Gaugris.qxd 2004/04/15 06:28 Page 9

The vegetation of Tshanini Game Reserve and a comparison with equi- valent units in the Tembe Elephant Park in Maputaland, South

J.Y. GAUGRIS, W.S. MATTHEWS, M.W. VAN ROOYEN and J. du P. BOTHMA

Gaugris, J.Y. , W.S. Matthews, M.W. van Rooyen and J. du P. Bothma. 2004. The veg- etation of Tshanini Game Reserve and a comparison with equivalent units in the Tembe Elephant Park in Maputaland, . Koedoe 47(1): 9–29. Pretoria. ISSN 0075- 6458. An analysis of the communities of the to be proclaimed Tshanini Game Reserve and their relation with selected similar plant communities of Tembe Elephant Park are presented. The study area lies 6 km due south of Tembe Elephant Park within the Maputaland Centre of Plant Endemism, which is part of the Maputaland-Pondoland region recognised by the IUCN as a centre of Plant Diversity. A Braun-Blanquet classi- fication on 51 random sample plots revealed eight distinct, mainly woodland plant asso- ciations. Vegetation distribution follows the same patterns as described in Tembe Ele- phant Park. Plant communities and subcommunities are mapped, described and com- pared to similar plant communities in Tembe Elephant Park to detect differences between protected and unprotected areas. Results from the qualitative comparison showed >50 % similarity between equivalent units and results from the quantitative comparison showed >25 % to 50 % similarity between equivalent units. Mueller-Dom- bois and Ellenberg (1974) regard 25 % as the lower threshold value to indicate similar associations. The observed differences may relate to animal and human utilisation. Key words: plant communities, Sand Forest, similarity indices, Tembe Elephant Park, Tshanini Game Reserve

J.Y. Gaugris and J. du P. Bothma, Centre for Wildlife Management, University of Pre- toria, Pretoria, 0002 Republic of South Africa; W.S. Matthews, Tembe Elephant Park, Ezemvelo KwaZulu-Natal Nature Wildlife, Private Bag 356, KwaNgwanase, 3973 Republic of South Africa; M.W. van Rooyen Department of Botany, University of Preto- ria, Pretoria, 0002 Republic of South Africa.

Introduction The Tshanini Game Reserve was established in 2000 on some land of the community of The Tembe Elephant Park was proclaimed in Manqakulane, but it has not yet been regis- 1983 after negotiations between the then tered. The reserve’s creation developed from KwaZulu Bureau of Natural Resources and the initiative of the local community, from as the Tembe Tribal Authority in consultation far back as the creation of Tembe Elephant with the local communities of northern Park. The ideal of the Manqakulane people Maputaland, KwaZulu-Natal, South Africa. was to create their own reserve on their own The park boundaries were subsequently land to counteract high unemployment and to fenced and animal numbers started to raise funds for the development of the com- increase. The fence has kept the utilisation of munity through ecotourism. For the past 20 renewable natural resources by the local years, the land where the Tshanini Game communities at bay for the past 19 years. In Reserve has been established was utilised by this period, the vegetation of the park has the local people for wood harvesting, live- been utilised only by the indigenous fauna, stock grazing, hunting, fruit and honey gath- but it has been affected by management deci- ering, and subsistence cultivation where this sions and possibly also regional environmen- was viable. It is presumed that large ungulate tal changes. numbers have always been low in Tshanini

ISSN 0075-6458 9 Koedoe 47/1 (2004) Gaugris.qxd 2004/04/15 06:28 Page 10

Fig. 1. Map of Maputaland with the locations of Tembe Elephant Park and Tshanini Game Reserve in northern KwaZulu-Natal, South Africa.

Game Reserve, due to the absence of stand- reserves lie close to each other with the ing water. However, hunting pressure has northern boundary of Tshanini Game affected the larger mammal populations, and Reserve some 6 km due south of the south- only small, secretive antelope such as the ern boundary of Tembe Elephant Park. Both suni Neotragus moschatus and red of these conservation areas lie in the core of Cephalophus natalensis have survived there. the Maputaland Centre of Plant Endemism as described by Van Wyk (1996). In both The Tembe Elephant Park and the Tshanini reserves a rare forest type, known as Sand Game Reserve are underlain by the same Forest occurs (Van Wyk 1996; Van Rensburg type of soil and they share the same broad et al. 1999; Kirkwood & Midgley 1999; vegetation types. Moll (1977) distinguished Matthews et al. 2001; Van Wyk & Smith the Pallid and the Sand Forest 2001). The Sand Forest in this particular area Zones in the area, whereas Low & Rebelo is further defined as the eastern Sand Forest (1980) recognised the Subhumid Lowveld by MacDevette et al. (1989) and Kirkwood Bushveld of the Savanna Biome and the & Midgley (1999). In South Africa and in Sand Forest of the Forest Biome. The two , this forest type is restricted to

Koedoe 47/1 (2004) 10 ISSN 0075-6458 Gaugris.qxd 2004/04/15 06:28 Page 11

the ancient coastal dune cordons of northern Study area KwaZulu-Natal, and it is home to many rare Tshanini Game Reserve is 2 420 ha in extent and unusual plant and animal species (Kirk- and its northern boundary lies 6 km due wood & Midgley 1999; Van Wyk & Smith south of the southern boundary of Tembe 2001). Sand Forest in South Africa covers Elephant Park (Fig. 1). The area consists only 354 km², of which 44 % is conserved mainly of a sandy plain that is interspersed (Low & Rebelo 1998) with the largest pro- with ancient littoral dunes, and is covered by tected area in South Africa in the Tembe Ele- an open to closed woodland with patches of phant Park (McGeogh et al. 2002). The short to tall Sand Forest (Fig. 2). major part of this vegetation type is found in Mozambique but the extent of it is unknown The area covered by Tshanini Game Reserve is unfenced and therefore subject to human at present, because of a lack of detailed stud- influence. Cattle grazing with small herds, ies. Cattle grazing, firewood demands and rarely exceeding 20 heads, and wood collect- elephant Loxodonta africana impact are the ing, either for construction or fire, are the main identified threats for the Sand Forest most important forms of natural resource (Van Wyk 1996; Van Rensburg et al. 1999). utilisation. Other less intensive forms of use The human population growth rate in include fruit gathering, the collecting of Maputaland is high, increasing both by nat- honey and thatching grass, and hunting of ural growth and immigration, despite a high birds and mammals. Until 1995, the eastern mortality rate from the AIDS epidemic. portion of Tshanini was still utilised for agri- Roads have been constructed through the culture, with some large fields cut into the communities, and safe drinking water is pro- woodland or forest. The Manqakulane com- munity resided on the eastern boundary of vided. Although these are laudable rural the reserve until 1990, when they moved to developments, they may become a threat to 3 km east of the eastern boundary of Tshani- the conservation of the remote and less ni Game Reserve. The relocation followed exploited areas of northern Maputaland. the installation of a governmental clean Matthews et al. (2001) did a complete vege- water supply to the local communities. The tation survey of Tembe Elephant Park, and new location of the human settlements on the suggested that their data could be used for southern part of the Muzi Swamp has a bet- management and land use planning for other ter soil potential for agriculture. The com- similar areas in Maputaland. However, the munity of Manqakulane today controls an questions are to what extent has human pres- area of just over 10 000 ha, with a population of approximately 700 people in 110 house- sure affected the vegetation in comparison holds that are concentrated around the Muzi with wildlife use; can the vegetation of Swamp. Tembe Elephant Park be regarded as repre- sentative of similar areas in Maputaland and how well did the Sand Forest fare outside a Topography formal conservation area? In this paper we Tshanini Game Reserve lies centrally in the aim to provide ecological and floristic data southernmost portion of the Mozambican specifically for the Tshanini Game Reserve Coastal Plain between the Lebombo Moun- by identifying, characterising, interpreting tains in the west and the Indian Ocean in the and mapping the major vegetation units that east. Ancient littoral dunes that run in a occur there. Further objectives are to com- broadly north to south direction are inter- pare the vegetation of Tshanini Game spersed with inter-dune depressions. To the Reserve with similar units in Tembe Ele- west of the reserve a seepage line forms the phant Park in terms of species composition source of the Nhlole River, a non-perennial and abundance, and to relate the perceived seepage line winding to Makane’s Drift and changes to differences in utilisation pressure. flowing into the Pongola River. The seepage

ISSN 0075-6458 11 Koedoe 47/1 (2004) Gaugris.qxd 2004/04/15 06:28 Page 12

Fig. 2. The vegetation of the Tshanini Game Reserve and surrounding area in Maputaland, northern KwaZulu-Natal, South Africa.

Koedoe 47/1 (2004) 12 ISSN 0075-6458 Gaugris.qxd 2004/04/15 06:28 Page 13

line runs from the southeast to the north- west. The Tshanini Game Reserve has a mean alti- tude of 80 m above sea level, and two main ancient littoral dunes occur in it. The dune on the eastern boundary is the southern extension of Nhlela Ridge that extends into the western portion of Tembe Elephant Park. A beacon is positioned on top of the ridge at an altitude of 103 m above sea level. A sec- ond dune runs along the western boundary at an altitude of 95 m above sea level. Between the two main dunes, a third but smaller dune formation is found that starts in the middle of Tshanini Game Reserve and increases in height towards the southern-central part of the reserve. The Nhlole River lies to the east Fig. 3. Climatogram of the Sihangwane Weather Sta- of the western dune cordon, and the lowest tion, Tembe Elephant Park, Maputaland, northern point of the reserve is 64 m above sea level KwaZulu-Natal, South Africa (b = height above sea in the northeastern corner. level, c = duration of observations in years, d = mean annual temperature in °C, e = mean annual precipita- tion in mm, f = mean daily minimum of the coldest Climate month in °C, g = lowest temperature recorded in °C, h = mean daily maximum of warmest month in °C, Rainfall and temperature data for the i = highest temperature recorded in °C, j = mean weather station at Sihangwane at the daily temperature variation in °C, m = dry season, entrance of Tembe Elephant Park n = humid season, o = mean monthly rainfall (32.42388 ºE, 27.04316 ºS) appear in Fig. 3. > 100 mm) following Walter (Cox & Moore 1994). Maputaland lies in a transition zone between the tropics to the north and subtropics to the south. The summers are hot and the winters cool to warm. The summer months are the wettest, although rainfall occurs throughout the year. In the drier winter months, morning coastal plain date back approximately 5 mil- mist is a common phenomenon especially in lion years to the early Pleistocene. The the swamp areas. The relative humidity of recent dunes are as young as 10 000 years, the air in summer is higher than in winter from the late Pleistocene to 500 years ago (Matthews et al. 2001). for the current coastal dunes. The dunes of Maputaland are among the most recent geo- logical formations that are found in southern Geology Africa (Hobday 1976). The underlying geology of Tshanini Game Reserve consists of Cretaceous siltstone Soils units that form the base of the Mozambican Coastal Plain. This formation is overlain by The soils of Tshanini Game Reserve are sim- sediments of Miocene and Pleistocene ori- ilar to those described by Matthews et al. gin. The Maputaland Group overlies the lat- (2001) for Tembe Elephant Park, and ter. On top of this formation lie the marine, specifically so for the dunes and inter-dune littoral and coastal dune deposits that were depressions. The soils have a thin, organi- formed by successive marine transgressions cally enriched A-horizon underlain by a and regressions. The oldest dunes of the sandy subsoil. The clay content is attributed

ISSN 0075-6458 13 Koedoe 47/1 (2004) Gaugris.qxd 2004/04/15 06:28 Page 14

to the weathering of labile minerals over a map of the Tshanini Game Reserve. As it appeared long period. The soils along the two main impossible to digitise every vegetation unit and sub- dunes are similar to those described for the unit separately, some units were grouped in mosaics. Nhlela Ridge in Tembe Elephant Park. The These mapping units are indicated in Fig. 2. soils are well-drained profiles with less than The homogeneous vegetation units that are common 5 % clay, have a high base status, and are to Tembe Elephant Park and Tshanini Game Reserve classified as Hutton or Clovelly forms (Soil were then analysed further to determine the degree Classification Working Group 1991; of similarity or dissimilarity in terms of species com- position and abundance by using similarity indices Matthews et al. 2001). The inter-dune as described by Mueller-Dombois & Ellenberg depressions are underlain by moderately to (1974). The phytosociological table created for well-drained yellowish Clovelly or grey Tshanini Game Reserve and the table obtained from Fernwood soils (Soil Classification Working the study of Matthews et al. (2001) for Tembe Ele- Group 1991; Matthews et al. 2001). phant Park were used to calculate the indices. Sorensen’s qualitative index of similarity (Mueller- Dombois & Ellenberg 1974) that is based on the absence or presence of plant species was used to cal- Methods culate indices of similarity in plant species composi- tion. To be able to do so, a similar number of com- Aerial photographs were obtained from the Survey- parable relevés were compared in each of the vege- or General in Pretoria, South Africa. The appropriate tation communities. The following equation was 1:50 000 scale photograph was scanned and digitised used: by using the Geographical Information Tool Cartalinx (The spatial Data Builder, Version 1.1. Sorensen’s Index of Similarity = (2c x 100) / A + B Clarks Laboratories, 1998) to create a Geographical where: Information System (GIS) based geo-referenced map of the homogeneous physiognomic and physio- c = number of species common to both communities graphic units of the reserve. The map was then refined by using the ArcView GIS tool (Arc View A = total number of species in community A Gis, Version 3.1, Environmental Systems Research Institute 1998). B = total number of species in community B. In all, 51 sample plots, of 30 m x 30 m each, were The similarity index of Motyka et al. (Mueller-Dom- distributed in a stratified random manner throughout bois & Ellenberg 1974) was used to calculate a quan- the study area. Vegetation sampling was done from titative similarity index in terms of species abun- January to May 2001. The terminology that was used dances. The equation is: to name and describe the vegetation structure of the Motyka et al.’s Index of Similarity = (2 Mw x 100) / plant communities follows Edwards (1983). For (M + M ) each sample plot, a relevé of the floristic composi- a b tion and cover-abundance values for each plant where: species, according to the Braun-Blanquet cover- abundance scale (Mueller-Dombois & Ellenberg Mw refers to the sum of the smaller quantitative val- 1974), were recorded. Environmental and ecological ues of the species that are common to the two information collected for each sample plot included communities; slope, aspect, altitude, topography, rock cover, ero- sion percentage, cover for canopy height classes of Ma is the sum of the quantitative values of all the species in one community; herbs, grasses, shrubs and trees and biotic influence. Unidentified plant species were sampled for later Mb is the sum of all the quantitative values of all the identification at the herbarium of the Tembe Ele- species in the other community. phant Park. To calculate the values of Mw, Ma and Mb, the The floristic data were classified with Braun-Blan- Braun-Blanquet cover-abundance values in the phy- quet procedures using the TURBOVEG and MEGATAB tosociological tables of both reserves were convert- computer packages (TURBOVEG for Windows version ed into mean cover percentages. To accomplish this 1.97, Hennekens & Schaminee 2001). This vegeta- each cover-abundance value on the Braun-Blanquet tion classification system was used to refine and scale was replaced by a cover value, whereafter a interpret the preliminary map of homogeneous phys- mean for the community could be calculated. iognomic units to produce a more detailed vegetation Although this is a crude method it does allow quan-

Koedoe 47/1 (2004) 14 ISSN 0075-6458 Gaugris.qxd 2004/04/15 06:28 Page 15

Table 1 Classification of the vegetation communities for Tshanini Game Reserve in northern KwaZulu-Natal, South Africa 1. The Urelytrum agropyroides-Sapium integerrimum open woodland 1.1. The Albizia versicolor-Diheteropogon amplectens sparse woodland 1.2. The Indigofera podophylla-Albizia adianthifolia ecotonal sparse woodland 1.3. The Antidesma venosum-Urelytrum agropyroides open woodland 1.4. The Fimbristylis complanata-Diheteropogon amplectens open woodland 2. The Sclerocarya birrea- madagascariensis closed woodland 2.1. The Cordia monoica-Sclerocarya birrea open to closed woodland 2.2. The Melinis repens-Sclerocarya birrea closed woodland 3. The Ptaeroxylon obliquum-Hymenocardia ulmoides sand forest 3.1. The Rothmannia fischeri-Ptaeroxylon obliquum low to short sand forest 3.2. The Strychnos henningsii-Ptaeroxylon obliquum tall sand forest.

titative comparisons to be made. When comparing form mosaics that grade into each other and vegetation communities, the similarity in terms of the boundaries are difficult to map on an aer- absence or presence of species in two communities is ial photograph. However, the plant commu- more important than the quantitative contribution of each species to a community (Mueller-Dombois & nities are relatively easily recognised in the Ellenberg 1974). The latter can be used to refine the field. comparisons. Description of plant communities Results and discussion 1. The Urelytrum agropyroides-Sapium integerrimum open woodland Classification This open woodland is the most widespread Three major plant communities and eight community in Tshanini Game Reserve. It is sub-communities were identified for the associated with the dunes and inter-dune Tshanini Game Reserve (Tables 1 & 2, depressions on grey-brown to orange-grey Fig. 2). The Tshanini Game Reserve lies on dystrophic regosols. These soils have a low regic sand and no major geological feature clay content and are associated with a low could be identified that plays a role in the pH value of 5.5 (Matthews et al. 2001). The differentiation of the plant communities. open woodland covers 37.7 % of Tshanini Matthews et al. (2001) postulated that clay Game Reserve and represents a surface area content, moisture level in the soil and vege- of 912 ha. tation dynamics were major determinants of vegetation differentiation in Tembe Elephant This plant community is characterised by Park. However, clay content and moisture diagnostic species of Species Group E. Plant level differences in the soil of various areas species of Species Group K form the link in Tshanini Game Reserve are slight and between this open woodland community and could not be directly related to the differen- the closed woodland community. Diagnostic tiation between plant communities in Tshani- plant species are the tree Albizia adianthifo- ni Game Reserve. Except for the sand forest, lia, the grasses Diheteropogon amplectens the plant communities do not have sharp and Urelytrum agropyroides and the shrub boundaries. The open and closed woodlands Acridocarpus natalitius var. linearifolius.

ISSN 0075-6458 15 Koedoe 47/1 (2004) Gaugris.qxd 2004/04/15 06:28 Page 16 Table 2 Table Phytosociological table of the vegetation of Tshanini Game Reserve in northern KwaZulu-Natal, South Africa Game Reserve in northern KwaZulu-Natal, South Phytosociological table of the vegetation Tshanini

Koedoe 47/1 (2004) 16 ISSN 0075-6458 Gaugris.qxd 2004/04/15 06:28 Page 17 Table 2 Table (continued)

ISSN 0075-6458 17 Koedoe 47/1 (2004) Gaugris.qxd 2004/04/15 06:28 Page 18 Table 2 Table (continued)

Koedoe 47/1 (2004) 18 ISSN 0075-6458 Gaugris.qxd 2004/04/15 06:28 Page 19 Table 2 Table (continued)

ISSN 0075-6458 19 Koedoe 47/1 (2004) Gaugris.qxd 2004/04/15 06:28 Page 20 Table 2 Table (continued)

Koedoe 47/1 (2004) 20 ISSN 0075-6458 Gaugris.qxd 2004/04/15 06:28 Page 21

Prominent species with a constancy of tree layer is underlain by a short to tall grass- >50 % include the shrub Vangueria infausta, land up to 2 m tall. the grass Andropogon gayanus and the trees Acacia burkei, Strychnos madagascariensis, 1.2 The Indigofera podophylla-Albizia sericea, molle and adianthifolia ecotonal sparse wood- Sapium integerrimum. The dwarf shrubs land Pavetta schumanniana, Eugenia mossambi- censis and the forb Corchorus junodii have a This subcommunity is also found on the high constancy in this community (Table 2). dune crests, slopes and inter-dune depres- sions on the western portion of the Tshanini A mean of 45 species was recorded per Game Reserve on acidic orange-grey dys- relevé. Of the plant species that were record- trophic regosols. The subcommunity is not ed in this community, 18 species are endem- characterised by its own distinct Species ic to Maputaland (Van Wyk 1996). Struc- Group but shares Species Group B with turally, according to the classification of community 1.1. Prominent species in this Edwards (1983) this community is a wood- community are found in Species Groups E, J land that varies from sparse to open. The and K. This subcommunity forms a transi- non-continuous canopy can reach heights of tion zone linking the sparse and open wood- 10–12 m, but a height of 6–8 m is the norm. lands. Diagnostic species are the tree Com- This community is subdivided into four sub- bretum mkuzense and the shrubs Garcinia communities based on the density, structure livingstonei and Indigofera podophylla. The and floristic composition of the vegetation. prominent species include the trees Albizia adianthifolia, Sapium integerrimum and Strychnos madagascariensis, the grasses 1.1 The Albizia versicolor-Dihetero- Diheteropon amplectens, Urelytrum agropy- pogon amplectens sparse woodland roides, Perotis patens, Digitaria eriantha This subcommunity is found on the dune and Pogonarthria squarrosa, the shrub crests, slopes and inter-dune depressions of Acridocarpus natalitius var. linearifolius and the western side of the Tshanini Game dwarf shrubs Pavetta schumanniana, Euge- Reserve and occurs on acidic orange-grey nia mossambicensis and Albertisia dela- dystrophic regosols. The subcommunity is goensis. A mean of 46 plant species was characterised by the diagnostic species of recorded per relevé. According to the classi- fication of Edwards (1983), this subcommu- Species Group A and common species of nity is a short sparse woodland, with a few Species Groups E, J and K. The diagnostic tall trees that reach a height of 6–8 m. The species are the tree Albizia versicolor, the tree layer is underlain by a short to tall grass- forb Oxygonum robustum and the shrub land up to 2 m tall. Grewia inaequilatera. Prominent species are the trees Albizia adianthifolia, Sapium inte- gerrimum and Strychnos madagascariensis, 1.3 The Antidesma venosum-Urelytrum the grasses Diheteropogon amplectens, Ure- agropyroides open woodland lytrum agropyroides, Perotis patens and This subcommunity is found on the dune Pogonarthria squarrosa, the shrub Acrido- crests, slopes and inter-dune depressions on carpus natalitius var. linearifolius and the the western portion of the Tshanini Game dwarf shrubs Pavetta schumanniana, Euge- Reserve. It occurs on acidic orange-grey nia mossambicensis and Albertisia dela- dystrophic regosols and is characterised by goensis. A mean of 45 plant species was diagnostic species of Species Group C with recorded per relevé. According to the classi- common species of Species Groups E, I, J fication of Edwards (1983), this subcommu- and K. The diagnostic species are the tree nity is a short sparse woodland with a few Antidesma venosum, the forbs Oxygonum tall trees that reach a height of 6–8 m. The dregeanum and Hoslundia opposita, the

ISSN 0075-6458 21 Koedoe 47/1 (2004) Gaugris.qxd 2004/04/15 06:28 Page 22

sedge Cyperus obtusiflorus and geophyte 2. The Sclerocarya birrea-Strychnos mada- Hypoxis hemerocallidea. The prominent gascariensis closed woodland species are the trees Albizia adianthifolia, In the Tshanini Game Reserve the closed Sapium integerrimum and Strychnos mada- woodland community is associated with the gascariensis, the grasses Diheteropogon dunes and inter-dunes depressions and amplectens, Urelytrum agropyroides, occurs on grey dystrophic regosols. The soils Eustachys paspaloides and Perotis patens, have a low clay content associated with a the shrub Acridocarpus natalitius var. lin- low pH value of 5.5 (Matthews et al. 2001). earifolius and dwarf shrubs Eugenia The closed woodland covers 19.1 % of the mossambicensis and Diospyros galpinii. A Tshanini Game Reserve and represents a sur- mean of 52 plant species was recorded per face area of 463 ha. relevé. This makes this plant subcommunity the second richest unit sampled in the This plant community is characterised by Tshanini Game Reserve. According to the species of Species Group H. Diagnostic classification of Edwards (1983), this sub- plant species are the trees Sclerocarya bir- community is a short open woodland with a rea, Ziziphus mucronata and Bridelia few tall trees that reach a height of 6–8 m. cathartica, the grass Eragrostis superba, the The tree layer is underlain by a short to tall shrub Gymnosporia senegalensis and forb grassland up to 2 m tall. Priva cordifolia. Prominent species with a constancy >50 % include the shrubs Vangue- ria infausta and Diospyros dichrophylla, the 1.4 The Fimbristylis complanata- trees Strychnos madagascariensis, Strychnos Diheteropogon amplectens open spinosa, Euclea natalensis and Tabernae- woodland montana elegans and the grasses Digitaria This subcommunity is found on the dune eriantha, Perotis patens and Pogonarthria crests, slopes and inter-dune depressions on squarrosa (Table 2). the western portion of the Tshanini Game A mean of 52 plant species was recorded per Reserve. It occurs on acidic orange-grey relevé. Of the plant species that were record- dystrophic regosols (Matthews et al. 2001). ed in this community, 15 species are endem- The subcommunity is characterised by the ic or near endemic to Maputaland (Van Wyk diagnostic species of Species Group D with 1996). According to the classification of common species in Species Groups E, I, J Edwards (1983), this community is a short, and K. The diagnostic species are the sedge open to closed woodland that can be distin- Fimbristylis complanata, the dwarf shrub guished by a semi-continuous to continuous Parinari capensis subsp. incohata, the forb canopy reaching heights from 6–10 m. A tall Tricliceras mossambicense and the grass to high shrub component, which forms dense Alloteropsis semialata. The prominent clumps around the base of trees, is associat- species are the trees Albizia adianthifolia ed with this community. The tree and shrub and Strychnos madagascariensis, the grasses canopies are underlain by a continuous Diheteropogon amplectens, Urelytrum mixed layer of short grasses and herbs. This agropyroides, Eustachys paspaloides, Pero- community is subdivided into two subcom- tis patens and Pogonarthria squarrosa, the munities based on the structure and floristic shrub Acridocarpus natalitius var. lineari- composition of the vegetation. folius and the dwarf shrub Eugenia mossam- bicensis. A mean of 37 plant species was recorded per relevé. According to the classi- 2.1 The Cordia monoica-Sclerocarya fication of Edwards (1983), this subcommu- birrea open to closed woodland nity is a short open woodland with a few tall This subcommunity is found on the dune trees that reach a height of 6–8 m. The tree crests, slopes and inter-dune depressions in layer is underlain by a short to tall grassland the eastern half of the Tshanini Game up to 2 m tall. Reserve on acidic grey-brown dystrophic

Koedoe 47/1 (2004) 22 ISSN 0075-6458 Gaugris.qxd 2004/04/15 06:28 Page 23

regosols. It is characterised by diagnostic senegalensis, Strychnos spinosa and Bridelia species of Species Group F with common cathartica and dwarf shrub Eugenia species of Species Groups H, I and K. This mossambicensis. A mean of 48 plant species subcommunity is usually found around was recorded per relevé. According to the ancient settlements and shows a slow recov- classification of Edwards (1983), this sub- ery from human impact. Old settlements can community is a short closed woodland with still be located and trees grow back on the a semi-continuous to continuous canopy former cultivated lands around these reaching a height of 8–10 m. The tree layer emplacements. The outline of the old fields is underlain by a mixed short layer of grass- is still distinguishable from surrounding veg- es and herbs as well as tall to high shrubs. etation. The diagnostic species are the shrubs Cordia monoica, Lantana rugosa and Lippia javanica, the herb Helichrysum nudifolium 3. The Ptaeroxylon obliquum-Hymenocar- and the grass Setaria sphacelata. Prominent dia ulmoides sand forest species are the trees Sclerocarya birrea, Zizi- The sand forest is a major community in phus mucronata, Tabernaemontana elegans Tshanini Game Reserve. It is associated with and Strychnos madagascariensis, the grasses the dunes and inter-dunes depressions on Eragrostis superba, Digitaria eriantha and orange-brown arenosols and dystrophic Perotis patens, the shrubs Gymnosporia regosols. The soils have a low clay content senegalensis, Strychnos spinosa, Bridelia associated with a low pH value of 5.6 cathartica and dwarf shrub Eugenia (Matthews et al. 2001). The sand forest cov- mossambicensis. A mean of 56 plant species ers 43.2 % of Tshanini Game Reserve and was recorded per relevé, which makes it the represents a surface area of 1 045.5 ha. richest plant subcommunity sampled in the Tshanini Game Reserve. According to the This plant community is characterised by classification of Edwards (1983), this sub- diagnostic species of Species Group O. Con- community is a short, open to closed wood- sistent diagnostic species are the tree land with a semi-continuous canopy, reach- Ptaeroxylon obliquum, the shrub or liana ing heights of 8–10 m. The tree layer is Uvaria caffra, the understorey shrubs underlain by tall to high shrubs as well as a Drypetes arguta and Toddaliopsis bre- continuous grass layer up to 1 m tall. mekampii. Prominent species with a constan- cy of >50 % include the shrubs Croton 2.2 The Melinis repens-Sclerocarya pseudopulchellus, Hyperacanthus micro- birrea closed woodland phyllus, junodii, Salacia leptocla- da and Croton steenkampianus, the trees This subcommunity is found on the dune Cleistanthus schlechteri, Hymenocardia crests, slopes and inter-dune depressions in ulmoides, myrtifolia, Brachylae- the eastern half of the Tshanini Game na huilensis and Balanites maughamii. The Reserve. It occurs on acidic grey-brown dys- understorey shrub Synaptolepis kirkii has a trophic regosols. This subcommunity is high constancy in this community (Table 2). characterised by diagnostic species of Species Group G, with common species in A mean of 40 plant species was recorded per Species Groups H, I and K. The diagnostic relevé. Of the plant species that were record- species are the tree Trichilia emetica, the ed, 20 species are endemic to Maputaland shrub Phyllanthus reticulatus, the herb with 10 species restricted to the Sand Forest Nidorella auriculata and the grasses Melinis type (Van Wyk 1996; Matthews et al. 2001). repens and Eragrostis ciliaris. Prominent According to the classification of Edwards species are the trees Sclerocarya birrea, Zizi- (1983), this community is classified as a for- phus mucronata, Tabernaemontana elegans est, with a closed canopy reaching heights and Strychnos madagascariensis, the grasses from 5 m for the low sand forest subcommu- Eragrostis superba, Digitaria eriantha and nity to 12 m for the tall sand forest subcom- Perotis patens, the shrubs Gymnosporia munity. This community is subdivided into

ISSN 0075-6458 23 Koedoe 47/1 (2004) Gaugris.qxd 2004/04/15 06:28 Page 24

two subcommunities based on the structure occurs on low clay content, acidic orange- and floristic composition of the vegetation. brown arenosols and dystrophic regosols. This subcommunity is characterised by diag- nostic species of Species Group M and com- 3.1 The Rothmannia fischeri-Ptaeroxy- mon species of Species Groups N and O. The lon obliquum low to short sand diagnostic species are the trees Strychnos forest henningsii, Manilkara concolor and Wright- This subcommunity is found on the dune ia natalensis, the shrubs Dovyalis crests, slopes and inter-dune depressions longispina, Commiphora neglecta and Tri- throughout the Tshanini Game Reserve on calysia capensis. Prominent species are the acidic orange-brown arenosols and dystroph- trees Ptaeroxylon obliquum, Newtonia hilde- ic regosols. It is characterised by diagnostic brandtii, Haplocoelum gallense, Ochna species of Species Group L and common arborea, Brachylaena huilensis and Mon- species of Species Groups N and O. The odora junodii, the lianas Uvaria caffra, diagnostic species are the tree Rothmannia Uvaria lucida and Hippocratea delagoensis, fischeri, the creeper Tecoma capensis and the the shrubs Ochna natalitia, Drypetes arguta, shrub Tarenna littoralis. Prominent species Toddaliopsis bremekampii, Salacia leptocla- are the trees Ptaeroxylon obliquum, Balan- da, Monanthotaxis caffra and Hyperacan- ites maughamii, Brachylaena huilensis, thus microphyllus. A mean of 39 plant Pteleopsis myrtifolia and Hymenocardia species was recorded per relevé. According ulmoides, the lianas Uvaria caffra, Uvaria to the classification of Edwards (1983), this lucida and Hippocratea delagoensis, the subcommunity is structurally a tall forest shrubs Ochna natalitia, Boscia filipes and with a closed canopy that reaches a height of Hyperacanthus microphyllus. The understo- 12–15 m for the largest and oldest trees. An ry shrub Synaptolepis kirkii has a high con- intermediate canopy of smaller trees and stancy in this subcommunity. A mean of 48 high shrubs reaches heights of 5–6 m. The plant species was recorded per relevé. ground level is covered with a semi-continu- According to the classification of Edwards ous layer of low to short herbs. The forest is (1983), this subcommunity is a low to short relatively open and easy to walk through. forest with a continuous canopy that reaches Izidine et al. (2003) suggest that such vege- a height of 5–7 m. The understory is dense tation types should be named Licuati forest. and nearly impenetrable with no intermedi- The Nhlole seepage line area was not sam- ate strata between the canopy and ground. At pled and has to be investigated further. It ground level a low, semi-continuous layer of appears that the seepage line may have a mixed grass and herbs is found. duplex soil and that the vegetation type There is a controversy regarding the forest along the Nhlole should be similar to the status of the low sand forest. The low canopy Spirostachys africana-Berchemia zeyheri associated with the absence of secondary closed woodland on duplex soils described intermediate canopy strata suggests that this by Matthews et al. (2001). This area has subcommunity should be classified as tall been mapped as woodland on clay in Fig. 2. thicket as defined by Low & Rebelo (1998). Izidine et al. (2003) suggest that such vege- Similarity indices tation types should be named Licuati thicket. The concept of some degree of similarity or dissimilarity between plant communities is 3.2 The Strychnos henningsii-Ptaeroxy- difficult to quantify. It is however, possible lon obliquum tall sand forest to use mathematical models to describe such This subcommunity is found on the dune relationships, but because of the mathemati- crests, slopes and inter-dune depressions cal nature of these models, arbitrary limits throughout the Tshanini Game Reserve. It have to be set (Mueller-Dombois & Ellen-

Koedoe 47/1 (2004) 24 ISSN 0075-6458 Gaugris.qxd 2004/04/15 06:28 Page 25

Table 3 A comparison between the vegetation of Tembe Elephant Park and Tshanini Game Reserve by means of the Sorensen Similarity Index based on species composition and that of Motyka et al. (In: Mueller-Dombois & Ellenberg 1974) based on species abundance Vegetation type Species composition: Species abundance: presence-absence data cover-abundance data Open woodland 50.9 % 31.8 % Closed woodland 53.4 % 26.8 % Sand Forest 54.6 % 39.0 %

berg 1974). Several indices have been used - The Acacia burkei-Euclea natalensis in the past to quantify the degree of similari- closed woodland and thicket subcommu- ty between homogeneous vegetation units or nity plant communities. Jaccard’s Similarity Index is a simple mathematical expression - The Salacia kraussii-Themeda triandra that is based on the ratio of the plant species open woodland on sand subcommunity that are common to two communities, and The similarity indices indicate that the three the total number of species present. main plant communities of Tshanini Game Sorensen’s Similarity Index is an evolution Reserve are similar to their equivalent plant of Jaccard’s and is also used to compare veg- communities and subcommunities described etation communities on the basis of the for Tembe Elephant Park (Table 3). These absence or presence of species. The Similar- results therefore confirm the prediction made ity Index of Motyka et al. (Mueller-Dombois by Matthews et al. (2001) that the plant com- & Ellenberg 1974) is a modification of the munities that were described for Tembe Ele- Sorensen Similarity Index and is designed to phant Park can be extrapolated to other parts treat data on vegetation structure by quanti- of Maputaland and can be used for further fying each species that is present (Mueller planning and development of Tshanini Game Dombois & Ellenberg 1974). To interpret Reserve and any other similar areas. similarity indices based on the absence or presence of species, Mueller-Dombois & The qualitative similarity in species compo- Ellenberg (1974) suggest that a 25 % up to sition between the two sets of data is excep- 50 % overlap in species indicates similar tionally high (Table 3). However, the quanti- vegetation communities, with 25 % as the tative similarity indices suggest that there are lower threshold below which communities more pronounced differences between the are considered as different. Similarity two reserves. The Tembe Elephant Park has indices exceeding 50 % show that the two been protected from human utilisation since vegetation communities are so similar that 1983, although various wildlife species were more fieldwork might be necessary to obtain introduced or re-established there. The ecologically meaningful information Tshanini Game Reserve on the other hand (Mueller Dombois & Ellenberg 1974). has been devoid of most forms of wildlife for a similar period of time, but it has been sub- The vegetation units from Tembe Elephant ject to human utilisation, cattle grazing and Park that were used in this comparison are the associated fire regime, especially in the described by Matthews et al. (2001) as: sparse to open woodland areas where grass- - The Drypetes arguta-Uvaria lucida land occurs. The leaders from the communi- subsp. virens sand forest community of ty of Manqakulane still allow cattle grazing the deep sandy areas (dry sands) on the land of the Tshanini Game Reserve.

ISSN 0075-6458 25 Koedoe 47/1 (2004) Gaugris.qxd 2004/04/15 06:28 Page 26

As a result, small cattle herds still roam the Concerning the closed woodland the similar- reserve, especially in the open woodlands. ity index is high in terms of plant species The cattle are herded to and from the centre composition but is greatly reduced in terms of the community through small tracks cut of species cover indicating a definite differ- through the bush, usually avoiding the sand ence in closed woodland structure between forest vegetation type. The cattle herders Tshanini Game Reserve and Tembe Elephant have also been observed repeatedly burning Park. The mean species cover value for the the open woodland areas in Tshanini Game closed woodlands in Tshanini Game Reserve Reserve to keep the grass green for the cat- is significantly higher than for Tembe Ele- tle. phant Park (t = 7.62, df = 9, p < 0.001). The The high similarity in plant species composi- larger mammals found in Tembe Elephant tion between the two reserves, despite the Park are unlikely to have used the Tshanini different utilisation regimes over a period of Game Reserve intensively at any stage in the 19 years is noteworthy. Although no factual past due to the absence of standing water. data on human population numbers and the Utilisation of Tshanini Game Reserve was intensity of their use of natural resources are probably limited to migratory incursions available, various authorities in the northern from larger forms of wildlife, which must Maputaland region agree that both have have been restricted in time and impact. increased over the past 19 years. The degree Small cattle herds, which have supplanted of similarity that still exists in the plant the wildlife in Tshanini Game Reserve, pre- resources suggests that either the human fer the more open woodland areas as presence and utilisation of these resources in opposed to the closed woodland and sand Tshanini Game Reserve has been consistent- forest. The cattle herds were mostly ly low, or that the wildlife utilisation in observed walking through the closed wood- Tembe Elephant Park has been high, or that land from one open woodland area to both forms of utilisation lead to the same another. It therefore appears unlikely that the pressure on the species. cattle herds could create a noticeable impact on the vegetation of the closed woodland However, the lower similarity between the two reserves when a quantitative measure is community. The lower cover values in the used suggests that there are differences in closed woodland community in Tembe Ele- terms of cover of the species. The sand for- phant Park may indicate that this vegetation est community in Tshanini Game Reserve type has been subjected to a high pressure has a significantly higher mean cover value from the wildlife since the complete fencing per species than the equivalent community in of the park in 1989. A similar phenomenon Tembe Elephant Park (t = 5.65, df = 8, p < was reported in grazing exclosures where 0.001). Although the species composition species diversity remained similar, but plant has remained relatively unchanged in both density and cover increased (Lenzi-Grillini reserves, the sand forest vegetation in the et al. 1996). In areas that are subjected to Tshanini Game Reserve is denser than in animal utilisation and especially elephant Tembe Elephant Park. The assumption made presence, a decrease in vegetation cover and by Van Rensburg et al. (1999) that the sand density is often observed. This is especially forest around the park suffers the same from true when the stocking density of animals is human pressure as from elephant impact higher than the ecological capacity, or the inside the park may not be true as the dis- areas are close to natural or artificial water- tance from the forest to human settlements ing points (Brits et al. 2002, Mapaure & increases. The present study suggests that Campbell 2002, Mosugelo et al. 2002). Fur- elephants may be affecting the sand forest in ther research is needed to compare utilisation Tembe Elephant Park and that they have pressure and patterns by cattle herds in already changed the plant cover significantly Tshanini Game Reserve with wildlife in but not the species composition. Tembe Elephant Park.

Koedoe 47/1 (2004) 26 ISSN 0075-6458 Gaugris.qxd 2004/04/15 06:28 Page 27

The similarity between the open woodland ture immediately around the park to an of the two reserves is once again noticeable increase in human utilisation of these in terms of plant species composition, but resources. The changes recorded mainly suggests a difference at the species abun- involved the opening up of the canopy in the dance level. The mean species cover value sand forest, an action they deemed was com- for the open woodlands in Tshanini Game parable to the opening up of the canopy by Reserve is significantly higher (t = 12.45, elephants in Tembe Elephant Park. df = 18, p <0.001) than for Tembe Elephant Park. This difference can possibly be linked In the rural community of Manqakulane the to the absence of wildlife in Tshanini Game situation appears to differ from that Reserve combined with the improper use of described by Van Rensburg et al. (1999), fire. It is unlikely that the small cattle herds possibly because of the remoteness of the that roam the open woodland of the Tshanini Tshanini Game Reserve as opposed to areas Game Reserve now have had a noticeable adjacent to Tembe Elephant Park. The latter impact on the vegetation. However, the high- areas are served by the main road that has er density of the woody vegetation in Tshani- become an extraction route for construction ni Game Reserve may also be interpreted as wood and especially for firewood. The vege- a sign of bush encroachment. The repeated tation adjacent to the road is also harvested burning of the grass layer of the open and for the sale of wood. Human settlements are sparse woodland by the cattle herders to found up to 2 km south of the main road that stimulate out of season growth might pro- passes along the southern boundary of the vide conditions favourable for bush Tembe Elephant Park. Impact on vegetation encroachment (Van Rooyen 2002). Cattle surrounding human settlements can be and wildlife have notoriously different observed as far as 2 km from the settlement requirements and impact on the vegetation location. (Bothma 2002). Moreover, the absence of The interpretation of the vegetation data sug- the whole range of grazers and browsers in gests that the present human pressure on the Tshanini Game Reserve combined with the Tshanini Game Reserve in northern Maputa- absence of tree damaging animals could also land is still low. There is no doubt that the provide conditions favourable for the pressure is increasing because of the popula- increase in density of the woody component tion increase over the past 10 years (H. Els, in the open woodland. The improper use of Centre for Indigenous Knowledge, Universi- fire and to a lesser extent cattle grazing as ty of Pretoria, 0002 Pretoria pers. comm.). opposed to high intensity wildlife utilisation However, local people prefer to remain close are suggested as probable causes for the dif- to the water resources, main roads and arable ference observed in species cover between lands. A low potential for agriculture, the the open woodland of the two reserves. distance from a major road and a lack of The trends observed in this study differ from water have all kept people away from those that were suggested by Van Rensburg Tshanini Game Reserve. These factors have et al. (1999) and verified by McGeoch et al. therefore contributed to the preservation of (2002) working with a suite of dung beetle this environment. The distance of the Tshani- species as bio-indicators. Van Rensburg et ni Game Reserve from the community cen- al. (1999) suggested that the impact of ele- tre, where most people of the community of phants on sand forest structure inside Tembe Manqakulane have settled, is less than 3 km. Elephant Park was comparable to the human However, this distance in such a poor com- impact outside the park. When selecting sites munity where vehicles and other means of outside conserved areas, Van Rensburg et al. transportation of goods and resources are not (1999) and McGeoch et al. (2002) concen- readily available, seems to be sufficient to trated on the populated areas within close prevent an excessive impact on the reserve proximity of Tembe Elephant Park. They from human utilisation. Most people have to attributed the change in the sand forest struc- carry the material which they harvest them-

ISSN 0075-6458 27 Koedoe 47/1 (2004) Gaugris.qxd 2004/04/15 06:28 Page 28

selves, often as loads on their head, and this Acknowledgements seems to have limited the utilisation of We are grateful to Thabani Mthembu, who assisted resources further than the immediate vicinity with the fieldwork, the people from the community of the settlements. The availability of ade- of Manqakulane for authorising the research to be quate material for human utilisation in the conducted on their land, the management staff from immediate vicinity of human settlements Tembe Elephant Park and the Ezemvelo KwaZulu- may also have contributed to the protection Natal Wildlife for the use of the research facilities of land further away. However, there are and the logistical support, the Centre for Wildlife indications that a limit of sustainable utilisa- Management, Prof. H. Els and the Centre for Indige- tion has been reached in the vicinity of the nous Knowledge from the University of Pretoria for the material and financial support of the project. This human settlements, as suggested by Van material is also based upon work supported by the Rensburg et al. (1999), and that the people in National Research Foundation under Grant Number this community may soon seek to utilise the 2047386. plant resources further away (Gaugris unpublished data 2003). References

Conclusions BOTHMA, J. DU. P. 2002. Combining wild and domestic herbivores. Pp. 201–208. In: BOTHMA, A vegetation map for Tshanini Game J. DU P. (ed.). Game Ranch Management. Preto- Reserve and its surrounding areas has been ria: Van Schaik. provided. Moreover, the plant communities BRITS, J., M.W. VAN ROOYEN, & N. VAN ROOYEN. have been described, and were compared 2002. Ecological impact of large herbivores on the woody vegetation at selected watering points with similar vegetation units in Tembe Ele- on the eastern basaltic soils in the Kruger phant Park. This comparison between simi- National Park. African Journal of Ecology 40: lar vegetation units shows that species com- 53–60. position is very similar but that plant cover COX, C.B. & P.D. MOORE. 1994. Biogeography: an values are significantly higher in Tshanini ecological and evolutionary approach. 5th ed. Game Reserve than in Tembe Elephant Park. Oxford: Blackwell Scientific Publishing. EDWARDS, D. 1983. A broad scale structural classifi- These differences will have to be investigat- cation of vegetation for practical purposes. Both- ed in more detail to establish which condi- alia 14: 705–712. tion should be regarded as representative of HENNEKENS, S.K. & H.J. SCHAMINEE. 2001. TUR- pristine conditions. The present study sug- BOVEG, a comprehensive data base management gests that the sand forest in Tshanini Game system for vegetation data. Journal of Vegeta- tion Science 12: 589–591. Reserve has not suffered degradation by HOBDAY, D.K. 1976. Quarternary sedimentation and human pressure, but that elephants may be the development of the lagoonal complex, Lake affecting the sand forest in Tembe Elephant St Lucia, Zululand. Annals of the South African Park and that they have already changed the Museum 71: 93–113. plant cover significantly but not the species IZIDINE, S., S. SIEBERT & A.E. VA N WYK. 2003. composition. In general the high intensity Maputaland's licuati forest and thicket, botanical exploration of the coastal plain south of Maputo wildlife utilisation, notably by elephant, in Bay, with an emphasis on the Licuati Forest Tembe Elephant Park is suggested as proba- Reserve. Veld & Flora 89: 56–61. ble cause for the difference observed in KIRKWOOD, D. & J.J. MIDGLEY. 1999. The floristics species cover between equivalent communi- of Sand Forest in northern KwaZulu-Natal, ties of the two reserves. The Manqakulane South Africa. Bothalia 29: 293–304. community is still reasonably small and nat- LENZI-GRILLINI, C.R., P. VISKANIC & M. MAPESA. 1996. Effects of 20 years of grazing exclusion in ural resource use is restricted to the vicinity an area of the Queen Elizabeth National Park, of the settlements. This has apparently pre- Uganda. African Journal of Ecology 34: served large areas from been overexploited. 333–341.

Koedoe 47/1 (2004) 28 ISSN 0075-6458 Gaugris.qxd 2004/04/15 06:28 Page 29

LOW, A.B. & A.G. REBELO. 1998. Vegetation of South 36 year period in northern Chobe National Park, Africa, Lesotho and Swaziland. Pretoria: Depart- . African Journal of Ecology 40: ment of Environmental Affairs & Tourism. 232–240. MACDEVETTE, D.R., D.K. MACDEVETTE, I.G. GOR- MUELLER-DOMBOIS, D. & H. ELLENBERG. 1974. Aims DON. & R.L.C. BARTHOLOMEW. 1989. Floristics & Methods of Vegetation Ecology. New York: of the Natal indigenous forests. Pp. 124–144. In: John Wiley. GELDENHUYS, E.C.J. (ed.). Biogeography of the SOIL CLASSIFICATION WORKING GROUP. 1991. Soil Mixed Evergreen Forests of Southern Africa. Classification a taxonomic system for South Pretoria: Foundation for Research Development. (Ecosystems Programmes Occasional report; Africa. Memoirs of the Agricultural Natural no 45.). Resources of South Africa No. 15. Pretoria: MAPAURE, I. & B.M. CAMPBELL. 2002. Changes in Department of Agricultural Development. woodland cover in and around Sengwa VAN RENSBURG, B.J., M.A. MCGEOGH, S.L. CHOWN Wildlife Research Area, , in relation & A.S. VAN JAARSVELD. 1999. Conservation of to elephants and fire. African Journal of Ecolo- heterogeneity among dung beetles in the gy 40: 212–219. Maputaland Centre of Endemism, South Africa. MATTHEWS, W.S., A.E. VAN WYK, N. VAN ROOYEN. Biological Conservation 88: 145–153. & G.A. BOTHA. 2001. Vegetation of the Tembe VAN ROOYEN, N. 2002. Veld Management in the Elephant Park, Maputaland, South Africa. South savannas. Pp. 571–620. In: BOTHMA, J. DU P. African Journal of Botany 67: 573–594. (ed.). Game Ranch Management. Pretoria: Van MCGEOGH, M.A., B.J. VAN RENSBURG & A. BOTES. Schaik. 2002. The verification and application of VAN WYK, A.E. 1996. Biodiversity of the Maputa- bioindicators: a case study of dung beetles in a land Centre. Pp. 198–207. In: VAN DER MAESEN, savanna ecosystem. Journal of Applied Ecology 39: 661–672. L.J.G., X.M. VAN DEN BURGT & J.M. VAN EDENBACH DE OOY MOLL, E.J. 1977. The vegetation of Maputaland—a M R (eds.). The biodiversity of preliminary report of the plant communities and African . Dordrecht: Kluwer Academic their future conservation status. Trees in South Publishers. Africa 29: 31–58. VAN WYK, A.E. & G.F. SMITH. 2001. Regions of MOSUGELO, D.K., S.R. MOE, S. RINGROSE & C. floristic endemism in southern Africa. A review NELLEMAN. 2002. Vegetation changes during a with emphasis on succulents. Pretoria: Umdaus.

ISSN 0075-6458 29 Koedoe 47/1 (2004)