van staden woud.qxd A floristic analysis of forest and thicket vegetation of the Marakele National Park P.J. VAN STADEN & G.J. BREDENKAMP Van Staden, P.J. and G.J. Bredenkamp. 2006. A floristic analysis of forest and thicket vegetation of the Marakele National Park. Koedoe 49(1): 15–32. Pretoria. ISSN 0075- 6458. One of the major plant communities identified in the Marakele National Park was for- est. It became clear that this major forest community contained various forest and thick- et communities. Relevés compiled in the forest were classified by TWINSPAN and Braun-Blanquet procedures identified six communities that are hierarchically classified. The forests dominated by Podocarpus latifolius and Widdringtonia nodiflora represent Afromontane Forests, whereas the Buxus macowanii-dominated dry forests and Olea europaea subsp. africana represent Northern Highveld Forests. A further group of com- munities represent thickets on termitaria with floristic affinities to both savanna and for- est. The floristic composition and relationships of the forest and thicket communities are discussed. Key words: Afromontane forest, Afrotemperate forest, classification, National Park, ter- mitaria, thicket, TWINSPAN. PJ van Staden, Centre for Wildlife Management, University of Pretoria, Pretoria, 0001 Republic of South Afirca; GJ Bredenkamp , Department of Botany, University of Pre- toria, Pretoria, 0001 Republic of South Afirca (george.bredenkamp@.up.ac.za). ISSN 0075-6458 15 Koedoe 49/1 (2006) Introduction In an overview of the vegetation of Marakele National Park, Van Staden & Bredenkamp (2005) recognised five major plant commu- nities, of which one represents forests. These forests are mostly restricted to moist, shel- tered valleys, and represent typical Marakele Afromontane Forest (Von Maltitz 2003). Although Du Preez et al. (1991) classified this type of forest as typical Afromontane Forests, the Marakele forests exist under much drier conditions than in the Drakens- berg (Von Maltitz 2003). These forests occur in specific niches in deep valleys, protected gorges and ravines along the eastern and western slopes of the Drakensberg mountain range. Indeed, some of the Marakele forests are very dry and represent a very rare and restricted plant community classified by Von Maltitz (2003) under Northern Highveld Forests. Van Staden (2002) indicated that various plant communities occurred within these forests, though they are not yet described. This paper aims to describe the forest plant communities of the Marakele National Park, to provide an inventory of the biodiversity of the plant community at the plant species level. Study area The study area was described in detail by Van Staden (2002) and Van Staden & Bre- denkamp (2005), and are therefore only briefly mentioned here. The study area cov- ers 290.51 km² in the southwestern part of the Limpopo Province between 27°30'E– 27°45'E and 24°15'S–24°30'S. Marakele National Park is situated mainly in the Water- berg Moist Mountain Bushveld (Low & Rebelo 1996), in the Savanna Biome (Rutherford & Westfall 1994). The vegeta- tion of the study area includes Acocks’ van staden woud.qxd 2006/04/17 08:38 PM Page 15 (1988) Sour Bushveld (Veld Type 20), Mixed Bushveld (Veld Type 18), Sourish Mixed Bushveld (Veld Type 19) and North- Eastern Mountain Sourveld (Veld Type 8). The underlying parent rock of the biggest part of the study area consists of sandstone of the Kransberg Subgroup, Sandriviersberg Formation. Sandstone of the Matlabas Sub- group, Aasvoëlkop Formation, occurs in the southwestern and southern parts; shale and mudstone of the Matlabas Subgroup Aasvoëlkop Formation, Groothoek Mud- stone Member; a conglomerate outcrop of the Matlabas Subgroup, Aasvoëlkop Forma- tion are found in the western part. These rock formations resulted in the complex mountain topography of the Waterberg. Within this complex topography, the forests are restrict- ed to the relatively moist kloofs, mostly with small rivers. The soils that have developed on the parent materials range from shallow sandy to deep sandy soils on sandstone and clayey soils on diabase and mudstone. The rainfall varies from 556 mm to 630 mm per annum and occurs mainly during the summer months. This is considerably lower than the rainfall limits described for forests in summer rainfall area by Rutherford & Westfall (1994). The study area experiences warm, wet summers with temperatures of up to 32 ºC and cool dry winters with frost in the low-lying areas. Methods Analysis In the overview of the vegetation of the Marakele National Park, Van Staden (2002) used 130 sample plots, located in a stratified random manner in physiographically and physiognomically homoge- neous units, delineated on 1:50 000 scale aerial pho- tographs (Bredenkamp & Theron 1978; Westfall 1981; Gertenbach 1987). The number of sample plots for each delineated physiographic-physiog- nomic unit was determined according to the size of each delineated unit. Termitaria were not included in the placing of the sample plots. Although they could be recognised on the aerial photographs, they were too small to delineate. Additional sample plots were subjectively placed within the termitaria vegetation. For the purpose of this study, a standard sample plot size of 10 m x 20 m was fixed and used throughout the study area. This plot size is considered adequate for surveys in savanna vegetation by Coetzee (1975), Coetzee et al. (1976), Westfall (1981), Van Rooyen (1983) and Gertenbach (1987). At each sample plot a list is compiled of all the species that occur. A cover-abundance value was given to each species according to the cover-abun- dance scale, used by Braun-Blanquet and given by Mueller-Dombois & Ellenberg (1974) and Werger (1974), and modified by Barkman et al. (1964): 5 - Any number of plants, with cover >75 % of the sample plot. 4 - Any number of plants, with cover > 50–75 % of the sample plot. 3 - Any number of plants, with cover > 25–50 % of the sample plot. 2a - Covering between 5–12 % of the sample plot area independent of abundance (indicated as A in the tables). 2b -Covering between 13–25 % of the sample plot area independent of abundance (indicated as B in the tables). 1 - Numerous, with cover of 5 % or less. + - Individuals with cover of < 1 %. r - Usually a single individual with a cover of < 1 %. Although Van Staden (2002) recorded various habi- tat variables, only the following habitat classes are used in this study (see Table 1): The altitude of each sample plot was recorded using an altimeter and is given in meters. The slope of the terrain of each sample plot was measured in degrees, using an optical clinometer. The following classification of slope units (Westfall 1981), were used in this study: Symbol Description Class L level 0.00° - 3.49° G gentle 3.50° - 17.62° M moderate 17.63°- 36.39° S steep 36.40° The aspect of the terrain where each sample plot is situated was determined using a compass. Aspect is given in the eight compass directions, namely: N - North S - South NE - Northeast SW - Southwest E - East W - West SE - Southeast NW - Northwest Koedoe 49/1 (2006) 16 ISSN 0075-6458 van staden woud.qxd 2006/04/17 08:38 PM Page 16 ISSN 0075-6458 17 Koedoe 49/1 (2006) The surface rock cover in each sample plot was esti- mated as a percentage of stones (> 20 mm diameter), boulders and rocky outcrops. The following five classes were used, based on its potential influence on mechanical use (ploughing) (Van der Meulen 1979; Westfall 1981): Symbol Class Description O < 1 % No limitation on mechanical utilisation L 1 - 4 % Low limitation on mechanical utilisation M 5 - 34 % Moderate limitation on mechanical utilisation H 35 - 84 % High limitation on mechanical utilisation V 85 - 100 % No mechanical utilisation possible Land type was read from the Land Type maps. Synthesis The full dataset was captured on the mainframe com- puter of the University of Pretoria, in the BBNEW software package. It was then exported to be used in the software package BBPC (Bezuidenhout et al. 1996). After a TWINSPAN classification (Hill 1979), the output of the resulting classification was import- ed into a spreadsheet, for refinement by Braun-Blan- quet procedures (Behr & Bredenkamp 1988). The final classification of the relevés was then interpret- ed for identification of major communities for the entire park. The relevés representing the forests and thickets were then extracted, subjected to TWINSPAN (Hill 1979) and exported to be used in the software package BBPC (Bezuidenhout et al. 1996) for refinement by Braun-Blanquet procedures. The results are summarised in a phytosociological table (Table 1). Results The phytosociological classification of forests and thickets in the study area resulted in the identification of five forest communi- ties with one of the communities having two sub-communities. Olea europaea subsp. africana-Diospyros whyteana Major Community Van Staden & Bredenkamp (2005) identified the forest and thicket vegetation that occur on the Marakele National Park as the Olea europaea subsp. africana-Diospyros whyteana Major Community. Obvious variation in plant species composition and dominance was observed within this major community. The floristic composition of the Olea europaea subsp. africana-Diospyros whyteana Major Community is given in Table 1. Van Staden & Bredenkamp (2005) described this major community as having a large number of diagnostic species, all being trees or shrubs, which distinguish this forest vege- tation from any other plant community occurring in the park. These diagnostic species included Pappea capensis, Mimu- sops zeyheri, Olea europaea subsp. africana, Cussonia paniculata, Diospyros whyteana, Euphorbia ingens, Rhus leptodictya, Cryp- tolepis transvaalensis, Podocarpus lati- folius, Maytenus undata, Zanthoxylum capense, Myrsine africana, Canthium gilfil- lanii, Ficus sur, Calpurnia aurea, Olea capensis and Grewia occidentalis. Although different plant communities can be recognised within these forests and bush clumps, certain species, e.g., Olea europaea subsp. africana, Ficus sur, Cryptolepis transvaalensis, Cussonia paniculata and Euphorbia ingens were prominent through- out the range of this major community. The forest communities occur in the kloofs, or as thickets in bush clumps on south and east-facing slopes and also as dense thickets on termitaria on the plains. The kloofs are the most sheltered of the geomorphology classes found in the study area, with water in the watercourses draining down the kloofs. These kloofs are also protected against the veld fires that occur frequently in the adja- cent savanna or grassland vegetation. Geldenhuys (1994) described a similar situa- tion for forests in the Western Cape. Coetzee (1975) described a related commu- nity from the Rustenburg Nature Reserve as Hypoestes verticillaris-Mimusops zeyheri Forests, and Westfall (1981) described a sim- ilar community from the Farm Groothoek in the Waterberg as Kloof Forest Communities on moderately deep soils in moist, sheltered van staden woud.qxd 2006/04/17 08:38 PM Page 17 Koedoe 49/1 (2006) 18 ISSN 0075-6458 Ta bl e 1 P hy to so ci ol og ic al t ab le o f t he fo re st s an d bu sh c lu m ps o n M ar ak el e N at io na l P ar k van staden woud.qxd 2006/04/17 08:38 PM Page 18 ISSN 0075-6458 19 Koedoe 49/1 (2006) Ta bl e 1 (c on tin ue d) van staden woud.qxd 2006/04/17 08:38 PM Page 19 Koedoe 49/1 (2006) 20 ISSN 0075-6458 Ta bl e 1 (c on tin ue d) van staden woud.qxd 2006/04/17 08:38 PM Page 20 ISSN 0075-6458 21 Koedoe 49/1 (2006) Ta bl e 1 (c on tin ue d) van staden woud.qxd 2006/04/17 08:38 PM Page 21 Koedoe 49/1 (2006) 22 ISSN 0075-6458 Ta bl e 1 (c on tin ue d) van staden woud.qxd 2006/04/17 08:38 PM Page 22 ISSN 0075-6458 23 Koedoe 49/1 (2006) habitats. Du Preez et al. (1991) classified this type of forest as typical Afromontane Forests, where these forests occur in specific niches in deep valleys, protected gorges and ravines along the eastern and western slopes of the Drakensberg mountain range. Coetzee et al. (1976) described termitaria bush clump communities from the Nylsvley Nature Reserve. These have diagnostic species sim- ilar to the bush clump communities on ter- mitaria in the Marekele National Park. In the hierarchical phytosociological classifi- cation, the following plant communities are classified under the Olea europaea subsp. africana-Diospyros whyteana Major Com- munity (Table 1): 1 Olea capensis-Cheilanthes viridis Forest 1.1 Podocarpus latifolius-Curtisia dentata Forest 1.1.1 Widdringtonia nodiflora- Podocarpus latifolius Moist Short Forest. 1.1.2 Podocarpus latifolius-Rothmannia capensis Moist Tall Forest. 1.2 Buxus macowanii-Kirkia wilmsii Dry Low Forest. 2 Pappea capensis-Ziziphus mucronata Thickets and Bush Clumps 2.1 Rhus leptodictya-Mimusops zeyheri Termitarium Thickets. 2.1.1 Rhus leptodictya-Carissa bispinosa sub-community 2.1.2 Rhus leptodictya-Berchemia zeyheri sub-community 2.2 Olea europaea subsp. africana- Calpurnea aurea Tall Closed Woodland. A dendrogram showing the habitat relation- ships of the various plant communities is shown in Fig. 1. 1. Olea capensis-Cheilanthes viridis Forest This community represents all the forests that occur in moist kloofs or sheltered val- leys in the Marakele National Park. Based on species composition, this vegetation can be regarded as Afromontane Forest (White 1978). According to the classification of Fig. 1. A dendrogram showing the habitat relationships of the plant communities classified under the Olea europaea- subsp. africana-Diospyros whyteana Major Community. Olea europaea- Diospyros whyteana Major Community 1. Dense Forest Pristine Ib Land Type Shallow lithosols > 50 % Rock cover 2. Dense Thicket Degraded Mostly Ad Land Type Mostly Yellow Apedal Soils 0-30 % Rock Cover 1.1 Moist & Cool Perennial/Seasonal Streams Southern & Eastern Slopes 1.1.1 Steep Ravines 1.1.2 Sandstone Slopes 2.1.1 Large Bush Clumps 25 m diameter 2.1.2 Small Bush Clumps 11 m diameter 2.1 Flat Plains Ad Land Type Deep Yellow Apedal Soils Termitaria 0 % Rock Cover 1.2 Dry & Warm Seasonal Stream Northern & Western Slopes 2.2 Southern Slopes Ad & Ib Land Types Shallow Rocky Soils 30 % Rock Cover van staden woud.qxd 2006/04/17 08:38 PM Page 23 Koedoe 49/1 (2006) 24 ISSN 0075-6458 South African forests by Von Maltitz (2003), the forests of the Waterberg area represent Marakele Afromontane Forests and Northern Highveld Forests, both these types being classified under the Northern Afrotemperate Group of forests. The Olea capensis- Cheilanthes viridis Forest represents a north- western island of Afromontane (White 1978) or Afrotemperate (Von Maltitz 2003) forests. Within the study area this community is characterised by species group E (Table 1), with the trees Olea capensis and Canthium gilfillanii and the ferns Cheilanthus viridis and Pellaea calomelanos as the diagnostic species. Olea capensis, in particular, can be regarded as a typical forest species, though the other three species may occur wide- spread on rocky areas of the Waterberg and Bankenveld (Bredenkamp & Brown 2003; Van Staden & Bredenkamp 2005). These forests can be subdivided into Podocarpus latifolius-Curtisia dentata For- est, restricted to moist, sheltered valleys, and representing typical Marakele Afromontane Forest (Von Maltitz 2003), and Buxus macowanii-Kirkia wilmsii Dry Low Forest, a very specialized dry forest, representing a very rare and restricted plant community classified by Von Maltitz (2003) under Northern Highveld Forests. 1.1 Podocarpus latifolius-Curtisia dentata Forest These forests represent the typical moist short to tall forests of the Waterberg area, Tall Forest is found at 1300 m–1730 m a.s.l. along perennial or seasonal streams and moist kloofs. This forest community is a rep- resentative of North-Eastern Mountain Sourveld (Acocks 1988), Afromontane For- est (White 1978) and typical Marakele Afromontane Forest (Mucina & Van Staden 2003), classified under the Northern Afrotemperate Forest Group (Von Maltitz 2003). This is a local forest type found only on the Waterberg, Marakele National Park (Van Staden 2002) and some kloofs of Magaliesberg (Coetzee 1975). These Northern Afrotemperate Forests are, according to Von Maltitz (2003), restricted to mountains and low ridges (Strydpoortberg, Waterberg, Pilanesberg, Witwatersrand, Magaliesberg, Suikerbosrand, Sekhku- khuneland) interrupting the relatively flat northern highveld grassland or bushveld. The forests are found in kloofs on the north- ern and eastern flanks of the Drakensberg. The diagnostic species of Podocarpus lati- folius-Curtisia dentata Forest are listed under species group C, with the trees Podocarpus latifolius, Curtisia dentata, Rothmannia capensis, Faurea saligna, Rhus dentata, Myrsine africana and Ochna holstii, the liana Secamone alpinii and the sedge Cyperus albostriatus as diagnostic species. The tall-growing Podocarpus latifolius is mostly the dominant species, locally reach- ing a cover of up to 70 %. 1.1.1 Widdringtonia nodiflora-Podocarpus latifolius Short Forest Typical: relevé 130 Habitat The Widdringtonia nodiflora-Podocarpus latifolius Short Forest is found at high alti- tudes, 1500 m–1730 m a.s.l. along perennial streams and deep, sheltered, moist kloofs, on gentle to moderate (6–22°) east and south- facing sandstone slopes within the Ib Land Type. The soils are of the Mispah Form derived from sandstone of the Sand- riviersberg Formation, and the soil is rela- tively shallow (< 500 mm) (Land Type Sur- vey Staff 1988). Although this community is not totally pro- tected from fire, the woody vegetation and structure are seldom damaged due to the poorly developed grass layer and the high percentage rock cover that provides shelter against fire. Large rocks cover more than 76 % of the soil surface. During the rainy season and for some time thereafter, water constantly seeps from the sandstone, creat- ing an ideal moist habitat where this com- munity is typically found. van staden woud.qxd 2006/04/17 08:38 PM Page 24 ISSN 0075-6458 25 Koedoe 49/1 (2006) Floristics and structure The Widdringtonia nodiflora-Podocarpus latifolius Short Forest is differentiated by the following diagnostic plant species (species group A, Table 1): the woody Widdringtonia nodiflora, Pittosporum viridiflorum, Ilex mitis, Halleria lucida, Syzygium cordatum and Cliffortia linearifolia; the shrubby Clu- tia pulchella and Phylica paniculata; the ferns Blechnum giganteum, Pteridium aquil- inum, Asplenium splendens and Cheilanthus hirta; and the forb Gerbera piloselloides. This community is represented by five relevés and an average of 19 species was recorded per sample plot. The tree layer is on average 8.4 m tall with an average canopy cover of 79 % (Table 2). Widdringtonia nodiflora is totally dominant, whereas Podocarpus latifolius may locally be very prominent. Apart from the diagnostic woody species, other prominent trees are Curtisia dentata, Canthium gilfillanii (species group C), and Olea capensis (species group E). The latter species has a wider distribution within various forest com- munities in the park (Table 1). The shrub layer is on average 1.4 m tall, with an average canopy cover of 22 % (Table 2). Prominent shrubs occurring in this commu- nity are Myrsine africana and Ochna holstii (species group C) (Table 1). The woody liana Secamone alpinii (species group C) is fre- quently found in the tree and shrub strata. The herbaceous layer has an average height of 0.5 m and an average canopy cover of 11.0 % (Table 2). It is dominated by the diag- nostic ferns Blechnum giganteum, Pteridium aquilinum and Asplenium splendens (species group A) and the sedge Cyperus albostriatus (Species Group C) as ground cover in the more open areas. The tree fern, Cyathea dregei, which was not recorded in any sam- ple plots, also occurs as isolated individuals in the forests in the study area. General The Widdringtonia nodiflora-Podocarpus latifolius Short Forest has many characteris- tic species in common with the Ilex mitis- Pittosporum viridiflorum Forest Community described by Coetzee (1975) and the Kloof Forest Communities described by Westfall (1981). This community occupies the most mesic habitats of sheltered parts of kloofs on the eastern side of the Waterberg Mountain. The dominant woody species Widdringtonia nodiflora and Podocarpus latifolius are, however, absent from the forest communities described by Coetzee (1975). Communities 1.1.1 and 1.1.2 are floristically related through the mutual presence of the species of species group B, emphasised especially by Podocarpus latifolius, Curtisia dentata, Rothmannia capensis and Myrsine africana. These forests are also remotely related to Buxus macowanii-Kirkia wilmsii Low Forest (Community 3), through species group E (Table 1). 1.1.2 Podocarpus latifolius-Rothmannia capensis Tall Forest. Typical: relevé 109 Habitat The Podocarpus latifolius-Rothmannia capensis Tall Forest is found at 1300 m– 1660 m a.s.l. along perennial or seasonal streams and moist kloofs on sandstone slopes (4–30°), facing southeast, north, northeast and east. It is represented by five relevés and an average of 26 species was recorded per sample plot. This forest com- munity is a representative of Acocks’ (1988) North-Eastern Mountain Sourveld, with the structure as a tall forest (Edwards 1983) and it all occurs in the Ib Land Type (Land Type Survey Staff 1988). During the rainy season and for some time thereafter water constantly seeps from the sandstone, creating an ideal moist habitat, characteristic for this community. The soils are of the Mispah Form derived from sand- stone of the Sandriviersberg Formation, and the average soil depth is relatively shallow (< 500 mm) (Land Type Survey Staff 1988). Large rocks cover more than 70 % of the soil surface. van staden woud.qxd 2006/04/17 08:38 PM Page 25 Koedoe 49/1 (2006) 26 ISSN 0075-6458 Floristics and structure The Podocarpus latifolius-Rothmannia capensis Tall Forest is differentiated by the diagnostic following plant species (species group B, Table 1): the woody Combretum moggii, Strychnos usambarensis, Pterocelastrus echinatus, Celtis africana, Tetradenia bre- vispica, Solanum giganteum, Dovyalis zey- heri, Englerophytum magalismontanum, Pavetta gardeniifolia, Diospyros lycioides and Nuxia congesta; the herbaceous Aspara- gus setosum; and the grass Oplismenus hirtellus. The tree layer is on average 12 m tall with an average canopy cover of 79 %. Other promi- nent trees are Pittosporum viridiflorum (species group A), Podocarpus latifolius and Curtisia dentata (species group C), Cussonia paniculata, Ficus thonningii, Olea europaea subsp. africana, Euphorbia ingens, Ficus sur (species group K), Mimusops zeyheri (species group J), and Olea capensis (species group E). The tree fern Cyathea dregei, which was not recorded in the sample plots, also occurs in this community. The shrub layer is on average 1.7 m tall with an average canopy cover of 49 %. Apart from the diagnostic species, other prominent shrubs occurring in this community are Myr- sine africana (species group C), Canthium gilfillanii (species group E), and Diospyros whyteana and Grewia occidentalis (species group K). The woody liana Secamone alpinii and Cryptolepis transvaalensis are frequent- ly found in the tree and shrub strata. The herbaceous layer has an average height of 0.5 m and an average canopy cover of 22 %. It is dominated by Cyperus albostria- tus and Oplismenus hirtellus. General This community is closely related to the Celtis africana-Osyris lanceolata Kloof For- est described by Westfall (1981). These two communities have many distinctive habitat features and many, distinctive species in common. The Podocarpus latifolius-Roth- mannia capensis Tall Forest is floristically related to the Widdringtonia nodiflora- Podocarpus latifolius Short Forest, both being classified under the Podocarpus lati- folius-Curtisia dentata Forest. 1.2 Buxus macowanii-Kirkia wilmsii Low Forest Typical: relevé 64 Habitat The Buxus macowanii-Kirkia wilmsii Low Forest is found at 1110 m–1250 m a.s.l. The Buxus macowanii-Kirkia wilmsii Low Forest is restricted to hot, dry kloofs of north fac- ing, east and northwest-facing slopes (2– 29°) of the study area (Fig. 1) This low for- est occupies a relatively drier and warmer habitat than the Podocarpus latifolius-Cur- tisia dentata Forest. The structure of this forest community is a low forest (Edwards 1983) and it occurs in the Ib Land Type (Land Type Survey Staff 1988). The soils are of the Mispah Form derived from sandstone of the Sandriviers- berg Formation, and the soil is relatively shallow (< 500 mm) (Land Type Survey Staff 1988). Rocks cover more than 50 % of the soil surface. Floristics and structure The Buxus macowanii-Kirkia wilmsii Low Forest is represented by four relevés and an average of 23 species was recorded per sam- ple plot. It is differentiated by the following woody plant species (species group D): Buxus macowanii, Kirkia wilmsii, Apodytes dimidiata, Ochna pretoriensis, Croton gratissimus, Euphorbia turicalli, Acacia ataxacantha and Heteromorpha trifoliata. The tree layer is on average 7.8 m tall, with an average canopy cover of 35 %. Other prominent trees are Mimusops zeyheri, Berchemia zeyheri, Dombeya rotundifolia (species group J) and Olea europaea subsp. africana, Cussonia paniculata and Euphor- bia ingens (species group K). The shrub layer is on average 2.3 m tall, with an average canopy cover of 84 %. Prominent shrubs occurring in this community are van staden woud.qxd 2006/04/17 08:38 PM Page 26 ISSN 0075-6458 27 Koedoe 49/1 (2006) Ochna pretoriensis, Croton gratissimus, Euphorbia tirucallii and Acacia ataxacantha (species group D), Diospyros whyteana, and and Heteropyxis natalensis (species group K). The herbaceous layer has an average height of 0.3 m and an average canopy cover of 7 %. The herbaceous layer is poorly repre- sented and is dominated by the xerophytic ferns Cheilanthes viridis and Pellaea calo- melanos (species group E). General Westfall (1981) described an Erythrina lysis- temon-Celtis africana Kloof Forest with Buxus macowanii as a conspicuous woody species with more than 5 % mean canopy cover and occurring in more than 50 % of the relevés. This community is similar to the community found in the study area. These shrubs are often gregarious, forming pure stands (Table 1). The distribution of the tree/shrub, Buxus macowanii, is restricted to warm valleys and coastal dunes in the East- ern Cape, with the Limpopo Province popu- lation as an unexpected outlier (Coates- Palgrave 1983). This could indicate a former Afromontane forest biogeographic connec- tion via the Waterberg and Stydpoort Moun- tains to the northeastern escarpment and also to the Blouberg and western Soutpansberg. The Buxus macowanii-Kirkia wilmsii Dry Low Forest, a very specialised dry forest, representing a very rare and restricted plant community classified by Von Maltitz (2003) under Northern Highveld Forests. According to Von Maltitz (2003), these latter forests are restricted to Suikerbosrand (Bredenkamp & Theron 1978, 1980), Witwatersrand (Behr & Bredenkamp 1988; Ellery et al. 2001), Magaliesberg (Coetzee 1974, 1975), Water- berg (Westfall 1981; Westfall et al. 1985; Van Staden 2002), Sekhukhuneland (Siebert 2001), and some other ridges occurring throughout the western Central Bushveld region (Van Vuuren 1961; Van Vuuren & Van der Schijff 1970; Van der Meulen 1979). These forest patches are imbedded within an ecotone between sourveld grasslands and subtropical savannas of the Central Bushveld. 2. Pappea capensis-Ziziphus mucronata Thickets and Bush Clumps Dombeya rotundifolia subsp. rotundifolia, Berchemia zeyheri and Ziziphus mucronata (species group J) show strong floristic affini- ties between these thickets and the Buxus macowanii-Kirkia wilmsii Dry Low Forest. 2.1 Rhus leptodictya-Mimusops zeyheri Termitarium Thickets Typical: relevé 18 Habitat The Rhus leptodictya-Mimusops zeyheri Ter- mitarium Thickets are found from 1160 m– 1300 m a.s.l. This widespread community occupies an even drier habitat than that of the Buxus macowanii-Kirkia wilmsii Low Forest, though the termitaria are moister than the surrounding savanna (Van der Meulen 1979). They are therefore classified with the forests. The bush clumps on termitaria are restricted to the Ad Land Type on level and gentle slopes (Land Type Survey Staff 1988) (Fig. 1). Before proclamation of Marakele National Park, this community was subjected to heavy grazing by cattle and the field layer varies in height and cover depending on soil salinity and the amount of grazing (Coetzee et al. 1981). The vegetation developed on huge termite mounds to form this specific plant community. The size of these termitaria determines the type of vegetation that occurs. The trees are usually very tall because of the depth and aeration and the fine texture and higher nutrient status of the soil (Coetzee et al. 1976). This community is a representative of Acocks’ (1988) Sour Bushveld, with the structure as a short closed woodland (Edwards 1983) Floristics The Rhus leptodictya-Mimusops zeyheri Ter- mitarium Thickets are differentiated by the following diagnostic plant species (species group G, Table 1): the woody Rhus lepto- van staden woud.qxd 2006/04/17 08:38 PM Page 27 Koedoe 49/1 (2006) 28 ISSN 0075-6458 dictya, Euclea crispa, Grewia flavescens, Flueggia virosa, Ficus thonningii and Pavet- ta zeyheri; the liana Asparagus transvaalen- sis; the grass Eragrostis curvula; the herba- ceous Schistostephium heptalobum; and the weedy annual Tagetes minuta. 2.1.1 Rhus leptodictya-Carissa bispinosa variation Typical: relevé 18 Habitat The Rhus leptodictya-Carissa bispinosa variation is found at 1160 m–1300 m a.s.l. This variation occurs on level ground or/and gentle slopes (1-160º). The soils are yellow, apedal and well drained and relatively deep (> 500 mm), belonging to one of the follow- ing forms: Inanda, Kranskop, Magwa, Hut- ton, Griffin and/or Clovelly (Land Type Sur- vey Staff 1988) (Fig. 1). The average diameter of the Rhus lepto- dictya-Carissa bispinosa variation bush clumps is 25 m. It is represented by seven relevés and an average of 35 species was recorded per sample plot (Table 1). Floristics and structure The Rhus leptodictya-Carissa bispinosa variation is differentiated by the following diagnostic plant species (species group F, Table 1): the woody Carissa bispinosa, Gymnosporia buxifolia, Vangueria infausta, Clerodendrum glabrum, Adenia glauca, Dichrostachys cinerea, Gardenia volkensii, Psiadia punctulata, Aloe marlothii, Acacia karroo, Maerua angolensis and Pterocarpus rotundifolius; the forbs Pavonia burchelli, Sida dregei, Kalanchoe rotundifolia, Schkuhria pinnata, Lippia javanica, Sida cordiifolia, Anthospermum hispidulum, Ipo- moea magnusiana, Hemizygia pretoriae, Solanum panduraeforme, Scadoxus puniceus; and the grasses Panicum maximum, Aristida congesta, Eragrostis rigidior, Eragrostis lehmanniana and Cynodon dactylon. Many of these diagnostic species indicate previous over-utilisation and trampling. The tree layer is on average 7 m tall with an average canopy cover of 79 %. Other promi- nent trees are Pappea capensis and Schotia brachypetala (species group I), Olea capen- sis (species group E), Mimusops zeyheri, Dombeya rotundifolia subsp. rotundifolia, Berchemia zeyheri and Ziziphus mucronata (species group J), and Olea europaea subsp. africana, Ficus sur, Cussonia paniculata and Grewia occidentalis (species group K). The shrub layer is on average 1.4 m tall, with an average canopy cover of 46 %. Prominent shrubs are Grewia flavescens, Flueggia virosa (species group K) and Calpurnia aurea (species group H). The herbaceous layer has an average height of 0.4 m and an average canopy cover of 20 %. It is dominated by the grass Panicum maximum and Eragrostis rigidior, the forbs Pavonia burchelli, Schkuhria pinnata and Lippia javanica (species group F). General Coetzee et al. (1976) described similar com- munities from flat bottomlands with termi- taria at Nylsvley Nature Reserve, namely the Pappea capensis-Acacia tortilis variation which shows strong affinities with the Rhus leptodictya-Carissa bispinosa variation, with the most common plant species being Carissa bispinosa and Pappea capensis. 2.1.2 Rhus leptodictya-Berchemia zeyheri variation Typical: relevé 16 Habitat The Rhus leptodictya-Berchemia zeyheri variation is found at 1250 m–1360 m a.s.l. The soils are the same as for the Rhus lepto- dictya-Carissa bispinosa variation, but here the bush clumps are much smaller, with an average diameter of 11 m, whereas the aver- age diameter of the bush clumps Rhus lepto- dictya-Carissa bispinosa variation is 25 m. Floristics and structure This variation is represented by four relevés and an average of 35 species was recorded per sample plot. Although it seems that no van staden woud.qxd 2006/04/17 08:38 PM Page 28 ISSN 0075-6458 29 Koedoe 49/1 (2006) diagnostic species occur in this variation, it can be distinguished from the Rhus lepto- dictya-Carissa bispinosa variation by the absence of species in the species group F and the presence of species in species group H (Table 1), with Maytenus undata, Calpurnea aurea and Calodendron capense being dif- ferential species. The tree layer is on average 6.5 m tall with an average canopy cover of 66 %. The shrub layer is on average 1.6 m tall with an average canopy cover of 46 %. The herbaceous layer has an average height of 0.4 m and an aver- age canopy cover of 10 %. 2.2 Olea europaea subsp. africana-Calpurnea aurea Tall Closed Woodland Typical releve: relevé 97 Habitat The Olea europaea subsp. africana- Calpurnea aurea Tall Closed Woodland is found from 1180–1500 m a.s.l. in deep kloofs on southwest-facing and south-facing sandstone slopes (2–250º), usually along seasonal streams (Table 1, Fig. 1). This bush clump community is a representative of Acocks’ (1988) Sour Bushveld, with the structure as a tall closed woodland (Edwards 1983). It occurs on the Ad Land Type and the Ib Land Type (Fig. 2.7) (Land Type Survey Staff 1988). The soils are of the Mispah Form derived from sandstone of the Sand- riviersberg Formation (Land Type Survey Staff 1988) and large rocks cover more than 31 % of the soil surface (Fig. 1). Floristics and structure The Olea europaea subsp. africana- Calpurnea aurea Tall Closed Woodland is characterised by the presence of species group H and absence of species group G. No diagnostic species group was identified, though the species of species group H are differential, it also indicates a floristic rela- tionship with the Rhus leptodictya- Berchemia zeyheri variation (community 2.1.2). The differential species are the woody Maytenus undata, Calpurnia aurea, Calo- dendron capense, Gymnosporia polyacan- tha, Vepris lanceolata and Tricalysia lanceo- lata, the herbaceous Asparagus virgatus and the grass Setaria lindenbergiana. The tree layer is on average 10.5 m tall with an average canopy cover of 56 %. Other prominent trees are Celtis africana (species group B); Rhus leptodictya, Euclea crispa, Ficus thonningii (species group G); Pappea capensis, Ziziphus mucronata, Zanthoxylum capense, Schotia brachypetala, Euclea natalensis and Acacia caffra (species group I); Mimusops zeyheri and Berchemia zeyheri (species group J); and Olea europaea subsp. africana, Ficus sur and Euphorbia ingens (species group K). The shrub layer is on average 1.8 m tall with an average canopy cover of 41 %. Cryp- tolepis transvaalensis is prominent. The herbaceous layer has an average height of 0.4 m and an average canopy cover of 21 %. The herbaceous layer is dominated by the grass Setaria lindenbergiana and the forb Hypoestes forskaolii (species group K). General The Olea europaea subsp. africana-Calpurnia aurea Tall Closed Woodland has floristically strong affinities with the Rhus leptodictya- Mimusops zeheri Termitarium Thickets (species group I) and especially the Rhus leptodictya-Berchemia zeyheri variation, indicated by species group H. The habitat difference between these two bush clump communities is that the Olea europaea subsp. africana-Calpurnia aurea Tall Closed Woodland occurs in deep, sheltered kloofs on rocky substrates without termitaria, while the Rhus leptodictya-Berchemia zeyheri variation occurs on termitaria on the plains without rocks. Discussion According to White (1978) the majority of the tree species in the Afromontane forests are very widespread. Amongst them are Podocarpus latifolius, Ilex mitis and Halle- ria lucida that occur in the study area. These van staden woud.qxd 2006/04/17 08:38 PM Page 29 Koedoe 49/1 (2006) 30 ISSN 0075-6458 species could almost be used to define the Afromontane region as a whole. Not one species occurs throughout, but most species of the assemblage are represented on virtual- ly every ‘island’ of Afromontane forest. According to Von Maltitz (2003) the North- ern Highveld Forest type is a relict forest type in a progressed stage of ‘erosion’ from the original Afromontane composition. They are considered remnants of the Magaliesburg Extension (White 1978)—an Afromontane (forest) intrusion linking the forests of Northern Escarpment to those found in shel- tered sites of the Central Bushveld and Northern Highveld regions. These forests are Afrotemperate (Afromontane) in character, however Holocene climate changes have caused their confinement to fire-sheltered habitats and a concomitant decline in forest elements, but an increase in woodland ele- ments (Von Maltitz 2003). This fragmentary Afromontane type forest forms transitions to surrounding savanna and to the riverine forests with Combretum ery- throphyllum (Van Rooyen 1983; Von Maltitz 2003). There is also floristic similarity between the kloof forest communities and thickets found on termitaria. Termitaria are known to increase water infiltration and per- colation, as well as nutrient availability, compared to the surrounding savanna wood- lands (Van der Meulen 1979). Therefore, ter- mitaria do support dense thickets with savan- na species, however some forest species are also present. Low diversity of the woody element places this type close to Drakensberg Montane Forests and Northern KwaZulu-Natal Mist- belt Forests (Von Maltitz 2003). The distribution of the shrub Buxus macow- anii is of biogeographic interest. This species is restricted to coastal forests of the Eastern Cape and forms an outlying population in the Waterberg region (Coates Palgrave 1983). Acknowledgements South African National Parks is thanked for the opportunity to use the offical research project for a MSc degree. References ACOCKS, J.P.H. 1988. Veld types of South Africa, 3rd edn. Memoirs of the botanical Survey of South Africa 57: 1–146. BARKMAN, J.J., J. MORAVEC & S. RAUCHERT. 1964. Kritische Bemerkungen und Vorschläge zur quantitativen Vegetationanalyse. Acta Botanica Neerlandia 13: 394–449. BEHR, C.M. & G.J. BREDENKAMP. 1988. A phytosoci- ological classification of the Witwatersrand National Botanical Garden. South African Jour- nal of Botany 54: 525–533. BEZUIDENHOUT, H., H.C.BIGGS & G.J. BREDENKAMP. 1996. A process supported by the utility BBPC for analysing Braun-Blanquet data on a personal computer. Koedoe 39(1): 107–112. BREDENKAMP, G.J. & L.R. BROWN. 2003. A reap- praisal of Acocks’ Bankenveld: Origin and diversity of vegetation types. South African Journal of Botany 69(1): 7–26. BREDENKAMP, G.J. & G.K. THERON. 1978. A syneco- logical account of the Suikerbosrand Nature Reserve. 1. The phytosociology of the Witwater- srand geological system. Bothalia 12: 513–529. BREDENKAMP, G.J. & G.K. THERON. 1980. A syneco- logical account of the Suikerbosrand Nature Reserve, Part II: The phytosociology of the Ven- tersdorp geological system. Bothalia 13(1&2): 199–216. COATES PALGRAVE, K. 1983. Trees of Southern Africa. 2nd ed. Cape Town: Struik. COETZEE, B.J. 1974. A phytosociological classifica- tion of the vegetation of the Jack Scott Nature Reserve. Bothalia 11: 329 – 347. COETZEE, B.J. 1975. A phytosociological classifica- tion of the Rustenburg Nature Reserve. Bothalia 11: 561–580. COETZEE, B.J., F. VAN DER MEULEN, S. ZWANZIGER, P. GONSALVES & P.J.WEISSER. 1976. A phytoso- ciological classification of the Nylsvley Nature Reserve. Bothalia 12: 137–160. COETZEE, B.J., P. VAN WYK, W.P.D. GERTENBACH, A. HALL-MARTIN & S.C.J. JOUBERT. 1981. 'n Plantekologiese verkenning van die Water- berggebied in die Noord-Transvaalse Bosveld. Koedoe 24:1–23. DU PREEZ, P.J., G.J.BREDENKAMP & H.J.T. VENTER. 1991. The syntaxonomy and synecology of the forests in the eastern Orange Free State, South Africa. I. The Podocarpetalia latifolii. South African Journal of Botany 57: 198–206. van staden woud.qxd 2006/04/17 08:38 PM Page 30 EDWARDS, D. 1983. A Broad-scale structural classifi- cation of vegetation for practical purposes. Bothalia 14 (3 & 4): 705–712. ELLERY, W.N., K. BALKWILL & K. ELLERY, R.A. REDDY. 2001. Conservation of the vegetation on the Melville Ridge, Johannesburg. South African Journal of Botany 67: 261–273. GELDENHUYS, C.J. 1994. Bergwind patches and the location of forest patches in the southern Cape landscape South Africa. Journal of Biogeogra- phy 21: 49–62. GERTENBACH, W.P.D. 1987. 'n Ekologiese studie van die Suidelikste Mopanieveld in die Nasionale Krugerwildtuin. D.Sc proefskrif, Universiteit van Pretoria, Pretoria. HILL, M.O. 1979. TWINSPAN – a Fortran program for arranging multivariate data in an ordered two way table by classification of individuals and attributes. Ithaca, New York: Cornell University. LAND TYPE SURVEY STAFF. 1988. Land types of the maps 2426 Thabazimbi, 2428 Nylstroom. Mem- oirs of the agricultural natural Resoures of South Africa No.10: 1–431. MUCINA, L. & P.J. VAN STADEN. 2003. Marakele Afromontane Forests. Pp. 99–104. In: VON MALTITZ, G. (Project leader). Classification sys- tem for South African indigenous forests: An objective classification for the Department of Water Affairs and Forestry. Pretoria: Council for Scientific and Industrial Research. (Environ- mentek report ENV-P-C 2003-017). MUELLER-DOMBOIS, D. & H. ELLENBERG. 1974. Aims and Methods of Vegetation Ecology. New York: Wiley. RUTHERFORD, M.C. & R.H. WESTFALL. 1994. Bio- mes of southern Africa: An objective categoriza- tion. Memoirs of the Botanical Survey of South Africa 63: 1–94. SIEBERT, S.J. 2001. Vegetation of the ultramafic soils of the Sekhukhuneland Centre of Endemism. Ph.D. thesis, Universty of Pretoria, Pretoria. VAN DER MEULEN, F. 1979. Plant sociology of the Western Transvaal Bushveld. South Africa: Syn- taxonomic and synecological study. Vaduz: Cramer. (Dissertationes Botanica 49.) VAN ROOYEN, N. 1983. Die Plantegroei van die Roodeplaatdam-natuurreservaat II. Die plantge- meenskappe. South African Journal of Botany 2: 115–125. VAN STADEN, P. J. 2002. An Ecological Study of the Plant Communities of the Marakele National Park. M.Sc. thesis, University of Pretoria, Preto- ria. VAN STADEN P.J. & G.J. BREDENKAMP. 2005. Major plant communities of the Marakele National Park. Koedoe 48(2): 59–70. VAN VUUREN, D.R.J. 1961. 'n Ekologiese studie van 'n noordelike en suidelike kloof van die Maga- liesberge. M.Sc. thesis, Universty of Pretoria, Pretoria. VAN VUUREN, D.R.J. & H.P. VAN DER SCHIJFF. 1970. 'n Vergelykende ekologiese studie van die plantegroei van 'n noordelike en suidelike kloof van die Magaliesberg. Tydskrif vir Natuurweten- skappe 10: 16–75. Von Maltitz, G. (Project leader). 2003. Classification system for South African indigenous forests: An objective classification for the Department of Water Affairs and Forestry. Pretoria: Council for Scientific and Industrial Research. (Environ- mentek report ENV-P-C 2003-017). WERGER, M.J.A. 1974. On concepts and techniques applied in the Zürich-Montpellier method of vegetation survey. Bothalia 11(3): 309–323. WESTFALL, R.H. 1981. The plant ecology of the farm Groothoek, Thabazimbi District. M.Sc. thesis, University of Pretoria, Pretoria. WESTFALL, R.H., N. VAN ROOYEN & G.K. THERON. 1985. The plant ecology of the farm Groothoek, Thabazimbi District. 2. Classification. Bothalia 15: 655–688. WHITE, F. 1978. The Afromontane region. Pp. 465–513. In: WERGER, M.J.A. (ed.). Biogeogra- phy and ecology of Southern Africa. The Hague: Junk. ISSN 0075-6458 31 Koedoe 49/1 (2006) van staden woud.qxd 2006/04/17 08:38 PM Page 31