South African Journal of Botany 2002, 68: 475–496 Copyright © NISC Pty Ltd Printed in South Africa — All rights reserved SOUTH AFRICAN JOURNAL OF BOTANY ISSN 0254–6299 Vegetation ecology of Sekhukhuneland, South Africa: Combretum hereroense–Grewia vernicosa Open Mountain Bushveld SJ Siebert1,2*, AE van Wyk1 and GJ Bredenkamp1 1 Department of Botany, University of Pretoria, Pretoria 0002, South Africa 2 Present address: SABONET, c/o National Botanical Institute, Private Bag X101, Pretoria 0001, South Africa * Corresponding author, e-mail: [email protected] Received 18 September 2001, accepted in revised form 30 August 2002 A hierarchical classification, description, and ecological sub-associations. Many new syntaxa are described and and floristic interpretations are presented on one of the ecologically interpreted. For each of the plant commu- six major vegetation types of the Sekhukhuneland nities the species richness, endemism and conserva- Centre of Plant Endemism, namely the Combretum tion status were determined. A selected set of environ- hereroense–Grewia vernicosa Open Mountain mental factors is provided to aid with the delimitation of Bushveld. Relevés were compiled in 91 stratified ran- plant communities. The floristic information, proposed dom plots. A TWINSPAN classification, refined by classification, general description and vegetation key Braun-Blanquet procedures, revealed 20 plant commu- can be used for future identification of conservation nities, which are divided into six associations and 19 areas, land-use planning and further research. Introduction Although phytosociological studies have been conducted on Siebert et al. (2002) recorded six major vegetation types the savannas of ultramafic substrates in southern Africa for the SCPE during a comprehensive vegetation and floris- (Werger et al. 1978, Van der Meulen 1979), several vegeta- tic survey of the savannas and grasslands of the 4 000km2 tion types associated with this type of substrate remain poor- of ultramafic rock of the eastern Rustenburg Layered Suite. ly investigated and documented. An example of ultramafic This was probably the most extensive survey of vegetation vegetation which has never been studied in detail, is the in a single ultramafic region since the comprehensive sur- Mountain Bushveld identified by Siebert et al. (2002) on the veys conducted by Jaffré (1980) on the serpentine flora of norite, pyroxenite and anorthosite hills and mountains of the New Caledonia. It is envisaged that the identification, classi- Sekhukhuneland Centre of Plant Endemism [SCPE] (Van fication and description of the various vegetation units will Wyk and Van Wyk 1997, Van Wyk and Smith 2001). contribute to the knowledge of the ecological and floristic Ultramafic mountains and hills are floristically noteworthy in intricacies of the SCPE’s Mountain Bushveld and ultramafic that they usually harbour many endemic plant species that are regions in general. restricted to, and communities that are associated with, this This paper characterises and interprets the vegetation particular geological substrate (Brooks and Malaisse 1985, units and the associated habitats of the Open Mountain Iturralde 1995, Madulid and Agoo 1995, Siebert et al. 2001). Bushveld [OMB] of the Sekhukhuneland Centre (Siebert et Various savanna vegetation types have been recognised al. 2002). An assessment of the plant diversity, endemism in areas adjacent to the SCPE, on the dry dolomitic hills and and threatened taxa of the various plant communities of the mountains of the northeastern Drakensberg Escarpment OMB is supplied to aid future conservation actions in the (Matthews et al. 1992), the western Rustenburg Layered region. A vegetation key is presented to assist with the easy Suite (Coetzee 1975) and the northeastern plains of the identification of the syntaxa of the OMB in the region. Habitat Springbok Flats (Breebaart and Deutschlander 1997). The data are provided to aid with future, more comprehensive SCPE shares many near-endemics and shows a definite ecological interpretations. floristic affinity (Siebert 1998) with these areas. Acocks At a local level the data provided with this study are basic (1988) mapped the savanna in the SCPE as three major to the formulation of management policies, as phytosociolo- veld types, namely Mixed Bushveld, Sourish Mixed gy has a regular application in ecosystem management, for Bushveld and North-Eastern Sandy Highveld Grassland. A an adequate database of natural features, such as the vege- more generalised classification of the same region’s vegeta- tation and physical environment, is essential for effective tion is given by Low and Rebelo (1996), which recognises land-use management and implementation (Kent and Ballard only one major vegetation type, namely Mixed Bushveld. 1988, Bedward et al. 1992, Rhoads and Thompson 1992). 476 Siebert, Van Wyk and Bredenkamp Study Area value assigned to each according to the Braun-Blanquet scale (Mueller-Dombois and Ellenberg 1974), namely R = The study area comprises the central part of the SCPE rare, + = <1%, 1 = <5%, A = 5–12%, B = 12–25% and 3 = (Figure 1), between 24°40’ and 25°20’S latitude; 29°50’ and 25–50% cover. 30°30’E longitude. The whole SCPE is defined and dis- Phenological changes over the sampling period influ- cussed in Siebert et al. (2002), where it is emphasised that enced the species cover and consequently the data of the the SCPE is more or less restricted to the Rustenburg relevés (Fischer 2000). Plant species names follow Retief Layered Suite, one of the stratigraphic units of the Bushveld and Herman (1997) and general practice followed at the Complex. The extent of occurrence of the Mountain HGWJ Schweickerdt Herbarium (PRU), University of Bushveld is approximately 2 500km2 and is characterised Pretoria. Terminology to describe vegetation structure is by considerable diversity in geology (Kent 1980) and phys- based on Edwards (1983). Environmental data recorded in iography (Land Type Survey Staff 1989). each sample plot include terrain type (Land Type Survey The average annual rainfall for the Mountain Bushveld Staff 1987, 1988, 1989), aspect, slope, geology (Visser et al. region is 550mm (South African Weather Bureau 1998), but 1989), soil type (MacVicar et al. 1991) and rockiness of soil the rainfall pattern is strongly influenced by the local topog- surface. Longitude and latitude readings were recorded for raphy (Siebert 1998). Rainfall varies from as little as 400mm each sample plot using a Global Positioning System. All in some of the valleys, to an estimated 600mm on the east- relevé data are stored in the TURBOVEG database ern slopes of mountains (Erasmus 1985). Temperatures for (Hennekens 1996a) and located at the Department of the Mountain Bushveld region range from 0°C to 38°C, with Botany, University of Pretoria (Mucina et al. 2000). a mean daily temperature of 20°C (South African Weather A first approximation of the vegetation classification, based Bureau 1998). Temperatures vary at different localities with- on a total floristic data set of 415 relevés for the SCPE, was in the area, also correlating strongly with physiographic fea- obtained by the application of Two-Way Indicator Species tures, being higher in the low-lying valleys and lower on the Analysis (TWINSPAN) (Hill 1979). This first step of an objec- high-lying plateaus (Buckle 1996). The northern and western tive multivariate classification identified several major vege- parts of the study area are on average warmer and drier tation types. The result was then used to subdivide the data than the southern and eastern parts (Siebert 1998). set into six phytosociological tables, each representing one of Soils of the study area are typical for ultramafic areas the major vegetation types of the SCPE (Siebert et al. 2002). (Siebert et al. 2001), as certain areas exhibit Mg/Ca ratios One of these, consisting of 91 relevés, was again subjected (>1) similar to that of serpentine, they are rich in chromium, to standard TWINSPAN cut levels and the subsequent divi- cobalt, iron, nickel, and deficient in the nutrients calcium, sions were adopted as homogeneous groups of relevés rep- molybdenum, nitrogen, phosphorous and potassium resenting plant communities of the OMB. (Johnston and Proctor 1981, Brooks 1987). However, a low Using Braun-Blanquet procedures in the MEGATAB com- Mg/Ca ratio (<1) has also been recorded along lower moun- puter programme (Hennekens 1996b), the species groups tains slopes of certain areas in the SCPE (Siebert 1998), a were further refined. The partial subjectivity of this process phenomenon presumably caused by a reduction in the bind- is controversial in that it is not fully formalised, but serves to ing strength of Mg due to topsoil acidification and subsequent find coincidence patterns which carry ecological information. Mg leaching (Roberts and Rodenkirchen 1995). Soils of the Species groups were only recognised if the species was rep- study area have a moderate pH (6–8) (Loock et al. 1982). resented in a group of relevés more often than outside the The Mountain Bushveld of the SCPE can broadly be group and if the resultant group could be ecologically inter- described as savanna of undulating norite and pyroxenite preted. All syntaxa were hierarchically classified according hills and mountains, which is bordered by a northeastern to Westhoff and Van der Maarel (1978) and formally named Sandy Highveld Grassland-Mountain Bushveld ecotone in according to Weber et al. (2000). the south and a