deposits or sand banks (depositional bars), and large rocks are quite common on both the macro- channel banks and in the active channel. This section of the active channel is charac- terised by pools, and a few rapids. In certain areas mining activities have seriously dis- turbed the adjacent terrestrial vegetation, while planted pastures that stretch down to the macro-channel replaced the natural grassland. Diagnostic species are, inter alia, the grass- es Heteropogon contortus, Eragrostis gum- miflua, Elionurus muticus, Alloteropsis semialata, Aristida congesta subsp. conges- ta, Aristida sp., the forbs Bidens formosa, Acalypha caperonoides, and the fern Pterid- ium aquilinum (Table 5). The grass layer is completely dominated by Themeda triandra, followed by Cynodon dactylon, Aristida sp., patches of Phragmites australis and Hypar- rhenia tamba. Dominant sedges are Cyperus longus var. tenuiflorus and Cyperus lati- ISSN 0075-6458 47 Koedoe 47/1 (2004) Table 5 Diagnostic species of the Heteropogon contortus- Cyperus longus var. tenuiflorus grassland (M.C.C. - mean canopy cover) Species name Growth Constancy M.C.C. form (%) (%) Heteropogon contortus grass 71 2 Eragrostis gummiflua grass 71 <1 Elionurus muticus grass 43 <1 Bidens formosa forb 43 <1 Alloteropsis semialata grass 29 <1 Acalypha caperonioides forb 29 <1 Aristida congesta subsp. congesta grass 29 <1 Aristida sp. grass 29 3 Pteridium aquilinum forb 29 <1 folius. These species, and Phragmites aus- tralis, are, however, limited to the waterside and alluvial deposits (depositional bars) in the active channel. The grasses Brachiaria brizantha, Hemarthria altissima and the forb Persicaria lapathifolia are also associ- ated with the lower-lying, moister sections of the macro-channel banks. The shrub component is represented by Rubus fruticosus, Rhus gerrardii and Diospyros lycioides subsp. sericea. The lat- ter shrub is associated with stony and rocky substrates, similar to the surrounding terres- trial Bankenveld substrates. The average canopy cover of these three species is less than 1 %, while the total average canopy cover of the Heteropogon contortus-Cyper- us longus var. tenuiflorus grassland is 56 % (Table 1). 5. The Eragrostis plana-Cyperus fastigia- tus grassland The Eragrostis plana-Cyperus fastigiatus grassland is represented by five relevés and occurs both on the macro-channel banks and the terrestrial area above the banks. This plant community is associated with the Olifants River in the vicinity of the Mid- delkraal farm and it also occurs in the vicin- ity of the Vandyksdrift farm (Fig. 1). This grassland is associated with the Bb and Fa Land Types (Land Type Series 1985a) at an altitude of about 1 600 m a.s.l. Typical underlying rocks are, inter alia, sandstone, shale, layered mudstone and rhyolite from the Karoo and Transvaal Sequences. The soil depth varies (1 000–1 200 mm), and the soil texture ranges from sandy clay loam to loamy clays throughout this community. Soil texture varies from 21 % to 55 % clay. The macro-channel of this section of the Olifants River is, on average, 16 m wide with a single active channel (Fig. 6). The river typically has steep macro-channel banks. Above ground rock occurs only in a few isolated areas, either on the banks or in the active channel. Locally, alluvial deposits or sand banks (depositional bars) are found in the active channel. Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 47 The grass Bromus catharticus and forbs Cir- sium vulgare and Crabbea acaulis are diag- nostic for this grassland community (Table 6). The grasses Eragrostis planicul- mis, Brachiaria brizantha, Bromus carthar- ticus, the sedge Cyperus fastigiatus and the forbs Rumex crispus and Persicaria lapathi- folia are dominant on the steep macro-chan- nel banks and at the waterside. The grass Eragrostis plana and to a lesser degree Era- grostis curvula and Themeda triandra are dominant above the banks. Only three dwarf shrub species are found in this community and Clutia natalensis has the highest average canopy cover. The dwarf shrubs Gomphostigma virgatum and Gom- phocarpus physocarpus were recorded only in one relevé and are poorly represented. The average total canopy cover of the Eragrostis plana-Cyperus fastigiatus grassland is 39 % (Table 1). 6. The Rhus gerrardii-Hemarthria altissima grassland The Rhus gerrardii-Hemarthria altissima grassland is, as is the case with the Era- grostis plana-Cyperus fastigiatus grassland (community 5), associated with the Bb and Fa Land Types (Land Type Series 1985a), but it occurs at an altitude of about 1 550 m a.s.l. (Fig. 1). The Karoo and Transvaal Sequences represent the geology of this area. Soil depth varies from 600 mm to 1 200 mm with a clay content of higher than 55 %. This vegetation is represented by six relevés. Koedoe 47/1 (2004) 48 ISSN 0075-6458 Fig. 6. River profile of the Eragrostis plana-Cyperus fastigiatus grassland. Table 6 Diagnostic species of the Eragrostis plana-Cyperus fastigiatus grassland (M.C.C. - mean canopy cover) Species name Growth Constancy M.C.C. form (%) (%) Bromus catharticus grass 40 <1 Cirsium vulgare forb 60 <1 Crabbea acaulis forb 40 <1 The macro-channel has a single active chan- nel varying between 19 m and 22 m in width, with pools and rapids succeeding each other (Fig. 7). The river flows through rocky ridges, consequently the above-ground rock cover is high in both the active channel bed and on the macro-channel banks. Stone sizes vary from large stones (> 250 mm) to boul- ders (>1 000 mm). It may be assumed that this high rock cover would play a stabilising role on both the macro-channel banks and in the active channel bed, and that it would Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 48 influence the water flow pattern, and conse- quently, that this section of the river would not be influenced as much as other sections by floods. The riparian zone is characterised by the presence of a shrub component. The exotic, though naturalised, tree Salix babylonica and forbs Galium capense subsp. garipense, Argyrolobium tuberosum, Helichrysum mundtii and Polygala hottentot- ta are diagnostic for this grassland commu- nity (Table 7). The dwarf shrubs, Gom- phostigma virgatum and Gomphocarpus physocarpus, the grasses, Miscanthus junceus, Brachiaria brizantha and Hemarthria altissima and the forb Persi- caria lapathifolia are dominant at the water- side and in the lower-lying parts of the macro-channel bank. Rhus gerrardii is the only shrub recorded. It has an average canopy cover of 3.91 %. The tree Salix babylonica was recorded at only two sites. It has an average canopy cover of ISSN 0075-6458 49 Koedoe 47/1 (2004) Fig. 7. River profile of the Rhus gerrardii-Hemarthria altissima grassland. Table 7 Diagnostic species of the Rhus gerrardii- Hemarthria altissima grassland (M.C.C. - mean canopy cover) Species name Growth Constancy M.C.C. form (%) (%) Galium capense subsp. garipense forb 50 2 Argyrolobium tuberosum forb 67 <1 Helichrysum mundtii forb 33 <1 Polygala hottentotta forb 33 <1 Salix babylonica tree 33 <1 < 1 %. Digitaria eriantha, Brachiaria brizantha, Eragrostis curvula, Hemarthria altissima, Miscanthus junceus, Themeda triandra and Eragrostis plana are dominant in the grass stratum while the forb species Galium capense subsp. garipense, Persi- caria lapathifolia and Tagetes minuta are commonly found. The average total canopy cover of the Rhus gerrardii-Hemarthria altissima grassland is 38 % (Table 1). 7. The Salix mucronata subsp. wilmsii- Eragrostis curvula grassland The Salix mucronata subsp. wilmsii-Era- grostis curvula grassland is represented by six relevés. This plant community is associ- ated with the Bb Land Type (Land Type Series 1985b) at altitudes ranging between Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 49 1 425 m and 1 450 m a.s.l. This grassland is limited to narrow, easily distinguishable dia- base dykes and intrusions crossing the Olifants River, immediately north of the Doringpoort Dam, in the vicinity of the town Witbank (Fig. 1). The shrub Salix mucronata subsp. wilmsii and the grass Ischaemum fasciculatum are diagnostic for this community. The commu- nity is divided into two variants, namely the Paspalum scrobiculatum-Miscanthus junceus variant (variant 7.1) and the Acacia dealba- ta-Eragrostis curvula variant (variant 7.2). The most significant difference in the habi- tats of these two variants lies in soil depth and above ground rock cover. The Paspalum scrobiculatum-Miscanthus junceus variant is found on deep soil with a low rock cover, Koedoe 47/1 (2004) 50 ISSN 0075-6458 Fig. 8. River profile of the Paspalum scrobiculatum-Miscanthus junceus variant. Table 8 Diagnostic species of the Paspalum scrobiculatum- Miscanthus junceus variant (M.C.C. - mean canopy cover) Species name Growth Constancy M.C.C. form (%) (%) Paspalum scrobiculatum grass 100 2 Chamaecrista comosa forb 100 <1 Imperata cylindrica grass 50 7 Plantago virginica forb 50 <1 Ipomoea crassipes forb 50 <1 Cynoglossum lanceolatum forb 50 <1 while the Acacia dealbata-Eragrostis curvu- la variant is found on shallow soil with a high rock cover. There is also a great difference in the vegeta- tion structure of these two variants. The woody component of variant 7.1 is limited to a few dwarf shrub and shrub species with a low average canopy cover, in comparison with the woody component of variant 7.2, which consists of tree, shrub and dwarf shrub species. The tree Acacia dealbata (a declared alien invader,) and shrub Sesbania punicea (a declared weed), dominate the macro-chan- nel banks of variant 7.2 and contribute large- ly to the total canopy cover of the woody component of this variant. However, these two alien problem species are not found in the Paspalum scrobiculatum-Miscanthus junceus variant. There is also a huge differ- ence in the floristic composition between Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 50 these variants that are limited to the diabase intrusions. 7.1 Paspalum scrobiculatum-Miscanthus junceus variant The Paspalum scrobiculatum-Miscanthus junceus variant is represented by two relevés only, and the extent of this unit is limited. The macro-channel is up to 22 m wide and deep, stagnant pools of water are predomi- nant (Fig. 8). The macro-channel banks form flat to convex terraces and the above ground rock cover is limited to a few pebbles in the active channel bed. The sandy clay loam soil is deep (≥ 1 200 mm) and has a clay content of 21 % to 35 %. This variant has six diagnostic species (Table 8), and the grass Paspalum scrobicu- latum and forb Chamaecrista comosa, have been recorded only on the macro-channel banks. The shrub layer of this variant is dominated by Salix mucronata subsp. wilmsii. The shrubs Diospyros lycioides subsp. sericea and Rhus gerrardii are poorly represented in this variant. Hyparrhenia hirta, Paspalum scrobiculatum, Ischaemum fasciculatum, Imperata cylindrica, Setaria sphacelata var. sphacelata, Hemarthria altissima and Mis- canthus junceus dominate the grass layer. The grass species Miscanthus junceus together with the sedge species, Cyperus lat- ifolius, Cyperus marginatus, Cyperus longus var. tenuiflorus are limited to the lower-lying terraces on the macro-channel banks and the areas bordering the active channel. The variant represents the vegetation associ- ated with the macro-channel bank as well as the vegetation above the macro-channel bank. There is no single dominant forb species. All the forb species present have an average canopy cover of less than 1 %. How- ever, the joint, average canopy cover of the forbs is 6 %. Forb species that make the largest contribution are Crabbea acaulis, Berkheya radula, Ipomoea bathycolpos, Ipo- moea crassipes, Helichrysum rugulosum and Tagetes minuta. The total average canopy cover of the Paspalum scrobiculatum-Mis- canthus junceus variant is 48 % (Table 1). 7.2 Acacia dealbata-Eragrostis curvula variant The Acacia dealbata-Eragrostis curvula variant is represented by four relevés. The macro-channel divides in certain areas to form two active channels that are between 23 m and 25 m wide (Fig. 9). These active channels are characterised by shallow pools ISSN 0075-6458 51 Koedoe 47/1 (2004) Fig. 9. River profile of the Acacia dealbata-Eragrostis curvula variant. Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 51 followed by rapids, and by the presence of a few islands on which woody plant species are found. The riparian zone is clearly visi- ble with a strong woody component. The soil on the macro-channel banks is 150 mm to 200 mm deep and the soil texture varies from loamy sand (11 % -15 % clay) to clayey soil (>55 % clay). This section of the river is characterised by a high percentage rock cover. The stones vary from pebbles limited to the active channel bed to large stones and rocks that occur mostly on the macro-channel banks. This variant has 14 diagnostic species (Table 9). The declared alien invader Acacia dealbata and the declared alien weed Sesba- nia punicea have the highest average canopy cover. These species should be removed, since they both grow near the waterside and are a source of seed that may spread to downstream areas. The Acacia dealbata-Eragrostis curvula variant is characterised by the presence of a strong woody component. Dominant woody species are the tree Acacia dealbata, the shrubs Sesbania punicea, Diospyros lycioides subsp. sericea, Salix mucronata subsp. wilmsii, Rhus gerrardii, and the dwarf Koedoe 47/1 (2004) 52 ISSN 0075-6458 Table 9 Diagnostic species of the Acacia dealbata-Eragrostis curvula variant (M.C.C. - mean canopy cover) Species name Growth Constancy M.C.C. form (%) (%) Acacia dealbata tree 75 12 Sesbania punicea shrub 100 2 Deverra sp. forb 75 <1 Acacia karroo dwarf shrub 75 <1 Bidens pilosa forb 50 2 Dicliptera clinopodia forb 75 1 Lepidium virginicum forb 75 <1 Ranunculus multifidus forb 50 <1 Sida rhombifolia dwarf shrub 50 <1 Vangueria cyanescens shrub 50 <1 Maytenus heterophylla shrub 50 <1 Chenopodium ambrosioides forb 50 <1 Typha capensis grass 50 <1 Achyranthes aspera var. aspera forb 50 <1 shrubs Clutia natalensis and Artemisia afra. The grasses Era- grostis plana, Miscanthus junceus, Hemarthria altissima, Eragrostis curvula and Paspalum distichum, and the sedges Cyperus marginatus and Cyperus latifolius dominate the grass layer. The grass species Miscanthus junceus, Hemarthria altissima, Paspalum distichum, Phragmites australis, and sedge species Cyperus mar- ginatus and Cyperus latifolius are mostly limited to the lower-lying parts of the macro-channel bank close to the active channels. The macro-channel banks are dis- turbed in local areas. The high average canopy cover of the opportunistic weedy forbs Tagetes minuta, Verbena bonariensis and Bidens pilosa confirms that disturbance. The high weed cover probably results from the high woody cover and its shadow effect—the alien tree Acacia dealbata, with an average canopy cover of 12 %, being the greatest contributor. The negative shadow effect is clearly visible on the grass layer. The total average canopy cover of the Acacia dealbata-Eragrostis curvula variant is 68 % (Table 1). 8. The Echinochloa crus-galli-Paspalum distichum grassland The Echinochloa crus-galli-Paspalum dis- tichum grassland is represented by four relevés and it occurs south of Witbank Dam in the vicinity of the Wolwekrans farm and downstream of the Doringpoort Dam (Fig. 1). The plant community is associated with the Bb Land Type (Land Type Series 1985b) and is found at altitudes of 1 450 m to 1 500 m a.s.l. The typical underlying geol- ogy is derived from the Loskop Formation and the Karoo Sequence, Group Ecca, and mostly consists of sandstone, shale and con- glomerate. The soil is mostly 1 200 mm deep and varies from sandy clay loam (21 %– 35 % clay) to heavy clayey soil (>55 % clay). Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 52 The width of the macro-channel varies between 26 m and 32 m with steep banks up to 4 m high (Fig. 10). A single active channel has a rocky appearance in certain areas, and divides to form two active channels with an island covered by a dwarf shrub component. The grasses Echinochloa crus-galli and Echinochloa jubata, the sedge Juncus effusus, the forb Cotula anthemoides and the alien tree Morus alba are diagnostic for this grassland community (Table 10). The tree component of this community is represented by Acacia dealbata, Morus alba and Rhus lancea. However, the latter two species have low average canopy covers. The woody species have a low constancy throughout. The grasses Eragrostis plana, Phragmites australis, Setaria sphacelata var. sphacelata, Echinochloa jubata, Cyperus latifolius and Paspalum distichum and the sedge Cyperus fastigiatus dominate the grass layer, while Verbena bonariensis, Persicaria lapathifolia ISSN 0075-6458 53 Koedoe 47/1 (2004) Fig. 10. River profile of the Echinochloa crus-galli-Paspalum distichum grassland. Table 10 Diagnostic species of the Echinochloa crus-galli- Paspalum distichum grassland (M.C.C. - mean canopy cover) Species name Growth Constancy M.C.C. form (%) (%) Echinochloa crus-galli grass 75 <1 Echinochloa jubata grass 50 10 Juncus effusus grass 50 <1 Morus alba tree 50 <1 Cotula anthemoides forb 50 <1 and Tagetes minuta dominate the forb layer. The sedges Cyperus latifolius and Cyperus fastigiatus, the reed, Phragmites australis, the grasses Echinochloa jubata, Paspalum distichum and Hemarthria altissima and the forb Persicaria lapathifolia are mostly limit- ed to the watersides of the macro-channels. Phragmites australis is also associated with the active channel bed and islands. The total average canopy cover of the Echinochloa crus-galli-Paspalum distichum grassland is 62 %, with grass species making the largest contribution (Table 1). Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 53 Discussion and conclusion The spatial scale of 1:250 000 at which the stratification and sampling took place, caused the vegetation on the macro-channel banks and the vegetation on top of the banks to be seen and described as a single unit. The species composition, representing both macro channel and terrestrial vegetation, therefore, may seem strange, unless it is taken into account that specific plant species are associated with the lower-lying sections bordering the waterside and that other plant species are associated with the banks or even on top of the banks. The locality and distrib- ution of the species were however recorded and described. The riparian vegetation of the macro-channel in the Grassland Biome is not clearly dis- tinguishable from the surrounding vegeta- tion and is mainly represented by terrestrial grass species. A few indigenous woody species occur in the rocky areas associated with the macro-channel. Floristic differences occur within and between communities due to changes in habitat but also as a result of different land use practices and intensities of utilisation. In certain areas the river was diverted due to mining activities and the nat- ural vegetation replaced with planted pas- tures. The alteration of riverbanks and the occurrence of exotic woody species in cer- tain areas modify the stabilising role of the indigenous vegetation and influence hydro- logical processes. Acacia dealbata and the declared alien weed Sesbania punicea pose a threat to the biodiversity of the macro-chan- nel and should be removed. These species grow near the waterside and are a source of seed, spreading to the downstream areas. Although time consuming and costly, it is important to obtain geo-referenced baseline data for these longitudinally narrow ecosys- tems, in order to manage and monitor vege- tation change over time. Acknowledgements This project was made possible by funds donated by the Water Research Commission and the use of the facilities at the Agricultural Research Council. References ACOCKS, J.P.H. 1988. Veld types of Southern Africa. (3rd. ed) Memoirs of the botanical Survey of South Africa 57:1–146. LAND TYPE SERIES. 1985a. Land Type Map 2628 East Rand (1:250 000). Pretoria: Government Printer. LAND TYPE SERIES. 1985b. Land Type Map 2528 Pretoria (1:250 000). Pretoria: Government Printer. MYBURGH, W.J. 2000. Oewerplantegroei van die Olifantsriviersisteem—'n Ekologiese perspek- tief. Water Research Commission. ISBN No 186845 5688. MYBURGH, W.J. 2001. Die ekologie en plante- groeibestuur van die Olifantsriviersisteem. PhD thesis, University of Pretoria, Pretoria. NAIMAN, R.J., H. DECAMPS, J. PASTOR & A.C. JOHN- STON. 1988. The potential importance of bound- aries to fluvial ecosystems. Journal of the North American Benthological Society 7:289–306. NILLSON, C., A. EKBLAD, M. DYNESIUS, S. BACKE, M. GARDFJELL, B. CARLBERG, S. HELLQVIST & R. JANSSON. 1994. A comparison of species rich- ness and traits of riparian plants between a main river channel and its tributaries. Journal of Ecol- ogy 82:281–295. OLIFANTS RIVER FORUM. 1995. Save water—or face wars. On track (October/November): 60–61. RUTHERFORD, M.C. & R.H. WESTFALL. 1986. Bio- mes of Southern Africa—an objective catego- rization. Memoirs of the botanical Survey of South Africa 54: 45–65. TOWNSEND, C.R. 1989. The patch dynamics concept of stream community ecology. Journal of the North American Benthological Society 8:36–50. VANNOTE, R.L., G.W. MINSHALL, K.W. CUMMINS, J.R. SEDELL & C.E. CUSHING. 1980. The river continuum concept. Canadian Journal of Fish- eries and Aquatic Science 37:130–137. WARD, J.V. & J.A. STANFORD. 1987. The ecology of regulated streams: past accomplishments and directions for future research. Pp. 391–409. In: Craig, J.F. & J.B. Kemper (eds.). Regulated Streams. Advances in Ecology. New York: Plenum Press. WESTFALL, R.H. & M.D. PANAGOS. 1988. The Plant Number Scale—an improved method for cover estimation using variable belt transects. Bothalia 18(2): 289–291. WESTFALL, R.H., G.K. THERON & N. VAN ROOYEN. 1997. Objective classification and analysis of vegetation data. Plant Ecology 132: 137–154. Koedoe 47/1 (2004) 54 ISSN 0075-6458 Myburgh & bredenkamp.qxd 2004/04/08 01:57 Page 54