Threatened Limestone Fynbos plant communities of Andrew’s Field and Tsaba-Tsaba Nature Reserve, Western Cape

The vegetation of inland plains and hills of the Andrew’s Field and Tsaba-Tsaba Nature Reserve, Bredasdorp District, Western Cape was classified using TWINSPAN and Braun-Blanquet procedures. The resulting four plant communities and nine subcommunities were described and interpreted ecologically. The vegetation was sampled using 97 randomly stratified plots. The floristic composition, Braun-Blanquet cover-abundance of each species, and various environmental variables were recorded in each sample plot. The relationship between the vegetation units and the associated environmental gradients was confirmed by ordination, using the DECORANA computer program, applied to the floristic data set. The conservation priority of each vegetation unit was determined by taking the occurrence of Red Data List species, limestone endemic species and Cape Floristic Region endemic species into consideration. TTie distribution of the plant communities can mainly be ascribed to differences in the clay/sand content of the soil and the degree of exposure of the vegetation to the dominating winds (southeastern and northwestern) of the area.


INTRODUCTION
The Andrew's Field and Tsaba-Tsaba Nature Reserve is situated in the Bredasdorp/Riversdale Centre of Endemism (CowHng 1992). This centre refers to a welldefined group of plants confined to the limestones of the Bredasdorp Formation (Moll et al. 1984) and associ ated colluvial deposits (Heydenrych 1994). Limestone Fynbos is one of the most threatened vegetation types in the Cape Floristic Region (Hilton-Taylor & Le Roux 1989). Factors threatening Limestone Fynbos vegetation include alien plants (considered to be the biggest threat to the natural environment), land clearing, resort devel opment, incorrect fire management and over-harvesting of flowers (Heydenrych 1994). Cowling et al. (1988) and Rebelo et al. (1991) indi cated that very little is known about the vegetation ecol ogy o f Limestone Fynbos. Both these authors attempted vegetation classifications based on a combination of structural characters (Campbell 1986) and dominant spe cies. Vegetation classification and description using total floristics were avoided due to the highly variable floristic structure and the very high number of species in Fynbos communities (Cowling et al. 1988). Detailed vegetation studies based on total floristics are therefore very limited or do not exist at all in Limestone Fynbos. Furthermore, 21% o f all Limestone Fynbos species are not conserved in state-owned nature reserves (Heydenrych 1994). The conservation importance o f Limestone Fynbos, which dominates the inland plains and hills, makes the effec tive management of the area all the more important, and justifies a detailed vegetation study based on total floris tics.

Location
The study area is situated in the Bredasdorp District, Western Cape, between Struisbaai North in the south and the De Mond State Forest in the north (Figure 1). The area is bordered in the west by the Bredasdorp/ Struisbaai road, and in the south by the sea. According to the geographical division of Heydom & Tinley (1980), the study area is situated on the south coast, which is a transitional zone, between tropical and temperate waters. Offshore is the major retroflex area of Agulhas Current waters, recurving eastwards and landwards (Heydom & Tinley 1980).
The study area is ± 979 ha in extent and consists of Andrew's Field (± 129 ha) and Tsaba-Tsaba Nature Re serve (± 850 ha), which are adjacent areas on the same farm (Portion 7 and the remainder of Portion 8 of the Farm Zoetendalsvlei No. 280). Approximately two thirds of the study area consists of inland plains and hills of Limestone Fynbos; the rest o f the area consists of coastal thickets. According to Low & Rebelo (1998), two veld types occur on the inland plains and hills: Limestone Fynbos and South-and Southwest Coast Renosterveld, whereas Mustard et al. (1997) also mentions Dune Asteraceous Fynbos for this coastal area. According to Mucina et al. (2005), the area consists of Overberg Dune Strandveld and De Hoop Limestone Fynbos.

Geology
Two formations of the Bredasdorp Group namely the Strandveld Formation and the Waenhuiskrans Formation are found in the study area. The Bredasdorp Group is described by Malan et al. (1994) as Cenozoic sediments of marine and marine related origin, stretching up to 22 km inland from the current coastline. The rocks are dis cordantly laid down on marine-paved rocks o f the Table  Mountain, Bokkeveld and Uitenhage Groups (Malan et al. 1994).
The Strandveld Formation, found along the coast, consists of white to pale grey dune sands with a high percentage of shell fragments. Partial cementing of sands with a high calcium carbonate content took place. The lithology of this formation can be described as white dune sand with finely divided shell and alluvial stones (Malan e/fl/. 1994).
The Waenhuiskrans Formation forms outcrops next to the current coastline. The Waenhuiskrans stratiotype is 12.4 m deep, overlain with 1 m thick calcretes locally, and consists o f medium granulated cross-layered calcarenite with well-rounded quartz and a few glauconite grains. Large-scale aeolic cross layers are character istic of the unit. The lithology o f this formation can be described as partially calcified dune sand (Malan et al. 1994).
A small portion o f the study area's mother mate rial consists of sedimentary rocks (pale grey to pale red sandy soil), underlain by the Waenhuiskrans Formation (partly calcified dune sand with calcrete lenses) (Malan etal. 1994).

Physiography
The study area is situated on the Agulhas Plain, which is a coastal lowland. Most of the study area falls below the 10 m contour with a ridge running roughly north-south reaching an ahitude of 31 m in places (Jeffery 1996). The area mostly consists of gradual, sloped lime stone hills and flat to gradual, sloped plains.

Soil
The soil is shallow and sandy, overlying limestone, or deeper and sandy, overlying clay, silt and gravel. Four different soil forms have been distinguished in the area. Coega (Orthic A on hard bank carbonate horizon), fam ily Marydale (lime containing A-horizon); Immerpan (Melanie A on hard bank carbonate horizon), family Kalkpan (lime containing A-horizon); Brandvlei (Orthic A on soft carbonate horizon), family Koike (signs of wet ness in the carbonate horizon); and Namib (Orthic A on regie sand), family Beachwood (containing lime within 1 500 mm from the soil surface) (MacVicar 1991).
The Coega soil form is found on the limestone hills and shallow-soil limestone plain. The Immerpan soil form is found on the deeper soil o f a proteoid-dominated limestone plain. The Brandvlei soil form is found in the marsh area, as well as on the renosterveld plain, whereas the Namib soil form is found on the dune plain and the deep sand found on proteoid-dominated foothills and plains.

Climate
The main factors affecting the climate are the con trasting sea surface temperatures of the two major ocean currents and the inshore circulation (Tinley 1985). The south coast has a warm temperate climate with all sea sons and bimodal equinoctial rainfall (Strydom 1992).
The study area is situated in the Southern Overberg, which is a transition between the winter rainfall region and the nonseasonal rainfall region in the east (Mustard et al. 1997). According to Heydom & Tinley (1980) the study area is situated in a low rainfall region, on an arid inland tongue reaching the coast. The low rain fall regime appears to be due to the occurrence of cold inshore waters, which inhibit shoreline rains. The average annual precipitation of the area is 444 mm. Precipitation mostly takes place in the form of rain, but also occurs in the form of fog. The maximum precipitation is in June, and the minimum precipitation during February and December. A climate diagram ( Figure 2) was compiled from data obtained from the Agulhas Weather Station.
The average annual minimum and maximum temper atures for the area are 6"C and 13.3®C, respectively. The maximum and minimum temperatures for the area were 36. r C in February and 3.9®C in June respectively.
The wind along the south coast is bi-directional; southeasterly winds alternate with northwesterly and southwesterly winds (Heydom & Tinley 1980).  The following habitat characteristics were recorded in each sample plot: altitude, topographical position, geomorphology, exposure to sun and wind, slope angle, slope direction, geology, soil, percentage rock cover and biotic influence.
All data were incorporated into a vegetation database created in TURBOVEG (Hennekens & Schaminee 2001). To obtain a first approximation of the plant communities of the area, releves were extracted from TURBOVEG into MEGATAB (Hennekens 1996) and then classified using the TWINSPAN algorithm (Hill 1979) contained within MEGATAB. These results were then refined by application of the classical Braun-Blanquet methodology (Behr & Bredenkamp 1988), using MEGATAB as the table editor. The results obtained from the classification are presented in a phytosociological table (Appendix 1). Plant names conform to Germishuizen & Meyer (2003).

METHODS
Releves were compiled in 97 plots, placed in a strati fied random way in relatively homogeneous areas, repre sentative o f particular plant communities. A plot size of 10 X 10m was used, which is considered as large enough to ensure that all species of regular occurrence in the stand are present in the sample plot (Rebelo et al. 1991). Sample plots were placed to ensure that each plot ade quately represented the structure of the particular vegeta tion (Werger 1974). A list o f plant species found in each sample plot was compiled. The cover-abundance of each species in the sample plot was assessed, using the Braun-Blanquet cover-abundance scale (Mueller-Dombois & Ellenberg 1974;Werger 1974), with the modification proposed by Barkman et al. (1964) and Westhoff & Van der Maarel (1978). Vegetation structure, based on height and plant cover classes, is described according to the structural classification of Edwards (1983).
The Edwards (1983) height classification is as fol lows:

Classification
Six main habitat types are found in the study area, namely wetlands, limestone hills, shallow limestone plains, deep-soil limestone plains, undulating dune plains with neutral sands and Renosterveld.
The plant communities are classified as follows: A diagrammatic presentation of the hierarchical clas sification and associated environmental interpretation of the plant communities is given in Figure 3. The commu nity numbers in Figure 3 correspond with the plant com munity numbers used in the text.
The distribution of the communities is shown in a vegetation map (Figure 4). Structure: high, closed grass Phragmites australis in the wetland, with high Phragmites australis, tall Juncus kraussii and some short shrubs in the area surround ing the dam. The surrounding marshy area contains tall, closed Juncus kraussii, with some short shrubs in between.

Description o f the communities
The community is situated near the comer of the northern and western borders of the Reserve. It is found in the flat wetland/marsh area close to a dam, at an alti tude o f 3 m. Small, irregular, white-grey limestone stones, originating from the Waenhuiskrans Formation of the Bredasdorp Group, and orange-brown ferricrete pebbles associated with Table Mountain Rocks (Malan et al. 1994), cover 10% o f the soil surface. Soils are of the Coega Form.

Chondropetalum microcarpum-Metalasia pungens
Low Closed Ericoid and Restioid Fynbos Community Structure: varies fi-om predominantly low, closed eri coid and asteraceous shrubs, with low restioids in abun dance and with short shrubs sparsely scattered in the area.
This widespread and diverse community, consisting of four subcommunities, is found at the western and north ern borders of Tsaba-Tsaba Nature Reserve and also in a mosaic of scattered patches in the central parts o f the study area. It occurs at an altitude of 3-23 m, at a gra dient of 0°-45°, on various habitat types, including the shallow-soil limestone plains, undulating dune plains, sandy footslopes and southerly slopes o f limestone hills.
Small, irregular, white-grey pebbles, stones and rocks from the Waenhuiskrans Formation cover 0-40% on the limestone plains and 2-60% on the limestone hills. Small limestone pebbles originating from the Strandveld Formation cover 0-5% on the dune plains.
Soils are of the Coega and Brandvlei Forms on the limestone plain, the Namib Form on the dune plain and the Coega and Immerpan Forms on the limestone hills.
This restio-dominated fynbos community is floristically different from the Restioid Fynbos recognized by Cowing et al. (1988) or Rebelo et al. (1991). It should, however, be mentioned that Rebelo et al. (1991) com bined structural and floristic characters in their analysis, and only dominant species were surveyed. This may account for some of the floristic differences between the plant communities from the current study and those from the Riversdale Plain. Many floristically different plant communities are, however, expected to be found in Restioid Fynbos from the limestone areas, as four sub communities are recognized in this study: Structure: low, closed ericoid and asteraceous shrubs, with low restioids in abundance; short shrubs are also sparsely distributed.
The subcommunity is found scattered throughout the study area and is situated at an altitude of 3-10 m on a gradient of 0°-8°, on the shallow-soil limestone plain, where pebbles, stones and rocks cover 0-30% of the soil. Soils are of the Coega and Brandvlei Forms.   Structure: mostly low to short, closed ericoid and asteraceous shrubs; taller shrubs scattered throughout the area; some tall and short restios in abundance.
The subcommunity is found scattered over the entire study area at an altitude of 0-12 m, at a gradient of 0°-26°, on the dune plains where pebbles cover 0-5% and the sandy, shallow-soil limestone plains with pebbles covering 7^0 % . Soils are of the Namib Form on the dune plains and the Brandvlei Form on the limestone plain.

Empoditun gloriosum-Chondwpetalum microcarpum
Low Structure: mostly low to short, closed ericoid and asteraceous shrubs; some taller shrubs scattered through out the area; some tall and short restios in abundance.
This subcommunity is situated near the western bor der and in the eastern middle part o f the Reserve. It occurs at an altitude of 3-23 m, at a gradient of 0°^5°, on the southeastern-and southwestem-facing slopes of limestone hills. Pebbles, stones and small rocks cover 2-60% o f the soil. Soils are of the Coega and Immerpan Forms.

Eriocephalus kingesii-Chondropetahm microcarpum
Low Structure: mostly low to short, closed ericoid and asteraceous shrubs; taller shrubs scattered throughout the area with tall and short restios in abundance.
The subcommunity occurs scattered over the entire study area on footslopes and slopes o f limestone hills and on dune plains at an altitude of 3-21 m, with gra dients of 0°-16°. Pebbles cover 1-35% of the soil. Soils are o f the Namib Form on the dune plain and Immerpan and Coega Forms on the limestone hills. The community occurs from the comer formed by the northern and western borders of the Reserve up to the middle o f the study area, running along the western border, and on the eastern section o f the northern bor der, at an altitude of 1-31 m, at a gradient of 0°-16°, on the shallow and deeps soils of the limestone plains, along the limestone hills, on the northern and northeast ern midslopes of limestone ridges, on the deep sands of footslopes as well as in the dune troughs between dunes.

Leucadendron mehdianum-Protea obtusifolia
Large, irregularly shaped, white-grey rocks, with some smaller stones and pebbles, as well as rock sheets, cover 3-65% of the soil on the ridge. Small, irregu lar white-grey pebbles and stones cover 0-7% of the limestone plains. The sand in the dune troughs have no rocks. The limestone originated from the Waenhuiskrans Formation of the Bredasdorp Group. Soils are of the Brandvlei Form on the shallow-soil limestone plain, the Immerpan Form on the deep-soil limestone plain, the Immerpan and Coega Forms on the limestone hills, and the Namib Form in the dune through areas.
The tall-, high-, short-, and low shrub strata have an average canopy cover o f 21%, 16%, 11% and 11% respectively, and an average height of 2.5 m, 1.5 m, 0.75 m and 0.4 m respectively. The forb stratum has an aver age canopy cover of 4% and an average height of 0.3 m. The high-, tall-, short-, and low restioid strata have an average canopy cover o f 0.5%, 6%, 2%, and 14% respectively, and an average height of 2.5 m, 1.5 m, 0.75 m and 0.4 m respectively. The sedge stratum has an average canopy cover o f 0.7% and an average height of 0.15 m. The short-and low grass strata have an average canopy cover o f 1% and 3% respectively, and an aver age height of 0.5 m and 0.3 m respectively. Rebelo et al. (1991) emphasized the conspicuous high cover of proteoids and restioids in this type of vegetation.
Five subcommunities could be distinguished. The diagnostic species groups of particularly subcommuni ties 3.2, 3.3 and 3.4 are rather weakly defined, though TWINSPAN delimited these subcommunities on the basis of total floristic composition, and the diagnostic species, though present with rather low constancy, are to a large extent restricted to the particular subcommunity. The subcommunity is situated near the comer of the northem and westem border o f the Reserve, adjacent to the marsh area. The subcommunity is also found at the northwestem inland border, as well as near the middle of the westem border of the Reserve, at an altitude of 3-31 m, at a gradient of 0°-16°, on shallow-soil limestone plains and along the limestone hill summit.

Amphithalea alba-Leucadendron meridianum
Large irregularly shaped, white-grey rocks, smaller stones and pebbles cover 3-5% of the soil on the lime stone hills, while pebbles are present on the limestone plain. The limestone originated from the Waenhuiskrans Formation of the Bredasdorp Group. Soils are o f the Brandvlei Form on the limestone plain and the Coega and Immerpan Forms on the limestone hills. Structure: tall, closed, proteoid-dominated shrubland fynbos.

Erica abietina-Leucadendron meridianum
The subcommunity is found along the northern bor der o f the Reserve, near the marsh area and at the northwestern border, as well as near the middle of the Reserve, at an altitude of 3-31 m, at a gradient of 0°-16°, on the summit o f limestone hills, deep soils of the limestone plains and the deep sand on the footslopes of the limestone hills. Small, irregular, white-grey pebbles and stones on the limestone plains, and large, irregular, white-grey rocks, with some pebbles and stones, as well as sheets of rock on the limestone hills cover 0-65% of the soil. Soils are of the Immerpan Form on the lime stone plains, the Immerpan and Coega Forms on the limestone hills, and the Namib Form in the sand area. The subcommunity occurs in patches from the middle section o f the northern border to the central part o f the Reserve; it also occurs in the centre of the whole o f the study area, near the western and northern borders, at an altitude o f 3-21 m, at a gradient o f 0°-16° on the north ern and northeastern midslopes and upper slopes o f lime stone hills and also on the sandy, deep-soil limestone plain. Irregular white-grey limestone pebbles, stones and small rocks and rock sheets cover 1-10 % o f the soil on the limestone plain. Large, irregular, white-grey rocks, with pebbles, stones and rock sheets, cover up to 60% on the limestone hills. Soils are of the Immerpan Form on the limestone plain, and the Immerpan and Coega Forms on the limestone hills. Structure: tall, closed, proteoid-dominated shrubland.

Metalasia muricata-Leucadendron meridianum
The subcommunity is found in patches near the cen tral part of the whole study area and in patches near the southeastern border of the Reserve, at an altitude of 1-14 m, at a gradient of 0°-8°, on the sandy sum mits o f limestone hills, and on the deep-soil limestone plain and on dune troughs. Small, irregular, white-grey pebbles and stones cover 3% of the soil on the limestone hill, but no rocks occur on the limestone plain and in the dune troughs. Soils are of the Namib Form in the dune area, the Immerpan Form on the limestone plain, and the Immerpan and Coega Forms on the limestone hills.
The absence of the widespread species of Species Group M is a further diagnostic feature. An obvious explanation for the lack of diagnostic species is the very low species richness o f this subcommunity. The species present in this subcommunity are from Species Group G (diagnostic species of the Leucadendron m eridiamim -Protea obtusifolia low to high Limestone Fynbos com munity), and other widespread species (Species Groups N, P and Q).
Prominent species: the shrubs Helichrysum patulum (Species Group G) and Passerina paleacea (Species Group P) and the restios Ischyrolepis eleocharis and Chondropetalum microcarpum (Species Group N).
The subcommunity is situated in patches along the northern border of the Reserve, adjacent to the m arsh area, in the comer of the northern and inland northwest-em border and in patches near the centre o f the whole study area, at an altitude of 3-25 m, at a gradient of 3°-8°, on the southern and southwestern deep sands on the footslopes of limestone hills and on the deep-soil lime stone plain. There are no rocks in this subcommunity. Soils are of the Namib Form on the sand areas and the Immerpan Form on the deep-soil limestone plain.
A further characteristic feature of this subcommu nity is the absence of the widespread species of Species Groups M and N. Rebelo et al. (1991) also identified a community with Protea sussanae from the Riverdale Plain, while Cowling et al. (1988)  Structure: low, closed renosterveld shrubland.
The community is found in patches near the middle of the western border of the study area, as well as adjacent to the wetland area, at an altitude of 3-18 m, at a gradi ent o f 0°-13°, on the northern and northwestern slopes of limestone hills, and on the Renosterveld plain. Small, irregular, white-grey pebbles and stones cover 1-10%.
The tree stratum has an average canopy cover o f 6% and an average height o f 2.5 m. The high shrub stra tum has an average canopy cover of 6% and an average height of 1.5 m, while the medium shrub stratum has an average canopy cover of 7%, and an average height of 0.75 m and the low shrub stratum has an average canopy cover o f 32% and an average height of 0.4 m. The high, medium and low restio strata have an average canopy cover o f 2%, 0.6% and 4% respectively, and an aver age height of 1.5 m, 0.75 m, and 0.20 m respectively. The medium grass stratum has an average canopy cover of 17% and an average height of 0.6 m while the low grass stratum has an average canopy cover of 2% and an average height of 0.25 m. The sedges have an average canopy cover of 0.5% and an average height of 0.10 m. The forb stratum has an average canopy cover of 3% and an average height of 0.15 m.
The absence of widespread Limestone Fynbos species listed in Species Groups M, N and P (Appendix 1) is a feature o f this Renosterveld community. This community is considered to be rare and threatened by encroachment from alien Acacia species (Zietsman & Bredenkamp 2006). A similar community was recognized by Cowling et al. (1988)

DISCUSSION
A large variety o f plant species grouped into various plant communities occur on the Reserve. This study indi cated that a total floristic approach can be successfully applied in the species-rich Limestone Fynbos communi ties. It is also shown that floristically derived plant com munities can be mapped at the scale of nature reserves, and that this can be applied in conservation management plans. It is, however, recognized that a total floristic approach for regional studies remains difficult within the Limestone Fynbos area (Van der Merwe 1977;Cowling et al. 1988;Rebelo et al. 1991). The use of TWINSPAN and the application o f Braun-Blanquet procedures for refinement were successfiil to delimit the communities and classify them in a hierarchical system. Four plant communities, with nine subcommunities, were identified. The plant communities could all be related to specific environmental conditions and are therefore floristically and ecologically distinguishable and interpretable. These plant communities are interpreted as separate ecosystems (Bredenkamp & Brown 2001), they therefore represent different management units, and can be incorporated in a veld management program.
The communities are well defined by diagnostic spe cies, though diagnostic species for the subcommunities often have low constancy values. However, these sub communities are retained because they were delimited by TWINSPAN, based on total floristic composition, and the diagnostic species are largely restricted to the particu lar subcommunities.

CONSERVATION
All the plant communities of the inland plains and hills are considered to have a high conservation value. Community 1, the Phragmites australis-Juncus kraussii tall to high closed Reed Wetland is a sensitive wetland system that could easily be disturbed and should there fore be protected (Doust & Doust 1995;Van Wyk et al. 2000). The conservation o f wetlands is furthermore enforced by law (Environment Conservation Act, 1989: Act No. 73 of 1989and National Water Act, 1998. 36 o f 1998).
Among the terrestrial plant communities identified in the study area, nine species encountered are listed in the Red Data List o f southern African plants (Hilton-Taylor 1996) and a further four are listed in the later Southern African plant Red Data List (Golding 2002), giving a total of 13 Red Data plant species. Furthermore, 23 lime stone endemic species were present. From a PRECIS list (SANBI 2006), 38 from 238 species collected from the 3420CC grid, were Red Data species. The Agulhas plain, on which the study area is situated, is considered, from a botanical viewpoint, as an area of high irreplaceability and high vulnerability, being rich coastal lowland with remnant patches o f coastal renosterveld and low land fynbos, which are considered among the highest priorities for conservation in South Africa and globally (Schwegler 2003).
Heydenrych (1994) listed 110 plant species endemic to limestone outcrops. The Agulhas Plain Centre, one of the six phytogeographic centres of the Cape flora, contains a total o f 1 374 species (Goldblatt & Manning 2000). O f the 285 species recorded during the plant surveys in the Reserve, 138 species are endemic to the Cape, giving a figure o f 48% endemism for the study area.
It can clearly be seen that the study area is one of exceptional species richness and conservation impor tance. Communities 2, 3 (Limestone Fynbos) and 4 (Renosterveld) should be protected due to the endemic and Red Data plant species present. Furthermore, large areas o f renosterveld have been destroyed, mainly because of agricultural land clearing (Low & Rebelo 1998).

CONCLUSIONS
Andrew's Field and Tsaba-Tsaba Nature Reserve are o f great conservation significance. This area comprises an important part o f a unique natural floral heritage found at the southern tip o f Afiica, with several endemic and Red Data species, and it should be protected and conserved to maintain its biodiversity for future genera tions.
The resulting classification could provide a useful tool, not only for the management of the plant communi ties o f Andrew's Field and Tsaba-Tsaba Nature Reserve, but also for similar vegetation areas, e.g. the Agulhas National Park, found in the surrounding region.

ACKNOWLEDGEMENT
Mrs F. Siebert is thanked for her assistance during data processing.